UNIV. OF
fORONTO
LIBRARY
or
BINDING LIST VIAR 1 5
PSYCHOLOGICAL RXVUW PUBLICATION
Psychological Review
SDITKD BY
HOWARD C WARREN, PRINCETON UNIVERSITY
JOHN B. WATSON, NEW YORK CITY (J.ofExp. Psyckol.)
JAMES R. ANGELL, 522 FIFTH AVENUE, NEW YORK (Monograph)
SHEPHERD I. FRANZ, Govr. HOSP. FOR INSANE (Bullttin) AND
MADISON BENTLEY, UNIVERSITY or ILLINOIS (Indtx)
ADVISORY EDITORS
R. P. ANGIER, YALE UNIVERSITY; MARY W. CALKINS, WELLESLEY COLLEGE ; H. N.
GARDINER, SMITH COLLEGE; JOSEPH JASTROW, UNIVERSITY OF WISCONSIN; C. H.
JUDD, UNIVERSITY OF CHICAGO ; ADOLF MEYER, JOHNS HOPKINS UNIVEPSITY ; W. B.
PTLLSBURY, UNIVERSITY OF MICHIGAN; C. E. SEASHORE, UNIVERSITY OF IOWA ; G. M.
STRATTON, UNIVERSITY OF CALIFORNIA; MARGARET F. WASHBURN, VASSAR COLLEGE.
VOLUME XXVII, 1920
PUBLISHED BI-UOMTHLT BY
PSYCHOLOGICAL REVIEW COMPANY
41 NORTH QUEEN ST., LANCASTER, PA,
AMD PRINCETON, N. J.
f«l«i »rl M ••rood -riiMt matter July 13, 1897. at the post-office At T «nra«t«r, P«u, M<Ur
Act of Congress :rf March 3. 1879
P7
I/.J-7
PRESS OF
THE NEW ERA PRINTING COMPANY
LANCASTER. PA.
CONTENTS OF VOLUME XXVII
January
The Absolute Limits of Color Sensitivity and the Effect of Intensity of Light on the
Apparent Limits. C. E. FERREE and GERTRUDE RAND, i.
Some Factors in the Perception of Relative Motion. A Preliminary Experiment. H.
A. CARR and M. C. HARDY, 24.
A New Objective Test for Verbal Imagery Types. SAMUEL D. ROBBINS, 38.
A Functional Interpretation of Human Instincts. J. R. KANTOR, 50.
Immobility: An Inquiry into the Mechanism of the Fear Reaction. J. P.
M'GoNiGAL, 73.
March
Changes in Some of Our Conceptions and Practices of Personnel. WALTER DILL
SCOTT, 81.
An Analysis of Effort. JOHN J. B. MORGAN, 95.
A Comparison of Complete versus Alternate Methods of Learning Two Habits. J.
F. DA SHI ELL, 112.
The Tonal Manifold. R. M. OGDEN, 136.
Is Lack of Intelligence the Chief Cause of Delinquency? CURT ROSENOW, 147.
May
Manifold Sub-theories of " The Two Factors." C. SPEARMAN, 159.
General versus Group Factors in Mental Activities. GODFREY H. THOMSON. 173.
Suggestions toward a Scientific Interpretation of Perception. J. R. KANTOR, 191.
Instinct and Purpose. EDWARD CHACE TOLMAN, 217.
Brain Mechanisms and Mental Images. S. BENT RUSSELL, 234.
July
The Modification of Instinct from the Standpoint of Social Psychology. WALTER S.
HUNTER, 274.
The Nature of the Rhythm Experience. ELCANON ISAACS, 270.
A New Point of View in the Interpretation of Threshold Measurements in Psycho-
physics. GODFREY H. THOMSON, 300.
The Correlation between Interests and Abilities in College Courses. JAMES W.
BRIDGES and VERONA M. DOLLINGER, 308.
Visual Phenomena in the Dreams of a Blind Subject. RAYMOND H. WHEELER, 315.
September
The Physical Basis of Nerve Functions. LEONARD THOMPSON TROLAND, 323
Theories of the Will and Kinaesthetic Sensations. RAYMOND H. WHEELER, 351.
A Pursuit Pendulum. WALTER R. MILES, 361.
The Limits of Color Sensitivity: Effect of Biightness of Preexposure and Surround-
ing Field. C. E. FERREE and GERTRUDE RAND, 377.
iii
IV
CONTENTS
November
Do We Think in Words? ARTHUR S. OTIS, 399.
A Behavioristic Account of Sleep. CHARLES H. WOOLBERT, 420.
The Compensatory Function of Make-believe Play. EDWARD S. ROBINSON, 429.
The Control of Attitude in Psychophysical Experiments. EDWIN G. BORING, 440.
The Physical Measurement and Specification of Color. LOYD A. JONES and PRENTICE
REEVES, 453.
Suggestions Looking toward a Fundamental Revision of Current Statistical Pro-
cedure, as Applied to Tests. SYDNEY L. PRESSEY, 466.
VOL. 27. No. i January, 1920
THE PSYCHOLOGICAL REVIEW
BY C. E. FERREE AND GERTRUDE RAND
Bryn Mawr College
INTRODUCTION
In describing a general plan of investigating the chro-
matic sensitivity of the peripheral retina in an earlier paper
(1) the following were mentioned as two of the problems
which we wished to take up: (a) a point to point determination
of comparative sensitivities to the different colors from the
center to the periphery, and (b) an investigation of the limits
of sensitivity. The former of these problems has been made
the subject of a recent paper (2). The latter will be treated
of here.
The investigation of the limits of sensitivity may be
considered from two points of view. As indicated in the
first of the papers referred to above, (a) it may be made a
part of the investigation of the comparative sensitivities of
the peripheral retina to the different colors; and (b) it may
be considered more specifically in relation to points of theory.
In the former case the limits should be obtained with stimuli
equalized in energy. The results will then represent positions
on the retina at which the stimuli for one of the intensities
which it is possible to employ have the same or nearly the
same threshold value.1 In the latter case the problem con-
1 Strictly speaking the threshold value may be considerably less at this point
than the intensity of the stimulus employed, because the stimulus may be increased
much above the threshold value in the far periphery of the retina without changing the
limits by a detectable amount. That is, the stimulus value of the just noticeable
I
2 C. E. FERREE AND GERTRUDE RAND
sidered in relation to its historical development divides into
two, — a determination of the relative or apparent limits and a
determination of the absolute limits. In the second of the
papers referred to above it was shown that the determination
of the apparent limits was given an undue importance in
relation to theory by Hess and his followers because of their
failure to realize that great irregularity and not uniformity
characterizes the decrease of sensitivity from the center to
the periphery of the retina. The details of that demonstra-
tion need not be repeated here. The determination of the
absolute limits of sensitivity, however, does sustain an im-
portant relation to theory, especially to theories of the
paired process type; for if it be found that sensation can
be aroused farther out from the center of the retina for one
of the paired colors than for the other, that fact must tell
against the theory unless some supplementary concept is
provided to explain the discrepancy. For one thing we have
undertaken, therefore, to determine the limits of sensitivity
with stimuli any further increase in the intensity of which
tends to decrease rather than to increase the chromatic
component of the response. For another we have determined
the effect of a given range of variation of intensities on the
apparent limits. Our reason in part for doing this was to
supplement at higher intensities the work of former papers (3)
in which we called attention to the large variations that are
required in any of the factors influencing the chromatic
response (intensity being the most effective of these) to
change the limits of sensitivity as much as I degree especially
when a certain degree of excentricity has been reached,
pointing out in particular with regard to the work of previous
writers, (a) the importance of taking into account deviations
of 1-3 degrees from coincidence of limits when conclusions
with regard /to comparative sensitivities are to be drawn
from the results, and (b) the futility of making a brightness
equalization of the stimuli, with its attendant disadvantages,
difference in limits is much greater than the stimulus value of the difference limen of
intensity. In other words, a given point in the peripheral retina may be considered
the limit for a range of stimulus intensities, varying in magnitude with its degree of
excentricity.
ABSOLUTE LIMITS OF COLOR SENSITIVITY 3
for the determination of the limits with lights of medium and
high intensities and perhaps for any but intensities so low as
to give very narrow limits. For the latter point three
reasons may be given, (i) The brightness equation does
not equalize the stimuli in power to arouse the chromatic
response, the only subjective equation, so far as we can see,
that could rightly be given a place in a determination of the
limits of chromatic sensitivity, and this only in a determina-
tion of whether or not the same ratio of sensitivity holds for
the limits as for the center or other point at which the equa-
tion was made. (2) It does not equate them in intensity
(the equation is merely of the very selective achromatic
response to the stimuli). And (3) so far as the effect of the
achromatic on the chromatic component of the excitation is
concerned (the final variable that might be considered), it
has already been shown by one of the writers in a previous
paper (4) that there is not enough of this effect for the colors
ordinarily used to change the limits by a detectable amount.
The particular bearing of the present work on this question
is to give a clearer and more definite idea of just how much
difference in intensity or equivalent influence is required to
change the limits by detectable amounts in the mid and far
periphery of the retina. In the beginning this was in fact
our chief incentive to undertake the work.
THE PROBLEM
The investigation was given the following form, (i) An
attempt was made to find out whether by means of our
spectroscopic apparatus, which was designed especially to
give high intensities of light, stimuli could be obtained which
could be sensed as color to the limits of white light vision.
(2) The effect on the extension of the limits of sensitivity of
varying the stimuli through quite a wide range of intensities
was investigated. And (3) the determination of the limits
was made in 16 meridians with all of the lights made equal
in energy to the blue of the prismatic spectrum employed
and with 1/32 of this amount of energy.
4 C. E. FERREE AND GERTRUDE RAND
CONDITIONS UNDER WHICH THE WORK WAS DONE
The conditions under which the work was done fall
under five headings: (i) the wave-lengths of light employed
and the means used of getting greater purity of light than is
found in the prismatic spectrum; (2) the energy content of
the stimuli used and the method of measurement; (3) the
control of the brightness of preexposure and surrounding
field; (4) the control of the general illumination of the optics
room; and (5) the method of rendering the amount of light
entering the eye independent of variations in the size of the
pupil, without the use of an artificial pupil. These conditions
are so nearly identical with those used in the work of the
immediately preceding papers that at the request of the
Editor space has not been taken for their repetition in the
present paper. For a description of the conditions the
reader is referred to 'Chromatic Thresholds of Sensation
and their Bearing on Color Theory, Part I.,' this journal,
1919, 26, pp. 18-25.
The stimulus used was the circular aperture of the cam-
pimeter, 15 mm. in diameter, filled with light by the focusing
lens. At a distance of 25 cm. from the pupil of the eye, on
which the light from the objective slit of the spectroscope was
focussed, this aperture subtended a visual angle of 3° 26'. The
time of exposure was I sec. and the interval between exposures
varied between 3-5 min. depending on circumstances and
the need for precautionary measures. If the stimulus was
sensed in its proper color at any time during the I sec. in-
terval of exposure, the retina was called color sensitive at
that point. (At the limits of white light vision the red stimu-
lus, for example, of the intensity used was sensed as a tint of
red.) The field in the 16 meridians was always mapped for
one color before the work on another color was begun.
Systematic results were obtained for all of the points of
the work for only one observer. This was the observer
whose results were published in the immediately preceding
papers: 'Chromatic Thresholds, etc., Parts I. and II.' For
data with regard to the various ways in which the normality of •
both the chromatic and achromatic sensitivity of this observer,
ABSOLUTE LIMITS OF COLOR SENSITIVITY 5
central retina, and chromatic sensitivity, peripheral retina,
has been confirmed, the reader is referred to pp. 26-32 of the
first of the papers noted above. Data on additional points,
important in a general specification of the ocular condition
of the observer, have also been published in various places:
e.g., on the dioptric or refraction condition and power to
sustain acuity in Trans. Ilium. Eng. Soc.y 1915, 10, p. 1128,
and in other papers by us on lighting in relation to the eye;
on muscle strength, muscle balance, muscle lag, photopic
acuity, near point, range of accommodation, and refraction
condition (more recent), Trans. Amer. Ophthal. Sor., 1918,
66, pp. 142-163; and on scotopic acuity and amount and
rapidity of scotopic adaptation, Trans. Amer. Ophthal. Soc.,
1919, 67 (in press). The more important points such as
the coincidence of the limits of red, yellow and blue with
the limits of white light vision; the narrower limits for green;
the interlacing of limits for stimuli of medium intensity of
equal energy, or of the same general order of intensity; and
the large differences in amount of light required to change
the limits of sensitivity by a detectable amount in the mid
and far peripheral portions of the retina, have been con-
firmed in a less detailed and systematic way by one or more
check observers.
RESULTS
The following results were obtained, (i) It was quite
easy to obtain an intensity of light for the red, yellow and
blue wave-lengths that could be sensed to the limits of white
light vision. In fact these wave-lengths in the spectrum
employed were considerably above the threshold at the
limits of white light vision in the sixteen meridians investi-
gated. The limits of the green of this spectrum, however,
fell far short of the limits for white light; nor could the zone
of sensitivity be widened as much as I degree by increasing
the current in the Nernst filament from 0.6 to O.8 ampere.
The energy entering the eye from the spectrum of the Nernst
filament operated by 0.6 ampere of current with the width
of collimator slit employed was for the red 9096.639 x io~10
o C. E. FERREE AND GERTRUDE RAND
watt; for the yellow, 4065.624 x io~10 watt; for the green,
1562.388 xio~10 watt; and for the blue, 882.025 x io~10 watt.
The energy value of the threshold at the limits of white
light vision in the nasal meridian, for example, was for the
red 277.836 x io~10 watt; for the yellow, 268.95 x IO~10 watt;
and for the blue, 264.368 x io~10 watt. The intensity of
light for these colors in the 0.6 ampere spectrum was, there-
fore, strongly supra-liminal at the limits of white light
vision, as is stated above. In the 0.6 ampere spectrum,
the energy of the green light, it will be noted, was greater
than the energy of the blue, but less than the energy of the
red and yellow. It was, however, nearly six times as great
as the threshold value of these colors at the limits of white
light vision. Moreover, when the current was raised to 0.8
ampere this value was considerably increased, but there
was still no detectable extension of the limits. Since
then the sensitivity to green at the center of the retina
and for several degrees towards the periphery is approxi-
mately the same as to blue and considerably greater than
to red and to yellow, and since so large an increase in the
energy value of the stimulus made no detectable difference
in the limits and any further increase lessened rather than
increased the chromatic component of the response, it seems
highly improbable that the limits could by any means what-
soever be extended the 20—35 degrees needed to make them
coextensive with the limits of white light vision. It seems
fairly certain, therefore, that while the far periphery of the
retina is only deficient in its chromatic sensitivity to red,
yellow and blue, the blindness to green for the observers used
is absolute.
(2) In the investigation of the effect of changes of in-
tensity on the limits of sensitivity eight intensities were
used, sustaining to each other the following relations: I, 1/2,
1/4, 1/8, 1/16, 1/32, 1/64 and 1/128. The highest intensities
were taken respectively from the prismatic spectrum of a
Nernst filament operated by 0.6 ampere of current and
from a spectrum made equal in energy to the blue of this
spectrum. These spectra will be designated as Spectrum
ABSOLUTE LIMITS OF COLOR SENSITIVITY 7
A and Spectrum B. The reductions were produced by means
of an aluminum sectored disc of 180, 90, 45, etc., degrees
open sector. The energy values of the different intensities
of light, as has already been stated, were obtained by radio-
metering the highest intensities and computing the lower
from the simple law of the disc. It had been our intention
to make the investigation systematically with the eight dif-
ferent intensities in the sixteen meridians of the retina.
However, for the purpose of the present paper a briefer
substitute plan has been adopted. A preliminary investi-
gation was made with the eight intensities of Spectrum A
in two meridians of the retina, the nasal and the temporal,
which meridians represent opposite extremes with regard to
breadth of zone of sensitivity, in order to get some idea of
the amounts of reduction that would be needed to be effective
in changing the limits. It was found, for example, that a
reduction of the red light to 1/32 of its value at intensity A
was not sufficient to narrow the limits in the nasal and
temporal meridians, the meridians designated in the tables
and charts as 90 degrees. At this value the stimulus was
still slightly supra-liminal in these meridians at the limits of
white light vision. This amount of reduction, however, was
sufficient to narrow the limits for the other stimuli by quite
considerable amounts. Also a further investigation showed
that it was enough to narrow the limits for red in 12 out of
the 16 meridians employed. It was decided, therefore, to
make the final determinations in the 16 meridians with the
full intensities A and B and with 1/32 A and B. The amount
of narrowing for the yellow of the prismatic spectrum in the
different meridians produced by this reduction ranged from
3-11 degrees; for the green from 5-17 degrees; for the blue,
from 10-18 degrees; and for the red, from 0-8 degrees. For
the equal energy spectrum the amount of narrowing for the
yellow ranged from 5-18 degrees; for the green, from 5-15
degrees; for the blue, from 4-18 degrees; and for the red from
3-25 degrees.
(3) In case of the equal energy spectrum of the higher
intensity, all of the lights with the exception of the green
C. E. FERREE AND GERTRUDE RAND
were seen in their proper color to the limits of white light
vision in each of the 16 meridians. Made equal in energy
to the blue of the prismatic spectrum (882.025 x io~10 watt)
the red and yellow were considerably less in energy value than
was the green of the prismatic spectrum, still the red and yellow
were sensed to the limits of white light vision while the green
which represented a considerably greater amount of energy
fell short of those limits by amounts varying from 20-35
degrees in the different meridians. There can be no reason-
able doubt, we believe, that the difference found here repre-
sents an actual difference in sensitivity. It obviously can
not be attributed to the relative intensities of the stimuli
employed.
Landolt has also investigated the effect of high intensities
on the extension of the limits of sensitivity. Writing of this
work (5), he says: "In ein absolut dunkles Zimmer fiel nur
durch eine kleine Offnung im Finsterladen directes Sonnen-
licht. Dieses wurde auf das ausserste Ende des Perimeter-
bogens gelenkt. Wahrend wir unser Auge ins Centrum des
Bogens setzen, bracht man in die kleine, intensive beleuchtete
Stelle farbige Papiere von moglichster Intensitat der Farbung.
Nun bewegtet sich das Auge langsam vom entgegengesetzen
Ende des Bogens nach Scheitelpunkte zu und es zeigte sich
dabei, dass wenigstens mit der innern Netzhautpartie alle
Farben schon bei 90° erkannt wurden. Die Grosse des
Objectes betrug weniger als I cm2.
"Als dieselben Priifungen auch mit Spectralfarben zu
machen, entwarfen wir ein Sonnenspectrum im sonst dunkeln
Zimmer und liessen es durch eine achromatische Linse auf
einen Ende des Perimeters befindlichen Schirm fallen.
Dieser hatte eine verandliche Spake, mittelst welcher man
die einzelnen- Farben aus dem Spectrum isolieren konnte.
Wahrend wir nun wiederum nach langer Adaptation, und
bei verbundenem zweiten Auge das eine Ende des Bogens
fixierten, wiirde von einem Assistenten irgendeine Farbe des
Spectrums auf die Spalte gelenkt, und wir drehten nun,
unter stehter Fixation unserer Fingerspitze, welche sich auf
dem Bogen bewegte, das Auge allmahlig der Farbe entgegen.
ABSOLUTE LIMITS OF COLOR SENSITIHTY
Es zeigte sich auch hier wiederum dass alle Farbe schon bei
90° erkannt werden, wenn sie intensiv genug sind."
Landolt's investigation was made, it will be noted, in a
dark room while ours was made in a light room. We have
TABLE I
A. THE EFFECT OF INTENSITY OF STIMULUS ON THE LIMITS OF SENSITIVITY,
PRISMATIC SPECTRUM
In this table are given the results of a preliminary investigation in two representa-
tive meridians to show how much reduction is needed to produce a significant change
in the limits of sensitivity. Eight intensities of stimulus were used: A, I/2A, I/4A.
1/8 A, etc.
Meridian Investi-
gated
Stimulus
Limits of Sensitivity for
Inten-
sity
A
Inten-
sity
JA
Inten-
sity
JA
Inten-
sity
*A
Inten-
sity
AA
Inten-
SP
Inten-
sity
AA
Inten-
sity
i*iA
Nasal
Red
(670 MM)
92
92
92
92
92
92
88
86
Yellow
(581 MM)
92
92
92
92
91
89
88
88
Green
(522 MM)
69
69
69
66
63
62
59
56
Blue
(468 MM)
92
92
87
83
79
78
77
76
Temporal
Red
(670 MM)
61
61
6l
61
61
61
46
44
Yellow
(581 MM)
61
61
61
61
55
47
46
45
Green
(522 MM)
45
45
45
42
34-5
32.5
30
29
Blue
(468 MM)
61
61
56
45
43-5
43
43
43
B. THE ENERGY VALUES OF THE STIMULI USED
Total energy of light at campimeter opening and at eye
(watt X 10-'°)
Intensity
Red (670/1/1)
Yellow (sSiw.)
Green (saaMM)
Blue (468,1,,)
A»
9096.639
4065.624
1562.388
882.025
1 The energy values of 1/2, 1/4, 1/8, 1/16, 1/32, 1/64 and 1/128 A may be obtained
by dividing the above values by the appropriate factor.
The energy density at the campimeter opening (watt per sq. mm.) may be ob-
tained by multiplying the above values by 0.005659; the energy density at the eye,
by multiplying them by 0.303.
not as yet had opportunity to repeat the work of the present
paper with the dark adapted eye. However, determinations
somewhat rougher and less detailed than those described
10 C. E. FERREE AND GERTRUDE RAND
here have sufficed to show that for our observers the far per-
iphery of the retina is color-blind to green also with the dark
adapted eye. With reference to the relative insensitivity
of the peripheral retina to green, it may further be noted
that in our results with the Hering papers with a different
set of observers the limits for green fell much nearer to the
center of the retina than for red, yellow and blue. The
results represented in Fig. 5, for example, were taken from
this series of observations. That the limits for green are
narrower than for red, yellow and blue with stimuli of the
same order of intensity has, moreover, been verified many
times in the work of our undergraduate laboratory.
In Table I., A, are given the results of the preliminary
investigation in the nasal and temporal meridians to find
out whether an intensity of light may not be gotten sufficiently
high to make the limits of color sensitivity coincide with the
limits of white light vision, and once this intensity is attained
how much reduction is needed to produce a significant
narrowing of the limits. We have already indicated in this
and in previous papers the large changes of intensity that
are needed to change the limits by a significant amount when
a certain degree of excentricity has been reached. How very
large these changes have to be for the far periphery of the
retina is shown in this table.
In Table I., B, is given a specification of the energy values
of the stimuli used in making the determinations represented
in Table I., A. Four energy values may perhaps be consid-
ered of importance for each determination: The total value
at the campimeter opening, the density per sq. mm. at the
campimeter opening, the total energy entering the eye, and
the density per sq. mm. at the eye. For the sake of brevity,
however, only qne of these values is given in the table,
namely, the total energy entering the eye; and the factors
needed to convert this value into density at the eye and at
the campimeter opening are appended in a footnote. Since
all of the light from the campimeter opening is focused into
the image on the pupil, the figures expressing the total energy
at the eye and at the campimeter opening are the same.
ABSOLUTE LIMITS OF COLOR SENSITIVITY
II
The most important of the four specifications noted are
probably the total amount of light entering the eye and the
density at the campimeter opening. The latter value, for
example, sustains a fixed but unknown ratio to the density
of the image formed on the retina.
TABLE II
THE BRIGHTNESS VALUES OF PREEXPOSURE AND SURROUNDING FIELD
In this table are given the brightness values of preexposure and campimeter
screen in candlepower per square inch1 for the determination of limits given in Table I.
In all cases in which it was possible the brightness of the preexposure and campimeter
screen was made equal to that of the stimulus at the limits of sensitivity.
Meridian
Stimulus
Brightness Value of Preexposure and Campimeter Screen for
Intensity
A
Intensity
JA
Intensity
JA
Intensity
»A
Intensity
AA
Intensity
AA
Intensity
AA
Intensity
i».A
Nasal
Red
(670 MA*)
O.O5O88
0.05088
0.05088
0.05088
0.05088
0.05088
0.03093
3.O2II6
Yellow
(581 MM)
O.OJOSS
0.05088
0.05088
0.05088
0.05088
0.05088
0.03663
3.02686
Green
(522 MM)
O.05O88
O.05088
O.O5O88
0.03663
0.03663
O.O2686
0.02II6
0.01384
Blue
(468 MM)
0.05088
0.05088
0.03663
0.02686
0.01262
0.01140
0.00643
0.00578
Temporal
Red
(670 MM)
0.05088
0.05088
0.05088
0.05088
0.05088
0.05088
3.03093
O.OI79I
Yellow
(58 1 MM)
0.05088
0.05088
0.05088
0.05088
0.05088
0.05088
0.02686
0.02116
Green
(522 MM)
O.O5O88
0.05088
0.05088
0.03663
O.O2686
0.01384
O.OII4O
0.01140
Blue
(468 MM)
0.05088
0.05088
0.03663
0.03093
0.01262
0.01140
0.00716
0.00643
In Table II. are given the brightness values of the pre-
exposure and campimeter screen for the work represented in
Table I. As stated earlier in the paper the preexposure and
the campimeter screen were selected from the Hering series
of standard papers. In case of the higher intensities of light
used, No. I of this series (the standard white, coefficient of
reflection about 75 per cent.) reflecting the light of the room
was not as bright as the stimulus light. These cases may be
identified in this and the following tables by the brightness
value of this paper under the illumination of the room, namely,
0.05088 cp. per sq. in.
1 The above values may be converted into millilamberts by multiplying by 486.8.
12
C. E. FERREE AND GERTRUDE RAND
In Table III. are given the limits of sensitivity in 16
meridians of the retina for the highest intensity of light used
for the work of Table I., intensity A, and for 1/32 A, an
intensity representing the order of reduction needed for all
of the colors to produce any considerable narrowing of the
TABLE III
THE EFFECT OF INTENSITY OF STIMULUS ON THE LIMITS OF SENSITIVITY, PRISMATIC
SPECTRUM
In this table are given the limits of sensitivity in 16 meridians of the retina for
intensity A and 1/3 2A of Table II. The upper vertical meridian is numbered o°
and the lower vertical 180°. Reading down to right or left they are 25°, 45°, 70°,
90°, 110°, 135°, 155°, and 180°.
Limits of Sensitivity for
Meridian Investi-
Intensity A
Intensity ^SA
gated
Red
Yellow
Green
Blue
Red
Yellow
Green
Blue
(6yoMM)
(58iMM)
(522/1/1)
(468/1/0
(670/1/0
(581/1/0
(5221111)
(468/iM)
0°
65
65
36
65
64
57
29
5°
Nasal 25°
86
86
49
86
83
79
36
72
45°
90
90
52
90
89
81
38
76
70°
92
92
67
92
92
86
59
77
90°
92
92
69
92
92
89
62
78
110°
9i
9i
68
9i
90
87
59
78
135°
88
88
61
88
87
84
49
75
155°
86
86
47
86
84
78
35
67
1 80°
57
57
36
57
53
47
27
42
Temporal 25°
65
65
44
65
64
SS
31
52
45°
6S
65
SO
65
57
56
33
Si
70°
62
• 62
46
62
SS
50
35
44
90°
61
61
45
61
61
47
32-5
43
110°
58
58
37
58
57
49
3i
48
135°
55
55
32
55
55
48
27
20
155°
54
54
30
64
5°
43'
25
41
limits at the extreme periphery of the retina. The 16
meridians used are designated as follows. The upper vertical
meridian is marked o and the lower vertical 180 degrees.
Beginning with o and reading down to left or right they are
o, 25, 45, 75, 9p, no, 135 and 180 degrees. The specification
of the energy of the stimuli at intensity A and 1/32 A is
given, it will be noted, in Table I., B. For all of the stimuli
at intensity A, No. I of the Hering series of papers was used
as preexposure and campimeter screen, as has already been
noted. This paper illuminated by the light of the room was too
dark (0.05088 cp. per sq. in.) for all of the four colors at the
ABSOLUTE LIMITS OF COLOR SENSITIVITY
»3
limits of sensitivity. However, for the want of a suitable
pigment surface of higher reflection coefficient it was used.
For intensity 1/32 A, it was darker than the yellow stimulus
TABLE IV
A. THE EFFECT OF INTENSITY OF STIMULUS ON THE LIMITS OF SENSITIVITY, EQUAL
ENERGY SPECTRUM
In this table are given the limits of sensitivity in 16 meridians of the retina for
stimuli all made approximately equal in energy to the blue of the prismatic spectrum
used in Table I., and for 1/32 of this intensity, — intensity B and 1/32 B.
Limits of Sensitivity for
Meridian Investi-
•? atcd
Intensity B
Intensity ^B
Red
Yellow
Green
Blue
Red
Yellow
Green
Blue
(6/OMM)
(58IMM)
(S«MM)
U68MM)
(670^/1)
(58lMM)
(5"MM)
<4«W)
0°
65
65
34
65
62
56
29
50
Nasal 25°
86
86
47
86
74
70
34
72
45°
90
90
SO
90
74
78
36
76
70°
92
92
63
92
78
85
SO
77
90°
92
92
69
92
81
87
59
78
110°
91
91
64
91
79
85
58
78
i3S°
iS5°
88
86
88
86
8
88
86
It
80
74
42
32
I5
67
1 80°
57
57
31
57
46
39
25
42
Temporal 25°
65
65
39
65
57
54
30
52
45°
65
65
39
65
47
55
32
51
70°
62
62
42
62
46
48
30
44
90°
61
61
41
61
40
45
30
43
110°
58
58
36
58
45
45
29
48
135°
SS
55
28
SS
49
41
25
40
155°
54
54
26
54
47
38
24
B. THE ENERGY VALUES OF THE STIMULI USED.
Total energy of light at campimeter and at eye
(watt X icr10)
Intensity
Red (670 fifi)
Yellow (581 w.)
Green (5221 /»M)
Blue (468 MM)
B1
891.050
882.510
884.946
882.025
and approximately equal to the green and red. For the blue
stimulus Nos. 9-14 of the Hering series (0.01404-0.0114 cp.
per sq. in.) were used as needed in the different meridians.
The photometric values for these intensities and intensity
1 The energy value of 1/32 B may be obtained by dividing the above values by
the appropriate factor.
The energy density at the campimeter opening (watt per sq. mm.) may be ob-
tained by multiplying the above values by 0.005659; the energy density at the eye,
by multiplying them by 0.303.
C. E. FERREE AND GRETRUDE RAND
B and 1/32 B have not been given in detailed tabular form
because of the large number of repetitions that occur.
In Table IV., A, are given the limits of sensitivity in
16 meridians of the retina for the four stimuli all made equal
in energy to the blue used in Table I. and for 1/32 of this value.
These values are, as we have already indicated, designated in
70
7C
155
155
FIG. i. The effect of intensity of stimulus on the limits of sensitivity, prismatic
spectrum. In this chart are represented the limits of sensitivity for intensity A of
Table I.: red 9096.639, yellow 4065.624, green 1562.388, and blue 882.025 watt X io~10.
the table as intensity B and 1/32 B. In Table IV., B, are
given the energy values of the stimuli used for Table IV., A.
For the higher intensity of these stimuli, intensity B, No. I
of the Hering series of papers (0.05088 cp. per sq. in.) was
used for the preexposure and campimeter screen. Again it
was darker than all of the four colors at the limits of sensitivity.
For intensity 1/32 B it was slightly darker than the yellow.
ABSOLUTE LIMITS OF COLOR SENSITIVITY »5
For the green of this intensity the no. 2 gray of this series was
used (0.0366 cp. per sq. in.); for the red nos. 10-14 (0.01384-
0.0114 cp. per sq. in.) varying for the different meridians;
and for the blue, nos. 7-14 (0.01791-0.0114 cp. per sq. in.).
A graphic representation of the results of Table III. is
given in Figs. I and 2. In Fig. I are shown the limits of
•45
/O
133
160
FIG. 2. The effect of intensity of stimulus on the limits of sensitivity, prismatic
spectrum. In this chart are represented the limits of sensitivity for intensity 1/32 A
of Table I.: red 284.27, yellow 127.051, green 48.825, and blue 27.563 watt X io~10
sensitivity to the four stimuli in the 16 meridians for the inten-
sities represented in the prismatic spectrum A. The limits
for the red, yellow and blue stimuli at this intensity are,
it will be remembered, coincident with the limits of white
light vision. All four limits may be represented, therefore,
by a single tracing, an unbroken line in black. The limits for
green are represented by a broken line. In Fig. 2 are
i6
C. E. FERREE AND GERTRUDE RAND
represented the limits of sensitivity for the four stimuli at
the intensities represented in the prismatic spectrum, 1/32 A.
In this case the zone of sensitivity to blue is outlined by an
unbroken line and the zones for the other colors by broken
lines as indicated in the figure. An inspection of this figure
will show that the degree of excentricity of the limits is in the
order of the intensity of the stimuli. In discussing the
70
135
155
IS5
FIG. 3. The effect of intensity of stimulus on the limits of sensitivity, equal
energy spectrum. In this chart are represented the limits of sensitivity for intensity
B of Table IV.: red 891.05, yellow 882.51, green 884.946 and blue 882.025 watt X lO"10.
significance of *he crisscrossing or interlacing of the limits
obtained with the Hering pigment papers in previous work,
this is what was pointed out would occur if there were a
significant difference in the intensity of the stimuli. That is,
if the zone of sensitivity to red, for example, is in one meridian
wider and in another narrower than to green, etc., it can not
ABSOLUTE LIMITS OF COLOR SENSITIVITY
'7
be due to any difference in the intensity of the stimuli; for
such a difference, if significant, would make one zone con-
sistently wider or narrower than the other in all meridians.
A graphic representation of the results of Table IV. is
given in Figs. 3 and 4. In Fig. 3 are shown the limits of
sensitivity to the four stimuli in the 16 meridians for the
intensities represented in the equal energy spectrum B.
•45
no
155
ISO
FIG. 4. The effect of intensity of stimulus on the limits of sensitivity, equal energy
spectrum. In this table are represented the limits of sensitivity for intensity 1/32 B
of Table IV.: red 27.845, yellow 27.578, green 27.655, and blue 27.563 watt X io~10.
Again the limits for the red, yellow and blue stimuli coincide
with the limits of white light vision and are represented by a
single tracing, the unbroken line in black. The limits for
green are represented by a broken line. In Fig. 4 are shown
the limits for the four stimuli at the intensities represented
in the equal energy spectrum 1/32 B. With regard to this
i8
C. E. FERREE AND GERTRUDE RAND
figure the following points may be noted, (i) With stimuli
of equal energy the limits of no one of the colors, red, yellow
and blue, are consistently wider than the others. That is,
their limits are characterized by frequent crisscrossing or
interlacing. The limits for all three colors, however, are
consistently wider than for green. And (2) Fig. 4 sustains a
45
70
.1C
135
•55
155
FIG. 5. The limits of sensitivity to red, yellow, green and blue of the Hering
series of pigment papers, intensity of illumination, vertical component, 390 foot-candles.
somewhat striking general similarity to the charts obtained
for the Hering pigment papers. One of these showing the
limits with a surrounding field and preexposure of the bright-
ness of the colors employed is given in Fig. 5. While no
conclusion can be drawn from this similarity with regard to
the relative energies of the wave-lengths dominantly reflected
by these papers; still it suggests that they may all, roughly
speaking, be somewhere near the same order of value, at
.IDSOLUTE LIMITS OF COLOR SENSITIVITY 19
least much more nearly so than are these colors in the pris-
matic spectrum. The red, yellow, green and blue of the
prismatic spectrum gave, it will be remembered, rather
widely concentric, not crisscrossing limits.
In this general connection it may be of interest also to
note the close correlation which obtains between the results
of this investigation and those of the previous investigation
of the sensitivity of the peripheral retina by the threshold
method. That is, wherever the thresholds are found to be
low the limits are found to be wide, and conversely wherever
the thresholds are high the limits are found to be corre-
spondingly narrow. Some interesting results follow from
this. For example, in a given meridian the threshold curve
for a given color is found to be very irregular, rising in some
places slowly, in others quickly, and still in others dropping
and rising again. These fluctuations in the curve are,
moreover, different in the different meridians. This means,
of course, that the shape of the zones of sensitivity for this
color should change with the intensity of the stimulus em-
ployed, which is found to be the case. Furthermore, in the
same meridian the threshold curves for the different colors
differ from each other widely in the direction and amount of
the irregularity: and this difference in ~turn varies from
meridian to meridian. The result of this is that a crisscrossing
or interlacing of limits must take place whenever stimuli of
such relative intensities are used that the limits are of the
same general order of excentricity. In other words, as was
pointed out in our discussion of this phenomenon in earlier
papers, crisscrossing can mean only that there- is a lack of
uniformity in the relative sensitivity to the different colors in
the different meridians. For example, when it occurs in the
limits for blue and yellow, it indicates that the ratio of sen-
sitivity to blue and yellow changes in passing from meridian
to meridian. In short any investigation at all comprehensive
either of the thresholds or limits of sensitivity shows that
striking irregularity and not uniformity characterizes the
distribution of chromatic sensitivity in the peripheral retina.
This is in direct opposition, it will be remembered, to the
20 c. E. FERREE AND GERTRUDE RAND
claim made by Hess (6) that constancy of ratio of sensitivity
to the paired colors prevails throughout the retina which
claim, it will be remembered, was advanced by Hering (7) in
support of his own theory and in refutation of Pick's (8)
and Leber's (9) modifications of the Helmholtz theory to
explain the color blindness of the peripheral retina. So far
as we are able to determine no one intensity or set of conditions
will give coincidence of limits in all meridians for any two
colors inside the limits of white light vision.
In conclusion it may not be out of place to point out the
bearing of these results on the work of the clinic. In the prac-
tice of perimetryas applied to diagnosis it is commonly accepted
that the field of vision for the normal eye may be divided con-
centrically from periphery to center in the following order :
white light and form, blue, red and green. It is obvious from
the fore-going results (a) with stimuli taken from the prismatic
and equal energy spectra and (b] from the effects obtained by
varying the intensity of the stimuli that the responsibility for
such a rating of the color fields rests for the greater part with
the relative intensities of the pigment stimuli used in the
work of the clinic. That is, the limits of sensitivity to
red, yellow, blue and white light for stimuli of high intensities
are coincident; for stimuli of lower intensities taken from the
prismatic spectrum they are rather widely concentric; and
for stimuli of equal energies of the order of intensity of
27.563 x io~10 watt they are interlacing.
Another feature of interest is the claim that has been
made by certain clinicians, but not generally accepted, we
believe, that the interlacing of the limits for blue and red
indicates a pathological disturbance in the relative distribu-
tion of sensitivities. While we are not disposed to dispute
this conclusion because of a too meager knowledge of all of
the data that should be taken into consideration in its evalu-
ation, still we do think it fair to note that pathological
disturbances are only one set of factors that may contribute
to such a result and that widely different results may be
gotten with the same eye with no greater differences in the
test conditions than may occur from time to time in the same
ABSOLUTE LIMITS OF COLOR SENSITIVITY *i
clinic or laboratory unless a clear understanding is had of
the factors which affect the apparent powers of response of
the peripheral retina and adequate means are exercised for
their control. These factors are, so far as we are able to list
them, composition, area, intensity of the stimulus and dura-
tion of the stimulation, breadth of pupil, the intensity of the
general illumination and the state of adaptation of the retina,
and the brightness of the preexposure and surrounding field.
Obviously if the determination of the apparent limits is to
be given clinical significance the work should be done under
conditions of work which have been most carefully standard-
ized, for the apparent limits are a resultant of these conditions
as well as of the actual distribution of sensitivities.
The degree of importance that is attributed by at least
one clinician to the absolute and relative distribution of
sensitivities over the retina may be indicated by the following
quotation from a recent work on perimetry. "Contraction
of the form fields shows the degree of disease of the visual
tract. It is better evidence of the real condition of the
visual path than an ophthalmoscopic study can possibly
furnish. The evidence is minute and analytical. The color
fields and color changes moreover furnish a more delicate test
in the early stages of the disease and at times furnish a clue
to the seat of the trouble before an appreciable change has
taken place in the form field" (10).
SUMMARY OF RESULTS AND CONCLUSIONS
The more significant features of the above results may be
summarized briefly as follows:
1. The far periphery of the retina is not blind to red,
blue and yellow. It is merely deficient in sensitivity to
these colors. That is, with stimuli of sufficient intensity
the limits of red, blue and yellow coincide with the limits of
white light vision. The blindness to green, however, is for
our observers absolute.
2. The amount of change of intensity required to produce
a detectable change in the apparent limits of sensitivity in
the more remote parts of the retina is very great. This
22 c. E. FERREE AND GERTRUDE RAND
amount changes very irregularly from center to periphery of
the retina in a given meridian and from meridian to meridian
as might be expected from the great irregularity in the
distribution of sensitivity in the peripheral retina. (Cf.
'Chromatic Thresholds of Sensitivity from Center to Peri-
phery of the Retina and their Bearing on Color Theory/
PSYCHOL. REV., 1919, 26, pp. 16-42.)
3. Two other important phenomena may also be men-
tioned as a result of this irregularity, (a) The shape of the
zone of sensitivity to a given color changes with the intensity
of the stimulus employed in making the determination.
And (b) when stimuli of equal or of the same order of intensity
are used the limits for red, yellow and blue are found to
interlace or crisscross each other irregularly rather than to
coincide in complementary pairs as was reported by Hegg,
Hess and Baird in a more limited investigation of the retina's
powers of response. The former phenomenon is the direct
corollary of the difference in the rate of decrease of sensitivity
to a given color in passing from the center to the periphery
of the retina in the different meridians; the latter, to the
change in the ratio of sensitivity to the different colors from
meridian to meridian. The lack of uniformity of grading
of function from point to point in the periphery of the retina,
reported in this and previous papers, while striking, can
scarcely be considered as surprising. It is in fact just what
might be expected of those parts of a sense organ which are
little used and poorly developed.
4. The responsibility of the accepted clinic rating of
limits in the order from widest to narrowest of blue, red and
green doubtless for the greater part rests with the relative
intensities of the pigment stimuli used in the work of the
clinic. With stimuli of high intensity the limits for red,
yellow and blue coincide with the limits of white light vision;
for stimuli of lower intensities, taken from the prismatic
spectrum, they are rather widely concentric; and for stimuli
of equal energies of medium intensities they are interlacing.
5. The interlacing of limits for red and blue is a normal
result for stimuli of equal energy of medium intensities. It
ABSOLUTE LIMITS OF COLOR SENSITIVITY »3
may not therefore be due to pathological disturbances in the
distribution of sensitivities as has been claimed by certain
clinicians. In all responsible work on the determination of
the apparent limits it is obviously of great importance to
bear in mind that the results are dependent both upon the
actual distribution of sensitivity and the numerous factors
which affect the apparent powers of response of the peripheral
retina.
BIBLIOGRAPHY
1. FERREE, C. E., AND RAND, G. A Note on the Determination of the Retina's
Sensitivity to Colored Light in Terms of Radiometric Units. Amer. Jour, of
Psych., 1912, 23, pp. 328-332.
2. FERREE, C. E., AND RAND, G. Chromatic Thresholds of Sensation from Center
to Periphery of the Retina and their Bearing on Color Theory. PSYCHOL. REV.,
1919, 26, pp. 16-41; 150-163.
3. Ibid., pp. 152-153; Rand, G. The Factors that Influence the Sensitivity of the
Retina to Color: A Quantitative Study and Methods of Standardizing. Psychol.
Man., 1913, 15, No. I, pp. 117 ff.
4. RAND, G. Psychol. A/on., 1913, 15, No. I, pp. 97-110.
5. LANDOLT UNO SNELLEN. Ophthalmometrologie. Handbuch der ges. Augenheilk.
von Graefe und Saemische, 1874, 3, p. 70.
6. HESS, C. Ueber den Farbensinn bei indirectem Sehen. A.f.O., 1889, 35, pp. 1-62.
7. HERING, E. Ueber die Hypothesen zur Erklarung der peripheren Farbenblindheit.
A.f.O., 1889, 35, PP- 63-83.
8. PICK, A. Zur Theorie der Farbenblindheit. Arbeiten aus dem physiol. Laborat.
der Wurzburger Hochschule, pp. 213-217.
9. LEBER, T. Ueber die Theorie der Farbenblindheit und uber die Art und Weise,
wie gewisse, der Untersuchung von Farbenblinden entnommene Einwande
gegen die Young-Helmholtz'sche Theorie sich mit derselben vereinigen lassen.
Klin. Monatsbldtter f. Augenheilk., 1873, n, pp. 467-473.
10. PETER, L. C. The Principles and Practice of Perimetry. N. Y., 1916, pp. 97-98-
SOME FACTORS IN THE PERCEPTION OF REL-
ATIVE MOTION. A PRELIMINARY
EXPERIMENT
BY H. A. CARR AND M. C. HARDY
The University of Chicago
The perception of motion is relative. The observer per-
ceives the motion of one object in relation to some stationary
object or set of conditions. In the perception of motion an
appreciation of the stability of the one object is just as essen-
tial and important as the appreciation of the motility of the
moving object. The observer himself constitutes the center
of reference in most perceptual acts.
The perception of motion involves two aspects, an appre-
ciation of the fact of motion, and an appreciation as to which
of the two objects is stationary and which is in motion.
These two aspects are to some extent independent variables.
One's appreciation of them may be based upon entirely
different sets of conditions. One may correctly perceive
the fact of motion, but judge erroneously as to which of the
two is moving. If a fixated visual object is moved toward
an observer, there results a change in the sensory conditions
which mediate his judgment of distance. But exactly the
same changes will be induced by a movement of the observer
toward the object. Changes of these distance criteria as
intensity, size of retinal image, accommodation and con-
vergence, etc., serve to induce in both cases an appreciation
of a change of the distance between the object and the
observer, — an appreciation of the fact of motion. But these
sensory changes give no clue as to which object moved and
which was stationary. Judgments of the relative motility
of the object and the observer depend upon other factors,
presumably certain characteristics of either one or both of
the two objects concerned.
When the organism constitutes one of the objects in the
24
PERCEPTION OF RELATIVE MOTION 25
perceptual situation, the judgments concerning the motility-
stability relation are based in part upon certain sensory
aspects of the observer. The nature of these organic stimuli
has been pretty well determined. The judgments of relative
motion are based upon the presence or absence of such
factors as the intention or expectation of moving, the sight
or feeling of muscular activity, friction, air currents, the
static sense, organic sensitivity, etc. As a rule these factors
mediate correct judgments, but exceptions occur as in the
illusions of the haunted swing, moving trains, etc. But
these organic factors do not constitute the whole of the con-
ditions which influence the judgments of relative motion.
These judgments are also based in part upon certain charac-
teristics of the observed object. The haunted swing illusion
is probably due in part to the fact that one does not, on the
basis of past experience, expect buildings to undergo rotary
motion. Likewise the perceptual situation is frequently
confined to two observed objects when all spatial relation
to the perceiving organism is pretty well excluded. The
cloud and moon illusion illustrates such a perceptual situation.
It is extremely improbable that this illusion is due primarily
or mainly to any erroneous judgment as to the spatial relation
of the observer to either of the objects. Rather it seems
that the judgment of relative motion in this case depends
upon certain peculiarities of the objective situation.
The existence and nature of these objective factors have
never been adequately considered. It was the purpose of
this experiment to attempt a preliminary investigation con-
cerning the possible influence of certain features of the objec-
tive situation. The experiment presented a condition some-
what similar to the cloud and moon illusion. The observer
was seated in a dark room and was required to judge as to
the relative motion of two small lights whose intensities were
such that no other objects were visible. By this procedure
it was hoped that the judgments might be based mainly
upon certain characteristics of the two lights rather than
upon any perceptible relation between them and the observer.
Two 2 c.p. electric lights were each enclosed within a
26 H. A. CARR AND M. C. HARDY
small wooden box containing a small circular diaphragm
opening covered with ground glass. Each box was placed
upon a horizontal double rod track. The vertical distance
between the centers of the two lights was 72 mm. Between
the two tracks was a piston which was attached to a metal
disc, and this disc was rotated by a motor with a worm gear
speed reducer. By an appropriate device either one or both
of the light boxes could be easily and quickly attached to or
disengaged from the moving piston without stopping the
motor. The motion imparted to the lights was thus a vibra-
tory or pendular one.
The observer was seated at a distance of 8 ft. from the
lights with his line of vision at right angles to the direction
of motion. A cloth curtain screened the lights from view
while the apparatus was being adjusted and set in motion.
The screen was then removed and the observer was requested
to fixate a specified light during four complete vibrations.
The lights were again covered while the report of the subject
was recorded and the apparatus adjusted for the next expo-
sure. Twenty such judgments constituted a day's test for
each subject. The subjects had been informed that either
one or both of the lights might move in an irregular temporal
order. As a matter of fact but one light was moved at a time,
and the two lights were moved an equal number of times in
each day's test of 20 trials. The observer was requested to
report the movements as perceived rather than to attempt to
guess at the objective situation. The reports were scored
as correct or illusory. The judgment was correct when the
moving light was perceived in motion and the stationary light
as stable. Illusory judgments were of two sorts: Both
lights were perceived as moving in opposite directions.
In this case the two movements might not be equal in rate or
extent. In the second case the moving light was perceived
as stationary and the stationary light as moving. The
relative number of the two types of illusion was not a function
of the variable factors studied and hence separate tabulations
are unnecessary.
The five factors of the size and intensity of the lights,
PERCEPTION OF RELATIVE MOTION *7
their extent and rate of motion, and the direction of fixation
were chosen for study. The size of the lights was controlled
by iris diaphragms. Three magnitudes were arbitrarily
chosen, whose diameters were 4 mm., 15 mm., and 32 mm.
These magnitudes will be termed A, B, and C respectively.
The intensity of each light was controlled by a rheostat.
Three intensities were chosen and these will be referred to
as I, 2, and 3 in their order of brightness. Intensity I was
chosen as near the limen of visibility as possible without
inducing discomfort in the observer. Intensity 3 was the
maximum possible without illuminating the room to the
point of visibility. Intensity 2 was approximately a mean
between the other two. Since the lights were fed by a
storage battery giving a constant current, it was possible
to reproduce these intensities approximately by calibrating
the rheostats. The extents of motion chosen were */2 inch,
i inch, and 2 inches. The amplitude of the movement was
controlled by varying the point of attachment of the piston
along the radius of the rotating disc. The rate of motion
was controlled by varying its amplitude and by altering the
speed of the motor by a rheostat. The rates chosen for
investigation were */4 inch, */2 inch, I inch and 2 inches per
second. The conditions were arranged so that the moving
and the stationary lights were fixated an equal number of
times. The upper light was fixated during one day's test of
twenty trials and the lower light was then fixated on the
succeeding day. Since the upper light moved in one half of
the trials, fixation was distributed equally between the sta-
tionary and the moving lights. It was thus possible to deter-
mine which direction of fixation was the more conducive to
correct perception.
Some illusions were obtained for practically all experi-
mental conditions. The efficacy of any factor must thus be
determined from the relative frequency with which correct
judgments were obtained as that factor was varied. The
percentage frequency of correct judgments was determined
for one condition and this value was compared with that
obtained for a second condition. For the individual records
28
H. A. CARR AND M. C. HARDY
we utilized Yule's formula, standard deviation = ^pqfn,
where n is the number of judgments and p and q represent
the percentages of correct and wrong responses. Unless
otherwise stated, the number of judgments for each condition
was 80. In general any difference of at least 12 between two
percentage frequencies for an individual is significant. The
averages of the individual records for the group are also
given in the tables. A difference of 10 in these values indi-
cates a high degree of probability.
i. Effect of Size, — The percentage frequencies for the
variations of size are given in Table I. The first vertical
column specifies the magnitudes employed. In the first
condition the magnitude of both lights was A. In the
second condition the size of the lower light was A and that of
the upper light was B. In the third condition the magnitude
of the upper light was changed to C. The horizontal rows of
figures are the percentage values for the individuals and the
average values for the group. Below are specified the
various conditions which were kept constant as the magni-
tudes of the lights were altered. The intensity of both
lights was i, the amplitude of movement was I inch, the rate
of motion was */2 inch per second, and each of the two lights
was moved and was fixated the same number of times.
TABLE I.
PERCENTAGES OF CORRECT PERCEPTIONS WITH VARIATIONS OF SIZE
Sizes
J.
K.
L.
M.
s.
Ave.
I. A-A. .
4.7
22
4-Q
69
34
44
2. A-B
ti
29
75
82
32
54
3. A-C
50
52
66
77
46
58
Constant conditions: Intensities, i; Amplitude, I in.; Rate, % in. per sec.; Equal
number of movements and fixations.
Conditions' 2 and 3, as compared with I, are evidently
conducive to perceptual accuracy. With those conditions
the individual percentage values are the larger in nine of the
ten comparisons, and five of these are statistically significant.
The average values are also the larger, the differences in both
cases being significant.
PERCEPTION OF RELATIVE MOTION
29
Comparing conditions 2 and 3, two of the individual
records favor the third condition and both of these are sig-
nificant. The average values also indicate the greater
efficacy of the third condition, but the difference between
them is slight. The records do not permit of any very con-
fident statements. Either we may say that the two con-
ditions do not differ in efficacy, or that the third condition
is the more effective with certain individuals.
The comparative data are ambiguous in one respect. It
is impossible to determine from this experiment whether
the number of correct perceptions is a function of relative or
absolute size. On the one hand it is possible to assume that
the perceptual accuracy is greater when the two objects are
unequal in size than when they have the same magnitude,
and that the accuracy of some individuals is proportional to
the degree of this inequality. On the other hand it is equally
valid to conclude that perceptual accuracy varies directly
with the magnitude of the combined areas of the two objects.
2. Influence of Intensity. — Three intensity conditions
were investigated. In the first the intensity of the lower
light was I, and that of the upper was 2. In the second
condition the intensity of both lights was 2, while in the third
condition the intensity of the upper light was changed to 3.
The percentage frequencies of correct judgments for the
three conditions are given in Table II.
TABLE II
PERCENTAGES OF CORRECT JUDGMENTS WITH VARIATIONS OF INTENSITY
Intensities
B.
C.
Ha.
Jo.
P.
Ave.
I. 2-1 . .
C3
35
81
54
61
57
II. 2-2
22
19
56
27
rj
1?
III. 2-3
31
27
SS6
44
SO
41
Constant conditions: Magnitudes, 2; Amplitude, I inch; Rate # inch per sec.;
Both lights fixated and moved an equal number of times.
Condition I. gives the maximum of perceptual accuracy.
Its values are the largest for each individual and for the group
as a whole, and of these ten individual comparisons six are
significant.
3° H. A. CARR AND M. C. HARDY
Condition III. is slightly more effective than II. Three
individuals secured the larger values for this condition and
one of these comparisons is significant. The remaining two
individuals of the group gave practically identical records for
the two conditions.
Perceptual accuracy is thus a function of both relative and
absolute intensity. On the one hand an inequality in the
intensity of the two lights gives more correct judgments than
does equality. When both lights are unequal in brightness,
the greater number of correct responses are secured with the
lower illumination. Of the two factors, the degree of illumi-
nation is possibly the more effective one.
3. Amplitude of Motion. — The percentages of correct
judgments with variations of the amplitude of movement are
given in Table III. Our apparatus permitted but two ampli-
tudes for each rate of motion.
TABLE III
PERCENTAGES OF CORRECT JUDGMENTS WITH VARIATIONS OF AMPLITUDE
Amplitudes
Ba.
D.
G.
H.
Mi.
Ave.
I. When rate of motion is X inch Per sec-
yi inch
72
70
47
42
26
24
2O
31
16
IS
3^
36
I inch
2. When rate of motion is I inch per sec.
I inch
79
94
4i
49
27
27
42
II
26
22
43
41
2 inch
Constant conditions: Intensities, 2; Sizes, A; Equal distribution of movements
and fixations between the two lights.
The ability to perceive correctly does not appear to depend
in any pronounced manner upon the amplitude of motion.
Six of the ten individual comparisons indicate the greater
efficiency of the smaller extents of motion and one of these
is significant; three comparisons indicate that more correct
perceptions are possible with the larger movements and one
of these is significant; one comparison favors neither assump-
tion. The results of but one of the five individuals (Mi.)
consistently favor either assumption; both comparative values
PERCEPTION OF RELATIVE MOTION 31
of this subject indicate the greater efficacy of the lesser
amplitudes, but the differences in both comparisons are
quite small. From these data one is not justified in asserting
that perceptual accuracy is dependent upon the amplitude
of motion so far as this factor was varied.
4. Influence of Rate of Motion. — Table IV. gives the
percentages of correct judgments for variations of rate of
motion. Our apparatus permitted of but two variations of
speed for each amplitude of movement.
TABLE IV
PERCENTAGES OF CORRECT JUDGMENTS WITH VARIATIONS OF RATE OF MOTION
Rate* of Motion
Ba.
D.
G.
H.
Mi.
Ave.
I. With amplitude of y* inch
% inch per sec
57
. 72
45
47
2
26
IS
20
2O
16
27
36
Ji inch per sec
2. With amplitude of i inch
J^ inch per sec
70
79
42
41
24
27
31
42
14
26
36
43
I inch per sec
3. With amplitude of 2 inches
I inch per sec
94
Si
49
66
27
62
II
16
22
I?
4i
4»
2 inch per sec
Constant conditions: Sizes, A\ Intensities, 2; Equal distribution of movements
and fixations between the two lights.
In each of the first two series the faster rate gives the
greater number of correct judgments in four of the five
individual comparisons; of the eight favorable comparisons
four are significant. In the third series three of the indi-
vidual comparisons favor the faster rate. Two individuals,
G. and H., were invariably more accurate with the faster
rates. Individuals Ba. and D. were more accurate with the
faster rate in two of the three series. Mi. was more accurate
with the faster rate in but one of the series.
The facts indicate that the rate of motion is effective at
least with some individuals, and that in the majority of cases
perceptual accuracy is greater for the faster rates of motion.
H. A. CARR AND M. C. HARDY
5. Individual Differences in Perceptive Ability. — In Table
V. the individuals of each group are ranked according to
their ability to perceive the objective situation correctly.
Group I was composed of five subjects. These individuals
were tested for three series in which the size of the lights was
varied (See Table I.). These individuals were ranked ac-
TABLE V
RANKS OF INDIVIDUALS IN RESPECT TO NUMBER OF CORRECT JUDGMENTS
Group I
M.
L.
J.
s.
K.
I
2
3
4
5
I
2
3
4
5
I
2
4
5
3
Total
3
6
10
13
13
Group 2
Ha.
P.
Jo-
B.
C.
I
2
3
4
5
I
2
3
4
5
I
2
3
4
5
Total
3
6
9
12
15
Group 3
Ba.
D.
G.
H.
Mi.
2
5
4
3
2
3
4
5
2
4
3
S
3
4
2
5
2
3
S
4
3
I
2
s
4
Total. .
8
12
21
2"?
26
cording to the number of correct judgments. Subject M.
secured the highest percentage in all three tests and is ranked
first three times with a total score of 3. Subject L. stood
second in every case with a total score of 6. From these
data it is obvious that individuals differ in their ability to
perceive the situation correctly. In group 2 the five subjects
maintained the same relative ranking in all three tests.
Some individuals are able to perceive a relative movement
situation with a high degree of accuracy while other indivi-
PERCEPTION OF RELATIVE MOTION
33
duals are more prone to perceptual illusions. It is possible
that this experiment might constitute a good test for deter-
mining the relative suggestibility of different individuals.
6. Influence of Direction of Fixation. — In any series of
80 judgments, the moving light was fixated 40 times and the
stationary light was fixated an equal number of times.
The percentages of correct judgments for each kind of fixation
were computed for each individual for all experimental
conditions. These percentage values are given in Table VI.
TABLE VI
PERCENTAGES OF CORRECT JUDGMENTS WITH VARIATION or FIXATION
Fixation
L.
M.
G.
Moving
50 70 60
6? QO 72
2? 27 27 27 2 4C
Stationary
48 80 72
72 75 82
22 25 27 27 2 57
Ha.
Jo.
Ba.
Moving
6<; <;7 7?
1 47 42
cc CQ 7C 87 67 2
Stationary
47 S6 87
SO 40 65
85 95 82 ico 47 100
S.
B.
H.
Moving
C, I? 4.O
7 12 •;;
2O C 12 C C, 5
Stationary
62 50 47
37 SO 70
42 35 72 17 25 27
J.
c.
Mi.
Moving
C7 77 72
1O C2 62
c 2 O O 1O O
Stationary
38 30 27
727
22 30 52 45 10 35
P.
K.
D.
Moving
4-2 4? 1C
i<; 27 42
32 45 17 30 30 35
Stationary
60 55 87
3O 3O 62
52 50 65 67 60 97
Subject L was tested for three experimental conditions.
His percentages of correct perceptions when the moving light
was fixated were 50, 70 and 60. These values are to be
compared with the percentages of 48, 80 and 72 secured
when the stationary light was observed. Each percentage
value of the table is based upon 40 judgments. With this
number of cases a probability that direction of fixation is an
effective factor in the perception of relative motion will be
indicated by any difference of 20 or more.
With four subjects, L., M., G., and Ha., the direction of
fixation did not appreciably influence the number of correct
34
H. A. CARR AND M. C. HARDY
perceptions. The comparative values do not consistently
favor either mode of fixation, and none of the differences are
large enough to be significant.
With two subjects, J. and C., fixation of the moving object
is the more conducive to perceptual accuracy. This mode of
fixation is favored in all comparisons and all of the differences
are significant.
With the remaining nine individuals, the greater percep-
tual efficiency was attained by observing the stationary object.
For these subjects 36 of the 39 comparisons favor this mode
of fixation, and of these 36 comparative values 27 are sig-
nificant.
IABLE VII
PERCENTAGES OF CORRECT JUDGMENTS WHEN EACH LIGHT WAS STATIONARY
Subject
Stable Light
Experimental Condition
i
2
3
4
5
6
Are.
Ba
Upper. .
60
80
30
55
45
2
30
32
22
5
72
85
27
55
42
12
42
42
30
22
62
82
42
52
42
10
IS
25
25
7
80
35
30
60
2
2
IS
IS
40
o
95
92
27
70
40
IS
12
IO
25
2O
52
SO
57
75
60
42
12
2O
17
17
70
71
IS
61
38
H
21
2i
26
12
D
Lower
Upper. .
G
Lower
Upper. .
H
Lower
Upper. .
Mi
Lower
Upper. .
Lower
Constant conditions: Lights equal in size and intensity; Equal number of fixations
and movements.
7. The Influence of Position. — In Table VII. are listed
the percentages of correct judgments when the upper light
was stationary as compared with the percentages when the
lower light was immobile. The upper and lower lights were
equal in size and intensity for all six experimental conditions.
The two lights were fixated an equal number of times. The
data are given only for the third group of subjects, since the
two objects were not always equal in size or intensity for
groups i and 2. In the first experimental condition, Ba.
correctly perceived 60 per cent, of the 40 cases in which the
upper light was stationary and the lower moving. He
perceived correctly 80 per cent, of the 40 cases in which the
PERCEPTION OF RELATIVE MOTION 35
lower light was stable and the upper one was moving. This
comparison was instituted to determine whether either
position tends to be associated with stability. A difference
of 20 in the individual values and of n in the average values
indicates a probability of the efficacy of the position factor.
Position did not exert any appreciably consistent effect
with subjects Ba. and H.
Subject D. consistently perceived the situation more
correctly when the lower light was stable and the upper in
motion than in the reverse case. Five of the seven compari-
sons are statistically significant. With this individual the
lower position is associated with immobility and the upper
one with motility.
Subjects G. and Mi. exhibited more correct perceptions
when the stationary light occupied the upper position. Both
subjects were consistent in this preference with one exception
in which the two percentage values were equal. These
individuals tend to associate stability with the upper position.
8. Summary. — The subjects were required to judge con-
cerning the relative motion of two lights, one of which was
always stationary.
Perceptual accuracy was promoted either by a difference
in size of the two lights or by an increase in their combined
area. Possibly both conditions may have been operative.
Perceptual accuracy was favored by an inequality of
brightness of the two lights and by a decrease of their com-
bined illumination.
The amplitude or extent of motion so far as this factor
was varied did not exert any effect upon perceptual ability.
Perception was generally more accurate with the faster
rates of motion.
The individuals exhibited consistent differences in their
ability to perceive the situation correctly.
The majority of the subjects were able to perceive the
situation more correctly with stationary eyes, i.e.t when the
stationary light was fixated and the moving light was per-
ceived with indirect vision. Some individuals gave the
better records when the moving light was fixated. Other
36 H. A. CARR AND M. C. HARDY
individuals were able to judge equally well for both conditions
of fixation.
Some individuals were able to perceive more correctly
when the stationary light occupied the upper position, while
the opposite condition obtained for other subjects. The
position of the stationary light was not effective for other
observers.
The influence of such factors as the degree of illumination
and the rate of motion upon perceptual ability is obvious.
The influence of fixation indicates that certain individuals
solved the problem in part by an indirect reference of one of
the objects to the organism. It is practically impossible to
exclude all spatial reference of the objects to the observer.
Some subjects were apparently able to detect the moving
object by means of the eye sensitivity involved in following it.
Others were able to identify the stationary object by the
absence of eye movement while fixating it. Individual
differences in this respect do not admit of a ready explanation.
We have no explanatory suggestions to offer for the in-
fluence of inequalities of size and brightness.
The influence of the position of the stationary light may
be explained by the supposition that the subjects apprehended
the perceptual situation much as though the two lights
constituted the ends of a swinging pendulum. Individuals
who exhibited any tendency to conceive the situation as
similar to a simple suspended pendulum will obviously
associate the upper position with stability and hence will
make a greater number of correct judgments when the sta-
tionary light occupies this position. On the contrary those
who preferred to regard the situation as analogous to a
metronome will associate the lower position with immobility
and give the better records when the lower light is stationary.
Certain subjects were not influenced by the position of the
stationary light. This fact may be explained by supposing
that these subjects conceived the situation as a compound
pendulum with the center of rotation located between the
two lights, that they were able to assume equally well either
of the first two attitudes, or that they failed to adopt any
pendular attitude.
PERCEPTION OF RELATIVE MOTION 37
Individual differences in perceptive ability may likewise
be explained in part by differences in apperceptive attitude.
Individuals who are able to adapt the first two attitudes to
the objective sequence will naturally make unusually high
scores. Subjects inclined to regard the movements as similar
to a compound pendulum will obviously make poor records.
No introspective records were taken in regard to these
apperceptive attitudes during the experimentation. The
hypothesis is suggested on the basis of its a priori plausibility.
The possibility that the subjects may adopt some pendular
attitude toward the situation suggests that it is advisable
in future experiments to eliminate this complicating condition
by locating the two lights in a horizontal plane.
A NEW OBJECTIVE TEST FOR VERBAL IMAGERY
TYPES
BY SAMUEL D. ROBBINS
From the Psychological Laboratory, Harvard University
The purpose of this experiment was to devise an objective
test by means of which the vividness of auditory, visual, and
kinaesthetic verbal imagery of persons who cannot be de-
pended upon for intelligent introspection may be quickly and
accurately determined.
I had for subjects six students who were studying psy-
chology. Four of these subjects were men, two were women.
Two other women were subjects only in the first test described.
No subject was under twenty years of age or over thirty.
All of the women were born and educated in this country
and had all their lives spoken English fluently. All the men,
on the other hand, were born and educated in foreign lands,
had spoken English but a few years, had a distinctly foreign
accent, and had limited vocabularies of English words differing
much from each other.
A number of series of card pairs containing a single
column of five monosyllabic nonsense words of three or four
letters were very carefully typewritten (double spaced) so
as to look exactly alike except for one word which differed
from the corresponding word on the other card in a single
letter. The following types of nonsense groups were tabu-
lated separately, the card pairs being shuffled so that the
subject could not predict the nature of the next change:
(i) the spelling was changed without altering the pronun-
ciation, as veek-veak; (2) two long vowels were exchanged, as
zoke-zake; (3) a short vowel was exchanged for the same long
vowel, as koss-kose, there being another change, a visual one,
in most cases; (4) two short vowels were exchanged, as meb-
mib; (5) two consonants were exchanged which gave quite
different auditory and kinsesthetic impressions, as aze-ane;
38
OBJECTIVE TEST FOR VERBAL IMAGERY TYPES 39
(6) two consonants were exchanged which sounded much
alike but gave quite a different kinaesthetic impression in a
trial series, as miz-niz; (7) two consonants were exchanged
which gave about the same auditory and kinaesthetic im-
pression in a trial series, as bim-pim. As some changes
necessitated four-letter words, at least one four-letter word
appeared on each pair of cards. There was an equal number
of changes at the beginning, middle, and end of the word
and for each of the five positions of the word in the column;
hence the position of the change could not be predicted.
The nonsense words were given to the subjects in three
ways: (i) They were read loud by the experimenter, the
subject being careful not to repeat, spell, or visualize the
word; (2) the subject repeated each word as it was read to
him before the next word was read, being careful not to spell
or visualize it; and (3) the cards were exposed in a tachisto-
scope and the subject was asked to whisper each word.
(Subjects were asked to whisper rather than to read aloud
because most stammerers can whisper without stammering,
and these results are later to be compared with those obtained
from stammerers.) The first method permitted the subject
to employ auditory imagery alone, the second permitted both
auditory and kinaesthetic imagery, and the third auditory,
kinaesthetic, and visual imagery. There was an interval of
five seconds between the reading of the first card of a pair
and the first word of the second card, during which the subject
was requested not to think of any of the words on the first
card.
The subjects followed this direction remarkably well, and
formed surprisingly few associations with these nonsense
words. They forgot the nonsense words on the earlier cards
before the later cards were read, hence memory of words
on the first test played little if any part in the recogni-
tion of changes by the second and third methods; changes
which were detected when the cards were read by the first
method were frequently not detected when the same cards
were subsequently read by the same subject by a different
method. In the few cases where a change was remembered
40 SAMUEL D. ROBBINS
from a previous test, the letter 'A' was written on the record
so that this change could be omitted in compiling the results
if enough associations were formed to alter the averages
materially. Each series was read at the speed which each
subject found most satisfactory on a trial series.
The subjects were asked to report on a scale of o to 3
whether there was a change, o indicated that there was no
change or that the subject did not know whether there was a
change. I meant there might have been a change of which
the subject was not at all sure. 2 denoted that the subject
was pretty sure there was a change. 3 showed that the
subject was positive there was a change. If a subject re-
ported the right position of the change, he was credited the
score he reported even if he gave the wrong word or could
not remember the word. If he reported there was no change
when there was a change, his score for the change was o.
As a matter of fact, every card was changed, except when a
subject reported so many changes in succession that I feared
he would mistrust my instructions that some pairs of cards
had one change and others had no change.
The subject was also required to report what the changed
word was on both cards, stating how sure he was of the word
he reported on the same scale of o to 3. For his word score,
he was given the score of the word of which he was less sure.
If he forgot one of the words or reported one or both words
incorrectly his word score was o. If, on the other hand, he
detected the main change correctly but made the same mis-
take in some minor letter of both words, he was given full
credit.
Each subject was given a long trial series1 by each method
before the cards representing the scores compiled in table I
were used. Subjects were therefore familiar with the expe-
riment before the series reported was used.
Table I. shows the average score each subject obtained
in each of the seven groups of changes by each of the three
methods of presentation. The four columns in division 6
1 One trial series consisted of sense words. This series was discarded because the
words proved unequally familiar to the men of foreign birth.
OBJECTIVE TEST FOR VERBAL IMAGERY TYPES
4«
£ E
<
O ^s to to O ^s is, ^^
** •*• ^ t*NO VO VO VO
00 «0
t* 00
*^
J
ss^a^^s
*t ***
^
K
«««««MOO
- M
••
J
- -
M
|ji
•i
H5H5I8
VO «
*
If
5:'
»* «0<O. Is. H O €0 vo
« ON« fOQ ls.00
M _,,_,, f< O O
i!
!
-| a.
pi
^.885^.00^
z&
^
9
.
VOOO vo to vovO t* CO
VOOO « ONOO K -+O
K» to
3
If
fl«("4^«^-OO
M M
•*
J
---OOOOO
0 -
*•
IJ
j
oooooooo
r* Is.
t* vo
5.
^
.
O» *• NO ON ^00 ONOO
»o r< Is. Is. ON vo ^»vO
oT8
s
-«<"«- ~oo
M M
M
-«-ddddd
0 -
<
to tooo r» to to Q f
O x « ON l>- VO ON O
5^
$
_
8«o ls.vO H \O NO r*
O ON ONOO ts.O O
^9
to
C*NP«i-i»«i-iOl
— —
M
n
••
I
.
O vo vo O O vovovo
•«*• vo ONOO M Is. ON O
ON«"
<§
I
.
VO|S.^QQVO O « O
fctf
M
0
*
W«t«~t*r«0«
M t*
r*
\i
r*fifi««H«-
- H
M
s
Jj
.
8O O vo O vo 10 Q
11 M VO vo O ONOO
o oo
«.
pi
•* »*• ONNO vo t* OO to
oooo
— OO
to
vo
««««««oo
M M
•-1
N««MMMM00
M M
**
.
vovoOOOOOO
ts. ts. ^ to vo 1-1 oo c*
OO t*
tooo
vS
I
.
OO •-• ^ N OO VO NO t*
ONOO O O vo to O O
Is. to
tooo
NO
«««c*««««
— —
"*
V
o
8S88S8S-S
vo O
to ON
VO
-.=
>
Is. NO •* O O O + •*
tovo
to to
5
DC
tot* to tot* « « «
t* N
W
0 -
0
1
.
80 0 Q Q 0 0 0
O O O O OO -*00
vo O
N O
VO
vi X
6C-5
5 N "
.
°*v88S£-S3.£
to to
O oo
to
i
to to to to tor* « «
f* to
ci
ri — — - — tl O O
M M
•*
1
pi
83-88883-3.
vovo
M OO
VO
vo
Pi
Ooo to rs to o ^s
vo oo O vovO tovo O
VOVO
P^
I
tor. to to to- t* c*
C« «
**
°*
M«OOOOOO
O «
0
J
80 0 Q O O O O
OO O -*-O -*oo
vo vo
VOOO
O
3J
.
^Q^JsON-^gto
O O
to «-•
o
(A
to to to toe* f* r« <*
« t*
N
--d-dddd
O -
o
^
is. Q to to vo Q Ooo
VO OO ON O-OO OO to O
VOOO
V&
g
Svjss'irs'as
vo to
OO vo
*
t*c*t*t*c«t*t*t*
f* C*
N
it
0 -
"*
-
>
^8^8885-^
vo ON
VO
.
OOOtoQtoQQ
¥• t* r* 1000 >- OOvC
tooo
-
|
t* tot* to to tot* «
t* t*
N
•> B
M«t*«««00
— -
*•
&
to O *^* ir* °^ O ^^ l^>
t*^ tN. O^OO u*> fs. O^ N
vo «
5;
K.H
.
«R«JJfSR^o-
roOO
to •+
ON
N««t*«««M
f* C*
M
c*«0«0000
O •«
0
R.R888RVOVO
O vo
vo oo
M
15
U
.
~ gfc-SSSO to
O to
vo •-
a
«t* to^tot* t*«
t* c*
**
«--00000
0 -
O
M
?|S
I
oooooooo
vo vo ON Is. Is. ^vO to
VO O
vovo
• 5
OO
vn
. a
. V
2
i
•
lii-jiiy
fj
b
c
u
1
V)
11
«2«2
I
"t
u
^wpiSQH<ity3
<<
<
^U(^^QE-<<a>
<
42 SAMUEL D. ROBBINS
give the average word scores for all of the groups taken as a
whole; the other columns give the average change scores.
In this table there are four columns each in divisions 2, 3, 4,
5, 6, 7, 8 and 9. Of these four columns the one headed L
contains each subject's average score in the first test where he
listened to the words as they were read to him; that marked R
contains those in the second test where the subject repeated
each word after the experimenter; that marked W contains
those in the third test where the subject whispered the words
exposed in the tachistoscope; and that marked Av. contains
the average of the other three columns. Division I is not
divided into four columns because the subject only whispered
those sets in which the changed word was spelled differently
yet pronounced exactly like the first word exposed in the
tachistoscope. At the bottom of this table are given the
average scores in each test for the men, for the women, and
for all eight subjects. Subjects D, E, M and W are women,
and subjects A, R, S and T are men.
This table shows that the women as a class did much
better in every test than the men and that subjects who did
especially well in detecting some change also did especially
well in remembering exactly what word appeared on each card,
the correlation1 being -f- 0.929.
This table shows also that long vowel changes are much
more readily detected than short vowel changes; it makes
little if any difference whether a long vowel is changed to
the same short vowel or to a different long vowel. Con-
spicuous consonant changes are noted almost as readily
as short vowel changes, inconspicuous ones far less frequently.
The whispering test gave the highest and the most uniform
scores for all subjects. The subject of vowel versus consonant
changes will be more thoroughly discussed in a later paper
comparing the verbal imagery of normal speakers with that
of stammerers.
Columns I, 2, 3 and 4 of Table II. give the vividness of
each subject's verbal imagery on a scale of o to 3 as
1 All correlations were derived from the formula, r = . = , where x and
y are the deviations of the two traits from their mean in any single individual
OBJECTIVE TEST FOR VERBAL IMAGERY TYPES
43
determined from his ability to detect changes by auditory
imagery, kinaesthetic imagery, and visual imagery. These
ranks are determined as follows: The visual rank is the
average score of the subject in detecting by whispering changes
TABLE II
COMPARATIVE EFFICIENCY OF DIFFERENT TYPES OF VERBAL IMAGERY
Subject
i
a
3
3-
4
5
6
7
«
9
Auditory
>
1
If
S.|
w«
Copying
Long Non-
sense
Word*
Copying
Perverted
mnd In-
verted Print
Writing
Perverted
and
Inverted
Column 1
of Table i
w..
2-5
2.8
i-9
2-3
2.1
1-9
i-5
2.0
2.4
1-7
3
1-7
0.4
0.9
i.i
O.I
O.I
5-7
S-2
5-2
5-i
ii
0.82
10.12
i-33
6.03
7.25
22.05
5
49
13
I
25
43
28
77
34
80
157
49
3
7
201
20
171
2.20
i-97
1.76
1.82
1.02
1. 06
R
T
D
S
A
Av. for women. . . .
Av. for men
2-5
2.2
1.6
1.6
i-5
0.4
S-6
4.1
3-43
10.19
3
32
21
1 06
I2S
SO
2.01
i-45
Av. for all O's
2-3
1.6
0.9
4.9
7-93
23
64
75
1.64
Correlation with co
1-4.
+0.90
without R
+0.94
without R
+0.90
+0.83
for men
+0-9S
in words that are pronounced alike but spelled differently,
such as zeat-zeet. The auditory rank is the average score
obtained by the subject in detecting vowel changes in all
three vowel groups when these are read aloud to him. Trial
experiments showed that kinaesthetic imagery played little
if any part in these groups and that it actually reduced the
score in many cases. The kinaesthetic rank is the average
score the subject obtained when he repeated after the expe-
rimenter those cards in which changes had been made in
words which sound much alike but give quite a different
kinaesthetic impression (such as vad-vab), and in which he
failed to detect changes when they were read aloud to him.
The kinaesthetic and auditory imagery are so closely connected
that I could find no better way to separate them. Trial
series showed that there is little likelihood the subject will be
able to detect a second time a slight auditory change which
he does not detect the first time by the same method. The
44 SAMUEL D. ROBBINS
changes in this group were so chosen from the introspections
on a trial series as to reduce to a minimum the auditory
element, and to make the kinaesthetic element as prominent
as possible. There is a close correlation, +0.950, between
the average ranks of subjects for the entire series as recorded
in the average column in division 8 in Table I. and the total
ranks obtained by adding together their three ranks in
columns I, 2, and 3 of Table II. as recorded in column 4.
It seems to make little difference, therefore, what type of
imagery is employed for the test as a whole; the subject
possessing the greater sum total of verbal imagery attains
the higher rank in this experiment.
Before beginning any tests, each subject was asked to
answer the questionnaire given on pages 195-200 of E. B.
Titchener's 'Experimental Psychology, Student's Manual,
Qualitative.' A careful examination of the question blanks
showed there was no correlation between verbal and non-
verbal imagery. It is incorrect to assume, therefore, that
because one has very vivid non-verbal imagery of a given
type he must also have very vivid verbal imagery of the
same type.
The four experiments which follow were performed to
confirm the reliability of the preceding test.
After the test with nonsense words had been completed,
each subject was asked to spell backward twenty words of
increasing length, beginning with two letters and ending with
twenty-two, and to report the method he employed for both
the short and the long words. The experimenter kept a
record of the time spent in spelling each word and of the
number of errors. It is obvious that the efficiency of a
subject in performing such a task depends upon both the
speed and the accuracy with which it is accomplished; the
greater the speed and the fewer the mistakes, the higher the
efficiency. If this efficiency is represented by the product
of the time in which a unit of the task is performed by the
number of errors made, the lower product will denote the
greater efficiency. These products for the six subjects who
performed this test will be found in column 5 of Table II.
OBJECTIVE TEST FOR VERBAL IMAGERY TYPES 45
It will be seen that, with the exception of R, the lower
product corresponds in every case with the higher sum total
of verbal imagery recorded in column 4, the correlation
without R being -f- 0.90. This shows that efficiency in this
test is proportional to the total amount possessed by the
subject of verbal imagery of types useful in performing the
task. R reported that he employed visual imagery alone for
this test because he believed it would increase his speed. As
he possessed the strongest auditory imagery of the six subjects
and as the trial series showed him to have about the weakest
visual imagery, he should have employed auditory imagery or
a combination of both, rather than visual imagery.
After spelling backward, each subject was asked to copy
fifteen long nonsense words of from twelve to twenty-eight
letters, to make a dash every time he looked at the copy, and
to report how he copied these words. The experimenter
kept a record of the time spent in copying each word and of
the number of errors. Column 6 of Table II. gives the relative
efficiency of the subjects in this test, this being the product
of the average time in seconds it took the subject to copy
each word by the average number of mistakes per word; and,
as in the spelling backward test, the lower the product, the
greater the efficiency. With the exception of R, the lower
product corresponds closely with the higher sum total of
verbal imagery, the correlation being + 0.94 without R. R
employed auditory imagery alone, looking at the first twelve
words but once and spelling them as he pronounced them.
As long groups of consonants made it impossible to pronounce
these words, he naturally made many mistakes.
After copying these long words, each subject was asked
to copy two or three lines of each of six type-written selections,
arranged in the following ways: perverted, backward per-
verted, inverted, backward inverted, perverted and inverted,
and backward perverted and inverted. By backward I
mean spelled from right to left instead of from left to right.
By perverted I mean written so as to be read in a mirror.
By inverted I mean written so as to be read by one facing
the writer. The subjects introspected as in the previous
46 SAMUEL D. ROBBINS
tests, and the experimenter kept a record of the time it took
to copy each line and of the number of mistakes. Column 7
of Table II. gives the relative efficiency of subjects in this
test. This efficiency is denoted by the product of the average
time per line in minutes by the total number of mistakes;
and here again the less the product, the greater the efficiency.
With the exception of T, the lower product corresponds in
every case with the higher sum total of verbal imagery, the
correlation being -j- 0.90 even with T included. T used an un-
fortunate method on the first line which caused him to make
nearly as many mistakes as he made on the other twelve lines
combined; when this one line is discarded, his score becomes
practically equal to D's.
After copying these unusual kinds of printing, the subject
was asked to write 'United States of America' so that it
would be read normally by a person facing him; 'Harvard
University' so that it would be read normally through the
paper without inverting it; 'Cambridge, Massachusetts' so
that it would be read normally through the paper inverted;
and 'European War' so that it would be read forward through
the paper without inverting it but with each letter perverted.
The subject introspected as before and the experimenter
kept a record of the time required to write each phrase and
of the number of errors. Column 8 of Table II. gives the
relative efficiency of the subjects in this test. Here the
product of the average time per letter in seconds by the
total number of mistakes gives the efficiency; so, as usual,
the smaller product denotes the greater efficiency. With
the exception of the women, who found this test far more
difficult than did the men, there is a fairly good correlation,
+ 0.83, between the efficiency of each subject in this test and
his total verbal imagery score.
As all of the spelling, copying, and writing tests just
described require the cooperation of auditory, kinaesthetic,
and visual verbal imagery, the high correlation between the
efficiency of a subject in these tests and his total verbal
imagery score (which is the sum of the scores of the separate
types of verbal imagery used by a majority of the subjects in
OBJECTIVE TEST FOR VERBAL IMAGERY TYPES 47
performing each of these tasks), shows that the efficiency of
subjects in performing a given task is proportional, approxi-
mately, to the sum of the scores of the separate types of
verbal imagery commonly employed in performing that task.
Any one wishing to use a similar but shorter test can
score auditory imagery by reading aloud to his subjects thirty
cards such as I used containing the following vowel changes:
koss-kose, nume-nurm, dack-dake, girn-gine, neff-neaf, afc-
ofe, eag-oag, ane-une, ipe-epe, fese-fose, bose-buse, nafe-nufe,
vife-vefe, obe-ube, zoke-zake; ank-enk, ald-uld, nef-naf,
ilt-ult, nop-nep, baf-bof, eft-uft, ilm-ulm, zep-zop, alk-olk,
maz-muz, oln-uln, mev-muv, tig-teg, zab-zib. These cards
should be well shuffled with one another and with a second
group containing at least twenty consonant changes which
give a distinctly different kinaesthetic impression such as the
following: ubs-uds, vab-vad, nis-niz, gem-gen, bis-tis, vot-vob,
sef-zef, tov-pov, ips-ids, zup-zut, pax-dax, mub-nub, abf-atf,
ost-ozt, emk-enk. These groups should then be repeated
after the experimenter, and in the case of the kinaesthetic
score the record should be kept only of those consonant
changes detected that were not noticed when the experi-
menter first read the second group. The third group, used
for scoring visual imagery, should contain at least ten non-
sense words which are spelled differently yet pronounced
alike such as mije-mige, bick-bik, fis-fiss, gax-gaks, veet-veat,
doxe-doax, klab-clab, zoll-zol, boze-bose, sibe-cibe. After
the subject has thus repeated the first and second groups
after the experimenter, the third group should be shuffled with
ten pairs from other groups and exposed in a tachistoscope
long enough for the subject to read each card aloud. The
experiment should be conducted and the results scored as
outlined at the beginning of this report.
SUMMARY
1. Women possess more vivid verbal imagery than men.
2. Long vowels receive more attention than any other
letters.
3. Short vowels receive much less attention than do long
vowels.
48 SAMUEL D. ROBBINS
4. Consonants receive less attention than vowels.
5. A change of one letter in a pair of nonsense words can
be detected most readily if the subject reads the words aloud
from copy.
6. Those persons will perform a given mental task most
efficiently who possess the most vivid types of verbal imagery
commonly employed to accomplish that task.
NOTE ON THE VERBAL IMAGERY OF STAM-
MERERS AND NORMAL SPEAKERS.1
I stated that the subject of vowel versus consonant changes
would be discussed in a later paper comparing the verbal
imagery of normal speakers with that of stammerers. As my
experiment shows that the average verbal imagery for twelve
normal speakers and twelve stammerers is practically iden-
tical, a brief note will suffice to summarize my results.
The same objective test described in the above named
article was given to four additional speakers, two of them men,
and two women, and to twelve stammerers, six of them men,
and six women; and the average score for the twelve normal
subjects and the twelve stammering subjects was compiled as
in my earlier article.
The average change score showed that the stammerers de-
tected changes a very little ^better than the normal speakers
when the card pairs were read to them, 1.7 compared with 1.6,
and that the normal speakers detected changes a little more
readily when they looked at the card pairs and whispered
them, 2.0 compared with 1.8. They averaged the same, 1.5,
when they repeated the card pairs after the experimenter, and
averaged the same, 1.7, when all three methods of presenta-
tion were averaged together.
The vividness of each subject's visual, kinaesthetic, and
auditory verbal imagery was determined as in the earlier paper
and the stammerers and normal speakers were found to have
the same average auditory imagery, 2.2, and the same average
kinaesthetic imagery, 0.8. The visual verbal imagery of the
1 Prepared after the article was in type.
OBJECTIVE TEST FOR VERBAL IMAGERY TYPES 49
stammerers was but 75 per cent, of that of the normal speakers,
however, 1.2 compared with 1.6. As few persons employ
visual verbal imagery to any extent in speech, there is obvi-
ously nothing in verbal imagery of any kind to account for
stammering. There is a suggestion, however, that certain
letters may attract the stammerer's attention unduly; this
problem I am now investigating. The following facts came
out in this investigation.
The stammerers did not detect the long vowel changes so
readily as the normal speakers, 2.4 compared with 2.7 in the
listening test and 2.2 compared with 2.6 in the repeating test;
yet they detected the short vowel changes more readily, 1.9
compared with 1.6 in the listening test and 1.7 compared with
1.4 in the repeating test.
Averaging the three series composed of the three types of
vowel changes I found that the stammerers and normal speakers
scored alike, 2.2, on the listening tests, and that the stammer-
ers did not do quite so •well as the normal speakers on the re-
peating tests, 1.9 compared with 2.1, or on the whispering
tests, 2.4 compared with 2.5.
Averaging the three series composed of the three types of
consonant changes I found that the stammerers excelled in the
listening tests, 1.15 compared with 0.95, and on the repeating
tests, 1.05 compared with 0.95, but scored only 1.44 compared
with the normal speakers' score of 1.68 on the whispering tests,
the difference being due, no doubt, to the better visual imagery
of the normal speakers. The average for the three methods of
presentation was the same for normal speakers and stammer-
ers, 1.19. The maximum and the minimum scores averaged
the same for stammerers and normal speakers in the vowel
changes and the maximums averaged the same in the con-
sonant changes, but the normal speakers' minimums averaged
twice as low as the stammerers in the consonant changes. It
would seem, therefore, that stammerers pay more attention to
consonants than do normal speakers; this is confirmed by the
work I have done in connection with the correction of stam-
mering.
BY J. R. KANTOR
University of Chicago
Recent developments in the study of human behavior
make it possible to begin a reinterpretation of instincts and
related phenomena which today admittedly constitute the
darkest chapter in psychology. In this paper the writer
attempts to suggest a functional interpretation of human
instincts and their integration into instinctive conduct.
The functional psychologist aims to start from an unbiased
naturalistic standpoint and therefore hopes to achieve some
progress in the understanding of some of the adaptationai
equipment of human beings. At the very inception of such
a study we observe the imperative necessity for a scrupulous
discrimination between the acts which are properly called
instincts, and the more complex reactions developed from
them which we will call instinctive conduct or behavior.
The Nature of an Instinct. — An instinct is a comparatively
simple and direct response to a specific stimulating object or
condition. It is in fact the functioning of a connate potential
reaction system1 which is organized from simple psychophysio-
logical dispositions or tendencies to respond to stimuli.
That instincts are so highly spontaneous may be accounted
for by the fact that the specific way in which the reaction
system functions, depends upon the stimulating conditions.
It is this moulding of the response by the surrounding con-
1 A reaction system is a complex function involving cognitive, conative, affective,
muscular, glandular and neural factors. Cf. Kantor, 'Conscious Behavior and the
Abnormal,' /. of Abnorm. Psychol., Aug., 1918. An example of a reaction system is the
response, 'August, 1914' (with all its accompanying organic resonances) to the stimula-
tion, 'when did the hostilities of the great European War begin?' This response is
potential in all those who have acquired the informational reaction.
50
HUMAN INSTINCTS S1
ditions which is the source of the many marvellous tales of
intelligence among the lower animals.
An instinct being a primary act and therefore entirely
"undebauched by learning," must be looked upon as one
of the primary functional elements in the embryological
development of the human organism. For the instinctive
reaction patterns are functions of animal adaptation de-
veloped from the simple functions of organized matter.1
Owing to this development instincts may be classified as
(i) food-getting, and waste eliminating responses, (2)
sexual reactions, (3) expressive acts, and (4) protective
responses. These classes represent specific adaptations to
particular adjustment-situations, that is to say concrete
actions, and with the random movements and reflexes form
the matrix of the entire series of human behavior.
The function of human instincts is to adapt the person
to the various surroundings in which he is found, pending the
development of the intelligent responses usually required
for such adaptations. These modes of instinctive response
develop in the species of organism during its interaction with
its environment; consequently there is an entirely natural
genesis of the instincts paralleling the growth of the human
being in the evolutionary course of the animal species to
which he belongs. Every organism possesses a series of
these reaction systems which in the presence of adequate
stimuli become responses. The response and the stimulus
together constitute an act, that is to say, a specific adaptation.
From a definitively psychological standpoint the individual
at any particular moment is this series of reaction systems.
If we have correctly described the origin and development
of instincts, we have sufficiently indicated that the instincts
of the human organism are very different from those of the
lower animals. The reaction systems, as the units of the
organism on the action side, must naturally be just as
diverse in dissimilar organisms as are the structural parts.
Thus we find differences of a wider or narrower sort in both
the mental and physiological factors of the specific functions.
1 These are usually described by the zoologist as irritability, metabolUm, repro-
duction, motility, etc.
52 /. R. KANTOR
Obviously, the most striking variation between human and
animal instincts is the extreme modifiability of the former.
In fact, human instincts are so distinctly transitory in char-
acter that they disappear very early from the reaction equip-
ment of the human organism, and in the adult individual are
completely absent. These human instincts become inte-
grated into more complex types of responses, while the animal
instincts remain as permanent acquisitions of the organism,
and change only by becoming more adaptable through prac-
tice to the situations in which they frequently function.
The Nature of Instinctive Behavior. — In contrast to the
instincts, instinctive conduct comprises adjustments which
are essentially acquired tendencies of response, and in most
cases constitute intelligent behavior. It must be noted,
however, that the reaction systems of instinctive conduct,
which, by the way, include the greatest portion of our actual
responses, are developed as elaborations of a prominent core of
organized innate reaction patterns. In all cases of instinctive
conduct we have integrations of concrete human acts; so
that if, for example, we start with the walking act of a child,
the exigencies of the surrounding conditions may condition
that initial act to become a locomotor response to the call of
the parents, or to any other stimulating circumstance acting
upon our illustrative child. The results of observations of
human behavior demonstrate that the rapidity and com-
plexity of the integrations are owing to the responsiveness of
surrounding objects; that is to say, a responsive object
forces the individual to apprehend the possibility and neces-
sity of varying his response, and therefore to learn to react
with a meaningful behavior to the stimulating response of
the other object. This sort of interaction with the environ-
ment constitutes the basis for social phenomena of various
types and is excellently illustrated by the constant inter-
stimulation between two boys during the preliminaries of a
fistic combat. In this situation each individual is intently
posed in an anticipatory attitude, requiring only the slightest
sign of change in position on the part of the opponent as an
effective stimulus to bring about a telling response. When
HUMAN INSTINCTS 53
we recall that a single individual can serve both as stimulating
and responding object we can appreciate the importance of
this self-stimulation as a factor in the rapid integration of
the simpler forms of behavior.
Since the natural environment of the human organism
consists primarily of responsive objects we see why the human
adult has no instincts, that is, performs no acts which are
actualizations of exclusively innate dispositions, but always
responds with a partially acquired reaction pattern. The
view that man has more instincts than the lower animals, for
which James1 is in part responsible, could only obtain credence
so long as the precise nature of a conscious act remained un-
analyzed. That James did not entirely ignore the facts
concerning instinctive conduct is manifested by his observa-
tion that human instincts do not remain blind.2 The phy-
siological viewpoint, which always influenced James, pre-
vented him from fully appreciating the psychological changes
which transform instincts into more complex actions. To
think of the non-rational activities of the human organism in
terms of reflexes which are somehow coupled with impulses,
means the capricious disavowal of the variety and richness
of the instinctive forms of behavior.
The contrast between instincts and instinctive behavior
is made clearer by dispelling somewhat the confusion existing
in the conception of the differences between the instincts and
the more simple reflexes which differ widely from the former
in organization and function. The reflex action involves
the functioning of a more definite and fixed reaction system
than does the instinct, and the result of the stimulation is a
genetically simpler form of behavior. The relative rigidity
of the reflex response allows comparatively little room for
adjustment between the organism and the stimulating
conditions while the action system is functioning.
Further, it has been frequently observed that instincts
involve a much larger conscious function than is the case
with reflexes,8 since the latter are on the whole much simpler,
i Cf. 'Principles,' pp. 393, 441.
* Ibid., 390.
» Stout, 'Manual of Psychology,' 1915, p. 343.
54 /• R- KANTOR
but we must guard against the idea that reflexes are merely
neuro-muscular actions.1 While Stout is entirely correct in
his assertion that * instinctive conduct does, and reflex action
does not presuppose the cooperation of intelligent conscious-
ness,' he is mistaken in supposing that the absence of intelli-
gent consciousness implies the complete absence of a conscious
factor in the response.2 Instincts and reflexes imply, then,
the functioning of two distinct types of connate reaction
patterns, both of which are to be distinguished from instinctive
conduct which is never the functioning of a purely innate
reaction pattern, although it is to a certain degree developed
always from instincts.
II
The Range of Instinctive Conduct. — The distinction be-
tween instinctive conduct and instincts paves the way for
the consideration of the large place which the former holds
in human life. We have already suggested that most of
our ordinary behavior is instinctive conduct, but this does
not mean in any sense that complex actions such as we perform
are the expressions of a few inborn impulses. Such a manner
of thinking represents a vestige of scholastic simplicity which
is genuinely subversive of all understanding of human
behavior. What is meant is that even our very complex
actions are in great measure conditioned by the instincts from
which they have developed. To be sure, the simplest
instinctive conduct is very largely the functioning of an
innate reaction system, although conditioned by acquired
factors. The proportion of innateness in the reaction pattern
is measured by the directness of the connection between the
stimulus and the response, or in other words by the character
of the appreciation which the individual has of the meaning
or significance of the stimulating object. In the simplest
case the meaning of the object does not emerge as a striking
factor in the act; it merely represents a modification in the
response owing to a previous contact with the stimulating
object. In a general way, we may very properly consider
1 Cf. Stout, Brit. J. of Psychol, 3, p. 244.
* Stout's separation of the conscious and movement components of a response
clearly exemplifies the difference between his position and a functional viewpoint.
HUMAN INSTINCTS 55
the simplest instinctive behavior as called out by the environ-
ment, and largely controlled by it, and not by the organism.
As examples we may quote all those activities usually de-
scribed by psychologists as subconscious or unconscious,
which are very prominent in manual learning, and technical
operations of all sorts.
On the other hand, the more complex instinctive conduct
is more independent of the stimulating object and includes
in its reaction systems a larger component of acquired factors.
Here the meaning of the object serving as a definite foresight
of the act, functions in a more precise manner, and in still
more developed behavior includes an effective appreciation
of the consequences of past responses to stimulating objects.
The instinctive behavior at this stage may involve an elab-
orate series of memorial and thought functions, and when so
complicated its specific characteristic as an instinctive be-
havior is the fact that it is perceptually stimulated, that is,
the act is not initiated by a problematic situation. In this
last class we may place all the involved social behavior which
constitutes many of our daily responses. We must conclude,
then, that instinctive conduct composes a considerable
portion of practically all adjustments from the simplest to
the most complex.
The Intelligence in Instinctive Behavior. — We may sum
up the essential characteristics of instinctive behavior by
pointing out the invariable presence in it of at least the rudi-
ments of intelligence. Thus in many cases the reaction
system, although a response to an immediately presented
perceptual stimulus, is still carried out by predominantly
acquired reaction factors, as is convincingly exemplified by
much of our socially restricted behavior. Such acts are
spontaneous responses to definite perceptual stimuli, but
they are performed in roundabout ways and in many instances
tend toward concealment. The openness and frankness with
which such acts are originally performed are by virtue of
social disapproval more or less successfully repressed.
Distinction of Instinctive from Rational Conduct. — The
great variety and complexity of instinctive responses make
56 J. R. KANTOR
it necessary to distinguish them from rational acts, a dis-
crimination which is all the more pertinent when we consider
that in the final analysis all of our acquired reaction systems
are at some level integrations of elementary instinct acts.
As a response to a problematic situation the rational act is
probably always initiated by an indirect stimulating object
through some highly developed meaning function. Unlike
complex instinctive behavior the rational act is not only
guided to its conclusion by intelligent functions, but is
originated by a reflective consideration of ways and means.
Thus it becomes the basis for all transformative conduct,
that is, action which remakes the environmental conditions
through some function of creative imagination, while in the
case of instinctive conduct the result is usually merely an
adaptation to those conditions.
Ill
The Specificity of Instincts. — Whether or not instincts
are specific in their functioning is a crucial inquiry for the
understanding of them, and a problem which may throw
considerable light upon the distinction between instinctive
behavior and instincts. It is important to note that since
instincts are simple and immediate responses to specific
stimuli which bring innate action systems into function,
they presumably must be specific in their results.
This view, however, is not generally held by psychologists,
although some adhere to it so tenaciously that the obser-
vation of the indeterminateness and indefiniteness of human
behavior influences them to deny the existence of instincts
in the human being. While it is entirely demonstrable that
mature persons possess no instincts, this must not be inter-
preted to mean,>s Stout does, that human behavior in general
has no instinctive foundation in the form of concrete action
patterns.1 To believe in the absence of instincts in the human
1 Our interpretation of Stout's position is in no wise invalidated by his reluctant
inclination, expressed in the third edition of his 'Manual' (p. 360), to make the term
instinct refer to general capacities, such as 'innately organized interest,' 'attention,'
and 'power of learning by experience in certain directions.' On this basis he asserts
that 'the whole development of human minds has its root in connate tendencies of this
sort and is inexplicable apart from them.' From our standpoint in appears that
Stout is here avoiding the essential problem of instincts.
HUMAN INSTINCTS 57
individual because instinctive conduct is contrasted with
intelligent conduct is to overlook entirely the facts (i) that
we are studying concrete conscious behavior, and for that
reason we need not think of an instinct as a permanent spring
of action, the absence of which at the present time indicates
that it was never present; and (2) that intelligent behavior is
developed by the integration of simple types of action, a fact
which enables us to understand how the reaction pattern of an
instinct becomes elaborated and developed into a complex
intelligent response.
An inquiry into the views entertained concerning the
definiteness of instincts reveals the fact that what is frequently
meant by an instinct is a neuro-muscular function. Thus
Stout, for example, describes an instinct as a "purely biologi-
cal adaptation comparable to the prearrangement of structure
and function which in human beings subserves the digestion
of food."1 Upon examining this conception we are impressed
with its inadequacy to represent human behavior, although
we are in hearty agreement with Stout in rejecting such a
view as that of Bergson-Carr, stated by Stout as the belief
that there is a special form of psychical activity which re-
quires the technical name of instinct.2 We insist that not
because human behavior has no instinctive basis do we not
find instincts, but because the latter have become developed
into intelligent behavior in the course of the individual's
contact with his surrounding conditions. This fact Stout
could have seen had he not been prevented by his general
psychological standpoint from appreciating that the psy-
chologist is interested in modes of response to stimuli, and
not in expressions of mentality. Apparently, Stout assumes
the specificity of instincts, and from such a premise he
concludes that there are no instincts in the adult human
being because he does not find man performing acts which
express a mental process, through innately coordinated motor
mechanisms.3 Stout consequently fails to appreciate the
1 Brit. J. of Psychol., 3, 243.
*Ibid.
» Cf. McDougall, Brit. J. of Psychol., 3, 269 ff.
58 . /. R. KANTOR
large place which instinctive conduct plays in the life of the
human individual.
When we turn to the work of Thorndike,1 who is attempt-
ing to investigate the * original nature of man/ we find much
to commend in his description of the specific instinct re-
sponses. Beginning with the admirable intention to describe
concrete facts of behavior, he scouts the viewpoint which
makes of instincts generalized tendencies to bring about some
vague result presumed to be beneficial to the organism.2
Thorndike stands upon firm scientific ground when he looks
upon instincts as specific types of unlearned responses to
definite kinds of stimulating situations, but his work presents
us with grave difficulties. Conceived in neuro-biological
terms, it implies that man's 'original nature' remains forever
a prominent part of his behavior equipment. From this
fact arise several implications tending to misconstrue the
actual character of instinctive behavior.
In the first place, such a viewpoint cannot escape the
implication that the human individual acts precisely as does
the animal, since the former is fitted with a similar sort of
neuro-muscular structure, and secondly, a more serious diffi-
culty is that such a position leaves no room for the develop-
ment of behavior.
The first difficulty must be understood as referring to
the obvious faultiness of the attitude that human behavior is
permanently like that of the lower animals. It is true that in
the case of infants the acts are like those of the simpler or-
ganisms, but this is because we are observing simple in-
stincts. In older children and adults the behavior has
become integrated into intelligent conduct and is thus quali-
tatively different.
In answer to the possible reply of Thorndike that a suffi-
cient differentiation of conduct in man and animals is allowed
for by the combination of neural elements, we might suggest
that such a way out of the difficulty would only result in
describing complex abstractions instead of observable be-
havior. Human conduct is infinitely more complex in every
1 Educational Psychol., 1913, Vol. i.
2 Cf. James ('Principles,' 1890, II., 183), whom Thorndike follows.
HUMAN INSTINCTS 59
phase of adaptational character than can be accounted for
on the basis of the combination of neural elements. Ob-
viously the neural connections are essential mechanisms in
all behavior, and since the activities of man are more complex
than those of animals, these mechanisms must necessarily
be more elaborate, but the nervous function cannot do
anything more than mediate the spontaneous movements of
the individual.1 It is because the neural hypothesis was
developed in connection with work on* animal instincts that
it has any significance as an explanatory principle, inasmuch
as the animal instincts are very simple activities and so
lacking in intelligence as to be almost mere biological func-
tions.
We cannot agree with Drever,2 who is essentially a follower
of McDougall, in his criticism that the lack of differentiation
in Thorndike's theory between human and animal instincts
points to the nonspecificity of instincts. Drever insists*
that there is no genuine specificity in Thorndike's instincts,
since, for example, the 'instinct to escape from restraint* is
so complex as to involve in the case of a little child ' stiffening,
writhing, and throwing back the head and shoulders' and
in the older child also 'kicking, pushing, slapping, scratching
and biting.' Drever declares that the instincts mentioned
belong with the six others enumerated by Thorndike4 under
McDougall's heading of pugnacity, and that the precise factor
of unity is the accompanying emotion of anger. As a further
argument against the specificity of instincts, Drever indicates
that in some cases we cannot predict what a specific response
will be, and the individual may try many different ones in
succession. Thus, for example, under some conditions of
stimulation the person may respond by flight or concealment,
and in some cases by both reactions in turn.
1 It is unfortunate that psychologists appear to overlook the fact that constructive
biologists do not think in terms of isolated nerve functions, but in terms of neuro-mus-
culo-glandular systems. In this connection it appears that if Thorndike has avoided
'mystic potencies' ('Educ. Psychol.,' p. n) he has done so only by translating them
into neural terms.
1 'Instinct in Man,' 1917, p. 155.
» Op. ft/., p. 166.
4 Op. cit., p. 68 ff.
60 j. JR. KANTOR
The writer is satisfied that instead of proving the non-
specificity of instincts, what Drever really shows is that
human beings respond only by means of instinctive behavior
and not with instincts. To repeat, instead of responding
merely with an innately organized reaction system, the
individual reacts with a complex acquired reaction pattern,
which in the course of his development has had an increased
knowledge and affective factor added to it. It is for this
reason that the anger or fighting situation calls out such a
wide and varying series of actions. In order to explain such
conduct it is entirely unnecessary to invoke a dubious inter-
pretation involving an unwarranted conception of the nature
and function of the emotions as Drever following McDougall
does.
Drever seems to realize that human behavior is a com-
plex function developed in interaction with stimulating cir-
cumstances, when he writes that "behavior will be largely
determined, first of all, by the circumstances of the case, by
what kind of response will best secure safety. It will be
determined, in the second place, by the intensity of the fear
aroused, and two individuals may behave in two entirely
different ways in response to the same situation, according
to the degree of fear aroused."1 This unimpeachable obser-
vation, which certainly controverts Thorndike's position
that an instinct (instinctive behavior) is the functioning of a
neuro-muscular apparatus, should have led Drever to see
that the actions which he quotes from Thorndike's descrip-
tion are phases of intelligent instinctive conduct, and not
the expressions of a mysterious 'general' instinct. Were
Drever thinking in terms of concrete behavior he would
easily see that instinctive conduct is not the functioning of
an 'end,' or 'instinctive impulse' with an intelligence entity
to carry it out,2 but that it is a definite response to a stimulus
which involves in its specific mode of action the integration
of numerous previous experiences. In all cases of actual
instinctive behavior the 'end' is gratuitously imposed upon
1 Op. cit., p. 163.
* Ibid., p. 122 ff.
HUMAN INSTINCTS 61
the situation.1 What actually happens is that at any par-
ticular time certain combinations of surrounding circum-
stances stimulate the person to perform definite acts, provided
that he has the necessary equipment of reaction systems.
The adequate consideration of the stimulating auspices of
behavior entirely removes the necessity of postulating teleo-
logical powers in the organism.
The spectator may profit by Drever's attack upon
Thorndike's position by observing that on the one hand in
his endeavor to avoid "mystic potencies" Thorndike refuses
to interpret behavior as it actually occurs, preferring to lean
upon unreal if not mystic potencies, while on the other,
Drever,2 following McDougall,3 describes behavior in a more
acceptable manner, but does not hesitate to explain it as the
result of metapsychological agency.
The second difficulty with Thorndike's view of instincts,
namely, that it disregards the development of behavior, may
be considered as a derivation from the first difficulty. It
results in the misinterpretation of human action, which as
we have seen has as its primary characteristic the process
of integration. A critical study of such behavior indicates
conclusively that not a single act of an adult person4 is an
original response, but always a complex development of
acquired reaction systems. It appears that Thorndike must
think of instincts in the adult as drives or potencies of some
sort, that is to say, at this point they have lost their specific
character. In his failure to distinguish between instincts
and instinctive conduct Thorndike vitiates his original
excellent intention to describe actual psychological occur-
rences. Consequently, his interpretation leaves unfulfilled
1 In this connection it is extremely edifying to observe the highly moral ends that
are sometimes imposed upon the instincts, such as 'heavy and unremitting toil on
behalf of the offspring' in the case of the parental instinct. Cf. McDougall, 'Soc.
Psychol.,' 1916, p. 269.
2 "Instinct is the 'life impulse' becoming conscious as determinate conscious
impulse," op. «'/., p. 88.
1 "For I hold that the instincts are essentially differentiations of the will to live
that animates all organisms and whose operation in them makes the essential difference
between their psychophysical activities and the physical processes of inorganic nature."
Brit. J. of Psychol., 3, p. 258.
4 Excluding the reflexes, of course.
62 /. R. KANTOR
his original functional promise, and ignores therefore one of
the extremely important factors in conscious behavior,
namely, the stimulating circumstances. In not allowing for
an interpretation of the actual responses which an organism
makes in adapting itself to surrounding conditions, Thorn-
dike's position results in an inert structuralism which pro-
longs the intellectual tradition of a permanent self.1
The most zealous advocate of the non-specificity of
instincts is probably McDougall who approaches the problem
from the angle of social behavior. This author, impugns the
theory of social action which assumes 'that man is a reason-
able being who always intelligently seeks his own good and
is guided in all his activities by enlightened self-interest.'2
Unfortunately McDougall's easy victory over such a vul-
nerable position has resulted in his substitution of another
absolute spring of action as the basis for all human behavior,
namely the series of instincts. The ubiquity and persistence
of certain types of action no doubt has influenced him to
propound the hypothesis that the human 'mind' is consti-
tuted by the sum of innate tendencies which bring about
the specific actions of the individual. As will appear in the
course of our discussion these tendencies McDougall believes
to be permanent psychic entities. The assumption that
instincts are the 'essential springs or motive powers of all
thought and action' necessarily implies therefore that they
are general capacities to bring about certain actions, for
otherwise there would be required an infinite number to
account for all the variety of social behavior.3 The suppo-
sition of the non-specificity of instincts in turn creates a
presumption in favor of the perseverance of dispositions as
permanent tendencies of human actions.
An impartial investigation of behavior clearly demon-
strates the extravagance of assigning any absolute foundation
for human conduct. Thus for example, to insist upon
instincts as the exclusive springs of action is to lose sight of
1 As a series of physiological mechanisms.
»'Soc. Psychol.,'p. ii.
» "Lightly to postulate an indefinite number and variety of human instincts is a
cheap and easy way to solve psychological problems." 'Soc. Psychol.,' p. 26 ff.
HUMAN INSTINCTS 63
the actual fact that many human actions are in a genuine
sense rationally motived. As a consequence of seeking an
absolute factor in human behavior, McDougall reaches the
same result as Thorndike, namely, a form of abstractionism
which adds little to the comprehension of such behavior.
The impuissance of McDougall's conception of instincts
as an interpretation of conduct is instructively intimated in
the existence of an uncrossable barrier between his exposition
of instincts and his discussion of social behavior. Although
he starts out with the assertion that an instinct is a psycho-
physical disposition, not only to act but also to perceive,
attend, and feel, that is to say, a concrete action,1 he really
thinks of it as an enduring condition or faculty of some sort.2
The hypostatic nature of McDougall's thinking appears in
its most overt form in his protest against using the term in-
stinct to denote an action.3 There is apparently no way in
which such instincts can develop into complex social behavior
excepting by some form of crude mechanical agglomeration.
In all fairness to McDougall it must be said that he realizes
the appalling chasm which separates his instincts from the
complex behavior of the social type, for he develops a theory
to account for the fact that instincts,^.while substantial ele-
ments, can still be the basis for all complex human action.
This theory assumes that an instinct can be divided into
'three corresponding parts, whose activities are the cognitive,
the affective and the conative features respectively of the
total instinctive process.'4 Now the emotional factor is
assumed to be unmodified throughout all the various changes
which involve the other two factors;5 so that not only can
you find the same dispositions in animals as in man, but in
man they can develop to any possible degree. Unfortunately
for McDougall this theory glaringly exposes his indefensible
position. For note, he allows for so much development in
*'Soc. Psychol.,' p. 26 ff.
1 Brit. J. of Psyckol., 3, 253. It seems clear that McDougall does not hold
that the enduring condition of an instinctive act is a definite potential reaction system,
that is to say, a concrete response pattern which will function when stimulated.
• Brit. J. of Psychol., 3, 253.
«'Soc. Psychol.,' p. 32.
• Ibid., p. 34.
64 /. R. KANTOR
the dispositions that he almost gives up the idea that the
springs of human actions are innate. In extricating himself
from this difficulty McDougall further weakens his position,
since in making the emotional aspect of the instinct the sole
innate spring of action he runs counter to the fact which he
himself admits,1 namely, his inability to point out definite
actual emotions in any but the 'principal' powerful instincts.
The precariousness of McDougall's position is not at all
mitigated by his highly questionable identification of the
affective component of an instinct with an emotion.2
The conclusion that we may draw from the imperfection
of McDougall's view is that he does not fully realize that he
is attempting to interpret instinctive conduct, which is an
entirely different matter from demonstrating the function of
instincts in all the complex actions of human beings. He
therefore starts from the wrong premises and is easily led to
the bizarre idea of the substantial mental character of in-
stincts. It is an egregious error for McDougall3 to think
that he is alone in believing that instincts are at the founda-
tion of our mental life. It is almost a universal conviction
among psychologists that all human behavior is based upon
instincts as a foundation, but the important point is, that this
foundation as an actual phenomenon is only a transitory
phase of a maturation process.4 For instance, when we
observe the fighting reactions in the child and in the adult
we are severely impressed with the qualitative difference of
the respective reactions. In the first place, the specific
responses in each case are different, implying that no enduring
nervous basis can be inherited for the purpose.5 Again, the
stimulating situations may be absolutely different not only
in the developing individual at different stages, but also in.
the same stage of growth at different moments, and in dif-
ferent individuals at the same moment. Not only will a fear
situation in any of these cases call out different sorts of
1 Ibid., p. 46.
8 Cf. Drever, op. cit., p. 156 ff.
3 Brit. J. of PsychoL, 3, 260.
4 In the same sense as the foetal structure which is the foundation for the adult
physique is integrated in the course of development.
* Cf. 'Soc. Psycho!.,' p. 29.
HUMAN INSTINCTS 65
responses in the individual, but it may call out the same sort
as an anger situation. In all cases the response of the indi-
vidual will depend in part upon the multiplicity of circum-
stances immediately surrounding him. The fighting reac-
tions, for example, will depend upon the presence or absence
of onlookers and the regard one has for them if they are
present, as well as upon the thing at stake in the contest.
Such reaction will also be conditioned by all sorts of technical
information and convictions one has acquired relative to
fighting in general or to fighting under these specific circum-
stances. When we observe a complex social action we are
convinced that even such elaborate and significant sugges-
tions1 which the penetrating students of human conduct
enumerate, cannot fully cover its conditioning influences,
although of course for some definite purposes not all of these
influences are relevant.
McDougall's discussion involves the gratuitous assump-
tion that the substantial instinct entity can be aroused under
very different kinds of circumstances. For instance, at
one time the instinct can be aroused by a natural stimulus and
at another by a complex social situation, all through the
medium of an emotion entity. There is apparently more
than a smack2 of the old faculty psychology in McDougall's
thinking, a fact which is genuinely surprising when we con-
sider that at certain points he almost realizes the distinction
between instincts and instinctive behavior, as for example,
in differentiating between the specific and general tendencies.*
McDougall's insistence upon the generality of instincts
is based therefore upon the dubious premise that there are a
few innate springs of all human conduct, rather than upon
the observation that human behavior is a complex inter-
action of an experienced and intelligent person with a multi-
plex environmental situation. It is the very complexity
of the total situation that seems favorable to the arbitrary
analysis of it into a few constant factors. This is familiarly
illustrated in the case of the complicated social and institu-
1 Such as are found in Veblen, 'The Leisure Class,' etc.
1 Cf. Drever, op. cit., p. 16.
» ' Soc. Psychol.,' p. 20.
66 /. R. KANTOR
tional circumstances which are reduced to a few simple
activities of the ' economic man.' McDougall1 has gone only
a step farther than Cousin, whom he severely criticizes, in
the interpretation of the conditions of human activity,
because the former fails to see that the dispositions to human
action are all complex acquired functions and not a com-
pound system of original sensori-motor arcs,2 plus some type
of antecedently functioning mental activity.3 The explana-
tion of McDougall's doctrine of generalized instincts seems
to be the fact that he stands for a theory of psychological
predestination,4 and so he makes of the human individual a
machine fitted with definite powers which require only an
indifferent stimulus to make them perform whatever seems
necessary to be done.5 Although he condemns the practice
in others, McDougall ascribes to the functioning of an instinct
any frequent or constant form of action. Thus, the acqui-
sition and building up of large estates are attributed to the
acquisitive instinct.6 It is a queer doctrine of magical
potencies which can describe the development of such elabor-
ate institutions7 as we have in our complex life to the func-
tioning of a dozen or so of instincts. And more anomalous
still is the presentation of such a doctrine in face of the
overwhelming facts pointing to the shaping of our instinctive
behavior, by the lives and acts of persons and institutions.8
The entire controversy concerning the specificity of in-
stincts is made possible only by an inclination toward a
structural psychological position. When we take concrete
1 Ibid., p. 12 ff.
1 'Soc. Psychol.,' p. 29.
8 Conative tendency — cf. Brit. J. of Psychol., p. 261 ff. "The instinctive impulses
determine the ends of all activities and supply the driving power by which all mental
activities are sustained." 'Soc. Psychol.,' p. 44.
* "I hold to the reality of teleological determination of human and animal be-
havior." 'Soc. Psychol.,' Preface, second edition.
6 The writer wonders whether McDougall considers the instincts as such absolute
springs of action that they function either as determining the ends of all actions or
merely by being suppressed. Cf. McDougall's discussion of the parental instincts,
'Soc. Psychol.,' p. 267 ff.
6 Cf. 'Soc. Psychol.,' p. 323.
7 "These impulses are the mental forces that maintain and shape all the life of
individuals and societies." 'Soc. Psychol.," p. 44.
8 Cf. Woodworth, 'Dynamic Psychol.,' p. 72 ff.
HUMAN INSTINCTS f
human behavior to be the province of psychology we are
very soon impressed with the fact that instincts are neces-
sarily specific in their functioning, but that the adult indi-
vidual has no instincts. Furthermore, the obvious generality
and unpredictability of adult behavior should lead us to ob-
serve that instinctive conduct is general because the environ-
ing conditions to which it is responsive are incessantly
variable in their stimulating capacities.
IV
Relation of Instincts and Emotion. — The study of instinc-
tive conduct has in recent years resulted in the almost uni-
versal agreement of psychologists that a very close relation
exists between such behavior and emotions, although there
are several doctrines as to the precise details of this relation-
ship. It is held, on the one hand, that emotions are of instinc-
tive origin and occur when the instincts are checked or in|
conflict, while on the other, it is believed that emotions are
the correlates of instincts in some form. It must be granted
that both these views are based upon observable conduct,
and especially the fact that in many cases of instinctive
behavior a powerful feeling element is involved; the impor-
tance of the data, however, intensified by the lack of uniform-
ity in interpretation, demands a more adequate analysis.
The view that emotions are correlates of instincts is ably
championed by McDougall, who, as we have seen, believes
that the primary emotion is the affective element of the
instinct. The primary objection to such an interpretation,
as we have also seen, is that many instinctive actions do not
involve emotions,1 and that many emotional situations da
not have such instinctive associates as are so convincingly
discussed in the cases of anger and fear. As we have re-
marked above, the insistence upon the invariable presence of
an emotion in every instinctive act is to reduce emotions in
many cases to simple affective states.2 It is significant that
Drever, who closely follows McDougall, is forced to the
1 Cf. Shand, 'The Foundation of Character,' p.. 6, 370.
* Cf. Drever, op. cit., p. 155 ff.
68 /. R. KANTOR
conclusion that only some instincts have emotional accom-
paniments.
The conflict theory, insofar as it insists upon a conflict
situation as the basis of an emotional behavior, meets with
few if any exceptions in fact, but the question arises whether
the conflict is in all cases a conflict of instincts. The critical
analysis of the emotional situation indicates that this is
not true. Before proceeding to such an analysis of emotional
behavior it is well to describe its chief characteristics.
An emotion is an interrupting form of response to a
suddenly presented stimulus in which various organic pro-
cesses are put into function, which in turn facilitate the imme-
diate performance of a new act. Among the outstanding
features of an emotional action are the confusion and excite-
ment which pave the way for a new act by inhibiting the
behavior which is taking place when the emotion-exciting
stimulus is presented. Naturally such an act is replete with
affective and organic resonance, and here we find the clue to the
relationship between the emotional and the instinctive types
of behavior.
What happens in the case of the emotional situation is
that a dissociation of the reaction systems of the person
takes place; so that in the most violent type of emotion the
person is left only with the capacity to act with almost purely
physiological (reflex) behavior. From this extreme case we
find a gradation of emotional conduct in which the disor-
ganization leaves free to function a series of reaction patterns
ranging from the simplest to the most complex instinctive
behavior. The fear and anger situations offer excellent
examples of the disturbances of behavior which leave com-
paratively simple forms of instinctive response to adapt the
individual. ..A person may be walking along through a
wood, perhaps thinking over some problem, when suddenly
there is a cessation of the thought activity and the person
finds himself in a state of great excitement and unpleasant-
ness, and in readiness to run from a tiny creature madly
scurrying through the brush. In this illustration, the simple
instinctive, danger-avoiding response might appear, as the
HUMAN INSTINCTS 69
most serviceable form of behavior under the circumstances.
We see, therefore, that not because an emotion is a component
of an instinct or a conflict between instincts is it closely related
to instinctive behavior, but because under certain conditions
of stimulation the reaction systems are so disorganized as to
leave only some instinctive mode of behavior to function.
Upon the basis of such an interpretation we can under-
stand the more refined emotional responses, namely, those in
which no violent instinctive reaction is involved. In the
functioning of the more subtle emotions1 the environmental
circumstances are such as to disturb only the most elaborate
and definitely focused acquired reaction systems, for example,
rational conduct, and thus leaves free to function such
complex forms of behavior as almost entirely to dispel the
appearance of a shock or conflict. The resumed activities in
such cases are of course only slightly different from those
interrupted.
V
In conclusion we might point out three cognate obstructive
tendencies, which persistently hinder psychological thinking
concerning instincts, and which prevent the scientific interpre-
tation of instinctive behavior, namely (i) metapsychological
speculation, (2) biological abstractionism, and (3) psycho-
logical simplification.
Metapsychological Speculation. — This motive has always
been a prominent factor in discussions of instincts, and
strangely enough is still responsible for the many inaccuracies
and trivialities of those studies. The unmistakable theologi-
cal implication of this attitude is manifested by the explana-
tion of instinctive behavior in terms of a mysterious original
force implanted in animals to carry out some primary end
of life, as for example, the preservation of the species.2 We
have already had occasion to refer to the hardly less objec-
tionable aspect of the metapsychological view, which makes
1 Not the diffused feelings.
1 Represented in psychology today by McDougall and Drever, who stand in the
von Hartmann-Bergson line of development, cf. Drever, op. cit., p. 89.
70 ]. R. KANTOR
instinctive behavior stand for everything that is considered
unknown or persistent in conscious behavior.1
Much the worst disservice of the metapsychological
attitude is precisely that it maintains unknowables which
prevent the adequate investigation of psychological pheno-
mena. To assume that instincts are ultimates of animal
nature and to seek to describe these putative elements,
precludes the conception of instincts as definite forms of
concrete responses.
Biological Abstractionism. — We may consider the histor-
ical rise of biological abstractionism as a protest against the
extreme vagaries of the speculative psychologists. The
biological influences in psychology transformed instincts into
simple psychological phenomena explicable in terms of the
nervous structure of the organism. Thus we find the state-
ment that the instinctive factors in behavior 'depend entirely
on how the nervous system has been built up through heredity
under the mode of racial preparation which we call evolution.'2
Instincts are consequently considered to be specific arrange-
ments of neural mechanisms; so that James, following Spencer,
spoke of them as conforming to the general reflex type.
An unacceptable issue of physiological abstructionism
is the tendency to overlook actual phenomena to such a
degree as to allow for no difference between such widely
varying behavior as we find in man and animals. Our study
of instinctive conduct has afforded sufficient intimation that
much of the unsatisfactory interpretation of such behavior
can be traced directly to the fact that it is the animal type of
reaction that is 'uncritically employed as an exclusive basis
of description. As a consequence this comparatively simple
behavior is resolved into hypothetical neural elements which
can in no way account for so conspicuous a variation as the
rapid development of human instincts into intelligent
conduct, and the practically stationary condition of the
1 So that the success of a politician or business man is attributed to the presence of
political or business instincts, the desire of a nation to govern itself to the functioning
of a self-governing instinct, and the building of cities to the presence of a gregarious
instinct.
2 Morgan, Brit. ]. of Psychol., 3., p. 220.
HUMAN INSTINCTS 7'
animal instincts. It is small wonder, then, that the upholders
of the physiological view fail to observe that the human
individual merely passes through the stage of animal conduct
just as he passes through the stage of simpler structural
developments, and that the mature person is equipped with
an entirely different series of reaction patterns than the animal
or child. And so we find that, contrary to our expectation,
the fact that the complete absence of instincts in the human
individual forces us to resort to animal behavior in order to
study them, does not influence the biological abstractionist
to reflect upon the differences in the two kinds of behavior,
but instead he is led to interpret human conduct in the
same abstract terms as the simpler kinds of behavior.
Incidentally, the failure to distinguish between the instinc-
tive conduct of man and the instincts of animals results in
the ascription of a degree of intelligence to animal conduct
which is really not found there.
Probably the most serious defect of biological abstrac-
tionism is that it obscures the extremely dynamic character
of human behavior. The principle of rigid neural functions
is entirely inapplicable to the spontaneous and developmental
aspects of our conduct, and favors the neglect of the stimu-
lating circumstances which greatly modify it.
Psychological Simplification. — The unfortunate phase of
the protest against biological abstractionism is the psycho-
logical simplification of human behavior, which reduces
instinctive conduct to the functioning of psychical dispositions
or impulses. As represented by McDougall and his followers,
this view stands as a justifiable criticism of physiological
abstractionism, but in its espousal of the subjective position
as over against the objective, that is to say, the position of
action and behavior,1 it is hindered from interpreting instinc-
tive conduct as it actually functions. Psychologists who
are influenced by this viewpoint are unable to depart from a
structural or content description of human behavior; they
are prevented from conceiving of the complex non-rational
conduct of man as the product of an intricate give and take
1 C/. Drever, op. cit., p. 16.
72 /. R. KAN TOR
process between persons and the social institutions which
constitute their milieu. When such complex behavior is
interpreted as an empirical consequent of numerous human
conditions, we can readily see that religious conduct1 cannot
be 'a very complex and diversified product of the cooperation
of several instincts,' that is to say, a 'compound' of simpler
emotions. To describe religious behavior as the manifes-
tation of a complication of simple mental elements is to forego
the scientific advantage of observing the ramified inter-
action of persons with their surrounding political, economic
and cultural institutions. The unwarranted simplification
of human behavior means that instead of analyzing the
social process in which are developed the deep-seated action
patterns, the latter are gratuitously assumed as permanent
elements of human character. The situation is not at all
improved by asserting that complex 'impulses' are developed
from simple 'impulses.'2 To deny that instinctive conduct
is socially formed reaction systems is to revert to the simplicity
of the rustic who protests against the daylight-saving law as
an interference with God's time.
A functional viewpoint of behavior, we submit, avoids
completely the three insidious tendencies which we have just
examined. Since the functional psychologist assumes the
data of psychology to be concrete adaptational responses to
surrounding things, he can whole heartedly reject all putative
powers and elements, and confine his labors to the analysis
of verifiable materials, such as human reaction systems are.
Abiding by such a policy, a psychologist is ipso facto barred
from an impatient out of hand solution of difficult problems.
Especially in the matter of instinctive conduct, a functional
viewpoint may lead to a scientific and significant, if tentative
interpretation of an important series of psychological adap-
tations.
1 Or religious emotion, cf. McDougall, 'Soc. Psychol.,' pp. 81-82, 89.
2 "If we accept the doctrine of the evolution of man from animal forms, we are
compelled to seek the origin of religious emotions and impulses in instincts that are
not specifically religious." — McDougall, 'Soc. Psychol.,' p. 89.
IMMOBILITY: AN INQUIRY INTO THE MECHAN-
ISM OF THE FEAR REACTION
BY J. P. M'GONIGAL
Judging by the uncritical spirit in which psychologists have
received recent physiological conceptions, it would seem
that the long discussion of emotion precipitated by the
James-Lange theory had exhausted the interest of the psychol-
ogists about the time that it secured the attention of the
physiologists. The current conception of fear is a case in
point. To Darwin, Ribot, Mantegazza, Mosso, Lange and
others fear, with its dual impulses of flight and immobility,
was the most baffling of the major emotions. Today one
finds it generally spoken of as a complex of the impulses of
flight and attack, the once perplexing impulse of immobility
receiving scant, if any, attention. This shift in opinion is
traceable to the demonstrations of the physiologists that
under the influence of either fear or rage the organism under-
goes profound physiological modifications which prepare the
body for sustained motor activity.
Now, it is not the business of physiologists to analyze
subjective states and psychologists have no right to quarrel
with such expressions as "the animal was either frightened or
enraged." If they were not prepared for such inexactness in
matters of importance to them, the^very tone of the physiologi-
cal literature should have warned them. The physiologists
have been, very properly, absorbed in tracing the interrela-
tions of certain obscure mechanisms which produce, among
man/ other intricate effects, emotional reactions, the particu-
lar kind of emotion being of secondary interest. Not so
the psychologist. To him, fear neither looks nor feels like
rage. Fear is a more primitive reaction, it appears earlier
in the life of the human infant, it is more uniform and definite
in its manifestations, less adaptive and more widely dis-
tributed throughout the animal kingdom. Such being the
73
74 /• P- M'GONIGAL
case, it would be strange, indeed, if the mechanism of fear
were not sharply differentiated from that of anger. In the
present paper the attempt is made to define the physiological
basis of fear, to show that it is distinct from that of anger and
to vindicate the earlier psychologists in their belief that the
fundamental impulse in the emotion is that of immobility.
For the purposes of comparative study we shall define fear
as the typical reaction to situations of imminent or potential
danger. Tracing this danger reaction downward from man
to the lower orders, it will be found that the impulse to hide,
crouch down, stand still, etc., becomes more and more pro-
nounced until, in some insects, we reach the trance-like
state of the death feint. The term 'death feint' is not only
inaccurate but is used with misleading looseness to cover
both the tetany of the lower orders and the remarkable state
of relaxation seen in birds and animals. The reaction is,
of course, not a feint at all. There is seldom an/ resemblance
between the postures of. a feigning animal and one really
dead, while the implication of a purpose to deceive is obviously
absurd. The reaction is, in fact, one of the least adaptive
and most difficult to account for teleologically in the whole
range of behavior. In the first place, the strength and dura-
tion of the response bears no relation to the stimulus. In
some insects the slightest stimulus will cause a profound
feint which cannot, thereafter, be produced by stimuli of
any strength. In other species the insect can be made to
feign as many as fifty times in succession, the duration of the
feints increasing for the first few times and then rapidly
decreasing, the stimuli being uniform throughout the ex-
periment. The distribution of the reaction among the species
is as erratic as its character. In closely allied forms, one will
possess the reaction and others will not. This last observa-
tion applies also to the relaxed form of the feint found in
birds and the higher animals. Some species of all orders
possess the characteristic, but it must, on the whole, be
considered an exceptional manifestation. The very fact of
this wide and erratic distribution would seem to indicate the
existance of some mechanism common to all animals, which
IMMOBILITY 75
under certain conditions assumes the form of the death
feint, but which in most cases, we must suppose, functions in
some less conspicuous way.
The possibility of such a hidden existence of the mechan-
ism is shown by an experiment of Holmes. In the lower
orders, contact stimuli very generally cause reversal of
tropisms. Holmes found this to be true of Ranatra when in
the water. On land, however, the same stimulus produced
the death feint, upon revival from which the insect showed
the usual reversal.1 In its customary environment there
would have been no reason to suppose that Ranatra possessed
the ability to feign. That a similar mechanism may, in this
way, lie hidden behind the simple hesitancy of the startled
individual of the higher orders is at least suggested by the
gradation of the response in birds. Chicks, ducklings and
plovers will, upon being startled, crouch down, pheasants stop
dead and hold rigidly the posture in which they happened to
be at the moment of fright; terns, peewits, land- and water-
rail are 'feigners' that lie limp and allow themselves to be
handled without showing signs of life. We shall assume,
then, for the time being, that an unitary principle runs through
all the various forms of danger response and proceed to take
advantage of the opportunity afforded by its extreme ex-
pression in the death feint to study its mechanism.
As it appears in the lower orders, the death feint is not a
state of mere quiescence. The muscular tension is often
tremendous. A feigning Ranatra may be held horizontally
by one of its slender legs, the whole weight of its body being
sustained by the tiny muscles. Here, clearly, is a sharp
redirection of nervous energy, which, since the posture may
be retained for long periods, must depend upon some pro-
found modification of the physiological state. Evidence of
the same nature is given by Holmes, Towle, Yerkes, Parker
and the Severins who have observed that upon revival the
feigning insects show reversal of tropisms, indicating a change
in sensitivity difficult to account for if the primary reaction
is a mere reflex. Pointing in the same physiological direc-
1 Holmes: 'Studies in Animal Behavior,' p. 104.
76 /. P. M'GONIGAL
tion are the results of the Severins on the duration of successive
feints in Belostoma. In a group of ten insects the average
duration of the first thirteen feints were as follows: (time in
minutes) 15.0, 32.4, 37.2, 24.0, 16.0, 15.9, 19.5, 12.8, 15.0,
9.1, 9.1, 5.4, 5.5. Specimen A of this group was made to
feign fifty times for a total of 382 minutes, of which 256 were
accounted for by the first ten feints. This distribution of
time is typical.1 Graphs of the feints of specially good
feigners (made by the present writer) all show sharp peaks
at the beginning and then a slow decline with a rhythmic
tendency, the whole picture suggesting a reaction conditioned
by some autocoid of which the supply is limited as in the case
of adrenin. A similar suggestion is supplied by the reaction
in birds. Lloyd Morgan mentions a water-rail that changed
in an instant from a state of complete inanimation to the
most violent and frenzied rushing about, while Holmes makes
the same observation in regard to terns. Holmes could
never get the birds to feign a second time. Remembering
that the passage from the feint to the frantic struggles and
blind flight occurs spontaneously and without any transi-
tional stage, it would appear that the second state resulted
from the exhaustion of some form of energy which served
to maintain the first.
These various conjectures (of which we have no definite
confirmation) that the death feint is the result of an unusual
functioning of a mechanism common to all animals, and that
it is due to the redirection of energy caused by some specific
agent of the nature of an autocoid, are given support by the
fact that under certain abnormal conditions a reaction
strikingly similar, even in details, can be observed in the
higher animals and man. This is the tetanic state produced
by strychnine poison and tetanus toxin. Under the in-
fluence of these poisons the body becomes tense and rigid,
the paroxisms, which are most easily induced by some form
of contact, lasting, in man, usually about fifteen seconds.
Two theories are advanced to explain the action of the poisons.
1 Severin, Henry H. P., and Severin, Harry C: 'An Experimental Study of the
Death-Feigning of Belostoma (-Zaitha Aucct.) flumineum Say and Nepa apiculata
Uhler,' Behavior Monographs, No. 3, p. 1 1.
IMMOBILITY 77
The first states that the toxin increases the conductivity of
the nerves to such an extent that coordinated activity be-
comes impossible, the chaos of currents resulting in tetanic
rigidity of the muscles. The second theory, that of Sherring-
ton, maintains that the reciprocal inhibitions normally main-
tained by the central nervous mechanisms are changed by
the poisons to excitations. By using carefully graded doses of
strychnine Sherrington was able to produce a state in which
reflex relaxation was diminished and not replaced by ex-
citation. "This phenomenon," he says, "shows well how
little competent is the view of lowered spinal resistance to
really explain the action of strychnine, for at this stage the
stimulated arc that normally acts on the extensor muscle by
inhibition is less able to affect it than before, so that on the
spinal resistance view the resistance at this stage is actually
heightened."1 Sherrington's theory, in other words, attri-
butes this pathological 'immobility' in the higher forms to
a redirection of nervous energy analogous to that which the
general nature of the reaction and the fact that it is accom-
panied by the reversal of tropisms, led us to assume in re-
gard to the lower orders. Now, while the use of inducted
poisons is a legitimate method of uncovering the mechanism
of immobility, we have, fortunately, a more direct avenue of
approach through the disorder of tetany, a disturbance
similar to tetanus but milder in form. Tetany is exhibited
in many diseases and is now generally believed to be of
parathyroid origin. This view is based upon the undis-
puted fact that deprivation of the parathyroids produces
tetany and that administration of parathyroid extract re-
lieves the disorder. Schafer, who represents the later work
on the endocrine organs, believes that this may be explained
by the cholonic or restraining action of the parathyroid
secretion which offsets in the normal animal the action of
an autocoid of opposite tendency produced by the thyroid
(thyrine).2 Here we have an autocoid which, by redirecting
the normal flow of nervous energy, produces a state of rigidity
1 Sherrington: 'Integrative Action of the Nervous System,' p. 108; vide p. 106-13,
292-7.
1 Schafer: 'The Endocrine Organs,' pp. 23 and 40.
7§ /. P. M'GONIGAL
and muscular tension that precisely duplicates the conditions
(known and surmised) of the typical death feint of the lower
orders.
In the healthy organism, of course, the existence of this
mechanism of immobility is concealed by the action of the
parathyroids. Is it merely concealed? It might be sug-
gested that the action we have just described is totally abol-
ished in the normal individual and plays no part in the danger
response of the higher orders. Yet, undoubtedly, animals
and men are occasionally 'paralyzed with fear' and a com-
parison of the readily recognizable 'expressions' of fear and
the known effects of thyrine will show that in the less violent
EXPRESSIONS OF FEAR1
(Expressions due to the skeletal musculature are omitted)
Quickened heart beat Thyrine
Arrested respiration Thyrine
Pallor Adrenin
Perspiration Thyrine
Immobility Thyrine
Tremors Thyrine
Dry mouth Thyrine
Arrest of menstruation Thyrine2
Arrest of lactation Thyrine2
Protruding eyeballs Thyrine3
Dilated pupils Thyrine3
'Authorities: For expressions of the emotions — Darwin: 'Expressions of the
Emotions in Man and Animals;' Ribot: 'The Psychology of the Emotions.' For
effects of the autocoids — Biedl: 'The Internal Secretory Organs;' Schafer: 'The En-
docrine Organs;' Gley: 'The Internal Secretions;' Cannon: 'Bodily Changes in Pain,
Hunger, Fear and Rage;' Crile: 'Man — An Adaptive Mechanism.'
2 While there is very little known in regard to these points there is an undoubted
connection between the thyroid and the phenomena of lactation and menstruation.
Ott and Scott state that iodothyrine inhibits lactation while Hertoghe claims the
exact opposite. (Bell: 'The Sex Complex,' p. 89.) During menstruation there seems
to be evidence of thyroid deficiency, since psychic disorders due to the disturbance are
relieved by the administration of thyroid extract. Biedl believes that the tetany
sometimes associated with menstruation, lactation and pregnancy is due to metabolic
disturbances set up by the functioning of the female organs. (Biedl: 'The Internal
Secretory Organs,' p. 44.) The arresting of menstruation in fear may possibly be
due to the diversion of the thyrine from the genital organs. This, however, is pure
conjecture. Ott and Scott also claim that adrenin inhibits lactation. I know of no
record of any effect of adrenin in delaying or promoting menstruation.
3 Also characteristic of adrenin but in a less degree. These are the most marked
and distressing features of exophthalmic goiter, a disease of hyperthyroidism.
IMMOBILITY 79
forms of the emotion there is clear evidence of thyroid ac-
tivity. In making this comparison we should be on the
lookout for manifestations of adrenal activity, for it is to
these glands that current theory ascribes the organic basis
of both fear and anger.
The chain of evidence seems complete. Starting with
the death feint as an exceptional primitive form of the fear
reaction, we were able to discover traces of a similar mechan-
ism through the various orders up to man. In man our mater-
ial pointed to the identity of this mechanism with the thyroid.
Then by a comparison of the known expressions of fear and
the known effects of thyrine we were able to show that the
thyroid plays an overwhelming part in the production of the
emotion, and, since the unrestrained functioning of the
thyroid produces a state of complete immobility, it follows
that the fundamental 'impulse' of the emotion must partake
of this character. To complete our argument for a specific
organic complex we shall proceed to a similar comparsion
of the expressions and autocoids observed in anger.
EXPRESSIONS OF ANGER
Flushed face Thyrine
Deep respiration Adrenin
Muscular contraction Adrenin
Salivation Adrenin
It is significant that the visceral expressions of anger are
meager as compared with those of fear. Ribot says, "Anger
and fear form an antithesis, but the former has, both physio-
logically and psychologically, a more complex character."1
Darwin makes a similar assertion and later observers have
been baffled by the paucity of the expressions of anger. The
effects of adrenin betray a similar complexity. On the
whole, it is accurate to say that adrenin speeds up the con-
version of energy and prepares the organism for violent
effort, increasing the contractibility of the muscles, immuniz-
ing against fatigue, promoting coagulation of the blood, etc.
It places the body, as Cannon tersely puts it, "upon a war
footing." It energizes and exhilarates, whereas thyrine
depresses and disorganizes.
1 Ribot: 'Psychology of the Emotions,' p. 220.
8o y. p. M'GONIGAL
It is true that in regard to the constriction and dilation of
the facial vessels these tables show a contradiction. It is
difficult to account for this, though Cannon has suggested
that the effect may be due to a purely mechanical result of
blood pressure.1 However, it would not affect our argument
if it were shown that there were even more crossing of effect
than this. It is not contended that the immobility reaction
wholly accounts for the affective state of fear in the higher
orders. An impulse to flee is distinctly recognized and in
flight the energizing qualities of adrenin would be of decided
service. Under the circumstance the organic complexes
of the two emotions are surprisingly clear cut.
1 Cannon: 'Bodily Changes in Pain, Hunger, Fear and Rage,' p. 227.
VOL. 27, No. 2 March, 1920
THE PSYCHOLOGICAL REVIEW
CHANGES IN SOME OF OUR CONCEPTIONS AND
PRACTICES OF PERSONNEL1
BY WALTER DILL SCOTT
President of The Scott Company, Philadelphia and Chicago
The term personnel implies a contrast relationship to
material. Just as material includes all the material equip-
ment owned or used by an institution or organization, so
personnel signifies all the individuals connected with such
an institution or organization. If we define applied psy-
chology as the science and art of the control of human
behavior, a large proportion of the field of applied psychology
is denoted by the term personnel administration.
Those of us who are engaged directly in personnel adminis-
tration in such organizations as educational institutions, the
army, the navy, industry and commerce, use the term per-
sonnel administration to include securing, testing, selecting,
hiring, placing, training, supervising, disciplining, stimulating,
directing, transferring, discharging and promoting each indi-
vidual concerned; and in developing the morale, increasing
the esprit de corps, and creating and sustaining a contented
and efficient group of individuals.
In this sense personnel administration is as old as human
institutions. The practice of personnel could not be post-
poned awaiting an adequate theoretical foundation. Every
bad personnel practice doubtless resulted from a false con-
ception and reflects itself in strengthening rather than in
weakening the error. This vicious circle is counterbalanced
by the fortunate circumstance that advanced personnel prac-
1 Address of the president, before the American Psychological Association, Cam-
bridge Meeting, December, 1919.
Si
82 WALTER DILL SCOTT
tice results from true conceptions and reflects itself in
strengthening and purifying the truth. Theory and practice
here as elsewhere are inseparable.
Because of the importance of personnel administration,
because of its complexity and difficulty, and because of the
very large amount of attention given to its various phases,
no profound and lasting changes in either the conceptions
or the practice of personnel are brought about suddenly or
by a single one of the numerous workers in this fceld. How-
ever, significant developments are being brought about from
the combined efforts of progressive workers.
During the nineteenth century many advances were made
in our conception of the material world and in our practice
of dealing with its various factors. The twentieth century
is characterized by an appreciation of the personnel problem,
by the possession of the behavioristic point of view in psy-
chology, and by the presence of numerous trained experts
devoting their energy to the development of the concepts
and practice of personnel. Constantly increasing numbers of
the members of the American Psychological Association are
making contributions to the field of personnel. It may be
contended that no other group of experts equals the members
of this Association in the number and in the importance of
the contributions made — certainly to the theory, and possibly
to the practice of personnel. It is too early in the century to
present the various contributions with adequate prospective,
and to estimate their importance. My aim at this time is to
call attention to certain typical changes that have come
about in our conceptions and practices of personnel and to
encourage the members of this association to renewed energy
in producing further changes.
The first of these changes to which attention is called is
the conception of the Equality of Men and the practices
associated with that conception. In the ancient Athenian
thought emphasis was put on the superiority of the Athenians
and the comparative inferiority of all other races. The
associated practice is found in the enslavement of foreign
races and in the custom of according equal but unrestricted
CONCEPTIONS AND PRACTICES OF PERSONNEL 83
honor to all the citizens. No citizen was assumed to be
superior to any other. Accordingly the casting of lots decided
which one from among all the citizens was to be chosen to
judge a dramatic or musical festival, to preside at court, to
be a legislator, or for one day at least to hold the highest
office in Athens, or to be an archon for a year and then a
permanent member of the Araeopagus.
This same insistence on the equality of men was dominant
in the writings of many of the great thinkers for two centuries.
It found expression even in the writings of such a Royalist
as Thomas Hobbes, as is indicated by the following quotation
from his " Leviathan ": "The question who is the better man
has no place in the condition of mere nature; where, as has
been shown before, all men are equal. . . .
"Nature hath made men so equal in the faculties of the
body and mind, as that, though there be found one man
sometimes manifestly stronger in body and quicker in mind
than another, yet when all is reckoned together the difference
between man and man is not so considerable as that one man
can therefore claim to himself any benefit to which another
may not pretend as well as he. For as to the strength of the
body, the weakest has strength enough to kill the strongest.
... As to the faculties of the mind, I find yet a greater
equality amongst men than that of strength. For Prudence
is but Experience; which equal time equally bestows on all
men, in those things they equally apply themselves unto.
That which may perhaps make such equality incredible, is
but a vain conceit of one's own wisdom, which almost all
men think they have in a greater degree than the Vulgar,
that is, than all men but themselves, and a few others, whom
by Fame, or for concurring with themselves, they approve.
. . . But that proveth rather that men are in that point
equal, than unequal. For there is not ordinarily a greater
sign of the equal distribution of anything, than that every
man is contented with his share."
It was no mere accident that Thomas Jefferson wrote into
our first official document the doctrine that "All men are
created equal." He was doubtless thinking primarily of men
84 WALTER DILL SCOTT
in their political relationship, but there seems no reason to
interpret his meaning as confined to political relationships.
The associated practice of universal citizenship must not be
thought of as an isolated phenomenon but is to be grouped
with all practices arising in connection with the theory of
the equality of talents, rights, and responsibilities.
The theory of the equality of men as related to education
is strikingly presented in the following quotation from Adam
Smith: "The difference of natural talents in different men is,
in reality, much less than we are aware of; and the very
different genius which appears to distinguish men of different
professions, when grown up to >maturity, is not upon many
occasions so much the cause, as the effect of the division of
labor. The difference between the most dissimilar char-
acters, between a philosopher and a common street porter,
for example, seems to arise not so much from nature, as
from habit, custom, and education. When they come into
the world, and for the first six or eight years of their existence,
they were, perhaps, very much alike, and neither their
parents or playfellows could perceive any remarkable dif-
ference. About that age, or soon after, they came to be
employed in very different occupations. The difference of
talents come then to be taken notice of, and widens by
degrees, till at last the vanity of the philosopher is willing to
acknowledge scarce any resemblance."
The conception of the efficacy of education in modifying
the inherent equality of men is responsible for much of the
good and much of the bad in American educational practices.
Among these may be cited, on the one side, our universal
compulsory education and, on the other side, the introduction
into the colored schools of the South of educational practices
from the classical preparatory schools of the North.
An unquestioned acceptance of the concept of the equality
of men results in inefficiency wherever applied. In the army
it results in seniority promotion. In labor unions it results
in an insistence upon an equality of wages for all the workers
of a craft. In popular thought on matters of social control
it leads to communism and syndicalism. In industry it
CONCEPTIONS AND PRACTICES OF PERSONNEL 85
results in the shaping of jobs to suit the capacity of the
average man, with the consequent elimination of adequate
stimulus to action for the superior individuals. The concept
of the equality of all normal adult men is a psychological
error that has perverted the thinking and weakened the action
of all peoples inspired with a true and worthy ideal of
democracy.
Possibly the greatest single achievement of the members
of the American Psychological Association is the establish-
ment of the psychology of individual differences. You have
discovered that normal adult men differ greatly in all human
capacities and attainments. You have demonstrated that
such differences are much greater than had ever been imag-
ined. You have found that individual differences are rela-
tively small in such matters as height, weight, physical
strength, and reaction-time, but that normal adults differ
enormously in the so-called higher mental qualities. Guided
by this new conception of individual differences you have
entered the schools and insisted that pupils be grouped by
their mental ages rather than by their chronological ages.
You have entered the army and urged that enlisted men be
assigned according to their fitness for army tasks rather than
by the location of their place of enlistment. You have
insisted that commissioned officers be promoted according to
merit rather than by seniority. You have cooperated with
progressive labor unions in developing a conception and prac-
tice adequate to provide protection for the weak and oppor-
tunity for the strong. You have entered industry and insisted
that applicants be accepted according to fixed standards;
that workers be promoted according to attainments and that
each employee be inspired by the particular stimulus most
effective for him. Your gospel of diversified talents is per-
meating our national thought and indicating, on the one
hand, the wisdom of a democracy utilizing experts in all
fields and, on the other hand, the hazard of all methods of
social control based on the assumed equality of normal adults.
A second change in our conception and practice of per-
sonnel administration is seen in the decreasing importance
86 WALTER DILL SCOTT
ascribed to Reason as a factor in determining human action.
For many centuries man was defined as the reasoning animal.
Aristotle's " Logic " was the standard textbook for all students
desiring to learn the best method of influencing and controlling
men. For persuasion the syllogism was believed to be the
most perfect tool. Arguments to be effective must be an-
alyzed and presented in a logical form. The hearer was
supposed to criticize appeals by the most rigorous of logical
standards.
The change in this point of view has come about so
gradually that we fail to appreciate its extent. At the hands
of certain authors the importance of reason is minimized
by an emphasis upon suggestion as descriptive of the process
of influencing men. Others contrast reason and instinct,
urging the important part instincts play not only in the
behavior of young children but also in the more important
acts of adults.
Still other scientists stress the significance of sentiments
and emotions, of impulse and habit, or of other forms of
human response not reducible to any standard type of reason-
ing. This change in our concept of the importance of reason-
ing is observable in the writings of modern psychologists,
and is reflected also in the practices of the more progressive
leaders in personnel administration.
The folly of treating workers merely as reasoning animals
but the wisdom of recognizing the importance of sentiment
is strikingly illustrated by the following instance:
The workers in the men's clothing industry, in Chicago,
were discontented last spring, because of various conditions
in the industry. To reduce this discontent some of the com-
panies increased wages 10 per cent. Company X. posted a
notice that on July I, each worker who remained loyal to the
firm until June 1 3, would receive " a special extra pay envelope."
This promise failed to change the attitude of the workers. A
few weeks after the posting of this notice the drive was on
for the sale of Liberty Bonds and the President of Company
X. purchased $34,000 worth of the bonds as a gift to his
employees. Each worker was given a coupon good for his
CONCEPTIONS AND PRACTICES OF PERSONNEL 87
share of the $34,000 worth of bonds. The workers manifested
no appreciation of this gift. On July I, each worker received
a special extra pay envelope containing a sum of money equal
to that which he had received on the second week in May—
a typical week. This generosity resulted in expression of
discontent among the rank and file of the workers. The
president of the company was much disappointed by the
failure of his program and called into conference on the subject
the local labor leader. I was asked to be present also. The
following is the substance of the conversation between the
president of Company X. and the labor leader.
President X: "I can't understand the lack of appreciation
of my men. I gave them $34,000 worth of Liberty bonds
and a special extra pay envelope of a full week's wages. The
union agreement has now put all the firms on an equal wage
basis. Although I did not increase wages 10 per cent, for
the period preceding the union agreement I have given my
men more than any other company by the extra pay envelope
and the Liberty bonds. I can't see what more they want."
The Labor Leader: "Yes, Mr. X, you have done all you
say and your people are not contented as the people are at the
other houses. They wanted the 10 per cent, and felt that
they had deserved it."
President X: "No, I did not give them the 10 per cent.,
but I did give the extra pay envelope and the Liberty bonds
which amounted to much more than the 10 per cent."
Labor Leader: "Yes, I have figured it up and you gave
them in extra pay and bonds somewhat over $10,000 more
than they would have received by the increase they asked.
But that is not what they wanted. They do not want the
gift of the extra pay envelope and of the bonds but they do
want the 10 per cent, even if it is less than the extra pay and
the bonds. I believe they would be willing to refund the
$34,000 worth of bonds if you would give them the $24,000
in what they regard as earned wages."
President X: "Very well, I will gladly make the exchange
for I shall thereby gain $10,000.
Labor Leader: "I think the discontent will be greatly
88 WALTER DILL SCOTT
reduced by the exchange. I will take it up with the people
at once."
The proposition was presented to the workers and was
accepted enthusiastically even though it entailed a recognized
monetary loss of $10,000. However, it restored their offended
pride and left them happy.
The President reasoned as follows:
Major premise, All wage earners prefer the greater to the
lesser amount of money.
Minor premise, The extra pay and the bonds is greater than
the 10 per cent, increase.
Conclusion, Therefore, the workers prefer the bonds and
the extra pay.
The experienced labor leader recognized that working
people are influenced as much by pride and by sentiment as
by the logic of the greater gain. He knew that strikes, and
demands for more pay and shorter hours are frequently but
a defense reaction against offended pride, and that the
rational interpretation placed on such action is usually as false
as the interpretation of President X upon the actions of
his men. The industrial leader who seeks to comprehend
and to lead his men to-day finds little assistance in Aristotle's
logic or in any conception that stresses the logical reasoning
ability of man. He does, however, receive great assistance
from the newer emphasis on the non-rational aspects of
human actions, as expounded by the members of this Asso-
ciation.
The third betterment in personnel administration to which
attention is called is that of the change in our conception and
practice of Education. To the popular mind education is
frequently assumed to be identical with learning, with the
acquisition of information more or less useful, with the com-
mitting to memory of the deeds and thoughts of forefathers
more or less worthy, and with the perpetuation of the classical
culture whether that be interpreted to mean Greek, Roman,
Hebrew, Turkish, Chinese, or Germanic.
Education is thought of as esoteric, — as a thing quite
apart from everyday life. It manifests itself as culture for
CONCEPTIONS AND PRACTICES OF PERSONNEL 89
culture's sake, as art for art's sake, or as pure science un-
contaminated with any possible practical results. A deep
gulf is assumed to separate learning and doing, theory and
practice, the school and the shop.
All these conceptions and distinctions cease to be signifi-
cant as soon as we take the modern behavioristic point of
view and define education as profiting by experience. Train-
ing is the result of the reactions made by the individual
being trained. The laws of Solon may be 'imparted' to a
phonographic record, a parrot may repeat certain phrases
of the Koran and an imbecile can commit to memory the
significant dates of Roman history. These instances are not
descriptive of education as we think of it today because they
are not instances in any real sense of profiting by experience.
In playing with fire the child secures training which in the
fullest sense is education. He learns to set up a series of
withdrawal reactions and to profit thereby in his increased
ability to establish this particular form of reaction when
facing similar situations.
The youth receives training in solving a mathematical
problem if in solving it he has acquired a new way of analyzing
a mathematical situation and can make use of it when such
situations arise again. He receives training in the reading
of Homer if in the study he acquires a new form of reaction
and is enabled to profit by this new possibility of action
whether it be in additional reading, in appreciating literary
style in another author, in improved diction on his own part,
or in comprehending human action.
The mechanic at the bench may be receiving an education
if he is profiting by his experience. In solving a mechanical
difficulty he may be acquiring a new form of thought that
may be repeated more readily when a similar situation is met
again. As a member of a group of workers he may acquire
a form of social response that may appear more readily and
be more effective with each repetition. As a member of a
shop committee his contact with the employer's representative
may change his entire point of view on industrial, social, and
political philosophy. In dealing with subordinates he may
90 WALTER DILL SCOTT
accidently or thoughtfully acquire a type of reaction that
fits him for more important executive duties. The reading
of a magazine article, the action of an associate, the chance
juxtaposition of two pieces of material may cause him to
think a new thought or perform a new act which may better
equip him to meet new and important duties. The signifi-
cant thing about these new reactions is that they are new and
can be repeated with benefit.
According to this conception there is no rigid demarcation
between school and society, between the pupil's desk and the
employee's bench or counter. In all these instances the
individual has experiences. Whether such experiences should
be classed as education or not depends less on the particular
geographical location than on the response resulting from
the experience.
When education is defined as profiting by experience, the
personnel director is faced with the double responsibility,
first, of providing educative experience, and second, of assist-
ing the individual to obtain the maximum of profit from the
experiences provided. The worker in repetitive manufactur-
ing processes may not be provided with adequate educative
experiences and the student in college may not be profiting
sufficiently by his experiences no matter how rich they be in
potentialities.
Our educators in institutions of learning are aware that
richness of content does not guarantee educative response on
the part of the student, so adequate responses are sought as
an essential part of all courses of instruction. In business
organizations need for varied experiences for each individual
is beginning to be recognized. Steps are being taken to
provide this variety by teaching the worker not merely one
job but by teaching him many jobs or by providing richness
of content in other ways.
In planning his training program the personnel director is
coming to see that his responsibility is not met by providing
formal classroom continuation school instruction for the
youths who are ambitious or for others compelled by law to
take such instruction. His is a greater responsibility and
CONCEPTIONS AND PRACTICES OF PERSONNEL 9l
ideally demands that he should supply each employee with
richness of experience and provide also that each worker
should profit continuously by his experiences as an individual
worker, as a member of the entire body of employees, as a
prospective junior executive, as a member of a family, and
as a citizen of the state.
A fourth change to which attention should be called is
that of the emphasis of the Biological Relationship existing
between the worker and his work. For a long time we have
used the evolutionary-biological point of view in interpreting
the relationship of man and the world in which he lives. We
have gradually ceased to think of man and his environment as
two contrasted and more or less independent entities. We no
longer think of him as the result of a special creation coming
into his world and subduing it. On the contrary, we think
of man as a result of an evolutionary process in which man
and the world were mutually involved. The general bio-
logical point of view is stressed in anthropology and sociology
but rarely in attempts to interpret the industrial worker.
This very necessary point in evaluating the concepts and
practices of personnel administration is clearly expressed in
extracts from an unpublished report of the Scott Company
Laboratory on 'A Point of View in Industrial Personnel.'
"In order to attain an insight at all adequate into the
field of industrial personnel, we must abandon any statement
which contrasts, or which even sets off against one another,
men and jobs. On the contrary, we must see clearly that
the industrial situation, the productive complex, is organically
not two things, but one; not men and over against them jobs,
but in reality worker s-in-their-work. In this active, shifting
unity, this intangible, ever-changing reality, the vital prob-
lems of industrial adjustment exist.
"There is a basic difference between this concept and the
older point of view toward personnel. The latter notion is
perhaps most crudely stated as 'Putting square pegs into
square holes.' For the sake of brevity and concreteness,
let us call this the 'square peg' concept. Clearly, there is a
family resemblance between the square peg concept and the
92 WALTER DILL SCOTT
statement that personnel work consists of man-analysis, job-
analysis and the bringing of man and job together. Man-
analysis is essentially discovering the shape of the peg;
job-analysis is essentially discovering the shape of the hole.
The phrase The Right Man in the Right Place is the slogan of a
personnel philosophy of the square peg variety.
"The inadequacy of the square peg concept arises from
its sharp discrimination between man and job — man on the
one side, job on the other side, with an act of bringing man
and job together. Here we have drawn for us two separate
entities, static, self-sufficient; our task seems to be to make
as good fits as possible. The coldness, the rigidity, the
sterility of this point of view is evident; it is not surprising
that it results so frequently in a mechanically impersonal,
jigsaw-puzzle attitude toward the problems of industrial
personnel. To fit the right man in the right place, to put
the square peg in the square hole, these are relatively fruitless
concepts for meeting the plastic, living problems of industrial
adjustment. . . .
"Our point of view differs from the square peg concept in
that it shows the worker-in-his-work as a unity, a developing,
living situation, as a productive complex, organically one.
We do not think of the hiring of the worker as the connecting
of a man with a job; it is the creation of a worker-in-his-work
situation, the birth of a new productive complex. We do
not think of the release of a worker as the separating of a
man from his job; it is the destruction of a worker-in-his-
work situation, the death of a productive complex. Square
peg philosophy regards as fundamental, structural diversity,
the man and the job, the worker and the work. Our point
of view regards as fundamental, functional unity, the man-
in-his-job, the worker-in-his-work. . . .
"The problems of industrial adjustment are conceived
to be not so much those of fitting together worker and work,
but rather those of securing the healthy development of an
organic unity, the worker-in-his-work. The worker-in-his
work is viewed as a living and changing situation, as the
functional element of the great industrial and social organism,
plastic and unstable. . . .
CONCEPTIONS AND PRACTICES OF PERSONNEL 93
"Personnel work involves the shaping of the growth of
this productive complex in forms of greatest economic effec-
tiveness and ultimate social value."
As a final point attention is called to the change that is
being brought about in our conceptions and practice of Voca-
tional Guidance. The caste system of India, and allied systems
in other lands, the European guilds of the middle-ages,
inheritance, opportunities for jobs available in the vicinity,
social approval and disapproval of certain occupations — these
are among the factors that have been dominant in vocational
guidance. Such general, non-commercial, and unadvertis-
able systems have been unable to retain a monopoly on
vocational guidance. There have sprung up special systems,
proporting to be infallible, in the guidance or the selection
of individuals. These systems have had great vogue in all
ages and today many of them are being sold to the so-called
hard-headed business men in every city in America. In the
list of such so-called "infallible systems" are included the
following: astrology, augury, chance as manifested in drawing
of straws, casting of lots or the flipping of a coin, chirography,
chiromancy, clairvoyance, character analysis, divination, for-
tune-telling, horoscopes, hypnotism, intuition, magic, med-
iums, mind reading, necromancy, omens, occultism, oracles,
palmistry, phrenology, physiognomy, premonitions, psycho-
logical tests, soothsaying, sorcery, sortilege, subconscious
hunches, stigmata, talismans, trade tests and telepathy.
When none of the systems here cited have been depended on
we commonly resort to the judgment of the maiden school
teacher, of the indulgent mother, of the ambitious father,
of the mercenary employment agent, of the hustling labor
scout, or of the listless recruiting officer. Vocational guidance
has been on an utterly unscientific basis and has been wholly
inadequate. However, no great improvement could be ex-
pected until a comprehensive job analysis of available voca-
tions had been made, until a technique of testing individuals
had been provided by experimental psychology and until the
point of view of the biological unity of the worker-in-his-
work had been recognized. During the past few years fairly
94 WALTER DILL SCOTT
adequate job analyses have been made of most positions in
many of our industrial and commercial organizations. Year
by year progress is being made by our laboratories in measur-
ing the talents, capacities, and skill of individuals. Great
advance has been made in our understanding of human
nature and in the creating of a practical biological point of
view of the worker-in-his-work. The advance in vocational
guidance may be adequately symbolized by the change from
the stage in which dependence was placed on the casting
of lots to the stage in which dependence is placed on the tono-
scope and similar instruments of precision; or from the stage
in which results were expressed in the ambiguous mutterings
of mediums to the stage in which results are expressed by
coefficients of correlation and by regression coefficients or by
other exact statistical formulations.
Of the changes in our conceptions and practices of per-
sonnel administration mention has been made of five that
are more or less typical of the many that might be cited.
The importance of these changes is very great, both for the
development of the science of psychology and for the welfare
of the human race. It has been estimated that during the
nineteenth century the power of the human race to produce
food, clothing and shelter was doubled by the application of
increased knowledge of the material elements of the universe.
All the significant advances in knowledge of the material
world were brought about by possibly a few thousand pro-
gressive minds devoted to that study.
It is quite probable that the productive power of the
human race is being doubled again during the present century.
The benefits of this advance will be divided between better
adjustments of the material world to the needs of man, and
the better adjustments of man to man. Such an increase in
the efficiency of the race will probably be due to the advance
in our knowledge of personnel rather than to further increase
in our knowledge of the material universe. If a few thousand
men in their study of the material world served their science
and the race so effectively, those of us who are engaged in the
study of personnel may get a glimpse of the responsibility
and the opportunity that is ours.
AN ANALYSIS OF EFFORT
BY JOHN J. B. MORGAN
University of Minnesota
The concept of will as outlined by the early faculty
psychologists has been discarded, and rightly so, by modern
genetic psychology. The will as a spiritual entity which
served to redirect, supply and control energy for the different
responses of an organism not only failed to explain many of
the phenomena which gave rise to the formulation of the
will concept, but in addition lent an air of mystery which in
itself inhibited the scientific investigation of the processes
it purported to explain.
The concept of dynamogenesis, which took the place of
the will concept, assumes that a sensorial stimulus not only
tends to find a direct response of some special sort but
affects more or less remotely other parts of the organism.
This diffusion of energy from an incoming impulse depends
upon the condition of the organism as determined by its
entire past history, upon other stimuli acting at the same
time, as well as upon the nature of the stimulus itself. Psy-
chologically the previous stimuli and experiences of the
organism may be considered as setting the organism in a
certain state of readiness or unreadiness. The task (Aufgabe)
—the set of the organism, or the directions received — facili-
tates one reaction and inhibits another. The directions to
add two numbers facilitate the adding of the numbers and
inhibit any tendency to multiply or subtract. If a man has
been placed in an attitude of anger this state of his organism
will cause him to become highly irritated by a relatively
mild stimulus. Such a simple reflex as the knee jerk is
affected by other stimuli acting at the same time, such as
martial music, the cry of an infant or other significant noises.1
1 Lombard, W. P., 'The Variations of the Normal Knee-jerk and their Relations
to the Activity of the Central Nervous System,' Amer. J. of Psych., 1887, i, 5-71.
95
96 JOHN J. S. MORGAN
Finally, a single stimulus if pleasant will cause a different
reaction from a single unpleasant stimulus. One may smack
his lips over a bit of candy but his whole organism may
respond in an effort to eject some disgusting bite.
By very carefully marshalling the facts of stimulus and
response, psychology has been able to explain most of human
behavior by a direct relation of each response to some stimulus
more or less remote. Some difficulties have arisen in connec-
tion with stimuli which elicited no immediate observable
response and with responses for which there were no imme-
diately preceding adequate stimuli. Many of the seeming
difficulties disappear when the human organism is regarded
as an integrating mechanism, and the working concept of
modern psychology is no doubt the concept that every
organism is a highly balanced system of forces, whose integrity
depends upon the maintenance of a certain balance. Each
stimulus is received into this system and is modified by the
condition of the balance at that time and its response can
only be interpreted in the light of its relation to all its previous
experiences as well as to all the other forces acting upon it
at that time.
Now, the question at the crux of this whole situation is,
what is this tendency to maintain a balance? The term used
in the will psychology to express this fact was effort. The
individual was presumed to make an effort to maintain his
'personality.' If a hostile stimulus was received he made an
effort to overcome it. The moral crisis was the place where
this manifestation appeared in its supreme state. It is the
sort of conflict which James describes in his classic fifth
type of decision.1 Why is it that "when a dreadful object
is presented, or when life as a whole turns up its dark abysses
to our view, the worthless ones among us lose their hold on
the situation altogether . . . and collapse into yielding masses
of plaintiveness and fear"; while the 'heroic mind' holds
itself erect, faces the situation, and 'makes himself one of
the masters and lords of life'? This question is still present
even when the situation is restated in behavioristic terms,
1 'Principles of Psychology/ 1899, II., 534-535.
AN ANALYSIS OF EFFORT 97
but it is simplified if one recognizes that such a question
resolves itself into a query as to why individuals differ in
their capacity to maintain their balance. We should expect
the trait to vary; what we wish to know is: What is this
trait? What is effort? What is meant by maintaining one's
balance?
This paper is an attempt to show that the concept of effort
is an elementary principle of all organic life which is as funda-
mental as the most firmly established reflex. To do this we
will first indicate where effort is best seen, we will show facts
which indicate its elementary nature and finally try to de-
scribe how the more complex situations in which effort is
revealed are developments from this elementary background.
The customary definition of effort is : ' the result or display
of consciously directed power.' This definition is itself a
product of will psychology and in order to get away from the
implications connected with the phrase 'consciously directed'
we will have to change our definition to read 'the result or
display of organic power.' This is not doing violence to the
term. The central idea is the display of power; organic
has been used in place of consciously directed. Power may
result in either of two situations, (i) When a stimulus is
received whose natural response would cause a loss of balance,
the stimulus does not give rise to its normal reaction.
Either 'power' is exerted by the other forces to divert the
stimulus to an unusual response, or it is merely inhibited.
(2) In the second case the organism may need to make a
certain response in order to maintain its balance. Any
opposition to this response will be opposed. The combined
forces of the organism will tend to break down this resistance.
It is effort in these two senses and not with its will implications
which we will attempt to analyze.
MANIFESTATIONS OF EFFORT
Effort may be seen when an organism is learning to react
to a novel situation. When an organism encounters a situa-
tion, the situation is in itself a stimulus to activity. If the
activity produces a favorable impression the organism con-
9$ JOHN J. B. MORGAN
tinues to respond in the same manner to that situation. If
the result is not satisfactory a new reaction is tried. This
may be considered an effort to meet this situation — it is the
first stage of effort — the connection between stimulus and
response is simple and direct; but the effort we mean to
analyze is a more complex affair. Suppose the organism is
placed in a new situation and makes the traditional random
movements in reacting to this situation. If all the movements
fail the organism could do one of two things; it could stop
its activity and go to sleep or it could increase the violence
of its responses to the point where it would be frantic in its
struggles. The increase in power displayed in the activity
of the organism when its responses fail to solve the situation
is what we mean by effort. The situation is the stimulus to
random movements; the situation plus the failure of the
random movements is the stimulus to effort or the display of
power. The failure is the added resistance in the situation
and the effort is the reaction of the organism to meet the
resistance.
As a specific example, suppose a dog tried to gain entrance
to this room. He could come to the door and make several
reactions to the fact that it is closed. He could whine,
scratch, push at the crack with his nose and bark. After
failing to gain access he could either go lie down or could
increase the violence of his efforts until the whole neighbor-
hood becomes disturbed by the intensity of his howling.
In this sense effort is a direct response to a stimulus and the
stimulus is the amount of resistance that the organism en-
counters.
Effort may be the response to a specific condition of the
organism, such as fatigue, sickness, etc., which renders it
difficult for it to make the normal response. An illustration
of this role of effort is seen in the case of an athletic contest.
The athlete starts the foot race as a response to the presence
of the crowd and his competitors. Gradually fatigue sets
in and the stimuli of his associates and the crowd weaken
until he is impelled to stop. He feels as though he would die
if he goes any further. He has no stimulus within or without
AN ANALYSIS OF EFFORT 99
to keep up this useless running but he keeps up impelled by
nothing but the stimulus of the opposition his organism is
receiving and the habits he has formed of persisting in spite
of opposition.
In many different psychological analyses the phenomenon
of effort is given a prominent place. In discussing the effects
of practice the statement is made by investigators that effort
must be directed toward improvement: that, if effort is
directed toward increase in accuracy, an increase in accuracy
will appear; if effort is directed toward an increase in speed,
an increase in speed will appear; or, if effort is directed
toward improvement in both speed and accuracy, both speed
and accuracy will be improved. Investigators of memory
make the statement that the learner must direct his efforts
toward memorizing in order to make the memorizing effective.
All discussions of attention point out two kinds of attention;
involuntary and voluntary. Voluntary attention requires
effort on the part of the individual. Successful reasoning is
conditioned upon the reasoner holding the problem to be
solved against irrelevant suggestions and distractions. This
process requires the direction of effort against stimuli which
are unfavorable for a proper solution.
Finally, effort is seen in the moral conflicts of the indi-
vidual. The instincts or the acquired complexes of the indi-
vidual tend to make him react in some particular way to
specific situations. Society says he cannot act in this way
without receiving punishment therefor. If the threatened
social punishment is able to exert force enough the tendency
to act may be inhibited; there may be a balancing of forces
with no apparent activity. James describes a case in which
social pressure is not sufficient to inhibit activity, where the
individual feels that he either will not be caught or that he
will be excused in case he is caught. He has nothing to lose
by refraining from an act except his moral integrity, and
even this loses its restraining potency. The individual feels
that by his own wilful act he is making the decision. This
is an illustration of the conflict of stimuli and must be carefully
distinguished from the opposition to response. Where there
100 JOHN J. B. MORGAN
is opposition to the normal response it was stated that the
situation plus failure in response is the stimulus to effort.
Here we have a balance between impulses neither of which
can gain the ascendancy. When a decision is made it is
usually with a violent impulse and not the slow response
which is typical of a nearly balanced physical movement.
The decision means the total inhibition of one or the other
stimuli. It is evidence that the balance, the withholding of
response, is itself the stimulus for the organism to do something.
This stimulus arising from the conflict throws itself finally
on one side or the other and a response occurs.
THE ELEMENTARY NATURE OF EFFORT
Suppose that all the organisms in a group reacted in
exactly the same way to the same stimulus, in other words
suppose that there were no such thing as biological variation.
In such a case it would be only through accident that organ-
isms could ever become selective in their reactions. Suppose,
however, that within a group two organisms responded just
a little differently to the same stimulus. If this stimulus
was favorable to the integrity of the organism the one which
gave the most ready or the strongest response would tend
to preserve its integrity better than the other. Suppose,
again, that the stimulus was light and that light was very
favorable for the organism in question. Those organisms
which responded favorably and most vigorously to a light
stimulus would have a better chance to survive than the ones
which were indifferent to light. Such selection would even-
tually give rise to a species which would tend always to
respond positively to light. (Positively phototropic). We
might say tha't the organism 'tries' to stay in the light, that
it * exerts' itself in an 'effort 'to keep where it is light, but
it can readily be seen that this is but a crude way of stating
that through biological variation and selection a positively
phototropic species has been evolved.
Again, suppose that from seeds which tended to throw out
sprouts indiscriminately a variation arose through which
some seeds tended to send sprouts toward moisture. If
AN ANALYSIS OF EFFORT ">'
moisture was favorable to the integrity of the life of the seed
the ones which showed the strongest tendency would survive
those in which the tendency was lacking. It can easily be
seen how through a long period of selection a species could
be evolved which would force its roots into the crevices of
rocks, even raising tons of weight to do so.
These principles of variation and selection form the
groundwork for the development of any biological trait no
matter how complex. The question is, can effort be ex-
plained as a biological development of this sort? If we can
show that effort is a fundamental trait of organic matter at
different levels, we believe that biological selection will ac-
count for its presence. Biological evolution has been so
elaborately expounded in the literature and is so widely
accepted that we scarcely need to defend it. All we need
to do is to show that effort is a definite response to a certain
condition and biological evolution will explain its existence.
i. Jennings1 has shown that the behavior of lower organ-
isms depends not only on the external stimulus but on what
he calls the physiological state of the organism. Physio-
logical states are of two kinds, those depending on the progress
of the metabolic processes of the organism, and those other-
wise determined. The latter are the ones which concern us.
The 'physiological state' is a dynamic condition and not a
static affair; it tends to produce movement. "This move-
ment often results in such a change of conditions as destroys
the physiological state under consideration. But in case it
does not, then the second tendency of the physiological state
shows itself. It tends to resolve itself into another and
different state. Condition I passes to condition 2, and this
again to condition 3. This tendency shows itself even when
the external conditions remain uniform." For example the
stentor2 is capable of greatly different reactions under the
same external stimulation. If the stentor is subjected to a
stimulus which would not be injurious unless applied for a
long time, if the stimulus and other external conditions remain
the same the organism will respond by a series of reactions
1 'Behavior of the Lower Organisms,' Columbia University Press, 1906, 282-299.
* Ibid., p. 176.
102 JOHN J. B. MORGAN
becoming more and more pronounced in character, until by
one of them it rids itself of the stimulation. The changes in
behavior may be summed up as follows:
"i. No reaction at first: the organism continues its
normal activities for a short time.
"2. Then a slight reaction by turning into a new position
— a seeming attempt to keep up the normal activities and yet
get rid of the stimulation.
"3. If this is unsuccessful, we have next a slight interrup-
tion of the normal activities, in a momentary reversal of the
ciliary current, tending to get rid of the source of stimulation.
"4. If the stimulus still persists, the animal breaks off its
normal activity completely by contracting strongly — devoting
itself entirely, as it were, to getting rid of the stimulation,
though retaining the possibility of resuming its normal ac-
tivity in the same place at any moment.
"5. Finally, if all these reactions remain ineffective, the
animal not only gives up completely its usual activities, but
puts in operation another set, having a much more radical
effect in separating the animal from the stimulating agent.
It abandons its tube, swims away, and forms another one in
a situation where the stimulus does not act upon it."
This situation can be clearly translated into the terms of
our thesis. The external stimulus was the cause of the
reacting movement; the same external stimulus plus the
fact that the movement was not effective in the removal of the
stimulus formed the stimulus for a different movement until
finally the organism made a very pronounced reaction, in-
volving its whole body, to a stimulus which at first caused no
noticeable response. While the variation in reaction at the
different stages is important for psychology we are primarily
concerned with the fact that there is an increasing intensity
in the responses until the stimulus is removed.
2. A simple illustration will show that the same thing
is present in animals of a higher order. A child upon being
rebuked for pulling the cat's tail replied that he 'was simply
holding and the cat was doing the pulling.' A stimulus upon
the cat's tail which at first will cause only a slight reaction
AN ANALYSIS OF EFFORT 103
will if that slight reaction fails to remove the stimulus give
place to the most violent reactions.
3. At birth a child shows the same reaction. Nothing
will arouse an infant to struggles and cries of seeming rage
as quickly as holding its arms close to its sides so that it
cannot move them. The child will resist movements of parts
of its organism and will increase the intensity of its resistance
if its first movements do not serve to remove the undesirable
stimulus.
4. A living muscle adapts itself in its contraction to the
resistance it meets. This fact has been known for some time
to physiologists. Luciani says;1 "According to the observa-
tions originally made by Fick, and afterwards confirmed by
others, when the weight applied to the muscle is not great,
and particularly when an elastic resistance is opposed to the
muscle, so that its tension increases constantly during con-
traction, the shortening is greater when the weight and the
initial resistance are increased. This paradoxical phenomenon
is a specific property of the substance of living muscle, and
shows that the sudden pull of the muscle and increase of
tension during shortening act as a stimulus on the contractile
substance, and increase the effect of the electrical stimulation."
5. It has been found in experiments with human subjects
that the force used in pulling a weight is determined by the
magnitude of the load.2 "After one has been pulling a weight
of 2,440 grams with what he supposes to be the maximum
force he is able to exert, when unexpectedly a weight of
7,770 grams is substituted for the lighter one, his force at
the very beginning of the pull is on the average 2.5 times as
great as the supposedly maximum force previously used."
The time taken for this adjustment ranges from 25 sigma to
91 sigma with an average of 54 sigma. This is much shorter
than the simple reaction time, which under the most favorable
circumstances can scarcely be reduced to 100 sigma. Since
1 'Human Physiology,' Trans, by F. A. Welby, Macmillan, Vol. Ill, pp. 13, 15,
and 46. See also 'The Speed and Accuracy of Motor Adjustments,' by the writer
of this paper, Jour, of Exper. Psychol., 1917, a, 225, 248.
1 Morgan, J. J. B., 'The Overcoming of Distraction and Other Resistances,'
Archives of Psychol., 1916, No. 35, Chap. VII.
104 JOHN J. B. MORGAN
this adjustment is so rapid it cannot be a conscious reaction.
It must be a reflex or a local muscular adjustment. In either
case the adjustment is certainly elementary.
6. In experiments on the distraction of attention it has
been found that individuals oppose distractions with increased
effort as well as by introducing other factors into their work
which will help them to overcome the distractions. These
adjustments are not made consciously but reflexly; the sub-
jects often asserting that they do not use the help that their
reactions indicate. For instance a customary reaction was
to articulate the material used in the process involved.
Breathing records were taken which indicated articulation
and the subjects were watched through a peep hole. Some
subjects who actually moved their lips in articulating denied
that they had made any such movements. In such a process
as overcoming distractions the exertion of effort is uncon-
scious.1
We have seen that the force exerted by an organism is a
direct response to resistance encountered. This adjustment
is seen in an organism as elementary as the stentor, in higher
animals, in infants at birth, in a nerve muscle preparation,
in muscle intact in the organism, and finally in complex
human activities such as resisting distraction. Using the
definition of effort given above we can say that effort is an
immediate response to the stimulus of failure. Failure is used
to mean the persistence of an unsatisfying situation in spite
of the normal reaction to that situation.
Let us now see whether complex types of mental effort
can be analyzed as derivatives of this elementary adjustment.
EFFORT IN PRACTICE
Let us take first the case of the effort involved in long
continued practice. In periods where no improvement is
made and a plateau appears in the learning curve, the ordinary
incentives fall off and subjects feel tempted to stop. In fact
many learners do stop at such points and those who persist
do so by what they call sheer will power. In this case no
1 Ibid., Chaps. I-VI.
AN ANALYSIS OF EFFORT 105
resistance is added but the customary drive to activity dis-
appears. When the man starts he sets before himself the
goal of becoming proficient in the line of work in which he is
practicing. His rapid progress at first enthuses him and
stimulates him to work for greater improvement. When
improvement comes to a standstill doubts assail him as to
the possibility of his ever becoming proficient. His goal seems
to fade away and he has no motive to continue except stub-
born persistence.
Now if an individual did not have the power of continuing
work in the face of obstacles all work would stop at such times.
The only way in which we could train ourselves to work would
be to learn which incentives are efficacious and how to keep
them before us. We all need incentives to start us on a task
and we need incentives to tide us over hard places; but as
we grow older we should need fewer and fewer incentives,
we should have developed this primitive trait of resisting
opposition to the point where we can surmount an obstacle
without any outside aid.
The ability to surmount obstacles is developed in some
such way as this. A young child innately opposes any
restraint upon its body or any forced movement of its mem-
bers. It wants to be free to move without restraint and
nothing will arouse a fighting, struggling reaction as quickly
as pressing its arms to its sides and holding them there.
As it gets older and wants something it will resist any inter-
ference in the way of its obtaining the thing it wants. The
wise parent or teacher will not fight back when the child
struggles for something, but if it should not have the desired
article will substitute something else for it, thus reinforcing
and redirecting rather than inhibiting the tendency to resist
interference. Such training strengthens the capacity of the
child to oppose force against resistance. Finally the child
comes to the point where it should learn some uninteresting
subject. Here the teacher presents proper incentives, at-
taches derived interests to the work and thus induces the
child to pursue the subject. The time will come when the
interests will fall off and the child will want to stop. The
106 JOHN J. B. MORGAN
teacher will add other incentives, give the child a helping
hand over the hard place and in so far has trained him to
oppose force against resistance. If help is not given at the
proper time the child may give up and in doing so has trained
himself to submit rather than resist.
It may be thought that the teacher in giving the extra
incentives is simply helping the child to form a habit which
will enable him to do this particular task. This is not the
case. The child's whole growth is a struggle between the
tendency to continue a thing once started and the opposition
of outer circumstances against continuance. In ordinary
cases the tendency to increase effort with an increase in
opposition wins the day; but when the opposition becomes
too great and the child is about to give in, this is the time the
teacher should give the help. Too much help will leave this
tendency in a dwarfed state, too little will train it to retreat
at the slightest opposition.
Viewed in this light punishment is not as beneficial a
factor in training as rewards. Punishment inhibits a certain
response and hence works against the tendency to carry
through a thing once started. Rewards are a reinforcement
to the tendency to carry through a thing regardless of opposi-
tion. Punishment if given at all should be coupled with
positive direction. Not only should an act in a specific
direction be inhibited but at the same time a substitute should
be provided and the offender encouraged in this alternate act.
As the child grows up under such training he is learning
to persist in certain types of activity and to give up in other
lines. He has learned that if he persists he can overcome
resistances and come off victorious. What is there in per-
\sistency in practice more than this? If there were not an
original tendency to persist upon which the teacher can
<build such training would be impossible; with such a tendency
and proper training nothing else is needed. No mysterious
exercise of a subtle faculty of effort is needed to explain such
persistent practice on the part of the learner any more than
it is necessary to say that when the child opposes restriction
of its limbs it is consciously saying, "I will not have this man
AN ANALYSIS OF EFFORT 107
holding my hands, I will exercise my will power and show
him I will not be thus dominated." An adult makes such
statements and thinks he is using a special gift of will power
when he does so. If he acts on his resolution he does so
because he has been schooled in using effort against resistance.
If he has not so schooled himself his asseverations will be
as idle as for a novice at typewriting to say, "I will use this
machine." He can use it if the resolution marks the begin-
ning of long practice; and a novice at mastering situations
can become master if his resolution marks the beginning of
practice in so doing. In both cases practice must begin with
simple problems. We are all familiar with the 'jack of all
trades,' the fellow who has never learned to persist in any
one trade long enough to become master of it. Such a man's
greatest lack lies not in the fact that he has not mastered
a trade but that he has never schooled himself in the meeting
of obstacles by persistence in some one trade. He has started
a dozen and continued until he struck a 'snag' and then
stopped.
EFFORT IN ATTENTION
Spontaneous attention is based upon the reflex response
of an individual to an adequate stimulus. If the stimulus
causes a response without involving conscious control it is
a simple reflex, if the individual is vividly conscious of the
response we say that he attends to it. If the first response
is followed by successive responses to other details of the
same object we have continued involuntary attention. An
object which is able to cause a series of responses to its
different details is said to be interesting to the individual.
Attention to such objects involves no effort, it is a native
response. It is necessary in order for one to get a proper
course of training to fit him for modern social life to attend
to a number of things which are not naturally interesting,
to objects to which he would not naturally attend. One
does this at first by attaching some outside interest to the
object. The drive from this outside interest makes us attend
to otherwise uninteresting objects and we still are exerting
no effort to do so. At times, however, the derived interests
io8 JOHN J. B. MORGAN
lose their potency, objects assail our senses which have more
interest for us than the objects to which we should attend,
and we do attend to them only by intense effort to do so.
Here the effort takes the form of reinforcing the desired
subject and inhibiting irrelevant subjects. Ability to exert
oneself in this way is only acquired after practice. One
cannot natively inhibit irrevelant stimuli which interest him.
A child's attention will waver from one subject to another,
and at the instant you think the child is listening to what
you are saying he will break out with the remark, "Daddy,
what is in your pocket?" A feeble-minded individual who
has never been trained to resist irrelevant stimuli shows the
same lack of control. Now if an individual did not have a
tendency to oppose effort against distractions he would never
learn to do so. All through life it would be absolutely neces-
sary to attend to the most interesting thing and the only way
to keep one's attention on a subject would be to make it
more interesting than the distractions. When one starts to
master a lesson the primitive tendency to persist in a line of
activity once started comes in to dispel irrelevant distractions.
When the distractions become too strong one can either
remove them or add extra derived interests to his task to
'boost' him over the hard place. Each time he succeeds in
dispelling a strong distraction it is easier for him to do the
same thing the next time. He learns tricks to help him do
so, such as straining his muscles, articulating words, etc.
He uses these aids to help him over the difficult places but in
so far as he is holding his attention by sheer force he is using
a modification of the inherent characteristic of all organisms
to oppose disturbances by increased effort.
OPPOSITION TO INNATE TENDENCIES OR LEARNED HABITS
A man may awaken to the fact that he has an irresistable
tendency to do some thing which is contrary to the moral
codes of the society in which he is living or the doing of which
would result in injury to himself. This tendency may be
due to inheritance or due to some habit he has formed, the end
result is the same. His whole organism impels him to do
AN ANALYSIS OF EFFORT 109
the act, the system of forces of which he is composed is
horribly unbalanced, all the weight is in favor of the act that
would harm him; still he knows that if he does it he will
receive the unanimous condemnation of his fellows or will
suffer in some other way. If the realization of the conse-
quences is vivid enough the tendency to do the act will be
counterbalanced and the man will refrain. This is the usual
type of activity in so-called moral decisions, when the indi-
vidual is combating an inherent desire to do something or is
fighting a habit he has learned. Victory comes without
display of effort by making the consequences so forceful as to
inhibit the performance of the act. In some cases however
when some specific temptation comes the individual loses
sight of the consequences, he believes he will not be caught,
his organism and the outer stimuli affecting it all urge him
to gratify his impulse. He feels that he will gain nothing
by resisting except the maintenance of the integrity of his
character. It is a battle royal with the odds greatly against
the man and only the exceptional man will win. When the
victory is won the victor feels that it was only through the
greatest effort of his will-power. Such a man is the heroic
type of which novelists write and poets sing. He is the ideal
to set before the young, he is the standard by which we
measure ourselves. How did he get the ability to display
such force — to resist when all help had deserted him? Such
a victory would never be won by a man who had never been
schooled for the battle. The teacher of morality takes great
pains to see that the child is given a proper chance to show
its resistance to temptation but is also careful to help it
across the hard place. No one would expect a child to fight
a difficult moral battle, one never blames the child if it
fails but blames the elders for their lack in properly guiding
and helping the child. The adult who is weak morally is the
one who has been so shielded that he never has had to fight
a battle, or the one who was never helped so that he never
won one. If there were no innate tendency to meet opposi-
tion with increased effort it could never be developed.
no JOHN J. B. MORGAN
SUMMARY
Every organism tends to maintain its integrity in the
face of situations which would destroy it. Whether one
explains the facts of the struggle which ensues from this
state of affairs by reference to the concept of will or by
reference to a tendency which developed through an evolu-
tionary process, the nature of the struggle which the organism
must put forth remains unexplained. Attempts to resolve
all response into terms of stimuli must account for a variation
in response with the same objective stimulus. This account
has been made by reference to the physiological condition
of the organism on the theory that every stimulus, besides
the direct discharge which it causes, diffuses energy through-
out the organism and leaves it in a different condition. This
difference in condition results in the second response to the
same objective stimulus being different from the first. What
is meant by physiological condition (or the correlative psycho-
logical term dynamogenesis) needs to be more carefully
defined and this paper is an attempt to define one phase of
this explanation.
The thesis of this paper is that effort in the sense of a
tendency to oppose any stimulus which would destroy the
integrity of the organism is a reflex response. Given an
inimical stimulus plus the failure of the normal response
of the organism to that stimulus to remove it, an increased
effort will result.
The paper proceeds on the assumption that the reason for
the presence of any trait can be referred to biological varia-
tion and selection, the thing which is required is to show that
the trait exists at various stages of organic life and that
complex manifestations Can be referred to the elementary
forms and explained as developments from them.
Following this line of reasoning it is shown that the in-
crease of effort to the stimulus of failure appears in an organ-
ism as elementary as the stentor, in higher animals, in infants
at birth, in a nerve muscle preparation, in muscles intact in
the organism, and finally in complex human activities such
as resisting distraction. From these facts the conclusion is
AN ANALYSIS OF EFFORT "«
drawn that effort is an immediate response to the stimulus
of failure.
The exhibitions of effort in such complex mental processes
as solving novel problems, persisting in practice, effort in
attention and in moral conflicts are then traced as develop-
ments of the reflex tendency present at birth to oppose re-
sistance to free activity.
Much work has been done in determining the laws of
retention and reproduction because it was felt inadequate to
explain present activity by a general reference to past im-
pressions. Just so, psychology cannot rest with the explana-
tion that the response depends upon the physiological state
of the organism. The physiological state (set of the organ-
ism, directions received, or any other term which may be
used) certainly depends upon certain laws which are discover-
able through proper research. This paper in an attempt to
formulate one of these laws.
A COMPARISON OF COMPLETE VERSUS ALTER-
NATE METHODS OF LEARNING TWO
HABITS.
BY J. F. DASHIELL
The University of North Carolina1
In the main, the study of habit formation has been limited
to the study of one or another single kind of habit. True,
in human psychology interest has widened from a technical
study of a single learning process to include the investigation
of two or more, and in animal psychology a few isolated
researches have been devoted to one aspect of the matter;
but the early researches upon the interference between habits
and the more extensive studies of the problems of transfer of
training seem not to have stimulated much activity along
still other lines of possible investigation in the matter of the
development of more than one habit. Putting it very gener-
ally, one might divide the past lines of interest in multiple
habit formation into (i) the investigation of the relations
between one process of learning and a simultaneous process
of learning, and (2) that of the relations between earlier and
later learnings (of different problems). The former interest
has been little shown in animal psychology, and in human
psychology only after the latter interest has been rather
exhaustively handled. The latter embraces well-known
questions as to the transfer of training.
The work here to be reported was a preliminary com-
parison between the practice methods referred to in (i) and
in (2) above. For a subject that is to learn two different
habits is it more economical to practice on them both at the
same time as nearly as may be, thus learning them together
or 'alongside' each other, or to practice on one only after
the other has been completely learned by itself? To approach
a reliable answer to this rather general question different
1 The experimental work was done in the Oberlin College Laboratory.
112
"3
types of subjects and of habits were used. This paper will
summarize these separate studies under the headings:
I. Maze Running by Rats;
II. Maze Running by Children;
III. Maze Running by Adults;
IV. Card Sorting by Adults;
V. Adding by Adults.
While this research was in progress Pyle's brief article
appeared1 in which he shows that experiments in card distri-
bution lead to the inference "that it is not economical to
form at the same time two mutually inhibitory sets of habits.
The better procedure is to form one, and then the other."
The present study may then be taken as a research similar
to Pyle's but extended over a wider range of habits.
I. MAZE RUNNING BY RATS
Nothing, so far as the writer is aware, has be'eV done with
animals on the question as to the relative economy of Com-
plete and Alternate methods in double habit formation.
The general method used by the writer embraced two
procedures: (i) that of training one group of white rats in a
single maze R, and when that was completely learned in a
second maze Z,; (2) that of training a second group of rats in
the same R and L mazes alternately, e.g., if maze R be used
on Monday, Wednesday, Friday, and Sunday of a given week,
maze L was used on Tuesday, Thursday, Saturday, and the
next Monday. The first method of training will be called
"Complete," the second, "Alternate." As implied, single
daily trials were set. The animals had been previously
accustomed to a feeding at 4:30 P.M., the hour adopted for
the experiments, and for a few days before the beginning of
the experiment were fed in the food box to be connected with
the maze. Uniformity of hunger conditions from day to
day were obtained by letting the animals feed as long as they
would (less than half an hour) after each day's run, then
removing them to the nest box where they had no food until
the next run, twenty four hours later.
1 W. H. Pyle, 'Transfer and Interference in Card-Distributing,' Journal of Edu-
cational Psychology, 1919, 10, 107-110.
H4 J- F. DASHIELL
The mazes were of cork-composition flooring, galvanized
iron partitions, and glass tops; and all runways and alleys
were 4 inches in height by 4 inches in width. The entrances
were closed behind the animals by raising a galvanized iron
piece hinged at the floor.1
The particular patterns of mazes used are shown in Fig. I.
They were exactly inverse to each other. It was believed
FIG. i.
that the identity of pattern when inverted would be unim-
portant to the animal subject, and it had the advantage of
making the mazes equivalent in difficulty. It will be ob-
served incidentally that the eight culs-de-sac furnished a
variety in position (a) with reference to runway, and (&) with
reference to each other. As preliminary to further studies
into the respective influence of different sorts of culs-de-sac
notes were taken regarding the functioning of these in this
experiment, but the data will not be presented in this place.
In order to provide as complete controls as possible a
single litter of white rats was used, one half of the litter
serving as the controls for the other half. This necessarily
made the number of subjects small, six being used, three in
each group. They were exactly eight weeks old at the be-
ginning of the experiments.
1 For details of construction of these and other mazes in the Oberlin laboratory
see Psychol. Bull., 1919, 16, 223-230.
METHODS OF LEARNING TWO IL 4 BITS
The data obtained for both groups in time consumed and
errors made for the various runs are given graphically in
Fig. 2. The 28th trial marks the first practice by the Com-
plete group (group learning by the Complete method) in
maze L.
n6 J. F. DASHIELL
Comparison of the numerical records ,of the two groups
would show little difference between the methods of Com-
plete and of Alternate learning in regard to total number of
trials required. The Complete group shows respectively,
37, 37, and 36 trials (total, no); the Alternate group, 30,
49 +, and 32 (total, in +).
As to type of curve it is to be said that not as much dif-
ference was found as would perhaps have been expected.
The curves for the subjects learning the R maze Completely
before being given the L maze conform in a general way to
the type found in numerous experiments with a single maze,
if we take them as far as the twentieth trial. The only
unusual feature is the highly increased time and errors at
the twelfth trial — attributable to rat C's relapse alone.
In the twenty-first to twenty-fourth trials the slight rise of
curves is due to the fact that rat /, having learned maze R,
was resting, and the average represents only the achievements
of the other two rats. The twenty-eighth trial was the first
one on the new maze L, and the curves are accordingly high.
That they are not decidedly higher is evidence of the transfer
generally found in the learning of one habit just after the
learning of a somewhat similar one. The average time taken
for the first trial in maze L is not as great as that for the
eighth trial in the previous maze jR; the average number of
errors in the first trial in L is only slightly greater than that
for the fifth in R. To what extent were transferred factors
operating and to what extent was there interference? The
former seem to have been more in evidence than the latter.
Apparently, the need is here indicated for a careful analysis
of the degrees and sorts of transfer of different features of
the whole maze learning procedure.
For the Alternating group the time curve shows again the
general form of the learning curve for the white rat in the
maze problem. The error curve is to a lesser degree of that
type. As is to be expected, these curves for learning Alter-
nately two mazes show poorer records than do those for rats
learning the R maze only, i.e., longer times and more errors.
This difference in amounts of scores made, however, is not
very great.
METHODS OF LEARNING TWO HABITS n7
A feature that appears more definitely is the difference in
regularity and irregularity in the curves for the two groups of
rats. It is especially evident in comparing the performances
in trials 2 to 20 inclusive. The curves for the Complete
group are fairly regular and smooth, those for the Alternating
group decidedly more irregular, and the time curve so through-
out its length. That this feature was not due to a difference
in difficulty in the two mazes in alternate use is shown by the
fact that the changes are not in the form of a regular alter-
nation between better and poorer scores. Consider such
successive changes as in trials 5-6-7-8, in 9-10-11, in 18-19-
20, or in 23-24-25-26. To obtain a more definite statement
of the differences in amount of regularity in the work of the
two groups the writer hit upon the following method:1
(a) finding the amount of improvement or loss between each
two successive trials for each animal in time and in errors;
(b) summing all changes of both sorts (both gains and losses)
for each animal; (c) averaging these totals for each group;
(d) determining the average number of trials taken in each
group; (e) writing the number obtained by (c) over that
obtained by (d), to express the group average of changes
between each two successive runs. (The same method can,
of course, be adapted for application to individuals.) Let
us call this the " Index of Irregularity." The relation between
the two groups may then be exhibited in the form of a fraction
or a ratio between the two indexes of irregularity. In this
way the Alternate group shows the greater irregularity in
reduction of time consumed by the ratio 24.51: 21.58. The
same group shows a very slightly greater irregularity in
elimination of errors, 3.48: 3.41.
The writer cannot refrain from remarking that had the
first trial of subject / of the Complete group been not so
disproportionately great in time and errors, the differences
1 Application of the method of average of deviations from the average of per-
formance for each group (the well known "A. D.") was tried and discarded at once —
the two groups gave the same A.D. for errors, 4.32. The large difference in the initial
trials of the two groups (time and errors for Alternate group being 233.0 and 14.3 to
365.6 and 23.6 for the Direct group) is enough to offset a very large amount of greater
irregularity by the Alternate group, if the ordinary A.D. be the measure used.
ii& /. F. DASH I ELL
between the two groups would have been materially increased.
Time on first trial for each subject of the Alternate group
was -£-92, !T-4O2, ^-205; for subjects of the Complete group,
H-ioSj 7-845, C-I44; errors on first trial for Alternate
group were £-8, T-if, ^-18; for the Complete group,
#-9, J-54, C-8.
We would seem to be warranted in concluding that for
white rats learning two different mazes with one run daily,
it is more economical to practice one Completely, then the
other, than to run them Alternately. This greater economy
is shown in the form of a greater regularity of performance.
As to a difference in total number of trials required by the
two methods the data are not conclusive.
II. MAZE RUNNING BY CHILDREN
Surely one important justification and raison d'etre of
the science of animal psychology is to be found in the possi-
bility that principles and laws empirically arrived at in this
field may be found applicable in some degree to human psy-
chology and education — fields more complex and difficult of
experimental as well as theoretical analysis. It may be
safely stated in this connection that such application from
one department of psychology to another will be the more
warrantable as the materials and methods involved in the
two cases are the more nearly identical. It was in accordance
with this principle that the author made a study of maze
learning by young children.
The eight children used were attending a kindergarten
and were in their fifth or sixth years (four or five years old).
In this experiment it was necessary because of lateness in
the school year to have the child make two runs daily about
twenty minutes apart.
The material used was a multiple unit set of screens
4 feet by 3 feet hooked end to end to form the partitions,1
and was set up indoors on the floor of a large room. To
serve as incentive, silk flags on a small upright stand were
shown the children and then 'hidden somewhere inside/
1 Described elsewhere, cf. supra, note p. 114.
METHODS OF LEARNING TWO HABITS
119
and the children were sent in to find them. This artificial
incentive quickly gave place to the * problem solving' interest
which persisted throughout in good strength.
On account of the limited floor space available it was
impracticable to have two entirely separate mazes erected;
and the same condition necessitated the employment of a
simpler pattern than the two used for the rats. Figure 3
K
FIG. 3.
shows the ground plan of the kindergarten maze. As one
problem the subjects learned to run this maze forward, in
direction indicated by arrow heads and passing culs-de-sac
in alphabetical order (called problem F); the other problem
consisted in learning the reverse path, passing the culs-de-sac
in the order F-E-D-C-B-A (called problem B). Similarly
to the plan employed with the rats, the method here used
was to have one group of children practice problem F only
until learned Complete (indicated by three successive error-
less runs), then the problem B only; and to have the other
group practice Alternately on the two problems, taking
problem F first on each day and B second.
The data obtained for both groups in errors and in time
consumed for the runs are plotted graphically in Fig. 4.
The seventeenth trial marks the first practice by the Complete
group upon the problem /?, the subjects who had learned
problem F earlier continuing to run on it until this trial.
120
/. F. DASHIELL
Comparison of the numerical records of the two groups
would show some slight advantage for the Complete method
in so far as the number of trials required is the criterion, the
children learning by this method requiring respectively
33, 32, 28, and 23 trials as against 34, 37, 30, and 35 + trials
for the other group.
METHODS OF LEARNING TWO HABITS I"
The type of curve is again of the well-known negative
acceleration kind for both groups.
The feature that again strikes the eye is the difference in
regularity or irregularity in the curves for the two groups of
subjects. Using the fraction described above in connection
with work with rats, it is found that the respective "indexes
of irregularity" compare as follows: the Alternating group
again shows greater irregularity over the Complete group
in the reduction of time consumed by the ratio, 6.62: 3.63;
the same group shows a greater irregularity in elimination of
errors, by the ratio, 7.00 : 3.44.
Another difference between the work of the two groups
is to be found in the amount of errors and time shown for
single trials. Consider especially the differences in errors in
the trials numbered 2 to 8, n to 15, 22 to 26, 28 to 32, and
the differences in time in trials numbered 2 to 4, 6 to 16,
1 8 to 20. The group learning by the Complete method
showed fewer average errors in 25 out of 32 trials and shorter
average time in 22 out of 32 trials.
III. MAZE RUNNING BY ADULTS
Another approach to the general problem of the research
was made by using adult human subjects with pencil mazes,
and with a somewhat different program.
Four students of college grade in a summer session were
enlisted as subjects. They will be denoted by initials,
H, K, B, and W.
The mazes were constructed as follows: the design of
each maze (see Fig. 5) was laid out on cardboard and then
with a narrow band saw cut out of * beaver board.' This was
given two coats of shellac to produce sufficient hardness and
smoothness of edges. The * beaver board' was nailed firmly
over a piece of cardboard placed upon a wooden base. The
runways were */4 inc^ wide and */4 inch deep, and had the
smooth cardboard for their floor. Two of the mazes (S and s)
were designed on a rectangular and straight line plan, to
make a suitable problem for the development of two habits,
and were respectively 8 X 91/* inches and 83/4 X 91/* inches
122
J. F. DASHIELL
in
outside dimensions. The other two (R and r) were
designed on lines similar to one another but quite distinguish-
able from the first two: they were of circular outside plan
with radius of 4:/s inches, and the runways were arcs of the
same curvature. A stylus was fashioned by rounding the
end of a 3/i6-inch round brass rod, 6 inches long. A rubber
band was wound around the stylus 2 inches from the lower
end. The stylus moved easily in the runways when held
vertical, but was of sufficient diameter to prevent too great
looseness and loss of contact with sides of runways.
The program of the experiments was arranged so that
each subject might serve as his own control, by having him
learn one pair of mazes by the Alternate method, the other
METHODS OF LEARNING 1'WO HABITS 1*3
pair by the Complete method. Instead of the trials being
distributed over a long series of days, as had been done with
children and especially with rats, the maximum of 40 trials
allowed to each problem was given all at one sitting (of 45
to 60 minutes). The order of use of the two pairs of mazes,
and the order of use of the two methods was varied, as seen
in Table I.
TABLE I
SHOWING PROGRAM OF SITTINGS
Subject Tuesday Thursday
H Mazes R and r, Complete method Mazes S and /, Alternate method
K Mazes R and r, Alternate method Mazes S and j, Complete method
B Mazes S and j, Complete method Mazes R and r, Alternate method
W Mazes S and /, Alternate method Mazes R and r, Complete method
The subjects were uninformed throughout as to the object
and the exact methods of the experiment and as to the order
in which the mazes were to be used. After every trial the
experimenter removed the maze and made a "business as of
changing mazes" before placing maze for the next trial.
This procedure was followed whether the same or the alternate
maze was to be used. Remarks or questions by the subjects
as to the apparent sameness or difference of mazes used in
successive trials were not answered. Thus in addition to
learning mazes in the usual sense the subject had to discover
inductively their number and the order in which they were
set. (As mentioned above, the entrance paths to all mazes
were different — to furnish a clue at the beginning of each
run that could sooner or later become definitely recognized
and used.) In operation this was found to have consider-
ably increased the difficulty of the learning and to have
introduced important ideational elements that may be par-
tially responsible for the minor differences in results between
this and preceding parts of this research.
For the experiment the subject was seated at a table upon
which the maze was placed, with a screen fixed above it to
shut off vision of it. The instructions given the subject
were as follows: "You are to take this brass stylus (shown)
in your fingers much as you would a pencil, but holding it
I24
/. F. DASHIELL
r i
+OT
S 10 IS VO Z5 30 35 10
5 /O IS 20 25 JO JS 40
METHODS OF LEARNING TWO HABITS 125
vertical. Do not place fingers below the rubber band.
As you hold it I will insert the point at the beginning of a
groove cut in a wooden floor, which you are not to see. You
are then to move the stylus within this groove until you reach
a point at which I say, 'Stop!' Time will be taken for the
trial but you are not to feel hurried. Three points are to be
remembered: Do not lift the point of the stylus from the
floor of the groove; always keep the stylus in strictly vertical
position, never let it slant; never let your fingers or any part
of the hand touch the floor." In actual operation the
subject's stylus upon reaching the exit went down off the
edge of the one inch base board.
The results of this experiment are best presented graphi-
cally in Fig. 6. Only curves for the errors are shown. The
time and error curves were in every case practically identical,
as the human subjects did not show the rapid pick-up of
speed shown by the rats, i.e., a relatively greater elimination
of surplus time than of errors. It will be seen that with the
subjects practicing by the Complete method first (H and B)
a great loss of time and a great number of errors accompanied
the first runs on each maze learned by this method. The
slightly better performance by these same subjects when
using the Alternate method later is, however, possible of
interpretation in terms of practice — practice on the first two
mazes being advantageous for practice on the later two,
being an evident case of transfer at least of the more general
elements in the learning situation. This is confirmed by
inspecting the curves for K and Wy who used the Alternate
method first. In both of these subjects the improvement in
learning two mazes Alternately is slow and, what is most
TABLE II
SHOWING INDEXES OF IRREGULARITY IN ELIMINATION OF ERRORS
Using C. Method First:
By C. Method By A. Method
2.25
2.00
4-25
11.37
5-35
l6.7i
H
1 ^'
B
4 16
Using A. Method First:
K
7/fi
. . 1 .01
W
. . 1.42
2.4S
126 y. f, DASH I ELL
striking, very irregular. Remembering its limitations, we
may again employ our index of irregularity to bring out the
last point. See Table II. It is also instructive to note how
rapidly the second habit by the Complete method was
learned, in all cases.
IV. CARD SORTING BY ADULTS
So far the question as to the relative efficiency of Complete
and of Alternate methods in practicing two habits has been
studied in connection with maze learning. Certain general
principles have been found to hold for maze habits for
different kinds of subjects with different kinds of maze
materials. The question arises, can the findings be demon-
strated for other sorts of habits? Since the running of
a labyrinth is a typical sensori-motor or perceptual-motor
habit, it occurred to the experimenter that it might be
enlightening to apply the same methods of approach to
some other style of perceptual-motor habit. The one chosen,
card sorting, possessed the advantage of being a familiar
one in psychological literature, having been already studied
somewhat with regard not only to questions concerning
single learning processes but also to some questions with
respect to the formation of multiple habits.1
In the present series2 the general program was similar
to that used in maze-running experiments with adults.
The material needed was two packs of cards of quite different
kinds. For one a 'flinch' deck was used which, with all
numbers of II and higher discarded, furnished ten cards
each for the numbers I to 10, one hundred in all. For the
other pack, one hundred blank cards of size, shape, and
general 'feel'* similar to the 'flinch' cards were obtained,
and upon them were printed autographs in script by means
1 For studying the interference of habits it has been used by Bergstrom, Brown,
Culler, Pyle. The study of Pyle, referred to above, which is almost identical in some
regards with this section of the present paper, appeared while these experiments were
in progress. The corroboration of his findings has its own value, and hence this
section is included in the paper.
2 For the data on the card sorting experiment the writer is indebted to Miss Helen
G. Smith.
METHODS OF LEARNING TWO HABITS 127
of rubber stamps obtained from men on the campus, one
autograph for each ten cards, ten autographs for the whole
pack. The general procedure was to have each subject use
one of the packs for learning to deal to two different distri-
butions or lay-outs on the table by one method (Complete
or Alternate), and to use the other pack for learning two
different distributions by the other method. See Table III.
TABLE III
SHOWING DISTRIBUTION PATTERNS USED
With "flinch "cards:
Pattern F Pattern /
3847 10 28 10 53
69152 71496
With autographed cards:
Pattern A Pattern a
JSCLM GBNLF
NGRFB CJSMR
The subjects used were four college students, Juniors and
Seniors, referred to as Br, Bu, P,'and H. They all cooperated
well throughout.
They worked at the experiment daily for ten days.
They were given a total of twenty-five deals in which to
learn each lay-out pattern; but these deals were arranged in
the two different orders, the Complete and the Alternate.
Table IV. shows in detail the program of the work as carried
out. The letters denote the lay-out patterns, as given above
in Table III., the figures indicate the number of deals on each
day to that lay-out.
The subjects were instructed to make each deal as rapid
as possible and were warned that time would be taken.
A misdeal had to be corrected before continuing.
The graphic method again recommends itself as the
clearest way of presenting the results. See Fig. 7, in which
the individual records are shown separately.
It is to be observed at once that the Alternate method of
practice in dealing to two different distributions is unques-
tionably inferior to the Complete method. This is strikingly
true in the matter of actual amounts of time taken, shown by
the lower vertical positions on the graphs. What is less
128
J. F. DASH1ELL
r i GURU 7
H 1 — I \-
IO 15 SO 25 JO Jf
7" >/.«/•
METHODS OF LEARNING THO HABITS
129
TABLE IV
SHOWING PROGRAM or CARD-SORTING EXPERIMENTS
Sub-
ject*
Fir»t Day
Srronil
Dmy
Third Day
Fourth
D.y
Fifth Day
Sixth Day
Seventh
Day
Eighth Dmy
Ninth Day
Tenth Day
Br.
Bu.
? ..
«..
7"
A"
A* and a5
altern.
F* and /'
altern.
A* and a»
altern.
/"• and /»
altern.
/">•
J»
/•»
^10
^* and a»
altern.
F» and /•
altern.
A • and a»
altern.
F* and /»
altern.
/-10
^10
f» then/'
^•then a»
A* and <?»
altern.
/** and /«
altern.
y^1 and a*
altern.
f» and /»
altern.
F» then/*
//'then a*
/»
a>«
^» and a»
altern.
F* and /*
altern.
A* and a»
altern.
/*and/'
altern.
/"
aio
f*
a'«
y/* and a*
altern.
/"» and /*
altern.
A1 and «•
altera.
F* and /•
altern.
/'•
a"
apparent in the curves but is shown in the numerical data
is also a greater irregularity in the rate of progress in learning
by the Alternate method. Table V. gives the warrant in
figures for both these conclusions.
TABLE V
SHOWING SCORES IN CARD SORTING
Subject
Method
Average Time Taken
(Seconds)
Index of Irregularity
Br
Complete
6l.O
c.7
Bu
Alternate
Complete
74-8
ci.6
9-5
«;.7
P
Alternate
Complete
68.6
67.4.
6-9
A. 2
H
Alternate
Complete
74.0
e87
5-3
A.f
Alternate
72-5
5-1
What has been demonstrated for maze learning is found
to hold true also for card sorting. It would seem to follow
that the principle would be found to apply to all forms of
true sensori-motor or perceptual-motor habit formation.
V. ADDING BY ADULTS
If certain principles are found to hold true of a particular
region in the whole field of learning, an important question
then arising is, will they hold true of all learning in general?
To make one further step in this logical direction, the author
sought an answer to the fundamental question of this research
in connection with learning on a 'higher' plane than the
perceptual-motor. As a more 'purely mental' process that
13° /. F. DASHIELL
is yet sufficiently of the habit type to be easily recorded and
measured objectively, numerical computation suggested itself.
Addition was taken as a particular form of computation the
improvement in which might serve as an interesting task to
students, especially in view of the practical value of adding
ability.
The subjects were ten summer session students of college
grade, members of the writer's class in the psychology of
training. The adding work was given during the first few
minutes of each class hour, the class meeting usually five days
in the week. The students were fully informed of the prob-
lem in hand; in fact, it was treated as a concrete side of the
material of the course. Their daily individual and group
records were exhibited at the following meeting, both numer-
ically and graphically. The writer was convinced, and was
so assured by the students, that their interest in the adding
was throughout high, generally being keen.
The material used consisted of mimeographed addition
blanks, with thirty examples on each sheet, each example
consisting of ten two-place numbers in vertical order. Six
sets of the material were provided, and the sets used in rota-
tion.
The experiment was conducted as follows: One copy of
the examples to be used for that day was given face down to
each student at his desk. At a signal, all turned their papers
over and immediately set to adding, continuing until the
stop signal was given aloud at the end of five minutes.
Both in order to eliminate some of the elements of psy-
cho-physical adjustment to the work, in a sense somewhat
extraneous to the problem at hand, and in order to have some
basis for dividing the class into two groups as equal as pos-
sible, a preliminary series of trials in simple addition was
given. In these trials the subjects added the numbers
vertically digit by digit and column by column. The
results were taken in terms of single columns or half columns
added. This was done for the five-minute period at thirteen
successive class hours. On the basis of the individual scores
made the experimenter divided the class into two groups of
METHODS OF LEARNING TWO HABITS 13*
five each, for the formal experimental series. Fig. 8 gives
the average scores of the individuals as so grouped.
For approaching the question as to the relative efficacy of
learning two kinds of addition by the Complete or by the
Alternate procedures, it was necessary to fix upon addition
methods that, while having some elements in common, would
10 // IZ /3
yet differ in important ways. Moreover, the addition methods
had to be novel. The plan hit upon was to use for the one
method or habit to be learned, the adding of two-place num-
bers horizontally, from left to right, adding the units first and
then the tens. The other habit decided upon was the adding
of the two-place numbers vertically, but by grouping them:
adding first the odd-place numbers together and noting down
the sum, then the even-place numbers likewise. Thus with the
I32 J- F. DA$HIELL
blank partly given in Table VI, the horizontal method would
involve adding successively the digits 4-5-2-9-8, etc., for
unit place in the sum, and 8-3-5-7, etc-> for the tens and
hundreds; the vertical method would require adding the
digits 6-9-7-5-1 and 3-5-7-4-9 for the total 308 of the odd
numbers, and then adding 8-4-2-3-4 and 6-3-6-2-8 for the
total 271 of the even numbers. As in the simple addition
in the preliminary series a complete sum obtained horizontally
counted as two columns; a sum of only odd-place or of only
even-place numbers obtained vertically counted as one column.
TABLE VI
SHOWING PART OF SAMPLE ADDITION BLANK
84 35 52 79 18 44 63 85 59 27
9* — — —
13
62 — Etc.
77
34
59
68
36
The program followed was for one group (C) to practice
the horizontal habit for seven trials in succession then to
practice the vertical method for the remaining seven trials;
and for the other group (A) to practice at horizontal adding
on the first, third, fifth, etc., days, and at vertical adding on
the second, fourth, sixth, etc., days.
For simplicity's sake the results of the formal series will
be given in terms of number of columns added, corrected for
accuracy by deducting a half column for each error.
It was early observed that these two habits were not of
equal difficulty (as had apparently been the case in all the
preceding experiments), the vertical habit being clearly the
harder. The gross average number of columns added ver-
tically by group A was found to be only 83.8 per cent of the
number added horizontally, the corrected average of the
number added vertically being 84.3 per cent of the corrected
average of the horizontal additions. For group C the cor-
responding uncorrected and corrected averages bore the
METHODS OF LEARNING TWO HABITS
133
ratios 89.6 per cent and 89.9 per cent, respectively. In
order, then, to be able to show the relations between the
two habits it was necessary to make them more commen-
surable by using the above ratios of corrected averages for
the two groups as bases for weighting. Thus, the group A
average for each trial by the vertical method was considered
as 84.3 per cent and raised to loo per cent; the group C
^--.
Z3- -
/5-
k
»
Id-
i'7
V
16
15
/f
----- Q T 0 U. p C
/ Z 3
76 9 10 Jl IZ 13 /+
456
T r ;*./*
average for each trial by the vertical method was similarly
increased from 89.9 per cent to 100 per cent.
The resulting data for both groups by both habits are
plotted graphically in Fig. 9. Here we see a very clear
superiority in the work done by the group adding by the
Complete method. The curve showing their horizontal
adding is consistently high. What is more striking, however,
is the extreme rapidity of improvement in vertical addition
134 J> F. DA SHI ELL
when once it was undertaken and practiced without inter-
ruption. One doubtful feature of the record is the inter-
pretation of the relatively good performance by the C group
at the very first trial. If our preliminary scores for the two
groups are reliable (see Fig. 8) such high initial score is not
due to greater initial general ability in adding.
In the matter of regularity in improvement the curves
speak more clearly than numerical figures. The index of
irregularity found for group A is 1.3, that for group C is 1.5.
Unquestionably, this difference does not speak for a lesser
irregularity in improvement for A so much as for the great
drop by the C group in starting the learning of the second
habit. In any case the difference is small, and the outstand-
ing feature of the results is the much more rapid progress
shown by the group learning one habit at a time.
SUMMARY
We have approached the question as to the relative
efficacy of learning two habits by practicing them alternately
(the Alternate method) or by getting one to some extent
fixed before practicing the other (the Complete method).
The approach was made with the use of mazes for rats,
children, and adults, then extended to include another
perceptual-motor habit, card sorting, and further still to
include a habit involving very little of the motor element,
addition.
The particular technique of the different experiments
was intentionally varied considerably: (a) in temporal dis-
tribution of trials, (&) in stage at which shift was made
from one to the other habit by the Complete method, (c)
in arrangement of controls — division of subjects into groups,
(d) in methods of scoring, (e) in incentives used, (/) in sub-
jects' previous familiarity with the habits to be learned,
(g) in subjects' knowledge of the number and order of the
habits to be learned, (k) in subjects' knowledge of the nature
of the problem investigated. Thus, the general results found
may be considered as independent of particular details of
technique and to be of general bearing.
METHODS OF LEARNING TWO HABITS 135
For results, it has been found that in all the forms of
double habit formation studied, learning by the Complete
method is more economical than learning by the Alternate
method. This is indicated in the different sets of experiments
in terms of the different criteria of efficiency respectively
applicable. They include: (a) the number of trials necessary
to fix a habit, (b) the degree of regularity in improvement,
(c) the average amounts of scores made on individual trials,
(d) the rate of acceleration of improvement.
THE TONAL MANIFOLD
BY R. M. OGDEN
Cornell University
Psychologists have often resorted to graphic representa-
tions, both bi- and tri-dimensional, in endeavoring to express
the interrelationships which obtain among the concurrent
aspects of elementary sensation. The most successful of
these attempts has been in the field of vision where the color
triangle, the color pyramid and the color cone are well
known as means of setting forth the dominant features of the
chromatic and the achromatic series of visual sensations.
Similar schemes have not been wanting for the other senses,
but no very useful representation has become current to
elucidate the characteristic features of auditory sensation.
Sometimes the tonal manifold has been represented as a
straight line to suggest the rise of pitch from low to high tones,
and sometimes a spiral has been used in order that the
recurrent similarities of the octave might be indicated by
points directly above one another in the spiral progression.
But the latter succeeds only in emphasizing the recurrent
likeness of octaves; for it fails to distinguish the similar
relations obtaining between other consonant intervals. In
the light of recently acquired knowledge concerning the
volume and the intensity of sound, the relations of pitch and
consonance can no longer be regarded as the dominant
features of the tonal manifold, and it is now possible to regard
the octave-quality of a tone as perceptual in its origin. Al-
though tonal "consonance is still a debatable question, it
seems fairly obvious, upon analysis, that the octave, or any
other consonance, subsists in the relation of a musical interval;
and that its elemental nature is therefore either an implicit
relation, as Stumpf understands it to be, or the product of an
implicit or an explicit act of perception in which not one tone
alone but two different tones are involved. In neither case
136
THE TONAL MANIFOLD »37
does a quality of consonance or an octave-character attach
itself directly to the simple element of tonal experience, as
does, by contrast, its pitch, its intensity, or its volume.
Unless we revise our whole conception of sensory analysis,
the characteristic aspects of an elemental sound are now
recognized to embrace at least four attributes; namely, pitch-
brightness, volume, intensity, and duration. Although each
of these is a variable, the particular degree of each which
attaches to any given sound is determined once and for all
by the psycho-physical conditions under which it exists. The
octave-quality, on the other hand, is a characteristic which
attaches equally to all tones within the musical range when
an appropriate reference has been made to another tone of the
series. Furthermore, a particular tone, though it can have
octaval relationship with but two other tones, one below and
one above it in the series, may establish numerous consonances
with various tones both above and below it in pitch.
In addition to the octave-quality of a tone there is also a
debatable quality which is supposed to enable one to assign
a tone to its appropriate place with reference to certain fixed
regions of the scale. This would explain the occasional
ability of a person to judge the 'absolute pitch' of a tone, and
it has also been thought to explain certain outstanding
regions of pitch with which the vocal qualities seem to be
associated. Whether a c-ness, d-ness, e-ness, etc., of tones
is immediately apprehended without involving a somewhat
complicated act of perception is a question we shall not here
attempt to decide, but in view of the fact that Oriental
peoples employ scales in which these harmonic designations
have no significance, it seems best to reserve judgment for the
present and test the possibility of some other explanation
before we proceed to base the elusive phenomena of absolute
pitch upon a universal quality inherent in tones which enables
us to assign them to their appropriate places in the musical
scale. All things considered, the case for the outstanding
octaval regions with which the chief vowel-sounds are asso-
ciated is a better one; but even here there are other possi-
bilities of interpreting the phenomena, like the one suggested
138 R. M. OGDEN
by Watt,1 in accordance with which the vocal apparatus is
assumed to find the utterance of sound at one region of pitch
easier than at another; hence the tendency to give prominence
to vocalization at this region and likewise at other regions,
above and below, which are in octaval relationship with it.
For our present purposes we shall ignore the conflicting
claims as to the qualitative aspects of tone other than the
features of pitch-brightness, volume, intensity and duration
which have already been mentioned, and shall confine our-
selves to the first three of these in the graphic representation
which accompanies this paper.
We have before us, then, a representation of the psycho-
logical aspects, or attributes, of a series of pure tones in a
progressive manifold extending throughout the range of audi-
bility. It should be noted that the dimensions of our graph
are measured in terms of psychological and not in terms of
physical components, and though we may refer to vibrational
frequency and vibrational amplitude, these are to be under-
stood as the conditions under which the psychological entities
of our manifold are controlled and produced; and not as
being themselves involved in the scheme.
The particular tones we have chosen to represent are the
successive octaves conditioned by vibrations ranging from
1 6 to 32,768 per second. Each tone is pictured with a certain
spread on the base-line to suggest its volume; thence rising
to a peak which indicates its pitch. The height of this
peak above the base-line measures inherent intensity, and
denotes the relative sensitivity of hearing at different degrees
of vibrational frequency. Duration, since it involves move-
ment, is not included in the scheme.
It will be observed that the total spread or volume of the
lowest audible tone comprises within its range the volumic
emplacement of all higher tones, the extreme upper point of
emplacement being identical for all tones. This accords with
Watt's theory2 and seems to be justified on the grounds he
has advanced. But Watt's further assumption, that 'when
1 Cf. Brit. J. Psychol., 1914, 7, pp. 12-13.
z 'The Psychology of Sound,' 1917.
THE TONAL MANIFOLD *39
octaves are played, the upper tone coincides with the upper
half of the lower tone . . .,' l is no longer acceptable in the
light of Rich's determination of the threshold for volume.2
If the increments of vibrational frequency necessary to pro-
duce liminal differences of volume depend upon a constant
fractional increase in the middle range of the scale, then
octaves within this range must differ by a constant number
of steps, which precludes the possibility of Watt's assumption
that the volume of the upper tone of an octave should always
be half the size of the lower tone.
Decrease in the spread or size of volume by a constant
amount is indicated for octaves throughout the musical range
of the tones here pictured (64 to 2,048 vibrations). Both
above and below this range, however, the fractional increase
is presumed to vary. The volumes of the lowest tones are
represented to be greater than the normal increase would
warrant, and their pitches are displaced to the right, indicating
the known tendency of low tones to appear higher in pitch
than they should. Similarly the highest tones are shown
decreasing more rapidly in volume than they do at the middle
range; while their peaks are displaced to the left — indicating
the tendency to regard tones of the four-accented octave and
above as flat. The total range of volume has been divided
somewhat arbitrarily into 228 steps, each step representing a
discernible interval as determined by a clearly defined dif-
ference of volume. According to the investigations of Rich,8
the threshold of volume is approximately .02 to .03 of the
vibrational frequency. Since the interval of the semitone is
about .06, we have taken .03, or the quarter-tone, as being
the threshold for interval-distance, and have plotted the
curves with abscissae measuring 228 just noticeable differences
of interval from the lowest to the highest tones.
In the middle range of the scale the volume for each octave
is so plotted that it diminishes at the constant rate of 24
quarter-tone intervals. In the highest and lowest ranges of
I0p. tit., p. 212.
* Amer. J. of Psychol., 1919, 30, pp. 122 ff.
1 L.c. and /. of Exper. Psychol., 1916, i, pp. 13 ff.
14° R- M. OGDEN
the scale, however, judgment of intervals is known to be less
certain. Tones below 40 vibrations appear to be a little
higher than would be warranted by the rate of vibration;
while in the upper range tones of 3,000 vibrations and above
seem flat, and at about 4,000 vibrations, according to von
Maltzew,1 accurate judgment of intervals breaks down com-
pletely. Equal decrease of volume as a basic feature in the
determination of octaves and other musical intervals extends,
therefore, only through tones that range from about 50 to
about 3,200 vibrations in the second. The volumes of the
lower tones are made relatively larger, and of the higher tones,
relatively smaller, than the normal variation of the middle
register would allow.
Turning now to the pitch of tones, this is indicated by the
central point or salient in the upward-rising mass of volume.
It will be noted at once that as volume decreases the pitch
becomes more salient, or pointed. This suggests the bright-
ness characteristic. As pitch rises it emerges more and more
clearly; it becomes more and more salient. The upward
trend from the base-line also suggests the variation of inherent
intensity attaching to tones of different pitch-levels. The
curve which circumscribes the salient peaks of these progres-
sive tones is the one determined by Max Wien in his study
of auditory sensitivity for tones of different pitch.2 According
to Wien's investigation sensitivity to tones increases rapidly
from the lowest audible tones to those of about 2,000 vibra-
tions when it begins to diminish, first slowly, and then more
rapidly.
This curve of sensitivity is of especial interest because of
the indication it gives as to differential sensitivity for pitch.
In the lower range, successive tones coincide to so large an
extent that the sensitivity for pitch is not much greater than
the sensitivity for volumic differences. The pitch-salients of
low tones are vague and indefinite, and an appreciable distance
or interval is therefore requisite before one pitch can emerge
distinctly from another. At a higher level this is not the
1 Zsch.f. Psychol., 1913, 64, pp. 161 ff.
2 Cf. Pfliiger's Archiv., 1903, 97, pp. I ff.
THE TONAL MANIFOLD 14*
case, for with salient tones one pitch distinguishes itself from
another even though there is no perceptible volumic difference
upon which a judgment of interval can rest. Thus the num-
ber of discriminable pitches within an octave increases steadily
until we reach tones in the region of 2,000 vibrations when it
begins to decrease. Decrease of sensitivity in the upper
range, together with inability to judge volume accurately,
both correlate with a falling off in ability to discriminate pitch;
although the absolute difference of vibrational frequencies
required for a given interval being progressively greater in
the higher range, may therefore occasion a larger number
of discriminable pitches per interval than is to be found for
the same interval at a lower level of the scale.
We have thus represented in our figure the progression of
tones throughout the range of audibility and have indicated
in a general way the course taken by volume, pitch, and
intensity. It remains to add a few words regarding brightness.
In a previous paper on the attributes of sound,1 I have sug-
gested that brightness be added to the list of auditory attrib-
utes. It is obvious enough that sounds are characterized
not only as big or little, loud or soft, long or short, high or
low, but also as piercing or dull. But it is still an open ques-
tion whether this latter characteristic, variously referred to
as brightness-dullness; shrillness-mellowness and sometimes as
vocality, merits consideration as an independent variable.
Rich, who made some study of brightness in his recent experi-
mental investigation upon the attributes of tone, comes to
the tentative conclusion that while the term is valid for
descriptive purposes, brightness is not independent of pitch,
since each appears to have the same threshold for differential
judgments. He therefore suggests that pitch-brightness would
be a more appropriate description for a single attribute
hitherto called pitch.2 This is very probably the case as
regards the tonal manifold, but as mentioned above brightness
has also been linked with vocality, and it is in this connection
that I still desire to recommend further investigation before
1 Cf. PSYCHOL. REV., 1918, 25, pp. 227 ff.
1 Cf. Artur. J. ofPsychol., 1919, 30, p. 157 f.
142 JR. M. OGDEN
we discard it as unnecessary to the complete description
of sound.
Thus far in this paper we have confined ourselves to but
one species of sound — namely, tone. But there are at least
two other perceptual objects of sound: the vocable and the
noise. The latter may perhaps be dismissed from our present
reckoning, for noise is commonly regarded as a complex
sound all of whose components can be described either as
tones or as vowels. Although such a conclusion is by no
means certain, an additional attribute which may furnish a
basis for the perception of noise is not at present under
consideration.
In the case of vocalic sounds, however, we have a type of
percept which offers some interesting features when it is
compared with the perception of tones. The investigations
of Kohler,1 Miller,2 and Schole3 have indicated the regions
of vibrational frequency which seem to characterize the chief
vowels. Yet Kohler's conclusion that the vowel is defined
by a certain pitch has not been confirmed by other investi-
gators. Jaensch,4 has made an apparently successful attempt
to produce vocalic sounds synthetically. His results indicate
that a compound of pendular-formed vibrations, varying but
slightly from, one another in frequency, possesses a vocalic
character which passes over into noise as the mean variation
of the vibrational components increases beyond a certain
point. Unfortunately Jaensch's method does not permit us
to determine just what vibrational frequencies combine to
give an optimal vocalic effect, and as the investigation has
not been repeated, his results have thus far received less
attention than they seem to merit.
It is highly desirable that this matter should be re-investi-
gated, for if it" be true that the characteristic feature of a
vocalic sound is obtained by compound regional vibrations,
rather than by a simple pendular-formed wave, the difference
between vocalic sounds and tones might be manifest when a
1 Cf. Zsch. f. Psychol., 1910, 58, pp. 59 ff.
2 Cf. 'The Science of Musical Sounds,' New York, 1916.
3 Cf. Arch.f. d. ges. Psychol, 1918, 38, pp. 38 ff.
* Cf. Zsch.f. Sinnesphysiol., 1913, 47, pp. 219 ff.
THE TONAL MANIFOLD 143
variation in brightness takes place without an alteration of
pitch. An investigation of Baley1 demonstrates that the
pitch of such a combination is indeed that of its mean tone,
and if the components of the sound-mass are very near one
another in frequency, it would certainly be impossible to
analyze them perceptually. Hence the sound must be of
elemental simplicity. The question is, does it possess the
vocal character which Jaensch has assigned to it? If so,
this change must be occasioned by an attribute other than
pitch, intensity, volume, or duration. Such a variant is
indicated by the term brightness, represented in our graphic
scheme by the pointedness or salience of pitch; for salience
must be reduced when a number of different tonal components
of slightly varying frequencies are combined. While the pitch
does not vary, brightness must necessarily decrease and pass
into dullness; thus the characteristic of the vocalic sound is
presumed to appear.
I have myself been unable to make but a casual observa-
tion with the aid of an Appunn tonometer upon the effect
of compounding tones of slightly varying pitch. If one adds
successive tones of slightly varying pitch on this instrument
the sound becomes rapidly confused and noisy, though the
tonal character of the whole does not readily disappear.
But there is also noticeable a nasal quality which seemed to
be most clearly manifest after the addition of a single tone.
The nasal quality impressed me as vocalic and since the
smallest increment on the tonometer was four vibrations, I
suspect that the threshold for vocalic sounds would lie within
this range of vibrational difference; though possibly the
number of components necessary to reduce the saliency of
pitch is more important than the range of the components.
Until further investigations have been made the whole
matter is merely one of conjecture. But Jaensch's experi-
ments are certainly suggestive, and I find nothing in the
results of Kohler, Schole, or Miller to discredit them. Hence
I conclude that even if brightness and pitch correlate directly
with increase of vibrational frequency in the upward trend
1 Zsch.f. Psychol., 1913, 67, pp. 261 ff.
R. M. OGDEN
of simple tones, brightness may nevertheless manifest a
variability independent of pitch if the conditions of sound-
production are such that a varied number of closely graded
vibrational components unite to produce a series of sounds;
for under these conditions a series is conceivable which would
pass gradually from tone through vowel to noise without a
noticeable change of pitch or intensity; and, at least so far
as concerns the appearance of the vowel sound, without an
alteration of volume.
As the sounds become noisy, we would be dealing with
10
101-
90 120
INTERVALS
FIG. i
210 228
components which no longer hold together in a single uniform
impression, after discriminable pitch, intensity and volumic
differences had one or all made themselves felt. In this
connection the variation of inherent intensity with difference
of pitch is also worth noting, for excepting the region of
2,000 vibrations and thereabouts, one cannot alter vibrational
frequencies very much without introducing a noticeable dif-
ference of intensity which would tend to destroy the uni-
formity of effect even if the objective conditions of vibrational
amplitude were to remain fairly constant.
THE TONAL MANIFOLD '45
ADDENDUM
Since the preceding article was submitted for publication
an opportunity has been afforded me to discuss the physical
aspect of Jaensch's theory with Professor Dayton C. Miller.
Although he had not read Jaensch's papers on the subject,
Professor Miller finds no evidence for the "mixed sine curves"
of Jaensch in his physical analysis of the vowels. But
although he accepts the Helmholtz hypothesis, rather than
Hermann's formant theory, his analyses show that the char-
acteristic of a vowel is a fixed region of resonance, within
which region must appear some partial of the fundamental
tone upon which the vowel is uttered. Lacking an appro-
priate partial the vowel is not given. Furthermore the
region of resonance in each case extends over a considerable
range of pitch and the physical aspect of the significant
partial is not that of a sharply defined tone, but rather that of
a distribution of energy in which the amplitude of the partial
in question is limited by the form in which the energy is
distributed over the fixed region of resonance. Thus, if the
partial falls near the middle of the resonating region of a
certain vowel, its utterance is more pronounced than if it falls
near the extremes of the region. But in any case the adjacent
parts of the tonal region are also involved, since the distribu-
tion of a fairly constant amount of energy over the entire
region is requisite to produce the vowel.
Miller's results are therefore not entirely at variance with
Hermann's theory of the formant or fixed tone, which Jaensch
also accepts; although Miller regards the production of the
formant as a phenomenon of sympathetic resonance, while
Hermann refers it to an independent "anaperiodic" blowing of
the mouth resonator.
As regards the point at issue, whether the vowel is differen-
tiated from tone by an attributive variation, Miller's analysis
seems to show that not one but a number of auditory recep-
tors are involved in its production. This is likewise the
view of Jaensch. Whether the "mixed sine curves" are
products of the sound wave or functions of the ear mechanism
146 R. M. OGDEN
is of secondary importance if the mode of stimulation is in
either case such that what we hear is a mixture of adjacent
pitches. If, as Miller agrees, a "sharp" tone in the character-
istic region of resonance would tend to destroy the vowel
effect, then a dull tone aroused by stimulating a region of
resonance rather than a single resonator is the phenomenal
basis of the vocalic sound. Our tentative assumption that
brightness may vary independently of pitch is therefore
feasible, and the way is open to determine the threshold of
this variation by devising experimental means for securing a
regularly graded transition from the simple, sharply defined
resonance of a single vibrational frequency to the regional
resonance involving a more extended series of receptors.
The experimental problem is perhaps none too simple, for
the variations may involve shifting amplitudes and differences
of phase rather than a direct attack upon several receptors
at once. In a brief paper presented before the Sixth German
Congress for Psychology,1 Jaensch suggests that the synthetic
effects which occasion vocalic sounds are a result of successive
waves which introduce complications of amplitude not sub-
ject to the Fourier analysis; yet the brevity of his report
leaves me uncertain that I clearly understand what he
means. However, despite all technical difficulties in the
control of the experimental conditions, empirical methods
would probably yield a means of demonstrating whether or
not brightness and dullness can be varied without a corre-
sponding alteration of pitch; and we must await such an
experiment before we can be assured that brightness and
pitch are separable aspects of sound.
1 Cf. Sericht. Leipzig: Earth, 1914, pp. 79 ff.
IS LACK OF INTELLIGENCE THE CHIEF CAUSE
OF DELINQUENCY?
BY CURT ROSENOW, PH.D.
Biometrist to the Juvenile Psychopathic Institute, Chicago, Illinois
A great deal of controversy has raged about the question
of the relative importance of intelligence (or the lack thereof)
as a causal factor of delinquency. In the present paper I
wish to discuss the bearing which actual modern statistical
findings have upon this issue, in order to clear up a certain
amount of confusion and misconception which seems to me
to exist. Of course, in the last analysis, the answer arrived
at through statistical or other methods comes back logically
to the definitions of intelligence and delinquency explicitly or
implicitly used by some particular author. But, to some
minds, the fact that the statistician deals as a rule with objec-
tive fact, and uses methods which, as methods, are logically
beyond criticism, means that the findings arrived at are also
beyond cavil. It seems worth while therefore to point out
that the very opposite is true. Statistical findings need not
only all of the scrutiny and criticism so lavishly given to
conclusions arrived at by less objective methods, but in addi-
tion there is need of the careful checking of the conclusions so
far as they are interpretations of the statistical constants.
As a case in point let us take the statistical findings of
Goring.1 Goring found a correlation of +0.66 between
criminality and mental deficiency, and this coefficient is
considerably higher than any he found to exist between
criminality and any of the other factors which he investigated.
He sums up as follows: "Our final conclusion is that English
Criminals are selected by a physical condition, and a mental
constitution which are independent of each other — that the
one significant physical association with criminality is a
generally defective physique; and that the one vital mental
1 'The English Convict.'
'47
148 CURT ROSEN OW
constitutional factor in the etiology of crime is defective
intelligence."1 ". . . our interim conclusion is that, rela-
tively to its origin in the constitution of the malefactor, and
especially in his mentally defective constitution? crime in this
country is only to a trifling extent (if to any) the product of
social inequality, or of adverse environment, or of other
manifestations of what may be comprehensively termed
'the force of circumstances.'"3 While the second conclusion
is stated tentatively, Goring makes it plain that he believes
that it should be accepted in the absence of contrary evidence
based on better data. His position seems to me to be well
stated by Miner4 as follows: "While there is always a possi-
bility of finding some other factor closely related to delin-
quency and independent of capacity, nevertheless we should
hardly urge this possibility at the present time as overweighing
the accumulation of negative evidence which has been as-
sembled in recent years, especially at the Galton Laboratory."
In other words, Goring guards himself sufficiently against
the possibility that further research may reverse his findings
through the discovery of new and better evidence. He does
not seem to see that the future must reverse his conclusions,
or, better, that his conclusions simply do not follow from his
statistics. He is correct when he states that the Intelligence-
Delinquency relation is the most important relation measured
so far statistically. He is absolutely wrong when he claims
that, in the absence of other data, we must accept his con-
clusions as the nearest approach to the truth attained thus
far. For if the coefficient of the correlation between intel-
ligence and delinquency is + 0.66, the correct conclusion to
be drawn is that it is exceedingly probable that factors other
than intelligence are of greater importance as determinants
of crime than intelligence.
This conclusion follows from the following considerations.
If we have a number of variables one of which is of special
interest, it is possible to express the relation of these variables
1 'The English Convict,' p. 263.
2 Italics mine.
8 Ibid., p. 287.
4 ' Deficiency and Delinquency,' p. 228.
INTELLIGENCE AND DELINQUENCY H9
in a single equation. Let D stand for delinquency, / for
intelligence, and N for a combination of all factors other than
intelligence. The equation will then take the form1
. P tfd.tnl ffd- in "j
- i\ rdi.n - \ + n\ rdn.i-
L ffi-dnj L ffn-dij
In order to understand the meaning of this equation a
certain amount of explanation will be necessary for the reader
not versed in this form of mathematics. The explanation
however has been confined to the minimum which is absolutely
necessary, and, if the reader will take the mathematical
assertions for granted, he will be able to follow the argument
of the rest of this paper. The expressions in brackets are
the coefficients of partial regression. They are, in effect,
the measures of the relative importance of intelligence and of
the other factors. For the sake of simplicity of statement
let us assume that the sum of the factors other than intelli-
gence can be summed up under the term economic status.
Then if the measure of the intelligence of any given individual
is known, and if that measure is multiplied by the coefficient
of partial regression of intelligence on delinquency, the result
will be the measure of the delinquency of that individual
which one would expect from his intelligence, independent of
his economic status; and if the measure of the economic
status of that individual is multiplied by the coefficient of
partial regression of economic status on delinquency, the
result will be the measure of the delinquency of that indi-
vidual which one would expect from his economic status,
independent of his intelligence. The sum of the two terms
will be the total expected delinquency of the individual.
The expression rd,-.n is the coefficient of partial correlation
of intelligence and delinquency. It measures the correlation
of delinquency to intelligence, independent of economic
status. If, for example, the entire population were arranged
into classes according to economic status, so that all indi-
viduals within any one group were equal in wealth to all
other members of their group, then, within any such group,
1 In this equation, D, /, and N are expressed as deviations from their respective
means in terms of their respective standard deviations.
150 CURT ROSEN OW
differences in delinquency cannot possibly be related to
differences in economic status, because, within any such group,
there are no differences in economic status. Therefore,
within any such group, the correlation of intelligence to
delinquency is independent of economic status. Similarly,
the coefficient of correlation rtn-i is the measure of the corre-
lation of economic status to delinquency, independent of
intelligence.
The reader should be careful not to confuse coefficients of
correlation with coefficients of regression. The one is a
measure of relation; the other is a measure of relative impor-
tance. Consider for example the effect of the moon, the sun,
and the planet Jupiter upon the height of the tides. If we
were able to measure with absolute accuracy the influence
of each of these heavenly bodies, the partial correlation
between the position of Jupiter and the height of the tides
would be well nigh perfect, but the relative importance of
the position of this planet would be negligible.
Having explained the meaning of the terms of our equa-
tion, we may return to the consideration of the equation itself.
In any given actual case, the right and left sides of this equa-
tion will not be exactly equal. For example, if we estimate
the degree of delinquency of any given individual from his
known intelligence and from the other known causes, the
estimated delinquency will be likely to differ from the actual
delinquency because we are dealing with imperfect measures
of intelligence, environmental influence, etc., and because we
are sure to have left some of the causal factors out of account.
If however we imagine ideally perfect conditions — if all the
causes of delinquency were known and accurately measured —
the two sides of the equation would be exactly equal. Fur-
ther, both rdn.'i (the coefficient of the relation of delinquency to
factors other than intelligence, intelligence being constant)
and Tdi-n (the coefficient of the relation of delinquency to
intelligence, all other factors being equal) would be equal to
unity, because, under the ideal conditions imagined, the degree
of intelligence would of course be a perfect measure of de-
linquency, so far as caused by lack of intelligence, and "all
INTELLIGENCE AND DELINQUENCY
'5'
other factors" would measure perfectly "all other resultant
delinquency."
The above considerations open up the possibility of sub-
jecting Goring's conclusions to quantitative criticism. For the
present I would direct the attention of the reader to the
following quantitative problem. Given a correlation of
+ 0.66 between lack of intelligence and delinquency, under
what circumstances will the correlation to delinquency of a
combination of all other causal factors be greater than
+ 0.66?
TABLE I
I
DN-I
2
DI-N
3
DI
4
IN
5
Dtf
I.OO
.00
0.66
— O.2O
0.604
.00
.00
0.66
—O.I4
0.652
.00
.00
0.66
-0.13
0.659
.00
.00
0.66
-O.I2
0.667
.OO
.00
0.66
O.OO
0.7SI
.00
.00
0.66
O.2O
0.868
.00
I.OO
0.66
0.40
0-953
.00
1. 00
0.66
O.6o
0.997
.00
0.66
0.66
0.66
I.OOO
Table I. supplies the answer to the problem. Column I.
shows the coefficients of partial correlation between delin-
quency and the factors other than intelligence, intelligence
being constant. The coefficients are i.oo in every case under
the assumed ideal conditions. Column 2 shows the coeffi-
cients of partial correlation of delinquency to intelligence,
all other factors being constant. These coefficients are i.oo
in every case except in the special case where rdn is i.oo,
when rji-n will take the value of + 0.66 for reasons which
will be easily understood. Column 3 shows the coefficients
of correlation of delinquency to lack of intelligence, which
are + 0.66 in every case by definition. In column 4 there
are a number of assumed coefficients showing a number of
possible values of the correlation of lack of intelligence to all
other factors. Column 5 shows the coefficients of correlation
between delinquency and all factors other than intelligence
which would necessarily result from the conditions assumed
in the other 4 columns. Thus, taking the first horizontal
152
CURT ROSEN OW
line, if the coefficient of correlation of lack of intelligence to
delinquency is + 0.66, if the coefficient of correlation of
intelligence to the other factors — 0.20, and if the analy-
sis is ideally complete as indicated by the coefficients of
partial correlation of columns I and 2, then the coefficient
of correlation of the factors other than intelligence to de-
linquency will necessarily have a value of + O.6O4.1 And
from the rest of the table we see that it will have a value equal
to or greater than + O.66 if the coefficient of the correlation
of lack of intelligence to the other factors is between — 0.13
and + 0.66.
Common sense would indicate (in a situation involving
only 3 variables), that the above also shows the limits within
which the factors other than intelligence are of greater impor-
tance than intelligence. But as a quantitative statement of
the coefficients of importance (regression) may be of some
interest, I have tabulated them in table II, which is merely
an extension of table I with the first three columns omitted.
The reader will remember that the coefficient of correlation
of intelligence to delinquency is 0.66 in every case.
TABLE II
Correlation
Correlation
6
Regression
Regression
IN
DN
DI-N
DN'I
— O.2O
0.604
0.8134
0.7667
—0.14
0.652
0.7658
0.7587
-0.13
. 0.659
0.7586
0.7577
— 0.12
0.667
0.7505
0.7567
O.OO
0-7SI
0.6603
0.7513
O.2O
0.868
0.5068
0.7667
0.40
0-953
0.3306
0.8197
O.6O
0.997
0.0968
0.9391
0.66
I.OOO
o.oooo
1 .0000
Columns 4 and 5 are identical with columns 4 and 5 of Table
I. Column 6 shows the coefficients of partial regression of in-
telligence on delinquency, all other factors being constant,
and column 7 exhibits the coefficients of partial regression of
1 The coefficients of column 5 were computed by solving the conventional equation
for a coefficient of partial correlation for ran- For this formula as well as the other
formuke utilized in this paper, see Yule's 'An Introduction to the Theory of Statistics,'
Chap. XII.
INTELLIGENCE AND DELINQUENCY »53
the factors other than intelligence on delinquencey, intelli-
gence being constant. Thus under the conditions of the
first horizontal line of Tables I. and II., the importance of
intelligence is to the importance of the other factors as
+ 0.8134 is to +0.7667. We see that "the other factors"
begin to be of more importance than intelligence when their
correlation with intelligence is — 0.12, and, as this last relation
becomes positive and increases in value, the other factors
become twice, and even ten times as important as intelligence.
Although I do not mean to enter into the psychological
aspects of the case in the present paper, I may remark in
passing that these figures become highly suggestive if one
believes that character is closely related to intellect without
being in any sense identical with it.
At any rate, it follows from these figures that lack of
intelligence is not the most important factor in the causation
of delinquency unless we have a right to assume that the coeffi-
cient of correlation of intellignece to the other factors is negative
and greater than — 0.12. But according to the best of our
present knowledge, the very opposite is very probably true.
We may divide the causal factors of delinquency other than
lack of intelligence into environmental factors and factors
peculiar to the individual. So far as the environment is
concerned, we know that poverty, absence of parental care
and supervision, and other evils in and out of the home pre-
dominate amongst the ignorant. Indeed it is often urged
by those who wish to minimize the importance of the environ-
ment that the poor are stupid not because they are poor,
but that they are poor because they are stupid. For our
present purpose, we are not concerned with this issue except
to note that ignorance and poverty and other environmental
factors go together and cannot therefore be negatively corre-
lated. Therefore, so far as we have gone, we have no reason
for believing that Goring has shown that lack of intelligence
is the most important factor in the causation of delinquency.
Our evidence, so far, points in the opposite direction. So far
as the factors peculiar to the individual are concerned nothing
very definite can be said in the present state of our knowledge.
154 CURT ROSENOW
Definitions in this field are either lacking or are so vague
and general that they tend to confuse rather than clarify
issues such as the one we are considering. Intelligence, for
example, is usually defined as the ability to adapt to the
environment, and, inasmuch as anything which makes
for detected delinquency necessarily makes for maladaptation,
other factors would be ruled out by definition. It is however
worthy of note that many authors hold to the existence of
factors other than intelligence and therefore depart from the
all comprehensive definition of intelligence, implicitly at any
rate. Obviously then it is impossible to say anything very
definite about the probable sign of the correlation existing
between intelligence and other causes of delinquency. In
the near future the writer hopes to show that intelligence
and moral character, when subjected to functional psycholog-
ical analysis, have many factors in common, so that there is
every reason for believing the relation between them to be
positive. At present it must suffice to point out that there
is no reason for believing it to be negative. And, as there is
good ground for believing that the relation of intelligence to
the environmental factors is positive, there would seem to be
good grounds for holding that lack of intelligence is not the
most important cause of delinquency, and no grounds at all
for holding that it is.
But even though lack of intelligence is not of greater
importance than all other factors taken in the aggregate, it
may be urged that a correlation of + 0.66 shows that it is
likely to be the largest single factor. The following hypo-
thetical example will show that that is not at all likely to be
the case. Suppose that in addition to lack of intelligence
there are six other known causes of delinquency, Xi, Xz, X3,
X^ X&, and X6, or seven causes altogether. Let the corre-
lation of each of these seven causes to delinquency be + 0.66,
and let all the possible intercorrelations of the seven causes
with each other be + 0.50. Let the relation of delinquency to
these seven causes be expressed in a single equation, as on
page 149. Then suppose that the degree of delinquency and
the intensity of the seven causes under consideration is
INTELLIGENCE AND DELINQUENCY '55
known with perfect accuracy for a very large number of
individuals and that, for each individual, these values are
entered into the equation. Then, if our seven causes have
furnished us with a complete analysis of the causes of delin-
quency, the right and left sides of the equation will be exactly
equal in each and every individual case. If the analysis is
incomplete, that is if there are causes not included under
our seven, there will be differences between the actual and
the estimated degrees of delinquency. Now if the coefficient
of the correlation existing between the actual and the esti-
mated values be determined, we have in that coefficient a
measure of the closeness of our approach to completeness of
analysis. We shall designate this special coefficient by the
symbol R. If the analysis is complete, R will be equal to
one. Otherwise it will be less than one and the amount by
which it falls short will indicate the importance of the causes
left out of account.
Now in our hypothetical example R is equal to -f- 0.873.
In other words, our analysis is incomplete even though we
have taken into account six other causes of delinquency as
highly correlated with delinquency as Goring found lack of
intelligence to be. An idea of the incompleteness of the
analysis may be gained from the fact that the average dif-
ference between the actual and the estimated delinquency
would be half of what this difference would be if R were zero.1
Indeed, if I had not shunned the labor of computation, I
could easily have taken twenty causes related to delinquency
as highly as our seven and still have reached an R unmistak-
1 R would be zero only if the coefficients of correlation to delinquency of each of
our seven "causes" were zero. In that case they would, of course, not be causes at all.
Nevertheless our regression equation would still yield the most reasonable estimate
possible in the circumstances of the degree of delinquency of any given individual.
Having no knowledge at all of the causes likely to produce delinquency, it would
be most reasonable to estimate the degree thereof as the average degree. For example,
if one were to estimate the height of John Doe, John Doe being any individual what-
ever about whom nothing at all is known except that he lives in Chicago, it would be
most reasonable to take his probable height to be the average height of the male
citizens of Chicago. Now if we were to estimate the degree of delinquency of each
and every individual by means of the regression equation based upon our seven as-
sumed causes, our average error would be half as great as if we were to estimate that
degree of delinquency to be the average degree of delinquency of the entire population.
156 CURT ROSENOW
ably below one. The truth of this last statement can be
strongly suggested by showing how much each additional
cause adds to the value of R in our hypothetical example.
I have done so in Table III.
TABLE III
Rd(i) = 0.660
Rd(iXl) = 0.762
Rd(ixlx2) = O.8O8
Rd(iXlx2xs) = 0.835
Rddx^x.x,) = 0.852
Rd(iXlxtxM) = 0.864
Rd(ixlXMxbx6) = 0.873
The first R is computed on the basis of one cause only; the
second on the basis of two; the third on the basis of three,
etc. It will be seen that R increases rapidly at first and
then more and more slowly. It seems that a great many
more causes would be needed to reach a value of one, if
indeed it can be reached at all, for R may be approaching a
limit of less than one. That is, if causes are intercorrelated
to the degree which we have assumed in our hypothetical
example, it may be impossible to reach completeness of
analysis even with an infinity of causal factors.1
It would be possible to keep ringing the changes on the
various hypothetical combinations of causal factors which
might be formed. I trust what has been done will be enough
to validate our claims. Summing up, they are:
1. The claim that Goring's statistics prove lack of intel-
ligence to be the chief cause of delinquency, at any rate until
better data are at hand, is due to a mistaken interpretation
of his statistical results.
2. His results do show that very probably lack of intel-
ligence is of less importance than all other factors combined.
3. They show, also, that lack of intelligence is probably of
less importance than one or more other factors, taken singly.
In conclusion I wish to say that, in the present paper,
terms such as intelligence, delinquency, etc., are used un-
1 Cf. Pearson, Biometrika, Vol. 10, p. 181.
INTELLIGENCE AND DELINQUENCY 157
critically and naively. Whatever definitions of these terms
may be implicit in Goring's data are accepted by me without
question. For the present paper addresses itself only in-
directly to the larger problem of the real significance of the
causal factors of delinquency, and is concerned mainly with
the correct interpretation of statistics. In the near future
I hope to be able to publish some views on the more important
psychological analysis of the factors which are summed up
under such headings as intelligence and character, so far as
they are causes of delinquency.
VOL. 27, No. 3 May, 1920
THE PSYCHOLOGICAL REVIEW
MANIFOLD SUB-THEORIES OF "THE TWO
FACTORS"1
BY C. SPEARMAN
PRESENT SITUATION
Few competent judges would dispute that among the
most unexpected events in the psychology of the last dozen
years has been the sudden spring of * general intelligence'
from an almost universal incredulity to a no less universal
investment with the highest importance. In this movement
I venture to claim having been among the earliest partici-
pators. Already in 1904, I went so far as to put forward an
explanatory theory, namely, that of 'Two Factors.' 2
The purport of this theory is that the cognitive perform-
ances of any person depend upon : (a) a general factor entering
more or less into them all; and (b) a specific factor not enter-
ing appreciably into any two, so long as these have a certain
quite moderate degree of unlikeness to one another.
The proof offered was that, on eliminating from any set
of mental tests any that happened to be very obviously like
others in the same set, then the intercorrelations may still
remain large, but will now admit of being tabulated in a
'hierarchy.' This mode of describing the proof was soon
afterwards reduced to the exact mathematical equation:
where a, b, p, and q indicate any of the tests and r is the
1 In this paper I owe a double debt to Professor Nunn: firstly, for stimulating me
to wrife; and secondly, for pointing out to me several serious obscurities when I had
done so.
1 'General Intelligence,' objectively determined and measured, Amer. J. Psychol.
15, 1904-
'59
160 C. SPEARMAN
(product moment) correlation between them.1 This equation
was said, both to hold good of experimental results, and also
to prove the said theory. A corollary of the equation is
that in any table of correlations as ordinarily set out, every
column will have a perfect correlation with every other one.
Thus the three criteria, the 'hierarchy,' equation (A), and
a perfect intercolumnar correlation all indicate the same
theory.
Now, this theory has had a curious fate. It was soon
backed up by a seemingly overwhelming mass of evidence.2
Above all, even the data expressly brought forward to refute
it were without any exception obliged, Balaam-like, to give
it their complete support. In particular, a very valuable,
but certainly not friendly, investigation of Simpson was shown
on more careful reexamination to fulfill the hierarchy just
as exactly as all the results obtained before.3 And as for
any attempt at disproving this fulfillment by either the data
of Simpson or any others that had been adduced, I have
laboriously searched the vast literature of mental tests for
such in vain. In spite of all this, mirabile dictu, hardly any
writer (outside of those working in more or less intimate
connection with myself) has so far uttered a sign of being con-
vinced!
Of the reasons that might be alleged for this obduracy,
some are obviously improbable, such as that so many psy-
chologists should be biased by their preconceived doctrines,
or that they would decline to make themselves acquainted
with the newer mathematical aids to research.
A more plausible explanation would be some widespread
belief that, although the truth of the theory would necessitate ,
the fulfillment of equation (A), yet this fulfillment might
not, inversely, necessitate the truth of the theory. But even
1 This equation was at first expressed in a slightly different form, see Zeit. /.
Psychol., 1906, pp. 84-5. Then the form (A) was communicated by the present writer
\ to Mr. Cyril Burt, who employed it in his well-known admirable paper, Brit. J. Psychol.,
3, 1909-
2 See in particular, 'General Ability, its Existence and Nature,' Brit. J. Psychol.,
5, 1912, and 'The Heredity of Abilities,' The Eugenics Review, p. 8, 1914.
3 See 'The Theory of Two Factors,' PSYCHOL. REV., 21, 1914.
MANIFOLD SUB-THEORIES OF THE TWO FACTORS 161
this suggestion now fails, since a recent luminous paper by
Garnett has shown that also this inverse relation holds good.1
As his demonstration is rather long and difficult (but pro-
portionately instructive), the following very simple one may
be of service. It has been used by myself for many years,
but never published.
Let rxy denote the correlation between two variable tests,
x and y. It can be written as /(/>), where p is any one of the
elements entering into x and contributing to its correlationsT i
the remaining elements being regarded as parameters. Simi- ;
| larly, rxt may be written as <j>(q), wljere q is any of the elements '
entering into and contributing to the correlations ofyJ But,
by equation (A),
f(p)/<t>(q) = rxv/rxz = constant.
Hence, p and q cannot possibly be independent of each other;
there can be only one independent; and this necessary single-
ness is at once extensible to the whole set of functions in
question.
"A HIERARCHY WITHOUT A GENERAL FACTOR"
Faced, then, on all sides by this elusively silent 'passive
resistance' to the theory of Two Factors, it was a great
pleasure to myself, anxious to get the matter settled, when
at last one (and only one) writer did step into the open field
and challenge the above-mentioned evidence on definite
grounds. This was G. H. Thomson, the tenor of whose
argument was to admit that the theory is proved whenever
equation (A) is satisfied exactly, but to deny that it even
approaches being proved whenever (A) is only satisfied with
close approximation.2
Very possibly, indeed, it is just this paper which has been
in large measure responsible for the said 'passive resistance.'
This would appear to be indicated by such statements as
the following:
"Thomson has shown in a recent paper that Spearman's
•^ ! 'On Certain Independent Factors in Mental Measurements,' Proc. Roy. Soc.,
A, 96, 1919.
1 'A Hierarchy without a General Factor,' Brit. J. Psychol., 8, 1916.
i62 C. SPEARMAN
method of calculation of data which led to his conclusion of
the existence of a general intellective factor — a 'general
ability' as against special abilities — is open to the gravest
criticism. Thomson attacks the concept on purely mathe-
matical grounds, but his reasoning would appear to be un-
questionably accurate." l
As to whether this final comment can be accepted without
reservation we shall see shortly. But in any case Thomson's
method of proof is a notable contribution to the subject and
will serve, I believe, to clear up much obscurity and misunder-
standing.
He constructed tables of correlations artificially, by an
extension of the method of Weldon.2 Each function — here
representing the marks obtained in a single test by a single
individual — consisted in the total throw of a set of dice. Of
course, there was a separate throw made for each individual.
But there was not a completely separate throw for each test;
instead, some of the dice were marked beforehand and their
resulting points were counted for two or more tests in common.
The result is a correlation amounting, as Weldon showed, to
the proportion of dice used in common; thus, if 4 dice were
counted to two tests in common, whilst 6 other dice were
thrown for each test separately, then the correlation between
the two would on a large number of throws tend towards
4/10 = .4. Moreover, instead of going through the tedious
operation of making the throws actually, the result of an
infinite number of them can easily be obtained theoretically;
it is given by
r = i/(c + m)(c + n}
where c is the number of dice used in common, while m and
n are those used each for one only of the two functions.3
Proceeding in this manner, Thomson constructed the fol-
1 Fernberger, /. of AppL Psychol., x, p. 197, 1917.
2 Cited by Edgeworth, Encycl. Brit., loth ed., XXVIII., p. 282.
3 The equation at once got from the 'Correlations of Sums or Differences,' Brit. J.
Psychol., 5, p. 419, equation (2). The term on the right hand becomes:
e(r = I) + S(r = o) _
2S(r = o) Vc + n + 2S(r = o)
MANIFOLD SUB-THEORIES OF THE TWO FACTORS 163
lowing arrangement, in which 36 of the elements (dice) 'run
through more than one test each, but never through all.'
TABLE I
I
•
3
4
5
6
7
8
9
IO
II
13
«3
'4
*5
16
»7
18
1
(
X
X
X
x
X
x
X
X
d
x
X
x
x
X
X
X
t
x
x
X
x
X
X
/. .
X
X
X
X
x
x
X
h
x
X
X
k
T
T
z
/
X
X
X
»9
ao
31
22
33
•4
»S
96
27
28
39
3°
31
32
33
34
35
36
s
a
X
X
z
X
T
X
T
X
X
X
0
b
X
X
X
X
X
0
c
X
X
T
T
X
T
X
X
X
X
X
I
d
X
T
X
T
X
X
X
X
X
X
3
e
x
T
X
X
X
X
q
/ -•
X
T
T
X
'4
£. .
X
X
X
X
X
16
t.
X
X
T
20
k
X
X
X
22
/
X
X
X
24
In this table, the heading numbers denote the dice thrown,
while the letters are the tests to which they were counted.
Under the heading S, are given the number of throws counted
only to a single test, namely that indicated by the letter in
the same row.
Now, from this arrangement of the dice there resulted the
following set of intercorrelations of the tests [Table II.]
In this Table, he says, the columns correlate with one
TABLE II
a
i
c
d
e
/
e
h
k
/
a
867
73°
593
356
174
167
120
116
112
b
867
6«>
ceo
341
143
137
088
084
082
c
71O
6i;o
500
292
143
091
088
084
041
d
CQ7
i^O
coo
244
095
091
088
042
041
t
3VJ
356
341
292
244
O93
089
043
041
040
/. .
174.
143
143
095
093
044
042
040
039
P. .
f
167
117
OQI
091
089
044
040
039
037
t:
1 20
088
088
088
043
042
040
037
036
k
116
084
084
042
O4I
040
039
037
034
/
112
082
O4I
041
040
039
037
036
035
164
C. SPEARMAN
another, not indeed exactly, but quite as well as in the cases
adduced by me on behalf of the theory of a general factor,
and yet in his arrangement given above there is 'nothing
approaching a general factor.' From this he concludes that
the question as to whether mental tests really do demonstrate
the existence of a general factor will require 'a very much
more extensive set of experiments than has yet been at-
tempted.' And as he remarks that even a thousand cases
would be insufficient, the outlook for further research would
appear to be arduous.
This demonstration, unfortunately, was brought forward
at a time which to me appeared inopportune for scientific
controversies. Hence, my reply consisted — unpardonably,
save for the extraordinary circumstances — in a brief note,
not so much giving my arguments, as indicating what general
lines they would in due course follow.1
/' For one thing, it was indicated that this new view could
be shown not to rest upon a solid foundation by the fact of
its proposing to settle the point through more extensive
experiments; my note suggested, on the contrary, that not
even a million cases could possibly produce a hierarchy dis-
tinguishable from such as could be constructed by the new
method. This suggestion I will now endeavor to justify,
showing that by such a method even the most perfect hierarchy
can be constructed. Suppose, for instance, that we wanted
to get the following tables of correlations between the tests (or
other functions) a, b, c, d, and e, where the columns intercor-
relate quite perfectly and equation (A) is satisfied exactly.
TABLE III
a
b
c
d
e
a
4O
3°
20
10
b
4°
24
16
08
c
1O
24
12
06
d
2O
16
12
04
e
IO
08
06
04
Following the new method precisely, we can at once get this
Table III. if we put together our dice or other elements in
the following manner [Table IV] :
1 Brit. J. Psychol, 8, 1916, p. 284.
/v
MANIFOLD SUB-THEORIES OF THE TWO FACTORS
165
TABLE IV
Dl
ce
Tests
I tO 90
si to
35
3610
47
48 to
57
58 to
65
66 to
7'
ft to
76
7L'°
81 to
83
84 to
85
86 to
9«
92 to
'°5
106 to
129
130 to
163
a. . .
X
X
X
X
b. ..
X
X
X
X
X
c . . .
X
X
X
X
X
d. ..
X
X
X
X
X
e. . .
X
X
X
X
X
The above table means that, for instance, the first 20 dice
have entered into tests a and b only; the next 15 dice into
a and c only; similarly, the others. Obviously enough,
purely specific dice may be added to any extent and in any
manner, without disturbing the hierarchy at all.
Thus, the new argument against the theory of Two Factors
proves too much. By its results — since the theory is ad-
mittedly true when (A) is satisfied exactly — it comes to
suicide.
THE DICE HIERARCHIES FURNISH No EVIDENCE
But the preceding conclusion leaves us with a paradox.
How are we going to explain the seeming fact that the arrange-
ment given in Table IV. does not assign any of the dice to all
the tests, or indeed to more than two? My original note
asserted that Thomson had, after all, let in a general factor
by a 'back door.' Where is there any such thing? And of
what nature is the general factor really? And of what
nature is it?
Moreover, some justification is needed for the charge in
my note against the dice arrangements as being "arbitrary."
We will take this last point first. Let Max and Mpx denote
the total marks obtained by the individual x in the tests a
and p respectively, so that
Max — AiQt\x T" ^12^2* T" * ' ' T~ AmCimx T~ -fVirj
and
r ' ' • ~T *m<Xmx T -Rpxi
where the a's denote the elements (dice) common to any two
or more qf all the tests belonging to the domain in question,
any of the ^'s or P's may be either + I or o, and the R's
1 66 C. SPEARMAN
denote the sums of the elements specific to the two tests
respectively.
Let each M and also each a be measured from its arith-
metical mean for all individuals as origin; and let the units of
measurement be such that the sum of the squares of M for all
individuals is in each test = I. Then, Ah-Pk- ^x(ahx -akx)ln = o,
if the elements h and k are different; but if they are the same,
it may be written as ^//-/y-S(o:2)/w, where S(a2) is constant
for all elements. Hence, we get for the correlational coeffi-
cient between a and p:
ra =
ap
.Pi' + • • • + ^«'-/V), (B)
Using analogous symbols for another test q, and assuming
the validity of equation (A), there results immediately:
A*' -PI + • • • + Amr-pnr = xw-ft' + • • • + Am'Qmf], (o
or
^i'(Pi' - X&') + • • • + Am'(Pm' - \Qm'} = o. (D)
Let us now, keeping the tests p and q constant, allow a
to change its constitution. For every variation of it we
get, if (A) holds good, another equation of the same form as
(D). But there cannot simultaneously subsist a greater
number of independent equations having that form than
vib — I, where m* is the number of ^'s having any of
their coefficients in these equations not zero. Hence, the
variation of a satisfying (A) are confined to m' — I points,
and therefore cannot possibly cover any continuous (or even
'dense') region at all, however minute; for any such has an
infinity of points. We cannot even take m' to be very large,
for then the luck of the throwing would be equalized for the
different individuals, giving them all equal average marks, so
that no correlation would be determinate at all. This proves
that, whatever import the dice hierarchies may possess, at
any rate they do not touch the evidence arrayed on behalf
of the Theory of Two Factors, for there the test a was by no
means confined to a set of selected points, but, in general,
varied at random. The objection against arguing from values
selected arbitrarily appears, then, to be substantiated.
MANIFOLD SUB-THEORIES OF THE TWO FACTORS 167
WHERE THE GENERAL FACTOR LIES
But although the dice hierarchies are thus once more
put out of court as regards the controversial employment
attempted, nevertheless their intrinsic interest may well claim
for them some further examination. In particular, we have
not yet found out what actually constitutes the general
factor in tables such as II. and III.
Now, for Max really to consist of a latent general factor
together with a specific factor (as maintained by the theory
in question) it is necessary if these factors are additive, and
it is sufficient in any case, that we should be able to re-write
Max aS Wa'gx + Sax = «V/*z'g + Sax, (£)
where gx is some function of the a's, g is its mean value and
thus constant throughout, hx is constant for the individual x
with all tests, wa is constant for the test a with all individuals,
whilst Sax varies from individual to individual independently
of both gx and SbX (b indicating any further test).
Let us see, then, whether and how the above conditions
for g2, Sax and wa are fulfilled by equation (A).
We require, first, r^,^ = o.
But this coefficient = f^^ir^-g^ = <r0-rao — «v<702, expanding
by known formula1 and remembering that
Hence, our condition will be fulfilled by rao = zua •<?<,, which
we may write as wa'.
We also want f^^ = o. But this equality, as we see
by similarly expanding r0& into «v«v<702 + att • <r,4 • r(tM), will
hold good when
Tab = Wa-Wb'ffg* = Wa' 'Mb (F)
Actually by equation (A\ on the other hand,
p-rbgyi^rpq-^] (G)
for all combinations of p and qj it therefore may be written
as Va-Vb and thus has precisely the form required by (F).
1 ' Correlation of Sums or Differences/ Brit. J. Psych., V., 1913, p. 419.
1 68 C. SPEARMAN
There only remains, then, to so determine the function gx
that war or rag = va. This done, wa can at once be taken as
Va/ffg-
Such a determination is readily supplied by the sum of
the scores of the individual x in all the tests. For then,
expanding as before,
fag" = f2(o)(o+6+...+ z)
= [i + rab + • • • + rax]2l[z + 2SPOap) + 2Spa(rpfl)],
where p and 5- take all values different from each other and
from a, whilst z is the number of tests,
= [(z - l)0 - 2)rap-ra_q + 2(z - l)rTp + (z - i)r*^ + i]
-r [(z - i)(z - z)r^ + 2(z - i)7^ - i]
= Tap-raqlrpq = vaz, again just as required, (H)
where the scorings indicate means for all combinations of p
and q, in the case of the equation (A] holding good for any
finite domain (as it actually does, according to the empirical
evidence), so that z becomes infinitely large and its lower
powers vanish in comparison with the squares.
Even in the case of equation (A) applying only to some
isolated points, our expression for gx remains an approxima-
tion, and can easily be made exact by appropriately changing
the relative weights of the a's entering into it.
On the whole, then, g may be regarded as the unit of
"General Intelligence," hx indicating how much of it is
possessed by the individual xt whilst wa shows what relative
scope for its influence is afforded by the test a.
This analysis of the general factor and its properties
derives no little interest from the fact that our whole argu-
ment, although based on the dice correlations, is at once
capable of generalization to all product-moment correlations
whatever, merely by removing the restriction of the A's, 5's,
etc., to the values of + I and o. This follows readily if we
(with Bravais himself) assume that any of the functions con-
cerned can be represented with sufficient approximation by
Taylor's expansion to the first differentials. For then we get,
using/ as the symbol of function:
/(ai, «2, • • • a,) = Ai-ai + A2'af2 + • • • + Az-az,
MANIFOLD SUB-THEORIES OF THE TWO FACTORS 169
where the ^'s are now the first differential coefficients and
may have any values.
FORM OF EXPRESSION
There was yet a fourth pledge in my original note, namely
to show that the dice hierarchies, besides re-introducing the
general factor unnoticed, did so in a form that was psycho-
logically altogether untenable.1 This pledge also, I will now
try to redeem, but with the qualification that such psycho-
logical untenability must not be taken to imply psychological
lack of importance and interest.
If we consider generally the form and the fact of de-
pendence, the difference between them appears to be pro-
found. As regards the fact, a function contains, in general,
a determinate number of totally independent variable ele-
ments; this number measures its grade of freedom, its
dimensionality. But the function's form of expression, on
the contrary, can introduce any number desired of what may
be called partially independent variables. For example, x
can always be transformed into a\x\ + #2*2 + • • • + anxn
where #1, #2) *«-i as also a\t a* > • - an, may have any values
whatever, so long as the power is retained of connecting them
together again by means of the still remaining xn. The latter
need only be made = (x — atfi — ... — an-ixn-i)/a. Or,
instead of introducing xn among the variables, some condition
imposed on them (as done by equation A) can have just the
same effect. Again, any function of any two variables,
x and y, can equally well be expressed as functions of two
fresh variables, for instance, of r and 8 where r2 = x2 + y2
and tan 6 = y/x; r and 6 are just as independent of each
other as were x and y.
In many cases, the choice of form between the infinitely
numerous possible alternatives is dictated merely by con-
venience or lucidity. But in other cases, the form, even if it
does not positively state, at any rate strongly suggests, some
facts. In physics, for instance, all the units employed are
mathematically reducible to those of mass, length, and time.
1 Brit. J. Psychol., 1916, 8, p. 284.
17° C. SPEARMAN
Yet in practice, such highly significant units are usually
employed as ' force' (= MLT~2), or of 'work' ( = MZ,2^-2),
or even of 'power' (= MLZT~Z). Here, whilst the number
of totally independent elements constitute the fundamental
theory, the form of expression implies a sub theory.
If, now, we look at the properties of our general factor
shown to be latent in the dice hierarchies, we see from (D)
that they happen to be of the class that do imply facts. For
they are such that any of the a's entering into any test
indicated by a can disappear when the a passes to b indicating
another test, but then some other a is obliged to arise in its
place in order to keep the correlation with any further test
b still in accordance with equation (Z)).
Does not this imply that, the more the test b resembles
the first test indicated by a, the more it must resemble the
second test so indicated however unlike the first one?
And is not this result absurd? The answer to both
questions, I think, must be in the affirmative, so long as
the a's denote any phenomenological characters whatever.
And by 'common elements' psychologists do, as a matter of
fact, habitually mean something phenomenological. But sup-
pose that, instead, we transcend the universe of phenomena
and take the a's to denote portions of some hypothetical
underlying 'force' which is equally effective for varied phe-
nomena. Upon this basis, there is no longer any absurdity,
but a necessity, that for any lost a, there should always
become available another one. With such interchangeable
force, however, we have arrived at neither more nor less
than the hypothesis of a common fund of energy (and inci-
dentally, we have in large measure transferred the burden
of further propf from psychology to physiology).
It may be observed that essentially the same position
reoccurs in the recently published view of Otis, according
to which each test has its own particular degree of 'spread'
of elements.1 For on this view, the fact of test p 'spreading'
so as to contain many common elements with test a\ involves
its also 'spreading' so as to contain many common elements
with any other test a2.
1 J. Educ. Psychol., 1918, 9, p. 345.
MANIFOLD SUB-THEORIES OF THE TWO FACTORS I71
Another form of expression may perhaps be derived from
the very interesting work of McCall.1 He finds that in
several cases the intercolumnar correlations, far from ap-
proaching to -f i.oo, come very nearly to the exact opposite,
— i.oo. This, he urges is in polar contradiction to the
Theory of Two Factors. But such an argument rests upon
a mere misapprehension (quite possibly due to unprecise
wording on my part) ; for it has long been noticed by Webb2
that equation (A} is just as well satisfied by negative as by
positive intercolumnar correlations, so long as they approach
unity. Moreover, it is even perfectly true to say that the
intercolumnar correlation is always + i.oo, so long as the
reservation is made that the units of measurements should
be chosen suitably. By this reservation, the positive sign
can at once be restored throughout the intercolumnar corre-
lations of both Webb and McCall. But even then, there
may be indications of a new and important sub theory.
Yet another instance of a sub theory is in the course of
being elaborated by the investigator already quoted, Thom-
son; this is that of a hierarchy arising from "Two Levels"
of mental process.3 At present, his only published evidence
is 'an arrangement actually arrived at in a trial': and the
value of such a single arbitrarily selected case in a statistical
investigation seems to me, I must admit, open to gravest
question. But when he has supplemented this by a deter-
mination of the amount of intercolumnar correlation produced
by his conditions on an average, and has found that this
average approaches to unity — thereby showing that the
general factor has come in once more — then his view may
lead to very interesting developments.
To close, a word may be allowed as regards my own
attitude towards all these sub theories. For a long time,
I have suggested their possible diversity. And — although
my present preference may have made itself manifest — I
1 Teachers College, Columbia University Contributions to Education, No. 79,
1916. The very newness of his results as compared with those obtained by other
researches, however, must make us demand a proportionately ample corroboration.
» 'Character and Intelligence,' Mon. Suppl., Brit. J. Psychol., 1915, p. 56.
» Brit. J. Psychol., 9, 1919, P- 337 ff.
172 C. SPEARMAN
certainly never have claimed that any considerable evidence
has been produced as yet (either by myself or by any one else)
for or against any one of them. Nor, I am inclined to
think, will they ever admit of absolutely decisive evidence,
since they appear to involve hypotheses incapable of more
than relative probability; these hypotheses, like those
involved in the theory of light, or of electricity, may always
— even after a long reign of acceptance and usefulness — be
suddenly upset again.
But as regards the fundamental theory, I venture to
maintain that this has now been demonstrated with finality.
It involves no hypothesis whatever, but is a direct mathe-
matical deduction from equation (A). Nothing I can con-
ceive will shake it, unless it be detecting some flaw in the
mathematical logic, or convicting all those investigators
whose work supports (A) of a vast conspiracy. So it would
seem as if psychologists have now got definitely to accept
this Theory of Two Factors; it becomes a Bed of Procrustes,
into which all our doctrines must somehow or other be made
to fit, even though the so doing may at times involve a not
unpainful surgical operation upon them.
GENERAL VERSUS GROUP FACTORS IN MENTAL
ACTIVITIES
BY GODFREY H. THOMSON
Armstrong College, University of Durham
I. OBJECTS OF THE PAPER
The objects of this paper are to give a general summary, in
non-mathematical language, of various scattered papers which
in the writer's opinion prove the invalidity of the reasoning
upon which Professor Spearman has based his Theory of
General Ability, or Theory of Two Factors: to submit an
alternative theory, which may be called the Sampling Theory
of Ability, explaining the known facts of mental correlation
at least equally as well as does Professor Spearman's theory:
and, while admitting that the existence of general ability
is still possible, though unproven, to give reasons in support
of the greater probability of the Sampling Theory. The
two theories are not necessarily exclusive of one another.
II. HISTORICAL
The controversy as to whether ability in any individual
is general, or specific, or in groups or 'faculties' is a very old
one, but for the purposes of the present article it is not
necessary to go back prior to the above mentioned paper,1
the first of a series in which Professor Spearman has developed
his Theory of General Ability, or Theory of Two Factors, as
it is alternatively named.
Professor Spearman's method in that paper was to measure
a number of mental abilities, some of them school subjects,
others artificial tests, in a number of persons, and calculate
the correlation coefficients of each of these activities with
each of the others. These correlation coefficients, he then
noticed, had a certain relationship among themselves, a rela-
1 'General Intelligence objectively determined and measured,' C. Spearman,
Amer. J. Psycho!., 1904, 15, pp. 201-293.
174 GODFREY H. THOMSON
tionship which may be called hierarchical order, and is ex-
plained in detail in the technical papers on the subject. He
saw, quite rightly, that the presence of a general factor
would produce this hierarchical order among the coefficients
and, reversing this argument, he concluded that the presence
of hierarchical order proved the existence of a general factor
A number of experimental researches on these lines, in
some of which Professor Spearman himself took part, fol-
lowed during the next eight years, but with very conflicting
results, some experimenters finding the hierarchical order
among the coefficients, others finding no such order. Two
of the best articles of this period are those of Mr. Cyril
Burt,1 who found practically perfect hierarchical order, and
Dr. William Brown,2 who found no trace of such order. The
experimental work in each case was psychologically of a high
degree of excellence.
Things were in this very unsatisfactory state when an
important article by Professor Spearman, in cooperation with
Dr. Bernard Hart, appeared in I9I2.3 In this article the
difficulty of making an unbiased judgment as to the presence
or absence of hierarchical order was recognized, and a form
of calculation was given for obtaining a numerical criterion
of the degree of perfection of hierarchical order, which criter-
ion would be independent of any bias on the part of the
calculator. This criterion ranges theoretically from zero, for
absence of hierarchical order, to unity, for perfection of
hierarchical order. But their formula can, arithmetically,
exceed unity.
The authors applied their criterion to all the experimental
work available, work dating from various periods, and repre-
senting the researches of 14 experimenters on 1,463 men,
women, boys, and girls. From beginning to end the values
of the criterion were positive and very high. The mean was
1 Cyril Burt, 'Experimental Tests of General Intelligence,' Brit. J. Psychol., 1909,
3, pp. 94-177-
2 William Brown, 'Some Experimental Results in the Correlation of Mental
Abilities,' Brit. J. Psychol., 1910, 3, pp. 296-322.
s 'General Ability, its Existence and Nature,' by B. Hart and C. Spearman,
Brit. J. Psychol., 1912, 5, pp. 51-84.
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES 175
almost complete unity. That is to say, Dr. Hart and Pro-
fessor Spearman claimed that all the data then available
showed perfect hierarchical order among the correlation coeffi-
cients, even the data of workers like Dr. Brown and Professor
Thorndike, who had been unable to detect any such order.
The reasons why the hierarchical order among the correla-
tion coefficients was not obvious at a glance were, according
to these authors, two. In the first place, their theory did
not entirely deny the presence of group factors of narrow
range, and tests which were too similar were, according to
them, to be pooled, before the hierarchical order would be-
come apparent. Only in very few cases, however, did they
find it necessary to pool tests in the data used. In the
second place, the obscuring of the perfect hierarchical order
was, according to them, due to the fact that only a small
sample of subjects is examined. For this error allowance is
made in the formula for calculating their criterion.
Dr. Hart and Professor Spearman therefore considered
their 'Theory of Two Factors' proved. This theory con-
siders ability in any activity to be due to two factors. One
of these is a general factor, common to all performances.
The other is a specific factor, unique to that particular per-
formance, or at any rate extending only over a very narrow
range including only other very similar performances. "It
is not asserted," they say, "that the general factor prevails
exclusively in the case of performances too alike, but only
that when this likeness is diminished, or when the resembling
performances are pooled together, a point is soon reached
where the correlations are still of considerable magnitude,
but now indicate no common factor except the general one."
In the same paper Dr. Hart and Professor Spearman
consider, and in their opinion confute, two other theories, (a)
the older view of Professor Thorndike, viz., a general inde-
pendence of all correlations, and (b) Professor Thorndike's
newer view of 'levels,' or the almost universal belief in
'types.' If the former were true, their criterion would, they
consider, show an average value of about zero: if the latter,
a low minus value.
176 GODFREY H. THOMSON
Many experimental researches were inspired by this paper
of Dr. Hart and Professor Spearman, of which, as a good
example, may be cited one in 1913 by Mr. Stanley Wyatt.1
It is I think not too much to say that in practically all of
these the application of the Hart and Spearman criterion
gave values closely approximating to unity and therefore
supporting the Theory of General Ability. But complica-
tions began to arise, of which the first of importance will be
found in Dr. Edward Webb's monograph on 'Character and
Intelligence,' in 191 S-2 Dr. Webb considered that he had
found (in addition to Professor Spearman's General Ability),
a second general factor, which he calls 'persistence of motives.'
Other writers began to find that their data required for their
explanation large group factors, of wider range than those
contemplated in the original form of Professor Spearman's
theory.3 Quite recently Mr. J. C. Maxwell Garnett, discuss-
ing the data of a number of workers with the aid of mathe-
matical devices which he has introduced for the purpose,
concludes that in addition to the single general factor of
Professor Spearman, there are two large group factors which
are practically general4 (one of them being indeed almost
identical with Dr. Webb's second general factor), which he
calls respectively 'cleverness' and 'purpose,' both distinct
from general ability.
It is clear therefore that in any case the simple original
form of Professor Spearman's theory is becoming complicated
by additions which tend to modify it very considerably.
Meanwhile, however, the present writer had come to the
conclusion that the mathematical foundations upon which it
was based were in fact incorrect. Before developing the
line of argument which led to this, it will be well to re-state
Professor Spearman's case in its simplest terms in a few words.
1 Stanley Wyatt, 'The Quantitative Investigation of Higher Mental Processes/
Brit. J. Psychol., 1913, 6, pp. 109-133.
2E. Webb, 'Character and Intelligence,'^'/. /. Psychol., Monog. Suppl., 1915,
3, pp. ix and 99.
3 See especially E. Carey, 'Factors in the Mental Processes of School Children,
Brit, J. Psychol., 1916, 8, pp. 170-182.
4J. C. Maxwell Garnett, 'General Ability, Cleverness, and Purpose,' Brit. J.
Psychol., 1919, 9, PP- 345-366-
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES 177
It is entirely based upon the observation and measurement
of hierarchical order among correlation coefficients. It states
that after allowance has been made for sampling errors this
hierarchical order is found practically in perfection. And
it finally states that such a high degree of perfection can only
be produced by a general factor, and the absence of group
factors, which would mar the perfection.
III. THE CASE AGAINST THE VALIDITY OF PROFESSOR
SPEARMAN'S ARGUMENT
It is possible, by means of dice throws or in other ways, to
make artificial experiments on correlation, with the immense
advantage that the machinery producing the correlation is
known, and that therefore conclusions based upon the correla-
tion coefficients can be confronted with the facts. Working
on these lines, the present writer made, in 1914, a set of
imitation 'mental tests' (really dice throws of a complicated
kind) which were known to contain no general factor. The
correlations were produced by a number of group factors
which were of wide range, and, unlike Professor Spearman's
specific or narrow group factors, they were not mutually
exclusive.
These imitation mental tests, containing no general factor,
gave however a set of correlation coefficients in excellent
hierarchical order, and the criterion was when calculated
found to be unity, so that had these correlation coefficients
been published as the result of experimental work, they would
have been claimed by Professor Spearman as proving the
presence of a general factor.1
In a short reply Professor Spearman laid stress on the
fact that this arrangement of group factors which thus pro-
duced practically perfect hierarchical order was not a random
arrangement, and that it was exceedingly improbable that
this one special arrangement should have occurred in each
of the psychological researches of many experimenters, so
improbable indeed as to be ruled entirely out of court.2
1 Godfrey H. Thomson, 'A Hierarchy without a General Factor,' Brit. J. Psyckol.,
1916, 8, pp. 271-281.
1 C. Spearman, 'Some Comments on Mr. Thomson's Paper,' Brit. J. Psychol.,
1916, 8, p. 282.
178 GODFREY H. THOMSON
It is clear that Professor Spearman did therefore definitely
admit that at any rate one arrangement of group factors
existed which would give hierarchical order of sufficient perfec-
tion to satisfy completely his criterion. He did more than
this, however, for he claimed already to have published,
without proof, in an earlier paper, what the effect of a really
random overlapping of all the factors in his opinion is, namely
that in this case his criterion will be of the same value as
the average correlation between the tests.1 Now the average
correlation between the tests employed in the psychological
researches under consideration is not as a rule low. Indeed
in those tests which really play an important part in the
calculation of the criterion it is usually very high. So that
this criterion would, if Professor Spearman's admission be
correct, apparently be high on the random overlap theory;
that is to say sheer chance would produce considerable though
not perfect hierarchical order. This already puts the proof
of the Theory of General Ability into a very different position
from that which it appeared to occupy immediately after the
publication of the paper of Dr. Hart and Professor Spearman
in 1912. For in that paper the alternative theories gave
values of the criterion which were either zero or negative,
and the fact that it actually came out to be almost unity
seemed conclusive. But now the comparison is much less
definite, for here is a theory which may give high positive
values. The criterion must not merely be high and positive
to prove the Theory of Two Factors, it must be absolutely
unity. True, in Professor Spearman's calculations it does
come to unity with most remarkable regularity. But if it
can be shown that these calculations are in any way erroneous,
then the fact- that the comparison is with a theory which
can give a high criterion, and not merely with theories which
give zero or less, is of great importance.
One reply which Professor Spearman might make to this
step of the argument is contained by implication in the
footnote on page 109 of the already quoted 1914 article in the
1 C. Spearman, 'The Theory of Two Factors,' PSYCHOL. REV., 1914, 21, p. 109.
See also E. Webb, Character andlntelligence,' Brit. J. Psychol. Monog. SuppL, 1915,
No. 3, on page 57 and Appendix, page 82.
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES iJ9
PSYCHOLOGICAL REVIEW. He appears to think that on the
random overlap theory the criterion and the average correla-
tion, though equal, will both be zero or very small on the
average. In other words, on this view random overlap will
produce hierarchical order if it produces correlation at all,
but usually both will be zero. To return to the reply of
Professor Spearman to the * Hierarchy without a General
Factor,' the reply namely that this special arrangement would
doubtless give such order, but was too improbable to be
seriously considered, and that a random arrangement of Group
Factors, though it might give some hierarchical order, would
not give it in the perfection actually found: the obvious way
to find out if this is so or not is to try it, with artificial ' mental
tests' formed of dice throws. This the present writer did in
November and December of 1918, after an unavoidable delay
of some years. Sets of artificial variables (analogous to the
scores in mental tests) were made, in each of which the ar-
rangement of group factors was decided by the chance draws
of cards from a pack.1 It was found that in every case a
very considerable degree of perfection of hierarchical order
was produced, quite as high as that found in the correlation
data of experimental psychology. A further test was made
on that set of data, from among these artificial experiments,
which appeared to yield the least perfect hierarchical order.
The true values of the correlation coefficients being known,
the true degree of perfection of hierarchical order could be
correctly calculated, and was 0.59 (perfection being repre-
sented by unity). Dice throws were now made to obtain
experimental values of the same correlations, and Professor
Spearman's criterion applied. As it has done in the case of
so many experimental researches in psychology, it gave the
value unity.2 This set of correlation coefficients, therefore,
if it had been published as the result of experiments on mental
1 Godfrey H. Thomson, 'On the Cause of Hierarchical Order among the Correlation
Coefficients of a Number of Varieties taken in Pairs,' Proceedings of the Royal Society
of London, 1919, A, 95, PP- 400-408. See also, by the same author, 'The Hierarchy of
Abilities,' and 'The Proof or Disproof of the Existence of General Ability,' in Brit. /.
PsychoL, 1919, 9, pp. 321-344.
1 Godfrey H. Thomson, 'On the Degree of Perfection of Hierarchical Order among
Correlation Coefficients,' Biometrika, 1919, Vol. 12, pp. 355-366.
i8o GODFREY H. THOMSON
tests, would have been claimed by Professor Spearman as
additional proof of the existence of a general factor, although
in fact there was no such general factor present, and the
correlations were due to randomly selected group factors.
The conclusions which appear reasonable from this are
(a) that hierarchical order, unless perhaps when it is abso-
lutely perfect, is no proof of the existence of a general factor,
and (b) that the Hart and Spearman criterion for hierarchical
order is somehow incorrect, and exaggerates the degree of
hierarchical order present.
The errors which cause this exaggeration are pointed
out in the last cited article in Biometrika, and are mainly two.
In the first place, Dr. Hart and Professor Spearman assumed
certain quantities to be uncorrelated when they are really
strongly correlated, though in a peculiar manner. This error
causes the possible values of the criterion to be distributed,
not from zero to unity, but from zero to infinity. In the
second place, they employ a 'correctional standard' which
rejects all the values greater than about i|. The possible
range for the accepted values is in practice from about ^ to
i^, and their average is naturally about unity. In other
words, the remarkable regularity with which this criterion
gives the value unity is not a property of the investigated
correlation coefficients at all, but is a property possessed by
the criterion itself, due to errors and the action of the 'correc-
tional standard.'
In the writer's opinion the work outlined in this section
of the present paper finally proves the invalidity of Professor
Spearman's mathematical argument in favor of the Theory
of Two Factors. If this be so that theory returns to the
status of a possible, but unproven, theory.
IV. HIERARCHICAL ORDER THE NATURAL ORDER AMONG
CORRELATION COEFFICIENTS
The fact is that hierarchical order, which Professor Spear-
man was the first to notice among correlation coefficients,
is the natural relationship among these coefficients, on any
theory whatever of the cause of the correlations, excepting
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES 181
only theories specially designed to prevent its occurrence.
It is the absence of hierarchical order which would be a
remarkable phenomenon requiring special explanation; its
presence requires none beyond what is termed chance.
An analogy from the simple repeated measurements of a
linear magnitude may help to illustrate this. Indeed it is
rather more than an analogy, being in fact the same phe-
nomenon in its simplest terms and dimensions. It is well
known that many measurements of the same quantity, made
with all scientific precautions, under apparently the same
conditions, and with an avoidance of all known sources of
error, nevertheless do not give a number of identical values.
The values are all different, but are not without law and order
in their arrangement. They are grouped about a center from
which the density decreases in both directions, and it is found
that this grouping is for most practical purposes closely
represented by the Normal or Gaussian Curve of Error.
Experimenters are not surprised to find their data obeying
the Normal Law, nor do they require a special theory to
explain it. On the contrary, it is the departures from the
Normal Law which if wide would cause alarm and require
special investigation, and if confirmed would require a special
theory. In the same way hierarchical order among correla-
tion coefficients should not cause surprise, though any marked
variation from this order would demand investigation.
Correlation coefficients are ^themselves correlated, and n
correlation coefficients form an n-fold or w-dimensional corre-
lation-surface. The particular and convenient form of tabu-
lation of correlation coefficients adopted by Professor Spear-
man and followed by most other psychological workers brings
to light, in the form of "hierarchical order," one of the
properties of this correlation-surface of the correlations.
It is true that in the ordinary form of the theory of corre-
lation of correlations,1 the variations in the correlation coeffi-
cients to which the correlation of these coefficients refers,
are variations due to sampling the population; i.e., to taking
1 K. Pearson and L. N. G. Filon, 'On the Probable Errors of Frequency Constants
and on the Influence of Random Selection on Variation and Correlation,' Phil. Trans %
of the Roy. Soc. London, 1898, 191 A, pp. 299-311.
182 GODFREY H. THOMSON
in our case a class of only perhaps 50 English grammar school
boys of age 12, instead of all such boys: whereas the hier-
archical order we desire to explain is already found in the
true 'theoretical' correlation coefficients. This difference is
however one of point of view only. It was left partially un-
explained in the above cited article1 although it was referred
to. Further consideration leads to the following resolution
of the difficulty.
Suppose that n variates (in our work the scores in mental
tests) are so connected by factors that the correlations are all
equal and positive. Then let a small sample of the popula-
tion be taken. The observed correlations will show departures
from equality, and will be found to be in hierarchical order.
This hierarchical order is due to sampling the population.
Now consider why the correlations do not come out at their
true values. They give of course the true values for the
sample. The reason of their departing from the true values
of the whole population is that (a) some of the factors which
really are links between the variates (the mental activities)
happen to have remained steadier than usual during the
sample. In the limit a factor might happen to retain exactly
the same value through the various individuals of the sample.
That is, some of the linking factors do not in reality come into
action, or not in their full force, (b) On the other hand,
some factors which are really different and unconnected may
happen by chance to rise and fall together, throughout the
sample, and more or less to act as one. That is, fictitious
linking factors are created, which would disappear with a
larger sample.
Clearly therefore a hierarchy of correlation coefficients,
caused by sampling the population, is due to chance having
caused a change in the apparent factors acting. It follows
that if we make a real change in the factors acting, we shall
get a hierarchy, and this is what we do when we choose the
mental tests to be employed in any research. Each mental
test is a test of a sample of abilities.
The laws governing the correlation of correlation coeffi-
1 Godfrey H. Thomson, Proc. Roy. Soc. London, 1919, A, 95, pp. 407 and 408.
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES 183
cients which vary because of sampling the population can,
in fact, be applied without hesitation to the relationships
between 'true' correlations in the whole of any population
simply because any such population is itself a sample. Eng-
lish grammar school boys of 12 are themselves a sample of a
larger boyhood; the whole human race indeed is a sample
of 'what might have been,' selected by the struggle for
survival.
The whole question clearly has philosophical bearings on
the degree of reality of causal connections; for on this view
those chance links in a small sample which were a few para-
graphs ago termed 'fictitious links, which would disappear
with a larger sample,' do not differ except in degree from the
'real' causal links which we only term real because they
persist throughout the largest sample with which we are
acquainted.
In another direction there are connections with the differ-
ence, which is one of degree only, between what is called
'partial' correlation and 'entire' correlation.1
The conclusion to be drawn from this section of the present
paper is that hierarchical order is the natural order to expect
among correlation coefficients, on a theory of chance sampling
alone, and that therefore, by the principle of Occam's razor,
its presence cannot be made the criterion of the existence of
any special form of causal connection, such as is assumed in
the Theory of Two Factors.
V. A SAMPLING THEORY OF ABILITY
In place therefore of the two factors of that theory, one
general and the other specific, the present writer prefers to
think of a number of factors at play in the carrying out of
any activity such as a mental test, these factors being a
sample of all those which the individual has at his command.
The first reason for preferring this theory is that of
Occam's razor. It makes fewer assumptions than does the
1 See Karl Pearson, 'On the Influence of Natural Selection on the Variability
and Correlation of Organs,' Phil. Trans. Roy. Soc. London, 1902, A, 200, pp. 1-66.
Godfrey H. Thomson, 'The Proof or Disproof of the Existence of General Ability,'
Brit. J. Psychol., 1919, 9, pp. 321-336.
184 GODFREY H. THOMSON
more special form of theory. It does not deny general
ability, for if the samples are large there will of course be
factors common to all activities. On the other hand it
does not assert general ability, for the samples may not be
so large as this, and no single factor may occur in every
activity. If, moreover, a number of factors do run through
the whole gamut of activities, forming a general factor,
this group need not be the same in every individual. In other
words general ability, if possessed by any individual, need
not be psychologically of the same nature as any general
ability possessed by another individual. Everyone has prob-
ably known men who were good all round, but Jones may be
a good all round man for different reasons from those which
make Smith good all round.
The Sampling Theory, then, neither denies nor asserts
general ability, though it says it is unproven. Nor does it
deny specific factors. On the other hand it does deny the
absence of group factors. It is this absence of group factors
which is in truth the crux of Professor Spearman's theory,
which is not so much a theory of general ability, or a theory
of two factors, as a Theory of the Absence of Group Factors.
And inasmuch as its own disciples have begun to require
group factors to explain their data, its distinguishing mark
would appear in any case to be disappearing.
Such group factors as are admitted by Professor Spearman
are of very narrow range, and are mutually exclusive, that is
they do not overlap. Both these points follow from the
sentence used in the 1912 article with Dr. Hart, where it is
said that, in the case of performances too alike, 'when this
likeness is diminished, or when the resembling performances
are pooled together, a point is soon reached where the correla-
tions are still of considerable magnitude, but now indicate
no common factor except the general one.'
Since this point is soon reached, the group factors must
be narrow in range. Since pooling a few performances will
obliterate any group factors, they must be exclusive of one
another. For if A, B, C and D are four tests, in which A and
B have a group factor common to them, and C and D another,
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES 185
then of course by pooling A with B and also C with D we can
obtain two pools AB and CD which have no link. But if
A, B and C have one group factor, and C and D have another
then these group factors cannot be separated into specific
factors. In fact, a specific factor is a separated group factor,
and Professor Spearman's theory asserts that group factors, if
any, are separable and mutually exclusive. This is to the
present writer the great stumbling block in the way of the
acceptance of the Theory of Two Factors, unless 'specific
factor' is interpreted in the way suggested later in this
article.
It is a fact which will be admitted by most that the same
activity is not performed in the same way by different indi-
viduals, even though they are equally expert. Not only are
specific factors therefore required by this theory for every
separate activity, excluding only any which are very closely
similar; but also specific factors of different psychological
natures are required for each individual. Further, the same
individual does not always perform the same activity in the
same way. A man using an ergograph will, as he tires, begin
to employ muscles other than those naturally used at the
outset. When we are returning from a cycle ride muscles
are used in a different manner from the style adopted at the
start, indeed sometimes deliberate changes are made to give
relief. And in the same way a mental task is performed by
different methods at different times. Does this then mean a
different specific factor for each way of doing a task? All
these difficulties appear to argue against the Theory of Two
Factors, and seem to the present writer to be considerably
cleared up by the Sampling Theory.
Finally, the Sampling Theory appears to be in accordance
with a line of thought which has already proved fruitful in
other sciences. Any individual is, on the Mendelian theory,
a sample of unit qualities derived from his parents, and of
these a further sample is apparent and explicit in the indi-
vidual, the balance being dormant but capable of contributing
to the sample which is to form his child. It seems a natural
step further to look upon any activity carried out by this
1 86 GODFREY H. THOMSON
individual as involving yet a further sample of these quali-
ties.
VI. THE DIFFICULTY OF 'TRANSFER OF TRAINING'
Although Professor Spearman's Theory of Two Factors
has been chiefly based by him on the line of argument which,
it is suggested, has now been proved invalid, viz., the 'hier-
archy' argument, yet there is another and powerful form of
reasoning which can be brought up to its support, based
upon the fact that, according to some experimenters, improve-
ment in any activity due to training does not transfer in any
appreciable amount to any other activity, except to those
very similar indeed to the trained activity. And even those
workers who do not agree that this is an experimental fact are
usually content to take a defensive attitude and say that
transfer is not disproved. Few if any will say that it is
proved.
This certainly seems to point to the absence of group
factors, and to support Professor Spearman's theory, which
only needs to add to itself the assumption that the specific
factors are, while the general factor is not, capable of being
improved by training, to fit the case admirably. Of course,
if transfer really occurs, the argument proves the opposite.
And although psychological experiment points on the whole
to the absence or the narrowness of transfer, yet popular
opinion among business men, schoolmasters, and others is in
favor of transfer to a considerable extent. Assuming no
transfer, however, how can the Sampling Theory, with its
numerous group factors, explain this?
It is necessary to assume that the group factors are all
unimprovable or only slightly improvable by training, though
they may change with the growth and development of the
individual. The improvement which certainly takes place
when we practice any activity is due, it may then be assumed,
not to improvement in the elemental abilities which form the
sample, but to a weeding out, and selection of these. The
sample alters, mainly no doubt is diminished, though addi-
tions are also conceivable. It becomes a more economical
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES 187
sample, and waste of effort in using elements which are un-
necessary is avoided. Improvement in any mental activity
may on this view be compared with improvement in a manual
dexterity, in which it is notorious that the improvement
consists largely in the avoidance of unnecessary movements.
When another activity is then attempted, the elemental
factors are just the same as they would have been had the
practice in the first activity not taken place. The new ac-
tivity will be performed by a new group of factors, which
sample will as in the first case be in the beginning wasteful
and will include many unnecessary elements. Transfer of
improvement gained in the first activity will therefore not
take place except insofar as the second activity is recognized
as a mere variant of the original one, in which case the weeding
out process which has taken place in the first case may be
done at the very first attempt, at any rate to some ex-
tent.
To use another analogy, the improvement which takes
place when a football team practices playing together for a
series of matches is due more to team work than to individual
improvement. A new team, even though it contain a large
proportion of players from the first team, will not have this
unity of action. There will be little transfer of improve-
ment.
According to the view here developed, it is the weeding
out of the sample of elemental abilities which is specific.
The team work is specific, though the players play for several
clubs. This would appear to enable a reconciliation to be
affected between the almost universal belief in 'types' of
ability (to which Professor Spearman refers) and the experi-
mental facts concerning both correlation and transfer. If
there be a general factor at all, it might be the power to
shake down rapidly into good team work, in a word, educa-
bility. But there seems no objection to assuming that this,
instead of being a general factor, is a property of each ele-
mental factor, varying from factor to factor.
To sum up this section: if transfer of training really does
not occur to any great extent, then it has to be admitted that
1 88 GODFREY H. THOMSON
the Theory of Two Factors readily explains this. But the
Sampling Theory can also do so, in a manner which is perhaps
not so easy to set forth, but which nevertheless appears to
the present writer to be more illuminating and less artificial
than the alternative theory.
VII. THE 'FACULTY FALLACY'
Since the group factors spoken of in this Sampling Theory
are, in the fact that they are supposed to come into play in
many different activities, similar to the banished 'faculties'
of the mind (though the writer conceives of them as being
smaller units than were those faculties) it is probably neces-
sary to defend the theory against the charge of committing
what is known as the 'Faculty Fallacy.' This defence is
easy. It is only necessary to point out (a) that a person who
believes in 'faculties' or 'types' or 'levels' does not neces-
sarily commit the above-mentioned fallacy, and (b) that any
charge of being 'faculties' which may be brought against the
group factors can of course also be brought against the general
factor.
The clearest account of the faculty fallacy known to the
writer is given in the older edition of Professor G. F. Stout's
'Manual': "An effect cannot be its own cause, and cannot,
therefore, afford its own explanation. But it is a fallacy of
not infrequent occurrence to assign as a cause what turns out
on examination to be only the effect itself, expressed in
different language. . . . The classical instance of this con-
fusion is the answer of Moliere's physician to the question:
'Why does opium induce sleep?' 'Opium,' he answers, 'pro-
duces sleep because it has a soporific tendency.":
Now it is to be clearly noted that there is no logical
objection to the physician saying either that opium produces
sleep, or that it has a soporific tendency. All that he must
avoid doing is to give the one as the cause of the other. And
in a similar way there is no logical objection to anyone believ-
ing, on the ground of experiment, that if a man has a good
memory for historical matters he will, as a fact, have a
good memory for all other matters. But if he believes this
GENERAL VERSUS GROUP FACTORS IN MENTAL ACTIVITIES 189
without any other ground than that the name memory is
given to these diverse activities, then indeed he is committing
the fallacy in question. Even if a man uses the form of
words: "Robinson will be a good man for this post, because
he has a good memory," he is not necessarily committing
any logical fallacy. He may very well mean by this short
statement something like the following: "I have noticed
that a man who remembers one class of facts well is also
frequently good at remembering other classes of facts. I
know that Robinson can remember such and such things
easily and accurately, therefore I think it very probable that
he will be above the average in this job, which requires the
memorizing of certain facts." And in this there is no fallacy,
whether the conclusion be true or false.
The existence of the group factors spoken of in this paper
is deduced with more or less probability from the known
experiments. Their existence is an hypothesis which explains
these facts, though it is not the only hypothesis to do so.
If, as is very probable, the language used in any part of this
paper is open to an interpretation which would involve the
fallacy, then it can only be said that this is not the interpreta-
tion which is intended.
VIII. CONCLUSIONS
Professor Spearman's Theory of Two Factors, which as-
sumes that ability in any performance is due to (a) a general
factor and (b) a specific factor (group factors being absent,
or at any rate very narrow in range and mutually exclusive)
is based chiefly on the observed fact that correlation coeffi-
cients in psychological tests tend to fall into 'hierarchical
order.' It has been shown, however, that the criterion
adopted for evaluating the degree of perfection of hierarchical
order present is untrustworthy and has led to overestimation.
Such hierarchical order as is actually present is in fact the
natural thing to expect, and it is the absence of such which
should occasion surprise. The proof of the Theory of Two
Factors which is based on the presence of hierarchical order
therefore falls to the ground. The theory remains a possible
190 GODFREY H. THOMSON
explanation of the facts but ceases to be the unique explana-
tion. As an alternative theory there is here advanced a
Sampling Theory of Ability, in which any performance is
considered as being carried out by a sample of group factors.
This theory is preferred because it makes fewer and less
special assumptions, because it is more elastic and wider, and
because it is in closer accord with theories in use in biology
and in the study of heredity.
SUGGESTIONS TOWARD A SCIENTIFIC INTER-
PRETATION OF PERCEPTION
BY J. R. KANTOR
University of Chicago
Much of the criticism directed toward the results of psycho-
logical investigation might serve as a direct challenge to psy-
chologists to clarify their interpretations of psychological
phenomena; for a study of those criticisms amply reveals the
bizarre views attributed to psychologists. Although this pe-
culiar situation obtains with respect to all psychological
descriptions, it is especially striking in the case of per-
ception. Thus, a recent writer1 finds it necessary to point
out that an object is not merely a thing which 'starts a chain
of vibrations which eventually results in its own creation.'
To the present writer this specific criticism does not really call
for a defense of the psychologist's position, since the critic
holds substantially the same view as most psychologists, but
the very fact that a writer will find much to criticize in any
one who supports a similar doctrine is a symptom of a confusing
situation which demands at least a restatement of perception.
Naturally enough the confusions mentioned reach deeper
than the mere matter of exposition and in fact arise directly
from the types of conceptions held concerning the process under
discussion. A careful reading of psychological literature on
perception creates the suspicion that the descriptions fail to
tally with the actual facts in the case. As a striking example
we find that perception is described as in some sense a creative
process which functions in the organization of the discrete
qualities constituting the objects of our reaction. In effect,
we find practically all current perceptual doctrines very
strongly reminiscent of Berkeley's subjectivism albeit modified
somewhat a la Reid; the latter modification results in the
view that there exists a percept as well as an object of percep-
1 J. B. Pratt, /. of Phil., Psychol., etc., 1919, 16, 596 ff.
I92 /. #• KAN TOR
tion. Psychologists cannot but consider the problem of per-
ception as crucial, since the admission of a non-scientific sub-
jectivism at this point will bring disastrous consequences into
the entire science of psychology. In this article the writer
attempts to suggest a description of perception, which, so far
as it goes, consistently complies with the rigorous canons of
natural science.
I.
General Description of Perception. — Perception is the con-
scious behavior through which are developed the meanings of
objects and relations which operate in the adaptation of the
individual to his surroundings and in the control of them. It is
precisely in the process of perception that the individual, in
direct contact with objects, develops reaction patterns enabling
him to differentiate and distinguish the various objects affecting
him.
At the outset it must be noted that the act of perception1
is an adjustmental reaction, an actual interaction of one natural
object with another. But the precise difference between this
kind of interaction and some other is, namely, that one of the
interacting objects is a psychophysiological organism to whom
the results of the present interaction will become significant
in influencing future contacts of this object (person) with the
same or a similar object. Consider, that what was formerly
a mere interconnection between objects becomes what we
might now call a knowledge process because the reaction
becomes a means to some other form of reaction; that is to
say, the first natural contact with an object is the basis for
the development of an anticipatory reaction system. If the
person is once burned, the object which produces this effect
will upon a future occasion stimulate a touch inhibition reaction
rather than a touch response. An empirical fact it is, there-
1 While the writer is in complete sympathy with Watson in his revolt against
subjectivism, and in his assertion that functional psychology is just as guilty in this
respect as the structural view, he cannot assent to Watson's implication that perception
among other processes is not properly the subject matter of a non-subjectivistic
psychology. Nor indeed does Watson omit perception when he is interested in
4 integrations and total activities of the individual.' His rejection of the terms is
obviously to allow room for a predominantly physiological tone to his discussion.
SCIENTIFIC INTERPRETATION OF PERCEPTION 193
fore, that all developed1 perceptual responses operate as knowl-
edge reactions, for in this way only do we learn to discriminate
between objects, and to anticipate the specific response we
should make to a particular object. But it is of extreme
importance to notice that the perceptual reaction is not in its
primary occurrence a knowing. To overlook this fact is to
fall into the error of finally resolving the objects of our reactions
into knowledges of some sort, and the history of psychology
stands to witness that on the basis of such premises we in-
variably land in a mentalistic world in which objects are
reduced to sensations, and the world of fact and science
disappears in our description.
Only upon the assumption that the perceptual reaction is a
natural psychophysiological response, the writer submits, can
we achieve a natural science interpretation of the development
of discriminative meanings. By thus investigating all the
components of an act we may hope to obtain a scientific
description of the total response and escape the arbitrary and
confusing concept of a mental content, which is an unavoidable
consequence of the presupposition that perception is a knowl-
edge process.2
We must, then, look upon the perceptual reaction as a
complex adjustment from which is derived the significance of
objects through the integration of reaction patterns. This
meaning of objects we shall see may be resident in the response
pattern, or it may be more remotely connected with it, even
to the point at which the act is no longer a perceptual but a
conceptual reaction; in the latter case we observe that the
meaning is detached from any overt act, and as a matter of
fact we find that such detached meanings constitute the im-
plicit functioning of the original reaction which ultimately
generated the conceptual meaning.
Primary Perception and Simple Apprehension. — Upon the
basis of the specific operation of meanings we may distinguish
two definite forms or degrees of perceptual response which we
1 Note the distinction drawn between perception in development and perception
in use on another page of this paper.
1 Expressed in the statement that perception is the consciousness (awareness)
of an object present to sense.
194 /• R- KAN TOR
will call primary perception and simple apprehension. In the
former case, the meaning of the object responded to resides
within the reactional movement of the person,1 as illustrated
by the perceptual process of an instinctive act. The meaning
of a 'danger' object for the person is merely the startled jump
which constitutes the operation of a connate reaction pattern.
It must be observed that in this situation the neuro-muscular
and neuro-glandular factors in the response are very prominent,
and as a record of fact, the cognitive component merely consists
of a simple appreciation of the presence of the stimulating
object.2
In simple apprehension the meaning becomes more and
more detached from the immediate condition of response.
Instead of the mere presence of an object calling out a specific
reaction, the object may serve as a symbol for some action.
In consequence, the discriminated significance of the object
will be attached not to the direct movement as in primary
perception, but to another response which is to follow. Evi-
dent it is that a meaning of this type is an implicit response
in the form of an anticipatory process similar to that we in-
variably find as an important factor in all delayed responses,
whether simple acts or chains of acts. This capacity to detach
meanings gives the person a greater control over the objects
of his environment, for, if the meaning of the object is appre-
ciated before an overt response is made, the type of response
can be widely varied between limits.3 In contrast to primary
perception the meaning in simple apprehension is always cor-
related with an awareness-attention process.
Implicit perception functions in adaptational situations in
which there are more definite appreciations of the surrounding
objects. We, might take the case of meeting a friend in which
there is a complete and definite meaning element. Conse-
quently the overt action which takes place is more conditioned
1 We might just as well say the meaning is in the object, but it is clear that unless
there is an action involved the problem even of the location of meaning does not arise.
2 The components spoken of are, of course, abstracted from the actual response by
logical analysis.
3 Cf. my brief suggestion concerning the detachment of meanings, PSYCHOL. REV.,
1919, 26, 2 ff.
SCIENTIFIC INTERPRETATION OF PERCEPTION *95
by the meaning component. If he is an American friend, I
may merely shake hands with him, but if he is a foreigner, I
will probably also raise my hat and bow. Clearly the entire
course of my behavior in this situation presupposes my
familiarity with the person. It must not be overlooked that
we do not exclude from our description of simple apprehension
the simpler immediate reactions which occur in primary per-
ception. For the fact is, that since simple apprehension is
always the development of an act of primary perception, it
involves therefore an integration of the simpler acts. Of prime
importance here is the fact that it is precisely through the
integration of the simpler acts that a person profits by past
experience. For instance, my reaction to this person is condi-
tioned by the numerous integrations of responses representing
my previous contacts with him.
Thus through the constant growth of the reaction pattern
does the perceptual process undergo a continuous development.
Not only does a given response serve at any specific time as
an adaptational function, but also as a developing potentiality
for some future contact between the person and the object.
Analytic Description of Perception. — Although the percep-
tual response is a thoroughly organic process, we can never-
theless analyze it into a series of specific stages or act com-
ponents which we can tabulate as follows:
1. The attention function in correlation with contact media
(light rays, for example).
2. Functioning of a reaction pattern which involves
(a) Discrimination and appreciation of specific qualities
and relations of objects coupled with conative and
affective factors.
(b) Neuro-musculo-glandular processes.
3. Emergence of meaning (new).
4. Overt adaptation follows.
i. The attention factor is the selective process which serves
to prepare the individual for a new reaction. At any moment
of time innumerable possibilities for action naturally exist
because of the previous acquisition of many reaction systems.
The change in the surrounding medium or media of the person,
I96 /. R* KANTOR
which occurs when the individual comes into the presence of
new objects, or when objects change their positions with respect
to the person, puts him into a condition of readiness to react
to some new object. It must be observed that the attention
processes depend not only upon the stimulating object and
its setting, but also upon the condition of the organism at the
time, that is to say, the selection process depends very directly
upon what the activities of the person were prior to the present
contact. Such activities condition also what precise phase of
an object we will react to at a given time. Thus, for example,
the problem as to why at one time our attention is attracted
to a red solid instead of a smooth surface, when both form the
phases of a book, is solved by an investigation of the previous
activities of the person.
2. Following the selection function, the reaction pattern
is brought into activity, and we find thus a highly coordinated
series of processes taking place. These may be enumerated
separately, although they constitute merely descriptive phases
of a unitary process. Here we find the discriminative process
which enables the organism to distinguish the various qualities
and relations of things. This phase may be thought of as the
cognitive aspect of the reaction system, and to a degree we
may look upon this phase as conditioning the mode of operation
of the entire complex. The conative factor in this complex,
being very closely connected with the attention function, may
be considered as the aspect which conditions the occurrence
of a response at all. Of primary importance are the affective
processes, which in part predispose the organism to act. Every
reaction pattern involves of course also the elaborate function-
ing of musculo-neural and neuro-glandular processes, which
are so prominent as to convince some observers that they
constitute the total reaction pattern.
3. As a result of the operation of the reaction pattern a
new effect is or may be produced upon the organism. Should
the object or person reacted to, with all the involved relations,
remain constant, no new reaction is called out; that is, the
previously developed reaction pattern remains unmodified
despite the present contact. The object, then, will not take
SCIENTIFIC INTERPRETATION OF PERCEPTION
on any new meaning and the overt act following the apprecia-
tion of the identity of the object may be precisely like one that
has previously occurred. We can readily determine this to
be the case of perception in use. On the other hand, should
the previously developed reaction system prove inadequate
for the purpose of the present contact, new features may
develop. Instead of involving some given system of receptors
in connection with certain neural and muscular processes,
additional factors may be put into operation. Thus, for
example, should the apple previously sound and firm to touch
now offer no resistance, it will call out different muscular
responses. Similarly, should it now present color surfaces
varying in hue, turning from red to brown, the object will
take on new meaning, and we will react in a different way to
the now deterioriated apple. Thus, indefinitely many modi-
fications are developed in the course of the exercise of so
intricate a psychophysiological response pattern.
4. Following upon the operation of the definitively per-
ceptual reaction system, the person performs some sort of
overt act. •The latter is directly conditioned by the emergence
of the meaning brought out through the course of the specified
contact with the object. It must be observed that the specific
perceptual process is a coordinate process with some other
type of reaction system. Thus, we should look upon the
perceptual function as a part of a perceptual-instinct, per-
ceptual-emotional, or perceptual-voluntary action, etc. To
look upon it in this way obviates the dangerous view that in
the actions mentioned we have isolated activities. As a pre-
liminary or partial action the perceptual process represents an
evaluation of the object which leads to a definite overt response.
It is at this point that the perceptual reaction becomes a
knowledge function, since it stands for some actual adjust-
mental act. Whether the apple of our illustration will be
eaten, or thrown away, depends upon the information elicited
through the operation of the perceptual reaction system and
its modifications. At this point, we must not overlook the
fact that the appreciation of an edible or non-edible meaning
depends upon the surrounding conditions of the object. Even
I98 J. R. KANTOR
if the knowledge elicited from the object itself is favorable to
its consumption, that event will not occur unless conditions
are otherwise favorable. We mean to point out here that the
specific kind of response patterns that will act as a series in
any given situation will depend upon that situation. This
fact indicates the close interaction between stimuli and re-
sponses.
An important reservation to the above description of the
perceptual activities must be made in the light of our distinc-
tion between primary perception and simple apprehension.
It is only in the case of simple apprehension that the distinct
series of factors are found; for it is only there that a definite
meaning factor is isolated in the total act. In primary per-
ception the overt act is identical with the original system, and
the perceptual process itself constitutes not exclusively a
definite knowledge factor in an adjustment, but it is the whole
adjustment itself.
Perception in Development and Use. — Of primary importance
for the understanding of the perceptual reactions is the dis-
tinction between perception in development, and in use. In
the former type of reaction with objects meanings are de-
veloped; that is to say, a definite form of reaction pattern is
acquired; so that the future contact with this object will be
of a definite and peculiar sort, because the reaction pattern
developed will then be put into use. The distinction made
indicates the extremely complex and constantly varying char-
acter of the perceptual reactions and points to the mechanism
of elaboration of such functions.
Since clearly the original perceptual contacts with objects
occur in the instinct stage of development, we may date the
origin of a meaning or reaction pattern from the first instinct
contact of an organism with any given object. The point is,
that the hypothetical, original contact of an organism with
an object is the result of a direct arousal of a connate reaction
pattern through the instrumentality of various physical media
such as light rays or air waves. If we dare speak of a meaning
possessed by an object at this stage, it is merely that of 're-
sponse eliciter.' This contact is as mechanical as a conscious
SCIENTIFIC INTERPRETATION OF PERCEPTION »99
behavior act can be, and here we find the full significance of the
statement that we have innate tendencies to discriminate colors
and other physical qualities. The fact is that our connate
reaction patterns are brought into function by the stimulation
of the specific receptor systems whose activities form a part
of them. At this stage the simple psychophysiological response
as a whole, symbolizes the meaning of the object. Now,
when the action just mentioned occurs, some effect will be
produced upon the organism; so that the next contact with
this object will involve a modified reaction system, or we might
say, the object has taken on a new meaning. The perceptual
processes thus represent a constant integration of a reaction
pattern depending upon the number of contacts with the same
physical object under varying conditions of surrounding aus-
pices. In general, it is clear that the perceptual reactions are
entirely genetic in their functioning, hence only by studying
them in their development can we hope to understand them.
Another form of integration in the development of per-
ceptual reactions is the establishment of a definite interactional
relationship between the stimulating object and the reaction
system. Not only must there be a coordination of specific
factors of a response system, such as for example, visuo-
muscular, visuo-glandular and neuro-muscular processes, but
there must also be a connection between this total reaction
pattern as a functional representative of the organism at the
time and the stimulating object.1 Just how this intimate
relationship between stimuli and response systems, which is
the essential factor in perception in use, is established, can be
experimentally studied through various types of conditioned
reflexes. Excellently is perception in use illustrated by the
story of the discharged veteran, quoted by Spencer, who had
had the auditory object * attention' so integrated with a par-
ticular response system as to lose his pie when a practical
joker uttered the command.2 When the integration has been
accomplished, the reaction pattern can be stimulated by one or
more of a large series of phases of the object, which become
1 This connection between the response pattern and its simulating object consti-
tutes the primary and fundamental type of psychological association.
''Psychology,' i, p. 499.
200 /. R. KANTOR
differentiated because of the different media through which the
contacts between the organism and the objects are made.
Thus, a reaction pattern involving a ball-meaning may be put
into action by either a visual, auditory or tactual stimulus.
As an illustration of the arousal of a complex system of per-
ceptual responses through the mediation of a simple type of
stimulus, we can take the case of the visual contact with ice,
which arouses coldness, smoothness and hardness meanings
at the same time. The effective adaptation of the organism
depends to a considerable degree upon the complexity of the
two sorts of integration described.
Because perception in use as just described involves putting
a complex reaction pattern into operation by .some phase of an
object, we find in such adjustments the beginnings of a dif-
ferentiation between the explicit and the implicit functioning
of a reaction system; the latter case gives us the detached
meaning. The implicit functioning of a reaction pattern is
clearly discerned in the many cases in which the visual contact
is the only direct one; and the meaning of the object, which
may be very elaborate, though not attached directly to an
immediate response, is most certainly acting. A striking ex-
ample of the implicit functioning of reaction patterns is the
situation in which a banker, while otherwise preoccupied, for
a moment will begin to respond as though at a director's
meeting, when stimulated by the crumpling of a crisp paper.
Again, the 'wave of feeling' brought on by the perusal of a
literary description indicates the living over of some crucial
situation by the incipient operation of reaction systems. It is
this implicit functioning of reaction patterns in perception
which shows the way toward the development of the con-
ceptual and memorial processes.1
From our description of perception in development and in
use it must appear that these are not two distinct operations,
but rather two mutually interrelated processes. Since the
perceptual activities are constantly developing we have in
practically every new operation of a perceptual reaction system
1 As a matter of fact, so far as psychophysiological mechanism goes, there is oniy
a difference in degree between perception and thought, but from the standpoint of
results effected through these reactions the variation is or may be enormous.
SCIENTIFIC INTERPRETATION OF PERCEPTION 201
a more complex integration of the component action elements
with the stimulating situation. If we consider the perceptual
reaction as the use of meanings stimulated by direct contact
with objects, we find that the distinction between the develop-
ment of perception and its use, depends upon the amount of
direct stimulation which is required to elicit the response.
Perception in development requires a relatively larger series
of direct contacts to effect an equally complex response than
is true in the case of perception in use, since in the latter case
the meaning attaches to an incipient reaction pattern. We
repeat, the development of perception is a process of so in-
tegrating acts that only a minimum of receptors may be
necessary to effect the appropriate response. If we remember
that this development never ceases, provided that we have
occasion to react to the given stimulating object, then it is
clear that perception in use is merely the condition of respond-
ing on the basis of a previously acquired reaction system,
pending its modification by the present contact with the object
in question.
The Specific Mechanisms of Perception. — A more penetrating
analysis of perception than we have yet made will yield informa-
tion as to the specific integrations which operate in the per-
ceptual reactions. To a certain point we can trace the precise
organization of the component processes, such as the muscular,
cognitive, glandular, neural, etc. Our ability to do this is
made possible by the fact that underlying all these modifica-
tions is a simple psychological law which may be formulated
as follows. Every integrative modification of a reaction pattern
is a direct function of a differential contact with actual things.
By far the most important problem of perception arises
just here, namely, what are the specific means of contact be-
tween the organism and objects? The interest in this problem
emerges because of the inevitable incorrect inference from the
customary psychological premises, namely, that the cognitive
qualities are existential processes somehow aroused in 'con-
sciousness'which bring about the movements of the organism.
Now as a matter of fact, it is easily seen that in any description
of perception the qualities mentioned (odors, colors, etc.) are
202 /. R. KANTOR
abstracted from the objects. The discrimination of these
qualities as it occurs in the actual response is in part the per-
ceptual act; that is to say, the discriminative process con-
stitutes part of the perceptual act as distinguished from the
overt action which follows it. The discriminative factors are
thus seen to be phases of concrete psychophysiological pro-
cesses, and this means in effect the total extrusion from the
perceptual act of any substantial mental or subjectivistic
quality.
Responsible for the view of the existence and primary
functioning of sensation qualities, is the psychological tradition
which makes knowledge the differentia between biological and
psychological acts. Taking conscious behavior as our starting
point, we may catch a glimpse of the true significance of the
perceptual reaction as a knowledge process which brings about
adequate, psychological adaptations, and still keep our descrip-
tive analysis of the facts within the range of observational
interpretation. The favorable prognosis for the scientific de-
velopment of psychology depends in large measure upon the
rejection of a theory implying that the adjustmental responses
of the individual are due to a mystic potency resident in
'consciousness.' In place of such a theory should be sub-
stituted a verifiable interactional mechanics of natural things.
Upon the basis of such an interactional mechanics it is possible
to avoid the assumption that perceptual responses are pri-
marily cognitive operations or that they are 'consciousness,'
that is to say, awareness of something, rather than adjustment
acts.
Thus, the problem of the contact of the individual with
objects is reduced to the description of the precise manner in
which a reaction pattern or system is put into operation by
the stimulating object. Here we have to assume that the
reaction is that of a conscious organism, which has the capacity
to react to colors and other qualities. As a matter of fact, the
notions we have of such qualities are historically developed
through the discriminating evaluations of such conscious beings.
Now, although it may be impossible to develop a detailed
analysis of all that takes place in a perceptual reaction, we can
SCIENTIFIC INTERPRETATION OF PERCEPTION *<>3
isolate series of systems which play their part in such reactions.
These systems are logically ordered sequences of events which
occur when a perceptual reaction is made. An example of one
of these systems is the cycle beginning with the reflection of
light rays of definite sorts which set up differential processes
in the retina, followed by definite happenings traced out in
the neural pathways and in the cortical areas of the brain.
The completion of the cycle involves the consideration of
changes taking place in the association tracts and the motor
localities of the cortex, the happenings in the efferent trans-
mission system and in the effectors located in muscles and
glands.1
Of extremely great consequence is the series of appreciative
and feeling processes which are factors in the operation of the
total reaction system under discussion. The important point
here is that the perceptual reaction must be looked upon as
one of the ways in which a psychophysiological machine is
operating. Above all, what we wish to avoid is the conception
that the physico-neural functions constituting part of the
perceptual act, are the causes or the parallels of conscious
action. A very simple means to avoid this confusion is to
remember that we are dealing here with two phases of a
natural happening which for scientific purposes are differently
classified, but never separated, and also that no process is
any more tangible than another. Physical processes are not
tangible physical substances, nor are the physiological factors
biological material; neither are both of these functions abso-
lutely distinct from the mental processes which naturally do
not reduce themselves to mentality, a substance the existence
of which we all join in denying. What we must describe here
is a psychophysiological reaction, for it is only such a reaction
which can be the object of our observations. While observing
1 The reader who is interested in a more detailed discussion of the mechanisms,
of conscious behavior is referred to Watson's recent volume, 'Psychology from the
Standpoint of a Behaviorist,' which contains the best description in psychological
literature of the behavioristic components of a reaction system. Because of the
author's resolute attempt to suppress the mentalistic components of the reaction
pattern, the book contains merely suggestions, though frequently very important
ones (especially in the chapter on Emotions), concerning those phases of a conscious
response.
204 /• R- KANTOR
a psychophysiological organism we can discriminate between
acts involving a response pattern of predominantly mental
factors and others having the physiological factors more promi-
nent. It is the former type of psychophysiological act which
is usually called subjective, and which is in part responsible
for the inexcusable separation of the mentalistic and behavior-
istic phases of a unitary act.
Since we can analyze many of the isolated factors of a per-
ceptual reaction we can describe specific correlations between
the qualities of objects and the particular phases of the reaction
pattern. Thus colors, sounds, tastes, hardnesses, etc., can be
coordinated with specific receptor systems, because during the
evolutionary development of the organism the receptor systems
became differentiated in sensitivity to particular kinds of
stimuli, which objects initiated. For example, the retinae are
normally sensitive only to light rays reflected by the colored
surfaces of objects, and the cochlea to air vibrations, which
emanate from sounding bodies. In passing, we might point
out that our analysis has provided no basis for the assumption
that 'objects as perceived' are synthesized in some form out
of qualities produced in the mind or in the organism by stimuli
set up by objects. After many detours this view just men-
tioned has seeped into current psychology from the Berkeleyan
head waters, and for a long time has been effective in preventing
the conception of psychological phenomena in a scientific way.
In contrast to the Berkeleyan view, we must look upon the
stimuli which constitute the middle link between objects
and organisms as natural predisposing conditions, mediating
changes in the activities of the latter, much after the fashion
in which an electric current produces changes in a machine.
The undesirable consequences of thinking that in perception
there is a synthesis of objects is well illustrated by the concep-
tion that space and time are somehow compounded by some
additional attribute of the mental 'contents' called sensa-
tions.
The Relational Character of Perception. — Observations upon
the perceptual interaction with things convince us that not
only are all perceptual reactions not merely responses to specific
SCIENTIFIC INTERPRETATION OF PERCEPTION 205
qualities, but also that they are not confined to isolated objects;
they are more than either of these descriptions indicate, namely,
responses to a complete object in all its setting. We might
generalize this fact by saying that we always perceive situa-
tions, not isolated things, and of course our conduct is condi-
tioned accordingly. Thus a chair which ordinarily would be
responded to by being sat in, will not call out such a response
when it is occupied by some object or when there are indi-
viduals present before whom it is impolite to make such a
response. In every such case the meaning of the object will
depend upon the contextually related objects. When the chair
itself is reacted to, we respond to a unified object, and not to
simple elements (back, seat, legs, etc.); that is to say, we
react to an object to-be-sat-in, and not to isolated fragments
which require to be somehow connected. This relational char-
acter of perception is excellently illustrated by our responses
in which words and not letters are the stimulating objects, and
in which the words are directly and inseparably attached to
other words.1
That we can immediately appreciate a complex situation
apparently comprising many diverse elements is owing to this
relational character of perception. Thus, in looking at a land-
scape the objects all seem to be in their proper places; distances
are correctly located and the lights and shadows properly
distributed. The total situation is the customary object of our
reactions and is thus the stimulus to a unified primary response
or simple apprehension. The meaning of the total situation
can be readily and completely confounded by placing ourselves
in a position incapacitating us for response, such as looking at
the landscape with our head upside down. Much the same
effect is produced by looking at an inverted painting. In such
situations what happens is that the series of integrated reaction
systems are thrown out of their customary harmonic organiza-
tion and must be reorganized before the object can be correctly
perceived. Experiments on space perception have shown that
1 James points out with his characteristic description the unnatural aspect which
a word takes when looked at in protracted isolation. "It stares at him from the
paper like a glass eye with no speculation in it. Its body is indeed there, but its soul
is fled." 'Princ. of Psychol.,' Vol. II., p. 81.
206 /. R. KANTOR
by practice disorganized response systems can readily be
reintegrated.1
The difference in the responses to objects when they appear
in different contextual relations illustrates the extremely subtle
interaction between the stimulating object and the reacting
person, and also shows the operation of perception as an ad just-
mental reaction to surrounding objects. Pliableness of the
individual in this sense constitutes an important factor of
general intelligence and exemplifies the law of integrated modi-
fications of reaction systems mentioned above.
The Interpretative Function of Perception. — Since every psy-
chological phenomenon is a product of two factors, namely,
the stimulus and the response, our discussion of the influence
of the stimulating circumstances upon our perceptual reactions
naturally leads to the consideration of the influence of the
individual's stock of reaction patterns upon any given present
reaction. An observable fact it is, that the reaction systems
which the individual has developed in his constant contact
' *-*
with objects, play a large part in any present reaction; for in
a genuine way such reaction patterns constitute the individual
at the moment. And since as we have indicated, these response
patterns have been developed in the individual's previous
experience, every perceptual reaction may be thought of as
an interpretative function. In effect, this means that the
person will respond to objects much as he has been accustomed
to do under previous conditions of contact with similar objects.
It is this fact which gives origin to the idea that perception is
a kind of habit.2 Being equipped with a response system to
react to stimulating objects, is fundamental as a condition of
every recognition behavior. The element of novelty conies
into a response situation precisely at the point at which the
person is unable to offer a complete response to the present
stimulating object. Since the meaning of the object is not
fully comprehended the person can respond in a way which is
only a partial reproduction of a previous form of response.
The lack of complete recognition means that the person is not
1 Cf. Stratton, PSYCHOL. REV., 1897, 4, 341-360, 463-481.
2 C/. Angell, 'Psychology/ 1910, p. 157.
SCIENTIFIC INTERPRETATION OF PERCEPTION
supplied with a reaction system to respond immediately to
the object in question. In such a case the pressing need for
a response to the object results in an incipient trial and error
process ending in a clear-cut appreciation of its meaning and
a consequent thinking reaction.
The interpretative function of the perceptual reactions is
observable in many instances of daily occurrence. In the
case of reading and speaking we find that there is very little
stimulating material, but the response is not at all interfered
with. In listening to a familiar voice, or familiar written
material, we can easily demonstrate to ourselves that our
response patterns are aroused by no considerable amount of
excitation. No doubt the explanation for this lies in the
individual's possession of dispositions organized for particular
forms of situations and any prominent feature of those situa-
tions will set off the reaction patterns. It is here that we
find the bases for the incorrect or unexpected responses com-
monly called illusions. For the same reason a person^with a
limited experience will be ready upon fewer occasions to respond
to objects, and on those occasions will be slower to make the
reaction. It has been aptly said,1 that 'the artist sees details
while to other eyes there is a vague and confused mass; the
naturalist sees an animal where the ordinary eye sees only a
form/ That the child reacts to objects in monotonously
similar ways, is true because it has been impossible for him
to build up many reaction systems. And so the significance
of the pony and what can be done with it are the same as
in the case of the dog, with only a variation ;n size. The
classical illustration of the observable facts in this case is
found in the name response (big dog) which the child makes
to the pony.
The Elaboration of the Perceptual Functions. — The constant
development of the perceptual response serves as one of the
individual's important means for a growing mastery over his
environment. As the reactions to an object multiply, that is,
as the number of responses which it calls out increases, the
object takes on more and more meaning. It is owing to this
1 Lewes, 'Problems of Life and Mind,' 3d series, p. 107.
208 y. R. KANTOR
increasing elaboration of the perceptual response systems
through the addition of meaning factors that the organism is
enabled to make its way with greater facility through the
maze of its surrounding objects. This facility is further in-
creased by the fact that this elaboration of the perceptual
response systems makes it possible for the person to adapt
himself to many situations without invoking a definite problem
of adaptation. Because of the absence of such a specific
problem, and the consequent exclusion of a thought function,
the simple form of the perceptual reaction allows for an im-
mediate response to objects.
As a hypothetical illustration of the growth of the perceptual
responses we might consider the reactions of a child to a type-
writer. Allowing for a definite development already attained,
the machine may be at first merely a thing which can produce
a series of sounds when the keys are pressed. The machine,
then, as soon as it is seen, has merely the sound-making
meaning in the immediate response. With a more extended
acquaintance with the machine the child learns that it can
stimulate different and additional responses, and it thus has a
different meaning when perceived. Finally, the machine takes
on the complete set of meanings which are derived from all
the responses the child can make to it. The point is, that
what sort of perceptual reaction an object will call out at
any time, or what it will mean, will depend upon the sum total
of the person's contacts with the object in question. The
perceptions of persons grow continually, and the growth de-
pends upon the addition of new features to the response
patterns and of completely new patterns of response.
The development of perceptions by the growth of responses
is well illustrated when we are at the point of substituting an
object for another in the face of an immediate need. Thus, a
chair becomes a barricade, or step ladder, or typewriter table.
As a consequence of the person's being forced to make new
and unusual responses to objects, the latter become endowed
with a range of new meanings. In the above illustration we
also observe the active relating function as it occurs on the
perceptual level. The similarity between objects is of course
SCIENTIFIC INTERPRETATION OF PERCEPTION 209
a fundamental causative factor in the perception building
activity, since otherwise the possible reactions to objects would
be at such variance as not to admit of any correlation.
II
If the brief description of the perceptual reaction which we
have essayed is correct, it obviates some of the most salient
errors in current discussions of perception, and places the
interpretation of such processes upon a definite natural science
level. Let us first observe, then, that the perceptual reaction
is always a reaction and not a thing, namely a complex organiza-
tion of subjective qualities. Moreover, a perceptual reaction
is a psychophysiological reaction as all data of psychology are.
That is to say, the perceptual act is not in any sense the act
of an ego, or mind of whatever description, nor of a nervous
system, but a complex reaction system involving all the func-
tions of a conscious being. Notice that the vexing problem
of a self, vexing, that is, once it is allowed, plays no part
whatever in the interpretation we have made above. For the
sum of the reaction systems which adjust the individual con-
stitute the person, and since each person because of his par-
ticular interaction with things and persons, has developed
definite types of reaction patterns, the problem of character
or personality is thereby solved. Since psychology is interested
only in such reaction systems there is naturally a perfect co-
ordination between psychology and the other sciences of the
individual, such as anatomy, for example, which is interested
in the structure whose functions form part of the reaction
pattern.
That we cannot assume that in the perceptual act we have
besides an object stimulating the organism, and the organism
(frequently taken to be merely the nervous system) also an
object of perception, that is, a sum of mental qualities, we
indicate by the statement that there are only two interacting
things in a perceptual as in any psychological act, namely,
the organism and the physical object. The fact is that the
physical object contains all qualities, colors, sounds, tastes,
hardnesses which we can ever analyze out of it, and the organ-
310 /. R. KANTOR
ism learns to distinguish these and to name them because of
specific psychophysiological effects which contact with objects
brings about in organisms.
Our view may be illustrated by the following example.
When we perceive a blue object, in no sense is there started
up a * consciousness' of blue by an antecedent or accompanying
neural activity. As the matter is stated by practically all
psychologists there comes to be at this point a blue conscious-
ness or a blue sense quality. Now we maintain that the only
blue involved is a blue object, independent of a perceiver and
in no wise modified by the specific perceptual act; any change
in the object must be effected by the overt action following the
perception. What occurs in the above illustration of a per-
ceptual act, is that the light rays set into function a complex
reaction system which involves the specific meaning of this
object, in the sense that the immediate effect produced by the
object on the person may now result in a specific act, perhaps
in the exclamation, 'I see a blue flower.' The effect upon the
person, we repeat, is muscular, neural, glandular, cognitional
and perhaps affectional. Let us remember that at this stage
we must consider the activity as perception in use, which has
developed through a series of previous contacts with the object;
for otherwise many kinds of direct contacts besides those
mediated by light rays would be necessary, in order to arouse
so definite a meaning of the object as to be followed by a
definite act.
Clearly, the specific perceptual act is an abstraction from
an empirical interaction of a person and an object; that is to
say, the perception proper is abstracted from the preceding and
following acts of the person, while the object is abstracted
from its setting which includes many other objects and persons.
The description of a perceptual act is always a deliberate
rationalization of a complex event, a fact which is at least
implicitly recognized by all psychologists; even those who
despite their Berkeleyan adherence to mental states agree
that the perception of blue is an abstraction from a blue object
(of perception).1 This abstracting process can be made out
1 The writer refers here to the statements by psychologists that sensations are
always abstractions.
SCIENTIFIC INTERPRETATION OF PERCEPTION an
most clearly perhaps by a thoroughgoing analysis of the
development of the naming reaction performed in denoting
things.
While it is almost impossible to describe so intricate an
organic activity as a perceptual act, and at the same time
avoid completely falling into a logical instead of a psychological
analysis, it is still possible so to guard one's description as to
prevent an essential misconception of the behavior. But un-
fortunately such misconstruction is rarely guarded against, for
most analyses of perception merely amount to the isolation of
the qualities of an object and the transformation of them into
sensations, which in their totality are presumed to constitute
the known object as over against the material object, which
by hypothesis must remain forever beyond the pale of mental
things.1
At the basis of the current, primarily logical analyses of
psychological phenomena which may be taken as symptomatic
of the unscientific character of such description, lies the pre-
judice deep rooted, that psychology is the study of mental
states, a kind of stuff (masked by the veil of process)2 which is
different from physical material. In no matter what form this
subjectivistic view is presented it must be looked upon as a
vestige of religious thought in psychological dress. Today, it
must be rigidly extruded from scientific thinking, since it is a
prejudgment of facts to be observed. On the contrary, genuine
scientific thinking must start with observable phenomena, and
naturally enough when we start in this way, we never meet
with mentality or physicality as the psychologist deals with
them.
The immediate development from this false dualism is that
the domain of psychology is that of knowing, for consciousness
is thinking stuff.3 Now thinking or knowing is assumed to be
1 Sometimes the percept is considered as distinct from sense qualities as in the
statement by Stout, "The general possibility of the transition from sense-impression
to percept depends upon the existence of the percept as something distinct from the
sense-presentation to which it seems as a rallying point and center of connection,"
'Analytic Psychology,' Vol. II., p. 31.
1 In spite of the veil no other interpretation is possible of a thing which has
attributes.
1 The trend of modern thought as influenced by Descartes.
212 /. R. KANTOR
the most intangible and inaccessible stuff or process, and thus
has arisen the esoteric psychology of introspection. Clearly
there can be no science which has as its subject matter in-
tangible and invisible subjectivistic states, and for this reason
the history of psychology mirrors much groping about for some
concrete material with which to work. Finally, psychologists
seized upon the nervous system as a tangible basis for the
intangible consciousness. In our own day the behavioristic
movement at least in one of its phases assumes that it is really
the nervous system with which psychology has to deal and
not at all with consciousness. This behavioristic view, though
clearly mistaken in that it is still based upon a dualism one
phase of which is rejected, should be credited with much
scientific acumen, since it must be taken as a protest against the
obviously unscientific character of a mentalistic psychology.
For no science can be built upon things or processes which are
not observable.1 When we consider a perceptual act as an
adaptational response to some natural object, we find no
necessity for the dual interpretation of psychological phe-
nomena such as leads to the problem how a mental state can
be made to know or refer to an external object. For the
functional psychologist there can be no such problem; what
he is concerned with is the way a definite sort of interaction
takes place between two natural things, a person and some
other object which may or may not be a person. Thus, the
functionalist does not create for himself the question as to how
a conscious state can be initiated by a previously or simul-
taneously occurring neural process.
Berkeleyan and Reidean influences in psychology are main-
tained by the confusion of the products of logical analysis and
the concrete facts of conscious behavior. Thus, the relational
and interpretative character of a psychophysiological reaction
1 It is not to the point to argue as Stout ('Manual,' 1915, p. 18) does that 'mental
dispositions' must be assumed to exist in the way that 'mass' and 'energy' exist,
though not directly observable, for as he himself points out, physical things and
processes are inferred from directly observable phenomena. This can not be said for
his mental dispositions, which are not descriptive of actual facts. And furthermore,
mass and energy are obviously useful categories of physical science, but mental dis-
positions are only necessary because of an erroneous subjectivistic interpretation of
human behavior.
SCIENTIFIC INTERPRETATION OF PERCEPTION
is assumed to be the growth of a mental state which is called
a perceptual object. From this standpoint, animals and pos-
sibly infants are presumably supposed to have no perception
because they cannot possibly have the knowledge which a
human adult has. In detail, a perception is assumed to be
a complex organization of sensation qualities with meanings
attached. Thus meaning is further assumed to be the definite
self-conscious interpretation of the sensation qualities, clearly
an epistemological view. In contrast to the above view, we
have already suggested that what occurs in nature is the build-
ing up of a reaction-system, which at first is simple, that is,
the object has little meaning, and later further contact with
the object complicates and expands the reaction system, which
fact is interpreted as giving more meaning to the object by the
increased number of possible reactions which can now be
made to it.
The consequences of the view that perception is a mental
structure, are clearly brought out in the issue raised by James1
concerning the confusion of the object perceived with sensations
or perceptions of the object. James saw clearly the fallacy
of Stumpf's analysis of the sensation of oil of peppermint into
the sensations of taste and temperature; for James would
have it 'that we perceive that objective fact, known to us as
the peppermint taste to contain those other objective facts
known as aromatic or sapid quality, and coldness respectively.'
We cannot sympathize with Stout's2 fear that the view of
James involves 'the impossibility not merely of the "analysis of
presentation " but of all analysis properly so called,' although
we do agree with him that the psychologist's interest is in the
psychological and not particularly in the physical object. We
cannot agree with Stout, however, against James, (i) because
for the former the psychological process must be purely mental,
and (2) because he assumes that a perception must be a com-
pound of sensations such as can be analyzed. The issue
between a functional and a structural view is definitely brought
out here. Stout thinks that because he can remember that
oil of peppermint has certain definite qualities of taste and
1 'Principles,' I., p. 521 ff.
1 'Analytic Psychology,' 1896, p. 56 ff.
214 /. R. KANTOR
temperature he has analyzed a purely mental thing. Now as
a matter of fact the memorial behavior is primarily the implicit
functioning of a reaction system developed in direct contact
with an object and is therefore most certainly a psychophysio-
logical action. It is precisely because Stout does not see that
a reaction system, that is to say the system built up in per-
ceptual contact with objects, can be put into function by a
substitution stimulus that he means to perpetuate the men-
talistic tradition in psychology. If we assume that what is
studied in psychology is the development of the complex
reaction patterns and the means whereby they are put into
complete or incipient function by various types of stimuli,
we need never invoke any mysterious or inscrutable entities.
The literature on space perception clearly demonstrates the
hopelessness, from a scientific standpoint, of the mentalistic
doctrines. For, the problem of space in mentalistic psychology
is the problem of building up or constructing space instead of
the observation of the specific means whereby a person per-
forms space reactions and adapts himself to objects variously
placed. When space reactions and not geometry is made the
subject matter of the psychology of space the problem of the
genetic or a priori character of space drops out of sight. In
the observation and interpretation of space reactions there
can be no question of innateness or acquisition of knowledge
of space, for a space reaction is not essentially knowledge as
we have indicated in our description of the perceptual reaction.
There is no doubt, however, that our knowledge of space is
derived more or less directly from the space reactions, but this
is a problem of logic and not of psychology. The study of the
literature on space perception shows us clearly how the psy-
chologists persist in forcing into their science epistemological
problems which should have no place therein. Curiously
enough the epistemological view gains impetus from an at-
tempt to give psychology a scientific setting as is familiarly
illustrated by the influence of Helmholtz's ideas of mathe-
matical space upon the development of the psychology of
space perception.
The ascription to current psychologists of a subjectivistic
SCIENTIFIC INTERPRETATION OF PERCEPTION
heritage from Berkeley and Reid may call for some explanation.
The statement that we are still working and thinking in the
Berkeleyan tradition does not exclude the fact that current
introspectionism was established and elaborated by the work
of the German physiologists. It is of course a matter of
common knowledge that the introspective view was made
possible and plausible by the physiological experiment which,
dealing with isolated physiological functions, had to assume a
correlated mental state to complete the description of the
reaction observed. It is thus that the work of the German
writers from Herbart through Fechner to Wundt, although
designed to place psychology upon a sound scientific basis
has in reality, because of its maintenance of the subjectivistic
tradition, accomplished the opposite.
The proposed interpretation of the perceptual reactions
suggests the extrusion of the separation doctrine from psy-
chology and thus makes toward the removal of what is probably
the greatest hindrance to the thorough establishment of psy-
chology as a science. For, as long as psychology deals with
conscious or mental states of any sort whatsoever it cannot
ever attain to the dignity of a science as Kant long ago asserted.
This statement holds whether consciousness is taken as an
attribute of the psyche or mind, or of the states of consciousness
and unconsciousness which are presumed to be the mind.
In conclusion, we might point out that although the organic
conception of psychological phenomena appears to some psy-
chologists as widely accepted,1 the manifest predominance of
the mentalistic and behavioristic views would seem to indicate
the contrary. The apparent prevalence of the organic con-
ception may be accounted for on the principle that insofar as
a psychologist is to describe some actual psychological fact,
the description must in some fashion correspond to the fact,
regardless of the private view of the writer. Thus, much of
current practice may be organic, but the question remains
whether psychology can make much progress toward scientific
1 Cf. Carr, PSYCHOL. REV., 1917, 24, 182. "The conception is unorthodox only
in relation to prevailing definitions of psychology. To my mind it is essentially in
harmony with the dominant point of view of the science, and it is not wholly incon-
sistent with much of current practice."
j. R. KANTOR
stability if psychologists do not fully appreciate the character
of the materials with which they deal. While it is certainly
true that definitions may linger far behind practice, the scien-
tific practice in which this occurs, lacks much in desirable
effectiveness. Even if scientists were forced to recognize all
the component functions of a reaction, they might still be
lacking in a full appreciation of the organic interpretation of
such a reaction. That there is little genuine interest in the
psychophysiological view among psychologists is evidenced
by the general paucity of articles written from that standpoint.1
Undoubtedly true it is, that the biological influence in psy-
chology has fostered the unitary conception of organisms, but
it has not resulted in any complete modification of viewpoint.
In fact, the rise of the behavioristic movement urges the belief
that there is no general tendency to look upon psychological
phenomena as they naturally function but as they are tradi-
tionally supposed to operate. A sympathetic acceptance of
the objective functional view must result in the description of
the complete actual psychophysiological reaction pattern, and
the consequent rejection of the exclusively mental or physical
interpretation.
xTwo notable exceptions must be here referred to, namely, Carr, 'The Relation
between Emotion and its Expression,' PSYCHOL. REV., 1917, 24, 369, and Peterson,
'The Functioning of Ideas in Social Groups,' PSYCHOL. REV., 1918, 25, 214.
INSTINCT AND PURPOSE
BY EDWARD CHACE TOLMAN
University of California
This paper will be roughly divided into two parts. The
first part will present a definition of instinct. The second
part will use this definition in a psychological analysis of
purpose. The discussion will be throughout from an objective,
external standpoint, that is, the interest will be in how purpose
works rather than in how it feels.
By way of introduction let us liken the human being to a
slot-machine. The pennies will represent the stimuli, i.e., the
sights, sounds, printed symbols and the like which we may
apply to the machine, and the resulting pieces of candy the
words, action, and movements which issue forth. If the penny
be a word of praise, the answering candy may be a blush or
sparkle of the eye. If the coin we apply be an insult or a blow,
the resulting packet will probably contain vituperation. If
the penny be the word 'white 'spoken suddenly and in no par-
ticular connection, the answering phonograph sound will in all
probability be the word 'black.' In every case, if we but knew
the mechanism well enough, we could predict a particular
action as the result of a particular stimulus.
But let us see in what ways this picture of the simple
slot-machine is inadequate. We may note that sometimes
when a stimulus is presented to the human machine, nothing
externally observable issues forth; or, again, that something
quite different from a previous response to the very same
stimulus comes out. These facts of the uncertainty and
changeableness of response in the human machine, though
one and the same stimulus be presented, require the assumption
of two principles not contained in the simple machine so far
described. The first of these principles is that the nature of
the response on any given occasion, or whether in fact there
is any overt response at all, is dependent upon the general
217
21 8 EDWARD CHACE TOLMAN
internal adjustment of the organism at the moment. To make
a slot-machine adequate to such a situation we would have to
imagine a complex machine capable of various different adjust-
ments such that, when one adjustment was in force, the
succeeding pennies produced musical sounds, when another
was present, the same pennies introduced into the same slots
produced different kinds of candy, and so on for each different
adjustment. Finally, we would have to assume in addition
that some of these internal adjustments might act like partial
locking devices such that, when they were in force, no response
at all would be produced from some single penny or for some
particular group of pennies.
These internal adjustments would sometimes arise primarily
as the result of just preceding external stimuli and sometimes
as due almost wholly to automatic changes within the organism
itself. If a man refuses food (i.e., if the eating response is
locked), it may be because of a preceding stimulus such for
example as a slap in his face which has aroused the internal
adjustment which we call anger (which locks the eating re-
sponses) ; or it may be because of some automatic physiological
condition (e.g., lack of hunger) which, though not positively
locking, at least does nothing to unlock the eating responses.
If a man responds to one and the same book on one day with
tears and on another with laughter, the change in internal
adjustment bringing about the change in response may be due
either to a specific preceding stimulus or to some mere physio-
logical rhythm.
The second principle which the simple slot-machine lacks
and which it should possess, if it is to adequately represent the
human organism, concerns the changeableness of response in
the human organism which is due to learning. Everywhere
we find that, simply as a result of previous experience, the
organism exhibits new responses to stimuli. When a child
learns to play the piano it acquires a series of finger movements
in response to black marks on paper to which, before, it did
not react at all. When one learns to read and write, to play
tennis, to open and close doors, to lace and unlace shoes;
when, in short, one learns any of the million and one things
INSTINCT AND PURPOSE 219
which one does learn, one is merely attaching responses to
stimuli which did not originally call them out.1
No easy way of representing such alteration in the case of
the slot-machines, however, suggests itself. We will, therefore,
have to think of the latter as illustrating but single stages in
the human organism and imagine a new and improved machine
as a result of each acquisition of new traits and habits.
To sum up: the human being is a mechanism which makes
responses to external stimuli. The nature of these responses
and whether there is any overt response at all, however, is a
variable matter. This changeability depends, first, upon the
possibility of different internal adjustments (either called out
by specific external stimuli or as the result of internal physio-
logical rhythms), and second, upon the changes in the internal
structure of the organism due to learning.2
Our task must now be a more specific classification and
description of such responses and internal adjustments. Dif-
ferent classification would no doubt be possible, but for our
interest, which is concerned primarily with a definition of
instinct, the necessary classification is simple. It contains but
three groups: (i) independent reflexes, (2) subordinate acts,
and (3) determining adjustments.
By an independent reflex we shall mean any response to a
stimulus which takes place always in the same manner and
relatively independently of what the rest of the organism is
doing. The kick of the foot in response to a tap on the knee,
winking in response to a movement before the eyes, sneezing
in response to tickling the nose, yawning in response to certain
internal sensations, are examples. These always occur in
much the same way and each is relatively complete in itself
and independent of what the rest of the organism may be doing.
Activities on the other hand, such as biting, chewing,
1 And this holds, be it noted, not only for actual overt responses, but also for the
internal adjustments we have just discussed. Thus, for example, the internal adjust-
ments originally appropriate to and only aroused by such things as loud sudden noises
and really startling objects may as a result of training get attached to a whole series
of secondary associated objects such as the dark, strange faces, etc.
1 Also upon changes resulting in the course of natural growth, e.g., the appearance
of new sets of instincts as the child matures.
220 EDWARD GRACE TOLMAN
swallowing, which form part of larger wholes — in this example,
eating — would be classed in the second group: i.e., termed
subordinate acts. The members of this group are almost
infinitely numerous. The leg movements of walking, the hand-
lings of curiosity, the cries and shouts and strugglings of anger,
the sighing and tears of sorrow, the facial expressions, words
and gesticulations of love, would all be examples. In fact
all the things we do, not as separate and independent reflexes,
but as parts of bigger groups of activity, belong to this second
class.
Finally we have as our third group what we called deter-
mining adjustments. These are, in fact, to be considered as
identical with the internal adjustments just described in our
picture of the slot-machine. They determine and set in readi-
ness the subordinate acts. Whether one responds to one and
the same stimulus with the subordinate acts of handling and
manipulation, those of destruction, or those of rejection, de-
pends upon which particular determining adjustment has first
been aroused — whether one of curiosity, one of anger, or one
of fear.
It is to be noted that determining adjustments often occur
in hierarchies. What may be called the lowest one of the
hierarchy is then the immediate determining adjustment for
the actual subordinate acts. The next higher one of the
hierarchy releases this lowest one. A still higher one releases
that, and so on. For example, we may suppose that on a
given occasion an individual's leg and foot movements are
directly subordinate to what may be called the walking adjust-
ment. This walking adjustment, however, we may assume is
subordinate to an anger adjustment. (The man may be on
his way to confront a business opponent.) This anger adjust-
ment will then be subordinate to a business adjustment and,
finally, this business adjustment itself may be assumed to be
subordinate to what may be called the man's general socio-
domestic adjustment. In the case of such a hierarchy of
adjustments it is obvious that the function of all, save the
lowest one in the sequence, consists in a release of a lower
determining adjustment rather than in a release of actual
subordinate acts.
INSTINCT AND PURPOSE 221
One further point. In the case of activities such as eating,
running, walking, is it legitimate to talk of a determining
adjustment as something existing in addition to the individual
walking or eating movements themselves? The reason I as-
sume that there is a distinct walking adjustment rather than
that the individual walking movements are released directly
by the next higher adjustment (for example, the anger of the
above illustration) may be indicated first by the case of the
child. In the case of a baby, the individual walking move-
ments are obviously very irregular and variable. Yet (when
the child is in the 'walking vein') they are all walking move-
ments: they all fall within that one general class. Now,
wherever these two phenomena occur, of variability within
a class of movements and persistence of the class as a whole,
my thesis will be that we must assume a specific determining
adjustment.
In an adult the situation (in the case of walking) would
seem to be somewhat different because of the added influence
of habit. With the advent of habit there come fixed and
invariable sequences (in the case of walking, fixed and invar-
iable sequences of foot and leg movements). This being the
case, the assumption of an immediate walking adjustment to
release and maintain walking movements, as such, would not
seem so necessary. The total complex of movements is nearly
equivalent to a single act and as such would seem a candidate
for the immediate control of a higher adjustment, such, for ex-
ample, as the anger of the preceding illustration. It may be
noted, however, that in unusual situations such as unevenness
or obstacles in the path, this unitary and automatized character
of walking may break down, in which case the original walking
adjustment would seem again to have to come to the fore to
release further walking movements not part of the automatized
act.1
This discussion has brought out three important points
1 This last point, however, I do not necessarily desire to stress. I would be willing
to admit the possibility that with the growth of habits the original determining adjust-
ment upon which these habits are built up recedes and may even entirely disappear
(waning of instinct). And, if such is the case, the total habit becomes an alternative
act directly at the service of higher adjustments.
322 EDWARD CHACE TOLMAN
concerning determining adjustments which it will be well to
summarize, (i) The determining adjustment sets in readi-
ness a particular group of subordinate acts. One and the
same external or internal stimulus may call out quite different
groups of subordinate acts according to the particular deter-
mining adjustment which happens at the time to be aroused.
(2) Determining adjustments often occur in hierarchies, the
higher ones calling out the lower ones and the lowest one of
all calling out the actual acts. (3) The essence of the deter-
mining adjustment and the reason for it consists in the varia-
bility of the subordinate acts. If such variability has dis-
appeared, as is the case where habits have developed, the
determining adjustment tends to atrophy and may, perhaps,
even disappear altogether.
Let us here stop and assert that determining adjustment
'as thus characterized is a definition of instinct.
It would appear that such a definition tends to differ
primarily from most others by virtue of its two-level conception
(determining adjustment underneath, subordinate acts on top).
Most other definitions seem to reduce in the last analysis to
the assumption that an instinct (on the objective side and
before learning has affected it) is a definite and stereotyped
action (i.e., that it is an inherited reflex pattern).1
Two authors, however, I have found who suggest views
similar to mine. They are Woodworth2 and Kempf.3 The
former's conception of 'drive and mechanism,' and the latter's
'autonomic and projicient systems,' both suggest a two-level
account.4 My idea of determining adjustment and subordinate
JTo take an introspectionist on the one hand and a behaviorist on the other:
such a view seems to be that of both McDougall and Watson. See W. McDougall,
'Social Psychology,' p. 29 and following, and J. B. Watson, 'Behavior,' Chs. IV. and VI.
2 R. S. Woodworth, 'Dynamic Psychology,' Chap. II.
8 E. J. Kempf,' 'The Autonomic Functions and the Personality,' Nervous and
Mental Disease Monograph Series, No. 28. See also an article by G. A. de Laguna,
PSYCHOL. REV., 1919, 26, especially page 419, for a discussion of emotion significant
for the present theory.
4 Woodworth might object to his theory being called two-level, since he seems to
hold that one and the same apparatus may function either as a drive or mechanism
according to circumstance (i.f., that the difference between drive and mechanism is
functional rather than structural). But from the point of view of function, if not
structure, his is a two-level theory.
INSTINCT AND PURPOSE 223
act would differ from either of theirs only, first, in leaving
speculation as to the actual mechanism of the thing open;
and, second, and most importantly, in emphasizing the varia-
bility among the subordinate acts.
It is this variability which I now particularly wish to
emphasize. It will be found especially significant when we
turn to the analysis of purpose.
By way of introduction to that analysis, let me now quote
two concrete descriptions of animal behavior. First an account
of nest building by Prof. Swindle.1 The bird observed was a
male, one of a pair of Brazilian birds, in a large outdoor cage
in a zoo.
"Early in April, I noticed that as this bird walked about
in its cage, it occasionally bit in the air as if it were grasping
an object. At times, however, it bit the bars of its cage, a
branch of a tree, and even the naked earth. Sometimes, it
sprang and ran rapidly, and it very often flew to a one-and-a-
half meter post on the top of which a wide shallow basket was
fastened. On April 18, 1915, the following movements were
observed: The bird stood at first motionless in the basket,
shortly it began biting in the air as if attempting to seize
something, and occasionally it seized, lifted, and then dropped
certain branches which lay in the basket. It threw a stick
out of the basket onto the ground, gazed for a few seconds
at the sky, grabbed still another twig which it had previously
thrown out of the nest, flew back into the basket with these, beat
them quickly here and there without releasing them, let them
fall in the basket, bit and arranged them there, and then
remained for many minutes by the side of the female which
was then also in the basket. Presently the male, half-springing
and half-flying, reached the ground, ran quickly to and fro
in the cage, gazed for a while at the female as he walked round
and round the post, and finally sprang and seized a twig which
projected from the basket. This twig was unfortunately so
badly tangled with the others of the nest that a great many
were drawn out with it. But in spite of the fact that the nest
was occasionally mutilated by the builder, a neat nest was
eventually constructed."
1 E. P. Swindle, Amer. J. of Psychol., 1919, 30, 180.
224 EDWARD CHACE TOLMAN
The startling thing about this account is that it indicates
that, even in the case of those supposedly perfectly adaptive
instinctive activities such as nest-building, careful observation
may show a considerable amount, and in this case indeed a
positively shocking amount, of variability, the very point we
wish to emphasize. If the reader is inclined to doubt the
validity or general applicability of this one case, I may quote
a word or two more.
The author says, further: "The fact deserves emphasis
that birds often work very crudely while building the nest.
It is really astonishing how often a bird allows objects of
building material to fall, apparently without responding further
to them. A bird frequently stands or walks among objects
which it could well use in constructing its nest but suddenly
runs or flies away without grasping any of them. I have
observed the Blue Jay to tear the leaves, branches, and feathers
from another bird's nest before it seized an object of the foreign
nest and flew to the one which it had started; and it often
seemed to arrange the objects on the foreign nest as if it were
preparing to deposit its own eggs there, which it did not do." l
And, again: "Many birds can build their nests at only
certain portions of the trees which are adapted to hold the
collected objects, and generally, these places are located by the
birds only after a number of trials. That this fact is not well
known seems to be due to the circumstance that it is very
difficult to observe a bird with sufficient scientific accuracy in
freedom. It should also be mentioned in this connection that
birds occasionally start two or three nests simultaneously and
later destroy some of them to obtain the material for a single
nest." 2
Let us turn, now, to our second case; to wit, Prof. Thorn-
dike's classical- experiment of a hungry kitten shut up in a
cage with food outside. We quote his words: "When put
into the box the cat would show evident signs of discomfort
and of an impulse to escape from confinement. It tries to
squeeze through any opening; it claws and bites at the bars
1 Op. cit., p. 178.
2 Op. cit., p. 183.
INSTINCT AND PURPOSE 2*5
or wire; it thrusts its paws out through any opening and claws
at everything loose or shaky; it may claw at things within
the box. It does not pay very much attention to the food
outside, but seems simply to strive instinctively to escape from
confinement. The vigor with which it struggles is extraordi-
nary. For eight or ten minutes it will claw and bite and
squeeze incessantly." l It may be added that in the experi-
ment as arranged by Thorndike the kitten usually got out
eventually because one of its strugglings quite accidentally
hit upon a release mechanism arranged to open the door.
These mechanisms were always very simple: a hanging loop
of wire which required but the slightest clawing, or a wooden
latch easily lifted by the nose. It is to be observed that with
the opening of the door, a new final act occurred, the cat went
out and ate.
I draw attention to the significant feature of both illustra-
tions, the variable or random character of the subordinate acts.
In the one case, squeezings, bitings, clawings; and in the other,
picking, dropping, carrying. First one act and then another
occurs in a perfectly haphazard order. The whole thing seems
to be mere chance.
Let us stop, however, and analyze the thing further. Is
there any principle underlying the actual order of these, to
all outward appearances, purely random acts? It must be
supposed that there is. To return to our slot-machine, we
must suppose that each one of them is set off by some particular
penny, if we could but detect the penny. Now, undoubtedly
the pennies are in part internal conditions such as the senations
arising from muscle strain. In addition, however, it is equally
certain that they are in part external objects. The stimuli to
which the cat's clawings, biting, etc., are the responses are in
part particular features of the cage itself. In so far as they
are such features, a definite principle underlying the succession
of the responses can be actually observed. The cat, at any
given instant, responds to a feature of the cage with which its
just previous reaction has brought it in contact. In other
words, it carried out a definite train of movements. One
1 E. L. Thorndike, 'Animal Intelligence,' 1911, p. 35.
226 EDWARD CHACE TOLMAN
feature of the cage calls out one response. As a result of the
movement made by this response the cat is brought in contact
with another feature of the cage. This new feature calls out
still another response and so on. If, in between these responses
to the successive parts of the cage as such, we imagine a few
responses to purely internal conditions, we shall have a fairly
exact picture of the cat's total behavior. An identical account
could be given of the bird's nest-building activities. Although
to all outward intents the acts are purely haphazard and
random, still they follow definitely traceable sequences.
Finally, at the risk of hammering the point to excess, we may
again emphasize that in each case the particular subordinate
acts are what they are, rather than other equally possible
responses to the same stimuli, by virtue of the particular
sensitizing effect of the determining adjustments.
The second feature about the process, to be emphasized,
is that the individual random responses continue until some
one of them presents a new stimulus, the final response to
which, removes the condition or stimulus which was the original
cause of the determining adjustment itself. In the case of the
bird, the activity continues until a nest eventually gets built.
When this happens we have a new stimulus, the completed
nest. And the responses to the completed nest, those of laying
and setting, are such as to remove the internal physiological
condition which was the original cause of the nest-building
adjustment. In the case of the cat, its random acts eventually
open the door, whereupon food is presented and the response
to food is such as to remove the internal sensation of hunger
which was the stimulus to the original food-getting adjustment.
In each case the determining adjustment continues, until some
of its subordinate acts removes the stimulus and with it the
adjustment itself.
We have here a fundamental phenomenon. In as much
as only one act will remove the adjustment and the adjustment
continues until that act occurs and in as much as further the
adjustment sensitizes and hence, so to speak, supplies the
group of acts among which this final successful one appears,
it seems to me that we have a situation which may truly be
INSTINCT AND PURPOSE 2*7
characterized as one of purpose. A determining adjustment
provides the purpose. The subordinate acts (which the adjust-
ment sensitizes) are the means which the organism adopts to
fulfill that purpose and the removal of the determining adjust-
ment itself (as a result of one of these subordinate acts) con-
stitutes the fulfillment of that purpose.
If this analysis be accepted, then the goal of this paper, an
objective psychological analysis of purpose, is in sight. All
that remains to be done is to transfer the account to man.
Before attempting this latter, however, let me draw attention
to the relation between this definition of purpose, and that of
Professor Perry.1
In Professor Perry's analysis the emphasis is put upon
learning, upon the fact that with repetition the cat gradually
learns the successful act. It is in the fact that the successful
act is selected (learned) and the other acts rejected, that
Professor Perry sees the primary justification for calling the
situation teleological. I, on the other hand, believe that even
without learning, the situation is teleological. Even though
the cat showed absolutely no evidence of learning to get out
in a shorter time on succeeding trials, the mere fact that on
each single trial it hits about until it gets out, seems to me to
be sufficient to characterize its activity as purposive. The
cat hits about in order to get out, for the sake of getting out —
expressions which Professor Perry himself designates as the
'most unmistakably and unqualifiedly teleological expressions
in common use.'
It will be noted that such situations do not imply anything
essentially non-mechanical. Given the environment and the
total condition of the organism, the complete response (i.e.,
the particular succession of subordinate acts and the time of
the appearance of the final successful one) can all be predicted
in a wholly deterministic way. This, however, is no criticism
of the definition. When we are talking mechanism we would
be very much upset to find something which was not mechan-
ical.2
1R. B. Perry, 'Docility and Purpose,' PSYCHOL. REV., 1918, 25, 1-20.
1 It should be emphasized that my definition differs from Prof. Perry's principally
in not going quite so far. The essential idea for my conception was for the most
228 EDWARD CHACE TOLMAN
To return, now, to our further problem: we wish to show
how this mechanism of determining adjustments (or instinct)
and subordinate act works in man. One preliminary remark,
however, may not be amiss. In the preceding descriptions
we spoke as if the random strugglings of the cat or the bird
always ended in success. As if, that is, when the activity
ended, it was always because a response was finally made which
removed the initiating stimulus for the determining adjust-
ment. But such an ending, though from the point of view of
purpose the successful one, is by no means the only mechan-
ically possible one. Instead of the cat's getting out and eating
the food, some other powerful adjustment may intervene and
replace the food-getting. Thus it may become frightened by
the bruises and bumps that it receives as a result of its strug-
glings so that a. fear adjustment gradually sets in and replaces
the original food-getting one. Hence when the door finally
does open, the cat either continues to struggle or runs off and
hides instead of eating. In such a case the original food-getting
adjustment has not been satisfied but merely replaced by
another. A second way in which the original determining
adjustment may not be 'satisfied' may be as a result of exhaus-
tion. The cat may become so utterly exhausted that all
responses cease to be made. 'It gives up trying.' Only if
some subordinate act takes place which was released by the deter-
mining adjustment and which removes the stimulus to that adjust-
ment can the -purpose as such be said to be fulfilled.
Let us now turn to man. In the case of a single purpose I
believe that the situation is essentially identical to that repre-
sented by the cat struggling to get out of the box or of the bird
struggling with sticks and straws. An instinct or determining
adjustment is aroused. This facilitates and sensitizes one
particular class of subordinate acts. Some one or group of
these, if they occur, remove the stimulus to the original instinct
and, if they do thus occur, we say, speaking in purposive
part drawn from Perry's discussion. It may be noted, however, that determining
adjustment and subordinate acts, though analogous to, are not exactly identical with
Professor Perry's 'higher propensity' and 'lower propensities.' I believe, indeed,
that my two concepts are, for the purposes of behavioristic treatment, more precise
and definite than his.
INSTINCT AND PURPOSE «9
language, that our problem has been solved, that the right
means have been chosen.
We may make the issue more concrete by an example.
Imagine a man trapped in a burning hotel. He may rush
madly about in the same blind fashion as does the cat in the
cage. If so, his behavior and that of the cat would seem to be
entirely identical. It may happen, however, that instead of
thus rushing blindly he stops to think. If such be the case,
he does not attack all the exits of his trap indiscriminately,
but only some one which is apparently suggested to him by his
'thoughts.' We have in these thoughts a new principle which
does not hold or, if so, to an infinitesimal extent in the case
of the cat.
What, now, we may ask, is this thought and when and
why does it occur? It will be remembered that in the initial
statement of our program we declared that it was an objective,
behavioristic account rather than an internal subjective one
that we should aim to achieve; that we were interested not
in how purposes felt, but in how they worked. Can we, now,
shift our point of view and begin to talk about apparently
internal subjective things such as thoughts? My answer is
that thoughts, or at least the kind of thought with which we
are here concerned, can be conceived from an objective point
of view as consisting in internal presentations to the organism
(on a basis of memory and association) of stimuli not actually
present but which would be present, if some hypothesized
action were carried out. Such a definition says nothing about
the subjective 'immediate-feel' side of thoughts as such. It
is concerned with thought simply in so far as the latter has
significance in an objective, behavioristic, stimulus and response
account. A complete treatment of thought on its subjective
(immediate-feel) side and of its epistemological significance
we can leave to the combined researches of introspective
psychologist and philosopher.1 The one point we here mean
1 In what follows I present one sample of the way in which 'thoughts' may, it
seems to me, be properly introduced into what claims to be a purely behavioristic
(stimulus and response) account. It is my belief that such examples might be indefi-
nitely multiplied and that a whole system, properly to be called behavioristic psychol-
ogy, might be built up in which thoughts (on their behavioristic side) would still find
as much place as do sense-stimuli.
23° EDWARD CHACE TOLMAN
to make is that over and above whatever these functionaries
may have to say, a significant behavioristic aspect of thought
still remains.
To return to our definition itself. What we mean by
thought in this particular case as 'an internal presentation to
the organism (on the basis of memory and association) of
stimuli not actually present but which would be present, if
some hypothesized action were carried out,' can be made clear
perhaps, only by a concrete example. We come back to our
man in the hotel. Instead of trying all possible parts of his
burning trap, we find him stopping to think and then on the
basis of that thought reacting to certain parts only. What is
this stopping to think in behavioristic terms? It consists,
I would assert, in what may well be called, not random sub-
ordinate acts, but random subordinate thoughts-of-acts.
You will remember that the cat reacted to a part of the
cage with which the just preceding response had brought him
in contact. As a result of each successive response the cat
was automatically provided with a stimulus for another re-
sponse. How now in the case of the man? He sees a door
but instead of actually responding, he merely thinks of re-
sponding. He hypothesizes a response and on the basis of this
hypothesized response he achieves mentally, a new stimulus,
i.e., the mental image of what is beyond the door. (This
mental image is provided by memory or association. He may
actually remember that this door leads to a corridor or merely
by association based upon the general position be merely led
to imagine that it leads to a corridor.) In either case this
idea or mental image of what is beyond acts as a stimulus to
a new thought-of-act. He thinks of going down the corridor
and this will lead, mentally, to still a new stimulus of what is
at the end of the corridor, perhaps stairs or an elevator. These
will lead in a flash to a mental image of an open road to the
outside. When the image of the latter occurs he will in all
probability actually react. If no such vision of an open road
beyond stairs or elevator occurs, he will mentally rush back,
or perhaps more truly speaking jump back, to his starting point
and mentally attack some other feature of his environment.
INSTINCT AND PURPOSE 231
One point in need of immediate further elucidation emerges.
Why is it that the man thinks in the first place? We have
already said that sometimes he does not think but merely
rushes blindly about as does the cat. Stopping to examine
the matter more carefully, it would seem that those times
when he does not think, but does thus rush blindly about, are
times when he is excessively frightened. Occasions, in short,
when the original determining adjustment is especially potent.
If, therefore, a particularly potent adjustment produces action,
I should suggest that when action does not occur, it is when
some inhibiting or checking process which works against or
controls the determining adjustment is also acting. Just what,
physiologically, this inhibiting or checking process may be,
I shall not venture to say, though I shall assume that like all
other physiological processes it follows perfectly definite me-
chanical laws. The significant thing, for us, is simply that it
works against the determining adjustment in such a way as
to shunt the latter's activities, so that instead of producing
subordinate acts, it, the determining adjustment, produces
merely though ts-of-acts. For this reason, let us call it the
thinking or rationalizing adjustment. We may note in passing
several interesting things about this tendency. First it seems
to be much stronger in some individuals than in others and
secondly, to be very much subject to training and practice.
In fact it can sometimes be so over-trained as to result in an
almost complete inability to act at all, a condition which is
sometimes supposed to be characteristic of the typical college
professor.
We may now ask, how is it that this thinking, rationalizing
tendency, having once got going, ever ceases, in order to allow
action to take place. In answer, we would posit the general
principle that action eventually occurs because of what we shall
call a prepotent stimulus. A stimulus may be prepotent for
either of two reasons: (a) because it is the stimulus to an
act to which the original determining is particularly favorable
or (b) because it releases some other favorable adjustment. To
return to the example of the man in the hotel. The first case
would be represented when as a result of his trains of thoughts-
232 EDWARD CHACE TOLMAN
of-acts he arrives at the mental image of open road beyond
stairs or elevator. Such a sight of the outside, if present in
perception, would be the one stimulus to which the man would
have been most sensitive. If the door in the cat's cage had
been left open, the response of going out would have been
first to occur. It would have taken precedence over all other
responses such as those of squeezing, clawing, biting. So, here,
the thought of the open road beyond stairs or elevator is the
stimulus to which the given determining adjustment makes the
man most susceptible, so susceptible in fact that the impulse is
enough to break down the inhibiting effect of the thinking
propensity and action results.
Turn, now, to the second way in which the stimulus may
be prepotent: the case in which it is prepotent because it
releases a second adjustment favorable to the given action in
addition to the original adjustment. Suppose that as a result
of social intercourse our man had acquired a general maxim
to the effect that stairs and elevators become perfect sr^u~
stacks and that much the best thing to do in case of fire is to
run to the window and call for help. Simply on the basis of
his original determining adjustment alone the thought c
window would tend to call out the subordinate act of
and calling. If, now, in addition such an act is support d by
what we may call a general 'social-subservience' adjustment,
a tendency to do those acts recommended by society, this act
becomes doubly ready to go off, so much so that it does actually
occur.
To sum up: thought ceases and action supervenes when-
ever thought arrives at the image of a prepotent stimulus.
And a stimulus is prepotent either (a) because it tends to call
out a subordinate act which is especially favored by the original
determining adjustment or (V) because it tends to call out in
addition to the original determining adjustment some other
adjustment which is also favorable to the act.
This is all there is to a case of single purpose.1 An original
determining adjustment provides the purpose. Subordinate
1 The problem of what happens in the case of a conflict of purposes is more com-
plex, but the general principles of explanation would be the same.
INSTINCT AND PURPOSE
acts are either actually called out or merely thought of.
Eventually one occurs which removes the stimulus to the
determining adjustment and the purpose is satisfied. Or, if no
such subordinate act occurs, it remains unsatisfied until, per-
haps, mere exhaustion causes the determining adjustment to
disappear.
In conclusion, we may briefly enumerate the more impor-
tant points we have advanced and which we most wish to
emphasize: (i) a two-level (i.e., determining adjustment —
subordinate act) theory of instinct; (2) Purpose as interaction
of determining adjustment and subordinate acts; (3) images
of memory and imagination (thoughts) as properly included
in a behavioristic non-introspective account; (4) the satis-
faction of purpose as consisting in the removal of the stimulus
to the determining adjustment as a result of one of the sub-
ordinate acts which the determining adjustment itself releases.
BRAIN MECHANISMS AND MENTAL IMAGES
BY S. BENT RUSSELL
St. Louis, Mo.
When you think of an absent friend you may seem for an
instant to see his face, i.e., you have a mental image of his face.
This is commonly said to be due to memory. But what then
is memory? There may be readers of this article who have
considered memory only from the subjective point of view.
Let us for the moment now consider memory apart from the
matter of images and consider it from the objective point of
view. It can be shown that memory is accounted for by the
operation of nerve processes. There are pathways along nerve
fibers for nervous impulses. Some of these are known as
association pathways. There is good reason to believe that
changes of resistance at the points of junction between nerve
fibers known as synapses, determine the course of nervous
impulses along one path or another. The frequency and re-
cency of previous impulses determines the synaptic resistance.
Hence the nerve paths are developed by individual experience
and thus memory is evolved.
The dual common path theory furnishes an explanation of
associative memory. The common path is open to impulses
coming from two tributary or private paths. An impulse
from one lowers the synaptic resistance for an impulse from
the other. We need not go further into this theory in this
discussion. Let us take it as an explanation of associative
memory for the purposes of this demonstration.
Before you can have a memory image of an object you must
at one time be conscious of the object itself. Let us consider
briefly how perception can be explained in terms of nerve
processes. A man is never conscious of an object unless there
be communication from the object to his brain. We may say
further that there must also be a molecular change at some
brain center or centers that bears a correspondence to the
234
BRAIN MECHANISMS AND MENTAL IMAGES
object. In other words, the object causes a nerve impulse from
a sense organ to a brain center and the impulse causes a
response (molecular) at the brain center. To mention a special
case, when a point on the retina is stimulated by a ray of
light from some object, there is an impulse sent by way of a
certain nerve fiber to a certain nerve center in the cerebral
cortex, where it provokes a molecular response that corresponds
to the retinal stimulation. We see that each visible point of
the object has a line of communication by way of the retina
and a particular nerve fiber to a brain center. All the points
together produce a pattern molecular response in the brain.
This joint response simulates the object in view.
Let us for convenience use the term 'mimetic response' to
express the molecular response that simulates the object as
represented at a sense organ.1
To put it another way, there is a registering mechanism in
the brain center that transforms the afferent impulse into light
equivalent, heat equivalent or other sensation equivalent as
the case may be. That is, it in effect reverses the transforma-
tion that occurred in the sense organ. When a ray of red
light for example falls on the retina, there is a change in the
brain center the same as if a red light had penetrated it.
Let us term this hypothesis 'the mimetic theory of percep-
tion.'
This explanation of perception is not complete unless we
make allowance for the effect of association mechanisms. You
will see this if you think of a trained musician listening to a
well-known melody on one hand, and a man without any
musical training listening to the same melody for the first time
on the other hand. The brain correspondence will be far
greater in the former case. Every individual has to be trained
to see and hear things as they are. Brain correspondence
increases with knowledge, i.e., with the development of associa-
1 The writer has presented an explanation of mechanisms of associative memory
and of their functions in intelligent and purposive behavior and in consciousness in
the following articles in the PSYCHOLOGICAL REVIEW. 'The Effect of High Resistance
in Common Nerve Paths,' 1916, 23, 231-236. 'Compound Substitution in Behavior,'
1917, 24, 62-73. 'Advance Adaptation in Behavior,' 1917, 24, 413-425. ' Communica-
tion, Correspondence and Consciousness,' 1918, 25, 341-358.
236 S. BENT RUSSELL
tion mechanisms. In other words, a man's habits determine
the definiteness of his perception.1
We must also make allowance for the effect of language and
other forms of expression upon perception. Naming and num-
bering are great helps in perception. In learning to describe
his environment, a man develops association nerve paths that
serve to increase his brain correspondence with the world
he lives in.
Keeping these matters in mind, we can say we have in the
mimetic theory, a satisfactory explanation of consciousness
as it reflects a man's present environment. The purpose of
this discussion is to bring out an explanation of the memory
image in terms of nerve processes. Before attacking the prob-
lem, let us note the difficulties we have to meet. It is not so
hard to conceive how a ray of red light can cause a disturbance
in a brain center that is characteristic of redness as it is to
conceive of a nerve impulse aroused in the ear by the sound of
the word red, for example, producing a disturbance in a brain
center that is so characteristic of redness that the subject will
for an instant see a red color in his mind. That is, he is con-
scious of red when there is in reality no red to be seen. There
is, you will allow, an apparent paradox. Why should an
impulse coming from the ear, arouse an image of light? We
can think of a ray of red light penetrating a brain center and
causing a characteristic molecular disturbance like a light ray
acting on a photographic plate. We can go another step and
think of a ray of red light stimulating a sensory nerve ending
and so causing a nerve impulse which goes to the brain center
and by aid of a special mechanism reproduces there the dis-
turbance at the sensory ending caused by the red ray and
characteristic of redness. But why should some other kind of
stimulus produce a molecular disturbance that has a quality
belonging to red?
In other words, if we seek explanations in terms of nerve
processes, the one for consciousness of the past is far more
difficult than that for consciousness of the present.
On account of the obvious resemblance, the memory image
1 W. B. Pillsbury, 'The Essentials of Psychology,' 1911, p. 157.
BRAIN MECHANISMS AND MENTAL IMAGES 237
is often considered as composed of sensations that are centrally
aroused instead of coming from the effect of the environment
acting directly as in perception. In these terms we would
note that an explanation of a centrally aroused sensation by
nerve processes is by no means as simple as in the case of a
true sensation. If the reader can see the point of difficulty
it will help him grasp the demonstration given herein.
As the next step let us take for illustration the case of a
young man of normal mind sitting in a boat and drifting down
a river. As he drifts along he observes different objects along
the bank and of course is conscious of each in turn. Let us
say it is the second time he has made this voyage. He will
have from time to time mental images of objects that have
passed out of view and sometimes he will have an image of
an object ahead of him that will soon come into view. Suppose
at one time he passes a gravel bar on which he sees a flock of
white cranes. Later on he passes another gravel bar and has
a memory image of the white cranes. At one time he passes
under a railroad bridge and is startled by the roar of a train.
Later on he passes another bridge and has a memory image of
the roar. Let us see if we can find an explanation of these
memory images in terms of nervous mechanisms.
In a book published some years ago, Professor Kirkpatrick
describes images as the result of the functioning of brain centers
that are made active in perception by impulses coming from
sensory centers, while in imaging they are made active by
impulses coming from some other direction.1 The theory that
will now be advanced does not conflict with Dr. Kirkpatrick's
view.
Returning to our illustration, when the man observes an
object in the first instance, there is, as we have already stated,
a molecular change in the cerebral cortex 'that simulates the
object and which we term the mimetic response. That is
not the whole story, however, as there is another process too.
In the case of vision for example, the mimetic response only
lasts while the retina is stimulated by the object or for a
moment more. There is a second process that is a recording
1 E. A. Kirkpatrick, 'Genetic Psychology,' 1909.
238 S. BENT RUSSELL
process. Let us call this record making process the 'tuning
process.' It is a specific molecular change that is more perma-
nent than the mimetic response. There is one tuning process
for light, one for sound, one for heat, etc. After the tuning
process has occurred in a certain center, any nerve impulse
that reaches the center will cause a molecular disturbance
that simulates the original stimulus disturbance in the sense
organ. In this way we have correspondence of a brain center
with a past environment. This simulating process in the
nerve center is, we will say, a conditioned molecular response.
Let us term it the 'sounder response.'
In our illustration when the man is conscious of his present
environment, there is a series or procession of impulses through
certain brain centers. In other words the mimetic response is
a series of molecular changes in the brain centers that bears a
correspondence to the environment. The tuning process is
also a series of molecular changes that bears a correspondence
to the environment. At the same time it leaves an impression
on the molecular structure that also bears a correspondence
to the environment. We may liken the mimetic response to a
gust of wind passing over a sheet of water and causing a
ripple that is soon gone and we may liken the tuning process
to a gust of wind passing over a smooth sand bed and leaving
the familiar ripple marks that remain for a time.
The sounder response is conditioned by the tuning process.
It is a series of molecular changes that bears a correspondence
to the environment that caused the tuning process. There
must of course be a specific sounder response that corresponds
to light and another that corresponds to heat and another for
sound and another for taste, etc. When a man has a memory
image of a noise he heard a minute ago, there is in one of his
brain centers a "sounder response that is provoked by a nerve
impulse coming, we will say from some association nerve path.
The effect is the same as if there were communication from the
noise-making object to the brain center that took a minute
to reach the brain center. It is something like the thunder peal
that reaches the ear some seconds after the lightning stroke
that made it.
BRAIN MECHANISMS AND MENTAL IMAGES 239
In the same way when a man has a memory image of an
object that has recently passed out of view, there is in one of
his brain centers a sounder response that is provoked by an
impulse from some association nerve path. The effect is the
same as if the object were still in communication with his brain.
In his mind's eye he sees perhaps a red triangle although there
is no red triangle now present to stimulate his organs of vision.
Let us now consider the tuning process again. A harp
string can be tuned by turning the peg that holds it and thus
changing the tension. With a given string a certain tension
will produce a certain note, say C. Now a certain tension
means a certain relation of the molecules to each other. The
tuning operation produces a certain molecular arrangement
that corresponds to the note C. As long as this molecular
arrangement is preserved the string when struck will give the
sound of C. The note C is of course a vibration at a certain
rate and thus we see the molecules are so arranged by the
tuning operation as to give when struck, a certain response.
We may use the harp string as an illustration of a brain center.
We may suppose that a nerve mechanism exists by which a
certain brain center can be brought into a certain molecular
arrangement so that thereafter any nerve impulse will produce
a particular response or movement.
The tuning process is, we have stated, a change in the
molecular structure. We may compare it to the alteration in
litmus paper that is changed in color from red to blue when
it is wet with an alkaline liquid. We note that after the change
the paper gives off blue rays when exposed to light. We may
also compare the tuning process to the case of paper that is
tinted with some color that is not fast. If a ray of sunlight
falls on it, a faded spot remains that will be seen whenever
there is any light on the paper and so is a sort of memory
image of the sunlight ray. Hence we may think of a bright
object in view of a man as causing a little faded spot in his
brain that is brought out by any passing nerve impulse that
comes later on.
Let us remember that the man is never conscious of the
mimetic response or the sounder response. He is conscious
240
S. BENT RUSSELL
of the object itself that furnished the stimulus, whether it be
in the present or past environment. In the case of an image,
the effect on consciousness is the same in kind as if the object
were still in communication with the brain center.
It may make these explanations more clear to you if they
are illustrated by a diagram or table like the one below.
Let the features of the object seen in the first instance be
represented by the letters A, B, C, D, E, and the impression
on the retina be represented by the letters Ai, Bi, Ci, Di, Ei,
and the mimetic response by A2, B2, C2, etc., and the tuning
process by A3, 63, C3, etc., and the sounder response by
A4, 64, C4, etc. The table follows :
Object
A
B
c
D
E
Object
Retinal image
Ai
Bi
Ci
Di
Ei
Mimetic response
A2
B2
Cz
D2
£2
Perception
Tuning process
A3
BS
Cs
D-j
Ei
Sounder response
*
A*
3
rj
D4
FI
Mental image
Let us now go back to our case of the man drifting down
the river. We will take the experience of the bridges. There
is a certain center in the man's brain that is reached by a
certain nerve path coming from the ear. It is also reached
by a certain association path that has one connection with the
ear and one with the eye. When the man sees the first bridge,
there is an impulse from the eye via the association path and
when he hears the roar of the train there is an impulse from
the ear via the association path and one via the other path
from the ear. The sound impulses cause a tuning process in
the brain center. Later on the man observes the second
bridge and there is an impulse from the eye that follows the
association path more easily as it is now more open because
the two impulses in succession, caused by the experience at
the first bridge, have made it so. When the impulse reaches
the brain center, there is a conditioned molecular response
and the man has a sound image of the train roar at the first
bridge. Please note that the response simulates a sound al-
though the provoking impulse originates in the eye. This is
the apparent paradox that we remarked in the beginning.
BRAIN MECHANISMS AND MENTAL IMAGES M1
You may comprehend the case better when you see it
presented in a diagram as in Fig. I. The circle V represents
the visual organ, the circle A^ the auditory organ and the circle
R, the brain center. At the sight of the first bridge an impulse
travels from the eye at V by the dotted line path C to the
center at R. The roar of the train sends a second impulse
from the ear at A by the common path C
to R, and a third impulse by the dotted
line path B to R. The last two impulses
provoke a tuning process in R. After the
first and second impulses have passed over
path C it is more open to later impulses.
The sight of the second bridge later on
sends an impulse from the eye at V by the
path C as shown by a heavy line, to the
brain center R. There is now a sounder
response at R that simulates the roar of the train at the first
bridge. A path line on the diagram of course represents a
plural number of nerve fibers.
On consideration you will find that the explanation just
given for a particular case will answer for all memory images.
There are many cases of course, where a large number of
association nerve paths are required to produce a definite
memory image. The larger the number, the higher will be
the degree of correspondence with the environment at the
time of observation.
The association nerve paths must be developed by indi-
vidual experience. A baby cannot have mental images like
an older person. A baby must learn to talk before he can
hear words in his mind. He must learn to sing before he can
hear a song in his mind. He can only coordinate impressions
that have become familiar. He learns to form images step by
step. He must learn to walk before he can image distance
relations and space relations. We have previously observed
that perception is determined by a man's habits to an important
extent and the same thing is true of mental images. The
study of language and expression, also, must have an important
effect in developing the mechanisms for making mental images.
242 S. BENT RUSSELL
A man who has learned how to write a good description of a
landscape has no doubt increased the definiteness of his mental
images. You are perhaps able to look at a new model of an
aeroplane and memorize it so as to have a good memory image
of it a week afterwards and have never remarked that your
ability to do this is the result of long and repeated practice on
similar objects. On consideration you will see that these
statements fit in with our theory that the image is due to a
conditioned molecular response that simulates a past environ-
ment at one time in communication with the brain centers
responding.
We find then that a typical memory image accompanies a
compound sounder response and is due to the operation of
tuning process mechanisms and association mechanisms, the
latter serving to coordinate the former.
We have found an explanation of how the brain can simulate
a past environment. Let us pass on to the case where the
image corresponds with a coming environment. Let us return
to our illustration of the man on the river. Suppose he sees
in the distance ahead of him a column of smoke and then has
a mental image of a landing pier at a village around the next
bend. At first thought this appears to be an image of a
future environment but on second thought we find that the
man's mental image is the result of his previous voyage over
the same route. The same rule will apply to all cases where
the image appears to anticipate the dbject. The mental
picture of things to come is made up of elements derived from
past experience.1 A man can imagine an aggregate that he
has never seen but it will be made up of familiar units. The
power of combining mental images is acquired by degrees.
It must be largely due to social environment and to language
associations. What one man has observed another man can
image by the medium of language. The younger are taught
analysis and synthesis by their elders.
It is evident then that an image of a coming environment
is due to the same nerve mechanisms that serve for the image
of a past environment. All mental images are really memory
images in kind.
1 Pillsbury, op. cit., p. 131.
BRAIN MECHANISMS AND MENTAL IMAGES
In considering these brain processes, one should keep it in
mind that the power to form an image diminishes with time.
You perhaps have a clear mental image of a particularly fine
cigar that was given you yesterday, but had it been a week
ago you probably would have no image to speak of. The
longer the time, the less is the conductivity of the association
nerve paths and sooner or later the mechanisms will fail to
coordinate so as to simulate the past environment.
The kinaesthetic impulses from actual and incipient mus-
cular movements have much to do with provoking sounder
responses. Acting, talking and thinking are all linked together
by association mechanisms. Behavior habits and language
habits serve to prevent the untimely occurrence of sounder
responses as a rule. It is not always true, however, as a man
can sometimes be reading aloud and at the same time, have
mental images having no connection with the words which he
is subconsciously repeating.
When a man is awake but resting quietly in an unchanging
environment, he is usually unconscious of his environment or,
as we say, he is lost in thought. From our point of view there
are no mimetic responses in his brain centers and his nerve-
muscle system is occupied with a series of incipient muscular
movements accompanied frequently by a series of sounder
responses in the brain centers. There is a constant flow back
and forth of efferent and afferent (kinsesthetic) impulses which
penetrate one cortical center after another. Those centers
that have been tuned by past experience give a characteristic
conditioned molecular response. It is thus that mental images
transpire more commonly. The man may be thinking of
future events but it is always in terms of what he has seen
and heard before.
Having reached these conclusions regarding the mental
operations of a man, it will not be amiss to consider briefly the
mental processes of animals lower than man. Do monkeys,
dogs and other intelligent animals have mental images? We
know that the nervous system in these animals is much the
same as in man and in behavior also there is much resemblance.
The brute learns by experience as does the man. In man we
244 S. BENT RUSSELL
find great superiority in expression as he has an articulate
language which the brute has not. There is, however, com-
munication with the brutes. The baying of a beagle on the
trail of a rabbit, no doubt has in its changes more meaning to
the other beagles than even to the experienced hunter. The
available evidence indicates that the nervous mechanisms in
the animal are of the same kind as those in man. Therefore
it is probable that a squirrel has a mental image of a nut that
he has hidden away for future use, the same as a man would
have, and that a fox has a mental image of the bone that he
buried and will one day dig up. A bird probably has a mental
image of her nest that she built herself with the aid of her mate
just as a man has of a hut he has constructed. In the man's
brain the association mechanisms are far more complex. The
number of association nerve paths is much greater.
In conclusion let us say that a mental image can only occur
when there is a coordinated molecular response in a brain
center that is conditioned by a former environment. The
response is conditioned by means of registering mechanisms
in the center and of association mechanisms. The response
is the same in kind as if the environment in question were still
acting upon the sense organs and so upon the center. The
man is not conscious of an image as such but of the environ-
ment as it was. He sees it around the corner, one might say.
Usually, of course, the response is very much fainter than in
the case of direct stimulus from the sense organ in perception.
The registering mechanism at the brain center when in
communication with the environment, acts by retransforming
the nervous impulses into specific molecular changes that simu-
late the environmental action upon the sense organ. The
result is a permanent set of the molecules that is a factor in
subsequent responses to excitation by any nervous impulses
that reach the center. The responses in the registering centers
are further conditioned and also coordinated by mechanisms of
association. These mechanisms function in the same way as
they do in the case of conditioned reflexes which cause muscular
movements. The same sort of association mechanisms govern
the responses in the registering centers as govern the responses
in the muscles and other effectors.
BRAIN MECHANISMS AND MENTAL IMAGES 245
The knowledge that a highly intelligent man has of the
world about him is due to the brain mechanisms that have
gradually become organized and developed from his birth up.
This applies also, of course, to a man's knowledge of himself.
Finally we cannot escape the conclusion that in a man's
consciousness of past and future, brain mechanisms are an
essential factor. Without brain mechanisms a man would
have no more mental power and no more consciousness than
an apple tree.
VOL. 27, No. 4 July, 1920
THE PSYCHOLOGICAL REVIEW
THE MODIFICATION OF INSTINCT FROM THE
STANDPOINT OF SOCIAL PSYCHOLOGY
BY WALTER S. HUNTER
The University of Kansas
In the discussion which follows I have attempted a rela-
tively comprehensive statement of the modification of instinct
with particular reference to the topic as incorporated in
social psychology. The formulations have been made with
reference to a statement of general principles rather than with
a view to a summary of such experimental data as might be.
available. I think most social psychologists approach the;
topic of instinct with the feeling that while it should be of
great importance, from an a priori standpoint, yet that the
actual developments which the topic receives fail somehow
to justify the expectation. Social psychology is, of course,
as interested in the experimental facts concerning instinct
as is normal human adult psychology, but it seeks more
insistently to put the data together in a manner significant
for the understanding of human nature so far as this is modi-
fied by its social environs.
In section I., I have sought merely a reformulation (pos-
sibly more detailed than usual) of the modification of instinct
on its afferent or efferent sides, or on both simultaneously.
This section is preparatory for those that follow. In sections
II. and III., types of instinctive modification are considered
which are new to social psychology and all but unsuggested
in the other fields of general psychology. These sections
discuss (l) the fact and significance of the temporal position
of the modifications as occurring before or after the first
247
248 WALTER S. HUNTER
instinctive performance, and (2) the modification of the
biological purposes of inherited responses.
I. MODIFICATIONS OF THE SENSORY AND MOTOR ASPECTS
OF INSTINCT
Previous writers have attacked the question of the
modification of instinct at three chief points: (a) indicating
that an increase in perfection of response through practice
does take place; (b) disentangling (partly) the separate roles
of maturation and use in the increasing perfection; and (c)
pointing out that modifications concern either the stimulus
or the response side of the instinct. It is this latter point that
we wish to formulate in the present section.
Shortly after birth an individual will, through heredity,
manifest the fear reaction upon the presentation of certain
stimuli. By virtue of associations, these stimuli may later
become ineffective and new stimuli be secured which were
previously indifferent. Thus birds on desert islands show
no fear of man until the frequency of his appearance, coupled
with effective stimuli for fear, finally endows the perception
of man himself with the capacity to arouse fear. Studies of
the conditioned reflex are laboratory observations of this
same phenomenon. The protective reflex of the finger,
e.g., has as its unconditioned (inherited) stimulus injury to
the finger; but by a frequent simultaneous presentation of
sound and injury, sound also becomes an effective stimulus
producing withdrawal of the finger. The internal mechanism
of this need not concern us in the present discussion. It
should be stated, however, that habits as well as inherited
forms of response are susceptible to this type of modification,
the distinction being that we deal with conditioned reflexes
directly when the changes effected are made from the original
stimulus rather than from stimuli which in themselves may
be one or more removes from the hereditary status of the
response.
From the side of changes in effector activities proper,
the same statements are true although the term conditioned
reflex seems not to have attached to such modifications,
MODIFICATION OF INSTINCT *49
undoubtedly due to the accident governing the choice of
laboratory procedure. The protective reflex and the salivary
reflex, i.t.t the effector activities proper, have been kept
constant in such studies and experimentation directed toward
the analysis of stimulus changes. However the physiological
changes effected are presumably no different from those
which occur in the contrary case when experience changes
the type of response while the stimulus remains constant.
The illustrations of this are legion. One may cite the changes
which occur in the "expression" of fear and anger as the
human individual matures in a social environment, or one
may consider the modifications which occur in animal be-
havior during the process of learning. In the latter case, a
total situation is presented to a white rat placed in a visual
discrimination box, calling forth exploratory movements;
but under the influence of punishment, reward, and frequency,
the exploratory movements are inhibited and give way to
well-defined food responses. One may state such an outcome
either as the inhibition of an instinctive response to a given
stimulus by acquired responses, or as the acquisition by the
food-getting response of a new stimulus. Perhaps both
are involved.
The social values of the above types of change in instinct
have been so widely recognized that we need not elaborate
the problem further. This is not true, however, in the case
of those modifications termed sublimations. The sublimation
of instinct in the human individual is an example of the
simultaneous modification of the afferent and efferent phases
of the response. Anger becomes righteous indignation by
the substitution of a new and (in this case) an ideal stimulus
for the sensory (animal) one and by the conversion of the
gross bodily attack into the response of denunciation, pur-
chasing Liberty Bonds, or longer hours of labor. Sex im-
pulses may be sublimated in artistic activity, in dancing, in
religious activity, or, when joined possibly with the parental
impulse, in social service. Instances of sublimation are
those where the inherited impulses are placed at the service
of activities which bear little or no resemblance to the activity
250 WALTER S. HUNTER
which normally embodies the impulse. The cases are not
due to the suppression or elimination of the instinct in its
entirety; only the somatic, skeletal responses are inhibited
while the visceral continue probably at full intensity. The
individual may entirely fail, and usually does fail, to identify
the persistent behavior complex, because to the uninitiated,
instincts are identified in terms of their somatic components.
It is this difficulty of identification which permits the sublimation
to proceed unimpeded by emotional conflict, and unthreatened
by the failure which would almost surely be its lot did the subject
realize the origin of his impulses in their proper (unconditioned}
instinct.
Although the non-technical use of sublimate means to
purify, or to idealize, the preceding analysis would indicate
that the physiological mechanisms involved need not include
the equivalent of ideals. The stimuli for artistic activities,
for dancing, for charity and social service may be as concrete
as for the arousal of any other form of modified instinctive
performance. The presence of syncopated music and mem-
bers of the opposite sex initiates dancing, and the awareness
of suffering and poverty calls out charitable relief in those
individuals possessing the sublimated behavior. And so,
although one would hesitate to apply this term to animals
below man, the understanding of instinctive modifications
is better when one realizes the essential continuity of the
phenomena. Thus a dog can by training be made angry by
whistling, and the instinct can then be modified on its effector
side by training the dog to vent his pugnacity in some unusual
manner. Behavioristically, the instinct is as trully subli-
mated as in man, although the social significance of the
change may be infinitesimal.
It is proper that we should place beside the above state-
ments the following remarks by Woodworth1:
"Freud's 'sublimation' is an attractive concept. It is
'nice' to believe that crude motives, that cannot be allowed
their natural outlet, can be drained off into other activities,
so that a libidinous infatuation, sluiced out of its natural
1 Woodworth, R. S., 'Dynamic Psychology.' New York: 1918, p. 175-6.
MODIFICATION OF INSTINCT ^S1
channel, can be made to drive the wheels of an artistic or
humanitarian hobby. But there is no clear evidence that this
can be accomplished. What does happen sometimes is that,
in the effort to escape from, and distract oneself from a
strong but unwelcome impulse, one turns to some other
activity capable of enlisting interest; and, since the unwelcome
impulse is not easily resisted, one has to become as absorbed
as possible in this other activity. Under such conditions,
interest in this other activity may grow into a strong motive
force and effectually supplant the unwelcome impulse.
But this is distinctly not making the unwelcome impulse do
work foreign to its own tendency. This impulse is not drawn
into service, it is resisted. If there were no other and con-
trary motive force, the impulse in question would have its
own way. We did see that the tendency towards a 'con-
summatory reaction' acted as the drive to other mechanisms,
but these were mechanisms that subserved the main tendency,
whereas 'sublimation' would mean that the tendency toward
a certain consummation could be made to drive mechanisms
irrelevant or even contrary to itself. There seems to be really
no evidence for this, and it probably is to be regarded as a
distinctly wrong reading of the facts of motivation."
We must agree with Woodworth that compelling evidence
of sublimation is difficult to secure. We do believe however
that the psycho-analysts have made a good case for its
existence; and when we remember the introspective difficulties
besetting the identification of visceral components of response
and of minor somatic responses in general, we are tempted to
conclude that the case will always lack that clear-cut evidence
which is desirable. However, the James-Lange theory of
emotion meets the same type of difficulty and yet has managed
to survive its severest critics because of the intrinsic merit of
its claims. We shall indicate schematically in a following
paragraph how the neural processes may proceed in sublima-
tion; but here, in the light of Woodworth's remarks, renewed
emphasis should be placed upon points already stressed.
i. Sublimations do not arise suddenly in an effort to
control an unruly impulse that is recognized as undesirable;
252 WALTER S. HUNTER
they are the end-products of modifications whose formation
has probably extended over several years. The behavior
which may be said to undergo this modification may indeed
never make its actual appearance, due to the fact — which we
shall emphasize later in the paper — that certain habits or
customs have been fixed upon the individual before the
normal time for the instinct to appear. Therefore when
the instinct manifests itself, it does so from the very beginning
in modified form.
2. The visceral responses which constitute the physical
basis of the impulse and emotion of the sublimated behavior
can be identified by skilled introspection as closely similar to
the visceral core (or "feel") of the unsublimated form of the
response. Indeed this is a chief reason for insisting that such
behavior as righteous indignation, e.g., is a refined and derived
form of animal anger. Or again, the alleged similarity of
the emotional quales is a prominent reason for the insistent
attempts to identify the sex and religious activities.
3. One need not assume, as Woodworth does, that in
sublimated forms of behavior the "drive" does only work
foreign to its natural purpose. On the contrary, an intro-
spective description of the cases would suggest that, did we
have adequate recording methods, widespread visceral and
somatic responses would be found present at low intensity in
contrast to the high intensity marking the untransformed
behavior. What is important is that the behavior initiated
by the sublimated impulse shall not impress the observer as a
surviving (or anticipating) part of the original instinct.
The uninitiated subject may only feel the restlessness due to
visceral change.s without recognizing in any degree the total
response to which this restlessness normally belongs. He may
therefore proceed to make use of this impulsive tendency in
some socially acceptable behavior, the frequent repetition
of which may constitute his idiosyncracy or even his profes-
sion. We shall see later that many instinctive impulses may
be made to work out purposes other than those for which
the instinct was apparently designed. In sublimation the
situation is the same, a behavior component becomes trans-
MODIFICATION OF INSTINCT *53
ferred from one total response to another through the so-called
conditioned reflex type of association and so does duty in the
service of a purpose not originally its own.
A formulation in terms of the neural diagram of Fig. I
may help give definiteness to the preceding account. In-
B,
FIG. i. Schematic representation of the neural elements involved in the modifi-
cation of instinct. NS, the central nervous system. Si, the original stimulus; St, an
acquired stimulus; Rt, the original somatic response; Rt, a new or modified somatic
response; X, the visceral sensory component of the stimulus; and IR, the internal, or
visceral, response. The mutual relations of these elements are discussed in the body
of the paper.
stincts belong to either of two classes: those having a con-
spicuous visceral component in the stimulus and those that
do not. This visceral component corresponds on the phy-
siological side to the appetite or desire prominent in food-
getting and sex, e.g., relatively absent in fear and anger, and
totally absent in the simpler instincts (reflexes) of walking,
standing, grasping, and even in such responses as collecting,
curiosity, and others. This visceral component is repre-
sented in the figure by X. Normally the instinctive behavior
RI is produced by the unconditioned stimulus Si acting alone
or in conjunction with internal stimulus X. In many cases
these afferent conditions also produce visceral effects, IR.
Modifications of this original inherited equipment, so far as
the elements of the neuro-physiological mechanisms are
concerned, may be thought of in any one of the following
ways, or in combinations of these: (i) St acting alone or in
conjunction with X may by use become an effective stimulus
for the responses RI (somatic) and IR (visceral). The
254 WALTER S. HUNTER
organism now fears some new object, has adopted some new
article of diet, or (as is beautifully illustrated for animals
below man in Craig's work with pigeons) has acquired some
new sexual object. The internal appetite is still present, the
responses of the skeletal muscles are unmodified, the visceral
effects underlying the consciousness termed emotion are in
full vigor, only the external stimulus has changed, although
it may have changed to something which no longer suggests
Si to the experiencing subject. (2) Si, in the cases where
by heredity the cooperation of X is necessary to give the
afferent activity control of the final common path to R\ and
IR, may by use secure the power to arouse R\ when X is
absent. Here belong the cases where an instinct is aroused
in the absence of the normal appetite or desire, jaded instincts,
in a word. (3) Modification 2 may occur after S2 has become
the effective stimulus. (4) By use, or through the absence of
the proper S, X may become so vigorous, so intense, so volumi-
nous, that in the absence of an effective S, or even of any
discoverable S, it may secure possession of the final common
path to RI and IR. As examples we may cite: Breed's
chicks, when they gave the drinking reaction in the air with
no observable outside stimulus present; the case of a starving
man or one perishing with thirst who swallows totally inade-
quate and normally non-effective stimuli; unreasoning,
groundless fears; and finally cases of gluttony, alcoholism,
and abnormal sex hunger. (5) Si or S2 may by practice
secure the power to arouse R\ not only in the absence of X
but without involving any noticeable visceral changes, IR.
This is the instinctive behavior devoid not only of normal
appetite but of the normal emotional satisfaction which
accompanies its exercise. Again the most striking illustra-
tions come from the field of food and sex responses. (6) The
modification of the instinct may proceed with Si and X
unchanged but with the response shifted from R\ to R2, —
or from a clumsy and unskilled RI to an efficient performance
of the same response (as, e.g., in Breed's experiments).
Again it should be noted that R2 may be so different from
RI that an observer not knowing the genetic facts would be
MODIFICATION OF INSTINCT 255
unable to detect a relationship between the two activities.
(7) The final case of modification occurs when the effects of
practice, or use, have substituted S2 — jRj for the original
behavior with or without abnormality in X and IR. These
are the typical cases of sublimation; and, if X and IR are
unmodified, they are the cases where the desires and emotions
(satisfactions) of one original response are put at the service
of, or incorporated into, derived forms of behavior. Stated
in this manner and placed in relation to other forms of sti-
mulus and response changes, sublimation loses any mystical
character it may have been thought to include and stands
forth as a peculiarly important type of the modification of
instinct.
II. THE TEMPORAL POSITION OF THE MODIFICATIONS
So far our analysis has concerned those phases of instinc-
tive modification which can be formulated in terms of change
in the elements of the stimulus-response situation. Two
other problems now remain to be emphasized, problems which
although of fundamental importance in the modification and
control of behavior are unnoticed in the social psychologies,
and are at the best treated only by implication by the tech-
nical students of instinct. These problems are: (i) The
temporal position of the modification, whether coming prior
or subsequent to the first instinctive performance; and (2)
modifications of the biological purpose, or end, involved
in the inherited behavior.
The modifications of instinctive performance are not all
variations (of the stimulus, of the somatic response, or of
the visceral response) produced after the instinct first appears.
Instances which do belong here we have already illustrated.
Other modifications occur because of influences at work before
the instinct makes its initial appearance. These changes
will clearly affect the instincts in proportion to the length of
the interval between birth and the instinct's appearance
and in proportion to the social value inherent in a modification
of the instinct in question. The dates and order of the
appearance of the various instincts are sufficiently known
*56
WALTER S. HUNTER
to serve our present general purpose. Figure 2 indicates for
man the early appearance of the responses of feeding, fear,
anger, and vocalization, the final appearance of the sex and
parental responses, and the intermediate appearance of such
responses as play, acquisition, locomotion, construction, etc.
We do not mean to imply by the use of this diagram any more
FIG. 2. Diagram indicating: the early appearance of the instincts of feeding,
fear, anger, and vocalization; the intermediate appearance of such instinctive responses
as play, acquisition, locomotion, etc.; and the late appearance of sex and parental
behavior. The curves for each instinct suggest the appearance of component elements
before the complete instinct matures and is active. No emphasis is to be placed either
upon the relative order of the intermediate instincts or upon the form and length of
the several curves.
than concerns our immediate purpose. The instincts and
instinctive tendencies in man are as a rule too indefinite in
their manifestations to enable a very satisfactory listing,
and the question of their temporal order of appearance is one
calling for much additional experimental work. Particularly
is it important for the problem of the modification of instinct
that the early traces be noted of instincts which appear late.
Our very simple diagram (Fig. 2) enables us to visualize
clearly the possibility of the temporal aspects of the modifica-
tions above mentioned. It also serves to suggest that the
instincts which will be most open to change by virtue of
preexperience will be the ones listed farthest to the right.
Feeding, fear, and anger, e.g., appearing as they do practically
at birth, offer no other possibility than modifications sub-
sequent to their appearance; while the temporal interval
antedating sex, e.g., makes possible the acquisition of many
MODIFICATION OF INSTINCT *S7
responses which will serve to modify (and control) that
instinct.1
Although our chief interest does not lie in the historical
aspects of our subject, it will be worth while to indicate the
types of comment and experiment already available in the
literature so far as they may concern the modification of
instinct by preexperience. We derive the first suggestion
from the writings of Karl Groos on play (1895) where the
following statements occur:2
"... there can be no doubt that instinct plays a part
in all this adaptation for the struggle for life and preservation
of the species, so necessary in man and other animals. Further
... it would be entirely in harmony with other phenomena
of heredity if we found that these instincts appear at that
period of life when they are first seriously needed. Just as
many physical peculiarities which are of use in the struggle
for the female only develop when the animal needs them;
just as many instincts that belong to reproduction first
appear at maturity; so the instinct of hostility might first
spring up in the same manner only when there is real need
for it; and so it might be supposed with other instincts in
connection with related activities. The instinct for flight
would only be awakened by real danger, and that of hunting
only when the animal's parents no longer nourished it, and
so on. In this case it would be necessary for the special
instincts to be elaborated to their last and finest details.
For if they were only imperfectly prepared, and therefore
insufficient for the real end, the animal might as well enter
on his struggle for life totally unprepared. . . .
"Without play practice it would be absolutely indispen-
sible that instinct should be very completely developed, in
order that the acts described might be accurately performed
1 In what follows we shall have much to say concerning sex behavior, but this
must in no way be interpreted as an adoption of the Freudian point of view that sex is
the dominant instinct. Our emphasis upon this response results because (a) of the
strength of its impulse, (b) of the lateness of its appearance, and (c) of the fact that it
assumes a more definite instinctive form than other late instincts.
2 Groos, Karl, 'The Play of Animals.' Eng. trans, by Baldwin. New York:
1898, pp. 73-4; 79.
»58 WALTER S. HUNTER
by inherited mechanism, as is also the case with such in-
stinctive acts as are exhibited but once in a life time."
"... instincts are not so perfectly developed, not so
stamped in all their details on the brain, as they would have
to be if their first expressions were to be serious acts. There-
fore they appear in youth, and must be perfected during that
period by constant practice."
One need not accept Groos's theory of play in order to
admit the essentials of the above quotations. There is a
playful exercise of those elements of an instinct which appear
prior to the complete appearance of the inherited behavior,
and this exercise does, by the law of habit, have an effect
upon subsequent arousals.
Lloyd Morgan, writing in 1900, speaks much more
definitely on our present problem:1
"Even in the case of the very first exhibition of such a
deferred instinct as the moor-hen's dive, although that or-
ganized sequence of acts which constituted the behavior as a
whole had never before occurred, although there was no
gradual learning how to dip beneath the surface, and to swim
under water, still many of the constituent acts had been often
repeated; experience had already been gained of much of
the detail then for the first time combined in an instinctive
sequence. So that if we distinguish between instinct as
congenital and habit as acquired, we must not lose sight of
the fact that there is continual interaction, in a great number
of cases, between instinct and habit, and that the first per-
formance of a deferred instinct may be carried out in close
and inextricable association with the habits which, at the
period of life in. question, have already been acquired."
This point of view Morgan continues to discuss down to
191 2,2 without, however, so far as I am aware, commenting
upon its social significance or attempting any general analysis
of the problem. I can find no discussion of this temporal
aspect of the modification in the current textbooks of psy-
1 Morgan, C. Lloyd, 'Animal Behavior.' London: 1900, p. 106.
8 Morgan, C. Lloyd, 'Instinct and Experience.' New York: 1912, chs. I and 2.
MODIFICATION OF INSTINCT «59
chology, nor in the social psychologies1 and more special
treatises on instinct and behavior. Undoubtedly in the
observational literature on instincts many instances could be
unearthed. We shall cite but one, perhaps the best one,
however, and then proceed with our comments on the general
problem.
This illustration is drawn from the work of C. O. Whitman
on pigeons, and is as follows:2
"If a bird of one species is hatched and reared by a wholly
different species, it is very apt when fully grown to prefer to
mate with the species under which it has been reared. For
example, a male passenger-pigeon that was reared with ring-
doves and had remained with that species was ever ready,
when fully grown, to mate with any ring-dove, but could
never be induced to mate with one of his own species. I kept
him away from ring-doves a whole season, in order to see
what could be accomplished in the way of getting him mated
finally with his own species, but he would never make any
advances to the females, and whenever a ring-dove was seen
or heard in the yard he was at once attentive."
("It may be remarked by the editor that the discovery of
this principle furnishes the key to Professor Whitman's
success in hybridizing the various species of pigeons. A novel
and important principle of behavior is here involved. The
range of stimuli to which an instinctive tendency will respond j
may be modified by habits acquired long before the first *
expression* of the instinct. The first expression of a delayed
instinctive tendency may thus be in part a function of all
that the organism has previously acquired.")
While we cannot agree with Carr that the principle of
behavior involved in Whitman's work is novel, yet we must
agree that it is important far beyond any recognition yet
1 Baldwin skirts the edge of the problem in his account of social heredity as de-
veloped in "Social and Ethical Interpretations," New York: 1906, 4th edition, Pt. I,
ch. 2; but he seems not to have stated explicitly that social heredity may get in its
work on the individual before the instinct (physical heredity) has appeared in that
individual.
* Whitman, C. O., 'Orthogenetic Evolution in Pigeons,' Vol. 3, 'The Behavior of
Pigeons.' Edited by Harvey Carr. Carneg. Inst. Washington, Publ. No. 257, 1919,
p. 28. I have added Carr's editorial comments in parenthesis.
«6o WALTER S. HUNTER
given it. This modification of instinctive behavior by expe-
rience encountered prior to the first appearance of the in-
herited response may be conceived in any of the following
ways: (i) The early appearance of component elements of
the final total behavior (as indicated by the curves of Fig. 2)
may involve their own modificaton on the basis of use in
such a manner that when the total instinctive response ap-
pears it does so in a manner not entirely determined by
heredity. This modification may be either on the stimulus or
on the motor side of the early appearing tendencies, and is in
this respect a case under section I. of this paper. This would
be the type of case covered by Groos's theory of play and by
Morgan's description above quoted. (2) Perhaps indepen-
dently of any early component tendencies of later instincts,
the individual may be instructed in the nature of the socially
accepted stimuli and forms of response so that when occasion
arises he will respond in the socially accepted manner.1
So thorough and far reaching may these modifications be,
that the individual may never know the animal form of the
instinct; and yet we must believe that this persists, in the
form perhaps of synaptic connections, because something
very like it appears when the bonds of social restraint are
relaxed.2
The responses of feeding, fear, and anger, as we have
said, appear too early in the individual's life for this general
type of modification; but such responses as play, constructive-
ness, sex, display, and the parental instinct, occurring later,
offer the individual and the social group an opportunity to
determine prior to the onset of the behavior the stimuli which
1 This instruction which precedes the maturing of relevant interests is undoubtedly
very uneconomical from the standpoint of the laws of learning; but the vital problem
is not the speed and efficiency of the acquisition, it is rather the mere fact of acquisition,
the importance of building up controls while the organism is yet young.
2 It seems hardly believable, in view of this last fact, that any psychologist should
deny that man possesses true instincts. Present-day society has so modified the
individual and his environment, that the individual seldom experiences the sheer animal
form of the response — indeed some may never do so. However, occasionally in
moments of great stress, the individual is literally swept off his feet by a gust of animal-
like passion. Perhaps once or twice I have approached such an experience. My own
testimony would be that in such a case one is for the moment an all but unconscious
automaton.
MODIFICATION OF INSTINCT «6i
shall ordinarily arouse it and the form which it shall take.
Play activities vary in their content in dependence upon the
social environment, as do constructiveness and sex also.
Long prior to the maturing of the latter instinct and even
longer before its usual manifestation, society has set before
the individual a pattern which, like the Great Stone Face of
Hawthorne's tale, shall serve more or less unconsciously to
instruct and guide him in the accepted stimuli and responses
of that behavior. Religious training likewise can, and does
in many cases, take the young individual and so shape his
religious symbols and responses that when religious activities
do appear definitely in adolescence, it shall seem but natural
to turn to one sect or one religion for their gratification.
Society in this type of modification is giving the individual
the benefit of its own experience, not by permitting the
instinct to manifest itself in the crude animal form and then
modifying it, but by building up the proper controls prior
to the emergency.
III. MODIFICATIONS OF THE BIOLOGICAL PURPOSE
The final problem which we have set ourselves is now at
hand. Modifications of the instincts are not only of the
sorts which have been outlined above, but they may involve
essential changes in the biological purpose of the response.
By the biological purpose of an instinct, I mean the adaptive ^
purpose which it secures or tends to secure. Thus the
biological end in view in the case of fear is the removal of the
organism from the dangerous stimulus; in the case of anger,
it is the injury of the offending object; in sex, it is the repro-
duction of the species, etc. There is, as I understand, no
dispute on this point, viz., that instincts are adaptive forms
of response. This statement carries no implication that the
purpose is a conscious one or that it has been instrumental
in molding the behavior. The statement is a straightfor-
ward, scientific, objective formulation, implying nothing of
vitalism or of other speculative interpretations of the place
of purpose in nature.
Inasmuch as the animals below man give as yet no evi-
262 WALTER S. HUNTER
dence of possessing the behavior equivalent of thought
processes, it is a probable assumption that they never possess
an awareness of the purpose of their acts. At some time in
the evolution of man, therefore, the consciousness of the
purpose served by his responses has appeared. At first,
undoubtedly, only the more obvious purposes have been
grasped, such as those found in the protective reflexes,
in hunting, in display, and in parental behavior. Particularly
in the case of the sex instinct there is reason to believe that
the race has only recently, i.e., recently as one estimates
time in terms of man's existence on the earth, discovered the
connection between the sex instinct and reproduction. By
way of illustration, we may quote from Spencer and Gillen's
account of the tribes of Central Australia. Writing in 1899,
they say:1
"... we have found amongst the Arunta, Luritcha, and
Ilpira tribes, and probably also amongst others such as the
Warramunga, the idea firmly held that the child is not the
direct result of intercourse, that it may come without this,
which merely, as it were, prepares the mother for the recep-
tion and birth also of an already formed spirit child who
inhabits one of the local totem centers. Time after time we
have questioned them on this point, and always received
the reply that the child was not the direct result of inter-
course."
Writing again in 1904, they say:2
"Indeed Mr. Roth's latest work in Queensland shows
clearly that the idea of spirit children entering women, and
that sexual intercourse has nothing to do with procreation,
is a very widespread belief amongst the Australian aborigines,
and is by no means confined to the tribes amongst whom its
existence was -first described by us" (p. xiii).
"The cermonies [of initiation] can never have had any
reference directly to procreation, for the simple reason that
the natives, one and all in these tribes, believe that the child
1 Spencer, B. and Gillen, F. J., 'Native Tribes of Central Australia.' New York:
1899, p. 265.
2 Spencer, B. and Gillen, F. J., 'Northern Tribes of Central Australia.' New
York: 1904, Pp. xiii and 330-331.
MODIFICATION OF INSTINCT 263
is the direct result of the entrance into the mother of an
ancestral spirit individual. They have no idea of procreation
as being directly associated with sexual intercourse, and
firmly believe that children can be born without this taking
place. There are, for example, in the Arunta country
certain stones which are supposed to be charged with spirit
children who can, by magic, be made to enter the bodies of
women, or will do so of their own accord. Again, in the
Warramunga tribe, the women are very careful not to strike
the trunks of certain trees with an axe, because the blow might
cause spirit children to emanate from them and enter their
bodies. They imagine that the spirit is very minute, —
about the size of a small grain of sand, — and that it enters
the woman through the navel and grows within her into the
child."
In all cases a definite and accurate formulation of the
adaptive value of the behavior has waited upon a clear
perception of cause and effect relations among objects and
events, which in many cases means waiting upon scientific
analysis. Until the individual and society know the biologi-
cal purposes of instincts, only accident can identify the
purposes which society approves and fosters with those
which heredity is seeking. But once this knowledge is
forthcoming, society and the individual may proceed con-
sciously and definitely to foster the purpose, or they may
change the environment in such a way that the biological
purpose can give way to a new purpose, or, finally, the biologi-
cal purposes may be satisfied incidentally so far as the con-
scious plans of the individuals are concerned.
Nor should the present type of instinctive modification
be confused with the voluntary exercise of a response that
may at times be automatic and inherited. Such a case
would occur when one winks voluntarily at a joke, and so
might apparently be said to have modified the biological
purpose of protection normally subserved by this response.
In order to subsume the winking response under this third
type of modification, the winking would have to be produced
by the individual's voluntarily placing himself in front of a
264 WALTER S. HUNTER
stimulus which would automatically bring about the response
and then for a social purpose which might or might not be
the same as the biological one. Perhaps in the last analysis
so-called voluntary activity is precisely of this nature, con-
sisting of a highly elaborated conditioned reflex whose
stimulus is an idea. But for the purposes of the present
discussion there is an active participation and a feeling of
control in voluntary activity which contrasts strikingly with
the automatic, impulsive, compelling characteristic of the
instinctive response (characteristics which are as definitely
present when the instinct is "used" for social purposes as
they are when it accomplishes purely biological ones).
The two great modifications which have been made in
biological purposes appear to be these: (i) purposes which
are inimical to civilized social life are supplanted by new
and more acceptable ones; and (2) the biological purposes in
all of the more powerful instincts are occasionally or habi-
tually secondary to the use of the instinctive behavior as a
pleasure giver. To be sure, in so far as the original synaptic
connections persist — and it is my opinion that they are
rarely if ever lost — the original biological end of the behavior
will tend to remain and be satisfied, although perhaps only
surreptitiously.
Table I has been drawn up with reference to the two types
of cases suggested above. Here an attempt is made to state
the biological purposes subserved by certain of the instincts
and to place over against these the recognized social purposes
which usually or occasionally dominate them. In some cases
the two will be identical, due at times to accident and at
times to foresight. The principle involved in this third
type of modification of instinct is not dependent for its
validity upon the accuracy of the analysis of Table I.; it is
dependent rather upon the fact of variation between the two
types of purpose whose detailed nature is there suggested.
There are certain features of Table I. which invite definite
comment. In column three I have placed only what have
seemed to be social purposes that are widely recognized in
social practice. No attempt has been made to indicate
MODIFICATION OF INSTINCT
TABLE I
Iniiinct Biological PurpoM
Anger Defense of organism by removing
offending object.
Fear Defense of organism by removing
it from offending environment.
Acquisitiveness. .Accumulation of food and nest
supplies.
Vocalization
.Stimulation of certain instincts
and habits in associates.
Social Purpose Definitely Pottered
*Used in hostility and competi-
tion to stimulate great en-
deavor. Put at service of
customs.
*Used for taboos in maintenance
of social organization.
*Accumulation of objects posses-
sing general value or power to
satisfy human wants. Fos-
tering prestige.
'Communication of ideas; stimu-
lation of any instinct or habit
in self or others.
'Recreation, health, and prestige.
'Domination in all fields of
activity.
'Nourishment, pleasure, and so-
cial solidarity.
Pleasure, and reproduction. Be-
getting of offspring in order
that parents may be cared for
in sickness and old age.
'Protection of young.
'Sex excitant, arousal of fear in
others, prestige, creation of
caste.
Protection from "Great Dan-
ger," protection of morals,
social service.
The * indicates that biological purpose is not specifically combated.
Present occidental society fosters all instincts in some degree for health and
pleasure as well as for the social purposes above enumerated.
the vast multiplicity of purposes for which the instincts may
on occasion be used. With the appearance in man of idea-
tional processes and ideational methods of behavior control,
it has become possible to use the instincts not for their biologi-
cal ends alone but for almost any end that the manipulator
may have in mind. The demagogue and the propagandist
by placing certain stimuli before the crowd may utilize
the resultant fears, angers, acquisitivenesses, or religious
activities to satisfy ulterior purposes of much or little merit.
This is a matter of great social importance, but what we have
indicated in the table differs in at least two vital ways from
Hunting Securing of food and mates.
Rivalry Domination, particularly in sex
and play activities.
Feeding Nourishment.
Sex Reproduction.
Parental Protection of young.
Display Sex excitant, arousal of fear in
others.
Religious Protection
ger."
from "Great Dan-
866 WALTER S. HUNTER
the uses of instinct made by the individual social manipu-
lator. In the first place, the social purposes or utilities
there listed are definitely sanctioned by present occidental
society; and in the second place, the individual in whom the
instinct manifests itself may be, and usually is, well aware of
the social purpose to be attained, inasmuch as much social
or group effort is directed toward instructing him on this very
point.
So far as our analysis can reveal, the social purposes permit
the accomplishment in a more or less incidental manner of the
biological purposes without any attempt to combat these
purposes save in the case of the sex instinct and the religious
tendency. In the hunting instinct, e.g., the purposes of
recreation, good health, and prestige are not incompatible
with the food- and mate-getting end; nor does society repress
the latter. The occasions on which the instinct appears
may be limited by law, but when it does appear the biological
end to be attained is laudable. This is true also in the cases
of fear, anger, and the other responses whose social purposes
are indicated with an asterisk. Society definitely favors the
use of display (in clothing and physical prowess) as a sex
excitant as well as an enhancer of prestige and a creator of
class distinction. The original form of the stimulus and re-
sponse is usually modified, and sublimated instincts may have
been -added to the complex, but when the instinct appears
its biological purpose is approved. In the case of the religious
tendency, on the other hand, society is tending to negate
the biological purpose of protection from great danger or the
mysterious threat (or however one may formulate the unseen
characteristic of objects with which primitive man seeks to
get en rapport. through definite religion and magic).1 In its
place it is putting social service and the maintenance of moral
conduct as the proper goal of the religious impulse. The
change is not that of stimulus and response or of the accretion
of other instinctive impulses alone, nor is it a limitation of the
1 1 do not know, of course, that this is the biological purpose, nor am I certain
that the religious tendency is instinctive. The response is, however, coextensive with
social groups, and the apparent purposes subserved at the lowest level are here stated
as biological.
MODIFICATION OF INSTINCT 167
occasions upon which the impulse may manifest itself. This
is not to say that the use of the religious tendency as a defense
mechanism against the imperfections of the present does not
receive great social sanction; it is to emphasize that much of
the time, and in some groups most of the tim.e, when the
behavior appears, its biological purpose is combated.
Before extending our comments to include the sex instinct,
we may best return and take up the thread of our argument
as left on page 264 where it was stated that the second funda-
mental manner in which biological purposes are modified is
the use of the instinct as a pleasure giver. It should be noted
that with all instincts (not merely with that of sex) there is a
pleasure and satisfaction m the experiencing of inherited
muscular and glandular activity where the experiencer is free
to turn his attention to the response as opposed to the stimulus.
In the arousal of the instinct under conditions that realize or
tend to realize the biological (and certain social) purposes of
the response, the attention of the individual is definitely
focused upon the stimulus which initiates and controls his
behavior. Thus in a fire where the individual is in danger, it
is not the emotional thrill which is in the focus of conscious-
ness but the dangerous aspects of the situation. The by-
standers, on the other hand, who have congregated, can
enjoy the thrills of fear aroused by the fire because in the
background of consciousness is the understanding that, so far
as they are involved, it is all make-believe. It is beyond our
intention to offer an explanation for this enjoyment of inherited
forms of response under the conditions described; it is enough
to indicate the fact and its implications for the modification
of instinct. Within the limits of the apparently harmless,
society sanctions the arousal of instincts for purposes of
pleasure. Forms of art vie one with another in subtle stimu-
lations of the instincts, while in the fringe of the beholder's
consciousness the feeling of make-believe permits him to
enjoy the resultant behavior. The individual confronts
himself upon the stage and the screen with stimuli for all of
the instincts — fear, anger, hunting, acquisitiveness, religion,
sex, etc. — and then enjoys the result much as a child in play
a68 WALTER S. HUNTER
will pretend the existence of hobgobblins in order to enjoy
the thrill of fear, or wiggle a sore tooth or finger for the pleasure
of the resultant pain.1
It so happens that the sex instinct is through heredity
accompanied by a greater pleasure than pertains to the
exercise of any other instinct, and it is therefore not unex-
expected that the history of the modifications of this instinct
should be peculiar. In the animals below man, where there
is no awareness of the biological end, the instinct functions
solely for reproduction. No social purpose exists. The use
of sex for pleasure, so far as I know, has its first beginnings
among the monkeys, although here the probable absence
of thought processes would count against its conscious use
for that purpose. Moreover, among primitive men and even
among peoples as advanced as the Central Australian natives,
the biological purpose of reproduction is unknown (undoubt-
edly because of the great temporal interval between the
activity of the instinct and the birth of the offspring), and
yet there is sufficient development to insure the presence of
definite social purposes. The result is that the sex instinct is
recognized by society as a type of behavior whose purpose is
the production of pleasure. Women are property, and the
violation of chastity is the violation of a property right.
With the development of man to the point where the
biological purpose of sex is understood, comes the possibility
that society and the individual may definitely sanction the
biological purpose. This they have done. Certain indi-
viduals and certain groups have maintained that the only
conscious purpose to be sanctioned is the biological one;
and yet in practice society at the present time sanctions the
modification of 'this instinctive behavior by utilizing it in the
ancient manner as a pleasure mechanism. This it does
through emphasis upon birth control and the make-believe
stimulation of the instinct on the stage and in certain phases
of art in general.
1 G. T. W. Patrick has made extensive use of the pleasurable aspects of instinctive
activities as they appear in playful form. See his 'Psychology of Relaxation.' New
York: 1916.
MODIFICATION OF INSTINCT 269
In the case of the food-getting instinct society and the
individual do not at present combat the biological purpose,
although they do relegate it to the background and satisfy
it incidentally in many cases. In instances of perversion,
however, the nutritive purpose has been definitely combated.
Thus it is said that the Roman voluptuaries practised arti-
ficial vomiting in order that their banquets might proceed
unhindered by the limited capacity of the individual. While
our own banquets lack this interesting feature, nevertheless
they are conducted for pleasure and not for the purposes of
nutrition. Custom has from time immemorial recognized
the effect on social solidarity of "breaking bread" together,
utilizing the pleasurable aspects of feeding in the creation of
consciousness of kind. This and similar uses of instinct to
satisfy social rather than biological purposes is fundamental
in understanding social phenomena.
IV
Summary. — The social significance of instinct cannot be
brought out by analyses of the nature of specific forms of
response, but must come largely from a consideration of the
types of modification that instinctive forms of behavior under-
go. These variations come fundamentally from the influence
of habits and other forms of intelligent behavior. The present
paper has elaborated the topic with reference to the following
points: (i) Modifications of the structural elements, including
(a) changes of the stimulus in its external or internal aspect,
(&) changes of the somatic or of the visceral response, and
(c) combinations of these in sublimated behavior; (2) the tem-
poral position of the modification as it occurs before or after
the initial appearance of the instinct; and (3) modifications
of the biological purpose or adaptive value of the response.
THE NATURE OF THE RHYTHM EXPERIENCE
BY ELCANON ISAACS
University of Cincinnati
Recent years have witnessed a marked decrease in the
number of contributions to the experimental psychology of
rhythm. Dunlap, writing in 1916, remarks, "It is a signifi-
cant fact that experimental work on the perception of time
and of rhythm has nearly ceased. One research on time, a
statistical study of speech rhythm and a minor study on
rhythm effects are all that have appeared in the last two
years" (7, p. 206). With the exceptions of one or two studies,
this statement is still applicable; in fact, the subject of time
and rhythm has been dropped from the reviews in alternate
years.
The reason for the discontinuance of scientific work in
this field is not a knowledge of all the facts about rhythm, for
such is not the case, but the lack of a working hypothesis
for the nature of the rhythmic experience. The tentative
bases which have existed up to this time have been given up,
for the greater part, as the result of a tendency to eliminate
time as an element of rhythm perception, and of the recogni-
tion of rhythm in others than the auditory field. Although
these factors are the necessary result of each other, it is not
until recently that they have been recognized, and that more
than one aspect of the rhythm experience has become subject
to analysis.
The earliest introspective and empirical studies,1 and the
modern theories of literary scholars, although characteris-
1Briicke, 'Die physiologischen Grundlagen der neuhochdeutschen Verskunst',
1871 (5).
Riemann, ' Katechismus der Musik,' 1888, p. I (36).
Lobe, 'Katechismus der Musik,' 25 Aufl., 1893, p. 4 (20).
Sully, 'The Human Mind,' Vol. i, 1892, p. 271 (54).
Lanier, 'The Science of English Verse, '1880, p. 62 (19).
Gurney, 'The Power of Sound,' 1880, p. 127 (12).
270
THE NATURE OF THE RHYTHM EXPERIENCE *7J
tically vague,1 have not seriously questioned the necessity of
an absolute regularity in the recurrence of the objective
elements. The view represented and assumed by them, has
given way in the case of the former, under the influence of
studies pursued under experimental conditions, until in 1903
MacDougall wrote: "There is properly no repetition of
identical sequences in rhythm. Practically no rhythm to
which the aesthetic subject gives expression, or which he
apprehends in a series of stimulations, is constituted of the
unvaried repetition of a single elementary form" (22, p. 319).
Under experimental conditions, rhythm was also removed
from the field of objective stimulation,2 to the field of sub-
jective perception,3 and firually to that of motor experience
in time.4 There has likewise been a tendency to limit the
definition of rhythm greatly, from the cosmic recurrences of
the universe (3, p. 146), to the field of human experience
(38, p. 305), and here to the voluntary as opposed to the
organic rhythms (27, p. 3).
Although most of the experiments with the voluntary
1 Guest, 'A History of English Versification,' 1882, p. I (10).
Schipper, 'History of English Versification,' 1910, p. 3 (42).
Gummere, 'A Handbook of Poetics,' 1885, p. 134 (xx).
Alden, 'An Introduction to Poetry,' 1909, p. 156 (i).
Mayor, 'Chapters on English Meter,' 1901, p. 4 (24).
Omond, 'A Study of Meter,' 1903, p. 2 (30).
Saintsbury, 'Historical Manual of English Prosody,' 1910, p. 291 (40).
Matthews, 'A Study of Versification,' 1911, p. 12 (23).
1 Kostlin, 'Aesthetik,' 1869, p. 90 (17).
Riemann, 'Elemente der musikalischen Aesthetik,' 1900, p. 135 (37).
Cf. Wundt, 'Grundziige der physiologischen Psychologic,' III., 1903, p. 158 (64).
1 Squire, 'A Genetic Study of Rhythm,' A. J. Psy., XII., 1901, p. 586 (50).
Meumann, 'Untersuchungen zur Psychologic und Aesthetik des Rhythmus,'
Phil. Stud., X., 1894, pp. 272, 304 (25).
* Miner, 'Motor, Visual and Applied Rhythms,' PSY. REV. MONOG., V., 1903
(^7).
MacDougall, R., 'The Relation of Auditory Rhythm to Nervous Discharge,'
PSY. REV., IX., 1902, p. 466 (21).
Stetson, 'A Motor Theory of Rhythm and Discrete Succession,' PSY. REV.,
XII., 1905, p. 258 (52).
Ruckmich, 'The Role of Kinsesthesis in the Perception of Rhythm,' A. ]. Psy.,
XXIV., 1913, p. 305(38).
Weld, 'An Experimental Study of Musical Enjoyment,' A. J. Psy., XXIII.,
1912, p. 298 (59).
Patterson, 'The Rhythm of Prose,' 1916 (31).
272 ELCANON ISAACS
rhythms were in the auditory field because of the comparative
ease of experimentation, it is a curious fact that writers of
general accounts and text-books consider rhythm possible
only in certain fields, whereas special investigators are con-
fident that it is theoretically possible in every field. Thus
Miner says: "There would be no reason a priori why a series
of stimuli addressed to any one sense should not produce an
experience of rhythm. I am quite confident that they would
under proper circumstances; that rhythms of smell, taste,
touch or vision are just as possible as rhythms of hearing"
(27, p. 40). So Meumann, with temporal qualifications (25,
p. 261); and according to Wundt, 'no series of impressions
exists that cannot in some way be comprehended as rhythmic*
(64, p. 62). Ruckmich asks whether a rhythm cannot be
induced which shall be perceived principally in terms of
those sensations that correspond directly to the nature of the
stimuli given (39, p. 247); and Woodrow: "To produce an
impression of rhythm, it is necessary to have a series of
stimuli," enumerating some (60, p. 5); and Dunlap: "The
facts seem to be that all sorts of sensations lend themselves
to serial grouping" (6, p. 350).
Such pervasiveness of rhythm, however, is not understood
by writers of general accounts and elementary texts. For
them, rhythm is limited to definite fields of sensation, to the
auditory, kinsesthetic, tactual and visual fields (34, p. 329);
to the first three (29, p. 301); to the first two (8, p. 484);
to the first (18, p. 389); and according to Titchener, there is
rhythm in one and only one field, the kinsesthetic (56, p. 345).
From this, it can be seen that the special investigators did
not hesitate to generalize from their limited results, but that
the general accounts were more cautious, and did not theorize.
Among neither group, however, is there agreement. The
result has been a condition chaotic in the extreme. In 1917
Ruckmich wrote: "Much experimentation has been done
in the last two and a half decades on the general subject of
rhythm. Theories have almost equalled these attempts in
number, and difficulties have arisen out of all proportion to
the facts discovered" (39, p. 326).
THE NATURE OF THE RHYTHM EXPERIENCE *73
THE RHYTHM OF PROSE
In the latter part of 1916, however, appeared a new and
important contribution to the rhythm of prose (31), in which
it is stated unequivocally that a 'new standard' is established
'for passing judgment upon the rhythm of a sentence or
paragraph' (p. 13). Although the title refers to the rhythm
of prose, the book is concerned with the nature of rhythm
in general.
From the example of syncopating rhythms of Indian
music, where there are two levels of rhythmic stimulation,
often in different times, as a melody in three-time against
a tom-tom accompaniment in four-time (p. xx), it is suggested
as the 'new standard,' that all rhythm is composed of two
levels, an objective level and a subjective temporal measuring
scale of 'unitary pulses,' 'elastic* in their nature (p. xx).
"The 'boom! boom! boom!' of subjective time-units, such as
rattle along in the consciousness of an aggressively rhythmic
person, may be accelerated or retarded, within certain limits
defined for each such individual, without destroying their
value as a subjective foot-rule with which to correlate all
experience" (p. 47).
The origin of the subjective time-unit may be the breath-
ing rate. " Segments of breath-waves, each segment marked
by a slight reinforcement in the flow of air, and measured,
in turn, by so many concomitant heart-beats — when these
are consciously felt — may easily register for us our mental
seconds. It is only by such mental time-beats or 'unitary
pulses' that we are able to make anything like accurate
judgments of time. Suppressed articulation usually assists
us in counting; our memory images record the number"
(32, p. 259). Or the origin of the subjective time-unit may
be the walking rate, and the memory of it, the basis of
measuring. 'Perhaps, as each step is taken, the concomitant
tension of some obscure muscle of the head occurs, which
thus furnishes the means of repeating the walking-rate, with-
out carrying the innervation as far as the legs' (31, p. 65).
Both levels, that of the objective stimulation, and that
274 ELCANON ISAACS
of the subjective time-units are characterized by (i) accelera-
tion and retarding (pp. 3, 47), (2) substitution of one long
time-interval for several equivalent short ones, or vice versa
(p. 3), and (3) syncopation, by which is meant the correlation
of two sets of time-intervals, concomitant but not coincident
(p. 4), as in the case of the Indian music mentioned above.
"The impression which results from the combination in con-
sciousness of the auditory (syllabic) sensations (including
their effect upon attention) and the subjective time-units,
may be compared to a melody and its accompaniment, with
attention focussed, not upon the pitch relations so much as
upon the relations of time and stress" (p. 69). The 'new
standard' makes rhythm a temporal affair in that the sub-
jective 'unitary pulses' measure time. "The ultimate basis
of all rhythmic experience, however, is the same. To be
clear-cut, it must rest upon a series of definite temporal
units" (p. xxii).
VISUAL AND AUDITORY RHYTHM
Almost side by side with this study appeared a new con-
tribution by Ruckmich in the field of visual rhythm (39,
p. 231). Miner in 1903 found that "the experience of rhythm
in the field of vision is identical in its essentials with that in
the auditory field. Since the experience is novel, it is at first
more vague than with sounds, but it becomes quite precise
with practice" (27, p. 71). Koffka has concluded that "no
essential difference between auditory and visual rhythm has
shown itself," that 'series of visual imagery can be the sole
associates of the experience of rhythm,' that 'for rhythmical
experiences visual and auditory images are equivalent through-
out' (16, pp. 96-97).
From these and other considerations of visual rhythm
(39, p. 23 2f.), it seems clear that visual rhythm cannot be
placed in a category different than that of the auditory.
"All of the special studies which have been undertaken in
this field are at one in pointing out that visual rhythm does
not differ essentially from any other kind of rhythm" (p. 236).
Nevertheless, Ruckmich has found that a purely visual
THE NATURE OF THE RHYTHM EXPERIENCE «75
rhythm can be experienced as such (p. 247), that "it is
possible to obtain rhythmical perceptions from stimuli that
are visually presented and that differ objectively only in
color quality. It is furthermore possible to obtain from such
stimuli experiences of rhythm which are visual in their very
essence, i.e., in which no other processes play an important
part" (p. 253).
TIME AND RHYTHM
Although it may not appear so at first sight, there is a
distinct conflict in the field of rhythm between the results
of Patterson and Ruckmich. It is in the application of the
'new standard' to visual rhythm that difficulty arises. It is
not clear the part that 'unitary pulses of subjective time'
whose rate is 'about .7 sec.' (31, p. 67), would play in a
rhythm induced in terms of stimuli that are not auditory,
but visual. Visual rhythm can result from stimuli in terms
of difference of color quality (39, p. 253), of difference of
intensity of the members (38, p. 356), and what is of greatest
importance, in terms of spatial structure,1 as well as from the
duration (27, p. 71), and temporal arrangement of the mem-
bers (16, p. 104). Because of the nature of rhythmic stimula-
tion, time is an element but it is ancillary to the rhythm.
It is a prominent item in auditory rhythm, since, because of
the nature of the end-organ, only series of discrete stimuli
can be presented. Visual rhythm, likewise, is temporal in the
aspect of serial stimulation, in that, since two stimuli cannot
occupy the same place at once, one must follow the other in a
time sequence.
While, therefore, it is true that recurrence is present in
rhythm, and takes place in time, it does not follow that the
perception of rhythm is due to a measurement of the recur-
rence by a 'subjective foot-rule.' Recurrence measured by
time as an integral part of rhythm involves a confusion of
one of the physical factors of rhythm with the nature of
rhythm. According to Squire, 'Temporalness, in its connota-
1 Mcumann, op. cit., p. 262 (25).
Koffa, op. cit., p. 97 (16).
Miner, op. cit., p. 43 (27).
276 ELCANON ISAACS
tion of regular succession, is the basal principle of rhythm.
This, however, is quite another thing than saying that the
character of the grouping is dependent upon the time order"
(50, p. 541). As Stetson points out: "The time judgment is
much too vague to determine rhythmic intervals, and accurate
judgments of time founded on rhythms are secondary and
derived" (51, p. 258).
If absolute periodicity were indispensable in rhythm, a
subjective temporal measuring scale would be necessary to
ascertain whether rhythm were present or not. It has been
shown, however, that regularity of recurrence is not char-
acteristic of rhythm (22, p. 319). What function, then, can
a temporal measuring scale have, unless to show that the
recurrence is irregular? Yet regular recurrence is not of
itself felt as unrhythmical. Nor can the time 'pulses' be a
measuring scale as to whether the rhythm can be attended to.
The action of attention is organic in its nature; in any case,
its duration need not be subjectively measured. The only
function a temporal measuring scale could have would be
as a test for the presence of rhythm, and then only for
irregular rhythm. An illustration may make the relation
clear. A clock is a temporal measuring scale for the recur-
rence of night and day. But night and day are not dependent
on the clock for their recurrence. Rhythm takes place in
time, but time is not rhythm. According to Brown: "The
regularity of the motor performance and the equivalence of
the resulting feelings lead naturally to the introduction of
the impression of temporal regularity; but that impression
is really subsequent to the rhythm itself" (4, p. 44).
There is a further objection to a rhythm based on time
measurement as determined by the 'new standard.' Patter-
son says that on the appearance of objective stimuli 'sufficient
in number to suggest serial grouping' there is an adjustment
of the 'pulses' by means of instinctive processes (31, p. 66).
This would mean that a constant experience of rhythm would
continue during the greater part of waking consciousness from
presented stimuli, whether organized or not. This is not the
case; the experience of rhythm is unique and unmistakable.
THE NATURE OF THE RHYTHM EXPERIENCE a 77
Finally, visual rhythm may take place in space. "The
perception of rhythm may be aroused by visual impressions,
whether by simple series of discrete stimuli, presented under
laboratory conditions, or by the sight of rafters on a corridor
ceiling, or of the recurring ornaments of a facade" (56, p. 345).
Miner says, "There seems to be good reasons for believing
that the perception of groups among repeated decorative
figures, lines, etc., is a real rhythmical experience depending
upon the repetition of a like accompaniment of strain sensa-
tions" (27, p. 43). Lastly Koffka found that visual rhythm
may "easily assume, especially in the higher groups of beats,
a spatial structure" (16, p. 97). "We will merely say this,
that according to the result of this study, we can speak about
rhythm in space just as in poetry or music. Aside from
motor and auditory impressions and independent of them,
purely visual impressions can serve as the sensory basis of
group formations, and as a starting point of that inner
activity which conditions accent and thereby rhythm. In
the space arts, we find this realized in the repetition of an
ornament to a considerable extent. The eye moves along
and keeps meeting the same forms, and in this way, rhythm
arises" (p. 109).
If visual rhythm differs from the auditory so as to have a
different basis for its experience, it will be necessary to assume
a heterogeneous nature for rhythm. Many factors point,
however, to its homogeneity. First of these is the pre-
dominant presence in both of kinsesthesis (39, p. 249). Then
there is the appearance, in both, of the secondary character-
istics of rhythm, accentuation and grouping (p. 245). There
are also similar illusions in visual rhythm, "apparently the
same as those that have been noticed by other observers
for sounds. They include the lengthening of the interval
between groups, the intensive accent, and the shortening
of the time between unaccented units in the three-group"
(27, pp. 67-68). Ruckmich says, "Many of the phenomena
which accompany other kinds of rhythm manifested them-
selves. Intervals were under- and over-estimated; attributes
were subjectively assigned to the members; subjective rhyth-
278 ELCANON ISAACS
mization occurred; and redistribution of the groups was
common" (39, p. 254).
KlNJESTHESIS IN RHYTHM
Tentatively eliminating time as an integral part of the
rhythm experience, we must return to the predominant
elements which are characteristic of rhythm in general.
Titchener, believing that rhythm was homogeneous in its
nature, had come to the conclusion that there was one and
only one rhythm, the ultimate basis of which was kinaesthetic.
"The author was formerly disposed to attribute a separate
rhythmical perception to hearing, but recent observation has
convinced him of the existence of kinsesthetic sensations due
to the contraction of the tensor tympani of the middle ear"
(56, p. 345). As to rhythm aroused by visual impressions, "in
the author's opinion, this rhythm is always kinaesthetic, based
upon eye-movement, upon slight movements which tick off
the successive impressions, or upon some other form of inter-
mittent kinsesthesis" (p. 345).
The importance of the kinaesthetic factors had, however,
been pointed out as early as in Bolton's work. "Each im-
pression as it enters into consciousness tends to find expression
in a muscular movement ..." (3, p. 325). Even before,
Gurney, than whom there were few keener observers, had
called attention to them — "let the sounds become regular,
and instantly the impulse comes to tap the hand or move the
foot concurrently with them" (12, p. 128). They have been
pointed out numerous times since. "By far the greater
number of investigators and systematic writers on the subject
of rhythm emphasize the primary importance of kinaesthesis
and of motor response in rhythmical perceptions" (38, pp.
308-9).
Ruckmich, however, who says "in point not only of
frequency of occurrence but of the importance of the part
played, motor factors are almost indispensable items in the
rhythmical consciousness" (39, pp. 246-7), states that after
rhythm is initiated it may continue in the absence of kinaes-
thesis (38, pp. 342, 359). Bolton, however, says that if
THE NATURE OF THE RHYTHM EXPERIENCE «79
movements were attempted to be restrained in one muscle
they were likely to appear elsewhere (3, p. 234). Meyer
(26, p. 37) and MacDougall (21, p. 466) found that the
activity need not be visible in order to give feelings of move-
ment, and Miner states that one subject who gave 'no re-
sponse whatever to the metronome beat with her hand, head
or body' showed considerable reflex response to the beats
under hypnosis (27, p. 2fi.}. Lastly Weld found "when
actual movements were inhibited one of three things usually
occurred. In some cases the rhythmic effect was decreased;
in others a tendency to movement appeared in some other
part of the body; or, again a motor image or a visual image
served as a substitute for the actual movement" (59, p. 265).
We can therefore disregard Ruckmich's statement that
rhythm may continue in the absence of kinaesthetic processes.
Mere kinaesthesis, however, as Titchener thought, is not
sufficient of itself to explain rhythm. It limits rhythm to the
kinaesthetic field as Patterson's theory limits it to the auditory
field. Ruckmich has demonstrated a visual rhythm distinct
from both. In addition, as Ruckmich has pointed out, the
kinaesthetic factors although most prominent in the organs
to which the stimuli are directed, are present in other fields
as well as in the field of stimulation (39, p. 246), and the
clearest part of the perception are the sensations and images
corresponding to the stimuli given (p. 254). It is for this
reason that he rejects the kinaesthetic basis of rhythm.
Since, however, the kinaesthetic factors are present in all
rhythms, it may be well to inquire into their nature. Prac-
tically no work has been done on the nature and degree of
stimulus as affecting the motor response in rhythm. The
contributions which treated of the latter simply recorded
they were present and made little or no attempt to localize
and measure them as to comparative rate, intensity, quality,
or to differentiate rates as manifested through different organs.
Bolton suggested there must be different degrees of muscular
activity depending on intensity of stimulus (3, p. 235), and
Weld recorded larger muscular movements corresponding to
the musical phrase (59, p. 266). Accurate work has, how-
«8o ELCANON ISAACS
ever, been done on the reflex response in another connection,
and a grading of intensity found (47, p. 71). But the most
important item and the basis of the rhythm experience is
found in the following: "The rhythm of the reflex has prac-
tically the same frequence whether the reflex be excited
strongly or feebly: thus, whether the amplitude of the con-
tractions be great or small, they recur with practically the
same frequence" (p. 122).
THE BASIS OF RHYTHM
A consideration of the essential elements in the various
theories of rhythm formulated from Bolton to Patterson
will show that a single hypothesis was in every case at
the basis of their demonstrations. Preliminary experiments
show that the motor response except for simple forms and
certain rates of rhythmic stimulation, is independent of the
rate of the stimulus. This has been widely recognized, but
not acted upon.1 It is clear that the reflex response in rhythm
'represents a relatively undifferentiated type of reaction' in
response to stimulus (21, p. 474), but there are several
elements of reflex response which have been overlooked when
it was advanced as an explanation of rhythmic activity.
First of these is that in the reflex arc conduction, as shown
by Schafer, the rhythm of the discharge of the motor cell is
totally different from that of the action induced in the
afferent cell by stimulation (41, p. 613). In the nerve trunk
conduction, on the other hand, there is a close correspondence
between the rhythm of the stimulus and the rhythm of the
end-effect. It was this latter correspondence which was
erroneously made the basis of the motor theories of rhythm.
But if, on the contrary, the motor reaction is a serial reflex
response resulting from stimulation, the rate of response of
the organ cannot depend on the rate of the stimulus given.
This is the point of departure from all theories of rhythm
heretofore.
1 MacDougall, PSY. REV., IX., p. 474 (21).
Stetson, Harv. Psy. Stud., I, 1903, p. 458; PSY. REV. MONOG., IV., 1903, p. 458
(Si).
Sherrington, op, cit., p. 71 (47).
THE NATURE OF THE RHYTHM EXPERIENCE} »8i
It has long been recognized that each organ has a rate of
response characteristic and constant for its particular activity.
This rate has usually been called the 'natural rate.'1 Scrip-
ture defines it as the rate in which one 'can perform the
greatest number of movements with the least fatigue' (44,
p. 181), and cites the 'route' step on long marches where
each man chooses his own step (45, p. 107). Smith says that
every one has his own rate which is variable within set limits
(48, p. 82), and Patterson tries to correlate the different rates
for various activities in the case of each individual (31, p. 148).
As Weld found, 'we estimate tempo in terms of our momen-
tary ability to make that motor response which seems to be
most fitting for the particular composition which constitutes
our stimulus' (59, p. 268), and according to Squire, the
'natural rate of the individual' is the basis of the pleasantness
of rhythm (50, p. 588). Scripture, furthermore, has shown
that the natural rate varies with practice, fatigue, time of day,
general health, and external conditions of resistance (46, p.
528). The determining elements of this rate are certain
structural and physiological factors.
STRUCTURAL FACTORS
The rhythmical reflex, the response sometimes resulting
from continuous stimuli, as, for instance, the scratch reflex,
is subject to a certain periodicity in its functioning. It is
not only at a frequency independent of the rate of stimulation,
but does not change for various modes of excitation, for
grouped succession of stimuli, or for variation in the intensity
of the stimulus (47, p. 45f.)« In other words, its period of
vibration is constant (p. 122). The rhythmical reflex because
of its periodicity may be said to be pendular in its character.
Confirmatory of this is the periodic nature and constancy of
the rate of response (p. 122), its tendency toward regularity
regardless of the number of stimuli (41, p. 613), its inde-
pendence of the tempo of the rhythm and the amplitude of
1 Stevens, 'On the Time-sense,' Mind, XL, 1886, p. 393 f. (53).
Scripture, 'The Law of Rhythmic Movement,' Science, IV., n.s., 1896, p. 535 (43).
Scripture, 'The New Psychology,' 1899, p. 181 (44).
282 ELCANON ISAACS
the movement (47, p. 122). In terms of the law of the
pendulum, the amplitude of the reflex varies directly with
the nature of the stimulus, but the period of vibration char-
acteristic of the organ remains constant.
The periodicity of the rhythmical reflex is bound up with
another aspect of reflex movement, i.e., the refractory phase
of muscular contraction, the period in which stimuli are with-
out effect. In all reflexes which are rhythmic and not tonic
in their nature, the refractory period is of importance for the
maintenance of the movement. "The reflexes of which the
refractory phase constitutes a prominent feature are those
concerned with cyclic actions occurring in rhythmic series;
such as the scratch-reflex, reflexes of swallowing and blinking,
and probably the rhythmically recurring reflexes concerned
in the stepping of the limbs" (p. 97).
It has been shown that the maintenance of the organic
rhythms over long periods is due to the refractory phase of
the muscular contraction, and that in voluntary movements,
if a sufficient interval is allowed between the contractions,
no fatigue is apparent (13, p. 49). A similar phenomenon is
observable in the nerve cell (p. 139). There may be a relation
between the pendulum rate of response and the refractory
phase of the nerve cell. It is not, however, the rate of the
nerve impulse which determines the rate of response, inasmuch
as the different organs are subject to wide variations depen-
dent on their structure. Furthermore, as Sherrington has
shown, when one group of motor cells, that of the scratch
reflex, is stimulated to produce a weak reflex, and another
distinct group, that of the flexors of the hip, is then stimulated
alternately with the first, although the second group can
respond to a quicker rhythm than that of the first, neverthe-
less, the rhythm appears of greater amplitude, but un-
quickened and unaltered, without even a break or interference
in it (47, p. 122).
In addition, because of the pendulum nature of the
response, increase in the intensity of the stimulus does not
affect the rate of the rhythm or the length of the refractory
phase. "Increase of intensity of the reaction does not show
THE NATURE OF THE RHYTHM EXPERIENCE 283
itself in increase in frequency of the rhythm of the reflex, or
shows itself very slightly in that form, the refractory period
being hardly curtailed at all. The increase reveals itself as
greater amplitude of the individual beat of the rhythmic
contraction. . . . The beats in response to a strong stimulus
may have six times the amplitude of those evoked by a weak"
(P- 70.
The motor response in rhythm, since it is also a reflex
response, and operates through the same elements and ex-
ternal factors, would tend to show the same characteristic
of regularity as the rhythmical reflex. The rhythmical reflex
may be identified with the reflex response, with the difference
between them that the latter is simpler and uncoordinated
in its character; for many forms of stimulation, as some
musical stimulation, are continuous, yet give rise to a reflex
response, while, on the other hand, the rhythmical reflex is
not affected by grouped succession of stimuli (p. 48f.).
Whether or not identical, the same periodicity has been
found to govern the reflex response. As in the case of the
rhythmical reflex, the basis of the rate and its regularity is,
to a large degree, the result of the mechanical factors involved.
Among these, the most important is the length of the member
and of its parts; as in walking, the rate of time varies inversely
with the length of the limb (41, p. 270). Wundt speaks of
the principle of the isochronisms of like amplitudes of the
limbs, and defines rhythmical movements as ones in which
the voluntary energy of the muscles is operative only so far
as is required to set the limbs oscillating in their joints and to
maintain the movement (63, p. 174^)- Miner recognized the
importance of this factor — "it must be some structural ar-
rangement of our body by which a series of like impressions
diffusing to the muscles produces not a separate wave for
each impulse, but a longer wave corresponding to a group of
impulses" (27, p. 34).
The structural element, here likewise, results in a pendulum
movement. Stetson has described the mechanism which
gives rise to the rate of muscular response as a contraction-
relaxation process working between the positive and negative
284 ELCANON ISAACS
muscle-sets of the limb (52, p. 268f.). As in the case of the
pendulum, the limb is carried past the point where the
generating force is lost by momentum alone (p. 262). The
force in the case of the limb is the contraction of the muscles
involved. "Thus the limb is thrown back and forth, and
caught in turn at the limits of its movement by the positive
muscle-sets" (p. 262).
It seems clear from the governing effect of the structural
factors and their relation to the refractory phase of contrac-
tion, that the periodic reflex response, like the rhythmical
reflex, is not coordinated with the objective stimulation, but
is dependent on the pendulum rate of the member of response.
This is further indicated by the periodicity and constancy of
its character (p. 263-4), ^ts independence of the rate of
stimulation and of the variation in the number of stimuli
(27, pp. 36—37). Stetson found: "An obstacle against which
the limb strikes does not affect the character of the movement;
at the end of the normal interval the negative muscle-set
contracts and withdraws the limb, as if the limb had shot to
the end of the course unimpeded" (52, pp. 263-4).
The fact that the basis of rhythm is motor response,
and that this motor response is periodic in its nature, and
similar to the pendulum in its movement, leads to the thesis
that rhythm may be defined as the experience arising from the
periodic, pendular, reflex response of characteristic organs to
objective stimulation.
This definition has been the foundation of so many
theories and systems that it is difficult to see why it has not
been formulated before. Each of the various theories which
were scientific in their nature, recognized one or more of the
elements, but "were inadequate because they made the one
element unduly prominent because of the type of apparatus
employed, the sole basis. Bolton recognized that regular
muscular response resulted from stimulation, but because of
the key-board arrangement which he used, thought it was
to every stimulus (3, p. 235). Miner saw this was not the
case, that these movements were 'something more than one
respnose to each stimulus' (27, p. 30), but because he found
THE NATURE OF THE RHYTHM EXPERIENCE 285
that a single response took place for a group of stimuli in the
case of the metronome, assumed therefore that the muscular
response was the basis of grouping (p. 12). Both are correct
in what they saw, but neither explanation is the basis of
rhythm. Stetson found that the duration of the muscular
contraction was 'strikingly uniform,' and 'independent of
either the tempo of the rhythm or the length of the stroke*
(52, p. 261), that an obstacle against which the limb strikes
does not change the normal interval or the character of the
movement (p. 263), but he explains this on the ground of
experience. "It is experience alone which teaches us to
guide the ballistic stroke" (p. 263). To Patterson, 'syncopa-
tion' was emphasized; it is the basis of his rhythm — synco-
pation between the objective stimulation and the 'unitary
pulses of subjective time' (31, pp. 4, 67).
Each of these contains an element of the nature of rhythm,
but each is a theory of the apparatus. Bolton used a key-
board arrangement; Miner used a metronome; Stetson used
a baton; Patterson saw the Indians dance; and all made the
peculiarities which were emphasized, the basis of a system.
Only MacDougall recognized that the muscular response was
independent of the rate of stimulation, and yet not limited
to one organism (21, pp. 466, 474), but he thought the recur-
rent stimulation exerted an inhibitive influence 'if its periodic
phases are in opposition to those of the natural rhythm of the
sensori-motor process itself (p. 474). He also believed that
in addition to the reflex response there was a physiological
rhythm 'in the functioning of the central nervous system,' —
'functional facilitation and reflex motor discharge, I conceive
to be represented in the conditions which support the impres-
sion of rhythm' (p. 466). The first factor, the possibility of
lack of coincidence of stimulus and response, was also recog-
nized by Stetson: "What happens when a sound occurs
out of place, early in the phase of relaxation, or just before
or just after the climax of the contraction phase? Does it
make it impossible to establish the coordination, or destroy
it if already established?" (51, p. 458).
It is evident that at certain rates there may be two
286 ELCANON ISAACS
opposing tendencies, that of the periodic, pendular response,
and that of the rate of stimulation. Weld found that when
music seemed too fast, it was 'too fast for the particular
motor reaction which seemed most natural to the observer'
(59, p. 267). When not of identical occurrence, but when
within favorable limits, the response may tend to approximate
the rate of stimulation. Stevens found that intervals of a
subject who beat time to a metronome, and continued after
the metronome had stopped, agreed only when a particular
interval was used (53, p. 401). Even here, however, there
may be a gradual divergence, which, when it becomes appre-
ciable, requires an adjustment. "The introduction from
time to time of a single extra tap, with the effect of transposing
the relations of the motor accompaniment to the phases of
the metronome, has been here interpreted as arising from a
periodically recurring adjustment of the reaction process to
the auditory series which it accompanies, and from which it
has gradually diverged" (22, p. 338).
When there is regular presentation and regular reflex
response, if a favorable organ is available, there will result a
correlation between the response and a certain number of
stimuli. But the relation may not always be an essential
one, and even Miner, who has made this relation the basis of
his whole system, says, 'the length of the group does not
increase proportionately to the number of elements in it,'
and farther on, 'we know that the same individual varies
greatly in the length of the group he chooses' (27, pp. 36-37).
Due to the many forms of reflex response, there is no one
unit 'to correlate all experience,' no one basic rate of measur-
ing as Patterson has ascribed to the walking rate (31, p. 64).
Miner recognized that it is misleading to claim there is a
'standard length of group or that the normal group depends
on respiration, fatigue or any particular physiological rhythm
as determining its natural length' (27, p. 39). The walking
rate which Patterson uses as the basic rate for the whole
rhythmic experience, is only one manifestation of the motor
response and is dependent on the pendulum rate resulting
from the factors involved in walking, just as the nodding rate
is dependent on the factors involved in moving the head.
THE NATURE OF THE RHYTHM EXPREIENCE 287
There is still another characteristic of reflex action which
influences rhythm, the after-discharge. It is usually a tetanic
contraction after the cessation of stimuli, and is affected by
increase in the number of stimuli and increase in the intensity
of the stimulus (47, pp. 28, 30). The after-discharge throws
some light on the pause in verse. In the rhythmical unit in
verse, it has been found that the final element has greater
duration and intensity than the other elements. Thus Snell
records that the word or syllable in a verse immediately
preceding a pause is marked by greater duration and probably
intensity (49, pp. 39, 47). The discharge of an unusually
strong impulse leaves the nerve cell exhausted and a certain
time to be recharged is required. So Stetson found that in
lyric verse, the verse pause was from one fourth to one third
longer than the foot pause (51, p. 443), and that since the
end of the verse is the natural climactic position, rhyme was
also preferable at the close (p. 429). Snell found that in
lyric verse, the end-pause is twice as long as the internal pause.
In some verse rhythm, the rhythmic unit is also dependent
on the summation of effect, when there is not a complete
relaxation after each response (cf. infra).
THE INITIATION OF RHYTHM
In the initiation of rhythm, according to Patterson there
is an adjustment by means of instinctive processes of the
elastic unitary pulses and the objective auditory stimuli,
sufficient in number to suggest serial grouping (31, p. 66).
Practically all other investigators consider the kinsesthetic
processes the basis of the initiation of rhythm. Ruckmich
thought the kinaesthetic processes were references for its
interpretation when it is first heard (38, p. 351), and that in
the initiation of a difficult rhythm they may be even the most
prominent item (39, p. 247).
The initiation of rhythm it is suggested presents the
following phenomenon. Unless there is a preconceived pat-
tern of response, stimuli not strong enough of themselves to
evoke a reflex response may, when repeated, result in a sum-
mation of stimuli and produce contraction. This continues
288 ELCANON ISAACS
till the pendulum rate is organized, and the adaption to the
refractory phase established. The organizations of most
poetic meters is on a basis of the reflex pattern. If, however,
irregular stimulation in the absence of a pattern of response
is presented, confusion results at first, and until the adjustment
is made; or if it is never made, or if the recurrent stimuli are
too frequent and intense, the rhythm is never initiated.
Ruckmich states: "Should the rhythm be more than
moderately difficult, and should it, therefore, not become
definitely fixed, or should the mental set of 0 be such that
he cannot make the rhythm 'fit in,' the pleasant affection
may never be reported, and strain sensations may continue
in a vague degree to the end. . . . Then, ordinarily, sensa-
tions of strain gradually die away, attention drops in level,
kinsesthesis grows less intensive and extensive, and finally
vanishes completely or becomes irrelevant to the rhythm.
The rhythm is heard merely in terms of auditory perceptions"
(38, p. 342). Scripture found that in the beginning of an
experiment on a rhythm with a new period, the subject is
quite at a loss for a few beats and can tap only spasmodically
until he obtains a 'subjective judgment' of the period (46,
pp. 527-8). Smith says, "It is doubtful if a rhythm is really
perceived before a certain degree of facility or skill in the
movement is attained" (48, p. 289).
The reflex pattern may result through the pendulum rate
asserting itself, or through the establishment of the latter
by presented schedules in its terms. Patterson found that
on the first hearing a large number of observers found all the
records which he used elusive and more or less irregular
(31, p. 2). When organization through schedules was pre-
sented, various degrees of satisfaction were obtained (p. 64).
Even when there is a schedule, however, confusion may result
when there is a maladjustment of the pattern to the stimula-
tion, as in an attempt to read anapaestic meter when the
motor response is adjusted to iambic. Wallin found that
schematic arrangement was an aid to such an extent as to
differentiate prose and poetry (57, p. 64).
The case of involuntary movement is interesting in this
THE NATURE OF THE RHYTHM EXPERIENCE 289
connection. Miyaki found that "arhythmic movements have
a constant tendency to become rhythmic, notwithstanding
the voluntary effort of the subject to execute the movements
at irregular intervals. The subjects of the experiments in-
variably agreed in confessing that the arhythmic tapping
required strenuous effort and that the performance was very
fatiguing" (28, p. 4). Voluntary irregular movement necessi-
tates a disturbance in the refractory phase involved and the
pendular aspect of movement.
SYNCOPATION
The phenomenon of syncopation to which Patterson has
drawn experimental attention, "in itself, involves a complex
of mental processes. The most essential part of the phe-
nomenon seems to be that we keep our impression of a series
of subjective time-intervals, regular, accelerating or retarding,
but find a pleasure in marking the beats objectively, either by
different forms of motion, such as foot-taps alternating with
hand-taps, or by what appears at first as omission of objective
marking for certain beats. As a matter of fact, this is usually
nothing but the interpolation of some concealed form of motor
reaction such as eye, throat, tongue, or breath movement,
which alternates with a more visible movement, such as
nodding or tapping or dancing" (31, p. 4).
Stetson has described it in much the same terms. "Along
with this precision of all the movements comes a tendency to
beat a new rhythm. This accompanying rhythm is simpler
and broader in character; it is a kind of long swell on which
the speech movements ripple. This second rhythm may
express itself in a new movement of hand, head, foot or body;
when it has become more conscious, as in patting time to a
dance or chant, it develops complicated forms, and a third
rhythm may appear beside it, to mark the main stresses of
the two processes. The negro patting time for a dance beats
the third fundamental rhythm with his foot, while his hands
pat an elaborate second rhythm to the primary rhythm of the
dancers. . . . This regulation of the movement by the co-
incidence of several rhythms is the cause of the striking
regularity of the temporal relations" (51, pp. 465-6).
290 ELCANON ISAACS
In Patterson's definition (31, p. 4), syncopation is appar-
ently manifested by the performer of the rhythm. Syncopa-
tion is used by Patterson in three senses, (i) as any full motor
response (p. xix), (2) motor response in the performer of the
rhythm (p. 4), (3) a correlation of the 'unitary pulses' and
objective stimulation in the observer (p. 91). This analysis
has shown that while there is coexistence, there is not neces-
sarily correlation and rarely coincidence of the objective
stimulation and the reflex response. Syncopation in the
third sense exists, but it is limited to a comparatively small
field of rhythm.
Full motor response is not so evident in modern rhythm.
As Patterson says, "Modern sophistication has inhibited
many native instincts, and the mere fact that our conventional
dignity usually forbids us to sway our bodies or to tap our
feet when we hear effective music, has deprived us of un-
suspected pleasures" (p. xix). Patterson concludes: "What
is left, then, but to conclude that the sentence which has in
its structure the possibility of a maximum of rhythm must
be capable of evoking in us a maximum of motor response?
To test it, therefore, we must tap to it, nod to it, walk to it,
sway to it, chop wood to it, if necessary. . . . If it is easy for
us to nod or tap, or, for that matter, hoe potatoes to these
salient 'drum-songs' . . . the first degree of rhythmic excel-
lence is obtained" (p. 15).
The contortions of the polar bear which Patterson has
called 'prose' merely present syncopation of the muscular
responses of various organs due to the pendulum rates of the
organs. They are not, however, 'harmoniously but intri-
cately regulated by the incessant unitary "flap! flap! flap!"
of those great white feet' (32, p. 261). Each is as independent
in its own sphere as the walking movement is in its sphere.
The large body of literature on rhythm, then, is not invali-
dated by the 'new standard.' On the contrary, it is enriched
by the hitherto experimentally unrecognized field of syncopa-
tion.
THE NATURE OF THE RHYTHM EXPERIENCE 291
THE FACTORS OF ATTENTION
Although rhythm is intimately bound up with attention,
the unsatisfactory state of knowledge about the latter pre-
vents a wholly satisfactory correlation of the two. Ruckmich
says that during the rhythm experience attention is at a
high level (38, p. 342), and he believes that there is a typical
rhythmical consciousness (p. 341). Bolton ascribes grouping
and accentuation to a 'sequence of acts of attention' (3,
p. 21 1), and with this position there is substantial agreement.
Squire says: "One group corresponds to one pulse of atten-
tion, and the regularity of the subjective rhythm is due to the
regularity with which the pulses of attention succeed one
another" (50, p. 575). MacDougall posits a kinaesthetic level
due to changes in attention, 'those elements which are em-
phasized being likewise more clearly attended to' (21, p. 468).
Meumann says that rhythm may be regarded as an unlike
energy division of the attention, an alternation of attending
and not attending (25, p. 304). Arps and Klemm found that
the greatest degree of attention occurs at the accented sound
and the least at the second unaccented sound (2, p. 5181.).
Rhythm was at one time thought to be solely a matter of
attention (35, p. 164; Cf. 34, p. 330).
In repeated stimulation resulting in rhythm, it is clear
that there are two kinds of presentation, regular and ir-
regular. There is no essential difference between them from
the point of view of rhythm, other than that of degree.
There are, however, two classes of stimulation which are
different in their nature, that objectively accented and pos-
sibly grouped, and that undifferentiated. An example of
the first class is most poetry or music; an example of the
second class is the ticking of the metronome or the puffs
of a locomotive. Neither grouping nor accent, however, are
necessarily a part either of the objective stimulation or of
the periodic response.1 Neither are present in organic rhythm
nor at certain rates in presented rhythm. It is evident that
1 Cf. Patterson, op. cit., p. 4 (31).
Squire, op. cit., p. 540 (50).
.Wallin, 'Experimental Studies of Rhyrthm and Time,' PSY. REV., XIX., 1912,
P- 297 (58).
292 ELCANON ISAACS
they do not cause nor are they the result of the periodic
response. Grouping and accentuation, it is suggested, are
the result of the organic rhythm to which attention itself is
subject (V. 33, p. 70). Wundt in speaking of the periodic rise
and fall of attention says it may become regular in its periods
when there are special considerations favoring rhythmical
succession (63, p. 255). Titchener says: "As for the effect
of the anticipatory image, it is clear that, the more nearly
the excitation correlated with the given stimulus coincides
with a psychophysical excitation already in progress, the more
easily will it make its way within the nervous system and the
more dominant will it become" (55, p. 205).
The results of attention also appear with undifferentiated
stimulation and give rise to accentuation and grouping. It
is in the case of undifferentiated stimulation that the verifica-
tion of the suggestion must be found. "It is the fact of
periodical differentiation, not its particular direction, which
is important. Further, as we know, when such types of
variation are wholly absent in the series, certain elements
may receive periodical accentuation in dependence on phases
of the attention process itself, and a subjective but perfectly
real and adequate rhythm arise" (22, p. 320).
The operation of rhythm can be thought of as, on one
part, objective stimulation, regular or irregular; on the other
part, regular serial reflex response. Bridging the two is
attention, which acts in its own way. Rhythm arises from
the reflex response; accent and grouping are the result of
attention.
Supplementary evidence that this is the case is furnished
by the illusions of the durations of the undifferentiated
member and its contiguous intervals. "The effect of both
intensity and duration in rhythm may be generalized as
follows. If every second or third sound is made more intense
or is made shorter, the effect on grouping is the same as if
the interval immediately preceding that sound were increased
relative to the other intervals. The effect of the more intense
sound, when all the sounds are of equal duration, or of the
shorter sound when all the sounds are of equal intensity, is a
THE NATURE OF THE RHYTHM EXPERIENCE 2 93
V
relative overestlmation of the interval preceding the more
intense or the shorter sound" (60, p. 66).
Ruckmich, following Miner, lays great stress on kinaesthesis
as the basis of grouping and accentuation. He says, "Three
points are certainly clear: (i) the kinaesthetic complex
changes for accent and non-accent, (2) kinaesthesis on the
accent is more intensive and is felt as strain or tension, while
kinaesthesis on the non-accent is less intensive and is felt as
relaxation, and (3) kinaesthesis, prominent as it is, may be
temporarily or entirely replaced by visual or auditory com-
plexes" (38, p. 336). "To the writer the group appears
to be a complex of perceptions organized in terms of imaginal
and kinaesthetic processes on the basis of affectively toned
organic processes" (39, p. 254). MacDougall, however,
places the kinaesthesis of attention on a different level than
that of the motor accompaniment although he says it is
concomitant with the sensory series (21, p. 467). The ques-
tion arises whether this particular kinaesthesis of strain or
tension is not of this nature and dependent on attention
(cf. 3, p. 21 1).
POETIC RHYTHM
A word may here be said on the rhythm of poetry. The
rhythmic experience arising from poetry is more satisfactory
than that of prose although Patterson would consider it of an
inferior order. "The aggressive 'timer,' of course, gets his
keenest delight from prose in the fact that he feels no
trammels" (31, p. 84; cf. 21, p. 478). That this is not the
verdict of experience is shown by the fact that all peoples in
all times have chosen poetry as a vehicle to express their most
satisfactory experiences. In poetic rhythm, there is the possi-
bility of greater correlation between the regularity of the
periodic response and of the occurrence of the objective
stimuli. This does not mean, however, that they are co-
incident. Patterson says that poetry is the result of co-
incidence of the unitary pulses and the accented syllables
(31, p. 91). Considering the unitary pulses as equivalent
to one form of response, this may be true for a limited body of
poetry, but it would be at the sacrifice of attention and
294 ELCANON ISAACS
interest. According to Weld, movement is not unison with
the Takt, but is in accordance with the musical phrase (59,
p. 266).
At the same time the methods of the 'stresser' and all
schematic classical systems of scansion which Patterson so
greatly condemns (31, p. 83), have been of use and still are
in classifying certain forms of poetry. The unfortunate result
of their use was to render difficult of analysis the nature of
poetries, in which the motor response is not so nearly cor-
related with the objective stimulation. For this reason, the
nature of biblical meter was obscured for nearly twenty
centuries although scholars had worked on it steadily during
that time. Its nature recently indicated (14, p. 20), shows
as close an approximation to the motor rhythm of other
poetries, but not through stereotyped metric forms. The
motor response arises in connection with the normal unit
for recitative. Its basis is the 'word-foot,' so that there is an
identification between the word and the unit of response
(p. 41). This is possible through the similar lengths of the
words (p. 41). The rhythmic unit corresponding to the verse,
however, is of definite and invariable lengths, consisting of
three units and, in a certain form of poetry, of two units
(p. 44). Furthermore, owing to the intense form of the
poetry, there was not a relaxation after each response,
but a simple and clear case of summation of effects resulted,
which gave rise to the parallelistic structure (15, p. 114).
In the adult reading of the modern verse, the characteristic
reflex response seems also to be in relation to the point of
maximum emphasis rather than in any indispensable relation
to the uniform metrical foot. It is dependent on the form
of reflex response. Brown found that the verse in English
poetry seems to be divided into short phrases which are fairly
uniform in their length while the feet are not (4, p. 51).
ELEMENTS OF THE DEFINITION OF RHYTHM
(a) AFFECTIVE TONE
Whether affective tone should be included in the definition
of rhythm has been put in issue between the extremes of Smith
THE NATURE OF THE RHYTHM EXPERIENCE «95
who says that rhythm no longer exists when affective tone
becomes unpleasant (48, p. 287), to Squire who says that
feeling is not essential to the perception of rhythm (50, p. 587).
There are many intermediate views, and some that cannot be
placed at all. Wundt defines rhythm as an emotion arising
from the feelings of expectation and satisfaction (61, p. 311;
63, p. 200), and says the pleasantness of rhythm depends on
the repetition of feelings of tension and the contrast between
feelings of tension and relaxation (64, p. is8f.). Meumann
says that the affective tone of rhythm depends on the mood
of a given time (25, p. 266), and according to Smith, although
she gives no citation, defines rhythm as an emotion dis-
charging itself in ordered movements (48, p. 292). Ebhardt
places the main stress on affective tone and makes it the
sine qua non of rhythm (9, p. 127). Ruckmich says rhythm
may change from slight unpleasantness before it is grasped,
through pleasantness when it is thoroughly perceived, to
unpleasantness when it continues without change (38, p. 359).
Throughout these theories there runs an unconscious dis-
tinction between affective tone as a result of the rhythmic
experience and affective tone as an element of it. If the
latter is true, then the statement that 'there is no poor
rhythm' (48, p. 292) is correct. The weight of the evidence,
however, is against this. Rhythm may be unpleasant; at
times it may be 'dreadful' (3, p. 221; 48, p. 285).
Affective tone is the result of rhythm, but since rhythm is
a continuing phenomenon, the affective tone aroused by the
feelings of repeated, perfected movement, has been thought
to be the cause of the rhythm coming. Squire says: "The
affective tone increases in proportion as the summation of
excitation increases, till a state bordering on ecstasy may be
reached. Ecstasy, when it follows upon rhythmical stimula-
tion, is due to a spreading of the excitations to a greater and
greater number of centers, till the body and the whole of
consciousness are set in co-vibration" (50, p. 588).
(&) COMPLEXITY
Every investigator whether or not he includes affective
tone is his defiiaition of rhythm is very certain and unequivocal
296 ELCANON ISAACS
as to its complexity. This is practically the only point that
all are agreed upon. Thus Meumann says: "The error must
be emphatically combated, that on defining any one of these
elements, even of the so-called equality of the beats, we have
defined the nature of the rhythm," but enumerates the various
factors which are necessary for the experience of rhythm.
"In this manner, in the rhythmical impression we shall have
to seek for the elements of time, accent and pitch. By the
side of these there must be distinguished a number of higher
intellectual factors, whose operations we must seek in the
inner comprehension, in the additions of subjective accent,
in the strain and relaxing of the attention, the relating of the
rhythmical groups to one another, their perception as repeti-
tion of the preceding and preparation for the following im-
pressions" (25, pp. 305-6). So Wundt (62, pt. 2, I, p. 3761.);
and Ruckmich says, "... the rhythmical perception is an ex-
ceedingly complex affair. . . . The grouping effect of a
rhythm in any case may depend on visual patterns, on audi-
tory imagery, on organic complexes, on changes of clearness,
on alterations of temporal arrangements, on verbal ideas, on
motor responses, and on many similar items" (39, p. 247).
Likewise Patterson is careful to point out this item of the
experience. "Rhythmic experience may thus be roughly
described as a complex of perception, emotion, and sensation,
with all three elements subjected to the moulding processes of
attention, both voluntary and involuntary" (31, p. 91).
There is ample complexity here, yet elsewhere he says, "The
final impression of rhythm derived from a sentence is, to a
large extent, a fusion of elements, in which actual pitch, tone-
color, thought, mood, capricious or logical attention, etc.,
enter as factors in addition to duration, stress, and the dim
elements of pitch, actual or purely subjective, implicated
in the drum-beat tune" (p. 70).
SUMMARY
Rhythm is the experience arising from the periodic,
pendular, reflex response of characteristic organs to objective
stimulation. There are four elements in the impression of
THE NATURE OF THE RHYTHM EXPERIENCE *97
rhythm, the perception of the objective stimulation, the
experience of the periodic reflex response, accentuation and
grouping resulting from attention, and the affective tone
arising from repetition of movement.
The pendulum rate is the rate at which an organ vibrates
in the absence of voluntary factors, and is the result of the
length of attachment of a member, and of the refractory
phase of the muscles involved.
The reflex response is the result of, but independent of the
stimulation, and depends on the pendulum rate of the member
responding. Because of the periodic nature of the reflex
response, regularity was read into the objective stimulation,
and it was thought that the latter must be chronometrically
proportionate. This belief gave rise to elaborate and some-
times artificial systems of meter, and prosody was erected
into a science.
The objective stimulation has one prerequisite, that the
discrete stimuli recur so as to give rise to a serial response.
With this qualification, the stimulation may be regular or
irregular, accented or unaccented, grouped or ungrouped.
BIBLIOGRAPHY
1. ALDEN, R. M. An Introduction to Poetry. 1909.
2. ARPS, C. F. AND KLEMM, O. Der Verlauf der Aufmerksamkeit bei rhythmischen
Reizen. Psychologische Studien, 1909, 4.
3. BOLTON, T. L. Rhythm. Amer, J. Psychol., VI., 1894.
4. BROWN, WARNER. Time in English Verse Rhythm. Arch, of Psychol., No. 10,
1908.
5. BRUCKE, E. W. Die physiologischen Grundlagen der neuhochdeutschen Verskunst.
1871.
6. DUNLAP, KNIGHT. Rhythm and the Specious Present. /. Phil., Psychol. and
Sci. Mtth., 1911, 8.
7. DUNLAP, KNIGHT. Time and Rhythm. Psy. Bull., 1916, 13.
8. EBBINGHAUS, H. Grundzuge der Psychologic. 1902.
9. EBHARDT, KURT. Zwei Beitrage zur Psychologic des Rhythmus und des Tempo.
Zeit. fur Psychol., 1898, 18.
10. GUEST, EDWIN. A History of English Versification. 1882. .
11. GUMMERE, F. B. A Handbook of Poetics. 1885.
12. GURNEY, EDMUND. The Power of Sound. 1880.
13. HOWELL, W. H. A Text-book of Physiology. 1917.
14. ISAACS, ELCANON. The Metrical Basis of Hebrew Poetry. Amer. J. Sem. Lang.,
1918, 35-
15. ISAACS, ELCANON. The Origin and Nature of Parallelism. Amer. J. Sem. Lang.,
1919, 35-
ELCANON ISAACS
16. KOFFKA, K. Experimental-Untersuchungen zur Lehre vom Rhythmus. Zeit.
fur Psychol., 1909, 52.
17. KOSTLIN, K. Aesthetik, 1869.
18. KULPE, O. Outlines of Psychology, trans. E. B. Titchener.
19. LANIER, SIDNEY. The Science of English Verse. 1880.
20. LOBE, J. C. Katechismus der Musik, 25 Aufl. 1893.
21. MACDOUGALL, ROBERT. The Relation of Auditory Rhythm to Nervous Dis-
charge. PSYCHOL. REV., 1902, 9.
22. MACDOUGALL, ROBERT. The Structure of Simple Rhythm Forms. Han. Psy.
Stud., i, 1903; PSY. REV. MONOG., 4, 1903.
23. MATTHEWS, J. BRANDER. A Study of Versification. 1911.
24. MAYOR, J. B. Chapters on English Meter. 1901.
25. MEUMANN, ERNST. Untersuchungen zur Psychologic und Aesthetik des Rhyth-
mus. Phil. Stud., 1894, 10.
26. MEYER, E. B. Beitrage zur deutschen Metrik. Die neuere Sprachen, VI, 1896.
27. MINER, J. B. Motor, Visual and Applied Rhythms. PSY. REV. MONOG.,
1903, 5-
28. MIYAKI, ISHIRO. Researches on Rhythmic Action. Yale Psy. Stud., 1901, 9.
29. MYERS, C. S. A Text-book of Experimental Psychology. 1911.
30. OMOND, T. S. A Study of Meter. 1903.
31. PATTERSON, W. M. The Rhythm of Prose, 1916. 2d ed. 1917.
32. PATTERSON, W. M. New Verse and New Prose. North Amer. Rev., 1918, 207.
33. PILLSBURY, W. B. Attention. 1908.
34. PILLSBURY, W. B. The Fundamentals of Psychology. 1916.
35. PUFFER, E. D. The Psychology of Beauty. 1905.
36. RIEMANN, H. Katechismus der Musik. 1888.
37. RIEMANN, H. Elemente der musikalischen Aesthetik. 1900.
38. RUCKMICH, C. A. The Role of Kinaesthesis in the Perception of Rhythm. Amer.
J. Psychol., 1913, 24.
39. RUCKMICH, C. A. Visual Rhythm. Titchener Commemorative Volume. 1917.
40. SAINTSBURY, G. Historical Manual of English Prosody. 1910.
41. SCHAFER, A. E. A Text-book of Physiology, II. 1900.
42. SCHIPPER, J. History of English Versification. 1910.
43. SCRIPTURE, E. W. The Law of Rhythmic Movement. Science, n.s., 1896, 4.
44. SCRIPTURE, E. W. The New Psychology. 1899.
45. SCRIPTURE, E. W. Observations on Rhythmic Action. Yale Psy. Stud., 1899,7.
46. SCRIPTURE, E. W. Elements of Experimental Phonetics. 1902.
47. SHERRINGTON, C. S. The Integrative Action of the Nervous System. 1906.
48. SMITH, M. K. Rhythmus und Arbeit. Phil. Stud., 1900, 16.
49. SNELL, A. L. F. Pause; a Study of its Nature and its Rhythmical Function in
Verse, Especially Blank Verse. Univ. of Michigan Contributions to Rhetorical
Theory. 1918.
50. SQUIRE, C. R. A Genetic Study of Rhythm. Amer. J. Psychol, 1901, 12.
51. STETSON, R. H. Rhythm and Rhyme. Harv. Psy. Stud., 1903, i; PSY. REV.
MONOG., 1903, 4.
52. STETSON, R. H. A Motor Theory of Rhythm and Discrete Succession. PSYCHOL.
REV., 1905, 12.
53. STEVENS, L. F. On the Time-sense. Mind, 1886, n.
54. SULLY, JAMES. The Human Mind. 1892.
THE NATURE OF THE RHYTHM EXPERIENCE
299
55. TITCHENER, E. B. The Psychology of Feeling and Attention. 1908.
56. TITCHENER, E. B. A Text-book of Psychology. 1919.
57. WALLIN, J. E. W. Researches on the Rhythm of Speech. Yale Psy. Styd., 1901, 9.
58. WALLIN, J. E. W. Experimental Studies of Rhythm and Time. PSYCHOL. REV.,
1912, 19.
59. WELD, H. P. An Experimental Study of Musical Enjoyment. Amer. J. Psychol.,
1912, 23.
60. WOODROW, H. A Quantitative Study of Rhythm. Arch, of Psychol., No. 14,
1909.
61. WUNDT, W. Vorlesungen uber der Menschen- und Thier-seele. 1897.
62. WUNDT, W. Volkerpsychologie. 1900.
63. WUNDT, W. Grundriss der Psychologic. 1901.
64. WUNDT, W. Grundzuge der physiologischen Psychologic, III. 1903.
BY GODFREY H. THOMSON, D.Sc.
Armstrong College, University of Durham
I. INTRODUCTION
I propose in this paper to examine some of the ideas under-
lying the interpretations hitherto given of threshold measure-
ments, and in particular (a) the measure used in comparing
the sensitivity of one individual with that of another, (b) the
measure of sensitivity used in testing Weber's Law, (c) the
origin of this law, whether it be psychological or physiological,
(d) the Idea of a psychometric function, and (e) the notion
of the probability of a judgment. These points cannot be
treated separately, but must be considered in conjunction
with one another. The ideas which I wish to express have
their origin, however, chiefly in an extension of the last-
mentioned point.
II. THE DIFFERENCE THRESHOLD
To fix ideas, I shall discuss difference thresholds only,
leaving the question of absolute thresholds aside. Further,
I shall use the case of experiments on lifted weights, by the
Method of Right and Wrong Cases.1 In this form of experi-
ment, comparison weights are contrasted frequently, one by
one, by lifting them in a specified way, with a standard
weight, and on each is expressed a judgment lighter, undecided,
or heavier. When a sufficient number of judgments have been
collected, the tliree categories are found to occur with varying
frequency with different comparison weights, and curves
similar to those shown in the accompanying diagram can
be constructed.2
1 Otherwise termed the Method of Constant Stimuli, and the Methode der drei
Hauptfalle.
2 For fuller details of the determination of this and other thresholds, and for
references to experimental memoirs, the reader may consult G. T. Fechner, Elemente
300
INTERPRETATION OF THRESHOLD MEASUREMENTS 3OX
In this diagram the points marked on the #-axis represent
the comparison weights. The height AB represents unity,
and the three curves give the proportional frequency of
answers lighter, undecided and heavier respectively, the un-
decided curve being a bell-shaped curve, and the other two
being what Galton termed ogives.* The difference threshold
B
A 7
FIG. i
is then decided by the positions of the points S and S' where
the lighter and heavier curves cross the halfway line.2 The
distance (S — S')/2 or some closely similar quantity is what
is called the difference threshold, and this quantity is com-
monly used in comparing the sensitivity of different subjects.
The smaller (S — S'), the more sensitive the subject is said
to be.
This distance however depends entirely on the subject's
der Psychophysik, 1869; G. E. Mueller, Die Gesichtspunkte und die Thatsachen der
psychophysischen Methodik, Wiesbaden, 1904; W. Wirth, Psychophysik, Leipzig,
1912; and F. M. Urban, The Application of Statistical Methods to the Problems of
Psychophysics, Philadelphia, 1908.
1 The actual experiments of course do not give curves but only points through or
among which the curves are then drawn smoothly, either by hand, by a flexible ruler,
or by assuming mathematical equations for the curves. In the latter case if the
equations contain sufficient constants, the curves can be made to go exactly through
the points; otherwise the best fitting curves are drawn by some method such as the
method of least squares or the method of moments.
* These points on an ogive are analogous to the medians of bell curves. Some
experimenters, instead of calculating these medians, calculate points analogous to
means. These differ slightly from S and S' but these differences have no bearing on
our present argument. Questions of space and time errors also* arise.
302
GODFREY H. THOMSON
readiness to give the answer undecided. It measures there-
fore rather a moral characteristic than a physical sensitivity.
It is a question of quickness of decision on small evidence
just as much as a question of difference of amount of evidence.
If persons A and B differ widely in the number of undecided
answers which they give, it may just as well be their habits
of forming decisions which differ, and not their power of
actually discriminating the weights. The moral character
of the measure S — Sf is above all seen from the fact that any
subject who wishes may reduce it to zero, whatever may be
his actual sensitivity, simply by determining that he will
never give an answer undecided. Difference thresholds there-
fore are unsuitable for comparing the sensitivity of different
subjects.
III. THE INTERQUARTILE RANGE OF THE POINT OF
SUBJECTIVE EQUALITY
There is however another measure which has been used
for this purpose. This can be most conveniently described
by considering first a case in which a subject gives no un-
0.75
0.60
0.25
— , »• STIMULUS
FIG. 2
decided answers. In such a case the central curve of the
former diagram, that is the bell-shaped curve, disappears,
and only the two ogives are left, as in the accompanying Fig. 2.
The thresholds S and S' have come together and on the
previous plan the subject's sensitivity would be considered
INTERPRETATION OF THRESHOLD MEASUREMENTS 3°3
infinite, and all subjects giving no undecided answers would
be given the same infinite sensitivity, whereas clearly the
subject's sensitivity is connected with the rapidity with which
the curves pass from zero to unity or vice versa, and two
subjects may differ very much in this respect even although
they may both give no undecided answers. Under these
circumstances a measure which has been used is the distance
Q — Q' which is sufficiently explained by the diagram. This
measure (though in another guise) was used by Fechner also
for the cases where undecided answers were given. In such
cases he reduced the three curves to two by sharing the
undecided answers equally between heavier and lighter. This
measure has the advantage that the subject cannot increase
his apparent sensitivity at will, as was the case with the
threshold measure. Q — Q' is the interquartile range of a
hypothetical point of subjective equality. It and the differ-
ence threshold measure distinctly different things and subjects
placed in order of merit by the one will be found in a different
order when graded by the other. The measure Q — Q' is
more physiological than the threshold measure.
IV. WEBER'S LAW
Although these two measures, the difference threshold
and the interquartile range of the point of equality, measure
different things and give quite different results when we are
comparing separate subjects, yet in one and the same subject
each of them obeys Weber's Law in so far as that law is
obeyed by any measure. It would appear that some light
might be thrown on the question whether Weber's Law is
physiological or psychological in its origin by a comparison
of the accuracy with which these respective measures obey
that law, although such a comparison would not be crucial.
If, however, the difference threshold were found to obey
Weber's Law more exactly than does the probable error of
equality, then I should consider this distinctly in favor of a
psychological explanation of the law. But I should myself
anticipate the other measure to obey the law more exactly,
which, although by no means proving a physiological explana-
tion to be correct, would nevertheless point in that direction.
304 GODFREY H. THOMSON
V. PSYCHOMETRIC FUNCTIONS
Turning now temporarily aside from the considerations
which have occupied us up to this point, let us consider some
questions arising out of what is known as the search for the
* psychometric function.' These functions are the mathe-
matical equations which best fit the bell-curves and ogives
into which psychometric data fall. They are clearly error
functions of some sort or other, and there is no particular
advantage in calling them psychometric functions. For our
purpose here it is sufficient to consider the possibility of the
"psychometric function" being the normal curve of error,
and to leave the elaborations necessary when more compli-
cated functions are used. What is here said about the normal
curve is typical and true of other suggested functions such as
Professor Pearson's types.
Consider Fig. I. The suggestion has been made by many
that the extreme curves, the ogives, are integrals of the normal
curve of error. The bell-curve in the center is deduced by
these writers by first forming the outer curves and then
subtracting their sum from unity at each point. All the
usual arguments, however, which support the view that the
outer curves are integral normal curves would lead one to
expect, when applied to the central curve, that it is a normal
curve as it stands. But this is impossible. Two such normal
ogives added together and subtracted from unity, ordinate
by ordinate, do not give a normal bell curve.
VI. EXTENSION OF THE NOTION OF THE PROBABILITY OF A
JUDGMENT
This difficulty can I think be overcome if we recognize
the fact that these curves are not each of them complete
error functions, but represent one error function divided into
three parts (the division itself however being no doubt in
turn subject to an error distribution).
This involves an extension of the idea of the probability
of a judgment. In its simple form this idea compares the
giving of a judgment heavier, undecided, or lighter, with draw-
ing a ball from an urn containing say red, white, and blue
INTERPRETATION OF THRESHOLD MEASUREMENTS 3°5
balls, and ascertaining its color. For each stimulus the urn
is supposed to contain different proportions of the colored
balls.
In place of this I suggest the following. For each stimulus
imagine an urn containing an infinite number of balls divided
between black and white in a proportion varying in some way
with the stimulus. A judgment is to be compared with
taking not one but a handful of balls from the urn, and the
kind of judgment is to depend on the proportion of black
balls in the handful.
VII. THE RESULTING CURVES
To clear our ideas let us take a concrete case with small
numbers. Suppose an urn contains seven tenths black and
three tenths white balls, and that a judgment corresponds to
drawing four. The possibilities that can occur in such a
lottery are five in number, namely
Occurrence Frequency
4 black, no white 0.2401
3 " I " 0.4116
2 " 2 " 0.2646
I " 3 " 0.0756
o " 4 " 0.0081
I. oooo
where the frequencies are the terms of the binomial expansion
(0.7 + o.3)4.
Now suppose that either 4 or 3 balls correspond to an
answer heavier, 2 to an answer undecided, and I or o to an
answer lighter. For the stimulus in question the relative
frequencies of these will be:
heavier 0.2401 + 0.4116 = 0.6517
undecided = 0.2646
lighter 0.0756 + 0.0081 = 0.0837
I. oooo
Imagine now this process carried out for a number of
urns, thus
306
GODFREY H. THOMSON
1
Vequency of Answer
i
Heavier
Undecided
Lighter
o.o black
O.OOOO
O.OOOO
I.OOOO
O.I
o.oo-?7
0.0486
O.Q4.77
0.2
O.O272
O.IC^6
0.8192
CM
O.oSn
j-5;;
0.2646
O.6?I7
0.4.
O.I79O
o.-?4i;8
O.47i;2
o.c
O.'?I2C
O.'J7CO
O.3I2C
oi .:::::::::::::::::::
O.4.71; 2
O.T4.C8
O.I7QO
0.7
O.6CI7
0.2646
o.o8-?7
0.8
0.8192
o.ic-?6
O.O272
O.Q
0.04,77
0.0486
O.OO37
I.O
I.OOOO
O.OOOO
O.OOOO
These numbers are the basis of the three curves in Fig. 3
which it will be recognized are very similar to those actually
found in psychological experiment. Skew curves are easily
obtained by placing the points of division into the categories
heavier, undecided, and lighter at unequal distances from the
o.i 0.2
0.3 0.4 0.5 0.6 0.7
> BLACK BALLS IN URN
0.8 0.9
1.0
FIG. 3
center of the underlying error curve. The actual form of
the curves depends on the way in which the #-axis is measured
out. In this diagram the simplest assumption is made,
namely, that the proportion of black balls in the urn is a
linear function of the stimulus, and, between the limits at
which the subject answers heavier or lighter with certainty,
directly represents it. The number of balls taken in the
sample which represents a judgment is also a factor. But
ultimately this would be made very large, and Type III.
curves would replace the binomials.
INTERPRETATION OF THRESHOLD MEASUREMENTS 3°7
On this point of view, the standard, the variable stimulus,
and the physiological make-up of a subject decide the propor-
tion of black balls in the urn, but the decision as to what pro-
portion in the sample is to be called heavier, what undecided
and what lighter depends upon a conscious act of the subject,
and can be varied, if he be so disposed, at his whim; and will
vary with his mood at the moment.
The difference between this point of view ^nd the older
one may prove to be academic only,1 for the disentangling
of the factors which are here distinguished may in practice
be impossible. Nevertheless the idea seems illuminating, and
this sketch is put forward in the hope that some of the result-
ing curves may prove of interest to mathematical statisticians,
and of use in psychophysics.
1 Indeed it may be that mathematically the one reduces to the other, at any rate
under certain conditions.
THE CORRELATION BETWEEN INTERESTS AND
ABILITIES IN COLLEGE COURSES
BY JAMES W. BRIDGES AND VERONA M. DOLLINGER
Ohio State University
It is obvious that achievement in any vocation depends
partly upon ability to do the work and partly upon interest in
that particular kind of work. The latter factor is not less
important than the former, for it supplies the necessary in-
centive to the greatest endeavor. The problem for vocational
guidance is therefore to measure ability and interest.
The measurement of ability whether considered as special
aptitude or as general intelligence presents no great obstacles.
To obtain a test or group of tests whose results correlate highly
with actual achievement in any vocation is a relatively simple
psycho-statistical problem. The measurement of interest is a
much more difficult matter. No objective method is available.
The evaluation of interest must therefore depend entirely
upon subjective estimate; and of course this estimate cannot
be made until the subject has some knowledge of the work.
If it can be shown that interest is highly correlated with
ability, the problem of vocational guidance will be simplified;
because on the one hand, any interest on the part of the
individual will give him definite assurance of ability in that
direction, and on the other hand, the presence of a definitely
determined ability will surely indicate an interest which
perhaps merely awaits experience to be awakened. If how-
ever, the correlation is low, as the results of this study seem
to indicate, then the practical problem in vocational guidance
will be to determine the general ability and special aptitudes
of the individual, to advise him as to the several occupations
suited to his intelligence level and specific abilities and let him
select from these on the basis of his interest, together with other
considerations such as social status of the work and remunera-
tion.
308
INTERESTS AND ABILITIES IN COLLEGE COURSES 3°9
The problem of the relation between interest and ability
has received considerable attention from educators from time
to time; but the only experimental study of the subject known
to the writers is that reported by Professor E. L. Thorndike
of Columbia University.1 His subjects, college students, ar-
ranged the courses of their curriculum, namely — mathematics,
history, literature, science, music, drawing, and other hand-
work, in the order of their interest and then later in the order
of their ability in them. The correlation between these two
orders was found to be as high as .89, and the conclusion was
drawn that "A person's relative interests are an extraordinarily
accurate symptom of his relative capacities."
It seems quite probable that the correlation obtained, .89,
is much higher than the actual correlation between interests
and abilities, for a subject's ranking of his courses for the one
is likely to be influenced by his ranking for the others. Thorn-
dike points out that it would be "Better to get the measure-
ments of relative interest and of ability . . . not from indi-
vidual reports alone, but from objective tests." A satisfactory
objective test for interest is not yet available; but an objective
measurement of ability would seem to be afforded by college
grades, especially where the proportions of the various grades
assigned conform approximately to the normal distribution.
It seemed, therefore, desirable to determine the correlation
between interests, evaluated subjectively by rankings of
courses, and abilities, measured objectively by grades obtained
in these same courses.
With this object in view, several hundred students were
requested at the beginning of the semester to fill out and return
blanks upon which were printed the following instructions:
"Arrange the courses you are studying this semester according
to your interest in them. Place first in the list the course you
are most interested in, then the others in order. Please make
your judgments carefully and deliberately, and try as far as
possible to avoid influence by class grades or preference for
1 'The Permanence of Interests and Their Relation to Abilities," Pop. Sci. Mo.,
1912, 81.
'Early Interests, Their Permanence and Relation to Abilities,' School and
Society, 1917, 5.
310 JAMES W. BRIDGES AND VERONA M. DOLLINGER
instructor." Just below a space on the blank for listing the
courses a second instruction was given: "Now arrange the
subjects you are studying this semester according to your
ability in them. Try to make your judgments independent of
your interests and of any class grades you may have received."
Since these data were obtained near the beginning of the
semester, the influence of class grades upon the students*
rankings was probably negligible. At the end of the semester
the grades actually made by each student on the courses
ranked were obtained from the registrar's office. Ranging in
order from highest to lowest, the grades used in the university
are: M, G, A, P, and F, which are assigned approximately
in conformity with the proportions: 5, 20, 50, 20, and 5.
Records were obtained from over five hundred students,
and the number of courses ranked varied with different students
from four to eight. In order to simplify the calculations only
those records with five courses ranked were used since they
furnished the bulk of the material.
The relationship between these interests and abilities was
determined by means of Pearson's formula for mean square
contingency1 since this seems best adapted to such rough forms
of evaluation of traits. The coefficient of mean square con-
tingency (C) thus obtained is somewhat smaller than the
coefficient of correlation (r), for in the case of a five by five-fold
classification the former (C) cannot exceed .894. However,
with coefficients as low as those actually obtained the difference
becomes relatively negligible.
The relationship between rank in interest and grade earned
was determined first for a specific course of study, namely,
elementary psychology. All records that reported psychology
were used irrespective of the sex or college year of the student.
The data are presented in full in Table I., an inspection of which
plainly corroborates the low coefficient obtained (.22). The
relationship between rank in interest and rank in ability (as
reported by the subject on his record blank, hereinafter referred
to as estimated rank in ability) was next determined for the
same course, and a coefficient of .59 obtained. Similar coeffi-
1 See H. D. Rugg, 'Statistical Methods Applied to Education,' pp. 299-307.
INTERESTS AND ABILITIES IN COLLEGE COURSES
TABLE I
RELATION BETWEEN RANK IN INTEREST AND GRADE EARNED IN PSYCHOLOGY'
Grade
Number
Rank in Interest
F
P
A
G
M
Students
I
3
8
14
IO
5
40
2
1
IO
IO
7
9
59
1 . .
2
ii
11
17
8
69
I
21
30
11
7
72
4.
IO
27
i I
4
59
Number students
13
60
132
61
33
209
1 The letter grades, F, P, A, G, and M, are assigned approximately in conformity
with the proportions: 5, 20, 50, 20, and 5; and they have the following meanings:
fail, pass, average, good, and merit.
cients were calculated for the English course with very similar
results (see Table V.).
Coefficients might have been calculated in the same way
for each course reported by the students; but it seemed more
desirable to obtain a coefficient that would express the relation-
ship between rank in interest and grade for all courses com-
bined. With this object in view Table II. was prepared from
TABLE II
RELATION BETWEEN RANK IN INTEREST AND GRADE EARNED. ALL COURSES
COMBINED
Grade
F
P
A
G
M
Total
I
6
A.2
OO
58
C¥
2?6
2
7
48
111
c.7
11
•*/-
2^6
1. .
12
c-a
TIC
ce
21
•>,
2CO
4. .
11
76
IO1
CO
1.1
**J
2CO
?. •
2C
78
£.
IOO
17
IO
2e
2^6
Total
63
297
536
257
127
1,280
the records of two hundred fifty-six students of both sexes and
all college years. The table shows the distribution of grades
for the two hundred fifty-six courses placed first in interest,
then for the two hundred fifty-six courses placed second in
interest, and so on for the third, fourth, and fifth places. The
grand total of twelve hundred eighty must accordingly be
read as student-courses. The coefficient calculated from this
table is .25.
312
JAMES W. BRIDGES AND VERONA M. DOLLINGER
The relationship between rank in interest and estimated
rank in ability is shown in a similar manner in Table III.
Each row of figures gives the distribution in ranks of estimated
ability for a given rank in interest for 291 students. The
coefficient calculated from this table is .57, which agrees very
closely with the similar coefficients for the psychology and
English courses.
TABLE III
RELATION BETWEEN RANK IN INTEREST AND ESTIMATED RANK IN ABILITY.
ALL COURSES COMBINED
Rank in Interest
Estimated Ability
Total
5
4
3
2
i
I
9
18
37
63
164
19
44
69
102
57
49
72
69
65
36
63
94
66
45
23
151
.63
SO
16
ii
291
291
291
291
291
2
3- •
4. •
<>• .
Total
291
291
291
291
291
i,455
The low correlation so far indicated between rank in interest
and grade and the relatively higher correlation between rank
in interest and estimated rank in ability point to the mutual
dependence of judgments of interest and of ability. A low
correlation between estimated rank in ability and grade might
accordingly be expected. Data presenting the actual relation-
TABLE IV
RELATION BETWEEN ESTIMATED RANK IN ABILITY AND GRADE EARNED.
ALL COURSES COMBINED
Grade
Estimated Rank in Ability
F
P
A
G
M
Total
I
26
07
<;8
4.4
228
2 1 .
7
4.8
82
62
2O
228
3 . .
II
AA
112
4-1
2O
228
4. . .
12
61
IOC
•5Q
II
228
«C . .
2O
76
86
08
8
228
Total
53
255
482
238
112
1,140
ship are given in Table IV, and the coefficient was found to be
.28, which is very nearly the same as the coefficient for rank in
interest and grade.
INTERESTS AND ABILITIES IN COLLEGE COURSES
3'3
The relationship between rank in interest and grade and
between rank in interest and estimated rank in ability was
also determined for each sex separately, and for freshmen and
upper class-men separately. The results show no significant
differences. Indeed, all the coefficients calculated are sur-
prisingly uniform — a fact which would seem to indicate their
general validity. Table V, which sums up all the results,
TABLE V
COEFFICIENTS OF MEAN SQUARE CONTINGENCY (C)
Between Rank in Interest
and Grade
Between Rank in Interest and
Estimated Rank in Ability
C
No. Cases
C
No. Cases
Psychology
.22
.27
•2!
.26
.26
•25
.28
315
394
256
171
85
157
99
•59
•57
3
•54
•57
•5°
I58
194
291
201
90
Ǥ
106
English
All courses
All courses (males)
All courses (females)
All courses (i yr.)
All courses (2, 3, 4, yr.)
will facilitate comparison. The coefficient obtained and the
number of cases used in each calculation are given.
It will be noted that the contingency coefficients for rank
in interest and grade range from .22 to .28, and for rank in
interest and estimated rank in ability from .50 to .59. The
product-moment coefficient of correlation in the former case
would probably not be over .30 and in the latter not over .65.
This latter figure is much lower than that obtained by Pro-
fessor Thorndike (.89) in the experiment referred to above;
but, as a measure of the relationship between interest and
actual ability, it is probably much too high, and is merely a
result of the subjective method of evaluating ability. When
ability is measured by a more objective means, namely college
grades, a very low correlation between interest and ability is
obtained, so low in fact that one might well be justified in the
statement: A person's relative interests are an extraordinarily
twaccurate symptom of his relative capacities. It might also
be inferred from data here presented that a person's estimate of
his ability is an extraordinarily inaccurate symptom of his real
ability, for the correlation between the students' rankings of
their courses in ability and the grades obtained is only .28.
314 JAMES W. BRIDGES AND VERONA M. DOLLINGER
There are, to be sure, many obvious objections to the use
of college grades as measures of ability. First, a grade is also
in part a measure of interest since the persistent application
which earns the higher grades is based very largely upon
interest. The effect of this would be to increase the correlation.
That is to say, in so far as the grade earned depends directly
upon the incentive supplied by interest, the coefficients given
above are too high as measures of the relationship between
interests and ability. Secondly, college grades are not suffi-
ciently discriminative; and, consequently, a student may ob-
tain the same grade in all his courses when his actual abilities
are perhaps not so even. In so far as this is true, the coeffi-
cients reported are too low. Thirdly, grades are also dependent
upon general intelligence (if there be one). This would tend
to make the grades of each student uniform; and would have
the same effect upon the correlation as the factor last men-
tioned. Finally, grades are also affected by other factors,
such as personal relation between student and instructor, out-
side activities of the student, home environment of the student,
and so forth. All these factors would, presumably, affect the
correlation adversely.
The writers are therefore not ready to draw any certain
conclusions from this short study regarding the actual relation-
ship between interests and abilities. The problem is an
extremely complicated one; and it cannot be solved until a
more objective method of evaluating interests, as well as
abilities, is forthcoming. Achievement, as has already been
said depends upon both interest and ability. If these are not
highly correlated, the conclusion of practical importance for
vocational guidance is that both must somehow be evaluated
separately.
VISUAL PHENOMENA IN THE DREAMS OF A BLIND
SUBJECT
BY RAYMOND H. WHEELER
University of Oregon
The subject whose dreams are herein reported was a
student at the University of Oregon from 1915-1918. He
was a trained introspector and at the time of this investiga-
tion he had had several courses in psychology, including
laboratory. After a preliminary period of training in re-
cording dreams he found it possible to describe the important
details in note form, in American Braille, immediately upon
waking. These notes were subsequently edited by the
writer with the help of the reagent, great care being taken to
omit all uncertain or otherwise questionable details. The
reagent lost his sight by accident when eleven years old.
He is now twenty-seven.
Although the primary purpose of this paper is to report
visual phenomena in the dream life of a blind subject after
sixteen years of blindness, it is interesting to note, in addition,
that in his dreams we find a peculiar association between
visual images and images in other modalities. In his waking
life the subject has very complicated synaesthetic phenomena.
It has been noted in the literature1 that in certain instances
those individuals who possess associations between visual
and other sensations in their perceptual processes also asso-
ciate these modalities in a similar fashion in their imaginal
processes. This is also true of our reagent. As far as the
writer knows, however, such phenomena in dreams have not
been described in the literature. For the convenience of
the reader the reagent's descriptions of these associations are
printed in italics.
DREAM i
First I had a rather confused visual image of a portion of a room including one
large window and the surrounding walls. The light which came through the window
1 To be reported in a subsequent paper.
315
31 6 RAYMOND H. WHEELER
was dim, giving the appearance of a heavy fog or thick dust which seemed to fill the
room. I could distinctly "see" the rays of light penetrating through the fog. The
space in the room, penetrated by the light, seemed to be about six feet wide and three
feet deep. Accompanying this visual imagery was a marked unpleasantness, a vivid
organic and kinaesthetic experience consisting of a tension in the muscles of my arm,
of a tightening in the vocal cords and of contractions in the muscles of my jaws, the
latter resulting in a state of marked rigidity. Then I had a sense of "half pressure and
half buzzing" in my ears and a diffuse and vague tension in the muscles of the brows,
forehead, neck and chest. The unpleasantness and kin&sthetic tensions were linked, in
consciousness, with the foggy, yellow light. The second group of experiences constituted
an awareness that I was near the walls of the room.
Then I suddenly found myself in a second room in a house some distance away
and to the south of the first. Here my imagery was somewhat similar to the preceding
but lacking the yellow light and the affective accompaniments. I infer that this
house was strange for there was nothing familiar about it. I was then conscious of
looking from where I now stood toward the direction of the house I had previously
been in. Simultaneously with this change in my line of regard I had a vague flash of
straw-colored light. This meant "south" to me, which I innervated in vocal-motor
fashion: "south." (My notion of "south" is associated, in waking consciousness,
with straw-colored light.) I then noticed two friends in the room. Both were sitting
to my right and a little in front of me. To my left and very near me was localized a
very dark mass, somewhat of the size and general shape of a person, with a thick, hazy
area as a fainter background. The lines of his shoulders, chest and arms to the elbows
were the only distinct features of the figure. Instantly I had the faintest tendency to
turn in his direction. This was the first intimation that I had had of being in this
house with a companion. / was unable to recognize him for the color of the imagery was
too faded. (In waking life I always identify people, in imagery, by their color.)
The friend to my right, seated nearest me was a vague form in sitting position, colored
a very deep shade of blue. The only features which approached distinctness were
outlines of his head, arms, legs and trunk. I imagine that the form appeared much
the same as a person might look through a thick blue lens with the object much out of
focus. The other figure was fawn-colored yellow, of medium brightness and rather poor
saturation. I identified both of these persons by their colors.
Then there appeared a very irregular image of a large oak chair finished with
yellow wood and black leather upholstering. The imagery was localized and dis-
tributed in space in a position corresponding to the details of a chair which are visible
to the eye when one is sitting. I also had tactual imagery of the "feel" of the up-
holstering and kinsesthetic imagery of the bodily position assumed in sitting in this
style of furniture.
I was then conscious of saying: "Hello boys; when did you get in?" There was
no answer. My attention then became more rigidly fixed upon the two visual forms
as I thought to myself in verbal imagery: "Why don't they answer?" I then spoke
again: "I want you fellows to come over and see me while you are here." Again no
answer. I then became very angry; I was conscious of intense tightening of muscles
in my arms and chest together with a characteristic tension in the back of my scalp.
(This latter is characteristic of anger in waking life.) With the growing anger the
colors of the two forms became very much brighter. I then turned to the figure at my left,
which represented my companion, and said: "Let's go back." I then found myself
back in the first house but this time in a different room, for the window was on the
VISUAL PHENOMENA IN DREAMS OF BUND SUBJECT 3*7
west side and there was a telephone on the wall. The window curtains were streak*
of drab-grey, which, I suspect, is my representation of white. The room was large,
which I interpret from the fact that I experienced no pressure images. My awareness
of the telephone consisted of visual and tactual imagery. I "saw" the brownish
wooden box containing the mechanism of the telephone. In the middle of the trans-
mitter was a circle of rather brilliant light which told me the exact place into which
to direct my voice. The hard rubber mouthpiece was visualized as black; the receiver
hook was a shining steel-gray as were also the bells. The outlines of these objects
were fairly distinct but fleeting. I then had the vocal-motor: "I will call up the boys
and ask them over to see me." I had kinsesthetic and tactual imagery of taking down
the receiver and of holding it to my ear, with a distinct image of coldness as the edge
of the receiver came in contact with my ear. No further imagery appeared until the
vocal-motor: "They won't answer," whereupon I was extremely disappointed and
angry. Here I had the characteristic sinking experiences localized in the region of the
diaphragm, inhibition of breathing and tensions in the throat and chest. I then had
the vocal-motor: "You can go straight to the devil." At this moment I became con-
scious of a person in back of me, visualized as a bright and silvery form, which meant to
me that the form was a woman. I then had auditory imagery of her voice: "He
has drowned in the creek." Simultaneously with the "he" there appeared a dark
visual schema to my right, very indistinct and not sufficiently colored for identification.
I next found myself searching for the creek. I was visualizing myself walking up a
slope along a winding path. My consciousness of surrounding objects consisted only
of kinaesthetic imagery of shrinking, dodging or otherwise avoiding branches, rocks and
trees. I was distinctly conscious of an awkwardness with tensions in the trunk, legs,
face and shoulders, all of which contributed to an awareness that I was uncertain where
to step. I then found myself at a gate. I stepped back and watched a visual image of
myself looking over the gate up the hill. Suddenly I lost the "visual me" and was
looking at the gate at close quarters. The transition was almost instantaneous. I
now visualized the two gate posts distinctly, together with the braces attached to the
top of each post. Just beyond the braces was the woven wire, silvery in color, as if
it had been galvanized recently. I was then conscious of fingering the wires, at which
instant the visual imagery became more distinct. I was aware, next, that just be-
yond the gate was a hill. I did not visualize the hill distinctly; it was merely a brownish
haze — a color which represents rank undergrowth to me; but I was distinctly conscious
of tactual and auditory imagery of being in the shadow of a hill. The tactual experi-
ences refer to changes in temperature and the auditory to changes in echo. I retraced
my steps down the hill and while on the way noticed a group of oak trees which I
had "seen" on my way up. I visualized their peculiar dark brown trunks, knotted
and gnarled, and had tactual imagery of running my hand over the bark, covered with
moss and lichens. I could see upward as far as the first branches but beyond that the
trees faded into a hazy background, thence into nothingness. The brown of the
trunks was faded and dim, the color irregular in distribution, giving the effect of a
pouring rain on a window pane — a "wiggly" appearance.
At this point in the dream I was conscious of the person who went with me to
the second strange house. This consisted of a colored visual schema at my left, too vague,
however, to identify, and also of a peculiar motor "start," characteristic of a sudden
consciousness of a person near you. We had a short conversation concerning the oak
.trees, the details of which I cannot remember. I recall that both his words and mine
were in my own verbal imagery.
3i8 RAYMOND H. WHEELER
I next found myself at the edge of the creek, vividly aware that I was facing the
creek and that the first strange house I had entered was back of me. This was present
in terms of a faint brownish haze off at the horizon back of me. The water in the creek
was sluggish and a dark opaque, oily green. The water almost overran the banks.
Next to the edge of the stream I visualized masses of dead, pale, straw-colored grass,
most of which was very hazy except for the larger tufts in which latter I seemed to
"see" some of the individual blades. The grass lined both banks of the stream and
extended over into the water. Suddenly I "saw" a collar and necktie floating down
the stream, too far out to be reached. I viewed them indifferently until I had the
vocal-motor: "It is A's collar!" Upon experiencing the verbal imagery I was con-
scious of intense grief, represented by marked feelings of stuffiness in my chest and by
tendencies to sob. I awoke and actually found myself sobbing. Although I did
not notice it particularly at the time, I am certain, on recalling the imagery of the
collar, that it was white. The tie was blue with large brown bands running diagonally
across it. I could not "see" those portions of the objects below the surface of the
water.
DREAM 2
I was seated in a passenger coach, midway between the ends, on the right side of
the car and next to the window. I could "see" the vague form and outline of the sun-
bleached red plush seat directly in front of me. For some little distance in front I
could distinguish the yellow woodwork between the windows but farther down the
coach the woodwork became very indistinct. A brilliant light was shining through
the windows, illuminating the coach. Mixed with the visual images of the plush
seats were tactual images of the friction of my trousers as they adhered to the plush,
preventing me from making slight movements with ease. I also had clear tactual
imagery of my arm as it rested in the sill of the car window.
The objects in the surrounding country were vague and fleeting and seemed to
pass by very rapidly. They were of the brightness and tint of vegetation in the sandy
desert. Now and then I was conscious of vague outlines of the rolling hills, stretching
out in the far distance. I was travelling northeast, indicated to me by a large area of
dense blackness which was projected beyond the side of the car off to the front and to the
left. This dark or black horizon meant north. The relation of the black schema to
the side of the car I interpreted to mean northeast. Then I noticed that my position
in the coach had changed. I was riding backward. The dark cloud was now at my
back and to my right. I became conscious that my brother was with me. This was in
terms of a gray splotch about the size of a man, in sitting position opposite me. My
brother said: "We are coming to the Blue Mountains." Although he seemed to be
speaking these words, for I was listening in his direction, the words themselves came
to me in my own verbal imagery.
Looking out of the window to my left I visualized a range of steep mountains
rising abruptly from the desert. Here the brightness of the landscape greatly increased;
various vaguely outlined hills passed by the window rapidly; the hills toward the
south were dark blue, fading into a dull grey the nearer they were to me. Mixed with
the awareness of this blue schema was a tactual image of tall thick grass soaked with
rain. As the number of hills increased I was conscious of a distinct feeling of relief,
referred to muscular relaxations about the brows, eyes and jaws. The thicker the
hills became, the more marked became the relief until it merged into wonder and sur-
prise. The muscles of my chest became tense; I was conscious of tendencies to smile.
I noticed that on the sides of the nearest hills there were dark patches. Then I vocal-
VISUAL PHENOMENA IN DREAMS OF BLIND SUBJECT 319
ized my brother's words as he spoke again: "The dark places are where they graze
their sheep. The dry and sandy places are where the coyotes live." Following
these latter words I had fleeting visual images of a comparatively level stretch of
ground, covered with innumerable greyish rocks, scattered thickly over a faded,
yellowish-brown soil.
DREAM 3
I was in the middle of a river, astride a log. All about me the water for twenty
or thirty feet was jet black and so dense that I could not "see" below the surface. The
log was about eight inches in diameter and quite distinctly visualized as an old alder
snag which was smooth, with no bark and a very light grey in color. The size of
the log is an interpretation partly from the visual image and partly from the extent
to which it protruded out of the water. Its smoothness was present to consciousness
partly as an interpretation from the visualized surface and partly from the fact that
my legs were adjusted in such fashion as to prevent my slipping on the surface.
I had very distinct motor imagery of balancing myself. This involved my legs,
trunk and arms. I judge that I was about one third of the distance back from the
front end of the log which I gather from the manner in which the log responded to my
movements. I was going down stream rapidly. This latter consciousness came to
me in terms of a visual image of a dark, ribbon-like streamer indicating how the water
was being "cut" by the passage of my feet through it. I also had tactual imagery
of water rushing rapidly by my hand, an experience similar to the sensations one re-
ceives when placing his hands in the water as he rides in a fast boat. Together with
this latter visual imagery was auditory imagery of the swishing of the water. I do
not know how fast the current was flowing but I seemed to be travelling faster than
the stream. I had tactual-kinaesthetic imagery of being hurled rapidly through space
in the direction of down-stream.
All about me, especially to the left and in front, small fish, about ten inches long,
were continually jumping out of the black, inky water. They leaped only a few inches
above the surface, looked at me and disappeared with a croak. I could not visualize
the fish clearly but merely got their general shape and size. Sometimes they would
appear only on the surface of the water; they would extend their heads upward until
their gills appeared when they would utter a peculiar sound and disappear again. In
many instances I had auditory imagery of their croaks together with a flash of yellow
light, hovering, temporarily, about the region of their mouths. In other instances I was
aware only of the yellow light, which, however, meant to me that they were making their
peculiar sounds. Some of the fish were striped grey and black, the stripes running
across their sides from the lower front to the upper back, thus making diagonal bars
across their bodies. I was next conscious of the vocal-motor: "These are singing carp."
Then I had a visual image of a fish net lying upon the log in front of me, and the vocal-
motor: "I have a net and will get some."
DREAM 4
I was under the south end of a bridge. I was facing the south with the stream in
back of me. I have a very distinct remembrance of visualizing the supports of the
bridge as I looked through them toward a cloud of yellow light off in the distance. This
cloud of yellow light meant south. I did not visualize the bridge above me but was
aware in auditory terms of the faint echo and in tactual terms of the nearness of the
bridge to me. I then had visual and kinaesthetic imagery of piling rocks into a gunny
sack. I was greatly disturbed both by the fear of getting caught and by my bull-
32° RAYMOND H. WHEELER
dog which kept barking, and pulling vigorously at my trouser's leg. Here the tactual
and auditory imagery was profuse and clear. The imagery of the dog's form was in-
distinct save for a bright straw-colored schema which always represents this dog to me. I
was holding one corner of the sack with my left hand (tactual, motor and visual im-
agery) and the sack itself was a dark yellowish-brown. I could not "see" the weave
and did not notice any tactual imagery of it. Then there appeared the verbal imagery:
"I want A to see these before anyone finds me." Then I was vividly aware of the
fact that D might find me. This consisted of tenseness about the chest and abdomen,
with "sinking sensations" in the region of my stomach. D was represented to me in
terms of peculiar flashes of color which corresponded to his voice. I then awoke.
Three striking peculiarities stand out in the dreams given
above. First is the appearance of associations between
visual and other sense modalities, which, in every instance
conform to synaesthetic phenomena in the reagent's waking
life. Schematic forms of persons are identified by color.
Persons' voices are recognized by their color; directions of
the compass, the "croaking" of the singing carp, the bull
dog, are all identified by certain colors. The reagent's
auditory imagery is exceedingly deficient both in dream
and in waking life but this deficiency is largely compensated
for by visual associations or synsesthetic phenomena.
Secondly, our subject has the peculiar tendency to vocalize
the spoken words of another person's voice in terms of his
own vocal-motor imagery. Colors provide the necessary
qualitative differences for identification. This is true both
of waking and of dream consciousness. Verbal imagery is
very rarely syncopated or abbreviated and is very definite
in details of enunciation and of expression.
Thirdly, there is to be noticed a peculiar tendency, in the
subject's dreams, to visualize himself at a distance. Various
details, which the writer has been able to obtain on this
point, indicate that the subject often has a "visual me"
in dreams. This "visual me" is evidently a product of
repeated tendencies on the part of the subject to visualize
himself as he walks about in order to ascertain, if possible,
whether he showed his blindness in any peculiarities of walk-
ing. This "visual me" appears to him frequently, in dreams,
if he is conscious of being watched. The visualized figure
consists mostly of hands and feet, separated and in move-
ment as if in walking. Peculiar feeling complexes are associ-
VISUAL PHENOMENA IN DREAMS OF BUND SUBJECT 321
ated with this visual imagery but otherwise the experiences
have not been definitely analyzed. Occasionally this "visual
me" appears close to the subject but it is usually localized
some distance away.
Visual imagery still outnumbers other modalities both in
the subject's dreams and in his waking life. This imagery
has now become vague in form and outline. For example,
persons lack outlines of eyes, mouth, ears, fingers, and de-
tails of trunk and legs. The size and duration of the imagery,
however, have remained practically normal. Visual imagery
of small objects similarly lacks definiteness but possesses
a greater degree of clearness than does visual imagery of
larger objects. This is undoubtedly due to the fact that the
act of handling small objects makes it easier for the reagent
to visualize them. In fact, when the subject wishes to vis-
ualize an object clearly he always endeavors to explore it
in tactual-motor fashion. In the first dream it will be
noticed that as soon as the reagent found himself fingering
the wire in the fence, its color increased in vividness and the
outlines of the visual imagery became more distinct.
All visual imagery lacks details as well as form and out-
line. The room visualized in the first dream was devoid of
furniture. In like manner were lacking the details of the
telephone, the lace curtains, the chair, the scenery along the
path, the trees, bushes, clothing, etc. A hill is a mass of
color with light and dark patches; mountains are clouds of
color; water is lacking in detail of wave or brightness; trees
are visualized only in part, and so on.
Taking the place of details in visual imagery are auditory-
vocal-motor, tactual and kinaesthetic-organic experiences.
For example, the visual imagery of oak trees in dream i
is pieced out with tactual imagery of exploring the surface
of the trunk. The visual imagery of being seated in a chair
is supplemented by tactual imagery. The same is true of
talking through a telephone.
Kinaesthetic imagery is exceedingly clear and persistent.
In many instances the reagent has hesitated in calling these
experiences "images" owing to their vividness, and dis-
322 RAYMOND H. WHEELER
tinctness of localization. Such imagery is very readily
recalled. Organic complexes constitute an important feature
of our reagent's dreams. While experiencing joy, for instance,
he feels the characteristic changes in respiration, the tenden-
cies to smile, and the pleasantness; while in fear he is able
to detect the tensions in the throat, the organic changes from
the regions of the stomach and diaphragm, and the altered
breathing. Again, in anger he is aware of the jaw tensions,
the tightness of arms and hands, and the changes in facial
expression, all of which seem to be exact copies of experiences
in waking life. Organic processes seem to be recalled with
greater readiness than visual details.
We believe that the above descriptions demonstrate that
the introspective method — at least the terminology — can be
applied in the description of dreams. Whereas it may be
contended that the chief interest in dreams pertains to their
function rather than to their content, it is obvious that the
significance of dreams can be much better understood if
their content is first described in minute detail. Such descrip-
tions should be obtained by using a method patterned after
introspection rather than after interpretation.
VOL. 27, No. 5 September, 1920
THE PSYCHOLOGICAL REVIEW
THE PHYSICAL BASIS OF NERVE FUNCTIONS
BY LEONARD THOMPSON TROLAND
Harvard University
I. INTRODUCTION
Modern experimental investigations of the laws governing
nervous action, combined with theoretical studies based
upon these investigations, provide us with the materials for a
very definite and satisfactory conception of the physical
mechanism of the process. It is the purpose of the present
article to sketch a rough picture of the probable nature of
this mechanism, in the hope that the sketch may prove valu-
able to some psychologists who may not have followed closely
modern developments in this field. I do not wish to claim
any essential originality for the general ideas to be presented,
although it is my belief that the total picture which I shall
outline has not been offered heretofore and that some of the
suggestions which will be made concerning details are new.
My views are founded mainly upon the writings of Nernst,1
R. S. Lillie,2 and Lucas.3
II. THE GENERAL MECHANISMS OF EXCITATION AND
STIMULATION
It was for a long time suspected that the nerve impulse
is essentially an electrical phenomenon, since the electrical
1 Nernst, W., 'Zur Theorie des elektrischen Reizes,' Arch.}, d. ges. Physiol., 1908,
122, 275-315; and other papers.
1 Lillie, R. S., 'The Relation of Stimulation and Conduction in Irritable Tissues
to Changes in the Permeability of the Limiting Membranes,' Amer. J. of Physiol.,
1911, 28, 197-223; and other papers.
3 Lucas, K., 'The Conduction of the Nervous Impulse,' London, 1919. Also many
papers.
323
324 LEONARD THOMPSON T ROLAND
action current of a nerve provided the only direct evidence
of the existence of the impulse. Paradoxically enough, this
index of excitation consists not in the appearance of a state
of electrification in the nerve but rather in the partial dis-
appearance of such a state which is already present in the
resting cell; and which manifests itself through the 'current
of rest' or 'demarcation current,' obtained when the cut
end or injured surface of the nerve is connected through a
galvanometer to the nerve sheath. The current flows from
the sheath to the injured portion, indicating that the former
is positively charged with respect to the latter. When the
nerve cell is excited this positive electrification of the sheath
decreases momentarily.1
The electrical theory of matter, and in particular the
so-called theory of 'electrolytic dissociation'2 offers us the
materials for a simple explanation of the resting electrification
of the nerve. A nerve fiber, like almost any other living
cell, consists of an aqueous solution of many substances, both
organic and inorganic. Among these are substances which
become ionized in solution, that is, their individual mole-
cules are broken up into electrically charged particles, positive
and negative ones being produced in equal numbers. Promi-
nent among ions of this sort are to be counted those of carbonic
acid, the substance which is produced by the respiratory
function of the cell. The ionic particles, in common with
all of the molecules of the cell, are endowed with a rapid
vibratory motion which represents the temperature or
thermal energy of the substance. In accordance with the
general theory of solutions, the motion of the dissolved
particles among the molecules of the solvent, water, resembles
that of the molecules of a gas in free space. This means
that they must exert a pressure upon any surface which tends
to interfere with their free diffusion.
1 A simple account of the electrical phenomena exhibited by nerve and muscle
will be found in Howell, VV. H., 'A Text-Book of Physiology,' 1909, Chap. IV, pp.
96-110.
* For a simple presentation of the essentials of the modern electrical theory of
matter, reference may be made to the book by the writer and D. F. Comstock, 'The
Nature of Matter and Electricity,' 1917. On 'electrolytic dissociation' see pages
139-141, and on pressure due to molecular motion see pages 106-109 of this work.
THE PHYSICAL BASIS OF NERVE FUNCTIONS 3*5
The liquid mass of the cell is contained within a mem-
brane which may be conceived as a more or less solid en-
velope. This membrane is not necessarily to be identified
with the myelin sheath in the case of the neYve fiber, and it
may simply consist in some peculiar condition of the mole-
cules at the surface of the general cell mass. In general
the bounding surfaces of liquid masses behave like actual
membranes. Now it is obvious that the ions in the general
cell mass will exert an outward pressure on this membrane,
especially as there is probably an aqueous medium outside
as well as inside of the cell. If the membrane is absolutely
tight or impermeable to these ions there will simply be a
tendency to distend it, but if on the other hand the membrane
is somewhat porous certain of the ions may pass through.
It is clear that small ions will pass through more readily
than large ones, and in the case of those of carbonic acid the
hydrogen ions will almost certainly be the smaller. Hydrogen
is the smallest of the atoms and its positive ion is probably
the smallest of all known physical particles. If bare hydrogen
ions occur in solution it is probable that no membrane could
possibly exist which would be capable of holding them back,
since the membrane must itself have a molecular structure.
The carbonate ions must be very much larger than the ions
of hydrogen, not only because they contain a number of
atoms in combination instead of one, but because the atoms
themselves are many times larger than those of hydrogen.
It is, therefore, almost a priori certain that if a membrane
is relatively impermeable the hydrogen ions will still pass
through it, while the carbonate ions will be held back.
Since the hydrogen ions are by nature positively charged,
the outcome of this selective diffusion through the cell
membrane must evidently be to form a layer of positive
electrical particles on the outside of the cell. On the inside
of the cell there will be a corresponding layer of negative
particles which are held in position by the attraction exerted
upon them by the external positive layer. This attraction
is mutual and not only causes the diffused positive ions to
remain in the immediate vicinity of the membrane but limits
326 LEONARD THOMPSON TROLAND
the number of them which can pass through the membrane,
since when their concentration on the outside reaches a certain
limit there will be just as many returning into the cell per
unit time as leave it. Thus a definite state of equilibrium
is established which involves an 'electrical double layer,' of a
magnitude depending upon the exact permeability of the
membrane and the internal concentration of the ionized
substance. The existence of such an electrical double layer
at the membrane may be said to constitute a polarization of
the latter, this polarization being capable of variations in
degree under different conditions.
It is clear that the polarization of the cell membrane just
considered explains the demarcation current, even the
direction of this current corresponding to that theoretically
deduced. When an electrical connection is made between
an injured and an uninjured portion of a nerve, this con-
nection is virtually established across the cell membrane so
that the mechanism of the membrane acts in a manner
similar to an electric 'battery.' The energy of this 'battery'
is derived from the compression of the ionized substance
inside of the cell, and the continued flow of the demarcation
current must involve a diffusion of some substance, viz..,
hydrogen, through the cell wall. This process is similar to
that which occurs in the special form of electrical 'battery'
known to physical chemists as a 'concentration cell.'
The next problem which we face is that of determining
the mechanism of stimulation of the resting cell. It is a
familiar fact that an externally produced electric current or
voltage provides a very ready means of stimulating nervous
tissue. The great sensitivity of such tissue to electric cur-
rents is by itself almost conclusive evidence of the essentially
electrical nature of the nerve process. The stimulating
power of electrical currents depends, however, very radically
upon the exact time relations of the current. An alternating
current of the right frequency stimulates more readily than a
direct current, but if the frequency is very high there may be
no stimulation at all, even with very large currents. It is
found experimentally that the intensity threshold for a given
THE PHYSICAL BASIS OF NERYE FUNCTIONS 327
nerve is proportional to the square-root of the frequency.
Nernst1 succeeded in explaining this relationship mathe-
matically on the basis of an assumption as to the physical
conditions underlying the threshold. He supposed that in
order just to stimulate a nerve cell, the current acting upon
the cell must polarize the cell membrane to a certain critical
degree, this polarization being a natural result of a current
flow, which piles one kind of ions up on one side of the mem-
brane and draws them away from the other side. By com-
bining this assumption with the known laws of electrical
action and diffusion Nernst arrived at a law connecting the
time required to stimulate with the intensity of the stimulus.
Although Hill,2 and others have found it necessary to com-
plicate Nernst's original theory by further details, the
quantitative success of his original hypothesis stands out
as a unique achievement in the application of mathematical
physics to biological phenomena.
It is evident, however, that Nernst's assumption that the
physical basis of the nerve threshold is the establishment
of a certain degree of polarization in the cell membrane must
be modified to fit the fact that the membrane is already
polarized before stimulation. It was pointed out by R. S.
Lillie3 that this modification might take the form of a mere
change of algebraic sign; in other words, stimulation of the
nerve may require a critical ^polarization of the membrane,
this depolarization although definite in amount not necessarily
being complete. Lillie further pointed out that this assump-
tion is in harmony with many facts concerning conditions of
stimulation of nervous tissue. It harmonizes firstly with
Pfliiger's law of electrical stimulation,4 according to which
there is stimulation at a negative pole (cathode) upon closing
the circuit and at the positive pole (anode) upon opening
the circuit, when both of these poles are applied to the
1 Loc. cit.
1 Hill, A. V., 'A New Mathematical Treatment of Changes of Ionic Concentration
in Muscle and Nerve Under the Action of Electric Currents, with a Theory as to their
Mode of Excitation,' /. of Pkysiol., 1910, 40, 190-225.
* Loc. cit.
4 For a statement of Pfliiger's law or laws, see Howell, loc. cit., pp. 89-91.
328 LEONARD THOMPSON TROLAND
outside of the nerve. The cathode may be regarded as
spraying the nerve with electrons,1 which are negatively
charged and which consequently combine with the positive
particles on the outside of the nerve, neutralizing them and
thus depolarizing the membrane. The anode, on the other
hand, sucks electrons away from the outside of the cell,
thus increasing its coating of positive particles at this point.
Facts to be considered below lead us to suppose that under
these conditions the membrane will react in such a way
as to compensate for this increased polarization, so that when
the circuit is opened and the auxiliary polarizing force of
the anode is removed a depolarizing action will occur. It
is a well recognized principle that anode stimulation at the
'break' occurs less readily than cathode stimulation at the
'make.'
Stimulation of nerve by many agents other than the
electric current can also be explained on the assumption that
the essential condition of stimulation is a certain depolar-
ization of the membrane. One way in which to depolarize
the membrane would be to place the cell in a solution con-
taining positive and negative ions very similar in physical
character to those within the cell itself. Under these con-
ditions there would be just as much of a tendency for positive
ions to pass into the cell as to pass out, so that the produc-
tion or maintenance of polarization by the mechanism of
selective diffusion above described would be impossible.
This explains the stimulating power of various 'electrolytes.'
It also explains why these substances, including carbonic
acid, eventually bring about a permanent depression of
activity or 'narcosis' in the nerve.
Another method which is obvious a priori for stimulating
a nerve is to injure or destroy the membrane, which depolar-
izes the latter because negative as well as positive ions are
permitted to diffuse through it. So-called mechanical stimu-
lation of the nerve is evidently of this sort, depending upon a
gross maceration of the cell envelope. The action of certain
organic narcotics, such as chloroform, ether, alcohol, etc.,
1 Cf. Comstock and Troland, loc. cit., pp. 24-25.
THE PHYSICAL BASIS OF NERVE FUNCTIONS 329
which first excite and then depress the nerve, probably
depends upon a chemical destruction or impairment of the
membrane. The stimulating effect of other conditions,
such as desiccation and heat, can be explained in similar
terms.1
III. THE SPECIFIC MECHANISMS OF THE THRESHOLD IM-
PULSE PROPAGATION, AND OTHER NEURAL PROPERTIES
It is evident that the Nernst theory of stimulation in
the modified form outlined by Lillie demands a 'negative
variation' of the 'current of rest,' as empirically found,
since depolarization would necessarily manifest itself in
this way. It does not follow, however, from any statements
made heretofore that this negative variation will be propa-
gated along the cell or fiber from the point of stimulation.
Lillie has pointed out that to explain propagation of the
variation we must suppose not only that the polarization of
the membrane depends upon its differential permeability
but that its relative impermeability depends upon its degree
of polarization. Polarization and permeability, in other
words, are the two essential factors in the nerve process,
and they are related to one another in the propagation of
the nerve impulse in a manner analogous to the relation
of pressure and displacement in the propagation of sound2
or the relation of the electric and magnetic vectors in the
propagation of light.3 These relations are all such that a
change in one of the quantities always involves a change in
the other, the locus of the secondary change not coinciding
with or falling wholly inside of that of the primary change.
This relation necessitates propagation.
It is easy to see why the degree of polarization of the
nerve cell membrane should depend upon its degree of perme-
ability, since it was from assumptions regarding the nature
1 On the various means of exciting nervous tissue see Schafer's 'Text-Book of
Physiology,' 1900, pp. 459-468.
* See, e.g., Duff, A. W., 'A Text-Book of Physics,' 4th ed., pp. 518-519; or Poynt-
ing and Thomson, 'A Text-Book of Physics,' 1904, 2, pp. 12-14.
3 See, e.g., Houstoun, R. A., 'A Treatise on Light,' 1915, Chap. XXII. This is a
mathematical exposition; a simple, qualitative treatment is difficult to find.
33° LEONARD THOMPSON TROLAND
of this permeability that we were enabled to deduce the exist-
ence of polarization. It is not so obvious, however, why the
permeability of the membrane should alter when some external
force, such as an electric current, brings about its depolar-
ization. A moment's consideration, however, will show that
such artificially produced depolarization should have some
effect upon the constitution of the membrane. When the
polarization is present the substance of the membrane is
subjected to electrical stresses which are represented by
parallel lines of force connecting the positive charges on the
outside of the membrane with the negative charges on the
inside. In accordance with the general electrical theory of
matter these electrical stresses will necessarily produce some
distortion of the molecular or atomic structure of the mem-
brane, which distortion will have a direction determined by
the impressed electrical forces. When these forces are
removed by depolarization this distortion will tend to dis-
appear. The distortion may well be of such a character as
to render the membrane less permeable than would be the
case if such distortion were absent. In this event, the
depolarization would have the effect of increasing the perme-
ability.
One conception of the molecular structure of the membrane
which has occurred to me, pictures the membrane molecules
as being considerably longer than they are wide, with opposite
electrical charges on either side in the middle of the molecular
length. The individual molecules are conceived to be
rotatable, like compass needles, about their centers of mass.
The electrical field produced by the polarization will then
tend to dispose all of these molecules with their long axes
at right angles to the field, that is parallel to the plane of
the membrane, and in this position they will offer maximal
obstruction to any particles tending to move perpendicular
to this direction. When the depolarization field is removed,
however, the molecules will tend to swing into position at
right angles to the one just considered, on account of the
mutual attractions and repulsions of their charges. In this
position they will offer much less resistance to the passage
THE PHYSICAL BASIS OF NERVE FUNCTIONS 33 l
of particles perpendicular to the plane of the membrane.
This mechanism is of course only symbolic, but illustrates
the general principles which are involved.
Apart from purely physical reasoning, we have excellent
biological grounds for believing that stimulation of a living
cell will result in an increased permeability of its enveloping
membrane. It is through modifications in the permeability
of this membrane that the cell regulates the income and
outgo of chemical materials. Food substances absorbed
from the environment and waste products excreted must pass
through the membrane, and it is to be expected that both of
these transfers of substance will be increased during a state
of excitation of the cell, since this latter state involves an
increased chemical activity. Lillie has cited a number of
cases in which an increased permeability of the cell mem-
brane resulting from stimulation can be clearly demonstrated.
The recent work of Tashiro on the liberation of carbon
dioxide by stimulated nerve fibers is evidence in the same
direction, and also corroborates the idea that the ions which
are involved in the polarization of the nerve cell membrane
are those of carbonic acid.
Tashiro1 finds that a small amount of carbon dioxide is
given off by the nerve fiber even in the resting state but that
this amount is markedly increased during stimulation.
This indicates that the normal permeability of the mem-
brane is such as to permit a slight diffusion of carbonate ions
and that the depolarization accompanying excitation is the
result of an increased permeability to these ions rather than a
decreased permeability to the positive hydrogen ioqs. Our
initial theory regarding the cause of the polarization of the
cell membrane demands that an increase in the permeability
should result in a decrease in the polarization provided the
permeability is already practically perfect for the positive
ions. That this requirement is fulfilled is indicated not
only by the slight diffusion of carbon dioxide during the
resting state but also by the free permeability of the mem-
brane at all times to water and the ions of water.
1 Tashiro, S., 'Carbon Dioxide Production from Nerve Fibers when Resting and
when Stimulated,' Amtr. J. of Physiol., 1913, 32, 107-136.
33 2 LEONARD THOMPSON T ROLAND
The facts which we have just been considering show that
the reciprocal relationship between permeability and polari-
zation is a mutual or 'symmetrical' one. In other words,
these facts prove that a decrease in polarization beyond a
certain critical point results in an increase in permeability.
The analogy between the principles of propagation of nervous
energy and those of acoustic and radiant energy is thus very
close, a change in either one of the two principal variables
resulting in a reverse change in the other variable. The
reciprocal relationship between these two factors in the
case of the nerve impulse is not, however, absolutely sym-
metrical. It is almost certain that any initial change, no
matter how small, of the permeability will result in a reverse
change of the polarization, but the fact of the threshold
indicates that an alteration of the polarization does not
bring about an increase in the permeability until a certain
critical depolarization has been developed. Indeed certain
phenomena which we shall soon consider indicate that before
this critical point is reached, there is a tendency for the
permeability of the membrane to decrease below normal as a
result of depolarization. The polarization and the perme-
ability of the membrane appear to enter into a system in
which there is a point of unstable equilibrium.
The nature of this system can be illustrated by a mechani-
cal analogy which is effective as a classroom demonstration.
Suppose that we tip a chair gradually forward by applying a
finger to the back of the chair. At first the weight of the
chair resists the tipping force, but when the center of gravity
passes over the very small base of support the chair parts
company with the finger and falls to the floor with a crash.
The tipping of the chair up to the point of unstable equi-
librium represents the changes in the character of the mem-
brane system which must be produced in order to pass the
threshold, while the crashing of the chair to the floor, once
the equilibrium point has been passed, represents the libera-
tion of the internal energies of the nerve cell which constitutes
the state of excitation. The degree of energy thus liberated
is evidently dependent almost wholly upon the inherent
THE PHYSICAL BASIS OF NERFE FUNCTIONS 333
nature of the system and not upon the intensity of the
stimulus or tipping force.
The laws of the nerve membrane system, formulated in
more exact terms, would probably read somewhat as follows.
An increase in permeability always produces a decrease in
polarization. An initial decrease of polarization, however,
first results in a decrease in permeability which tends to
compensate for the initial change. But if the depolarization
reaches a critical or threshold amount the 'sign' of the
change is reversed, so that an increase in permeability results.
As soon as this increase begins, the initial depolarization is
further augmented by the law which makes polarization
depend reciprocally upon permeability, so that the equi-
librium of the nerve membrane system is completely upset,
the polarization now decreasing to a minimum while the
permeability increases to a maximum.
The supposition that the first reaction of the membrane
to a decrease in polarization is a decrease in the permeability
rather than an increase is in harmony with a variety of
facts concerning the nerve function. It explains among
other things DuBois-Reymond's law of electrical nerve
stimulation, according to which a nerve is stimulated only
by a change in an electrical current and not by a steady flow
of electricity. The investigations of Waller,1 Hill,2 Lapicque3
and others have shown that there is a certain rate of appli-
cation of an electrical voltage which stimulates the nerve
with the least total expenditure of energy. The failure of
a slowly applied electric voltage to stimulate — at the cathode
—can be understood if the first action of the membrane
is in the direction of adaptation or compensation, the mech-
anism of which involves a decrease of the permeability
below normal; thus tending to increase the polarization
above normal or to maintain it at normal in the face of an
external depolarizing agency. Such an action is evidently
a physical possibility if, as we have found reason above to
1 Waller, A. D., 'The Characteristic of Nerve,' Proc. Roy. Soc., 1899, 65, 207-222.
1 Loc. cit.
^ * See Lapicque L., and Legendre, R., 'Relation entre le diametre des fibres ner-
veuses et leur rapidite fontionnelle, Comptes rendus, 1913, 157, 1163-1165.
334 LEONARD THOMPSON TROLAND
suppose, the normal permeability does not completely inter-
fere with the diffusion of the negative ions but simply impedes
their movement to a certain degree. If the external de-
polarizing agency is applied very rapidly the membrane does
not have time to develop its compensating reaction, so that
the threshold depolarization is reached and excitation results.
We have seen that the 'break' stimulation at the anode
is best explained on the assumption that the membrane
tends to compensate for the excess of positive ions at the
anode by decreasing its permeability at this point. This
is apparently another instance of the general tendency of
the membrane to adapt itself to a stimulus by permeability
changes so as to maintain the resultant polarization constant.
Stimulation at the anode upon interruption of the current
will result only if this interruption is sufficiently quick, since
if it is slow the membrane will have time to compensate again
and the resultant change in the polarization will be insufficient
to cause stimulation.
These views regarding the changes in the condition of the
membrane at the cathode and anode respectively should
lead us to look for some further specific effects of such modi-
fications. The phenomena of 'electro-tonus'1 are of this
sort. It is to be anticipated that the stimulability, or the
stimulation threshold, of the nerve- will depend upon the
condition of the membrane, in particular upon the dynamics
of the equilibrium between its permeability and its polariza-
tion, so that a compensating change in the permeability will
result in an alteration of the threshold. It is a familiar fact
that the stimulation threshold decreases at a cathode while
it increases at an anode, indicating in terms of our theory
that decreasecl permeability results in increased stimulability
and vice versa.
The above deductions are further borne out by a study
of the alterations in stimulability which take place during and
after any given process of excitation in the nerve fiber. It
is well known that after such a fiber has once been stimulated
it cannot again be set into excitation until a certain interval
1 See Howell, he. cit., pp. 88-89.
THE PHYSICAL BASIS OF NERyE FUNCTIONS 335
has elapsed, this interval being called that of refractory
phase.1 Refractory phase is evidently represented in our
physical hypotheses by the state of collapse of the membrane
which results from the attainment of the threshold depolar-
ization. If excitation consists in an increase of permeability
to a maximum, then it is clear that further excitation is
impossible until the original impermeability has been at
least partially restored. So long as no restoration whatever
has occurred the cell remains in what is known as absolute
refractory phase.
The fact that this condition is only temporary proves that
a mechanism of restoration of the normal condition of the
membrane exists, this mechanism probably being identical
with that which enables the membrane to react in a com-
pensating manner to small depolarizing forces. The recovery
is gradual and while it is in progress the cell is in a so-called
state of relative refractory phase. In this condition its
stimulability is depressed below normal to a degree which
is greater the less the recovery which has been achieved at
any instant. Since the recovery must consist essentially
in a progressive decrease in the permeability of the membrane
these facts evidently correspond with the view that stimula-
bility is inversely proportional to the permeability in question
or that the threshold of stimulation is directly proportional
to it.
It is an empirical finding that the relative refractory
phase is followed by a phase of hyper-excitability during which
the threshold of stimulation is lower than normal. This
effect can evidently be attributed to an 'over-shooting' of
the permeability decrease which constitutes the recovery of
the membrane. Such 'over-shooting,' according to our
assumptions, should be accompanied by an increase in the
stimulability of the nerve above that obtaining in the normal
condition. The 'over-shooting' may be viewed as a delayed
consequence of forces of compensation set into action by
the initial operation of the stimulus.
1 On refractory phase and the course of excitability after stimulation, in general,
consult Bayliss, W., 'Principles of General Physiology,' 1915, pp. 389-390.
33 6 LEONARD THOMPSON T ROLAND
The above considerations are evidently in harmony with
the facts which indicate that a nerve impulse set up during a
state of refractory phase has a smaller amplitude or intensity
than a normal impulse, while one which is generated in a
phase of hyper-excitability possesses an amplitude or in-
tensity greater than normal. The 'amplitude' of a nerve
impulse must stand either for the amount of change in
permeability which results from stimulation or for the
amount of depolarization which is a consequence of this
change. It should be clear why these amplitudes vary with
the exact level of permeability which exists at the instant
of stimulation. If a nerve impulse is represented as a wave
of permeability change referred to normal permeability as a
base line it is evident that a decreased amplitude is repre-
sented by an elevation of the troughs of the waves above
the base line, whereas an increased amplitude must be
referred to a depression of these troughs below this line.
The 'all or none' principle, which we shall consider more
specifically below, may be interpreted as to mean that the
crests of all waves lie on a constant locus which represents
the invariable maximum permeability of the membrane.
IV. THE ENERGETICS OF NERVE PROCESSES
It is almost a necessary consequence of the above out-
lined theory of the physical nature of nervous activity that
such activity should involve metabolism. It seems almost
inevitable that the process of restoring the nerve membrane
to its rest condition after excitation should require the
expenditure of new energy. This energy would naturally
be obtained by the oxidation of some substance present in
the cell. It is' a fact that oxygen is required in order to main-
tain the nerve in a state of excitability. A nerve which is
caused to function in the absence of oxygen eventually falls
into a state of permanent refractory phase, exactly what
would be expected if oxygen is necessary in order to rebuild
the membrane.
It is a remarkable fact that although oxygen is required
in the nerve function there is apparently no generation of
THE PHYSICAL BASIS OF NEWE FUNCTIONS 337
heat due to the activity of the nerve.1 Very sensitive heat
detecting instruments capable of recording the heat pro-
duced by the oxidation of a single molecule in a portion of
space easily visible under the microscope have failed to
indicate any heat production whatsoever. This fact has
led some investigators to believe that the oxygen is not
required for metabolic purposes but has some other function,
such as one of catalysis. They do not make clear, however,
why in a non-metabolic function the oxygen should require
constantly to be replenished. The idea that the oxidation
actually is employed in an oxidative process may possibly
be reconciled with the absence of heat production in the
following way. The process of excitation, according to the
theory herein considered, involves a diffusion of carbonic
acid through the cell membrane and this diffusion, being of
the general nature of evaporation, should have a cooling
effect upon the cell. This effect is immediately followed
by the oxidative change, and the heat generated by this
latter change may be only just sufficient to counteract the
cooling produced by the diffusion. It would be quite reason-
able to suppose that these two quantities would almost
exactly neutralize each other, owing to the generally cyclic
nature of the process.
The apparent indefatigability of nerve tissue is probably
a consequence of the fact that only a very small amount of
energy is required in the nerve function, so that the substance
of the nerve cell can supply this energy by oxidation during a
very long period. It is clear that the oxidation which we
have assumed to occur in the rebuilding of the cell membrane
will tend to compensate for the loss of carbon dioxide from
the cell which occurs during excitation, so that fatigue will
set in only when the fundamental oxidizable material of the
cell is exhausted. In vivo this substance will naturally be
replenished, while in vitro the nerve becomes unfit for experi-
mental tests due to other causes, such as desiccation, long
before its metabolic fuel is exhausted.
Some views of the nature of the nerve impulse have
1 On these points, cf. Bayliss, loc. cit., pp. 378-379 and 390-391.
338 LEONARD THOMPSON TROLAND
apparently regarded it as a true cyclic process in the thermo-
dynamic sense.1 Such a process, resembling the propagation
of light in free space or of sound in a perfectly elastic medium,
would involve no loss or gain of energy at any point, the
original energy of the stimulus simply being transmitted
from one part of space to another. It is inconceivable,
however, that all cases of nerve functioning should be thermo-
dynamically cyclic. The original process of stimulation
apparently does not consist simply in an absorption of the
energy of the stimulus by the nerve; on the contrary, the
stimulus apparently serves merely to operate a trigger which
releases energies latent in the nerve itself. Moreover, if
propagation depends simply on the principle that one portion
of the nerve can be stimulated by the excitatory state of an
adjoining portion, it is natural to suppose that this trigger
process is repeated at all points in the nerve during the
propagation of the impulse.
It is, of course, conceivable that the energy released at
the point of stimulation is simply passed along the nerve
during the propagation without involving any further
expenditure or release of energy. There are certain cases of
propagation, however, to which this supposition can scarcely
apply. One of these is propagation through a so-called
* region of decrement' in which the amplitude of the nerve
impulse suffers a progressive diminution, but one which is
completely recovered from when the impulse emerges from
the region in question. We must suppose that the reduction
of the amplitude of the impulse in such a region involves a
loss of energy, and the restoration of the impulse to its
original magnitude when it passes out of a region of decrement
must involve the expenditure of new energy. Conduction
through synapses apparently involves processes of the same
general character.
Other reasons for refusing to believe that the nerve
impulse is thermodynamically cyclic lie in the metabolic
character of all other vital activities and in the nature of
1 On the nature of a thermodynamically cyclic process consult Lewis, W. C. McC.,
'A System of Physical Chemistry,' 1916, Vol. 2, pp. 29 ff.
THE PHYSICAL BASIS OF NERVE FUNCTIONS 339
the specific physical mechanism which we now believe is
involved in the nerve process. This mechanism requires a
change in the structure of the physical substance of the
nerve membrane, a re-arrangement of its constituent atoms
or molecules, and such changes always involve a degradation
of energy. Nervous tissue undoubtedly has a lower degree
of metabolism than any other living tissue, but this is not
equivalent to saying that its metabolism is zero.
V. THE BASIS OF THE ALL-OR-NONE PRINCIPLE
Modern studies, mainly those of Lucas and Adrian,1 have
made it quite clear that the action of the individual nerve
fiber follows a principle of 'all or none.' If a nerve cell is
set into excitation at all its excitation is ipso facto the greatest
which is normally possible for it. A stimulus of the highest
intensity can cause no greater response than one of threshold
intensity. The experimental demonstration of the validity
of this principle for the individual nerve fiber has involved
work of a very intricate and ingenious kind, but on the
theoretical side it requires very little effort to see that the
response of nervous tissue should be of the 'all or none'
type. If what the stimulus does is to upset a condition of
unstable equilibrium in the nerve membrane, the response of
the nerve must depend upon its own inherent nature and not
upon that of the stimulus. The old 'train of gunpowder'
analogy for the nerve impulse involved an 'all or none'
action, and the modern substitute for this classical mechanism
makes the same theoretical demands.
It is interesting to note that the 'all or none' principle
as applied to nerve activity forces us to think of such activity
in terms of fixed units of energy, so that we have a system
somewhat resembling that necessitated by the modern
quantum theory of radiation.2 The nerve process is quantal,
or to use a more biological term, it is isobolic. The concep-
tions of atomism and discontinuity seem to be creeping into
every branch of scientific analysis; not only are the chemical
1 See Lucas, loc. cit.
* See Comstock and Troland, loc. cit., pp. 46-49 and 182-189.
34° LEONARD THOMPSON TROLAND
elements atomic but also electricity, light, the determinants
of heredity, and finally the activities of nerve and muscle.
The acceptance of the 'all or none' principle for nerve
action does not, however, imply that all nerve impulses are
of the same magnitude. The magnitude or amplitude of a
nerve impulse depends upon the characteristics of the nerve
substance in which it occurs. These characteristics differ
for different nerve fibers in their normal condition and for a
single fiber in various abnormal, subnormal and super-
normal conditions. They are different at synapses from
what they are in the non-synaptic portions of the nerve.
In general, however, variations in the quantity of nervous
energy transmitted along a given nerve in unit time must be
conceived to depend upon the number of impulses or neural
quanta which pass through a cross-section of the nerve during
the time in question. This number is the nerve impulse
frequency, and there cannot be the slightest doubt that the
concept of impulse frequency is an absolutely fundamental
one for the theory of nerve action. It is probably as im-
portant for the understanding of such action as is that of
wave-length for the understanding of radiation phenomena.
However, the characteristics of a given nerve current are
not completely determined by a specification of its frequency,
since the amplitude, the length and the form of the individual
nerve pulse are not determined by frequency, although they
may serve to limit the latter.
VI. THE MECHANISM OF THE SYNAPSE
Facts summarized in a masterly way by Sherrington1
indicate that nerve conduction through a reflex arc differs
radically from conduction between two points in a single
nerve fiber. A reflex arc always involves one or more synapses
and all of the evidence points to the synapse as the locus
of the factors which differentiate reflex arc conduction from
simple nerve fiber conduction. Certain experiments of
Lucas2 provide us with facts which make it easy to construct
1 Sherrington, C. S., 'The Integrative Action of the Nervous System,' 1911, pp.
I4/
2 Loc. cit., pp. 17-22.
THE PHYSICAL BASIS OF NERVE FUNCTIONS 34 '
a theory of the nature of the synapse. These experiments
relate to the conduction of the nerve impulse through a so-
called 'region of decrement/ Such a region is provided by a
stretch of nerve fiber which has been narcotized, that is,
which has been subjected to the action of a narcotic, e.g.,
alcohol. An impulse in passing through a narcotized stretch
of fiber decreases progressively in amplitude, the total
decrease being proportional to the length of narcotized fiber
through which it has passed. If the narcosis is sufficiently
deep or the length of the region sufficiently great, the reduc-
tion of the impulse amplitude may be such as completely
to extinguish it. If, however, the reduction does not carry
the amplitude below a certain critical or threshold magnitude
the impulse, upon emerging from the narcotized stretch,
regains its normal amplitude and continues as if it had not
passed through the narcotized region at all.
All of the evidence at hand points to the view that the
action of a narcotic on a nerve cell consists essentially in a
permanent increase of the permeability of the cell membrane,
a physical state resembling that of refractory phase. Nar-
cotics are apparently membrane destroyers and the depth
of the narcosis is represented physically by the extent to
which the membrane has been injured or made permeable.
That this is a correct picture of the state of affairs in a
narcotized region is indicated by the fact that such a region
is electrically negative with respect to a non-narcotized
portion of the nerve. It is also shown by the fact that
narcotic substances, alcohol, chloroform, ether, etc., in
general increase the diffusion of materials through the cell
boundaries. The exact physical basis of 'conduction with a
decrement,' however, is less easy to picture. Such conduction
apparently demands that the degree of response of one por-
tion of the nerve fiber should depend upon the intensity of
the stimulus supplied by an adjoining portion. In other
words, in a region of decrement the self excitation of a nerve
fiber does not follow the 'all or none' principle. At present I
see no plausible explanation of this change in law.
It will be recalled that if a nerve is restimulated during
342 LEONARD THOMPSON TROLAND *
the relative refractory phase an impulse is generated which
has an amplitude less than normal, while if it is restimulated
during the phase of hyper-excitability the resulting impulse
has an amplitude greater than normal. The experiments
of Lucas show that the ability of the nerve impulse to pass
through a region of decrement without being extinguished
depends upon its initial amplitude. A normal impulse may
pass successfully while a subnormal one generated during
relative refractory phase may fail to pass. In another case,
both normal and subnormal impulses may fail while one of
supernormal amplitude generated during a phase of hyper-
excitability may succeed. These facts are employed by
Lucas to explain inhibition and summation on the assump-
tion that synapses, in connection with which these two pro-
cesses are most commonly found, are actually regions of
decrement. A synapse, according to this view, is a naturally
narcotized or auto-intoxicated portion of the nerve circuit.
The so-called 'resistance' of the synapse is an expression of
this condition. The 'resistance' is really a 'leakage,' but
owing to the 'all or none' character of the nerve function
the resistance cannot permanently lower the intensity of
individual pulses but can only determine whether a pulse
will pass through the synapse or not. The success or failure
of a pulse in attempting to pass through a synapse will
depend upon its initial amplitude and upon the depth of
auto-narcosis of the synapse. •
It is clear that if the individual impulses reaching a
synapse are separated by distances so great that the con-
dition of the nerve is restored to normal between each suc-
cessive impulse, the frequency of the pulses can affect their
ability to penetrate the synapse only by some accumulation
of effects in the synapse itself. However, if the frequency is
sufficiently high so that one impulse falls in the phase of
hyper-excitability of the preceding impulse, then this in-
creased frequency will aid the nerve current to pass through
the synapse. A further increase in frequency, however,
which causes one impulse to fall in the relative refractory
phase of the preceding one, will render the nerve current
THE PHYSICAL BASIS OF NERPE FUNCTIONS 343
less able than normal to penetrate the synapse. In case
the synapse is of such a 'resistance' that only supernormal
impulses can pass through there will be a relatively narrow
range of nerve frequencies which will be capable of being
conducted through the nervous arc, and the law governing
the relation of conductibility to frequency will be similar
to that of resonance, since impulse frequencies both higher
and lower than the available range will fail entirely. If,
on the other hand, normal impulses are conducted, nerve
currents of all frequencies up to a certain critical frequency
will pass the synapse but above this frequency there will be a
complete block.
Adrian1 makes it clear that these principles are adequate
to explain many of the facts of inhibition as well as of sum-
mation. An increase in the frequency of any impulse above a
certain critical value will evidently result in the inhibition
of this impulse and any process depending upon it, provided a
synapse is involved in the nerve circuit. Inhibition of one
nerve current by another may occur if the second current
impinges upon the same synapse as the first one and is of a
sufficiently high frequency. If two currents of different
frequencies combine, the resultant current must have a
frequency at least as high as that of the highest component,
and if the synapse blocks the high frequency component it
will also block the low frequency one. This explanation of
inhibition is clearly in harmony with the effects produced by
strychnine, rabies, tetanus toxin, calcium salts, and other
similar agents upon the nervous system, if we suppose that
these agents decrease the permeability of the nerve membrane
below normal or act in a direction opposite to that of nar-
cotics. Upon this assumption strychnine would tend to
obliterate synapses, in the physiological sense of the term;
regions of decrement would be wiped out, and consequently
all nerve impulses of whatever frequency would be trans-
mitted. Inhibitions would be converted into excitations and
the slightest stimulus would set the entire nervous system
into action. It will be recalled that previously we have
1 In the final chapter of Lucas's book, already quoted.
344 LEONARD THOMPSON TROLAND
associated hyper-excitability with a decrease in the perme-
ability of the nerve membrane below normal. The converse
fact that narcotics acting upon the nervous system "as a
whole tend to convert excitations into inhibitions is also
clearly in harmony with the given account of the synaptic
function.
It is improbable, however, that the mechanisms of sum-
mation and of inhibition proposed by Adrian and Lucas on
the basis of their study of conduction through regions of
decrement are the whole story. Nevertheless most of the
differences between reflex arc and nerve trunk conduction
can be accounted for if we suppose the synapse to have a
mechanism not differing qualitatively from that of the plain
nerve fiber membrane, although differing quantitatively from
the latter to a very considerable degree. There are un-
doubtedly certain physical constants which determine the
processes of the nerve membrane. Among these are the
threshold depolarization required to stimulate, the magni-
tude of the excitation or maximal depolarization, the rate of
recovery of the membrane, etc. At the synapse these con-
stants appear to suffer a radical change in magnitude, of
such a character that all of the processes are retarded; the
latent period is much longer, as is also the refractory phase,
and if we suppose the synapse to have a phase of hyper-
excitability this also is probably much prolonged. By a
proper choice of the values of the various membrane con-
stants the majority of the characteristic features of synaptic
functions can be explained.
The synaptic process is evidently a membrane process,
which means that it is localized in a region of space having
the general form of a thin sheet. It should be noted, how-
ever, that this is equally true of the general process of con-
duction in a nerve. The synaptic function is probably more
complicated than that of plain conduction, since it involves
the combined properties of two membranes in juxtaposition.
Some of the peculiar characteristics of conduction through
synapses which may not be explicable by the postulation of
mere quantitative differences between the synapse and the
THE PHYSICAL BASIS OF NERFE FUNCTIONS 345
nerve trunk may perhaps be accounted for on the basis of
quantitative differences between the two membrane elements
in the synapse. For example, the fact that a synapse nor-
mally conducts in only one direction can be attributed to
a difference between the intensities and thresholds of exci-
tation for the two adjoining neurones, such that a more
afferent neurone is able to stimulate a more efferent one but
not vice versa. It is possible also that specially ionized
substances may exist in certain synapses which introduce
properties characteristic of these synapses.
VII. THE MECHANISM OF THE RECEPTOR
Our account of the physical nature of nerve stimulation,
conduction, and synaptic transfer needs to be supplemented
by considerations bearing on the receptor function. Re-
ceptors, in general, are recognized to be especially differ-
entiated cells, often more epithelial than neural in character,
which lower the threshold of a given nerve path for certain
forms of stimuli while raising it for others. It is probable,
moreover, that in addition to being stimulus selectors, re-
ceptors are essential to the continued stimulation of nervous
tissue by any fundamental force. DuBois-Reymond's prin-
ciple tends to make the effect due to the direct action of
any force upon the nerve a merely momentary one. In
spite of the phenomena of sensory adaptation, stimulation
through receptors produces relatively continuous excitation
of the afferent nerve path. Adaptation itself appears to be
attributable mainly to fatigue of the receptor process rather
than of the nerve path.
In a previous article1 I have suggested a plausible physical
account of a manner in which the visual receptors may be
conceived to produce a continued stimulation of the optic
nerve fibers. This account is based upon the empirical
finding that the retina is electrically negative, rather than
positive as would be expected, with respect to the cut end
of the optic nerve. This negativity of the retina, which
Ireland, L. T., 'The Nature of the Visual Receptor Process/ /. of the Opt. Soc.
of Amer., 1917, i, 9-13.
346 LEONARD THOMPSON TROLAND
is increased by the action of light, seems to imply that the
essential ionized substance of the rods and cones has a negative
ion which diffuses through the enclosing membranes of the
receptor cells more readily than does the positive ion, pro-
ducing a state of polarization opposite in direction from that
of ordinary nerve fibers. It is clear that a receptor cell thus
polarized and in contact with a normal nerve fiber, would
tend to depolarize the latter at the place of junction, thus
setting the fiber into excitation. The excitation would
probably be mutual, resulting in a reduction of the polariza-
tion both of the nerve membrane and of the receptor mem-
brane. We have reasons for supposing that the inertia of
the receptor process is much greater than that of the nerve
function, so that in all probability the nerve would recover
from refractory phase considerably in advance of the re-
cuperation of the receptor cell polarization. However, when
this latter recuperation has reached a certain point, a second
stimulation of the nerve fiber would result and this process
would be indefinitely repeated. The result would be a
stream of quantal impulses sent along the optic nerve fibers,
and having a frequency determined by the rate of recupera-
tion of the receptor cell.
In the case of visual response it is easy to see how this fre-
quency can be caused to vary as a result of variations in the
stimulus. The direct effect of light upon the sensitive sub-
stance of the rods or cones is probably one of increased ioniza-
tion and we should expect the rate of r<?polarization of the
receptor cell membrane to increase with increase in concen-
tration of the ions within the cell. The optic nerve impulse
frequency would thus tend to be augmented by the action of
radiation on the retina and to a degree greater the greater the
intensity of radiation of any given wave-length.
It is certain that the processes by which sensory stimuli
excite receptors are as varied as there are different forms of
adequate stimuli. However, it is conceivable that the mode
of interaction of the receptor cell and the conducting nerve
fiber is always of the same general sort. It may be a general
characteristic of receptor cells to have a negative polarization,
THE PHYSICAL BASIS OF NERVE FUNCTIONS 347
so that they tend constantly to depolarize and to excite the
nerve fiber. A stimulus acting upon the receptor cell would
in this case so operate as either to increase or to decrease its
negative polarization. It is easy to imagine physical or
chemical mechanisms through which such changes could be
brought about by the action of almost any conceivable
agency.
Rhythmic depolarization mechanisms of a sort similar to
that above discussed may possibly be found in the central
nervous system as well as in the periphery. The rhythm
of the breathing center and other centers such as those
determining certain types of peristaltic action, muscle tonus,
etc., may be controlled by such mechanisms.
VIII. PSYCHOPHYSIOLOGICAL APPLICATIONS
The ultimate physical analysis of the nerve function must
provide us, according to my view, with the fundamental
materials on the physiological side for the construction of an
ultimate psychophysical theory. The traditional doctrine
of the interdependency of 'mind and body' teaches us that
consciousness depends upon the existence and nature of
central nerve processes. If we reject vitalistic fancies we
must be forced eventually to describe these central processes
in physical terms and thus to conceive them as certain con-
figurations and changes in configuration of electrical particles.
Some modern thinkers, often officially catalogued as
psychologists, subscribe to the view that there is no such
thing as consciousness, and they of course cannot be expected
to take much interest in the psychophysical problem. These
same thinkers often appear to believe in the existence of a
peculiar characteristic of physiological activity called 'the
operation of the organism as a whole.' If, however, we
define consciousness as simply any given experience we dispose
of any difficulty about the non-existence of this entity, and
the more we analyze the operations of 'organisms as a whole'
the clearer it becomes that these operations are simply con-
catenations of many part processes. The ultimate account
is one which expresses any process, no matter how compli-
348 LEONARD THOMPSON TROLAND
cated, in terms of the operations of irreducible physical
elements.
The central nerve process, which in the traditional
theory acts as the determinant of consciousness, consists
essentially, according to the modern idea, of synaptic func-
tions. Such functions, however, apparently differ only
quantitatively from those of simple nervous conduction, and
if consciousness is associated with synaptic processes it
probably is also correlated to some degree with the simpler
processes of conduction. Both of these related functions,
as we have seen, are membrane processes, and it would
therefore appear that the physical correlate of consciousness
is localized in some definite configuration of sheet-like regions
of space. This 'region of determination of consciousness'
is almost without doubt located in the association areas of
the cerebral cortex in the case of the human introspective
field, which is the main object of study for pure psychology.
The introspective analysis of consciousness provides us
with certain psychical elements, attributes, and modes of
relationship, for each of which it is desirable to find definite
physiological correlates. A careful study of the implications
of ordinary laboratory psychophysics — which determines the
relationships between stimuli and their conscious reactions —
with the theory of nerve action should ultimately enable us
to work out some of these direct psychophysical correlations.
Our theory of the receptor process in the case of vision sug-
gests that nerve impulse frequency, or at any rate the mag-
nitude of the nerve current, is the determinant of what we
call sensory intensity. The characteristic qualities of differ-
ent 'sensations' are correlated by the traditional doctrine of
specific energies with the identity of the nerve path which is
excited. Mere abstract identity, however, will scarcely
suffice. We must suppose that these qualities depend upon
peculiarities, either structural or functional, in the cerebral
synapses which are set into operation by different afferent
nerves. In another paper1 I have suggested that many of
1Troland, L. T., 'A System for Explaining Affective Phenomena,' /. ofAbnorm.
Psychol., 1920, 14, 376-387.
THE PHYSICAL BASIS OF NERFE FUNCTIONS 349
the facts about affection, or pleasantness and unpleasantness,
can be explained on the assumption that its neural correlate
lies in the rate of change of synaptic conductance in that par-
ticular portion of the cerebral gray matter which is re-
sponsible for the introspective consciousness at any given
moment. Clearness and other fundamental characteristics
of factors in consciousness will eventually find their proper
correlates in the cerebral nerve process.
It is certain that the physical correlate of the simplest
introspectively discriminable factor of consciousness must,
from the physical point of view, be enormously complicated.
A point visual sensation, for example, can scarcely be attri-
buted to the function of any cerebral component smaller
than a single synapse, and it is certain that a single synaptic
mechanism involves the simultaneous cooperation of millions
of physical atoms, electrons, electrical and magnetic fields,
etc. The coexistence in a single moment of consciousness
of a multitude of sensory elements arranged in a definite
pattern must depend upon the concurrent and unified func-
tioning of a large number of cerebral synapses. Unless we
are to employ assumptions which suggest a non-physical or
spiritual factor in the determination of consciousness, we
must suppose that the unity of consciousness depends upon
some sort of unity in the total nerve process upon which
consciousness depends.
I am attempting to work out plausible solutions for some
of these fundamental psychophysical problems in connection
with a general metaphysical theory which I have called
'Paraphysical Monism.'1 This doctrine provides us with
an explanation of the facts of psychophysical parallelism
which eliminates the fundamental dualism of the Leibnitzian
preestablished harmony, permitting us to combine the facts
of physics and psychology into a unified system. Psy-
chology needs all of the stimulus which it can derive from the
advances of physical and physiological science. Indeed,
this need is so dire as almost to warrant the suicidal pro-
mulgations of those 'psychologists' who call themselves
, L. T., 'Paraphysical Monism,' Philos. Rev., 1918, 27, 39-62.
35° LEONARD THOMPSON T ROLAND
'behaviorists.' The psychology of the soul is dead, and
that of consciousness is suffering murderous attacks. What-
ever we may think of the former, the latter is assuredly worth
saving, and it is my impression that the data provided by
modern nerve physiology will provide us with means for
resuscitating the true science of physiological psychology as
it was conceived by Fechner, Miiller, Helmholtz, and other
pioneers.
THEORIES OF THE WILL AND KIN^ESTHETIC
SENSATIONS
BY RAYMOND H. WHEELER
University of Oregon
Theories of the will in psychology owe their origin to the
development of ethical views among the early Greeks. These
theories, together with those held by the early Church Fathers
and the Scholastics, may be classified as intellectual and abso-
lute. Both of these groups made will a faculty of the soul.
The intellectual theories derived their name from the impor-
tance which they ascribe to the faculty of reason as a con-
stituent of willing; the absolute theories made all other faculties
of the soul subordinate to the will. In the Aristotelian view
the will consisted of a l desire' to which a 'goal' or an 'end'
was supplied by the reason. Desire was analyzed as a per-
sistent state of unrest or striving and was the essential active
or dynamic feature of mental life. During the middle ages
theories of the will were concerned largely with the problem of
determinism versus freedom, hence the relative importance
of the two faculties, the intellect and the will, as agents in
controlling human behavior, was a subject of paramount
importance.
As a consequence of empirical and inductive methods, we
find in England from the time of Hobbes to Hume the develop-
ment of a new type of theory, the emotional theory: Hume
regarded the feelings as essential constituents of every voli-
tional process. Locke had previously held that disquietude
or uneasiness constituted the origin and the dynamic feature
of volition, a view which in its essence was but a rehabilitation
of Aristotle's potential desire. On the continent from Des-
cartes to Herbart the tendency was gradually becoming defini-
tized of conceiving the will as the active side of mental life
in general, or as a general striving tendency.
Up to the time of more modern psychology, then, there
35'
35 2 RAYMOND H. WHEELER
were in general three views of the will. In a broad sense of the
term the will included all mental activity as such, an activity
characterized as a potential desire or as a state of unrest.
This tendency culminated in a definitized theory of conation.
Secondly there was a voluntaristic tendency to employ the
term will itself to mean this active feature of mental life or
the source and cause of action. Thirdly there was a tendency
to bring the will into intimate relations with the feelings where
the feelings were regarded as essential constituents of volitional
processes.
Modern theories may be classified according to the feature
most characteristic of each, namely its emphasis upon the
reductivity or the non-reductivity of will to a complex group-
ing or mode of functioning of simpler constituents. Those
theories which reduce will to a peculiar order of sequence of
sensations, images and affections and which do not emphasize
the elementary nature of any one constituent we may call the
totally reductive. On the other hand those theories which admit
the existence of an elementary mental content — an elementary
and unique volitional process — may be called non-reductive.
In this latter group the will has an elementary content which
is not necessarily independent, functionally, from other con-
tents. Intermediate between these two groups of theories is
another which we may call the partially reductive. Here the
will is reduced partly to the functioning of the traditional
elements of consciousness and partly to a unique elemental
content. The latter might be said to constitute a structural
and functional criterion or an essential conscious concomitant
of the volitional consciousness. This group may be subdivided
thus: (i) those theories which posit a conative element; (2)
those theories- which posit an intrinsically active ego; (3)
those which regard the feelings as the essential constituents
of will; (4) those which ascribe to mental processes a force or
general innervation mechanism. A final group of theories
may be called motor or behavioristic owing to the fact that
emphasis in them is laid upon the principle of stimulus and
response. Here the will, reduced to its simplest terms, becomes
a system of coordinated reflexes or motor responses and the
innervation mechanisms are not mental but physiological.
THEORIES OF THE WILL AND KINASSTHETIC SENSATIONS 353
Non-reductive Theories. — Lotze1 conceived the will to be
an unanalysable psychic process which functioned chiefly in
choice and resolve. Since it was not constantly functioning
as a datum of consciousness it was relegated to the domain of
the subconscious where it found a resting place when not
actually in operation. In the more recent literature Ach2 and
Michotte8 might be mentioned among those who, in conse-
quence of their failure to analyze the volitional process experi-
mentally and in consequence of finding a 'feeling of mental
activity' in 'genuine' volitional acts, make a certain part of
the volitional consciousness, at least, elemental and non-
sensory. James'4 most often quoted 'fiat' consciousness was
a subjective experience sui-generis which could be designated
but not defined.
Partially Reductive Theories. — Herbart's5 theory of the will
is an early example of the conative theory. Here the will in
its elemental form was to be found in the striving of ideas for
existence in consciousness or for the possession of the conscious
level. This striving process began in the realm of the un-
conscious. Volition consisted first of desire, a product of re-
sistance between striving ideas and secondly in the opposition
between groups of associated ideas (goals of purposes). Lipps6
assumed a striving process not only in ideas but in feelings
and sensations as well. Both affective processes and feelings
of activity were essential constituents of willing and of the
volitional consciousness. The conationists, among whom may
be mentioned Stout,7 especially, ascribe to each conscious
state an inherent tendency to pass beyond itself into another
conscious state, a striving process which is directly labelled
conation and which, in itself, is present to consciousness.
Writing from the point of view of a self psychology Calkins
described the content of willing as an active 'consciousness of
'Medicinische Psychologic,' Leipzig, 1852.
'Ueber den Willensakt und das Temperament,' Leipzig, 1910.
'Etude experimental sur le choix volontaire et ses antecedents immediats,'
Arch de psychol., 10, 1910, 113-321.
'Principles of Psychology,' 1890, Vol. II.
'Lehrbuch der Psychologic,' Leipzig, 1850.
'Von Fiihlen, Wollen und Denken,' Leipzig, 1902.
'A Manual of Psychology,' London, 1913.
354 RAYMOND H. WHEELER
my active connection with other selves or with other things '
('First Book in Psychology,' 3d rev. ed., p. 226). The voli-
tional consciousness is said to involve the essential non-sensory
factor of the 'self-as-willing.' Ach and Michotte found an
immediate and unanalyzable consciousness of the self or an
intrinsically active ego in all genuine volitional acts. Meu-
mann1 assumed that the process of 'accepting' a goal idea, in
a volitional act, involved an immediate consciousness of the
self. Both partially reductive and totally reductive tendencies
can be found in the analyses of volition which have been made
by these latter writers.
The emotional theory of the will may be illustrated by
reference to Bain2 and Wundt.3 Bain defined the will as all
mental and physical activities insofar as they were guided or
impelled by the feelings of pleasure or pain. Wundt's theory
(which is voluntaristic as well as emotional) makes will the
original energy of consciousness, the first and primary form
of which is a simple impulse motivated by pleasure or pain.
When an organism experiences a simple sensation there arises
a feeling process which develops to a maximal state of intensity
beyond which it overflows into movement. Such a sensation-
feeling-action series of experiences and events is called a simple
or primary act of will. Feeling processes, therefore, possess
an innate capacity toward willing. This is either a capacity
to arouse physical movements or to initiate into consciousness
other mental processes. In more complicated forms of will
the feelings or emotions may be observed to increase gradually
in their intensity, beginning with pleasantness or unpleasant-
ness, eventually developing into strain or excitement. These
feeling states fuse into a 'total feeling of activity' which is an
essential conscious concomitant of complex or secondary acts
of will. In this category are found such acts as choosing and
performing difficult tasks. Secondary volitional acts may be
motivated by ideas which are associated with the feelings.
As an example of a 'force' theory may be cited Fouillee4
1 'Intelligenz und Wille,' Leipzig, 1913.
2 'The Emotions and the Will,' London, 1899.
3 'Grundziige der Physiologische Psychologic,' Leipzig, 1903.
4 'L'evolutionisme des idees-forces,' Paris, 1893.
THEORIES OF THE WILL AND KINJESTHETIC SENSATIONS 355
for whom sensations and feelings are at the same time conscious
states and mental forces according to the viewpoint taken in
regarding such mental processes as contents or as acts.
Totally Reductive Theories. — Miinsterberg1 held that the will
as a datum of consciousness was a goal-idea which had come
to be associated with other ideas or with muscular movements.
This goal-idea involved the anticipation of an end. In other
words it is a mental process of sensory origin having to do
with preparatory motor adjustments. Ebbinghaus2 in like
manner held that volition consisted in the capacity to foresee
the end of action by associating an image with an act in such
fashion that the image would function as the stimulus for the
subsequent act. In the views of Ach3 and Meumann4 we
might have found totally reductive theories had it not been
for the fact that both ultimately drag in a non-sensory experi-
ence pertaining to the self and feelings of activity. Meumann
described the will as a selective process brought about by
means of * accepted' goal ideas, while in a similar fashion Ach
found the clue to a volitional act in the acceptance of an
Aufgabe. The capacity of goal-ideas to so control subsequent
mental processes depended upon the previous forming of asso-
ciations between the foresight of an end and the act which
attained that end.
Behavioristic Theories. — The modern trend of descriptions
of the will is obviously behavioristic, where the emphasis is
laid upon the coordinated responses of the organism to its
environment and not upon the mental contents as such. But
owing to the lack of experimental evidence accurate accounts
of the volitional process cannot be presented. According to
Ribot6 the will is to be regarded as the sum total of the organ-
ism's responses to environment. All mental processes tend to
express themselves in some form of overt action and would
succeed in doing so were it not for processes of inhibition.
The continuity of mental states can be expressed only in terms
of the continuity of these organized motor responses.
1 'Die Willenshandlung,' Freiburg, 1913.
8 'Grundzuge der Psychologic,' Leipzig, 1911.
1 Loc. cit.
4 Loc . cit.
4 'The Diseases of the Will' (trans.), Chicago, 1903.
35 6 RAYMOND H. WHEELER
Notwithstanding the numerous descriptions and interpreta-
tions of the will which have appeared in the literature we
find little genuine progress in ascertaining the exact char-
acteristics of the volitional consciousness. It is evident, how-
ever, that it cannot be described adequately in terms of struc-
tural contents alone. Functionally, two distinct problems are
involved, namely those of the will in a broader and in a narrower
sense. In the broader sense the problem of the will should be
identified with the problem of the general sequence of mental
processes. Such discussions seek to answer the question, how
may the sequence of mental processes be envisaged in mental
terms? Or stating the problem in objective terms: how may
the continuity of the organism's responses to environment be
best described? In the narrower sense the will should be
identified with the problem of a particular portion or order of
sequence where the question is asked how may one group of
mental processes exert an apparent directing influence over
subsequent mental processes ? Stated in objective terms: how
may one response lead inevitably to the making of a subsequent
response ?
Theories which have been formulated from a subjective
point of view have reduced the will, in the broader sense, to a
potential desire (Aristotle), to a state of uneasiness (Locke),
to a striving process (Herbart, Lipps), to, a conative tendency
(Stout, Baldwin and others) to an intrinsically active ego
(Calkins, etc.), to innervating properties of the feelings or
emotions (Bain, Wundt) and to an alleged dynamic force
(Fouillee). Other writers have concerned themselves with the
general problem of sequence but from an objective point of
view; hence the will has become the sum total of the organism's
motor responses (Ribot and others).
Still others who have faced the same problem and who
have borne in mind both its structural and functional aspects
have failed to find any structural clue to the general problem
of sequence, hence for them the problem becomes one of a
unique or particular order of sequence. Here, again, failing to
find elementary structural clues they have formulated theories
of a totally reductive character. In such theories the im-
THEORIES OF THE WILL AND KIN ESTHETIC SENSATIONS 357
portant role is assigned to a goal-idea or Aufgabe which condi-
tions the sequence of mental processes by means of productive
or selective influences. On the other hand there have been
many attempts to state clearly the problem both in its broader
and narrower aspects. For example, the will in its narrower
sense (volition proper) was envisaged by Aristotle as a rational
desire; by Hume, Bain and Wundt as a highly organized
sequence of potential emotional processes associated with ideas;
by the conationists as a highly organized conative system the
distinctive feature of which was the foresight of an end. An-
other view involves an assertive attitude of the self (Calkins).
The problems have been the same throughout all these dis-
cussions; but how different have been the solutions!
The confusion found in these descriptions can be traced
obviously to widely different points of view. Advocates of
partially reductive theories have been obliged to appeal either
to a volitional constituent in the feelings, to an alleged conative
element, to an intrinsically active ego or to forces inherent in
sensation and affection in order to account for the general
conscious continuum in purely subjective factors while adhering
at the same time to an atomistic conception of mind. It is
the problem of getting elementary states of consciousness back
into a working, active system again after they have been as-
signed the role of units or * atoms.' It is the problem of making
a river out of a succession of barrels in the stream bed. Witness
the attempts of those writers who adhere to an atomistic and
structural conception of mind but who have failed to discover
any elemental conscious experience whose constant presence
in mental life will explain the continuity of conscious states.
Such writers (James,1 Brentano,2 Witasek3) have resorted to
other factors such as to a differentiation between transitive
and substantive states or between act and content. From a
more objective point of view the recent motor movement in
psychology seems to be based essentially upon an attempt to
solve the general problem of sequence and unbroken continuum.
1 Loc. cit.
2 'Psychologic vom empirischen Standpunkte,' 1874.
1 'Grundlinien der Psychologic,' 1908.
358 RAYMOND H. WHEELER
As for the narrower problem — the determination of subse-
quent processes by an antecedent process — one finds the same
divergence in points of view and in results. The issue has
been definitely sharpened by positing the existence of a deter-
mining tendency or directing of the course of the stream of
consciousness. But this determining tendency may be either
a driving, a vis a tergo exerted by the Aufgabe (Ach) or a leading
— an attracting — by goal ideas (Meumann). Auxiliary prin-
ciples such as associative tendencies, perseveration tendencies
and constellations have been appealed to in an endeavor to
formulate the problem and its solution more clearly. But
the traditional principles of association have borne the burden
throughout all these more recent discussions. The objective
or behavioristic attack upon this same narrower problem in-
cludes the reflex-arc concept (Dewey)1 and the principle of
ideo-motor action.
The solution of these problems has been delayed, also,
because of the very meager introspective analyses which have
so far been made under experimentally controlled conditions.
It is extremely doubtful whether a genuinely vigorous volitional
act has ever, in the past, been subjected to adequate intro-
spection.
It seems to the writer that insofar as the partially reductive
theories have been based upon introspective evidence, the
trouble has been in a failure to reduce conation, striving process,
feelings of activity, etc., to a common process. We have
every reason to believe that many of Wundt's 'feelings,' for
example, are forms of kinsesthetic sensations. Is it not quite
probable that what the conationists have called the immediate
conative experience is kinaesthetic sensation? The writer is
firmly convinced that the 'feeling of mental activity' described
by Ach and Michotte is a complex of kinsesthetic sensations.
Moreover is it not also possible that Meumann's consciousness
of the self in the acceptance of a task or that Professor Calkins's
intrinsically active ego are interpretations unwittingly based
upon an immediately experienced but complex and diffuse
kinsesthetic background and nothing else? Within the last
1 'The Reflex Arc Concept in Psychology,' PSYCHOL. REV., 1896, 3, p. 357.
THEORIES OF THE WILL AND KINAESTHETIC SENSATIONS 359
few years there have appeared several elaborate introspective
descriptions1 of various conscious processes but in none of
these do we find the slightest hint of such elements. Kinaesthe-
tic sensations are with us always in mental life. Recent
introspective works and more especially those of Fernberger
and the writer show the importance of such processes. In
fact, as the writer has shown in an introspective study of
choosing,2 kinaesthetic processes are essential and as a type
the only essential form of elemental conscious process in the
act of choosing.
It does not seem unwarranted, therefore, to conclude that
the extreme variations in past descriptions of the will conscious-
ness both in its broader and narrower aspects have been due
to various interpretations of a consciousness which is so largely
made up of kinaesthetic sensations. From these experiences
we get our notions of striving, strain, activity, force, conation
and the like. It may be open to question, also, whether the
Freudian wish and its various cousins are not veiled and
unconscious interpretations unwittingly based upon a con-
sciousness of kinaesthetic strain.
To sum up, theories of the will in the history of psychology
represent successive attempts to describe the conscious con-
tinuum as a whole and to describe the process of control in
any given portion of the conscious continuum. The chief
cause for the great variability of these descriptions lies in a
further attempt to find evidence of this continuity in some
unique mental process. Where such an elemental process has
been found lacking we have been obliged to resort to the prin-
ciple of act and content. Various points of view have added
to the confusion. And in modern psychology, inadequate
introspective data has led to inadequate interpretation. The
unique mental process, we believe, is nothing more than
kinaesthetic sensation.
1 E. L. Woods, 'An Experimental Analysis of the Process of Recognizing,' Amer.
J. of Psychol., 1915, 26, 313-387. S. C. Fisher, 'The Process of Generalizing; and its
Product, the General Concept,' Psychol. Mon., 1916, ax, No. 2 (Whole No. 90). S. W .
Fernberger, 'An Introspective Analysis of the Process of Comparing,' Psychol. Mon.,
1919, 26, No. 6 (Whole No. 117).
2 R. H. Wheeler, 'An Experimental Investigation of the Process of Choosing,'
University of Oregon Publications, 1920, Vol. I, No. 2.
360 RAYMOND H. WHEELER
This compels us to settle upon a point of view. It shows
us the futility of searching for introspective evidence of con-
tinuity and places us in the position of the other sciences where
we rightly should be — the position of finding continuity in
data treated from an objective point of view. We should look
for the solution of the problem, therefore, in behavioristic
principles. The cry at present is, in some circles, to do away
with introspection. But in the other sciences observations
are made via the senses. Why can we not observe our own
behavior in the same way? Introspection as sensory evidence
of our own behavior ought to be as valid as sensory observation
of any movement external to our bodies. Let the 'feel' of a
response be as ample evidence of the existence of that response
as the 'sight' of it in another person or an animal. To be sure
introspective evidence should be verified wherever possible
both by objective instruments and by similar reports from large
numbers of observers. Our view is that in attempting to
solve psychological problems in the future, — problems which
were formally considered purely subjective — we are warranted
in drawing behavioristic interpretations, in part at least, from
introspective data.
A PURSUIT PENDULUM1
BY WALTER R. MILES
Nutrition Laboratory of the Carnegie Institution of Washington. Boston, Mass.
One of the measurements used at the Nutrition Laboratory
on the aviation candidates in the spring of 1917 was to record
the adequacy of ocular-pursuit movements in following the
swing of a pendulum. The subject was seated at a head-rest
with the left eye covered. A polished metal bead suspended
by an invisible cord was arranged to swing through a visual
angle of about 40°. The pendulum made a double swing in
2 seconds. Its release was synchronous with exposure of the
eye to the recording beam of light, after the manner of Dodge's
photographic technique.2 The repeated instruction was to
watch the bead intently every moment of its swing. Six or
more successive trials by a subject were photographed side
by side on one plate. These records do not easily provide
an exact quantitative score for accuracy of pursuit. How-
ever, it is convenient to rank these photographic records
showing the reaction time occurring at the start of the
pendulum's swing, together with the number and size of
abrupt horizontal movements by which the subject supple-
ments his inadequate pursuit, into five grades or groups of
excellence. Such grouping gave a positive correlation of 0.40
with the subsequent progress of these men in learning to fly.3
1 In abbreviated form this paper was read before the American Psychological
Association, Cambridge, December 30, 1919.
1 Diefendorf and Dodge, Brain, 1908, 31, pp. 451-489. See Plate II for illustra-
tive records showing fully the characteristics of this type of eye-movements. For a
description of the eye-movement recording apparatus as used on the aviation candi-
dates, see Benedict, Miles, Roth, and Smith, 'Human Vitality and Efficiency under
Prolonged Restricted Diet,' Carnegie Ihst. Wash. Pub. No. 280, 1919, pp. 159 ff. and
pp. 184 ff.
'Our subjects, the first groups of candidates to attend the Aviation Ground
School of the Massachusetts Institute of Technology, were a very superior lot of men.
Nearly all graduates of our best universities, these men had been prominent in athletics
and many of them on their own initiative and at their own expense had taken some
361
WALTER R. MILES
At the time this result was found hardly any single test
indicated a higher correlation with flying. Officials advised,
however, that the ocular-pursuit measurement, as carried out
photographically, was too complex for any general use in the
preliminary selection of candidates for pilot training.
These details have been recited as they account for the
simplicity of the device described below. The pursuit pendu-
lum was an effort to meet a definite situation. Care was
exercised to exclude all electrical and photographic or other
graphic features, to make the apparatus its own gravity-
operated chronometer and such that it could be used nearly
anywhere and would give an immediate quantitative score
for the accuracy of the eye-hand coordination in pursuit
movement. Although an opportunity never came after the
development of the test to try it on a group of aviators or men
who were in this training, the possible general usefulness of
the measurement to other laboratory workers and in industry
may warrant the description of the pursuit pendulum, together
with illustrative data for initial performance, improvement
with practice, and changes in efficiency, e.g., as produced by
a superimposed nutritional factor such as alcohol.
From a suitable wall bracket a pendulum carrying a
reservoir is arranged to swing over a sink or table, a small
stream of water flowing from the lower extremity as the
pendulum swings. The individual under test, seated before
the sink, attempts to catch the water in a cup of limited
diameter. A separate cup is used for each double swing
and the volume of liquid collected represents quantitatively
the adequacy of pursuit.
The bracket, A (see Fig. i), extends from the wall about
45 cm. and is fairly rigid. The pendulum, B, 140 cm. long,
training in aviation. They were keenly interested in aviation problems and cooperated
whole-heartedly, as did also the officials at the ground school. Although about 65
men were measured, the government found it imperative to send many of these to
Europe immediately upon their having finished at the ground school, and they therefore
received the flying training abroad. Most energetic efforts were made by Professor
E. L. Thorndike to secure the flying scores for these men on whom Drs. H. E. Burtt,
L. T. Troland, and myself had worked. Scores for 26 were finally obtained and it is for
these that the correlation mentioned was found.
A PURSUIT PENDULUM
363
FIG. I. Diagram of the pursuit pendulum. A, wall bracket; B, pendulum
carrying reservoir, E. Gauge, F, determines volume of liquid in reservoir. C,
adjustable weight regulating pendulum's period. C, wooden frame clamped to sink,
D. Nozzle, H, of pendulum held by catch, AT, against rubber tubing, /, until released
by fall of hammer, L, hinged at M. Short section of chain, N, determines lift of
hammer. 0, cup of limited diameter, in which the expelled liquid is to be collected
by the subject under test. P, position at which cup, 0, is held at start. Q, short-
stop for ending the catch.
364 WALTER R. MILES
is suspended by two screw eyes. Its shaft is continuous
through the reservoir and is very stiff, being made of sections
of galvanized-iron pipe (regular |-inch inside diameter). The
reservoir, E, a i-gallon galvanized-iron oil can, surrounds
the shaft and is firmly secured and made leak-tight by the
use of a "railing flange" soldered to its bottom. As the
reservoir is located about midway the length of the pendulum,
the head of water changes but little with the decreasing level
in the can. An adjustable weight, G, of about 4 kilograms
allows for regulation of the pendulum's period and makes
the position of the center of mass much less dependent upon
the exact amount of liquid in the reservoir. Openings are
arranged in the pipe shaft on a level with the floor of the can
and air vents are placed above. The water flows very freely
from the reservoir, and at the lower end of the shaft is reduced
to a stream 3 mm. in diameter by the nozzle, H.
A simple arrangement for retaining and releasing the
pendulum is shown in Fig. I and separately illustrated by a
top view in Fig. 2. The wooden frame, C, is clamped to one
end of the sink, D, at such a height that when the orifice, H,
is slipped up on a cushion, made of a short horizontal section
of rubber tubing, /, a closure is made which is practically
leak-tight. A catch at K retains the pendulum in this position
until the fall of the rubber-headed hammer, Z,, hinged at M.
The hammer is lifted by the operator and held in a nearly
upright position, determined by a short section of chain
(see N, Fig. i). It is released on verbal signal from the
subject and requires 0.3 second to fall and start the pendulum.
This method of release corresponds to common industrial
operation and the reactor, especially a subject without psycho-
logical training, likes it better than having the start occur at
some arbitrary and more or less unexpected time beyond his
control.
The cups in which the subject is to catch the expelled
liquid (see 0, Fig. i) are made of thin-walled brass tubing,
19 mm. (regular f-inch tube) inside diameter and nearly 22
cm. long. At the start a cup is held at position P, against the
wooden frame, nearly vertical from and about 2 cm. lower
A PURSUIT PENDULUM
365
than the orifice, H. No water leaks into the cup and it is
possible to begin the hand
movement almost in register
with the pendulum. While
the subject follows the pen-
dulum to the right, the oper-
ator turns the short-stop, Q,
from position I into position
2. (See Fig. 2.) This metal
screen stops the return move-
ment of the cup at a distance
of 2 cm. in front of position
P and thus gives the operator
a little space within which to
catch the pendulum and re-
place it on / without spill-
ing liquid into the cup, as
otherwise an error would be
introduced in the result.1
With two liters of water in
. i . . i . , r .1 F'G- 2. Top view of the arrangement
the reservoir the period of the for retaining and releasing the j^™
pendulum for a double swing H, pendulum nozzle pressed by the rubber
is just 2 Seconds. Naturally tube» /» to form leak-tight closure; K,
this time will increase some- hook by which pendulum is held in posi-
tion until A is depressed by the hammer,
what as water is lost and the
center of gravity lowered.
Practically, the change is of no
consequence to the test, as we
find that with two liters of wa-
Scalc
10
L, which is hinged at M; short-stop, Q, at
an appropriate time is turned into posi-
tion 2 in order to cut the return pursuit
slightly short and allow space in which the
experimenter may catch the pendulum.
ter 20 v. d. require 40 seconds, while with one liter 40.3 seconds
are required, and when the reservoir and shaft are empty
40.9 seconds are required. Thus, for testing purposes the
period of swing may be regarded as independent of the
1 Another source of error must be guarded against with the cooperation of the
subject. He should be cautioned not to slip the cup up over the nozzle of the pendulum
or, indeed, to bump the cup against the nozzle and thus interfere with its motion.
The stream of liquid does not spray out and there is no advantage, from this standpoint,
in having the mouth of the cup very near the pendulum. This requirement not to
touch the pendulum with the cup is a part of the coordination, but unfortunately
does not show in the objective results.
366 WALTER R. MILES
amount of liquid in the tank. The amount of head of the
liquid and the size of the orifice were arranged with the idea
that 50 c.c. should be the possible catch per double swing.
With 2 liters in the reservoir at the start, it is found that at the
first double swing the subject can catch, possibly, 50.4 c.c.
and at the tenth succeeding catch 49.7 c.c. For eight succes-
sive trials the amount delivered is thus within I per cent,
of 50 c.c. It has seemed satisfactory to replenish the water
every five or ten trials. If the subject is catching nearly all
the liquid expelled, the opening in the cup should be reduced.
A gauge, F (Fig. i), on the side of the tank makes it a simple
matter, when introducing water at the opening in the top
of the reservoir, to determine that the volume of water shall
be up to 2 liters. During the swing the orifice of the pendu-
lum, as used in the collection of the data presented below,
moved a horizontal distance to the right of 70 cm. This is a
fairly large excursion, but most adults can follow the move-
ment without swaying of the body, if they so desire.
More complex arrangements of such pursuit apparatus
naturally suggested themselves, for example, the pendulum
might be made the long arm of a siphon. An orifice, not a
part of a pendulum, might be carried on a belt and given a
complicated series of movements, prolonging the pursuit and
requiring coordination for forward and backward as well as
for lateral displacements.1 After the experience with the
ocular-pursuit measurement it was assumed, however, that
in trying to contribute to the problem of selecting aviation
pilot material the simpler the test apparatus the more service-
able it might possibly become. Therefore this model was
made independent of electrical features, did not require
running water GT a sink, could be filled by hand from a pitcher
and could be arranged over a table or inclined trough, as
the operator's conveniences might permit.2
1 In a personal communication Professor Carl E. Seashore informs me that, after
trying the original test at the Nutrition Laboratory, he has arranged a very successful
combination for testing motor ability to perform circular pursuit movements, by using
a phonograph motor, a time-interrupted circuit, and an electric counter.
2 A criticism which may be raised against the quantitative score which the appa-
ratus makes possible is that this score is not a sufficiently graduated result. Prac-
A PURSUIT PENDULUM 367
In practice successive catches can follow each other rather
rapidly, their speed being largely determined by the quickness
of the subject's motions in replacing and taking up the cups
and the promptness of his verbal signals for release of the
pendulum.' Twenty-five trials are easily made in five min-
utes. The 25 cups stand in order, as at R in Fig. 3, being
conveniently held in a box frame, S. Each fifth cup has a
black band near the top serving in the test as a signal to the
operator to replenish the water in the tank. If there is no
time immediately after the test to measure the results, the
cover, V, is placed over the open ends of the cups, the name,
date and hour are noted at £7, and the box is set aside.
In measuring the results it is tedious to empty each cup
separately into a small graduate and so determine the volume
of liquid. Since the cups are all of the same inside diameter
(as nearly so as brass tubing is commercially made) and all
have the same inside depth, a graduated scale, W, cut from
thin aluminum sheet attached to a cork float can be intro-
duced into the mouth of each cup as these are held con-
veniently side by side. in the box frame, S, and the volume of
liquid can thus be very quickly determined for the successive
catches. After the catches have been individually measured,
provided that such analytical data are desired for securing a
measure of variability, the frame, S, is grasped in such a way
that the rubber tube, T, attached to the hinged door on the
front, presses against all 25 cups so that their contents may
be emptied and drained at once into the inclined V-shaped
trough, X, and so into the large graduate, Y. The total
score is in this way very readily secured.
If a good subject reaches such a degree of skill that, for
example, he regularly catches more than 80 or 90 per cent.
tically, it makes no difference whether the pursuit is so accurate that the stream of
water goes in at the center of the opening in the cup or over at one side, just so long
as the whole stream is collected. Since the edge of the cup is sharp, tapered from the
outside, if the stream strikes here it will be divided, part collected and part lost.
With a slower-moving pendulum, a somewhat larger orifice, and a cup with the mouth
the same size as the orifice, probably an arrangement could be made so that the subject
could always catch a part of the liquid and thus small inaccuracies of pursuit would
more properly be represented in the result.
368
WALTER R. MILES
w
0
CENTIMETERS
FIG. 3. Arrangement of the cups in a way convenient for measuring the quantity
of liquid, by simple accessory apparatus. R, bank of cups in carrying frame, S; F,
cover for cups; U, pad for noting name, date, and hour; T, rubber tube attached to
hinged door of frame, S, which is pressed against cups in emptying; W, measuring
float graduated in c.c.; X and Y, inclined trough and large graduate for receiving the
contents of a whole bank of cups emptied at once; Z, collar to reduce the opening
at the mouth of the cups, for especially expert subjects.
A PURSUIT PENDULUM 369
of the possible catch, the task may of course be made more
difficult by decreasing the effective opening in the cups. A
collar, Z, can be slipped into and withdrawn from each cup
by the subject as he uses them in turn and thus the opening
may be reduced to 10 mm. or to whatever size is deemed
desirable to make the task satisfactorily difficult.
It is probable that any investigator who arranges such
an apparatus as is here described will not make it an exact
duplicate. Nevertheless illustrative data are of value in
supplementing the description of the apparatus, as they give
an idea of the type of results that may be expected from its
use. In December, 1918, and January, 1919, considerable
data for this test were obtained on a group of staff members
of the Nutrition Laboratory, including ten women and eight
men.1 The pursuit test was given on 35 days, usually suc-
cessive except for Sundays, and the amount of practice was
20 catches per day. At that time the equipment of cups
consisted of two banks of ten each. It was hardly feasible
that each individual should be tested at exactly the same
time on each day, but care was taken not to measure subjects
when they were fatigued or otherwise indisposed.2
The average results for a group of 18 adults are shown
graphically in Fig. 4. Each plotted point on the curves
represents 360 catches, i.e., 20 catches by each of 18 subjects.
Each of the two groups of ten catches made by a subject on a
single day was dealt with separately when obtaining the
average and standard deviation. This was done to show the
progress made during the day. As might be expected, the
second ten catches almost invariably averaged I or 2 c.c.
higher than the first ten and the variability was usually
smaller. For this brief paper we have averaged the two means
1 The collection of these data, including its tabulation and elaboration, was success-
fully accomplished by an assistant, Mr. E. S. Mills, whose care and cooperation are
gratefully acknowledged.
1 In this early practice experiment it was thought that the subject should execute
the pursuit by an arm movement not supplemented by a body movement. Therefore
two rods were arranged to extend from the sink and to be brought in contact with the
individual on both sides, somewhat above the waist. These rods, while not hindering
the trunk from twisting, obviated the subject's swaying from side to side. Probably
this restriction is unnecessary. It has not been used in later measurements.
37°
WALTER R. MILES
and the two coefficients of variability secured for each of the
1 8 individuals on each day and have employed these 18
quantities to obtain the average represented by eachjplotted
point on the curves shown in Fig. 4.
GROUP PRACTICE ON PURSUIT PENDULUM.
50
48
46
44
42
40
38
36
I
/
S7T
V
i Average cc per catch
Av. for men
Av« for women
Variability in percent.
^AA
Oavs.2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 3436
FIG. 4. Curves showing average results for ten women and eight men tested on
35 days with 20 catches per day practice.
The heavy broken line in the figure indicates the coefficient
of variability (standard deviation divided by mean) in terms
of per cent. The heavy solid line gives the average catch
in cubic centimeters per day. On the first day the individual
averages ranged from 8 to 29 c.c. per catch with a grand
average of 15 c.c., which represents 30 per cent, of the possible
catch. First trials by a number of other adults confirm this
figure as about what may be expected for an initial per-
A PURSUIT PENDULUM 37 »
formance, when the subject seriously tries the test and
consistently makes 20 or 25 catches. The curve showing
the average catch per day rises rapidly on successive days to
20, 23, 25, 26, 28, and, on the tenth day, to 31 c.c. representing
62 per cent, of the possible catch. Thus, on the tenth day
of 20 trials the average efficiency has doubled over what it
was at the start. After 25 days more of such practice the
average increase above this level is only 7 c.c., bringing the
figure to 38 c.c., which is about 75 per cent, of the possible
catch. The practice curve is very regular in form and shows
no definite indication of orthodox plateaux, and it is evident
that the chief part of the rise due to practice can be quickly
worked off by 200 or 300 catches, if it is desirable to bring
the individual up toward the stage of a practice level. The
curve for the coefficient of variability is practically an exact
counterpart in form to that for the average catch. At the
beginning the variation between catches equals about 34
per cent, and at the tenth day, when the average catch has
doubled, the variability has decreased to 19 per cent, or
not far from one half, and by the end of the series has de-
creased to about 14 per cent.1
It is recognized that the group of subjects employed in
this experiment was relatively small and it is hardly justifiable
to draw conclusions regarding such matters as the difference
between men and women in their efficiency in executing such
a pursuit movement. If the individuals are ranked on the
basis of their total average catch per day, it is found that of
the better nine there were six men and three women, while in
the poorer half of the group there were two men and seven
women. There were three women poorer than the poorest
man, but only one man did better than the most efficient
woman. The average difference between the groups of eight
men and ten women is shown in Fig. 4 in the light line curves,
which are above and below the curve for the general average
1 The coefficient of variability for other neuro-muscular tests may be found by
referring to Benedict, Miles, Roth, and Smith, Carnegie Inst. Wash. Pub. No. 280, 1919,
pp. 551 et seq. Examples which may be mentioned are: eye-movement speed, 9 per
cent.; eye reactions, 19 percent.; word reactions, 9 per cent.; and electrical threshold
about 6 per cent.
37* WALTER R. MILES
catch (heavy solid line). The curves for the men and for
the women are fairly smooth, and maintain about a uniform
distance apart, the men on the average catching 4 cc. more
than the women.1
Individual practice curves, such as illustrated in Fig. 5,
are naturally less smooth than the average for the whole
group. In Fig. 5 results for one of the most efficient and
also for one of the least capable subjects have been com-
bined. Subject C made very rapid progress, starting with
21 c.c. and rising to 28, 30, 32.4, 35.8, and 36.7 c.c. on the
next six successive days. (See Curve i.) There is a slight
decline on the seventh and eighth days and quite a definite
decrease on the ninth day. The fluctuations usually range
from i to 3 c.c. Undoubtedly these variations in the average
would have been smaller had the number of trials on each
day been larger. For a fairly long period, that is, from the
tenth to the twenty-ninth day, the average for Subject C is
very close to 37 c.c., which is 74 per cent, of the possible
catch. The performance during this period may conceivably
be classed as a plateau, for there is undoubtedly an indication
of a definite stage of improvement following it, during the
last six days. The coefficient of variability for Subject C
(Curve 2) shows rapid improvement at first, corresponding to
his improvement in the amount of the catch up to the seventh
day. Beyond this time there are fluctuations, some of them
quite large. From the seventh to the twentieth day, inclu-
sive, the average variability is about 14 per cent., while
from the twenty-first to the thirty-sixth day, although there
are several instances as low as 8 or 10 per cent., the average
is 12.5 per cent.
Subject R'did poorly at the start, with an average catch of
8.5 c.c. (see Curve 3) and a variability of 54 per cent, (see
Curve 4). Furthermore, poor learning ability is demon-
strated by the results for the third and fifth days, when the
1 Nothing extensive has been done with children. Probably in working with
them the excursion of the pendulum should be reduced somewhat from the 70-centi-
meter swing employed with adults. However, preliminary trials with the apparatus
as arranged for adults indicate that a nine-year-old child can catch at the beginning
from IO to 12 c.c. and a six-year-old child from 5 to 6 c.c.
A PURSUIT PENDULUM
373
average was in each case slightly lower than on the preceding
days. The results show considerable progress between the
fifth and the twentieth days, a change from 15.5 to 29 c.c.
with some decrease in the variability, although the fluctua-
tions here are quite large. The average for the last five days
g&INQIVIDUAL PRACTICE ON PURSUIT PENDULUM.
12
10
8
Days 2 46 6 10 i * 14. re 18 20 22 24 2ff 28 30 2 34 36
FIG. 5. Individual practice curves showing comparison between two subjects.
Curve I shows the average catch per day and Curve 2 the average variability per day
for one of the most skilful subjects. Curves 3 and 4 show the average catch and
variability per day, respectively, for one of the least capable subjects.
is 32.3 c.c., or approximately 65 per cent, of the possible
catch, with an average coefficient of variability for these same
days of about 18 per cent., in contrast to the 75 per cent,
catch and 14 per cent, variability for the total group of adults.
374
WALTER R. MILES
To illustrate a series of measurements made on an indi-
vidual for the purpose of investigating the effect of some
introduced factor on neuro-muscular efficiency the data in
Table I are given. The pursuit-pendulum test was one of a
number of measurements used in a recent alcohol experiment.
This fragment of data is introduced for the sole purpose of
illustrating the pursuit pendulum results and in no sense as a
contribution to alcohol literature. The complete data are
being elaborated for later publication as an alcohol research.
The series of eight tests required 30 minutes. Following a
light lunch, the subject carried through this series two times
in succession. At the end of the second period he drank I
TABLE I
THE PURSUIT PENDULUM AS A TEST OF NEURO-MUSCULAR EFFICIENCY.
RESULTS SHOWING THE EFFECT OF ALCOHOL
Date
Successive Half-hour Periods
i
2
Drink
3
4
5
XT19'9
Nov. 5
c.c.
1,142
1,170
1,190
1,190
I,2OO
C.C.
1,120
1,152
1,184
1,178
1,180
I liter water
u « «
« « «
« « <(
C.C.
I,IIS
1,176
1,192
1,178
I,2OO
C.C.
,HS
,191
,200
,140
,193
C.C.
1,120
I,l8S
1,190
1,148
1,192
Nov. 8
Nov. 19
Nov. 21
Nov. 22
Av
1,178
1,1
1,163
70
1,172
+ 2
i,i74
+4
I,l67
-3
Nov. 6
1,178
1,156
1,172
1,173
1,172
,150
,166
,185
,173
,208
I liter
2.75 ale.
27.5 grams
« «
,092
,137
,146
,130
,107
1,108
1,122
1,130
1,043
I,IO2
1,096
i, 066
1,096
1,065
i,i35
Nov. 7
Nov. 10
Nov. 17
Nov. 18
Av
1,170
1,1
5.9 per cen
1,176
73
t.
1,122
-51
53
I,IOI
-72
76
1,092
-81
78
Av. loss,
1,172
4-5
1,172
6-5
1,172
6.7
liter of water, or i liter of water in which 27.5 grams of ethyl
alcohol had been diluted. The quantity and temperature of
the liquid were not varied. Fifteen minutes were quite suffi-
cient in which to drink the liquid. After the liquid was taken,
A PURSUIT PENDULUM 375
the series of measurements was repeated three times, that is,
periods 3, 4, and 5 of the day. Table I shows data for five
normal days, on which only water was taken, and for five
other days on each of which I liter of a 2.75 per cent, alcohol
mixture was consumed. No effort was made to disguise the
taste of the alcohol. The subject, an abstainer by habit,
was in the best of physical condition. The values in Table I
give the total catch in cubic centimeters for 25 cups, as
measured by the method of emptying the whole bank of cups
at once. (See Fig. 3.) The two preliminary periods for the
five days on which water only was taken show total averages
of 1,178 c.c. and 1,163 c-c'j or a combined preliminary average
of 1,170 c.c. (46.8 c.c. per catch), which compare favorably
with the two preliminary values, i.e., before alcohol was taken,
on the alcohol days, namely, 1,170 c.c. and 1,176 c.c., or a
combined average of 1,173 c-c- (46.9 c.c. per catch). Periods
3, 4, and 5 show only minor differences (+ 2 c.c., + 4 c.c.,
and — 3 c.c.) from the combined preliminary average in the
case of normal days. On the alcohol days the differences are
all minus, that is, less water was caught after the alcohol was
taken by — 51 c.c., — 72 c.c., and — 81 c.c. for the total
averages. Subtracting the alcohol differences from those for
normal days, it is found that 53 c.c., 76 c.c., and 78 c.c.
represent the alcohol effect for periods 3, 4, and 5, respectively.
These decrements between normal and alcohol performance,
on the basis of 1172 (the grand average for all preliminary
trials on both groups of days), equal 4.5, 6.5, and 6.7 per cent.,
or an average loss of 5.9 per cent., which represents the
alcohol effect on this test of coordination. The subject im-
proved somewhat between November 5 and 22, but since
there is only slight improvement within the day and the water
and alcohol experiments alternate with each other, this prac-
tice change is not troublesome. Indeed, the data are very
consistent, e.g., at no time following ingestion of alcohol was
the subject able to catch as much as he had in the poorer
preliminary period for the same day.
A pursuit coordination test, such as has been here de-
scribed, not only possesses the advantages of requiring very
376 WALTER R. MILES
simple apparatus and of securing quantitative results which
are immediately available without the painstaking reading of
records, but it appears to have elements comparable to many
industrial operations where the task not only requires quick
reaction but also that a movement or movements be executed
according to a fairly definite pattern. A reaction is usually
only the beginning of coordinated movement towards some
end or of a series of such coordinations, and probably in most
instances in practical life the adequacy with which the series
of coordinations is carried through is fully as important, if
not more so, than the mere matter of speed in initiating them.
Especially would this appear to be the case in movements for
compensating, directing, aiming, or otherwise tending any
moving object, where the pace and pattern are not set entirely
by the subject himself. Quickness, precision, and steadiness
of movement have long been tested in reference to stationary
objects. The pursuit pendulum provides a simple means of
measuring these factors in reference to a moving object and
thus supplements the general psychological measurement of
motor control and capacity. The pursuit test invariably
challenges a subject's interest, but practically every one finds
it more difficult than he at first expects.
THE LIMITS OF COLOR SENSITIVITY: EFFECT
OF BRIGHTNESS OF PREEXPOSURE AND
SURROUNDING FIELD
BY C. E. FERREE AND GERTRUDE RAND
Bryn Mawr College
INTRODUCTION
The difficulty of getting reproducible results in deter-
minations of the color sensitivity of the peripheral retina is a
common complaint among clinic workers. This difficulty
is so great as to lead many seriously to question the value
of such determinations in the work of diagnosis. Their value
in diagnosing and in checking up the course of some of the
most serious affections of the eye is readily conceded, how-
ever, provided the needed precision can be attained. The
need of greater precision of working in the laboratory, while
less important to human welfare, is no less insistent. These
combined needs led us several years ago to make a study
of the variable factors which influence the chromatic response,
the details of which are still in progress. Some of these
factors pertain to the control of the stimulus, some are peculiar
to the response of the eye itself. All may be standardized
and controlled. The normal eye is highly sensitive and
complex in its responses but not inherently erratic. While
the abnormal eye may be more erratic, one of the symptoms
it may be of its abnormality, there should be so far as we can
see no essential difference in the technique of the study and
of the testing of its functioning. In fact a characteristic
difference in this regard, which can be determined with
certainty only when other variable factors are controlled,
may well be found to serve as a clue to an early diagnosis of
its abnormality.
The variable factors which influence the chromatic
response of the retina are, so far as we have discovered, the
wave-length and the purity of the stimulus, the intensity of
377
378 C. E. FERREE AND GERTRUDE RAND
the stimulus and the visual angle, length of exposure of the
eye, accuracy and steadiness of fixation, general illumination
and state of adaptation of the retina, breadth of pupil, and
the brightness of the preexposure and of the field surrounding
the stimulus. We have already published considerable data
on the effect of these factors in earlier papers (1). It will
be the special purpose of the present paper to deal with the
last two, the brightness of the preexposure and of the sur-
rounding field. A detailed explanation of the effect of these
two factors on the amount of the chromatic response has
been given in the second of the papers referred to above (1).
A brief explanation and statement of principles will suffice
here.
I. When a small colored stimulus, surrounded by a field,
for example, of white or black is viewed, a sensation is given
which consists of the color mixed with black or white, due
to a contrast sensation induced from the surrounding field.
The effect of fusing a color with white or black is twofold.
(a) There is a quantitative effect due to the inhibition of the
chromatic excitation by the achromatic. In general, in the
central retina at medium and high illuminations, white
inhibits color the most, the grays in order from light to dark
next, and black the least. Also the amount of the inhibitive
action varies with the different colors, with the part of the
retina at which the stimulation takes place, and the state of
brightness adaptation of the retina. The amount of induc-
tion depends upon the difference in brightness between the
stimulus and the surrounding field; it increases with the
distance from the fovea and with decrease in the general
illumination; and, with a given difference in brightness
between the stimulus and the surrounding field, it is greater
with a white than with a black field — also the amount of
increase of induction with decrease of illumination and with
increase of distance from the fovea is greater with a white
than with a black field. And (b) there is also a qualitative
effect. The hue of certain colors is changed by the action
of the achromatic excitation. The change is greatest when
the stimuli are blue and yellow. For example, yellow when
THE LIMITS OF COLOR SENSITIVITY 379
mixed with black gives a greenish yellow which with the
right proportion of components may become an olive green;
and blue when mixed with white or light gray gives a sensa-
tion of reddish blue.
2. When making the color observation in the peripheral
retina, the observer is given a short period of preparation
before the stimulus is exposed, in which to obtain and hold a
steady and accurate fixation. This introduces the factor of
preexposure for, during this period of preparation, the area
which is to be stimulated by color receives a previous stimu-
lation. This previous stimulation, when it differs in bright-
ness from the color, gives a brightness after-image which
mixes with the color sensation and both reduces its saturation
and modifies its color tone. If the preexposure is lighter
than the stimulus color, it adds by after-image a certain
amount of black to the succeeding color impression; if
darker, it adds a certain amount of white. Since both white
and black as after effect reduce the sensitivity to color,
the eye is rendered more sensitive when no after-image is
given, that is when the preexposure is of the same bright-
ness as the color. The preexposure should, therefore, be a
gray of the brightness of the color. No brightness after-
image will then be added to the succeeding color impression
to modify either its saturation or color tone. The only
brightness change acting upon it will be due to the slight
adaptation to this gray during the short time of preexposure.
Even closing the eye, as is frequently done before stimulating,
is equivalent to giving a black preexposure.
The general principle then is clear. There remains only
to explain why in the peripheral retina the short preexposure
which takes place while the eye is obtaining a steady fixation
has so much effect on the color stimulation immediately
following. Two reasons are found for this, (a) The after-
image reaction of the peripheral retina is extremely quick.
While some slight variation is found at different angles of
excentricity, in general the peripheral after-image seems to
reach its maximal intensity with a few seconds of stimulation.
This amount of time is usually consumed in obtaining fixa-
C. E. FERREE AND GERTRUDE RAND
tion and preparing for the stimulation, hence in each observa-
tion there is fused with the color sensation about as strong a
brightness after-image as can be aroused. For this reason
alone it is readily seen why the brightness of the preexposure
is of so much greater consequence in the peripheral than in
the central retina, where the maximal intensity of after-
image is, roughly speaking, obtained from a stimulation of
40-60 seconds or longer. (&) There is apparently no latent
period in case of the peripheral after-image. It flashes out
at full intensity immediately upon the cessation of the stimu-
lation. Thus there is no possibility of escaping the full
effect of the brightness after-image on the stimulus color as
might happen in the central retina where the latent period
obtains, if there were a very short exposure to stimulus color.
CONDITIONS UNDER WHICH THE WORK WAS DONE
The determinations were made in an optics room of the
type described in previous articles (2). The illumination
was kept constant at a value at the point of work of 42 foot-
candles, vertical component; 31.2 foot-candles, 45 degree
component; and 12.5 foot-candles, horizontal component.
Three investigations were conducted.
I. A determination was made of the effect on the apparent
limits of color sensitivity of variations in the brightness of
the field surrounding the stimulus. Three fields were used:
the standard white of the Hering series, giving a surface
brightness at the intensity of illumination used of 0.0209
candle-power per sq. in.; the standard black of the series,
giving a surface brightness of 0.00094 candle-power per sq. in.
and grays of the brightness of the color at the limits of sensi-
tivity in each of the meridians investigated. These grays
ranged in brightness in the different meridians from 0.00350
to 0.00395 CP- Per scl- 'in' f°r red; °-OI445 to 0.0189 for
yellow; 0.01058 to 0.01185 f°r green; and 0.00289 to 0.00366
for blue. In order to study the effect of brightness of sur-
rounding field in separation, the preexposure was in each
case made of the brightness of the color at the point of
investigation.
THE LIMITS OF COLOR SENSITIVITY 381
2. A determination was made of the effect on the apparent
limits of sensitivity of varying the brightness of the preex-
posure. Again three brightnesses were used: the standard
Hering white; the standard Hering black; and grays of the
brightness of the color at the limits of sensitivity in each of
the meridians investigated. The photometric value of the
white, black and the range of grays for each of the colors are
given in I above. In this series of experiments the surround-
ing field was made in each case of the same brightness as the
color at the point of investigation.
3. A determination was made of the combined effect of
preexposure and surrounding field on the apparent limits of
sensitivity. The same three brightnesses were used as in
the preceding investigations. In these cases, however, the
surrounding field and preexposure were both made of the
same brightness, i.e., both white, black or grays of the
brightness of the color at the limits of sensitivity in the
meridians investigated.
Since the results obtained were meant only to be com-
parative of the effect of varying given factors, it was deemed
sufficient to make the determinations with pigment stimuli.
So obtained the results are moreover more nearly what may
be expected in the work of the clinic. The standard red,
yellow, green and blue of the Hering series of papers were
used. The work was done with the rotary campimeter
described in previous papers (3). With the control of
surrounding field afforded by the campimeter, this apparatus
combines the rotary features of the perimeter. Without
some apparatus combining both of these features we have
not found it possible to make a determination of the apparent
limits of sensitivity with an adequate control of the bright-
ness of the surrounding field and of the preexposure. The
need of an apparatus in the clinic by means of which this
control may be accomplished is obvious. Not only is it
impossible to secure an adequate control of these two im-
portant factors by means of the standard perimeter, but a
very great practical difficulty is encountered in daylight work
in getting an equal illumination of the pigment stimulus at
382 C. E. FERREE AND GERTRUDE RAND
different points in the field of vision and a constant illumina-
tion from sitting to sitting. In case of artificial illumination
the latter difficulty can perhaps be eliminated with care;
but the task of securing an equal effective illumination of the
stimulus from point to point in the same meridian and of
corresponding points in different meridians is practically
impossible in case of any perimeter now in use, because of
the unequal shading of the moving stimulus by the observer,
the varying inequalities of the incident and reflecting angles,
etc. In case of the instrument used by us these difficulties
are minimized by using a stationary pigment surface, 20 x 20
cm. placed with special reference to evenness of illumination
at some constant distance (in the present work 45 cm.)
behind the stimulus opening in the campimeter and by
securing the excentric stimulation by shifting the fixation
from point to point along an arm specially constructed for
the purpose. For other points of criticism of the perimeter
as an instrument of precision for either light or dark room
work the reader is referred to former papers. The preex-
posure was secured by inserting the appropriate pigment
surface between the stimulus card and the stimulus opening
in the campimeter. The duration of the preexposure was
kept constant at 2 seconds. The stimulus opening in the
campimeter was 15 mm. in diameter. At the eye, 25 cm.
distant, this subtended a visual angle of 3° 26'.
The more important results given in this paper have been
confirmed repeatedly both in the graduate and under-
graduate work in our laboratory. The determination of the
effect of the brightness of preexposure and surrounding field
on the apparent limits of color sensitivity has in fact formed
a part of the-drill work in the undergraduate laboratory for
several years. Space will be taken here for the results of
only one observer — the observer whose results have been
given in the preceding studies on the color sensitivity of the
peripheral retina.
As has already been indicated, the effect of brightness of
the preexposure and of the surrounding field falls under the
general heading of the inhibitive action of the achromatic
THE LIMITS OF COLOR SENSITIVITY 383
excitation on the chromatic. This action takes place how-
ever the achromatic excitation is aroused — by the admixture
of white light, by after-image, by contrast, etc. It may be
strikingly and conveniently demonstrated to large numbers
at once in the following lecture room experiments, (a) Set
up side by side on three color mixers discs made up of 180
degrees of color, e.g. blue, and 180 degrees of white, 180
degrees of blue and 180 degrees of gray of the brightness of
the blue, and 180 degrees of blue and 180 degrees of black.
When mixed, although the eye receives the same amount of
colored light from each set of discs, the mixture with black
seems to have lost but very little, if any, color; the mixture
with white is a lavendar with but little color; and the mix-
ture with gray of the brightness of the color, in this case a
very dark gray, is less saturated than the mixture with black.
When different grays are used the saturation decreases appar-
ently in graded steps as white is approached. The demon-
stration can be made on a single color mixer by compounding
the color disc with white, black and gray discs of different
breadths or radii. When rotated this gives the effect of a
surface made up of three concentric zones or rings, one in
which the color is mixed with white, one with gray and the
other with black. The demonstration may be made roughly
quantitative by determining the proportions of color required
to give the chromatic threshold in black, white and the grays;
also by determining the proportions of color and the achro-
matic series to give equal saturations.
(b) Prepare a preexposure surface, half white and half
black, 50x60 cm. Expose the eye 15-20 seconds and pro-
ject the after-image on a colored surface, e.g., blue, of the
same dimensions. The half of the field preexposed to black
will appear a very pale unsaturated lavendar, while the half
preexposed to white will be a dark strongly saturated blue,
although the eye receives the same amount of light from
both halves of the field. As the after-image dies away the
two halves of the field become more and more nearly alike
in saturation and color tone. If desired, the preexposure
surface may be made of white> black and a series of graded
384 C. E. FERREE AND GERTRUDE RAND
grays, appropriately arranged. When this is done the
graded loss in saturation due to the different brightnesses of
the after-image may be observed. This demonstration also
may be made quantitative by finding the threshold of color
after the eye has been preexposed for 15-20 seconds to white,
black and the grays.
(c) Prepare contrast discs with narrow rings of color and
inside and outside surfaces of black, white and a gray of the
brightness of the color, respectively. Set up on color mixers
side by side and rotate to smooth out all margins. The
colors are lightened and darkened respectively by contrast
induced by the black and white fields. The effect of these
achromatic excitations on the hue and saturations of the
colors is similar to that obtained in the former experiments.
A more striking effect is produced if a mixed color, <?.g.,
orange, is used. The quantitative features noted above can
also be utilized in this demonstration by employing for the
contrast ring in each case a gray of the brightness of the
color and enough of the color to give the threshold of color
sensation when acted upon by the white and black inductions.
The effect of induction and after-image, it will be remem-
bered, are not nearly so striking in the central as in the
peripheral retina. Much more induction with a given bright-
ness difference between the inducing and the contrast field,
for example, is produced in the peripheral retina, and only a
short period of preexposure (2-3 seconds) is required to give
a strong after-image with no latent period.
RESULTS
The following results were obtained: (i) The widest
angular limits of the color zones were obtained when the
preexposure and surrounding field were of the same bright-
ness as the color. (2) When the brightness of preexposure
and surrounding field were different from that of the color,
the effect of surrounding field was less than that of preexpo-
sure; and the effect of either is always less than the com-
bined effect of both. (3) In some meridians the effect of
surrounding field alone narrowed the limits as much as
THE LIMITS OF COLOR SENSITIVITY 385
II degrees; the effect of preexposure alone, as much as 17
degrees; and the combined effect of preexposure and sur-
rounding field, as much as 20 degrees.
(4) The amounts the limits were narrowed for red,
yellow, green and blue, respectively, by a white preexposure
alone ranged in the different meridians1 from 4-15 degrees,
2-17 degrees, 3-15 degrees, and 4-12 degrees; by a black
preexposure from 3—11 degrees, 3-10 degrees, 4-13 degrees,
and 2-12 degrees; by a white surrounding field 1.5-10 de-
grees, 2-9 degrees, 2-11 degrees, and 2-10 degrees; by a
black surrounding field 1-8 degrees, 1-8 degrees, 2-10
degrees, and 1.5-9 degrees; by a combined white preexposure
and white surrounding field 5-19 degrees, 2-20 degrees,
4-20 degrees, and 5-17 degrees; by a combined black pre-
exposure and black surrounding field 4-17 degrees, 5-12
degrees, 7-18 degrees and 5-18 degrees. When the effect
of a white or black surrounding field alone was wanted, the
preexposure was made of the same brightness as the color
at the point of investigation; similarly when the effect of a
white or black preexposure was wanted, the surrounding
field was made of the same brightness as the color at the
point of investigation. The value of the limits with a pre-
exposure and surrounding field of the same brightness as the
color served in each case as the standard value in terms of
which to estimate the amounts the limits were narrowed by
the white and black preexposures and surrounding fields and
their combinations.
These values, it will be remembered were obtained with a
very precise control of the illumination of the working
surfaces. It is obvious that a much greater variability of
result should be expected had there been no better control
of the constancy of illumination than is ordinarily exercised
in office and clinic work, and too often in laboratory work.
The effect on both the limits and hue of the color of such
variations in the daylight illumination of the working surfaces
as are apt to occur over long periods of time when no especial
control is exercised, will be given in a later paper.
1 In the order shown in the tables.
386
C. E. FERREE AND GERTRUDE RAND
In order to realize how profoundly the powers of chromatic
response must have been affected to change the limits of
sensitivity by the amounts represented by the above figures
one must bear in mind how abruptly sensitivity falls off in
the far periphery of the retina. A determination of the
thresholds of color in the temporal meridian with preexposure
TABLE I
LIMITS OF COLOR FIELD FOR RED
Showing the Effect of Brightness of Preexposure, Brightness of Surrounding Field, and
the Combined Effect of Brightness of Preexposure and Surrounding
Field on the Apparent Limits for Red
Meridian
Effect of Preexposure1
Effect ot Surrounding
Fields
Combined Effect of
Preexposure and Sur-
rounding Field
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Upper o°. ,
58
49
49
47
43
47
50
Si
60
73
79
85
89
89
85
75
45
43
43
43
38
$
47
53
59
64
75
83
82
78
62
47
43
41
42.5
37
42
45
47
56
68
74
80
85
85
82
65
58
49
49
47
43
47
5°
£
73
79
85
89
89
85
75
4*
46
46
45-S
4i-5
43
48
48.5
55
66
70
80
85
84
81
65
5°
46
44
44-5
40
43
47
48.5
57
70
76
82
88
86
83
68
58
49
49
47
43
47
5°
&
73
79
85
89
89
85
75
40
41
38.5
41
38
41
45
46
52
55
60
69
80
80
77
60
41
39
37-5
40
38
42-5
^
46
|6
62
72
78
84
83
Si
64
Nasal 25°
45°- • •
70s
90°..
110°..
135°-.
155°
Lower 180°
Temporal 25°
45°o
70°
! 9°°o
110°
; I3so°
155°
and surrounding field of the same brightness as the color for
red, yellow, green and blue at 5 degrees, 3 degrees, 2 degrees
and i degree respectively from the limit shows the following
values: for red, 132, 150, 250 and 320 degrees; for yellow,
100, 150, 240 and 330 degrees; for green 130, 145, 260 and
345 degrees; and for blue 130, 145, 200 and 310 degrees.
1 In determining the effect of the different brightnesses of preexposure, the bright-
ness of the surrounding field was made equal to that of the color at the point of in-
vestigation.
2 In determining the effect of the different brightnesses of surrounding field, the
brightness of the preexposure was made equal to that of the color at the point of
investigation.
THE LIMITS OF COLOR SENSITIVITY
387
For red thus there was an increase of 172.7 per cent, in the
threshold in passing to the limit from a point 5 degrees from
the limit; for yellow an increase of 260 per cent.; for green
an increase of 207.7 per cent.; and for blue an increase of
207.7 Per cent. For a more detailed experimental analysis
of the effect of preexposure, surrounding field, intensity of
TABLE II
LIMITS OF COLOR FIELD FOR YELLOW
Showing the Effect of Brightness of Preexposure, Brightness of Surrounding Field, and
the Combined Effect of Brightness of Preexposure and Surrounding
Field on the Apparent Limits for Yellow.
Meridian
Effect of Preexposure >
Effect of Surrounding
Field'
Combined Effect of
Preexposure and Sur-
rounding Field
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Upper o°
47
42
4*
46
44
46
50
48
59
65
73
87
89
89
87
72
41
39
37
42
42
42
46
44
5»
48
63
70
75
81
80
60
37-S
38
36
40
40
38
45
44
54
55
70
84
85
86
|4
65
47
42
42
46
44
46
5°
48
59
65
73
87
89
89
87
72
41
40
40
44
42
43
4<
46
53
58
68
79
80
83
82
63
39
39
38
42
41
41
47.5
46
I6
61
72
86
87
87
85.5
67
47
42
8
8
50
48
I9
65
73
87
89
89
87-
72
38
38.5
37
42
42
4*
46
43
47
11
69
72
80
78
59
36
37
35-5
39
38.5
37
45
43
52
I3
67
80
84
85
84
63
Nasal 25°
45°. •
70°..
90'..
lioV.
135°. •
155°
Lower 180°
Temporal 25° .
< '
7°l •
9°I '
110° .
'«: •
155 •
the illumination of the visual field, amounts of induction
with different brightness relations of surrounding field to
stimulus at different intensities of illumination, etc., and the
effect of all of these on the thresholds of color and the limits
of sensitivity the reader is referred to the first two papers
cited in the appended bibliography (1).
5. In those meridians in which the limits are wide there
is a general tendency for the white preexposure and surround-
ing field to narrow the limits more than a black preexposure
1 Brightness of Surrounding Field: gray of the brightness of yellow.
* Brightness of Preexposure: gray of the brightness of yellow.
388
C. E. FERREE AND GERTRUDE RAND
and a black surrounding field. We have stated in our intro-
duction that the amount of inhibition of the chromatic by
the achromatic excitation varies with the color, the part of
the retina stimulated and the state of adaptation of the
retina. This statement applies also to the relative effects
of white and black. In the central retina at medium and
high illuminations white inhibits color much more than black.
TABLE III
LIMITS OF COLOR FIELD FOR GREEN
Showing the Effect of Brightness of Preexposure, Brightness of Surrounding Field, and
the Combined Effect of Brightness of Preexposure and Surrounding
Field on the Apparent Limits for Green
Meridian
Effect of Preexposure 1
Effect of Surrounding
Fields
Combined Effect of
Preexposure and Sur-
rounding Field
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Upper o°. .
36
35
38
39
39
37
37
33
37
37
f
61
69
65
57
44
26
30
30
34
35
3i
32
3°
32
30
34
11
S3
42
39
29
27
28
31
33
31
29
29
31
32
36
53
60
56
44
37
36
35
38
39
39
37
37
33
37
37
42
61
69
65
57
44
28
31
3A
36
37
33
34
31
34
34
32
56
60
58
46
4i
31
29
30
31
35
33
3i
30
33
35
40
57
62
61
47
39
36
35
38
39
39
37
37
33
37
37
42
61
69
65
57
44
27
26
29
32
33
3i
3i
29
28
28
30
47
S2
46
37
35
22
21
24
27
28
30
25
26
26
26
33
SO
S3
So
39
34
Nasal 25°
45°
70°
Q0»
!»•::
I3c°..
155°
Lower 180°
Temporal: 25°
<
7°o
9°°
110°
I3s:
155°
At these illuminations therefore a black preexposure and
surrounding- field are much more unfavorable than white.
At lower illuminations this difference in effect becomes less
pronounced. In the far periphery of the retina the following
are some of the conditions which contribute to make black as
preexposure and surrounding field give wider limits of
sensitivity, (a) A condition of low illumination and a state
of low illumination adaptation, (b) A darkening of all of the
1 Brightness of Surrounding Field: gray of the. brightness of green.
2 Brightness of Preexposure: gray of the brightness of green.
THE LIMITS OF COLOR SENSITIVITY
389
colors, particularly red and yellow (the Purkinje shift of the
peripheral retina). This brings the brightness of the color
nearer to black than to white and the stronger relative
darkening of red and yellow than of their neutral or colorless
preexposures and surrounding fields, increases the contrast
TABLE IV
LIMITS OF COLOR FIELD FOR BLUE
Showing the Effect of Brightness of Preexposure, Brightness of Surrounding Field, and
the Combined Effect of Brightness of Preexposure and Surrounding
Field on the Apparent Limits for Blue
Meridian
Effect of Preexposure *
Effect of Surrounding
Field*
Combined Effect of
Preexposure and Sur-
rounding Field
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Gray of
Brightness
of Color
White
Black
Upper o°
52
45
4*
46
52
5°
II
70
70
79
86
9i
91
89
80
40
39
44
41
42
46
47-5
48
63.5
62
7i
78
86
85
84
75
46
32
46
41
42
45
47-5
46
62
65
73
82
85
I*
83
75
52
45
48
46
52
5°
52
58
70
70
79
86
9i
9i
89
80
42
41
45
44
48
47-5
50
&
65
73
80
89
88
86
77
48
42
46
44
47
47
49
$
64-5
68.5
75
84
89
88
85
77
52
45
4*
46
52
SO
52
58
70
70
79
86
9i
9i
89
80
35
38
40
41
42
It
43
61
I6
65
77
84
83
83
75
its
40
41
40
43
46
42
59
I?
69
80
84
83
83
74
Nasal 25°
4<;°. .
70°
S>
110° .
I-.C0
'"a
155°
Lower 180°
Temporal: 25° .
45°0 '
7° '
90° .
110° .
'«: •
iss .
and after-image effect for white and decreases it for black.
The darkening of red and yellow in passing to the far periphery
of the retina is very great. In the nasal half of the retina
with its wide limits, the effect of this darkening on the results
of our determinations was, of course, the most pronounced.
As colors darken, there is, when a certain point in the process
is reached, varying with the color, a tendency for them to
lose their saturation very rapidly, (c] Achromatic induction
increases very strongly with decrease of illumination and
therefore increases in passing from the center to the periphery
1 Brightness of Surrounding Field: gray of the brightness of blue.
1 Brightness of Preexposure: gray of the brightness of blue.
390
C. E. FERREE AND GERTRUDE RAND
of the retina. It increases much faster for white than for
black.
In the meridians in which the limits are narrower the
situation is more nearly as it is in the central retina. Here
the tendency is for the limits to be narrowed more by a black
45
70
flO
(35
155
180
FIG. I. Effect of brightness of preexposure on the limits of the color field. In
this chart are shown the apparent limits for red with preexposures respectively of white,
black, and gray of the brightness of the color at the point of investigation, surrounding
field in each case gray of the brightness of the color at the point of investigation.
than by a white preexposure and surrounding field. In
some meridians the amount of narrowing is approximately
equal for both. Another factor which tends to make the
effect more nearly the same in these meridians for all back-
grounds and preexposures is the more abrupt falling off in
sensitivity. That is, more effect on sensitivity is required
here to change the limits by a detectable amount than is
THE LIMITS OF COLOR SENSITIVITY
39*
required in those portions of the retina where the sensitivity
grades off more slowly.
A detailed representation of the results is given in Tables
I-IV. and a graphic representation of a part of the results
in Figures 1-6. In the tables results are given separately
45
45
7C
• 55
ISO
FIG. 2. Effect of brightness of surrounding field on the limits of the color field.
In this chart are shown the apparent limits for red with a surrounding field respectively
of white, black, and gray of the brightness of the color at the point of investigation, pre-
exposure in each case gray of the brightness of the color at the point of investigation.
for the effect of preexposure, surrounding field and com-
bined effect of preexposure and surrounding field for each
of the four colors: red, yellow, green and blue. In case of
the figures, however, space has been taken to represent
separately the effect of preexposure and surrounding field
for only one of the colors, red — Figs. 1-3. Figs. 3-6 show
392
C. E. FERREE AND GERTRUDE RAND
the combined effect of preexposure and surrounding field on
each of the four colors. In our previous papers the repre-
sentation of results has been in terms of position on the retina.
In this paper the representation has been made in terms of
field of vision.
25
Z5
45
70
(35
155
ISO
FIG. 3. The combined effect of brightness of preexposure and surrounding field
on the limits of the color field. In this chart are shown the apparent limits for red
with both preexposure and surrounding field respectively of white, black, and gray of
the brightness of the color at the point of investigation.
CONCLUSION
It is quite obvious from the preceding data that repro-
ducible results can not be hoped for in perimetric or campi-
metric determinations of the sensitivity of the peripheral
retina unless the variable effects of preexposure and surround-
ing field be eliminated from the conditions of work. This
can be done completely only by making the brightness of
THE LIMITS OF COLOR SENSITIVITY
393
the preexposure and surrounding field in each case the same
as that of the color employed and working under constant
intensity of illumination. Among the effects of a variable
intensity of illumination on the results of a perimetric or
campimetric determination the following two may be men-
25
Z5
45
no
135
155
155
FIG. 4. The combined effect of brightness of preexposure and surrounding field
on the limits of the color field. In this chart are shown the apparent limits for yellow
with both preexposure and surrounding field respectively of white, black, and the gray
of the brightness of the color at the point of investigation.
tioned. (a) When the color stimulation is given by light
reflected from pigment stimuli of a given coefficient of reflec-
tion the amount of colored light obtained depends upon the
intensity of light incident on the reflecting surface. And (b)
a brightness match of preexposure and surrounding field with
the stimulus surface will not hold at different illuminations
(the Purkinje phenomenon).
394
C. E. FERREE AND GERTRUDE RAND
We have worked out in previous papers the conditions
under which the desired standardization of intensity and
color value of illumination and control of brightness of pre-
exposure and surrounding field may be obtained in labor-
45
45
00
(35
155
ISO
FIG. 5. The combined effect of brightness of preexposure and surrounding field
on the limits of the color field. In this chart are shown the apparent limits for green
with both preexposure and surrounding field respectively of white, black, and gray of the
brightness of the color at the point of investigation.
atory campimetry (4). These conditions however are scarcely
feasible for the work of the office or clinic. We have there-
fore more recently devised and constructed a perimeter by
means of which equal illumination of the stimulus is received
at every point on the perimeter arm in all meridians and the
effect of brightness of preexposure and surrounding field can
be eliminated with an ease and speed of manipulation which
THE LIMITS OF COLOR SENSITIVITY
395
should be feasible for office and clinic work and with a com-
pleteness of result that should be adequate for this type of
work. We have in fact constructed two types of perimeter
either one of which provides for the uniform illumination of
the arm of the perimeter. The perimeters will be described
in a later paper.
tj
ts
70
155
155
(SO
FIG. 6. The combined effect of brightness of preexposure and surrounding field
on the limits of the color field. In this chart are shown the apparent limits for blue
with both preexposure and surrounding field respectively of white, black, and gray of the
brightness of the color at the point of investigation.
COMMENT
A much more detailed study of the quantitative relations
of the chromatic and achromatic components of the visual
sensation for different intensities of stimulus and for different
states of the reacting eye is needed. There are many im-
portant practical bearings of the knowledge that would be
396 C. E, FERREE AND GERTRUDE RAND
gained by such a study. For example, it is often deemed
sufficient to give a colorimetric specification of a light at one
intensity alone in spite of the fact that the saturation, even
the hue of the color, changes with the intensity as well as
the composition of the light. We are all familiar in a general
way with the fact that even the sensation aroused by a
spectrum band of light begins as achromatic or colorless at
very low intensities, passes through saturation and hue
changes with increase of intensity of light and finally becomes
colorless again at high intensities. We have pointed out
many times in connection with problems of lighting (5)
that while a specification of the composition of light is in-
dependent of intensity, a true colorimetric specification
may not, depending on the method used, be definite un-
less it is accompanied also with a specification of intensity.
Filters designed to give a certain coloration of light can not
be depended upon to give this subjective coloration at all
intensities even though the wave-lengths transmitted are
in the same proportions. Indeed when used in connection
with the same intensity of source the coloration of the il-
lumination of an object as seen by the eye, particularly
the saturation, will vary at different distances from the
source. The lack of realization of this dependence of the
color of light on its intensity as well as its composition has
doubtless played no small part in the popular confusion
which exists as to the comparative color values of different
artificial lights and of the closeness of approximation of
certain artificial lights to daylight. The surface of a Wels-
bach mantle, 0.7 per cent, ceria, viewed directly is, for
example, whitish; but the reading page illuminated by it to
ordinary working brightness appears distinctly yellowish
green. Again the illumination given by the blue bulb lamp
may be judged of different color values depending upon the
intensity of light falling on the illuminated object. Comple-
mentary colors combined to gray at medium or high intensi-
ties may not be seen as colorless at low illuminations, <?.g.,
the gray produced by combining the Hering standard blue
and yellow under daylight of good intensity becomes dis-
THE LIMITS OF COLOR SENSITIVITY 397
tinctly lavendarish under the same light at low intensities.
Daylight itself is popularly said to become bluish at low
intensities. Examples may thus be multiplied indefinitely
of the apparently peculiar complexity of the selectiveness of
the eye's chromatic response to intensity.
In addition to the practical bearings of the shifting of the
quantitative relations of the achromatic and chromatic
components in the visual sensation, with no change in the
composition of light, there is the interesting problem of
explanation. Many factors, it may be, are operative in the
production of this phenomenon: a selectiveness of response
to intensity, perhaps even a change in the range of the eye's
chromatic response to wave-length with change of intensity?
in case of spectrum lights; this and slight variations for
change of intensity, in the cancelling proportions of the com-
plementary colors and in the mutually inhibitive actions of
the non-complementary colors, in case of mixed lights; a
direct action of the achromatic excitation on the chromatic,
for both simple and mixed lights; etc. It seems not only
reasonable but necessary to infer this latter action because
the same type of effect is produced on the color when the
achromatic component of the sensation is varied in all of the
following ways: by keeping the composition of the light the
same and varying its intensity, by adding colorless light, by
adding white or black to the sensation as after-image or con-
trast, and by the achromatic changes in adaptation. No
other explanation seems possible when the phenomenon is
produced as an effect of preexposure and surrounding field or
as we commonly say by after-image and contrast, as has been
the case in the work reported in this paper.
BIBLIOGRAPHY
I. RAND, GERTRUDE. The Effect of Changes in the General Illumination of the
Retina upon its Sensitivity to Color, PSYCHOL. REV., 1912, 19, 462-491;
The Factors that Influence the Sensitivity of the Retina to Color: A Quanti-
tative Study and Methods of Standardizing, PSYCHOL. MONOG., 1913, 15, No.
62, i66+xl; FERREE, C. E. AND RAND, G., The Absolute Limits of Color
Sensitivity and the Effect of Intensity of Light on the Apparent Limits, PSY-
CHOL. REV., 1920, 27, 1-24.
398 C. E. FERREE AND GERTRUDE RAND
z. FERREE, C. E. AND RAND, G., An Optics Room and a Method of Standardizing its
Illumination, PSYCHOL. REV., 1912, 19, 364-373; A Simple Daylight Photom-
eter, Amer. J. of Psychol., 1916, 27, 335-340; Chromatic Thresholds of
Sensation from Center to Periphery of the Retina and their Bearing on Color
Theory, Part I, PSYCHOL. REV., 1919, 26, 16-42. FERREE, C. E., RAND, G.
AND HAUPT, I. A., A Method of Standardizing the Color Value of the Day-
light Illumination of an Optics Room, Amer. J. of Psychol., 1920, 31, 77-87.
3. FERREE, C. E., Description of a Rotary Campimeter, Amer. J. of Psychol.,
.1912, 21, 449-453. FERREE, C. E. AND RAND, G., A SpectroscopSc Appar-
atus for the Investigation of the Color Sensitivity of the Retina, Central and
Peripheral, /. of Exper. Psychol., 1916, i, 246-283.
4. Op. cit.: also FERREE, C. E. AND RAND, G., A Substitute for an Artificial Pupil,
PSYCHOL. REV., 1916, 23, 380-383.
5. FERREE, C. E. AND RAND, G., Some Experiments on the Eye with Different Illum-
inants — Part I, Trans. Illuminat. Eng. Soc., 1918, 13, 1-18; Part II, ibid.,
1919, 14, 107-133; etc.
VOL. 27, No. 6 November, 1920
THE PSYCHOLOGICAL REVIEW
DO WE THINK IN WORDS?
BEHAVIORIST vs. INTROSPECTIVE CONCEPTIONS
BY ARTHUR S. OTIS
Stanford University
I. Purpose of the Discussion. — The following discussion
takes its departure from the reading of Dr. John B. Watson's
'Psychology from the Standpoint of a Behaviorist.'
It is the purpose of the writer to discuss certain hypotheses
which are put forth with seeming conviction in the text but
which are believed by the writer to be false.
To discuss the whole subject of the Behaviorist point of
view, in relation to the more generally accepted points of
view in psychology would be quite impossible in the scope of
this article. That a text in psychology should be written
in which the author not only purposefully avoids the mention
of such concepts as perception, ideation, association of ideas,
consciousness, attention, will, etc., but even goes so far as
to claim that these concepts are useless for purposes of
psychology, is of course quite a source of wonder. The
indispensability of the concepts avoided by Behaviorist
psychology and of the use of introspection will be apparent,
we believe, from the discussion of but one 'assumption'
which it makes. We shall confine this article to the discussion
of this assumption.
The hypothesis referred to is 'the point of view that has
been advocated throughout the text, namely, that thought
is the action of language mechanisms' (p. 316). The mean-
ing of the expression, language mechanisms, is carefully
defined by the author as referring to any of those muscles
399
400 ARTHUR S. OTIS
of the body which actuate to produce words whether spoken,
written, or gesticulated (as by deaf mutes). The meaning
of the word, thought, as used in this hypothesis is not ex-
plicitly stated, but may be inferred with confidence from
various passages which we shall quote1 and is here taken to
be the same as the meaning of thought when used by those
who are not Behaviorists.
To be sure, the author states in the preface that "the
terms thinking and memory have been carefully redefined
in conformity with Behaviorist psychology." On page 14
we find in italics the expression: '"thinking," by which we
mean subvocal talking.' This may constitute the re-defini-
tion, but if so it obviously begs the question which we are
discussing; namely whether 'thinking' as ordinarily under-
stood does consist of subvocal talking. We shall therefore
leave this re-definition out of account.
It will be realized that the adjustment of an individual
to his environment may involve acts requiring mental
activity of all degrees of consciousness, from the most auto-
matic habitual or instinctive acts requiring little or no
consciousness, such as moving the eyes toward an object it
is desired to see, to the solving of problems requiring the
1 "A man may sit motionless at his desk with pen in hand and paper before him.
In popular parlance we may say he is idle or 'thinking,' but our assumption is that his
muscles are really as active and possibly more active than if he were playing tennis.
But what muscles? Those muscles which have been trained to act when he is in
such a situation, his laryngeal, tongue, and speech muscles generally" (p. 15).
"We manipulate vocally" (when trying to think of the name of a familiar person)
"by running over the names beginning with each succeeding letter of the alphabet,
or by saying 'black hair,' 'blue eyes,' 'six feet tall,' and the like" (p. 305).
"The explicit and implicit language habits are formed along with the explicit
bodily habits and are bound up with them and become a part of every total unitary
action system that the human organism forms. . . . They are present in the simplest
types of adjustment that he makes. We can see the functioning of language habits
only slightly in certain activities, as, for example, in swimming, tapping on the table
with a pencil, while in certain other types they form an integral part ..." (p. 309).
"Our view is that overt language develops under social training. It is thus
absorbed into and becomes a part of every total integration of the individual. Hence
when he is making adjustments in the absence of other like beings language remains
as part of the process" (p. 323).
"... the maiden thinks of her lover in words the beautiful thoughts
of the idealist for mankind as a whole or of the mother for her child ... are couched
in words or their equivalent" (p. 325).
DO WE THINK IN WORDS? 40 1
most concentrated mental effort. It is hardly conceivable
that the Behaviorist would claim that all such adjustment
involves language mechanisms, as the passages quoted would
imply if taken literally. To simplify matters, however, we
will limit our discussion to that type of adjustment ordinarily
referred to as 'thinking,' namely, those mental processes of
the problem-solving sort which require some degree of con-
scious mental effort, since these are open to introspective
investigation. That even these processes of adjustment do
not necessarily require language we shall attempt to show by
appeal to logic and common experience, omitting arguments
ad hominem.
2. Examples of Thinking. — Let us consider one or two
simple cases of problem solving and subject them to critical
psychological analysis in order to determine whether they
involve language.
Suppose I have unfolded a new map and am attempting
to fold it again as it was. I have no complete habit, not
having folded a map exactly like this before. Let us see
what happens. Surely there is a better way than to let
someone watch me and report his inferences. He would
merely see me look at the map and, let us say, try to fold it
in one way but fail and then try another way and succeed.
He might infer that my method was the so-called 'trial and
error' or 'perseverance' method. Or if my lips have moved
he might infer that I arrived at the solution of the problem
by means of the action of the muscles of my lips and other
speech organs. This appears to be the method of the
Behaviorist.
Let me introspect and report from direct observation
what happened from the point of view of one looking on
from the inside. The writer does not wish to be misunder-
stood as assuming that introspection is infallible. One's
testimony is not infallible even when he observes with his
own eyes an incident which transpires directly before him.
Relatively speaking however, introspection is far more
direct and reliable than inference based upon observation
from the outside.
402 ARTHUR S. OTIS
On introspection I report as follows regarding my action
with the map. More or less mechanically, as we say, that
is, while thinking about what I had seen on the map, I
began to fold the map along one of the creases. After a
moment I became aware that the map was not falling into
its accustomed folds. I then became aware of the need of
finding the correct way to fold the map and I unfolded it in
order to begin again. I recalled from previous experience
that the crease on which the first fold must be made is one
which runs entirely across the paper. I therefore looked
for such a fold and on finding it folded the paper on it and
repeated the process until the map was entirely folded up,
making no further error.
Now this adjustment which I have made to the problem
of folding the map was ( thinking,' alike in the popular
usage and that of the psychologist. The Behaviorist claims
that thinking is the action of language mechanisms. Let us
go over this example of thinking again and examine it very
minutely to see whether there is any necessary connection
between language and the solving of the problem.
First of all, how do I become aware that the map is not
falling into its accustomed folds? If I go slightly back of
this awareness I note a feeling of contradiction between a
subconsciously expected feeling of flatness and the experienced
feeling of bulginess. This contradiction, we may say, caused
me to become aware of the improper folding of the map —
caused the shift of my attention from the thoughts of what I
had seen on the map to the matter of folding the map.1
How did I then become aware of the need of finding the
correct way of folding the map? The experience suggests
no other explanation than merely to say that the idea of
contradiction 'called up' or 'suggested' the idea of need.
This idea in turn called up the idea of beginning again.
We may explain this process by saying that it was probably
1 That a subconscious awareness of contradiction may give rise to an idea of
need, together with an affective state which effects a shift of attention (clear aware-
ness) to the need, is a matter of so frequent observation in structural psychology as to
be considered a scientific fact. Such a fact, however, is of course quite unthought of
in Behaviorist psychology, being wholly outside of its scope.
DO WE THINK IN WORDS? 403
the result of a previously formed habit. One has learned
in such cases that it is best to begin again. When the idea
of unfolding the map again has come to occupy more or less
of the whole of consciousness, 'the thought takes form in
action.' Behaviorist psychology concedes such a phenom-
enon, so we need not attempt to explain it. Having unfolded
the map I recalled previous experiences regarding the folding
of large sheets of paper. We will say that the perception
of the paper before me and the idea of need of folding, to-
gether served to bring forth from my memory store those
ideas which came to my mind. These together with the
perceptions of the map during the process of folding served
to educe that train of ideas which guides the folding to a
successful termination.
Now what is the material of all this mental activity?
What do these ideas consist of? They consist of images,
visual, tactual, kinaesthetic, etc., of maps, and of certain
aspects of these images such as creases, folding movements,
flatness, bulginess, etc. They have nothing to do with
language, necessarily. The idea of flatness is tactual or
visual or both, the idea of a folding movement may be visual
or kinsesthetic or both. The idea of the length or direction
of a crease is visual or kinsesthetic or both. Possibly other
types of imagery enter to a slight extent. But no language
need be involved.
Let us now consider another type of thinking. Let the
reader ask himself why it is more difficult to play a game of
chess blindfolded than with the chess board visible before
him. Obviously the answer is that the perception of the
relative positions of the chess men is a great aid to the mental
manipulation which constitutes the basis of the study of
moves. Moreover, anyone who has played chess or checkers
will immediately appreciate the aid that would be derived
from actually making the trial moves that are contemplated,
in more clearly appreciating the relations that such moves
would introduce. If the thinking were done by means of
subvocal language it would seem that seeing or not seeing
the chess board would make no difference. The obvious
404 ARTHUR S. OTIS
answer is that the thinking is done by means of the per-
ceptions of the board and men as they are, the mental imaging
of the movement of the men into new positions and the
appreciation of the spacial and temporal relations between
the pieces and their possible moves as introduced after the
mental manipulation. No language whatever is required.
As we shall show, a person may indeed talk to himself while
contemplating moves, but this activity is entirely secondary
and supplementary.
3. The Material of Thought. — Thinking, as an adjustment
of the individual to his environment, as the solving of prob-
lems, consists of the evolving of new ideas, concepts, or
meanings, from old. This is accomplished by recombination
of the elements of the old into new patterns. By ideas,
concepts, and meanings are meant image patterns, whether
they be of words, objects seen, sounds heard, things felt,
tasted, sensed in any manner whatsoever, or any quality,
attribute, or aspect of such image patterns as may be con-
ceived separately by abstraction, such as shape, color, surface,
volume, extent, duration, intensity, symmetry, movement,
similarity, difference, causality, symbolism, abstractness or
affective quality; or of whatever degree of clearness or
attenuation or incipiency the images or image aspects may be.
We may think, therefore; that is, we may evolve new ideas,
concepts, meanings, in terms of image patterns of any kind
whatsoever, or of the consciousness (idea) of any relationship
whatsoever between these image patterns.
For example, I am thinking when I am effortfully engaged
in composing a piece of music. I sit at the piano with music
paper at my side. My mind is occupied with perceptions
and images of- tones, tone combinations, tone sequences, tone
relationships, tone emotional effects, tone symbols (dots on
paper) the making of these symbols, etc.
My effort consists in the maintenance of my attention to
the work, the calling up of sequences of tone images,1 the
1 Strictly speaking I adopt the mind set that will result in the calling up of tone
images, or that is calculated to do so. (Sometimes I may succeed better than at
other times.) We cannot call up an image necessarily at will. Generally it is a case
of taking a certain mental attitude ordinarily called 'trying to think' which usually
results in the recall of the idea desired.
DO WE THINK IN WORDS? 4<>5
comparison of these, the appraisal of their respective aesthetic
values, the choice of one or another, the calling up of the
proper symbols of notation in which to write down the
musical ideas, and the writing of these. No language is
involved in any of this thinking (except perhaps a final
translation of the results of thought into symbols). In this
case also I may compose without the piano; but this is more
difficult, since I am compelled to make my judgments upon
images only, whereas with the piano I may employ the per-
ceptions of the tones themselves in my judgments. If my
musical thinking were all done by means of the action of
speech muscles we do not see that it would make any differ-
ence whether the piano were struck or not.
Similarly, one is thinking when he is creating a new archi-
tectural design, or a drawing or painting or statue or stage
setting, or conceiving of a new dance movement or inventing
a new mechanical contrivance or playing tennis or searching
for the cause of engine trouble. The material of one's
thoughts in all these cases is in the form of images, which
need be only visual, auditory, tactual, kinaesthetic, may be,
in fact, of any kind whatever according to the requirements
or to one's ability to call forth such images. As we have
said, one may do any amount of talking to oneself while
thinking — which is merely putting one's thoughts in words
after they are thought — but the talking is not the thinking.
It is supplementary to it in exactly the same way that
describing a landscape is supplementary to seeing it.
Thinking may be called the controlled association of ideas,
in contradistinction to the free association of ideas. In the
free association of ideas, by which we refer to what is ordi-
narily called day dreaming or revery, ideas follow one another
in a more or less unguided manner, yet in a fairly rational
way as compared with the incongruous manner of idea se-
quence sometimes experienced in dreams. Doubtless there
is some sort of control even in 'free association' though it
may be the general interest in the subject of thought or the
control occasioned by thought habits. However, in what we
have called controlled association of ideas characterizing
406 ARTHUR S. OTIS
thinking, the ideas are guided in their sequence by some
conscious aim, e.g., a problem to be solved. Irrelevant ideas
are discarded (attended from), relevant ideas are attended to.
That which does the controlling is often also in the form of a
definite idea. This is best illustrated when one is given
two digits written one above the other: if told to add them,
under the influence of this guiding idea they call forth their
sum; if told to subtract one from the other, under the control
of this guiding idea they suggest their difference. The same
stimuli give rise to either one or another idea according to
the nature of an additional and controlling idea.
Similarly, we may have occasion to think of the opposite
of a given concept, or of a subordinate, or the superordinate
or the symbol of a given concept. In any case one idea calls
up a second under the guidance of a third.
We may not only occupy our minds with ideas of the
color, size, shape, etc., of objects, as referred to above, but
we may compare two objects as to color, size, shape, weight,
motion, acceleration, symmetry, etc., and judge which is
the best suited to our needs. Of two individuals we may
compare the good looks, cordiality, sincerity, hospitality,
integrity, adaptability, intelligence, etc., as conceived in
ideas of conduct, feelings, appearance, facial expression, and
of the many circumstances under which the impressions were
gained. All these mental activities dealing with ideas as
material — their association, recall, generalization, abstrac-
tion, comparison, judgment, etc., are elements in adaptation?
yet they may be experienced or accomplished quite inde-
pendently of words. The idea of a color is not a word. The
idea of one color being more intense than another need not
have anything to do with language. The choice of this or
that color for an aesthetic purpose does not require language,
nor does the act which the choice calls forth. Yet all this
is adaptation.
4. Words may be the Material of Thought. — As has been
suggested throughout the discussion, words may be the
material of thought. The place of words in the range of
material of thought may be stated as follows. The material
DO WE THINK IN WORDS? 407
of thought, as explained below, begins with perceptions;
then come images resembling perceptions, then more and
more attenuated images or aspects of images singled out by
abstraction, and finally symbols. By symbol is meant any
concept which is used in place of another. The best illustra-
tion of thinking in symbols is in the case of the number symbol
system used in arithmetic and algebra. The idea of eight
(not in the word but the number: ********) [s represented
by the symbol: 8. The idea of seven (seven things: *******)
is symbolized by the figure: 7. Now if we have the problem
of finding the sum of these numbers (******** and *******)
we may do so by translating them into their respective
symbols and give our attention to the symbols only. Having
previously formed an association between the symbols, 7
and 8, and the symbol of their sum: 15, the symbol 15 is
called up when the symbols, 7 and 8, and the guiding idea
of summation are in mind. We may then proceed to make
other arithmetical computations in terms of number symbols
only, letting these call up the number idea (**** — ) when
needed. Similarly in algebra we may let x represent one
number with which the problem deals and let y represent
another number, etc., and then by means of habits established
in connection with these symbols we may do thinking of a
simple type in lieu of what would be far more difficult if done
with the original concepts of number. This type of thinking
is exemplified in the following algebraic reasoning:
If xz — yz = 2 then (x — y)(x + y) = z.
There are of course many kinds of symbols. In addition
to the number symbols just mentioned there are the symbols
of operation upon numbers such as those of addition, multi-
plication, integration, involution, etc., there are the symbols
of musical notation, symbols of punctuation (?, !, ", *, -),
symbols on maps representing roads, trees, buildings, bridges,
tunnels, etc. (an engineer can think very effectively in these
symbols). There are even symbolic facial expressions used
by actors to portray emotions which off the stage would not
be expected to produce those expressions. A skull and
cross bones symbolizes danger. The flag symbolizes country,
408 ARTHUR S. OTIS
etc. Last, and most important, of course, words and sen-
tences symbolize thought of every description. Occasionally
we feel that we have experienced some thought or sensation
or feeling which cannot be expressed in words. But in
general all ordinary thoughts and feelings can be represented
by some word or sentence.
We see therefore that language constitutes only one of the
various kinds of symbolization, and symbols constitute only one
type of material of thought.
Words are themselves the material of thought under many
circumstances. Whenever we have to communicate thoughts
to another or learn the thoughts of another through language
we have to deal with words. By far the greatest use of
language of course is in the calling up of language symbols
to represent meanings or the calling up of meanings repre-
sented by language. Occasionally however we may think
in terms of language almost exclusively, as when dealing with
the rhyme and rhythm of poetry. In the case of syllogistic
reasoning we may be truly said to be thinking in words, when
the expression "All A is B and all B is C" calls up the language
idea: "All A is C," or when part of a sentence suggests the
rest as "All is not gold that ."
5. Language ike Symbolization of Meaning. — We have
attempted throughout this discussion to distinguish clearly
between a meaning and the language by which it is symbol-
ized. We cite the following illustrations to bring out this
distinction still more clearly.
If one says : " I saw John Jones on the street this morning "
the hearer will get the meaning of the sentence at once.
"Getting the meaning" means to the ordinary person getting
an image, more or less faint perhaps, of the speaker looking
at Jones on the street. But let us take another sentence.
Here is one in which the meaning of a new (coined) word is
stated. Every word in the statement of the definition except
the new one is perfectly intelligible and familiar and the
statement is a perfectly logical and meaningful one, yet we
are confident that the reader will not get the meaning from
the language on first reading. This is the sentence: "Let
DO WE THINK IN WORDS? 4<>9
us define the word, incration, as meaning the increase in the
number of feet per second per second by which the motion
of a body is accelerated." Anyone who has gotten the
meaning of this sentence clearly should be able to point out
immediately the error in the following statement, which if
correct would follow as a corollary to the above definition:
"The unit of incration is one foot per second per second."
(The correction is indicated in a footnote.) If the reader is
unable to point out the error it is merely because he has not
gotten the meaning of the definition, which is something quite
apart from the words by which it is symbolized and consists
of images either of the motion of a body or of the path of its
motion. Without such images we are confident the meaning
can in no way be appreciated.1
As has been said, the utterance of sentences or of parts
of sentences or of analogous statements often helps to bring
out the meaning, that is, helps to call up the imagery neces-
sary to build up the meaning, or helps to fix the meaning in
mind by symbolizing it after it is appreciated. But the
meaning may exist entirely independent and apart from
any utterance, either overt or implicit.
One often hears the expression from pupils in school:
"I know but I can't tell." This is generally a simple case
of having a meaning or idea without the ability to symbolize
it in language.
Moreover, as has been stated, even an adult may have
experienced perceptions, ideas, or feelings which he will
declare cannot be expressed in words. Even if they could,
1 The correct statement is: The unit of incration is one foot per second per second
per second.
A reasonable comprehension of the meanings of these statements may be built
up with the help of the following leading statements. The rate of motion of a body
is the number of feet per second which it moves. The unit of rate is one foot per
second. The acceleration of a moving body is the increase in its rate, that is, the
increase in the number of feet per second which it moves in succeeding seconds. The
unit of acceleration is one foot per second per second, that is, one foot per second every
second. And again, the incration of a moving body is the increase in its acceleration,
that is, the increase (from second to second) in the number of feet per second by which
its rate is increased, or in other words, it is the number of feet per second per second
by which the motion of the body is accelerated. The unit of incration is the unit of
acceleration every second, that is, it is one foot per second per second per second.
410 ARTHUR S. OTIS
his inability to do so testifies to the independence of the
thoughts from the language by which it would be symbolized.
We say of our memory of a sunset that it was 'indescribable'
and 'would have to be seen to be appreciated,' which is
entirely true. We say, 'Words fail me' regarding the
expression of our thoughts of past emotional experiences.
The emotional experiences of love, hate, fear, anger, etc.,
may be the material of thought just as well as the experience
of perceiving the color blue. Even our best attempts to
express our thoughts of experiences in words often fail to
carry full meaning to the hearer unless he has had a similar
experience. Thus, one may say that an experience was like
that of a sudden drop in an elevator or like flying in an
aeroplane, but unless the hearer has had an analogous
sensation or experience the expression is devoid of essential
meaning to him. What can the expression, 'like being struck
by lightning' or 'like finding oneself caught under the ice'
mean to one who has not had the experience in comparison
to what it means to one who has had the experience! Yet
the language is identical in the two cases. One and the
same expression from the lips of an oratorical person may
convey a meaning to one hearer which will call forth tears, a
meaning to another hearer which will call forth anger, and a
meaning to still another hearer which will call forth laughter.
Such a phenomenon would be of course entirely impossible
if there were but one meaning to the expression, a meaning
inherent in the language itself. It is a platitude that the
meaning of language is something which is brought to it
from the experience of the hearer, that it does not reside in
the language.
6. The Genesis of Language. — Not only may we have
thoughts for which we cannot think of the existing appro-
priate language, but we may often have an idea for which
there is no corresponding word or phrase. Indeed language
is built up by the coining of new words and phrases that are
needed to symbolize new thoughts for which no corresponding
language exists. The word, automobile, for example, did
not come into existence until after there had been made,
DO WE THINK IN WORDS?
or at least conceived, a machine which would move itself,
and which needed a name. Similarly the expression, 'carry
on,' came into use in response to the need of a name for an
action which was well comprehended but for which no con-
venient language equivalent existed. Ideas originate first;
afterward they are named — symbolized in language.
Perhaps the clearest example of the temporal relation
between the genesis of ideas and their symbolization is the
case of the naming of persons. First the child, which we
perceive, is born. Afterward it is given a name. Why do
we give a child a name? It is, of course, for the reason that
to refer to it always by description would be cumbersome
and inaccurate. When we think of a person deliberately we
think of his form and features, his speech, manner, expression,
etc. When we think of him more fleetingly, however, our
imagery becomes attenuated, even perhaps to the extent
exemplified by the representation of Roosevelt by merely a
pair of glasses and a row of teeth. But for the purpose of
one person conveying the thought of an individual to another,
such attenuated imagery is inexpedient. We therefore
symbolize the whole picture or idea of the individual by a
single word (a name). The name then, in cases of rapid
thinking, may nearly take the place of the concept of the
individual as imaged. The name, however, still does carry
with it something of the original imagery. The idea of
'James' to one person carries with it something which char-
acterizes his brother, James. The idea, James, to another
person, carries with it something which characterizes his
uncle, James, a different individual. This additional some-
thing is needed to constitute the difference in meaning be-
tween 'James' for the one person and 'James' for the other.
To a third person who knows no one by the name of James,
the idea is merely a word, known to refer to some individual.
On the other hand the perception of a person whose name
is not known does not carry with it any idea of a name, nor
need any word come to the mind. When I see a man I do
not think 'man.' When I think of a crowd of persons I do
not think of a crowd of words! Language is as distinct from
412 ARTHUR S. OTIS
the ideas it represents as the name of a person is distinct
from the perception of the person himself.
An infant of course does not use language habits until a
year or so after birth. Yet an infant can think — perceive,
compare, judge, choose, decide, act upon decision, etc.—
before language habits begin. The material of his thoughts
is the sights, sounds, smell, tastes, feelings, etc., which he
experiences throughout his waking state.
To illustrate the genesis of language habits we must go
back to the early days of a child's life when it is just beginning
those 'Abbreviated and short-circuited actions (which) be-
come a necessity if it is to hold its own in that environment
and make progress' (p. 319). The child's perception of its
doll, its desire for the doll, its idea of searching for the doll
when not in sight, its idea of creeping toward the doll when
seen, its idea of reaching for the doll when within reach, its
idea of grasping and the new ideas which arise from the
manipulation of the doll — assuming these ideas sufficiently
well fixed by habit that the actions have reached an ' abbrevi-
ated and short-circuited' stage and are purposeful and
adaptive — these ideas constitute the beginnings of thought
(conscious adjustment to the environment).
The stimulus, 'tata' (p. 320), cannot call up the concept,
doll, before the concept, doll, is in existence. Nor can it
create the idea. The child must have some idea of the doll,
formed from perceptions of the doll itself, before the idea of
'tata' can be associated with it. Language in general
bears just the same relation to thought in general that the
idea, 'tata,' as a word, bears to the idea of the doll, as some-
thing seen, touched, etc.
Similarly the idea of number is formed before any symbol
representing number can be associated with it. We may
talk about number in the hearing of a child for months after
it has begun the use of language, but until the child has by
observation and comparison become conscious of the two-
ness of its hands, of the two-ness of its feet, and of the two-ness
of many other separate things, so that the idea of two-ness as
an abstraction becomes a separate idea in the child's mind —
DO WE THINK IN WORDS? 4*3
until this time the sound of the word 'two* is meaningless
to the child, and only attains meaning when finally associated
with this abstract idea of the two-ness of any two things.
The meaning comes first; afterward the language symbol
(word) is associated with the meaning and may be substi-
tuted for it when occasion demands. One has but to attempt
to teach a child, who is just learning to talk, to count and
deal with number concepts to see how absolutely meaningless
the words are to the child until he has had opportunity in the
course of his daily experiences to make the abstractions neces-
sary to form these ideas. We may teach a three-year child
to pronounce perfectly the sentence: "The square root of
twenty-five is five," and if the action of language mechanisms
constituted thought we should expect the child to understand
perfectly the meaning of what he had said! Further com-
ment seems unnecessary.
It will be noted that in this discussion it has been neces-
sary to use concepts which are not found in Behaviorist
psychology. These are the concepts of 'meaning,' 'idea/
'concept,' 'conscious,' 'purposeful,' 'association of ideas,'
'abstraction,' 'symbolization,' etc. Yet these are funda-
mental to structural psychology and from the above dis-
cussion we deem it apparent that a consideration of the
acquisition of language habits, their function in thinking,
and the material of truly non-language thought, is totally
inadequate without these concepts. To be sure we find
passages in a Behaviorist psychology attempting to state
what goes on in the mind of an individual. Thus (p. 305) :
"We manipulate vocally" (in attempting to recall the
name of a familiar person) "by running over the names
beginning with each succeeding letter of the alphabet, or by
saying 'black hair,' 'blue eyes,' 'six feet tall,' and the like."
This seems to the writer to be one of several excursions quite
outside the realm of Behaviorist psychology. He does not
know whence these ideas came but judges that it was by
some sort of inference, partly because the Behaviorist does
not use introspection and partly because he is unable to
corroborate them by introspection.
414 ARTHUR S. OTIS
7. The Inadequacy of the Behaviorist Conception of Thought.
— To illustrate what is believed to be the wholly inadequate
conception of thought as entertained by the Behaviorist,
we cite the soliloquy postulated on page 332. "The implicit
word processes (aroused by whatever previous stimulus)
'it's a fine day, I think I will go to the races; it's twelve
o'clock now, I have just time to catch the train,' serve to start
you to get your hat and field glasses. Some unfinished
work meets your eye or other conflicting word processes are
aroused, as 'but I have to write those letters and I have a
luncheon engagement with X.' These tend to drive the
organism as a whole into some other form of action; for a
time there is a conflict (inhibition). Finally when the
conflict is over the final word act issues, 'Well, I guess I'll
have to give up the races and write those letters and keep
my engagement with X.' Here we see implicit word pro-
cesses tending to arouse overt acts and actually arousing
the initial steps. But since the human individual is a com-
pletely integrated affair, associated word processes arise
which may drive the organism into a totally different form of
activity from that which was first initiated."
The writer contends that the 'previous stimulus' together
with the mental activity which called forth this soliloquy
would be sufficient to start one to get his hat and field glasses,
without the accompaniment of any action of language
mechanisms, and that such action itself would not suffice.
The reasoning is as follows. Let us suppose the previous
stimulus to be the perception of the green grass and sun-
shine and warmth of the outdoors. This perception called
forth by association the memory of previous days when
races were attended and of the accompanying pleasure.
These memories contained the urge to renew the pleasures.
They gave rise to the decision which is expressed in the
language: "It's a fine day, I think I will go to the races."
The decision made, the thought took form in vocal expression.
At this point either the clock struck twelve, this serving as a
stimulus, or the idea of going to the races naturally called
up the idea of when to go, which in turn suggested the
DO WE THINK IN WORDS? 415
idea of looking at the clock, resulting in the perception that
it was just twelve o'clock. None of this mental activity
required language. The perception of the time of day
having been made in one way or another, the idea called
up the words which would express it and the individual added,
'It's twelve o'clock now.' What happened next? Presum-
ably at this point came the idea of going to the races by
train, followed immediately by the idea which if expressed
in words would be, 'When does the train go?' which in turn
called forth the memory that the train goes (let us say) at
twelve-fifteen. Immediately there came to the mind the
idea of the preparation which is necessary to catch the
train and a judgment is made as to how long this will take,
based upon past experience. The individual must also go
through a certain mental operation of determining how much
time there is available before train time and make a com-
parison between these lengths of time in order to make the
decision which when expressed in words is, 'I have just time
to catch the train.' This idea possibly suggests the idea of
haste which together with the idea of going to the races calls
forth ideas of the appropriate preparation, getting the hat
and field glasses, etc. These latter ideas take form in action.
In view of the obvious necessity for the mental activity of
perception, judgment, decision, etc., intervening between
the advents of the ideas which took form in the language
quoted, we submit that, as stated above, it is impossible
that the soliloquy postulated could of itself have given
rise to the getting of the hat and field glasses. Moreover,
the ideas themselves which suggested the soliloquy could
have given rise to the acts and there need have been no
language, explicit or implicit, involved whatever. Thus,
the ideas of time may have been conceived in visual imagery —
the imagery of the face of the clock and the movement or
path of the minute hand. No language is required. The
ideas of preparation for the train would consist of memory
of the acts of getting hat and field glasses, walking or riding
to the station, buying the ticket, etc., these consisting chiefly
of visual and kinaesthetic images. No language is necessary.
416 ARTHUR S. OTIS
The ideas involved in the judgment of distance (to the
station) or of time required for preparation and traversing
the distances, etc., would be kinsesthetic or visual or other
ideas of space and motion, the comparison of these ideas of
space and motion, etc., resulting in ideas of the relations
between them. No language is required. Moreover, the
ideas which take form in the acts of getting the hat and field
glasses are visual, tactual, and kinaesthetic and are quite
independent of language. The fact is, one could conceivably
note the weather, decide to go to the races, make preparations,
board the train, hand the conductor a ticket, note the progress
of the train, get off at the race track, pay the entrance fee,
and watch the races, all with mental activity and acts in
no way involving language. Any amount of soliloquy or
conversation may accompany the expedition, but this is
wholly incidental, secondary, and unessential.
8. Introspection. — We believe that Behaviorist psychology
is entirely sound within its own sphere, that is, so long as it
confines its study to the behavior of the individual as seen
from without. A psychology so limited, will, of course,
necessarily leave untouched a vast field of useful knowledge
which can in time be made scientific where not already so,
after extensive investigation, comparison of findings, deter-
mination of general tendencies, and the careful observation
of everyday experiences. But should one desire to explore
the realms of psychology outside the scope of Behaviorism,
he must then supplement his external observation by as
thoroughgoing, extensive, and careful an examination of that
which takes place within the mind — as seen from within —
as is possible by highly practiced and trained introspection.
To direct -the attention to the color of an object is a
very easy matter. To direct the attention to the difference
in shade between two colors may be slightly less easy but it
is entirely possible. To direct the attention to the idea
of the aesthetic value of the colors requires perhaps appre-
ciably more practice, but it is none the less possible. How-
ever, to direct the attention to the nature of the mental
process of choosing between two colors, and to the manner
DO WE THINK IN WORDS? 417
in which the choice gives rise to appropriate acts, may be
quite difficult, not to say impossible, for the inexperienced
person. Yet these phenomena are available for observation
no less truly than the habit of typewriting is open to acquisi-
tion or the length of a rail is capable of being measured to
the thousandth of an inch. These accomplishments require
long practice or minute observation, but we do not say they
are impossible. An unpracticed person cannot direct his
attention to the less tangible aspects of thought any more
than he can play a theme on the piano. Because it is difficult
however, one does not forego the learning of piano playing,
if he desires to learn to play. Again, it is possible that no
two observers might obtain the same measurement of a
rail to the thousandth of an inch. Nevertheless we do not
say that measurement is of no use in physics. Observers
can nearly all agree on the length of a rail to the tenth of an
inch, and on the length of a needle to the hundredth of an
inch not to say to the thousandth.
Similarly in psychology, those inexperienced in intro-
spection may not be able to distinguish between middle C
on the piano and the C an octave above, or to observe that
they see objects double which are not focused upon. More-
over, persons highly trained in introspection may not always
be able to distinguish between the perception of a very faint
sound (as the distant ticking of a watch) from the auditory
image of the sound (imagined sound) nor to state just what
constitutes the mental element of difference between the
emotions of fear and anger. But there should be little differ-
ence of opinion between persons of extended experience in
introspection as to whether the material of our thoughts,
when we create new ideas, and conceive new modes of activity
in the fields of music, art, drama, mechanics, etc., is in the
form of language or in the form of tones, visual pictures, etc.
9. Summary. — There may be no experimental proof
whether or not thinking — conscious adjustment to the
environment — is invariably accompanied by the actuation
of some language mechanism as the larynx, lips, fingers, etc.,
in the incipient production of some form of language, spoken
418 ARTHUR S. OTIS
or written, but the evidence would seem to favor the belief
that no such invariable accompaniment is necessary. One
uses the eyes in observing objects attended to almost through-
out the waking state. It would seem more plausible to
assert therefore that some form of "implicit" eye movement
is a necessary accompaniment of all thinking. Much evi-
dence such as that from the observation of a chess player
studying his moves could be brought forth in support of
this view. All this, however, is quite beside the point.
The claim is made by the Behaviorist that "thought is the
action of language mechanisms" (italics mine). Certainly
the evidence against such an assertion is overwhelming.
Man is an organism highly adapted physiologically to his
environment, provided with sense organs of sight, hearing,
taste, smell, touch, pain, heat, cold, muscle movement,
body position, etc. Each of these sense organs is capable
of giving rise to sensations which take the form, in the
mind, of images or image patterns. (The word image is
used in a very broad sense as shown below.) The organism
has at its disposal any or all of these incoming percepts or
stored images or image patterns as material for thought, for
working over into new combinations, new thoughts, which
will give rise to new actions, new adaptations to the environ-
ment. In the event of the bringing together of two or more
concepts — images — or of the dividing of one concept into
two or more (as when a child first separates from the concept
ball the concept roundness) — in the event of this working
over of concepts, they are necessarily abbreviated, composited,
exemplified, attenuated, or substituted for by others. If the
substituted concepts are of a kind remote from the kind
for which they are substituted but are more or less definite
and commonly understood, we call them symbols. A careful
description of the manner in which thought material is
abbreviated, composited, attenuated (even to a point which
is considered by some psychologists to be 'imageless'), etc.,
is of course quite impossible within the limits of this article.
However, the mental activity which brings forth a new
act may be the result of the combination or division or other
DO WE THINK IN WORDS? 419
working over of any type of mental material — the bare sensa-
tion, the fresh vivid full percept, the fairly vivid memory
image, or the image or image pattern when abbreviated, or
attenuated, or composited, or exemplified, or in any manner
generalized or particularized, or finally in the form of symbols.
And language, as has been shown, is but one general type of
symbol system.
In conclusion, then, let it be said that we may think in
words, and when we do, the thinking may be accompanied
by the action of language mechanisms. But thought —
even conscious mental adjustment — is not restricted to the
material of language any more than it is restricted to the
material of musical tones or of architectural designs or of
facial expressions, nor is it restricted to the action of language
mechanisms any more than it is to the mechanism of hearing
or of sight or of locomotion.
A BEHAVIORISTIC ACCOUNT OF SLEEP
BY CHARLES H. WOOLBERT
Division of Public Speaking, University of Illinois
The phenomenon of sleep does not lend itself conveniently
to explanation in terms of sensation, image, and feeling.
Accordingly structural psychology can offer little by way of a
description of the state of sleep. The behaviorist, on the
other hand, believing that his definition of consciousness
offers a statement not only of what consciousness is, but of
what it is not, is in a position to explain sleep. If, as be-
haviorism asserts, mind is a matter of reflex connections always
involving the movement, tension, or tonicity of muscles and
always correlated with the activity of glands, then conscious-
ness, in its varying degrees of clearness, is a matter of degrees
of complexity and ordination among systems of muscular
action. A high degree of consciousness thus becomes synony-
mous with an intricate and ordered complexity of tonicity in
muscular systems; while a low degree of consciousness is
equivalent to a complexity of low degree and an ordering of
simple texture. Consciousness thus may be, speaking in a
paradox, scattered, involving perchance abundant activity —
by way of muscular tonus in various muscle systems — but
activity of a low degree of intricacy in organization.
If consciousness, then, be a matter of the degree of com-
plexity of interacting muscle systems, non-consciousness is a
lack of activity or else a lack of this complexity. In either
case the factor of complexity is vital and needs describing.
Behaviorism's explanation of this concept is based on the
continuative function of the sense endings within the muscles.
Always stimulated by any muscular event, they afford the
means for causing a single inaugurating stimulation to rever-
berate through a long series of tensions, or else to provide a
continued hardening of some one set of muscles. The type
of tensions running in a series is what is called the chain
420
A BEUAFIORISTIC ACCOUNT OF SLEEP
reflex; the continued is that called circular. Chain reflexes
have much to do with that complexity of interacting muscle
systems that makes consciousness. Carried on by means of
proprioceptor organs of stimulation, they have a right of way
and a special kind of clearance. This insures them continuity
and lends a measurable degree of stability to the complexity
of structure that makes up consciousness.
This description recognizes, first, that the muscularity of
the body occurs in fairly well-defined systems: as those of
the back, the legs, the head, the face, and the throat; and,
secondly, that these muscle systems are set off one by the
other and in a certain order. This order is based on priority.
Priority of muscle systems is, in general, a matter of pre-
cedence in the development of working efficiency in reflex
arcs. Those systems that have an early development history
come to have a pronounced control over systems developed
later, in that the systems developed later get their initial
determinations from the workings of the habitual responses
of the earlier. Their most intense determinations are, both
earlier and later, then, conditioned by the determinations of
the systems already habituated. Their capacity for quick
and valuable response depends thus very largely upon their
close coordination and cooperation with habitual reactions
of systems determined at an earlier stage of development.
An understanding of sleep requires a description of this
development order, a description of gross muscle systems
and a statement of their superordination and subordination.
This is found in an account of the operation of Pawlow's
Law, as manifested in the conditioned, chain, and circular
reflexes.
By Pawlow's Law a reflex arc may be a factor in new
activities by the stimulation of sense endings imbedded in the
muscles that are contracted by the operation of such an arc.
Every motor process stirs a muscle; this stirring starts new
impulses; and these impulses seek a new outlet. Two direc-
tions they can take: they can go around back by way of the
motor nerve that stimulates this muscle, and so stimulate it
again; or they can take some other motor nerve leading to a
422 CHARLES H. WOOLBERT
quite different muscle. The former of these processes is the
circular reflex, the latter, when carried on through a seriesj
the chain reflex. By means of circular and chain reflexes in
various combinations the organism has within itself the
machinery for carrying on activity, for a while, at least,
without the intervention of peripheral stimulations. Only in
this way can the continued activity of involuntary organic
acts be accounted for, activities like heart-beat and breathing.
In this way also is given a satisfactory account of the mechan-
ism of such activities as catatonia, catalepsy, emotional com-
plexes, fixed ideation — in fact, repetitive and continuative
actions of all kinds.
To get the full significance of these circular and chain
reflex systems in the coordination and superordination of
actions, it is necessary to envisage them in connection with
the development order of muscular systems. Systems de-
veloped early in the life of the organism are obviously deter-
mined strongly; in the case of the very earliest, heart-beat
and breathing, no stimulus short of that adequate to stop
life can divert them. Moreover, their continued activity is
but little dependent upon other stimulation than that pro-
vided by the circular reflexes that keep them regular; at
least so long as the muscles concerned receive nourishment
from the blood. Systems developed one stage later, like
those involving walking, reaching, turning the head, have
within themselves much of the same continuative mechanism.
Their ability to function, however, they have gained largely
as an adjustment from the successful functioning of systems
developed earlier. At the start of their functioning they do
not possess a full measure of self-determination; they are of
necessity dependencies, subject to the caprice of superiors
holding power by a rule of seniority. Certain things they
can do so long as the older systems go about their business
in an orderly fashion; certain other things, under the same
conditions, they are not privileged to attempt. Thus the
use of the eyes, the hands, the legs, is conditioned very
materially by the regularity of the beating of the heart and
of breathing. Let, once, something go wrong with either of
A BEHAyiORISTIC ACCOUNT OF SLEEP
these older activities, and the organism loses complete control,
eventually, of hands, legs, eyes, and of all other muscle
systems. Under similar circumstances any determinations
that are ordinarily well established give way to a recrudes-
cence of the wildest of random movement.
Thus the muscle systems operate in a kind of hierarchy,
with jurisdictions fairly distinct, though not exclusive. Most
firmly enthroned of all are the primary reflex systems con-
trolling heart-beat, flow of blood, operation of vital organs,
and breathing. Next come those developed in the organism's
earlier days, use of arms and legs, back, torso, and neck
muscles; later, and probably overlapping the earlier systems,
muscles of the eyes, ears, face, and head; lastly — coincident
with the development of speech — the muscles of jaws, lips,
tongue, and throat. Thus consciousness as complexity of
muscle systems is a pyramid with the organic systems at the
base and the muscles of thinking, reasoning, and speech at the
top. Or, changing the figure, it is a hierarchy with the
organic systems as autocrats and the other systems holding
office on a descending scale of self-government, dependent
always upon the commands of the autocrats ruling by virtue
of prior possession of power.
This hierarchy operates to provide the difference between
sleep and waking consciousness. Without the tension of head
and face systems there is not complexity enough for conscious-
ness. So vital are they to clear cognition that they are
easily confused with the totality of consciousness; remove
them altogether from the systems active at any one time, and
unconsciousness occurs. Yet they are not autonomous; when
fatigued and free from intense peripheral stimulation they
normally yield easily to the relaxing of the lower systems and
go out of function along with them. Any condition in which
they refuse to stop functioning when free from peripheral
stimulation or when fatigued, and when the lower systems
have relaxed, is looked upon as abnormal. In fact psycho-
pathic conditions can be described either in terms of an
actual lack of the upper systems, or in terms of their failure to
cooperate with the activities of the lower. Sleep is accounted
424 CHARLES H. WOOLBERT
for in the formula : Remove the higher systems from activity,
and consciousness departs altogether; weaken the lower, and
consciousness is in a precarious condition, especially if the
higher systems are affected by fatigue. When the lower
systems are thrown out of function, the higher circular reflexes
either stop at once or, in abnormal cases, ultimately wear
themselves out; in either case consciousness breaks up.
Remove the lower entirely, and death is instantaneous.
A prime requisite of easy and deep sleep is freedom from
stimulation for the eye, ear, nose, tongue, and parts of the
skin not constantly pressed by clothes; forms of stimulation
that have little to do with the organic systems. Yet sleep is
possible even in the face of such stimulations; but only in
cases where great fatigue throws lower systems, like those of
leg muscles, back, and neck, out of commission. The chief
power of estopping other systems, especially under conditions
when fatigue is present, is authoritatively appointed to the
organic systems; because they get their determination at a
time when the organism is in its most plastic state — in its
early stages.
Thus sleep becomes behavioristically a matter of the
efficient domination of the upper systems by the lower,
operating through the relaxing power of fatigue; while wake-
fulness and insomnia always imply that the higher and later
systems are assuming dominance over their precursors.
Wakefulness, so, is characteristically the dominance of the
lower systems by the upper when fatigue is not present.
Accordingly when wakefulness exists at the same time that
fatigue is present, the condition is abnormal.
This means that when the muscles of the back, legs, and
neck are relaxed, a powerful stimulator is lost to the muscles
of the arms, hands, feet, and head. When arms, hands, feet,
and head muscles in turn are relaxed, there is lost a powerful
source of stimulation to the muscles of the face, jaw, tongue,
and throat. While systems are undoubtedly more finely
differentiated than this, still their hierarchical interdependence
is on just such an order — and sleep can be explained by gross
characterizations as well as by those more minute. In the
A BEHAFIORISTIC ACCOUNT OF SLEEP
inducing of sleep much significance must be attached to the
order in which muscle systems go out of function. When the
organism follows the development sequence, sleep is easy;
when the order of relaxation is in any way reversed, restless-
ness and wakefulness follow. In cases of complete reversal
of the order, we get such states as hypnosis, temporary high
degress of attention, manic conditions, and forms of insanity.
The beginning of sleep then normally is the relaxation of
the muscles that hold the body erect. As soon as these
muscles are relaxed, the prime determiner of higher systems
is taken away, the proprio-ceptor foundation, and the higher
systems then are kept in function by only a veritable bom-
bardment from the outside world or from very strongly deter-
mined circular or chain reflex arcs within their own system.
Among these latter are emotional states, fixed ideas, tunes
running through the head, repeated attempts to solve a
problem, rhythmical verbalizing, and thinking in circles.
The next step in normal sleep is the sequential relaxation
of each of the systems hierarchically dependent upon the
erect-holding systems. Finally through the sufficient dissolu-
tion of the complexity that makes consciousness, sleep comes.
Consequently once a person lies down, relaxing the muscles
of legs, back, and neck, the beginning is made of sleep.
Providing there is no interference from outside stimulations —
chiefly those acting upon the sense endings in the head — and
also so long as there is no intense circular reflex process
going on in the muscles of the jaw, lips, tongue, or throat, —
'thoughts that will not go out of one's head' — such a be-
ginning once made leads to complete sleep and loss of con-
sciousness.
Certain easily-made empirical observations as to sleep
confirm this account, (i) Sleep is characteristically accom-
panied by relaxation of muscles. (2) Characteristically also
it takes place with the body in a horizontal position, a position
that induces first of all a relaxation of the muscles of the legs,
back, and neck, all of which must maintain a high degree of
tonicity to maintain erect posture. (3) When the muscles
of the back, legs, and neck are relaxed, the muscles of the
426 CHARLES H. WOOLBERT
head, face, jaw, and throat all tend to relax in a short time
ensuing. (4) These muscles also are most easily relaxed
when freed from stimulation of the head sense endings, in the
dark and in silence and free from intense taste or smell. (5)
All these relaxations occur parallel with a scattering of con-
sciousness, a defocalizing of attention; and the greater the
degree of relaxation, the less the ability to perform any act
implying a high degree of concentration. (6) Organic activi-
ties, though, are kept up despite any relaxation of the volun-
tary muscle systems. (7) Again, deep sleep implies complete
relaxation; also it connotes rest and recuperation from
fatigue, a retoning of muscles for future work. (8) The
degree of sleep involved conditions the number and vividness
of dreams had; deep sleep implying few dreams and light;
light sleep implying many dreams or dreams that are vivid.
(9) Deep sleep also leaves little recollection of dreams, except
for the moment when one is coming out of sleep to conscious-
ness. (10) Great numbers of dreams, or dreams that are
vivid, it is generally assumed, are equivalent to defective rest,
and restlessness is always accompanied by the inability to
stop thinking or by numerous and intense dreams, (n)
During widespread muscular activity there is no such thing
as sleep; as during walking, eating, reading, talking, giving
active attention in any way.
These general observations, and many others of similar
nature, point clearly to the close relation existing between
sleep and the movement or tonicity of muscles.
In the subjective terms of ideation, sensation, and feeling
it is difficult to explain what happens to conceptual thought
during sleep. The behaviorist, by assuming that thought
of all kinds and in all degrees is a matter of muscular tonus,
precisely as in walking or standing erect or moving the hands
or talking, can give an account of sleep that fits in with his
whole program. Sleep to him is nothing but a disorganiza-
tion of muscle systems which in waking consciousness are
closely interrelated hierarchically, the action of each deter-
mined in part by the continued action of the others. When
conditions are set for the relaxation of sundry systems of
A BEUAyiORlSTIC ACCOUNT OF SLEEP 427
muscles, consciousness begins to be more scattered, system
after system drops out of function, and ultimately, in the
soundest sleep, nothing is left by way of muscular activity
but the functioning of the organic systems.
Dreams are clearly the result of systems involving throat,
face, tongue, and lip muscles which remain in function when
other systems have been thrown out of gear, systems which, if
combined, would make consciousness. The Freudian dream
psychology presents agreement with obvious facts in that it
recognizes the existence during sleep of mental processes
which seem very like others that go on in waking conscious-
ness, yet which at the same time are partly unlike them.
The behavioristic explanation of this is that in so far as a
dream is a matter of the activity of muscle systems involving
a high degree of complexity and coordination, in so far it is
similar to waking consciousness; and so has a way of seeming
logically coherent. The illogical dream, on the other hand,
the freakish dream, the dream that seems to forbid explana-
tion and interpretation, can be explained broadly as a type
of organization and coordination not met with in that form
in waking life; so that, speaking generally, the more unusual
the combination left operative during sleep, the more fan-
tastic the dream. From these suppositions can also be found
the reason why dream analysis cannot be a matter of accurate
interpretation and why the Freudians who assume to inter-
pret all dreams give promises in reality beyond powers of
observation to fulfill. So entirely beyond inspection and
prediction can be the permutations and combinations of the
hierarchy of muscle systems, that they can defy all powers
of analysis.
From these observations follow certain therapeutic in-
ferences worthy of note, most of them current already through
the experience of the race. If you would sleep soundly, exer-
cise much, in particular the muscle systems of the body below
the head; for if the 'body' is tired, the 'mind' will rest also.
If restless in sleep, study how to relax, first of all the muscles
of the legs, back, and neck. Then reduce the breathing rate:
high tension almost invariably is accompanied by rapid breath-
428 CHARLES H. WOOLBERT
ing; low tension by slow-breathing. Also hands and feet,
fingers and toes, must be inert. If thoughts crowd thick
and fast and will not leave, let the jaw drop, make the
muscles of the cheeks and lips flabby, avoid screwing up
the muscles around the forehead and the eyes, see that the
tongue lies limp in the mouth, and make certain that the
muscles of the throat are not in any way tensed. These
last-named muscles, together with those of the jaw, tongue,
and lips, are more likely than any other to get in the way of
sound sleep. Make sure to observe the right order of relaxa-
tion of systems; gross lower systems first, then the finer sys-
tems below the head, and finally the fine systems of the head.
Sleep is synonymous with carrying out the following order
in relaxation: Reduce the breathing rate; then relax legs,
back, abdomen, and neck; then arms, hands, fingers, and
toes; next the muscles around the eyes, forehead, scalp, and
ears; and finally those around the mouth, jaws, tongue, and
throat — the muscles of speech and conceptual thought.
THE COMPENSATORY FUNCTION OF
MAKE-BELIEVE PLAY
BY EDWARD S. ROBINSON
The University of Chicago
It is the purpose of this paper to present in outline a view
of play which will usefully supplement those theories which
are generally entertained. Play is, of course, a phenomenon
of extreme complexity and, for that reason, its complete
explanation should hardly be looked for in any single state-
ment. The suggestions have been made that the play of
children is a chronicle of race activities, that it gives practice
or preparation in functions of coming importance, that it
furnishes an outlet for surplus energy, and that it is a recrea-
tional agency or means of relief from fatigue induced by other
occupations.1 Undoubtedly all of these things are true to
some extent, but more important than any of them is the fact
that play is essentially a compensatory mechanism having
the same origin and impetus as the day-dream or fantasy.2
A compensatory function is especially evident in that type
of play which involves the element of conscious shamming or
make-believe. While it is possible to demonstrate that other
types of play may operate as compensation,3 1 shall confine the
following discussion to play which is clearly make-believe
in character.
The child is driven by many inherited and acquired im-
pulses, some of which are adequately and easily expressed
and some of which find no direct outlet. These latter create
a situation demanding compensation, and this compensation
1 For one of the latest discussions of the various theories of play see Reaney, M. J.,
'The Psychology of the Organized Group Game,' Brit. J. of Psychol., Monog. Supple-
ment, 1916, IV.
* The compensatory nature of the day-dream or fantasy is clearly brought out by
Freud and other writers of the psychoanalytic school.
1 Dr. Reaney, for example, holds that organized group games may have a com-
pensatory function. Op. cit.
429
430 EDWARD S. ROBINSON
is as a rule secured through make-believe activities. Most
common among such activities are play and fantasy. A child
would fight, hunt, and make a home as particular stimuli
arouse him. He is seldom in such an environment, however,
and he is practically never so organized by inheritance or
training that these undertakings can be fully carried out.
There are inexhaustible inhibitors around him and within
him which check free expression. And so he plays at, or
has day-dreams of, fighting, hunting, and home-making. I
have no desire at this time to say which of the unsatisfied
impulses of childhood are inherited and which acquired; but,
however they arise, we find that they are many and urgent,
and consequently that every normal child must find com-
pensation for their inhibition.
There are a number of factors which may act as inhibitors
of the behavior tendencies of children. These may con-
veniently be divided into the extra-organic and the intra-
organic, according to whether they are in the nature of en-
vironmental interferences or interferences which arise out of
the child's own organism.
During his development the child is constantly running
into extra-organic or environmental facts which are incom-
patible with the satisfaction of his desires. He may want to
hunt. Perhaps the family cat supplies him with a stimulus
to make this impulse felt. But this hunting impulse has
become a particularized affair. Hunting is shooting, and
he can not shoot because he has no gun. Instead of ignoring
a stimulus to which he can not react adequately, he points a
stick at the cat and shouts 'Boom!' He may then, and
perhaps to his sorrow, try to drag in his 'dead' game by the
hind legs. But the main and incontestable point is that the
child is compensating, by means of his pretensions, for the
inadequacies of the situation. He would like only too well
to shoot a real gun and drag in game which is really dead,
but his environment does not supply the appropriate circum-
stances. And so he plays.
Among the more important extra-organic factors which
limit the child's expression are the people around him. Just
COMPENSATORY FUNCTION OF MAKE-EELIEVE PLAY 43*
as he discovers the splendid interior of his father's watch,
someone takes the watch away from him. Just as he dis-
covers the importance of certain corners of the pantry, some-
one carries him away to another room. Everywhere there
are people and they are constantly interfering with his
behavior.
As I have intimated, it is not only the lack of a physical
world fitting in with every whim which causes the child to
play rather than to act in earnest. He has also his intra-
organic interferences arising out of his own complex little
nature. For the pure joy of it he would, at times, like to
bring down a stout club upon the head of his playmate —
that is, he would like to do this if it were not for the dis-
concerting facts that he would not like to hear his playmate
cry in pain, and that he would not like to feel the blows of his
playmate's revenge. And so the two boys will play at
fighting. Often, too, a child is hindered from acting as he
would because of a realization of the smallness of his body
and the slightness of his muscular strength. In such cases
we are apt to have a mimicry of feats of strength and daring.
It is evident that there are instances of make-believe play
and fantasy which apparently, at least, are not primarily
compensatory. A child may straddle his hobby-horse, not
because it is the best substitute for a real horse he would ride,
but simply because he has been taught to do so by his parents.
There is little doubt, however, but that the average child
enjoys his playing the more where he perceives its symbolic
relationship to a more serious pursuit. The fact that chil-
dren's play is given much of its specific form by adults, does
not, in the last analysis, indicate that it is therefore less
compensatory. By custom and tradition we initiate various
make-believe performances for children, but something in
the nature of childhood must explain why children take to
the make-believe with such enthusiasm. When we first teach
a child to ride a hobby-horse he may be unaware of any
connection between this activity and the actualities of horse-
back riding. But as he learns about real horses and real
riding, his play will become more and more clearly com-
432 EDWARD S. ROBINSON
pensatory in function. In other words, the rise of certain
impulses in children is so inevitable that their compensatory
expression may be provided for by the customs of the race.
In the case of any one child a compensatory activity may be
set up before the need for that particular compensation arises,
but we may still consider the activity a typical product of
child life and its characteristically incomplete adjustment.
Just as in certain individual cases a compensatory make-
believe may arise before the need for that particular com-
pensation, so specific habits of play and fantasy may be
retained after the apparent need for their compensatory
service is past. I know of successful men who find great
pleasure in day-dreams of achievements which they would not
care to have realized in any tangible fashion. In some of
these cases the day-dreams express real desires which are
denied direct expression because they run counter to other
desires of a more powerful sort. In other cases, however,
it is quite possible that day dreams which once had a com-
pensatory function now operate as old habits and are retained
because of their own repetition rather than because of any
important compensation which they still render.
Play and fantasy are frequently concerned with situations
more painful and disagreeable than any we should choose
to meet in real life. A natural question arises as to the sense
in which such make-believe can be considered compensatory.
Children do not want to be in railroad wrecks nor to receive
bullet wounds, and yet they enjoy pretending they are in
such straits. So it appears on first thought, but, as a matter
of fact, children do wish that just such things would happen
to them, providing they might happen without pain or other
ill consequences. In regard to railroad wrecks, if we could
read a child's impulsive nature completely, we should prob-
ably find that he wishes he could be in a wreck and, at the
same time, hopes he will not. He is in the same predicament
as the boy who would like to club his companion and yet
would not like to. And like that boy he compensates for
his conflict by playing. In other words there are few, if any,
situations in life which appeal to us in a purely negative way.
COMPENSATORY FUNCTION OF MAKE-BELIEVE PLAY 433
We do not, as a rule, want to suffer great misfortunes; yet
there are certain factors, such as affectionate demonstrations
on the part of our friends, the joy of being in the public eye,
and the like, which give the majority of unfortunate circum-
stances a considerable amount of positive appeal.
Distinctly unpleasant play and fantasy may also provide
for the compensatory expression of negative impulses. There
is little reason to believe that fears, for example, do not
require expression of some sort as urgently as more positive
tendencies. Playing, day-dreaming and the telling of stones
involving ghosts and goblins may well serve to express fears
which must be inhibited in the world of actuality.
Holding the older view that childhood is a period of
happiness and serenity, one could hardly accept an explana-
tion of play in terms of compensation for incomplete or faulty
adjustment — in terms of the partial resolution of conflicts
between the child and his environment or between contra-
dictory factors within his own character. I believe, however,
that there is little need to argue against that older view.
Childhood is primarily a period of incomplete adjustment,
and we remember it as peaceful because we have forgotten
its sorrows and because problems of great consequence to us
in childhood mean little to us now. Full of impulses to do
actual things, the child is equipped with a physique and
surrounded by an environment which are constant obstacles.
I do not believe, like some, that it is desirable, if possible, to
remove these obstacles and make childhood a comparatively
easy and comfortable state. Human life requires, and gets
much of its value from, an abundance of nice adjustments
which can come only as the result of long and necessarily
arduous training. The child comes into the world with an
inherited behavior equipment, but at best this equipment is
an uncertain affair. Each impulse tends to operate in in-
appropriate as well as in appropriate situations. Each im-
pulse, if the child is to become prepared for adult life, must
be defined, and definition implies inhibition. The child must
live through a period of paradoxes before he can become an
individual of discrimination. If he were a perfect mechanism
434 EDWARD S. ROBINSON
and if educational stimuli were perfectly coordinated, it is
possible that he might be trained without being constantly
thrown out of adjustment. Then, too, if the life for which
society prepares him were more simple in its requirements, he
might be spared some maladjustment. But the human
organism is not perfect, and, while educational practice
improves from time to time, the world rushes forward into
new complexities. One who has any faith in the present
direction of progress can hardly do other than accept the
essentially incomplete adjustment of the young as a necessary
product of that progress. The happy fact is that the con-
flicts of youth can be so adequately compensated for by the
play and fantasy mechanisms.
While they are fundamentally natural and necessary phe-
nomena of child life, play and fantasy can result in patho-
logical as well as in normal compensations. And as normal
compensations the forms which they take may modify the
development of character to a marked degree. For these
reasons their exhaustive study, as but different manifesta-
tions of a single process, is essential. Indeed it seems to me
that few fields may be more profitably explored either by
those interested in child life in general or by those interested
in some particular child. From this compensation process,
studied for what it really is, we may hope to gain some new
and useful knowledge about the stresses and strains of human
development.
Still, even at the present time, it is possible to point out
some of the principles which operate in the compensatory
behavior of children.
The distinction between play and fantasy is, of course, a
distinction between overt and ideational behavior. Play,
in so far as it is pretending, is never without an element of
fantasy, but we may arbitrarily confine the application of the
latter term to those forms of pretending which are lacking
in overt bodily accompaniments. It will then be possible to
distinguish between these two types of compensation and to
note their interrelations.
Although we cannot be certain of it, play probably pre-
COMPENSATORY FUNCTION OF MAKE-BELIEVE PLAY 435
cedes fantasy in the child's life. The latter does appear
quite early, however, in some children at least, and before
the school age is reached both are clearly present. I re-
member the interesting evidence for the early rise of fantasy
given to me by a little girl of not more than four, who said,
placing her two chubby hands before her face, "Let's shut
our eyes and play we're at Gran'ma's."
In play and fantasy there are two factors, which may or
may not be consciously recognized by the child, determining
to a large extent the nature of his pretending activities.
In the first place, there is a tendency toward breadth and
freedom of expression. The child must express impulses
which are often clearly incongruous with his world of actual-
ity, and the greater this incongruity the more lively will be
the flights of imagination to which they give rise and the
more apt will the child be to engage in private fantasy
rather than in overt play. In the second place, the satis-
faction which is derived from compensatory behavior depends
to some extent upon its being within the limits of the child's
own credulity. The impulses which drive the child are aimed
at an actual world, and their indirect expression itself must
not get too far beyond the realms of that actuality. Thus,
we may think of these two main determinants of play and
fantasy as (i) the child's natural tendency toward free expres-
sion, and (2) his need for a certain credibility in experience.
The tendency toward free expression leads to the estab-
lishment of all sorts of fictitious characters and objects within
the playground. Toys and playmates which do not fit in
with the completer, fancied world may be put aside. I
remember that even up to the age of sixteen I frequently
judged congeniality in terms of the readiness of others to
disregard reality in favor of a world of pretty definite and
well defined fancy. I always preferred to knock grounders
with one particular lad because he cooperated so well in con-
verting the procedure into the pretensions of a big league
game. The same was true in boxing. Having read and
memorized the details of most of the historic ring battles, we
repeated many of these almost blow for blow upon the floor
436 EDWARD S. ROBINSON
of my mother's laundry. And many were the Harvard-Yale
football games in which I engaged with one other actual
player, both of us, as often as not, playing on the same side.
In cases of this sort, the meaning of ordinary play activity
is widened by the liberal use of fantasy.
In the course of an individual's development many im-
pulses arise which can not be expressed to any satisfactory
extent in a cooperative fashion. Often a child is afraid of
being laughed at for the world he would live in. Under such
circumstances there may be a withdrawal from play to pure
fantasy with its wider possibilities for pretending. Indeed,
one of the signs of coming adulthood is the giving up of
overt play and the switching over to compensatory behavior
of a more private sort. Adults seldom play in the childhood
sense of that term, unless it be in art. In the adult, com-
pensations through pretending are more likely to be worked
out in private day-dreams. The fact remains, however, that
less compensation of any kind is necessary in the general
run of adult lives, so that we may safely assume that fantasy
as well as play is more common during childhood.
Along with this tendency toward free expression, we have
a tendency to make that expression as realistic as possible.
Children are constantly recognizing inconsistencies in their
play life and trying to patch them over as best they can.
When, as a very small boy, I played with tin soldiers and
miniature locomotives, I always felt the inappropriateness
of the size of my own body. The device which I hit upon to
get around this difficulty I called Playing You Are Nothing.
Every playfellow who entered into the world of my tiny
armies and railroads was introduced to the proposition of
suspending all interest in his own body. The running of the
trains and the marching of the troops were to be considered
as events independent of ourselves. There was one youngster
who could not push a locomotive across the floor without
playing he was the engineer. His fate was obvious. I never
invited him to play unless I could get no one else; and, when
he did come, it was to be made miserable by my constant
insistence that he must play he was nothing. Our dis-
COMPENSATORY FUNCTION OF MAKE-BELIEVE PLAY 437
agreement, of course, grew out of the fact that each of us in
his own way was striving to give the play a more vivid
atmosphere of reality.
Just as overt play often passes over into private fantasy
owing to a struggle against the limitations of the actual social
and physical world, so private fantasy often passes over into
overt play in the interests of greater credibility. As a child
I was full of baseball fantasies. Although I played baseball
a great deal, these games did not satisfy certain standards
set up by reading athletic stories and watching older and more
skillful players. But the fantasies, too, often became un-
satisfactory on account of their intangibility. As a result I
formed the habit of laying out a diamond upon the lawn and
there, without ball or playmates, carrying out the overt
movements of an heroic baseball performance. Many a time,
I pitched nine long innings to baffled athletes who swung
immaterial bats at my imaginary curves. Here was fantasy
improved and made realistic by the actuality of its muscular
accompaniments.
The topics of private fantasy are perhaps even more apt
to find increased tangibility by being brought into contact
with a real social world. The child knows that his day-
dreams are unreal, but the insistence of that fact becomes
less troublesome if only he can get some one else to believe
or act as though he believes in the reality of those imagined
events. Many of the lies of children arise out of such circum-
stances. A boy longs for a pony and a box of tools. He
fancies these things in his possession, and before a great while
he somehow feels driven to tell his friends either that he
already has the things he desires or that he has been promised
them. An acquaintance of mine spent her earliest years on
a farm which was more or less out of touch with the livelier
affairs of the world. Now it so happened that an older sister
in this household was sent to town to finish her education.
Upon her return she had much to say of her experiences.
These tales thrilled the younger sister and stimulated her to
day-dreaming. Soon after this the little girl began her own
education at a neighboring country school. As she tells of
438 EDWARD S. ROBINSON
it now, almost her first intercourse with her school mates
was marked by her own spectacular reports of what she
had seen and heard while sojourning in the town which really
she had never been near.
It is interesting to note here that the literary make-believe
of adults contains within it evidence of the tendencies toward
free expression and credibility, which I have mentioned as
such significant factors in child life. Written fiction, for
example, may be thought of as an instrument for free expres-
sion and the spoken drama as an instrument for giving human
fancies increased tangibility. It is hardly necessary to point
out the importance of artistic appreciation and production for
the compensatory life of children.
This view of play as a compensatory mechanism does not
pretend to refute the more familiar theories which, by the
way, were not formulated with special reference to the make-
believe. That theory which describes play as a recapitula-
tion phenomenon simply states that the primary impulses
expressed in play appear more or less spontaneously at set
periods in the child's life, and that the child's activities during
successive periods of his life are definitely reminiscent of the
typical periods of racial development. Most of us would
probably admit that there is a rough similarity between
individual and racial development. But the view that play
is a compensatory activity demands neither the acceptance
nor the rejection of this theory. One need not know nor
try to guess the exact origin and analogies of a child's im-
pulses to realize their variety and the conflicts among them
which demand the compensatory service of the make-believe.
The theory that play prepares a child for later life, if
broadly enough" interpreted, fits in quite well with the notion
which has been developed here. Many impulses arise during
childhood which, while they can not be directly expressed at
that time, still demand preservation. A boy may be inter-
ested in machinery. If he is permitted or even encouraged
in his play and fantasy to concern himself with machines, a
very useful interest may be preserved for the time when it
can find adequate expression. If it were not for compensatory
expression through play and fantasy, it is quite conceivable
COMPENSATORY FUNCTION OF MAKE-BELIEVE PLAY 439
that many such early rising interests or impulses would suffer
repression and thus be lost as far as useful functioning is
concerned.
The theory that play furnishes an outlet for surplus
energy is somewhat vague, but as far as it goes it meets with
no contradiction from the conception that play is compensa-
tory in function.
Much the same may be said for the recreational theory,
which really finds some little support in the type of facts
which I have been presenting. A boredom which longs for
some impossible or impractical distraction is often indirectly
relieved by a compensatory make-believe. The school boy,
tired of his lessons but afraid to dash from the class room,
may partially satisfy himself with a fantasy of the swimming
hole. The worried business man, whose unused muscles
would not tolerate exertion, may yearn to play ball and
take his yearning out in fantasy.
In conclusion, play, the more private forms of fantasy,
much lying and story telling, and the appreciation of stories
all serve the same fundamental purpose in human life. They
are compensatory mechanisms. They are more typical of
children than of adults, because it is in children that the most
incongruity exists between different impulses and between
impulses and the surrounding world of actuality. The nature
of play and the other compensatory mechanisms is deter-
mined by the need of imperfectly adjusted organisms to
express their impulses as freely as possible without too greatly
straining the possibilities of their own belief.
It is essential, if not self-evident, that play should not be
thought of as behavior which is usually undesirable or patho-
logical simply because its function is compensatory. Neither
should we think that, because play grows out of imperfect
adjustment, we should strive for a world in which play is
unnecessary. Simpler organisms than ourselves get com-
pensation through play. The ancients in a comparatively
simple civilization got compensation through play. And in
all likelihood the further humanity advances upon its present
path of progress, the more important will be play and its
related phenomena, especially for the young of the species.
THE CONTROL OF ATTITUDE IN PSYCHO-
PHYSICAL EXPERIMENTS
BY EDWIN G. BORING
Clark University
Dr. Godfrey Thomson has recently published some very
illuminating discussions of the mathematical logic of psycho-
physics, and I for one feel myself too much in his debt to
to indulge in the mere picking of flaws. Nevertheless his
paper, 'A New Point of View in the Interpretation of Thresh-
old Measurements in Psychophysics,' l gives me concern
because it seems to cast aspersions upon what I have regarded
as the most promising direction of development in psycho-
physics. It is true that Dr. Thomson's 'offences' are implied
rather than actual and that ultimately I may discover him
in agreement with my thesis; nevertheless his article furnishes
a reason for placing the point of view that I have in mind
squarely before psychologists. I may add that I conceive
that I am merely explicating an idea that arose within the
Cornell Laboratory. If there be any credit for its origin it
is due Cornell, though the responsibility of the present
exposition is mine.
Dr. Thomson's suggestion is as follows: In the test of
Weber's law (he is thinking of lifted weights as the example)
we may take the differential threshold, which is half the
distance between the upper and lower limens (i.e., half the
'interval of uncertainty'), as the measure of sensitivity;
and such has teen the usual practice. Dr. Thomson, how-
ever, prefers to take as a measure of sensitivity the inter-
quartile range of the point of subjective equality. The
threshold is proportional to the distance between the two
points where the psychometric functions for 'greater' and
for 'less' cross the 50 per cent, abscissa, and its amount
1 Godfrey H. Thomson, 'A New Point of View in the Interpretation of Threshold
Measurements in Psychophysics,' PSYCHOL. REV., 1920, 27, 300-307.
440
THE CONTROL OF ATTITUDE 44 1
depends upon the number of judgments that fall within the
third category of * undecided' judgments (as Dr. Thomson
styles them) or 'equal' judgments. Dr. Thomson would
take these 'undecided' judgments and divide them equally
for every stimulus value between the 'greater' and 'less'
categories, thus establishing two new psychometric functions
(instead of the original three) which of necessity will inter-
sect upon the 50 per cent, abscissa and give zero limens. The
interquartile range is the distance between the two points of
intersection of these new psychometric functions with the 75
per cent, abscissa. It is independent of the 'undecided'
judgments and dependent upon the measure of precision (K)
of the psychometric functions, since the steeper the curves
the less the interquartile range, and vice versa.
Dr. Thomson's preference for the interquartile range as a
measure of sensitivity lies in his distrust of the relative
frequencies of the 'undecided' judgments upon which the
threshold depends. The threshold, he writes, 'depends en-
tirely upon the subject's readiness to give the answer un-
decided. It measures therefore rather a moral character than
a physical sensitivity.' 'The moral character of the measure
S — S' is above all seen from the fact that any subject who
wishes may reduce it to zero, whatever may be his actual
sensitivity, simply by determining that he will never give
an answer undecided.' Thus the interquartile range 'is more
physiological than the threshold measure.' The threshold is
to be mistrusted because 'the decision as to what proportion
. . . is to be called heavier, what undecided and what lighter
depends upon a conscious act of the subject, and can be
varied, if he be so disposed, at his whim; and will vary with
his mood at the moment.' It is such a designation of things
constant and measurable as 'physiological' and of things
inconstant and uncontrolled as 'moral' or psychological that
moves a psychologist to reply.
It is not my present purpose to inquire how much of
Dr. Thomson's argument is actually new. Fechner wished
to measure Unterschiedsempfindlichkeit by h and divided his
442 EDWIN G. BORING
zzveifelhafte cases between the 'greater' and 'less' categories.1
G. E. Miiller opposed Fechner, arguing that the limen must
be used as the basis of Weber's law.2 But Fechner stuck to
his guns,3 as did also Miiller.4 In 1904 Miiller was wondering
'wie in aller Welt kann man ohne weiteres voraussetzen'
that the measure of precision, which is independent of many
factors entering into the lifting of weights, could constitute
the basis of Weber's law. Dr. Thomson may need to meet
Miiller's argument against Fechner.5 I belong to the younger
generation to whom Weber's law and Unterschiedsempfindlich-
keit are less sacred than they once were, and I am willing to
admit that the matter may well wait for supporting facts.
Experimental studies are wanted that show both the threshold
and the interquartile range as functions of the absolute
magnitude of stimulus, and then we can determine how
each fits the Weber-Fechner formula. But I am not willing
to let Dr. Thomson dissuade us in advance from an interest
in the threshold because its 'morality' can not be controlled.
There is enough experimental work to indicate, so it seems
to me, that accurate control of the third psychophysical
category is possible and scientifically necessary.
THE EXPERIMENTAL CONTROL OF THE PSYCHOPHYSICAL
JUDGMENTS
I. In the first place, if we are to gain accuracy of definition
of the psychophysical categories, we must exclude the doubt-
ful judgments.
Dr. Thomson has ample historical ground for including
them. They have been left in from the first. Fechner and
Miiller called them the 'z-cases' (zzveifelhaft), as we have seen.
Miiller raised the question as to whether 'doubtful' and
'equal' judgments ought not to be separated, but concluded
*G. Th. Fechner, 'Elemente der Psychophysik,' 1889, I, loiff.; 'Revision der
Hauptpunkte der Psychophysik,' 1882, 2^f.
2 G. E. Muller, 'Zur Grundlegung der Psychophysik,' 1878, 28f., 33-36.
3 Fechner, ' Revision,' 48f .
4 Muller, ' Die Gesichtspunkte und die Tatsachen der Psychophysischen Metho-
dik,' 1904, 104-109.
B E. B. Titchener summarizes the controversy: Experimental Psychology, 1905, II,
ii, 278-285.
THE CONTROL OF ATTITUDE 443
on experimental evidence that the positive impressions of
* equal' were rare, especially in trained observers, and that
separate treatment of them was therefore not necessary.1
Titchener in 1905 grouped both 'doubtful' and 'equal' cases
under the heading 'uj (uncertain) or '?'.2 Urban in 1908
had his subjects guess when in doubt which weight of the
lifted pair was the heavier. Thus he obtained the categories
'heavier-guess' (hg) and 'lighter-guess' (Ig), which he saw
fit later to combine as 'equality' judgments.3 In such a
setting it is natural to call the region in which the 'equality'
judgments are most frequent the 'interval of uncertainty.'
Urban's pupils have followed his final practice. Fernberger
wrote: "The equality judgment was more complex [than the
other judgments] as it not only included cases of actual
subjective equality . . ., but also all those cases where it
was impossible for the subject to give either a lighter or a
heavier judgment, usually termed doubtful cases." 4
It is under this practice that Dr. Thomson would reject
the interval of uncertainty as a measure of sensitivity because
'any subject who wishes may reduce it to zero . . . simply
by determining that he will never give an answer undecided.1
He is undoubtedly thinking of results like Fernberger's on
the effect of attitude on the interval of uncertainty; but
plainly he does not accept, as I think he should, Fernberger's
contention that attitude can be controlled in the laboratory
and that, in view of this state of affairs, it must be controlled.5
Fernberger's proposal was that attitude should be rendered
constant by explicit instructions and observational training,
1 Miiller, 'Methodik,' I2f.
'Titchener, op. cit. II, i, io7ff; ii, 268.
*F. M. Urban, 'The Application of Statistical Methods to the Problems of
Psychophysics,' 1908, sf., 15, 99.f., io6f., nof., etc.
* S. W. Fernberger, 'On the Relation of the Methods of Just Perceptible Differences
and Constant Stimuli,' Psychol. Monog., 1913, No. 61, 16.
1 Fernberger, 'The Effect of Attitude of the Subject upon the Measure of Sensi-
tivity,' Amcr. J. Psychol., 1914, 25, 538-543. Fernberger refers here to another case
in his own experiments and to Warner Brown's results. He has since called my
attention to a study where three of eight subjects give no interval of uncertainty,
presumably because of failure of attitudinal control: A. L. Ide, 'The Influence of
Temperature on the Formation of Judgments in Lifted Weight Experiments,' 1919,
25 pp. (Univ. Pennsylvania thesis).
444 EDWIN .G SORING
but I am going further in insisting that the instructions and
experimental setting must assure a practically complete
elimination of doubtful judgments. This is the point of
George's study,1 which merits, I think, considerable attention
from psychophysicists. George's thesis is that in a psycho-
physical experiment we are dealing with a series of mental
states which are a continuous function of the given series
of stimuli: we vary the stimuli and note the concomitant
mental variation. Under the rules of scientific experiment
everything else must be kept constant including the attitude
and psychophysical constitution of the subject; if they are
inconstant we can no longer tell of what our judgments are
a function. Moreover, if we find that any particular cate-
gory is in itself an indicator of a change in attitude, we must
so arrange the experiment that this form of judgment will
not occur or rule it out from the results if it does occur, since
the information that it yields is beside the point of the psycho-
physical problem, which always proposes the establishment
of the dependence of judgment upon varying stimulus.
George finds doubt and the doubtful judgments to be the
great offenders against constancy. His method is the estab-
lishment of an Einstellung for constancy of attitude and the
determination of (i) what categories the maintenance of the
Einstellung rules out, (2) what categories, conversely, are
noted by the observers as interfering with the Einstellung
when they do appear, and (3) the objective evidence of
attitudinal shift furnished by reaction times. George's article
must be its own summary; I can not do it justice here. It
is scarcely more than preliminary and his method itself may
yet be called in question. It would be a slender weapon with
which to combat the main body of psychophysical practice
were it not for the fact that, like Dr. Thomson, many psycho-
physicists already know that something or other must be
done to gain greater constancy of results, that the case
against the doubtful judgment is plausible a priori [is a
doubtful difference ipso facto less than an undoubted one?],
1 S. S. George, 'Attitude in Relation to the Psychophysical Judgment,' Amer. J.
Psychol., 1917, 28, 1-37.
THE CONTROL OF ATTITUDE 445
and that the attitude that George proposes as an ideal has
already gained some slight support from experimental usage.1
My personal prejudice for George's conditions I shall mention
presently.
2. Doubt is the most persistent offender and the one pre-
sumably most responsible for Dr. Thomson's strictures upon
the interval of * uncertainty,' but other attitudinal seducers
must also be dismissed. George makes a case against expec-
tation, a case which indicates among other things that a
haphazard method of presentation is to be preferred to a
serial method with partial knowledge (e.g., the method of
least perceptible differences).
* Reflective' judgments are the general class under which
inconstancy is apt to occur. George does not find that the
reflective attitude necessarily means uncontrolled variability
but merely that it favors it. The reflective judgment is
very often equivocally determined; the subject is in a dilemma
as to his report and decides what to report. He may be
judging on the basis of more than one criterion, and may
find in a given case different categories concurrently indicated
by different criteria. If he must make a univocal report for
such an equivocal situation, he must make some decision,
even though the decision may be for the category 'undecided.'
It is the occurrence of such uncontrolled 'decisions,' I take it,
that makes Dr. Thomson wish to give up the threshold as a
measure of sensitivity; but his rejection of the threshold is
not necessary on this score, since reflective judgments can
be avoided.
George shows that doubtful judgments tend to be reflec-
tive. The doubting subject is resolving a dilemma in favor
of one category or the other, and does justice to the unre-
ported category by labelling his judgment 'doubtful.' The
'or-judgments' (e.g., 'equal-or-less') are compromises in which
neither side has won out. The category 'no-difference' is
also often a reflective compromise. Reflection and hesitation
are, however, not synonymous; a judgment may be long
1 L. B. Hoisington, Anur. J. Psychol., 1917, 28, $88ff.; M. Kincaid, ibid., 1918,
29, 227-232; A. M. Bowman, ibid., 1920, 31, 87-90; C. C. Pratt, ibid., 1920, 31.
446 EDWIN G. SORING
delayed and yet come as a simple report of mental process
without any indication of resolution or of attitudinal in-
constancy. In fact it seems probable that the time of forma-
tion of the judgment is one of the mental factors which under
a constant attitude is a serial function of the stimulus and a
feasible subject for exact psychophysical investigation.
The type of judgment furthered by ordinary psycho-
physical procedure is shown in the introspective analyses in
Fernberger's recent monograph.1 These descriptions he ob-
tained under conditions analogous to Urban's except in so
far as the introspection itself interfered. Compared with
what I have in mind as the ideal, they show a relatively
complex — often very complex — process of comparing. I
should not on the basis of them expect even the degree of
constancy which one actually does get. The equality judg-
ments are typically reflective in that they involve a 'verifica-
tion process,' which Fernberger, as if in support of George,
seems to equate to 'doubt.' 2
Of course what is needed to support my argument, and
what is lacking, is the companion introspective study made
under George's conditions. Perhaps these consciousnesses
would not prove so simple as I think. My conviction that the
two consciousnesses would, however, be very different affairs
is dependent, I must in honesty confess, upon personal
experience. I was one of George's observers and I have
observed in other experiments under his conditions. I have
also observed with lifted weights on the turning-top table
under conditions patterned after Urban's. And the two con-
sciousnesses are to me almost unbelievably different. Under
the conventional procedure I am constantly forced into resolu-
tions, verifications, decisions, like those that Fernberger
describes, and thus I am led into doubt and discomfort,
and thence into a naive uncritical attitude which affords no
assurance of rigorous constancy. Under George's conditions
enough effort is required to be sure, but the mental process
is kept simple or else rejected. I do not have to report a
1 Fernberger, 'An Introspective Analysis of Comparing,' Psychol. Monog., 1919,
No. 117.
2 E.g., ibid., p. 160.
THE CONTROL OF ATTITUDE 447
complex situation by an inadequately simple word, nor an
equivocal setting by a univocal judgment. It seems reason-
able, does it not, that a report under George's conditions
should be worth more than a report under Urban's, because it
leads to attitudinal constancy or the detection of inconstancy
when it does occur.1
3. There is as much to be gained by the establishment of
definite serial criteria of judgment as by the elimination of
unsuitable categories of judgment. In fact, these two reme-
dies are obverse and reverse. One can not get rid of doubt,
for instance, unless the judgments are based upon a univocal
criterion. Two criteria that may conflict are fatal to con-
stancy since their resolution is left to chance.
The 'stimulus error' is a term which among other things
characterizes judgments where a definite mental criterion is
not established but judgment is left to chance habits. The
term is undoubtedly much maligned as there are cases where
the direction of the attention to the stimulus is psychologically
useful, as in the preliminary investigation of a new perceptual
field. But in general the stimulus attitude means indefinite-
ness and instability of criterion, as George pointed out.2 A
recent series of studies on the various criteria that may
underlie the judgments of cutaneous duality shows how
fundamental to accurate psychophysical work an avoidance
of the stimulus attitude is.3 In lifted weights Friedlander's
1 In justice to Urban it must be said that I do not believe he would experiment
now as he did in 1908. C/., e.g., the 'Statistical Methods,' 1908 (op. «'/.), with his
'Ueber einige Begriffe und Aufgaben der Psychophysik,' Arch.f. d. ges. Psycho!., 1913,
30, 113-152. It is a pity that he has not yet been able to return to experimental work
since writing his systematic articles. Moreover, both Urban and Fernberger have been
under a special disadvantage in that they were dealing with a peculiarly refractory
material, lifted weights. Isolation of a homogeneous series of univocal mental cor-
relates of the stimulus is very difficult in the lifting process, although it has been
attempted: H. Friedlander, 'Die Wahrnehmung der Schwere,' Zeitsch. f. PsychoL,
1920, 83, 129-120.
* George, op. cit., 35f.
* E. J. Gates, "The Determinations of the Limens of Single and Dual Impression
by the Method of Constant Stimuli,' Amer. /. PsychoL, 1915, 26, 152-157; Titchener,
'Ethnological Tests of Sensation and Perception,' etc., Proc. Amer. Philos. Soc., 1916,
55, 206-215; E. deLaski, 'Perceptive Forms below the Level of the Two-point Limen,'
Amer. J. PsychoL, 1916, 27, 569-571; C. L. Friedline, 'Discrimination of Two Cutane-
ous Patterns below the Two-point Limen," ibid., 1918, 29, 400-419.
448 EDWIN G. BORING
experiments, although they leave much of psychophysical
accuracy to be desired, seem at least to show that different
numerical results follow when attention is upon the weight
and when it is upon the sensory aspects of the lifting.1
The more general ground for the control of criteria is the
one which Fernberger took in defense of the interval of un-
certainty.2 The subjects must be constantly and effectively
eingestelltj and the test of an effective Einstellung lies in
preliminary trials, the taking of introspections, and the
observer's full and repeated characterizations of their attitude.
4. The need for the isolation of the single judgment within
the series is perhaps worth especial mention since the matter
has just been [implicitly] brought to fore by Fernberger's
measurement of the effect of one member of a series upon a
succeeding member.3 Under Urban's conditions with the
turning-top table Fernberger found that the judgment
'lighter' of a pair of lifted weights tends to be succeeded by a
judgment 'heavier,' and vice versa. 'Fixing' a series so that
one kind or the other of sequences predominates produces as
startling effects upon the form of the psychometric functions
as anything that Dr. Thomson is complaining of. Plainly
some sort of expectational or rhythmic effect is operative;
attitude is not remaining constant. Fernberger's solution of
the difficulty is to balance one sort of succession against the
other, and trust that they will cancel. For myself, I could
not feel secure in such a procedure; an algebraic cancellation
where the factors are so little understood can not be so satis-
factory as an actual elimination. The members of the series
should be separated by an interval — a distracted interval if
necessary — so that the intraserial effects are broken up. This
course slows down the rate of experimentation and robs the
turning-top table of much of its charm; but, even when rela-
tive frequencies are aimed at, I do not conceive that numbers
of cases can be allowed to weigh against rigorous scientific
control.
1 Friedlander, op. cit.
2 Fernberger, Amer. J. Psychol., 1914, 23, 538ff. (op. cit.).
8 Fernberger, 'Interdependence of Judgments within the Series for the Method
of Constant Stimuli,' /. Exper. Psychol., 1920, 3, 126-150.
THE CONTROL OF ATTITUDE 449
5. Lest the argumental sauce obscure the meat, let me
summarize. I recommend (i) that every judgment in the
psychophysical experiment stand absolutely independently in
its own right. One member of the series must be separated
from the others in time, and by the instruction to the subject
that he judge it without reference to any other member.
Haphazard presentation should be the rule; at any rate serial
presentations with partial knowledge should not be allowed
since they connect the members and interfere with their
individuality. I urge further (2) that the criteria of judg-
ment be laid down explicitly and univocally in psychological
terms. These psychological terms will be sensory in the
class of experiments especially under consideration. Judg-
ments of stimulus are often desirable, but they are not the
final ideal since they are ipso facto equivocal.1 The univocal
character of the criteria must be tested by introspection and
by the subjects' report. (3) The total Aufgabe under which
the subject judges must be made definite in instructions, and
must be more fully determined by means of repeated char-
acterizations by the subject of his attitude and procedure.
This latter check is important since much of the subject's
instruction is apt to be a self-instruction. (4) The subject
must be both instructed and trained to maintain a constant
attitude throughout the experiment and to report lapses from
this attitude. When he has learned the full meaning of this
instruction, he will not give doubtful judgments nor ordinarily
be doubtful, provided his task is made sufficiently easy for
him by the means of the three foregoing rules. He will in
like manner avoid other reflective judgments that violate the
constant attitude. (5) He will probably tend to give imme-
diate judgments, and he will be greatly helped if he is en-
couraged to report quickly. His times, however, will vary
ll have met psychologists who smile superiorly when I mention the 'stimulus-
error' and even have something to say in reply about 'bigoted introspectionism,' so
I know that I ought not, on purely diplomatic grounds, to bring the stimulus-error in;
but unfortunately for diplomacy it belongs in. I hope some day to show that the
stimulus-error is not a figment of an epistemologizing or a quibbling mind, but that it
is a very real scientific devil. In the meantime let those whom numerical measures
alone will impress, see Friedline, and Friedlander, opp. citt.
450 EDWIN G. BORING
and there will occasionally be long delays [they seem like
* inhibitory jams'!] without gross shift of attitude.
Undoubtedly research will bring more means of control
to the fore, but the observance of these five rules alone will,
I think, give Dr. Thomson data for which the thresholds will
show the degree of constancy that he desires. If in the course
of doing all this he thinks that we have made the thresholds
'more physiological' and less 'moral,' well and good. It is
of this sort of stuff, nevertheless, that psychology is made.
THE NATURE OF THE PSYCHOMETRIC FUNCTIONS
In the article which has caused me to write this paper, Dr.
Thomson supplies us with certain suggestions as to the nature
of the psychometric functions — an exposition that employs
the familiar device of the urn and balls. He gains comfort
from this analogue, I take it, because within it he can show
how the number of black balls necessary for a given category
'depends upon a conscious act of the subject, and can be
varied, if he be so disposed, at his whim.' He gains support
for his analogue (i) from the fact that it gives psychometric
functions that are in accord with present statistical analysis1
in that they are not normal curves (they are not the phi-
function of gamma nor the normal bell) and (2) from the
fact that the three psychometric functions are founded upon
a single underlying error curve. I am quite ready to be
convinced of (i), but (2) seems to me scarcely a reason, since
there are other ways of founding a set of psychometric
functions upon a single error function. I have suggested such
a derivation elsewhere,2 and I desire here to raise the question
which of these two analogues better represents our notion
of the psychophysical organism, and, furthermore, whether
experiment and curve-fitting may not ultimately decide
between the two and thus throw light upon the nature of
sensitivity.
1 Cf., Thomson, op. cit., 304-307; 'The Criterion of Goodness of Fit of Psycho-
physical Curves,' Biometrika, 1919, 12, 226-229.
2 Boring, 'A Chart of the Psychometric Function,' Amer. J. Psychol, 1917, 28,
465-470.
THE CONTROL OF ATTITUDE 45'
Dr. Thomson's parable runs thus. A stimulus (or a
stimulus-pair) is like an urn with black and white balls
in it in a given proportion. The experimental trial is the
drawing of a given (constant) number of balls from the urn.
Something varies so that different numbers of black balls
turn up in the successive drawings. The number of black
balls drawn represents the impression; there is a series of
impressions possible for a single stimulus all the way from
no black balls to all black balls. (At least so I interpret Dr.
Thomson.) The subject reports on this series in terms of
(say) three categories. Dr. Thomson thinks that the subject
decides, 'if he be so disposed, at his whim' what proportions
of black balls drawn shall be reported by each of the cate-
gories, e.g., whether 40 per cent, to 60 per cent, shall be
reported as 'undecided' with the reports 'less' for under 40
per cent, and 'greater' for over 60 per cent. I urge that rigid
criteria should replace the subject's whim. The same thing
happens for the other stimuli, which are represented each by
other urns in which the proportion of black balls is different.
The series of urns, analogous to the series of stimuli, shows a
continuously increasing (or decreasing) proportion of black
balls, but the relation between the place of the urn in the
series (stimulus-value) and the proportion of black balls need
not be linear. Psychometric functions result that are similar
to those actually obtained in practice, although they are not
normal functions.
My own parable, adapted to the differential limen and the
urn, separates the stimulus from the organism. A given
stimulus supplies a fixed number of black and white balls,
since a stimulus is ideally constant. In stimulating the
organism its effect is facilitated or inhibited according to the
chance disposition of the organism, that is to say, the organism
is an urn from which an additional number of black and white
balls is drawn to be added to the fixed number determined
by the stimulus. The total resulting proportion of black
balls fixes the position of the impression in the impressional
series, and the report in terms of predetermined categories
occurs as it does in Dr. Thomson's reasoning. The constant
452 EDWIN G. BORING
tendency toward a given proportion of black balls is a func-
tion of the magnitude of the stimulus, and the variability
about this constant tendency is a function of the organism
(the urn). The proportion of black to white balls in the
urn may be anything at all but is the same for all stimuli.
The resulting ogive psychometric functions may, in a certain
simple case, be normal, but there is no reason why they
should be nor a priori presumption that they would be.
My first question is this. Is it not more reasonable to
ascribe constancy to the stimulus and variability to the
organism, and to assume a law of physiological variability
that is fixed and independent of the magnitude of the stimu-
lus? In other words, should not the organism be represented
by a single urn with contents of fixed composition? I confess
I am not clear as to what Dr. Thomson's urns symbolize nor
as to precisely where in his scheme variability occurs. He has
an urn of different composition for every stimulus. If
variability resides in the organism, then the law of varia-
bility changes for every stimulus, and such an occurrence
does not seem to me physiologically understandable.
To my second question I am not even prepared to suggest
an answer. May it not be^possible to determine empirically
what urn-and-ball analogy best fits the facts, and thus
analytically to learn something new of the laws of organic
variability? A determination would be ever so much better
than guesses — Dr. Thomson's or mine, — but for most psy-
chologists curve-fitting is precarious work. If we had the
proper data, would the mathematical solution be feasible?
Perhaps Dr. Thomson will tell us.
THE PHYSICAL MEASUREMENT AND
SPECIFICATION OF COLOR
BY LOYD A. JONES AND PRENTICE REEVES
Eastman Kodak Co., Rochester, N. Y.
A complete understanding of the subject of color involves
a knowledge not only of the nature of the sensation resulting
from the action of the radiant energy upon the retina, but
also of the exact physical composition of that radiation. The
physicist has been inclined to overlook the nature of the
sensation and to regard as sufficient for a complete specifica-
tion of color a determination of the exact physical composition
of the radiation considered. On the other hand, the psycholo-
gist has been prone to neglect the radiation factors and to
regard as sufficient a specification of the nature of the sensa-
tion resulting when this radiation acts upon the retina. It is
important for the sake of continued progress in this field that
both phases be given due consideration. In this paper it is
proposed to present an outline of the methods whereby the
various factors necessary for the complete specification of
color may be determined. It is the desire of the writers to
treat this subject so as to give a broad general survey of the
whole field in order that the interrelations between the various
factors may be emphasized rather than to deal in detail with
any specific phase.
A careful consideration of the subject as a whole leads to
the conclusion that for the measurement and specification of
the nature of the stimulus, i.e., radiant energy, analytical
methods must be employed, while in dealing with the sensa-
tion the methods are necessarily of the synthetic type. The
necessity of using analytical methods in dealing with the
stimulus is due to the fact that radiation in general is com-
posite in nature and must be separated into its component
parts in order that each may be measured. The most gen-
erally accepted theory of radiation postulates that radiant
453
454 LOYD A. JONES AND PRENTICE REEVES
energy is transmitted as a form of transverse wave motion in
which the wave-length or frequency may vary throughout
wide limits. When radiation of certain wave-lengths reach
the retina the sensation of light is produced. The wave-
length is usually specified in terms of millimicrons (MM)> the
micron being one thousandth (.001) of a millimeter. The
Angstrom unit (.000,0001 mm.) is sometimes used and is
one-tenth of a millimicron. The visual range is approxi-
mately from 400 to 700 nfj, although radiation of shorter or
longer wave-length may be perceived if sufficiently intense.
When any stimulus acts on the retina the resulting sensa-
tion gives no indication as to whether or not the stimulus is
simple or compound in nature. This indicates that the retina
is a synthetic receptor and does not recognize the individual
component parts of the radiation as such but receives the
mixed radiation as a single stimulus producing a single
sensation. The sensation produced may be specified by two
factors, brightness and color, the former being dependent on
the intensity, and the latter on the quality (wave-length
composition) of the stimulus. The Committee on Nomencla-
ture and Standards of the Illuminating Engineering Society
states: "Color of luminous flux is the subjective evaluation
by the eye of the quality of the luminous flux. Any color
can be expressed in terms of its hue and saturation." l
Since visual sensation is dependent upon the radiant
energy emitted by some luminous source, it will be well to
consider briefly the manner in which the quality of such
emitted radiation may be measured and specified. This is
most satisfactorily accomplished by separating the radiation
into its component parts and measuring the intensity of the
individual element. This factor is properly expressed in
watts (or ergs per second) per square centimeter of the emit-
ting source per unit difference in wave-length. Such values
when plotted as ordinates against the various wave-lengths
as abscissae result in a graphic representation of the spectral
energy distribution for the source considered. Such a curve
is commonly referred to as a 'spectral energy curve' or
1 Trans. Ilium. Eng. Soc., 1918, 13, 515.
\
PHYSICAL MEASUREMENT OF COLOR
455
'emission curve* and when determined for all wave-lengths
within the visible range constitutes a physical specification
of the quality of the emitted radiation, and hence of its color.
The measurement of the energy values is accomplished by
use of an instrument known as the spectro-radiometer in
which either a bolometer or thermopile is usually employed
as a receiving element. It will not be advisable to go into a
detailed discussion of such instruments and methods at the
present time, and for more complete information the reader
160
CYLINDRICAL ACETYLENE FLAME.
/
/
/
SPECTRAL) .
5 I
/
/
^•ou
O
(T
UJ
UJ
40
0
/
/
'
/
— . —
.-"
^
420 580 74
WAVE LENGTH
FIG. I
is referred to the literature on spectro radiometry.1 The
spectral energy curves of many sources have been precisely
determined and as a typical example, the curve of the cylin-
drical acetylene flame is shown in Fig. I. A black body is
probably the best standard for spectral energy distribution,
1 Nutting, P. G., 'Outlines of Applied Optics.' Philadelphia: P. Blakiston's Sons
& Co., 1912, p. 234, Chapter IX. Also see recent reports by P. D. Foote in Trans.
Amer. Inst. of Mining and Metallurgical Engineers.
456 LOYD A. JONES AND PRENTICE REEVES
but a black body is difficult to realize in practice so that a
standard acetylene burner operated under the conditions
recently specified by Hyde1 is practically identical with a
black body operated at 2360° K., i.e., 2087° C. Such a
source is easily reproducible.
The precise determination of the emission curve of the
source requires an elaborate equipment and considerable
experience in manipulation on the part of the operator.
However, if this function is known for one source that of any
other can be determined indirectly with considerable ease
by a method of spectrophotometry. The method consists
simply of a comparison and quantitative measurement, wave-
length by wave-length, of the intensity of the unknown
source in terms of that of the known throughout the visible
range. A spectrophotometer for this purpose consists essen-
tially of an optical system such that two beams of light, one
from the source whose spectral emission is known and the
other from the source which is being measured, may be dis-
persed into their component parts. From the spectra thus
formed narrow regions may be isolated and used to illuminate
the parts of some suitable photometric field. In the path
of one beam is situated some device such as a pair of Nicol
prisms, a rotating sector, or a slit of variable width by which
the intensity of that beam may be varied in a known manner.
By the proper adjustment of this device, a photometric
balance may be made and from the constants of the system
the ratio of the intensity of the known to the unknown is
determined. The measurement of this ratio at a sufficient
number of points suitably spaced throughout the visual wave-
length range provides the data from which the emission curve
may be plotted. As examples of the most commonly used
spectrophotometers may be mentioned the Lummer-Brodhun,
the Brace, the Hufner, Konig-Martins, each of which has its
peculiar advantages and disadvantages for the various special
purposes in the field of spectrophotometry. Space does not
permit a detailed discussion of this instrument and again the
1 Hyde, Forsythe & Cady, /. Frank. Inst., 1919, 188, 129-130. See also Coblentz,
ibid., 1918, 188, 299.
PHYSICAL MEASUREMENT OF COLOR 457
reader is referred to the literature on the subject for more
complete information.1
Non-luminous objects are visible only by virtue of the
radiant energy which they transmit, reflect, or otherwise
divert, in such manner that it enters the eye and falls upon
the retina. In case an object reflects or transmits to an
equal extent all wave-lengths of the incident energy within
the visible range, it is said to be visually non-selective or
colorless. This class of objects includes all true grays, which
form a scale varying only in intensity, and limited at the
extremes by black and white. However, when an object
transmits or reflects to an unequal extent the wave-lengths of
the incident energy it is visually selective and is colored.
Every known substance absorbs to some extent radiation of
some wave-length and nearly all absorb very strongly at
some particular wave-length or spectral region. The colors
of opaque objects depend upon the ratio of reflecting to
absorbing power for each wave-length.
A saturated color, that is a pure hue, reflects or transmits
a very narrow region of the spectrum, i.e., absorbs most of
the spectrum and is, practically speaking, monochromatic.
Such media are rare in nature and practice, as most objects
we meet are far from being monochromatic. The spectro-
photometer is used to determine the amount of transmission
or reflection at each wave-length. Fig. 2 shows the curve of
a green filter with wave-lengths as abscissae and density as
ordinates in one curve and transmission as ordinates in the
other. This type of curve is rather typical of ordinary colored
things with the maximum transmission in a certain spectral
region and the somewhat gradual sloping to zero transmission
on either side. A density of I allows ten per cent, of the
incident light to pass so the corresponding transmission is
O.I, a density of 2 transmits I per cent, and so on. Density
is the logarithm of the reciprocal of the transmission, i.e.,
D = log i/T. If two filters are taken together their com-
bined density is the sum of the separate densities and their
1 Nutting, op. cit., Chapter VIII. Also annual reports of Committee on Progress
in Trans. Ilium. Eng. Soc.
458
LOYD A. JONES AND PRENTICE REEVES
transmission is the product of the separate transmissions.
The physical law of absorption stated that the absorption at
any wave-length is an exponential function of the thickness
of the absorbing media or in other words if a unit thickness
transmits a fraction 7", absorbs (i — 71), then the next unit
thickness will transmit the same fraction of what remains
and a thickness X will transmit the fraction Tx. This law
applies only to homogeneous media and monochromatic
radiation.
2.2
to
o-
440
/?
\
\
500 560
WAVE LENGTH
FIG. 2
620
V)
.10
680
A spectrophotometric curve gives the relative transmission
of a filter but may be taken as indicative of color only when
the composition of the incident light is known and specified
as shown by the spectral energy curve. If we change the
nature of the incident light through a filter, or on a reflecting
surface, we also change the nature of the transmitted or
reflected light. Another important thing to bear in mind is
that equal energies do not produce equal brightnesses and
this is illustrated in Fig. 3 which shows the so-called visibility
PHYSICAL MEASUREMENT OF COLOR
459
curve for an average eye. This curve is determined by
measuring the relative amount of energy necessary at each
wave-length to cause equal sensations of brightness. If we
to
.9
£
.7
A
3
.f
J
.2
./
0
V/SIBt
.ITY C
URVfS
/J
^
NVv
//
\
ll
•'
I
I
II
1
1
"
1
fc
1
%
3
§ ;
§ L
I
V
•
Vx
?/
%
/
N
WAVE
LENG1
H
tf JO 3 Si Si & tt Sb 51 58 69 60 6/ hi 63 6f
- Author -j
--- Nutting') ••
----- lvci< - ., „ ,
--- Curve calculated from
FIG. 3
take the maximum sensibility as unity (the point in the spec-
trum being 556 /zju) the relative energy values for other wave-
lengths in the visible spectrum are represented by the ordi-
460 LOYD A. JONES AND PRENTICE REEFES
nates of the curve.1 This shape and position of the maximum
of the visibility curve varies with intensity of illumination
as we might expect from dealing with peripheral or foveal
vision. The product of spectral energy and visibility at each
wave-length gives luminosity and it is really the luminosity
curve that we consider when treating a visual stimulus.
Spectrophotometry gives the exact physical composition
of the radiation in question, i.e., analyzes the stimulus, but
does not provide any direct specification of the subjective
factors of the color sensation resulting when this radiation
impinges upon the retina. This brings us then to the syn-
thetic method by which the specification of the sensation is
made directly in terms of the subjective factors, brightness
and color. Brightness is dependent upon the intensity and
color upon the quality of the incident radiation. These
measurements are usually made by means of a colorimeter,
the action of which is based on the fact that any color can
be matched by a mixture in proper proportions of mono-
chromatic light of the correct wave-length, with white light.
The wave-length of the monochromatic light used is termed
the "wave-length of the dominant hue" and constitutes the
specification of the hue factor. The amount of white light
necessary to make a match taken as a percentage of the total
mixture is known as the per cent, white and represents the
saturation factor. In case the color to be matched is a non-
spectral color such as a purple or a magenta the complement
of the color is found and the specification stated in terms of
the complementary color. Brightness is measured by a suit-
able photometer which may or may not be an integral part
of the colorimeter itself. In the case of reflecting surfaces
the intensity factor is specified in terms of reflecting power,
in case of transmitting media, by the total transmission and
in case of emitting sources by the intensity of the source.
In the latest colorimeters it is possible to read hue to a
fraction of a wave-length in the visible spectrum and a
problem which arises from such a possibility is the determina-
1 Reeves, P., 'The Visibility of Radiation,' Trans. Ilium. Eng. Soc., 1918, 13, 101.
The most extensive work on this subject was published by Coblentz and Emerson.
Sci. Paper 303, Bureau of Standards, issued September 12, IQI7-
PHYSICAL MEASUREMENT OF COLOR
tion of a hue scale. If we desire to measure any quantity it
is first necessary to have a unit of measurement which is
constant throughout the entire scale. (For example, an inch
is an inch, or a pint a pint wherever it may be taken.) When
examining the visible spectrum, however, we find that equal
wave-length intervals do not produce equal color sensation
intervals at different parts of the spectrum, so the wave-
length is not a satisfactory unit in the establishment of a
scale of subjective color sensation and a hue scale established
with fixed wave-length intervals as a unit is therefore un-
satisfactory. The question as to whether or not the difference
limen is equal at all points on any sensation scale has been
discussed pro and con for years but from a physical stand-
point it seems permissible to make these points equal by
definition and call the least perceptible difference the sensation
unit and use it as such in establishing sensation scales. So
the experimental problem in establishing a fundamental hue
scale is to determine the relation existing between the wave-
length unit and the least perceptible difference in hue for the
entire range of visible radiation. Two methods of procedure
present themselves,1 one method measuring the least per-
ceptible difference at certain intervals, say every 5 or 10 ju/z,
throughout the spectrum and the other progressing step by
step and measuring each least perceptible difference in hue in
the scale. Fig. 4 shows the results obtained from the latter
method. The greater the least perceptible difference the less
the sensibility, so the sensibility can be taken as proportional
to the reciprocal of the least perceptible differences. Curve
A represents the hue sensibility curve obtained from the
reciprocal of the limens plotted against wave-length. Curve
B is the scale reading curve obtained by integrating the
sensibility curve. Or by taking least perceptible differences
as units the hue scale may be determined by direct measure-
ment and the sensibility curve obtained by differentiating
the scale reading curve. Examination of these curves shows
maxima and minima at different points in the spectrum.
These results agree very well with previous results obtained,2
1 Jones, L. A., /. Opt. Soc., 1917, i, 63.
1 Steindler, C., Sitzungsber. Wien. Acad. Wiss., 1906, 115, i.
462
LOYD A. JONES AND PRENTICE REEVES
though, of course, the final values must be the average results
from a large number of observers with normal vision.
For an exact determination of the hue scale, spectral light
must be used, as the development of transmitting media or
reflecting surfaces up to the present time has not furnished
either material with monochromatic properties. So an appa-
ratus which will illuminate independently the two parts of a
photometric field with monochromatic light of variable
\
7
400
70O
FIG. 4
quality and intensity must be used. It is quite essential
that the wave-length in each part be easily varied and meas-
ured, and also important that the intensity be accurately
controlled. In making a judgment of equality or difference
of hue with two fields, it is important to have the intensity
of those fields balanced so that an intensity difference does
not influence the hue judgment. The most satisfactory
apparatus used in our laboratory consisted of a Brace spectro-
photometer used in connection with a Hilger spectroscope
of the constant deviation type. The number of distinct
PHYSICAL MEASUREMENT OF COLOR
463
hues between 400 /*/* and 700 /UM was found to be 128 when the
observer started at a given point and proceeded step by
step through the visible spectrum. Several more steps (about
twenty) are added if we examine the non-spectral purples
and magentas. Hue sensibility is nearly independent of
brightness, although it is found that the sensibility is some-
what higher for brightnesses of medium value than for those
of extremely high and low values.
By mixing a pure hue with white and at the same time
maintaining a constant brightness, a series of colors will be
I
^04
— •
1
PURITY SCALE
-
**-^
^<
^v.
IMPURITY °)
•=. _ro w 4
^
N
x
\
\
\
) 20 40 60 80 100 WHITE
X) 80 60 40 20 0 HUE
FIG. 5
obtained differing only in saturation. A specification of
saturation, therefore, denotes the proximity of the color to a
condition of monochromatism. Various terms such as purity,
chroma, and tint are used, but the writers consider either
saturation or purity the most suitable. If we start with a
pure hue (monochromatic spectral light may be considered,
for purposes of measurement, as having a saturation of 100 per
cent.) and proceed to a saturation of zero, we find about
twenty steps. Fig. 5 shows the purity scale with the per cent,
white and the per cent, hue plotted as abscissae and the least
perceptible differences expressed as percentages, as ordinates.
At this point it might be well to review some of the factors
464 LOYD A. JONES AND PRENTICE REEVES
involved in determining the number of visual sensations and
compare the use of spectral light and physical controls with
the ordinary method of mixtures on a color wheel. In either
method the number of perceptible differences we find in a
hue scale or purity scale will depend on the brightness region
we choose for the hue scale and the initial saturation for the
purity scale. In a suitable apparatus using spectral light
we are able to select very narrow regions of the spectrum and
know accurately the region selected from a direct reading
scale. Spectral colors are saturated. When mixing white
light ("white" in this usage means any gray in color mixing)
we can get a very wide range, from the full intensity of the
light source by removing the Nicol prisms to total extinction
when ihe elements are crossed. Here again the physical
specifications are easily obtained from a scale reading. For
the purposes of comparison it is rather easy to obtain two
identical sources and, furthermore, results obtained in one
laboratory will be directly comparable with those obtained
in any other laboratory. With the color wheel, however, we
first meet the difficulty of obtaining monochromatic discs
and discs approaching a physical white or black. Then we
do not know the physical specifications of the mixtures ob-
tained from the various proportions of the discs used. Discs
from the same order may differ from one another and many
of them are unstable even in darkness. These variable
factors not only preclude intercomparisons but also make
separate results doubtful except for demonstrations. In any
case where discs are used it would be advisable to include in
published data the physical specifications of the discs used.
Many text books in psychology make the statement that
"yellow is distinctly lighter than green; violet is darker than
the other spectral hues," and consequently picture the spectral
belt in the color pyramid or color spindle as being tilted
upward at the yellow region and downward at the violet.
This type of statement is true only for certain spectra. The
prismatic spectrum of sunlight will certainly bear out these
statements but any number of other spectra show a different
set of conditions and it is possible to get equally bright yellows
and violets. We also find statements that violet and blue
PHYSICAL MEASUREMENT OF COLOR 465
are the most saturated colors. From a physical standpoint
it is possible to have any color 100 per cent, saturated and
while it is true that 100 per cent, yellow is quite similar to
white, so, too, other saturated colors seem to approach white
when the intensity of illumination is increased. In the spectra
ordinarily used the dispersion and luminosity factors easily
lead one to the statements mentioned above but much more
work must be done with spectral regions of equal saturations,
equal brightnesses and equal luminosities before we can be
positive as to the form and position of the spectral belt of the
color pyramid.
The Nutting colorimeter1 is one of the most satisfactory
monochromatic analyzers, and is especially adapted to the
determination of the dominant hue and per cent, white of
reflecting surfaces or transmitting media. The dominant
hue is read from a direct reading wave-length drum on a
screw which operates a constant deviation dispersing prism.
This prism controls the quality of the standard, the intensity
of which is controlled by a pair of Nicol prisms and the purity
by mixing white light from another source.
Another form of colorimeter is the trichromatic analyzer
which is partly analytical and partly synthetic in nature.
This method specifies a color by giving the relative intensities
of some arbitrary red, green, and blue which when mixed
together match the unknown. The red, green, and blue are
obtained by the use of filters of glass, stained gelatine or other
suitably colored material or by the isolation of narrow bands
of the spectrum. A representative of this type of instrument
is the Ives colorimeter.2 Many other so-called colorimeters
are in reality only color comparitors or tintometers and in
many cases are based on arbitrarily chosen standards.
Although this paper has been a hurried review of some of
the facts on color, it is hoped that the importance of accurate
control of the stimulus in color experimentation has been
emphasized. If results are to be duplicated in various labora-
tories and colored stimuli standardized some of the afore-
mentioned facts must be observed.
1 Bull. Bur. Stand., 1913, 9, i; Zsch. f. Instkund., 1913, 33, 2O.
*/. Frank. Inst., 1907, 164, 421. Ibid., 1915, 180, 673.
SUGGESTIONS LOOKING TOWARD A FUNDA-
MENTAL REVISION OF CURRENT STATIS-
TICAL PROCEDURE, AS APPLIED TO
TESTS
BY SIDNEY L. PRESSEY
University of Indiana
The past three or four years have been notable in psycho-
logical history for the remarkable development of statistical
methods as applied to the problems of mental measurement.
This advance is undoubtedly of the very greatest importance.
The writer has come to feel, however, that, with the first
enthusiasm in such work, there has been a tendency toward
over-elaborateness and diffuseness of treatment, and a lack
of directness and incisiveness in the statistical procedure.
And he wishes to point out certain limitations to the present
concepts of "reliability" and "validity" as applied to tests,
and certain objections to the customary use of the theory
of the normal curve in test building, which he feels to be of
distinct importance.
The situation can most readily be made clear by a very
concrete example. Suppose, then, that a high-school prin-
cipal desires to give a group test for measuring general intelli-
gence to his entering class, in order to pick out in advance
those who are likely to fail in their freshman work. He has
a number of scales under consideration. And he wishes
evidence as to the relative merits of these scales for this
purpose, — for the selection of potential failures. He will
very likely be given data with regard to the comparative
'reliability' and 'validity' of these scales; information may
also be produced with regard to the organization of the tests,
especially in respect to the normality of the distribution of
scores yielded. The present paper aims to show that no
one of these three sets of facts gives that close contact, which
is desirable, with the practical problem.
466
REVISION OF CURRENT STATISTICAL PROCEDURE 467
I. INADEQUACY OF THE PRESENT CONCEPT OF 'RELIABILITY'
The principal may be urged to use a particular scale
because the scale has a high 'reliability.' The exact meaning
of 'reliability' must, however, first be carefully looked into.
The meaning of the concept can best be understood by con-
sidering the way in which 'reliability' is usually measured.
The most common method is simply to give two duplicate
forms of the same test, one after the other, to the same sub-
jects. The ratings obtained by the subjects on 'Form A*
and 'Form B' are then correlated. And the closeness of the
correlation indicates the reliability of the test.
The significance, and the limitations, of a measure thus
obtained are fairly obvious. Two such limitations are espe-
cially important, (a) The measure is evidently a measure
of the reliability of the sampling, — of the particular type of
performance involved in the test. When one speaks of the
reliability of an instrument, one naturally thinks of its reli-
ability/or some purpose. Such a connotation must be guarded
against here. One must not come insensibly to think of the
reliability coefficient of a test of intelligence, for instance, as
indicating the value of the test, as a measure of intelligence.
Such a conclusion is sound only if a test is a simple sampling
of the ability which it is sought to measure; and this happens
much more rarely than might be supposed.1 The term 'con-
sistency' would, therefore, seem a more accurate term; the
'reliability' coefficient indicates only the extent to which a
test is consistent with itself. And it is entirely possible that
a test should yield highly consistent results which were,
nevertheless, not at all measures of the function which it was
desired to measure.2
1 It might seem, for instance, that the Courtis Scale B was a simple sampling of
ability in the fundamentals. But recent research has shown the situation to be by no
means so simple. (See Thorndike, E. L., and Courtis, S. A., 'Correction Formulae
for Addition Tests,' Teachers' College Record, 1920, at, 1-24.)
2 Thus, not so many years ago, cancellation tests were frequently included in
'batteries' of tests intended for the measurement of mental endowment. (See, for
instance, Pyle, "The Examination of School Children,' Macmillan, 1913.) It now
seems quite clear that cancellation tests are not good tests of intelligence. (See McCall,
' Some Correlations between Mental Traits,' Teachers' College, 1916.) But cancellation
tests appear to be quite 'reliable' measures, — they are simply not good tests of general
intelligence. They are, therefore, not 'reliable' for the purpose for which Pyle used
them.
468 SIDNEY L. PRESSEY
It must also be kept in mind (b) that such a measure of the
reliability of the sampling may be considered an adequate
measure for this purpose only if the scores obtained on ' Form
A' and 'Form B' may be considered entirely random sam-
plings of performance on such a test. Usually they cannot be
so considered. There may be an initial difficulty with direc-
tions at the beginning of 'Form A' and a slight fatigue
toward the last of 'Form B.' What is, with many of the
tests, more important — the method as described above tells
us nothing whatever about the 'consistency' of the results
from one examiner to another, one scorer to another, from
one day to another, or one time of the day to another.
To come back to the original problem, then: such a meas-
ure of the consistency of the test with itself, under certain
circumstances, tells the high school principal surprisingly
little as to the value which that test may have in distinguish-
ing his potential failures from the rest of their class.1 And
information with regard to the 'validity' of the scale is
naturally turned to, to settle this practical question.
II. THE ARTIFICIAL NATURE OF CURRENT CONCEPTS
REGARDING VALIDITY
The principal is, then, urged to use a particular scale
because the scale has a high 'validity' as a measure of general
ability. That is, data are presented showing that the scale
gives results having a high correlation with independent
criteria as to general intelligence, and congruence with current
theories regarding the nature of general intelligence, — there
1 The writer is inclined to feel that most problems of consistency can best be dealt
with in general terms. That is, what difference, in general, may one expect in test
results if one tests Monday instead of Friday, at 9 o'clock instead of 3 o'clock? What
difference, in general, may be expected, with a given type of directions, from one
examiner to another? What differences, with various scoring methods, may be
expected from one scorer to another? What differences may result in the score of an
individual as the result of fluctuations from one time to another, in general feeling
tone, energy, vigor, health? The writer believes that, until evidence to the contrary
appears, it may be taken for granted that such factors affect all tests in- more or less
the same way; certain general theorems with regard to their operation should, then,
be possible — or general precautions taken. The only problem of consistency that
needs specific determination for each test would then be consistency as it relates to the
subject matter of the test.
REVISION OF CURRENT STATISTICAL PROCEDURE 469
is a regular rise in score from year to year until maturity,
a relative freedom from the influence of specific training, and
so on. This concept of 'validity' is also, the writer feels,
beside the point, if not misleading, so far as the practical
problem of the high-school principal is concerned. And again
there are two difficulties.
In the first place (a), since the extent to which general
intelligence is the fundamental factor, in conditioning success
and failure in the Freshman year of high school, is not known,
the usefulness of the scale (even if proven a satisfactory
measure of general intelligence) is still an unknown quantity.
Stability of character, willingness to apply oneself even
though the restraints of grammar school supervision are now
removed, interest in the more mature subjects of the high-
school curriculum — such elements are probably more im-
portant than is often supposed, in the total situation.1
Differences in the adequacy of previous preparation may also
be of importance. So proof of the 'validity' of a scale as a
measure of 'general intelligence' is by no means proof of the
value of the scale in sorting out potential failures among these
high-school freshmen. In fact, it might almost be said that
in proportion as the scale measured one element only, in a
complex situation, to just that extent was the scale inadequate
for dealing with that total situation!2
It remains to be pointed out, however, that even though
1 For a discussion of this tendency to overestimate the comparative importance
of intelligence see Rosenow, Curt, 'Is Lack of Intelligence the Chief Cause of Delin-
quency?' PSYCHOL. REV., March, 1920.
2 The more extreme theories in regard to general intelligence surely make up, in
the aggregate, an extraordinary concept. It should surely be kept in mind that it is,
in the first place, an analytical concept and so dependent for its character upon the
methods of analysis employed. It should also be pointed out that such a concept
naturally receives successive accretions in the way of theory and may, by a mental
synthesis largely adventitious to the facts, acquire a reality which is very largely an
artifact. Scores on various tests are lumped and the aggregate used as a measure of
general ability. A more or less close relationship is naturally found between such an
aggregate and the average marks of the children in school. Teachers, especially in the
grades, naturally think of the child's work as a whole, and give marks showing high
correlations between abilities in different subjects. And the children come to this
attitude and react to their school work as a whole. And — the whole situation is
cumulative. One might, in fact, imagine the concept of general ability thus developing
even though abilties were, as a matter of fact, diverse and uncorrelated.
470 SIDNEY L. PRESSEY
the demonstration of a close correlation between the scale in
question and a general intelligence were supplemented by
evidence that general intelligence was the fundamental factor
in the situation, still the suitableness of the scale for the
particular problem would remain to be shown. That is,
(&) the usual method for stating relationships between two
variables — the correlation coefficient — does not express satis-
factorily the nature of that relationship, for diagnostic pur-
poses, at a particular point in the distribution. The problem
is: How unmistakably will the scale set off the lower 15 per
cent, or so in scholastic ability? A correlation coefficient is
only very general evidence in regard to this particular matter.1
And it is evidence with regard to such diagnostic efficiency
that the school principal should require.
III. THE IRRELEVANCY OF THE THEORY OF THE NORMAL
CURVE IN PRACTICAL PROBLEMS IN CLASSIFICATION
/
Proof of the validity of a scale as a measure of general
intelligence is, then, not proof of the value of that scale for
sorting out potential high-school failures, since failure is not
conditioned by general ability alone, and since the diagnostic
efficiency of a measuring instrument is not the same thing
as the general relationship of that instrument to the factor
concerned. A third set of facts may, nevertheless be intro-
duced in evidence of the value of the scale in question. It
may be pointed out that the tests of the scale are very care-
fully constructed so that equal increments on the scale
represent equal increments in ability, and so that the total
distribution of abilities yielded is closely similar to the dis-
tribution of abilities that would be expected according to
1 See, for instance/Thurstone, L. L., 'Mental Tests for College Entrance,' /. of
Educ. Psychol., March, 1919 and Pressey, S. L., 'Suggestions with Regard to Prof.
Thurstone's "Method of Critical Scores," ' /. of Educ. Psychol., December, 1919.
The writer has often wondered whether the early introduction of the Pearson
products-moments formula for calculating the correlation coefficient has not hindered
rather than helped the study of relationships, in psychology. There are, of course,
no right and wrong methods; methods are simply more or less adequate to the data
and the problem in hand. One could almost say, dogmatically, that the particular
type of data and problem to which the Pearson method is applicable were relatively
rare. Most practical problems require a two or threefold division.
REVISION OF CURRENT STATISTICAL PROCEDURE 47*
the theory of the 'normal curve.' Once more the writer
would object to the relevancy of the information to the
practical problem, and on two counts.
(a) Construction of the scale so that equal increments of
ability are related to equal increments in score means,
probably, transmutation of values in terms of the per cent,
passing different items into positions on the normal curve or
some such procedure.1 It need only be said here that items
which give a satisfactory scaling on such a curve need by no
means be the most diagnostic items. An item may appear
in a test because it is the only item appearing at 1.52 P.E.
(when scaled as mentioned above) or it may appear in a test
because most of the potential failures cannot pass it and most
potential successes can. The last criterion is obviously the
fundamental one if the problem in hand is to obtain a test that
shall most completely differentiate the potential failures.
(&) It may also be pointed out shortly that for the par-
ticular practical problem under consideration a normal dis-
tribution of scores is hardly to be desired. If a scale sets off
the potential failures very completely, it will lump the assured
failures at the bottom and the assured successes at the top,
and spread out the questionable cases in between. In short,
equal increments of ability and a normal distribution of scores
are not to be desired if the greatest efficiency, for the practical
problem postulated, is sought.
IV. DISCUSSION
Well — most of these points seem obvious enough, perhaps.
But the concept back of them indicates a fundamentally
different statistical attack, in the development and use of
tests. If differentiation of the potential failures in high school
is an important problem, why not build a scale specifically
for that purpose? Select items simply according to their
ability to make the desired division. Combine those items
so that such a lumping of cases at the two extremes is ob-
tained; the reverse of the normal distribution is the distribu-
1 Of which procedures, transmutation of percents passing at different chrono-
logical ages into supposed units of mental growth is surely more questionable still.
472 SIDNEY L. PRESSEY
tion to be desired.1 Then measure the value of the test by
measuring its 'efficiency' in dealing with the practical problem
for which the scale has been designed. Deal with each
important problem in some such empirical and concrete
fashion. And, if, out of a large number of such attempts,
there emerge certain unitary factors, — a general ability, a
series of character types, or what not, — well and good. But
the postulation of such elements in advance, with verification
primarily by reference back to these postulates, is both an
unscientific and a practically dangerous proceeding.
First a very specific problem; then, after that — everything
subservient to the solution of it! Every item chosen with
reference to that one problem, every method aiming only at
the most direct and empirical solution of that problem — no
hypotheses, as thoroughly empirical treatment as may be!
The result will be, the writer believes, an essentially new
statistical approach (methods now in use suggest something
of this sort, particularly the methods used in the development
of the army trade tests). Such a revision of methods is, the
wrker has come to feel, necessary, for a clarifying of the
total situation.2
1 Is this not really the solution of the problem of the normal curve in mental
measurement? (See, for instance, Boring, Amer. J. of Psychol., January, 1920).
The actual distribution of various traits is a matter of academic interest only. But
meantime, the distribution to be sought in test work will be determined by the problem.
2 And now the apology! There is little essentially new in the paper, of course.
(In fact, it should be said that a detailed discussion, with full use of the literature,
was first attempted, but was found to extend beyond reasonable limits.) The im-
portant thing, however, is the total implication of the various points presented. Our
statistical methods as applied to tests have been largely borrowed methods, — and
methods borrowed from the descriptive sciences. So the question has been: What is
the test measuring, and how accurately is this thing being measured? But mental
testing is not a descriptive, but a technical science. And the question should be,
instead: What are we trying to do, and how well are we doing it? The distinction is,
the writer believes, of the very most fundamental importance, involving fundamental
differences in statistical approach.
It remains to be mentioned that the points made apply equally to measures of
achievement in the school subjects or other like tests. Instead of measuring " ability in
arithmetic " in the eighth grade — and then commenting mildly on the extent to which
arithmetical ability in the eighth grade overlaps on the seventh, why not tackle a defi-
nite practical problem, — attempt to define the passing point in arithmetic for the
eighth grade? The distribution, again, should be bi-modal, not normal, — and the
other points mentioned follow.
BF
1
P7
v.27
Psychological review
For use in
the Library
ONLY
PLEASE DO NOT REMOVE
SLIPS FROM THIS POCKET
UNIVERSITY OF TORONTO
LIBRARY
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Books to Borrow
Open Library
TV News
Understanding 9/11