COLUMBIA LIBRARIES OFFSITE

HEALTH SCIENCES STANDARD

HX64101690
QP391 .H91 The delayed reaction

QF3dl

Columbia ^ttitier^ftp

Ifpartm^ttt of pijpwlogg

Digitized by tine Internet Archive

in 2010 witii funding from

Open Knowledge Commons (for the Medical Heritage Library project)

http://www.archive.org/details/delayedreactioniOOhunt

COLUMBIA UNIVpRSITV

OEPARTMENT OF PHYSI/)LOC

College of Physicians and Surgeons
437 west fifty -ninth street

NEW YORK

Behavior Monographs

Volume 2, Number 1, 1913 Serial Number 6

Edited by JOHN B. WATSON

The Johns Hopkins University

The Delayed Reaction in Animals
and Children

WALTER S. HUNTER

Published

at Cambridge, Boston, Mass.

HENRY HOLT & COMPANY

34 West 33d Street, New York

G. E. STECHERT & CO., LoDdon, Paris and Leipzig. Foreign Agents

The Journal of Animal Behavior

An organ for the publication of studies concerning the instincts,
habits and intelligence of organisms

The Journal contains a Departntent of Notes in which appear brief accounts of
especially interesting and valuable observations of behavior.

Published bi-montlily at Cambridge, Boston, Mass., by Henry
Holt and C!ompany, New York. Each volume contains at least 450
pages with plates and text-figures.

The subscription price is $3.00 per volume (foreign subscription
$8.60) postpaid.

Manuscripts, subscriptions, and all business correspondence should
be addressed to

THE JOURNAL OF ANIMAL BEHAVIOR

Emerson Hall, Cambridge, Massachusetts.

Behavior Monographs

For the publication of studies in behavior and intelligence which are

too lengthy or too expensive for acceptance

by The Journal of Animal Behavior

Published at irregular intervals at Cambridge, Boston, Mass., in
connection with the Journal of Animal Behavior, by Henry Holt and
Company, New York. Each volume contains approximately 450
pages with plates and text-figures.

The subscription price is $3.00 per volume (foreign subscription
$3.50) postpaid.

Monographs may be purchased separately at prices varying with
the cost of manufacture.

Manuscripts and inquiries concerning terms of publication should
be addressed to the Editor of the Behavior Monographs,
JOHN B. WATSON,

The Johns Hopkins University,

Baltimore, Md.

Subscription to The Journal of Animal Behavior and the Behavior
Monographs should be sent to Emerson Hall, Cambridge, Massa-
chusetts.

Behavior Monographs

Volume 2, Number I Serial Number 6

, Edited by JOHN B. WATSON

The Johns Hopkins University

Studies from the Psychological Laboratory
of the University of Chicago

The Delayed Reaction in Animals and Children

WALTER S. HUNTER, Ph. D.

Instructor in Philosophy, The University of Texas

Published

at Cambridge, Boston, Mass.

HENRY HOLT & COMPANY

34 West 33d Street, New York

G. E. STECHERT & CO.. London, Paris and Leipzig, Foreifn Agent*

ACKNOWLEDGMENTS

It is a pleasure to acknowledge my deep obligation to Prof.
Harvey A. Carr both for setting the problem and for the careful
supervision and criticism of the work here presented. To Prof.
James R. Angell also I owe much. Not only has he subjected
the monograph to a searching analysis; but, through that im-
palpable something called "laboratory atmosphere," he has
influenced the tone of the interpretations. Prof. John B. Watson
of Johns Hopkins University has read a first draft of the theo-
retical section of this paper. I wish to add a word of thanks
to Mr. W. R. Hough and Dr. H. B. Reed for the guidance given
me by the data which they had accumulated in an early study
of the present problem. Further notice of their work is given
below. Mr. De Vry, director of the Zoological Garden in Lincoln
Park, Chicago, and Dr. Homaday, director of the New York
Zoological Garden, have both given me information concerning
the habits of raccoons.

TABLE OF CONTENTS

Page

I. Introduction^ 1

II. Critical Review of Historical Data 2

1. Imitation 3

2. Use of tools 5

3. Dreams 8

4. Learning curve 8

5. Memory 9

6. Thorndike's test 10

7. Learning by being "put through" 14

8. Recognition 16

9. Rate of forgetting 16

10. Association by similarity 16

11. Reluctance and expectancy 17

12. Varying means to the same end ' 17

13. Reactions to a temporal series of colors 19

14. Washburn's cat on the stairway 20

III. Notes on the Animals and Children Tested 21

IV, Apparatus and General Method 22

V. Experimental Results 30

1. Tests with animals 30

A. Learning the association 30

(a) Rats 30

(b) Dogs 32

(c) Raccoons 33

(d) Summary 34

B. Maximal intervals of delay attained 35

(a) Group differences in maximal delay 35

(b) Effect of size of release upon delay 37

(c) Effect of backgrounds of different brightnesses upon

delay 38

(d) Effect of number of boxes upon delay 39

(e) Effect of other conditions upon delay 39

C. Methods of reaction after delay used by animals 39

(a) Orientation of whole or part of body 40

(b) Position in the box 47

2. Tests with children 52

A. Method of experimentation 52

B. Are the results obtained from animals and children com-

parable? 53

C. Learning 55

D. Differences between the learning of animals and children ... 55

E. Delayed reactions 57

VI. Theoretical Considerations 62

1. The cues essential for correct delayed reactions 62

A. Substitutes derived from the external environment 63

B. Substitutes derived from within the subject's body 67

(a) The type of internal cue used 67

(b) The mechanism by which internal cues guide behavior 70

2. The place of ideas in the grades of animal learning 73

VII. SUM.MARY and CONCLUSIONS 79

VIII. Appendix 82

A. Detailed records of two rats and two raccoons 82

B. Notes on raccoons 85

V

I. INTRODUCTION

The experiments in this monograph • aim at an analysis of
typical mammalian behavior under conditions where the de-
termining stimulus is absent at the moment of response. Asso-
ciations were first set up between movements that led to food
and a Hght which might be in any one of three boxes. ^ Controls
were used to make sure that the position of the light alone
determined the reactions of the subject. Tests were then in-
stituted in which the light was turned off before the reaction
was made. The subject thus had to respond in the absence of
the stimulus that hitherto had guided his reactions.

The nature of the present experiment may be further set
forth by contrasting it with the following type of adjustment:
A cat watches for a mouse and sees it appear at an open hole.
The mouse vanishes before the cat can react, yet the cat goes
over to the hole. There can be no question here but that the
determining stimulus is absent at the moment of response, pro-
vided possible olfactory stimuli be neglected. Our experiment
differs from this in complexity. If there were three holes that
differed only in their several directions from the cat, and if in
the past the mouse had appeared an equal number of times in
all three holes, the conditions would be "the same as in our tests.
A selection between the three holes would need to be made on
the basis of the immediately previous presence of the rat, if
a correct reaction were to occur. If an animal can manifest
behavior that does not lend itself to a "stimulus and response"
explanation, this is one type of situation in which that behavior
should appear. That, in fact, it is the situation par excellence
for the eliciting of this behavior will, I believe, appear as this
monograph progresses.

' Experimentation on the present problem was first begun in the University of
Chicago laboratory by a graduate student, W. R. Hough. The followmg year
the work was taken up and carried somewhat further by another student, H. B.
Reed. Both students worked with white rats. Although in each case the results
obtained were in strict harmony with those presented in this paper, m neither
case were they conclusive. The chief value of the work lay in its suggestiveness.
The apparatus used by Reed— Prob. Box D— is described below, ^^^e present
investigation was carried on in the same laboratory from October, 1910 to
April, 1912.

1

2 WALTER S. HUNTER |

In the present experiments, two main factual questions arise ;
(i) How long after the determining stimulus has disappeared
can an animal wait and still react correctly? (2) Does the
animal give any behavior cues as to its method of solving the
problem? If so, what are they? With these data given, there
remains the task of interpretation. If a selective response has
been initiated and controlled by a certain stimulus, and if the
response can still be made successfully in the absence of that
stimulus, then the subject must be using something that func-
tions for the stimulus in initiating and guiding the correct
response. Our investigation thus forces us to the considera-
tion of the functional presence of a representative factor in the
behavior of animals and children. Not only this, but the prob-
lem of the nature of this representative factor confronts us. Is
it an overt motor attitude, or not? If not, is it sensory or imag-
nal, i.e., ideational?

In the interpretative study, I shall proceed on the assump-
tion that an'mals are conscious. What the nature of this con-
sciousness is, it will be the task of this paper to help determine.
(If the reader does not choose to follow this line of interpreta-
tion, he may state everything in neurological terms without
marring the significance of this discussion.) But a propos of
the term "image" or "idea," let it be said once for all that
wherever these terms are used by the present writer with refer-
ence to animal consciousness, they should be supplemented by
the phrase "or functionally equivalent process." I use the
structural term chiefly for the sake of its brevity.

II. CRITICAL REVIEW OF HISTORICAL DATA .

In the interpretative discussion at the close of the present
monograph, we shall be confronted with the possibility that
images or ideas ^ may have guided the reactions of the sub-
jects. In that discussion, we shall assume that there is no
necessity that psychology postulate such a representative factor
save where successful reactions occur in the absence of the stim-

^ In the literature, it has been taken for granted that the question of the exist-
ence of images and that of the existence of ideas in animal consciousness is the
same. I shall proceed on the assumption that images are centrally aroused pro-
cesses, while ideas may be either peripherally or centrally conditioned. The essence
of the idea is not its origin (or content), but its function. This point will be ampli-
fied in the final divisions of this monograph. The present statement will suffice
for a definition of terms.

DELAYED REACTION 3

ulus (object) or movement that is represented. In the present
historical section the attempt is made: (i)To present the main
types of behavior that have been used as evidence of the exist-
ence of images in animals; and (2) to show that this evidence
is inconclusive, because the behavior may be stated in sensori-
motor terms. In the light of the assumption just made, there
are two reasons why this behavior may be so interpreted: (i)
The stimuli determining the reactions in question have not been
adequately known; and (2) inasmuch as this is true, one has
no right to assume the absence of the stimuli which are sup-
posed to be represented. All of the behavior that is summar-
ized in this section may have involved representative actors.
The point of my criticism will be that they need not have in-
volved such processes. It is not merely a question of the appli-
cation of the law of parsimony, as that is usually stated. It
is also a recognition of the fact that sensori-motor behavior
is genetically the more fundamental form.

The following are the types of evidence to be considered:
(i) Imitation; (2) use of tools; (3) dreams; (4) nature of the
learning curve; (5) memory; (6) Thomdike's test: The imme-
diate reaction to one stimulus before tjie appearance of a second
which has always accompanied the first after an interval of
a few seconds; (7) recognition; (8) learning by being "put
through;" (9) rate of forgetting; (10) association by similar-
ity; (11) reluctancy and expectancy of response; (12) vary-
ing means to the same end; (13) reactions to a temporal series
of colors, and (14) Washburn's cat on the stairway.

1. Imitation

Thorn dike ^ is an example of the comparative psychologist
who interprets the highest type of imitation as requiring the
presence of images or ideas. In his experiments, however, he
found no evidence for the existence of this kind of behavior.
The general character of the imitation experiment is well known.
An animal that has failed to solve a certain problem is con-
fined where it can see another work the mechanism and get
food. It is then given an opportunity to make the desired
reaction. If the trial succeeds or if there is a sudden improve-
ment in the animal's behavior, it is said to have profited by

'Thorndike, E. L. Animal Intelligence. 1911, New York, pp. 76-108.

4 WALTER S. HUNTER

viewing the other's performance. Haggerty * found this be-
havior in well controlled tests with monkeys. He divides the
stimulus for that imitative behavior into three parts: (i) A
fairly simple mechanism; (2) the perception of another animal
working at the mechanism, and (3) the perception of the other
animal getting food. Haggerty does not discuss the theoretical '
side of imitation and argue for the existence of images. He
simply presents the data. However, in view of the stand taken
by Thorndike and others on interpretation, we shall neverthe-
less use material drawn from Haggerty 's monograph for analysis.
The following criticisms tell against the use of this type of
behavior as crucial evidence for images: The animal that does
the imitating may make its improvement under the incentive
of a social impulse rather than of the apprehension of relations.
This receives confirmation by the fact that Haggerty 's monkeys
did better if the animal to be imitated was a stranger and thus
aroused the imitator more strongly than a familiar animal would
have done.^ This criticism is a variant on the following one,
(2) The fact that the sight of the other animal being fed was a
part of the stimulus makes it at least possible that the imitator's
attention was simply drawn very vividly to the spot to be
attacked. Of course the more complicated the reaction to be
made and the more exact the execution of this by the imitator,
the better the argument for an ideational perception of rela-
tions. However, the general objection will always be valid that
one can never say that the imitating animal was not guided
solely by stimuli present to sense. The determining factor in
the stimulus can not be said to be ideational. Once the animal's
attention is vividly focused upon the objective sequence "pull-
ing string — getting food," e.g., the necessary reaction may fol-
low by association when the opportunity presents itself. Mon-
keys in particular have such an enormous repertoire of reactions
that it is very tempting to analyze such imitative reactions as
Haggerty presents on a purely "stimulus and response" basis.
That author says of one of his typical experiments (Chute
Experiment A) : "In order to secure the food, the monkey must
leap from the wire part of the cage to the chute, and, while

* Haggerty, M. E. Imitation in Monkeys. Jour, of Camp. Neur. and Psy., 1909
vol. 19.

5 Op. cit., p. 436.

DELAYED REACTION 5

holding to it, must thrust a hand up inside and pull the string,
thereby releasing the small door in the top of the cage and
allowing food which had been placed on it to fall to the floor.
He must then descend to the floor to get the food." « After
the imitator's attention is focused upon the hole at the chute's end
due to the action of the imitatee, what is more natural than
that it should jump to the chute? The hole has "caught its
eye" before the jump, so the hand goes in almost reflexly, as
reflexly closes about the string and pulls. Food falls and is
eaten. If the animal is attentive to the movements during the
performance, the chances are that it will henceforth succeed.'

2. Use of Tools
A second argument for the presence of ideas is the use of
tools by animals. In view of the fact that such behavior has
at times been exclusively claimed for man, students of animal
behavior have long sought for conclusive evidence on this ques-
tion.* We shall take typical data presented by Hobhouse for
consideration. This will also lead to an understanding of that
author's attitude on the question of images. Hobhouse treats
such behavior largely under the heading "Articulate Ideas."
An example will best lead us to an understanding of his posi-
tion. The animals used were monkeys, one a Rhesus and the
other a chimpanzee. The latter had already learned to throw
a rug over food placed at a distance in order to rake in the
latter. He was then taught to substitute a stick for the rug.
Quotations from the author will now indicate the animal's prog-
ress. "Next day, the chimpanzee learnt to use a short stick
in order to reach a larger one, with which in turn he could reach
the banana." * "One day I gave him a rope with a noose to
throw over the box in place of his stick." (The banana was
placed in a cigar box.) "I did not give him any hint, but he
soon tried it in a vague way. He did not, however, understand
the matter very well, for when he succeeded in getting the

• Op. clt., p. .355.

' Vide Hobhouse, L. T. Mind in Evolution. London, 1901, p. 202 for place
of accident in learning.

' Lindsay, \V. Lauder. Mind in the Lower Animals in Health and Disease.
London, 1879, vol. I, chaps. 23 and 24.
Hobhouse, L. T. Op. cit., chap. 10.
Also many recent studies in imitation.

• Op. cit., p. 236.

6 WALTER S. HUNTER

rope round the box, he did not seem aware of his advantage,
flung it away, went off for his shawl, and used it very success-
fully. I then tied a block of wood to the rope to assist in throw-
ing it. He attempted this spontaneously, at first without suc-
cess. Presently, however, he happened to pitch the block right
into the box, which today was open, pulled it in, and got the
banana. Notwithstanding this signal success, he never took to
this trick." ^^ "In the experiment to which I have already
referred, when the box was tied to a rope, the further end of
which was passed over a stanchion several feet from the cage,
he^ failed, as I shall mention later, to find the right method,
but was fertile in devising wrong ones. He would shake the
rope violently, so that the banana would fall out of the box.
He would then swing the rope to and fro, swishing the banana
about from side to side, until by degrees it would come within
his reach, in a way which I should have thought beforehand to
be quite impossible." ^^ "In these tests," says Hobhouse, "it
was necessary that [the monkeys] should grasp how the stick
and the food stood in relation to them; that they should get
the stick at the food and beyond it." 1=' "A form of 'analogical
extension ' is also strongly marked in the use of substitutes
differing very widely in appearance and the manner of use
from the object first employed." ^^ These illustrations of the
use of tools are examples and proofs of the existence in monkeys
of what Hobhouse means by the practical judgment, where artic-
ulate ideas are employed.'^ The author explains articulate ideas
as follows : " By a more articulate idea, is meant one in which
comparatively distinct elements are held in a comparatively
distinct relation." ^^ The nature of the practical judgment is
set forth as follows: "It is more than assimilation, because
what is revived is an idea, a definite reference to something
unperceived. It is more than association, because relation be-
tween the 'revived' idea and the given perception is an essen-
tial part of it, and it is less than analytic thought, because the
relations involved are not dissected out as distinct elements in

i» Op. cit., pp. 237-8.
" Op. cit., p. 238.
12 Op. cit., p. 241.
15 Op. cit., pp. 241-2.
" Op. cit., p. 269.
1= Op. cit., p. 234.

DELAYED REACTION 7

consciousness." '" "The practical judgment is not independent
of associations, for association supplies the whole of its material.
But out of that material, it selects what it wants, and shapes it
as required."" This "material" that association presents is
ideational as the quotation from page 117 indicates. It seems,
though I would not be sure, as if Hobhouse assumed the exist-
ence of ideas and is only concerned with the possibility of their
functioning in behavior. As a basis for this imputation, I
point to his discussion of the association of ideas on page 114
and to that on pages 200-1. In the latter place, he makes
clear (?) that what he does not wish to attribute to animals is
the conscious analysis of the perceptual order. Such a dis-
tinction between existence and efficient functioning in behavior
seems to be theoretically permissible.^" Images, ideas may
exist sporadically as has been claimed in the case of animal
dreams. But comparative psychologists have neither the right
nor the need to assume the existence of such processes save as
that may be forced upon them by the evidences of behavior.

The question now is : Does the use of tools'" such as described
imply the presence of a centrally aroused factor? In the light
of the above analysis of imitation, the reply is no. Even if
Hobhouse 's results be fully accepted without raising the ques-
tion of careful controls, one need not accept his interpretations.
The behavior may have been controlled entirely by sensory
factors. As to the fertility shown by the animals in devising
methods, (see above, p. 6), this was very probably but a ran-
dom use of acquired co-ordinations. The use of the tools was
acquired as any habit and the sight of the individual objects
(ropes, etc.) aroused the type of reaction that had been taught.
This shows a higher grade of intelligent adaptabiHty in the
animal than if it had been Hmited to the use of one object, but
it does not prove the existence of a central conscious factor.
It may be that such animals as the primates are able to give
similar responses to different sensory stimuli on account of a
factor of "hyper-excitabihty." I hesitate to use such a term,
but the general type of case held in mind is illustrated by the

'•Op. cit., p. 117.
" Op. cit., p. 204.

>* We shall see below (p. 9) that Morgan makes the same assumption.
'» Hobhouse c.ticH fop. cit., p. 'i.W) only one well authenticated case of the mak-
ing of a tool by an animal. The tool in this case was, as he notes, a very simple one.

8 WALTER S. HUNTER

animal that is supposed to react to a certain stimulus, but, being
on the qui vive, reacts to anything that occurs at the proper
moment. This behavior is familiar in human reaction experi-
ments. If, now, we subtract the possible emotional disturb-
ances (and I doubt whether that even is necessary), we have
the type of case, I believe, that the experiments of Hobhouse

present.

^ 3. Dreams

A third argument used for ideas is the supposed fact of an mal
dreams. The usual criticism of this — to which I subscribe — is
that the law of parsimony forces one to recognise that the inter-
pretation of the facts may as well or better be physiological than

ideational.

4. Learning Curve

A fourth argument used in the discussion as to .the presence
of ideas is the nature of the learning curve. In 1898, Thorn-
dike^" presented data for cats which when plotted gave what the
author termed a "gradual slope." "The gradual slope of the
time curve, — shows the absence of reasoning." It represents
"the wearing smooth of a path in the brain, not the decisions
of a rational consciousness." There seems to be no doubt but
that Thorndike meant that had ideas been present to guide the
reactions that the latter would have succeeded within a few
trials. Hobhouse would seem to agree with- Thorndike that
such a curve as the latter claimed to present was evidence of
the absence of imagery. His criticism of Thorndike is to the
effect that the former's curves are not gradual — "unless the
slope of a church steeple is gradual."

The (apparently) common assumption of these writers has
been questioned effectively both by Watson" and by Hicks and
Carr." The criticism of the latter authors is factual and is
summed up as follows : " Our results indicate that any inference
from such a general characteristic of a curve is not feasible,
because we are dealing with a complex phenomenon due to
several independently variable factors. Our results indicate that

^"Thorndike, E. L. Animal Intelligence. Psy. Rev. Mon. Supp., 1898, vol.
2, p. 45.

^' Watson, J. B. Kinaesthetic and Organic Sensations. Psy. Rev. Mon. Supp.,
1907, vol. 8, pp. 23-4.

^^ Hicks, V. C. and Carr, H. A. Human Reactions in a Maze. Jour, of Animal
Behavior, 1912, vol. 2, pp. 116-118.

DELAYED REACTION 9

the rational status of a group of animals can not be inferred
from the slope of a curve in so far as this slope is dependent
upon the number of trials or the relative rate of elimination.
They indicate, moreover, that inferences as to intelligent status
are legitimate in so far as the slope is determined by the factor
of total values eliminated, but that the relation between the
abruptness of slope and the degree of rational ability is just the
inverse of that assumed by Thorndike and Hobhouse."

5. Memory

Arguments for the existence of ideas, have also been drawn
from behavior purporting to be guided by memory — in the
psychological sense. Let us use an example from Lloyd Mor-
gan. The quotation of a few sentences will adequately represent
his position when the Introduction to Comparative Psychology
was written. " In the first place we may notice that the exist-
ence of memory is implied in the association of ideas ; or rather
in the occurrence of ideas at all." "If, therefore, animals have
ideas at all — ^and if they have not we need not attempt to carry
any further our investigations into zoological psychology — they
must have memory, and there must be in them, as in us, some
anatomical and physiological basis for what is popularly termed
the retention of ideas." " By idea Morgan understands any
centrally aroused conscious process. Of course Thorndike 's
results showed long ago that the presence of ideas in an animal
requires vigorous proof rather than mere assumption. Now for
a concrete example: "When I was at the cape I used to take
my two dogs up the Devil's Peak, an outlying point of Table
Mountain. There were several places at which it was necessary
that I should lift them from ledge to ledge since they could not
scramble up by themselves. After the first ascent they always
remembered these places and waited patiently to be lifted up.
On one of our first ascents one of them put up a young coney
and they both gave chase. Subsequently, they always hurried
on to this spot, and though they never saw another coney there,
reiterated disappointment did not efface the memory of that
first chase, or so it seemed. I think the last time I took them
up must have been about three and a half years after the coney

** Morgan, Lloyd. Intro, to Comp. Psych. London, 1898, p. 117. Thus Mor-

f;an, as we noted for Hobhouse, seems to assume that ideas exist whether they
UDctioD in beliavior or not.

10 WALTER S. HUNTER

hunt: so long had the memory endured, and the association
remained uneffaced." ^*

This is a fairly well known example of the t3^e of proof used
by the "anecdotal psychologists." To some it may seem too
trivial either for serious analysis or notice. But such a judg-
ment is ill informed. We shall find similar arguments as late
as Cole's paper on the Intelligence of Raccoons. The obvious
criticism of Morgan's illustrations is that they may be simply
cases of sensory recognition of the commonest kind. A further
word will be said in connection with the criticism of Cole's work.

6. Thorndike's Test
A great many of the experiments which Thorndike presents
in his recent book on Animal Intelligence^' involve more or less
intimately the question of the existence of images. However,
I shall limit my analysis to the case in which the problem of the
fact of images is most crucially attacked. The case I choose
is the famous one reported in the first monograph of Animal
Intelligence. I shall term it the "hand-clapping test" with
cats. Thorndike's own words are such an excellent example of
scientific description that I shall quote them at length:^* "The
only logical way to go at this question and settle it is, I think, to
find some associations the formation of which requires the pres-
ence of images, of ideas. You have to give an animal a chance
to associate sense -impression A with sense-impression B and
then to associate B with some act C so that the presence of B
in the mind will lead to the performance of C. Presumably the
representation of B, if present, will lead to C just as the sense-
impression B did. Now, if the chance to associate B with A
has been improved, you ought, when the animal is confronted
with the sense-impression A, to get a revival of B and so the
act C. Such a result would, if all chance to associate C with
A had been eliminated, demonstrate the presence of represent-
tions and their associations. I performed such ari experiment
in a form modified so as to make it practicable with my animals
and resources. Unfortunately, this modification spoils the cru-
cial nature of the experiment and robs it of much of its author-
ity. The experiment was as follows:

^ Ibid, p. 118.

2' Thorndike, E. L. Animal Intelligence. New York, 1911.

='8/feid, pp. 110-112.

DELAYED REACTION 11

"A cat was in the big box where they were kept (see p. 90)
very hungry. As I had been for a long time the source of all
food, the cats had grown to watch me very carefully. I sat
during the experiment, about eight feet from the box, and would
at intervals of two minutes clap my hands four times and say,
'I must feed those cats.' Of course the cat would at first feel
no impulse except perhaps to watch me more closely when this
signal was given. After ten seconds had elapsed I would take
a piece of fish, go up to the cage and hold it through the wire
netting, three feet from the floor. The cat would then, of course,
feel the impulse to climb up the front of the cage. In fact,
experience had previously established the habit of climbing
up whene\-er I moved toward the cage, so that in the experi-
ment the cat did not ordinarily wait until I arrived there with
the fish. In this experiment

A=The sense-impression of my movements and voice when

giving the signal
B=The sense-impression of my movements in taking fish,

rising, walking to box, etc.
C==The act of climbing up, with the impulse leading there-
unto.

"The question was whether after a while A would remind
the cat of B, and cause him to do C before he got the sense-
impression of B, that is, before the ten seconds were up. If A
leads to C through a memory of B, animals surely can have
association of ideas proper, and probably often do. Now, as a
fact, after from thirty to sixty trials, the cat does peform C
immediately on being confronted by A or some seconds later,
at all events before B is presented. And it is my present opinion
that their action is to be explained by the presence, through
association, of the idea B. But it is not impossible that A was
associated directly with the impulse to C, although that impulse
was removed from it by ten seconds of time. Such an associa-
tion is, it seems to me, highly improbable, unless the neurosis
of A, and with it the psychosis, continues until the impulse to
C appears But if it does so continue during the ten seconds,
and thus get directly linked to C, we have exactly a represen-
tation, an image, a memory, in the mind for eight of those ten
seconds. It does not help the deni -rs of images to substitute

12 WALTER S. HUNTER

an image of A for an image of B. Yet, unless they do this,
they have to suppose that A comes and goes, and that after ten
seconds C comes, and, passing over the intervening B blank,
willfully chooses out A and associates itself with it. There are
some other considerations regarding the behavior of the cats
from the time the signal was given till they climbed up, which
may be omitted in the hope that it will soon be possible to
perform a decisive experiment. If an observer can make sure
of the animal's attention to a sequence A-B, where B does not
arouse any impulse to act, and then later get the animal to
associate B with C, leaving A out this time, he may then, if
A, when presented anew, arouses C, bid the deniers of represen-
tations to forever hold their peace."

First as to the data obtained, Thorndike's results indicate
only the magnitude of the interval between two stimuli which
association can bridge. Using his symbols, B and C have been
associated before this experiment was begun. The hand clap-
ping, A, now precedes B by ten seconds. At the end of from
thirty to sixty trials, the cat climbs up at A rather than waiting
until B appears. Now must we assume either that the "A"
neurosis, and hence the "A" psychosis, persists or that A has
resulted in the central arousal of B? Not at all. There is good
evidence to show that association in animals can bridge an
interval of ten seconds and more. Nearly all behavior experi-
ments cover at feast ten seconds from the beginning of the
test to the acquisition of food. Yet it is necessary that the
first and the last of the test be associated in order to provide
a motive for the complete reaction. In no case — Thorndike's
not excepted — ^is the ten seconds a sheer gap. (Thorndike did
not describe the behavior of the cats during the interval, although
he did refer to it.) The animal is reacting during the interval.
Motor attitudes at least are present to fill the gap. An animal
as high in the scale as a cat could certainly form this simple
association between a sound and a single reaction within sixty
trials. Moreover, it is to be remembered that only two cats
succeeded within this time. Two others were tested for one
hundred and thirty-five trials, but uniform reactions were, not
secured. The situation would have been quite different had
there been two or more signals, "^A's," and as many different

DELAYED REACTION 13

responses. But even under such conditions, one would still be
stud^'ing the association of stimulus and response. This would
remain true as long as it was the feeding and not the animal's
reactions that were delayed. On the other hand, if both the
feeding and the reactions were delayed after the stimulus had
been given, then if there were such states as ideas one would
expect them to function here. Furthermore the need for ideas
would increase with the number of different stimuli and reactions.
Just a word now concerning Thomdike's formula for the
study of imagery. The fact that he himself was unable to
carry out experiments in conformity with it, and that none
have been carried out since his attempt, does not speak very
well for the formula. I must confess my own failure as yet to
perfect a technique by which the formula might be applied to
animals. In order to associate A and B, it will be necessary
that they be followed by a reaction x. A, B and x are now asso-
ciated. B and C may now be linked together through a second
reaction y. Even granting the ideational character of reactions
carried out according to the formula, one would have to know
the following facts concerning the above test: (i) Did the
animal discriminate between A and B, between B and C, and
between A and C? If the first and second discriminations were
not made, A and B or C would have been directly associated
through X or y. If A and C were not discriminated, the asso-
ciation B-C would have been useless. Difficulties such as these
lead me to believe that the goal aimed at is unattainable. In
fact, Thorndike states that the formula is valid only when B
arouses no impulse to activity; this is the essential weakness
of the formula, for one can never be certain of the absence of
these intervening mediating motor tendencies. In fact their
presence is extremely probable. It cannot be too often reiter-
ated that structural psychology has no place in the study of
animal behavior. One must speak in terms of function. It is
impossible to tell whether an image is present or not. The
most that one can ever say is that some process other than
overt motor activity is present which functions as an image
might in human consciousness. This amounts to an accept-
ance of Hobhouse's statement (although I do not feel that he
always limits himself to this) that the ideas we deal with are

14 WALTER S. HUNTER

"practical ideas," understanding by this a function which does
for animals that which practical ideas do for human behavior."

7. Learning BY Being "Put Through"

Perhaps the most important and best known piece of work
on the presence of imagery in animals is that by Cole on the
Intelligence of Raccoons. Let us consider the evidence which
Cole presents. The argument derived from an animal's learning
a problem by being "put through" may be analyzed first. Cole
writes in particular reference to Thorn dike, saying " It would
seem that nine-tenths of the experimental evidence for the
absence of ideas in dogs and cats comes from their inability to
learn from being put through. "^^ Again, "If inability thus to
learn is evidence against the- presence of ideas, then ability to
do so should be equally strong evidence for it."^* In an earlier
paper, ^' I have discussed some aspects of this problem in the
light of experiments carried out upon the white rat. This phase
of the question need not be gone into more fully here. Whether
or not it seems probable, from a speculative point of view, that
an animal must use "free impulses" or images in order to learn
from being "put through," we need not consider. My conten-
tion in the paper mentioned is that the data so far at hand
do not warrant conclusions as to the presence of imagery.
Furthermore I indicated that the behavior could be explained
better in other terms. Now with reference to that type of
experiment in which the problem learned is that of working
latches rather than climbing into boxes, I believe the data
presented by Cole are conclusive, as far as the facts are con-
cerned. Some raccoons at least appear to learn by being "put
through." Whether all raccoons would do so is, of course,
quite another matter. But given the fact, it does not follow
that one must necessarily interpret it as an evidence of the
presence of images. Cole seems to have carried over this inter-
pretation rather uncritically from Thorndike. The entire pro-
cess can be adequately stated in sensational terms. Certain

" Hobhouse, L. T. Mind in Evolution. New York, 1901, p. 283. That Hob-
house does not limit himself strictly to this may be seen by reading the first few
sentences on p. 284.

^^ Cole, L. W. Concerning the Intelligence of Raccoons. Jour, of Comp. Neur.
and Psych., vol. 17, 1907, p. 249.

'^^ Hunter, W. S. A Note on the Behavior of the White Rat. Jour, of Animal
Behavior, vol. 2, 1912,

DELAYED REACTION 15

stimuli. X, y, and z, e.g., are made prominent by directing the
animal's attention to them. These stimuli occur in connection
^Yith one another and with certain movements, kinaesthetically
reported, and are followed by the acquisition of food. What
could be more natural, then, than that the cognizance of the
stimuli should set off the associated movements? The behavior
noted by Cole bears out this contention. All four raccoons,
both the pair put through and the pair not put through, solved
the problem of escaping from the box (No. 4) by working the
fastenings at one trial in one manner and at another in another
fashion.^" This need mean no more than that several responses,
as opposed to a fixed series, might follow upon certain stimuli.
This would be a higher type of behavior, to be sure, than where
only one response was given, but it would not therefore involve
a new type of conscious process. " If the act which he (the
raccoon) is put through is the one which will remain the easiest
and the most convenient for him throughout the tests, irre-
specti^'e of his position in the box, he will never vary from it.
If not, he will employ your act when his position makes it con-
venient and he is looking at the latch you began with."^' We
are not told whether the raccoon learned which was the easiest
way by trial and error or not. But it is to be inferred from
the behavior of raccoon No. 2 that such was the case." The
behavior thus described is interesting, but entirely inadequate
as far as the presence of imagery is concerned. It may well
be that "animals which, so far as we know at present, are
utterly unable to learn save by innervating their own muscles ' '
are devoid of ideas, without its following that if this type of
learning is present, the animal possesses imagery. Hence assum-
ing the facts that Thorndike and Cole present to be unquestion-
able, it need only follow that the raccoon exhibits more com-
plex sensori-motor behavior than the dog and the cat, and
not that it shows a new type of behavior, i.e., a type of behavior
involving the functional presence of a representative factor.

Cole adduces further evidence for the presence of imagery. ^^
These may be listed as he himself presented them; (a) Recog-
nition of objects; (b) forgetting; (c) variability; (d) associa-

" Op. cit., p. 243.
^' Op. cit., pp. 245-6.
" Op. cit., p. 246.
" Op. cit., p. 251, ff.

16 WALTER S. HUNTER

tion by similarity; (e) reluctance and expectancy; (f) varying
means to the same end, and (g) reactions to colors presented
in a temporal series, (c was treated under 7, above.) Most
of these arguments can be dismissed summarily. It should
never be forgotten that although almost any type of behavior
may involve imagery, the comparative psychologist is seeking
for behavior whose explanation requires the assumption of such
a function, even under the law of parsimony. Is the determin-
ing sensory stimulus present or absent at the moment of re-
sponse ? If it is present, why should the animal use a represen-
tative factor? These are the questions that every investigation
as to the presence of images in animal consciousness must face.

8. Recognition

It seems to me extremely obvious that the fact of the recog-
nition of a food bottle need not be interpreted as presupposing
imagery. In fact it is hard to understand how imagery would
function in such a situation! Recognition of this type does not
necessarily imply memory or the dating of an experience in
one's past. On this basis all animals must be granted the pos-
session of images.

9. Rate op Forgetting

The fact that some of the raccoons forgot the solution of
the boxes after an interval of three days does indeed indicate,
as Cole claims, that automatisms had not been set up. But
one must not infer therefore that images were involved. There
is no factual support for the assumption that imaginal forgetting
is more rapid than sensory. The same is true of the variable
nature of the raccoons' behavior discussed above. This very
probably indicates a high order of adjustive ability on the
sensori-motor level, but not necessarily an " imaginative " ad-
justment. Mere variability of response is present in all animals.
Do all animals, then, possess images?

10. Association by Similarity

"Association by similarity," or the fact that a raccoon will
attack a certain fastening even after its location in the box
has been changed, when contrasted with the activity of cats^*

3* Cole, L. W. Op. cit., p. 253.

DELAYED REACTION 17

and rats" that attack the old position rather than the old fasten-
ing, proves for raccoons only the superior importance of "ob-
jects,"— or the form, size and quality aspects of the stimulus, —
over'kinaesthetic space controls, i.e., the position aspects of the
stimulus. In addition, it should be noted that Davis in his
study of raccoons^" obtained data of the opposite nature. His
animals would claw at the spot where the fastening had been.
But aside from all this, I see no reason why "association by
similarity" should not be purely perceptual and hence be simply
a type of recognition. As a matter of fact, all animals have
responses (instinctive reactions for example) that are applied
to classes of objects. Someone also has well said that animals,
in cases like the present ones, simply fail to see the difference
between two objects and hence react as though the two were

the same.

11. Reluct ANCY and Expectancy

Is the "reluctancy" or the "expectancy" which appears to
be manifested in an animal's behavior toward a difficult and an
easy box respectively to be taken as evidence of the presence
of imagery? Cole, e.g., says "no one who saw the animals
resist being put into a box failed to credit them with a rather
distinct memory of the difficulty of escape.""" By "distinct
memory" Cole undoubtedly means an imaginal process. But
do the facts prove this? Is the case not perfectly amenable to
a "stimulus and response" explanation? The raccoon has asso-
ciated a certain box with a certain displeasure until the presen-
tation of the box arouses immediately the negative reaction.
The raccoon may have had images of his previous experiences,
but the facts do not prove it. One does not need images to
explain this behavior any more than to explain a child's refusal
•to take a second dose of bitter medicine.

12. Varying Means to the Same End
The data presented by Cole under the heading "varying
means to the same end" are just as inconclusive as that pre-
sented above, although they are more suggestive. We have
here the activity of four raccoons directed toward entering a

3s Richard.son, Florence. A Study of the Sensory Control in the Rat. Psych.
Rev. Man. Supp., vol. 12, 1909, p. .'^8, e.g. . . , t . „• a r

3« Davis, H. B. The Raccoon: a Study in Animal Intelligence. Amer. Jour.
PHyrh., vol. 18, 1907, p. 470.

"''Op. cit., p. 2.>i.

18 WALTER S. HUNTER

box containing an apple. The raccoons were accustomed to
reaching through a hole in the top of the box in order to pro-
cure the fruit. When a block with a steeple in it was placed
in the hole, one raccoon immediately clawed out the block and
ate the apple. " She seemed to work as if actuated by a thought
of the apple in the box. It was not done by random clawing,
nor could she smell or otherwise perceive the piece of the apple
in the box."^' We are not informed why the animal could not
smell the apple. The fact that the fruit odor was in the room
will not suffice. But even assuming this to have been con-
trolled, we need not attribute an image of the apple to the
animal. Habit got the raccoon to the hole and started her paw,
and the contact (?) of the staple initiated the claw reflex. This
plus the pleasurable results associated with the box are suffi-
cient to explain the activity. In a slightly different experi-
ment, the animals crawled through a hole in the top of the
box in order to procure the apple. I now quote Cole: "The
box had no bottom and instead of resting directly on the
floor it rested on a row of bricks. Removing one of these made
an opening under the lower edge of the box through which the
raccoon might crawl. The opening in the top was now closed
and nailed fast. No. i was freed, went to the top of the box
and tried to claw out the block. He then walked about the
room, then tried the block again. He then went to the opening
made by removing the brick, stopped a moment, then crawled
in. "3^ To argue that this means image of apple is certainly
naive, at least. Could the raccoon not sense the apple when
its nose was within a foot (see description of box i8, op. cit.,
p. 215) of it? Again where the animal climbed up and over a
roll of poultry wire in order to descend into the box, the possi-
bility of the presence of imagery is only suggested, not proved. •
The opening into the box, as well as the odor of the food, was
there impelling the raccoon to approach. What more natural,
then, than that the animals should climb the wire and thus
reach the food. Such behavior is what would be expected of
raccoons that lived in a wire cage. The case of raccoon No. 4
is somewhat different. With the box which possessed two
openings, he went directly into the lower of the two at . the

" Op. cit., pp. 254-5.
38 Op. cit., p. 255.

DELAYED REACTION 19

first trial. What influence his starting point in approaching the
box had upon his success, we are not told.

Logically, the position taken by Cole in his illustrations
would require him to argue that ideas are present wherever
"motor excess" in learning occurs. There a sensorially reported
situation calls out in succession the animal's repertoire of in-
stinctive and habitual acts. This is a variation of means toward
the attainment of an end, and is on a par with the "variability
of response" argument discussed above.

13. Reactions to a Temporal Series of Colors
It is only fair to Cole to note that his main emphasis does
not rest upon the above data, but upon a series of tests that
he made with colors presented in a temporal series. Three
colors, white, blue and red, were placed upon three levers which
in turn were secured by a single pivot on the back side of a
board one foot high. When the colors were presented in the
order W, B, R, the animal was to secure food by climbing upon
a box. When R, R, R was given, no reaction should be made.
Now since the terminal stimulus was identical in each case.
Cole argues, the only means by which the animals could react
discriminatingly is by remembering what colors of the series
had preceded. The fact that the raccoons clawed up the cards
from behind the screen, reacting only when the proper one
appeared, was also used as evidence of images. Believing that
these tests were almost absolutely uncontrolled and that the
interpretations were invalid as far as the data presented were
concerned, I set two graduate students, F. M. Gregg and C. A.
McPheeters at work upon this problem. Their purpose was:
(i) to duplicate as nearly as possible Cole's results under ade-
quately controlled conditions, and (2) having set up the dis-
crimination, to determine and not to assume its basis. Their
results will soon appear in the Journal of Animal Behavior under
the title vSome Reactions of Raccoons to a Temporal Series of
Stimuli. I shall only note here that they found; (i) that dis-
crimination was not based upon the cards — in fact the discrim-
ination was not even visual, and (2) that practically the entire
discrimination was made on the basis of the first lever and not
on the basis of all levers as Cole assumed. (Since all levers
were influential, in Cole's opinion, it had been necessary to

20 WALTER S. HUNTER

assume that the first two were represented imaginally when the
last was presented.)

The criticisms on Cole's entire work as outlined above reduce
to these: (i) The facts are either inconclusive or irrelevant.
And (2), there is no evidence of adequate controls. On the
positive side, the work suggests that the raccoon is more in-
telligent than the dog and cat, but it does not determine wherein
this superiority lies.

14. Washburn's Cat on the Stairway

There is one other type of behavior that deserves mention.
Again, it was not and need not be interpreted as necessarily
involving the presence of imagery. The illustration follows:
"A cat, indeed, once observed by the writer, did behave as a
human being would do to whom any idea had occurred, when,
on coming into the house for the first time after she had moved
her kittens from an upper story to the ground floor, she started
upstairs to the old nest, stopped halfway up, turned and ran
down to the new one. But errors of interpretation are possible
at every turn of such observations."^'

This is an excellent illustration of the t3rpe of argument that
would use "hesitation" and "wavering" as an evidence for
the presence of ideas. It is a mode of behavior that is found
almost everywhere in animal studies. A rat, e.g., hesitates at
a division point of the maze and finally selects the right path-
way, or it runs half the length of a blind alley and then turns
back. Was it guided by an ideational representation of the
movements to be made and their consequences ? . Not neces-
sarily. Accidental stimuli may have initiated the new reaction
and any conflict present may have been resolved on a purely
sensory -motor level. The experimental technique for the con-
trol of such reactions will be discussed below (see p. 74).

There is very little that needs to be said in the way of a sum-
mary of this historical review! All of the arguments for the
presence of imagery in animals that we have examined have
been found inconclusive. It is not that the various t3^es of
behavior may not have involved a representative factor. The
point is that this possibility is nowhere proved necessary. The
fault does not lie in the exhaustiveness of the data. The vari-

=*» Washburn, M. F. The Animal Mind. New York, 1909, p. 272.

DELAYED REACTION 21

ous methods have been made to yield ample returns for this
purpose. The crux of the matter is that the methods them-
selves are inadequate for the solution of the problem. Let
me re-emphasize the fact that if comparative psychology is to
postuFate a representative factor, it is necessary that the stim-
ulus represented be absent at the moment of response. If it
is not absent, the reaction may be stated in sensori-motor
terms. But in order to know that the stimulus is absent, it
is first necessary to determine carefully what the stimulus is.
None of the methods reviewed, I believe, meet these require-
ments. Whether the tests presented in this monograph do or
not, the reader himself may judge.

III. NOTES ON THE ANIMALS AND CHILDREN TESTED

Four classes of re-agents were used in the experiments whose
description is to follow: white rats, dogs, raccoons (Procyon
lotor) and children. A few words descriptive of these subjects
will not be amiss.

1 . Rats

Twenty-two rats w^ere used during the entire course of the
experiments. Five of these were normal adults and were used
only in preliminary tests in which the purpose was the perfec-
tion of a method. The remaining seventeen (normal) were all
started in the experiments when approximately four weeks old.
All were vigorous, healthy animals whose records may stand
as typical.

2. Dogs

The two dogs tested were mongrels in whom the rat terrier
strain was dominant. They were very bright and intelligent
looking, very active, playful and affectionate, — indeed they
seemed to possess all the qualities that are attributed to intel-
ligent dogs in countless anecdotes. This was the unanimous
testimony of many observers. The two dogs, Blackie and
Brownie, both females of the same litter, were secured from
an a,nimal dealer when they were small puppies and' were started
on the preliminary tests at about the age of five months. They
were usually kept in a kennel out of doors and remained in
excellent condition during the experiments. Of the two, Brownie
was the more aggressive and, to the ordinary observer, appeared
possibly the more intelligent.

22 WALTER S. HUNTER

3. Raccoons

Four raccoons, two males — Bob and Jack — and two females
— Betty and Jill, were tested. Bob and Betty had been pets
and were secured from their owner when about five or six months
old. Jack and Jill were caught in the woods when about two
and a half months old. Preliminary experiments were started
almost immediately with all four. The raccoons were and
remained in perfect health throughout the experiments. The
only physical defects possessed by any of them were the cata-
racts which developed in Bob's eyes about a month after his
purchase. Although this interfered with his accurate vision of
objects and resulted in his colliding frequently with them, he
was able to distinguish such changes in brightness as were
necessary in the experiments. This fact was demonstrated
conclusively by many control tests which will be described
later. Of the four, Betty was the quietest and most timid.
She was the least promising subject among them. Further
facts about these animals are given in the appendix.

4. Children

Five children were used in the course of the present tests:
two boys, Hd and L, and three girls, F, M, and H. H, Hd
and L were each approximately six years old. M was about
eight years old; and F, about two and a half. Hd and L were
in kindergarten work, and M and H were in the graded schools.
The indications were that they were children both of normal
ability and of normal intellectual advancement for their ages.
F was a bright little girl and made an excellent subject. All
of the children were more or less timid at first; but this was
overcome, in all save possibly H's case, before tests were begun.
Particular pains were taken with F. The experimenter was in
her company a great deal, and by the beginning of the tests
was a gladly accepted play-fellow.

IV. APPARATUS AND GENERAL METHOD

The plan of box A is presented in Fig. i. (This box was used
for the raccoons.) The box is made of ^" boards and is 2^' high
with doors 7I" wide and 13" high. The light stimulus came
from 3 c.p. 8 volt miniature carbon lamps, so wired that they
might be switched on one at a time. The current was obtained

DELAYED REACTION

23

from a 220 volt lighting circuit and was passed through a lamp
rheostat before reaching the discrimination box. The release
box R was raised by means of a cord passed over a pulley in
the ceiling and back to the experimenter at E. The first release
box had glass over the top and sides. The right and left faces
of the box were i2"xi5^". The front was 'jV^'^sV- With
this release box the distances to the entrances of the three light
boxes (L) were unequal. Those at the sides were each 19I", while
the distance straight in front was 20^". These various inequal-

FiGURE 1. Ground plan of Box A

ities were due to two causes: (i) The box had been planned
originally for a different type of test and was only later arranged
for the present experiment. (2) The release box, being covered
with glass would have been too heavy to handle had it been
made larger. During the course of the experiments, another
release was made. This one was covered with wire of Y' mesh.
Its sides were i4i"x 15^"; and its front was 9^" x 15^". The
distances to the light boxes were now equal and of the dimen-
sions indicated in Fig. i .

vSliding doors were placed at the points marked D in the
figure. They were controlled by strings which ran from them,

24

WALTER S. HUNTER

through screw-eyes on the top of the box, to the experimenter
at E.

A i6 c.p. Hght was suspended about four feet from the floor
over the center of the apparatus. Its intensity was diminished
approximately by two-thirds by wrapping the bulb in cloth.
The three light boxes were covered in order to prevent light
from entering them from above. Part of the entrance box
leading to the release, R, was covered by the switch-board, S,
and the remainder by paste board. This prevented the animals
from watching the experimenter.

Figure 2. Ground plan of Box C. D, exit door.

Box B was, in principle, like Box A. It was used in testing
the dogs. Its dimensions differed from those of A. Instead
of being 90 degrees, the extreme lateral angles were 75 degrees.
The angles on each side of the middle light box were about
145 degrees as opposed to 127 for Box A. Box B was only
2' high, and the length of its entrance box was i' 4". The re-
lease box was covered with wire, and each face was i' x 10".
The distances from the release boxes to the light boxes were
each \\' .

Fig. 2 is a ground plan of Box C. This box was used for
rats. In addition to the data there given, the following points
should be noted: The release box, R, was fastened by hinges

DELAYED REACTION

25

so that when it was raised the three faces cleared the floor prac-
tically the same distance. The faces and the front half of the
top of R are covered with glass ; the rest is of wood. The doors
leading into the light boxes are 4" x 3". Those leading out of
the light boxes, called exit doors here, are 3" x 2 V'. The switches
at S' are for the lights which are of the same intensity and
wiring arrangement as those in the two preceding boxes. The
switches at S turn the current into any or all cf the paths from
the release box to the light boxes. This current is obtained
from a dry cell and is passed through the primary coil of a Porter
inductorium. The strength of the current passing into the prob-

I'KiuUK .3. Box ('

lem box was regulated so that the animals ne\''er became fright-
ened by severe shocks. The current was never strong enough
to cause the animals to squeak and only rarely did they attempt
to jump over the strips. The brass strips from which the shocks
were obtained (only one group, marked B, is shown in this
figure), were very thin and lay flat on the lloor of the box.
Before being tacked down, they were given an acid l)ath ^^■hich
destroyed their lustre but left their conduct! \ity 'practically
unaffected for my purpose.

Fig. 3 should give a clear ])resentati()n of the essentials of
this box as well as of the others usetl in the course of this
research.

26

WALTER S. HUNTER

Box D, also used for rats, is similar to Box C in all save two
respects: (i) It was not wired for punishment. (2) The doors
leading from the light boxes could be closed with wooden slides.
The use of these slides was discontinued shortly after the ex-
periments began. Pieces of wire mesh were then used. These
admitted the light and thus offered less opportunity for the
animals to tell which box was open and which was closed.

The apparatus (Fig. 4) which was used with children was
constructed on the same principle as that described above for
the other problem boxes. Three boards, each one foot square.

L

T

Figure 4. Ground plan of apparatus used with children. D, door; W, window

were placed against the wall of a room 12' x 14'. The middle
square was seven feet distant from the release box, R. On
each of these boards was mounted: (i) A 4 c.p. miniature
electric light, L; and (2) just below the light, a push button.
All three boards were painted black and were exact duplicates
the one of the other. The front. A, of the release box was a
lever which could be raised by a handle at P. The experimenter
usually sat at E. T is a table which held: (i) The candy used
as a reward in the tests ; (2 ) the switches regulating the lights
and the buzzer; and (3) the buzzer. C is a curtain which hid
the experimenter from the subject's view when the latter had

DELAYED REACTION 27

left the release box. At first this curtain was also continued
between R and T. Subsequently, this was found to be both
inconvenient and unnecessary and its use was discontinued.
The apparatus was wired so that any light could be turned on
at will and so that any push button could be connected with
the buzzer. Moving the switches was done without the sub-
ject's knowledge. The light was always turned on over the
button that rang the buzzer. The child's problem was to find
this button at the first trial when the light was on (in the
learning series) and then (in the delayed reactions) after the
light had been turned off for a certain interval of time.

The general method of experimentation was as follows: The
animal to be tested was put in the release box, R, of problem
Box C, for example. If, now, the lighted box is the one on the
left, the exit doors of the others are closed and the switches at
S are so set that if the experimenter close the switch at S", the
animal will receive a shock if it steps on either the strips leading
to the box on the right or on those leading to the one in front.
The light is then turned on in the left box. The animal is re-
leased after five seconds, the time being taken with a metro-
nome. A careful, detailed record is kept of the direction in
which the animal is oriented'" when released and of just where
it goes after being released. In the case of the animals used in
Boxes A, B and C, they should go straight to the lighted box,
out through the exit doors and back to the entrance of the
release box where they are fed. The rats used in Box D were
fed a small morsel of bread and milk at the exit doors of the
lighted boxes. Theoretically the olfactory control was not so
good here as where no food at all was given in the light boxes.
Practically, there was no difference. The rat was given only
a bite, so almost no food fell on the floor; all the boxes were
used an equal number of times ; and all were frequently washed
out. Whatever odor was present was so distributed as to afford
no appreciable basis for discrimination. The results obtained
with these rats, when compared with those where the olfactory
control was bettTer, support this statement. In any event,

*" Orientation.s are spoken of as right and wrong, irrespective of whether an
aninnal may \)e said to defiend on them as cues or not. When the orientation is
"right," the animal is headed toward the proj^er box. Any other orientation is
"wrong."

28 WALTER S. HUNTER

olfactory inequalities would persist after the light was turned
out and would aid in delayed reactions only if associated with
the light. More attention will be given to this possibility later
in the discussion.

After the animal had been trained until it chose the lighted
box almost perfectly, delays were begun. The light was turned
off just as the animal reached the box. This was called the
first stage of delay. At the second stage, the light was turned
out when the animal was half way to the box. At the third
stage, the light was turned out just as the experimenter started
to raise the release box. Here there was a genuine delay, al-
though a small one. The first two stages served primarily to
adapt the animal emotionally to the sudden change from light
to darkness. The rats and dogs usually ran so fast that their
momentum was sufficient to carry them into the box when once
they were started toward it. In any case they only needed to
continue in the direction in which they were going. This, how-
ever, was not the case with the third stage. The light was put
out before the animal started. Throughout these three stages,
the animal was released promptly at the end of five seconds.
From this stage on, where the animal was detained one or more
seconds after the light was out before being released, it was
obviously necessary to let the animal see the light before this
was turned off. Occasionally, the interval thus required was
more than five seconds. In these higher stages of delay, I
always waited until I felt sure that the animal had seen the
light, and then turned off the current while the animal was
still oriented toward the source cf light. Record was kept of
any change in the orientation which an animal made after the
light was turned off. How detailed these records were will be
seen in the section on experimental results.

The delays were gradually increased in length until one was
found at which the animal failed. They were then decreased
until the animal was again making a high percentage of correct
choices, when the intervals were again increased. An animal
was thus tested twice for the limit of its ability to delay with
the backgrounds surrounding the entrances to the light boxes
all similar the one to the other. When this limit was found,
the wall of the box about the entrance to c was covered with
white cardboard; that about h, with a black; and that about

DELAYED REACTION 29

a,*' with a medium gray. If the animal's limit of delay was
no better or was worse with this arrangement than before, the
animal was dropped from the experiment. If the limit were
better, the different backgrounds were removed and the sim-
ilar ones used again. The limit of delay with these was then
re-determined. If this third limit were greater than the second,
the effect of training could be evoked as an explanation of the
fact. But if it were markedly less than the second, the only
cause could be an association between the backgrounds and the
lights. The significance of this type of association will be dwelt
upon in detail later in this paper.

One more point in general method remains to be considered.
This is the question of what percentage of correct choices shall
be taken as sufficient to justify further increase in the interval'
of delay. With three discrimination boxes, pure chance would
lead the animal to make 33i%of correct reactions out of a long
series of presentations. But series used in experimentation are
very rarely long enough for chance to operate as theory de-
mands. Besides there are various other influences which enter
in to determine an animal's behavior above and beyond the
influence of the stimulus proper to the test. One such influence
is the position factor. Try as I might to eliminate this, most
of the animals acquired at various times during the experi-
ments more or less pronounced preferences for certain boxes.
And these preferences varied from animal to animal. Some-
times they were so strong that the regular tests had to be
stopped until the habit could be broken up. In the light of
this, although on the whole each box was presented to an
animal an equal number of times, in any one stage of the tests
such an equahty might not be present. Thus an equality of
percentages among the various animals, in this case, would not
mean that they knew the problem equally well. Again hesi-
tancies and waverings must be noted in estimating how well
the animal is grounded in his appointed task. Further state-
ments concerning the value to be attached to the percentages
appear later in the paper (pp. 44, 46 and at intervals thereafter.)
Considerations such as these render it highly inadvisable to lay
down a rigid standard as to the number of trials and the per-

*' a, b, and c will be used in this paper to designate the right, middle and left
light boxes respectively.

30 WALTER S. HUNTER

centage of correct reactions to be required of all animals. As
will be seen in the following experiments some animals were-
advanced from stage to stage when but 70% of correct reactions
were made. Others were detained on one stage until 85 % or
95% were made. Sometimes an animal was given 5 or 10 trials,
sometimes 150 or 200 trials on one stage of delay. One general
fact, however, should be noted: The first set of rats. Bob and
Betty during their first year's work, and H, L, M, and Hd of the
children were rushed from stage to stage as fast as was at all
feasible. The emphasis here lay upon what the subjects could
do with their native equipment. With the other subjects used
and with Bob and Betty during the second year's work, a larger
number of trials were given on each stage of delay and the
advance in the length of the periods was more gradual. The
emphasis here was placed on what the subjects could learn to do.

V. EXPERIMENTAL RESULTS
1. Tests with Animals

A. Learning the Association, (a) Rats. — The method of ex-
perimentation used by Hough and Reed was retested with the
result that the general method outlined above was adopted.
These tests were made upon five normal adult rats, Nos. 18-22.

The regular experiments may be divided into two sets: (A)
Those in which reward only was employed; and (B) those in
which both punishment and reward were used. Nine rats,
Nos. 1-9 were tested in the first set and eight, Nos. 10-17, i^^
the second set. Of the rats in A, Nos. i, 3 and 8 were in poor
health, and their records will not be considered.

Table I gives the number of trials required by the rats of
Set A to learn the association between the light and getting
food. These rats were given 5 trials daily. All save No. 6
learned the association. Of the last 50 trials given this animal,
only 40% were successful; while of the entire 800, 54% were
correct.

TABLE I

No.

, of trials

Rat

on

learning

No. 2

176

No. 4

175

No. 5

505

No. 6

800

No. 7

361

No. 9

280

DELAYED REACTION 31

The rats of Set B were tested as follows: Nos. lo, ii, 14 and
17 were given 5 trials daily. Nos. 12, 13, 15 and 16 received 10
trials. All of these animals learned the association. The num-
ber of trials required for this learning is given in Table II. The

Five

TABLE I]

trials

Ten trials

Rat

No. of trials
on learning

No. of trials
Rat on learning

No. 10
No. 11
No. 14
No. 17

165
160
200
175

No. 12 440
No. 13 250
No. 15 220
No. 16 480

results here indicate: (i) That the use of 5 trials favors rapid
learning more than does the use of 10 trials, and (2) that rats
tested with 5 trials daily under conditions of reward and pun-
ishment will learn the association more nearly in the same
length of time than will rats given the same number of trials
but tested with reward only. The use of punishment, while it
may not shorten the length of time required by an "intelligent"
rat to learn an association does hasten the learning in the case
of the "dullards" such as Nos. 5, 6, 7 and 9. Rats will put
forth a maximum effort under punishment when they would
not do so under other conditions. These results are in harmony
with the work of Hoge and Stocking."

That the influences on time of learning exercised by the
number of trials and by punishment are not due to prefer-
ences for the light or for the dark room boxes is indicated by
Table III which gives the number of times each rat chose the

TABLE III
Rat No. 2 No. 4 No. 5 No. 6 No. 7 No. 9

No. of choices of light box. . 23 21 13 18 23 38

Rat No. 10 No. 11 No. 14 No. 17

No. of choices of light box . . 20 28 28 32

Rat No. 16 No. 15 No. 13 No. 12

No. of choices of light box . . 24 30 21 27

" Hoge, M. A. and Stocking, R. J. A Note on the Relative Value of Punish-
ment and Reward as Motives. Jour, of Animal Behavior, 1912, vol. 2.

32 WALTER S. HUNTER

lighted box during the first 50 trials. The data given for the
rats of each group were obtained under similar conditions (for
our present purpose) and are strictly comparable. The only
cases of marked preferences occur with rats Nos. 5 and 9. No.
5's preferences for the dark boxes will help to account for his
slow learning; but it would seem that No. 9 should have learned
very rapidly owing to its marked preference for the light box.
With the exception of these two rats, all the animals were on
an essential par as to preferences. Hence the differences noted
in their learning periods must have been due to the different
conditions under which they worked. Accidental individual
variations are not the causes of these differences because of
the number of rats used.

(b) Dogs. — ^The two dogs, Blackie and Brownie, were given
10 trials daily in Box B under conditions of reward only. The
former animal was given 560 trials on the association. Of
these, 408 trials or 72% were correct. Brownie was given 650
trials on learning the association. Of these, 396 trials or 60%
were correct. Ninety-five per cent of the last 100 trials were
successful. The relatively poor showing made by this dog was
due to the acquisition of habits that interfered with the proper
reaction. Twice during the learning series, extra tests had to
be given in order to break up a discrimination now against the
right and middle boxes, now against the left and middle boxes.
In general, it may be said of both dogs that their rate of learn-
ing is no better than that of the rats. Indeed it is much worse
than any of the rats tested with reward and punishment and
10 trials daily. The only rat that learned the reaction that
did not do much better than the dogs was No. 5. I would sug-
gest that the difference is due to the dogs' "helplessness" when
deprived of cues from the experimenter. The following para-
graph elaborates this point.

There are several interesting points relative to the dogs'
behavior in the learning series that deserve mention. Indeed
to a large extent, they are typical of their behavior throughout
the entire experiment. When the dogs were first put into the
problem box, instead of attempting to get out, they merely sat
down and howled. They were out of sight of the experimenter,
it will be remembered, and when placed in such conditions
seemed quite helpless. They gradually overcame this timidity,

'■ DELAYED REACTION 33

helplessness or lonesomeness but, as will be seen in the delayed
reaction tests to be described below, they never revealed a
high order of resourcefulness. We shall have further occasion
to see that their behavior was almost, if not entirely, on a par
with that of the rats. In learning the association, e.g., the
dogs would go back and forth between some two boxes for
many trials before investigating the third box. Often they
would stop and look around in an apparent attempt to find
the experimenter. This was never done by the rats and rac-
coons. Again, the brown dog fell into the most absurd habit
of going out of the release box, turning entirely around to the
left and then going to one of the boxes. This accomplished
no end that I could determine, yet it was persisted in for some
weeks.

(c) Raccoons. — Bob and Betty were each given lo trials
daily. Jack and Jill received 15. Reward only was used, al-
though running into the wrong box and having to back out
constituted no little punishment here as was also the case with
the dogs. The experimenter was practically out of sight of
the animals all of the time and absolutely so in certain control
tests. It will be remembered that the raccoon tests were made
in a room illumined by a single light suspended above the prob-
lem box. (See p. 24). This left the remainder of the room dark
and thus helped conceal the experimenter.

Bob required 120 trials to learn the association. Of these,
93 trials or 77% were correct. Of the last 50 trials, 96% were
successful. Betty was given 340 trials in the learning series.
The work was not properly controlled for her until the last 120
of these trials. She formed the habit early of looking into the
light boxes and seeing whether or not the wooden exit doors
were closed. This difficulty was finally obviated by using coarse
mesh wire for the doors. These could not be sensed until the
animal practically touched them. Of Betty's last 50 trials on
learning, only 79% were correct. These tests should have been
continued longer. Jack was given 540 trials on learning. Of
these, 428 trials or 79% were correct. Of the last 150 trials,
98% succeeded. Jill received 825 trials during the learning
period. Of these, 631 trials or 76% were correct. Of the last
150 trials, 99% succeeded. (The experiment box had wire
exit doors when Jack and Jill were tested.)

34 WALTER S. HUNTER

Although the nature of the learning process is not our spe-
cific problem, the differences in the total number of trials re-
quired for learning by Bob and the last two animals are so
great as to invite comment. The conditions under which the
records were obtained differ in the following points: (i) Jack
and Jill were younger than Bob by at least a year. (2) Work
was begun with them in July and with Bob in (the preceding)
October. This brought him nearer the period of hibernation
when his appetite would begin to fail. (3) Jack and Jill received
more trials daily and also were tested for more days. Con-
sidering the fact that when the experiments ceased Jack and
Jill were younger than the other two raccoons were when the
latter started and yet had accomplished as much, I do not
believe that age was an important factor in determining the
reactions. However, the actual fact of an age difference re-
mains. The second point should have little explanatory force,
since the animals were always eager to work. As to the third
point, one would expect this factor to work in a direction oppo-
site to that which the results indicate. However, such was not
true with the rats, and it may be that here also the lesser num-
ber of trials daily is more favorable to rapid learning. The fact
of a possible difference of preference for light and dark remains
to be considered. Of the first 50 trials. Bob made 56% correct.
Of the first 45 trials, Jill made 22% correct. Jack made 48%
successful responses out of the first 45 trials. Age differences,
differences of brightness preference and differences in the num-
ber of daily trials . are possible explanations for the varying
lengths of the learning records. Innate (?) individual abilities
must also be recognized. (Betty's results are not considered
because of their unreliability at this point of the experiments.)

Summary. — The only comparative statement with reference
to the learning periods of the different groups of animals
that the facts warrant is this : The raccoons Jack and Jill and
both dogs belong together in the class that learned most slowly.
The second class with reference to speed of learning is com-
posed of the rats. Bob is in a class by himself. He learned
more rapidly than any of the other animals. Further comments
upon this aspect of the learning process will be deferred to the
section on children.

DELAYED REACTION 35

B. Maximal Intervals of Delay Attained, (a) Group differ-
ences as to maximal delay. — The only conditions of experimen-
tation that varied from animal group to animal group were
these; (i) Number of trials daily. Rats 2, 4, 5, 6, 7 and 9
(Set A) and rats 10, 11 ,14 and 17 (of Set B) were given 5 trials
daily. Rats 12, 13, 15 and 16 (of Set B) received 10 trials.
The two dogs were given 10 trials daily. Of the raccoons, Jack
and Jill were each given 15 trials and Bob and Betty 10 trials
daily. (2) The rats of Set B were tested with reward alone.
As was mentioned above, the dogs and raccoons were very
much discomfited by having to back out of the light boxes
when a wrong choice had been made. This was even more
true in our tests than in most discrimination work owing to
the small size of the light boxes.

TABLE IV

Per cent of
Rat Maximal delay correct response

Set A

2

1 sec.

64

4

1 sec.

52

5

3rd stage

52

6

did not learn association

7

3 sees.

56

9

10 sees.

72

SetB

10

1 sec.

76

11

1 sec.

64

*12

1 sec.

72

*13

4 sees.

88

14

3 sees.

80

*15

1 sec.

86

*16

1 sec.

50

17

1 sec.

37

Set A tested with reward.
Set B tested with reward and punishment.

Those rats that are marked with a * received 10 trials daily. All the others re-
ceived 5.

Table IV gives the maximal delay attained by the rats. The
percentages are computed on a basis of 25 trials. It will be
seen from this that only four rats of the fourteen tested reached
intervals greater than one second. The percentage of correct
behavior for one of these (No. .7) was only 56. Rats Nos. 4,
II, 15, 16 and 17 were tested with a choice of two boxes as
opposed to three. This was done only after the animals had

36 WALTER S. HUNTER

more or less completely broken down in their reactions to one
box. Tests with two boxes were never made on any subject,
unless the animal lost the cue to one box. It was this box
that was dropped from the series. The maximum delays made
by the rats under these conditions were increased. Nos. ii,
15 and 16 each made 5 sees, at from 76 to 90%. This must
mean that it is easier to use two cues than three. Further
comments will be added below (see p. 39).

Of the dogs, Blackie made 10 sees, at 76% for 30 trials. She
then lost the cue to one of the boxes and was unable to react
successfully to that one. When tested with two boxes she
finally made a delay of 5 mins. for 5 trials at 80%; 86% had
been made at 3 mins. for 30 trials. Brownie was tested only
with three boxes. She made a delay of i sec. for 50 trials at
90% and 2 sees, for .70 trials at 68%. Both dogs were tested
for eight months. The results should be representative of what
dogs can do under the present conditions.

Of the raccoons. Jack made 14 sees, at 70% for 30 trials.
Tested with two boxes, he finally made 85% at 20 sees, for 40
trials. He was tested for eight and one -half months. Betty
made 82% at 7 sees, for 50 trials. Tested with two boxes, she
reached 10 sees, at 86% for 30 trials. She was tested for one
year and three weeks. Jill was tested only with three boxes.
She made 3 sees, for 45 trials at 93%. The tests extended
over a period of seven months. Bob made 90% for 10 trials
at 15 sees. Tested with two boxes, he reached a delay of 25
sees, for 20 trials at 90%. Bob was tested steadily for a
year and five months.

For fear that some critic would urge that the animals could
not have delayed longer than they did even though the light
had been constantly present, control tests were ttiade as fol-
lows: The animals were taken at the stage when they had
just broken down at some (for them) long interval of delay
and were held in the release for one minute with the light on.
When they were released, the light was still left on. The results
show that the animals made a very high percentage of correct
behavior. Their failure to make correct long delayed reactions
must have been due, then, to their inability to use some cue
by which to guide their reactions after such an interval and not
to mere restlessness caused by the long confinement.

DELAYED REACTION 37

Appendix A contains tables that show all the regular tests
given to the raccoons Bob and Jill and to rats 4 and 16. The
data are there given in the order in which they were obtained.
They are typical of the results of all the animals.

There is so much individual variation within the three groups

of animals whose results have just been given that any exact

correlation between length of delay and groups of animals is

unwise. It is to be borne in mind moreover that it is not the

length of delay but the methods of reaction after delay to which

the greater importance attaches. This topic will be discussed

later. To give some further idea, however, as to the relation

between the groups Table V gives the longest and the shortest

delay made by each class of animals. This table ignores the

different conditions under which the delays were obtained. It

presents the maxima and minima of the best reactions that the

individual animals of a group were able to make under the

present conditions.

TABLE V

Subjects Min. delay Max. delay

Rats Either no learning or 3rd stage 10 sees.

Dogs 2 sees. 5 mins.

Raccoons 3 sees. ' 25 sees.

(b) Effect of size of release upon interval of delay. — The
experiments so far described were all made with the small
release described in the section on apparatus. This release con-
fined the animal's activities to a small part of the apparatus.
It was thought if a release was used which would give the
animal the freedom of the whole interior of the box, that not
only might the maximal interval of delay be increased, but the
animal might reveal more clearly its method of solving the
problem, — indeed it might even develop a new and higher
type of behavior in response to the more complex situation.
These latter possibilities, we shall consider below.

The new release that was used fitted just inside the openings
into the light boxes. It was made of a continuous piece of wire
netting, thus making possible the simultaneous presentation of
all three boxes. It was now possible for the animal to go over
to the door of the lighted box and wait there during the interval
of delay. The light was always left on until the animal had
reached a position immediately in front of the lighted box.

38

WALTER S. HUNTER

Preliminary tests insured an absence of emotional disturbances
in the reactions.

The animals tested under these conditions were: Rats 13,

15, 16 and 17 (No. 17 was the only one tested with a choice of

•three boxes) ; both dogs (Brownie alone was tested with a

choice of three boxes) ; raccoons Jack, Jill and Bob (Jill alone

was tested with three boxes).

Table 6 gives a comparative statement of the animals' abil-
ities with large and small release. Both records for each animal
are for discriminations with the same number of boxes. It
will be seen from this that rat No. 13, the dog Blackie and the
raccoons Jack and Bob are the only ones that failed to delay
longer with the large release. This fact can be readily explained
in the case of Jack and Blackie. The experimenter was prim-
arily interested in discovering whether any new mode of be-
havior would appear under these conditions. When this ques-
tion was answered, tests were discontinued with all the animals
although in the case of Jack and Blackie the interval of delay
was being steadily increased. The increase in delays made by
the other animals is to be correlated with the changed condi-
tions of experimentation. This point will be amplified when
the detailed behavior of the animals is discussed.

TABLE VI

Small release

Large release

Per cent of

Per cent of

Animal

Delay

correct reactions

Delay

correct reactions

Rat No. 13

4 sees.

88

3 sees.

95

Rat No. 15

1 sec.

86

6 sees.

86

Rat No. 16

1 sec!

50

9 sees.

82

Rat No. 17

7 sees.

68

1 sec.

37

Blackie

5 mins

80

1 min.

80

Brownie

' 2 sees.

68

6 sees.

96

Jill

3 sees.

93

7 sees.

80

Jack

20 sees.

85

15 sees.

88

Bob

25 sees.

90

20 sees.

76

(c) Effect of backgrounds of different brightnesses upon the
interval of delay. — When the apparatus for the animals was
constructed, every effort was made to secure qualitative simi-
larity in the three light boxes. The backgrounds surrounding
the entrances to the light boxes were all of the same brightness.
During the course of the tests, the attempt was made to deter-

DELAYED REACTION 39

mine whether the animals could form an association between the
lights and some constant marked differences in the external
environment. In order to test this, three backgrounds of widely
differing degrees of brightness were used as described above in
the section on method. The animals tested under these condi-
tions were: Rats Nos. 2, 4, 5 and 7 of Set A and Nos. 10, 12
and 16 of Set B; the dog Brownie; and the raccoon Jill.

The results of this test were entirely negative. No case was
found where the animal made use of the different backgrounds
as cues for guiding its reactions. One or two of the animals
increased their interval of delay under the new conditions.
However, when the old conditions of similar backgrounds were
replaced, the long intervals of delay still continued. This ind -
Gated that the improvement was due to practice.

(d) Effect of number of boxes upon delay. — As stated above
in section (a), whenever an animal's reactions broke down upon
a particular box, that box was dropped from the tests and the
discrimination confined to the other two. The following ani-
mals were tested under these changed conditions: Rats 4, 11,
1 5 and 1 6 ; the dog Blackie ; and the raccoons Jack, Bob and
Betty. The results are given for convenience above under
section (a). Here it may be noted that all the animals delayed
longer with two than with three boxes. This can be readily
explained on the basis of the complexity of the problem. The
maximum delay for any animal is decided by the accuracy of
its response. With two boxes, this accuracy was increased and
hence the maximum delay recorded was greater.

(e) Effect of other conditions upon delay. — The results do not
indicate certainly any effect of punishment or of the number
of trials upon the length of the interval of delay.

The results of this section indicate that the following factors
influence the maximal amount of delay: (i) Different groups
of animals; (2) size of release, and (3) the number of light
boxes used. The following factors do not influence the amounts
of delay: (i) Punishment and reward; (2) number of trials
daily, and (3) backgrounds of different brightnesses.

C. Methods of Reaction After Delay Used by the Animals. —
There are three different methods of delay which might have
appeared and in point of fact did appear in our delayed reac-
tion experiments: (i) The animal may maintain an orientation

40 WALTER S. HUNTER

of all or part of its body during the interval of delay, i.e., it
may keep its head or even its whole body pointing toward a
certain box. (2) There would be the negative side of this,
where the experimenter could detect no orientation cues used
by the animal. In this case, no observable part of the animal's
body would remain in a constant position. (3) The animals
might rely upon position in the box for their cues, i.e., they
might actually go nearer to one box than to the others and
then wait to be released. (4) Any combination of these three
methods might occur. The discussion of methods i and 2
(orientation and non-orientation cues) will be combined and
will be followed by a consideration of method 3. The actual
existence of method 4 will be considered as occasion demands.

(a) Orientation of whole or part of body. — In addition to
what was given above in the section on method concerning
orientation, it will be well to make such additional comment
here as will indicate clearly the nature of the data secured.
Great pains were taken to insure accuracy and consistency in
the recording of orientations. Needless to say in such pro-
longed tests as the present ones, the experimenter soon becomes
expert in deciding whether an animal's movements are to be
interpreted as a change in orientation. Before an animal has
been tested long, the experimenter can pick out a certain range
of movement and call this the orientation toward a certain box.
The animal (dogs excepted) was in constant motion, but so
long as its activity was directed toward any one face of the re-
lease box, the orientation was recorded as unchanged. There
would seem to be some chance for doubtful cases when the
animal was pointed halfway between any two boxes. These
cases were never counted as changes in orientation. Record
was kept not only of the body position, but of whether any
observable part of the animal remained in a constant position
during the delay. Further note was taken of the gross amount
of the loss of orientation — i.e., whether the animal turned clear
around or not; and if not, then how far around — ^and of just
which reactions were preceded by apparently identical orien-
tations. The data were recorded quickly and easily by the use
of symbols.

Every rat at the moment of release, went in the direction of
his bodily orientation in 99 cases out of 100. (At times this

DELAYED REACTION 41

was not true because of position habits formed by the animal.)
(For the present purposes, it makes no difference whether the
reaction was correct or not.) The data supporting this state-
ment are so overwhelming that they need not be given here in
detail. The rat, when put into the release box during the de-
layed reactions, oriented immediately to the light with its entire
body and began a series of attacks on that side of the box in
an effort to get out. This attempt was kept up until the animal
was released, whereupon it went to the box that was straight
in front. Experiment served only to lengthen the period during
which they would attack any one side of the box. These state-
ments hold true for all rats.

Both dogs were dependent upon orientation for the guidance
of their successful reactions. They only differed from one an-
other in the length of time during which they could maintain
a certain orientation. The dogs differed from the rats in that
the determining cue was the direction of the head rather than
of the body. For the sake of concrete material illustrative of
this type of reaction, I shall give a summary of typical reac-
tions made by Blackie. Of 770 trials, given during a period of
two months, on delays less than 3 sees, long, 141 were unsuc-
cessful. On 116 of the 141, the dog had the wrong orientation
at the moment of release and followed it. On the remaining
25 reactions, the dog failed to follow its orientation and was
wrong. On 8 trials the dog had the wTong orientation at the
moment of release, i.e., was not headed toward the proper box,
and yet reacted correctly. However, only 3 of these trials
were. with a 2 sees, delay and may have been due to chance.
The remaining 5 trials were at the second stage of delay where
the light was on until the animal was halfway to it. Obviously,
these 5 reactions signify no great ability. These results indicate
that just to the extent that the dog was able to hold the proper
orientation during the delay, just to that extent it was capable
of reacting correctly.

Let us take another typical set of results from the same dog
obtained on delays from 1 5 sees, to 5 mins. extending over 30
days. Two hundred and eighty-five trials in all were given of
which 37 were incorrect. In all 37 trials, the dog had the wrong
orientation and followed it. Only once did she have the wrong
orientation and react correctly. When Blackie entered the

42 WALTER S. HUNTER

release box, she would turn clear around and then, before the
final breakdown came, would lay down facing the light. Occa-
sionally she would be distracted by some accidental noise inside
or outside the laboratory. In the great majority of these cases,
she only turned her head. A few times she got up and turned
clear around. But in any event, if she did not recover the
proper orientation, the reaction failed. If she could have lost
the orientation either completely, or almost so, and then have
returned to it and have reacted correctly, the fact would have
been strong evidence for believing that the dog recognized the
proper orientation when it was reinstated. But taking the last
125 trials as typical, and classifying the instances where orien-
tation was lost, recovered and a correct reaction made, it is
found that 33 times the orientation was changed but slightly
and three times completely changed. — By a slight change is
meant that the dog turned her head and not her body. By a
complete change is meant that the animal turned completely
around. — Of the three cases of this latter behavior, one was
very probably due to chance. The other two occurred on the
second day of the 5 min. delays when the final breakdown
was beginning. Moreover, the orientation that was lost and
recovered, — in one case at the end of 36 sees.; in the other at
the end of 4^ mins., — ^was the orientation toward the box at
the right. The reaction is thus not significant of some higher
process, for on the following day. the box at the right was the
only one to which the animal responded. The recovered orien-
tation, therefore, most probably indicates solely the growth of
the habit that, on the following day, resulted in the complete
disintegration of the reaction.

It is interesting in this connection to trace the change that
occurred in Blackie's behavior from the beginning to the end
of the experiments. It has already been noted how "helpless"
and inactive both dogs were when the tests were started. After
the work had progressed for several weeks and Blackie had
become quiet and attentive, she would stand in the release
box on all four feet and occasionally paw the wire in the direc^
tion of the light. A little later, she sat on her haunches during
the retention in the release, but still clawed at the wire in a very
calm manner. Toward the last of the tests described above*
when the delays were growing rapidly longer, she lay flat down

DELAYED REACTION 43

on the floor with her head pointed, usually, toward the light.
But she did not lie quietly. The long delay seemed very trying.
Blackie would whine, wiggle her body and pat the floor with her
fore paws in a fever of impatience, — yet never change the align-
ment of her body. Many times the dog held her orientation
almost until the last second of delay and then if, when the
release came, her head was (apparently) not more than half
an inch to the right of its earlier position, she w^ent in that
direction and consequently went to the wrong box. Surely this
is weighty evidence against the functional presence of any
higher processes. Everything points to the conclusion that
Blackie 's reactions were determined by the orientation of her
head at the moment of release.

It has just been indicated that the maintenance of an orien-
ation either of all or a part of the body was necessary in the
case of the rats and dogs, if their reactions were to succeed.
Such was not the case with the raccoons. Each of these ani-
mals could react successfully when the wrong orientation was
held at the moment of release and when, so far as the experi-
menter could detect, no part of the animal's body remained
constant during the interval of delay. The evidence in support
of this generalization is perfectly conclusive. It is only pos-
sible— and necessary — to present typical cases here. I use Jack
as an illustration, although he did not delay as long as Bob.

In Jack's first 800 trials of delays, 77 were wrong. In 7 of
these 77, the raccoon had the wrong bodily orientation, but
did not follow it. On 12 trials, he had the right orientation,
but did not follow it. On 58 trials, he had the wrong orienta-
tion and followed it. Of the 723 correct reactions, 167 were
made starting with wrong orientations. .

In the following 1066 trials, 149 reactions were incorrect.
This group of trials extends from a period of 8 sec. delays at
which 75% was made through a series at 11 sees, where 91%
was made. Forty-one errors were made when the wrong orien-
tation was held but was not followed. Thirty-seven were made
when the right orientation was held. Seventy-one were made
when the wrong orientation was held and followed. Out of
the 917 correct trials. Jack reacted 309 times correctly when
his orientation was wrong.

It is interesting to note that almost one-fourth of the last

44 WALTER S. HUNTER

mentioned errors were made despite the fact that the proper
orientation was being held. This is a t3^e of reaction that
almost never occurred with the rats and the dogs. It would
seem to indicate that this raccoon is less dependent on gross
motor attitudes than the other animals. But the most signifi-
cant behavior is that of reacting correctly when the wrong
orientation was held at the moment of release. From the first
set of figures above, it will be seen that of 232 trials when the
wrong orientation was held, 167 or 71% were correct. The
second set of figures shows that of 421 trials when the wrong
orientation was held, 309 or 73% were successfully carried out.
Such a high percentage places the results above the possibility
of explanation by chance. Again, the fact that this type of
behavior dominated for several days at a time indicates that
something more than chance was manifesting itself. At 8 sees,
delay, e.g.. Jack made 10 correct reactions in one day, starting
with wrong orientations. The following two days had 7 each;
and the following two, 4 each. The next day the delay was
increased to 9 sees, and 8 correct reactions with wrong orienta-
tions were made. Eight, 6, 4, 8 and 6 were the records of such
reactions for the following days. The next day after these, 10
sees, delay was given. Here 9 correct reactions were made
starting with wrong orientations. These are fair samples of the
prevalence of this type of reaction.

In order to emphasize further the fact that Jack did not need
to keep any observable part of his body in a constant position
in order to react correctly, I shall give one day's record in detail
(Table X). Often only the raccoon's hind feet — or one of them —
remained constant. The interval of delay in this illustration is
10 sees. N.c. means no part of the body kept constant. H.f.c.
means hind-feet constant. Rt.h.f.c. means right hind-foot con-
stant. If the letter designating the orientation held at the
moment of release is in italics, the animal oscillated between
two faces of the release. If in addition the letter is starred,
the animal oscillated from one to another of all three faces of
the release. On the 3rd and 5th trials. Jack was distracted by
some noise during the delay. A concrete statement of what
these symbols mean in one or more instances should make
the record perfectly clear. On the first trial, the light was
turned on in the right-hand box. It was turned off and the

DELAYED REACTION 45

animal held for lo sees. During this delay, Jack turned toward
all three boxes. No observable part of his body remained con-
stant. At the instant of release, he was headed toward the
proper box, a, and reacted correctly. On the 5th trial, Jack
was oriented toward the middle box, b, at the moment of release.
He went to a, then to b, then to c and back to the experimenter
for food. The reaction had failed. If to the large number of
correct trials made with wrong orientations be added the correct
trials made with correct orientations that had been completely
lost and then refound, the number of reactions significantly
different from those of the rats and the dogs reaches relatively

huge proportions.

TABLE X

Orient.

Raccoon

when released

Light box

Behavior

Jack

*a

a

a

n.c.

*a

a

a

h.f.c.

*a

a

a

h.f.c.

*a

b

b

rt.h.f.c.

*b

c

abc

n.c.

*h

c

c

rt.h.f.c.

c

b

b

n.c.

*c

a

a

n.c.

h

c

c

h.f.c.

*a

c

c

n.c.

b

c

c

h.f.c.

a

b

b

rt.h.f.c.

*b

. a

a

h.f.c.

b

b

b

n.c.

*a

b

ab

rt.h.f.c.

Two other types of reaction remain to be indicated in the
above table, (X). Orientation b was held 6 times; orientation
a, 7 times, and c, twice. Yet different correct reactions followed
from the same orientations. Jack was oriented to c once and
went to b correctly. The next trial he had the ?ame orientation
and went to a correctly. In neither case had he seemed to
keep any part of his body constant. In the 4 correct responses
to the light box b, the corresponding orientations were as differ-
ent as possible. Twice Jack was oriented to a and once each
to b and c, yet each time the correct reaction was made. In
other words, the same orientation did lead to different reactions
and different orientations to the same reaction. These, again,
are types of behavior never met with in the rats and dogs. This
behavior occurred so frequently with Jack and with Bob —

46 WALTER S. HUNTER

less frequently, however, with Jill and Betty — that it must be
described as a genuinely new type of reaction in these experi-
ments and not as the result of chance or of the foreknowledge
of what box was to be presented. (This latter possibility was
adequately ruled out by controls.)

Table XI gives a numerical statement of the importance of
orientation for the rats, dogs and raccoons. As far as this
factor is concerned, the animals within each class were on a
par with one another. For this reason, I give the results for
typical subjects and not an average for all members of a group.
The data given for each animal are calculated from comparable
groups of 800 delayed reactions each.

TABLE XI

Rat Dog Raccoon

No. 14 Blackie Jack

No. of reactions in accordance with orientation . . 698 trials 760 trials 614 trials

or or or

87% 95% 76%

No. of reactions not in accordance with orienta- 102 trials*^ 40 trials 186 trials
tion or or or

13% 5% 24%

15 trials

5 trials

167 trials

or

or

or

14%

12%

89%

No. of reactions not in accordance with orienta-
tion that succeeded . . . . :

This table indicates plainly the similarity of the behavior of
the dogs and rats as well as the wide divergence of the raccoons
from the other two groups of animals. The rats and dogs
almost never reacted in opposition to orientation. When they
did do so, the number of their successes was a negligible quan-
tity. That orientation was a strong factor with the raccoons
is evidenced by the 76% of reactions that followed it. This
fact makes the 89% of correct reactions starting from wrong
orientations of great significance. The reactions succeeded in
opposition to strong orientation influence. This statement is
supported by the facts above noted. It was shown there that
one of the raccoons (a typical one): (i) Made different correct
reactions from the same orientation, and (2) made the same
correct response from different orientations. Additional evi-

*^ The large percentage of reactions not in accordance with orientation made
by the rat, when compared with those made by the dog, is due to the acquisition
of a habit of holding orientation b and reacting to box a, i.e., to a position habit.

DELAYED REACTION 47

dence will be presented later when the effect of the size of the
release upon the methods of delay is considered.

One other line of evidence which points to the uniqueness
of the raccoon's behavior should be noticed here, viz., the growth
of the methods of delay during the course of the experimenta-
tion. Did the rats and dogs rely upon orientation from the
beginning of the tests? Was the non-orientation cue natural
with the raccoons or acquired under the stress of circumstances ?
An examination of the records reveals the fact that from the
beginning of the learning tests the rats and dogs reacted in
accordance with orientation. There was no development of a
new method as the delays were increased in length. There was
simply an improvement in the facility with which the animal
maintained its position for a given length of time. When the
orientation was lost, it was no easier to react correctly at the
last of the experimentation than at its beginning. In the case
of the raccoons also there was no initiation of a new mode of
response. There were, on the average, as many reactions per
day during the learning period that did not follow the orien-
tation as there were per day during the delayed reactions. The
improvement that took place as the experimentation proceeded
was in the accuracy of responses not in accordance with orien-
tation, when these responses were made after a long delay.
"Non-orientation reactions," therefore, seem to be natural with
raccoons and thus seem to differentiate their behavior from that
of the rats and dogs.

(b) Position in the box. — The data considered under the
above title are those indicating the effect that the large release
box had upon the methods of delay used by the various animals.
It will be remembered that the large release gave the animals
the freedom of the interior of the box and thus permitted them
to take up a certain position in the box as well as to maintain
a certain bodily orientation.

Four rats, both dogs and all of the raccoons save Betty were
tested with this large release. Only three animals had their
methods of behavior essentially modified : Rats 1 7 and 1 6 and
the raccoon Jack. In no case, however, did an animal that
had depended upon orientation for its cue begin the use of a
non-orientation factor. What the modifications were will come
out in the following descriptive summary.

48 WALTER S. HUNTER

An analysis of the behavior of the four rats when tested with
the large release, brings to light the following facts: Save in
four cases, No. 17 went in the direction in which his body was
pointed at the moment of release. This was true regardless of
his position in the box. Nos. 13 and 15 always went in the
direction of body orientation. Hence position was only effec-
tive as an aid to the retention of body orientation. During
the last week of the tests with this large release. No. 17 mani-
fested an exceedingly interesting type of behavior. He needed
to keep neither position nor orientation constant in order to react
correctly. I could find no observable body cue by which the
problem was solved. The first day of this particular week,
3 of the 10 trials were of this type, but for several days no more
instances were noted Then they were again in evidence until
the end of the work. The correct reactions made in this manner
never exceeded 3 per day. That I was much astonished at the
sudden appearance of this type of behavior in the rat goes
without saying. To think that at the end of eight months'
steady work, the animal should suddenly adopt a new mode
of behavior! It was not many days, however, before the accu-
mulated data made clear the explanation. The rat always
turned to its right and entered the first box that was encountered.
Thus the animal could be at the middle box, but with its nose
slightly to the right of the door to this box, and when released
it would whirl and go to the box on the right. In the same
manner, the rat might have its nose just to the right of the
door to the left box when the release was raised and turn and
enter the middle box. The behavior of rat No. 16 differed from
this only in the direction of turning. This rat always turned to
the left and entered the first box that it came to. From these
data, it is obvious that the animals were using motor. cues to
guide their reactions. Their behavior was practically, if not
entirely, automatic.

There is no need to detail the results obtained with the dogs
when the large release was used. Both animals always went up
close to the release in front of the lighted box and then waited
until released. Each followed the orientation of his head re-
gardless of which box was nearest.

Among the raccoons, Jack's change of behavior was a shift
from the use of a non-orientation cue to a position cue. The

DELAYED REACTION 49

change occurred just at the close of the tests with the large
release. His usual behavior as well as his new form may be
set forth as follows: Before the new release was used, Jack
was being tested with boxes a (the right one) and c (the left
one) only. When the large release was put on, the tests w^ere
still confined to these two boxes. This was necessary in order
that the results be strictly comparable. Of the 510 trials given
under these conditions, 56 were made: (i) In which no observ-
able part of the animal's body remained in a constant position;
and (2) in which the animal's position in the box was wrong at
the moment of release. In every one of these 56 trials, i.e., the
animal w^as both in front of the wrong box and headed away
from the right one when the instant of release arrived. Twenty-
six of the 56 were reacted to correctly. Chance will account for
this number of reactions as far as the mathematics of the prob-
lem are concerned. However, from the point of view of the
observer, the majority of the reactions looked like anything
but chance behavior. This was especially true when Jack
reacted correctly and yet was pointed directly away from the
proper box. There was a directness and sureness about his
reactions that hardly savors of chance. I do not urge these
instances on the reader as evidence of non-accidental reactions.
I simply note the fact of their presence and the impression that
they made upon me. It is well to remember in addition, though,
that since orientation was such a strong factor in determining
Jack's reactions (see above, Table XI) that any responses in
opposition to this must be given great weight.

Owing to the fact that only two boxes were used in this series
and that they were located far apart, the raccoon had every
incentive to rely solely upon his position in the apparatus for
the reaction cue. This he soon did to a large extent. In every
case where his position was constant (and correct) and his orien-
tation changed, he reacted successfully. In other words, posi-
tion was the determining factor. When this became evident,
the experiment was stopped. The raccoon had shifted the basis
of his response so that I could detect its nature by observation.
Jack's behavior at this point was thus on a par with the dogs
and rats. Had the tests been continued, all that could have
been expected was the perfection of a habit of staying near
the proper box. This did not appeal to me as a profital)le goal
of endeavor.

50 WALTER S. HUNTER

For further confirmation of the statements made above con-
cerning the general methods of delay, I shall present the rac-
coon Bob's record with the large release in some detail. Three
hundred and sixty trials were given on two boxes with this
release. Of the 27 reactions made with the wrong position and
correct orientation, 19 were correct. Of the 91 reactions made
with the right position and wrong orientation, 81 were correct.
Sixty -three times the reactions were preceded by positions and
orientations that were both wrong. There are two possibilities
here: (i) Orientation and position may favor the same box;
(2) they may favor different boxes. Forty of the 63 were reac-
tions of class I, i.e., were initiated by orientations and positions
that favored the same wrong box. Fifteen of the 40 were suc-
cessful. In other words, although both position and orientation
favored the same wrong box. Bob was able to overcome the
handicap and make 37% correct reactions. The remaining 23
reactions were of class 2. Of the 23, 16 were successful. Where
position and orientation were both wrong, but did not combine
to favor the same box, Bob made 69% correct reactions. Of
the 7 reactions of class 2 that failed, 5 were in accordance with
position and 2 in accordance with orientation, i.e., 5 times the
animal went to the box favored by position and the remaining
times to the one favored by orientation.

In order that the reader may have a perfectly concrete presen-
tation of the reactions with the large release, I will quote a
day's record from the diary. The reactions were made at 8
sees, delay and with the exit doors from the light boxes, h and
c, open. In the records obtained with the large release, some
new symbols were used in describing the data placed in the

TABLE XIV
Eight Seconds Delay with Large Release
Animal Orient. Light box Behavior

Bob ba-2 b b h.f.c.

b
b
bc-2

c
b

b

c

b

b

n.c.

h.f.c. (walked to a and back)

n.c.

c
b
b

cb-l
b

c
c

b
c
b

c

be

b

c

b

n.c.

n.c.
h.f.c.

b

c

c

n.c.

DELAYED REACTION 51

"orientation" column. The first reaction, e.g., is to be read
as follows : The light was turned on in b, the middle box. Bob
was held for 8 sees, during which time he pivoted on his hind
feet, swinging his body along in front of b. At the moment
of release he was still in front of this box, but was oriented with
his nose in the corner halfway to a, the box on the right. Dur-
ing the delay at the fourth reaction, no part of the raccoon's
body was constant. He walked over in front of a and when
released was in front of b, but his body was pointed to the comer
half way to c at the moment of release. At the instant of re-
lease for the eighth reaction, Bob's body was pointed three-
fourths of the way toward c. This brought his nose within a
few inches of the entrance to that box, yet the reaction succeeded.
In the tenth trial, as in the second, position and orientation both
favored the same wrong box. In each case the reaction was
correct.

The net result of the data presented here for Bob is an almost
entirely conclusive proof of the statement that he does not
wholly depend either upon bodily orientation or upon position
in the box for the cues determinant of the subsequent reaction.
The 37% made with the reactions of class i is very significant
when one does not lose sight of the fact that both orientation
and position were here combined against any other factor lead-
ing to a correct response. In class 2 where such a combination
was not present, the percentage is conclusive as to the presence
of some non-observable cue. This is but further data confirma-
tive of that already presented above for Jack with the small
and large release.

In view of the generally negative results obtained with the
large release, as far as developing new and higher types of be-
havior is concerned, some one may say: (i) That if all of the
animals had been started with the large release instead of with
the small one, they would not have been so likely to develop
gross motor cues to guide their reactions, and (2) that it is not
surprising that no new type of behavior appeared after the
animals were firmly in the grip of habits developed in the small
release. To these criticisms I can only reply that any experi-
ment must exhaust first one method and then another. Time
did not permit the use of both methods here.

52" " WALTER S. HUNTER

2. Tests with Children

A. Method of Experimentation. — Details of the method of
experimentation here used may be presented as follows : When
the subject was first brought into the room, the following in-
structions were given: We have a little game in here which
we are going to play. You will stand in here, (I indicate the
release box). When I release you by raising the gate, (I illus-
trate raising the gate), you are to go and push one of the but-
tons over on the wall. One of these buttons will make a noise.
If, after I raise the gate, you push the noisy button first, I will
give you some candy when you come back to me. But if you
go first to some button that isn't noisy, then you must try again
before you get the candy. So, you see, the game is to push
the noisy button -first and so get candy. Do you see now how
we are to play the game? Run over there and push on some
of the buttons. See, sometimes they make a noise and some-
times they don't. (I switch the buzzer on and off for all the
buttons. The lights are not on yet, nor has the subject's atten-
tion been called to them.) Now let's try the game and see if
you can push the noisy button first. (The child is placed in the
release box. The light is switched on over the noisy button.
The child is held 5 sees, in the release box before being set free.)
These instructions were memorized by the experimenter; and,
although parts were repeated several times to the subject,
nothing that is not given above was told to the child.

It was foimd that with all save F, 6 preliminary trials were
sufficient to familiarize the children with the apparatus and to
overcome their timidity. (H's exception will be noted below.)
F was given 35 trials extending over three days on the preHm-
inary work of learning that the noisy button meant candy.
By the end of that time the association was firmly established.

No fixed number of trials per day was given. The amount of
the day's work was adjusted to the child's disposition and to
the length of delays. No set number of trials was given at any
stage of delays. In other words much the same method was
used as for the first year's work with the raccoons. Bob and
Betty. Delays were increased continuously until an error was
made. At this point, they were either decreased at once or
continued at their existent value for several trials before further
change was made. The experimenter believes that so flexible

DELAYED REACTION 53

a method, when properly checked by careful observation of the
subject so that the task is changed in constant sympathy with
the subject's apparent needs, is excellently fitted to bring out
what the subject can do naturally as opposed to what it can
be trained to do. Of course the effects of training cannot, and
need not, be eliminated. However, they are not so great in
the method outlined above as they would be were many trials
given at each stage. The difficulty here is the same as that
mentioned when discussing the records for the raccoons viz.,
where only a few trials are given, the critic has a better chance
to claim that the results are due to chance. Our conclusions
^•ill seek to avoid this criticism. But in the last analysis, the
themselves must be their own justification.

B. Are the Results Obtained from Animals and Children Com-
parable f — In the light of the foregoing method and of careful
observation of the children, the following points suggest them-
selves as the essential considerations in a relative estimate of
the conditions under which the children and the other animals
worked: (i) Fear. — This was overcome in the animals by the
preliminary training; in the children, by kindliness and cheer-
fulness on the part of the experimenter and by the child's ex-
amination of the apparatus as described above. (2) Motive. —
Hunger and punishment insured a maximum of effort on the
part of the animals. Candy, words of praise from the experi-
menter and a desire to excel its companions incited the child
to do its best. (3) Knowledge of the reaction desired. — The
rats, dogs and raccoons had to learn everything by themselves.

(a) The preliminary series acquainted them with the fact that
there were three exits to the problem box and possibly also
with the fact that only one of these would be open at a time.

(b) In the regular learning tests, these animals had to associate
the light and the open box in such a manner that the light
became the sign of the open box. (c) In the delayed reaction
tests, again, they had to learn that the open exit was always
in the box which had been most recently lighted. If we turn
now to the children, we find the following situation: (a) They
were told of the push buttons which for their problem corres-
ponded to the exits of the other experiment boxes. Where the
animals had had to learn the fact of only one open exit by
trial and error, the children were at least aided by being told

54 WALTER S. HUNTER

that only one button would make a noise, (b) In the regular
learning series, the children had to acquire the association
between noisy button and light on their own initiation. It
must be remembered that the experimenter never mentioned
the word light and never directed the subjects' attention to
the lights as long as the experiments continued, (c) In the
delayed reaction, also, the children were thrown on their own
resources in the working out of the problem. (4) Treatment
during the delay. — During the interval of delay the rats, dogs
and raccoons were usually left strictly alone. Only in a few
control tests was any effort made to distract them. However,
uncontrollable noises occasionally intervened and disturbed the
dogs and the raccoons. During the period of delay, the children
were entertained by the experimenter by means of stories, the
drawing of pictures and, in a few of the long delays, by gifts of
candy. In his opinion, these distractions made the experiment
more difficult, although it is true that impatience and fretting
on the part of the subject were largely eliminated. It would
thus seem that the only objection to*this method is that possibly
the distractions served to urge the child to form a "purpose to
remember" sooner than would otherwise have been the case.
I cannot deny this as a possibility. It may have occurred with
the girl M. But from the fact that the others remained impa-
tient and complained of the delay until late in the experimenta-
tion, I do not believe they realized before that time that the
problem was to see how long they could remember. Question-
ing on this point at the close of the experimentation confirmed
this. (5) Possibility of the problem being talked over by the
subjects. — This, of course, has no bearing on the case of the
animals, but it presents a fairly serious possibility with respect
to the children. The possibility is all the greater because of
the fact that four of the children lived in the same neighbor-
hood. Although recognizing this, I feel that it played very
little part in the experiments; for where it might have been
effective, the tests continued but ten days. M was old enough
to do as she was told, rmless severe temptation came her way.
Such a possibility, however, was prevented by cold weather and
school keeping her away from the other children. As will be
pointed out later, talking could not have influenced H's con-
duct because she knew all about the experiment. F was under

DELAYED REACTION 55

the experimenter's control and never saw the other children.
Hd and L differed so in their attitudes toward the problem and
in their general behavior, that I cannot well believe they talked
about the problem when together. Further evidence why it
is improbable that the children planned with one another how
to work the problem is that there was a very keen rivalry as
to who should have the most candy beans at the end of the
day's work. When the work was finally dropped, the children
said that they had not talked the problem over with one an-
other. In the light of these considerations, I believe that the
children and the animals worked essentially on a par, so far as
extra-individual influences are concerned, throughout both the
regular learning series and the delayed reaction series. I say
"essentially on a par" because the social influence due to the
presence of another member of the same species was operative
in the case of the children while it did not appear with the
animals.

C. Learning. — A few words will suffice to describe the trials
given on learning the association between the lights and the
noisy button. With all subjects save F the light was first turned
on at a. (For F it was put on at h.) On this first trial all the
subjects failed. F went to a, then walked past 6 to c and thence
back to h and rang the buzzer. All the others went first to b
and then to a. This may have been due to the fact that h was
the nearest of the buttons. F was the only one that made an
error in this series after the first trial. The others learned the
association in one trial. Out of the i6 trials on the first day,
F failed on 9. The first test was missed the second day and
none the third day. In other words, no errors were made after
the 17th trial.

D. Differences Between the Learning of Animals and Children. —
Some of the difference between the above data and that for
the animals is undoubtedly due to differences in attitude toward
the problem, although the conditions were so arranged that this
should ha\'e been at a minimum. Five rats learned the asso-
ciation in from 160 to 176 trials. The two dogs, it will be re-
called, required more than 500 trials. Bob received only 120
trials. These figures present the number of tests after which
no errors were made. Over against these figures, 46 should,

56 WALTER S. HUNTER

stand as the largest number of trials given on learning to any
of the children, and this to F, the youngest.

I believe that the main factor that would make for non-
comparabiHty in these results is that of brightness preference.
It is not known what the value of this factor was for the chil-
dren. But leaving this possibility aside, it is to be noted that
the above records fall into two well defined groups; those for
the animals and those for the children. If the difference here
is correlated with grades of intelligence," one may well ask why
no such differences appear between the several classes of animals.
The problem is all the more interesting when it is pointed out
that there seems no certain correlation between the ability to
learn the association and the ability to delay when one considers
the various groups of animals. The dogs and the rats used the
same method in delay, yet Blackie delayed longer than any rat.
Jack and Bob used methods of behavior in delay quite different
from the other animals and their delays were far longer than
those of the rats and of Brownie. Furthermore two of these
raccoons delayed about the same period of time, yet varied
greatly in their times for learning the association. Among the
children, matters would seem to be different. They learned
rapidly and in delays reached relatively long periods of time
by what seemed the same method used by the raccoons. The
following answers suggest themselves with reference to the
question put just above: The present data indicate: (i) Either
that the different grades of intelligence among the animals were
not great enough to be registered in the learning rates although
the grades between an'mals and children were sufficient to mani-
fest themselves ; or (2 ) that the association can be learned with
a type of process that will not suffice for long delays where the
orientation is not maintained. If this latter alternative be
correct, it would seem that the children used the method re-
quired for the last mentioned t3^e of delay in their learning
period. The raccoons, on the other hand, used the same method

** V. C. Hicks and H. A. Carr *^ find no such correlation of the nunaber of trials
taken in learning a maze and the grade of intelligence of the subjects. The time
is not ripe for a statement of the type of learning present in maze problems when
compared with the present test. However, the data presented by the above writers
and by myself would indicate that the two problems do not involve the same means
of solution, at least to the same extent.

^^ Hicks, V. C. and Carr, H. A. Human Reactions in a Maze. Jour, of Animal
Behavior, 1912, vol. 2, p. 101.

DELAYED REACTION 57

in learning the association that the other animals did, but a
different method when it came to delays. There may thus be
a difference in intelligence due to varying abilities in the use of
some one instrument of adjustment or due to varying abilities
in the use of different instruments of adjustment. Blackie and
the rats would illustrate the first case; the children and rac-
coons in comparison with the other animals would illustrate
the second. So much for the possible meaning of the learning
times of the different animals and children

E. Delayed Reactions. — M, age 8 years, was given 38 trials
on delay. The first 2 only were at intervals less than i sec.
The others ranged from i sec. to 28 mins. and extended over
7 days. Fifteen trials were on intervals of 10 mins. and over.
Only two errors were made, one at 5 mins. and one at 20 mins.
The latter was probably caused by ill-humor. Longer delays
might doubtlessly have been secured had it been desired. The
only attempt was as follows: On four days (not consecutive
ones), M was asked which button she pushed last the day before.
Three times she answered correctly.

M did not need to keep any part of her body constant during
the delay. On the trials involving more than 10 mins. delay,
she was sent out of the room and put with the other children.
During the delays she conversed freely with the experimenter.
Any "purpose to remember" that the subject formed was formed
on her own initiative. Great care was taken — and I believe
effectively so — to insure that no suggestions be secured from
extraneous sources. As early as the 15th trial on delays M
volunteered the information that she remembered where the
light was in order to push the proper button. Several times she
reacted correctly and then volunteered that she had guessed
which button to push.

Several extra trials were given in which the problem was
changed. The light was now placed successively over two
different buttons, the last button being the noisy one. This
problem brought out a type of reaction very often seen in the
animals. M would start toward one, box and then turn and go
to another. In the regular series, hesitation occasionally oc-
curred and M wavered between two boxes.

Hd, aged 6 years, was given 47 trials on delays. Only the
first two were less than i sec. long. The others ranged from

58 WALTER S. HUNTER

I sec. to 33 mins. Fourteen trials involved intervals of lo
mins. or over. A total of 15 trials was given with intervals of
4, 5 and 6 mins. in length. In the 47 trials, 10 errors were made.
Two of these came in each of the delays of 4, 5 and 6 mins. One
came at 11 mins.; i at 12 mins.; and 2 at 20 mins. Hd, there-
fore, had no trouble in remembering the solution of the problem
until the intervals of delay reached 4 to 6 mins. Then the
difficulties that arose were mastered and did not reappear until
the periods 11 to 12 mins. and 20 mins. were reached.

Hd did not need to keep any part of his body constant dur-
ing delay. He and the experimenter exchanged stories con-
tinually. This subject differed from M in the number of errors
and in the greater frequency with which information was offered.
The former I attribute to difference in ability; the latter, to
natural garrulousness. The restraint during delay was a great
source of annoyance to Hd. He complained a great deal be-
cause of it. My diary notes contain many such passages as the
following: When light was turned off, he said: "Why hold
me so long, I may forget which it is." The subject evidently
realized that he was to remember where the light had been.
Many times during the delay, Hd would stop conversing and
say "O, I know which one it is" and would then point — not
always correctly — to the button he had in mind. Instances of
wavering and hesitations preceding acts of choice were noticed
with this subject.

L, age 6 years, was given 41 trials on delay. Only one of
these involved an interval less than i sec. long. The remainder
were from i sec. to 25 mins. in length. Twenty-one trials were
on delays from 4 to 9 mins. long. Nine were on delays over
10 mins. in length. Seven errors were made in the total 41
trials. Ail of these came in delays of 4 mins. or over. Two
were at 4 mins. and one each at the following intervals: 5, 7,
8, 15, 17 and 25 mins. L, a,s Hd, found most difficulty in the
period around 5 mins.

L did not find it necessary to keep either his body or the
direction of his attention constant in order to solve the problem.
He, too, conversed freely with the experimenter during delays.
L would often watch for the light out of the corner of his eyes.
After it appeared, he would apparently pay no further atten-
tion to the problem until released. L was less demonstrat ve

DELAYED REACTION 59

than Hd and did not remark so often during the delays that
he still knew the proper button. Like Hd, he was impatient
during the short delays at the first and the long delays at the
last of the experiments. At times he wavered and hesitated in
his reactions, saying that he had forgotten or that he wasn't
sure. (In such cases, of course, the experimenter gave no cue
to the solution.)

Each of the three subjects whose records have been discussed
formulated his own "purpose to remember." H was given trials
under the same conditions as the others, save that she was
told the purpose of the experiment. She was told that the
light would be over the noisy button. When delays were begun,
she was told to be sure and remember where the light had been.

H, age 6 years, received 15 trials on delays. Only the first
two involved intervals less than i sec. in length. The others
varied from 10 sees, to 35 mins. One error was made at 21
mins. At this trial, H walked half way to b, paused for 5 sees.,
wavered as though to go to c, but finally pushed b. She then
returned to the release and was told to try again. Again she
wavered at a point half way to b, but this time she went to
c. The light had only been turned on the one time. I did not
talk to H during the first 12 delays. She was always told before
the light went out to be sure and remember, but this was all.
In the subsequent trials, every effort was made to distract her
during the periods of delay. The results obtained with H, when
compared with those for Hd and L, indicate that it is an aid
to the subject to have the "pui-pose to remember" expressly
formulated for him.

F, age 2^ years, was given 507 trials on delay. Of these, 30
delays were less than i sec. long. The other 477 trials ranged
from I sec. to i min. Of the 477, 143 were wrong. The follow-
ing table (XV), gives the relative distribution of these errors.
It does not give the delays in the order in which they were
given to F. The table simply summarizes the number of trials
and errors at each stage. An advance was never made from
one stage to another until at least 80% correct reactions were
made for at least 5 successive trials. The only interval that
F did not finally master was i min. This is suiprising when
such a high percentage of correct reactions occurred at 50 sees.
On each of the two days when F was tested for the i min. delays,

60

WALTER S. HUNTER
TABLE XV

Delay

No. of trs.

Errors

Delay

No. of trs.

Errors

1 sec.

23

4

15 sees.

65

19

2 sees.

4

0

20 "

40

8

3 "

5

0

25 "

22

5

5 "

38

10

30 "

22

11

6 "

9

0

35 "

. 18

2

7 "

35

9

40 "

23

10

8 "

3

0

50 "

19

2

10 "

85

37

1 min.

45

26

12 "

10

0

the first one -third of the trials were at 50 sees. Reactions were
perfect. The last two thirds of the day's work was at i min.
and both times she fell below 50%. There were no known
extraneous factors to cause this. The conditions were as near
ideal as possible. The method adopted with F, after the first
three days, was one of slow advance from stage to stage. At
the higher delays, each day's work was begun — ^as just illus-
trated— with the longest interval that had been mastered. Some
5 trials were given here and only if the reactions were perfect
was an advance made. I believe that entire dependence can
be placed upon the results obtained. F was the only one of
the children that did not reach a delay of at least 20 mins.

F was distracted continually. She needed to keep neither
her attention nor any part of her body constant in order to
react correctly on delays up through 50 sees. She often hesi-
tated and wavered in making the choice of buttons. It is inter-
esting to note in this connection that this wavering, hesitant
behavior was only noted in the case of the raccoons and children.
Every individual of these gave many examples of it. This is
significant when it is borne in mind that these subjects used
apparently the same method in solving the problem. If that
statement is too broad, at least it may be said that they agreed
in not using gross motor orientation exclusively as did the rats
and dogs. This is not the very usual t3^e of behavior described
by Yerkes for the dancing mouse. He says: "I have at times
seen a mouse run from one entrance to the other twenty times
before making its choice ; now and then it would start to enter
one and, when half way in, draw back as if it had been shocked.
Possibly merely touching the wires with its forepaws was re-

DELAYED REACTION 61

Sponsible for this simulation of a reaction to the shock." *•
The above reaction, noted by Yerkes, and other similar cases
in the literature are reactions to present stimuli. Where the
mouse ran to the entrance of the box, but did not enter, the
explanation undoubtedly lies in the inability of the stimulus
to set off the proper reaction. As Yerkes suggests, the stopping
of the animals half way to the box was probably due to the
contact with the wires. The same general type of behavior
described above for the raccoons and children occurred during
certain periods of time with my own rats and the dogs. In
every case, however, it was due to a habit of turning around
when released. This habit was executed in the same fashion
no matter which was the proper box to choose. As opposed
to the tests where either present objective stimuli or habit are
involved, the reactions of the children and raccoons were, as
far as could he determined, perfectly spontaneous, i.e., deter-
mined by intra-organic conditions that varied for the different
boxes. These cases belong in the same class as that of Miss
Washburn's cat described above (p. 20).

F's father, a trained psychologist, informed me that the
child was just reaching the stage where her memory for objects
and events had begun to take on definite form. When brought
in from a ride in the park or a visit to a friend, she could very
seldom remember the details of the event, indeed not more
than half the time could she remember the gross fact of having
been somewhere. This occurred even with what were to F
very interesting experiences. Sometimes, it is true, the diffi-
culty lay with the lack of control of language ; but this was
not always the case. After her playmate had gone for some
time, if F was asked who had been there, not only could she
frequently not tell, but at times she was bewildered even by
the suggestion that anyone had been to see her at all. On the
other hand, some cases were noted where F remembered an
event for several days. Phrases, also, that she had heard but
once were often spontaneously repeated for the first time sev-
eral days later. The present tests thus found the child in a
\ery important stage of mental development. Definite memo-
ries, of the adult human kind, were still hanging in the balance
with the chaos of the preceding period. As far as the lengths

*• Yerkes, R. M, The Dancing Mouse. 1907, New York, p. 130.

62 WALTER S. HUNTER

of delay go, F ranks approximately half way between the other
children and the raccoons. As to apparent method of work
she — and the other children too — is superior to the raccoons in
that orientation played no discernible role in her reactions.

VI. THEORETICAL CONSIDERATIONS
1. The Cues Essential for Successful Delayed Reactions

After the above presentation of experimental results, there
remains the important task of determining as nearly as possible
just what cues the subjects used in their reactions. The follow-
ing paragraphs of this section will set forth in detail the various
possibilities of interpretation and indicate what seem to me to
be their relative validities.

Let us first formulate in a general way the conditions that a
reagent will have to meet in the solution of the present problem.
He is confronted with three boxes which offer as many known
possible points of egress. One of the boxes is lighted. In the
course of a series of experiments, the reagent has learned to go
through, or to the lighted box in order to reach food. In time
the stimulus will set off the reaction practically automatically.
Approximately at this period, I begin to turn off the light before
the subject reaches the box. It is possible that for several
stages of delay the box which has been lighted remains light
for a short interval of time due to the presence of after-images
in the subject's eyes. But as the delays increase in length, there
will come a point at which the problem shifts from "go to the
lighted box" to "go to the one of three dark boxes that was
most recently lighted." If this shift comes before the i sec.
delay, it will be less likely to involve relearning. If, e.g., the
change comes at the second stage, the momentum of running
will rnake it easier for the animal to continue into the dark
box than to turn and seek another. With some rats, the results
indicate that the change did come at this stage; for when the
light was turned out, the animals stopped short and went into
another box, even though that too was dark. With the other
animals and the children, it was impossible to tell just when
the shift came. The problem having once changed, however,
the question now is how long after the boxes all become dark
can the subject pick out the one which was lighted most re-
cently. Our special problem concerns only the solution of this

DELAYED REACTION 63

second problem. During the learning series, the light stimulus
acquires the power of releasing muscular activity applied in a
certain direction. Just what direction this shall be is deter-
mined by the spatial location of the light. A significant mean-
ing here attaches to the' object ve stimulus, i.e., the light plus
the definite location. (By the term meaning, I imply nothing
more than the fact that a certain stimulus evokes a certain
reaction under conditions that are not usually described as
involving mere habitual or reflex activities.)

As soon as the problem shifts to a choice of one of three sim-
ilar boxes, i.e., as soon as a stage is reached where the deter-
mining stimuli are absent at the moment of reaction, then it is
necessary, I assume, if the reactions are to succeed, that the
subject develop substitutes which shall take the place of those
stimuli as carriers of the needed meanings. In other words, the
substitutes must fulfill the function of the previous stimuli in
arousing the three appropriate movements. The substitutes
may secure this power either through association with the light
during the learning series, or during the delayed reaction tests
themselves through a process of trial and error. In view of the
fact that delayed reactions did succeed under the present con-
ditions, there can be no question as to the existence of the sub-
stitutes. Our next problem is that of determining their nature.

The substitutes or cues that determined the subject's reac-
tions may theoretically have arisen either within or without
the organism. We shall consider the latter first.

.4. Substitutes Derived from the External Environment. — Under
this heading, we shall consider two possibilities: (a) Were>
there three simultaneously present objective cues that may
have served to determine the subject's reactions? and (b) were
there three objective cues that varied from trial to trial with
the position of the light which may have determined the reac-
tions after the light was turned off? In any case there must
be two or three cues each determining one of three reactions.
We shall consider the possibilities in the above order.

(a) Simultaneously present objective cues. — Although every
effort was made to secure uniformity in the visual aj^pearance
of the three light boxes, they differed at least in spatial position
and in the nature of the part of the experimental room visible
above the walls of the apparatus. In addition there were pos-

64 WALTER S. HUNTER

sibly olfactory differences in the boxes used by the animals due
to the animal odor itself and (in the case of the rats and rac
coons) to a food odor set up by the milk that was on the fore-
feet of these animals. Suppose now that these simultaneously
present external factors be treated in their entirety for each box
and be represented by the letters x, y and z. It may be assumed
that the subjects learned to react not to the lights — -a, b and c
— ^alone,' but to the complexes ax, by and cz. Then when the
delays were begun, the reactions were made to x, y and z, either
immediately or after a period of learning. Here we would have
three cues governing as many reactions, and all three cues
would be presented simultaneously at the moment of release
whereupon the reaction would take place.

This hypothesis is not supported by the experimental facts.
If the reactions during the delay were determined by these
stimuli present at the moment of release, the animals should
have learned to delay for almost any interval of time. It should
be no more difficult to react to x, y and z, than to the lights,
since the reactions during delays were just as "precipitate"
and "headlong" as when the lights were present. Now it was
demonstrated that all of the animals could react perfectly when
held in the release i min. with the light on. Why then could
they not delay a minute with the factors x, y and z which were
also constantly present? Further, if the subjects succeeded in
reacting to x, y and z for intervals of 5 or 6 sees., why should
they be unable to reach 7 or 8 sees.? In other words, why,
if they reached one stage of delay on this basis, should the
subjects not go a little beyond and so up to a large delay ? There
is no answer to this, if one assumes, as we have done, that the
cues are simultaneously present at the moment of response.

In the section on experimental results, tests were described
where the entire sides of the problem box surrounding the
entrances to the light boxes were covered with cardboards of
widely separated grades of brightness. This device accentuated
(from the experimenter's point of view) the constant differences
between the hypothetical stimuli x, y and z, yet under these
conditions neither rats, dogs nor raccoons showed improvement
in their abihties to react. This series does not prove that no
objective factors x, y and z were influential in initiating behav-
ior ; but, in conjunction with the immediately following theoret-
ical discussion, it does make it very improbable that such fac-

DELAYED REACTION 65

tors should have exercised a determining influence on the
reactions.

Let us grant for the moment that such objective factors
were present, could they alone control successful reactions, i.e.,
could they be the determining factors in initiating correct de-
layed reactions? Since x, y and z are present simultaneously at
every response, they per se cannot be the bases for difi:erential
responses. The subject must select one of the boxes. Hence
the hypothesis would need to assume that the subjects are not
reacting merely to x, but to x-where-the-light-has-just-been.
Here we are forced face to face with the problem from which
we started: What represents, or is a substitute for, the light?
What is the element attaching to x that is equivalent to "where-
the-light -has just-been " ? There can be no question but that
X, y and z in some form constituted a part of the general stim-
ulus, for the subject must apprehend the different spatial loca-
tions of the boxes in order to react to them. The objection
may be offered that this contention as to the effectiveness of
these objective stimuli is inconclusive, inasmuch as the animals
may have been reacting entirely in terms of kinaesthesis, i.e.,
had learned the fixed order in which the three boxes were pre-
sented. This possibility was adequately eliminated by control
tests whose results are discussed on p. 67. The point to be
established here is that x, y and z could not have been the
crucial substitutes for the lights. These must have been fac-
tors which were not all simultaneously present, each in its
entirety, at every response, i.e., they must have alternated
from trial to trial, depending therefore on the position of the
light. The question now is: Were there any such factors in
the objective environment?

(b) Alternating objective cues. — Controls were instituted:
(i) To prove that during the learning series the light (an alter-
nating factor) was the determining cue for the reactions; and
(2), to prove that in the absence of the light no other external
factor took its place as a determinant of the reactions.

I. The reagents did not derive any cues from the experi-
menter. Screens were so arranged that the operator was never
visible to the dogs. Control tests were also made under these
conditions with the rats, raccoons and children. The constant
beating of a metronome covered up any noises due to the ex-

66 WALTER S. HUNTER

perimenter's breathing that the subjects might have utiHzed.
The dogs and raccoons did not rely upon the manner of being
released. This was proved by having different persons operate
the release. This control was not used with the rats. How-
ever, it is extremely improbable that these animals depended
upon cues from such a source. Moreover had they or any of
the reagents done so, their intervals of delay should have been
almost indefinitely great. It must be borne in mind that any
cues derived from the experimenter, in order to afford extra
aid in the reactions, must be present at the moment of release.
This would make it possible for the animal to avail itself of
the cue after a delay of any length. If the cues were given by
the experimenter only at the beginning of the delay, the prob-
lem confronting the animal would not differ from that of delay-
ing with reference to the light.

2. The reagents did not depend upon any after-glow of the
lights. When the lights were left on for one minute — & period
much greater than was ever used in the experiments proper —
and were then turned off, there was no appreciable after-glow
of the carbon filament that the experimenter could detect.
(In any event, such an afterglow would not persist long enough
to influence the longer delays of the reagents.) Hence any con-
tinued brightness of the boxes (considered from the reagent's
point of view) after the current was switched off must have
been due to the reagent's after-images. The possibility of
using these for cues will be considered below.

3. The reagents did not depend upon variations in the tem-
perature of the boxes in making their reactions. After the
lights had been turned on for one minute in any box, the tem-
perature of that box was never raised more than a degree centi-
grade. Indeed only occasionally could any change of tempera-
ture be detected. The "headlong" manner in which the ani-
mals reacted to the boxes, together with the fact that they
oriented toward the light immediately upon its appearance
indicate that they were not governed in their reactions by the
slight variations in temperature. Yoakum*' found that rats
could discriminate differences of 16° C, but even then their
behavior was the result of long special training. His squirrels

*' Yoakum, C. S. Some Experiments Upon the Behavior of Squirrels. Jour.
Comp. Neur. and Psych., 1909, vol. 19, p. 565.

DELAYED REACTION 67

— and probably the rats also, although he does not say — were
hesitant about entering the boxes even with a difference of
25° C. These data make it exceedingly improbable that the
animals in the present tests were influenced by temperature.

4. The animals did not derive cues from the doors at the
exits of the light boxes. Numerous control tests were made
in which all of the doors — and they were of large wire mesh,
and hence hard to see — were open. Under these conditions,
the animals reacted as though only one door had been open,
as was usually the case.

B. Substitutes Derived from within the Subject's Body. — The dis-
cussion of these intraorganic substitutes will be divided into two
parts: (i) What were the internal cues used. by the various
reagents?; and (2), how did these internal cues operate in order
to guide behavior?

(a) The type of internal cue used. — i. Did the subject antici-
pate the order of presentation of the lights, i.e., were the cues
to the reactions the individual responses of an habitual series?
The evidence is perfectly unambiguous in support of the fact
that the reagents did not rely upon the presentation order of
the boxes in making their reactions. The order for each group
of animals (not children) followed a series of 30 presentations
so arranged that each box recurred an equal number of times.
A new series was given whenever one box was eliminated from
the experiments. Control tests in which the regular order of
presentation was varied were introduced practically once every
two weeks with all the animals. Controls were also made in
which the lights were not used.

2. Did the reagents guide their reactions by after-images of
the light? Although the existence of after-images in animals
has not been demonstrated, we shall, for the sake of the argument,
disregard this fact and admit of their possibility. In order to
lead to correct reactions in our experimsnt, these after-images
must appear in the proper direction for each response. Their
directional position is a function of the orientation of the head
and eyes, and as a consequence the hypothesis can not explain
those correct reactions resulting from faulty orientations. After-
images can possess a possible function only in conjunction with
the maintenance of a constant orientation, and the hypothesis
would need to assume that these overt motor attitudes are but

68 WALTER S. HUNTER

subsidiary phenomena serving as a means for the effective func-
tioning of the after-image processes. But after-images can
hardly persist long enough to account for the maximum periods
of delay attained by this method of solution. Such delays
varied from lo sees, for the rats to 25 sees, for the raccoons
and 5 mins. for one dog. All of the conditions of the test were
distinctly unfavorable for any persistence of possible after-
images. The light was weak in intensity (3 c.p.). Its average
duration of exposure was approximately but 5 sees. Any ap-
proximation to steady fixation either during or subsequent to
the exposure of the light was the exception. The animals as a
rule were continually on the move, nosing and clawing at the
face of the release box both during the exposure and the period
of delay. Steadiness of fixation after the perception of the
stimulus is a very essential condition for the development of
after-images. Movements of the head or eyes tends not only to
prevent their appearance but also to destroy them when present.
The significance of these conditions is more apparent by re-
calling the fact that any extended duration of after-images is
an exceedingly rare phenomenon in the normal perceptual activ-
ities of humans. Fixation is too short and changeable for their
development. Since we are forced to argue from analogy with
human conditions, one must also distinguish between the pos-
sible presence of such processes and the ability to perceive
them. The mind tends in the interest of clear vision to over-
look and neglect such processes as it does in the case of entoptic
phenomena. With many people the ability to see after-images
involves a previous knowledge of their existence and some
degree of training and practice in their observation. In other
words, after-images as persistent objects of consciousness are
a product of the laboratory, and the assumption of their effec-
tive existence as guides to conduct in the normal perceptual
activity of an animal is exceedingly questionable. After-images
exhibit the phenomenon of intermittence. No high percentage
of 'correctness of response could result from such a cue, as its
presence at the moment of response would be a chance coinci-
dence. The theory of after-images, moreover, is entirely un-
necessary, as we have the possibility that these motor attitudes
of orientation may themselves serve as a sufficient guide to

DELAYED REACTION 69

conduct, and on this M-pothesis the assumption of the effective
presence of after-images is an explanatory luxury.

3. Motor attitudes of orientation as cues of response. The
data of the preceding section conclusively prove that main-
tenance of orientation during delay was an essential condition
for coiTect response with the rats and dogs, and that such motor
attitudes exerted a strong influence upon the behavior of the
raccoons. Either these attitudes serve as the substituted cues
and control conduct directly, or they function indirectly as a
means of support to some such cue as an after-image. The
evidence unambiguously favors the first supposition. These
orientation attitudes, like any sensory process, may be a stim-
ulus to definite movements. This tendency of the animal to
run in the direction of their orientation at the moment of re-
lease was natural and habitual. The tendency was present in
full strength at the beginning of the experiment. The tests
merely developed the maintenance of orientation for longer and
longer periods. This fact indicates that the motor attitude
functions directly upon subsequent conduct. If the attitudes
were but a means of support to some other cue, one would
expect that this relationship of means and end would need to
be acquired gradually during the experiment. An alternative
theory presents many theoretical and factual objections. Any
such roundabout and forced type of explanation is entirely
unnecessary when we know that motor attitudes are a natural
guide to the direction of subsequent responses. The mechanism
of such a cue may be entirely automatic and mechanical. It
requires nothing more for its explanation than does any habit.
A stimulus initiates a certain act whose completion is prevented
by external means. This initial activity persists unchanged so
far as its directional aspect is concerned until the raising of the
release permits it to function in a normal and habitual manner.

4. Some unknown intra-organic cue non-observable by the
experimenter. Our data prove conclusively that some such
cue was utilized by the raccoons and the children. Our proof
of this statement is based upon the method of exclusion and
the nature of such a factor must necessarily be defined at pre-
sent in negative terms. We have exhausted our ingenuity as
to objective possibilities of explanation, and as a consequence

70

WALTER S. HUNTER

are forced to conclude in favor of an intra -organic factor. The
possibility of a temporal series of habits was eliminated by
control tests. Neither orientation nor any distinctive motor
attitude could be detected in the children or in at least 2 5 % of
the responses of the raccoons. The after-image hypothesis is
entirely inadequate when orientation is faulty. In those reac-
tions of the raccoons resulting from wrong orientations, the
percentage of correctness was so great as to eliminate the possi-
bility of chance. The nature and mechanism of this factor
will be discussed in subsequent sections.

5. The following table (XIII) summarizes the cues used by
the different reagents. P.C. (possible cue) means that the
reagents so listed may have used that cue at times. N.C.
(necessary cue) means that the reagents so listed had to use
that cue or fail in a significant number of their reactions. R
stands for rats; D, for dogs; RA, for raccoons; and CH, for
children. For convenience of reference a classification is made

TABLE XIII

Cubs Used by the Re-agents

Internal

External

After-image

Orient. At.

Non-orient. At.

Idea

P.C.

N.C.

P.C.

N.C.

P.C.

N.C.

P.C.

N.C.

Never a

determining

cue

R
D

RA
CH

RA

R
D

RA
CH

RA
CH

in this table of those reagents that may have guided their re-
action by ideas. This phase of the table will not be clear until
the final section on the Place of Ideas in the Grades of Animal
Learning is read.

(b) The mechanism by which internal cues guide behavior. — As
already indicated the mechanism of orientation attitudes pre-
sents no difficulties. The light stimulus arouses the proper act.
This inhibited act persists unchanged so far as its essentials
are concerned during the delay. At the moment of release the
animal runs in the direction of its orientation. This tendency
of responding in conformity with orientation is natural and

DELAYED REACTION 71

habitual with the animal. The whole process is explicable on
the basis of habit. Maintenance of orientation is acquired
gradually by the trial and eiTor method as is any habit.

The case of the non-orientation cues presents more difficulty.
During the preliminary learning tests there was established by
the trial and error method an association between the Hghts
and the three acts of securing food. Between the two terms
of each of these three primary associations there was interpolated
probably by the trial and error method an intermediary link.
These three cues were associated with their respective lights on
the one hand and their respective acts on the other. Each light
will now awaken its corresponding cue and this cue will in turn
initiate the act with which it has been associated. In order to
insure correctness of response, the proper cue must be present
at or immediately after the release. As the interval of delay
between the light stimulus and the response is increased in
length, we have three possibilities as to the behavior of the
intermediary link or cue. (i) After being aroused by the light
stimulus, the cue may persist, or be constantly maintained,
during the interval of delay. All of the available evidence tends
to disprove such a hypothesis. The raccoons were frequently
distracted during the delays i by various laboratory noises, such
as the squealing of rats and the rattling of windows. Note was
made of these occurrences and still correct responses were pos-
sible in spite of these distractions. The raccoons were exceed-
ingly active during the delays, pawing and clawing and running
all about the release box. I often distracted the animals by
bending down over the release box and yelling at them at the
top of my voice. A typical case occurred when Bob was making
a delay of 15 sees. — with a very high percentage of correct
responses. These distractions during delay lowered his per-
centage approximately eight points. His behavior indicated
that this treatment actually diverted his "attention" from the
problem at hand. The emotional character of such a disturb-
ance makes the high percentage of correct behavior especially
significant. I also continually distracted the children during the
delays by engaging their attention with stories, drawing pictures,
conversation, etc. In fact the attention and interest of the
children were often engaged to the point of absorption by these
devices with no effect upon the correctness of their reactions.

72 WALTER S. HUNTER

The constant maintenance of the cue under. these conditions
of distraction and length of delay is highly improbable. Speak-
ing in conscious terms, it would require great concentration and
mental ability even for a human adult to keep any cue con-
stantly in "mind" during such conditions. (2) The cue might
be some intrinsically intermittent process such as an after-
image. Such substitutes, however, could not suffice to guide
reactions under the conditions of the experiment. Their pres-
ence at the moment of release would be purely accidental and
hence they can not account for the high percentage of correct
responses obtained. (3) We are forced to adopt the third hypo-
thesis that the cue disappears after being aroused by the light
stimulus, and is rearoused in some manner at the moment of
release. To explain the mechanism of this revival, we shall
assume that all three of the intra-organic cues have become
associated during the course of the experiment with some sen-
sory factor connected with the releasing of the animal. Hence
the release is a stimulus which tends to arouse all three intra-
organic tendencies. This revival, however, must be selective
and adaptive and this adaptiveness can be explained by two
additional assumptions. The presence of each light stimulus
at the beginning of delay excites its corresponding intra-organic
factor, and this excitement subsides after the disappearance of
the light. Although the release stimulus does tend to revive
all three factors, yet it will arouse that one most recently active,
viz., that excited by the light at the beginning of the delay.
The assumption that the predisposition of a tendency to response
depends upon its recency of functional activity is a recognized
principle of human psychology.'* With such a mechanism, it
would seem that the problem of delay should present no serious
difficulties. However, the time interval between tests, i.e., the
differences between the recency of excitation of the three fac-
tors, is small in many cases. Learning to enter the box most
recently lighted as opposed to the box most recently entered is
also no easy problem to master.

As we have indicated, such a mechanism would apply only
to the non-orientation cues used by the raccoons and children.
The type of function here involved is ideational in character.

^^ Ladd and Woodworth. Elements of Physiological Psychology. New York,
1911, p. 285. .y s J- 5J'

DELAYED REACTION 73

By applying the term "ideas" to these cues, I mean that they
are similar to the memory idea of human experience so far as
function and mechanism are concerned. They are the residual
effects of sensory stimuli which are retained and which may
be subsequently reexcited. The revival, moreover, is selective
and adaptive to the solution of a definite problem, and when
aroused, they function successfully as a necessary substitute
for a definite component of the objective stimulus aspect of
the problem. The question as to the content nature of these
cues, i.e., their sensory or imaginal character, is reserved for
the succeeding section.

2. The Place of Ideas in the Grades of Animal Learning

A survey of animal reactions from those of the protozoa to
those of the higher vertebrates leads one to the conclusion that
the simplest behavior, from a genetic point of view, is the ad-
justment of a certain movement to a certain object or situa-
tion. This adjustment may be either native or acquired. In
the former case, it is instinct; in the latter, the result of indi-
vidual learning. (As such, it does not seem to be present in
the protozoa.) The remarks in what is to follow are directed
solely to the latter case. In the literature this is referred to
as the stage of learning by experience on a sensori-motor level.
Over against this genetically simple learning, may be placed a
more complex form of behavior which involves a representative
function. This ideational or representative process arises out
of a genetically prior sensori-motor level of behavior. The
field of its functioning is limited, moreover, to the representa-
tion either of some aspect of the object (or sensory) side or of
some part of the movement and its consequences. In other
words, the representative process must stand for either the
sensory or the motor aspect presented in the genetically lower
level of behavior. According to the law of parsimony, the only
conclusive evidence in favor of the existence of such a repre-
sentative element is the case where successful adaptations occur
when that part of the sensori-motor process assumed to be
represented is known to be absent at the moment of response.
If the object or movement to be represented is present, why
assume a representative c)r ideational process? The adjustment
can be explained in terms of sensory stimulus and response.

74 WALTER S. HUNTER

This is the defect in all of the arguments and experiments which
we have examined in the historical section above. Ideas may
have been present, but since all of the behavior can be inter-
preted in terms of stimulus and response, the arguments are
inconclusive.

From this point of view, there are from the standpoint of
function two classes of ideas: — ideas of objects or those repre-
senting the stimulus aspect of the situation, and ideas of move-
ment or those, representing some aspect of the movement or
its sensory consequences. Theoretically, the problem of ideas
can be attacked from either point of view. Many discussions
and experiments do approach the topic exclusively from the
movement side, but practically such a procedure involves
almost insuperable difficulties. Washburn,^" e.g., makes the
number of ideas of movement possessed by an animal more or
less of a rough index of that animal's place in the scale of in-
telligence. But the presence of such ideas is as yet an assump-
tion of very uncertain validity. Furthermore, an experimental
technique that would isolate and control the movement factor
would be extremely difficult, if not impossible to devise. When,
e.g., an animal is brought along a path in a maze to a point
where two possible reactions are presented, both responses are
for the moment inhibited. But the two movements need not
be represented, they may be actually there, although in an
incipient form only, i.e., the conffict may be between the motor
impulses themselves and the conflict may be resolved on this
level without the influence of any factor representative of the
effects of the movements. I doubt whether experimental tech
nique can ever control this movement factor. Quite the reverse
is true with respect to controlling the presence of the object,
i.e., of the determining stimulus. This latter may be given or
withheld at the investigator's pleasure. It is the merit of the
experiments here set forth to have followed such a procedure.
In the present case, there seems to be no room for doubt that
the object reacted to was the light. Now if a representative
function were involved in the behavior of the reagents, as seems
to have been the case with the raccoons and children, it must,
in part at least, have been representative of the lighted box,
because all else — including the three possibilities of movement —

5» Washburn, M. F. Op. cit., pp. 279-284.

DELAYED REACTION 75

was constant from tr al to trial, whereas a selective response
must needs have an alternating cue.

In the light of the evidence in the present monograph, let
us grant the presence, in certain reagents, of a process represen-
tative of objects. The question now arises. Must this process
be imaginal or may it be sensory? We may treat the latter
possibihty in two ways: (i) There may be a sensation arising
from the reagents body — kinaesthetic, e.g., — that stands for a
certain reaction. Or (2), the substitute may simply consist of
a differential meaning attached to the perception of the par-
ticular light box. In this case, when the reagent apprehended
the box, he would simply recognize it as the one in which the
light had been most recently. This would be perceptual recog-
nition as the term is understood in human psychology.

With either of the above explanations, it must be remem-
bered, the question that we are now raising is one concerning
the content of the representative factor. There is no doubt in
my mind that the function is an ideational one. Even should
some critic claim that all of the present behavior is but per-
ceptual recognition, the fundamental difference between the
behavior of the class containing the rats and dogs and that
of the class containing the raccoons and children will still chal-
lenge explanation. If the behavior of both the above classes
of reagents is to be termed perceptual recognition, then two
orders of this must be admitted — one on a level with habit,
the other on a level with ideas.

One class of facts suffices to disprove the possibility of ac-
counting for the behavior of the raccoons and children on the
basis of perceptual recognition. These reagents did not stop
and search for the proper box. They started their reactions
immediately upon being released. The raccoons Jack and Bob.
e.g., might be headed away from the proper box at the moment
of release and yet whirl around in the proper direction and
react immediately. Bob, in particular, might go half way to
the wrong box and then turn suddenly and react successfully.
When the reagent does not follow his orientation, one would
expect him usually to look toward at least two boxes before
reacting, if the behavior were on the basis of perceptual recog-
nition. It is these reactions of children and raccoons that were
not in accordance with orientation and yet that were "head-

76 WALTER S. HUNTER

long" and "precipitate" that lead me to regard perceptual
recognition of the boxes as an inadequate explanation of the
facts.

The objections just stated do not apply to the assumption
that the representative process is an intra-organic sensation.
A positive justification of an ideational function whose content
is an internal sensation may now be elaborated. There is no
doubt but that in human consciousness a sensation may carry
a meaning that is woven into thought sequences. In reading,
e.g., all of the actually discernible conscious content may be
kinaesthetic sensations from the muscles of the throat or may
be auditory sensations, if the reading be aloud. Whether one
say that the sensation is the meaning or that sensation is there
plus a meaning, the case for our purpose is unaltered. Each
sensation and its meaning become incorporated in a train of
thought. A slightly different situation is presented where it
becomes necessary for the sensation to represent that which is
not there and then stimulating the sense organs. The following
illustration from Titchener is a case in point : "I had to carry
across the room, from bookshelf to typewriter, four references —
three volume numbers of a magazine, three dates, and four page
numbers. The volumes and years I said aloud, and then con-
signed to the care of the preservative tendencies. Of the four
page numbers, I held two by visual images, one by auditory,
and one by kinaesthesis."=i The case of the volumes and years
and that of the page numbers remembered by means of kinaes-
thesis are significant. Each is an instance (so far as we can
tell from the account) of memory without the revival of images.
The sensory cues, present at the time when the data were written
on the machine, elicited the proper material. It is a dangerous
procedure to complete another investigator's introspections. I
do not intend to do so here. I simply suggest the above as
a possible supplement to Titchener 's own brief statements.
Whether it be right or wrong in his case, the experience in
which a single sensory process represents an absent object is
sufficiently frequent to give us the suggestion for which we
are seeking. The suggestion may also be found elsewhere.
That which has become known as "conscious attitude" in the

^^ Titchener, E. B. Experimental Psychology of the Thought Processes. New
York, 1909, p. 202.

DELAYED REACTION 77

literature of imageless thought is identical with the first phe-
nomenon that we have had under discussion in this paragraph.
So far as the attitude itself is concerned, it might well be desig-
nated sensory thought as may other experiences also, such as
the reading aloud that was mentioned above. The memory-
cases also may go into the same category. The sensations, both
in the memory instances and in those of the conscious attitudes,
difter, if at all, only on their meaning side and not by the addi-
tion of any overt imagery. Already the reader may have sur-
mised, and rightly, that the writer is introducing the theory
that the representative function found in the raccoons and in
the child, F, at least, who was most nearly comparable with
them, is likewise sensory thought and essentially comparable with
the cases above given.

Let us come to closer quarters with this theory. The pro-
posed hypothesis is equivalent to making the so-called "image-
less thought" genetically prior to "thoughts with images" and
to placing the point of origin at least as low as the raccoon.
As opposed to this, current discusdon by advocates of imageless
thinking would seem to assume the opposite, viz., that " thoughts
with images" are prior." It is to be noted that the assumption
has no factual basis, but that it seems to be the result of preju-
dice or of temperamental leaning. My usage here takes for
granted that imageless thought can be analyzed into either
sensory or imaginal content which carries the meaning. It is
the former case that I use as the t3^e found as low as the rac-
coons. This thought is imageless, though not sensationless, in
the strict sense. In the light of this, I can see no intrinsic objec-
tion to the above theory, while the following points are in its
avor: (i) The evidence in support of the possession of sensa-
tions by animals is absolutely convincing to any save one who
denies all consciousness to animals. On the other hand, the
evidence for images is very meager and unsatisfactory. In-
deed, it is even more so that has been thought if, as I believe
to be true, the present theory of sensory thought will account for
all of the controlled behavior that has been used to support a
theory of images where that behavior is not open to a stimulus
and response explanation. (2) Raccoons and young children

« Buhler is an illu.stration of this. Referred to by Jas, R. Angell, Psych. Rev.,
vol. 18, p. 317.

78 WALTER S. HUNTER

are capable of reactions that seem explicable only on the assump-
tion of the functional efficiency of a representative factor. If
sensations can function in this manner, the law of parsimony
forbids the assumption of images. (3) That sensations can
function in this manner is indicated by the illustrations from
human psychology noted above. It is a different matter, I
know, to say that the sensations of animals also may function
in this manner. But in order to explain the behavior of the
raccoons, it is necessary to assume either the presence of images
or the presence of sensations that function in this manner. The
law of parsimony favors the latter type of process because sen-
sations are genetically earlier than images.

Inasmuch as the rats and dogs, as well as the raccoons, used
sensory processes in the solution of our problem, we are forced
to recognize that sensations may be placed in two classes on the
basis of function: (i) There are sensations that can initiate
a coiTect reaction from at least three possible responses only
when they have not been displaced, during the interval of delay,
hy other sensations of the same modality. Here would fall those
sensory experiences of constant orientation possessed by the
dogs and rats. (Those cases where these animals lost their
orientation and then regained it were few enough to be rated
as accidental.) (2) There are sensations that can initiate correct
responses, under the above conditions, even though they have
been displaced during the interval of delay by others of the
same modality. Here would fall the sensory experiences of the
raccoons and children. The evidence which indicates that no
sensory process need be constantly maintained in order that
these subjects may react correctly is unambiguous. It is this
type of sensory process that I have denominated sensory thought.

On the basis of human introspection, there is another grade or
kind of learning, viz., the stage of the functional efficiency of
images or centrally aroused conscious processes. Why there
should be both sensory and imaginal thought in human ex-
perience is very difficult to say. The most obvious suggestion
would be that imaginal thought, since it is genetically later,
could accomplish tasks which sensory thought could not. I
shall hazard no guesses as to what such tasks might be. At
the present stage of psychological theory, there seems to be no
question but that the distinction of sensory and imaginal thought

DELAYED REACTION 79

itself is valid. Future work must attempt the formulation of
the functional differences between them.

The result of the above theoretical considerations is a
classification of the grades of animal learning different from
those hitherto advanced. Washburn," e.g., has presented
three classes: "First, there is the condition where, so far as
we can see, the animal does not learn by individual experience.
* * * In the next place, we have the grade where the ani-
mal learns by experience, without having the power to recall
an image of its experience. * * * Finally, * * * we
have the possibility of an image." I would modify this scheme
by the insertion of a stage called sensory thought between the
second and third stages. The table would then read: (i) The
animal shows an absence of learning by experience. (2) The
animal is able to profit by experience, but has no higher capacity
than "trial and error" or the "stimulus and response" behavior
would indicate. The rats and dogs of the present tests come
here. (3) The animal can learn by the " trial and error " method.
Indeed, probably most of its reactions are on this basis, being
ruled by stimulus and response. But a new element now makes
its appearance, viz., sensory thought, which is a representative
function of strictly sensory content. The raccoons that I used
are in this class. Of the children, F is the most likely to belong
to this class. (4) The fourth grade reveals the presence both
of "stimulus and response" behavior and of sensory thought,
but added to these is the possibility of directing reactions by
images. The older children of the present tests very probably
belong here rather than in the third class.

VII. SUMMARY AND CONCLUSIONS

The following is a statement of the results and conclusions
that have been reached as a result of the foregoing experiments
and analyses:

I. The rats (one excepted), dogs, raccoons and children made
successful reactions in situations where the customary deter-
mining stimulus was absent at the moment of response. The
stimulus might appear in any one of three boxes. These boxes
were qualitativ^ely alike, but situated in different directions

" Op. cit., p. 270.

80 WALTER S. HUNTER

from the release box. At every trial, three possibilities of reac-
tion confronted the subject. A selection had to be made and
that box chosen in which the stimulus had appeared most
recently.

2. The conditions under which the maximal delay was tested
and the results obtained are indicated as follows:

(a) Different classes of subjects were used. Table XIV gives
the maximum and minimum delays that were obtained from
the different classes.

TABLE XIV

Subjects Min. delay Max. delay-

Rats either no learning or 3rd stage of delay 10 sees.

Dogs 2 sees. 5 mins.

Raccoons 3 sees. 25 sees.

Children 50 sees. 25 mins.

(b) Backgrounds of widely different grades of brightness did
not affect the intervals of delay.

(c) The use of a large release which gave the animals the
freedom of the interior of the box lengthened the intervals of
delay in the case of some subjects.

(d) The use of two boxes as opposed to three lengthened the
intervals of delay by increasing the accuracy of response.

(e) Neither punishment nor the particular number of trials
per day appear to have affected the interval of delay.

3- An analysis of the possible cues that may have been used
by the subjects in the solution of the present problem gave the
following results: (a) Overt orienting attitudes were the prob-
able cues for many reactions of the raccoons. These attitudes
must be assumed as cues for the rats and dogs in order to explain
their reactions, (b) Some intra-organic (non -orientation) factor
not visible to the experimenter must be assumed in order to
explain a significant number of the correct reactions of the
raccoons and all of the successful reactions of the children.
These cues fulfilled an ideational function, (c) All of the re-
agents were influenced by external stimuli that were constantly
present from trial to trial, e.g., those given by the box itself.
However, these could not be used as a basis for selective re-
sponses inasmuch as they were constant from trial to trial and
hence could not furnish varying, or alternating, cues.

DELAYED REACTION 81

4. No animal that had used overt motor attitudes in sohing
the problem when the small release and similar backgrounds
were used adopted another type of cue either when a large
release or when backgrounds of d fferent brightnesses were used.

5. The method used in the present tests for attacking the
question of the functional presence of a representative factor
in an'mal behavior is superior to that of imitation, use of tools
and others that have been used in the past, because here it is
possible to determine what stimulus controls the behavisr. It
is therefore possible to insure the absence of the stimulus at
the moment of response.

6. The representative factor for which search has been insti-
tuted in this monograph stands primarily for "objects" and
not movements. . A technique that would make certain a con-
trol of the latter factor so as to insure its presence or absence
at the will of the experimenter has not as yet been perfected.

7. From a consideration of the theoretical advantages to be
derived from interpreting this representative factor as sensory
rather than as imaginal, a decision was reached in favor of the
former alternative for all reagents save possibly the older chil-
dren, H, Hd, M and L. Illustrations were given from human
consciousness where a sensation performed a memory function
or served as a link in a train of thought. Such cases have been
termed "conscious attitudes" or "imageless thought." This
function, as considered in this paper, was designated sensory
thought.

8. The theory was advanced that such a function as sensory
thought represents the highest grade of behavior in raccoons
and probably also in children of some two and one-half years
of age. This theory is supported by the hardly-to-be-doubted
presence of sensations in animal consciousness and by the
assumption that these sensations can function as the illus-
trations indicate that such processes do in human behavior.
Such a theory seems more in accordance with the law of parsi-
mony than would a theory which made images perform the
representative function found in the raccoons and the child F.

9. From this theory, it follows that subjects may be put into
at least four classes on the basis of the highest type of learning
present in their behavior: (a) Absence of learning; (b) trial
and error; (c) sensory thought, and (d) imaginal thought.

82

WALTER S. HUNTER

VIII. APPENDIX

A. Detailed Records of Two Rats and Two Raccoons. — The
data given in the following two tables are self-explanatory.
They give the course of the delayed reaction tests as these
were presented to the animals.

TABLE

XV

Raccoons

•

Bob

Jill

Delays

Trials

%

Delays

Trials

%

1st stage

10

100

1st stage

15

100

2nd

"

10

100

2nd "

75

97

3rd

u

179

67

3rd "

75

97

Controls were

not perfected until

1 sec.

150

77

the

last 100 trials.

Of these

3rd stage

150

92

72% were ^

correct.

Of last 50

1 sec.

150

84

trials, 80%

were correct.

2 "

75

50

3 ;

sec.

20

80

1 "

30

40

3rd stage

5

100

3rd stage

120

87

1 1

sec.

7

28

1 sec.

155

83

3rd stage

4

50

2 "

105

86

1 1

sec.

10

30

3 "

45

57

learning

17

52

2 "

105

79

3rd stage

50

88

3 "

75

56

1 1

sec.

11

72

Backgrounds

were used now:

3

a

50

56

2 sec.

90

58

1

((

80

70

1 «

45

71

2

«

50

78

3rd stage

150

97

3

«

20

90

1 sec.

75

93

4

((

62

69

2 "

105

96

5

u

56

82

3 "

90

66

8

(1

20

55

2 "

45

55

5

tt

40

75

1 "

45

71

7

u

10

100

3rd stage

60

96

8

u

189

55

. 1 sec.

45

97

last

48

79

2 «

60

98

10

u

25

76

(backgrounds off for 30

100)

12

u

29

68

3 sec.

45

93

15

u

30

73

4 " .

45

62

5

u

1

100

Learning with light on 5 sec.

5-15

u

10

50

30

100

15

u

10

60

Learning with light on 1 min.

20

u

18

44

25

80

2

a

3

33

Wire release i

low, 10 trials daily:

20

u

10

20

Learning

80

95

2

It

3

33

2 sec.

50

98

5

u

9

55

4 «

50

60

10

It

13

46

2 «

100

89

5

It

4

50

last 50

96

7

It

10

33

4 "

50

88

A— 5

tt

49

75

5 «

50

70

7

It

20

75

last 30

90

8

u

36

80

6 "

40

90

10

tt

50

72

7 «

50

84

12

tt

10

90

last 30

70

15

tt

10

90

6 "

30

86

20

a

96

64

Small release

again :

last

, 46

71

6 sec.

30
(dropped)

56

DELAYED REACTION

83

TABLE XV— Continued
Raccoons

Bob

Bob

Delays Trials

%

Delays

Trials

%

Two boxes, a and c, from now on:

2nd stage

10

100

20 sec. 20

95

3rd

"

50

100

25 " 20

90

1

sec.

30

100

30 " 10

90

2

40

97

35 " 70

64

3

20

100

25 " 10

40

4

50

80

B— 15 " 10

60

5

100

85

5 " 10

33

6

50

88

Boxes a and b now used to break

7

100

83

up position habit.

8

100

70

. 5 sec. 60

30

7

100

81

3rd stage 10

40

8

110

89

Learning 295

97

9

80

81

2nd stage 240

86

10

50

86

last 51

93

11

SO

73

1 sec. 51

94

10

50

66

2 " 50

96

Boxes b

and c

from now on

3 " 51

60

10

sec.

40

100

2 " 50

66

12

"

30

100

1 « 10

40

15

"

50

96

3rd stage 51

86

20

"

30

90

Learning c alone

25

"

50

86

21

14

30

"

71

80

Learning all three

35

«

101

70 or less

C— 120

80

4

(t

30

96

last 100

100

5

It

30

100

Learning with light on

5 sec.

8

"

10

100

boxes b and c

15

u

20

55

30

86

8

u

50

86

last 20

100

12

It

90

84

Learning with light on

1 min.

20

"

80

70

30

90

first 50

76

Wire release, 3 boxes:

(dropped)

Learning 50

100

2 sec. 50

48

Two boxes, b and c:

2 sec. 50

94

84

WALTER S. HUNTER

lABLE XVI

Rats

Rat No. 4

Rat No. 16

Delays Trials

%

Delays

Trials

%

1st stage 5

100

1st stage

50

96

2nd " 10

80

2nd "

100

98

3rd " 30

50

3rd "

110

83

2nd " 5

100

last 50

76

3rd " 15

40

2nd "

100

99

2nd " 75

70

3rd "

50

100

3rd " 50

60

1 sec.

40

50

2nd " 75

85

3rd stage

20

60

3rd " 25

92

2nd stage

100

99

1 sec. 30

60

3rd "

50

96

3rd stage 25

88

1 sec.

30

36-

1 sec. 25

52

Different backgrounds now:

Different backgrounds now :

3rd stage

20

30

3rd stage 25

60

2nd "

90

96

2nd " 60

80

3rd "

50

58

3rd " 50

80

2nd "

100

99

1 sec. 25

64

3rd "

50

48

U " 25

48

2nd "

70

100

1 sec. 35

68

3rd "

50

68

li " 50

62

Two boxes now, b and c.

1 " 25

64

3rd stage

50

92

3rd stage 50

64

1 sec.

100

87

Old backgrounds now:

2 "

40

97

3rd stage 65

61

3 "

50

98

2nd " 75

94

5 "

50

90

3rd " 60

66

7 «

35

60

Two boxes now, a and c:

Series with wire release:

3rd stage 25

100

Learning

80

80

1 sec. 20

75

2 sec.

70

78

(dead)

3 "

50

88

4 "

40

92

5 «

80

78

6 "

60

76

7 "

50

90

9 "

40

82

10 «

30

(dropped)

50

DELAYED REACTION 85

B. Notes on Raccoons. — Davis'' and Cole" have given an
excellent description of the habits of raccoons in captivity.
In the main my observations substantiate theirs. A few differ-
ences, however, may be noted. Davis lays particular stress
upon one habit which his raccoons formed, viz., that of covering
their excrement. As well known this is contrary to their be-
havior in the natural state. My four animals were confined in
one cage, 10x10x14 feet, the floor of which was covered an
inch or more deep in shavings. Yet in all the long months
during which the animals lived there, they never formed the
habit of covering their faeces. These were always voided along
the walls of the cage, and not once have I found evidence of
an attempt to cover them. Moreover both Mr. DeVry of the
Lincoln Park Zoological Garden, Chicago, and Dr. Hornaday of
the New York Zoological Garden give an unqualified "no" in
answer to the following question: Will the tame raccoon bury
or cover his excrement, if given the opportunity?

No trouble was found in adapting the animals to a rather
monotonous diet of bread and milk, varied occasionally with
raw meat. But the amount of the rations to be given was more
difficult to determine. Throughout the spring, summer and
into the winter, the raccoon will eat voraciously. But by the
middle of January, unless strict precautions be taken, the ani-
mals will be so fat that they will refuse to work and will sleep
almost continually. My animals were kept on a back porch
which was not artificially heated, and which was but slightly,
if any, warmer than out of doors. Whether they would have
gone into a genuine state of hibernation had I not cut down their
rations in the fall, I cannot pay. However, in view particularly
of Betty's behavior, I am inclined to think that this would have
happened. P'^rom about the middle of January until late in
April, my notebook indicates that Betty lacked a motive for
working. Tests were made during this period in order: (i) To
prevent forgetting of the problem, and (2) to give the animal
food in order that it might not become too weak from a long
fast. The cases of the three other raccoons arc not so extreme
as this one, — yet all become less eager for food during the last

"Davis, H. B. The Raccoon: A Study in Animal Intelligence. Amer. Jour.
Pitych, vol. 18, 1907.

" Cole, L. W. CJlwervations of the Senses and Instincts of the Raccoon. Jour,
of Animal liHiavior, vol. 2, 1912.

86 WALTER S. HUNTER

two months of the winter. Davis found that all of his raccoons
hibernated during the first winter when they were kept out
doors. But during the second winter when they were kept
within doors, although unsupplied with artificial heat, they did
not hibernate. How he regulated the food supply at this period,
and whether the animals became sluggish, he does not state.
Further observations should be put on record before the con-
clusion is reached that the raccoon will change so fundamental
an instinct upon so slight a change of habitat. Both Mr. DeVry
and Dr. Hornaday inform me that their raccoons do not hiber-
nate in the winter, although the living quarters are no warmer
than outdoors. In addition. Dr. Hornaday replies as follows in
answer to the question of how to prevent hibernation: "By
constant feeding. Bears and raccoons hibernate because they
cannot find food in the deep snow. Our bears never hibernate
because they are constantly fed." I, too, fed my animals regu-
larly, although sparingly, in the fall, yet indications of a desire
to hibernate were observed. The subject of hibernation is
very poorly understood at the present time, even among the
biologists proper.

Both of the authorities above quoted inform me that raccoons
reach maturity at three years of age. I note this fact because
Cole's statement that "The year-old raccoons apparently are
not quite full grown "^'^ may be as misleading to some as it
was to me.

Two other observations may be noted in passing: (i) The
raccoon appears to have a very acute vision. I have seen sev-
eral individuals chase flies that were crawling upon the floor
of the experimental room whose illumination was extremely low.
(2) The price of keeping tame raccoons is eternal vigilance. In
the spring .when the "wanderlust" strikes them, they will gnaw
wood and tear wire, — ^anything to escape. And the usual reward
for attempting to catch a loose raccoon is a severe bite.

56 0p. cit., p. 213.

Behavior Monographs

Volume 1

1. The development of certain instincts and habits in chicks. By
Frederick S. Breed. Pp. ivj+78. $1.00.

2. Methods of studying vision in animals. By Robert M. Yerkes and
John B. Watson. Pp. iv+90. $1.25.

3. An experimental study on the death-feigning of Belostoma (-zaitha
aucct.) flumineum say and nepa apiculata uhler. By Henry H. P. Severin
and Harry C. Severin. Pp. iii+47. $.65.

4. The Biology of Physa. By Jean Dawson. Pp. iii + 120. $1.50.

5. The Function of the Vibrissae in the Behavior of the White Rat. By
Stella Bumham Vincent. Pp. iv+85. $1.15.

Volume 2

6. The Delayed Reaction in Animals and Children. By Walter S.
Hunter. Pp. v+86. $1.15.

7. The Canada Porcupine : A Study of the Learning Process. By Leroy
Walter Sackett. In preparation,

8. Audition and Habit Formation in the Dog. By Harry Miles Johnson.
In preparation.

QP391
Hunter

H91

The delayed reaction in animals
and children

imi

iaren // r