^: 4{i§«^' '■MiH' >>'«>JL^llLll r^rf^^lt'itlt ; .^-U-, N. E. Biology of c^ve spiders, 402, 467. Mitzmain, M. B. Behavior of fleas, 401, 467. Monkey, reactions of, 33. Montgomery, T. H. Behavior of spiders, 406, 467. Moore, A. R. Righting movements of the star fish, 398, 467. VI INDEX Morgan, T. H. and ShiiU, A. F. The life cycle of an aphid, 410, 4(37. Morgan, C. L. Instinct and intelligence, 457, 465. Morse, M. Rhythm in phototaxis, 399, 467. Mosquitoes, reaction of, to light, 29. Myers, C. S. Instinct and intelligence, 456, 465. Neger, F. W. Behavior of harvesting ants, 421, 46S. Nest-building behavior, literature on, 407. Nests and nest-building, in l)irds. Part 1, 159; Part 2, 244; Part 3, 336. Newt, behavior of, 305. O'Brien, R. A. Habits of tree-ant, 421, 468. Observations, on termites, in Jamaica, 193, Paramecium, learning in, 67. Parker, G. H. Reactions of sponges, 399, 467; smell in fishes, 438, 470; Mast's " Light and Behavior of Organ- isms," 461. Pearse, A. S. Influence of color on arthro- pods, 79. reactions of amphibians to light, 434, 470. Petrunkevitch, A. Behavior of spiders, 405, 467. Pien e, W. D. Nest-building habits of ant, 422, 469. Pieron, H. The slave-making -instincts of ants, 422, 469. Pigeon, habits of, 278. Polimanti, O. Memory in cephalopods, 399, 467. Porter, J. P. Imitation in birds, 446, 470. Prowazek, S. von. Reactions of protozoa, 400, 467. Rat, white, experiments on sensations of, 125. Ran, P. Behavior of Cecropia, 407, 467. Reactions, visual, in turtle, 1 ; in mosquitoes, 29; trial and error in mammals, 33; in ]iaramecivmi, 67; of the Lacrymaria, 229; of organisms to light, 461. Rose, M. Tropisms, 400, 467. Roubaud, E. The instincts of wasps, 453, <467. Sanders, G. E. Habits of Disognus, 412, 468. Santschi, F. Polyandry in ants, 422, 469. Schaeffer, A. A. Habit formation in frogs, 309; selection of food in stentor, 400, 467. Schimmer, F. Instinct of toleration in ants, 423, 469. Schmitz, H. Relations of beetles to hosts, 423, 469. Sensations, experiments on, in white rat, 125. Shepherd, W. T. Mental life of the mon- key, 435, 470. Shull, A. F. The life cycle of an aphid, 410, 467. Simroth, H. Locomotion of gasteropods, 400, 467. Spiders, literature for 1910, on, 401. Stimulus, relation of, to learning, in the chick. 111. Stout, G. V. Instinct and intelligence, 458, 465. Swift, W. B. Reactions of dog to sound, 437, 470. Termites, observations on, in Jamaica, 193. Thauzies, A. Distance orientation, 143, 470; behavior of homing pigeons, 444, 470. Trial and error reactions, 33. Tropisms, literature on, 401. Turner, C. H. Behavior of bee of family Stelidae, 374 ; literature for 1910 on behavior of spiders and insects other than ants, 401 ; color vision of the bee, 403, 468. Turtle, discriminative ability of, 1. V alues, relative, in learning, 138. Vertebrates, literature on, for 1910, 430. Vickery, R. A. Relations of ant to the root aphid, 424, 469. Viehmeyer, H. Lycaenids and ants, 425, 469. Vision, of turtle, 1; literature on, 403. Washburn, M. F. A discussion on in- stinct, 456. Wasmann, E. Symphilic instincts, 426, ' 469; social parasitism in ants, 428, 469. Watson, J. B. Literature for 1910 on behavior of vertebrates, 430; homing sense of terns, 444, 470. INDEX vil Waugh, K. T. Vision in tlie mouse, 435, 470. Webster, F. M. Effects of mite, 410, 468. Wellman, C. Parasitic habits of an Ephydrid, 428, 469. Wheeler, W. M. Ants, 74; literature for 1910 on behavior of ants, tlieir guests and parasites, 413. Witmer, L. Behavior of monkey, 445, 470. Woodworth, C. W. Control of Argentine rat, 429, 469. Y erkes, R. M. Wheeler on ants, 74; Brehm's Tierleben, 307. Yerkes, R. M. and Bloomfield, D. Reac- tions of kittens to mice, 439, 470. Zeliony, G. S. Hearing in kitten, 437, 470. JOURNAL OF ANIMAL BEHAVIOR Vol. 1. JANUARY-FEBRUARY, 1911. No. 1. THE DISCRIMINATIVE ABILITY OF THE PAINTED TURTLE D. B. CASTEEL Contributions from the Zoological Laboratory of the University of Texas, No. io6 FOUR FIGURES INTRODUCTION In the experiments herein recounted an endeavor has been made to obtain an accurate measure of the ability of turtles of the species Chrysemys marginata to discriminate between lines of different width and direction and patterns of different form. The evidence for or against such discrimination has been obtained by establishing in the minds of the turtles asso- ciations -between certain kinds of lines and patterns and certain pleasant or unpleasant experiences. The reward for successful choice was food; as punishment for failure the electric shock was used. Experiments by various investigators upon a fairly large number of animals have shown that associations between light and darkness, black and white and intermediate shades may be established in a number of cases. In relatively few instances have 'such discriminative associations been established for pat- terns of different form. To the systematic zoologist no phenomena in nature are more evident than the occurrence of those slight differences in shape and markings which separate species of the same genus and closely related genera. For the comparative psychologist the question must often arise, Are the higher animals capable of 2 D. B. CASTEEL visually appreciating these slight differences in form and mark- ing? Are they capable of using them as signs of specific iden- tity? Whether or not such visual discrimination, if it exists, is partly or entirely subordinate to ability to distinguish through the sense of smell, hearing or touch is another question. If we attempt to analyse the method by which an animal is able visually to recognize familiar objects it may well be sup- posed that the impressions which it gains are either general or particular. As men recognize familiar forms and faces at a distance without definitely considering or perceiving the finer details of feature and expression, so animals by observing gross outlines and characteristic movements may draw conclusions as to the nature of the forms around them. Such generalized recognition of familiar objects may be considered, in part at least, to have resulted from an educational process. Its acqui- sition is the result of long association with the objects in ques- tion. Throughout this association an appreciation of certain finer distinctive differences, whether consciously or subcon- sciously manifest, has been a fundamental factor in producing in the animal mind a concrete appreciation of the object's iden- tity in terms of the whole. With the human mind such processes are of common occur- rence. A stranger first met with is afterward recognized by the recollection of some peculiarity which has been impressed upon us. With further acquaintance we forget the peculiarity or it becomes subconscious, while recognition is based on the broader lines of general association. The systematic entomol- ogist, by laborious examination of detailed structure, differen- tiates between closely related species of insects hitherto unknown to him. But if he specializes upon the group, his recognition of the various familiar species forming it in time becomes almost intuitive, — he knows them " by sight " without consciously analysing their differential markings. In the realm of animal intelligence it might be difficult to prove directly that such generalized recognition of individuals and objects was the result of a gradual mental development in which discrimination by appreciation of detail gradually led to a more generalized appreciation of difference. It is possible, however, to subject the animal to tests which will indicate in how far it is able to make fine visual discriminations, and if DISCRIMINATIVE ABILITY OF THE TURTLE 3 ability in this respect is evident, it may be concluded that such visual powers have played some part at least in bringing to the mind of the adult animal a realization of the nature of its more intimate surroundings. With the results of such an investi- gation this paper is chiefly concerned. DESCRIPTION OF APPARATUS The apparatus used in all tests is similar in nature to the electric-box apparatus of Yerkes' though somewhat modified to make it more adaptable to the habits of turtles. It is shown in relief in Fig. I and in ground plan with boxes removed in Fig. II. The large oblong box in which all of the other appa- ratus is placed (except some of the electrical connections) is of heavy galvanized sheet-iron, 78 cm. long, 38 cm. wide and with sides 16 cm. high. It is divided into three regions: an entrance room at one end, an insulated area at the other, on which the boxes stand, and a runway between. The entrance room can be shut off from the runway by dropping the slide operated by a string which runs behind the screen. The insula- tion plate upon which the electric and food boxes stand is of beeswax. It stands 4^ cm. above the floor of the large galvan- ized box. The floor of the runway and entrance room is covered with clean, washed sand, which in the runway is banked up to form a gentle slope leading from the sand covered bottom to the level of the insulation. Water to a depth of 3 cm. covers the bottom of the entrance room and part of the runway. The shallow water and the sandy approach to the boxes are, of course, simply artificial environmental conditions used in an endeavor to simulate natural conditions. The long metal electrode extending in front of the boxes is kept in electrical connection with the water by means of the wet sand which surrounds it. The two metal plates embedded in the insulation are connected with the switch which is in turn connected with the rheostat. The other pole of the rheo- stat is connected with the water. By means of the switch the current may be directed to either one of the insulated elec- trodes. This switch is located behind the screen which conceals the operator from observation. A small peep-hole through the ' Yerkes, R. M., The Dancing Mouse. New York: The Macmillan Company. 1907, D. B. CASTEEL screen allows the operator to observe fully the actions of the animal without himself being seen or in any way influencing the animal's behavior. Upon each insulated electrode, well Fig. 1 — General view of apparatus. toward the back, is fastened a piece of cork behind v/hich meat may be placed. The cork is of such shape that it acts as a blind, DISCRIMINATIVE ABILITY OF THE TURTLE 5 making it necessary for the turtle to crawl to the back of the electrode before the food is discovered. f-^r Heoatat "^ electrode electr-Dclc Irvsuilafi-an. elscirad-e stsarvd- /aier ^v 5Ucij vva-ter- FiG. II — Ground plan of apparatus with boxes removed, showing electrical con- nections. The food and electric boxes are of tin, exactly similar in size and shape, their dimensions being, height 26 cm. width 18 cm., depth 17 cm. They are without bottoms, the lower 6 D. B. CASTEEL edges of their sides resting upon the insulation plate. The side toward the screen has an opening at the top sufficiently- large for introducing the hand in baiting the box. The side toward the runway has an opening at the bottom sufficiently large to allow the turtle to enter easily. The boxes fit loosely in the end of the large oblong box and can readily be removed or interchanged by the operator. They are protected from the sides of the large box by cardboard insulation. In a few of the earlier experiments the designs or shades used w^ere painted on the boxes, but for several reasons this method proved unsatisfactory and was discontinued. The boxes most generally used are as exactly alike as it was possible to make them. Their distinctive character is given them by the attachment of pieces of cardboard which bear the desired de- signs. Upon each box are placed two of these boards. One completely covers the outer surface which faces the runway; the other covers the inner surface directly opposite the opening through which the turtle enters in search of food. This latter is immediately behind the cork blind which conceals the food and is intended to give continued emphasis to the design after the outer surface of the box has passed from view. These design boards are fastened to the boxes with small clips and can readily be removed. In practice it was customary to change them from one box to another during or between a set of trials to avoid the chance that the boxes themselves, through some unnoticed peculiarity of form or odor, might aid in influencing the choice. During most of the experiments the electric current used was taken from a commercial circuit, being reduced to a proper degree by use of a rheostat. The severity of the shock used was varied considerably for different individual turtles. As a rule a shock of very slight degree, one not unpleasantly perceptible to the moistened fingers of the operator, was most effectual. GENERAL METHODS AND CHARACTERISTIC REACTIONS All of the turtles upon which observations were made were kept in an aquarium and tamed for from three to six months before experiments were undertaken. This is particularly neces- sary when dealing with a reptile. Chrysemys is probably one of the most tractable water turtles, yet some individuals never DISCRIMINATIVE ABILITY OF THE TURTLE 7 become sufficiently docile to allow of satisfactory experimen- tation. Most of my turtles, however, finally became so tame that they could be handled gently without alarm and would take food readily, even while being held in the hand. A well conditioned Chrysemys is a voracious feeder and if not allowed to gorge unduly will be eager for a daily ration of food. It is an active turtle and when well tamed will swim or crawl rapidly to the edge of the aquarium in hope of food whenever anyone approaches. For these reasons it makes a fairly satisfactory experimental animal. Feeding can be used as a reward in the formation of associations and tests can be conducted with a fair degree of rapidity. These statements, however, apply only to turtles which are well tamed and in a satisfactory physiolo- gical state. No animal is more unsatisfactory or taxes the patience of the operator more severely than a disgruntled turtle, whether its sulkiness arises from fear or from physical disability. In describing experiments the term "trial " is used to express one individual attempt to choose between electric and food box. A series of successive trials, usually ten in number and immediately following each other, is denominated a "test." As a rule one test per day was given and upon successive days. In certain cases they were given less often, particularly when a turtle's eagerness for food became lessened. As may be noted from the tables, twenty trials were sometimes given at a test. It was thought that a larger number of trials per day might give better results, but the records show no more than a pro- portionate acceleration in learning. In all cases an effort was made to subject the animal to tests and trials occurring with a fair degree of regularity. In practice the turtle to be experimented upon was gently removed from the aquarium and placed in the entrance room. The operator then retired behind the screen having assured himself that the boxes were properly placed and baited and the electrical apparatus adjusted as desired. Food was always placed in both boxes so that its odor, if appreciated, might not serve as a clue. By pulling a string from behind the screen the slide was then raised between the entrance room and the runway and the turtle thus given opportimity to approach the boxes. This might occur immediately or be somewhat delayed 8 D. B. CASTEEL according to the inclination of the animal. If the turtle was eager for food it would usually swim rapidly into the runway, crawl up the sandy "bank," making choice the while of the box into which it would enter. If it entered the "right " box it received no repulse but continued on to the back of the box, found the meat hidden behind the cork blind, turned around, scrambled out of the box and down the bank into the water. Here it might stop to eat the meat or might continue on to the entrance room, in which case the slide was immediately dropped and the operator rearranged the boxes. Ifit remained in the runway while it w^as eating the meat or afterward, it was urged to enter the entrance room and the slide was dropped before another trial. If the turtle chose the " wrong " box it received an electric shock at the entrance to the box immediately upon touching the electrode upon the floor of the box with its fore feet or snout, its hind feet being in contact \\ith the long electrode in front of the boxes or with the wet sand. As a rule this caused it to jump backward though it might or might not retire to the water. If the turtle remained in the runway it was, as a rule, replaced in the entrance room and the slide dropped before another trial was given. However, at the beginning of its training, a turtle would at times be allowed to remain in the runway until it found the " right " box and obtained the food from it, though the trial was counted as " wrong " if it at any time approached the electric box sufficiently near to receive an electric shock. It was considered that a turtle had learned any particular problem when it was able to make a perfect record of correct choices in three successive tests of ten trials each. The relative position of food and electric boxes was changed with irregular sequence during each series of trials so that learn- ing by an appreciation of regular alternation might be avoided. If no change of position was made between two trials the boxes were lifted and replaced that lack of noise might not give a clue. Full records were kept during the progress of each experi- ment showing the relative positions of the boxes at each trial, the time consumed in making each choice, the success or failure of the trial and miscellaneous data regarding the actions of the individual throughout its experimental history. The descrip- tions of the experiments and the appended tables are sum- marized from these records. DISCRIMINATIVE ABILITY OF THE TURTLE 9 The above account is a very general one. Many variations oc- curred in the behavior of turtles during the course of the experi- ments. Different turtles would differ greatly in the methods by which they seemed to attack the problem, and any individual turtle might vary its course of procedure greatly from time to time. Some of the more accomplished turtles would go through the apparatus with machine-like precision, going straight for a box when the slide was lifted and straight back again into the entrance room after getting their reward or electric shock. But such action was more characteristic of turtles that were nearing perfection in any particular experiment. Such automaticity probably resulted from several rather definite causes. The individual had. become, through numerous trials, well accus- tomed to travel and retravel a beaten path. As a rule its mis- takes were few and its mental poise was not continually dis- turbed by annoying shocks. The most tractable, possibly the most " intelligent " turtles, showed greatest regularity of behavior at this time. Seemingly nervous, erratic turtles, those which appeared never to become entirely tame, showed lack of auto- maticity. A direct return to the deeper water of the entrance room by a turtle that has succeeded in securing the bait is a natural action, since Chrysemys appears unable, or at least unwilling, to tear up and swallow its food unless its mouth is beneath the water. Unfortunately for the patience of the operator many of the turtles, particularly during the earlier tests of an experiment, were prone to hesitate and wander around in both the entrance room and the runway. Or an individual might leave the entrance room and remain quietly in the runway before the boxes for some time, seemingly deliberating before making a choice. That this was not always mental deliberation upon the par- ticular problem in hand the writer is fairly convinced, for the result of such hesitation was about as likely to be disaster as gain. Turtles upon receiving the electric shock were diversely affected by it. After a few such experiences some were ren- dered so nervous and wild that their withdrawal from experi- mentation was necessary. Others were at first thus affected but gradually became accustomed to the experience or reacted favorably under lessened stimulus. Some were simply rendered 10 D. B. CASTEEL sulky and refused to approach the boxes at all or even to leave the entrance room. Some, after each shock, would dart back quickly, turn and scurry away from the box, but were appar- ently eager to try the experiment again. Others would with- draw slightly from the electrode and would soon after, if allowed, try the same box again or slowly move over to the other box. Any individual might exhibit during the course of an experiment one or several of the idiosyncrasies above noted. But the most troublesome form of behavior from the stand- point of the operator was the development of a habit of right or left turning. This tendency of an animal always to go either to the right box or to the left, no matter how irregularly their relative positions were changed, might appear at any time during the course of its training. The habit might become firmly fixed or merely be transient ; but if not quickly overcome usually put an end to that animal's usefulness for experimental purposes. As a rule such turtles soon became sulky, since they received a large number of shocks. DESCRIPTION OF TURTLES The turtles used in all of the experiments were the Western Painted Turtle, Chrysemys marginata. This familiar turtle is semi-aquatic in habit, spending much of its time in the water where it may be found crawling over the bottom in search of food, darting through the water in pursuit of more active prey, or floating idly on the surface with head and neck extended. Upon warm days the turtles often leave the water, crawling up the bank or upon protruding logs where they lie basking in the sun. A more extended account of their habits is given by Newman^ Except for certain lines and markings Chrysemys is not a conspicuous turtle. The carapace is dark olive-green in shade with somewhat lighter markings separating the scutes. The marginal scutes bear dark red and yellow blotches. The plastron is bright yellow. The ground color of the head, neck and legs is similar to that of the carapace, dark olive-green; but the sides of the head and neck are conspicuously striped with bright yellow and red markings, longitudinal in direction, the yellow 'Newman, H. H., The Habits of certain Tortoises, Jour. Comp. Neur. and Psych., 1906, vol. 16, No. 2. DISCRIMINATIVE ABILITY OF THE TURTLE 11 predominating. The appearance of these conspicuous lines first suggested to the writer the possibility that Chrysemys marginata might readily distinguish between differences in line markings since a linear design forms so prominent a feature in the decora- tion of this species. All of the individuals used were collected around Ann Arbor, Michigan, where much of the work was done. Those shipped to Texas did not well undergo the hardships of the journey, nor have they since reacted as satisfactorily as before the trip. For this reason it has been necessary to discontinue the experi- ments somewhat earlier than had been planned, since their continuance with apparently abnormal animals would vitiate the results of the entire work. Fifteen turtles were used during the course of the experiments but the records of only seven will be given. For various reasons the results obtained from work on the others do not justify publication. Some sickened and died before the completion of a series of tests; others became sulky, lost appetite, were slug- gish or developed permanent habits of right or left turning. The following data refer to those directly mentioned in this paper: Length of Width of Turtle Number. Carapace. Carapace. Sex. 1 62 mm. 54 mm. Female. 4 106 mm. 82 mm. Female. 8 112 mm. 87 mm. Female. 9 92 mm. 80 mm. Female. 10 133 mm. 97 mm. Male. 11 98 mm. 76 mm. Female. 13 128 mm. 97 mm. Male. RESULTS OF EXPERIMENTS The experiments may be divided into three groups, each group being characterized by the nature of the test to which the animals were subjected. At the beginning of the work the writer desired to obtain some estimate of the ability of Chrysemys to form associations. For this purpose the Black and White Discrimination tests were tried, since similar experiments had been conducted by other workers with several other animals. The results of these tests do not bear so particularly upon the problem in hand but are here given for the sake of those who care to make comparisons. The other two groups, those show- 12 D. B. CASTEEL ing the results of tests for Pattern and for Line Discrimination, form the major portion of the work. BLACK AND WHITE DISCRIMINATION In these tests the boxes used were painted black and white respectively. Tests with turtles Nos. i and 2 were conducted without the use of the electric shock for punishment, food being the only incentive to success and forcible return to the entrance room without reward the only punishment. The apparatus used with these turtles was somewhat different from that already described, being less elaborate and, of course, lacking all elec- trical connections. Turtle No. i was given, as a rule, ten trials in succession every other day. On several days the turtle was sluggish and a fewer number of trials necessary. In a few instances more than one day intervened, necessitated by the animal's lack of eagerness and activity. It was fed in the white box only. In this experiment no preference tests were taken. The results, given in table i, show that an association between the white box and food was clearly established. The result of the tests for memorv is also to be noted. ml Turtle No. 11 was fed in the black box. It was induced to enter this box twelve times before the actual tests were begun. In all other respects it was treated in a manner similar to No. i. Its record, given in table 2, shows but two perfect tests although the general average of successful trials is good. The two memory tests, taken twelve and twenty-one weeks after the last test of the series, are not perfect though as good as that test. Turtles Nos. 8, 9, and 13 were tested with black and white boxes during the latter course of the work for the sake of obtain- ing additional evidence for or against this type of discrimination. Preference tests were made in all cases and the electric shock was used as punishment. Both No. 8 and No. 9 had, in a pre- vious experiment for pattern discrimination, been accustomed to entering black boxes on which white patterns were displayed. Their decided preference for black may thus be explained. In both cases the food box w^as white so that the former preference had to be overcome. No memory tests were taken. The records of these turtles for black and white are found in tables 3 and 4. Turtle No. 13 was decidedly the most unsatisfactory animal DISCRIMINATIVE ABILITY OF THE TURTLE 13 TABLE 1 Turtle No. 1, Black and White Discrimination Fed in white box Test. Right Wrong 1 8 2 2 2 3 3 10 0 4 2 0 5 7 3 6 10 0 7 9 1 8 9 1 9 8 0 10 9 1 11 8 2 12 9 1 13 8 2 14 10 0 15 10 0 16 8 2 17 10 0 18 10 0 19 9 1 156 19 TABLE 2 Turtle No. 11, Black and White Discrimination Fed in black box Test 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Memory tests: After 12 weeks, 8 light, 2 wrong; after 21 weeks, the ^.ame. Right Wrong 9 1 8 2 8 2 9 1 9 1 8 2 9 1 7 3 9 1 8 2 7 3 9 1 10 0 10 0 8 2 128 22 Memory test after 6 weeks, 10 right, 0 wrong. TABLE 3 Turtle No. 8, Black and White Discrimination Result of preference trials. Block 34, ivhite 16. Fed in white box Test Right Wrong 1. 4 6 2 3 7 3 4 6 4 8 2 5 7 3 6 7 3 7 5 5 8 10 0 9 10 0 10 8 2 11 9 1 12 10 0 13 ;.. 10 0 14 10 0 105 35 TABLE 4 Turtle No. 9, Black and White Discrimination Result of preference trials: Black 16, white 4. Fed in white box Test Right Wrong 1 3 7 2 2 8 3 3 7 4 4 6 5 8 2 6 6 4 7 8 2 8 5 5 9 7 3 10 9 1 11 10 0 12 10 0 13 8 2 14 8 2 15 10 0 16 10 0 17 10 0 121 49 14 D. B. CASTEEL TABLE 5 Turtle No. 13, Black and White Discrimination Result of preference trials: Black 16, white 24. Fed in black box Test Right Wrong 1 7 3 2 4 6 3 3 7 4 6 4 5 5 5 6 3 7 7 6 4 8 5 5 9 6 4 10 3 7 11 4 6 12 6 4 13 4 6 14 6 4 15 5 5 16 7 3 17 6 4 18 7 3 19 6 4 20 7 3 21 6 4 22 4 6 116 104 used in any of the experiments described. It was a large, slowly- moving, rather sulky animal, extremely difficult to tame and apparently in need of but little food. It early developed a habit of right turning, and at no time during its 220 trials showed consistent improvement. As compared with sofhe others, notably No. 10, it illustrates well the marked difference in behavior which different turtles may show. Its record is given in table 5. PATTERN DISCRIMINATION As the sequel will show the above heading shoiild rather read lack of pattern discrimination. The patterns used, reduced one-half in size, are shown in Fig. III. Their areas are equal. It may well be argued that the choice of patterns was unfor- tunate and that designs more markedly different should have been used. The history of but two turtles will be given here. Several others were tried but through seeming lack of ability DISCRIMINATIVE ABILITY OF THE TURTLE 15 to profit by their experiences they received so large d, number of shocks that they were rendered sulky and work with them had to be discontinued. In several cases, as shown in table 6, the white designs were placed on pendulums and caused to swing slowly back and forth in front of the black boxes in the hope that the moving object would attract and concentrate the attention of the animal. As far as could be observed the results were no more satisfactory. Preference tests were taken in all cases and the electric shock administered for failure. As is shown in tables 6 and 7 only negative results were obtained for discrimination between the two patterns used. A 3 Fig. Ill — Designs used in experiments on Pattern Discrimination. Reduced one-half TABLE 6 Turtle No. 8, Pattern Discrim- ination Result of preference trials: Pattern A, 19; pattern B, 21. Fed in pattern A box Test. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. Test Right Wrong 4 6 2 8 7 3 5 5 0 3 6 4 1 4 5 5 6 4 4 6 5 5 6 4 3 7 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. Right Wrong 6 4 7 3 3 7 6 4 Swinging pattern introduced. 4 4 2 7 7 7 1 7 7 6 128 6 6 8 3 3 3 9 3 3 4 130 16 D. B. CASTEEL TABLE 7 Turtle No. 9, Pattern Discrim- ination Result of preference trials: Pattern A, 19; pattern B, 31. Fed in pattern A box Test Right Wrong 1 10 5 2 7 3 3 5 5 4 5 5 5 4 6 6 4 6 7 4 6 8 5 5 9 6 4 10 9 11 11 9 11 12 9 11 13 11 9 88 87 LINE DLSCRIMINATION The most definite and satisfactory results were obtained in testing for this type of discrirrination. The experiments may be divided into two groups: I. Discrimination between two lined boxes on which parallel lines of equal width run horizontally on one box, vertically on the other. II. Discrimjination between two lined boxes on which the parallel lines run in the same direction on both boxes but are of different width. In all cases the amount of black and white on the faces of the boxes was the same. Great care was taken that the faces of the boxes exhibiting the lines should show no other difference which might aid the turtle in making a choice. The electric shock was used for punishment and preference tests were taken in all cases. Group I Experiments in line discrimination under the conditions described for Group I were conducted as follows: The boxes were faced with cardboards bearing parallel lines, black and white, each 8 mm. in width. On one of the boxes the lines ran vertically, on the other horizontally. Fig. I illus- trates the apparatus arranged for this experiment. Preference tests were taken and the box showing the lesser number of *?3 DISCRIMINATIVE ABILITY OF THE TURTLE 17 entrances . was chosen as the food box. Discrimination was then established by a series of tests using difference in the direc- tion of the parallel Hnes as the criterion of selection. If dis- crimination was satisfactorily established with the lines 8 mm. in width, the cardboards bearing these lines were then removed and boards bearing lines but 4 mm. wide were substituted on both boxes. If discrimination was fairly maintained after this reduction in the width of the lines they were then further re- Sinm. f 77* 77^. i mm. UntTn. imtri. Fig. IV — Graphic representation of the widths of hnes used in the experiments on Line Discrimination. The sides of the boxes facing the runway were covered with hnes of the widths shown. duced to 2 mm. on both boxes. No lines narrower than 2 mm. were used in the Group I experiments. At each reduction the number of black and white lines on the faces of the boxes was doubled though the amount of black and white surface remained the same. Fig. IV illustrated graphically the various widths of lines used in the Line Discrimination experiments. The histories of turtles Nos. 4 and 10 illustrate the results of the horizontal-vertical line experiments. No. 10 showed much greater speed in acquiring the association than No. 4. This latter turtle became sulky upon reduction in the width of the lines to 2 mm. No. 10 was in every way my most satis- factory animal. It learned the problem fairly rapidly, showed perfect memory for 8 mm. lines two weeks after learning them, 18 D. B. CASTEEL and was able to make distinction with a fair degree of accuracy when the lines were reduced to 2 mm. In the series of six tests (shown in table 8) with lines 4 mm. wide an average of 95 per cent of the trials were correct. When the lines were reduced to a width of 2 mm. on both boxes the first five tests resulted in 80 per cent of the trials being correct. With a continuance of 2 mm. line tests this percentage was lowered in five tests to an average of 68 per cent. It would thus appear that 2 mm. is about the limit of this turtle's discriminative ability for line markings running in different directions. Turtle No. 4 was long in learning the difference between the 8 mm. line boxes. Its first five tests with 4 mm line boxes gave a record of 80 per cent correct choices, and this was raised to 90 per cent in the five succeeding tests. With reduction of lines to 2 mm. the average of successful trials dropped to 54 per cent when the sulky disposition of the animal brought the experiment to an end. TABLE 8 Turtle No. 10, Vertical and Hori- f zoNTAL Line Discrimination Restdt of preference trials: Vertical 18, horizontal 22. Fed in vertical Lines 4mm. wide Test Right Wrong 1 10 0 2 9 1 3 9 1 4 10 0 5 9 1 6 10 0 Innes 8m m wide Test Right Wrong 1. 5 5 2. 2 8 3. 4 6 4. 4 6 5. 10 10 6. 11 9 7. 15 5 8. 16 4 9. 18 2 10. 9 ] 11. 10 0 12. 10 0 13. Memory test, 20 134 0 56 two weeks 10 0 57 Lines 2mm. imde 1 10 0 2 6 4 3 8 2 4 9 1 5 7 3 40 10 Lines 2mm. ivide (cont.) 1 6 4 2 8 2 3 4 6 4 8 2 5 8 2 34 16 DISCRIMINATIVE ABILITY OF THE TURTLE 19 TABLE 9 Turtle No. 4, Vertical and Hori- zontal Line Discrimination Result of preference trials: Vertical 43, horizontal 57. Fed in vertical box Lines Smm wide. Lines Amm. wide Test Right Wrong Test Right Wrong 1 1 9 2 1 9 3 2 8 4 4 6 5 5 5 6 5 5 7 5 5 8 5 5 9 1 9 10 2 8 11 4 6 12 1 9 13 3 ,7 14 5 5 15 4 6 16 5 5 17 4 6 18 8 7 19 8 2 20 6 4 21 10 0 22 10 10 23 9 1 24 5 5 25 17 3 26 15 5 27 18 2 28 19 1 29 19 1 30 9 1 31 7 3 32 17 3 33 16 4 34 9 1 35 10 0 36 10 0 37 10 0 1.. 8 2 2.. 8 2 3.. 7 3 4.. '9 1 5.. 8 2 40 10 Lines 4mm. wide (cont.) 1.. 9 2.. 9 3.. 9 4.. 9 5.. 9 45 5 Lines 2mm. wide 1.. 5 5 2.. 8 2 3.. 6 4 4.. 3 7 5.. 5 5 27 23 289 166 Group II While the results of the foregoing experiments indicate that turtles can leam to discriminate between two series of lines running in different directions, even though these lines are reduced in width to but 2 mm., they do not necessarily show that the turtles are able to discriminate between individual 20 D. B. CASTEEL lines of that width. Since in the previous experiments all of the parallel lines on one box ran vertically and all on the other horizontally, appreciation of difference might simply mean that the turtle discriminated between the general direction taken by the whole series of lines, and would be able to do this until the lines were so reduced in width that only gray surfaces were discerned. The results thus far given are of themselves inter- esting in that they indicate the ability of these turtles to appre- ciate the difference between fairly narrow line markings which run in different directions; but they do not serve as an accurate measure of the turtle's ability to distinguish between lines of different width, nor do they give us an exact idea of the extent of the animal's ability to distinguish relative widths of different lines. So it seemed desirable to conduct another series of experi- ments in which the lines on the faces and backs of the boxes all ran in the same direction, either vertically or horizontally, but differed in width on the two boxes. In these tests the turtle was first given choice between two boxes, one of which bore lines 8 mm. wide and the other lines much narrower (i mm. in one experiment, 2 mm. in the two others described). The records of turtles Nos. 10, 8 and 11 will be presented. Upon both of the boxes used in tests with No. 10 the parallel lines ran vertically; upon those used with Nos. 8 and 11 they ran horizontally. Unfortunately, after establish- ing discrimination between 8 mm. and 2 mm. lines with turtles No. 8 and No. 11, it was necessary to discontinue the tests. Their records are found in tables 11 and 12. Turtle No. 10 was carried much farther. It is possible that the former experiences with line boxes which this turtle under- went may, to some extent, have so familiarized it with a similar problem that its performance with these boxes would average better than the rule. No. 10 was started with boxes bearing lines 8 mm. and i mm. in width respectively. When this com- bination was learned the 8 mm. box was refaced with boards bearing lines 4 mm. wide. The choice then lay between a box faced with i mm. lines and one faced with 4 mm. lines. The I mm. lines had indicated the electric box in the preceding tests and were so retained, while the 4 mm. lines replaced the 8 mm. width on the food box. In fifty trials with these lines 96 per DISCRIMINATIVE ABILITY OF THE TURTLE 21 cent of the entrances were correct. The box with i mm. lines was then faced with boards bearing 2 mm. lines and a grade of 94 per cent correct was made in fifty trials. In this series the 4 mm. lines still indicated the food box while 2 mm. lines marked the electric box. The 4 mm. lines were next exchanged for a series 3 mm. wide, so that the problem before the turtle was to choose between two boxes bearing lines on their faces which differed but i mm. in width. The 3 mm. lines were on the food box, the 2 mm. on the electric box. The first fifty trials of this combination resulted in 82 per cent of the choices being correct and the next fifty trials of the same problem resulted in a grade of 92 per cent correct. As will be noticed, the changes of line width were alternated TABLE 10 Turtle No. 10, Discrimination Between Lines of Dif- ferent Width Lines 8, 4 and 3mm. were on the food box. All lines vertical Lines 8mm. and 1mm. wide Lines -imm. and 2mm. ivide Test Right Wrong Test 1 2 8 1 2 3 7 2 3 6 4 3 4 9 11 4 5 9 11 5 6 12 8 7 16 4 47 3 8 18 2 =— - 9 10 0 10 9 1 Lines Snun arui 2mm. wide 11 8 2 1 6 4 12 9 1 2 9 1 13 10 0 3 :.::: s 2 14 10 0 4. 9 1 15 10 0 5 9 1 Right Wro; NfG 8 2 9 1 10 0 10 0 10 0 141 59 41 Lines 4mm. and 1mm. ivide Lines 3mm. and 2mm. wide 1 8 2 1 10 0 2 10 0 2 9 1 3 10 0 3 10 0 4 10 0 4 9 1 5 10 0 5 8 2 48 2 46 22 D. B. CASTEEL from food to electric box so that the turtle always had one box similar to one in a previous series of tests. This arrangement doubtless aided the turtle in maintaining a high average of correct selections. The tests were carried no farther than the 2 mm. -3 mm. combination since it was felt that these lines were sufficiently similar in width to indicate a high degree of discriminative ability. The tabulated results of No. lo's tests are found in table 10. TABLE 11 Turtle No. 8, Discrimination Between Lines of Dif- ferent Width Lines on both boxes horizontal. 8mm. lines on food box Lines Smm. and 2mm. wide Test Right Wrong 1 5 5 2 7 3 3 6 4 4 3 7 5 8 2 6 6 4 7 7 3 8 7 3 9 8 2 10 14 6 11 6 4 12 9 1 13 5 5 14 8 2 15 7 3 16 9 1 17 9 1 18 10 0 19 10 0 20 10 0 154 56 TABLE 12 Turtle No. 11, Discrimination Between Lines of Dif- ferent Width All lines horizontal. 2mm. lines on food box Lines Smm. and 2mm Test Right Wrong 1 6 4 2 5 5 3 4 6 4 5 5 5.... 9 1 6 8 2 7 8 2 8 6 4 9 8 2 10 7 3 11 6 4 12 10 0 13 7 3 14 7 3 15 9 1 16 10 0 17 10 0 18 10 0 135 45 INCIDENTAL RESULTS As an aftermath to work on any problem in animal behavior, whether under natural surroundings or experimental control, there may always be gathered certain facts and conclusions not primarily sought in the investigation, yet often bearing impor- tant relations to it or yielding interesting data of a more general character. Such results will now be noted. Individuality. My turtles exhibited distinct individual differences DISCRIMINATIVE ABILITY OF THE TURTLE 23 in their mode of behavior under experimental conditions. A few of these have already been noted. Some turtles failed partially or completely to solve problems which others acquired with little seeming difficulty. Such cases of individual failure to learn appear to have been dependent upon two factors: indi- vidual lack of mental ability or lack of tractability. It is com- mon knowledge that an animal with a surly disposition cannot be taught tricks. Any turtle that refuses food, retires within its shell upon slight provocation and shows a pronounced dis- position to sulk is useless as an experimental animal. Almost any wild adult turtle will act in this manner when first handled but with proper care should gradually become more tractable. However, some of my turtles never became accustomed to captivity and these were finally discarded. Others would respond most satisfactorily for a time and then gradually or suddenly lose interest or become sulky. Such turtles might or might not change in disposition and be useful for later experiments. With- out doubt turtles also differ greatly in mental ability. Some appeared willing enough to learn, were cheerful and active, yet did not succeed in learning as rapidly as others. The persistence of individual characteristic movements. In addition to individual mental characteristics of a general nature some of the turtles showed certain definite peculiarities which were expressed by distinctive movements and reactions. No. lo almost invariably stopped upon entering the food box and insisted on biting the cork behind which the meat was placed. This was the one persistent " stupid " action of this turtle throughout its experimental history. Occasionally it would simply stop before the cork and crane its neck around the blind until the food came into view. Then it would either advance and proceed to take the food or would bite the cork again before doing so. It seemed continually to labor under the delusion that the reward might most easily be obtained by tearing away the intervening obstruction; at least such an explanation pre- sents itself as a plausible human interpretation of this habit. No other turtle acted in the same way. Directness or indirectness of approach and entrance into the boxes varied in different cases. As has before been intimated, physical quickness was not necessarily associated with mental alertness or deliberate movements with dullness. Neither was the converse true. The most active turtles were just as likely to choose unwisely as the deliberate ones, and vice versa. But this quality of agility of movement or its reverse usually mani- 24 D. B. CASTEEL fested itself throughout a turtle's performances and was char- acteristic of the animal. There were a few exceptions to this general rule, cases of sulkiness or sluggishness resulting from ill health or too frequent punishment. The tendency toward right or left turning was of frequent occurrence during the course of the experiments. Some turtles never acquired it, others for a short time only, but in most cases when it once became established its permanency was assured. The writer has no explanation to offer for the initial cause of such a habit, but merely suggests that once such a tendency is well started its continuance is assured by the potency of kinaesthetic sensation and association. Yerkes' long ago showed the facility with which land tortoises learn the windings of a simple maze, and in some early experiments with Chrysemys I obtained results which indicate that this water turtle learns such problems with comparative ease. As Watson'' has shown for the rat, the ability of an animal to learn a maze is largely dependent upon the kinaesthetic sense,-- it is a matter of mus- cular appreciation of distance and direction. Once it is estab- lished all other senses are subordinate to it. In the same way it may be supposed that once a tendency toward turning in the direction of a right or left box is (for some unknown reason) established, it will tend to persist even though other senses warn against such procedure. Mental instability. The mind of the turtle is erratic. A glance over the tables will show instances of pronounced failure closely following successful tests. One day the turtle will be near perfection and upon the next will record a series of failures. Such lapses may occur well toward the end of a series of tests otherwise excellent. I am convinced that these seasons of forgetfulness would continue to occur at irregular intervals no matter how long the training of the animal was continued. With any animal whose mental capabilities are no more highly devel- oped than those of the turtle it is scarcely to be expected that attention to the problem at hand will always be alert or that memory will always serve as a reliable guide. ^Influence of former experiences. Much has been written of recent years regarding the influence of past experiences upon animal behavior, and of the necessity of a thorough acquaintance with the past histories of animals subject to experiment. One ^Yerkes, R. M., The Formation of Habits in the Turtle, Pop. Sci. Mo., 1901, vol. 58, p. 519. * Watson, J. B., Kinaesthetic and Organic Sensations: their role in the reactions of the white rat to the maze. Psych. Rev. Monograph Supp., 1907, vol. 8, No. 2. DISCRIMINATIVE ABILITY OF THE TURTLE 25 section of the present work illustrates definitely the effect of such experience upon the animals concerned. This is found in the pronounced preference of turtles Nos. 8 and 9 for black boxes as the result of having participated in a former experi- ment in which the boxes were painted black to emphasize a white pattern. Had this former experience not been known, and had the turtles in the black-white tests been fed in black boxes without preference tests, the results would probably have given an erroneous idea of these turtles' mental ability. This is, of course, a rather special case, but it serves as a reminder that in dealing with such illusive qualities as mental character- istics the utmost caution must be taken if errors of judgment are to be avoided. On these grounds the results obtained with Nos. I and 11 in the black-white tests and also their memory tests are open to question since no preference tests were here taken. In discussing No. lo's record for discrimination between lines of different width it was suggested that the relative ease with which it discriminated between lines 2 mm. and 4 mm., and 2 mm. and 3 mm. in width might partly be explained by the fact that immediately before these tests it had passed through a large number of tests of a somewhat similar nature; in other words, that its former experience had to a certain extent pre- pared its mind to appreciate the latter problem. This is merely a suggestion and no definite claims are made that the animal's mind was more acutely receptive through having previously solved somew^hat similar problems. There is far from being sufficient evidence upon which to base statements of this turtle's abilitv to profit by such educational processes. The relation of time to successful trials. My work with Chry- semys shows conclusively that for this animal any calculation of mental ability based upon the time taken by the turtle on its trips from entrance room to food or electric box can give no accurate measure of its progress in learning. This is particu- larly true if several individuals are classed together and com- pared, and it also applies to a comparison of the performances of any one individual if compared at different periods during the X)rogress of an experiment. Not only do different individuals vary greatly in the speed with which they traverse the runway and make choice of the boxes, but any individual's speed may 26 D. B. CASTEEL vary enormously, often independently of the failure or success of its choices. It is true that first attempts are usually deliberate and that imsuccessful turtles may be rendered inert through continued punishment, but aside from these considerations the above statement is correct. Without attempting to make too detailed a classification, it may be said that some of the follow- ing factors determine the time consumed by a turtle in making its trips from entrance room to box: (i) temperament of the individual, which would remain fairly constant but would differ greatly if different individuals were compared; (2) physiological state, varying for the individual during a series of tests and dependent on general health, unavoidable fluctuations in temperature and amount of food taken, though efforts were made to equalize feedings ; (3) temporary mental state, varying for the individual during a series of tests, as sulkiness and inertness caused by an unfavorable physiological state or by undue sensitiveness to punishment. For the above reasons the time records are deemed of little value and are not given in the published tables though they were taken in all the experiments. SUMMARY AND CONCLUSIONS Turtles of the species Chrysemys marginata were tested for their ability to discriminate between, (a) black and white; (b)' two patterns of different shape; (c) two series of parallel lines of equal width but running in different directions; (d) two series of parallel lines of different width running in the same direction. Associations were established by the use of food and electric boxes except in two experiments where the electric shock was not used as punishment for failure. The apparatus employed was similar in construction to the electric-box apparatus of Yerkes, though considerably modified for adaptation to the habits of turtles. Negative results only were obtained in the experiments for pattern discrimination. It is felt that these results are not DISCRIMINATIVE ABILITY OF THE TURTLE 27 conclusive, since the patterns used are somewhat alike in gen- eral symmetry. In the black-white tests discrimination was established with four turtles, while one other turtle failed to show improvement in 220 trials. Two turtles learned to discriminate between two series of parallel lines 8 mm. wide, one vertical and the other horizontal in direction, and showed a fair degree of discrimination when these lines were reduced to 4 mm. in width. One of these tur- tles did well with a further reduction of the lines to a width of 2 mm. on each box. Two turtles learned to discriminate between two series of parallel horizontal lines 8 mm. and 2 mm. in width respectively.' One turtle learned first to discriminate between two series of parallel vertical lines 8 mm. and i mm. wide, next between lines 4 mm. and i mm. wide, then between lines 4 mm. and 2 mm. wide, and finally showed an excellent average of discrim- ination between lines 3 mm. and 2 mm. in width. Chrysemys does not learn rapidly. An average taken of all the experiments shows that about 183 trials were necessary to establish discrimination. If one experiment is omitted in which 455 trials were given, the average is reduced to 154 trials. In these calculations the three final perfect tests are not counted, neither are tests with reduced lines considered, only those line experiments being taken in which discrimination was first being established. Two of the memory tests are surprisingly good but are subject to question, no preference tests having been taken. The third, showing perfect memory two weeks after the vertical-horizontal line tests is unquestionably accurate. These results indicate a fair degree of retentiveness. The amount of evidence at hand does not justify general conclusions regarding the relative intelligence of turtles of dif- ferent age or different sex. The turtles studied exhibited marked individual differences in disposition and in mental ability. More particularly from the results of the line experiments it may be concluded that Chrysemys is able to appreciate differ- ences in the direction of line markings even though the lines be narrow, and differences in the width of lines even though these 28 D. B. CASTEEL be slight in amount. These results suggest that the discrimina- tive faculty thus shown may have been an important factor in the development of the mind of the growing turtle and that in the common activities of its daily life the individual's behavior may often be determined by its ability to appreciate differences of direction, though but lightly marked, and to distinguish accurately between slight differences of extent. THE REACTIONS OF MOSQUITOES TO LIGHT IN DIF- FERENT PERIODS OF THEIR LIFE HISTORY S. J. HOLMES From the Zoological Laboratory, University of Wisconsin Everyone is familiar with the quick wriggling downward of mosquito larvae upon one's approach. Pass your hand over a jar in which the larvae are hanging from the surface film and they will quickly dart towards the bottom. The movement is not an effort to get away from the source of the shadow. It is not simply a tropism. It is a specific reaction to shadows, or a sudden diminution of light intensity, by swimming downward. Nevertheless there seems to be an element of negative photo- taxis in the behavior of the larvae when stimulated by shadows, although otherwise they may show a positive reaction to light. These conclusions are borne out, I think, by the following experiments : A glass jar containing numerous larvae of Culex territans and C. pipiens was placed before a window through which the light fell upon the jar obliquely from above. A dark object was passed over the jar; the larvae quickly swam downward and toward the side of the jar away from the light. Numerous repetitions of the experiment gave the same result. When the larvae were brought into a dark room and illuminated from one side a shadow thrown upon them would cause them to go down- ward and away from the light as before. The direction of the light rays is therefore a factor in causing the direction of the movements of the larvae. I have often noticed that specimens in a jar before a window swam dow^nward and away from the light upon my approach, regardless of the direction in which I came. This led me to ascertain whether the larvae would swim away from an approaching object or would simply swim down- ward and away from the light when stimulated by a shadow or large object in their visual field. The latter was found to be true. In jars before a window the wrigglers would repeatedly swim directly towards my hand if it was moved toward them on the side away from the light. Specimens taken into the dark room and illuminated from one side by an incandescent electric lamp would react in the same way. A jar of larvae was then 29 30 S. J. HOLMES placed on a glass plate on a tripod and illuminated from below. When my hand was passed beneath the jar, thus throwing a shadow on the larvae from below, they would swim downward as before. Momentarily turning off the light produced the same effect. An object placed on a fine wire and brought near the wriggler from below when it was hanging from the surface film would invariably cause a sudden descent towards the ap- proaching object. When larvae are on the bottom, shadows thrown upon them either from above or below simply set them into commotion, which does not issue in any very definitely directed movements. Notwithstanding this peculiar reaction to shadows and the direction of the light rays the larvae of Culex often show a markedly positive photo taxis. Frequently I have seen them gathered on the side of their jar toward the window and when brought into a dark room and exposed to an incandescent, sixteen-candle power electric lamp they swim over toward the more illuminated side of the dish. Larvae vary greatly in their phototactic response. Many seem entirely indifferent to the light. Certain individuals swim toward it eagerly and will follow it about when it is passed from one side of the dish to the other, with great promptness. One lot of larvae three days old was strongly positive, the whole troop swimming over to what- ever side of the dish the light was held. The same larvae, how- ever, when suddenly struck by a shadow will go away from the light rays with equal readiness. Mosquito larvae are very sensi- tive to jars. Tapping lightly on their vessel sends them quickly downward. Placed before a window or illuminated from one side by an artificial light they swim downw^ard and away from the light whenever they are jarred just as they do when dis- turbed by a shadow. There is no orientation in the light response. This is in fact practically impossible owing to the peculiar method of swimming which consists in bending the body rapidly from one side to the other. The larvae swim sidewise or obliquely to the direction of the rays when they go toward the light, as they do in swim- ming away from it. When they have arrived at the positive side of the dish the position of the body bears no relation to the direction of the rays of light. The lack of orientation in the larvae may not be of significance in regard to the nature of the REACTIONS OF MOSQUITOES TO LIGHT 31 phototactic response, as any orienting tendency which may be present and which may perhaps account for the direction of locomotion would naturally be obscured by the contortions of the larvae during their progress through the water. Age makes little difference in the reactions to light. Experiments tried with larvae less than a day after hatching, elicited the same responses as in older larvae although they were a little less decided. The behavior of the pupae, up to even a short time before the emer- gence of the imago, is practically the same as that of the larvae both as regards reactions to shadows and positive phototaxis. The reactions of the larvae and pupae of Culiseta inornatus, Aedes fuscus and Aedes curriei are essentially the same as in the species of Culex above described. Larvae exposed frequently to shadows gradually fail to respond to them. A dish containing about thirty three-day old larvae was shaded by passing an object over it once every minute, the object being passed as nearly as possible in the same way and with the same degree of speed. At the first trial all (about twenty-five) which were at the surface went down. At the second shading about fifteen went down, leaving nine at the surface. In subsequent trials the number remaining at the surface gradually decreased. From the tw^elfth to the sixteenth trials none descended; in the seventeenth, six went down, after which numerous subsequent trials produced no response. In another lot all failed to respond after the seventh trial. Other experiments yielded very similar results. One lot of larvae was rather exceptional, however, in that they rapidly diminished in responsiveness in the first three shadings, but up to the fifty- fifth trial there was no further reduction in the number descending. The adult mosquitoes, like the larvae and pupae, show a peculiar combination of reactions to light. It is well known that they are much more apt to settle upon dark objects than upon light ones, and that people wearing dark clothes are more apt to attract them than those with light-colored apparel. Dur- ing the day they seek the shade, even when it is not hot or dry, and darkness brings them out of their haunts. Within doors, while occasionally found on the windows and the light areas of the room, they very frequently retreat into the shaded nooks. But like most moths and many other forms which seek the shade, they frequently show a decided positive phototaxis. 32 S. J. HOLMES They sometimes fly to lights at night, but some species manifest little tendency to go towards the light under any conditions. CuUseta inornaius, Aedes fuscus and two species of Culex, C. pipiens and C. territans, with which I have experimented, show a marked positive photo taxis. They keep on the sides of their cages nearest the window and in the dark room they follow the light in any direction. When collecting them in vials it is only necessary to hold the closed end towards the sun to be able to put in the cork without the insects escaping. This positive phototaxis is shown as strongly the first day after emergence as after specimens have been kept some weeks in confinement. After a full meal, although more sluggish, they are still positive. The males show the same positive reactions as the females. A STUDY OF TRIAL AND ERROR REACTIONS IN MAMMALS By G. V. HAMILTON, M. D. Montecito, California THREE FIGURES I. Introduction; Description of Apparatus; Description of Method 33 II. Description of Subjects 39 III. Tabulation of Preliminary Analysis of Results 44 IV. Determination of the Different Modes of Adjustment Manifested 48 V. Interpretation of Results 56 VI. Summary and Conclusions 63 I. INTRODUCTION; DESCRIPTION OF APPARATUS; DESCRIPTION OF METHOD The literature of animal behavior contains considerable experi- mentally obtained evidence that among the mammalia there are marked differences of ability to profit by experience, but these differences have been analysed almost exclusively in terms of sensory equipment and of quantitative measurements of reac- tion time and reactive errors. The trial and error mode of adjustment has thus come to serve the student of behavior as a conceptional unit in his analyses of mammalian reactions, and has been dealt with as genetically variable only with refer- ence to the degree of rapidity with which it leads to the formation of associations appropriate to given situations, and to the elimi- nation of useless activities. We speak of qualitatively different instinctive adjustments, but not of qualitatively dift'erent " try- try -again " (trial and error) efforts to meet a situation for which there is no specifically appropriate instinct, no opportunity for imitation of any kind, and no rational equipment. The present investigation seeks to collect facts of behavior which may lend themselves to qualitative interpretations of trial and error activities. In other words, it is concerned with the following problem: What, if any, are the qualitative differ- ences of reactive tendency that account for the fact that some mammals learn slowly, and with many errors, to meet situations which their fellows of superior age or race learn to meet quickly and with but few errors? We cannot know how to attack this problem, nor specifically what to look for, until we gain a general orientation concerning 33 34 G. V. HAMILTON the facts relevant to it. With this in mind I have made use of the apparatus and method described below. Description of apparatus. The reader will more clearly under- stand the purpose of the apparatus by picturing to himself a room which may be entered by a door capable of giving entrance only, and which may be gotten out of by means of a constantly varying one of four possible doors of exit. Let us then imagine EH ENtD Figure 1 — Floor plan of apparatus. EntD, Entrance Door; O, point of equidis- tance from exit doors; ExD 1, ExD 2, ExD 3, ExD 4, Exit Door number 1, Exit Door number 2, etc. that from within the room one may see on each of the five doors a distinctive inscription, as follows: Entrance door : cannot he used for exit. Exit door No. i : push against it; will afford exit unless it is locked. Exit door No. 2 : (same inscription). : (same inscription) . .: (same inscription) . The apparatus differs from this imaginary room in that its various doors are not labelled at all, and are all alike in appear- ance, so that a subject seeking exit from the apparatus must Exit door No. 3 : Exit door No. 4: TRIAL AND ERROR REACTIONS IN MAMMALS 35 gain by experience the information that the above inscriptions would have afforded. The floor of the apparatus is of wood, and the top of wire netting, and both are of the form shown in Fig. i . The narrow, rectangular part of this figure represents the entrance hall (EH, Fig. i). This hall can be entered by way of an entrance door (Ent. D, Fig. i), which will swing in the inward direction only, and which fits so snugly within its frame that, once it is closed, it cannot be opened from within the apparatus. It is hung a little out of plumb, so that it will always swing to the closed position when not actually held open by the subject. Figure 2 — An exit door, set within its frame, and equipped with button and button strings. B, Button; S^ String number 1; S^, String number 2. At the opposite, broader end of the apparatus are four exit doors (Ex D-i, Ex D-2, Ex D-3, Ex D-4, Fig. i). Each of these exit doors opens outward only, and when released from pressure swings to the closed position. Like the entrance door, the exit doors are hung out of plumb. The exit doors just described are equidistant from point O in Fig. I. Figure 2 is intended to show the manner in which the various exit doors can be " locked " or " unlocked " without the sub- 3b G. V- HAMILTON ject's knowledge. It represents the outer aspect of an exit door, set within its frame, and equipped with a button (B, Fig. 2) and two strings (S-i, S-2, Fig. 2). When the experimenter pulls the string attached to the inner end of the button (S-i, Fig. 2) the latter is brought to the vertical position and the door is thus unlocked. When the string attached to the outer end of the button (S-2, Fig. 2) is pulled, the button assumes the horizontal position, thus locking the door. The subject has nothing to do with the manipulation of these strings, which are carried in metal eyes to the under surface of the apparatus, whence they are carried in grooves, thoroughly concealed, to any part of the laboratory from which the experimenter can operate them without the subject's knowledge. Each of the four exit doors is thus equipped with a button and two strings. The sides of the apparatus are of wire netting. The inner surfaces of the exit doors are painted greyish-white, whilst the remainder of the interior of the apparatus is dark green. The exit doors have no individually distinctive marks. The dimensions of the apparatus are determined according to the following general rule: the subject's length being " i," the uniform inside height of the apparatus is "3;" the height of the entrance and exit doors is " 2.9 " and their width " 2." From point O (Fig. i ) to the mid-point of each exit door's lower margin is " 6." The various doors are sand-papered to prevent binding. In order to meet special conditions I have made various un- important modifications of the apparatus just described. The human subjects, by reason of their upright position in walking, required a relatively higher apparatus. For the horse an appa- ratus was provided which had " exit doors " merely wide enough to enable him to thrust his head and neck into a food box just beyond; after a trial he was led out of the entrance end of the apparatus. The monkeys so resented handling that their appa- ratus had to be built within a larger enclosure which, in turn,' adjoined their living rooms. In every case the following essen- tial condition was met: the subject, once within the apparatus, was unable to discover whether a given exit door was loclced or unlocked except by actually pushing against it. Description of method, (i) Preliminary training. It was found to be desirable to give the animal subjects thorough familiarity with the apparatus before their formal trials were begun, hence. TRIAL AND ERROR REACTIONS IN MAMMALS 37 a subject's first experience with the apparatus was merely that of entrance into it, with all the doors propped open and with food scattered about on the floor. After he had eaten he was coaxed out of one of the entrance doors. This was repeated until the subject was thoroughly familiar with the interior of the apparatus as a feeding ground, and with the open exit doors as the most convenient avenues of approach to a desirable locality. In the case of the horse this differed in that the feeding ground was, from the start, a place just beyond any of the exit doors. The next step in training the subject for the experiment was to close the entrance door as soon as he had entered the appa- ratus. As soon as this had ceased to produce any evidence of uneasiness the four exit doors were left partially closed, so that the subject was compelled to exert some pressure against any of the exit doors through w^hich he sought to escape. As the training advanced the exit doors were left more and more nearly closed until they were quite closed, although still un- locked. The subject was considered " trained " as soon as he had learned to seek the exit doors for escape from the apparatus, and to push against them without hesitation. (2) Formal trials. The description of these trials, which constitute the experiment proper, will be facilitated by speak- ing of the various exit doors as if they were numbered " i," "2," " 3 " and " 4," in order, from left to right. For the first formal trial of any subject three of the exit doors were locked by the experimenter, who pulled the appropriate button strings without the subject's knowledge. The remaining exit door, and only that one door was left unlocked. From within the apparatus, therefore, all four exit doors looked alike, i. e., closed. With the apparatus thus prepared, the subject was placed within it for his first trial, and was allowed to choose his own time for effecting his escape. As soon as he found the one unlocked exit door and effected his escape he was rewarded with food. For the second formal trial the one unlocked door of the first trial was locked, and another of the exit doors was unlocked. Thus, if door 4 were the unlocked door for the first trial, it would be one of the three locked doors for the second trial. Each suc- ceeding trial in a total series of 100 trials differed from the trial 38 G. V. HAMILTON immediately preceding it according to the same principle: the unlocked exit door of the immediately preceding trial was always one of the three locked exit doors during the present trial. A second condition of the formal trials was this: during each subject's loo trials, each exit door was left unlocked for twenty- five trials and locked for seventy-five trials, and care was taken by the experimenter to avoid a discoverable sequence in select- ing an ever- varying exit door to be left unlocked. Even the most sophisticated human subject was thus left in ignorance as to which of three inferentially possible doors of exit would open when pressed against until he had actually tried one or more of these doors. Ten consecutive trials were given each subject daily for ten successive days. The older human subjects were given no preliminary training, but were frankly told that no tricks would be played, and that they were expected to leave the apparatus by way of an exit door. The human infant was given essentially the same train- ing that the animals received, except that in his case toys were substituted for food when my commendation proved insufficient as a motive for reaction. Before we enter into a discussion of results, two obvious defects of the above method must be taken into account: (i) The situations may have had different sensory values for the different subjects. For example, the dogs of my experiment were unmistakably guided by local odor signs in their discrimina- tion of one exit door from another, whilst these signs were of no value to the human subjects, and probably of but little value to the cats and monkeys. On the other hand, the monkeys may have detected fine differences of visual appearance of the inner surfaces of the exit doors, whereas these doors certainly presented no such differences for the human subjects, and may have been visually indistinguishable to the cats, dogs, and horse. A fairly satisfactory solution of this difficulty was effected by the preliminary training, which taught the subjects to dis- criminate among the various exit doors by their differences of spatial position. (2) There was no adequate measurement of the reactive value of the motives supplied to the various subjects for escaping from the apparatus with the greatest possible speed and accuracy. Yerkes' and Dodson's (i) discoveries concerning the relation TRIAL AND ERROR REACTIONS IN MAMMALS 39 of strength of stimulus to rapidity of habit formation impose on the student of behavior a scientific obligation to regulate as accurately as possible the strength of motive-stimuli. Un- fortunately, the easily regulated painful stimuli (electric shock) that proved so useful in their work with the Dancing Mouse cannot be applied where one seeks comparable results from subjects who differ so widely in emotional responsiveness to frequently recurring discomfort as do, for example, dogs and monkeys. Even in the case of my naive human subjects, the possibility that the size of the reward or the degree of my appro- bation might be affected by stupidity or cleverness of reaction had to be ruled out in order to obtain uniform results. By observing, for each individual, the relation between strength and kind of motive on the one hand, and uniformity of reaction on the other hand, I was able to secure results which I believe to be quite safely comparable for the purposes of the present exploratory investigation. The animal subjects were never used for experiment until their hunger was partially appeased, and if they chose to lie down and sleep while within the apparatus they were merely urged to return to me for a bit of extra-tempt- ing food. With cats and dogs an atmosphere of lazy comrad- ship with their master greatly favored a steady, apparently unemotional quest of the unlocked door. The horse's natural habit of seeking food over prolonged periods of time rendered him an excellent subject. The monkeys would doubtless have given more uniform results had there been any way of over- coming their natural distractibility, but the food seeking aspect of the experiment proved to be most helpful in reducing the effects of their tendency to shift their attention from the appa- ratus to fortuitous sights and sounds. II. DESCRIPTION OF SUBJECTS The selection of subjects for experiment was made with refer- ence to the desirability of covering considerable ontogenetic and phylogenetic ranges without thereby impairing the value of the results for comparison. Since it was more relevant to the pur- pose of the investigation to explore for different kinds of adjust- ment than to make an intensive study of any particular mode of adjustment, no effort was made to obtain averages for large numbers of individuals belonging to a given age or species. 40 G. V. HAMILTON The many gaps in the age and phyletic series are due to the fact that the results obtained from many of my subjects had to be rejected for various reasons. Some of the subjects were stolen and some died before their trials were completed; in some cases I could not be reasonably sure that a possible diffi- culty in discriminating one exit door from another did not exist ; and many unavoidable interruptions rendered it impossible to make use of all the subjects that were available for experiment. It is much to be regretted that the list given below includes no human subjects between the ages of twenty-six months and eight years; the writer's wholly undeserved local reputation as a vivisectionist seemed to create a stubborn unwillingness on the part of parents to supply young children for experimental work. Human Subjects Man I. Age, 34 years. Native (Spanish-Indian) Californian. Ranch laborer in the experimenter's employ. A man of limited education, but of average intelligence for his class. He went through his trials in the stolid, unemotional manner that char- acterised his work in the fields. The " boss " wanted him to walk into and out of an enclosure 100 times, and he did so with- out asking questions or shirking his task. Boy 7. Age, 15 years. American, of original English descent. Grocer's boy. Countr}^ school education. He was shy and nervous at the beginning of the experiment, and always seemed to be more or less affected by a fear of appearing stupid. Boy 6F1. Age, 13 years. Father Italian, mother Swedish. Country schoolboy. He was less alert, mentally, than were his brothers, who are described below. Boy 5F1. Age, 11 years. Brother of Boy 6F1. Cotintry schoolboy. Volatile, alert, and rather distractible. Boy 4F1. Age, 10 years. Brother of Boys 6F1 and 5F1. Country schoolboy. Relatively precocious, and an excellent subject. Boy 3. Age, 10 years. Native (Spanish-Indian) Californian. Country schoolboy. Bright, but rather shy, and too nervously eager to please the experimenter. Girl I. Age, 10 years. American of Scotch descent. Until recently a student in the public schools of Cambridge, Mass. Her superiority of school training, and her very considerable TRIAL AND ERROR REACTIONS IN MAMMALS 41 degree of mental precocity, gave her a decided advantage over all the other human subjects, including, even, the adult Man i. Boy 7. Age, 26 months. American of mixed descent (son of the experimenter). At the time of the experiment he could walk; had a limited vocabulary of words, which he did not put into sentences; was able to find his way about the house; and understood many simple commands. Very quick to form new associations. Defective Man A. Age, 45 years. Native (Spanish-Indian) Californian. Ranch laborer in the experimenter's employ. Limited school education, but had read history and uncritical w^orks on socialism. He was a nervous, suspicious, " muddled " person, with a grievance against society in general, and a sur- prising fund of self-acquired misinterpretations relating to his social environment. He expressed a belief that my experiment was dangerous meddling with the human mind, and that it had some occult power of " making people crazy." His curiosity and his desire to argue matters rendered him available, but he seemed to be in constant dread of the apparatus, and always labored under a suspicion that it was not the simple structure that it pretended to be. Defective Boy A. Age, 11 years. English. With the excep- tion of occasional perfunctory lessons from a governess, his education was practically nil. He w^as barely able to read simple words, was unable to respect the conventions of conversa- tion usually recognized by a child of six years, and manifested an inordinate fondness for asking questions. His cooperation varied: at times he gave attention to the experiment, and did well enough; and at times he behaved in a dull, mechanical manner. His reactions are of considerable interest. * Monkeys Monkey 6.* Age, 15 years (estimated). Macacus cynomolgus. About 10 years in captivity. Tame, and an excellent subject. Monkey 4. Age, 1.5 years (estimated). Macacus rhesus. About 6 months in captivity. Very tame. After his 20th trial his vision became defective, so that he had difficulty in finding his way about the apparatus. His results do not appear in any of the averages. * The writer has followed Professor Yerkes' (2) method of designating male sub- jects by even numbers, and female subjects by odd numbers. 42 G. V. HAMILTON Monkey 3. Age, 5 years (estimated). Macacus rhesus. About 3 years in captivity. A truculent, untamable animal, but a fairly good subject. Monkey 2. Age, 1.5 years (estimated). Macacus (species undetermined). About i year in captivity. Timid, but exceptionally resourceful in meeting outdoor situations when he was given the freedom of the ranch. His timidity doubtless affected his behavior. Monkey i. Age, 1.5 years (estimated). Macacus rhesus. About I year in captivity. A comparatively stupid animal, but uniform in behavior during the experiment. Dogs The sixteen dogs of the following list range in age from thirty- six days to three years. With the exception of the two adults (Dogs I and 2), they are all descended from a common sire, the subject of my previously published "An Experimental Study of an Unusual Type of Reaction in a Dog." (3) This sire was a Boston Terrier of impure breed. The six puppies that are designated " F2 " in the list below were from a mongrel bitch of the small hound type. Their mother was unavailable for experiment. The eight puppies of the " Fi " group were from an English Setter bitch, — " Dog i " of the list. Dog. I. Age, 3 years. English Setter. Mother of dogs 3, 7, 8, 9, II, 12, 14 and 18 (one litter). She was untrained, hence a study of her behavior under natural conditions was easily possible. It was found that whenever she sighted or scented her prey she would inhibit every visible movement of her body but a slight tremor and a wagging of her tail, and would stand thus, in a rigid attitude, for several minutes before making a final dash to seize the object of her attention. A tendency to inhibit, momentarily, all active movements when prey is first discovered is not limited to any special breed of dogs: it is the prolongation of this momentary inhibition that so marks the behavior of the English Setter. This reactive tendency of the mother of my puppies is of much interest for the present investigation. Dog 2. Age, I year (estimated). Great Dane -mongrel. His tendency to look to his master for cues to action marked him as characteristicallv different from all the other dogs of my experi- TRIAL AND ERROR REACTIONS IN MAMMALS 43 ment. The other dogs sought cues to action in the situation, and were not so much affectec . by my presence. Dog 4F2. Age, 116 days. Boston terrier-mongrel. Dog 6F2. Age, 109 days. Boston terrier-mongrel. Dog 3F1. Age, 102 days. Boston terrier-English setter. Dog 8F1. Age, 95 days. Boston terrier-English setter. Dog 5F2. Age, 88 days. Boston terrier-mongrel. Dog 7F1. Age, 81 days. Boston terrier- English setter. Dog 10F2. Age, 74 days. Boston terrier-mongrel. Dog 9F1. Age, 67 days. Boston terrier-English setter. Dog iiFi. Age, 60 days. Boston terrier-English setter. Dog 12F1. Age, 53 days. Boston terrier-English setter. Dog 14F1. Age, 46 days. Boston terrier-English setter. Dog 13F2. Age, 43 days. Boston terrier-mongrel. Dog 16F2. Age, 43 days. Boston terrier-mongrel. Dog 18F1. Age, 36 days. Boston terrier-English setter. Cats Cat I. Age, I year. Manx. This animal was reared in the Harvard Psychological laboratory, and enjoyed the further distinction of having been one of Doctor Berry's (4) subjects in his studies of imitation. Although my apparatus has no technical resemblance to that used by Doctor Berry, Cat i explored it carefully, and for several trials persisted in her efforts to claw at imaginary loops of string beyond the meshes of the wire netting. Her close attention to the apparatus situa- tion, and her remarkable intelligence (as compared with that of common cats) was of much interest. Cat 2. Age, I year. Common house cat. Co-operated well. Cat 3. Age, I year. Same as Cat 2 except in sex. Cat 5. Age,' 56 days. Common cat. Co-operated well. Cat 7. Age, 70 days. Common cat. Timid and sluggish. Horse Horse 2. Age, 8 years. Gelding of western breed. Carriage horse. In view of the very poor showing made by this animal, the stableman's belief in his " smartness " is of some interest. On one occasion the writer was driving this horse at night, over an unfamiliar network of roads. The horse was guided in a wrong direction on the way home, and the writer became quite disoriented. When the horse was given a loose rein he 44 G. V. HAMILTON traced the way back without error, although he had been over the various roads involved less frequently than had his master. III. TABULATION AND PRELIMINARY ANALYSIS OF RESULTS It will be remembered that each of the subjects was given loo trials, and that during each of these trials one, but only one, of the four exit doors was capable of being opened. It will be further remembered that this " unlocked " door varied from trial to trial. Thus, if Door 3 were unlocked for any present trial it would surely be locked for the next trial. It is apparent, therefore, that any subject who avoided, during each present trial, the unlocked door of the immediately preceding trial, was apt to effect his escape during his 100 trials by trying the various exit doors 200 or 201 times: 200 times if he did not try all four doors during his first trial, otherwise, 201 times. Again, any subject who tried the various doors without refer- ence to the ever yarying one impossible door as such, or as a preferable door to try on account of its having just afforded the escape, would be apt to effect his escape from the apparatus necessary 100 times by trying the various exit doors 250 times. The " average " number of efforts to open exit doors would be affected by any of the following factors: (i) This number would be decreased by a tendency to try first, on entering the apparatus for a trial, the exit door that had been most remotely (in time, with reference to the present trial) an unlocked door. This, in spite of the fact that the ex- perimenter followed an irregular order in selecting exit doors for unlocking. (2) It would be increased by a tendency to prefer, as first choice of door to be tried during a given trial, the unlocked door of the immediately preceding trial. (3) It would be increased by a tendency to make more than one separate effort to open the same exit door during a given trial. In table i, I have tabulated the total number of separate efforts to open exit doors manifested by each of the various subjects during his 100 trials. In making current observations of my subjects' behavior I recorded as a separate effort to open an exit door the total activity of a subject from the time he attacked an exit door until he left its immediate vicinity. For example, during his second trial Dog 18F1 went to Door 4 and TRIAL AND ERROR REACTIONS IN MAMMALS 45 clawed it for several minutes, then turned away from this door and wandered about the apparatus ; one effort to open an exit door was recorded. Then he returned to Door 4 and again spent some time in alternately clawing and barking at it; when he left the door a second time a second effort was recorded. He returned to it, tried it and left it a third and a fourth time before attacking another door, hence his first four separate activities, all of which were definitely directed against the same unyield- ing door, were recorded as four separate efforts to open exit doors. With the above in mind the reader will be able to obtain a general orientation concerning the value of the situations for each of the various subjects. TABLE 1 Subject. Age. Man 1 34 years Boy 7 15 years Boy 6F1 13 years Boy 5F1 12 years Boy 4F1 10 years Boy 3 10 years Girl 1 10 years Boy 2F1 8 years Boy 1 26 months Man A 45 years Boy A 11 years Monkey 6 15 years Monkey 4 1-5 years Monkey 3 5 years Monkey 2 15 years Monkey 1 1-5 years Dog 1 3 years Dog 2 1 year Dog 4F2 . 116 days Dog 6F2 109 days Dog 3F1 102 days Dog 8F1 : 95 days Dog 5F2 88 days Dog 7F1 81 days Dog 10F2 74 days Dog 9F1 67 days Dog llFl 60 days Dog 12F1 53 days Dog 14F1 46 days Dog 13F2 43 days Dog 16F2 43 days Dog 18F1 36 days Cat 1 1 year . . Cat 2 1 year Cat 3 1 year Cat 7 70 days Cat 5 56 days Horse 1 8 years No. of efforts to open exit doors. 200 193 201 216 194 211 183 216 315 217 237 275 291 278 272 278 324 302 346 329 284 378 304 357 427 389 358 307 413 314 376 438 358 378 320 406 368 461 46 G. V. HAMILTON Discussion of Table i . In the light of ■ analyses that are to follow, this table has a largely negative value. It shows how misleading a single objective measure of ability to profit by experience may be. For example, the three-year old mother of the Fi puppies (Dog i) made 324 separate efforts to open exit doors during 100 trials, whilst her fifty-three-days old puppy (Dog 12F1) has a record of only 307 efforts to open doors. Her record is also greater than that of three of the other puppies (Dogs 3F1, 5F2, and 13F2). Table i also shows that immature Monkey 2 has a lower record than either of the two adult monkeys; and that the record of mature Cat 2 exceeds that of iifty-six-days old Cat 5 by ten efforts to open doors. In the case of the monkeys, the apparent inconsistency of onto- genetic findings becomes all the more striking when we take into account the fact that at 1.5 years of age the Macacque is scarcely half -grown, and sexually immature. When we enter upon a discussion of the different modes of searching for unlocked doors, and attempt to isolate the specific reactive tendencies to which these may be attributed, it will be seen that a genetically superior reactive tendency may lead, in some of its manifestations, to an actual increase in number of efforts to open doors. For the present, however, it is desirable to subject the data contained in table i to further analysis in order to discover whether or not there is a general tendency for increasing age and phyletic position to decrease the number of efforts to open doors. To this end the various subjects will be divided into age and phyletic groups, and the averages for each of these groups will be compared. Since the results obtained from the normal human subjects whose ages range from eight to thirty-four years do not present individual variations from their general average which can be clearly traced to age differences, these eight subjects will be included in a single group. The two defective human subjects — Man A and Boy A — cannot properly be classed together, hence their individual results will appear separately in the table of averages (table 2, below). The human infant (Boy i), the monkey whose failing vision affected his behavior (Monkey 4), and the horse should, for obvious reasons, appear separately in a table of averages. The mature dogs and mature cats each form an age group, as do also the kittens. In the case of the TRIAL AND ERROR REACTIONS IN MAMMALS 47 puppies, their considerable number enables us to form two age groups: the thirty-six-to-seventy-four days-old puppies and the eighty -one-to-one hundred and sixteen-days-old puppies. TABLE 2 No. of subjects in Groups of subjects each group 8-to-34-years-old normal human 8 Defective man A 1 Defective boy A. . . 1 Infant Boy 1 1 Mature monkeys 2 Immature monkeys 2 Defective Monkey 1 Mature dogs 2 81-to-116-days-old puppies 6 36-to-74-days-old puppies 8 Mature cats 3 Kittens 2 Horse 1 Average age for entire group 14 years 45 years 11 years 26 months 10 years 1 . 5 years 1 . 5 years 2 years 98.50 days 52.75 days 1 year 63 days 8 years Average number of efforts to open exit doors 201 . 75 217.00 237.00 315.00 276.50 275.00 291 . 00 313.00 333.00 377.75 352.00 387.00 461.00 Ontogenetic aspects of table 2. The normal humans, the dogs and the cats manifest a tendency toward increase in number of efforts to open doors with decrease in age. This is especially prominent in the case of the human infant, whose record exceeds that of the average for the older normal human subjects by 56.13 per cent. Defective Man A, whose record appreciably exceeds that of the eight-to-thirty-four-years-old normal human subjects, would doubtless have presented a much higher record had he not manifested a marked tendency to make first choice of the most remotely unlocked door during each present trial. It may be said, indeed, that his total modes of adjustment to the situations of the experiment were, on the average, of a lower type than those manifested by defective Boy A, whose record places him below Man A in the table 2 list. This additional evidence of the unreliability of the findings of tables i and 2 as measures of intelligence will be brought out more explicitly in sections IV and V. A general knowledge of the marked differences that obtain between the behavior of an adult monkey and that of a half- grown one would lead the student of behavior to expect that under similar conditions no year-and-a-half old macacque would find the unlocked door one hundred times with fewer efforts 48 G. V. HAMILTON than an adult macacque. The but slightly lower record of the two half -grown monkeys is, therefore, of some interest. Here, too, an explanation is to be found in differences of reactive tendency which require, for genetic evaluation, more than a single standard of measurement. Phylogenetic aspects of table 2. A comparison of the various adult groups shows that in order of ability to avoid useless efforts to open exit doors the older normal human subject stands first, the mature monkeys second, the dogs third, the cats fourth, and the horse fifth. Table i shows that the horse has the highest individual record of efforts to open doors that is manifested by any subject, regardless of age. It is of some interest that twenty-six-months-old-Boy i made a greater number of efforts to open exit doors than did any of the following subjects: All other human subjects, all of the monkeys, and dogs 2, 3F1, 5F2, 12F1, 13F2. If mere ability to avoid useless activities w^ere a measure of intelligence, this finding would imply that my son was less intelligent at twenty-six months than was one of my forty-three-days-old puppies (Dog 13F2). In view of the fact that at the time of the experiment Boy i gave unmistakable evidence of having " free ideas" (5), this is an absurd implication. IV, DETERMINATION OF THE DIFFERENT MODES OF ADJUSTMENT MANIFESTED It will be remembered that the conditions of the experiment require the subject who seeks escape from the apparatus merely to search for an unlocked door until he finds one. The present chapter seeks to classify the different modes of searching for this ever-varying unlocked door. In order to facilitate the descriptions that follow, the value of each of the four exit doors in a given trial-situation will be designated by one of the following terms : The impossible door. For any present trial this is the un- locked door of the immediately preceding trial; hence it is an inferentially impossible door of exit during the present trial for any subject who is able to appreciate that no one door is ever an unlocked door in two successive trials. The possible doors. No subject is able, during a given trial, to tell with certainty which one of three exit doors will yield TRIAL AND ERROR REACTIONS IN MAMMALS 49 to pressure until he tries it. Of these three inferentially possible doors of exit, two are actually locked, and one is unlocked. We shall speak, therefore, of " locked possible doors " and an unlocked possible door. ' ' The intention of our investigation requires a division of all the reactions of the subjects into two groups, only one of which can enter into the tables. These groups are: (i) Unclassified reactions. This group includes all reactions which led to the discovery of the unlocked door before all three possible doors were tried, and which did not include more than a single effort to open any given door during the trial. I have rejected from the tables as " unclassified " all reactions which met the above conditions, even though such a reaction in- cluded an effort to open the impossible door. This is justified, I believe, by the fact that none of the animal subjects seemed to have a consistent awareness of the impossible door as such. The description of classified reactions will disclose an additional reason for this exclusion. (2) Classified reactions. To fall within this group a reaction must meet one of the following* conditions : (a) Efforts to open each of the three possible doors; (b) more than one separately continuous effort to open a given door during the trial. For example, if during a present trial, door 3 were the unlocked door of the immediately preceding trial and door 2 the unlocked door of the present trial, a reaction which could be tabulated according to any of the following formulae would fall within the classified group: Efforts to open exit doors 4, i, 2 in the order given; or exit doors i, 4, i, 2 ; or exit doors 4, 4, i, 2, etc. The classified modes of searching for the unlocked door were found to belong to five objectively different general types, which may be described as follows: Type A. All three possible doors tried, once each; no effort made to open the impossible door. This is the most adequate possible type of classified reactions. Type B. All four exit doors are tried, once each, and in an irregular order. Type C. This reaction can occur only when the door to the extreme right (Door 4), or the one to the extreme left (Door i) is the unlocked door. It involves trying each of the four exit doors once, and in order from left to right or from right to left, 50 G. V. HAMILTON according as Door 4 or Door i is the unlocked door. Thus, if Door I be unlocked, the subject tries the doors in the following order: 4, 3, 2, i. Type D. More than a single continuous effort to open a given door during the trial; but between separate efforts to open the same door there must be an effort to open some other door. Thus, the subject tries doors in the following order: 4, I, 4> 3; or 4, I, 2, 4, 3, etc. Type E. This type includes various highly inappropriate modes of seeking escape from the apparatus which might be classed as separate types of reaction were it not that when collectively treated as a unit in the distribution curve they are seen to belong to a single general type. The various forms of Type E reactions are, — (a) during a given trial the subject tries a door, leaves it, then returns to it and tries it a second time without having tried any other door; (b) during a given trial the subject attacks a group of two or three locked doors tw^o or more times in a regular order; (c) during a given trial the sub- ject, without falling into either of the above two errors of reac- tion, persistently avoids an exit door, so that he makes at least seven separate efforts to open exit doors before effecting his escape. It is especially important to gain a clear understanding of the objective characteristics of the classified types of reaction described above, since they will appear as units in all of the analyses that are to follow. Table 3, given below, contains characteristic examples. TABLE 3 Examples of Classified Reactions Type E Ttjpe E 4, 3, 4, 1 3, 3, 1 (sub-type a) 2, 3, 4, 3, 2, 1 4, 2, 2, 2, 2, 1 (sub-type a) 2, 4, 3, 2, 1 4, 2, 4, 2, 1 (sub-type b) 3, 4, 2, 3, 4, 2, 3, 4, 2, 1 (sub-type b) 4, 3, 2, 4, 2, 3, 1 (sub-type c) 3, 4, 2, 4, 3, 2, 3, 4, 1 (sub-type c) Explanation of Table 3 — Each horizontally arranged group of figures describes a single trial; and each figure in such a group refers to an exit door tried. Thus, "4, 2, 3, 1" is descriptive of a trial during which the subject tried first to open door 4, following which he tried doors 2, 3 and 1 in the order given. The last figure in each horizontally arranged group refers to the exit door which afforded escape when the subject pushed against it. The examples under Type A obtain their true significance only when we assume that they describe trials during which door 3 was the impossible door. Type A Type B TypeC 4,2,1 4, 2, 3, 1 4, 3, 2, 1 2,4,1 3, 4, 2, 1 2, 3, 4, 1 1, 2, 3, 4 TRIAL AND ERROR REACTIONS IN MAMMALS 51 TABLE 4 Distribution of Classifiable Reactions Manifested by each Subject During His Series of 100 Trials Subject. Man 1 Boy 7 Boy 6F1 . . . Boy 5F1 . . . Boy 4F1 . . . Boy 3 Girl 1 Boy 2F1 . . . Boy 1 Man A . . . . Boy A Monkey 6 . . Monkey 3 . . Monkey 4 . . Monkey 2 . . Monkey 1 . . Dog 1 . . Dog 2 . . Dog 4F2 Dog 6F2 Dog 3F1 Dog 8F1 Dog 5F2 Dog 7F1 Dog 10F2. . Dog 9F1 . . . Dog llFl.. Dog 12F1 . . Dog 14F1.. Dogl3F2.. Dog 16F2.. Dog 18F1 . . Cat 1 Cat 2 Cat 3 Cat 7 Cat 5 . 34 . 15 . 13 . 12 . 10 . 10 . 10 . 8 . 26 . 45 . 11 . 15 . 5 1. 1. 1. . 3 1 .116 .109 .102 . 95 . 88 . 81 . 74 . 67 . 60 . 53 . 46 . 43 . 43 . 36 . 1 . 1 . 1 . 70 56 Horse 1 8 Age. years years years years years years years years months years years years years 5 years 5 years 5 years years year days days days days days days days days days days days days days days year year year days days years No. of classified reactions. 28 27 26 33 26 31 21 34 38 29 40 36 39 39 34 37 40 42 43 48 31 42 42 51 48 52 45 35 53 41 42 48 49 42 36 53 51 50 Distribution of classified reactions percentages of each subject's total number of the same. Type A Type B Type C Type D % m82.14 m70.37 m69.23 m84.85 m76.92 70.97 m85.71 m70.59 15.79 48.28 m62.50 19.44 23.08 17.95 14.71 16.22 12.50 11.90 11.63 16.67 25.81 4.76 7.14 17.65 16.67 15.38 20.00 11.43 7.55 17.07 14.29 12.50 12.24 4.76 8.33 7.55 5.88 8.00 % 17.86 25.92 23.08 15.15 23.08 19.35 14.29 17.65 5.26 m51.72 7.50 22.22 23.08 15.38 14.71 m29.73 12.50 30.95 16.28 27.08 m38.71 16.67 16.67 9.80 10.42 9.62 4.44 20.00 11.32 24.39 0.00 4.17 m30.61 11.90 13.89 1.89 19.61 4.00 % 0.00 0.00 7.69 0.00 0.00 3.23 0.00 5.88 18.42 0.00 30.00 19.44 20.51 12.82 m41.18 21.62 0 00 0.00 16.28 20.83 12.90 0.00 m28.57 9.80 2.08 13.46 m33.33 31.43 5.66 12.20 11.90 6.25 10.20 0.00 0.00 0.00 0.00 2.00 % 0.00 0.00 0.00 0.00 0.00 6.45 0.00 5.88 26.32 0.00 0.00 m27.78 m30.77 m28.21 14.71 21.62 30.00 m38.10 18.60 6.25 0.00 26.19 19.05 29.41 33.33 23.08 13.33 0.00 30.19 17.07 35.71 25.00 18.37 33.33 33.33 22.64 5.88 24.00 in TypeE % 0.00 3.70 0.00 0.00 0.00 0.00 0.00 0.00 m34.21 0.00 0.00 11.11 2.56 25.64 14.71 10.81 m45.00 19.05 m37.21 m29.17 22.58 m52.38 m28.57 m33.33 m37.50 m38.46 28.89 m37.14 m45.28 m29.27 m38.10 m52.08 28.57 m50.00 m44.44 m67.92 m68.63 m62.00 Discussion of table 4. The arrangement of the subjects in phyletic groups, and their arrangement within these groups according to age, enable us to gain from table 4 a general idea of the behavior of individuals as conforming to or varying from that which might be expected to obtain at a given position in the ontogenetic and phylogenetic scales. The highest per- centage for each subject is marked by an m in the table in order 52 G. V. HAMILTON to render it easy for the reader to tell at a glance the type of reaction most frequently manifested by any subject. The older normal human subjects and defective Boy A are seen to have " preferred " Type A reaction. Defective Man A preferred Type B, and the infant Boy i preferred Type E. The mature monkeys and one of the immature monkeys (No. 4) have their highest percentages in the Type D column, whilst immature Monkeys 2 and i have their highest percen- tages in Type C and Type B columns respectively. In view of the fact that reactions of Types B and C are more adequate modes of adjustment than is the Type D reaction, this finding is of considerable interest, as will be show^n in the next chapter, where a psychological interpretation of these types of reaction is attempted. Of the twenty -two animal subjects below the primates, only five preferred any other type of reaction to Type E. These exceptional cases require the following explanation. Dogs 2 and 3F1 were obviously more intelligent in meeting the situa- tions of everyday life than were their fellows. Dogs 5F2 and iiFi, who manifested a preference for Type C reactions, seem to have fixed upon this systematic mode of searching for the unlocked door much earlier in the experiment than is usually the case with dogs: after a variable number of trials (usually from 300 to 600) the average dog will manifest 100 per cent of Type C reactions. Cat I, who manifested a preference for Type B reactions, has already been described as an exceptionally intelligent animal (P- 43)- The evidences of marked individual differences contained in table 4 must be looked upon, of course, as a serious obstacle to any effort to deal with the results in terms of age and phyletic averages. It is obvious that such averages, to be available for conclusive interpretations, would have to be obtained from a far greater number of subjects for experiment than could be practically used in an exploratory investigation. However, since the averages obtainable from the above table may serve to attract attention to some interesting possibilities in genetic psychology, the writer is justified, I believe, in presenting them. TRIAL AND ERROR REACTIONS IN MAMMALS 53 TABLE 5 Distribution of Classified Reactions According To Age and Phyletic Averages Total Distribution of classified reactions classified in percentages of same. Average reac- TypeA TypeB TypeC TypeD TypeE Subjects. age. tions. % % % % % Older human (8) 14 years 226 76.11 19.47 2.21 1.77 0.44 ManA(l) 45 years 29 48.28 51.72 0.00 0.00 0.00 BoyA(l) 11 years 40 62.50 7.50 30.00 0.00 0.00 Infant (1) 26 months 38 15.79 5.26 18.42 26.32 34.21 Mature monkeys (2) 10 years 75 21.33 22.67 20.00 29.33 6.67 Immature monkeys (2).... 1.5 years 71 15.49 22.53 30.99 18.31 12.68 Mature dogs (2) 2 years 82 12.20 21.95 0.00 34.15 31.71 Older puppies (6) 98.50 days 357 13.61 19.84 14.79 17.51 34.24 Younger puppies (8) 52.75 days 364 14.26 10.16 13.74 23.08 38.76 Mature cats (3) 1 year 127 8.66 19.68 3.94 27.56 40.16 Kittens (2) 63 days 104 6.73 10.58 0.00 14.42 68.27 Horse (1) 8 years 50 8.00 4.00 2.00 24.00 62.00 Discussion of table 5. The significance of the findings of the table becomes more apparent when they are plotted in dis- tribution curves, as shown in figure 3 on the opposite page. Reference to these curves will disclose the following facts con- cerning the general effects of age and phyletic position on mode of adjustment: (i) The curve for the older human subjects attains its greatest height at point A, whence it descends rapidly to point B, thence to point C, where it closely approximates the base (zero) line. From C to D to E the curve descends continuously. (2) The infant's curve is relatively low at point A, whence it descends abruptly to point B, after which it ascends contin- uously and rather sharply imtil it reaches its maximum height at E. (3) Defective Man A's curve ascends from A to B; from B it descends directly to the base line, with which it is coincident at points C, D and E. (4) Defective Boy A's curve descends sharply from A to B, undergoes a marked secondary rise at C, and from there passes to the base line, with which it is coincident at D and E. (5) The mature monkey's curve ascends slightly from A to B; descends slightly from B to C; attains its maximum height at D, from which point it descends sharply to E, where it is near the base line. 54 G. V. HAMILTON Figure 3 — Curves showing the distribution of classified reactions. "A," "B," "C," D" and"E" refer to the types of classified reactions described in the text and the dots immediately beneath these letters indicate the percentages of the various reactions thus designated. TRIAL AND ERROR REACTIONS IN MAMMALS 55 (6) The immature monkey's curve differs radically from that of the miature monkey's in that it ascends continuously and rather sharply from A to B to C, attaining its greatest height at C; the descent from C to D to E is likewise sharp and con- tinuous. The curve is lower at A and higher at E than is the mature monkey's curve at these points. (7) The mature dogs' curve makes a considerable ascent from A to B; drops to the base line at C, from which it ascends to its maximum height at D; and makes a slight descent from D to E. In view of the fact that with prolonged experience the dog tends to manifest C reactions only, the absolute absence of these reactions, as reflected by the position of C in their curve of dis- tribution, is of considerable interest. (8) The older puppies' curve ascends from A to B; descends slightly from B to C, after which it ascends slightly to D; and ascends sharply to attain its maximum height at E. (9) The younger puppies' curve, when compared with that of the older puppies', affords us material of some value for ontogenetic interpretations, since the averages for these two groups include a considerable number of subjects. In both curves, A is relatively near the base line, but in the younger puppies' curve there is a descent from A to B, as compared with an A to B ascent in the older puppies' curve. From C to E both curves ascend continuously, but the younger puppies' curve makes a sharper ascent, and is higher at E. (10) The mature cats' curve bears a striking resemblance to that of the mature dogs, the only points in which they are essentially different being as follows: (a) The mature cats' curve is slightly above the base line at C, whilst in the case of the mature dogs it is coincident with the base line at C; (b) the mature cats' curve ascends from D to E, attaining its max- imum height at E, whilst the mature dogs' curve slightly descends from D to E. It is quite possible that averages for large number of cats and dogs would eft'ace these differences. It may be stated, however, that the writer's experience with these two classes of subjects leads him to believe that the average cat is more prone to manifest the Type E reaction than is the average dog. (11) The kittens' curve has many points of similarity with the older puppies' curve; it ascends from A to B; descends from 56 G. V. HAMILTON B to C, from which point it ascends to D; from D it ascends very sharply to E, where it attains its maximum height (12) The horse's curve is near the base line until it reaches C, from which it ascends sharply to D; from D it makes a still sharper ascent to E, the point of its maximum height. V. INTERPRETATIONS OF RESULTS. CONCLUSIONS. It has been shown, I believe, that the higher mammals man- ifest striking differences of modes of trial and error activity, and that these differences cannot be adequately expressed in terms of rapidity of habit formation alone. The results indicate, also, the need of intensive behavior-studies along hitherto unexplored lines of investigation. The present chapter seeks to relate the types of reaction that have been discussed in the foregoing to inferred reactive tendencies, and thus to assign psychological values to the curves of distribution shown in figure 3. This attempt must be prefaced, however, by a statement of the viewpoints from which the interpretations that follow have been undertaken. While the Comparative Psychologist has been almost exclu- sively concerned with a single psychological value — associative memory* — the field of Psychopathology has been revolutionized by a group of men who have shown that behavior is determined by a vast complexus of reactive tendencies which demand isola- tion and psychological estimation. These men, notably Freud (6), Jung (7), Bleuler (8), and Adolph Meyer (9), have opened up, by their activities along highly specialized lines of interest, new possibilities for the development of genetic psychology. They have shown, by implication, at least, that genetic psychol- ogy should not be solely concerned with the developmental history of the specifically adaptive instincts and of the mature human ability to think according to the traditional canons of logic. The dancing mouse's acquisition of a habit of selecting a spatially variable white labelled avenue of escape from pain, and the mathematician's solution of a problem in infinitesimal calculus are alike to be looked upon as end-products of reactive tendencies which have been variously subjected to the selective, suppressive, fixative and other corrective influences of experi- ' The writer assumes that functional* studies of the sense organs belong to Physi- ology rather than to Psychology. TRIAL AND ERROR REACTIONS IN MAMMALS 57 ence. We must have knowledge, therefore, of the genetic rela- tionships, not only of the adequately adaptive end-products in behavior, but of the reactive tendencies that lead to these end- products. The present investigation becomes intelligible only when account is taken of the fact that it seeks to deal with the funda- mental factors on which adaptation depends, rather than with the ultimate effects of experience on behavior. Figure III depicts, therefore, roughly determined curves of reactive ten- dency, and not curves of learning. Since their value as curves of reactive tendency will depend on the accuracy with which the objective facts are translated into terms of psychic entities, we are confronted by the difficult task of establishing an objec- tive criterion of the subjective. How may we recognize a reactive tendency in the behavior of our subjects? During the interpre- tations that follow, a tentatively constructed criterion will be adhered to: A mode of adjustment which appears, disappears and reappears consistently with ascent or descent of the age and phyletic scales may be looked upon as an expression of a definite reactive tendency. If this criterion be accepted, a tremendous amount of detailed investigation will be required to establish conclusively the ex- istence of even a few reactive tendencies as biological entities. With the above in mind as a qualification of what follows, we may proceed to an interpretation of results by first presenting a summary of the, values for reaction contained in the experiment : (i) All of the subjects brought to the formal experiment a more or less definite awareness of the four exit doors as possible means of escape from the apparatus, hence the demand for adjustment was essentially contained in the mere necessity of clawing, scratching or pushing at one or more previously mobile objects (all of the exit doors were left unlocked during the pre- liminary training) until activity proved successful. (2) Without exception, all of the subjects gave definite evi- dence of trying for success (escape from the apparatus) until success was attained. (3) For every subject a considerable percentage of trials led to more or less unsuccessful activity (trying locked doors). (4) In every case the individual trial was terminated by a definitely directed activity (trying the unlocked door). 58 G. V. HAMILTON (5) The relation of every present trial to its immediate pre- decessor was such as to exact a penalty of non-success for trying the unlocked door of the immediately preceding trial. (6) During a given trial, a second or third, etc., effort to open a particular door was invariably unsuccessful: a door which would not yield to a single, definitely directed attack during a given trial could not possibly be opened by the sub- ject during that trial. (7) The various doors were discriminable, one from another. Under the conditions just enumerated, the various types of reaction that we have isolated are capable of the following psychological interpretations : Type A. To conform to this type, the reaction must include a single, definite effort to open each of the three inferentially possible doors, and must not include an effort to open the infer- entially impossible door. It will be remembered that the impos- sible door varied from trial to trial. Now it is evident that only an awareness of the impossible door as such would enable any subject to manifest appreciably more than 50 per cent of Type A reactions out of his total number of classified reactions. If the impossible door were of indifferent value for reaction, either as impossible or as the object of the subject's latest successful activity, and if during no trial the subject were to make more than a single effort to open any one door, his record would tend to show 50 per cent of Type A reac- tions and 50 per cent of Type B or C reactions. Of course, a preference for the impossible door or a tendency to make more than a single continuous effort to open a particular door during a given trial would impair the subject's chance of approximating even 50 per cent of Type A reactions. What reactive tendency, then, would lead to more than 50 per cent of Type A reactions? It is obvious that this cannot be the primitive tendency to reduce diffuse activity-impulses to a definite attack upon a single object. The situation rendered it impossible for any subject to associate a simple object with successful activity and to obtain thereby a formula for invariable success. On the other hand, the establishment of a simple nega- tive association was not sufficient to enable any subject to avoid the impossible door. This spatially varying object-to-be-avoided could clearly obtain its true value for reaction only among the TRIAL AND ERROR REACTIONS IN MAMMALS 59 subjects who were able to elaborate a succession of previous experiences in such a manner as to associate, at each trial, the door last opened with an awareness equivalent to " during any present trial the door last opened is apt to be impossible as a means of exit." In other words, there would have to be an association in which one of the elements would be a complexly derived awareness of a principle deterrent to the activity in- volved in trying the unlocked door of the immediately preceding trial. The conscious avoidance of the impossible door as such may therefore be looked upon as due to a tendency to make rational inferences from a sequence of past experiences. For convenience of statement it will be referred to hereafter as " the rational inference tendency." Type B. This reaction involves trying all four doors, but once each, and in an irregular order. From the normal adult human viewpoint this adjustment contains but one error, viz., the inferentially and actually use- less effort to open the impossible door. In cases where there is clearly apparent a perception of the impossible door principle (i. e., where there is given a record of appreciably more than 50 per cent of Type A reactions), the manifestation of Type B reactions may be assigned either to mere inattention or to lapse of memory, or to the necessity of falling into the impossible door error a certain number of times before an awareness of the principle can be obtained. If we exclude these cases, the Type B reaction may be looked upon as an expression of a high type of searching tendency. Although it shows a lack of modifi- cation by the higher tendency to make rational inferences, it is of much significance as showing the absence of modification by the low^er tendencies to which we shall ascribe reactions of the C, D and E types. Of course the inclusion of an effort to open the impossible door may be due, in a certain number of cases, to the interference of a tendency to associate the last successful activity with the impossible door, in which case the searching tendency is not the sole reactive factor. But since we are deal- ing only with the reactive tendencies that precede the establish- ment of definite and habitual associations, we may, for con- venience, refer to Type B reactions as due to " the unmodified searching tendency." This, of course, only when the number of Type A reactions is so small as to exclude the possibility that 60 G. V. HAMILTON mere inattention or forgetfulness have interfered with an ade- quate expression of the rational inference tendency. Type C. This highly interesting mode of adjustment can occur, as has been explained, only when Door i or Door 4 is the unlocked door for the trial. It involves merely the act of trying (i) Doors I, 2, 3 and 4 but once each and in the order given when Door 4 is the unlocked door, or (2) doors 4, 3, 2 and I but once each and in the order given when door I is the unlocked door. The habit of starting at either the first or the last door from the left and working down the line of doors, striking each as it is passed until an unlocked door is found, deserves, in itself, a far more extensive investiga- tion than it has thus far received. My observations of the higher infra-human mammals, of children and of mentally defective or diseased persons lead me to believe that in this mode of adjustment we have the expression of a reactive ten- dency which has extensive genetic relationships, and which can be more easily recognized in behavior than can any other reactive tendency of which we have knowledge. In the discus- sions that follow Type C reactions will be referred to as due to ' ' the tendency to adopt stereotyped modes of searching. ' ' Type D. This reaction involves the error of making more than one separate, continuous effort to open a given door during the same trial, but always with an interruption of such repeti- tions of activity by an interval of effort to open one or more of the other doors. Since this mode of adjustment is objectively continuous with a form of Type E reaction (sub-type c, described below), and yet clearl)^ should not be made to include the latter, I have excluded from Type D all reactions involving more than six separate efforts to open doors during a given trial. A characteristic example of the Type D reaction will render its interpretation more intelligible. During his fortieth trial Dog 16F2 made a vigorous effort to open Door i, which was locked; failing in this effort, he tried to open Door 2, which also baffled him in his efforts to escape. Then he returned to Door I and made a second effort to open it by giving it two or three feeble scratches, after which he tried Doors 3 and 4, the latter of w^hich yielded to his attack. Anybody who has ever sought vainly, and with some irrita- tion, a lost collar button, will readily appreciate the inner sig- TRIAL AND ERROR REACTIONS IN MAMMALS 61 nificance of this behavior. One looks into every likely nook and corner, then remembers that on previous occasions he has discovered lost articles in a drawer which is seldom used because of its tendency to " stick." The drawer is opened and is found to be quite empty; one turns away and looks elsewhere, only to return in a moment to the empty drawer and again to open it in a stupid, unthinking manner. The impulse to open the drawer seems to have subsided with the first failure, only to come surg- ing back with most inappropriate persistence. Type D reactions will hereafter be referred to as due to " the searching tendency modified by recrudescent motor impulses." Type E. This type includes several different modes of behav- ior which have a common objective characteristic, viz., auto- matism. That is, the subject behaves in a relatively implastic, unadaptive manner. The objective characteristics of the various sub-types may be reviewed as follows: Sub-type a. The subject makes two or more successive but definitely separate attempts to open the same door during a given trial. Thus he tries Door 3, finds it locked, turns away from it, returns to this same door and makes a second effort to open it without having tried any other door in the meantime, etc. In view of the fact that in the majority of cases such per- sistence in returning to the same door during a given trial could not be attributed to the recency with which it had afforded escape (as compared with the recency with which the other doors had afforded escape), we are justified in assuming, I believe, that the sub-type a reaction is an expression of the unmodified primitive tendency to repeat an activity, once it is begun, until it leads quite definitely to pain or success. Sub-type b. During a given trial the subject tries a group of locked doors two or more times in an unvarying order. Cat I's ninety-fifth trial well illustrates this mode of adjustment. When this animal entered the apparatus to meet, for the ninety- fifth time, a situation which merely required that she find the one unlocked door, she tried the exit doors in the following order: 2-1-4 " 2-1-4 ~ 2-1-4 ~ 2-1-3- ^ have italicised each of three exactly similar cycles of activity in order to bring out more clearly the characteristic features of a mode of adjustment which seemed to spring from a persistent impulse to try Doors 2, I and 4 again and again, in an unvarying order. Now during 62 G. V. HAMILTON her experience with ninety-four previous trials she had found Door 3 unlocked twenty-three times, Door 2 twenty-four times, Door I twenty-four times and Door 4 twenty-three times. It seems that no elements in these experiences were sufficient to awaken an impulse to try Door 3 after she had given definite expression to the first impulses to try Doors 4, i and 2 ; and that these latter impulses continued to reassert themselves as a connected whole until a break in the fourth cycle of activities led her to try Door 3 instead of Door 4. Sub-type c. The subject, having avoided a given exit door dur- ing a trial, continues to avoid it while the other doors are tried at least six times and these six or more efforts to open the other doors do not contain errors of either the sub-type a or sub-type b kind. The ninety-fourth trial of Dog 9F1 affords an example of this behavior. On entering the apparatus she went to door 3, raised her paw as if about to strike it, then desisted without having touched the door. Following this she tried the doors in the following order: 1-2-4-2-1-2-4-1-2-3. The effect of the initial inhibition of the impulse to strike door 3 is apparent, I believe, in the reaction-formula just given; this inhibition persisted, so that whenever the subject passed door 3 she failed to include it in her list of doors to be tried. If the order in which she tried the other doors had suggested a mere persevera- tion of active motor impulses, or if she had tried these doors less than six times, the writer would not have felt justified in tabulat- ing her reaction as belonging to sub-type c. In view of the fact that many of the reactions manifested by the infant and the animals presented the characteristics of all three Type E sub-types, it has been found more satisfactory to deal with all Type E reactions as a unit for analysis and in- terpretation. This is justified, I believe, by the fact that the three sub-types are alike interpretable in terms of a single general tendency, viz., perseveration of impulses. Finer analyses of behavior than are possible in the present investigation would doubtless show that we are here dealing with a group of several distinctly different primitive reactive tendencies. As a psycho- pathologist, the writer finds much interest in the fact that a clinical phenomenon common to the dementia praecox group of psychoses is met with at certain points in the normal onto- genetic and phylogenetic scales; wherever a tendency to "per- TRIAL AND ERROR REACTIONS IN MAMMALS 63 severation " exists, it is apt to find expression in the continuous persistence of both useless activities and useless avoidances of activities. The katatonic who continues a motor impulse to the point of catalepsy, or who utters a single word for hours without interruption, is always a patient in whose behavior we expect also to find inordinate persistence of inhibitions. My investigation of normal behavior has disclosed the same asso- ciation of the one phenomenon with the other in the case of dogs, cats, a horse, and a human infant. The fifth trial of the infant affords an example of this; he tried the various doors in the following order: 3-1-3-4-1-4-3-4-4-4-3-1-4-3-2. Even more striking is the fifty-ninth reaction of Cat 5, who tried the doors in the following order: 3-4-1-3-3-1-3-4-4-1-3-1-3-4- 4-1-4-3-2. In each case there are apparent both types of per- severation. VI. SUMMARY AND CONCLUSIONS The above interpretations of Types A, B, C, D and E re- actions in terms of reactive tendencies to which they may be ascribed now enable us tentatively to assign psychological values to the genetic curves of distribution in figure 3. (i) The rational inference tendency (Type A) is clearly ap- parent in the behavior of only the eight normal human subjects whose ages range from eight to thirty-four years, and in the behavior of defective Boy A. Morgan's (10) law of parsimony as applied to interpretations of behavior, requires us to assign the behavior of all other subjects, including that of defective Man A, to lower reactive tendencies. We have not ruled out the possibility, of course, that with sufficient experience any of the subjects would manifest a sufficient percentage of Type A reactions to indicate the presence of the rational inference tendency; nor that the subjects who manifested relatively low percentages of these reactions were wholly uninfluenced by the tendency in the question. (2) The unmodified searching tendency (Type B) finds its most frequent expression in the behavior of defective Man A. Among adult animals, the monkeys rank first in this respect, the dogs second, the cats third, and the horse fourth. Of all the subjects, taken as individuals and without regard to age, the horse seems to have been least affected by the unmodified searching tendency. 64 G. V. HAMILTON The ontogenetic findings parallel the phylogenetic ; among human subjects, monkeys, dogs of three different age-groups, and cats, there is apparent a tendency for a sufficient decrease in age to decrease the percentage of Type B reactions (3) The tendency to adopt stereotyped modes of searching (Type C) seems to acquire its maximum phylogenetic value for the monkeys ; it is but slightly apparent in the behavior of the older normal human subjects and the other mature animals. The ontogenetic relationships of this reactive tendency are interesting. The immature monkeys and defective Boy A are especially affected by it; but it is more apparent in the be- havior of the older puppies than in that of either the mature dogs or the younger puppies. The latter circumstance is a matter of some perplexity to the writer, since, as has been said, the Type C reaction becomes an habitual mode of adjustment with mature dogs who have had sufficiently prolonged expe- rience with the situations of the experiment. (4) The searching tendency modified by recrudescent motor impulses (Type D reaction) regularly increases in frequency of manifestation as we descend the phyletic scale of mature sub- jects until we reach the mature dogs, at which point it attains its maximum frequency. This tendency decreases as we pass further down the scale through the cats to the horse. Ontogenetically, it increases in frequency of manifestation with descent from older childhood to infancy; but, with one exception, the younger animals are less affected by it than are their older fellows ; the younger puppies give a higher percentage of the Type D reactions than do the older puppies. (5) The tendency toward perseveration of active motor im- pulses and of inhibitions (Type E) increases regularly in the frequency of its manifestation as both the ontogenetic and phylogenetic scales are descended. It is of interest that during the total one thousand trials of the ten older human subjects (the two defectives being included) there was but one manifes- tation of this tendency — the first reaction of the much embar- rassed Boy 7, whilst 34.21 per cent of the infant's classified reactions may be ascribed to this tendency. The present investigation has afforded experimental evidence that the phylogenetic and ontogenetic differences of adequacy of mammalian adjustments are to be accounted for not merely TRIAL AND ERROR REACTIONS IN MAMMALS 65 in terms of sense-physiology and of associative habit formation, but in terms of reactive tendencies as well. Evidence has also been adduced to support the view that the elaboration of ex- perience involves an interplay of conflicting reactive tendencies, and that these are experimentally isolable according to criteria legitimate to genetic psychology. A final and still more general conclusion to be drawn from the above is as follows : Since the ideal development of genetic psychology demands the unravelling of long and intricately interwoven mental complexes, not only as they appear in a given species or at a given age, but throughout extensive phyletic and age series, our intensive studies of behavior are apt to be- come irrelevant to the broader issues in which they seek justi- fication if we do not explore, from time to time, for the general patterns according to which the threads of fact are arranged. The satisfaction derived from the accurate and conclusive determination of the quality and dimensions of a single thread as it appears in a carefully delimited part of its course is apt to blind the genetic psychologist to the historical significance of his own attitude. Darwin (ii), Spencer (12), Romanes (13), and Baldwin (14), among others, have sought to deduce from ex- tensive ranges of facts the more general principles of mental development. Their statements concerning general principles have led to the formulation of many detailed problems, and to methods appropriate to the investigation of these problems, with the result that the observations of behavior recorded by the older w'riters are now shown to have been made under insufficiently controlled conditions, and that their interpre- tations were often at fault. The obvious need of detailed in- vestigation as thus disclosed, everywhere finds recognition in the work of the younger students of behavior. But the equally obvious need of seeking direction anew from a general survey of comprehensive collections of facts at the expense, if need be, of some sacrifice of accuracy of detail, seems to have gained nothing more substantial than verbal recognition. The writer begs to acknowledge his indebtedness to Professor Robert M. Yerkes, Professor Adolf Meyer, and Mr. George R. Agassi z for many helpful suggestions and criticisms. 66 G. V. HAMILTON REFERENCES 1. Yerkes, R. M. and Dodson, John D. The Relation of Strength of Stimulus 1908. to Rapidity of Habit -Format ion. Jour. Comp. Neurol, and Psychol, vol. 18, No. 5, pp. 459-491. 2. Yerkes, R. M. The Dancing Mouse, p. 3. New York. 1907. 3. Hamilton, G. V. An Experimental Study of an Unusual Type of Reaction 1907. in a Dog. Jour. Comp. Neurol, and Psychol., vol. 17, No. 4, pp. 329-341. 4. Berry, C. S. An Experimental Study of Imitation in Cats. Jour. Comp. 1908. Neurol, and Psychol, vol. 17, No. 4, pp. 329-341. 5. Thorndike, E. L. Animal Intelligence. Psychol. Review, Monograph Series, 1898. vol. 2, No. 4. 6. Freud, Sigmund. Die Traumdeutung. Vienna. 1900. 1907. Zur Psychopathologie des Alltagslebens. Berlin. 7. Jung, C. G. Ueber die Psychologic der Dementia Praecox. Halle, a. S. 1907. 8. Bleuler, E. Affectivitat, Suggestibilitat, Paranoia. Halle a. S. 1906. 9. Meyer, Adolf. The Problems of Mental Reaction-Types, Mental Causes and 1908. Diseases. Psychol Bulletin, vol. 5, No. 8, pp. 245-261. 10. Morgan, C. L. An Introduction to Comparative Psychology, p. 53. New 1904. York. 11. Darwin, C. R. Expression of the Emotions in Men and Animals. 1872. 12. Spencer, Herbert. Principles of Psychology. 1855. 13. Romanes, G. J. Animal Intelligence. New York. 1883. 14. Baldwin, J. M. Mental Development in the Child and the Race. New York. 1903. A NOTE ON LEARNING IN PARAMECIUM By LUCY M. DAY AND MADISON BENTLEY From the Cornell Lahorutory for Comparative Psychology ONE FIGURE If Paramecium learns we may expect to find its learning based upon some slight modification of that " action system " which is brought into function by the common activities of the organism. Modification, however, is not in itself the equivalent of learning. It suggests learning only when it is preserved by the animal and used as an " acquirement " upon a later occasion. The experiments which follow sought to induce such a modifi- cation and to record — should the change persist — its subsequent effects. The protozoa were isolated and studied individually; for it seemed important to duplicate, so far as was possible, the experimental control exercised in the study of the higher forms of life. With this end in view, the writers made use of the capillary- pool, drawing a single Paramecium into a glass tube the diameter of whose lumen was a little less than the animal's length. Then by applying the lips to the larger end of the tube, the pool of culture-water containing the subject was drawn up and away from the tip of the tube and reduced to .5-2.0 cm. in length. The tube was then stuck with bits of wax to a long strip of glass for support and placed at once under the microscope at low power. The observer watched the pool continuously, recording with a key and kymograph backward and forward movements, wheeling movements at the meniscus, partial turns across the tube, and complete changes of direction (reversal). Paramecium succeeded in changing direction by bending its anterior end at the oral groove until the backward beat of the cilia carried it along the wall of the tube and to the rear.^ 1 Mter the method had been developed, an article on The Limits of Educa- bility in Paramecium, by Stevenson Smith, appeared in the Journal of Comp. Neur. and Psychol., vol. 18, p. 499, 1908. Smith observed reversal in the capillaiy-tube and remarked that the time was in some cases reduced in the course of 12 hours or more "from 4 or 5 minutes to a second or two" (p. 507). He seems not to have observed the process of reduction or to have controlled chemical changes in the tube. It is worthy of remark that the present writers detected an occasional turn towards the oral side. This unusual performance may have been due to con- tact with the tube; they are at present unable to say, 67 68 DAY AND BENTLEY Table I records the first fifteen reversals for eight individuals. Thus D turned in the tube twenty-five times before it brought its axis around more than 90° and swam to the opposite end of the pool. The next reversal came at the twenty-ninth turn, and so on. The figures at the bottom of the table show" that the first two reversals required, on the average, the greatest number of abortive, partial turns (22.6 and 15.5). After that, the averages fall into three groups: 3-4, 5-8, 9-15. The mean variation (last horizontal line) follows the same course. The averages and mean variation for the four groups are: 19.0 ±3.6, 8.i±i.5, 5-5±-3. 3-5±-5- TABLE I Number of Trials Necessary to Reversal for the First Fifteen Reversals Reversals I 2 3 4 S 6 7 8 9 10 II 12 13 14 Paramecia 15 D 25 29 8 4 3 1 10 3 1 1 2 1 2 1 1 E 39 17 2 2 5 3 2 7 2 2 1 1 1 3 1 G 13 44 10 37 5 19 5 2 6 1 3 5 8 2 3 H 26 19 17 3 3 1 13 7 16 4 10 6 5 21 20 5 1 2 11 3 12 4 12 4 11 10 12 1 6 K 5 N 20 6 2 3 1 2 1 3 3 2 1 3 1 2 1 0 31 5 9 4 3 1 2 1 3 2 2 2 1 4 1 X 8 3 18 7 2 2 2 3 8 1 2 4 5 2 5 Average 22.6 15.5 6.6 9.6 4.9 5.5 6.0 5.5 3.3 2.9 3.4 4.0 4.9 3.4 2.9 M. V 7.6 11.3 4.6 7.7 2.9 4.6 4.8 4.0 1.9 2.1 2.3 2.3 3.6 2.3 1.9 Table II gives the total times, in seconds, consumed between successive reversals. As an expression of change in behavior, it is less satisfactory than Table I, because the subjects some- times became inactive between " trials;" that is, they " wasted time," and since activity tended to diminish with the lapse of time, the periods of quiescence in the latter part of the table mask the actual increase of facility in reversing. A comparison of the two tables, line by line, will make this fact evident. The LEARNING IN PARAMECIUM 69 times do show, however, in spite of this circumstance, a marked decrease. Again, the averages fall roughly into groups: 1-2, 3-7, 8-15. The group-averages stand: 84. ±4.7, 45.8 ±9.3, 33.7 ±4.4. The irregularity in the mean variation is plainly due to wide individual differences of activity. Time in TABLE II Seconds Between Reversals Reversals I 2 133 78 202 56 32 41 26 16 3 33 5 95 21 7 41 82 61 4 5 6 7 8 9 12 13 14 Paramecia 10 11 15 D 117 139 77 65 99 60 102 50 24 17 182 65 80 46 12 83 21 9 56 62 37 36 8 16 4 13 137 42 66 39 6 16 36 3 75 45 204 23 8 17 28 39 28 105 50 30 6 16 9 8 49 18 42 41 22 82 2 22 35 134 51 27 7 16 20 2 51 65 66 17 7 16 2 3 64 78 52 46 9 32 12 2 98 65 97 35 5 21 6 11 42 87 2 32 15 15 4 4 49 E G H 50 31 19 6 4Q K 'n' 0 X Average 88.6 79.3 43.1 63.6 30.6 40.4 51.4 37.8 33.9 36.8 30.5 35.8 41.9 26.3 26.5 M. V 25.6 50.4 27.1 39.0 17.1 31.0 44.1 20.2 19.6 27.9 22.6 24.3 33.6 20.6 18.3 Thus, G, H, N, and X were either consistently or spasmodi- cally sluggish. The slow and uncertain changes in their beha- vior may be the direct result of their state, or they may be due to the long blank intervals of " unlearning " between successive reversals. Comparison with D, E, and 0 (long initial times) suggests the former interpretation.^ It was, now, evident to the observers that the apparent increase of facility in turning might be due to some change in the medium (such as increase in CO2 or decrease in oxygen) brought about by the paramecia themselves. Until this possibility had been ^ The tests were in all cases carried beyond the fifteenth full turn. The highest number of turns in this set was 123 (subject E). After the fifteenth, however, the subjects either continued to turn with little or no delay or else became inactive. For a further slight increase in facility, see Tables III and IV below. ro DAY AND BENTLEY eliminated by experimental control, it was impossible to ascribe the changes observed to conscious, or even to purely organic, conditions. A number of subjects were transferred, therefore, at the end of the test, to an open watch-glass of culture-water; then after an interval of ten to twenty minutes they were re- placed in the capillary-tube and observed as before. In this way the animal was brought for a second trial under the same external conditions as at first. If the first reduction of times and turns was owing to the environmental changes which we have just assumed to exist, then we might expect the second performance to repeat the first. On the other hand, if the individual had really " learned " during the first trial, we might well look for subsequent modification. The curves, figure i, give the results (number of partial turns before each successive reversal) for six new^ subjects. A, B, C, M, Q, and R. In the case of C, we succeeded in carrying the individual through a third test after a second interval of thirty minutes. Tables III, IV, and V are designed to interpret the curves of figure i. They give the number of reversals in each phase of the curve, together with the average number of partial turns or " trials " necessary to a single reversal. TABLE III Phase 1 Phase 2 Phase 3 Paramecia No of reversals. Ave. No. of trials. No. of reversals. Ave. No. of trials. No. of reversals. Ave. No. of trials. A B C M Q R 10 10 5 5 10 2 13.6 19.7 13.6 10.2 4.2 13.5 40 60 75 30 50 18 3.0 1.7 1.5 2.7 1.3 2.7 23 19 46 38 23 83 2.0 1.1 1.2 2.5 1.5 1-6 Gen. Ave. 12.5 2.2 1.7 M. V 3.5 .7 .4 LEARNING IN PARAMECIUM 71 TRIALS ':«- EXTEND TO 72 Fig. I. 72 DAY AND BENTLEY TABLE IV After Interval in Watch Glass Interval (in mins.) Phase 1 Phase 2 Paramecia No. of reversals Ave. No. of trials No. of reversals Ave. No. of trials A 10 15 20 1 3.4 21.0 59 6 1.2 B 1.3 C 20 10 2.7 40 1.1 M 10 2 15.5 5 1.8 Q 10 9 4.0 R 10 5 5.0 24 1.6 Gen. Ave 8.6 1.4 M. V 6.4 .2 TABLE V Paramecium C. after Second Interval (30 Min.) in Watch Glass Phase 1 No. of reversals Ave. No. of trials 6.0 Phase 2 No. of reversals 19 Ave. No. of trials 1.05 The curves and tables suggest a marked difference in the history of the first and of subsequent tests. The curves for the first test are much like the " trial-and -error " curves for mammalian forms. They begin high, drop quickly if irregu- larly, pass through a middle, saw-tooth phase, then drop to a low level with small fluctuation (cf. Table I). In the second test, the curve usually starts lower, drops, shows a reduction or absence of the second phase, and ends in a low flat line nearly parallel to the axis of abscissas.^ * M and Q are less satisfactory than the others. In the second test, these indi- viduals became very inactive and soon ceased to move. C is perhaps typical; for it maintained active movements nearly the whole time. LEARNING IN PARAMECIUM 73 A word should be added on the negative cases. We have examined many individuals not reported in the text. Not infrequently the work of whole days came to no issue. But failure was due either to poor cultures, to the technical diffi- culties of the experiment, or to the sluggish inactivity of the subject. A normal Paramecium once brought without injury into a pool of convenient size never failed — provided it remained active after the first few reversals — to show a modification of the type exemplified in the results. It appears from the experiments (i) that Paramecium is capable of modifying within a few minutes its usual avoiding reaction; the lateral turn is so increased and prolonged as to permit the animal to reverse its long axis in a narrow circular tube; (2) that the effect of this modification remains for some time independently of external changes induced. Observations of this nature are taken as evidence of learning. The evidence in this case is supported by the fact that during the process of turning, the animal had the appearance of doing a definite thing. Nevertheless, it is questionable whether this kind of "learning" involves consciousness; whether it is not as well interpreted as the result of purely organic processes. Further experiments designed to control more closely the con- ditions of the performance, and especially to ascertain the term of the modification induced, are in progress. WHEELER ON ANTS i By Robert M. Yerkes A series of pleasant pictures comes to mind as I write these words. I see myself, in attentive attitude, standing in the midst of a banana plantation in Colombia. On the ground at my feet I see a train, and isolated groups, of animated leaves moving across my pathway. My first thought is, what a beautiful instance of the mimicry of leaves by some insect. A moment later I detect beneath its green burden the huge head and heavy mandibles of a leaf-cutting ant. I had not expected this interesting sight and in consequence I was slow to appreciate its meaning. Yet another picture vividly presents itself. Gathered beside a drainage ditch at the edge of a road-side jungle is a group of native laborers from the plantation. Each is intently watching the ground between the ditch and the jungle. All are talking animatedly. Soon the group dissolves and I see myself standing where it had been. On the ground busily engaged in carrying bits of leaves from place to place, is an army of ants. They are apparently undis- turbed by being observed. Each seems intent on reaching his goal. There is a seriousness about the business which is impressive. I wonder what the natives were thinking and saying, what in the appearance and behavior of the insects caused them to attend. For me these vivid pictures typify the human interest in living things, and they constantly remind me that one need not be a student of the behavior, or structure, or mind of animals in order to be interested in ants. In his book Professor Wheeler treats alike skillfully and in an eminently interesting manner of the structure, the development, the behavior, and the psychology of ants. The work is thorough, although extensive; it is readable even for the amateur naturalist, although scholarly and indicative of careful analytic study of fundamental prob- lems of biology as they are presented in ants. The author is so evidently master of his subject and so deeply interested in making an excellent presentation of the facts that he should be permitted to speak for himself in this review. The book cannot be summarized; one can merely describe it briefly, characterize it in the light of his appreciations, and urge every student of animals to read it. ' It is a matter of common observation that the higher animals — those, namely, that in structure and behavior are most like ourselves — are also the ones which arouse our keenest interest .... The only lower animals that from immemorial time have retained a like interest for man are certain insects — the social bees and wasps, the termites and the ants. And among these what appeals so forcibly to the imagination is not the structure or activities of the individuals as such, but the extraordinary instincts which compel them to live permanently in intimate associations (p. 1). No other group of animals presents such a maze of fascinating problems to the biologist, psychologist and sociologist. It will suffice to mention the imrivalled material which they present for the study of variation and geographical distribution, both from the taxonouiic and experimental standpoints, the extraordinary phenomena of polymorphism, parthenogenesis ^ Wheeler, William Morton: Ants: their structure, development, and behavior. — The Columbia University Biological Series, IX. New York: The Macmillan Com- pany, 1910. Pp. XXV +663. With 286 text figures and extensive bibliography. 74 WHEELER ON ANTS 75 and sex determination; the wonderful cases of parasitism and symbiosis, and last, but not least, the great importance of these insects in the problems of instinct and intelligence." (pp. 11, 12) These sentences are quoted from chapter I, which carries the title " Ants as dominant insects," and serves as an admirable introduction to the more technical materials which follow it. The chapter is throughout quotable and exceptionally interesting to students of animal behavior and psychology. There follow chapters on the external and internal structure and the develop- ment of the insects. The descriptions are necessarily brief, but the materials are well chosen and clearly presented. Especially valuable to the naturalist, whether his interests be physiological, psychological, or sociological, is chapter IV, in which a general account of the nervous system is given. Of the curious chordotonal organs the author writes," Recent studies have shown that these structures, which are present in a great many insects, even in the larval stages, are typically compact, spindle-shaped bundles of sensillae, each consisting of a chitin-secreting gland and a nerve cell. These cells are arranged in a series at an angle to the integument and are stretched, like a tendon, across a cavity between opposite points in the cuticle, or between a point in the cuticle and some internal organ. The gland cell secretes and retains within its cytoplasm a peculiar cone or rod, known as the scolopal body. The chordotonal organs are supposed to be auditory in function, because they are most elaborately developed in the stridulating Orthoptera (crickets and katydids), and because their structure would seem to be adapted to responding like the chords of a musical instrument to delicate vibrations. In ants the development of these sense-organs is greatly inferior to that of the Orthoptera just mentioned but they are nevertheless very easily seen when one knows exactly where to look for them." (pp. 62, 63) In addition to the chordotonal organs, at least six other types of sense-organ are described; the tactile sensillae (organs of touch); the olfactory and gustatory sensillae (organs of smell and taste); the Johnstonian organ (probably an organ of hearing); the campaniform sensillae (whose function is unknown) the lateral eyes and the median eyes or stemmata (organs of vision). In a later chapter, XXVII, the functions of the organs of sense are further discussed under the heading," The sensations of ants." At the beginning of this chapter Professor Wheeler makes evident his attitude toward diverse methods of studying animals. He writes, " In endeavoring to gain an insight into the behavior of any animal, two courses are open to us. These may be designated as the intel- lectual and the intuitional, and it depends on the temperament and training of the observer which he will follow, or whether he will be inclined to follow both. The intellectual course is the one usually pursued by the scientist pure and simple, and is especially exalted by those most thoroughly embued with the spirit of our laboratories, where living organisms are best loved when they are dead, or, at any rate, when they can be subjected to the methods of investigation that have yielded such valuable results to the development of physics and chemistry The intuitionist, in dealing with the behavior of animals, proceeds along the path of aesthetic insight, sympathy and introspective knowledge of our own internal processes. His method is, therefore, essentially psychological and metaphysical. He does not deal with things or quantities, but with the living creative movement as immediately experienced in his own consciousness. He attempts to place him- self en rapport with the organism and to move in the stream of its vital current .... 76 ROBERT M. YERKES Both methods, when carried to extremes, lead to false or inane, or, at best, very partial interpretations — the scientific to a kind of animal phoronomy, hke the reflex-theories of Bethe and Uexkiill, the intuitional to the humanizing of animals and all the perversities of the American "nature fakers." If I decline to join the ranks of those whose only ambition is to describe and measure the visible movements of animals, and am willing to resort to a comparative psychology in which inferences from analogy with our own mental processes shall have a place, I do this, not because I believe that the former course would be altogether unfruitful or uninteresting, but because the latter seems to me to promise a deeper and more satisfactory insight into the animal mind." (pp. 505, 506, 507.) Polymorphism receives thorough and illuminating discussion, as do also such topics as the history of myrmecology, the classification and distribution of ants, and fossil ants. A fascinating chapter, XI, on the habits of ants in general prepares the way for detailed accounts of the habits and instincts of the pomerine ants, the driver and legionary ants, the harvesting ants, the fungus-growing ants, and the honey ants. I quote a fragment from chapter XI to indicate the nature of the treatment of habit. " Having previously described the nuptial flight the author continues his account of the behavior of the female thus. " On descending to the earth the fertilized female divests herself of her easily detached wings, either by pulling them off with her legs and jaws or by rubbing them off against the grass-blades, pebbles or soil. This act of dealation is the signal for important physiological and psychological changes. She is now an isolated being, henceforth restricted to a purely terres- trial existence, and has gone back to the ancestral level of the solitary female Hymenopteron. During her life in the parental nest she stored her body with food in the form of masses of fat and bulky wing-muscles With this physiological endo"muent and with an elaborate inherited disposition, ordinarily called instinct, she sets out alone to create a colony out of her substance. She begins by excavating a small burrow, either in the open soil, under some stone, or in rotten wood. She enlarges the blind end of the burrow to form a small chamber and then completely closes the opening to the outside world. The labor of excavating often wears away all her mandibvilar teeth, rubs the hairs from her body and mars her burnished or sculptured armor, thus producing a number of mutilations, which, though occurring generation after generation in species that nest in hard, stony soil, are, of course, never inherited "(PP- 184, 185.) The chapter consists of just such clearly drawn and interesting pictures of ant life as this. Of the special habits, and other activities and relations, which receive con- sideration, mention may be made of nest-building, compound nests, the relations of ants to plants, to other insects, and to one another. The chapters on parasitism and slave-making are especially valuable, for they make available in readable form a mass of information which is of extreme importance for a true appreciation of the social life of ants. Last, and for the student of the mind of animals most important, the chapters on instinct and intelligence may be characterized. They are full of facts, rich in penetrating analyses, stimulating and encouraging to those who despair of the solution of the problems which center about these concepts. Again the author may be permitted to speak for himself. " If ants exhibited merely the reflexes, or such brief and simple responses to sensory stimuli as we have been considering in the preceding chapter, their lives would flow on with the same monotonous regularity as those of many other insects WHEELER ON ANTS 77 and the lower invertebrates in general. In addition to these reflexes, however, ants manifest more comphcated trains of behavior, the so-called instincts; and both these and the reflexes may be affected with a certain modifiability or plasticity which, in its highest manifestations, has been called intelligence." (p. 518.) " In addition to instinct, two types of plastic behavior may be distinguished in ants: first, random behavior, like that observed by Jennings, Holmes, Yerkes and others in so many of the lower invertebrates and by Lloyd Morgan, Thorndike, Hobhouse and others in the higher animals. Random, or " trial and error " movements, occur, so to speak, in the very bosom of the instincts, as, for example, when an ant goes out to forage for food that has not as yet been located." "A second type of behavior is that in which the organism when confronted with a new situation does not proceed to make random movements, but at once adapts itself to the situation by a process which some authors (Loeb, Turner) have called associative memory. The nature of this process is, of course, a matter of conjecture and on this account it is differently conceived by different authors. Before con- sidering this matter, however, we may pass in review the main facts that compel us to postulate the existence of some form of memory in ants. These facts may be grouped under the heads of foraging and homing, recognition of nest mates and aliens, communication, imitation, co-operation and docility " (pp. 531, 532). " In conclusion it may be noted that all the activities of ants, their reflexes and instincts, as weU as their plastic behavior, gain in precision with repetition. In other words, all their activities may be secondarily mechanized to form habits, in the restricted sense of the word. This is tantamount to saying that even the reflexes and instincts are not so steorotyped but that they may become more so by exercise during the lifetime of the individual. And not only do ants thus form habits, but, as several myrmecologists have observed, these habits when once formed are often hard to break. It is certain that many instincts among the higher animals are at first incomplete or indefinite and are guided into their proper course by stimuli that effect the organism at a later period. This is probably true also of many formicine instincts. There is little doubt, more over, that the more fixed or sterotyped instincts are phylogenetically the older. This fact, and the close superficial resemblance of habits to instincts, has led many authors to derive the latter from the former. The views on the origin of automatic behavior, how- ever, are so diverse and conflicting that they cannot be satisfactorily considered without entering into a discussion of the doctrines of the NeodaiTvinians, Neo- lamarkians and those who believe in coincident, or organic selection. In my opinion we have little to gain at the present time from such a discussion It is, in fact, quite futile, to attempt a phylogenetic derivation of the automatic from the plastic activities or vice versa, for both represent primitive and fundamental tendencies of living protoplasm and hence of all organisms. As instinct, one of these tendencies reaches its most complex manifestation in the Formicidae, while the other blossoms in the intelligent activities of men " (pp. 543, 544). Professor Wheeler's book commands the attention alike of morphologists, physi- ologists, and psychologists, and for each group it has much of fact, interpretation, and theory that is of value. Its appearance has helped greatly to establish the scientific status of work in animal behavior and comparative psychology in An:ierica. Scientists who are also scholars and men of breadth and sanity of view are too rare for the work of one of them to escape the world's appreciations. " Ants"" stands as a comprehensive, reliable, eminently readable, thoroughly scientific account of one of the most important and interesting of organisms. The Animal Behavior Series as so far issued includes two volumes: THE DANCING MOUSE A Study in Animal Behavior. By Robert Mearns Yerkes, Ph.D. Editor of the Series. A record of minute observations through repeated tests of the abilities, natural and acquired of young dancing mice. He tests their discrimination, the efficiency of different training methods, studies the duration of their habits, memory and power to relearn, the sex and individual differences and the extent to which heredity is show in their behavior. Published in New York 1907. 290 pages. $1.25 net. THE ANIMAL MIND A Text-Book of Comparative Psychology. By Margaret Floy Washburn, Ph.D., Vassar College. "It is the first objective account founded upon years of patient research and upon discerning and expert analyses. For the first time it renders accessible to the studious lajmien in readable form the kind of data, the critical interpretation of results, and the source of the guiding principles that in the modern view are likely to bring some systematic understand- ing of animal psychology." Condensed from a long review in The Dial, Chicago. Published in New York 1908, ^^^ pages. $1.60 net. Other volumes of the series are in preparation Lectures on Human and Animal Psychology By Wilhelm Wundt. Translated from the Second German Edition by J. E. Creighton and E. B. Titchener. In this later edition of an early work Professor Wundt has eliminated everything not pertaining to the individual psychology of man and the animals, and has confined himself to the treatment of topics which he considers especially character- istic of the spirit and trend of modern psychology. The trans- lators have put this work before the English-speaking public with the hope that its comparatively popular and introductory character will render it acceptable to those beginning the study of psychology, and to others who desire some knowledge of the methods and results of the new psychological movement. Published in London, 1894. Fourth edition, 1907. Cloth, 459 pp., 8°, $2.60 net. 'X"" THE MACMILLAN COMPANY T.m JOURNAL OF ANIMAL BEHAVIOR Vol. 1. MARCH-APRIL, 1911. No. 2. THE INFLUENCE OF DIFFERENT COLOR ENVIRON- MENTS ON THE BEHAVIOR OF CERTAIN ARTHROPODS By a. S. PEARSE Contributions from tlie Zoological Laboratory of the University of Michigan, Xo. ijo THREE FIGURES TABLE OF CONTENTS TAGE Introduction 79 Description of Experiments 80 The crayfish 80 The spider crab 84 (a) Black vs. white discrimination 85 (b) Color discrimination 88 The larva of the caddis-fly 90 The crab-spider 92 (a) Color changes 93 (b) Reactions to flowers and colored backgrounds 95 (c) Reactions toward bees and wasps 99 Discussion of experiments 100 The crayfish 100 The caddis-fly larva: 102 The spider-crab 102 The crab-spider 103 General considerations 105 Coloration of arthropods 105 Color discrimination 106 Protective coloration 106 Reactions in relation to color environment 107 Adaptation ' 108 Conclusion 109 Bibliography 109 INTRODUCTION The protective resemblances of various arthropods have long been subjects of interest to naturalists, and a host of observa- tions have been made which bear directly or indirectly on such phenomena. There is no doul^t that many of these resemblances 80 A. S. PEARSE are beneficial to the animals possessing them, and it is quite generally believed at present that "natural selection has un- doubtedl}^ been the chief factor" (Kellogg, '05, p. 613) in de- veloping most of the striking cases that have been recorded. Not only have the colors of numerous animals been shown to correspond very closely to those found in their natural environ- ment, but many species are known to change their colors under certain conditions in a very striking manner, thus causing them to harmonize very closely with the background. Notwithstanding the amount of evidence which has been accumulated along these lines, little attention has been given to the reactions of protectively colored animals with respect to their color environment. The experiments described in this paper were undertaken to determine, if possible, whether the reactions of arthropods to colored backgrounds and colored objects are such as to bring them into the best surroundings; in other words, do the reactions of protectively colored arthro- pods indicate that such animals realize that their coloration is advantageous on certain backgrounds, but not on others. No attempt was made to demonstrate color vision in these ex- periments, i. e., the perception of colors per se. The following animals were studied : A crayfish, Cambartts propinquus Girard ; a caddis fly larva, said by Professor Charles T. Vorhies to belong to the genus Neuronia and probably to the species postica Walker ; a spider crab, Lihinia emarginata Leach; and a spider, Mis- umena aleatoria (Hentz) Emerton. The original experiments with these animals will first be considered and all general ques- tions left for discussion later. DESCRIPTION OF EXPERIMENTS The Crayfish, Cambarus propinquus Girard. — In order to ascertain the effect of subjecting crayfishes to various color environments for long periods of time, twelve individuals were selected which were as nearly alike in color and size as possible (56 to 66 mm. in length); half of these were males and half females. Owing to the fact that two of the females died after' the experiments were begun, only the males were used in testing reactions. A pair of crayfishes was placed in each of the six rectang- ular glass jars used for the experiments. These jars measured COLOR AND THE BEHAVIOR OF ARTHROPODS 81 20 cm. in height, and their other dimensions were 12 cm. and 15 cm. respectively; they were kept about one-fourth full of filtered water, which was changed frequently. Each of the six jars was, except at the top, completely enclosed within a tightly fitting wooden box (called a "color box" in these experiments) which had been painted a particuliar color on the inside. One box was painted black and another white ; the colors used in the others corresponded to the following shades in Klingksiegk et Valette's "code des couleurs":* Red, 8; yellow, 128; green, 303; blue, 376. A cardboard painted like the inside of the box was supported three-quarters of an inch above each in such a way that though ventilation was permitted and light allowed to enter, the color environment of the crayfishes in each jar was all of one shade. Figure i is intended to represent the general plan of a color box. a'pg' Figure 1 — Showing plan of a "color box" containing a glass jar partly full of water. , The crayfishes were placed in the color boxes on December 10, 1908, and kept under observation until January 19, 1909. During this time most of the individuals gradually changed color slightly *Ivlingksiegk et Valette's Code des couleurs will be subsequently indicated by the initials, C. C. 82 A. S. PEARSE SO that they more nearly corresponded to their environment. The dark median stripe down the abdomen was more prominent in those individuals kept in the red and black boxes; those in the blue had a decided bluish tint ; and those in the yellow were noticeably yellowish. Control animals of similar size and age, kept in open glass dishes did not show such variations. These results agree with those of Kent ('oi) who made similar experi- ments on a related species of the same genus of crayfishes. Such color changes are apparently not uncommon among crustaceans and many more striking examples might be cited. After the craA^fishes had been kept for seme time in the color boxes the next step was to ascertain whether such prolonged subjection to a monochrome environment would cause them L WINDOW 3LAC K Figure 2 — Ground plan of apparatus for testing the reactions of the crayfish to colors. A, runway where animals were placed; b, b', c, c', remo^'able colored cardboards. to move more often toward one color than another. This was tested by means of the apparatus shown in Fgure 2. It con- sisted of a rectangular glass dish (10 cm. high, 40 cm. long and 24.5 cm. wide) fitted up so that an animal could be placed at the beginning of the runway, A, and allowed to move down it until it must choose between the two sides b and b'. The only difference between these sides was in the colored screens c, c' and the cardboards b, b' which could be changed at will. The glass dish was filled with filtered water to a depth of four centimeters during the experiments, and the whole apparatus COLOR AND THE BEHAVIOR OF ARTHROPODS 83 was covered above with a flat black screen which was placed nver it immediately after each animal was introduced. The tests consisted in giving each individual ten chances to choose betw'een the color corresponding exactly to the tint by w^hich it had been surrounded for some time (at b and c) and each of the five other colors used for the experiments (at b' and c')- In order to avoid fatigue the different individuals w^ere used in TABLE I Reactions of six male crayfishes to colors after having been since December 10, IN A monochrome ENVIRONMENT December 19 Total No. of reaction 12 3 4 5 6 7 89 10 + — White vs. Blue Red Yellow Green Black 1 1 + + - + + + 1 + 11 + + + + + + — + + — ■f -f ' ' Total 23 27 Blue v^. White Green Yellow- Red Black + 1 1 + 1 1 1 + + + + 1 1 + + + + + + 1 + + 1 1 + + + + + 1 + + + 1 + 1 ++ 1 + 1 1 1 + 1 1 + + 1 1 Total 29 21 Green vs. White Blue Yellow- Red Black 1 1 1 + + + + 1 1 1 1 + 1 + + 1 + + + + 1 1 1 1 + ' + + - + + — + + + + + + + Total 24 26 Yellow vs. White Blue Green Red Black + + + + + + + + + 1 + + 1 LI + 1 1 + + + + + 1 1 1 + 1 1 + 1 ++ 1 + 1 1 + 1 + 1 1 + 1 1 + + 1 + + Total 29 21 Red vs. White Blue Green Yellow Black 1 + + + 1 + + 1 1 1 + + 1 + 1 + 111 + 1 + 1 + 1 + + + 1 1 + 1 + 1 1 1 ++ 1 + 1 1 1 1 1 + 1 1 + 1 Total 22 28 Black vs. White Blue Green Yellow Red — h -f — — + — + + — + — + — -f -f — + — + -t + + + — + — + + + — Total 22 28 Grand Total . . . 149 151 December 31 Total 12 3 4 5 6 7 8 9 10 + — Dead + 1 + 1 1 + + + + + 1 + + + 1 1 + + + 1 1 + + + + 1 1 1 1 + 1 + 1 1 1 1 + + + 1 1 1 1 + + 1 + 1 1 1 26 24 1 + + + + 1 + 1 + + M 1 + + + 1 + 1 + + 1 + 1 1 1 1 ++ 1 + 1 1 + + 1 + 1 1 1 1 + 1 1 + + + 1 1 1 24 26 + — + — + + + + + + + + + + + -f -f + — + + + + + + 25 25 + + + 1 1 + + 1 1 1 1 + + + 1 + + + 1 1 + + + 1 1 1 + 1 + + + 1 1 + + + + 1 + 1 + 1 1 1 1 + 1 1 1 1 25 25 + + 1 + 1 + +I 1 1 1 + 1 + 1 1 1 ++ 1 + + 1 1 1 1 1 1 1 1 1 + 1 + + 1 + + + -f + 1 ++ 1 + + + + + 26 24 126 124 Grand Total 23 27 55 45 48 52 54 46 47 53 48 52 275 275 84 A. S. PEARSE rotation and no more than ten successive reactions were recorded for any of them at one time. Possible errors due to a marked tendency of any individual to turn in a certain direction were avoided by interchanging the screens and cardboards, b, c, and b', c', after every five reactions. In recording reactions, animals which went toward the color corresponding to that of the box in which they had been kept were called " + " ; those in which the crayfishes went toward some other color were " — ". Two series of tests were made, on December 19 and December 31. The crayfishes tested had been, in both cases, in a monochrome environment since December 10. The results of the experiments are shown in Table I ; they indicate no effect due to the pro- longed sojourn of the crayfishes in a particular environment, for there were as many reactions toward other colors as the one by which the crayfishes had been long surrounded. Further- more, Table II, which is based on the same reactions as Table I, shows that there was no striking difference in the number of reactions toward any of the colors used, i. e., no " preference " for any particular color. TABLE II Number of Times Each Color Was Chosen by Six Male Crayfishes White December 19 48 December 31 32* Total 70* 99 92 98 89 92 To summarize, the experiments described show that though the colors of the animals may change to some extent, so that they more nearly resemble the background, the reactions of cray- fishes to colored backgrounds are not influenced by a prolonged sojourn (21 days) in a monochrome environment. The vSpider Crab, Lihinia emarginata Leach. — Lihinia emargi- nata is easily obtained at Woods Hole and the writer *The numbers in this column are smaller than the others because the male which had been in the white box died and, as is indicated in Table I, this animal, therefore could not be tested on December .31. Blue Green Yellow Red Black 58 47 57 41 49 41 45 41 48 43 COLOR AND THE BEHAVIOR OF ARTHROPODS 85 was able to experiment with it while occupying a room in the Marine Biological Laboratory during the summer of 1909. This species is of particular interest on account of its decorating habits. It takes various objects, such as bits of sea weeds, hydroid colonies, or in fact, almost anything that comes in its way, and sticks them on its back in such a way that it is very effectually concealed among the thick growths on the piles and sea bottoms. This crab is especially favorable for testing the question of selection with reference to colored backgrounds and a series of experiments was performed with this point in mind. All the animals used were smaller than the adult size for the species, none of them measuring more than eight centimeters in length. The method employed was to clean the back of a crab with a brush and then put it into a dish filled with sea water; after a short time several pieces of colored papers w^ere added so that a choice was offered between papers colored like the environ- ment and those which were not. The dishes were cylindrical in form, measuring 15.5 centimeters in diameter and seven in depth; each was completely surrounded on the bottom and sides by a monochrome paper. The bits of paper were of uniform size (one by fifteen millimeters) throughout the experiments. The behavior of the crab toward the colored papers was observed from time to time for about twenty-four hours. Under such circumstances, the crabs seldom failed to put some of the papers on their backs, and their interesting decorative maneuvers were often watched by the writer. Two kinds of experiments were tried; (a) th^^se to test black vs. white discrimination, and (b) those in which a variety of colors were involved. (a) Black vs. White Discrimination. Experiment 1 — July 6, 3.30 p. m., a clean Libinia was put on a black background. 3.35 p. M., ten pieces of black paper and ten pieces of white paper were added. July 8, 11.00 a., m., no reaction to papers, experiment discontinued. Experiment 2 — July 8, 11.25 ^^- ^^- Two Libinias were put on a black background and two others on a white background; 11.40, twenty pieces of white and twenty pieces of black paper were added to each dish. 86 A. S. PEARSE Black background 5.25 p. M. Larger individual had two white pieces on its back, smaller one had one white piece. White background 12.20 p. M. The smaller in- dividual bore four half white papers and three black; the larger one carried white paper. 5.25 p. M. Large one had no decorations; small one as at 12.20. 5.45 p. M. Water changed, both animals cleaned and all papers throAvn away; animals interchanged from black to white back- grounds and vice versa ; ten fresh strips each were put in of white and black papers. Black background 8.30 p. M. Large one had two black papers on its back. July 9 7.20 A. M. Large individual same as last night ; small one had half a black paper on top of head. 10.30. Changed water 1 1. 15. Large one same, small one had one white and one black paper on it. 1.25 p. M. Large one had added a white paper to its two black. White background July 9 7.20 A. M. vSmall animal had one piece of black on top of head. 8.30 A. M. Small one had one black and one white. 10.30. Changed water. 1 1. 1 5. Small one has two white and one black. 1.25 p. M. Small one had two black and two white; large one nothing. Experiment 3 — July 9, 2.00 p. m. Two clean Libinias were placed on a black background and two on white ; put ten pieces of black and ten pieces of white paper in each dish. Black background 3.00 p. m. Smaller individ- ual had one black paper on back. White background 2.25. Larger individual had one black piece on head ; small one nothing. COLOR AND THE BEHAVIOR OF ARTHROPODS 87 Black background 3.17. Both with one black paper. 4.45. Large one, nothing; small one had one black and one white paper. 7.50. Same. Put in fresh water and took all the papers off the small animal. July 10 8.20 A. M. Small one has one black, large one nothing. 2.00 p. M. Large one, six black; small one, one black. White background 4.45. Large one with two black and one white; small one, nothing. 7.50. Same. Fresh water added and all the papers re- moved from the large animal. July 10 8.20 A. M. Small individual with two white and one black ; large one nothing. 2.00 p. M. Large one, noth- ing; small one with four white and one black. ■ Experiment 4 — July 10, 2.07 p. m. Four clean Libinias were placed on white and four on black background; ten white and ten black papers were added to each dish. Animals in each dish numbered i, 2, 3, and 4. Black background 3.45. I, 2 and 3 nothing; 4 carried one white. 4.30-8.00. Same. 8.15. I, 2 and 3, nothing; 4 had two white, one black. White background 3.45. I and 2 nothing; 3 bore one white; 4 bore two black and one white. 4.30-8.00. Same. 8.15. I and 2, nothing; 3 bore two white ; 4, two black. Experiment 5 — July 11, 8.30 a. m. Same conditions as last experiment, except that the four Libinias which had been on the black and white backgrounds respectively, were interchanged. Black background 8.05 p. M. Nothing on any of the animals. Fresh water added. White backgwiind 8.05 p. M. Nothing on any of the animals. Fresh water added. 88 A. S. PEARSE Black background July 12 6.45 A. M. Nothing on any. 10.45. 0^^6 black paper on one individual; the others, nothing. 2.15. Same. White background July 12 6.45 A. M. Nothing on any of the animals. 10.45 A- M.-2.15 p. M. Same. Five other experiments of the same nature were performed and they are summarized in Table III. The results show little evidence of discrimination between white and black. There is a rather noticeable predominance of selections of black on a black background but this is due mostly to the selection of six black papers by one individual in Experiment 3. The matter seemed worthy of further investigation, however, and another set of experiments was carried out; these are described in the next section. TABLE III Showing the Results of Allowing Libinl\s to Decorate with Black and White Papers While Resting on an Entirely Black or White Background Background . . . Papers selected Experiment 1. Experiment 2. Experiment 3. Experiment 4. Experiment 5. Experiment 6. Experiment 7. Experiment 8. Experiment 9. Experiment 10 Total Black White Black 0 3 9 1 1 0 3 0 3 0 20 White Black 0 2 1 2 0 4 0 0 0 0 5 3 2 0 0 0 0 0 0 10 White 5 5 3 0 0 0 0 0 0 13 (b) Color Discrimination. — The experiments to be described here were carried out in the same manner as those considered under black vs. white discrimination (p. 85), except that papers of four colors were put in the dishes, and six dishes were used instead of two, four of the dishes being covered with colored papers. The colors used corresponded with the following numbers COLOR AND THE BEHAVIOR OF ARTHROPODS 89 in Klingksiegk and Valette's color code: Red, 7; yellow, 201; green, 306; blue, 426. During each experiment six animals were placed in six dishes, having black, red, yellow, green, blue and white backgrounds respectively; each crab was given a choice of six colors of paper for decorating purposes, these were equally divided among sixty pieces. The series of experiments extended from July 18 until August 4. Table IV shows that there was again no evidence that Libinia has any ability to select colors which correspond to the background on which it rests. From the experiments described it will be seen that Libinia showed no ability to discriminate colors. Professor S. O. Mast has carried out similar experiments and reached same con- clusion. Furthermore, the late Millet Thompson of Clark TABLE IV Showing the Number of Colored Papers Selected by Libinia on Variously Colored Backgrounds. B, Black; R, Red; Y, Yellow; G, Green; U, Blue; W, White Color of background Black Red Yellow Green Blue White Experiment 1 0 lU 0 lU 2B, IR, lY, IW 0 Experiment 2 . IR, IB, IW 4B, IG 2Y, 2R, 3U lY lU 2B, 2Y, 2G, 2R, IW ExDeriment 3 lU, IW IR, 2U 3R, 3Y, 3W, 2U lY, IR 0 lY IG i X , J.VJ, 2U, 2W Experiment 4 lU 3R, 3U, 4Y 2R, IW 0 2U, IR IG lU Experiment 5 0 0 IR, IG, lY, 2U, IW IR, lY, 3G 0 2B, 3G, 2Y, 2U, 2W Experiment 6 lU 0 0 IR, 2Y, 2G lU 2B, lY IW Experiment 7 IB, 4Y, IG, 3U, IW IG 0 2Y, lU, IW 2G, IW 3Y, lU, IR Total 2B, IR, 4Y, IG, 5U, 3W 4B, 4R, 4Y, 2G, 5U 6Y, 8R, IG, 7U, 5W 7Y, 3R, 5G, 2U, IW 2B, 2R, lY, 2G, 3U, 2W 4B, 3R, 9Y, 6G, 6U, 7W 90 A. S. PEARSE University, is said to have performed a series of experiments in which colored bits of hydroid and bryozoan colonies were used; he was unable to show that Libinia chose decorations similar to the environment. The Larva of the Caddis Fly, Neiironia postica Walker. — The case of this larva is built of slender bits of leaves finnly bound together to form a brown cylindrical tube. This is a familiar object to one who collects from the brooks about Ann nes GTI E£ /V Ti£D Q-REEM B Figure 3 — Plan of apparatus for testing the reactions of caddis fly larvae, A, section; B, ground plan. Arbor, for the species is abundant. The dull brown of the case harmonizes well with the muddy plant covered bottoms; hence it is usually inconspicuous. A series of experiments was performed to ascertain w^hether Neuronia larvae could be induced to select materials for the construction of their cases that would match color of the back- ground on which they rested. The color boxes described in considering the reactions of the crayfish (p. 8i, Fig. i) were used. On October 24, 1909, three larvae without cases were placed COLOR AND THE BEHAVIOR OF ARTHROPODS 91 in each of the six color boxes and left until November 5 ; on the latter date those in the green and yellow boxes were all found to be dead and only one was alive in each of the others (black, red, blue, white). On November 7, one fresh larva was put in both the green and yellow boxes ; at the same time twenty- four strips of colored paper, measuring one by fifteen millimeters, were placed in each of the four other boxes; these papers were divided equally among the six colors used. On November 14, twenty-four papers were also added to the yellow and green boxes. On November 14, 20 and 28 all cases which had been built were removed and the larvae were given a new start with twenty-four fresh papers. The colors of the papers from which these larvae made cases are shown in Table V. No evidence of selection of papers colored to match the background is indicated by the results. Even though the larvae did not select papers for their tubes which matched the background, it seemed possible that the TABLE V Results of Allowing Neuronia Larvae to Build Cases from Colored Papers While Resting on a Monochrome Background. W, White; U, Blue; G, Green; Y, Yellow; R, Red; B, Black Color of background Colors of papers composing cases Total Nov. 14 Nov. 20 Nov. 28 Black IW, 2U, 2R No case Dead, no case IW, 2U, 2R Red No case 3W, 3U, 2G, IR, 2B Dead, no case 3W, 2U, 2G IR, 2B Yellow % IW, 3U, IG, 2Y, 2R, 2B IW, 2U, 2B 2W. 5U, IG, 2Y, 2R, 4B Green IW. 3U, 2Y, IR, IB No case IW, 3U, 2Y, IR, IB Blue 2U, 3R. IB IW, 3U, 2R, IB 3U, 2R, IB IW, 6U, 4R, 2B White IW, 2U, IR, IB * No tube IW, 2U, IR, IB *.Not in a definite tube but most of the papers in the box were fastened together. 92 A. S. PEARSE prolonged sojourn in a monochrome environment might influ- ence their locomotor reactions; an experiment was therefore conducted to test this point. On December 5, two cases bearing larvae which had been in red and green color boxes, seven and twentv-eight days respectively, were tested in the apparatus shown in Figure 3. A rectangular glass dish containing water was placed in a box painted half green and half red. They were placed separately in the center of the dish and allowed to move toward either end. The larva from the red color box went to the red end of the dish six times and to the green end four times ; the one from green box gave three reactions toward the red and tw^o toward the green. No striking tendency to go toward either color was shown. From these experiments the conclusion may be drawn that the Neuronia larva selects the objects for its case without refer- ence to their color. However, it will nevertheless generally be protectively colored. The Spider, Misumena aleatoria (Hentz) Emerton *. — This crab-spider is common in the flowers of the fields about Ann Arbor and is readily collected with a beating net. Its particular interest for this paper lies in the fact that it has two striking color varieties, one white and one yellow^, Emerton ('02) says: " Whether spiders prefer flowers like themselves is an unsettled question; at any rate, Misumenas of all colors and both sexes have been found on white flowers." Naturally, the first question to be answered is whether there are more yellow Misumenas on yellow than on white flowers and vice versa. The results of collections made during the months of August and September in 1909 and 1910 are shown in Table VI. Some of the spiders collected could not be classed as white or yellow and two other varieties were made to include this compara- tively small number, i. e., green, for those with a greenish tint, and red, for those in which the abdomen was nearly covered with reddish brown blotches. The flowers from which collections were made were as follows; Yellow — golden rod (Solidago sp. ?) also a few from sunflowers {Helianthus sp. ?) and " butter and eggs " * -All the spiders used in these experiments were not accurately determined to belong to this species; all the adult females, however, agreed with Emerton's ('0-4) description of M. aleatoria; the smaller males are difficult to identify and some of them may have belonged to other species of the genus Misumena. COLOR AND THE BEHAVIOR OF ARTHROPODS 93 TABLE VI Showing the Total Number and the Colors of Spiders Collected with a Beating Net from White and Yellow Flowers Color of flower ^Vhite Yftllnw Color of spiders White Yellow Green Red White Yellow Green Red Number of spiders col- lected 215 14 14 12 46 42.3 15 12 Per cent, of total number 84 6 6 4 9 85 3 3 {Linaria vulgaris) ; white— boneset (Eupatorimn perfoliatum) , domestic buckwheat (Fagopyruni esculentum), and a few from milfoil (Achillea Millefolium). Table VI shows clearly that a majority of white or yellow spiders are found on the corres- pondingly colored flowers.* Having determined that Misumenas usually correspond in color to the flowers in which they lurk to capture their prey, attention was next directed to the causes determining this corres- pondence. In this connection three possible explanations were tested to some extent: (i) color changes, (2) positive reactions toward certain colors or flowers, (3) the elimination of individuals not matching their color backgrounds by predaceous enemies. (a) Color changes. Efforts were made to induce Misumenas to change from white to yellowy or from yellow to white, by keeping them on white or yellow backgrounds. Experiment 1. Spiders were placed in covered glass dishes surrounded on all sides but the top by colored paper. The colors were the same as those used in the experiments with Libinia (p. 89). On August 20, 1909, one white spider was * Although collections were not very extensive from flowers other than yellow and Avhite, the following results show that the majority of spiders were white. Color of spider White Yellow Green Red Epilohium angustijolium — purple 10 0 0 0 Eupatorium purpureum — purple 30 3 1 0 Polygonium sp.? — pink 1 0 1 0 Aster laevis — blue 10 2 0 0 Erigeron annuus — yellow and white 4 1 6 0 Total 55 5 8 0 Per cent 81 7 12 o 94 A. S. PEARSE placed in each of the six dishes (black, red, yellow, green, blue and white respectively) ; on August 2 3 two more white spiders were added to each dish. Daily notes were then made as to the condition of the eighteen spiders under observation until Septem- ber 6, a period of fourteen days (during which time eight indi- viduals died). Although four individuals changed a little and looked as though they might be about to take on a slightly different tint, none of the animals assumed a color that could be called anything but white. Experiment 2. September 3, 1909, five white spiders were placed on a bunch of goldenrod flow^ers {Solidago sp. ?), yellow, and five yellow spiders on a bunch of milfoil {Achillea Millefolium) , white. The flowers were in bottles which rested in large pans of water so that the spiders could not escape. Both dishes were placed before a window so that they were illuminated by the morning sun. The spiders were observed daily until all had been drowned. On the milfoil three were drowned after one day, one after four days, and the last after fifteen days ; on the yellow^ flow^ers, three lived four days, one eight, and one thirteen. None of these spiders showed any color changes. Experiment 3. For this experiment two of the color boxes described on page 81 were used (Fig. i). On August 31, 1909, seven yellow spiders were placed in a white box and a like number of the same color in a yellow box. All but one of these remained alive for more than seventeen days ; none of them changed color. On September 6, 1909, six white Misumenas were placed in a w^hite box, and a corresponding number of the same color in a yellow box. These were all alive on September 17 (10 days); on September 24, onh^ one was alive in the white box and three in the yellow; on October 7, there was no further mortality in the white box but all those in the yellow had died. None of these spiders changed color. Experiment 4. IVIethods like Experiment i. Up to November 27 the glass jars containing the spiders rested on a shelf in a laboratory room ; after that they were on a table before a window. September 19, 1910, six yellow^ males and two yellow females were put in two "white jars." Two of the males turned white between October 8 and 21, all the others died before that time except one male (who remained yellowish until he escaped on November 10). On November 24, the two surviving (then white) COLOR AND THE BEHAVIOR OF ARTHROPODS 95 males were placed in a yellow jar; on December i8, one died and the other was still white on December 31. September 22, 1910. Twelve white males and four white females were placed in four yellow jars. None of these Misumenas changed color, though five of the males were alive December 22. Experiment 5 was carried out in order to ascertain if light per se has any effect on the colors of Misumena. On September 13, 1 910, eight spiders were placed in four glass jars. Two of these jars contained two yellow spiders each, and the other two contained the same number of white individuals. The jars were so placed that half the spiders of each color were in the dark, while the others were exposed to direct sunlight before a window. No change could be observed to have occurred in the colors of any of the spiders on September 24. On October 8, all the white spiders died, also one of the yellow individuals. Light or its absence induced no observable color change during twenty-two da vs. From the five experiments described it is apparent that the few color changes which took place were too slow to be of much advantage to a spider in nature. Most of the Misumenas did not change color and those that did would hardly have had tiriie to adapt themselves to a particular flower before it withered. (b) The Reactions of Misumena to flowers and colored hack- grounds. As a preliminary step the reactions of Misumena to white light were tested by means of a horizontal beam from a one-glow^er Nernst lamp. During these experiments the lamp was placed at one end of a table and the spiders were allowed to run from a vial on to the table at a distance of forty centi- meters from the source of light. Five reactions for each of five individuals were taken. None of the spiders moved directly toward or away from the light but in every case they went ahead in a rather erratic manner and climbed up, or ran along the edge of the black side screens which were used to cut off their view from objects in the room where the experiments WTre performed. From these results it was assumed that ]\Iisumena's reactions to directive light were negligible so far as their influence on the reaction experiments which followed \^'ere concerned. This con- clusion was supported by the general behavior of spiders in the field and laboratory, no indications of marked reactions to light per se were ever noted. 96 A. S. PEARSE Experiment i. On August 21, 190Q, twenty-four white and twenty-six yellow Misumenas were collected and placed together in a large rectangular glass dish forty centimeters long, twenty- five centimeters wide and twelve centimeters deep. Except for the cover, one-half of this dish was completely enclosed in yellow, the other half in white. The fifty individuals were ex- amined two or more times a day until August 26. There appeared to be no tendency for yellow or white indi- viduals to come to rest more often on one color than on the other. There were always about as many white as yellow individuals in either end of the dish. This experiment then gave no evi- dence that Misumena seeks a background which matches its color. Experiment 2. In this experiment spiders were placed in a vial and allowed to go from it through a small hole in the side of a box (measuring 7 cm. high, 15 cm. long, and 11 cm. wide). This box was lined half with yellow paper (c. c. 176) and half with white paper in such a way that it \\'as divided vertically by the two colors at right angles to the long axis. The spider was admitted on the floor of the box and on the line of division between the white and yellow; its subsequent move- TABLE VII Showing the Direction of Movement T.\ken by Spiders in a Box Colored Half Yellow and Half White. "0" Indicates no Movement, or that the Movement Could not be Said to be in the Direction of Either Color. Color of spider Yellow White Direction of movement.. . Yellow White • 0 Yellow White 0 Number of reactions 46 49 5 10 11 4 ments were ob£er\'ed and recorded ; after that it was removed and another individual was tested. The results of these tests are given in Table VII ; they show no evidence of selection of color -environment to correspond with the color of the spiders. In fact the different individuals appeared to wander off in any direction after entering the box. COLOR AND THE BEHAVIOR OF ARTHROPODS 97 The Misumenas used in the experiments just described were individuals which had been kept in the laboratory for a day or two and it was thought that animals freshly collected might show different results if tested at once in the field. Accordingly, the "yellow -white" box was carried out on several collecting TABLE VIII Showing the Direction of Movement Taken by Newly Collected Spiders Tested in the Field in a Box Colored Half White and Half Yellow. "0" Indicates no Definite Movement with Respect to Either Color. Color of spider Yellow White Direction of movement.. . Yellow White 0 Yellow White 0 Number of reactions 35 31 11 14 8 3 trips and the spiders were tested as they were taken from the net. The results of these tests are given in Table VIII. Again there is absolutely no evidence that Misumena shows a positive reaction toward a background colored like itself. Experiment 3. This experiment was carried out on a table before an open window. Misumenas were allowed to crawl separately from a vial to the surface of the table at a point midway between two black screens ; the vial was then removed. The screens were inclined at an angle of about 7 5 degrees to the surface of the table and a cluster of flowers was fastened to each in such a way that it was exactly 9.4 centimeters from the spider. The two flower clusters were as nearly the same size as possible ; one was goldenrod {Solidago sp?) like that from which the yellow spiders had been collected, the other was milfoil {Achillea Mille- folium) from which some of the white spiders had been secured. Each spider as it reached the surface of the table was, there- fore, the same distance from a white and a yellow flower cluster. A total of fifty tests were recorded from five yellow Misumenas, ten from each individual. In four cases spiders went to the white flower, in five to the yellow flower; the other forty-one reactions were without apparent reference to the flowers. Several times spiders walked up one of the screens and passed within a centimeter of a flower without swerving. Experiment 4. On September 15, 1910, thirty-two freshly collected spiders, sixteen yellow and sixteen white, were tested 98 A. S. PEARSE in the field. An elongated space twenty-eight centimeters wide was cleared in the shadow of a tree trunk. On each side of this space at its middle a row" (13 cm. high and 30 cm. long) of flowers was placed, the flowers on the right being yellow {Solidago sp?), those on the left, white (Eupatorium perfnliatum) . The observer sat against the tree at one end of the cleared space, and there- fore his movements did not cause the spiders to move toward one bunch of flowers or the other. The spiders were allowed to hang from a thread, and were then placed half way between the two rows of blossoms. Their reactions are summarized in Table IX ; there is no evidence that the spiders went oftener toward flowers colored like themselves. TABLE IX Reactions of Yellow and White Spiders to Colored Flowers in the Field Color of spider Yellow White Color of flower Yellow White Yellow Wliite No movement toward flowers Number of reactions 7 9 6 7 3 Experiment 5. The behavior of Misumenas placed on yellow and white flowers was observed with some care in the field. It was thought that white spiders might be less active on white than on yellow flowers, and that yellow spiders might show a similar response on yellow flowers; i. e. that there might be some evidence that yellow or white spiders were less restless when surrounded by a background colored like themselves. On September 16 and 17, 1910, forty Misumenas were placed on goldenrod and buckwheat; half these were yellow and half white, and equal numbers of each color were placed on each kind of flower. Furthermore, every flower used was a part of a large field of the same kind. The behavior of each spider was carefully observed for two hours or until it had moved beyond the writer's field of observation. Both the days chosen for these experiments were clear with bright sunshine. The behavior of the spiders varied greatly; some individuals at once hid themselves beneath a spray of the flower, others chose a conspicuous place in an exposed situation; some re- COLOR AND THE BEHAVIOR OF ARTHROPODS 99 mained almost where they were placed for a long time, others at once spread gauzy aeroplanes and ballooned away to new fields ; no two spiders did similar things and it is impossible to tabulate the results. The writer quotes the following from his field note-book: "After two days I can see no difference in the way yellow spiders behave on white and yellow flowers; white spiders same." No difference that could be assigned to the influence of color environment was observed. (c) The Reactions toward bees and wasps. In order to ascer- tain whether ihe behavior of Misumena toward colors was such that most individuals would escape in the presence of predaceous insect enemies, some observations bearing on this question were made. On two occasions, two or three hours were spent in watching the behavior of Misumenas placed on flowers which were being frequently visited by bees and wasps. On September 14, 1910, three yellow individuals were placed on a bunch of fleabane daisies {Erigeron annuus), which was being prospected con- stantly by from fifty to seventy-five bees and wasps. The honey bee, Apis mellifica, was the most frequent visitor, and among the wasps, the commoner representatives were two sfjecies identified as Polistes pallipes and Philanthus solivagus by Mr. S. A. Rohwer, to whom they were referred by Dr. L. O. Howard. One of the Misumenas at once hid itself deep in a cluster of flowers and was not seen again during the two hours the observation lasted. One of the other spiders hung on the under side of a small flower cluster, and the third chose a position in plain sight on top of one of the highest sprigs of the fleabane. The individual which was hanging on the under side of the flowers avoided bees and wasps; once it moved aw^ay when a bee approached, but it usually remained perfectly motionless and concealed itself as much as possible when a winged disturber came near. On the other hand, the Misumena which chose the conspicuous situation behaved in quite a different manner; it rested with outstretched legs ready to attack; when one of the largest wasps {Philanthus solivagus) alighted near, it rushed toward the intruder with raised legs, and the wasp at once went elsewhere to forage. On September 15, 1910, five yellow spiders were put on boneset blossoms and watched from 2.30 p. m. until 5.00 p. m. One dropped to the ground at once, one wandered a good deal from 100 A. S. PEARSE one place to another, a third hid itself on the under side of a flower, the remaining two took positions on the tops of flowers in plain view and assumed a w^atchful attitude, apparently seeking prey. Neither of the two latter individuals seemed to be disturbed by the close proximity of bees and seldom changed their positions when approached. As a result of these and other similar observations made in the course of different experiments the writer was convinced that the behavior of Misumena is not finely adapted to enable them to escape predaceous hymenopterous insects. The reactions of different individuals apparently depend upon their physiological state. Some (hungry ?) spiders are pugnacious and ready to attack almost anything that approaches, other are secretive and remain in hiding. DISCUSSION OF EXPERIMENTS For the sake of conciseness the relation of the writer's experi- ments to the literature concerning similar investigations has been reserved for discussion at this place. This plan has the additional advantage of bringing all the conclusions together before we pass to the "general considerations" following. The arthropods used for the foregoing experiments will be considered separately. The crayfish. — Protective resemblance is common among crustaceans. Beebe ('09) points out a very striking case in the mangrove crabs on Trinidad Island, where a certain species shows a great variety of colors which correspond closely to the roots it frequents. Many crustaceans have been shown to undergo marked color changes which bring about a general correspon- dence with the colors in their environment, and careful studies have been made of these changes in certain Decapods and Mysidaceans by Keeble and Gamble ('00, '04), and in the crayfish by Kent ('01). More recently Franz ('10) has investigated the chromatophores of Pandalus and Crago. All of these investi- gators agree that the color of the background is an important factor in inducing color changes. There is evidence that color changes in the skin may be in- fluenced by stimuli acting through the eye and central nervous system (Frohlich, '10; Keeble and Gamble, '04), nevertheless, COLOR AND THE BEHAVIOR OF ARTHROPODS 101 color changes are without doubt commonly brought about by the direct effect of light on the chroma to phores which contain pigment (Keeble and Gamble, '04). Rynberk ('06) in his ex- cellent summary of the whole question says (p. 427) that such changes are not voluntary, and that they are induced not through seeing so much as by the changes brought about through nour- ishment. The experiments described in the present paper make it apparent that the reactions of the crayfish are not influenced by a prolonged sojourn in a monochrome environment even though there is a corresponding change in the color of the skin; i. e. crayfishes show no tendency to go toward the color which most nearly resembles their own. Keeble and Gamble ('00) make a statement concerning another crustacean, Hippolyte varians, which apparently does not harmonize with this con- clusion— they say (p. 601) " That the prawns exert powders of selection with respect to their weed, this will be readily realized from Pis. 32 and 2iZ^ ^Rs. i to 9, representing praw^ns placed in a dish with sea water, to which subsequently pieces of different coloured weeds were added. The prawns were left free to select their weeds, and, as will be seen in the figures, they succeeded in making wonderfully accurate color matches." Notwithstand- ing the striking similarities they present in their figures, Keeble and Gamble give no evidence to show that the prawns selected particular weeds on account of their color, and the selection may have been due, wholly or in part, to some other factor, such as food or a particular sort of tactile or chemical stimulation to which the prawns had been accustomed. The following quotations from Keeble and Gamble's paper support this view (p. 621) : " Its prime object in life is to anchor itself. Once fixed, rather than release its hold it will allow" the ebb tide to leave it stranded. By its immobility it has grown into its surroundings and become colored like them. Should it become separated from its favorite weed its movements become of an aimless sort." Hippolyte evidently becomes accustomed to a certain seaw^eed; it seems but natural that if it were separated from this and placed in a dish containing various plants it would choose the one to which it had become accustomed ; and further- more, it seems to the writer that such selection could not be assumed to be due to color alone unless it were shown that the 102 A. S. PEARSE prawns selected a particular color, without the presence of a particular weed. Until it has been proven that such is the case, we have not sufficient evidence, I believe, to permit the asser- tion that any crustacean selects an environment to suit its own color. The caddis-fly larva. — Poulton ('90, p. 77) says: "The well known cases of caddice-worms (Trichoptera) are partly for con- cealment and partly for defense, they are built of, . . . any suitable objects which are abundant at that bottom of the stream in which they live." He uses the caddis-fly case as an example of "adventitious protection" where "animals cover themselves with objects which are prevalent in their surround- ings and are of no interest to their enemies." The experiments described in this paper show that caddis-fly larvae do not select objects for their cases which will make their colors correspond with the general tint of the background. Nevertheless, they are protectively colored as a rule. The spider-crab. — In 1907 Minkiewicz published an account of the reactions of spider-crabs; in which he stated that he had induced individuals of several genera {Maja, Pisa, Inachus, Stenorynchus) to select certain strips of paper, from a variety of colors, which corresponded to the background on which they rested. Furthermore, Minkiewicz maintained that crabs which had selected decorations of a certain color showed a positive chromotropism toward the same color when they were placed in a particolored dish. These results seemed remarkable for Bateson ('89) had previously performed similar experiments with three species from the same genera used by Minkiewicz and reached quite different conclusions. He says (p. 214) : "There is certainly no disposition on the part of Stenorynchus dressed in any color, say green, to take up a position amongst green weed or indeed amongst weed at all, and so on, while some indi- viduals which have taken up their station among w^eeds do not dress themselves at all." Poulton ('90) also, quoting Bateson, says: "Stenorynchus does not betray any disposition to remain in an environment which harmonizes with its dress." The writer's experiments on Libinia support Bateson 's con- clusions. Before Minkiewicz 's results are accepted the species COLOR AND THE BEHAVIOR OF ARTHROPODS 103 he studied ought to be re-examined by another investigator. The evidence as it now stands cannot be said to prove that decorator crabs choose colors for concealing themselves which harmonize with their surroundings. The crab-spider. — Emerton ('02) says: " Whether spiders prefer flowers colored like themselves is an unsettled question; at any rate, Misumenas of all colors and both sexes have been found on white fiow^ers." From the results set forth in the present paper there can be little doubt that the majority of Misumenas are to be found on flowers colored like themselves. Thayer ('09) figures Misumenas as an example of "obliterative coloration," and in experiments described in this paper it is shown that its colors usually harmonize with the background. The question is, whether the presence of a majorit)^ of yellow spiders on yellow flowers and of white spiders on white flowers, is due to color changes in the spider itself, or to the selection of a particular background by each individual, or to some other influence. McCook ('89-'93, ^^^- 2, P- 341) says the color of spiders may be influenced by a variety of factors, among these moulting, ' advancing age, gestation, muscular contraction, sex difference and excitement may be mentioned as being sometimes important. Ne\'ertheless he says (vol. 3, p. 51), that there are no authentic cases of rapid color changes in American spiders, and (vol. 2, p. 271) because the color changes of Misumena are so slow, he says, "we are therefore compelled to the conclusion .... that the spider sought the flower and settled upon it, either accidentally or by choice." Beddard ('92, p. iii) mentions a rapid color change described by Heckel, in a spider which belongs to the same family as Misumena. This species was Thomisus onustus, which he says has three color varieties in the flowers of Con- * volvulus amensis, and two other colors in other flowers. These varieties correspond closely to the flowers and Heckel maintained that these spiders could change their colors in three or four days. ■ Beddard says, however, that the evidence for color change was not by any means conclusive. Davenport ('03) mentions a light colored sand-spider which became gradually darker when placed on grass. In the knowledge of the writer, the experiments described in the present paper are the first in which spiders have been allowed 104 A. S. PEARSE to remain for a long period of time on a background which might induce color changes. Out of the sixty spiders used only two showed any color changes and in both these cases more than a month elapsed before the skin had turned from yellow to white. Furthermore, McCook ('8g-'93, vol. 2, p. 325) demonstrated that spiders of a single species may show striking color variations in the same habitat. From these facts the conclusion is warranted, that Misumena does not change its colors rapidly nor with enough uniformity to make such changes of importance. This is apparently what McCook believed to be the case. If, however, we maintain that the color changes of Misumena are unimportant in relation to protective coloration, we must examine other alternatives with all the more care. McCook says: "We are compelled to the conclusion" that Misumena "sought the flower and settled upon it, either accidentally or by choice." Despite his cautious statement in the sentence quoted, he evidenth^ felt that spiders had some power of color discrim- ination, for he states (vol. 2, p. 367), without unfavorable com- ment, that Mrs. Treat found that Misumena would not stay on a background of a different color from its own, and he also says (p. 335) that the male spider is conscious of the colors of the female. The Peckhams ('87a) had previously shown that spiders showed a preference for certain colors when they w^ere allowed to choose from several on which they might rest, and they state distinctly ('95, p. 261) : " We, ourselves, are of the opinion that all the experiments taken together strongly indicate that spiders have the power of distinguishing colors." They also affirm that certain attids can see small objects distinctly at a distance of at least twelve inches. In the light of these observations it might reasonably be expected that Misumena would show a tendency to seek an environment colored like itself, but this w^as not the case, in the writer's experiments. When yellow and white spiders were given an opportunity to choose between white and yellow papers or flowers they did not do so either in the laboratory or in the field ; nor was any difference in the degree of activity on flowers of either color manifest. We are, therefore, forced to conclude that Misumena neither changes color rapidly to match its surround- ings, nor seeks an environment colored like itself. It is therefore not within the province of this investigation COLOR AND THE BEHAVIOR OF ARTHROPODS 105 to attempt to discover just what factors are responsible for a preponderance of yellow Misumenas on yellow flowers and of white individuals on white flowers. The writer has some evidence that such distribution may be due to the attacks of predaceous enemies, but it is not conclusive, and he rests his case here without attempting to discuss this or other factors. Whatever the cause of the general correspondence between Misumena and the colors in its environment, it is not due to color change nor positive chromotropism. GENERAL CONSIDERATIONS Coloration of Arthropods. Arthropods are sharply separated from all other groups of invertebrate animals; nevertheless, they possess certain common structural similarities which indi- cate a genetic relationship among the members of the different classes. There is great diversity among the different groups of arthropods, not only in structure but also in habits, and, if w^e compare the four chief classes, coloration is not by any means the least variable feature. The colors of Crustacea depend primarily upon chromatophore systems (Keeble and Gamble, 'oo, '04; Frohlich, '10; Franz, '10). These are usually deep seated and there is a migration of pig- ment granules in them to bring about more or less striking color changes which make the animals resemble their environment with varying degrees of accuracy (Beebe, '09; Keeble and Gamble, '00; Kent, '01). In many crustaceans the color phases are strongly periodic, appearing alternatingly with day and night, (Keeble and Gamble, '00), but the color changes are, nevertheless, chiefly induced by the presence or absence of light together with the tint of the background (Keeble and Gamble, '00; Kent, '01; Frohlich, '10; Franz, '10). The color changes of some crustaceans are apparently controlled in part by the nervous system, but there is no doubt that, even in such cases, changes may be brought about by the direct effect of light on the chromophores (Keeble and Gamble, '00; Frohlich, '10). The colors of Myriapods have not been studied, to the knowl- edge of the writer, and they are comparatively uninteresting, for the color of many species appears to be mostly in the chiti- nous exoskeletal covering. Spiders, the most common arachnids, present a great diversity 106 A. S. PEARSE of colors and color patterns. No indubitable cases of rapid color changes have been reported, and McCook, one of the fore- most students of spiders, points out ('89-'93, vol. 2, p. 325) that a single species may show a number of striking color vari- eties in the same habitat; the coloration is apparently not accurately adjusted to a particular background. The insects show striking adaptation to aerial life, and also to a great diversity of habitats. Numerous colors and color patterns have been developed along rather definite lines (Mayer, '97; Tower, '03). A host of insects show protective resemblance and a few have been observed to undergo slow changes w^hich make them more nearly resemble their surroundings (Poulton, '88; Davenport, '03). Kellogg ('05) in his work on American insects says (p. 600) the colors of insects are "fixed by the time they reach the adult stage," but a striking diurnal color change has recently (Schleip, '10) been demonstrated in Dixippus morosus. In this case there is a diurnal migration of pigment granules in a single layer of syncitial hypodermal cells. The chief factor which brings about this migration is the presence or absence of light, but the changes have a strongly developed diurnal periodicity and continue for as much as seventy-eight days in the dark. Color discrimination.* — Bateson ('89) and Merejowski ('81) maintained that there is no color discrimination * manifested by the reactions of Crustacea but Minkiewicz ('07) takes an opposite view. On account of the results of the experiments described in this paper (p. 88) the w^riter is disposed to agree with the results of the earlier investigators. The Peckhams ('87, '87a) firmly believe that spiders can discriminate colors, and Lubbock ('79), Lovell ('10) and Turner ('10) are of the same opinion in regard to hymenopterous insects. However, the whole question of color discrimination will bear further investigation. The present evidence is fragmentary, and some of it by no means conclusive. Protective coloration. — Di Cesnola ('04) has demonstrated that protective coloration may preserve an insect from the attacks of its enemies, and there is little reason to doubt that protective * Color "discrimination" is not intended to assume that arthropods see color as we see it, but only that they may be able to recognize a difference between colors or intensities of color. COLOR AND THE BEHAVIOR OF ARTHROPODS 107 resemblance is of value to most arthropods. However, Keeble and Gamble, ('04, p. 363) conclude from their exhaustive study of the coloration and color changes of crustaceans, that " The phenomena presented by these pigments are not exhaustively explained by any 'protective hypothesis,' " and Beddard ('92) maintains that color is not a protection against invertebrate foes. Although protective coloration is generally efficacious in preserving arthropods from the attacks of enemies it is not always perfectly adapted to its purpose (p. 99) and is often only effective for one particular enemy. Reactions in relation to color environment. — In the present paper it has been shown that none of the four animals tested, though protectively colored, show any tendency to seek the background that harmonizes with their own coloration. In fact, the writer knows of no ]:)ublished observation which proves that any arthropod does this. Minkiewicz ('07) maintains that Maja and other decorating crabs select colors for their backs which correspond with the tone of their surroundings, but Bate- son's ('89) experiments on the same kinds of crabs and the experiments with Lihinia described in this paper make his results seem doubtful; Keeble and Gamble ('00) believe that in their experiments Hippolyte selected the background which most nearly matched its own color, but they give no evidence to show that the prawns did not select a certain sea weed on account of some quality other than color. Such a careful observer as McCook ('89-'93, vol. 2, p. 335) concludes that spiders which conceal their nests with foreign objects do so without recog- nizing their protective value. In this connection it is interesting to note that Marshall and Poulton ('02, p. 323) say: " Insectiv- orous invertebrates are not capable of appreciating warning colors, but have to taste all their captures." Nevertheless, they believe (p. 424) that butterflies select a general habitat where they are well protected. Among the insects perhaps the best illustration of protective behavior coupled with absolute disregard for color environment is exhibited by the walking-sticks. Both Stockard ('08) and Schleip ('10) have shown that the behavior of these animals is suited in the highest degree to protect them except for the fact that they do not rest upon colors in their habitat which match 108 A. S. PEARSE their own. A walking-stick will maintain a difficxilt attitude for a long time rather than disclose its presence by the slightest movement, thus indicating by its reactions that it has some recognition (not necessarily recognition on the part of the indi- vidual, but recognition at least so far as the race is concerned) that it is protected, but it takes no cognizance of color. Further- more, Schleip ('lo) has shown that the color changes of the walking-stick have no relation to the color of the environment, but are induced chiefly by light. Adaptation. — Thayer ('09) would have us believe that all animal coloration is protective (concealing) ; not necessarily at every moment of an animal's life, perhaps only at some infre- quent moment of great need. His arguments are very convincing and many of his conclusions seem quite probable. If the Misu- menas described in this paper are used as an illustration, we can readily imagine that the presence of two color varieties which resemble the commonest flowers might be a valuable asset in the struggle for existence. Yellow or white flowers grow together in great fields and a spider would often find a suitable color background if it were in the proper habitat. It is possible that more yellow spiders are hatched year after year in large patches of goldenrod, and that white spiders are cor- respondingly more abundant where boneset abounds, but we can only surmise this, for nothing is known of the heredity of color in spiders, nor how much they wander from field to field. At any rate, if natural selection, having only two choices, picked out yellow and white to match the greatest number of flowers, it could not have chosen two colors that would be better for the locations where Misumena abounds. We have many striking instances of extremely refined pro- tective resemblances among the arthropods; examples like Kallima, Misumena and the walking-stick are familiar to every naturalist. Beddard ('92) mentions a spider which was so like a mass of bird excreta that it deceived the eye of a trained observer. Beebe ('09) in speaking of the mangrove crab says, "he grew to resemble his home root," and dwells at some length on the variety of mangrove roots and the accuracy with which the crabs imitate the patterns they present. Examples of this kind might be multiplied. In speaking of insects, Kellogg ('05, p. 613) says, "natural COLOR AND THE BEHAVIOR OF ARTHROPODS 100 selection has undoubtedly been the chief factor" in producing protective resemblances, and, though there are some cases apparently not readily explained in this way (e. g. the bright colors of some deep sea Crustacea, etc.), this statement seems to be generally applicable to arthropods as a whole. Further- more, Packard ('04) believes that the patterns and color mark- ings of arthropods have arisen through the operation of physical rather than biological factors. The evidence from the experi- ments described in this paper supports this view, for none of the animals appears to be able to take advantage of the colors in its environment in efforts to conceal itself. The coloration of arthropods shows various degrees of adapta- tion to the factors in the environment. In one animal a certain factor may be of chief importance in causing color changes and in another animal the same factor may have little influence. For examiple, the color of the background is most potent in changing the colors of the crustacean, Hippolyte varians (Keeble and Gamble, '00), but has no effect on those of the insect Dixippus morosus (Schleip, '10). CONCLUSION From the foregoing experiments and discussion the writer believes that it cannot at present be afBrmed that any protec- tively colored arthropod reacts toward colored objects or back- grounds in such a way that it can be said to have even an instinctive knowledge that it is protectively colored; i. e. arthropods do not choose the most favorable color environment on account of color. BIBLIOGRAPHY Bateson, W. Notes on the Senses and Habits of Some Crustacea. Jour. Marine 1889. Biol, Plymouth, N. S., vol. 1, pp. 211-214. Beddard, F. E. Animal Coloration. London, viii+288 pp. 1892. Beebe, C. W. and M. B. a Naturalist in the Tropics. Harper's Mag., vol. 118, 1909. pp. 590-GOO. Davenport, C. B. The Animal Ecology of the Cold Spring Sand Spit, with Remarks 1903. on the Theory of Adaptation. Univ. of Chicago Decen. PubL, S. 1, vol. 10, pp. 157-176. Di Cesnola, a. p. Preliminary Note on the Protective Value of Color in Mantis 1904. religiosa. Biometrika, vol. 3, pp. 58-59. Emerton, J. H. The Common Spiders of the United States. Boston and London. 1902. xviii+225pp. Franz, V Zur Struktur der Chromatophoren bei Crustaceen. Biol. Centralb., 1910. bd. 33, pp. 424-430. Frohlich, a. Farbwechselreaktionen bei Palaemon. Arch. Entw.-Mech., bd. 29, 1910. pp. 432-438, Taf. 13. 110 A. S. PEARSE Keeble, F. W. and Gamble, F. W. Hippobjte varians. a Study in Color-change. 1900. Q. Jour. Micr. Sci., vol. 43, pp. 589-698, pis. 32-36. 1904. The Colour Physiology of Higher Crustacea. Phil. Trans. London. S. B., vol. 196, pp. 295-388, pis. 18-23. Kellogg, V. L. American Insects. Xeiv York, ix -I- 674 pp. 1905. Kent, W. J. The Colors of the Crayfish. Ajiier. Xat. vol. 35, pp. 933-936. 1901. Klingksiegk, p., et Valette, T. 1908. Code des couleurs. Paris. 86 pp. LovELL, J. H. The Color Sense of the Honey Bee. Can liees Distinguish Colors? 1910. Amer. Xat., vol. 44, pp. 673-692. LiTBBOCK, J. Observations on Ants, Bees and Wasps. Pt. 5, Ants. Jour. Linn. 1879. Soc, vol. 14, pp. 265-290. McCooK, H. C. American Spiders and Their Spinning Work. Philadelphia . 3 vols. 1889-1893. Marshall, G. A. K., and Poulton, E. B. Five Years' Observations and Experi- 1902. ments (1896-1901) on the Bionomics of South African Insects, chiefly directed to the Investigation of Mimicry and Warning Colors. Trans. Entoniol. Soc. London, 1902, pp. 287-541, pis. 9-23. Mayer, A. G. On the Color and Color-Patterns of Moths and Butterflies. Proc. 1897. Boston Soc. Xat. Hist., vol. 27, pp. 243-330, pis. 1-10. Mere,iowsky, C. Les crustaces inferieurs distinguent-ils les couleurs? C. r. Acad. 1881. Sci. Paris, T. 93, p. 1160. MiNKiEwicz, R. Analyse experimentale de I'instinct de deguisement chez les 1907. Brachyures oxyrhynques. Arch. Zool. E.xper. (1907), T. 7, notes e. rev., pp. xxvii-Lxvii. Newbegin, M. I. Color in Nature. London, xii -1-344 pp. 1898. Packard, A. S. The Origin of the Markings of Organisms (Poecilogenesis) due 1904. to Physical rather than Biological Environment. Proc. Amer. Phil. Soc, vol. 43, pp. 393-450. Peckham, G. W. and E. G. Some Observations on the Special Senses of Wasps. 1897. Proc. Xat. Hist. Soc. Wisconsin, vol. 1, pp. 91-132. 1897a. Some Observations on the Mental Powers of Spiders. Jour. MorphoL, vol. 1, pp. 383-419. 1895. Sense of Sight in Spiders. Trnns. Wisconsin Acad. Sci., vol. 10, pp 231-261. 1898. On the Instincts and Habits of the Solitary Wasps. Wis. Geol. and Xat. Hist. Surr. Bull. No. 2, 245 pp., 14 pis. Petrunkevitch, a. The Sense of Sight in Spiders. Jour. Exper. Zool., vol. 5; 1907. pp. 275-310, pis. 1-6. Poulton, E. B. An Enquiry into the Cause and Extent of a Certain Colour- 1888. relation between certain Exposed Lepidopterous Pupae and the Surfaces which immediately surround them. Philosoph. Trans. Roy. Soc. London vol. 178B, pp. 311-441, pi. 26. 1890. The Colours of Animals. London, xvi -1-360 pp. Rynberk, G. van. Ueber den durch Chroraatophoren bedingten Farbenwechsel 1906. der Tiere. Ergebn. PhijsioL, bd. 5, pp. 347-571. Schleip, W. Der Farbenwecksel von Dixippus morosus (Phasmidae). Zool. Jahrb., 1910. bd. 30, pp. 45-133, Taf. 1-3. Stockard, C. R. Habits, Reactions and Mating Instincts of the "Walking Stick," 1908. Aplopusniatjeri. Extr. Puh. 103, Carnegielnst . Wash., pp. 43-59, pis. 1-3. Thayer, G. H. Concealing-Coloration in the Animal Kingdom. Xew York, xx-l- 1909. 260 pp. Tower, W. L. Colors and Color-patterns of Coleoptera. Decen. Publ. Univ. 1903. Chicago, vol. 10, pp. 33-70, pis. 1-3. Turner, C. H. Experiments on the Color- Vision of the Honey Bee. Biol. Bull., 1910. vol. 19, pp. 257-279. Wallace, A. R. Darwinism. London, xx -1-494 pp. 1905. THE RELATION OF STRENGTH OF STIMULUS TO RATE OF LEARNING IN THE CHICK By LAWRENCE W. COLE, (The University of Colorado) ] From the Harvard Psychological Laboratory ONE FIGURE The experiments described in this paper were undertaken in order to learn under what strength of stimulus chicks most rapidly learn to make, respectively, an easy, a medium, and a difficult discrimination. Yerkes and Dodson discovered, in the case of the dancing mouse, that when " discrimination is ex- tremely difficult the rapidity of learning at first rapidly increases as the strength of the stimulus is increased from the threshold, but, beyond an intensity of stimulation which is soon reached, it begins to decrease," while when " discrimination is easy, the rapidity of learning increases as the strength of the electrical stimulus is increased from the threshold of stimulation to the point of harmful intensity."' In other words, there appears to be an optimal strength of stimulus for each degree of difficulty of discrimination and the intensity of this optimal stimulus is lesf the more difficult the discrimination which is to be made. It was proposed, then, to test the chick's rate of learning to discriminate by a method simiilar to that which had been em- ployed with the dancing mouse. The work was done in the Harvard Psychological Laboratory and my thanks are due to Professor R. M. Yerkes for the plan of the investigation. The method of measuring the units of electrical stimulation and of calibrating the inductorium for that purpose is that of Doctor E. G. Martin of the Harvard Medical School.- The values of stimuli are relative, not absolute. Since the publication of the paper of Yerkes and Dodson, referred to above. Doctor Martin has discovered that certain corrections should be made which were not made for the original calibration published in the Yerkes and Dodson paper. All of the values of stimuli used in ^ Yerkes, Robert M. and Dodson, John D. The relation of .strength of stimulus to rapidity of habit formation. Jour, of Comp. Neur. and Psycli., 1908, vol. 18, pp. 459-482. ^ Martin, E. G. A quantitative study of faradic stimulation. I. The variable factors involved. Anier. Jour, of Physiol., vol. 22, pp. 61-74. IL The calibration of the inductorium for break shocks. Ibid., pp. 116-132. Ill 112 L. W. COLE their investigation, as given in their paper, -are relative as are those of the present paper. The chicks. In the experiments sixty-eight barred Plymouth Rock chicks were used, six in preliminary tests and sixty-two under the established conditions of the experiments. The eggs from which the chicks were hatched were all obtained from a single poultry breeder and were guaranteed to be of pure stock. It was necessary, however, to purchase six young chicks of another breeder, but these also were warranted to be pure barred Plymouth Rock chicks and they were kept until it was certain that they presented no marks of difference from the rest of our chicks. Six chicks were used in every series of tests except three. Under the medium condition of discrimination with the weakest stimulus which was employed four chicks were used, while in each of two other groups a chick became sick during the progress of the experiments. When the chicks were eight days old they were given two days of preliminary training (twenty trials) in order that they might learn the way through the experiment box. This was followed by twenty trials in order to ascertain whether the chick had a preference for either the lighter or the darker screen, thus the training series began in every case on the twelfth day after hatching. The training continued until the chick had made twenty consecutive choices of the darker screen. Thus the order of tests, for each chick, was (i) preliminary series, (2) prefer- ence series, (3) training series. Apparatus. Figure i represents in its essential details the apparatus w^hich was used in the investigation. The electrical connections are omitted and the electric key, K, was somewhat further to the right than appears in the figure. For convenience of description we may consider the apparatus as composed of three divisions or boxes, (i) The hover box, O; (2) the illumina- tion box which contains the electric lamps and has for its nearer end the two opal glass screens N-, and N3, and their frame or holder; (3) the experiment box which has the screens and holder for its remote end and consists of tw^o compartments, A. and C. The hover box, O, had dimensions of 100 x 27.5 x 21 cm.* Its floor w^as covered with sand and midway of its length was * All dimensions are given in the order length, width, and depth, and are inside measurements. 113 Figure 1 — The figure is symmetrical, hence the letters G, E, L, and S must be understood to designate both the parts which they respectively mark and also duplicates of these parts on the opposite side of the figure. 1. Hover box: O, hover box; G, inclined planes (which were replaced by wooden platforms) of mesh wire leading from the doors, E, of the experiment box to the hover box. 2. Illumination box: H, left compartment; P, right compartment; L, lamps; S, metric scales. 3. Experiment box: A, compartment in which chick was placed; C, com- partment in which it made choice of screens; B, gateway between A and C. D and D', electric passageways; N, and N.j, illuminated glass screens to be discriminated; E, openings to platforms at the sides of the experiment box; M and M', cardboard shutters for closing these openings; U and R, electric keys for extinguishing lamps in H and P, respectively; K, stimulus key; I, inductorium. 114 L. W. COLE an electric, i6 c.p., lamp (not shown in the figure) in a small box fitted with milk-glass windows. This lamp aftbrded light and warmth to the young chick during the intervals when it was in the hover box, and gave to a small area in the middle of this box approximately the temperature of the brooder in which the chicks were reared. The result of this was that chicks placed in O hovered near this lamp and thus rarely made, at either end of the box, any sounds which might influence the chick in the experiment box in its choice of a passageway back to O. The inclined planes, G, of box O were replaced, early in the experiment, by two small platforms at the level of the floor of the experiment box. From these platforms the chicks hopped down directly to the floor of box O. This change was made because it was found that while chicks very readily walk up an inclined plane it is very difficult and apparently unnatural for them to walk down such planes. This difficulty becomes evident if one tries to imagine a man descending a steep incline with his body leaning far forward. The inclined planes, therefore, to the inconvenience of the experimenter, served rather to toll the chicks in box O upward toward the small doors of the ex- periment box than to give a means of descent for the chick which was escaping from the latter box. The platforms obviated this difficulty. The illumination box, 107.8 x 40.2 x 23.2 cm., was divided lengthwise into two compartments by a light tight partition. The inside dimensions of each compartment were 107.8 x 19.3 x 23.2 cm. Each of these compartments held an incandescent lamp of the oval reflector type with frosted globe. These lamps were mounted on slides so that they could be moved easily along the millimeter scales, S. They were rated as of 50 c.p. When photometered at the close of the experiments the lamp in the right compartment had an intensity of 42.6 c.p., the one at the left 41.2 c.p. By moving the lamps along their millimeter scales they could be changed in position from 8.5 cm. to 103 cm. behind the opal glass screens, Nj and N3, so that a wide range of intensities of illumination was available. As already stated, three difi^erent conditions of discrimination were used. For the condition termed " easy " one screen was illuminated by a lamp 33.5 cm. distant, the other screen was not illuminated. For " medium " discrimination one lamp was RATE OF LEARNING IN THE CHICK 115 at 23.5 cm., the r)ther at 98.5 cm., and for difficult discrimination the lamps were placed respectively at 23.5 cm. and 53.5 cm. from the screens. The experiment box was, as shown in the figure, somewhat narrower than the illumination box. It was divided into two compartments, A, 30 x 16.7 x 21.3 cm., and C, 46 (from parti- tion B to the glass screens N 2 and N3), x 30 x 21.3 cm. A damp pad of felt was placed on the floor of compartment A during the experiments and a similar pad in compartment C extended from the partition, B, to within 2 cm. of the electric wires. These pads were used to moisten the feet of the chicks, for when dry the horny epidermis served to protect the animals perfectly from the electric stimulus. The opening, shown in the parti- tion, B, between the two compartments was closed by a mesh wire door which could be opened by lifting it vertically. That half of the floor of compartment C which was nearest the screens w^as wound with seventeen turns of phosphor bronze wire of No. 20 A.S. gauge. The distance between the successive wires was I cm. This wire was in circuit with the secondary coil of the inductorium, I, and the circuit could be closed by means of the electric key, K. A V-shaped partition divided the wired portion of this compartment into two passageways, D and D'. From these passageways two openings (of which one, marked E, is shown) gave means of egress for the chick to the platforms (see p. 113) and thence to the hover box, O. They were closed by the cardboard shutters M and M'. The two opal flashed glass screens, Nj and N3, (Ni was not used in the experiments) were each 12 cm. square. As already stated, the lamps ^^'ere placed at different distances behind the two screens so that the latter differed from each other in bright- ness. Their relative brightnesses when photometered were roughly as follows : For " easy " discrimination o' : 8 . 9 For medium discrimination i :i3 . 7 For difficult discrimination 1:5.1 While one screen w^as not illuminated under the condition of easy discrimination it had a surface of rather high reflecting power and, since the experiments were made in diffused day- "" This screen was not illuminated. The zero is meant to indicate nothing more than that fact. 116 L. W. COLE light, its value as perceived by the human eye was not dark- ness. This factor of reflected light was present throughout the experiments and made the dift'erence in brightness of the two screens, as judged by the experimenter, much less than that indicated by the above ratio. Subjectively estimated the bright- nesses of the two screens would stand, respectively, in the ratios I : 20 for easy discrimination, i : 4 for medium, and i : 2 for difflcult. A current of 2.1 amperes was supplied to the primary coil of the inductorium. The interruptions were 44 ± 5 per second. The positions of the secondary coil and the corresponding num- ber of units of stimulation appear in table i.^ s TABLE I Position of Units of secondary stimulation 6 220 5 350 4 480 3 590 Method of the experiments. As a- result of the experiments with the first group of chicks, Nos. 1-6 inclusive, it was found necessary to give all subsequent groups twenty trials in the experiment box in order that they might learn both ways of escape from it. The chick was first placed in compartment A of this box. The door in the partition w^as opened and it passed into compartment C. By drawing back the cardboard shutter M' the small door, E, was opened through which the chick escaped to the hover box. In the next trial it escaped at the right and so on until the preliminary series had been completed. There was no difference of brightness between the two screens during the preliminary tests. During the first five of such tests under the condition of easy discrimination there was no light behind either screen, during the second five trials both lamps were at 33.5 cm. and so on. During the first five tests of medium discrimination both lamps were at 98.5 cm., during the second five at 23.5 cm., and the distances 53.5 cm. and 23.5 cm. were similarly used in the preliminary tests of the difficult discrimi- nation. ^ For the calibration of the inductorium used in these experiments see the paper by Yerkes and Dodson, p. 467. RATE OF LEARNING IN THE CHICK 117 The experiments with the first group indicated also that chicks without preliminary training showed a very marked tendency to choose the more brightly illuminated screen. I therefore trained the chicks to escape to the hover box by choos- ing the darker screen. This was done also in order to make the results of my experiments more nearly comparable with those of Yerkes and Dodson, who trained their mice to select the TABLE II Positions of Darker Screen for Two Preference Series and Twenty- five Training Series Subject Date . Experiment . Tests Series Remarks 1 2 3 4 5 6 7 8 9 10 R W A r 1 r 1 r 1 r I r B 1 1- 1 r 1 r 1 r . 1 1 r 1 r 1 r 1 r 2 I r r 1 r 1 1 r r 3 r 1 r 1 1 r I r 4 1 1 r r r I r r 5 1 r I r 1 r 1 r 6 1 r I r r 1 r 1 r 7 1 1 I r r r 1 r 8 r 1 1 r 1 r 1 r 9 r r I 1 1 r 1 r 10 1 1 1 r r r r 1 11 1 r r r 1 1 1 r 12 1 r 1 r r 1 1 r 13 1 r 1 1 1 r r r 14 1 1 1 r r r r I r 15 1 r r r 1 1 1 r 16 r 1 1 1 r r r 1 I' 17 r r r 1 1 1 1 r 18 r 1 r r 1 1 r 1 r 19 1 r 1 r I r 1 r 20 1 1 r 1 r 1 r r r 21 1 I r r 1 I r r 22 1 r r 1 1 r r 1 r 23 1 1 1 1 r r r r 24 r 1 1 1 r r r 1 r 25 r r r r 1 1 1 1 r 118 L. W. COLE white box, since, in the preference tests, the dancers selected the black one in more than one-half the trials/ As described above (p. 1 12), the preliminary series were followed by tw^o series of ten trials each, called "preference series," and designated in Table II by the letters A and B. On the day follow- ing the completion of the "preference series," the training series was begun and they were continued until the chick had made twenty consecutive choices of the darker screen. The order of change of illumination of the two screens appears in table II. The letter 1 indicates that the screen at the left was the darker one, the letter r, that the one at the right was the darker. Since the preference series were preceded by the twent}^ pre- liminary trials, in which the chick escaped from the experiment box by going alternately through the right and left passage- ways, the preference, so-called, was interfered with by the par- tially formed habit. Untrained chicks chose the brighter screen uniformly. During the training series, if a chick chose the lighter passage- way, it received an electric shock, whereupon it usually retreated from the wires, the door of the darker passageway was opened and through that it escaped to the hover box. Under this stimulus the chicks quickly learned to choose the darker screen under conditions of easy and medium discrimination. A few chicks were unable, even after many trials, to learn to choose the darker screen under the difficult condition of discrimination. Results of the Experiments. The results of the experiments appear in table III. This table gives the three conditions of dis- crimination, easy, medium, and difficult, the relative strengths of the stimuli, the numbers by which the individual chicks were designated, and, opposite each of these, the number of trials which preceded twenty consecutive connect choices, or the number of trials "up to the point at which errors ceased." In order to spare the reader an annoying repetition of the phrases, "easy, medium, and difficult conditions of discrimin- ation," I shall sometimes refer to them, respectively, as great, medium, and slight differences of illumination or brightness of the two glass screens. It is evident from table III that under the condition of easy discrimination the rate of learning is more rapid the stronger " Yerkes and Dodson, loc. cit., p. 462. RATE OF LEARNING IN THE CHICK 119 TABLE III General Results of Experiments Condition of discrimination Units of stimulation Easy Medium Difficult Lamps at 33.5 Lamps at 23.5 Lamps at 23.5 cm. and darkness cm. and 98.5 cm cm and 53.5 cm. 220 No. 61s- 90 t rials Secondary " 62s- 90 a at " 651-150 a 6 " 66s- 90 a Av. 105 350 No 7-50 t rials No. 25-50 trials No. 43-230 trials Secondary a 8-30 U " 26-80 a " 44-180 " at ti 9-20 ii " 27-80 a " 45-110 " 5 a 11-60 ii " 28-40 a " 46-230 " ti 12-60 ii " 29-50 Av. 60 ii " 47-180 " " 48-100 " Av. 44 Av. 171.6 480 No 13-20 t rials No. 31-30 t rials No. 37-120 trials Secondary il 14-20 " " 32-50 ii " 38-140 " at u 15-30 a " 33-30 Ii " 39- 90 " 4 ti 16-20 a " 34-50 a " 40-220 " it t 17-20 a " 35-50 ii " 41- 80 " u Av. 18-20 ii " 36-30 u " 42-Failed. 21.66 Av. 40 Av. 130 590 No. 19-20 t rials No. 49-70 t rials No. 55-Died. Secondary " 20-10 a " 50-80 a " 56-Failed. at a 21-10 ii " 51-40 a " 57-70 trials 3 a 22-20 it " 52-30 a " 58-50 " u 23-10 ii " 53-50 ii " 59-40 " Av. 24-30 ii " 54-30 a " 60-Failed. 16.66 Av. 50 Av. 53.33 590 No. 67s-40 trials ] Secondary " 6SS-50 a [ Av. 40 at " 69S-30 '* 1 3 " 70i-80 " 71i-20 a Av. 53.33 " 72i-60 ti Av. 46.66 the stimulus. With a stimulus of 350 units an average of 44 trials was required before errors ceased, with 480 units 21.66 120 L. W. COLE trials, and with 590 units only 16.66 trials. The same relation holds time for medium discrimination and stimuli of 220, 350 and 480 units, but when a stimulus of 590 units was employed the number of trials required for learning to make the dis- crimination increased froin 40 to 50. In order to make certain that this increase in the number of learning trials was due only to the strength of the stimulus I repeated the test with a second group of six chicks and the average was practically the same, namely, 46.66 trials. With medium difference of brightness of the two screens, therefore, the optimal stimulus lies nearer the threshold than under the easy condition of discrimination. The responses of the chicks to the third, or difficult, condition of discrimination are less easy to interpret. With the weakest stimulus used for this condition, 350 units, none of the six chicks failed, with the medium stimulus one failed, and with the strong stimulus two out of five failed. Moreover, the utmost patience was required of the experimenter in order that all should not fail. Each trial also required much more time than in medium and easy discrimination. If, however, we consider only the chicks that learned to make the difficult discrimination the relation stated for easy discrimination appears once more, i. e., the stronger the stimulus the more rapid the learning. It seems clear, therefore, that, with difficult discrimination, the strong stimuli divided the chicks into two groups, (i) those which after a few trials ceased to try to escape and would no longer step on the electric wires, and (2) those which chose with greater and greater caution and, therefore, learned to choose correctly after a small number of trials, each of which consumed much time. To what shall we ascribe this dual result under the third condition of discrimination? It seemed possible that the chicks were divided into the two groups according to their sensitive- ness to the electric stimulus. That is, the more sensitive chicks might learn most rapidly under the influence of a weak stimulus, be slow to learn under the influence of a strong one, and fail completely when under the influence of both a strong stimulus and a difficult condition of discrimination. In order to answer this question twelve chicks were selected of which number six had a threshold of stimulation of 90 units and the remaining six of 150 units (relative values). The former RATE OF LEARNING IN THE CHICK 121 are designated by the letter s (sensitive) placed after their num- bers in table III, the latter by the letter i (insensitive). Tests were then begun with three sensitive chicks, Nos. 6i, 62, and 66, and with three insensitive ones, Nos. 63, 64, and 65, under the medium condition of discrimination and with a weak stimulus. Unfortunately, Nos. 63 and 64 died before the tests were com- pleted. No. 65, however, required 150 trials for perfect dis- crimination while each of the sensitive chicks required exactly 90 trials. The loss of the two insensitive chicks makes a definite conclusion impossible, yet all our work with weak stimuli agrees with the result of the records of these four chicks. It is prob- able, therefore, that the chicks which were most sensitive to the electric stimulus were the ones which learned most rapidly under the influence of weak stimuli. Let us turn now to the results of strong stimulation. Should the sensitive chicks be those which failed under the difficult condition of discrimination and strong stimuli they should be slowest to learn with the same stimuli and medium difference of illumination of the two screens, since it was already proved that a strong stimulus increased the learning rate under this condition. Three sensitive chicks (Nos. 67, 68, and 69) and three insensitive ones (Nos. 70, 71, and 72) were, therefore trained under this condition. An examination of their records shows that the sensitive chicks required an average of 40 trials for learning to discriminate between the two screens, while the insensitive ones required 53.33 trials. Evidently, therefore, sensitiveness to the stimulus was not the condition which pre- vented rapid learning under a strong stimulus. At the close of these experiments with sensitive and insen- sitive chicks there seemed to be no explanation for the divergent results under the third, or difficult condition of discrimination. The behavior of the chicks indicated, however, that the pain stimulus impressed the memory of those that failed so deeply and permanently that, after a few experiences of it, they avoided the electric wires completely and would no longer attempt to escape from the experiment box. This observation, based on the chicks' behavior, receives striking confirmation from the records. The records of the successful chicks in the group 37-41, inclusive, show that in their first fifty trials each chick received an average of 20.4 pain stimuli, while chick 42, which failed, 122 L. W. COLE received in the first fifty trials 30 such stimuli. These additional stimuli seemed to inhibit completely the impulse to enter the electric passageways. In the case of chicks 56-60, inclusive, only the average number of pain stimuli received during the first forty trials can be considered as chick 5 6 would not attempt to escape after the fortieth trial. In the first forty trials chicks 57, 58, and 59, which succeeded, each received an average of 15.3 pain stimuli. Chicks 56 and 60 received an average of 20.5 such stimuli and failed, while chick 55, which went to the wrong passageway in nine of the first ten trials, flew from the door of escape with such violence that he was injured in alight- ing. Those chicks failed, therefore, which made more wrong choices in their early trials and consequently received more pain stimuli than their successfid companions. The additional repetitions of the stimulus seem to have stamped in the impression of the pain and to have caused the failures rather than a native differ- ence of brain plasticity as I had supposed on observing the marked difference of behavior between successful and unsuc- cessful chicks. Here, as elsew^here, repetition seems to be pre- potent in determining memory, if these smooth brained and extremely stupid creatures may be said to have memory. The difference betw^een arousing extremely slow and cautious dis- crimination and inhibiting all efforts to escape lies, I believe, in the added number of pain stimuli given in early trials to the chicks which failed. Records were kept of the sex of all the chicks used in the experiments but they revealed no correlation between sex and rate of learning. In fact the slow and rapid learners were dis- tributed rather evenly between the two sexes. Under the conditions of the experiments, it seemed probable that the heavier chicks received stronger electric stimuli than the lighter ones and therefore learned the more rapidly. But the weights of the chicks of several groups were recorded every three days during the period of experimentation without reveal- ing differences between the heavier and the lighter individuals either in behavior or rate of learning. Again, there was no correlation between weight and sensitiveness to the current in the chicks whose threshold of sensitiveness was determined before training them. I have shown that, for easy discrimination, increase of the inten- RATE OF LEARNING IN THE CHICK 123 sity of the stimulus is followed by decrease in learning rate, while, for medium discrimination an optimal intensity of stimulus is found, increase beyond which is followed by slower learning. Thus far my results and those of Yerkes and Dodson in the case of the dancing mouse seem to agree. In the case of the mouse under the difficult condition of discrimination it was found that the optimal stimulus approached much nearer the threshold than with medium difference of illumination between the two boxes. My results with chicks are in conflict with this unless, as has been done, the cases of failure to learn to discriminate are considered. Then it is found that, with the difficult condition of discrimination and the weakest stimulus, none, with the next greater strength of stimulus, one, and with the strongest stim- ulus two chicks failed. With slight difference of brightness between the two screens the strength of stimulus under whose influence no chicks fail to learn to discriminate is nearer the threshold than the optimal stimulus for the medium condition of discrimination. Perhaps this is as close agreement of the results for mice and for chicks as we should expect to find in animals so unlike. The behavior of the chicks was, however, the reverse of that of the mice. Yerkes writes :^ "The behavior of the dancers varied with the strength of the stimulus to which they were subjected. They chose no less quickly in the case of the strong stimulus than in the case of the weak, but they were less careful in the former case and chose with less deliberation and certainty." My chicks, on the other hand, chose quickly with weak stimuli, but only after long delay with strong stimuli. A chick would sometimes require ten or fifteen minutes to make a choice in the latter case. This difference might perhaps be accounted for by the fact that, with the mouse, a moveable cardboard partition was used by which the space in which the animal could move was gradually restricted. Thus a choice of one passageway or the other was finally necessary. This device could not be used satisfactorily with chicks. The record of one chick, w^hich appeared to be perfectly normal when I began experiments with it, but died before they were completed, deserves notice. Its training series on successive days were as follows: ' Yerkes and Dodson, loc. cit., p. 476. 124 Daily series of tests L. W. COLE Choices I Right 5 8 9 9 8 8 9 7 Wrong 5 2 2 -? I o 4 I c 2 6 2 7 I / 8 3 On the ninth day the chick was weak and would not choose either passageway. When I dissected it a large intestinal cyst was found in which there was much food and a fluid secretion. Such a cyst could have formed in a few days. But the important point is that the only sign of ill health in this chick for four days was the decrease in the number of right choices. On the fifth day physical signs of weakness appeared. In conclusion, it is evident that within the limits of the stimuli which I used, the number of trials required by the chick to learn to choose consecutively the darker of two unequally illuminated screens, when discrimination is easy, decreases with an increase of stimulus. Under medium difficulty of discrimination the above law holds true only for the lower intensities of the stimuli which were used, or, in other words, the optimal stimulus recedes toward the threshold from 590 to 480 units. The above law for the condition of easy discrimination holds true for that of difficult discrimination if we consider only the records of the chicks which succeed in learning to make the discrimin- ation. If, however, we consider only the chicks which fail, the optimal stimulus recedes once more to a point nearer the threshold of stimulation than in the case of medium discrimination. In other words, with the difficult condition of discrimination, strong stimuli divide the chicks into two groups, those which succeed in learning to discriminate by reason of more right choices at the beginning of the training series and consequently fewer pain stimuli, and those which fail because of fewer right choices and more pain stimuli in the earlier trials. So far as I determined the sensitiveness of the chicks it may be said that on the average the more sensitive chicks learned more rapidly both for strong and for weak stimuli. EXPERIMENTS ON TACTUAL SENSATIONS IN THE WHITE RAT By EMORY S. BOGARDUS AND FREDERICK G. HENKE From the Psychological Laboratory of the University of Chicago FOUR FIGURES The object of the present series of experiments was two- fold: first, to determine if possible the function of the tactual sensations of the white rat in learning a maze; and second, to ascertain the effect of the running of previous mazes upon the learning of subsequent alterations of the original maze by open- ing and closing definite pathways. In previous experiments by Watson,' it has been shown .that any one of the following senses may well be dispensed with by the white rat in learning the maze : ( i ) vision — one series of rats learned the maze in darkness, and another series with eyes removed; (2) olfaction — the rats having been made anosmic by an operation; (3) audition — sense of hearing temporarily eliminated by filling the middle ear with paraffine; (4) cutaneous sensation so far as vibrissae were concerned — vibrissae closely clipped. While in the above tests no rats were deprived of more than oiie sense at a time, Watson ^ also experimented with a young male rat whose vibrissae had been clipped and which at the same time was blind and anosmic. Notwithstanding that a certain lack of tonicity was observable, and that errors were eliminated more slowly, the rat learned the maze, and finally became the usual automaton. It is obvious that while these tests indicate that certain senses are not necessary for learning the maze, they do not show what sense-factors are normally utilized. Further, though Watson ^ anaesthetized the nose of an anosmic rat and found that " successive reactions w^ere not in the least disturbed," this experiment threw no light on the significance of the cutaneous sensations in learning the maze, since the animal had been previously trained. Like- wise, although in these cases the vibrissae had been removed 'Watson, J. B., Kinaesthetic and organic sensations: their role in the reactions of the white rat to the maze. Psychological Revieic, Men. Sup., 1907, vol. 8, No. 2. - Ibid., p. 98 f. ' Ibid., p. 77. 125 126 E. S. BOGARDUS AND F. G. HENKE and thereby certain cutaneous sensations had been eliminated, the question of the part played by actual nose and head contact in learning the maze remains open. It is this problem which we propose to investigate. 4 f h 5. T 8 FOOD BOX 0 i 1 J 2 k a h 3 C < — r ENTRA 1 1 Figure 1 — Maze I — Doors 3 and 6 open; other doors closed. Maze II — Door 5 put at 6; Doors 3 and 5 open, other doors closed. Maze III — Door 2 put at 3; Doors 2 and 5 open, other doors closed. Maze IV — Door 4 put at 5; Doors 2 and 4 open, other doors closed. Maze V — Door 1 put at 2; Doors 1 and 4 open, other doors closed. The maze used in the following experiments was one with the food-box in the corner instead of at the center as in the Hampton Court maze. The maze was covered with glass in place of the wire netting commonly used in similar experiments. TACTUAL SENSATION IN THE RAT 127 As indicated in figure i, the alleys were constructed with re- movable doors at i, 2, 3, 4, 5 and 6. In maze I, doors 3 and 6 were open while doors i, 2, 4 and 5 were closed. The first set of rats used in this series of experiments were normal animals, two males and three females. They were about three months old, and had never been used in experimentation. They were fed daily in the food-box of the maze until they became thoroughly tame. Their vibrissae were cut oft' two days before our experiments began. At the end of that time all emotional disturbances had disappeared and the rats acted in a perfectly normal way. This was done in order to facilitate Fig. 2 a b cJ_ — s Fig. 3 Figures 2 and 3 the observation of actual head and nose contact in turning the corners of the pathway. No attempt was made in this experi- ment to keep a record of contacts except at the corners. In general, it was very noticeable, however, that the animals at first kept in close contact with the sides of the pathway. Reference to figures 2 and 3 will show what we mean by " corners." In figure 2, if a rat touched the corner a at any point between b and c or at d, he was checked up with one corner touched. In no case was a rat checked up with more than one contact for a corner. Likewise, in figure 3, if a rat touched the corner a between c and e or at d he was checked up with one contact ; and if he also touched the side of the path- way again between e and f he w^as checked up with another 128 E. S. BOGARDUS AND F. G. HENKE contact. The two corners were never represented by more than two contacts. A record was kept of all the corners not touched, as well as of those with which the animals actually came into contact. Table i shows the average time, the average number of errors, TABLE I Showing Average Time, Average Number of Errors, Average Number of Corners Touched, Average Percentage of Corners Touched OF Five Normal White Rats in Learning Maze I Average time, in Number of trial minutes 1 17.73 2 4.53 3 1.85 4 90 5 1.32 6 2.40 7 98 8 69 9 1.65 10 83 11 87 12 60 13 64 14 50 15 66 16 85 17 92 18 39 19 40 20 48 21 : 42 22 56 23 46 24 35 25 41 26 42 27 46 28 46 29 27 30 36 31 36 32 31 33 31 Average Average number percentage Average of corners of corners errors touched touched 46.4 51.2 .77 16.2 43.4 .70 9.2 25.0 .54 6.8 18.2 .43 5.6 22.4 .56 15.0 30.0 .52 6.4 17.0 .37 4.6 13.6 .37 4.4 14.4 .41 7.2 16.4 31 5.8 9.4 .24 2.4 8.0 .23 3.0 8.8 .20 2.0 4.2 .13 2.6 5.8 .16 1.2 2.8 13 3.2 5.0 .17 1.2 2.4 .13 0.4 1.8 .12 1.6 2.4« .12 0.8 1.6 .05 1.2 3.4 .16 1.4 1.4 .11 0.6 1.8 .14 1.0 2.2 07 0.4 1.8 .06 0.8 2.2 .07 0.2 1.2 .04 0.2 1.6 .05 1.0 1.8 .06 0.4 1.0 .03 0.0 1.4 .05 0.4 0.8 .02 the average number of corners touched, the average percentage of corners touched, of the five normal rats in learning the maze. Reference to the table reveals the following facts : (i) The percentage of corners touched is high at the beginning TACTUAL SENSATION IN THE RAT 129 and gradually decreases as the maze is learned. The first figures, while high, do not render full justice to the situation. Figure 3 will serve to illustrate the point. When the rat came down alley g toward a, one contact at any point in the vicinity of e MISTAKES TRIALS ^' Figure 4 — Constructed from table I. Curve I. — Graphic representation^of number Jfof corners touched in learning maze I by five normal rats. Ordinates represent number of corners touched; abscissas represent number of trials. Curve II. — Graphic representation of errors under above conditions. Ordinates indicate the number of errors. might serve as a sufficient stimulus to make the turn success- fully. Thereby the rat went around two corners with only one contact. Had the corners been farther apart, two contacts would probably have been made, since this was usually the case. 130 E. S. BOGARDUS AND F. G. HENKE In other words, since two corners have been turned with only one contact, the rat has been checked up with only fifty per cent, of contact in these cases. Inasmuch as the two corners were so close together, they became one to the rat, and the fifty per cent, in such instances really represents one hundred per cent, of contact. Our method of counting the corners as given above was due to the fact that often the rats actually touched both corners. It was not feasible sometimes to count two corners as one and sometimes as two. Since there are a large proportion of these double corners in the normal pathway of maze I, the percentage of the corners with which the rat came into contact has been lowered considerably by our method of enumeration; ninety per cent, probably is not too high an estimate for the first run. This seems to indicate that in acquir- ing the kinaesthetic and organic sensations which the rat later utilizes in running the maze, tactual sensations are more impor- tant than smell or vision. (2) In the second place, the table shows that there is a striking correlation between the number of corners touched and the number of errors. Curves I and II, fig. 4, give a graphic repre- sentation, making the correlation more obvious. This indicates that as soon as the running ceases temporarily to be automatic and errors are made, the number of contacts forthwith is increased and tactual sensations are used until the animal has run at least a unit of the maze and_ the automatic kinaesthetic and organic control is re-instated. (3) There is also a general correlation between the increase and decrease of time and the number of contacts. In run 4 (see table I) the average time was .90 minute and the number of contacts 18.2; in run 6, the average time rose to 2.40 and the contacts rose to 30. The rise in time at the ninth run is accounted for by the fact that rat 5 halted and took four minutes to run the maze. This correlation is in harmony with the theory that tactual sensations are of first importance in learning the maze. (4) The table shows that when the maze is learned, contact at the corners is no longer necessary. According to our obser- vations, what holds true of the contact at the corners applies to the contacts with the sides of the pathway between the corners. After the five normal rats had learned maze I, six female TACTUAL SENSATION IN THE RAT 131 blind rats were trained to run the maze. These rats had some time previously learned a different maze. They were run in maze I only until each individual had learned the maze for herself. Average results were secured for twenty-one runs, as indicated in table II. These results, while obtained under different con- ditions, bear out, as far as they go, the four conclusions given above, and especially add weight to the contact theory of ac- TABLE II Showing Average Time, Average Number of Errors, Average Number of Corners Touched and Average Percentage of Corners Touched OF Six Blind Rats for the First Twenty-one Runs of Maze I Average Average Average number percentage time, in Average of corners nf corners Number o f trial minutes errors touched touched 1 7.89 31.6 69.5 .63 2 1 . 84 14.8 34.8 .61 3 1.17 5.6 18.5 .43 4 3.16 24.5 40.8 .40 5 1 . 39 5.1 15.5 .38 6 86 7.5 16.3 .32 7 68 2.1 6.5 .19 8 78 4.3 7.3 7.1 11.1 .19 9 99 .17 10 50 1.8 4.6 .17 11 56 2.1 4.8 .14 12 49 1.5 4.6 .19 13 35 0.6 1.0 .03 14 38 1,8 2.3 .07 15 35 3.1 1.1 .03 16 51 1.8 3.8 .12 17 66 1.0 4.0 .12 18 48 2.1 2.5 .07 19 61 3.3 4.5 .08 20 50 2.5 2.3 .06 21 53 0.1 4.1 .12 quiring the kinaesthetic-organic cues. We also subjoin a typical table (III) of an individual blind rat, which will likewise serve to corroborate our conclusions. Moreover, the percentage of contacts as shown in tables II and III does not begin so high as in table I, and throughout the learning process it remains lower. This indicates that the rats were probably influenced by the previous learning of a maze. One of the most obvious factors doubtless was that the blind rats did not have to learn that there was food in the food-box. Furthermore, they w^ere accustomed to running a not entirely dissimilar maze. 132 E. S. BOGARDUS AND F. G. HENKE After the five normal and six blind rats had learned maze I, they were taught mazes II, III, IV, and V (see fig. i) in succes- sion. Our object here was to study the function of the contact sensations in making readjustments to slightly altered con- ditions. The same general results obtained. All rats became confused in the new situation and were forced to make a new adjustment by the trial and error method. During the period of confusion the animals fell back upon the use of contact sensations and continued to rely upon them until they reached a familiar unit in the maze. The number of contacts again varies with the number of errors made. Table III gives a typical detailed record for a blind rat. In bringing to a close this description of the experiments, it is evident that tactual sensations of the nose and head are util- ized in learning the maze, and this implies that they are used in getting the kinaesthetic and organic cues. The facts which we offer in substantiation are: (a) the percentage of corners touched, beginning high, gradually decreases as the movements of the rat become automatic; (b) a striking correlation exists between the number of contacts and the number of errors; (c) a general correlation between increase and decrease of time, and of the number of corners touched; (d) tactual sensations are no longed used when the maze is learned; (e) when the kinaesthetic and organic cue is lost at any point in the maze, the rats rely upon head and nose contact; (f) the conclusions hold for both normal and blind animals and indicate a minimal effect of -vision. As indicated, the maze used in this experiment was constructed so that the pathw^ay could be altered in various ways. This type of construction was designed for the purpose of studying the effects of the maze experiences upon subsequent behavior in slightly altered conditions. While our experiments were con- cerned primarily with the function of contact sensations in learning to make adjustments to new or slightly modified situ- ations, yet they yielded some incidental results bearing upon the former problem which are of sufficient interest to merit a short discussion. A reference to fig. i will show the successive alterations effected. In maze II, door 5 was placed at 6, while door 3 was opened. Maze II was altered by placing door 2 at 3 and by opening 5. TACTUAL SENSATION IN THE RAT 133 m o H tc w H Z « O ^> b^ n « ^ « «< w m> S^-H t> . ;^S t— 1 Q Z^- <;h-i -* -^ l-H Q^ 5g hJ t= ^ H «)i^ 03 oo^>^>^^oo lo 00 ca CO CO 00 iM rt (M o o iM o o 5^ OJ ^COOO'-HCOOOCO'-f<^C^JI^C^OiMOOTfCO(N(Mi-iTfi N CI 1— I 1— ( 1— I C^ c3 ^^ i) 50C0l>i01^C0t^C0(MCC)^O'-Hl>'-ii— lOOOOGOM'TfiO-^CDtMOOOOOOOOOOO N^C0C0C0C0(MC0Tt<(N'*. CO 00 O OO 1-1 O -^ O t^ C5 lO >^ --4 O O ■§* O lO tp CO I> ^ O 1-1 O to lO o o o o o rt o S>^i-i"IOOOOO^'-iCOOO'-l o '-' « PL, 03 o « H o z o a 01 u:)coco-*lO(^lcoc^^c^^lO(^^(^^>0(^^Tf^(^^c^c^c^^(^lc^lr-Hl--lc^^--HCO(^^c<^c^c^'-lC!^^'^^l^^ N . I— I c3> 2 S a) i-i(Ml>IMC^^Oi-i00 05i-Hi-(0000'-H'-H^Ttt^cocot^coco'Oi^coiocotoi^cocooooooooococoo'C(Moooooooot^oooooo QO(Mi— ii— iCOC^i— ii— li— iCSi— 11— ii— ii— ilM^Hi— ii— ii— lOi— lOOi— i^HdCMi— lOOOOOOOO iCC^t^iCOiOt^OtviOOOt^QCCOCOOOOOOl^COOOt^t^t^Ot^COCOt^ ^^(NO(^^c^'— ii-<>oi— ic^c^ic^ii— iO'ii-oc^ii-ii-ir-i(Ni-ii— irHi— 1 (M l>.OiCt^Ol^l:^iOC0C0C0iOl^iMC0iO(vOt^C000L0t^(M00 COCOIMi— i(N^Hi— iC^i-Hi— ii— i(Ni— 11— ii^C^i— ii— 11— ii— lOiMi— ii-hO ^^CO^O^C^^GO^C^^ON.COt^lOlO COG0iMiMC^C0C^i-iCO00 cot-oooQCioco-^-t^cot^cocococ^jr^ioooocoioio 00 1-1 ^ S 3 3 o i-HCOOC50i-i(MCO^iO!nr^OOC50i-i(NCO-*iOcD i-l.-li-li-ii-ii-ii-ii-ii-l.-l(N(M(M(N(MiMiM