The Animal Mind: A Textbook of Comparative Psychology

was simplified ; thus the birds learned not to attack other parts of the box, to use the bill instead of the claws, and to stand on the floor beside the box instead of hopping upon it (611). In Rouse’s test of the pigeon by the puzzle-box method, it showed less aptitude than that displayed by the English sparrow (647). Small (684) tested his white rats with two boxes containing food. One could be entered by digging away the sawdust which was banked around the lower end of the box, if the digging was done in a particular place; the other, by tearing off strips of paper which held shut a spring door. The result of the earlier series of experiments with the firstmentioned box was that after an hour and a half on the first day one rat happened to dig in the right place and entered. The second day this rat took only eight minutes, and the thirteenth day only thirty seconds, to enter. With the second box there was always a tendency to begin by digging, and even in the thirteenth experiment, where the rat got in by biting off the papersin fifteen seconds, she began by two strokes of digging. In a later test with this box the rat chanced to be extremely hungry, and dug violently for several seconds, displaying a blunting of the discriminative powers by hunger, analogous to that which we have found in very low animals. The rats were later trained to discriminate between the two boxes, being sometimes presented with one and sometimes with the other.

In Thorndike’s work on cats and dogs, the investigator placed the animals themselves in the boxes, and food on the outside, so that the problem was not how to get in but how to get out. The getting out could be accomplished in various ways, such as pulling a wire loop, clawing a button around, pulling a string at the top of the box, poking a paw out and clawing a string outside, raising a thumb latch and pushing against the door, and so on (Fig. 13). The animals, on being first put into the box, made all sorts of movements in their struggles to get out; the right movement was hit upon by accident. Only very gradually, as the experiment was repeated again and again, were the useless movements omitted, until finally the right one was

Fic. 13. — Puzzle box used in Thorndike’s experiments on cats. performed at once (704). Wesley Mills criticised these pioneer experiments of Thorndike’s on the ground that the animals were under such unnatural conditions and in such an extreme state of hunger that they profited by experience more slowly than might otherwise have been the case (492); and this may have been to a certain extent true. In testing monkeys with puzzle boxes Thorndike placed the food on the inside and the monkeys on the outside. He found a marked difference between the speed of their learning and that shown by the cats and dogs. ‘‘Whereas the latter were practically unanimous, save in the cases of the very easiest performances, in showing a process of

gradual learning by a gradual elimination of unsuccessful movements and a gradual reénforcement of the successful one, these are unanimous, save in the very hardest, in showing a process of sudden acquisition by a rapid, often apparently instantaneous abandonment of the unsuccessful movements and selection of the appropriate one, which rivals in suddenness the selections made by human beings in similar performances”’ (708). Kinnaman further complicated the box tests with his Macacus monkeys by constructing ‘‘combination” fastenings, which required the performance of a set of actions in a certain order, and found that these seer ei aR rt No were mastered by the animals The figures indicate the order in

4 which the parts of the combination (401) (Fig. 14). had to be dealt with. Cole’s (134) work on the raccoon indicates that in speed of learning this animal stands ‘‘almost midway between the monkey and the cat,” while “in the complexity of the associations it is able to form it stands nearer the monkey.” The raccoons, like the monkeys, learned combination locks, .although they did not learn to perform the various movements involved in a definite order. They showed an interesting tendency to skip at once to the movement that immediately preceded the opening of the door. The porcupine also proved gifted with the ability to learn combination locks (651), while the squirrel’s puzzle-box exploits were limited to boxes which could be entered by the simple process of digging in sawdust (832). The learning

of combination locks probably involves the formation of systems of movement, as well as the dropping off of useless movements; the process of system formation will be discussed in a later section. The building up of systems of movements is an important part of the learning process in another method of studying the intelligence of animals, namely, the labyrinth or maze method. In the typical form of this ‘—| method, food is rfi] placed at the end c| of a pathway in- ‘| volving a number of turnings, in which

Fic. 15 — The Hampton Court maze. (a) taking a longer instead of a shorter route, (b) entrance into cul-de-sacs. The animal has to learn to run to the end of the path and secure the food in the shortest possible time, or by the most direct route. His progress in learning may be measured either by the total time he consumes in running the path in each trial, or by the number of errors he makes, or by the total distance he runs. The method in its developed form was first used by Small (685) in experiments on white rats, and is especially adapted to an animal so active as the rat. Small used a very complicated maze, a facsimile on a small scale of the one to be found in the grounds of Hampton Court Palace (Fig. 15). Such mazes, with high box walls, were a frequent feature of old gardens. Much simpler mazes have been used with other

animals. Where a maze consists of only two passages, requiring the animal to learn merely a single turning, the method may be practically merely a discrimination method : thus Yerkes’s (822) training of the earthworm made use of a maze with two passages only to choose between, a light and a dark one. In a pure maze experiment, however, there is no way of distinguishing between the passages except by experiencing the consequences of following them. Thus the crayfish was tested by the use of a maze with a single choice of paths. One end of the box communicated with the aquarium; about halfway down the length of the box a partition put in longitudinally divided it into two passages, one of which was closed at the end by a glass plate. In sixty trials the animals, which had originally chosen the correct passage 50 per cent. of the time, came to choose it go per .cent. of the time. A second series, with a single animal upon which more tests a day were made, resulted in the formation of a perfect habit in two hundred and fifty experiments. The glass plate was then shifted to the other passage, and the crayfish was naturally completely baffled for a time, but succeeded in learning the new habit (829). The crab Carcinus granulatus made progress in learning to traverse a labyrinth with two points where a choice of path had to be made, but did not wholly master it in fifty trials (804). For both the crab and the crayfish, the experience of getting back into the water was the influence relied upon to eliminate the useless movements; the slow learning of these animals indicates that the method was not well adapted to them. Ants showed some ability to learn a maze with several turning points, following the proper course even when their smell trail was obliterated (219). Fish have proved able to learn very simple labyrinths, but

With the green frog, a maze allowing two choices was used, and was learned in one hundred trials (805). \ With the turtle, a labyrinth distinctly more complex was used. It involved four blind passages, and led to the turtle’s comfortable, darkened nest. During the first four trips the time was reduced from thirty-five ; minutes to three or tT e minutes and thirty seconds; in the fourth trip the an- F c | imal took two wrong turns. The time of 4 s | | the fiftieth trip was | thirty-five seconds. : fi , In a second laby-

6 ~ rinth (Fig. 16), two My ; inclined planes were | NEST | introduced, up and Fic. 16.— Labyrinth used by Yerkes with turtles. down which the A, starting point; F, blind alley; 3, 4, 6, inturtles had to crawl. sl This labyrinth took them longer to traverse, and the time curve shows greater irregularity, rising, for instance, to seven minutes on the forty-fifth trial, after having been as low as two minutes and forty-five seconds at the thirty-fifth. The process of shortening the path was observed very prettily in connection with the inclined planes. The turtles had to turn about as soon as they had reached the bottom of the descending plane. They soon began to make the turn before they got to the bottom, and finally to throw themselves over the edge as soon as they reached the top (801).

Some of Thorndike’s (704) early experiments on chicks involved a very simple form of the labyrinth method, in that the chicks, which were confined in small pens, could escape by running to a particular spot, or up an inclined plane. Porter (610) found that the English sparrow quickly learned the Hampton Court maze, and that the vesper sparrow and cowbird learned a simpler form in twenty or thirty trials (611). Pigeons tested by Rouse acquired the ability to traverse four different labyrinths, 4nd it was noted that their experience with the earlier ones seemed to help them in the later ones (647). Hunter’s (349) experiments on the pigeon with the use of the maze will be mentioned later. White rats observed by Small learned the Hampton Court maze, in nine experiments made at intervals of two days, so well that they committed only two errors in the ninth test, but the significance of this time is obscured by the fact that the rats were allowed to run freely about the labyrinth every night (684). In Yerkes’s (820) study of the Japanese dancing mouse, the reactions to irregular and to regular labyrinths were compared, and it was found that a maze of the latter type, that is, one where left and right turns alternated, was more quickly learned and more perfectly mastered than an irregular one. Kinnaman (401) taught two Macacus monkeys the Hampton Court maze. . ‘ The feature of maze learning which interests us at thi point is the dropping off of useless movements. This probably occurs partly through the general tendency of useless movements to be omitted, and partly through the tendency of the successful movements to survive. It has been argued that if the shortening of the maze running is due to the fortunate consequences of the successful movements, then the errors which should be earliest dropped off are those at the latter part of the course, which come nearest in point of time to the final success, usually the

obtaining of food. Hubbert (346) does not find that this is actually the case, but Vincent (750) does, and on the \whole the evidence points to the conclusion that the errors pees the final success are the first ones eliminated. uch exceptions to this tendency as appear may well be ue to two causes: first, the natural tendency of useless movements to drop away even when a successful movement is not pushing them out, and secondly, the equally natural but wholly opposed tendency of movements to organize hemselves into systems, a tendency which will be considered in the next section. Watson (771) lays especial 1 Chacon on the fact that the successful movements in puzzlebox and maze experiments have the advantage of fre- , quency of performance. The successful movements are ‘always performed, in every maze experiment, simply because the experiment continues until they are performed ; there is no such necessity that any particular unsuccessful [estcn should be performed in every experiment. Thus

the successful movements, Watson thinks, owe their survival the law of repetition. It is quite probable that their inevitable performance once in each running of the maze may ' bea factor aiding their survival, although quite conceivably, | as Thorndike! has suggested, many unsuccessful movements ‘may actually be oftener performed, owing to the fact that they may be repeatedly tried in the same experiment. But Watson endeavors to reduce all learning through the dropping off of movements to the influence of the frequency with which the successful movements occur; and this can only be done by ignoring such cases of learning as those where the frog ceased in one or two trials to snap at food when the snapping led to harmful consequences, or where the spider learned not to disturb itself at the sound of a

tuning fork. Frequency cannot be a factor of importance here. Such cases show that there do exist in animals teny, dencies (a) to abandon movements which have no consequences of importance to the organism, and (6) to eliminat even movements that are important in favor of movement that have greater importance. It is not clear why Watson is unwilling to admit that the ‘‘sensory consequences,” the vital importance of the results of a movement, are | factors in determining its survival. He seems to think' that sensory consequences must be stated in terms of mental processes, and therefore must not be mentioned pi a consistent ‘‘behaviorist.” We, not being behaviorists but psychologists, are quite willing to talk about the pleasantness and unpleasantness accompanying the benefit and harm of reactions, but if we were behaviorists, we should certainly not feel obliged to deny that animals can be benefited or harmed by their own actions, because we feared benefit and harm might suggest pleasantness and unpleasantness to the minds of our readers.

Before we pass on to another aspect of learning, which is quite as important in the phenomena of maze running as the dropping off of movements, there are a few points to be noted with regard to the relation between punishment and reward, harm and benefit, as influences in learning. . Punishment appears to produce more rapid learning than reward, unless it is so severe that it attaches itself to tna whole learning situation. Punishment and reward combined give, probably, better results than either alone (328). Further, a movement that results in harm, and is therefore supplanted by the negative response of withdrawal, is mor completely eliminated than one which is merely useless and is supplanted simply by a state of rest. Evidence o this was obtained by Bogardus and Henke (59) in experi-

ments where, after rats had learned a maze, the path was altered, certain passages being closed and others opened. The rats found it decidedly harder to learn to enter former cul-de-sacs than to take those turnings which they had formerly omitted merely because they were a longer way around than the true path. The positively unpleasant experience of running into a cul-de-sac had more completely eliminated the movements that led to it than did the merely useless running of a longer passage. Finally, it is clear that we cannot draw a hard and fast line between a useless reaction and a harmful one. We have seen that A severe punishment like an electric shock may delay learn- g because it attaches itself to the learning situation as whole. And in the writer’s experiments with rabbits 786), a young rabbit of very nervous temperament was rendered unfit for further experimentation simply by happening to push repeatedly at the wrong or closed door of a box. He had been working well up to that time, but from that time on he ran away whenever he was confronted with the experiment box. It would appear that emotional factors in the animal may render movements positively \parmful which would ordinarily be merely useless.

We have now to consider another type of learning, diametrically opposed to that of the dropping off of movements. In this type, the movements which an animal makes sucessively become organized into a series. No movementof the series is dropped out as a result of the learning, but the oftener the series is run through, the more rapidly it is ‘performed. It is evident, if we consider our own learning \processes, that many of them are of this type. When we

say the alphabet or the multiplication table, the learning process has not involved the dropping out of any of the movements. It would profit us little to pass immediately from A to Z, dropping out the intervening movements, or to skip at once from the first to the last stanza of apoem. We find in such cases that repeated performance of the series of movements results in two changes. First, the movements follow each other more rapidly: this can be explained by the law of repetition, according to which the oftener the nervous process traverses a certain pathway the less resistance it encounters. Secondly, the movements of the series no longer need outside stimuli, but apparently each movement supplies the stimulus for the next. When one is playing a piece of music for the first or second time, each movement has to have the stimulus of the notes on the page; when a piece has been long practised, each movement sets up the next one ‘automatically.’ This really means that as one movement is performed, the sensory processes occasioned by the contraction of the muscles involved excite the motor pathway for the next movement, The stimulus for one movement is the kinesthetic excitations received from the preceding movement. The truth of this is evidenced by the fact that if we break down in playing a certain passage, we can recover ourselves by going back a little, so as to get the proper kinesthetic stimuli.

This type of learning obviously functions in learning a maze path. Here we have to deal not with the acquiring of a single ‘successful’ movement and the dropping off of all others, but with the establishment of a whole series of successful movements which must be performed in a cer tain order. Much experimentation has been performed to study the sensory cues involved in maze running. The j question resolves itself into three: (1) What sensory stimuli j does an animal rely upon after it has thoroughly learned ‘) the maze path? (2) What sensory stimuli does an animal naturally rely upon in learning the maze path? (3) What \ sensory stimuli can an animal do without, if necessary, in \veaming a maze path?

i ganized into a system such that the sole requisite stimulus for the performance of one movement is the kinesthetic Fatation resulting from the preceding movement. The best proof of this is furnished by experiments where the rats which had learned the maze from its beginning were started at points further along; they could not pick up the true path with any speed until, in running back and forth, they chanced to make two or three correct turnings. This set off the remainder of the maze-running process precisely as playing over the preceding passage would enable a pianist to proceed beyond the point of a breakdown. Similarly, when the maze was shortened by removing a section from its middle, the rats ran against the ends of the shortened passages; when it was lengthened by elongating certain passages, the rats tried to make the turns at the old points (118).

(2) and (3) It is difficult to make sure just what sensory stimuli function in enabling an animal to learn the maze. The only methods that suggest themselves are (a) that of depriving an animal of the use of one sense and seeing whether he can learn the path, thus answering question (3) above; or (b) that of supplying him | with cues especially appealing to a certain. sense, and noting whether his learning is accelerated. But neither ‘of these methods reproduces the condition of normal maze

learning. Watson (767) showed that rats deprived of sight, hearing, smell, and touch (the vibrisse or long whiskers being removed and the paws made anesthetic), could learn the maze. Yerkes (820) demonstrated that the Japanese dancing mouse does not necessarily depend on sight, smell, or touch for guidance. On the other hand,\ Vincent’s (747-749) experiments show that a maze in which the true path was painted black and the wrong paths white, or vice versa, was learned more quickly than an ordinary maze; that one in which the true path was smeared with | beef extract and cream cheese alternately (the two odors | being used to prevent olfactory adaptation) gave a greater | total accuracy; and that in a maze without sides, that is, an elevated pathway, the rats were much disturbed by’ the loss of the accustomed contact with the walls. Different animals are undoubtedly unlike in the use they make of sensory cues. The frog studied by Yerkes (805) in a very simple labyrinth showed a disturbance in its habit when red and white cards placed on either side of the passage were interchanged. ‘The pigeon (647), when required to go through a labyrinth in darkness, was obliged to relearn it. On the other hand, Small found that altering the direction of the light had little effect on the performances of his white rats. He also placed wooden pegs painted red, at each division of the paths, in the middle of the correct path, and caused the maze thus arranged to be learned by untrained rats. They did not learn it any faster because of the presence of these visual hints, nor, when it had been learned, were they at all discomposed by the removal of the pegs (684). Allen’s (4) guinea pigs did not alter their behavior when the position of colored cards in the maze was changed. Rouse (647) found that the pigeon could make use of auditory stimuli as cues. He arranged to have

an electric bell rung whenever the birds entered a wrong alley, and a wooden bell sounded when they emerged and took the right course. After they had learned the path under ‘these conditions the two kinds of sound stimuli were interchanged, and the result was a certain amount of confusion on the part of the birds. On the whole, in he case of active animals whose vision is not highly developed, such as the rat, the principal factor in learning a

aze appears to be the actual running of it. As the paths are traversed at random, the useless movements tend to be dropped off, and the successful ones not merely to survive, but to become organized into a system such that each movement itself provides the stimulus for the succeeding one. Vincent (747) found that while visual cues aided the learning of the maze, the final running was not so rapid as if the habit had been formed wholly under kinesthetic guidance.

Some curious results have been observed when the maze is rotated through angles of 90, 180, or 270 degrees. Since this has no effect on any of the paths, but only on the relation of the entire maze to its environment, it ought not to disturb animals which are depending entirely on their ‘own movements for their cues, yet apparently it does in some cases disturb them (767). Possibly the preliminary swings which the animal gets in being picked up and introduced to the entrance of the maze are the disturbing factor. Hunter (349) found that some of his pigeons were disturbed when the maze was rotated, while others were not, and concludes that the latter were guided by cues within the

Maze experiments are not the only observations on animals which reveal the existence of successive movement systems, or “kinesthetic memory.” If Piéron (579) is right, some species of ants are aided in their return to the nest by repeating all the turnings they took on the out- | ward journey. Watson reports the following observation — on the terns of the Tortugas. On an occasion after he had trained one of these birds'to use a nest raised a hundred centimeters above the ground, he moved the nest a hundred centimeters to the eastward; the bird, returning, hovered “in space, attempting to adjust to the nest in the air at its former position and height” (769, page 226). Rockwell (639) relates that a ground squirrel had made inside a cabin, a nest which it was accustomed to reach by climbing up the leg of a cot that stood in one corner of the cabin. When the cot was removed, the squirrel, entering the cabin, ran to the place formerly occupied by the cot, and went through the motions of trying to climb the nonexistent leg.

It must further be noted that useless movements not infrequently get organized into movement systems, and thus }. their elimination is delayed. Animals in running a maze form habits of going wrong which greatly interfere with the reduction of their time records. In the case of some salamanders which the writer vainly tried to teach a fairly complicated maze, each individual acquired quite an elaborate habit of making wrong turnings, and remained true

volve strongly prepotent movements, whenever, that is, the “motive” is weak, thenatural tendency of movements to organize into systems may take the place of the tendency to drop off the unnecessary ones. Such an influence as this is very likely one reason why errors in the maze are not eliminated in the exact order of their distance from the final, beneficial, and pleasurable goal. coats movement systems is, in our own experience and ‘probably in that of an animal, the diminishing of attention to outside stimuli, and the disappearance of such emotional states as uncertainty and the unpleasantness of errors; a feeling of confidence and security replacing them. There is an alleged case of learning on the part of certain marine animals which, if it exists, probably belongs under the head of the formation of successive systems. This is _ the acquisition of rhythmic reactions, to stimuli which occur at equal time intervals, and the persistence of such rhythms when the stimuli have ceased to act. Marine snails, sea-anemones, annelid worms, and hermit crabs show changes in the direction of their responses to light and to gravity which correspond with the state of the tides: the sea-anemone, for instance, opens at high tide and shuts at low tide. Now certain French observers, Bohn (80, go), Piéron (584), and Drzewina (192) report that when the animals are removed to the aquarium, they continue to show fluctuations in their light and gravity reactions at the times of high and low tide, although of course the actual stimuli which the tide gives them, for instance the mechanica] jarring of the waves entering their pool as the tide rises, are now wholly lacking. No American observer has been able to show such a continuance of the tidal rhythm in animals removed from direct tidal action (256, 293, 509 a, 551). The phenomenon suggests analogies from our own ‘experience; for instance, there is the case of ‘habit hunger.” We feel hungry at the time of day when we are accustomed to be fed; if we do not get food at this time, in half an hour or so the hunger sensations disappear, and we can go quite comfortably without food for some time nmger. The hunger sensations are due to movements of the stomach; now these movements were originally in-

duced by the presence of food in the stomach at a certain time. They have apparently become a part of a system of internal, organic movements, so that when these internal processes have continued for a length of time equal to that which usually elapses between meals, they produce the stomach contractions, in the absence of the original stimulus, the food. Thus the case seems to be like that of the running of the maze by a thoroughly practised animal; each act is the stimulus for the next, and outside stimuli are | unnecessary.

Tae MopIFICATION OF Conscious ProcESSES BY INDI- VIDUAL EXPERIENCE (Continued) in the homing of certain animals. As we have seen, the / evidence is conclusive that solitary wasps guide themselves | back to the nests they have made by “‘recognizing”’ certain } visual peculiarities of the surroundings. They are confused if the appearance of the nest or its vicinity is altered. On first leaving the nest in search of the prey with which to stock it, as food for the larva, they make an elaborate flight with many turnings in and out about the immediate neighborhood, which has been appropriately termed a locality survey. Now when the wasp has found and secured the caterpillar or spider which she seeks, she retraces her flight apparently with the guidance of the visual landmarks she noted on the outward journey. No one, it is true, has yet actually determined the homeward flight of the wasp in its relation to landmarks, but the probabilities are that such is her method of procedure. The peculi-/ arity of such learning is that it does not depend on repetition. The wasp makes but one nest in a given situation, and in the case of certain species at least she makes but one flight in search of food and but one homing flight. She then makes a new nest in a.new locality, impresses new landmarks upon her memory, and is guided in her next