192 STUDY OF HABIT FORMATION the sand in hot weather (p. 136). Thinking that the egg might be destroyed, the experimenter scraped away the sand covering the entrance so as to expose an inch of the opening. The bird alighted, passing by the door again and again. Two inches of the opening were then exposed. Under these circumstances the bird mounted the sand pile again and again and attempted to push its way through the meshes of the wire above the opening. Finally by accident it poked its head through the opening and squeezed the rest of the body through, making no effort to scratch. The time required for this whole trial was 1 hour. After allowing the bird 5 minutes on the nest it was driven away, and the sand piled up against the door to the height at which it had previously been successful. Most of the useless movements disappeared. The time required was only 1.16 minutes. On the third trial the sand was piled so as to co'inpletely cover the entrance. After 20 minutes of random movements with no success an inch of the open- ing was exposed. The bird came up and squeezed through the hole, making violent efforts, in 0.25 minute. On the following day no further tests were made, since the mate of this bird was on the nest. On the second day afterwards the sand was piled up to within 1 inch of the top of the door. The time for entrance was 2 minutes. On the next trial, once again the sand was piled so as to completely cover the entrance. After 38 minutes of random move- ments, while attempting to poke its head through the wire meshes above the opening, it accidentally poked its bill and head through the sand pile, withdrew the head, walked around and around the cage, as it came back to the hole poking its head into it 6 or 8 times. Success was attained at the end of 40 minutes from the beginning of the test. On the next test this animal required 7 minutes to effect this response. On succeeding tests it dived at once for the opening, made a small hole, pulled the head out, ran half way round the box, returned and squeezed through the opening. Time, 0.66 minute.
Motor habits: fish. — Thorndike has described in very general terms the labyrinth habit of a fish (Fundulus). His MOTOR HABITS IN FISH 193 plan was to confine the animal in the sunlight in a tank by means of glass partitions which had holes in them placed at different distances from the ends, thus converting the tank as a whole into a simple labyrinth. The animal first, e.g., has to swim up and to the left to cross the first parti- tion; up and near the center to cross the second; and up and to the right to cross the third. The animal when con- fined in the sunlight tries to get back to the opposite end of the tank which is. shaded. He reacts to the situation at first by swimming against the screen, and bumping against it here and there along the bottom. He may stop and remain still for a while, but he will occasionally rise up towards the top of the water. When he happens to rise up to the top at the left-hand side, he has a clear pathway in front of him, and thus can cross the first partition. We have few detailed studies of motor habits in fishes. Triplett has given an example of the formation of an interesting habit in the perch (Perca Americana). A glass tank was constructed 4 x 2 x IJ feet in size. The study was begun upon two of the perch. They were fed for several months upon live minnows 2 or 3 inches in length. At the begin- ning of the experiment in question a glass partition was placed in the tank and the food of the perch changed to angle worms. Three times a week, Mondays, Wednesdays, and Fridays, for nearly a month, between the hours of 4 and 5 in the afternoon the perch were fed, and after that on every day at that hour. In making the test the min- nows were put into their side of the tank for 30 minutes, during which time the behavior of the perch was observed. At the end of 30 minutes the minnows were removed until the next trial. Triplett 's note on their behavior follows: " Two minnows were placed in the tank at 4:30 p.m. The perch immediately began ramming the glass to get at them. Their actions became very violent as the minnows approached the partition. They ceased their butting and swam away from the glass for a few sec- onds after 7 minutes of continuous effort. A second trial lasting only 1 minute followed, and this was followed in turn by still shorter periods. The greater energy was shown always when the minnows turned their heads towards the perch, as it is only when they are 'head on' that the latter strike. At 4:55 the female was showing what seemed to be signs of anger and was striking the 194 STUDY OF HABIT FORMATION glass hard...By 5 o'clock both perch had left the glass and seemed to have given up the attempt completely.'* On the second day their efforts were not so long continued nor so violent as on the first day. There was a gradual lessening of the time and energy spent in striking at the minnows. This depended, though, somewhat upon the state of hunger. By the end of a month the perch had ceased to strike at the minnows when the latter were put into the cage. The second step in the experiment then followed. The glass partition was raised and the minnows allowed to swim into the compartment containing the perch. The male perch paid not the slightest attention to the minnows. The female moved towards them several times but did not harm them. On succeeding days the minnows were admitted in the same way. Several times the perch started to stalk the minnows, but within a short dis- tance of their heads they turned aside. In the early trials before admitting the minnows the appetites of the perch had been dulled a little by feeding them with worms. In the later trials, however, this was not done. The third step in the experiment was to leave the minnows in the tank continuously for a week, separated from the perch only by the glass partition. The perch soon ceased entirely to touch the glass. They would frequently take up a position near the glass and watch the minnows playing on the other side within 2 or 3 inches of them. When the partition was removed they did not attack them any more than in the case just mentioned.
The most interesting feature of Triplett's experiment and the one which bears most plainly upon the motor side of habit formation comes out in his discussion of Bateson's statements concerning the lack of habit-forming powers in fish. Bateson says: '•' None of the fish seemed to get any lasting appreciation of the nature of the plate-glass walls of the tank. The same fish will again and again knock its head and try to seize the objects moving on the other side. After repeated attempts to take food on the other side of the glass, they will desist, but some of the oldest inhabitants [plaice, pollock, and bi'eam) which have been living in the aquarium for ai)out a year will perseveringly try again the next time."
MOTOR HABITS IN EEPTILES 195 Triplett entirely disagrees with this. He says: " While these fish did not entirely cease striking the glass during the time when the trials were about 30 minutes long, their attempts grew very much more infrequent and their blows feebler. Later, in the changed form of the experiment, becoming accustomed to the sight of the minnows, they gave up striking the glass, merely continuing to watch them. This, in connection with their conduct towards the minnows when the glass was removed, suggested that they have at least a strong temporary appreciation of the obstacle."
The strongest proof of the formation of a habit shows in their conduct after the partition itself had been removed. On reaching the place where the partition had formerly rested they stopped and turned back, advanced again, made little jumps towards it as if expecting to strike the usual obstruction and were plainly at a loss. They then turned and swam down as if following the glass. Ten days later, with the same conditions, they swam out to the mark several times and then turned and swam back. On a later occasion, when the glass was taken out, they turned 3 times at the mark, but finally crossed it in a hesitating manner.
Motor habits: reptiles. — The turtle has been studied in the laboratory by Yerkes. It is extremely sluggish in its movements. The impulse or incentive to get the ani- mal to work was that of escape. Instinctively the animal attempts to hide in some dark secluded place and will try to escape from confinement and go towards such a place. This combination serves very well for a motive. Hunger could not be used for a motive since the reptile does not eat well in confinement and its time of eating is Yery ir- regular and hard to control. The maze consisted of a simple box 3 feet long, 2 feet wide, and 10 inches deep. It was divided into 4 portions by partitions also 10 inches deep. At different points in the partitions a hole 4 inches long and 2 inches deep was cut. This permitted the passage of the animal. After passing through the last partition the animal could get to its darkened nest of damp grass. A small speckled turtle learned this simple maze as fol- lows: After wandering about constantly for 35 minutes, it chanced to find the nest, into which it immediately 196 STUDY OF HABIT FORMATION crawled and remained there until taken out 2 hours later. Experiments were made every 2 hours. On the second trial the nest was reached in 15 minutes. There was much less wandering. The time for the third trial was 5 minutes; for the fourth, 3 minutes and 30 seconds. During the first three trials the course taken was so tortuous that records of it were hard to obtain. There was aimless wandering from point to point within each space, and from space to space. After the third trial the routes became more direct. The tenth trial was made in 3 minutes and 5 seconds, wdth only 2 mistakes in turning. The time of the twentieth trial was 45 seconds; that of the thirtieth, 40 seconds. In the last experiment the course was direct, as was also true in the case of the fiftieth trip, which was made in 35 seconds.
Habits of manipulation: mammals. — An interesting ex- ample of the manipulation of a complicated piece of ap- paratus appears in the tests on the rhesus monkey. This example illustrates the general features of habits of manipulation as well as the complexity of such habits. The animal was first allowed to learn a series of simple problem boxes similar to those shown on p. 94 ff. After it had mastered a number of the boxes with simple fastenings a box was constructed which could not be opened until a combination of fastenings (made up from the earlier learned simple fastenings) had been released in serial order. In the particular case mentioned the monkey was required to bear down on a lever at the left of the box; to push in a bar at the right, and to pull out a string behind. These were all so arranged that w^hen once moved they locked and could not be shaken back, and thus prevent the moving of the next part of the combination. The cut for this particular box is shown on p. 99. The monkey in at- tacking a box of this kind uses his hands incessantly. He tries first one and then the other in quick succession. One part worked, he tries all the others, though he may labor on one at a time and try often to move parts that have already been moved. A complete record of these move- ments is very instructive. In order to bring success these MOTOR HABITS IN BIRDS 197 Habits of manipulation: birds.— Porter describes forma- tion of the habit in the English sparrow of opening a small food cage, the door to which was held in place by a latch. To the latch was attached a string which led out over the box in such a way that the bird, in crawling over the box, would press the string in, and thus raise the latch. The door would spring open of its own accord (p. 98). The behavior of one English sparrow, B, is described somewhat in the following manner: B succeeded in opening the door first by pushing in the string with his claws. On the first trial he hopped up to the left of the door, on the far end covered with wire, and climbed down over the sides from the top. He next walked along the front side with one foot on the floor and one on the side of the box. Finally he accidentally struck the right place and the box opened. The records show that the unnecessary reactions get fewer in number with each successive experiment. In the tenth test he had left off all of them. At about this time he began to use his bill on the string. Instead of pushing it with his claws he struck it with his bill. On the twenty-seventh test he no longer clung to the sides of the box while striking the string with his bill, but pulled the 198 STUDY OF HABIT FORMATION string while standing on the floor. The time of the first trial was 10 minutes and 50 seconds; of the fifth trial, 3 minutes; of the twelfth, 1 minute; of the twentieth, 15 seconds; and of the thirty-fifth trial, 6 seconds. It will be seen that the above manipulation device is extremely simple in character. The structural relations and methods of living of reptiles, amphibia, and fish make it difficult to test their ability to form habits of manipulation. Up to the present time no such habits have been reported.
Sensory habits. — It would take us entirely too far afield to discuss sensory habits in detail as we have the two types of motor habits. Reference to p. 61, where methods of forming such habits are discussed, and to pp. 220 and 222, where the learning curves and the analyses of sensory habits, respectively, are treated, will give a fairly definite idea of how such habits are formed. For similar reason the delayed reaction is not treated separately here. Reference should be made to p. 224.
III. Curves of learning: motor habits. — Enough ex- amples at least of motor habits have been given to show their common characteristics. The first few trials always require a large amount of excess time; succeeding trials require less and less time. If the total distance the animal traverses in reaching the food in successive trials is re- corded instead of the time, the same situation appears. In the maze, e.g., the rat, on its first trial, may run 400 meters (when the distance from entrance to food box is 4.48 meters) before finally reaching the food. On his fiftieth trial he may not take a useless step. Investigators usually plot curves of learning in terms of time or distance and number of trials. Below is presented the curve for the learning of the maze by white rats (27). The number of trials is marked off along the abscissae and the units of time and distance along the ordinates. The average of the time and of the total distance is recorded at each suc- cessive trial. In general it may be seen that there is an enormous drop in both the time and the distance curves on the first 8 or 9 trials. All succeeding trials show a slow and steady decrease in time and distance, until at the CURVES OF LEARNING CURVES OF LEARNING seventieth trial the animal reaches the normal physiological limit of training. The habit is perfected. Individual animals differ somewhat as might be expected in the abso- «TO 1 SCO \ ZOO J ISO u I m 50 Time fH luniNSseo v iD/sfoflcef-H 1unit»=5c/m-50c/ra AiUlM AH ihil!
1> i '""iJ^' " '40 JO ^P rr«/s Fig. 36. Curve Showing Relation Between Time and Distance, Solid line shows total average time (seconds) on each trial of 27 rats in maze: dotted line the average total distance run (centi- meters). (After Hubbert, Jr. Animal Beh., 1914, IV, p. 60.)
lute time required to run the necessary distance (4.48 meters).
The peculiar shape of the learning curve, the initial abrupt slope, has led to much speculation. We cannot enter very profitably into a discussion of this subject be- cause at the present time our knowledge of learning curves is limited to a very few of those illustrating the simple sensory and motor acquisitions. It is probable that the curve of learning of very complicated acts would not have STUDY OF HABIT FORMATION this initial abrupt slope. Many other factors enter in to determine the contour of the learning curve: the number of trials necessary to learn the act as a whole; the absolute and relative rates of elimination of excess time and excess distance at each trial; the number of trials allowed each Errors CuEVEs Illustrating Rise of Sensoey Habits ^ R z= with reward alone P =: " punisliment alone The figures along the base line give the number of the series of trials necessary to form the habit. A series consisted of 20 trials.
day; the age of the animal; its sex, etc. Some of these factors will be considered in part on p. 235.
Curves of learning: sensory habits. — Even a superficial examination of the curve illustrating the acquisition of a sensory habit will show that the fall of the curve (i.e., the curve of error) is a very gradual one. In acquiring such 6 These curves are further discussed on p. 206.
CURVES OF LEARNING 201 habits where no initial preference is noted the number of wrong responses and the number of right responses ought to be approximately equal on the first few trials. Fig. 37 shows the growth of such habits (Hoge and Stocking). Gradually,' and usually very gradually, the number of right responses exceeds the number of wrong responses until finally all the responses are right and the habit is formed. Here again, the method of conducting the test alters the character of the curve, e.g., the number of tests made on a given day; whether punishment is given alone for wrong choice and food alone for right; or whether food is given for right response and punishment for wrong, etc. Many other factors are at hand which alter the general character of the learning curve, such as the amount and kinds of previous practice the animal may have had; differences in instinctive traits, such as boldness, fear; the amount and kinds of distrac- tion, etc.; the general method of attacking the problems; and finally the method of plotting the actual data may alter greatly the general appearance of the curve — telescop- ing of trials and time, etc. When one considers all of the factors which actually modify learning, and the different methods of treating the data on learning, it is not to be wondered at that there is no such thing as a characteristic and typical curve of learning. In conducting experiments, though, the various factors can be controlled; at least we can make the conditions of learning comparable for all the animals being experimented upon, and we can vary any one factor at will so as to bring out differences in the learn- ing ability of young animals and old, male and female, normal and defective, etc.
Motor habits in the human being. — Tests upon the acquisition of simple motor habits have been made upon human beings, e.g., learning to typewrite; to toss and catch balls, to shoot with the long bow, etc. The general features ■^ Some exceptions to this rule are to be noted even here. In Shepherd's work on the discrimination of colored foodstuffs by monkeys where wrong response was punished with a bitter taste, quinine, the habit arose very quickly (p. 347).
202 STUDY OF HABIT FORMATION of these curves of learning are quite similar to those of the animals. Usually there is an abrupt drop and then a somewhat gradual decline until the physiological limit is reached. Even here, though, the contour of the curve is dependent upon the complexity of the act, the method of treating the data, etc. The curves of Book, Swift and Bryan and Harter exhibit usually an abrupt descent. In the curve of archery which we have obtained the slope is very gradual. Improvement is slow at first and continues slow throughout. In an individual case there was no im- provement even at the end of the three hundredth trial. Some of these curves seem to exhibit what are called pla- teaux and breathing places, although not all of them do. These places — where the curve remains horizontal, showing neither gain nor loss — need explanation. It is probable that their explanation is to be found in the failure to con- trol the incentives (p. 204). No such plateaux or resting places are to be found in the curves illustrating the motor acquisitions of animals. When an animal has to work or remain hungry; to make a correct response at an alley or be punished, etc., the incentive may be said to remain at a maximum. The situation is clearly different when human beings are forced to learn to typewrite. The act is a very complex one and the stimuli leading to action are not compelling. If a man's food (reactions to sex stimuli, shelter, etc.) were dependent upon acquiring skill in a certain line — conditions which we can now control in the animals — such resting places and plateaux would in all probability disappear from his learning curves.
Comparative learning ability of different animals. — We have little accurate comparative data at present upon the differences in the learning ability of the different phyla. The statement is often made and is probably true that the monkey, with his superb eye-hand coordinations, can form habits of manipulation much more quickly than any of the other animals. From the work that has been pre- sented by several investigators it is probable that the raccoon stands next to the monkey, and after him the porcupine and then the cat. Probably the dog should LEARNING ABILITY 203 appear next, yet the rat is extremely apt in forming simple habits of manipulation. On the other hand, mammals like the guinea pig and the rabbit are clumsy in forming such habits. The bird class as a whole is notoriously poor in acquiring acts of skill.
On the other hand, the rat stands preeminent among the animals in his ability to thread a complicated maze, and the gray squirrel probably stands next to him in this ability. The monkey, however, acquires the maze habit readily. Birds, fish, reptiles, and amphibia form maze habits with very great slowness. Birds can, with great difficulty, learn complex mazes, but the amphibia, fishes, and reptiles have so far been known to form only very simple habits of this kind. Some experiments have been made for the purpose of comparing the relative learn- ing ability of blindfolded human adults, children, and blind white rats to learn a maze under approximately comparable conditions. While the conditions for rat and man differed sufficiently to make comparison difficult, it would seem that the absolute number of trials was less for the rats than for the human subjects. When one compares the rapidity of the elimination of excess time and distance (i.e., time over and above that required to traverse the maze when the habit is learned, and distance over that of the length of the true pathway from entrance to food) one finds that the human subjects make such eliminations much more slowly than do the animals. On the other hand, the ani- mals start with a much larger number of excess factors and show a very much greater variability all through their learning curves than do human subjects (Hicks and Carr). We have little comparative data on the relative ability of different varieties of animals to form sensory habits. Since the motor and instinctive equipments of the different ani- mal groups differ so enormously it is doubtful if we shall ever have any accurate means of grading animals with regard to the rapidity with which they can form motor habits. On the other hand, there seems to be no real reason why birds and monkeys, e.g., may not be accurately com- pared with respect to the rapidity with which they form, 204 STUDY OF HABIT FORMATION visual sensory habits, etc. The interpretation of such dif- ferences in learning ability (were such found to exist) would be as difficult as in the case of motor habits. Dif- ferences in learning might be due to differences in sen- sitivity to the stimuli rather than to differences in the rapidity with which the arcs are established (assuming that the response demanded is on a par in the two cases). Until our determinations of the range of effective stimuli, the relative stimulating effect of stimuli, etc., have been carried much farther, observed differences in learning among the different phyla do not mean very much.