Analysis of motor habits in birds. — The bird stands in marked contrast to the rat with respect to its behavior in the maze. In the first place it is very difficult to get the bird to learn a maze as complicated as that which the rat learns with ease. Consequently experiments have been car- ried out upon birds with much simpler mazes. While ex- perimental analysis is not so complete as in the case of the rat, it is quite evident that distance receptors play a very large role if not a preponderating one in the formation and execution of this habit. The bird is so constructed physio- logically that visual impulses are necessary for motor activ- ity of any kind. Hence it is impossible to exclude light, either by normal methods or operative methods, and get the animal to learn at all. This makes it extremely difficult to test the actual role of vision. On the other hand, birds show almost no trace of sensitivity to olfactory impressions (p. 403). Hence it is highly improbable that olfactory arcs function in this habit. In general it seems safe to conclude that analysis will show that visual arcs, auditory arcs, and kingesthetic arcs are all functional. It is perfectly clear from all the work upon the formation of maze habits that birds never become the automata that rats become. ^^ ^* In the more recent work of Bogardus and Henke it is claimed that the number of contacts (touches with snout) made at the corners by untrained rats whose vibrissse were removed is directly in proportion to the number of errors made in learning the maze.
^^ It is interesting to note that birds apparently never dash into partitions in work on the shortened and lengthened forms of the 220 STUDY OF HABIT FORMATION Analysis of sensory habits in mammals. — For purposes of analysis we will take some sensory habits which the rat forms when confronted with monochromatic lights. The method of conducting such tests is given on p. 61, the apparatus on p. 71. When this animal was stimulated simultaneously by the lights red, 655 mm, and green, 505/^a'; a habit developed. In this case the animal re- sponded positively to red and negatively to green. Ten to twenty trials per day were given. After 555 trials the animal became automatic in his responses. He was then considered ready for control tests. In the training series the energy of the two monochromatic bands was not known. As the results of the experiment stand we are not able to define the stimidus which led to response. The results might be interpreted in three ways: (1) the animal may have been responding on the basis of wave-length difference; (2) upon the basis of intensity difference; (3) only one stimulus was effective — the other stimulus lying wholly outside the sen- sory range of the animal. If the latter supposition were true, it remained to find out which light was stimulating the animal.^® Control tests to decide among these suppositions maze, or when obstacles are placed in the way. Some experiments were made in the Hopkins Laboratory (by Dr. Ulrich) to test whether pigeons, after hopping ( and " half flying " ) to a food box 1 meter away, would hop short or over the food, depending upon whether the box had been brought nearer to the starting point or moved farther away. As is well shown by Richardson, rats, when trained to jump, e.g., 20 inches for food, are unable to adjust for new distances without training. If, e.g., they are trained to jump 20 inches and the box is brought 12 inches forward, the animal jumps to the old distance, 20 inches. He jumps short if the food box is moved farther away, e.g.. to 28 inches. The pigeon, on the other hand, makes these adjustments with equal readiness regardless of the distance. Tests were made within the limits of.75 meter on either side of the mid position. All of the work so fnr carried out on the rat's motor habits show the almost complete dominance of the proprio- ceptive impulses over the skeletal muscles. In contrast to this it appears that the skeletal muscles in the pigeon are dominated by the impulses from the visual receptors.
^® It should be made clear that sensory habits may arise where two stimuli are present when only one stimulus is effective. Fur- thermore, the non-effective stimulus may be either positive or negative.
ANALYSIS OF SENSORY HABITS 221 were begun with rather striking results. A diary record of the control tests follows: March 7th. Red (positive) full intensity. Green cut to 11% of full intensity. 15 trials. 93% correct.
March 8th. Green full intensity. Red cut to 11% of full intensity. 15 trials. 100% correct.
March 9th. Red full intensity. Green cut to 2.7% of full intensity. 11 trials. 40% correct. Complete loss of habit. Then immedi- ately afterwards 9 trials were given with both stimuli full in- tensity. 100% of correct choices.
March 10th. Conditions as beginning of test on previous day. 15 trials.. 53% correct. Complete loss of habit. 5 trials wer^ then given with both stimuli at full intensity. 100% correct. From March 11th to March 16th certain experiments were tried to determine the effect of external illumination upon the dis- crimination. These experiments were so unsatisfactory that mention of the results is omitted. By June 5th the animals were again trained to react perfectly to red and green. Further control experiments to determine the nature of the effective stimulus were then undertaken.
June 5th. 8 normal trials were first given. 100% correct. Then red beam was cut out. Green left at full intensity. Went to red side (darkness) on every trial (10 trials).
June 6th. 8 normal trials. 100% correct. Green was cut out and red left at full intensity. Right position hahit developed im- mediately. Discrimination lost. Then 4 normal trials were given. 100% correct.
June 1th. 8 normal trials. 100% correct. Green out, red at full intensity. 14 trials given. Right position hahit developed im- mediately. 4 normal trials were then given. 100% correct.
J-mie 10th. 8 normal trials were given. 100% correct. Red cut out, green full intensity. Went to red side (darkness) on all 15 trials. 2 normal trials were next given. Both correct.
June 11th. 6 normal trials given. 100% correct. Green out, red full intensity. Left position hahit developed immediately. (9 trials.) Then 10 normal trials were given. 90% correct.
June 12th. 6 normal trials given. 100% correct. Both lights cut out for 15 trials. Right position hahit developed. 4 normal trials were given. 100% correct.^' These results seem to show perfectly that the green was the effective stimulus and that the red had no stimulating value whatsoever. The animal tvas reacting against a lighted compartment. One might decrease the intensity of the green light enormously and eliminate the red altogether, 17 In all of the control tests food might be had regardless of the stimulus reacted to. Food was always kept in both compartments even during the training series.
222 STUDY OF HABIT FORMATION without changing the accuracy of the responses. When, however, the intensity of the green approached the threshold, as on March 9th, a disturbance immediately manifested itself. Whenever the green was cut out the animal behaved exactly as though no light stimulus were present (contrast tests of June 11th with those of June 12th). These results seem to show that we are dealing with a defect in the receptor. Summarizing the results we find: (1) that the green was the effective stimulus; (2) that the red had no stimulating effect. The results harmon- ize with the hypothesis that the red chosen lay outside of the animal's spectrum.
It follows from this experiment that supposition (3) was correct. But from our Results we are still unable to define the nature of the stimulus — we do not know from them whether the animal was reacting on the basis merely of its intensity or whether its wave-length was a factor also. We might easily have carried the experiment farther and tested whether the animal would have responded as readily to a white light as to the green. Instead, hoAvever, we began experiments upon two other rats, using stimuli both of which lay in the sensory range of the animal, viz., yellow, 595mm? and blue, 418MM' ^^ this case No. I re- sponded positively to blue and No. II, positively to yellow. One interesting fact was brought out very clearly: when both stimuli were in the field the animal failed to form the habit within the time limits of the test. In order to hasten the growth of the habit only one stimulus (the posi- tive in each case) was used. As might have been predicted from our tests on red and green, the habit arose readily. Then (with energy of both stimuli known) we introduced the negative stimulus. As fast as possible, without breaking down the habit, the intensity of the negative stimulus was increased. The following table shows the results of this test. Column 1 gives the number of days the test endured, and column 2 shows the angular opening of the sector. An opening of 10° admits a beam of very low intensity. At 360° the full intensity of the beam was utilized (i.e., the two stimuli possessed equal energy, but did not contain ANALYSIS OF SENSORY HABITS equal stimulating value). Column 3 gives the percentages of correct responses (in 10 — 20 trials).
Shows that This shows almost perfect animal was reacting to inhabit at equal energy.
tensity and that hlus at the same stimulating effect as the yellow at full intensity.
A difference in the behavior of the two animals becomes at once apparent. No. I persisted in the habit when both lights possessed considerable intensity. No. II lost the habit when the angular opening on the stimulus was set at 20 — 30° (when the energy of the blue was 1/18 — 1/12 that of the yellow). The fact that the habit dis- appeared when the negative color bore 1/18 — 1/12 the 224 STUDY OF HABIT FORMATION energy of the yellow furnishes almost conclusive evidence that supposition 2 above is correct: This animal was re- sponding on the basis of the difference in intensity, i.e., when both stimuli were present, positively to the more in- tense; as soon, however, as the negative stimulus possessed the same stimulating effect, the basis for response disap- peared. The results of the experiments on No. I cannot be interpreted except by the use of control tests. "Without entering into the details of these it was found that the stimulating value of blue for this "animal was very high and that in order to get an equal stimulating effect from the yellow one must make its energy many times greater than that of the blue. When the stimuli were so manipulated that both possessed equal stimulating value, the habit disappeared for animal No. II just as it did for No. I. It was further shown that when the negative color was made to possess higher stimulating value than the positive, the response was reversed, i.e., the animal then reacted posi- tively to the light which it had formerly reacted against. The same condition obtained when a strong white light was presented in place of the color reacted against — the animal responded positively to the white light. Furthermore the animals responded positively to the negative stimulus when it was given alone. The conclusion that the stimulus to which the animals were responding was an intensity difference, and that a difference in wave-length possessed no stimulating value, seems valid (at least so far as concerns the responses to yellow-blue).
Analysis of the delayed reaction. — On p. 105 we dis- cussed the method by means of which the delayed response is obtained. The situation in brief is as follows: while the animal is held in restraint its visual receptors are stimu- lated by a light which may appear in any one of three boxes, left, in front, or at the right. The animal has previously been trained to go to the lighted compartment for food. After training the animal is not released until after the light has been turned out. Upon release it must go to the box in which the light last appeared. The experimenter may, at will, allow an appreciable time to elapse between ANALYSIS OF DELAYED REACTION 225 turning out the light and releasing the animal. Let us glance at first at the maximum delay which can be obtained in different types of animals. Hunter has tested rats, dogs, raccoons, and children. We may observe his results more conveniently in the form of a table.
Subject Maximum delay Rats 10 seconds Dogs 5 minutes Raccoons 25 seconds Children 25 minutes The problem here seemingly is to determine the mech- anism by means of which correct response may be made to a stimulus which is no longer present. We shall see later that this problem may be restated so as to bring it in line with the other types of response which can be explained in terms of the immediate functioning of neural arcs. Prior to experimental analysis there appear to be several possi- bilities of sensory stimulation which might lead to success- ful response after the period of delay: (1) the experimenter himself might give the stimulus in one of several ways, such as shifting his position, changing the breathing, etc.; (2) the order in the tests might not have been sufficiently varied; (3) the box in which the light last appeared might have risen in temperature; (4) the increase in the tem- perature of the light box might have increased the diffusion of the gaseous particles which arouse the olfactory recep- tors;^^ (5) the after-glow of the recently lighted lamp might have been responded to; (6) the after-images of the light might lead to response. These possibilities of sensory stimulation are not exhaustive. They are merely illustrative of the many which must be considered. We cannot take up in detail the various methods by means of which such possible stimuli were eliminated. In general they were found to be inoperative and the only extra- organic factor was the presence of the light in the box. Normal methods of elimination were used in reaching these results.
On what basis, then, is it possible to react to the position — ^^ Neither (3) nor (4) was sufficiently controlled by Hunter, 226 STUDY OF HABIT FORMATION of a stimulus which was present for a time and then dis- appeared? Obviously one possibility of explanation lies in the fact that the light when present releases a definite set of arcs which function the one after the other, as in every habit. It is easy to see that the light itself forces into action a definite set of arcs, which result in bodily orientation, such as turning the head and body to the light, crouching, maintaining that attitude, and the inhibition of other movements. The movements initiated by the light in turn arouse kinsesthetic and organic impulses which in turn inhibit movement, etc. The auditory and visual stimuli offered by releasing the animal immediately initiate the walking and the running movement. Since the animal is already oriented, naturally the movements take him to the box which was last lighted. Once there the habit of entering the box, eating the food, etc., is re-aroused. We might call this the first method by means of which the light could exert a determining effect although absent (i.e., through initiating the reflex arcs which function continu- ously and serially from the moment the light is presented until the animal arrives at the food).
In more highly organized subjects in which speech habits are adequate for all ordinary situations, the light may arouse a group of throat movements long since correlated with definite modes of response (e.g., the words ''go to the box in front when door opens "). With such a highly complex system of habits there is no need for the arc origi- nally aroused continuously to exert its function as in the first case considered. The animal with speech habits may begin to play or engage in any number of activities with- out producing a breakdown in the response when the door is opened. At the moment of release the visual stimulus of the opening door re-arouses the vocal impulses which lead to their natural sequences, viz., movements straight forward to the box in front. We will call this method two.
In the less highly organized animals the periods of delay are wholly determined bj^ the length of time it can retain bodily orientation. This will depend upon its organization (i.e., sensitivity to forms of stimulus other than kinassthetic) ANALYSIS OF DELAYED REACTION 227 and to the actual intensity of the sensory stimuli present (noises, changes in the light, bites of insects, etc.).
Observation of the various animals seems to show that the rat, in order to respond correctly, must maintain actual gross bodily orientation. It cannot move about after the light is turned out. If this overt orientation is lost for a moment any one of the three boxes may be entered. This statement holds true for all of the correct responses of the rat and the dog, and for most of those of the raccoon. On the other hand, several of the responses of the raccoons and all of the responses of the children offered a mode of behavior which cannot be explained in the terms of main- tained bodily attitude. It is possible for the raccoons and the children to lose all bodily orientation and then to re- spond correctly when the door is opened. It is possible, also, in the case of the raccoons, to introduce loud noises, etc., without disturbing the response. The case of the long delays of the children may be explained by the use of the language habit (p. 328). The behavior of the raccoons can- not be explained in this way since language habits are not present.^*^ There is no known mechanism of response which might account for this. It thus seems best to reserve our attempt at explanation.
Summary. — A habit is thus seen to involve always a series of definite arcs beginning in a receptor and ending in an effector. In sensory habits we isolate and define the stimulus which will initiate the impulses in these arcs. In motor habits by the interposing of certain conditions (set- ting of certain problems) we force upon the animal certain ^® Both Carr and Hunter would insist that there are other possi- bilities of explanation, i.e., that certain other proprio- and entero- ceptive processes might function in the same way that we have assumed that language habits function (i.e., so that continuous bodily orientation is not necessary). Were Hunter a less careful investigator, we might think that there was an actual error in observation. Most of the raccoons were admittedly making the cor- rect reaction upon the basis of maintained bodily orientation. What more simple than to suppose that those animals which were not disturbed by loss of bodily orientation had learned to make correct, response upon the basis of olfactory stimuli which were not suffi- ciently controlled in Hunter's work?
228 STUDY OF HABIT FORMATION types of synthesis of action of the effectors, temporally space their action, note the rapidity of growth of the combination, etc. It must not be forgotten, though, that even in motor habits the receptors are equally involved. If we examine such habits from this standpoint we find that: ( 1 ) all receptors may be involved in a given habit at ever}' stage, i.e., there may never be any complete reduc- tion to kina^sthesis; (2) in some animals the learning of certain habits requires the cooperation of all receptors, but the execution of these habits after learning requires the functioning mainly of the proprio-ceptors; (3) in other animals, both the learning and the final execution may be mediated by the proprio-ceptive system.
VI. Efficiency of training methods: motor habits. — In the animal world w^e have a better opportunity for testing the various methods of training than we have in the schools, for the human child. The subject of efficiency as a whole has not yet received the attention which it deserves. Our records are fairly complete at least for one species of animal — the white rat — so far as his motor habits are concerned. The various questions to which we can re- turn answers may be stated as follows: (1) what number of trials per day should be given an animal while it is ac- quiring one motor habit at a time, if our object is to secure a perfect habit with a minimum number of trials? (2) What is the effect upon the retention of the habit of this method as compared with a method which requires a larger number of trials? (3) Will the same method which has happened to serve the purpose of (1) serve equally as well when the animal is acquiring simultaneously more than one motor habit, e.g., three such habits?
In order to test these questions, Ulrich (whose report is as yet unpublished) has made a three-years' study upon the white rat. He allowed several groups of animals to learn box II, each group being given a different number of trials per day,-° all other conditions except the one ^^ By trial is meant the complete reaction from time of release in the confining cage until the animal has solved the problem and eaten the food. If more than one trial is given the animil is allowed to EFFICIENCY OF TRAINING METHODS 229 specified being kept constant. The groups and the trials given are as follows: One-in-one method. Three-in-one method. Five-in-one method. One-in-two method. One-in-three method.
Group I. One trial per day.
" II. Three trials per day.
" III. Five trials per day.
" IV. One trial every other day.
" V. One trial in three days.
In Figs. 45 and 46 are given the distribution curves of the learning of these separate groups, both from the stand- BY DAYS R W Fm n TRIAL IL DISTRIBUTION CURVE LATCH BOX TRIALS A M^ TRIALS n MALES 0 FEMALES BY TRIALS H Wi rrm n TRIAL THREE TRIALS FIVE TRIALS 30 ^5 50 II III point of the number of trials given, and from that of the number of days required to learn the problem. These curves may be easily understood, e.g., the distribution curve for Group I is shown in Fig. 45, I. It shows that 2 animals learned the problem after 11 trials; 2 after 13 trials, 3 after 14, etc. All 16 of the animals had learned the prob- lems by the end of the twenty- fourth trial. If now we pass to Group II (Fig. 45, II) we see that the first ani- mal learning the problem required 21 trials, while the last eat but sparingly. Food is then immediately given on the second and succeeding trials. At the end of the daily allotment of trials, the animal is allowed to eat its full ration of food in the box.
230 STUDY OF HABIT FORMATION animal required 36 trials. Similarly, in Group III (Fig. 45, III) the first animal to learn the problem required 30 trials; the last animal 50 trials. Group IV (Fig. 46, IV) gives 9 trials for the first animal to learn, and 22 for the DISTEIBUTION CURVE LATCH BOX BY TRIALS □ MALES n^ Ja FEMALES - ONE TRIAL '^gti ST V EVERY TWO DAYS ONE TRIAL Q n li ifi is 90 22 ^v EVERY OTHER DAY R W Fm n m last. Group V (Fig. 46, V) showed that the first animal learned in 9 trials and the last in 17.
We find in this experiment the answer to question (1) raised above. Within certain limits, which have not been determined yet, it is far more economical, so far as the number of trials is concerned, to give a few rather than a large number of trials per day. From the distribution curves it would seem that the one-in-three method is cer- tainly far more economical than the three-in-one or the five-in-one; indeed, it seems the most advantageous method of all. As regards our second question, i.e., the effect of a different method on the retention of the habit, the re- sults seem to show that no one method has any distinct advantage over the others. The five-in-one method and the one-in-two method gave the poorest results. After each animal had been trained to the problem it was not tested again until after 60 days had elapsed. During the non-training period the animals were forced to run daily to the end of a long straight-away to obtain food. They were allowed to eat the same amount of food as during the training period. This method kept them in training (pre- vented the laying on of fat, the onset of sluggishness,^ wildness, etc.). At the end of 60 days each animal was tried EFFICIENCY OF TRAINING METHODS 231 for 2 days on the original problem. For convenience are given the average of each group on the last trial of the training series, and the average of the first trial of the retention series.
AvEBAGE Time of Last Trial on Average Time on First Regular Practice Trial in Retention Test No. animals Minutes Minutes Av.
Av.
Av. M.V.
When we consider the number of" days required to form the habit, we find an entirely different situation. The distribution curves by days are given on p. 229, Fig. 45. The curve of Group I is repeated (Fig. 45, C) for the sake of convenience. In plotting these curves the num- ber of days required to learn the problem replaced the number of trials in the early curves, e.g., in the graph showing the five-in-one method (Fig. 45, A) it will be seen that 3 animals required 6 days; 2, 7 days; 1, 8 days; 3, 9 days; 2, 10 days. When all the curves have been considered it appears that within the limits of this experiment the larger number of trials given per day, the less the number of days required for learning.