IV. The rise of habits when incentives are controlled. — It can hardly be said that behavior has yet developed any very complete methods of controlling incentives. Pun- ishment in one form or another for wrong reactions and food for right response have been used as controls since animals have been trained. In training for circus work the two incentives have usually been combined. In such work the whip has been largely employed and still is the mainstay of the professional trainer. With them, how- ever, the whip serves the double purpose, — that of punish- ing when wrong response is made and that of giving the signals or cues to which the animals respond.
Apparently Elmer Gates ^ was the first person to employ the method of induction shocks. Since his experiment is not very well known and has not been largely quoted, we give a brief abstract of his experiment upon the dog. On the floor of a hall were placed colored metal squares. All of the squares, except those of a given color, were connected with an induction coil. The dog, in passing down the hall, had to jump from one square to another. By jumping only upon same colored squares he could pass safely through the hall without receiving punishment.
Yerkes has very greatly extended the use of Gates' method (p. 59). His work on behavior, coupled with that of Martin on the standardization of the Porter induc- torium, gives us our first approach to a real control of incentives. Some of the difficulties in the use of the CONTROL OF INCENTIVES 205 method have already been discussed. As we have pointed out, no scientific control of the other incentives seems pos- sible, and yet certain steps in that direction may be taken with food (p. 58). A good practical method, as we have already remarked, in regard to the use of food, is that of allowing the animal to eat sparingly of mixed necessary food after each trial, and then either to satiety at the end of the daily routine of tests or else for a specified length of time. In the discussion of the experiments which follows some such method of food control, either during the tests or after them, is always used. Even w^hen punishment is used alone, the control of food is a necessary factor.
Habits formed without the use of punishment. — Since the punishment method has come into favor very slowly and has some serious limitations, we find that much of the work upon both sensory and motor habits mentioned in the literature on behavior has been carried on in the ab- sence of punishment. Certain other incentives, of which W'C cannot make separate mention, have been used, such as escape from narrow quarters, escape from cold, from bright light, etc. Examples of habit formation under such conditions have been given on p. 193 and p. 195.
With punishment as the sole incentive. — It would seem to be established by experimentation upon sensory habits (1) that when the difference between the two di- rective stimuli is great, the rapidity of learning increases as the strength of the electric stimulus (punishment) in- creases from the threshold of stimulation up to the point where the current becomes dangerous to the tissue. This law may be looked upon as holding for what we may call easy habits. (2) Where the habit arises with some diffi- culty, i.e., where the intensity difference between the two stimuli is less, there seems to be at present no clear rela- tionship between the number of trials necessary to learn the habit and the strength of the punishment. In general it will probably be found that there is an optimal strength of stimulus for every difference in intensity between the two stimuli upon the basis of which a habit is to be estab- lished. Suppose, e.g., we are working with two lights.
206 STUDY OF HABIT FORMATION The difference in intensity maintained may be 10 c.p., 5 c.p., or 1 c.p. Only experimentation could decide what strength of stimulus should be used with any one of these differences in intensity. Yerkes has stated the general law as follows: '' As difficultness of visual discrimination in- creases, that strength of electrical stimulus which is most favorable to habit formation approaches the threshold. The easier the habit the stronger that stimulus which most quickly forces its acquisition; the more difficult the habit the weaker the stimulus which most quickly forces its acquisition. ' ' With punishment for wrong response and food for right. — So far only one set of results has been obtained upon this subject. It would seem from work upon the white rat, where a habit of responding positively to the brighter of two lights (2 c.p. and 16 c.p.) was desired, that punishment for failure and food for right response give the most advantageous situation for the rise of the habit. Fig. 37, p. 200, shows these results.^ This work needs confirming by the use of a much larger number of animals than was employed.
V. Analysis of reflexes involved in habit. — So far in our discussion we have spoken of habits as though they consisted of only a few well-directed and definite muscular movements; as though the complex stimulus (problem box, maze, etc.) called forth on first presentation a large number of random acts, among which acts were to be found those necessary to obtain food (to escape punishment, etc.), and on later presentations fewer and fewer excess movements. Finally the stimulus arouses the few neces- sary acts in serial order. Such an illustration fits very well those cases of habit where the movements are simple and definite in character, such as pulling a string, knock- ing up a latch, etc. But certain habits are enormously complex. Consider, e.g., the learning of the Hampton ^ A reference to this figure shows that where no punishment was used (R) the habit was not perfected within the time limits of the experiment; where punishment alone was used 610 trials were re- quired (P); where reward and punishment were combined only 530 t-rials w^ere necessary (R.P).
ANALYSIS OF MOTOR HABITS 207 Court maze, with its many possibilities of false turns. It is evident that in order to thread this maze as a whole it is necessary for the animal to have a series of simple habits (i.e., one wherever an error of turn is possible) of the character we have just considered. It becomes necessary in the prediction and control of behavior to make an analysis both of the simple types of habits and of the most complex. We should like to be able to state, were such possibilities of analysis ideal, (1) the number, location, and serial order of functioning of the reflex arcs from the initial response until food is reached; and (2) the stimulus which releases each arc. Furthermore there is some evi- dence that these systems of arcs change somewhat as the habit passes from the initial stage to the final or automatic stage. We may possibly make this latter point clearer by saying that in the early stages visual, auditory, tactual, etc., stimuli are all functioning (necessary for the release of the appropriate motor impulse), whereas in the later stage only the kingesthetic ^° are functional. There is less and less resistance in the synapses of the kineesthetic arcs and probable actual blockage in the synapses of the arcs of the distance receptors. The movement of muscle 1 (it will be remembered that such muscles contain sensory neurons as well as motor) sets up a sensory stimulus which, upon entrance into the proper place in the central nervous system, releases a motor impulse which brings about move- ment 2. This routine continues until the whole train of movements has been run off. It is thus seen that the muscle is at one and the same time the ejfector of the arc in operation and the receptor of the arc which is next to function. The notion that every habit is of this character is more or less current. That this need not be true may easily be understood when we consider that the stimulus to movement 1 may be olfactory, visual, etc., but move- ment 1, however released, may set up an auditory stimulus ^^ By kinsesthetic we mean the sensory impulse aroused by the contraction of the muscle. When the muscle contracts the muscle spindles (sensory nerve endings in muscles) are chemically stimu- lated. The endings in the tendons are aroused at the same time.
208 STUDY OF HABIT FORMATION (the animal may run over a trap which causes a door to drop, — auditory but no visual stimulus). The succeeding movement, movement 2, may be released now through at least two avenues: (a) the sensory impulse coming from the muscles used in springing the trap; (h) through audi- tory impulse aroused by the impact of the falling door. As we watch the behavior of a normal animal while executing a customary act we find ourselves wholly unable to state what impulses are functioning. Special types of experi- mentation are necessary for analysis and for tracing the history of a habit. Suppose (to take a very simple case) we train an animal to open box III, p. 95, with all sense organs intact. After training we may begin analysis: the auditory impulse may be blocked by eliminating the sense organ (operative method), or it may not be aroused at all, which may be effected by padding the door (normal method). We may find that the animal can still execute movement 2, i.e., run to the door, enter it, etc. On the other hand, we may begin our analysis in the learning stage by forcing the animal to learn with certain stimuli absent. Actual tests show that learning may proceed with the auditory stimulus absent, or in an animal with auditory organ destroyed; but the character of the learning curve may or may not be altered, depending upon the animal used. In the laboratory in order to evaluate results we have proceeded upon the assumptions (1) if the trained animal can as quickly and with as little excess effort exe- cute the habit under consideration after the auditory stimulus has been removed through normal or operative methods, we are justified in concluding that movement 1 gives the impulse which releases movement 2. On the other hand, (2), if, after the loss of the auditory stimulus, there is a serious break in the habit, we must conclude that the auditory stimulus functioned at least along with the kinesthetic. If, (3) after the loss of the auditory stimulus, the animal can relearn the problem, we conclude that the kinresthetic impulse may come to function alone, i.e., be completely substituted for the auditory. If the animal (4) deprived of the auditory stimulus, by whatever means, can ANALYSIS OF MOTOR HABITS 209 learn the problem in as short a time and with as little excess effort as the animal not deprived of the auditory stimulus, we can conclude that the stimulus of movement 2 is kinesthetic; whereas ( 5 ) if learning is more difficult, we assume that the auditory stimulus is functioning in normal learning. The right of making such conclusions from the premises stated has been questioned somewhat in the literature. Some authors would have us believe that the animals have a nervous system so delicately constructed that there can be an immediate shift in the arcs; e.g., if in cases where distance receptors are functioning in a well- established habit we block impulses over those avenues, they maintain that there may be an immediate shift to kinaes- thetic impulses with no break in the habit. Such a view must have some factual support before it need be seriously considered. This simple illustration makes it obvious that habit analysis forces us to resort to methods of elimination of stimuli. This may be considered under two heads; normal methods and operative methods.
Normal and operative methods of eliminating sensory impulses. — Since many investigators object to removing the sense organs in animals, various normal methods have been devised. The simplest way of excluding visual im- pulses, e.g., is to work in a dark room; to exclude audi- tory impulses, to work with noiseless instruments; to exclude olfactory impulses, thoroughly to cleanse and wash the apparatus after each experiment. Animals may be trained with these stimuli present until the habit becomes automatic. These stimuli may then be removed and the effect noted upon the execution of the act. On the other hand, we may force the animal to learii with several of such stimuli excluded. The objections to the use of normal methods are many and serious. We have proceeded usually upon the assumption that the sensitivity of the receptor of a given animal to a particular form of stimulation is the same as O'ar own.
The error of such a procedure appears clearly when we consider the olfactory sensitivity of many animals. "Wash- ing and even boiling the maze may not conclusively elimi- 210 STUDY OF HABIT FORMATION nate the olfactory stimulation, i.e., such as trails, etc., for certain animal forms. On the other hand, if we wish to ex- clude visual stimulation by working in a dark room, we eliminate, pari-passu, the experimenter's possibility of ob- serving the behavior of the animal.
The safest method, in the case of distance receptors, is to remove the organ. When this is done under proper operative condition, most animals are abnormal only with respect to those functions for which we desire abnormal- ity.^^ The normal method can give us only indications of the data, and can never be said to furnish scientific proof of their existence.^ ^ It must be obvious that while we are principally concerned with the analysis of motor arcs, methods of eliminating undesirable sensory stimuli must be constantly employed in the establishment of sensory habits, and habits of the delayed reaction type. On p. 63 we mention the necessity for control of this kind. Were it possible, in wor-k upon certain aspects of vision, to use animals in which only the visual receptors with the neces- sary effectors were functioning, our results could be ob- tained with far greater accuracy than at present.
Some results of experimental analysis of motor habits in mammals. — The simple case of habit analysis discussed on p. 208 enables us to examine in detail a much more complicated case — that which the rat forms in the Hampton Court maze (p. 103). One is struck immediately by the peculiar rapidity and neatness with which the rat learns to thread this maze. Fig. 38 gives the learning curve of 4 normal rats. It will be seen that on the average 29 minutes are required for a first successful trip and that by the end of the thirtieth trial this time has been reduced to approximately 30 seconds. As we see the rat threading this complicated maze with never a false step, we get no ^^ Birds offer an exception to this observation in so far as visual receptors are concerned.
^^ For obvious reasons we cannot enter into a complete treatment of the operative methods and technique. The reader is referred to the author's paper. Psychological Monograph No. 33. and to that of Florence Richardson, Psychological Monograph No. ^8. In order to avoid all criticisms it is best to have such operations made under medical, supervision, and where possible by physicians themselves.
ANALYSIS OF MOTOR HABITS indication of the system of arcs involved in the execution of the act as a whole. The sensory avenues may be very numerous. On early trials the animals cover every square inch of space, passing in and out of the cuts de sac, coming back to the entrance, advancing, returning again, etc. Soon the animal passes by the openings into the cals de sac at A z P TRIALS Fig. 38. Showing the Normal Process of Learning the Maze Based upon four normal male rats, about one year of age. In all of the following curves one division of the ordinate represents one minute, while one division of the abscissa represents one trial.
and B (Fig. 25). Gradually those further along are safely passed and finally the whole run is made without error. It is obvious that learning to avoid any one of these cids de sac may be looked upon, for convenience, as involving the formation of a habit. Since there are several culs de sac it is clear that we have several unit habits " chained " to- gether. After all these units are formed and concatenated, " 212 STUDY OF HABIT FORMATION learning is said to be complete. ^^ Two questions arise: (1) What process is involved in forming any one of these unit habits? We have the same condition at each cul de sac that we have in the simple habit considered above, random movements giving place finally to the right movement, i.e., making the proper turn or keeping straight ahead. We have already spoken of deferring this discussion to Chapter VII. The answer to these questions does not involve an analysis of the arcs employed. It gives a reason in mechanical terms for the more or less complete elimina- tion of all random movements. (2) What systems of recep- tors are involved in the various parts of the maze — the straight-aways as opposed to the turns, the culs de sac, etc.? We can conceive of the animal, deprived of all its distance receptors, walking or running from the entrance to the food box. Each movement executed would arouse new contact and kinesthetic impulses which in turn would release the succeeding movement. It is difficult to conceive of such a defective animal learning the maze de novo. When an animal, after leaving 0, arrives at A, the first cid de sac, the sensory situation becomes more complex than in the alleys. There are differences in the optical stimulation offered, differences in the olfactory, and dif- ferences in contact, since the vibriss^e on the left are in contact with the walls of the maze, while those on the right are not in contact. Furthermore, there are differences in the direction and intensity of the air currents and in temperature. Likewise the auditory stimulation is differ- ent— there are differences in the reflection of the sound of the animal's own footsteps, etc. — and finally kingesthetic impulses change. The possibility of a change in the form of stimulation at the entrance of the cid de sac as con- trasted with that in the alleys is very great. The question arises: are all of these impulses necessary for the release of the movement which carries the animal safely past the cul de sacf If all are not necessary, which ones are? Furthermore, all (or many) may be necessary in the early ^' On p. 261 we show that concatenation is not a part of the process; it appears ipso facto when errors are eliminated.
ANALYSIS OF MOTOR HABITS stages of the habit, but only the kinaesthetic in the auto- matic stage. These questions we have attempted to answer by experimental methods. (1) We first trained several animals in daylight and then tested them in darkness. The appended table shows that the animals executed the habit in the dark in as short a time as in the light. This is Maze in Light Rats I II III IV V Trials Min.
Min.
Min.
Min.
Min.
1 2 3 4 5 Maze in Dark 1 2 3 4 5 Averages for the Five Trials in the Light and the Five in Dark- ness— Each Rat Separately In light. In dark.
merely indicative of the fact that after the habit is learned it can be executed automatically without the use of the visual receptors. (2) We then forced several rats to learn the maze in the dark. They likewise learned in normal time. (Fig. 39.) Could we be sure that vision was com- pletely excluded in this learning process, we should have good evidence that vision is not necessary to the forma- tion of this habit. In order completely to control our work, we eliminated vision by removing the sense organs in two groups of animals, one (3) wholly untrained in the maze and (4) the other trained to it. When brought to test the untrained animals learned the maze as rapidly as did the normal animals. (Fig. 40.) The trained group of blind STUDY OF HABIT FORMATION Fig. 39. Leaening of Maze in Darkness by Normal Rats animals ran the maze with only a slight loss in efficiency which might have been predicted when we recall the fact MlNUT»« 8 TRIALS Fig. 40. Learning of Maze by Blind Rats that after the animals had been operated on, they were not again tested in the maze until 40 days had elapsed; con- sequently there was a slight loss in retention. (Fig. 41.)
ANALYSIS OF MOTOR HABITS It would seem that visual receptors are not involved in the formation of the maze habit at any stage.
We made no attempt to control the factor of smell by normal methods, but (5) removed the olfactory bulb in H-H-H Fig. 41. Showing the Effect of the Loss of Vision upon Rats Trained to the Maze with Sense Organs Intact several rats and forced them to learn the maze in the light. The learning time was normal in every sense and the learn- ing curve was identical with that of the normal group. (Fig. 42.) The anosmic animals were then (6) forced niNUTCS Fig. 42. Learning of Maze by Anosmic Rats to run the maze in darkness. No disturbance of the habit was noticeable. (7) Animals with defective auditory re- ceptors (drum membrane and auditory ossicles removed) were next tried. The learning was normal. (Fig. 43.) (8) Trained normal and defective animals with vibrissag removed were next tried. No disturbance was noted if the animals were kept in the cage for a few days after the removal of the vibrissse. (9) The maze is learned by ani- mals without vibrissag as readily as by those with them. (Fig. 44.) The effect on trained animals of (10) altering the direction and intensity of air currents and (11) of the temperature of the air at the culs de sac was without effect.
STUDY OF HABIT FORMATION (12) The effect of producing local anaesthesis (ethylcMo- ride) of the soles of the feet, snout, etc., was likewise nil. We seem forced to conclude that in the rat kinsesthetic MitiuTea Fig. 43. Leaening in Partially Deaf Rats arcs are the only ones necessary at any stage of the forma- tion of this habit. This does not mean that the distance receptors and the tactual are not being stimulated while Fig. 44. Leaening in Animals Whose Vibrissa Had Been Removed the animal passes through the entrance point to the food. It has later been shown that conditions can be so arranged in the maze that arcs other than kinesthetic are necessary to the formation of the habit. It may be objected that ANALYSIS OF MOTOR HABITS 217 the evidence so far advanced is purely negative, obtained by the exclusion of data other than kinsesthetic, conse- quently that we have no real proof of a positive character to show that the kingesthetic arcs really function in the way in which we maintain. Very definite and positive proof that our main thesis is right has been furnished by taking a maze that could be shortened or lengthened without alter- ing the relations of the turns. Fig. 26, p. 104, shows this form. Two groups of animals were trained upon this maze, one group in its shortened and the other in its lengthened form. Animals that had been trained with the lengthened form were tried with the shortened form. Q in the diagram is thus made to take the position Q'. Under this change the animals released at 0 gathered speed as they made the various turns and then ran with full strength into the wall at 1. With the kinsesthetic arcs functioning, the extra- organically aroused impulses (visual stimulation of walls, etc.) were unable to inhibit the customary movements. When the animals that had learned the shortened form were tested in the lengthened form, they tried to turn in at the places where formerly they had turned in. Again the evi- dence is conclusive. The turns had been made hitherto solely on the basis of proprio-ceptive impulses. Changing to the lengthened form brought no immediate change in these impulses, so the animal tried to turn into the solid wall. As soon, however, as the failure occurred, the old chain broke and the new habit was formed — i.e., of going forward several inches and then turning. Many other ex- periments furnished the needed positive proof that was lacking in the elimination method.
One objection has been raised to the resolving of the maze habit process into the functioning of a serially chained kinesthetic arc system. It may be stated as fol- lows: in man and in animals kinesthetic arcs function in perfect habits until some disturbance occurs, i.e., until rival impulses coming in over other receptors become suffi- ciently strong to produce inhibition of the customary move- ments. It is held that a similar situation exists in the case of the rat in the maze. The animal may be automatically 218 STUDY OF HABIT FORMATION traversing the maze at a high rate of speed when suddenly a loud noise, a strange odor, or an intense itching occurs. • The chain of movements is broken. How now does auto- maticity become reestablished? The human being under similar circumstances — when momentarily lost while exe- cuting a purely automatic habit — remains unoriented until supplementary distance sense data are at hand; e.g., he glances at the score if playing a piece of music. These rela- tions at hand, the reign of the kinaesthetic arcs is reestab- lished and the automatic character of the acts again becomes apparent. But in the case of the rat do distance sense data function in this way? We conclude, on the basis of a large amount of experimental work, that automaticity is reestablished for the rat solely from the distinctive kingesthetic impulses which function exactly as do the visual impulses in the case just stated for man. If the trained rat is put down in a part of the maze other than the entrance, he runs at first randomly. He may wander about, turn round and round in the alleys, but suddenly he darts off and traverses the remainder of the maze in the usual automatic way. We hold that during the period of random activity the animal passed over what we may call a '' kingesthetic unit," thereby arousing a certain sequence of the kinesthetic impulses which could not be aroused in any other part of the maze. This distinctive group of impulses is sufficient to reestablish automaticity. Exami- nation of any maze shows that there are several ways in which kinsesthetic impulses become grouped: (a) two run- ways are unequal in length; (5) they may be of equal length but occur in different positions of the total series, i.e., they are preceded by different conditions; (c) they may be alike in every respect except that one may be entered by a turn to the right and the other by a turn to the left; (d) the runs may be of the same length and be entered from the same direction but present possible differences in their stimulating effect by reason of the fact that they extend in different directions (180°, etc., rotation of the maze, which changes only the absolute direction of the run-ways, pro- duces marked disturbance in habit). When the unori- ANALYSIS OF MOTOR HABITS 219 ented animal passes over any distinctive part of the maze, the kingesthetic impulses arise, releasing the old movements.
We are far from maintaining that we have a complete analysis of the maze habit even in the case of the rat.^* Some experiments in the maze showed quite clearly that there still are unanalyzed factors. If the Hampton Court maze is rotated 180°, normal, blind, and anosmic animals are disturbed, i.e., a breakdown occurs in the execution of the habit. What this disturbance is due to is not clear. It may depend upon the disturbance of sensory impulses which are already known (i.e., no new systems of receptors are required for its explanation).