[Footnote: The writer hopes that no confusion will be caused by his use of several words to express this same meaning.

"Attachment of stimulus and response", "linkage of stimulus and response", "connection between stimulus and response", and "bond between stimulus and response", all mean exactly the same; but sometimes one and sometimes another seems to bring the meaning more vividly to mind.]

(a) _Subst.i.tute stimulus_ refers to the case where the natural response is not itself modified, but becomes attached to another stimulus than the one that originally aroused it. This new linkage can sometimes be established by simply giving the original stimulus and the subst.i.tute stimulus at the same time, and doing so repeatedly, as in the conditioned reflex experiment.

(b) _Subst.i.tute response_ refers to the case where the stimulus remaining as it originally was, a new reaction is attached to it in place of the original response. The conditions under which this takes place are more complex than those that give the subst.i.tute stimulus. A tendency towards some goal must first be aroused, and then blocked by the failure of the original response to lead to the goal. The dammed-up tendency then facilitates other responses, and gives trial and error behavior, till some one of the trial responses leads to the goal; and this successful response is gradually subst.i.tuted for the original response, and becomes firmly attached to the situation and tendency.

(3) New _combinations of responses_ occur, giving higher motor units.

Human Learning

To compare human and animal learning, and notice in what ways the human is superior, cannot but throw light on the whole problem of the process of learning. It is obvious {312} that man learns more quickly than the animals, that he acquires more numerous reactions, and a much greater variety of reactions; but the important question is how he does this, and how his learning process is superior.

We must first notice that all the forms of learning displayed by the animal are present also in the human being. Negative adaptation is important in human life, and the conditioned reflex is important, as has already been suggested. Without negative adaptation, the adult would be compelled to attend to everything that aroused the child's curiosity, to shrink from everything that frightened the child, to laugh at everything that amused the child. The conditioned reflex type of learning accounts for a host of acquired likes and dislikes. Why does the adult feel disgust at the mere sight of the garbage pail or the mere name of cod liver oil? Because these inoffensive visual and auditory stimuli have been a.s.sociated, or paired, with odors and tastes that naturally aroused disgust.

The signal experiment is duplicated thousands of times in the education of every human being. He learns the meaning of signs and slight indications; that is, he learns to recognize important facts by aid of signs that are of themselves unimportant. We shall have much to say on this matter in a later chapter on perception. Man learns signs more readily than such an animal as the rat, in part because the human being is naturally more responsive to visual and auditory stimuli. Yet the human being often has trouble in learning to read the signs aright. He a.s.sumes that a bright morning means good weather all day, till, often disappointed, he learns to take account of less obvious signs of the weather. Corrected for saying, "You and me did it", he adopts the plan of always saying "you and I", but finds that this quite unaccountably brings ridicule on him at times, so that gradually he _may_ come to say the one or the {313} other according to obscure signs furnished by the structure of the particular sentence. The process of learning to respond to obscure signs seems to be about as follows: something goes wrong, the individual is brought to a halt by the bad results of his action, he then sees some element in the situation that he had previously overlooked, responds to this element, gets good results, and so--perhaps after a long series of trials--comes finally to govern his action by what seemed at first utterly insignificant.

Trial and error learning, though often spoken of as characteristically "animal", is common enough in human beings. Man learns by impulsively doing in some instances, by rational a.n.a.lysis in others. He would be at a decided disadvantage if he could not learn by trial and error, since often the thing he has to manage is very difficult of rational a.n.a.lysis. Much motor skill, as in driving a nail, is acquired by "doing the best you can", getting into trouble, varying your procedure, and gradually "getting the hang of the thing", without ever clearly seeing what are the conditions of success.

Human Compared With Animal Learning

Fairly direct comparisons have been made between human and animal learning of mazes and puzzles. In the maze, the human subject has an initial advantage from knowing he is in a maze and has to master it, while the rat knows no more than that he is in a strange place, to be explored with caution on the odd chance that it may contain something eatable, or something dangerous. But, after once reaching the food box, the rat begins to put on speed in his movements, and within a few trials is racing through the maze faster than the adult man, though not so fast as a child. Adults are more circ.u.mspect and dignified, they make less speed, cover less distance, but also make fewer false moves {314} and finish in less time. That is in the early trials; adults do not hold their advantage long, since children and even rats also reach complete mastery of a simple maze in ten or fifteen trials.

The chief point of superiority of adults to human children, and of these to animals, can be seen in the adjacent table. It is in the _first trial_ that the superiority of the adults shows most clearly.

They get a better start, and adapt themselves to the situation more promptly. Their better start is due to (1) better understanding of the situation at the outset, (2) more plan, (3) less tendency to "go off on a tangent", i.e., to respond impulsively to every opening, without considering or looking ahead. The adult has more inhibition, the child more activity and responsiveness; the adult's inhibition stands him in good stead at the outset, but the child's activity enables him to catch up shortly in so simple a problem as this little maze.

AVERAGE NUMBER OF ERRORS MADE, IN EACH TRIAL IN LEARNING A MAZE, BY RATS, CHILDREN AND ADULT MEN

(From Hicks and Carr)

Trial No. Rats Children Adults

1 53 35 10 2 45 9 15 3 30 18 5 4 22 11 2 5 11 9 6 6 8 13 4 7 9 6 2 8 4 6 2 9 9 5 1 10 3 5 1 11 4 1 0 12 5 0 1 13 4 1 1 14 4 0 1 15 4 1 1 16 2 0 1 17 1 0 1

The table reads that, on the first trial in the maze, the rats averaged 53 errors, the children 35 errors, and the adults 10 errors, and so on. An "error" consisted in entering a blind alley or in turning back on {315} the course. The subjects tested consisted of 23 rats, five children varying in age from 8 to 18 years, and four graduate students of psychology. The human maze was much larger than those used for the rats, but roughly about the same in complexity. Since rats are known to make little use of their eyes in learning a maze, the human subjects were blindfolded. The rats were rewarded by food, the others simply by the satisfaction of success.

The puzzle boxes used in experiments on animal learning are too simple for human adults, but mechanical puzzles present problems of sufficient difficulty. The experimenter hands the subject a totally unfamiliar puzzle, and notes the time required by the subject to take it apart; and this is repeated in a series of trials till mastery is complete. In addition to taking the time, the experimenter observes the subject's way of reacting, and the subject endeavors at the end of each trial to record what he has himself observed of the course of events.

The human subject's behavior in his first trial with a puzzle is often quite of the trial and error sort. He manipulates impulsively; seeing a possible opening he responds to it, and meeting a check he backs off and tries something else. Often he tries the same line of attack time and time again, always failing; and his final success, in the first trial, is often accidental and mystifying to himself.

On the second trial, he may still be at a loss, and proceed as before; but usually he has noticed one or two facts that help him. He is most likely to have noticed _where_ he was in the puzzle when his accidental success occurred; for it appears that _locations_ are about the easiest facts to learn for men as well as animals. In the course of a few trials, also, the human subject notices that some lines of attack are useless, and therefore eliminates them. After a time he may "see into" the puzzle more or less clearly, though sometimes he gets a practical mastery of the handling of the puzzle, while still obliged to confess that he does not understand it at all.

{316}

Insight, when it does occur, is of great value. Insight into the general principle of the puzzle leads to a better general plan of attack, and insight into the detailed difficulties of manipulation leads to smoother and defter handling. The human "learning curve" (see Figure 50) often shows a prolonged stretch of no improvement, followed by an abrupt change to quicker work; and the subject's introspections show that 76 per cent, or more of these sudden improvements followed immediately after some fresh insight into the puzzle.

[Ill.u.s.tration: Fig. 50.--(From Ruger.) Curve for human learning of a mechanical puzzle. Distance above the base line represents the time occupied in each trial, the successive trials being arranged in order from left to right. A drop in the curve denotes a decrease in time, and thus an improvement. At _X_, the subject saw something about the puzzle that he had not noticed before and studied it out with some care, so increasing his time for this one trial, but bringing the time down thereafter to a new and steady level.]

The value of insight appears in another way when the subject, after mastering one puzzle, is handed another involving the same principle in a changed form. If he has seen the principle of the first puzzle, he is likely to carry over this knowledge to the second, and master this readily; {317} but if he has simply acquired motor skill with the first puzzle, without any insight into its principle, he may have as hard a time with the second as if he had never seen the first.

Learning by Observation

"We learn by doing" is a true proverb, in the sense that we acquire a reaction by making just that reaction. We must make a reaction in order to get it really in hand, so that the proverb might be strengthened to read, "To learn, we must do". But we should make it false if we strengthened it still further and said "We learn _only_ by doing". For human beings, at least, learn also by observing.

The "insight" just spoken of consists in observing some fact--often some relationship--and the value of insight in hastening the process of learning is a proof that we learn by observation as well as by actual manipulation. To be sure, observation needs to be followed by manipulation in order to give practical mastery of a thing, but manipulation without observation means slow learning and often yields nothing that can be carried over to a different situation.

Learning by observation is typically human. The adult's superiority in tackling a maze may be summed up by saying that he observes more than the child--much more than the animal--and governs his behavior by his observations. The enormous human superiority in learning a simple puzzle, of the sort used in experiments on animals, arises from seeing at once the key to the situation.

A chimpanzee--one of the most intelligent of animals--was tested with a simple puzzle box, to be opened from outside by turning a b.u.t.ton that prevented the door from opening. The device was so simple that you would expect the animal to see into it at once. A banana was put into the box and the door fastened with the b.u.t.ton. The {318} chimpanzee quickly found the door, and quickly found the b.u.t.ton, which he proceeded to pull about with one hand while pulling the door with the other. Without much delay, he had the b.u.t.ton turned and the door open. After about three trials, he had a practical mastery of the puzzle, showing thus considerable superiority over the cat, who would more likely have required twelve or fifteen trials to learn the trick.

But now a second b.u.t.ton was put on a few inches from the first, both being just alike and operating in the same way. The chimpanzee paid no attention to this second b.u.t.ton, but turned the first one as before, and when the door failed to open, kept on turning the first b.u.t.ton, opening it and closing it and always tugging at the door. After a time, he did shift to the second b.u.t.ton, but as he had left the first one closed, his manipulation of the second was futile. It was a long, hard job for him to learn to operate both b.u.t.tons correctly; and the experiment proved that he did not observe how the b.u.t.ton kept the door from opening, but only that the b.u.t.ton was the thing to work with in opening the door. At one time, indeed, in order to force him to deal with the second b.u.t.ton, the first one was removed, but he still went to the place where it had been and fingered about there. What he had observed was chiefly the place to work at in order to open the door.

We must grant that animals observe locations, but most of their learning is by doing and not by observing.

Here is another experiment designed to test the ability of animals to learn by observation. The experimenter takes two cats, one having mastered a certain puzzle box, the other not, and places the untrained cat where it can watch the trained one do its trick. The trained cat performs repeatedly for the other's benefit, and is then taken away and the untrained cat put into the puzzle box. But he has derived no benefit from what has gone on before his eyes, and must learn by trial {319} and error, the same as any other cat; he does not even learn any more quickly than he otherwise would have done.

The same negative results are obtained even with monkeys, but the chimpanzee shows some signs of learning by observation. One chimpanzee having learned to extract a banana from a long tube by pushing it out of the further end with a stick which the experimenter had kindly left close by, another chimpanzee was placed where he could watch the first one's performance and did watch it closely. Then the first animal was taken away and the second given a chance. He promptly took the stick and got the banana, without, however, imitating the action of the first animal exactly, but pulling the banana towards him till he could reach it. This has been called learning by imitation, but might better be described as learning by observation.

Such behavior, quite rare among animals, is common in human children, who are very observant of what older people do, and imitate them on the first opportunity, though often this comes after an interval. The first time a child speaks a new word is usually not right after he has heard it. When, on previous occasions, he has heard this word, he has not attempted to copy it, but now he brings it out of himself. He has not acquired the word by direct imitation, evidently, but by what has been called "delayed imitation", which consists in observation at the time followed later by attempts to do what has been observed.

Observation does not altogether relieve the child of the necessity of learning by trial and error, for often his first imitations are pretty poor attempts; but observation gives him a good start and hastens the learning process considerably. "Learning by imitation", then, is, more properly, "learning by observation followed by trial and error" and the reason so little of it appears in animals is their lack of observation.

_Learning by thinking_ depends on observation, since in {320} thought we make use of facts previously observed. Seldom, unless in the chimpanzee and other manlike apes, do we see an animal that appears to be thinking. The animal is always doing, or waiting, or sleeping. He seems too impulsive to stop and think. But a man may observe something in the present problem that calls previous observations to mind, and by mentally combining observations made at different times may figure out the solution before beginning motor manipulation. Usually, however, some manipulation of the trial and error sort is needed before the thought-out solution will work perfectly.

Sometimes mental rehearsal of a performance a.s.sists in learning it, as we see in the beginner at automobile driving, who, while lying in bed after his first day's experience, mentally goes through the motions of starting the engine and then the car, and finds that this "absent treatment" makes the car easier to manage the next day.

In summing up the points of superiority of human over animal learning, we may note that--

1. Man is perhaps a quicker learner, anyway, without regard to his better methods of learning. This, however, is open to doubt, in view of the very rapid learning by animals of such reactions as the avoidance of a place where they have been hurt.

2. Man is a better observer, and this is the great secret of his quick learning. He is especially strong in observing relationships, or "principles" as we often call them.

3. He has more control over his impulses, and so finds time and energy for observing and thinking.

4. He is able to work mentally with things that are not present; he remembers things he has seen, puts together facts observed at different times, thinks over problems that are not actually confronting him at the moment, and maps out plans of action.

{321}

The Learning of Complex Practical Performances

A great deal of light has been thrown on the learning process by psychological studies of the course of improvement in mastering such trades as telegraphy and typewriting.

A student of telegraphy was tested once a week to see how rapidly he could send a message, and also how rapidly he could "receive a message off the wire", by listening to the clicking of the sounder. The number of letters sent or received per minute was taken as the measure of his proficiency. This number increased rapidly in the first few weeks, and then more and more slowly, giving a typical learning curve, or "practice curve", as it is also called.