Trial and Error Learning

Psychologists agree generally that the fundamental method of learning--among men as well as among the lower animals--is that of "trial and error" response. Other schemes for the establishment of beneficial reactions to given stimuli rest ultimately upon the foundations laid by this procedure, together with the principle of the cortical recording of total afferent patterns in conjunction with concomitant motor innervations. Consequently, our first task in the analysis of the complete process of learning in human beings must be that of arriving at a clear understanding of the factors which are concerned in the production and selection of random forms of movement, in response to particular stimuli.

In our historical review of the study of "trial and error" learning, as well as in our chapter on "instincts," we have seen that random movements appear when an appetition is aroused. The mechanism of the production of these movements cannot be fully explained at this point in our argument. However, certain general causes can be assigned. In the first place, we should certainly expect some kind of motor reaction to follow as a consequence of the appetitional currents which are flowing into the cortex, if we assume that it is a general property of the cortex that it tends to generate motor innervations when sensory excitations are received. However, the reaction of the cortex to the appetitional currents is not qualitatively different from that to any sensory impulses. Stimulation of the eyes, ears, or other non-appetitional sense organs, may also be expected to arouse movements. What is called "play" in young animals may be regarded as random activity of this sort, based upon non-appetitional stimulation. If the question is asked as to why the cortical motor innervations are "random" the answer may well be that it is because no
systematic connections have yet been formed. The cortex is compelled to supply some sort of movements, and it may have to respond with something which is not pertinent to the given situation. In the preceding chapter, we have considered some of the factors which may actually determine the particular random innervation which appears at any given time. In the last analysis, of course, we cannot relegate the process purely to the domain of chance.

In accordance with this view--which harmonizes very well with that of Woodworth in the given respect--all stimuli act as "drives" to cortical motor innervation. The appetitional nerve currents may be regarded as being particularly powerful in this respect, since they all fall into the class of nociceptive or beneceptive processes which, according to Sherrington, are endowed with especial neural energy. Now, in experiments upon animals and human beings, the phenomena which may be actually observed are somewhat as follows. The animal exhibits a given kind of behavior and may continue this for some time, but soon it is replaced by another kind. This latter, in turn, may run on for some time but finally gives way to still another type. It would seem to be a satisfactory description of what happens here to say that the "trial" movement is inhibited, so that it is replaced by a different movement; but if this second "trial" does not lead to satisfaction of the appetition, it, also, is subjected to inhibition. If a particular kind of movement is inhibited, it naturally has a reduced tendency to appear, and in order to account, in a general way, for the elimination of the unsuccessful movements, we have only to suppose that this reduction persists over a long period of time. The only movement which is not inhibited will be the one which satisfies the appetition.

A theory somewhat of this character has been advocated by Hobhouse. He supposes that the movements which do not lead to success bring the animal under the influence of stimuli which set off reactions antagonistic to the original movements. Thus, a pedestrian who is endeavoring to cross the street, sees an automobile and is impelled to retreat; or a cat, in trying to get food inside of a cage, squeezes its nose between the bars, and, thus, receives a stimulus to withdraw. In this manner, the movements in the series of "trials" which are unsuccessful, come into association with opposite movement tendencies and tend to be "neutralized." A similar view, who characterizes the elimination mechanism by saying that "incongruent" tendencies will disappear from the responses of the animal to the given situation.

Now the "incongruity," which regarded as being responsible for the elimination of the unsuccessful movements, evidently develops with reference to the "final common path," or motor neurone system. However, learning requires a change in the adjustor mechanisms of the cerebral cortex; and since this adjustor system is a place of contact between the total afferent effect, and the total motor control, the influences which bear upon it must have afferent sources. The motor phenomena are merely expressions of these influences, not causes of them. Consequently, it would seem to be necessary to look for the real cause of the inhibition in the afferent nerve currents which are set off in the pedestrian when he sees the automobile, rather than in his motor reactions. If a chick picks up an ill-tasting caterpillar and is impelled to disgorge its prey, the inhibitory agency would appear to lie in the nociceptive taste impulses. The first thing which these impulses must accomplish when they arrive at the cortex is to stop the innervations which are carrying through the process of ingestion. Having done this, they can then proceed to set off the disgorging reaction. This agrees with the common sense interpretation of the chick's behavior, according to which the bird stops eating the caterpillar and spits it out because it tastes badly. The attempt to explain the inter. ference in terms of the motor side of the system is evidently due to an unwarrantable behavioristic bias.

Thereally critical point, however, type of explanation lies in the necessity that all reactions which are set off in the process of "trial" behavior should be incongruent with the "trial" movements, except when the latter are such as to lead to success. This necessity does not seem to be realized in the facts. Thus, it does not seem to follow inevitably that when an animal does the wrong thing in its quest for food, it will meet a stimulus which will warn it automatically of its error. When a cat presses its nose between the slats of a problembox, it will not necessarily be impelled to withdraw; such pressure upon the nose and mouth region may cause the animal to push harder, an action which can frequently be noted in petting a cat. It might happen, therefore, that the unfortunate animal would find itself pressed permanently against the slats, because of the appearance of a form of reaction which was congruent rather than being incongruent with the trial response. If we consider an animal running in a maze, we may explain its avoidance of cul-de-sacs on the theories that they are nose-bumping arrangements; but suppose that each cul-de-sac is a wonderfully comfortable place to sleep. This does not prevent the hungry animal from leaving them all behind.