Reasoning in White Rats

Front matter

COMPARATIVE PSYCHOLOGY MONOGRAPHS

MANAGING EDITOR

Knight Dunlap. The Johns Hopkins University

I. Historical

Since Koehler's work with chimpanzees the doctrine of "trial and error" has been considerably questioned even in the case of rats. Although insight has not been directly proved in their case, some experimenters have hinted that they are capable of it.

Higginson points out that there is a sudden readjustment to significant environmental changes. He taught his rats to go 6 feet past a closed door, in a circular maze, to the end of a "blind" and then to return and find the door open. After 100 repetitions the door was left open in the beginning. Five of the nine rats which he used, noted the change and took the open door without going to the end of the "blind" as usual, and from this point continued the run without error. This, he believes, was a visual adjustment to the changed environment. Although it was Higginson's purpose to demonstrate visual perception, it should be pointed out that he did not prove a visual adjustment. The rats may have reacted to a change made in the maze, but what he calls adjustment may have been curiosity. Tolman commenting on the same point says that there is a general tendency for a rat to take any new pathway. Hence it can only be said that the rats reacted to a change in the situation. Nor can one be sure of the cue to which the animals reacted. It may have been auditory rather than visual.

Because the rats continued the maze correctly from this point on, shows only that they got sufficient cues at this point.

Helson shows that rats respond to visual structures and not to elements. Thus animals trained to go to a certain light B instead of a less bright light A, go to light C instead of B, when C is brighter than B. The rat chooses the relatively brighter light rather than the absolute brightness with which the food has always been received. The relation brighter involves the whole structure and the association is made with the whole "bi-membral structure" rather than with an element in the situation. Koehler has shown this to be the case in apes and chicks, and Moody in the case of mice. This response to structure rather than to an element in the structure can hardly be considered proof of insight as Helson contends (p. 389). In the case of 2 of the 4 rats used, Helson believes he has a second criterion for insight. In order to reach the food the rats had to climb over a platform on which an electric shock was given in case the rat went wrongly. As the rats did not learn, in this case, which gray to choose for the food and the avoidance of the shock, they soon developed a negative association with the platforms of both compartments. To avoid the shock and still obtain the food, two rats now scaled the wall between the compartments and so reached the food. This may be an expression of insight as Helson contends, yet not necessarily so. As the food had a positive association and the platform a negative association, the response might have been the resultant of the two stimuli. An insect positively phototropic will crawl directly toward the light. If some of the legs are removed on one side of the body an adjustment is made which is not called insight. Or an ant on its way to food will successfully go around a disagreeable substance, placed in its path, without being credited with insight.

Thorndike found that cats solve "puzzle-box" problems by "trial and error" and from this he generalized that animals are only capable of this type of solution. Helson points out that this generalization is incomplete, because it does not necessarily follow that animals are only capable of this type of solution because "trial and error" is used to solve the "puzzle-box." This criticism may be valid, but Helson's experiments do not substantiate its validity.

Koehler has probably demonstrated insight or intelligent solutions in the case of apes. His method of "round-about-ways" to food presents the animal with problems in which there is a chance for intelligent solutions. The animal is allowed to become familiar with the whole situation and thus is able to apply its knowledge of that situation to the problem in an intelligent manner. Two considerations, however, raise some doubt; (1) the past experience of the animal was not controlled and (2) the question as to whether the solution was "clear-cut" or not was the important factor in the conclusions and depended on the interpretation of the experimenter rather than on objective measurements.

Miss Hertz working with 3 crows made some interesting observations which, however, did not deal with insight and so do not concern us here. In a second paper , using 1 crow, she reports experiments similar to some of Koehler's on the apes and with apparently positive results. Through holes in the cage she taught the bird to pass nuts which she then cracked for it. The bird had to respond not to a single stimulus, but to a whole situation-structure (nut—observer—hole) in order to reach the goal (nut-meat). The bird showed 3 tendencies (1) to hide the nut, (2) to go toward the observer, and (3) to pass the nut through the habitual hole. By varying her position she could see whether these 3 tendencies were integrated in a sensible manner.

Using 2 holes, the bird with the aid of coaxing, chose the hole before which the observer stood. On the third day this was done without the aid of coaxing (calling the bird toward her). Using now a third hole on the sixth day coaxing was again required. The following day no coaxing was required. On the tenth day she stood before a fourth hole and the bird passed the nut to her without coaxing.

Thus the bird learned to use the different holes so that on the tenth day the relation "hole nearest the observer" was responded to, rather than a particular hole. This responding to the relation was learned and need not be any different from the responding to the brighter of two lights.

She then did some "similarity experiments." Nuts painted in different shades of gray were still recognized. Small nuts covered with gypsum were also used. She concludes that the bird recognizes not only the walnut as it appears naturally, but also disfigured nuts, the more disfigured the less easily recognized. Thus the bird has different stages of recognition and can do more than make mere positive or negative responses. As she has not analyzed all of the cues one cannot state which complexes of cues call up a response toward the nut and which complexes do not.

In her "relation experiments" she finds that the bird chooses a hole that is large enough for the nut. As the bird has learned to respond to the whole nearest the observer and as it does not respond when she stands anywhere before the cage, a small hole does not call out the response of passing the nut through. This shows that the animal has good vision and is responding to the relation "size of nut to size of hole."

She states that the structure which gives the situation a definite sense for the bird is "nut inside, observer outside." Such a conclusion as to what is sense for a bird is an interpretation which is not forced by the facts.

The "obstacle experiments" approach those of Koehler. Three stones were placed before a door the middle one of which blocked a door which the bird had learned to open for food. In the experiment the bird walked back and forth before the door, the stones being in its path. The middle stone was then attacked, it resisted, the other two were then moved and only later the middle stone. One wonders whether the stones were attacked because they were in the way and the bird stumbled over them.

On repeating the next day all stones were again moved. She says, "Der Versuch ist ungünstig beeinflusst dadurch, dass das Ziel nicht recht 'zieht.'" Again an uncalled for interpretation.

Only on the fifth day is the correct stone removed first, but this time the two end stones were 70 cm. from the middle stone and so out of the situation.

Using a small and a large lever as latches which serve as obstacles for the door, she gets somewhat more significant results. The latches are moved by the bird, sometimes enough and sometimes not enough. The latter she excuses by saying that the latch was moved almost enough and that the bird could not make out so small a difference in the complex. Yet we know that the bird's eyes are very accurate from her relation experiments.

As the latch is a new element in the situation and as the door resists, it does not require an understanding of the situation for the bird to attack the new element. If clear-cut results are gotten after a few repetitions we do not know how much has been learned. The fact that a later experiment in which the stone was again the obstacle and the latches placed so as not to block, resulted in the bird's attacking the stone instead of the latch, does not prove understanding because the stone had been attacked in previous experiments and was now the most obvious object. Would the bird attack a lone stone lying near the door, but not being an obstacle? With the stones and latch both serving as obstacles she got a solution which was not clear-cut, but was followed by a clear-cut one. She excuses the first by saying that the tendencies were not organized.

Thus with one bird as a subject, with solutions that are not clear-cut, and with an observer that is decidedly prejudiced one can only express grave doubts as to whether birds have an understanding of problems, and whether there are such things as sense and non-sense situations for it.

Multiple choice, delayed reactions, and sudden drops in the learning curve all may have a close relation to intelligence. They do not, however, give us clear-cut cases of insight. Yerkes' orang-utan did so poorly in the multiple choice test that he believed it used "wrong ideas." Mastering the multiple choice problem involves many trials so that an original solution is never obtained. Delayed reactions involve more the problem of memory images (or some factor of substitutive value) and although memory images may be of great value in solving problems, insight or reasoning is not demonstrated if they are shown to be present. The sudden drop in the learning curve may be evidence of insight or abstraction; however, attention, anatomical differences in vision, etc., are such important factors that one can hardly consider such a drop as valid evidence in favor of insight.

As far as published evidence goes there is then little or no evidence that animals, except man and perhaps apes, have insight. The laws of association seem quite adequate to explain animal behavior, at least at the level of the rat, if not of the ape. The rat's great capacity for learning has, however, caused con- siderable trouble in the application of these laws. Shepard in studying the rat's learning of very complicated mazes and finding marked ability in their adapting to changes, has come to the conclusion that the mind of the rat is not nearly so simple as is usually supposed by the animal psychologist.

II. Statement of problem

The purpose of this investigation is to learn whether or not rats can solve problems without "trial and error." That is, can a rat, placed in a certain situation, adjust itself intelligently or adequately for its purpose, without previously having learned to respond in the situation concerned and without making a series of random responses that finally lead to an adjustment. Or further, can a rat be placed in a situation in which it can, by its solution, demonstrate insight? Is the "puzzle-box" a valid measure of animal intelligence? Does it fully test an animal's capacity or can a rat behave in a manner that might be called reasoning? If it can reason, then (1) it must adequately adjust itself to a new situation, (2) it must be able to apply essential relations from other experiences to the new situation, and (3) the solution must be a continuous whole directed toward a definite end. This end need not be consciously appreciated.