In the present experiment kindergarten children were used as subjects who were trained to choose the stimulus box containing a marble. The stimulus boxes were paired between those which differ in two dimensions, color and size, and presented simultaneously throughout the experiment. In the first learning situation, the Reversal Group was trained on a color discrimination with a certain color, white or black, reinforced, whereas the Nonreversal Group was trained on a size discrimination with a certain size, the larger or the smaller, reinforced. When a certain criterion was met in the first learning situation, the second learning situation followed without any instruction. In the second learning situation, only a pair of stimuli was presented to all subjects and the negative stimulus in the first learning situation was reinforced. Moreover, a control group which had not participated in the first learning situation was added in the second learning situation. When the criterion, 5 successive correct responses in the seconcl learning situation, was met, the third learning situation followed. In this situation, another pair of stimuli was added again and all subjects were trained on color discrimination, with the same color as that of the positive stimulus in the second learning situation reinforced, until 5 successive correct responses were achieved. The pair of stimuli unused in the second learning was presented at the first trial in the third learning situation.
Moreover, each of the experimental groups was divided into 4 subgroups respectively with respect to the criteria in the first learning situation of 3 and 5 successive correct responses, 10/11 correct responses and 10/11 correct responses plus 5 successive correct responses.
The results are as follows:
1) The more the amount of training during the first learning situation, the more difficult became the second and the third learning situations until the point where the criterion of the first learning was 3 or 5 successive correct responses, then they became easier beyond this point.
2) The comparison of the numbers of trials to reach the criterion in the third learning situation in Reversal Groups with Nonreversal Groups indicated the superiority of reversal shifts over nonreversal shifts.
3) The stimulus box choosen at the first trial in the third learning situation was the same to prior positive box in either dimension of color or size. It was recorded which dimension was choosen as relevant cue, i. e. which was abstracted. The result indicated that the Reversal Group trained on color discrimination during the first learning situation tended to abstract “color” with an increase of that training, whereas the Nonreversal Group trained on size discrimination during the first learning situation tended to abstract “size” with the increase of that training.
The results were explained in terms of connection procceses. That is, in the early period of discrimination learning, the immediate connection between physical stimuli and overt responses is dominant, thus the more pre-shift training the more difficult are reversal shifts as simple S-R theory predicts. With the progress of discrimination learning, however, the more complex connection at the higher level become dominant. This higher connection contains mediation-process between physical stimuli and overt responses, in which generalization between antinonyms or opposite cues appears. Consequently reversal shift become easier with the increase of pre-shift training in the later period of discrimination learning.
View full abstract