EFFECT OF DEGREE OF ORIGINAL LEARNING AND TIME INTERVAL BETWEEN TWO TASKS ON REVERSAL SHIFT.

Abstract

Sugimura (1962) reported that a nonreversal shift was easier with a 30-min. rest interval between two learning tasks than with no rest interval and this finding was dependent upon degree of original learning. The purpose of this experiment was to repeat the previous smdy using a reversal shift. Sixty boys, aged 15 to 19, were divided into 6 groups, 1 0 Ss each. S was trained on a simple 2-choice size or color discrimination task until one of the following 3 learning criteria was met: 3 or 5 successive correct responses, and 1 0 successive plus 2 0 correct responses. Immediately after reaching the respective criteria, half of the Ss were given the reversal task without further instructions. The remaining half were given the same task but with a 30-min. rest interval. The learning criterion of reversal task was 1 0 successive correct responses for all Ss. The median number of trials to criterion in reversal learning decreased as the criterion of original learning was made more rigorous under the no-rest condition: 13.5, 5.5, and 1.0 under the original criteria of 3, 5 correct and 1 0 plus 20 correct responses, respectively. The differences among 3 groups were significant ( p < .01) and agreed with our previous work (Iwahara & Sugimura, 1960) . Under the 30-min. rest condition, however, the median was 1.0 for 3 groups, regardless of original criteria. The difference between the 2 rest conditions was significant only when Ss were trained on the criterion of 3 correct responses in original learning ( f i < .01). When the median of trials to 1 0 successive correct responses in original learning (i.e., 5.0) was regarded as a baseline to evaluate the transfer effect, positive transfer occurred in the 3 30-min. rest groups and in the overtrained group without rest. Negative transfer occurred when Ss were trained to a criterion of 3 correct responses without rest. In conclusion, a rest interval between original and reversal tasks and overtraining of original task had facilitative effects on a reversal shift, as was shown in the previous nonreversal study.