The present study was aimed at investigating i) the developmental change of the ability for concept learning, especially the ability for retaining the criterion which was acquired through the practice of classification of individual cases; ii) the effect of practice without external reinforcement on concept learning.
Twenty-five 1st-graders, 22 3rd-graders, 18 5th-graders and 10 undergraduate students served as Ss. In the preliminary experiment Ss performed a simple concept learning task to acquaint themselves with the experimental situation of concept learning. Eight Ss could not reach the criterion of success in this task and were excluded from the main experiment that followed.
In the main experiment 32 cards were presented, on each of which was drawn a pair of stimuli that varied on 4 binary dimensions. Ss were asked to compare two stimuli (Xi, Xj) in terms of the relation, >, =or<. The rule was that “if Xi, Xj∈A or Xi, Xj∈A, then Xi=Xj” and that “if Xi∈A and Xj∈A, then Xi>Xj”. It was also informed that the positive value on one dimension defined class A, while two of 4 dimensions were distinctively irrelevant, the remaining two covaried in 7/8 of the trials, making it difficult to determine a relevant dimension.
Ss who reached the criterion of 14/16 correct responses before the termination of the reinforced practice after 4 round exposures were given trials without external reinforcement. They were tested as for the ability to judge pairs of new stimuli, for which none of the 4 dimensions correlated with each other, on the basis of one relevant cue disregarding the other, Some of them were also given “practice in conflict situations” -in which two apparently valid criteria would lead to the opposite conclusion-and were asked to verbalize “his” critetrion of a class (both without confirming information).
The results were as follows:
i) The proportion of Ss who reached the criterion was highly correlated with age. Of 23 1st-graders only 5 reached the criterion, while 13 of 19 3rd-graders and 13 of 16 5th-graders could identify the correct criterion. All of the undergraduates succceed in learning with about 1/10 of the average errors made by elementary school pupils.
ii) It was very difficult for elementary school pupils to disregard one of the two apparently relevant cues in the experiment. Only 4 out of 31 who had reached the criterion showed a consistent response pattern based on correct criterion from the beginning of trials without reinforcement. On the contrary, 8 of 9 undergraduate students showed such a pattern. This result might be interpreted that young children judged on cumulative intuitive impression, lacking a deliberately selected hypotheses. Children who could respond correctly when the relevant cue was explicitly present often based their judgment on another apparently relevant cue when values of relevant dimension were the same in both of the paired stimuli.
iii) In the process of judgment without confirming information, however, lack of consistency in children's response patterns gradually decreased. On test trials given after 36 or less trials without reinforcement, 11 out of 27 pupils could disregard irrelevant cue completely. This imrovement of performance without external reinforcement seemed to be essentially the same process as the acqusition of conservation by Smedslund's “practice in conflict situations”.
