Memory in concept identification

Recent quantitative theories of concept identification assume a limited memory during learning. To test this assumption, Ss recalled the stimulus dimensions and the correct responses on each stimulus card in 12 six-card problems. The expected recency effect was obtained but a larger, unexpected primacy effect was also observed. On those problems which were solved, the joint recall of the relevant dimension and the correct response to the cards was above chance. On unsolved problems, the recall of the relevant dimension was independent of the recall of its classification. Average recall on all problems was only slightly above the chance level. These results, with the exception of the primacy effect, support the limited memory assumption.