One hundred and sixty Ss, assigned randomly to cells of a 4 (conceptual rule: conjunctive, disjunctive, conditional, and biconditional) X 4 (memory condition: feedback retained for all instances, for positive instances only, for negative instances only, for none of the instances) factorial design, solved an attribute identification problem via the selection paradigm. Rules differed in difficulty and the display of feedback for previously selected instances facilitated performance. The relative effectiveness of retained instances differed, however, with the rule; e.g., positives were more informative in conjunctive and negatives in conditional concepts. There were differences in the stimulus selection patterns of better and poorer 5"s, which also interacted with rule and memory condition.

No abstract