On Retrieving Analogues When Solving Problems

Abstract

After criticism of the precision of previous experimental procedures for testing analogue retrieval, a new procedure that overcomes the proposed inadequacies is described. This procedure is then employed in two experiments that test aspects of the general hypothesis that base analogues that are semantically remote from a target problem (Duncker's radiation problem) are more difficult to retrieve than those that are semantically closer. Experiment 1 confirmed this hypothesis by finding that remote analogues are seldom retrieved relative to literal analogues. The results of Experiment 2 falsified the hypothesis that analogue retrieval is solely due to the recognition of an “identical element”. Finally, an ad hoc model of analogue retrieval is proposed based on Schank's dynamic memory theory, and its consistency with the evidence and more general implications are considered.