MACIFAC: A Model of Si~ i~ ari~-~ s~ Retrieval

We present a model of similarity-based retrieval that attempts to capture three seemingly contradictory psychological phenomena: (a) structural commonalities are weighed more heavily than surface commonalities in similarity judgments for items in working memory: (b) in retrieval, superficial similarity is more important than structural sirnjlarj~: ond yet (c) purely structural (analogical) remindings e sometimes experienced. Our model, MAUFAC, explains these phenomena in terms of o two-stage process. The first stage uses a computationolly cheap, nonstructural matcher to filter candidate long-term memory items. It uses content vectors, a redundant encoding of structured representations whose dot product estimotes how well the corresponding structural representations will match. The second stage uses SME (structure-mapping engine) to compute structural matches on the hondful of items found by the first stoge. We show the utility of the MAClFAC model through o series of computational experiments: (a) We demonstrate thot MAC/FAC can model patterns of access found in psychological data: (b) we argue via sensitivity analyses that these simulation results rely on the theory: and (c) we compare the performance of MAUFAC with ARCS, an alternate model of similarity-based retrieval, and demonstrate that MAUFAC explains the data better than ARCS. Finally, we discuss limitations and possible extensions of the model, relationships with other recent retrieval models, and place MACiFAC in the context of other recent work on the nature of similarity.