A PARALLEL RULE ACTIVATION AND RULE SYNTHESIS MODEL FOR GENERALIZATION IN CATEGORY LEARNING

This paper proposes a distinction between primary generalization (tran sfer from stored exemplars to perceived targets) and secondary general ization (transfer from inferred abstractions to perceived targets). Th is distinction is embodied in the parallel rule activation and rule sy nthesis (PRAS) model, a production model capable of exemplar-based and abstraction-based categorization. As an exemplar model, the PRAS mode l is related to the generalized context model (Nosofsky, 1984). Exempl ars are stored in memory encoded as condition-action rules. Working as an exemplar-based model, rules are activated on the basis of their st rength and their similarity to the current to-be-categorized instance. Similarity between a target and a stored exemplar is weighted for att ention to the dimensions of the psychological space. Depending on the value of a special parameter, the PRAS model is also able to operate a s an abstraction model. In the latter case, it attempts to construct g eneralizing productions, which are activated according to the same rul es as the exemplar-specific rules. The model is described in detail. I t is applied to a number of important observations described in the re search literature, and an experiment is reported that tested the usefu lness of the proposed secondary-generalization mechanism. Finally, the discussion elaborates on the implications of the present study for fu rther research.