Abstract

This paper investigates use of rules and exemplar memories in category learning by manipulating the salience of correlational rules. The resulis of two experiments showed that salient condition was better in categorization performance than nonsalient conditions (no symmetry cue and random arrangement conditions). An interaction between the two variables of salience and category size was also found in Experiment 2. In a salient condition, categorization appeared to be better in a larger category size than in a small category size. This implies the use of abstract rules. In a no symmetry cue condition, a small category size was found to categorize better than a larger category size, suggesting the involvement of memory-based knowledge. But a random arrangement condition was not affected by category size. In addition, the two experiments were used to test the relation between categorization and recognition. Salience of rules was found to affect categorization but did not affect recognition, which suggests a dissociation of the two tasks. These findings are consistent with the implicit/explicit learning that salient conditions employ analytic and explicit rules during category learning whereas nonsalient conditions are based on exemplar memories.