Document Type



A multiple-exemplar account of symmetric matching to sample was tested with 5- to 6-year-old children. Each child went through five similar phases. Each phase began with a two-choice arbitrary matching-to-sample task. This was followed by a test of symmetric responding. The first four phases ended with training of symmetric or asymmetric responding. This differed between two groups of 10 children. For example, after conditional relations A1 B1 and A2B2 had been trained, and symmetry had been tested, conditional relations B1A1 and B2A2 (Group Symmetric) or B1A2 and B2A1 (Group Asymmetric) were trained. The stimuli differed between phases. Symmetric responding increased as a function of tests in Group Symmetric; it decreased in Group Asymmetric. These findings support a multiple-exemplar account of stimulus equivalence.