摘要：Five-year-old children received training on four symbolic match-to-sample tasks with single-element stimuli (A 1-B1, A2-B2; P1-Q1, P2-Q2). Then they received training on four novel matching tasks with AB compounds as samples and X stimuli as comparisons. The children received positive feedback for selecting X1 when given compounds representing sample-correct comparison relations (A 1 B1-X1, A2B2-X1) and for selecting X2 when given sample-incorrect comparison relations (A 1 B2-X2, A2B1-X2). Finally transfer from AB-X to PQ-X was assessed (P1 Q1-X1, P2Q2-X1, P1 Q2-X2, P2Q1-X2). Most children did not learn the AB-X tasks unless they were first trained to respond to all compound elements, and they did not reliably show transfer unless they had first learned that the compounds (e.g., A1B1) represented sample-comparison relations (A 1-B 1).
