AN INTERACTION OF INSTRUCTIONS AND TRAINING DESIGN ON STIMULUS CLASS FORMATION - EXTENDING THE ANALYSIS OF EQUIVALENCE

Our previous research with subjects with mild mental retardation indicated different outcomes on stimulus equivalence tests when different training designs were employed. Subjects exposed to match-to-sample training involving four pairs of sample stimuli and one pair of comparisons (comparison as node) were more likely to show positive results on equivalence tests than subjects trained with four pairs of comparisons and one pair of samples (sample as node). We describe three studies aimed at analyzing the superiority of the comparison-as-node training procedure. The results suggest that (a) preexperimental individual differences and subject assignment do not account for the previous findings, (b) the greater effectiveness of the comparison-as-node procedure may be caused partially by an interaction of the procedure and instructions providing stimulus names, and (c) differential outcomes on equivalence tests with these training procedures may be demonstrated only with developmentally limited subjects. Several theoretical interpretations for the reason the comparison-as-node training method is more likely to produce positive equivalence test performances and the implication of these interpretations for understanding stimulus equivalence in general are discussed.