Establishing Equivalence Relations Using a Respondent-Type Training Procedure

Abstract

During three experiments, 35 human, adult subjects across seven experimental conditions (5 subjects in each condition) were exposed to a respondent-type training procedure in which arbitrary stimuli (i.e., nonsense syllables) were presented, one at a time, on a computer screen. In Condition 1, Experiment 1, instructions informed the subjects that the material to be presented during the first stage of the experiment (i.e., the respondent-type training procedure) was related to the second stage (i.e., the equivalence test). Nine nonsense syllables were presented to the subjects in the form of six stimulus pairs: A1 → B1, B1 → C1, A2 → B2, B2 → C2, A3 → B3, B3 → C3. The first stimulus of each pair was presented for 1 s (e.g., A1), the computer screen was cleared for 0.5 s (the within-pair-delay) and the second stimulus in the pair (i.e., B1) was presented for 1 s. The screen cleared for 3 s (i.e., the between-pair-delay) before the next stimulus pair was presented. All six stimulus pairs were presented 10 times in a quasi-random order across 60 trials. Subjects were presented with a standard matching-to-sample equivalence test that examined the six symmetry relations (i.e., B1-A1, B2-A2, B3-A3, C1-B1, C2-B2, C3-B3) and the three equivalence relations (i.e., C1-A1, C2-A2, C3-A3). All five subjects demonstrated equivalence responding after two, three, or four exposures to the training and testing. The remaining six conditions, across the three experiments, showed that the effectiveness of the respondent-type training procedure in producing equivalence responding was dependent upon (a) the presence of longer between-pair-delays relative to the within-pair-delays and (b) the sequence in which the stimulus pairs were presented.