Conditional control of equivalence and the relations different and opposite : a behavior analytic model of complex verbal behavior

Behavior analytic approaches to the explanation of verbal behavior have been criticized because of difficulty explaining verbal productivity---the ability to make novel verbalizations which are in some way appropriate to the context. Match to sample procedures have resulted in the formation of equivalence classes which allow productive responding to untrained stimulus combinations. The central hypothesis of this study is that arbitrarily applicable relations other than equivalence can come to control human responding in ways which are productive. A second-order conditional discrimination procedure was used to establish control over sample-comparison selections where samples and comparisons were arbitrary visual stimuli. Pretraining with non-arbitrary stimuli gave second-order conditional stimuli the function of signaling which relation—same, different, or opposite---was to control sample-comparison discriminations. These pretrained second-order conditional stimuli were used to establish networks of relations between arbitrary visual stimuli. It was predicted that the network of relations could come to control untrained responding to probes which presented second-order conditional stimuli, samples, and comparisons in novel arrangements. The predicted pattern of responding was derived from formal logic