Relational Learning in Stimulus Sequences

Abstract

This paper proposes a new analysis of generative performances derived from contingencies that establish the production of stimulus sequences. Neither simple chaining nor within-sequence conditional stimulus control can account for the production of novel, untrained sequences that have been the outcomes of a number of experiments. The alternative analysis we propose focuses on relations among stimuli within sequences and across sequences trained independently of one another, and behavioral tests that evaluate whether such relations possess the properties of an order relation (i.e., asymmetry, transitivity, and connectedness). Our approach is an extension of Sidman’s analysis of relations among stimuli in equivalence classes. It provides a framework for examining how equivalence and order relations interact to yield extensive repertoires of novel behaviors that occur under appropriate circumstances. The concepts and procedures we present outline a behavior analytic approach to studying aspects of phenomena like syntactic development and transitive inference.