Association for Behavior Analysis International

Stimulus equivalence has been an active field of research in Behavior Analysis, producing robust results and important applications. Old controversies and new research problems have kept the field alive and significant research continues to be produced. The four papers in this symposium, be researchers of three different countries, explore different avenues of current research on stimulus equivalence. Two basic studies deal with the relationship between stimulus equivalence and meaning: one of them investigates how meaningful stimuli influence the probability of equivalence class formation and the other uses a semantic differential to investigate to what extent equivalent stimuli share meaning. A translational study with cochlear implant users is also relevant to the question of meaning. It investigates the use of the equivalence paradigm to impart meaning to auditory stimuli in previously deaf children. The symposium concludes with a theoretical analysis of substitutability between equivalent stimuli, proposing how equal and unequal relatedness can coexist in equivalence classes.

When equivalence classes included a meaningful stimulus together with abstract ones, evaluations with a semantic differential showed that meaning transferred to the abstract stimuli. Some experimental parameters, however, interfered with transfer of meaning. The present study trained conditional relations AB, AC, and CD (three samples and comparisons each). A1, A2, and A3 were faces expressing happyness, neutrality, and anger, respectively. The other stimuli were abstract pictures. Semantic differential evaluations showed that D1, D2, and D3 acquired meanings similar to the equivalent faces. Conditional discrimination CD was then rearranged to C1D2, C2D3 and C3D1. Classes rearranged as expected. However, semantic differential evaluations showed that meanings of the stimuli did not reverse. Mean evaluations of the D stimuli now equivalent to the angry and happy faces became similar. Although the classes appeared to rearrange, on the basis of class formation tests, the stimuli that changed class membership did not acquire the meaning of the faces to which they were now equivalent. Class rearrangement is then another experimental condition that interferes with transfer of meaning between equivalent stimuli.

This study, conducted with children users of cochlear implant, isolated two components of dictated words as sample stimuli. Stimuli in set A were video clips of a person speaking words without any sounds (lips movements); stimuli in set C were dictated words presented through the computer speakers (sounds only, without orofacial cues); stimuli in the B set were pictures. The teaching procedure was simultaneous matching-to-sample with three comparison stimuli. The order of training was AB/CB for half of participants, and CB/AB for the other half. Tests evaluated the emergence of equivalence relations (AC/AC) and of two operant relations derived from the matching training: picture naming and echoic responding, with both trained and novel stimuli. All participants learned six conditional relations (3 AB and 3 AC). The acquisition of conditional discriminations was faster with auditory sample stimuli. All participants formed equivalent classes between pictures, dictated words, and lips movements. Echoic behavior and picture naming also increased for 3 participants, but echoic responding improved more than picture naming. The results confirm previous findings that, through the implant, hearing impaired children can learn arbitrary relations between sounds and other events, and that sounds acquire symbolic meaning through participation in equivalence classes.

The traditional view of equivalence classes is that all class members are substitutable for each other, and, by implication, equally related to each other. This assumption, however, does not fully or accurately describe the strength of relations among the stimuli in an equivalence class. A growing set of post-class-formation within-class, steady-state measures of preference demonstrates that the relational strength among the stimuli in an equivalence class are strongly influenced by at least two structural properties of equivalence classes. Specifically, the strength of relations between stimuli in an equivalence class is an inverse function of the number of nodal stimuli that separate the stimuli in the class. In addition, transitive relations are preferred to equivalence relations, which implies that relational strength is also an inverse function of the number of logical operators the define derived relations. These results then show clearly that the stimuli in an equivalence classes are not substitutable for each other and, by implication, are unequally related to each other. The equal and unequal relatedness of stimuli in an equivalence class actually coexist and are expressed by different types of tests. Performances indicative of equal relatedness are expressed in the presence of cross class tests, while performances indicative of unequal relatedness are expressed in the presence of within class tests. Thus, the two views of relatedness are complementary to each other and do not reflect the effects of procedural confounds.