Predicting taxonomic and thematic relational responding.

Pairs of pictures were classified by the authors and others as related by identity (A-A), basic taxonomy (A-B), superordinate taxonomy (A-C), or by theme (A-D). Two-choice matching-to-sample trial types were composed of these same picture pairs in which the sample was common to the two stimulus pairs in each configuration and, together with the sample, each comparison exemplified one of the relations in the picture pair; that is: A(AB), A(AC), A(AD), A(BC), A(BD), and A(CD). In five experiments, for each picture pair, college students classified the relation (as taxonomic or thematic) and rated its strength (Exps 1, 3) or its similarity (Exp 4); others matched to sample the foregoing trial types only (Exps 2, 5), or they classified and rated, too (Exp 3). With exceptions, students classed most pairs as the authors did. They also collectively ordered relational strengths from (1) identity, (2) basic taxonomy, and (3) theme, to (4) superordinate taxonomy based, in part, on the similarity of sample and comparison. Subjects chose the comparisons of the more strongly related picture pairs in the matching-to-sample task on 90 percent or more of the configurations. Subjects' selections in two-choice, matching-to-sample configurations using natural stimuli may be based on existing stimulus control topographies such as those exhibited by ratings of the relations in a configuration.