Analysis of apparent demonstrations of responding in accordance with relational frames of sameness and opposition by Alonso‐Alvarez and Perez‐Gonzalez (2018): A rejoinder

Alonso-Alvarez and Perez-Gonzalez (2018) describe two experiments on contextual control over derived relational responding (DRR), designed to investigate derived sameness and opposition relations as described within Relational Frame Theory (RFT; Dymond & Roche, 2013; Hayes, Barnes-Holmes & Roche, 2001). They argue that the outcomes of their experiments suggest that all previous empirical demonstrations of derived sameness and opposition might be explained more parsimoniously in terms of stimulus equivalence and exclusion. In what follows, we address limitations in the authors’ analysis and offer a rejoinder to their thesis. Their argument is first presented in a prior study by Alonso-Alvarez and Perez-Gonzalez (2017). In RFT-inspired studies into derived same and opposite relations, two arbitrary stimuli (hereafter denoted SAME and OPPOSITE, respectively) are first pretrained as cues for nonarbitrary same and opposite responding (e.g., Dymond, Roche, Forsyth, Whelan & Rhoden, 2007, 2008; Whelan & Barnes-Holmes, 2004). This procedure continues until participants select comparisons identical to the samples given SAME, and comparisons the most physically dissimilar from the samples given OPPOSITE, in unreinforced testing. Then, the participants are trained to form arbitrary stimulus relations in the presence of those cues (e.g., SAME-A1B1, SAME-A1C1, OPPOSITEA1B2, OPPOSITE-A1C2) and tested for derived relations (e.g., SAME-B1C1, SAME-B2C2, OPPOSITE-B1C2, OPPOSITE-B2C1). In the case of the examples just given, correct responding would be taken as evidence for derivation of combinatorially entailed relations of same and opposite. B1 and C1 should be responded to as the same, as both are the same as A1. B2 and C2 should be derived as the same, as both are opposite to A1 (i.e., combining two opposite relations should yield a same relation). B1 and C2 should be derived as opposites, as should C1 and B2, because both cases involve combining a same and an opposite relation (i.e., which always yields an opposite relation). Alonso-Alvarez and Perez-Gonzalez (2017) argued that a pattern of responding such as this might instead be explained in terms of equivalence, nonequivalence and exclusion responding. They suggest that “the cue for the selection of comparisons identical to the samples (SAME) could become a cue for selecting comparisons equivalent to the samples” while the “cue for the selection of comparisons the most dissimilar to the samples (OPPOSITE) could become a cue for the exclusion of comparisons equivalent to the samples (i.e., nonequivalent comparisons)” (p. 230). Following this, they argue that training Same-A1B1 and Same-A1C1 could produce the equivalence class A1B1C1, which would explain the choice of C1 given B1 and vice versa, with SAME as cue. In addition, training selection of “B2 and C2 [given] A1 and OPPOSITE would [make] B2 and C2 nonequivalent to A1”. Further, because “B1 and C1 would be equivalent to A1, then B2 and C2 would also be nonequivalent to B1 and C1. Thus participants would [select] C1 [given] B2 and OPPOSITE because C1 and B2 [are] nonequivalent.” Likewise, they would select C2 given B1 and OPPOSITE because C2 and B1 are equivalent. Finally, they would select “C2 given B2 and SAME by the exclusion of C1. If C1 and B2 [are] nonequivalent, then C1 could not [be] the correct choice [given] B2 and a cue for equivalence (SAME). Thus, participants [exclude] C1 and [select] C2” (p.231). They Address correspondence to: Simon Dymond, Email: s.o. dymond@swansea.ac.uk doi: 10.1002/jeab.555 JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2019, 112, 349–353 NUMBER 3 (NOVEMBER)