摘要：The present study was a modified replication of a paper-and-pencil format study by Eikeseth, Rosales-Ruiz, Duarte, and Baer (1997) on equivalence relations derived from instructionally induced conditional relations. The study consisted of three experiments, all with Dutch psychology students as subjects. After being instructed to memorize four printed examples of arbitrary A-B and B-C conditional discrimination tasks and completing A-B and B-C trials in the presence of these examples, the subjects received a series of probe trials (no access to the examples): baseline and symmetric transitivity (C-A) probes (Experiment 1), or baseline, symmetry (B-A, C-B), and symmetric transitivity probes followed by a sorting test (Experiments 2 and 3). Without the option to skip "impossible-to-solve" probe trials (Experiments 1 and 2), almost all subjects (99%) completed all training and all probe trials. Most subjects (87%) who responded correctly on the baseline training trials also responded correctly on the baseline probes. Most of these subjects responded correctly on the symmetry trials (87%), the symmetric transitivity probes (81%), and on the sorting test (76%). Symmetric transitivity was seen most often when tested after symmetry. The performances on the sorting test corresponded with the numbers of derived relations (symmetry and transitivity; symmetry or symmetric transitivity; no symmetry nor symmetric transitivity) rather than with equivalence per se. The introduction of the default option (Experiment 3) resulted in most subjects skipping and responding inaccurately on the symmetry and symmetric transitivity probes.
