摘要：In an earlier work (Gomez, Barnes-Holmes, & Luciano, 2001), it was found that although generalized break equivalence was achieved for 3 subjects (i.e., responding to new stimulus sets producing symmetry and transitivity but not equivalence; Cx-Ay, Cy-Ax) , only 1 subject showed this pattern of responding during the first test exposure, and 2 additional subjects did not produce the derived pattern. The present study attempted to develop a procedure that produces generalized break equivalence more effectively than in our previous research. To achieve this goal three experiments were conducted varying the contextual cues used for the abstraction of the Generalized Break Equivalence Pattern (GBEP). In Experiment 1, the main contextual cue consisted of one, two, three, or four asterisks to differentiate each type of relation (i.e., one for A-B, B-A; two for B-C, C-B; three for A-C; four for C-A). Four subjects produced the GBEp, and the most frequent errors were produced on the A-C relation. In Experiment 2 the contextual cue was that the C stimuli were formally different from the A and B stimuli. Three subjects produced the GBEP and the most common errors were produced on C-B relation. Finally, in Experiment 3 an "explicif' contextual cue appeared only on C-A trials. All subjects showed the GBEP (2 of them in the first test to which they were exposed). These data extend those found earlier (Gomez et aI., 2001) and support the generalized operant nature of derived relational responding. Applied implications are discussed.
