| Sumario: | Two subjects (Experiment 1) and one subject (Experiment 2) were given nonarbitrary relational pretraining that brought their responses under the control of four contextual stimuli; SAME, OPPOSITE, MORE-THAN, and LESS-THAN. Two further subjects in Experiment 1 were not exposed to this pretraining. All five subjects were then trained on the following six matching-to-sample tasks in the presence of three of the four contextual cues; SAME/A1-B1, OPPOSITE/A1-B3, SAME/A1-C1, OPPOSITE/A1-C3, MORE-THAN/A1-B2, MORE-THAN/A1-C2. Subject 5 in Experiment 2 was also trained on an additional two tasks. All five subjects were then tested for the following 10 derived arbitrarily applicable relations; SAME/B1-C1, OPPOSITE/B1-C3, SAME/B3-C3, OPPOSITE/B3-C1, MORE-THAN/B1-C2, MORE-THAN/C1-B2, LESS-THAN/B2-C1, LESS-THAN/C2-B1, LESS-THAN/B2-A1, LESS-THAN/C2-A1. All three pretrained subjects, but neither of the two nonpretrained subjects, passed these tests. Two of the stimuli from the relational network, B1 and B3, were then used to train two different self-discrimination responses using either detailed instructions (Subjects 1 to 4, Experiment 1) or minimal instructions (Subject 5, Experiment 2) on two complex schedules of reinforcement (i.e., subjects were trained to pick the B1 stimulus if they had not emitted a response, and to pick the B3 stimulus if they had emitted one or more responses on the previous schedule). All three pretrained subjects, and only one nonpretrained subject, showed the predicted transformation of self-discrimination response functions in accordance with the arbitrarily applicable relations of sameness and opposition (i.e., no response—pick C1, one or more responses—pick C3). Reprinted by permission of the publisher.
|
|---|