Analysis of the “discrimination-failure-hypothesis” in generalized matching and mismatching behavior

Three mentally retarded subjects were taught with matching-to-sample procedures and geometric shapes as stimuli to make matching and mismatching responses, respectively, in a two-component, multiple schedule. The subjects also made generalized matching and mismatching responses to unreinforced probe samplees in the respective components. During each trial the subjects were given the opportunity to choose the sample stimulus to which to respond. Choice responding was compared under conditions of no intertrial interval and 5-sec and 10-sec interial intervals. Generally, longer intertrial intervals resulted in a low probability of probe-sample selection and a high probability of reinforced-sample selection. Nevertheless, the subjects continued to display generalized matching and mismatching behavior when probe samples were selected, demonstrating that generalized matching and mismatching could be maintained concurrent with a discrimination between the reinforced and unreinforced sample stimuli as measured by the choice procedures. Further, the patterns of responding in the mismatch-reinforced component for all subjects were observed to change as intertrial interval conditions were changed across sessions. Under longer intertrial interval conditions, subjects responded to specific choice stimuli as mismatches for specific samples, but reverted to other strategies under no intertrial interval conditions. The results suggest that intertrial interval length may be an important variable in both matching-to-sample and choice behavior.