Abstract: | Four mentally retarded human subjects performed delayed identity matching-to-sample tasks under conditions of outcome-specific and nonspecific reinforcement contingencies. With outcome-specific contingencies, experimentally defined correct selections of different discriminative stimuli were perfectly correlated with one of two different consequences (e.g., different foods); with nonspecific contingencies, correct selections were equally likely to be followed by either consequence. Previous studies with pigeons have shown consistently higher accuracy on delayed-matching tasks with outcome-specific contingencies. Results of this study failed to replicate that finding. For three subjects, accuracy scores were similar in specific and nonspecific conditions at fixed delays ranging from 0 s to 5 s. For the other subject, delay duration varied within each session, and accuracy scores improved across sessions regardless of outcome condition. |