Schedules of interresponse time reinforcement

In the experiments reported in this study, data were obtained with rats on the effects of interresponse time (IRT) contingent reinforcement of the lever press response. It was found that (1) as the width of the reinforced IRT interval decreased, recoverability following exposure to other interval widths decreased; (2) exposure to one value of the lower bound of the reinforced IRT interval did not affect the final performance under a new value of the lower bound; (3) individual differences in the form of the IRT distributions were maintained to a considerable extent across different values of the lower bound; (4) in one set of experiments, homogeneous response chains intervened between lever presses; and decreases in the width of the reinforced IRT interval resulted in increases in the pause after reinforcement and decreases in the modal IRT; however in other experiments little chaining was observed and no relation between the pause after reinforcement or the mode of the IRTs and the width of the reinforcement interval was obtained; these discrepancies were tentatively attributed to differences in the magnitude of reinforcement; (5) the peakedness of the IRT distributions was a positive function of the width of the reinforcement interval and probability of reinforcement, a negative function of the value of the lower bound of the reinforcement interval, and unaffected by upper bounds which were no smaller than twice the value of the lower bound; (6) when the IRT distributions were computed with the class interval widths proportional to the lower bounds, the distributions tended to become congruent regardless of the value of the lower bound, although there was still a slight positive relation between the lower bound and the flatness of the distribution; (7) for schedules involving the reinforcement of IRTs in either of two intervals, as the probability of reinforcement in the first interval increased, the principal mode of the IRT distribution shifted from the lower bound of the second interval to that of the first interval, while at intermadiate probabilities the mode lay between the two lower bounds; and (8) at small widths of the first interval, a mode was present near both lower bounds.