Reinforcement loss and behavioral tolerance to d-amphetamine: using percentile schedules to control reinforcement density.

Two procedures were used to examine the impact of reinforcement loss on the development of behavioral tolerance to the effects of d-amphetamine on control over response number in rats. Under both procedures, trials consisted of at least one left-lever press followed by a single right-lever press. Consecutive left-lever presses on each trial comprised a "run". A targeted percentile schedule provided reinforcement if the current run length was closer to the target length (12) than two-thirds of the most recent 24 runs. This procedure differentially reinforced runs around 12 while holding reinforcement probability constant at 0.333. A second group acquired the differentiation under the percentile schedule, but was then shifted to a procedure which yoked reinforcement probability by subject and run length to that obtained under asymptotic percentile schedule performance. The two procedures generated roughly comparable, but not identical, control run lengths, response rates, reinforcement probabilities and reinforcement rates. Only under the yoked procedure, however, did drug-induced disruptions in run length produce decreases in reinforcement density. Acute administration of amphetamine produced dose-related decreases in run length and overall response rate under both procedures. Daily pre-session administration of 1.7mg/kg amphetamine persistently suppressed run length under the percentile procedure, but not under the yoked procedure. Run lengths under the latter gradually increased with repeated amphetamine to a level equal to or slightly below baseline levels. Response rate was suppressed initially, and tolerance developed inconsistently to this effect in both groups. Dose-effect curves obtained when doses of amphetamine were substituted for the chronic dose showed a larger shift to the right with the yoked than percentile group for run length, and a similar but smaller effect for overall response rate. These results indicate that reinforcement loss substantially contributes to the development of tolerance to the behavioral effects of amphetamine, even when the comparison behaviors are generated by reinforcement contingencies that under non-drug conditions control very similar rates and patterns of behavior and reinforcement. Future comparison of acute and chronic drug effects on behaviors maintained by the percentile and yoked procedures may prove very helpful in illuminating drug-behavior interactions and the dynamic interrelations typically engendered by more traditional reinforcement schedules.