Response elimination, reinforcement rate and resurgence of operant behavior

Abstract

The effects of reinforcement rate of alternative responding on resurgence were studied in six experiments with pigeons. In Experiment 1A, key pecking was maintained on a multiple variable-interval (VI) VI schedule in the Training phase. In the Response-Elimination phase, a variable differential-reinforcement-of-other-behavior (DRO) schedule was in effect in each component. Reinforcement rates were equal and then, higher in one (rich) component, and lower in the other (lean), than in the Training phase. More resurgence occurred in the lean component, but this could have resulted from response-rate differences between components in the Training-phase. Experiment 1B was a replication of Experiment 1A, but with experimentally-naïve pigeons. Response-Elimination phase reinforcement rates were manipulated systematically in subsequent experiments: In Experiment 2, reinforcement rate was equal, in one component, and lower or higher in the other, than in the Training phase. In Experiment 3, reinforcers were discontinued before differential reinforcement rates were effected. In Experiment 4, reinforcement rates first were differential and, then, equal to those in the Training phase. In Experiments 5 and 6, differential reinforcement rates were arranged by using fixed-DROs and VIs for pecking a different key, respectively. Even though resurgence was not obtained with every pigeon, at least some small-magnitude resurgence occurred in each experiment and was not related systematically to reinforcement rates of alternative responding. Schedule differences, response topography, order of conditions and the length of each phase were not sufficient to account for these results.