Effects of different histories of reinforcement and extinction on the behavioral variability

Abstract

This study attempted to (1) compare the behavioral variability produced by two contingencies, (2) verify whether the order of exposure to these contingencies might interfere with the acquisition and maintenance of variability, and (3) establish how extinction interferes in the reacquisition of variability. Reinforcement was delivered to fourteen male Wistar rats contingent upon four lever-pressing responses on two levrs (right -D and left -E). The behavioral unit examined was the sequence of four responses which, depending on the combination between D and E, could result in 16 different sequences (e.g., EDDE, DEDD, EEEE, etc.). While under LAG-5 contingency, the sequence was reinforced only if it differed from the five previous sequences. Under RDF contingency, the reinforcement was delivered only if the relative frequency of the sequence was equal to or less than 0.0625 (1/16). In addition, every response sequence that was emitted caused the frequency of every other sequence to be multiplied by a factor of 0.99, decreasing their frequency and, consequently, increasing their reinforcement probability. Therefore, when under RDF contingency, the lower the frequency and recency of the current sequence, the higher the probability of reinforcement; under LAG-5 contingency, the reinforcement probability was 1.0 as long as the current sequence differed from the last five, and 0.0 for those that did not meet this criterion. The subjects were divided into two groups, exposed to two contingencies of reinforcement and the extinction procedure. The groups differed from each other relative to the order of exposure to these contingencies. The experimental conditions were arranged according to an ABACA or a BABCB design, where A stood for the LAG-5 contingency, B for the RDF and C for the extinction. The results showed that (1) both contingencies produced behavioral variability, but to different extents: higher levels of variation were systematically observed under RDF schedule, whichever the order of exposure to the contingencies; (2) the condition of extinction generated a decrease in the response rate, but maintained the variability levels close to those obtained in the previous phase, with only a small interference from the reinforcement history; (3)the variability obtained under reinforcement in the last experimental phase was not affected by the previous exposure to extinction: the patterns of responding presented under LAG or RDF were typical of these schedules. These results indicate behavioral sensitivity to different reinforcement contingencies for variation, with its acquisition and maintenance controlled primarily by the contingency in effect, with little interference from the reinforcement history. Moreover, extinction induced variability, but did not interfere in its operant control. The processes of resurgence, optimization, and maximization have been considered for data analysis.