Association for Behavior Analysis International

This symposium will present current research on behavioral persistence and relapse. The persistence of problem behavior and the recovery of once-extinguished behavior following treatment are common issues in applied/clinical settings. Findings from laboratory settings suggest some variables relevant to the issues. Ribeiro, Kelley and Tanz point out that noncontingent reinforcement, which is a common technique for decreasing problem behavior, may actually increase its persistence in children diagnosed with autism. Ferreira and Canado show that behavioral persistence can partially be determined by the dependency between a response and the reinforcer that follows it. Abreu-Rodrigues, Canado, Ferreira and Siqueira, and Kuroda, Podlesnik and Canado, show that the manipulations of contextual stimuli (i.e., renewal procedures) also affect the degree of resurgence in rats and humans. Our discussant, Mark Bouton, will address implications of these studies. This symposium should be informative for both basic and applied researchers.

We evaluated the persistence-strengthening effects of noncontingent reinforcement (NCR) and some factors that may interfere with maintenance of treatment effects. Basic, translational, and applied research have demonstrated that the resistance of a given response rate to change depends on the baseline overall rate of reinforcement, regardless of response rate, such that the greater the rate of reinforcement the greater the resistance of behavior to change. These findings suggest that adding reinforcers to a context in which problem behavior has been reinforced (as it is usually the case with NCR) is likely to increase behavioral persistence when treatment is challenged. In the present study, we evaluated the effects of NCR on behavioral persistence during extinction (Experiment 1), with five children with autism, and during other disruption procedures (Experiment 2), with three autistic children. Basically, we compared responding during disruption following a reinforcement only condition and responding during disruption following a condition with both contingent and non-contingent reinforcement. Results from Experiment 1 were not consistent across participants and may be related to the procedure used to test for resistance. In Experiment 2, we found that behavioral persistence was greater after NCR for all three participants, which replicates previous findings on behavioral persistence.

A parametric analysis was conducted with four rats to investigate the effects of different percentages of response-dependent food on resistance to change. Lever pressing was maintained under a three-component multiple schedule with the same rate of food in each component. The percentage of response-dependent food was 10 and 100 in the first and second components, respectively, throughout the study. The percentage in the third (Alternative) component varied across conditions from 10 to 80 for each rat. In each component, interfood intervals were variable. Responding in the first component (10%) consistently was more resistant to change than that in the second component (100%). When the percentage of response-dependent food was 10, 20, or 30 in the third component, resistance to change approximated that in the first component. When it was 50 or 80, resistance to change was closer to that in the second component. Overall, resistance to change was an inverse function of the percentage of response-dependent food. Replicating previous findings, these results suggest that resistance to change is at least partially determined by aspects of the response-reinforcer relation.

The effects of contextual stimuli on resurgence were studied in two experiments with rats. In each, lever pressing was reinforced on a multiple variable-interval (VI) 20 s VI 20 s schedule in the Training phase. In the Elimination and Test phases, respectively, a differential-reinforcement-of other-behavior schedule (DRO) and extinction were in effect in each component. Contextual stimuli (houselight illumination patterns) were manipulated across components and phases. In Experiment 1, contextual stimuli were similar across phases in one component (AAA, with letters representing the context in each phase); in the other component (ABA), contextual stimuli changed from Training to Elimination phase, but in the Test phase, the context was similar to that in effect in the Training phase. In Experiment 2, the context changed from Training to Elimination phase in both components. In the Test phase, the context was that in effect in the Elimination phase in one component (ABB), and that in effect in the Training phase in the other component (ABA). Resurgence of greater magnitude occurred in the ABA than in the other component for all rats in Experiment 1, and for two of three rats in Experiment 2, indicating the relevance of contextual control for conceptualizations of resurgence.