Association for Behavior Analysis International

The studies in this symposium represent current approaches to the experimental analysis of resistance to change and relapse. Specifically, the effects of different schedules of reinforcement, of operant-variability contingencies and of reinforcement parameters on these phenomena were assessed. Viegas and Cançado report the results of a within-subject parametric analysis of resistance to change of responding maintained by VR and yoked-VI schedules of reinforcement. Aló and colleagues assessed the effects of relative-reinforcement rates on response rate, choice, and relapse under a renewal procedure. Chaparini and Abreu-Rodrigues studied the relapse of patterns of operant variability by combining the procedures used to study resurgence, renewal and reinstatement effects. Hernandez and Flores report the effects of variations on several reinforcement parameters (frequency, delay and magnitude) for the target response during the second phase of a resurgence procedure. The method of each study exemplify approaches to further understanding the controlling variables of both resistance to change and relapse, and their results extend those of previous studies on both phenomena. Collectively, as the results of experimental analyses do, the results suggest a broader understanding of resistance to change and relapse that encompasses the effects of understudied variables (schedules of reinforcement, operant-variability levels and reinforcement parameters).

Resistance to change usually is greater in variable-interval (VI) than variable-ratio (VR) schedules under similar reinforcement rates. In this experiment, we are assessing how parametric manipulations of the response-rate difference between VI and VR schedules affect resistance to extinction. Rats are being exposed to each of six conditions. In each, a multiple VI VR is in effect in baseline and extinction in the test, and two tests are being conducted: one in which sessions start with the VI and another in which they start with the VR. In the first condition, baseline response rates were similar between components. Resistance was greater in the VR than in the VI in the VR-first test (in the VI-first test, resistance was similar between components). In the second condition, a differential-reinforcement-of-low-rate schedule was programmed in tandem with the VI and response rates were lower in the VI than in the VR schedule. For each rat, resistance was greater in the VI but only in the VI-first test (in the VR-first test, there were no systematic differences between components across rats). The results suggest that response-rate differences affect resistance to extinction and these results are modulated by the first component in effect during test sessions.

The effects of relative reinforcement on response rates, choice, resistance to change, and renewal is being studied. During baselines, pigeons were exposed to a multiple-concurrent schedule with a concurrent VI 80 s VI 40 s (VI40-Rich) on one component and a concurrent VI 20 s VI 40 s (VI40-Lean) on the other component. All schedules were presented in different keys and using different colors. During satiation tests, free food was delivered before sessions. During baselines, response rates were highest either on the VI-20 or on the VI-40 Rich schedule and lowest either on the VI-80 or on the VI-40 Lean schedule. Choice for the VI-40 was greater when it was presented with the VI-80 rather than the VI-20 schedule. During the tests, choice for the VI-40 schedule approached in both components. Proportion of baseline response rates were not consistent with either absolute or relative reinforcement rates, during test session. Thus, response rates and choice followed relative reinforcement rates. Decreasing the value of the reinforcers by satiation, however, decreased sensitivity to relative reinforcement rates and produced mixed effects on Resistance to change. The renewal test has not yet been conducted.

Relapse has been defined as the recurrence of a response or pattern of responding, previously reinforced and later extinguished. Recurrence of behavioral variability was evaluated with a combination of three animal models of relapse: renewal, reinstatement, and resurgence. In the Training Phase (context A), rats were exposed to a multiple schedule. In the Lag 10 component, sequences that differed from the previous 10 sequences were reinforced; in the Yoke component, any sequence produced reinforcers such that variability was allowed, but not required. Reinforcer probability was equated between components. In the Elimination Phase (context B), a multiple Rep 3 Rep 3 schedule was in effect. In both components, a sequence was reinforced if it was equal to one of the three previous sequences (i.e., variability was extinguished). In the Test Phase (context A), extinction occurred in both components, and two response-independent reinforcers were delivered in each component. Relapse of recurrence time, a measure of sequence variability, was observed in the component previously correlated with Lag 10, but not in the one previously correlated with no variability requirements. The increase in variability in the Test phase indicated relapse of previously reinforced variability, rather than extinction induced variability.

Problem behavior can be decreased by means of extinction and reinforcement of an alternative behavior. However, when both types of responses are placed under extinction conditions, previously-decreased behavior can reappear. Resurgence has been observed when alternative-reinforcement contingencies (e.g. extinction, magnitude, delay and frequency of reinforcement) are degraded during the test. Under such procedure, target responding is decreased during the second phase by means of extinction. Even when target behavior can be decreased by means of extinction, responding can also be reduced by varying different reinforcement parameters. Rats were exposed to a resurgence procedure, in which target responding during phase 2 was decreased by means of extinction, magnitude, delay and frequency of reinforcement between conditions. For all rats, target-responding decreased as a result of changes in reinforcement parameters; however, responding decreased to near-zero responses only under extinction conditions. Similarly, during the latter condition resurgence was observed for all rats; whereas, resurgence was partially observed in the rest of the conditions for some of the rats. Our results suggest resurgence can be observed after target-responding is decreased by changes on reinforcement parameters. However, near-zero responding seems to be required for resurgence to occur once target and alternative responding are placed under extinction.