Association for Behavior Analysis International

Changes in contingencies of reinforcement results on recurring and variable responding. Treatment relapse has been assessed using a resurgence procedure in basic and applied settings. The first two studies will discuss strategies for mitigating treatment relapse in humans and rats. In the first study, Schieltz et al. evaluated the use of response latency as an incremental measure of long-term maintenance. In their study, initial demonstrations of long-term maintenance are prescribed by longer latencies to challenging behavior in comparison to shorter latencies to independent manding during treatment. Madrigal et al. assessed resurgence under conditions of extinction and decreasing frequencies of reinforcement, reporting an increase in resurgence as contingencies of reinforcement degraded. In a third study, Benavides et al., compared the levels of operant variability generated by a Lag n criterion with variability generated by VI and VR schedules with rats. Lastly, Gutierrez et al., manipulated the value of the Lag using a within session adjustable procedure with rats. Results on both the latter studies suggest that previous procedures in the study of operant variability have limited the maximum level of variability observed so far. These studies contribute valuable insights into understanding and addressing treatment relapse and operant variability.

Resurgence is defined as the recurrence of a previously reduce response when contingencies of reinforcement of an alternative response are worsened. Recently, Yensen et al. (2022) reported resurgence by worsening the contingencies of reinforcement for alternative responding. This worsening of conditions was arranged by transitioning from a VI to a FI schedule, and by decreasing the frequency of reinforcement. This current study builds upon Yensen et al.'s work, aiming to replicate their findings and account for unconsidered variables. Rats were exposed to a within-session resurgence procedure. In the first experiment, target and alternative responses were trained under VI30s and VI50s schedules, respectively, while in the second experiment, equal frequencies of reinforcement were arranged for both responses. Consistently and repeatedly, resurgence was observed in both experiments when alternative responding was placed under extinction conditions. However, as reported by Yensen et al., resurgence was only observed under the FI100s schedule after the target and alternative responses were trained with unequal reinforcement frequencies. Results are discussed by identifying some of the possible conditions under which resurgence can be observed.

There are two main sources of behavioral variability. One is its direct reinforcement, for example, through a Lag n procedure. In these procedures, a sequence of responses is reinforced only if it is different from n number of previous trials. The other is by induction through intermittent reinforcement. Given that both procedures are relatively different, there is no direct comparison between the variability generated by them. The purpose of these experiments was to compare the levels of variability generated by a Lag n criterion with variability generated by variable interval and variable ratio schedules with rats. A two-component multiple schedule was used, with one component in which a sequence of 4 responses distributed in 4 or 2 levers was reinforced with a Lag 5 criterion. The delivery of reinforcers in the second component was coupled to the performance in the first component, but with a criterion based on the intermittent reinforcement schedules. Higher levels of variability were found during the VI than during Lag, while higher variability was found during Lag than during VR. The results are partially congruent with both induced and operant variability literature. Possible limitations and future avenues of the procedure are discussed.