Association for Behavior Analysis International

Benjamin Franklin famously observed that the only certainties in life were death and taxes. Almost two hundred years later, Bob Dylan equally memorably observed how the times they are a’changin’. Both of these famous American aphorisms emphasize the inevitability of change in everyday life. The papers in this symposium address different aspects of how behavior adjusts following changes in reinforcement contingencies. Lattal’s paper addresses some of the methodological issues in assessing change in both laboratory settings and applications. Kincaid next shows the effects of adjusting changes in reinforcement rates as a function of the organism’s (pigeon’s) behavior to the dynamic situation. Phillips and Hagopian then will review how schedule thinning techniques have been applied in treatment. Finally, Daniels will discuss how organizations and individuals in organizations adjust to change, with a particular emphasis on the importance of shaping.

Reinforcement-based treatments for problem behavior often involve extremely rich reinforcement schedules that are impossible or impractical to implement long-term. A potential solution is to this problem is to decrease reinforcement, also known as “thinning” the schedule. Schedule thinning describes a variety of techniques for gradually and systematically conducting the rich-to-lean transition. The underlying assumption is that a more gradual transition between rich and lean schedules will result in more maintenance of behavior across the transition. When a schedule thinning procedure fails, a common solution is to back up to a richer schedule and then re-thin. We investigated an analog of these conditions in a reinforcement-rate titration procedure. Pigeons responded on two concurrent variable interval schedules programmed on the same response key in a changeover key arrangement. The programmed reinforcement rate of one of the component schedules was constant across sessions, while the reinforcement rate of the other schedule was adjusted daily based on performance. If responding was at or above 80% of the baseline response rate, the schedule was thinned. If responding was below this criterion, the schedule was enriched. Different rates of thinning were compared to determine whether more gradual shifts in reinforcement resulted in greater maintenance of behavior.

Reinforcement schedule thinning in the context of behavioral treatment offers an applied example of the effects of manipulating reinforcement density. Thinning is defined here as any procedure that involves decreasing the density of reinforcement within a treatment session. Examinations of the literature encompassing two common treatments for problem behavior, differential reinforcement of alternative behavior (DRA; specifically, functional communication training, FCT) and fixed time delivery of reinforcement (i.e., noncontingent reinforcement, NCR) reveal a number of methods for thinning. These include: delay schedules, chain schedules, multiple schedules, and in the case of FCT, response restriction. Although each method has been demonstrated to be effective, thinning can often result in increases in problem behavior and decreases in appropriate behavior. Several examples from our research will be discussed that highlight issues with the ways in which reinforcement schedules for problem behavior and alternative behavior are manipulated in the pursuit of a terminal schedule that can be implemented outside of the research environment. We will also describe supplemental treatment components which somewhat ameliorate the effects of thinning. However, important questions remain. What is happening during thinning? Does changing the density of the reinforcer produce local extinction in some cases? What do schedule thinning-induced changes in behavior reveal about the mechanisms by which different treatments reduce behavior? We will attempt to address these questions by describing possible directions for future research.

Behavior analysis traditionally is a science of the steady state, in which changes in a dependent variable are measured as the effect of an independent variable imposed on baseline behavior that has been trained for some time under a condition in which the independent variable was either not present or present at another value. Sprinkled throughout these numerous studies of steady-state behavior, however, are a few studies of behavior during the transition periods. These include studies of both response acquisition and extinction, adjunctive behavior, behavioral history, and resurgence. Some of these experiments involve a single transition, as in the transition to extinction, but others involve repeated transitions that result in a virtual steady state of transitional performance, as in the repeated acquisition of behavioral chains. In this paper I will review these various preparations for studying behavioral transitions and then consider some of the variables that attenuate and potentiate behavioral change when conditions change.