Basic and applied research on extinction bursts when treating problem behavior

Project Summary Results of epidemiological studies and meta-analyses indicate that treatments for destructive behavior based on functional analyses, like functional communication training (FCT), typically reduce destructive behavior by 90% or more and are much more effective than other treatments. However, these treatments involve extinction, which can produce extinction-induced adverse side effects. The most common side effect of extinction involves an increase in the frequency or intensity of the target response at the start of treatment, called an "extinction burst," which can place the patient and others at significant risk of harm. Prior theories on extinction bursts fail to account for the dynamic nature of these phenomena, and the basic behavioral processes that control bursts remain poorly understood. Our clinical and animal pilot data strongly suggest that our novel refinement of the generalized matching law, called the temporally weighted matching law (TWML), which incorporates reinforcement time and reinforcement history into the quantitative model, (a) fits our pilot data quite well, (b) resolves prior discrepancies between basic and clinical studies on bursts, and (c) identifies novel clinical procedures for preventing extinction bursts. The long-term goals of the project are to test our quantitative model of extinction bursts under typical clinical conditions to demonstrate its potential for improving treatment of destructive behavior and in the lab under a wider range of highly controlled conditions to provide a formal quantitative assessment of the predictions of the theory, evaluate its boundary conditions, and identify potential future clinical refinements. The specific aims are to demonstrate that large, discriminable drops in the rate (Spec. Aim 1), magnitude (Spec. Aim 2), and quality (Spec. Aim 3) of reinforcement when extinction or FCT are introduced increase the prevalence and magnitude of extinction bursts and that preventing such drops in the rate, magnitude, and quality of reinforcement will prevent or mitigate extinction bursts for both applied and basic research participants (i.e., humans referred for destructive behavior and rats, respectively). Specific Aim 4 will examine potential interactions between the variables of rate, magnitude, and quality of reinforcement predicted by the TWML to test current and inform future clinical refinements. The effects of reinforcement rate, magnitude, and quality will be manipulated and evaluated using a within-subjects design with the human clinical participants and using a randomized group design with lab rats, and the rate of destructive behavior (humans) and lever pressing (rats) will be the primary dependent variables.