Extinction-Induced Response Resurgence: A Selective Review

Resurgence refers to the recovery of previously extinguished responding when a recently reinforced response is extinguished (e.g., Epstein, 1983, 1985). Figure 1 illustrates resurgence using hypothetical data. In the first condition, only Response A is reinforced. Response A is extinguished in the second condition concurrent with the reinforcement of Response B. Response B is extinguished in the third condition, and the subsequent recovery of Response A is resurgence. The degree of recovery of Response A can be compared to a different (control) response that never was reinforced (Response C). Measuring Response C in this third condition serves as a measure of extinction-induced response variability (e.g., Antonitis, 1951). Thus, resurgence provides a more refined characterization of extinction-induced behavior. That is, responses predicted to occur in extinction are those responses that previously were reinforced under comparable stimulus conditions. Although resurgence and related extinction-induced phenomena were investigated in the 1970s (e.g., Mulick, Leitenberg, & Rawson, 1976; Pacitti, & Smith, 1977; Rawson, Leitenberg, Mulick, & Lefebvre, 1977), as well as earlier (see Epstein, 1985), Epstein (1983, 1985) usually is viewed as being the most forceful commentator regarding the utility of understanding the determinants and consequences of resurgence (see also Epstein & Skinner, 1980). Epstein argued that resurgence provides behavior analysts with a more descriptive means of understanding findings sometimes attributed to Freud's concept of regression. For Freud, regression involved an individual regressing to previously demonstrated behavior considered acceptable earlier in the individual's development. In addition to linking resurgence and regression, Epstein also summarized much of the experimental findings related to resurgence. The aim of the present paper, therefore, is to describe more recent experimental findings related to resurgence. By showing the relevance of the topic to a variety of topics in and outside behavior analysis (behavioral history, drug relapse, severe problem behavior, communication disorders, and cognition), this paper urges both basic and applied behavior analysts to continue the analysis of resurgence. Behavioral history Appealing to the behavioral history of an organism often provides behavior analysts with a means of explaining complex behavior (e.g., Barrett, 1986; Doughty, Cirino, Mayfield, da Silva, Okouchi, & Lattal, 2005; Lattal, & Neef, 1996; Tatham, & Wanchisen, 1998; Wanchisen, & Tatham, 1991). Broadly speaking, resurgence is a behavioral-history effect in that current behavior only can be understood by appealing to contingencies of reinforcement exposed to the organism previously. A primary issue, therefore, in the study of resurgence is: what are the necessary and sufficient conditions to produce resurgence, in terms of both past experiences and current environmental conditions? Lieving and Lattal (2003) addressed the question posed above across several experiments. In one experiment, key pecking of pigeons was reinforced via grain delivery in the first condition. In the second condition, key pecking was extinguished concurrent with the reinforcement of treadle pressing (also via grain delivery). In the third condition, treadle pressing was extinguished, and the resurgence of key pecking was measured. By repeating these three conditions , and consistently observing key-peck resurgence in the third condition, it was confirmed that resurgence is a general and replicable finding. In two subsequent experiments, the necessity of extinguishing treadle pressing in producing key-peck resurgence was investigated. In one experiment, instead of arranging extinction in the third condition, food delivery was made response independent. Not only was key-peck resurgence absent in this condition, but in a subsequent condition key pecking resurged when extinction replaced the response-independent, food-delivery schedule. …