Definition

Aversive and nonaversive interventions refer as much to a dynamic yet functional definition of both terms as to a set of intervention procedures. From a technical point of view, an aversive intervention involves the application of an aversive stimulus. This would include a noxious event that serves as a punisher when it follows behavior, one that evokes a behavior that has terminated the noxious stimulus in past circumstances, or one that functions as a reinforcer when it is removed after the occurrence of a behavior (Cooper, Heron, & Heward, 2007). A nonaversive intervention involves the application of positive reinforcement and/or extinction contingencies as a consequence to a behavior, or alteration of the intensity, duration, or magnitude of a behavior contingent upon the removal or presentation of an antecedent stimulus.

Historical Background

While treatment in autism has, over the years, had many controversies, perhaps none have been so heated as the discussion of the viability and appropriateness of aversive and nonaversive procedures to treat a variety of problems common to the disorder (and to those with other neurodevelopmental disorders as well). These controversies have pitted, in somewhat of a dichotomous fashion, empirical science against social validity. The result was at once unfortunate and the stimulus for a paradigmatic shift. When in its relative infancy, the science of the experimental analysis of behavior served a very important function: to prove that even the most recalcitrant of human behaviors are subject to the laws of learning and can be improved upon. For generations of clinicians raised on the belief that change was only possible in small increments for those severely affected by autism, and then only through rather drastic psychopharmacologic interventions, the opportunity to demonstrate progress in reducing self-injury, aggression, and other destructive behavior as well as to increase prosocial, adaptive behavior was a breakthrough. Applications of more basic operant conditioning principles such as positive and negative reinforcement, extinction, and punishment were tactics of choice during this period. Indeed, clinical significance was often defined only in terms of the magnitude of behavior reduction (the end product) but rarely so by the means of reduction.

As the 1960s progressed through to the late 1970s, however, means of intervention appropriately became a more prominent consideration. The seminal work on social validity by Kazdin (1977) and Wolf (1978) reshaped the narrative around three key points: not only must the outcome of intervention be socially valid but also the target of intervention (behavior to be changed) and also the means to achieve that outcome. As a construct, social validity imposes the requirement that all factors be considered before, during, and after treatment. This demand served several important functions. It posed the important question, “socially valid for whom?” Were targets, procedures, and outcomes socially valid for the client, the family, institutionally based caregivers? Social validity also raised the question of relativity. At different points in time, for different clients, and under particular circumstances, a treatment procedure or outcome might or might not be acceptable. But very importantly, at its base, the question of social validity also raised the issue of the generalizability of behavior change. While behavior analysis had evolved very good technologies of generalization and maintenance (Horner, Dunlap, & Koegel, 1988), things did not always work out as planned. So-called treatment failures continued to occur, often under the contingencies of more remote or diverse (and sometimes less well-understood) events. By imposing the demand to assess for social validity, interventionists had a tool to begin to predict potential functional relationships between change agents and the consumers of change and to begin to modify those contingencies that might interfere with long-term maintenance and generalization.

Within the span of a few years, however, a number of flashpoint events occurred that sharpened the issues concerning treatment of those whose autism placed themselves, and others, at the greatest risk. Highly publicized reports of the deaths of clients in the care of otherwise well-known residential programs following the use of contingent aversive procedures (e.g., white noise) changed the conversation from one of science alone to a discussion of human dignity and the right to effective treatment. Suffice it to say that while at times mean-spirited, personal, and derogatory, the power of the objectivity of science won out. Indeed, not only did the National Institutes of Health fund a number of collaborative research centers with the mandate to investigate and develop effective interventions that were nonaversive, but the NIH later convened a consensus conference (National Institutes of Health, 1991) in order to issue guidelines for the use of behavior reduction procedures (including punishment strategies) when treating destructive behavior in those with developmental disabilities. The efforts of established collaborative research centers, other scientists working in basic and applied settings, and the general understanding of the effects (and negative effects) of punishment have led over the past 20 years to a highly developed, evolving, evidence-based technology of behavior change based upon the use of antecedent and consequent control procedures that do not involve the use of aversive stimuli. To be certain, the controversy has not ended entirely, as those who empirically demonstrate the effective use of punishment procedures as a component of a comprehensive treatment package would argue (Axelrod, 1990). But, as importantly, the exceptional science being developed to understand the often complex functional motivators behind severe behavior continues as well and is especially visible in the efforts of those promoting positive behavior support initiatives in public schools.

Rationale or Underlying Theory

Given the extensive research base for both aversive and nonaversive interventions, it is reasonable to conclude that considerations about each are evidence-based. The important considerations, however, lie in the issue of negative effects and generalizability of effects. Both sets of procedures are based on the principles of operant conditioning earlier described by Skinner, with many decades of subsequent and substantive empirical extensions of that work. What has evolved over the years is a toolbox of intervention strategies, many working best as part of multicomponent procedures. While there may well be occasions for which a punishment procedure – in combination with positive reinforcement procedures designed to increase functionally equivalent, alternative prosocial behavior – is the least restrictive intervention option, intervention based on punishment alone is rarely advised.

Nonaversive interventions are broadly organized around antecedent strategies (those that occur before the problematic behavior is emitted), with the intention of altering the stimulus control and reinforcing value of the existing antecedent “triggers” for the behavior. Consequent procedures are those delivered after behavior has been demonstrated. They can include reinforcement-based procedures, extinction, and variants of interruption and redirection. In contrast, aversive interventions involve the application of an aversive or unpleasant stimulus immediately following the problem behavior, designed to discourage future occurrence of the behavior. In all cases, however, whether an intervention is aversive or reinforcing to a client is a functional question. If the application of a stimulus immediately following demonstration of a specific behavior increases the probability of that behavior occurring, the stimulus was reinforcing. If presentation of the stimulus immediately following the behavior reduces the likelihood of behavior reoccurrence, then the stimulus was aversive. Referring back to the discussion of social validity earlier, what is aversive to one person may be reinforcing to another. The only solution is to assess functionally before and during treatment implementation.

Ultimately, the rationale about which intervention strategies to employ in a particular case is a functional one, clarified by a thorough functional behavior assessment/analysis and subjected to rigorous outcome evaluation. In the final analysis, intervention must be effective, that is, it must be successful in its outcome and have minimal or no negative effects associated with it. Treatment strategies that are socially valid and empirically based will best serve the interests of persons with autism and related neurodevelopmental disorders.

Goals and Objectives

The selection of intervention strategies is based on behavioral function, not form. Function can be described in several ways. For example, behavior can serve to access positive reinforcement in the form of social attention or access to preferred materials. The behavior can be functionally reinforced by its ability to terminate an aversive or unpleasant event (negative reinforcement). These functions can be observed in the presence of others or when the client is alone. In this latter case, we suggest that the behavior can be maintained by the positive or negative reinforcing contingencies of sensory stimuli impinging on the client. In all cases, the stated goal of intervention should be to improve the behavior of the person with autism by teaching appropriate replacement skills while simultaneously reducing or eliminating the behavior that is problematic or that interferes with more adaptive functioning. Specific procedures to accomplish this are discussed below.

Treatment Participants

Treatment procedures for any given client are selected based upon the results of the functional assessment/analysis but may be modified to address the specific target behaviors selected, the learning history (history of reinforcement) of the client with the particular behavior, and the availability of resources and competencies of intervenors. Consideration is also given to such factors as severity, duration, pervasiveness, and frequency of the target behavior when determining priorities for intervention.

Treatment Procedures

Treatment procedures for nonaversive interventions can be broadly divided into two groups: antecedent interventions that occur prior to the behavior and consequent procedures that are implemented after the behavior has been emitted. Both seek to reduce the likelihood of behavioral expression in the future by emphasizing the use of positive reinforcement procedures as a key or collateral component of the treatment package. Most importantly, all treatment should be preceded by a thorough functional behavior assessment or analysis in order to determine which stimuli in the environment exert control over the target behavior.

Antecedent procedures include errorless learning, whereby the student is prompted to the correct response immediately after the presentation of the request; interspersing mastered or easy tasks with difficult tasks in teaching (Weber & Thorpe, 1992); the use of choice in the selection of tasks and reinforcers (Dyer, Dunlap, & Winterling, 1990); reducing the information-processing demands of the task or providing an alternative mode of task presentation; use of a high-probability request sequence (Zuluaga & Normand, 2008); functional communication training (Carr & Durand, 1985); stimulus change procedures, whereby a novel stimulus that is not an antecedent or a consequence to the behavior is interjected into a behavioral sequence, interrupting the response-reinforcer relationship (Carr, Robinson, & Palumbo, 1990); and environmental modifications such as use of visual schedules, curriculum adjustment, etc. (Flannery & Horner, 1994; Kern & Dunlap, 1998).

Consequent procedures with demonstrated efficacy include positive reinforcement, differential reinforcement, and its variants (differential reinforcement of other, incompatible, high rates, or alternative behavior); response interruption and redirection (Underwood, Figueroa, Thyer, & Nzeocha, 1989); extinction (Lerman & Iwata, 1996); and noncontingent reinforcement, whereby reinforcing stimuli are provided to a client independent of the client’s behavior (Carr, Severtson, & Lepper, 2009).

Aversive stimuli are noxious events that serve as punishers when following a behavior, evoke a behavior that has terminated the noxious stimulus in past circumstances, or function as a reinforcer when removed after the occurrence of a behavior (Cooper et al., 2007). While the function of an aversive stimulus is always to cause the cessation of a behavior, its forms are virtually limitless (Repp & Singh, 1990) and have included smelling aromatic ammonia, contingent water mist to the face, the application of “white noise,” and electric shock. It is noteworthy that while the NIH consensus conference clearly emphasized the importance of using treatment procedures based on positive behavioral supports, it also provided clear guidelines for the use of punishment procedures when they might be deemed clinically necessary.

Efficacy Information

The efficacy of antecedent strategies to treat behavior problems has been well documented in the research literature, and several in particular have been identified as evidence-based procedures (Cooper et al., 2007; Powers, Palmieri, D’Eramo, & Powers, 2011). It is important to remember, however, that the use of an antecedent (or any other) strategy does not guarantee success. Rather, the use of the procedure must be based on the results of the functional behavior assessment/functional analysis, must be implemented with fidelity, and must be evaluated accurately and objectively. Violation of any of these tenets can (and likely will) reduce the efficacy and efficiency of the correctly chosen treatment strategy.

Outcome Measurement

Objective and reliable measurement of treatment effects and outcomes is essential to the correct use of any procedure designed to increase desirable behavior or to reduce problem behavior. Fortunately, the use of single-subject experimental designs (SSEDs) have predominated in the literature (Kazdin, 1982), establishing a robust arsenal of potential designs for outcome measurement. When well used, SSEDs provide excellent internal and external validity, support the development of reliable observations, and ultimately contribute to the serial replication of findings. To this latter point, the aggregation of large numbers of individual studies, each with a small subject pool, can generate strong findings of efficacy (Reichow, Doehring, Cicchetti, & Volkmar, 2011).

Qualifications of Treatment Providers

While certainly effective when used correctly, the technology of intervention requires training in the principles and strategies of applied behavior analysis. Obviously, with behavior problems of greater significance (e.g., where personal safety of the client or others is at risk and where health status can/may be compromised), the demand for greater levels of sophistication and competency is critical. At a minimum, supervision of assessment and treatment protocols by an individual with Board Certification as a Behavior Analyst (BCBA) or by a clinician with equivalent training and experience would be appropriate. In cases where more extraordinary interventions are necessary, or where the risk of harm is greater, it is strongly advisable to have all clinical aspects peer reviewed and vetted by a human rights committee.

See Also

Board Certified Associate Behavior Analyst

Differential Reinforcement

High-Probability Requests