Experiential Avoidance and Behavioral Disorders: A Functional Dimensional Approach to Diagnosis and Treatment

Syndromal classification is a well-developed diagnostic system but has failed to deliver on its promise of the identification of functional pathological processes. Functional analysis is tightly connected to treatment but has failed to develop testable. replicable classification systems. Functional diagnostic dimensions are suggested as a way to develop the functional classification approach, and experiential avoidance is described as 1 such dimension. A wide range of research is reviewed showing that many forms of psychopathology can be conceptualized as unhealthy efforts to escape and avoid emotions, thoughts, memories, and other private experiences. It is argued that experiential avoidance, as a functional diagnostic dimension, has the potential to integrate the efforts and findings of researchers from a wide variety of theoretical paradigms, research interests, and clinical domains and to lead to testable new approaches to the analysis and treatment of behavioral disorders. Steven C. Haves, Kelly G. Wilson, Elizabeth V. Gifford, and Victoria M. Follette. Department of Psychology. University of Nevada: Kirk Strosahl, Mental Health Center, Group Health Cooperative, Seattle, Washington. Preparation of this article was supported in part by Grant DA08634 from the National Institute on Drug Abuse. Correspondence concerning this article should be addressed to Steven C. Hayes, Department of Psychology, Mailstop 296, College of Arts and Science. University of Nevada, Reno, Nevada 89557-0062. The process of classification lies at the root of all scientific behavior. It is literally impossible to speak about a truly unique event, alone and cut off from all others, because words themselves are means of categorization (Brunei, Goodnow, & Austin, 1956). Science is concerned with refined and systematic verbal formulations of events and relations among events. Because "events" are always classes of events, and "relations" are always classes of relations, classification is one of the central tasks of science. The field of psychopathology has seen myriad classification systems (Hersen & Bellack, 1988; Sprock & Blashfield, 1991). The differences among some of these approaches are both long-standing and relatively unchanging, in part because systems are never free from a priori assumptions and guiding principles that provide a framework for organizing information (Adams & Cassidy, 1993). In the present article, we briefly examine the differences between two core classification strategies in psychopathology syndromal and functional. We then articulate one possible functional diagnostic dimension: experiential avoidance. Several common syndromal categories are examined to see how this dimension can organize data found among topographical groupings. Finally, the utility and implications of this functional dimensional category are examined. Comparing Syndromal and Functional Classification Although there are many purposes to diagnostic classification, most researchers seem to agree that the ultimate goal is the development of classes, dimensions, or relational categories that can be empirically wedded to treatment strategies (Adams & Cassidy, 1993: Hayes, Nelson & Jarrett, 1987: Meehl, 1959). Syndromal classification – whether dimensional or categorical – can be traced back to Wundt and Galen and, thus, is as old as scientific psychology itself (Eysenck, 1986). Syndromal classification starts with constellations of signs and symptoms to identify the disease entities that are presumed to give rise to these constellations. Syndromal classification thus starts with structure and, it is hoped, ends with utility. The attempt in functional classification, conversely, is to start with utility by identifying functional processes with clear treatment implications. It then works backward and returns to the issue of identifiable signs and symptoms that reflect these processes. These differences are fundamental. Syndromal Classification The economic and political dominance of the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders (e.g., 4th ed.; DSM -IV; American Psychiatric Association, 1994) has lead to a worldwide adoption of syndromal classification as an analytic strategy in psychopathology. The only widely used alternative, the International Classification of Diseases (ICD) system, was a source document for the original DSM, and continuous efforts have been made to ensure their ongoing compatibility (American Psychiatric Association 1994). The immediate goal of syndromal classification (Foulds. 1971) is to identify collections of signs (what one sees) and symptoms (what the client's complaint is). The hope is that these syndromes will lead to the identification of disorders with a known etiology, course, and response to treatment. When this has been achieved, we are no longer speaking of syndromes but of diseases. Because the construct of disease involves etiology and response to treatment, these classifications are ultimately a kind of functional unit. Thus, the syndromal classification approach is a topographically oriented classification strategy for the identification of functional units of abnormal behavior. When the same topographical outcome can be established by diverse processes, or when very different topographical outcomes can come from the same process, the syndromal model has a difficult time actually producing its intended functional units (cf. Bandura, 1982; Meehl, 1978). Some medical problems (e.g., cancer) have these features, and in these areas medical researchers no longer look to syndromal classification as a quick route to an understanding of the disease processes involved. The link between syndromes (topography of signs and symptoms) and diseases (function) has been notably weak in psychopathology. After over 100 years of effort, almost no psychological diseases have been clearly identified. With the exception of general paresis and a few clearly neurological disorders, psychiatric syndromes have remained syndromes indefinitely. In the absence of progress toward true functional entities, syndromal classification of psychopathology has several down sides. Symptoms are virtually non-falsifiable, because they depend only on certain formal features. Syndromal categories tend to evolve changing their names frequently and splitting into ever finer subcategories but except for political reasons (e.g., homosexuality as a disorder) they rarely simply disappear. As a result, the number of syndromes within the DSM system has increased exponentially (Follette, Houts, & Hayes, 1992). Increasingly refined topographical distinctions can always be made without the restraining and synthesizing effect of the identification of common etiological processes. In physical medicine, syndromes regularly disappear into disease categories. A wide variety of symptoms can be caused by a single disease, or a common symptom can be explained by very different diseases entities. For example, "headaches" are not a disease, because they could be due to influenza, vision problems, ruptured blood vessels, or a host of other factors. These etiological factors have very different treatment implications. Note that the reliability of symptom detection is not what is at issue. Reliably diagnosing headaches does not translate into reliably diagnosing the underlying functional entity, which after all is the crucial factor for treatment decisions. In the same way, the increasing reliability of DSM diagnoses is of little consolation in and of itself. The DSM system specifically eschews the primary importance of functional processes: "The approach taken in DSM-III is atheoretical with regard to etiology or patho-physiological process" (American Psychiatric Association, 1980, p. 7). This spirit of etiological agnosticism is carried forward in the most recent DSM incarnation. It is meant to encourage users from widely varying schools of psychology to use the same classification system. Although integration is a laudable goal, the price paid may have been too high (Follette & Hayes, 1992). For example, the link between syndromal categories and biological markers or change processes has been consistently disappointing. To date, compellingly sensitive and specific physiological markers have not been identified for any psychiatric syndrome (Hoes, 1986). Similarly, the link between syndromes and differential treatment has long been known to be weak (see Hayes et al., 1987). We still do not have compelling evidence that syndromal classification contributes substantially to treatment outcome (Hayes et al., 1987). Even in those few instances and not others, mechanisms of change are often unclear of unexamined (Follette, 1995), in part because syndromal categories give researchers few leads about where even to look. Without attention to etiology, treatment utility, and pathological process, the current syndromal system seems unlikely to evolve rapidly into a functional, theoretically relevant system. Functional Classification In a functional approach to classification, the topographical characteristics of any particular individual's behavior is not the basis for classification; instead, behaviors and sets of behaviors are organized by the functional processes that are thought to have produced and maintained them. This functional method is inherently less direct and naive than a syndromal approach, as it requires the application of pre-existing information about psychological processes to specific response forms. It thus integrates at least rudimentary forms of theory into the classification strategy, in sharp contrast with the atheoretical goals of the DSM system. Functional Diagnostic Dimensions as a Method of Functional Classification Classical functional analysis is the most dominant example of a functional classification system. It consists of six steps (Hayes & Follette, 1992) -Step 1: identify potentially relevant characteristics of the individual client, his or her behavior, and the context in which it occurs through broad assessment; Step 2: organize the information collected in Step 1 into a preliminary analysis of the client's difficulties in terms of behavioral principles (e.g., reinforcement, stimulus control) so as to identify important causal relationships that might be changed; Step 3: gather additional information based on Step 2 and finalize the conceptual analysis; Step 4: devise an intervention based on Step 3; Step 5: implement treatment and assess change; Step 6: if the outcome is unacceptable, recycle back to Step 2 or 3. There are many problems with classical functional analysis as a functional classification system (Hayes & Follette, 1992). It is sometimes vague, often difficult to replicate and thus to test empirically, and is ideographic in the extreme. It may not adequately specify the relevant behavioral or contextual characteristics to identify or which behavioral principles should be applied and in what manner. Because it is vague, it may fall prey to the well-documented errors common to clinical judgment more generally (Dawes, 1994). Perhaps for all of these reasons, functional analysis has hardly progressed from its promising beginning (Salzinger, 1988). There are several possible solutions to these problems (Hayes & Follette, 1992). One of the most promising is the development of functional diagnostic dimensions. If one performs many individual functional analyses tied to the same dimension, these might then be arranged into a larger category with related assessment methods and treatment recommendations. The guiding principle behind these collections would be the identification of common processes of etiology or maintenance that suggest effective courses of action. This idea was prevalent early in the behavior therapy movement (e.g., Bandura, 1968; Kanfer & Grimm, 1977), but the suggested categories were too closely tied to a specific set of abstract principles and too little tied to the characteristic forms of behavior that clients display. For example, Kanfer and Grimm's major categories included "behavioral excesses" and "problems in environmental stimulus control." Such categories can, in principle, be applied to every case. Furthermore, the analytic categories themselves were so specific to behavior analysis that they were not useful unless one subscribed to this specific theoretical approach. Functional diagnostic dimensions may be most useful initially when they point to reasonably widely agreed-on pathological processes, without necessitating the adoption of an entire paradigmatic system. Classification based on such dimensions may integrate the contributions of various theoretical perspectives and allow for the validation of pathological processes tied to clear treatment implications. In the remainder of the article, we develop an example of a functional diagnostic dimension that, we argue, demonstrates these possibilities. Experiential Avoidance as a Functional Diagnostic Dimension Experiential avoidance is a putative pathological process recognized by a wide number of theoretical orientations. Experiential avoidance is the phenomenon that occurs when a person is unwilling to remain in contact with particular private experiences (e.g., bodily sensations, emotions, thoughts, memories, behavioral predispositions) and takes steps to alter the form or frequency of these events and the contexts that occasion them. We occasionally use terms such as emotional avoidance or cognitive avoidance rather than the more generic experiential avoidance when it is clear that these are the relevant aspects of experience that the person seeks to escape, avoid, or modify. We recognize that thoughts, memories, and emotions are richly intermingled and do not mean to imply any necessary rigid distinctions among them (although distinctions might be drawn by some theoretical perspectives without threat to the underlying principle of experiential avoidance), Furthermore, the term avoidance in this context explicitly includes both avoidance and escape in all of their forms, as long as they are used as methods of altering the form or frequency of experiences and the contexts that occasion them. In this article, we argue that many forms of psychopathology are usefully viewed as unhealthy methods of experiential avoidance. In what follows, we outline some of the perspectives and research that supports the concept of such a dimension and that shows its potential for integrating research across theoretical orientations and specific research areas. Convergence of Views on Experiential Avoidance Experiential avoidance has been recognized, implicitly or explicitly, among most systems of therapy. Foa, Steketee, and Young (1984) have noted that "the general phenomenon of emotional avoidance is a common occurrence; unpleasant events are ignored, distorted, or forgotten" (p. 34). Freud recognized the importance of the avoidance of private experiences and defined the very purpose of psychoanalysis as the lifting of repressions and making conscious material that has been too painful or threatening to be held in conscious awareness (Freud, 1920/ 1966). In client-centered therapy, "openness to experience" is a central therapeutic goal (Raskin & Rogers, 1989; Rogers, 1961). Rogers argued that, as the result of his therapy, the individual becomes more openly aware of his own feelings and attitudes as they exist in him at an organic level. . . . He is able to take in the evidence in a new situation, as it is, rather than distorting it to fit a pattern which he already holds. (1961, p. 115 ) Similarly, according to Gestalt therapists, the heart of many psychological problems is the avoidance of painful feelings or fear of unwanted emotion (Perls, Hefferline, & Goodman, 1951). Others from the Gestalt tradition suggest that "dysfunction occurs when emotions are interrupted before they can enter awareness or go very far in organizing action" (Greenberg & Safran, 1989, p. 20). Existential psychologists agree with the centrality of experiential avoidance, although they focus particularly on the avoidance of a fear of death to cope with these fears, we erect defenses against death awareness, defenses that are based on denial, that shape character structure, and that, if maladaptive, result in clinical syndromes. In other words, psychopathology is the result of ineffective modes of death transcendence. (Yalom, 1980, p. 47; cf. Becker, 1973) Somewhat unlike the aforementioned paradigms, behavioral and cognitive therapy have generally focused on changing (rather than accepting) private experiences. Traditional behavior therapy fought anxiety with relaxation, whereas cognitive therapy challenged irrational beliefs with more rational ones. Essentially, better forms of experiential avoidance were systematically trained as modes of intervention, Even within these domains, however, emotional and other forms of experiential avoidance has been recognized as a problem, and such recognition appears to be increasing. Indeed, recognizing and dealing with experiential avoidance has been a central theme of modern behavioral therapies such as dialectical behavior therapy (Linehan, 1993, 1994), acceptance and commitment therapy (Hayes, 1987; Hayes, Strosahl, & Wilson, in press; Hayes & Wilson, 1994; Strosahl, 1991), and revised forms of behavioral couples therapy (Koerner, Jacobson, & Christensen, 1994). Neimeyer suggests that modern cognitive therapy is also shifting to a position that is less interested in controlling negative feelings and more interested in interpreting negative affect as an essential aspect of experience (1993; see also Meichenbaum, 1993). Rational-emotive therapy has long embraced unconditional self-acceptance as a goal (Ellis & Robb, 1994). In the traditional cognitive therapies, however, experiential avoidance is not analyzed thoroughly and, to the extent that there is an associated theory of psychopathology, experiential avoidance does not play a central role. Why Would Experiential Avoidance Be So Pervasive? It appears that a concern with experiential avoidance is ubiquitous across otherwise divergent psychotherapeutic systems. We believe that the theme of experiential avoidance is so pervasive in various theories of psychopathology because it is such an aspect of human functioning. Experiential avoidance could be analyzed in a number of ways depending on one's theoretical orientation, but from our own contextual behavioral perspective (Biglan & Hayes, 1996; Hayes, Follette, & Follette,1995, Hayes & Wilson, 1993) there are several factors that contribute, including the bidirectional nature of human language, inappropriate generalization of control rules, cultural support for emotions and cognitions as causes of behavior, social encouragement, and modeling of experiential avoidance, among others. Self-Knowledge and the Bidirectional Nature of Language Let us begin an analysis of experiential avoidance with an even more general matter at the opposite end of the psychological continuum: Why is self-knowledge important? Following in the Socratic tradition, almost all schools of psychology have emphasized the importance of self-knowledge. For example, B. F Skinner suggested that "self-knowledge has a special value to the individual himself. A person who has been “made aware of himself’ is in a better position to predict and control his own behavior" (Skinner, 1974, p. 31 ). We agree, but we argue that the beneficial effect of self-knowledge depends entirely on the nature of human language (Hayes & Wilson, 1993). Human verbal behavior is arbitrarily applicable, bidirectional, and combinatorial (see Hayes & Hayes, 1992, for a review of the literature supporting these views). In more commonsense terms, when a human interacts symbolically with an event, the functions of the referent are partially present in the symbol and vice versa, and can, in some circumstances, spread from that symbol to others through a network of related terms. For example, in some contexts some of the characteristics of Tabasco sauce (e.g., heat, liquid, red, etc.) are partially psychologically present in the word "Tabasco" and vice versa (Hayes & Hayes, 1992; Hayes & Wilson, 1993). Because of the bidirectional quality of human symbolic behavior, when a human interacts symbolically with his or her own behavior, the psychological meaning of both the verbal symbol and the behavior itself can change as a result. This bidirectional property makes human self-knowledge useful, because it means that changes in how we view our behavior or the situations that occasion it can, in turn, change the functions of these situations and behaviors. However, this same bidirectional property also makes self-knowledge potentially aversive. Suppose a rat is trained to press one lever for food pellets if it has recently been shocked and another if it has not recently been shocked. In effect, the rat can "tell" whether it recently received an electric shock. This is readily done because, although the shock was aversive to the animal, the report of it is not: After all the report leads to food, not to shock. Simply stated, it is not aversive for a nonverbal organism to report an aversive event. Contrast this with the case for verbally competent humans. In this case, the report and the event reported are mutually related and therefore the functions of events are partially available in the symbolic description and vice versa. Because of this, for example, a survivor of trauma may reexperience pain simply in the verbal reporting of that trauma. In our own technical analysis, we refer to this phenomenon as the "bidirectional transformation of stimulus functions" and have contrasted it with the unidirectional nature of normal operant and classical conditioning (Hayes & Hayes, 1992; Hayes & Wilson, 1993). Verbal awareness of one's own experience can itself change the experiential quality of what is known. Human emotions provide an example. Human emotions seem to require a verbal label or appraisal as a cognitive component (Lazarus, 1982). These labels may carry with them implications for emotional avoidance. "Anxiety" is not just a fuzzy set of bodily states and behavioral predispositions (as in nonverbal organisms); it is an evaluative and descriptive verbal category that integrates a wide variety of experiences, including memories, thoughts, evaluations, and social comparisons, among others. The evaluative connotation of emotional labels alters the functions of private experiences that are so labeled. For example, in most contexts, anxiety is a "bad" emotion. The bidirectionality of human language can create the illusion that this "badness" is an inherent quality of the emotion itself We say "this is a bad emotion," not "this is an emotion and I am evaluating it as bad." These same referential or relational processes can increase the negativity of experienced events by relating the verbal label to still other verbal events in more complex ways. For example, if anxiety is said to be predictive of "loss of control" and "loss of control" is predictive of "social humiliation," then the aversiveness of felt (and labeled) anxiety could increase greatly because it is verbally related to humiliation. Because humans have been taught a wide variety of strategies for avoiding negative events, it is not surprising that these would be applied to "negative" emotions. Given our verbal abilities, experiential avoidance can function generally as a learned, negatively reinforced behavior (Gifford, 1994). We are arguing that experiential avoidance is built into human language and is, thus to some degree, a basic component of the human condition. The very bidirectional relations that make human verbal self-knowledge useful also can make that self-knowledge inherently difficult and actively resisted when its content is undesirable. In this view, psychopathological or maladaptive experiential avoidance is the dysfunctional range of a fairly "normal" behavior, and it is excessive entanglements with or fusion with thoughts that particularly supports the adoption of experiential avoidance strategies. Other Reasons Experiential Avoidance Is Ubiquitous There are several other reasons why experiential avoidance occurs so readily. First, the exercise of conscious, planful control strategies can be extremely useful in avoiding many of the hazards that life presents. If one drives cautiously in the rain, accidents can be avoided. A program of regular brushing and flossing can help one to avoid the dentist's drill. Because the exercise of such deliberate, verbally guided control strategies is effective in many contexts, it is only natural that we would attempt to use these same strategies with all evaluated targets, including aversive thoughts, emotions, and the like. It is not at first obvious that this is an inappropriate and unnecessary generalization of control rules to a domain in which they are notably ineffective. Second, there is a great deal of social encouragement and modeling of emotional avoidance. Children are often explicitly trained to suppress emotional responding (e.g., "stop crying or I will give you something to cry about"). From the perspective of the child, adults seemingly model this same behavior because they do not show emotions as readily. Regulating emotional displays is not the same thing as regulating private experience, however. The child who deliberately stops crying does not necessarily become happy-only silent. Third, emotions and cognitions are treated as socially valid reasons for behavior. As part of the socialization process, people are required and able to give verbal explanations for their behavior, even if its sources are unknown or obscure (Semin & Manstead, 1985). Thoughts and feelings are commonly used as reasons for behavior, and by extension, emotions and cognitions that "cause" bad behavior should be avoided. Unfortunately, this same process excuses destructive forms of experiential avoidance itself. For example, a client with agoraphobia might use anxiety as a socially valid reason for social withdrawal ("I would have been too afraid if I left the house, so I had to stay home"). Finally, the immediate effects of experiential avoidance are often positive, and short-term consequences are much more important than long-term ones. For example, the immediate effect of cognitive distraction or other forms of thought suppression is positive it is only over time that the increase in avoided thoughts appears (Gold & Wegner, 1995). Thus, experiential avoidance appears to work, even when it does not. Why Would Experiential Avoidance Be Detrimental? These several factors may explain why experiential avoidance is ubiquitous, but they do not explain why it may result in psychopathology. There are several possible reasons why emotional avoidance may be detrimental or why its converse psychological acceptance may be helpful. These analyses must be tentative, in part because data on these issues are relatively new. Until recently, the importance of psychological acceptance was emphasized in the less empirical traditions such as the humanistic or existential approaches (Greenberg, 1994). Only within the past 10 years have a number of empirically based treatment approaches emerged that are oriented toward different types of psychological acceptance (e.g., Barlow & Beck, 1984; Chiles & Strosahl, 1995; Cordova & Kohlenberg, 1994; Hayes, 1984,1987; Hayes, Jacobson, Follette, & Dougher, 1994; Hayes & Wilson, 1994; Jacobson, 1991; Koener et al., 1994; Kohlenberg & Tsai, 1991; Linehan, 1993; Marlatt, 1994; Strosahl, 1991). Again, the analysis of detrimental effects differs according to theoretical orientation, although the existence of such effects seems widely accepted. Our analysis of the destructive effects of some forms of experiential avoidance centers around certain well-established principles of learning, as well as on some emerging analyses of social-verbal contextual factors. The Process of Deliberate Avoidance Necessarily Contradicts the Desired Outcome In humans, deliberate avoidance usually involves formulating and following a verbal plan. In some areas of experiential avoidance, this is inherently problematic because the verbal regulatory process includes the avoided item. For example, cognitive avoidance is difficult because deliberately trying to rid one. self of a thought involves following the verbal rule that contains the thought. There is a significant and rapidly growing body of evidence that deliberate thought suppression and control may actually be counterproductive, in that an attempt not to think thoughts often creates those very thoughts (e.g., Wegner, Schneider, Carter, & White, 1987; Wegner, Schneider, Knutson, & McMahon, 1991). The Regulation of Private Events Is Largely Unresponsive to Verbal Control Many private experiences are classically conditioned, either directly or indirectly. For example, if painful emotional experiences have been associated with a particular event that is either directly present or present indirectly through verbal relations, it is likely that this event will now arouse negative emotions by association alone. In these circumstances, attempts at purposeful control may be relatively ineffective, because the underlying process or history is not readily verbally governed. The experimental evidence for the harmful effects of emotional suppression is not yet as great as that for thought suppression, but there seem to be many examples of these vicious cycles. For instance, suppose a person is extremely distressed about anxiety and tries to do everything to eliminate it because of its awful meaning and potentially terrible consequences. In this case, a small bit of classically conditioned anxiety may cue both purposeful attempts to avoid and reduce the anxiety, and additional anxiety associated with the verbal construction of highly aversive consequences. In essence, the person may become anxious about being anxious. This kind of harmful effect of experiential avoidance is most likely precisely when the seeming need for it is greatest, because it is then that the negative effects of failure would be most closely related to the avoidance strategy and even minor levels of avoided material would be threatening. Panic disorder may be an example of such a phenomenon (Chambless, Caputo, Bright, & Gafagher, 1984; Craske, Sanderson, & Barlow, 1987; Goldstein & Chambless, 1978; Hayes, 1987). Change Is Possible, But the Change Strategy Leads to Unhealthy Forms of Avoidance Suppose someone tried not to remember a given event. Memories are not simple voluntary behavior once an event has, occurred remembering it may occur relatively automatically in a wide variety of circumstances. Strategies to avoid memories might include avoiding all situations that might give rise to it, dissociating or other means. The problem with these strategies is that even if they are successful they create additional problems, such as constricting the person's freedom to be in otherwise valuable situations, or limiting conscious access to life events. This is one of the processes that has been most emphasized in the humanistic tradition: Emotional avoidance restricts the wisdom of humans by diminishing access to one’s own history and the “response tendency information” contained therein (Safran & Greenberg, 1988: cf. Greenberg, 1994). The Emotion-Laden Event Is Important Sometimes experiential control is put in the service of managing entirely appropriate reactions to unchangeable events, merely because the reaction is not pleasant. For example, a person may take the view that "I can't accept that my Dad was killed" and use drugs to avoid the grief associated with his death. Grief is a natural reaction to such losses, and no amount of drug consumption will alter either the situation or the loss. No effort to reduce or alter private events is called for here. When an unchangeable loss occurs, the healthy thing to do is to feel what one feels when losses occur. Healthy Change Produces Painful Experiences Change can be frightening. At times, what needs to be done is avoided because it is experientially difficult. This suggests a major reason experiential avoidance may lead to psychopathology: It restricts needed change. Instead of putting deliberate change efforts in the service of necessary and useful life changes, experiential avoidance emphasizes a secondary agenda that can too easily be accomplished in destructive ways. One cost is that a stenotopic condition a narrow range of adaptability to changes in environmental conditions ensues. We are not arguing that all forms of experiential avoidance are unhealthy. In many instances, there is nothing harmful about seeking relaxation or putting off distracting thoughts or avoiding physical pain. Ironically, however, these healthy forms of experiential avoidance apply most clearly when the experiences involved are not intense and clinically relevant. For example relaxation is much more likely to reduce normal stress than to eliminate panic attacks. In the present article, we limit our analysis to experiential avoidance that persists even when it is costly, useless, or life distorting. In these circumstances, experiential avoidance becomes pathological. Evidence for Experiential Avoidance as a Pathological Dimension A functional approach is explicitly inductive: The phenomenon and its context are the focus of interest. Unlike more deductively oriented approaches, where empirical findings are used to argue for and against complex theoretical positions, functional dimensions are organized more humbly around direct observations. Because of this, data from various schools of thought and research areas become relevant to each other. In the following sections, we review literatures as diverse as basic experimental analyses of human cognition and emotion, literature on coping styles, experimental psychopathology, clinical process, and psychotherapy outcome research. Although the purposes and orientation of these various researchers are quite divergent, the dimension of experiential avoidance can play an important role in organizing their data in a coherent way. We believe that this tendency to integrate data from both basic and applied research domains is characteristic of functional diagnostic dimensions (Gifford, 1995). Evidence From Basic Experimental Work One major source of data relevant to experiential avoidance is from the emerging thought suppression literature. A number of studies have demonstrated that when participants are asked to suppress a thought, they later show an increase in this suppressed thought, compared with those who are not given suppression instructions (Clark, Ball, & Pape, 1991; Gold & Wegner, 1995; Wegner et al., 1987; Wegner et al., 1991), A number of variations on this research have been done that have implications for an experiential avoidance perspective. For example, participants who have been asked to suppress a particular thought show rebound effects specifically in contexts in which the suppression took place (Wegner et al., 1991). These data suggest that if an individual successfully suppresses some aversive thought or memory in a particular situation, they might find the thought remaining at fairly low levels in contexts not related to suppression, but when they return to the original context they would not only be likely to re-experience the suppressed thought but would be likely to experience it at an even higher level than before. This suggests that cognitive avoidance is most likely to fail in the precise contexts in which its success is most important and desired. The paradoxical effects of thought suppression have also been linked to mood. In a study using mood induction, participants were asked not to think a specific thought while in a specific mood (Wenzlaff, Wegner, & Klein, 1991). In subsequent phases of the experiment, participants were assigned to either similar or different mood induction procedures. All participants showed a rebound effect but those who had the same mood induction in the second phase had a significantly higher rebound effect. This suggests that, if a person is avoiding a thought that is associated with anxiety, for example, the paradoxical increase produced by suppression is most likely to be seen when the person is anxious: precisely when a person struggling with an anxious thought would least like suppression to fail. Furthermore, a separate experiment showed that the suppressed thought will come to produce the mood that was present during the original suppression (Wenzlaff et al., 1991). Thus, a kind of self-amplifying loop can be created from precisely the kinds of mood, cognition, and suppression components likely to be experienced by those with clinical disorders. The paradoxical effects of suppression have also been demonstrated in a study examining somatic sensation (Cioffi & Holloway, 1993). Participants in a cold-pressor pain induction procedure were given one of three sets of instructions. They were told to think about their room at home (distraction), focus on the sensations in their hand, or eliminate thoughts about pain entirely. Recovery on discomfort ratings was slowest for the suppression instructions and most rapid for the focusing instructions. Later in the experimental hour, they were asked to rate the unpleasantness of an innocuous vibration. Participants from the suppression condition rated it as more unpleasant than did those with other instruction sets. This suggests that part of the unpleasantness of avoided emotions comes from the very process of avoidance, not from the emotions themselves. Evidence From the Coping Styles Literature Research into coping strategies has also demonstrated the detrimental effects of experiential avoidance. One of the most widely used methods of assessing coping strategies is the Ways of Coping Questionnaire (WOC; Folkman & Lazarus, 1988). Lazarus and Folkman (1984) have identified two dominant means of coping with stressful situations: problem-focused and emotion-focused coping strategies. Problem-focused strategies involve active attempts to alter distressing situations, such as "I made a plan of action and followed it." Many of the emotion-focused coping strategies, by contrast, clearly involve experiential avoidance as we have defined it here, such as refusing to think about disturbing events, supplanting bad thoughts with good ones, looking at the bright side of things, or telling things to oneself to help feel better. Similarly, factor analysis of another instrument, the Coping Inventory for Stressful Situations (CISS), identified task-oriented, emotion-oriented and avoidance-oriented coping strategies (Endler & Parker, 1990). Items assessing avoidance involve either distraction ("watch TV") or social diversion ("phone a friend"). Endler and Parker suggested that, although these activities are not focused on emotion per se, they are engaged in as a way to manage emotional responses to stressful situations. Consistent with an experiential avoidance perspective, both emotion-focused and avoidant strategies, measured using the WOC, the CISS, and similar instruments, have been found to negatively predict outcome for a variety of difficulties, including substance abuse (Ireland, McMahon, Malow, & Kouzekanani, 1994), depression (DeGenova, Patton, Jurich, & MacDermid, 1994; Bruder-Mattson & Hovanitz, 1990), and sequelae of child sexual abuse (Leitenberg, Greenwald, & Cado, 1992). A different but suggestive effect has been found in the study of chronic pain patients. Jensen and Karoly (1991) found that ignoring pain, diverting attention, and the use of coping self-statements was predictive of better psychological functioning, but this effect was found only for patients with low, not high, levels of pain. Thought suppression as a general style of coping seems to be detrimental. For example, research using the White Bear Suppression Inventory (WBSI; Wegner & Zanakos, 1994) has shown that thought suppressors tend toward depression and obsessional symptomatology. The WBSI includes such items as "I often do things to distract myself from my thoughts," and "I always try to put problems out of mind." Struggling with negatively evaluated emotions is also detrimental. For example, research using the Depression Sensitivity Scale (DSS; Zanakos & Wegner, 1993) has shown that emotional avoiders tend toward depressive symptomatology, particularly when combined with thought suppression as a coping strategy (Wegner & Zanakos, 1994). The DSS includes such items as, "when I start to feel sad I try to get rid of those feelings as quickly as I can." Evidence From Psychotherapy Process Research In a review of 1100 quantitative studies of the relationships between process and outcome variables, Orlinsky and Howard (1986), found that "self-relatedness" was "the most consistently positive correlate of therapeutic outcome" (p. 366), Clients high in self-relatedness were defined as being "in touch with themselves . . [and] open to their feelings" as contrasted with being "out of touch with themselves" (p. 359). The Experiencing Scale (Klein, Mathieu, Gendlin, & Kiesler, 1969; Klein, Mathieu-Coughlan, & Kiesler, 1987) is the mast widely used process measure in psychotherapy research (Hill, 1990). The Experiencing Scale measures "the extent to which . . . ongoing, bodily, felt flaw of experiencing is the basic datum of. . . awareness" (Klein et al., 1969, p. 1 ). In our terms, the Experiencing Scale measures, at least in part, the absence of emotional avoidance. High client level of experiencing has been consistently related to a good outcome in psychotherapy (Greenberg, 1983; Greenberg & Dompierre, 1981; Greenberg & Webster, 1982; Kiesler, 1971; Luborsky, Chandler, Auerbach, Cohen, & Bachrach, 1971; see Greenberg & Safran, 1989, for a review). Emotional willingness is a marker of good outcomes in the treatment of chronic pain (Geiser, 1992) and other types of pain (Khorakiwala, 1991; McCurry, 1991). Evidence From the Literature on Clinical Syndromes One approach to determining the utility of experiential avoidance as a functional dimension of psychopathology is to see if it provides a useful addendum to current syndromal categories. Many, if not most, psychopathological classifications include the presence of problems with private experiences, although the relevant features vary across specific syndromal categories. For example, the form of the avoidance may vary. A person who drinks to suppress anxiety may be an alcoholic; a person who hides for the same reason may be an agoraphobic. The avoided experience may also vary: For example, persons with a panic disorder may struggle with anxiety, whereas those with obsessivecompulsive disorder (OCD) may struggle with a thought. The source of the avoided material may vary. A person with posttraumatic stress disorder (PTSD) suffered a specifiable original trauma; a person struggling with depression may not. We are not claiming that all of these various syndromes are experiential avoidance disorders. Rather, we are suggesting that many topographically defined syndromes may include significant subgroups in which experiential avoidance has been a functionally important factor in the etiology and maintenance of these pathological patterns. We examine this idea in more detail with four diagnostic categories drawn from the current syndromal system: substance abuse, OCD, panic disorder with agoraphobia, and borderline personality disorder (BPD). This list is not exhaustive, indeed several other important disorders have been interpreted from an experiential avoidance point of view (e.g., depression; see Zettle, 1984). In each case, we examine evidence of (a) unacceptable and avoided private experiences as a component of this disorder, (b) ineffective avoidance strategies as a component, and (c) treatment implications of an experiential avoidance interpretation of some persons suffering from the disorder. Substance Abuse and Dependence Definitionally, substance abuse and dependence is not a complicated syndrome: it is defined by very straightforward behavioral criteria. Why people use drugs is not at issue. The literature suggests, however, that a subgroup of drug abusers are experiential avoiders (cf. Cooper, Russell, Skinner, Frone, & Mudar, 1992; Wanberg, Horn, & Foster, 1977). (The language of experiential avoidance applies equally well to stimulants as to depressants if one assumes that the felt absence of such stimulation is boring or otherwise aversive.) Evidence for drug use an ineffective avoidance strategy. There is no doubt that drug use per se is a highly effective short-term strategy for experiential change. Virtually all drugs of abuse have known psychoactive effects. Furthermore, drug abusers believe in these effects more than do others. In the area of alcoholism, for example, alcoholics exceed problem drinkers (with problem drinkers exceeding non-problem drinkers) in their expectancy that alcohol will enhance pleasure and reduce stress (Conners, O'Farrell, Cutter, & Thompson, 1986). Cooper et al. (1992) found that psychological stressors were related to poorer alcohol outcomes only among drinkers high in use of emotional avoidance coping strategies and high in expectancies that alcohol would lessen negative affect. Evidence./or avoided experiences in drug use. Because the effects of drugs are broad, almost any experience from boredom to anxiety attacks to withdrawal sensations can become linked to drug use. Among the private experiences considered important in various theories are expectancies and beliefs, cravings, aversive bodily states, and disagreeable emotions (Beck, Wright, Newman, & Liese, 1993; Childress, McLellan, Natale, & O'Brien, 1986; Meyer, 1986). Sanchez-Craig (1984) reported that, among 297 drinking episodes described by 70 participants, nearly 80% of the episodes involved drinking aimed at manipulating various subjective experiences (e.g., social discomfort, attenuation of negative emotions, increasing pleasure, etc.). Childress et al. (1986) demonstrated that emotions such as anxiety, anger, and depression function to trigger subjective experiences of craving and withdrawal among detoxified opiate addicts. Given this, it is not surprising that substance abuse is frequent among those with affective and anxiety disorders (e.g., Mirin, Weiss, & Michael, 1987; Pashko & Druley, 1987; Stockwell & Bolderston, 1987; Thyer et al., 1986). Researchers have suggested that some "with primary affective or anxiety disorders may abuse drugs in an attempt to alter undesirable mood states or to ameliorate intolerable anxiety" (Mirin et al., 1987, p.141). Similarly, because unpleasant private experiences seem most likely, given aversive histories, it is not surprising that the incidence of substance abuse among individuals with traumatic histories is high (e.g., Druley, Baker, & Pashko, 1987; Polusny & Follette, 1995; cf, Khantzian, 1985). Even if drug abusers did not start their patterns of abuse as a method of experiential avoidance, the effects of drugs on dysphoric or withdrawal states that are the result of excessive drug use may help maintain the pattern of abuse (Marlatt, 1985; Sher, 1987). Treatment implications. The psychological models applied to substance abuse and its treatment vary widely, but almost all are sensitive to the role of the abuser's private experience, in the development of the addiction or in relapse prevention, or in both. Most empirically oriented efforts to date have attempted to change the avoided content (e.g., through anxiety management, stress reduction, or pharmacotherapy) and thus reduce the need for drug use as a method of experiential manipulation (e.g., Miller & Hester, 1986; Stockwell & Bolderston, 1987). Others have tried to add conditioned aversive emotional states that are produced by reaching for tasting, or imagining drinking alcohol, so that experiential avoidance works for rather than against treatment goals (Rimmele, Miller, & Dougher, 1989). The direct use of acceptance methods in the empirical treatment of drug and alcohol abuse is in its infancy, although 12 step programs, meditative approaches, and others have long encouraged acceptance as a component of treatment (Marlatt, 1994; Wulfert, 1994). No data are currently available on the relative efficacy of acceptance versus control-based methods.