JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1998, 31, 137–156 NUMBER 1(SPRING 1998)

WHY BEHAVIOR ANALYSTS SHOULD STUDY EMOTION: THE EXAMPLE OF ANXIETY

PATRICK C. FRIMAN FATHER FLANAGAN’S BOYS’ HOME AND CREIGHTON UNIVERSITY SCHOOL OF MEDICINE

AND

STEVEN C. HAYES AND KELLY G. WILSON UNIVERSITY OF NEVADA

Historically, anxiety has been a dominant subject in mainstream psychology but an in- cidental or even insignificant one in behavior analysis. We discuss several reasons for this discrepancy. We follow with a behavior-analytic conceptualization of anxiety that could just as easily be applied to emotion in general. Its primary points are (a) that language- able humans have an extraordinary capacity to derive relations between events and that it is a simple matter to show that neutral stimuli can acquire discriminative functions indirectly with no direct training; (b) that private events can readily acquire discriminative functions; (c) that anxiety disorders seem to occur with little apparent direct learning or that the amount of direct learning is extraordinarily out of proportion with the amount of responding; and (d) that the primary function of anxious behavior is experiential avoidance. We conclude that the most interesting aspects of anxiety disorders may occur as a function of derived rather than direct relations between public events and overt and private responses with avoidance functions. Implicit in this conclusion and explicit in the paper is the assertion that anxiety is a suitable subject for behavior-analytic study. DESCRIPTORS: anxiety, emotion, avoidance, stimulus equivalence, relational frame theory

Anxiety is a topic of primary importance orders. There are several peer-reviewed jour- in most approaches to psychopathology. nals devoted exclusively to the topic (e.g., Anxiety co-occurs so prevalently with psy- Anxiety, Journal of Anxiety Disorders, Anxiety, chological distress that it, along with de- Stress, & Coping). The Diagnostic and Statis- pression, has been described as the psycho- tical Manual of Mental Disorders (4th ed., logical equivalent of fever (Carson, 1997). DSM-IV; American Psychiatric Association, Major theoretical positions ranging from 1994) lists 12 distinct anxiety diagnoses. A Freudian psychoanalysis (May, 1950; Stra- search for the term using the electronic chey, 1966) to existentialism (Yalom, 1980) psychFirst database produced 6,425 refer- describe anxiety as a central feature (some- enced articles from 1993 to 1997. In con- times the central feature) of behavioral dis- trast, contemporary behavior analysts have published very little on anxiety, or indeed on This research was supported by a grant to the first au- the scientific study of emotion in general. In thor from the Carmel Hill Foundation and a grant to the this paper we discuss reasons for this reluc- second author from the National Institute on Drug Abuse tance, describe recent developments in the (DA08634-02). Special thanks are extended to Keith Al- len, whose questions led to this project, and to Joe Sprad- experimental analysis of human behavior lin, Simon Dymond, Ray Miltenberger, Mike Handwerk, that make behavior-analytic study of anxiety Jack Finney, Winifred Ju, Michael Rapoff, and Bill Cook (and emotion in general) more tenable, and for comments on an earlier draft of the paper. briefly examine some applied implications of Address correspondence and requests for reprints to Pat- rick C. Friman, Youthcare Building, Boys Town, Nebraska these recent developments. The paper will 68010 (E-mail: [email protected]). conclude with a caveat pertaining to the

137 138 PATRICK C. FRIMAN et al. term anxiety and a recommendation for in- er metaphors with related usages such as creased research on the conditions that oc- butterflies in the stomach. The possibility of casion its use. butterfly disorders, butterfly management techniques, butterfly scales, and debates on REASONS FOR LIMITED the causal role of butterflies seems remote. BEHAVIOR-ANALYTIC RESEARCH Another obstacle to behavior-analytic As a research focus, anxiety is perilous for study of anxiety is the illusion of its status behavior analysts who wish to maintain fi- as an empirical phenomenon and the reality delity to their view of science. For example, of its status as a mere concept or category even Freud thought the term was imprecise (see Ryle, 1949). Categories are not empir- (Freud, 1917/1966). Precision refers to the ical events, that is, they cannot be observed number of ways that a specific phenomenon directly, and thus they are unappealing as may be construed verbally within a technical research targets for behavior analysts. Yet analysis; as the number goes down, precision conditions that govern category membership goes up and vice versa (Biglan & Hayes, are empirical events, and these have gener- 1995). Precise definition of terms is differ- ated abundant behavior-analytic study (e.g., entially valued across various approaches to Fields, Reeve, Adams, & Verhave, 1991; science; it is fundamental to behavior anal- Herrnstein, 1984) that has profitably con- ysis (Poling, Methot, & LeSage, 1995). De- verged with research from other fields (e.g., spite the enormous intellectual energy de- Medin & Smith, 1984; Rosch & Mervis, voted to anxiety since Freud, the number of 1975). A seemingly suitable topic for behav- verbal constructions organized under the ior-analytic study is thus the conditions that term seems to have grown larger and more occasion anxiety as a categorical response. ambiguous rather than smaller and more An apparent obstacle to this potentially pro- precise (cf. Hallam, 1985; Keedwell & ductive line of inquiry is the conventional Snaith, 1996; Levitt, 1967). but false assumption that anxiety, as a cate- One of the obstacles to precision is the gory, has an essential composition. metaphorical, idiomatic basis of the term This essentialistic position on anxiety is anxiety. Similar to other emotional terms, readily undermined with logical arguments anxiety was inaugurated as an ‘‘as if’’ de- (Wittgenstein, 1958a, 1958b). For example, scriptive idiom or metaphor (e.g., ‘‘it is as if a common definition of anxiety is physio- there is stuff that controls thinking,’’ Sarbin, logical reactivity to events with uncertain 1968, p. 413). Among its primary predeces- but potentially aversive outcomes. Walking sors was anguisse from the Old French which across a street that is free of cars and sud- became anguish in Middle English and anx- denly being overwhelmed by high-intensity iety in modern English (Sarbin, 1964, physiological responding (e.g., elevated heart 1968). Anguisse referred to choking sensa- rate, respiration, perspiration, and blood tions in the throat, which, in its subsequent pressure) is a well-accepted instance of anx- form, anxiety, was helpful in communicating iety (e.g., agoraphobia; American Psychiatric vague or excessive feelings associated with Association, 1994; Beck & Emery, 1985; aversive events (Oxford University Press, Hallam, 1985). Walking across the same 1971, p. 95; Sarbin, 1964, 1968; Skeat, street and almost being hit by a car produces 1963). The absurdity of the transition from the same physiology but is not an instance anxiety as metaphor to its current masquer- of anxiety. It is an instance of fear. Both are ade as technical term is revealed by posing related by physiology and avoidance and are the possibility of a similar transition for oth- separated by the actuality of the threat (Beck ANXIETY 139

& Emery, 1985; Epstein, 1967; May, 1950). 1992). In sum, anxiety appears to have no Being in the presence of a formerly reinforc- essential or nonreducible component, the ing commodity after an extended period of presence of which distinguishes participant deprivation (e.g., water, addictive substances, from nonparticipant events. This conclusion sex) also produces much the same physiol- seems to set the stage for a behavioral anal- ogy but is not an instance of anxiety or fear. ysis of the various usages of the term anxiety. It is an instance of craving (Wickler, 1973; One more obstacle remains, however, and it see also Pavlov, 1927, pp. 35–37). It has nei- is a theoretical one within behavior analysis ther the uncertainty nor the avoidance. Thus itself: Skinner’s analysis of emotion. physiology cannot be the essence of anxiety, Skinner’s analysis provided an opening for because it is part of many kinds of avoidance behavior-analytic study of emotion and si- (e.g., of events with known and unknown multaneously made stepping through it threat) and of many kinds of pursuit (e.g., seem unnecessary. Consistent with our com- for sex, drugs, fluids, food). Similar cases are ments above, he criticized traditional ways readily made for the other elements said to of speaking about emotion. In fact, Skinner’s compose anxiety. For example, demonstrably career-long opposition to mentalism in sci- dangerous events can evoke the physiology ence could have employed anxiety as a text- of fear but no avoidance and little uncer- book case. Mentalism commonly involves tainty (e.g., contact sports, skydiving). Neu- explaining behavior by appealing to inde- tral events that are conditioned to evoke the pendent variables that are inferred from the physiology of fear and avoidance (e.g., pho- behavior explained (Hayes & Brownstein, bias) sometimes produce neither (Rachman, 1986; Skinner, 1969; see also Wittgenstein, 1977, 1991). Some events are avoided or es- 1958a, 1958b). For example, in their influ- caped (i.e., avoidance or escape responses are ential book on anxiety and phobia, Beck and reinforced) but do not produce the physi- Emery (1985) state that anxiety disorders are ology of fear, nor are they associated with caused by an ‘‘upset in the cognitive system’’ danger (e.g., alarm clocks, obnoxious per- (p. 86). As a case in point, they describe the sons). most frequently occurring features of gen- Not surprisingly, given the persuasive log- eralized anxiety disorder (GAD), some of ical counterarguments, the essentialistic po- which involve upsets in the cognitive system sition on anxiety has generated little empir- (e.g., difficulty in concentrating, fear of los- ical support. The gold standard of psychi- ing control, etc.). The frequency of these atric diagnostic research is essentialistic; it features is then adduced as evidence of the involves identifying sensitive and specific bi- causal status of cognitive impairment. Yet if ological markers for psychiatric disorders these cognitive upsets are an intrinsic part of (Hoes, 1986). Yet the abundant research on GAD, using them to explain it is circular, numerous nominated markers, accompanied misleading, and mentalistic because the up- by experimental preparations appropriate for sets themselves remain unexplained. anxiety (e.g., salivary cortisol, lactate infu- Skinner went further than mere criticism sion, carbon dioxide inhalation, mitral valve of mentalistic use of terms such as anxiety, prolapse, monoamine oxidase levels), has yet however; he also offered a coherent behav- to yield a sensitive and specific (i.e., essen- ior-analytic alternative. In his landmark pa- tial) marker (e.g., Margraf, Ehlers, & Roth, per on the operational analysis of psycholog- 1986, 1988; McBurnett et al., 1991; Rapee, ical terms, Skinner laid the philosophical Brown, Antony, & Barlow, 1992; Sofuoglu, and theoretical foundation for a behavioral Dogan, Besim, Basturk, & Tanrikulu, analysis of emotion (Skinner, 1945). Con- 140 PATRICK C. FRIMAN et al. trary to caricatures of behaviorism that sug- ample, if a crying child complains that he or gest a denial of private events such as emo- she feels anxious and then runs away, most tion and cognition, radical behaviorism parents (and other persons including psy- ‘‘does not insist upon truth by agreement chologists) are likely to attribute the running and can therefore consider events taking to anxious feelings (i.e., the child ran because place in the private world within the skin’’ he or she felt anxious). This explanation, (Skinner, 1974, p. 16). In contrast to many however, is predicated on a behavior–behav- emotion theorists, Skinner downplayed the ior relationship and is thus incomplete. Feel- scientific value of research on the form of ing anxious is a behavioral event that cannot emotional responses and recommended be directly manipulated in a way that estab- study of the contingencies that govern emo- lishes a functional relationship between it tional talk instead (Skinner, 1945). To Skin- and running. Thus, although the role of feel- ner, the meaning of a term (e.g., anxiety) ing anxious as predictor can be readily es- resides in the functional relationship be- tablished, its role as controlling variable can- tween its use and stimuli that are antecedent not. To be complete in behavior analysis, an and consequent to the use. In other words, explanation must ultimately appeal to ma- understanding the meaning of the state- nipulable environmental events. ment, ‘‘I’m anxious,’’ requires knowledge of Skinner develops this logic using an ex- the context, both current and historical, that ample of a situation in which an electric occasioned the utterance. Skinner’s argu- shock has been reliably preceded by a bell. ments were elegant, plausible, and directly ‘‘The bell comes to elicit reactions, primarily pertinent to emotion, but they were derived in the autonomic nervous system, which are entirely from an analysis of direct contingen- felt as anxiety’’ (Skinner, 1974, p. 61). The cies. individual’s behavior ‘‘does not change be- Although Skinner’s approach provided a cause he feels anxious; it changes because of needed opening for scientific analysis of the aversive contingencies which generate emotion (and private events in general), his the condition felt as anxiety. The change in emphasis on direct contingencies limited the feeling and the change in behavior have a potential scope of research programs when common cause’’ (Skinner, 1974, pp. 61–62). In this and other instances, Skinner argues they were conducted with fidelity to his that emotional feelings are co-occurring views. Such programs would be simulta- products of the same contingencies that pre- neously viable and valid, yet not very pro- cipitate overt behaviors which the feelings ductive. Understanding how this conclusion are said to ‘‘cause’’ (e.g., running) and that was reached is an important step towards a they have ‘‘no functional significance, either more contemporary and potentially impor- in a theoretical analysis or the practical con- tant approach to behavior-analytic study of trol of behavior’’ (Skinner, 1953, p. 181). emotion. Thus, although Skinner did establish the sci- The two fundamental goals of scientific entific plausibility of analyzing private explanation in behavior analysis are predic- events, he also suggested that such analyses tion and control of the phenomenon under were theoretically and practically unneces- study (Biglan & Hayes, 1995; Hayes & sary. Brownstein, 1986; Skinner, 1969). To the layman (and many, possibly most, psychol- WHY EMOTION IS A VIABLE ogists outside behavior analysis), emotional BEHAVIOR-ANALYTIC TOPIC NOW talk itself is satisfactory as an explanation for The central theme of the previous section behavioral events involving emotion. For ex- was that anxiety is still not well understood. ANXIETY 141

Perhaps the rigor of behavior-analytic theory unidirectional relation will lead to a derived, and methods is needed for increased under- bidirectional relation. That is, given B, A standing of this seemingly ubiquitous, yet will be chosen, without additional direct misunderstood, phenomenon. Although the training, even in children as young as 16 initial promise of Skinner’s radical approach months (Lipkens, Hayes, & Hayes, 1993). to human behavior was a complete science By the age of 2 years, children learn even of private and public events, study of the more remotely derived stimulus–stimulus re- former has lagged far behind study of the lations. For example, if A-B and A-C rela- latter. Now, however, the field of behavior tions are trained, then B-C and C-B rela- analysis is changing, and, correspondingly, tions are derived (Devany, Hayes, & Nelson, the possibility of experimental analysis of 1986; Lipkens et al., 1993). These relations private events is increasing. The dangers for have been termed stimulus equivalence (Sid- science, discussed in the previous section re- man, 1971, 1994; Sidman & Tailby, 1982). main relatively intact. But the intellectual Recent research shows that many relations barrier created by Skinner’s position is grad- other than equivalence (e.g., greater than, ually giving way to empirical developments less than, opposition, difference, etc.) can be in the analysis of human verbal behavior learned, applied arbitrarily to stimulus that, in turn, expand the possibilities for be- events, and combined into networks of stim- havior-analytic study of emotion. ulus relations of almost unimaginable com- Skinner’s analyses of schedules of rein- plexity (Dymond & Barnes, 1995, 1996; forcement benefitted from a wealth of ex- Steele & Hayes, 1991). In addition, other perimental evidence (e.g., Ferster & Skinner, processes, such as stimulus generalization, 1957; Skinner, 1938). In striking contrast, can combine with derived relational re- his analyses of verbal behavior were virtually sponding to merge large and diverse classes devoid of experimental evidence (Skinner, of responding (Fields et al., 1991). Further- 1957) and thus did not benefit from the se- lective effects that data have on theorizing. more, the psychological functions of the el- In recent years, a large body of empirical ements in these ‘‘relational networks’’ tend evidence from the experimental analysis of to change or transform in accord with the human behavior has been published. Of par- underlying derived stimulus relation. In the ticular relevance to the analysis of emotion case of equivalence relations, several dem- is the literature on derived relational re- onstrations of this transformation are avail- sponding such as stimulus equivalence and able, and they include conditioned reinforc- the transformation of stimulus functions ing functions (Hayes, Brownstein, Devany, (e.g., Barnes, 1994; Hayes & Hayes, 1989, Kohlenberg, & Shelby, 1987; Hayes, Koh- 1992; Hayes & Wilson, 1993; Sidman, lenberg, & Hayes, 1991), discriminative 1994; see also Horne & Lowe, 1996). This functions of public (Hayes et al., 1987) and rapidly growing line of investigation may private (DeGrandpre, Bickel, & Higgins, fundamentally change the behavior-analytic 1992) stimuli, elicited conditioned emotion- view of verbal events and, with it, the anal- al responses (Dougher, Augustson, Mark- ysis of private events. ham, Greenway, & Wulfert, 1994), extinc- The research shows that humans readily tion functions (Dougher et al., 1994), and learn derived stimulus relations. If a lan- sexual responses (Roche & Barnes, 1997). In guage-able human, while in the presence of the case of relations other than equivalence one stimulus, A, learns to select an arbitrar- (e.g., opposition), demonstrations of corre- ily related stimulus, B, then this trained, sponding transformed stimulus functions are 142 PATRICK C. FRIMAN et al. also increasingly available (Dymond & behavioral accounts of complex behavior, es- Barnes, 1995; Roche & Barnes, 1997). pecially when private events are involved. A Studies documenting the transformation synthesis of just a few of the many related of stimulus functions propel a more contem- lines of research will underscore this point porary behavior-analytic account of verbal and set the stage for application of derived behavior and approach to emotion and other relational concepts to anxiety disorders. A private events. In addition, new behavioral large, long-standing, cross-disciplinary line terms might be needed (Hayes & Hayes, of research shows that generalized respond- 1992). For example, if a child is taught stim- ing to stimuli with discriminative function ulus relations between three stimuli, A, B, (discriminanda) spreads to novel stimuli that and C, that bear no formal similarity to each resemble the discriminanda (via stimulus other, and then B is given a discriminative generalization), resulting in large relational, function through direct training (e.g., rein- polymorphous, or fuzzy categories of re- forcement for waving in the presence of B), sponses (Fields et al., 1991; Herrnstein, the function will obtain for C (i.e., waving 1984; Medin & Smith, 1984; Rosch & is now likely in the presence of C) without Mervis, 1975; see also Wittgenstein, 1958b). direct training (Hayes et al., 1987). In this A more recent behavior-analytic line of re- example, B is a discriminative stimulus but search shows that events in relational classes C is not, because a greater probability of re- spread not just with stimulus equivalence ef- inforcement for waving in its presence than fects but also with these stimulus generaliza- in its absence has not been established. Wav- tion effects. For example, once Stimulus ing in the presence of C is also not an in- Events A and B are entered into a relational stance of stimulus generalization, because class, the events related discretely to A and there are no formal stimulus properties to to B via stimulus generalization can become account for the generalization. Nonetheless, related to each other via stimulus equiva- C functions like a discriminative stimulus, lence (Fields et al., 1991). Thus, large rela- but it is produced indirectly by the transfor- tional, fuzzy, or polymorphous categories mation of stimulus functions through de- created via stimulus generalization can rived stimulus relations, not by a history of merge with other large relational, fuzzy, or direct contingencies (as in the conventional polymorphous categories if a member from Skinnerian account). Recent papers have one is entered into an equivalence relation suggested that such functions are not just with a member from another. Another re- ‘‘verbal’’; they may be a defining property of cent line of behavior-analytic research shows verbal behavior (e.g., Barnes, 1994; Hayes & that private and public events can become Hayes, 1992; Hayes & Wilson, 1994, 1996; part of the same relational class (De- Roche & Barnes, 1997). Thus C, in the pre- Grandpre et al., 1992). In other words, these ceding example, might profitably be referred large relational categories can (and usually to as a verbal discriminative stimulus (Hayes do) include both public and private events. & Wilson, 1993). Although logically consis- Lastly, a rapidly expanding line of research tent, the argument for a modified nomen- shows that respondent (e.g., Dougher et al., clature has yet to recruit a consensus among 1994; Roche & Barnes, 1997) and other be- behavior analysts (e.g., Leigland, 1997; J. havioral functions (e.g., Hayes et al., 1987, Spradlin, personal communication, October 1991) participate in relational networks and 23, 1997). transform the behavioral functions of their The comments above suggest the breadth constituent responses. Collectively, these that derived relational responding adds to lines of research provide a preliminary basis ANXIETY 143 for penetrating the complexity of anxiety the concept of anxiety will require an ex- disorders. traordinary research effort. Anxiety disorders seem irrational because they primarily involve high-rate avoidance Implications for the Analysis of Anxiety of, or escape from, events that have no readi- A major implication of derived relational ly detectable direct relationship with punish- responding involves the potentially powerful ment (e.g., open spaces, doorknobs, harm- effect it can have on self-discriminations: It less insects). Analysis of anxiety-like re- can render them reactive (Dymond & sponses in animals suggests ‘‘a combination Barnes, 1995, 1996). A related issue is that of historically real and currently adventitious these effects are unlikely when nonverbal or- contingencies’’ (Sidman, 1960, p. 66). Thus, ganisms self-discriminate because they are even with a direct contingency analysis of not nearly as prepotent for derived respond- pertinent behaviors in organisms with much ing as humans are. In fact, whether animals simpler response patterns than humans, anx- can exhibit derived responding at all is a iety responses are puzzling (i.e., seem irra- contested issue (Devany et al., 1986; Hayes, tional). In the analysis of human anxiety re- 1989; Horne & Lowe, 1996). Even if re- sponses, the puzzles multiply in number and search ultimately unequivocally documents complexity because human response patterns derived relational responding in some non- are also strongly influenced by indirect rela- humans, the argument here remains intact. tionships between events and public and pri- That is, behavior resulting from derived re- lations requires an analysis that is somewhat vate responses to public and private events. different from that offered by Skinner. Thus, Of course, not all avoidance responding is when anxiety-like responses (e.g., avoidance) puzzling. Responses that are reinforced by in any organism (human or nonhuman) are avoidance of events with phylogenetic sig- traceable to direct contingencies, Skinner’s nificance (e.g., fire, large fur-covered carni- analysis of emotion seems to be sufficient. vores, heights) readily surrender to analysis. That is, when only direct contingencies are So too do responses that are reinforced by involved, it follows that ‘‘the change in feel- avoidance of seemingly nonsignificant events ing and the change in behavior have a com- when those events can be linked formally or mon cause’’ (Skinner, 1974, p. 62; see also functionally to aversive consequences. But Sidman, 1960). When the anxiety-like re- clinically significant anxiety appears to refer sponses cannot be traced to direct contin- to avoidance responses whose initiating con- gencies and the organism is prepotent for ditions are direct but very remote and whose derived relational responding (e.g., language- perpetuating conditions are mostly derived. able persons), additional analyses are re- The life of the clinically anxious person may quired. thus be influenced by iterations and reiter- Consider two situations: In one, a hungry ations of public and private events with re- rat can obtain food immediately but the de- active properties traceable to initiating con- livery is accompanied by a small shock; in ditions only through an almost fractal path- the other, a small delay is required for the way involving the processes of stimulus gen- food but no shock occurs. If the values are eralization, derived relational responding, set properly, the rat will consume the food and transformation of stimulus function (cf. immediately (i.e., impulsively) and will be Hayes & Wilson, 1993, 1994; see also An- shocked. If the rat is trained to press one derson, Hawkins, & Scotti, 1997). It seems lever for food if it has been shocked and safe to say that a true behavior analysis of another if it has not been shocked, it, in 144 PATRICK C. FRIMAN et al. effect, will have learned to report whether it example, mere thoughts about engaging in has been shocked (e.g., Lattal, 1975). The public speaking can instigate autonomic difference between a rat and a language-able arousal and self-limiting appraisals that are human is that the rat’s report (or self-dis- highly correlated with reluctance. As another crimination) is unlikely to have an effect on example, mere thoughts about engaging in subsequent similar (i.e., impulsive) respond- sexually related behavior can instigate arous- ing. That is, the report is unlikely to be bi- al of a different sort that is correlated with directionally related to the event being re- approach. A vast, albeit nonbehavioral, lit- ported; for rats and most (perhaps all) non- erature documents the adverse influence of human organisms, an event–report relation persistent self-deprecatory thoughts (Peter- does not entail a report–event relation. Ab- son & Seligman, 1984). Skinner acknowl- sent this bidirectionality, there is no known edged the behavioral impact of self-knowl- pathway for the report to influence subse- edge: ‘‘Self-knowledge has a special value to quent events similar to those reported. Sim- the individual himself. A person who has ilarly, there is no known pathway (absent ad- been ‘made aware of himself’ is in a better ditional experimental preparations) for the position to predict and control his own be- rat’s report of shock to become aversive it- havior’’ (Skinner, 1974, p. 35). However, self. The report would not be bidirectionally Skinner did not supply an account of how related to the event of shock, and thus it the behavioral effects occur. Such an account would not share its functions. Functionally, is not readily obtained from an analysis of the report is related to food, not to shock. direct contingencies (refer to the behavior of Reports of punishing events are simply not the rat above). When indirect responding aversive for organisms that are not prepotent through derived relations is added to the for derived relational responding. analysis, however, a verbal basis for the be- As indicated above, the responses of lan- havioral effects of self-knowledge is suggest- guage-able persons in similar circumstances ed. are different than those of simpler organisms Human emotion is a rich source of ex- such as rats. For the human, the shock and amples that suggest a verbal basis for many the report are likely to be bidirectionally re- important behavioral effects. Emotions are lated and thus share functions; some of the integrated into loosely defined and labeled effects of the shock are likely to occur when categories. Yet, loose definition and labeling the shock event is reported. In addition, dis- notwithstanding, under the right circum- tinct from the simpler organisms, a person’s stances, merely saying or thinking about the report of the behavior that led to the shock label renders self-discrimination of the la- is likely to affect the probability of being beled emotion a verbal (bidirectionally re- shocked in similar circumstances, because active) event (at least in part). To the extent verbal (i.e., derived) reports are bidirectional that this is true, it is not true that ‘‘the (event–report entails report–event). Thus, a change in feeling and the change in behavior person’s self-discriminations (or self-knowl- have a common cause’’ (Skinner, 1974, p. edge) can affect his or her behavior. To ap- 62), because both nonverbal and verbal con- preciate this point experientially, imagine tingencies are mixed in the control of the biting into a lemon. Some of the perceptual behavior and the feeling, and they may be functions of biting occur merely as a func- mixed in different ways. For anxious lan- tion of the thought, that is, the thoughts are guage-able persons, reports of anxiety are not reactive. Similar to thoughts about lemons, necessarily mere statements whose sole func- self-referential thoughts can be reactive. For tion is communication (for self or other) ANXIETY 145 about various contexts, bodily states, or be- responding, the feeling and the thought of havioral predispositions associated with anx- anxiety are strongly influenced by the prop- iety. The reports can also be reactive and erties of punishment. Illustrative examples thus generate untoward effects. That is, they are plentiful. Recounting a highly aversive not only describe the person’s behavior and event often instigates some (often much) of circumstances but can also alter the function the arousal and avoidance responding that of the behavior and circumstances described. was occasioned by the event itself. Merely For some persons, anxiety is a potentially de- thinking about the event can produce similar bilitating category whose members include a effects (Hayes, Strosahl, & Wilson, in press; variety of functional altering verbal compo- Hayes & Wilson, 1993, 1994). Complicat- nents (e.g., unpleasant memories, negative ing matters further is the expanding conflu- self-evaluations, unfavorable social compari- ence of functions that gather in anxious ep- sons, etc.) and related bodily states (e.g., au- isodes as circumstances change from episode tonomic arousal). In a colloquial sense, anx- to episode and the relational class grows. iety is what anxiety does, and what it does Highly anxious persons generally do not ex- includes what anxious persons say it does. hibit the pertinent responses (e.g., arousal, Thus, many public and private events are avoidance, escape, etc.) in the presence of integrated in the category of responding just one thing (American Psychiatric Asso- known as anxiety. A common observation ciation, 1994). Their responses occur in the about anxiety episodes is the appearance of presence of so many things that they some- surplus responding. That is, the related re- times behave as if they are ‘‘afraid of their sponses (e.g., arousal, avoidance, escape) are own shadow.’’ not fully explained by the direct contingen- Some of the complaints about anxious ep- cies in the present nor by those in the ap- isodes involve emotional subcategories and prehensible past. The derived relational ac- the phenomena integrated under them. The count seems to supply some of the missing labels for the categories and their associated information, however. As described previ- public and private events can, similar to anx- ously, through transformation of stimulus iety, produce clinical effects through direct functions some of the behavioral functions and derived relations. For example, a com- of one class of stimuli can carry over to an- mon complaint about some forms of anxiety other class, and this other class then begins involves a ‘‘sense of loss of control’’ (e.g., to exhibit behavioral functions that are sim- panic). Other forms involve ‘‘embarrass- ilar to those of the first class. ment’’ (e.g., social phobia). Some involve For example, if a person feels anxious and ‘‘social humiliation’’ (e.g., agoraphobia). a punishing event occurs (e.g., social rejec- There are numerous other examples. The tion), the functions of the feeling and the point here is not that these labels, or anxiety event may merge through bidirectional stim- itself, should be accepted into the argot of ulus transformation. Circumstances similar behavior analysis. As indicated earlier, anxi- to those that occasion the feeling, as well as ety is a vaguely defined, largely metaphorical the verbal report of the feeling, may subse- term and is unlikely to ever become a truly quently exhibit punishing properties, and technical term. If anything, these other cat- the punishing event may become more anx- egory labels are even more vague and met- iety provoking. This merging of function aphorical. Rather, the point it is that (a) the can create clinically debilitating response labels are important members of the lan- patterns (e.g., arousal, avoidance, escape, guage of the English-speaking culture, (b) etc.) especially if, through derived relational they refer to important psychological phe- 146 PATRICK C. FRIMAN et al. nomena, and (c) through derived relational pragmatists, and applied behavior analysts responding, they can have adverse influenc- are especially so. To recruit agreement with es. our view of emotion as an important topic There are other important implications of for applied behavior analysts, it is essential this verbal account of anxiety. For example, to show how study and treatment of clini- the account provides some support for talk cally significant behavior can be improved therapy. If verbal events are functionally re- when an analysis of emotional talk is con- lated to behaviorally and clinically important ducted. events in a client’s current and previous en- To begin this task, we will argue that be- vironments, it follows that knowledge of the havior resulting in avoidance of or escape relationships could help the therapist and from negative emotional states, as verbally client predict and influence important be- construed by an individual, is negatively re- havioral outcomes. The client’s self-reports inforced. Critical to the analysis is that some are likely to be informative on a number of (perhaps most) of the aversive properties of levels, including the direct communication the emotional states and reinforcing prop- of relevant information and the exhibition erties of the avoidance are primarily derived of clinically relevant psychological functions through arbitrarily applicable relational re- (Kohlenberg, Hayes, & Tsai, 1993; Kohlen- sponding. That is, emotional avoidance ap- berg & Tsai, 1991; Wulfert, Greenway, Far- pears to be an integral component of human kas, Hayes, & Dougher, 1994). Events that verbal behavior (Hayes, Wilson, Gifford, are presented verbally in clinical sessions Follette, & Strosahl, 1996). Our aim in (e.g., events ‘‘presented in imagination’’) may making the argument is to supply applied also generate effects similar to those gener- behavior analysts with theoretical and ap- ated by directly presented events. A well- plied reasons to step through the opening known case in point is the imaginal exposure Skinner provided in 1945. used in systematic desensitization (e.g., Wol- pe, 1990). ANXIETY DISORDERS AND More generally, a verbal account of emo- EXPERIENTIAL AVOIDANCE tion underscores the value (for clinical be- Conventional Diagnoses and Behavioral havior analysts) of attention paid to the cli- Alternatives ent’s private events. Those events include po- tentially maladaptive verbal and nonverbal A wide variety of anxiety disorders are responses that are the direct and indirect re- identified as distinct entities in the DSM di- sults of socialization within a client’s verbal agnostic system. Over time, the various re- community. The English language contains visions of the DSM have systematically in- myriad emotional terms, in part because the creased the number of purportedly distinct verbal community at large trains its mem- disorders. In the most recent half-step, from bers to use the terms to discuss their behav- the third revised edition to the fourth edi- ioral histories and to respond to those his- tion (American Psychiatric Association, tories in a socially acceptable manner. 1987, 1994), the number of diagnosable The preceding comments constitute an anxiety disorders increased from 9 to 12. argument for the theoretical importance of With no theory to restrain the increasing a behavioral account of emotion. Toward the proliferation of disorders, and with profes- end of the argument, applied implications of sional reinforcement for finer and finer dis- such an account began to emerge. This was tinctions in their formal properties, an infi- a natural progression. Behavior analysts are nite number becomes possible (Carson, ANXIETY 147

1996, 1997; Follette, Houts, & Hayes, two categories of phenomena whose avoid- 1992). These disorders are also often arbi- ance results in negative reinforcement: the trary with respect to function, which is evi- events and the responses to them (cf. Gif- dent in the following comparison of phobias ford, 1994; Hayes & Wilson, 1993; Hayes with sexual aversion. If a person is fearful of et al., 1996). This outcome may be specific and therefore persistently avoids snakes, to language-able persons. As indicated pre- shopping malls, blood, heights, social inter- viously, simpler organisms (e.g., rats) do not actions, elevators, or germs, he or she is di- exhibit bidirectional derived relational re- agnosed with an anxiety disorder. If the per- sponding, and thus there is no known path- son is fearful of and therefore persistently way for their responses to aversive events to avoids sex, however, he or she is diagnosed also become aversive. In other words, it may with sexual aversion, which is not an anxiety be true that only language-able persons ex- disorder. This theoretically unguided diag- hibit experiential avoidance or behavior nostic proliferation and functionally arbi- whose primary function is the elimination, trary diagnostic application could be re- minimization, or reduction in the form, fre- duced substantially through use of a func- quency, or situational sensitivity of various tional-dimensional approach to diagnostic private events (Hayes et al., 1996). Relat- classification. edly, it seems that all of the anxiety disorders The functional-dimensional approach has could be cogently classified as experiential recently been proposed as a behaviorally sen- avoidance disorders. Below is a brief review sible alternative to the DSM. The approach of a sample of DSM anxiety disorders; its involves organizing diagnoses according to purpose is to show the unifying effect of the behavioral processes (therefore functional), functional dimensional approach to diagnos- and these processes are seen as continua tic classification. (therefore dimensional rather than categori- Obsessive compulsive disorder (OCD). cal; Hayes et al., 1996). Experiential avoid- OCD involves the presence of unacceptable ance has been suggested as one functional thoughts (obsessions) and disruptive mal- dimension that could integrate a variety of adaptive responses (compulsions) that pro- diverse DSM diagnoses. vide temporary avoidance or escape from the thoughts and the physiological responses as- Experiential Avoidance as a sociated with the thoughts (Hollander, Functional Dimension 1993; Rasmussen & Eisen, 1992). Among Early in life, humans, and most other an- individuals diagnosed with OCD, unaccept- imals, learn a large and versatile repertoire of able thoughts dominate conscious awareness. strategies (e.g., vigilance, withdrawal) for These thoughts involve themes such as self- avoiding events (e.g., pain, danger). As hu- contamination, doubts about whether one mans develop, their verbal repertoires quick- has performed an important act (e.g., turned ly become quite elaborate, and, correspond- out lights, left doors open), or the desire to ingly, their responses to aversive events ex- engage in socially unacceptable behavior hibit verbal properties. That is, the responses (e.g., profanity, aggression). When such (e.g., bodily sensations, emotions, thoughts, thoughts occur to persons diagnosed with memories, behavioral predispositions) occa- OCD, they try to ignore them, distract sioned by the events begin to participate in themselves, or develop elaborate rituals to derived relations with the events. Through a avoid or escape from the thoughts. For ex- transformation of function, the responses ample, if the person’s thoughts involve un- themselves can become aversive, resulting in clean hands, he or she is likely to systemat- 148 PATRICK C. FRIMAN et al. ically and repetitively wash them. If inter- perceptions of physiological activity [that] rupted, the person is likely to become dis- can culminate in a panic attack’’ (Pauli et al., tressed, believing the interruption has 1991, p. 137; see also Clark, 1986; Ehlers, interfered with the quality of the washing, Margraf, Roth, Taylor, & Birbaumer, 1988; which amplifies the presence of the thought Taylor, 1995). In this ‘‘feedback loop,’’ bodily about having unclean hands. Therefore, states and other reactions occasion a verbal washing begins again and the cycle contin- formulation that links the present to an im- ues until the thought is absent or present in pending dire future (e.g., ‘‘I’m losing control. a much diminished form (e.g., ‘‘I’m almost I’m going crazy. I am about to die.’’). Unfor- clean’’). tunately for those afflicted with panic, the Consistent with our experiential avoid- bodily states occasioned by bona fide peril ance approach to anxiety disorders, the dis- and those that occur through derived rela- ruptive washing has important functional tional responding (i.e., their reactions to their properties: reducing, avoiding, or escaping reactions) are virtually indistinguishable. private events (experience). Thoughts are Thus, panic can be very behaviorally disrup- private verbal behavior. OCD-type responses tive. In addition, autonomic arousal (fear) oc- are reinforced by avoidance of related, highly casioned by the thought of panic is a stronger unpleasant thoughts (e.g., ‘‘My hands are negative predictor of treatment outcome than not clean’’) and the various other private is frequency or intensity of panic itself (Keij- events that occasion or follow these formu- sers, Hoogduin, & Schaap, 1994; cf. Chamb- lations. less & Gracely, 1988). In sum, panic disorder Panic disorder with agoraphobia. Panic in- with agoraphobia is cogently categorized as volves intense autonomic arousal that is not an experiential avoidance disorder. as related to actual events as it is to the re- Posttraumatic stress disorder (PTSD). action to those events; in a colloquial sense, PTSD is the most recent and the most wide- it is fear of fear (Taylor, 1995). For example, ly applied diagnostic category for trauma-re- persons with agoraphobia do not avoid pub- lated emotional disturbance. Previous de- lic places (e.g., the mall) per se; they avoid scriptors were more colloquial and specific the possibility of what might happen if they but appeared to refer to the same emotional enter a public place. Thoughts of a panic phenomenon (e.g., shell shock, rape trauma attack in the public place, and of the con- syndrome). PTSD involves direct exposure stellation of embarrassing responses that to a traumatic event and the subsequent would become public if the attack were to emergence of three clusters of symptoms: occur there, are the negative reinforcers for reexperience (nightmares, flashbacks), avoid- staying home (e.g., Amering et al., 1997). ance in active (avoiding trauma-related stim- This account does not fully explain panic, uli) and passive (numbing, disassociation) but it provides an advance when combined forms, and increased arousal (e.g., insomnia, with a contemporary behavioral view of ver- hypervigilance, exaggerated startle responses; bal events. American Psychiatric Association, 1994; Foa Persons diagnosed with agoraphobia be- & Meadows, 1997; Foa & Riggs, 1995). come extremely reactive to changes in their The pertinence of the experiential avoidance physiological state (Barlow, 1988). Small in- approach to classifying PTSD is evident in creases in heart rate may be interpreted ver- its symptom clusters. The first cluster in- bally as catastrophic (Pauli et al., 1991). This volves the unpleasant experience and reex- catastrophic interpretation has been theorized perience of the traumatic event, the avoid- to result in a ‘‘positive feedback loop between ance or escape from which reinforces re- ANXIETY 149 sponses in the second cluster: numbing, ac- DSM anxiety disorders, the redundancy of tive avoidance, and disassociation. The third such an effort would likely place an unnec- cluster involves persistent arousal that has an essary burden on the reader. Before moving apparently elicited basis, at least part of to treatment implications, we want to briefly which is due to formal similarities between reiterate two central points. The first point the traumagenic and current events. The is that when analysis of anxiety disorders in- pervasiveness of the arousal, however, seems corporates derived and direct learning pat- to unduly tax respondent conditioning as its terns, it encompasses vastly more of the rel- sole explanation. It seems likely that some evant phenomena than a solely direct con- (perhaps most) of the arousal is verbally pro- tingency analysis does, and it does so with- duced. out resort to hypothetical constructs (i.e., it The hyperarousal of PTSD often extends is strictly learning based). to conditions that bear no formal similarity The second point is that the functional to the original traumagenic settings. Situa- classification of behavior disorders is fostered tions that are only verbally or metaphorically by an analysis of verbal relations and that related, such as unknown or unpredictable the anxiety disorders have a dominant, ver- situations, may also generate hyperarousal. bally established function of experiential Although it is conceivable that such re- avoidance. As indicated above, a core con- sponses are due solely to higher order con- stituent of anxiety is avoidance. What is ditioning (cf. Forsyth & Eifert, 1996a, needed is a fuller understanding of all the 1996b), experiential avoidance and derived phenomena that are avoided. Anxious per- relational responding contribute to a precise sons are fearful not only of environmental account with considerably more scope. For events but also of their responses to those example, the respondent account cannot events. A person with panic disorder does readily explain why avoidance would lead to not just avoid public places; he or she avoids increases in hyperarousal (in fact, from a re- the full range of private behavior associated spondent perspective, decreases would be ex- with those places. Skinner’s contention that pected). Consistent with an experiential emotion and overt behavior are controlled avoidance perspective, PTSD is better pre- by the same events is thus incorrect or at dicted by the avoidance symptom cluster least incomplete. A fuller understanding re- (e.g., numbing, avoidance, and dissociative quires an analysis of the complex verbal con- symptoms) than by the aversive stimuli (fear tingencies that are involved in the human and horror) that were present at the precip- disposition to categorize arbitrary events itating event (Foa & Riggs, 1995). From this (e.g., a pounding heart) as negative emotions perspective, the reactivating role of the and respond accordingly (‘‘I’m feeling pan- avoidance cluster stems from a derived re- icky, I have to get out’’). An analysis of direct ciprocal relationship between avoidance, contingencies might reveal the basis for the guided by verbal formulations that increase person’s avoidance of public places, but it sensitivity to emotional reactions, and per- cannot readily account for the avoidance of ceived (imagined, verbally formulated) direct his or her thoughts and feelings about those consequences that instigate more reactions. places. The previous discussion was used to show These points yield the central themes of very briefly the application of the experien- this paper: A behavior analysis of human tial avoidance approach to three prevalent emotion is not only possible but is ultimate- anxiety disorders. Although we could readily ly necessary for understanding and treating apply the approach to all 12 of the current anxiety disorders. To complete the analysis, 150 PATRICK C. FRIMAN et al. an examination of both direct and verbal matic sensations or to engage in distraction contingencies is needed. tasks during exposure to feared settings. Re- sults showed that, although the distraction Treatment Implications group exhibited greater improvement post- The functional benefits from the study of treatment than did the focus group, the fo- emotion are perhaps most apparent in the cus group exhibited greater improvements at context of treatment. The dominant behav- the 6-month follow-up. Similar results have ioral approach to anxiety disorders is been shown in persons diagnosed with OCD straightforward: Expose the anxious persons (Foa & Kozak, 1986; Grayson, Foa, & Stek- to the feared phenomenon repeatedly and al- etee, 1982). low other behavioral processes (e.g., habitu- Thus exposure is crucial for effective treat- ation, positive reinforcement) to extinguish ment of anxiety, not merely in the brute the maladaptive avoidance response class. Al- physical sense (e.g., reinforce successively though many investigators recognize that closer movements to a feared object) but also private events may also be important, a com- in the psychological sense, which is consid- mon belief is that extinguishing public reac- erably more difficult. Experiential avoidance tivity to the feared object will concomitantly is an integral part of language itself, psycho- extinguish this private reactivity (e.g., Skin- logical avoidance is often readily available ner, 1969, 1974). A wide variety of exposure through distraction or other means, and ex- techniques have been developed that roughly posure-based procedures generate a signifi- accord with this view (e.g., systematic desen- cant rate of resistance or refusal (cf. Hayes sitization, response prevention). et al., 1996; McCarthy & Foa, 1990). Thus, If the present account is correct, however, exposure-based treatment requires special that view may often be insufficient. To be components that undermine covert avoid- fully effective, exposure may frequently have ance, maintain participation, and reduce to include all, or at least more, of the events avoidance of private verbal events. Absent a that functionally occasion maladaptive detailed analysis of the client’s emotional avoidance. Most anxious persons seek treat- constructions, effective programming of the ment to master being in the presence of the needed exposure will often be difficult and feared object or event while not thinking sometimes impossible. about or feeling fear. This means that the Recently, some behavior analysts have de- verbal aspect of fear is part of the feared veloped treatments whose primary compo- event. To incorporate more of the phenom- nent is effective exposure. For example, ac- enology of fear in treatment, knowledge of ceptance and commitment therapy (ACT) is the feared events and of the verbal behavior an exposure-based behavioral intervention occasioned by the events is therefore needed. that deliberately focuses on psychological ex- There are data from several laboratories posure and the weakening of avoidance of developing exposure-based treatments for private verbal events (Hayes et al., in press; anxiety disorders that illuminate this prob- Hayes & Wilson, 1993, 1994; Kohlenberg lem. An expanding body of evidence shows et al., 1993). Consistent with the experien- that attention to avoided private events im- tial avoidance perspective of anxiety, the proves outcomes in exposure-based treat- ACT perspective is that event-based unpleas- ments (Barlow, 1988). For example, Craske, ant emotion is not the essence of the client’s Street, and Barlow (1989) gave persons who dysfunction; it is the cluster of responses had been diagnosed with agoraphobia in- whose function is to reduce, avoid, escape, structional sets to either focus on feared so- or otherwise modify the unpleasantness. ANXIETY 151

ACT therefore strives to attenuate the rela- The highly successful extinction-based treat- tion between the problematic responses and ment involved reinforcing completion of their avoidance function. Among other math problems in the presence of live in- goals, the method attempts (a) to experien- sects. The boy’s most salient fears were of tially demonstrate the futility of the client’s spiders and ladybugs, which were relatively self-avoidance responses, (b) to redefine ex- unavailable, so live crickets, which were periential control as the problem rather than readily available at a local pet store, were the solution, (c) to reinforce the client’s ef- used instead. Eventually the boy completed forts to remain in the presence of the pre- high rates of math problems with insects viously avoided experiences, and (d) to re- present on his work table. Following treat- duce the linkage between overt behavior and ment, the boy’s aversion to crickets (part of private events. In a general clinical setting, treatment) and spiders and ladybugs (not clients of therapists who have been trained part of treatment) subsided to below clinical in these concepts and techniques improved levels. The reactive effects of classmates’ significantly more than did clients treated comments about insects were also eliminat- with mainstream methods (Strosahl, Hayes, ed. On the last day of the study, while he Bergan, & Romano, in press). was seated at his desk in class, the subject There is another reason that understand- saw a spider near his leg, picked it up with ing emotion may be important for treat- a tissue, placed it in a wastebasket, and re- ment. Fears (anxieties) can spread and recede turned to his schoolwork. in idiosyncratic ways. That is, fear can gen- Although the boy’s successes were eralize to events that bear no formal similar- achieved with a conventional approach to ity to the originally feared event and can di- insect phobia, explaining them seems to re- minish in the presence of feared events that quire an appeal to direct and derived contin- were neither the focus of treatment nor for- gencies. As described earlier, respondent mally similar to events that were. Solving the functions (which are central to emotion) apparent puzzle requires understanding the spread readily through equivalence classes derived relations between events. For exam- and other forms of derived stimulus relations ple, the panic-disordered person’s fear of el- (Dougher et al., 1994; Roche & Barnes, 1997). In the study mentioned above, the evators or small spaces may generalize to his effects of extinction appeared to spread from or her marriage, even though the two situ- accessible insects (crickets) to less accessible ations share no formal properties. Yet both insects and related verbal taunts (neither of situations may occasion the verbal formula- which were the focus of treatment), suggest- tion, ‘‘I feel trapped,’’ resulting in the func- ing that all were members of a fear-based tions of ‘‘trapped’’ affecting the elevator and equivalence class. The spread of effects from the marriage in similar ways. insect to insect may be attributed persua- A recent study of a boy with insect phobia sively to direct contingencies because of for- partially illustrates these points (Jones & Fri- mal resemblance (Fields et al., 1991; Herrn- man, 1997). At the beginning of the study, stein, 1984; Medin & Smith, 1984; Rosch the boy would not enter a room if he was & Mervis, 1975). To account for the spread merely told that an insect was present with- of effects to classmates’ taunts, however, ver- in. Classmates could instigate extreme dis- bal analysis seems necessary. ruptive behavior in the boy by saying they had seen an insect (thus the word insect and ACAVEAT FOR BEHAVIOR ANALYSIS actual insects appeared to be in an equiva- The first part of this article elaborated on lence class and to have similar functions). the vagueness of the term anxiety and the 152 PATRICK C. FRIMAN et al. phenomena to which it refers. Two caution- avoid using the word anxiety in articles per- ary points need to made here, however. First, taining to the phenomena associated with problems with the term should not deter be- the term and may avoid listing the term in havior analysts from study of an area of psy- a subject index. This type of linguistic con- chology that is so vast and important. Sec- servatism has some clear disadvantages, an ond, the problems should also not inhibit important one of which is the difficulty that use of the term in common social parlance. potentially interested nonbehaviorists would The problems merely indicate that the term have in finding the research. Expanding the is not sufficiently precise to be considered a influence of behavior-analytic research be- technical term. yond the boundaries of the field requires Skinner advocated not just forbearance communications that cross disciplinary and but acceptance of ordinary ‘‘mentalized’’ ideological boundaries. These communica- terms in everyday parlance: ‘‘The words they tions will sometimes involve use of vague use are part of a living language that can be terms, such as anxiety, that appear to be used without embarrassment by cognitive valuable virtually everywhere but the labo- psychologists and behavior analysts alike in ratory. their daily lives. But these words cannot be There is a final reason that behavior an- used in their science!’’ (Skinner, 1989, p. alysts should treat these terms seriously, even 18). This distinction needs to be stressed. if they are imprecise or mentalistic. The Employing the term anxiety in everyday most distorted lay concept may refer to dis- nonscientific life can help persons to know tinctions that are important within a tech- each other better, predict what the other nical account. For example, the concept of might do in given situations, and advance intention may underline the importance of relationships. Denying such use seems to verbal temporal frames in behavioral regu- have no advantage; quite the contrary, denial lation (Hayes & Wilson, 1993); the matter- would almost certainly instigate a long list of disadvantages, with professional segrega- spirit distinction may be understood as an tion, frequent misunderstanding, and occa- important and reasonably accurate distinc- sional ridicule near the top of the list (cf. tion between verbal content and verbal con- Bailey, 1991; Deitz & Arrington, 1983; text (Hayes, 1984). There are numerous oth- Foxx, 1996). er examples. We can never know a priori if In addition, words that are technically im- a lay distinction or concept is nonsense until precise (e.g., anxiety, fear, embarrassment, a technical account is attempted. Eschewing shame, stress) and thus unsuitable for the lab- lay terms and the domains to which they oratory may have great value for application, refer merely because they are in conflict with just as some words that are technically pre- a scientific account can limit scientific prog- cise (e.g., negative reinforcement, punishment) ress. The scientific task is to understand the are exceptionally ill suited for the language source of control over lay terms and their of application (Lindsley, 1991). Despite its use—not their dictionary meaning and lit- apparent technical opacity, the term anxiety eral conflict or agreement with preferred does have well-established functional value terms. Understanding lay terms often re- for virtually all of the social-verbal commu- quires a detailed examination of the content nity except behavior analysts. Thus, at the to which they refer. Some terms may refer very least, the term has an important role in to nothing of scientific value, but others clinical settings. (e.g., emotion) may. In the latter case, an Behavior analysts might be tempted to adequate behavioral account can sometimes ANXIETY 153 be aided by the cultural wisdom that is em- agoraphobia in panic disorder patients. Behavior Research and Therapy, 35, 517–521. bodied in language traditions. Anderson, C. M., Hawkins, R. P., & Scotti, J. R. (1997). Private events in behavior analysis: Con- CONCLUDING REMARKS ceptual basis and clinical relevance. Behavior Ther- According to poets, political pundits, and apy, 28, 157–179. Bailey, J. S. (1991). Marketing behavior analysis re- philosophers, we seem to be living in an ‘‘age quires different talk. Journal of Applied Behavior of anxiety.’’ To date, behavior analysts have Analysis, 24, 445–448. largely avoided the study of anxiety (and Barlow, D. H. (1988). Anxiety and its disorders: Na- ture and treatment of anxiety and panic. New York: emotion in general), whereas mainstream Guilford Press. psychologists have devoted enormous re- Barnes, D. (1994). Stimulus equivalence and rela- sources to it (but have avoided behavior tional frame theory. The Psychological Record, 44, analysis). Admittedly, anxiety is a vague met- 91–124. Beck, A. T., & Emery, G. (1985). Anxiety disorders aphoric term for a diverse array of prevalent and phobias. New York: Basic Books. psychological phenomena for which a widely Biglan, A., & Hayes, S. C. (1995). Should the be- accepted technical definition is not available havioral sciences become more pragmatic? The case for functional contextualism in research on (and we have not offered one here). This human behavior. Applied and Preventive Psycholo- vagueness is the primary reason for the be- gy: Current Scientific Perspectives, 5, 47–57. havior-analytic avoidance of the topic. Im- Carson, R. C. (1996). Aristotle, Galileo, and the precision of a term, however, is not a suffi- DSM taxonomy: The case of schizophrenia. Jour- nal of Consulting and Clinical Psychology, 64, cient justification for such avoidance when 1133–1139. the phenomenon to which it refers is so vast Carson, R. C. (1997). Costly compromises: A cri- and so central to the psychology of human tique of the DSM. In S. Fisher & R. Greenberg (Eds.), Psychopharmacology scientifically reappraised beings. In addition, relinquishing anxiety to (pp. 98–114). New York: Wiley. mainstream psychology virtually guarantees Chambless, D. L., & Gracely, E. J. (1988). Prediction that proper analyses of functional relations of outcome following in vivo exposure treatment between environment and behavior, the me- of agoraphobia. In I. Hand & H. U. Wittchen (Eds.), Panic and phobias: Vol. 2. Treatment and tier of behavior analysts, will not be con- variables affecting outcome (pp. 209–220). New ducted. The literatures on the experimental York: Springer. analysis of verbal behavior and experiential Clark, D. M. (1986). A cognitive approach to panic. Behaviour Research and Therapy, 24, 461–470. avoidance provide numerous productive Craske, M. G., Street, L., & Barlow, D. H. (1989). leads for those who wish to pick up the trail. Instructions to focus upon or distract from inter- While remaining ever vigilant of the philo- nal cues during exposure treatment of agorapho- sophical dangers presented by study of bic avoidance. Behaviour Research and Therapy, 27, 663–672. vaguely defined phenomena, behavior ana- DeGrandpre, R. J., Bickel, W. K., & Higgins, S. T. lysts should end their avoidance and begin (1992). Emergent equivalence relations between their empirical and theoretical approach. interoceptive (drug) and exteroceptive (visual) stimuli. Journal of the Experimental Analysis of Be- havior, 58, 9–18. Deitz, S. M., & Arrington, R. L. (1983). Factors con- REFERENCES fusing language use in the analysis of behavior. American Psychiatric Association. (1987). Diagnostic Behaviorism, 11, 117–132. and statistical manual of mental disorders (3rd ed. Devany, J. M., Hayes, S. C., & Nelson, R. O. (1986). rev.). Washington, DC: Author. Equivalence class formation in language-able and American Psychiatric Association. (1994). Diagnostic language-disabled children. Journal of the Experi- and statistical manual of mental disorders (4th ed.). mental Analysis of Behavior, 46, 243–257. Washington, DC: Author. Dougher, M. J., Augustson, E., Markham, M. R., Amering, M., Katschnig, H., Berger, P., Windhaber, Greenway, D. E., & Wulfert, E. (1994). The J., Baischer, W., & Dantendorfer, K. (1997). Em- transfer of respondent eliciting and extinction barrassment about the first panic attack predicts functions through stimulus equivalence classes. 154 PATRICK C. FRIMAN et al.

Journal of the Experimental Analysis of Behavior, N. S. Jacobson, V. M. Follette, & M. J. Dougher 62, 331–351. (Eds.), Acceptance and change: Content and context Dymond, S., & Barnes, D. (1995). A transformation in psychotherapy (pp. 218–222). Reno, NV: Con- of self-discrimination response functions in accor- text Press. dance with the arbitrarily applicable relations of Grayson, J. B., Foa, E. B., & Steketee, G. (1982). sameness, more than, and less than. Journal of the Habituation during exposure treatment: Distrac- Experimental Analysis of Behavior, 64, 163–184. tion versus attention focusing. Behaviour Research Dymond, S., & Barnes, D. (1996). A transformation and Therapy, 20, 323–328. of self-discrimination response functions in accor- Hallam, R. S. (1985). Anxiety. London: Academic dance with the arbitrarily applicable relations of Press. sameness and opposition. The Psychological Record, Hayes, S. C. (1984). Making sense of spirituality. Be- 46, 271–300. haviorism, 12, 99–110. Ehlers, A., Margraf, J., Roth, W. T., Taylor, C. B., & Hayes, S. C. (1989). Nonhumans have not yet shown Birbaumer, N. (1988). Anxiety induced by false stimulus equivalence. Journal of the Experimental heart rate feedback in patients with panic disorder. Analysis of Behavior, 51, 385–392. Behavior Research and Therapy, 26, 1–11. Hayes, S. C., & Brownstein, A. J. (1986). Mentalism, Epstein, S. (1967). Toward a unified theory of de- behavior-behavior relations and a behavior analyt- pression. In B. A. Maher (Ed.), Progress in exper- ic view of the purposes of science. The Behavior imental personality research (Vol. 4, pp. 1–89). Analyst, 9, 175–190. New York: Academic Press. Hayes, S. C., Brownstein, A. J., Devany, J. M., Koh- Ferster, C., & Skinner, B. F. (1957). Schedules of re- lenberg, B. S., & Shelby, J. (1987). Stimulus inforcement. New York: Appleton-Century-Crofts. equivalence and the symbolic control of behavior. Fields, L., Reeve, K. E., Adams, B. J., & Verhave, T. Mexican Journal of Behavior Analysis, 13, 361– (1991). Stimulus generalization and equivalence 374. classes: A model for natural categories. Journal of Hayes, S. C., & Hayes, L. J. (1989). The verbal ac- the Experimental Analysis of Behavior, 55, 305– tion of the listener as a basis for rule-governance. 312. In S. C. Hayes (Ed.), Rule-governed behavior: Cog- Foa, E. B., & Kozak, M. S. (1986). Emotional pro- nition, contingencies, and instructional control (pp. cessing of fear: Exposure to corrective informa- 153–190). New York: Plenum. tion. Psychological Bulletin, 99, 20–35. Hayes, S. C., & Hayes, L. J. (1992). Verbal relations Foa, E. B., & Meadows, E. A. (1997). Psychosocial and the evolution of behavior analysis. American treatments for posttraumatic stress disorder: A Psychologist, 47, 1383–1395. critical review. Annual Review of Psychology, 48, Hayes, S. C., Kohlenberg, B. K., & Hayes, L. J. 449–480. (1991). Transfer of consequential functions Foa, E. B., & Riggs, D. S. (1995). Posttraumatic through simple and conditional equivalence class- stress disorder following assault: Theoretical con- es. Journal of the Experimental Analysis of Behavior, siderations and empirical findings. Current Direc- 56, 119–137. tions in Psychological Science, 4, 61–65. Hayes, S. C., Strosahl, K., & Wilson, K. G. (in press). Follette, W. C., Houts, A. C., & Hayes, S. C. (1992). Acceptance and commitment therapy: Understanding Behavior therapy and the new medical model. Be- and treating human suffering. New York: Guilford havioral Assessment, 14, 323–343. Press. Forsyth, J. P., & Eifert, G. H. (1996a). The language Hayes, S. C., & Wilson, K. G. (1993). Some applied of feeling and the feeling of anxiety: Contribu- implications of a contemporary behavior-analytic tions of the behaviorisms toward understanding account of verbal events. The Behavior Analyst, 16, the function altering effects of language. The Psy- 283–301. chological Record, 46, 607–649. Hayes, S. C., & Wilson, K. G. (1994). Acceptance Forsyth, J. P., & Eifert, G. H. (1996b). Systematic and commitment therapy: Undermining the ver- alarms in fear conditioning I: A reappraisal of bal support for experiential avoidance. The Behav- what is being conditioned. Behavior Therapy, 27, ior Analyst, 17, 289–303. 441–462. Hayes, S. C., & Wilson, K. G. (1996). Criticisms of Foxx, R. M. (1996). Translating the covenant: The relational frame theory: Implications for a behav- behavior analyst as ambassador and translator. The ior analytic account of derived stimulus relations. Behavior Analyst, 19, 147–162. The Psychological Record, 46, 221–236. Freud, S. (1966). Anxiety (lecture XXV). In J. Stra- Hayes, S. C., Wilson, K. G., Gifford, E. V., Follette, chey (Ed.), The complete introductory lectures on V. M., & Strosahl, K. (1996). Experiential avoid- psychoanalysis (pp. 392–411). New York: Norton. ance and behavior disorders: A functional dimen- (Original work published 1917) sional approach to diagnosis and treatment. Jour- Gifford, E. V. (1994). Setting a course for change: nal of Consulting and Clinical Psychology, 64, The verbal context of acceptance. In S. C. Hayes, 1152–1168. ANXIETY 155

Herrnstein, R. J. (1984). Objects, categories, and dis- cortisol. Journal of the American Academy of Child criminative stimuli. In H. L. Roitblatt, T. G. Bev- and Adolescent Psychiatry, 30, 192–196. er, & H. S. Terrace (Eds.), Animal cognition (pp. McCarthy, P. R., & Foa, E. B. (1990). Obsessive- 233–261). Hillsdale, NJ: Erlbaum. compulsive disorder. In M. E. Thase, B. A. Edel- Hoes, M. J. A. J. M. (1986). Biological markers in stein, & M. Hersen (Eds.), Handbook of outpatient psychiatry. Acta Psychiatrica Belgica, 86, 220–241. treatment of adults: Nonpsychotic mental disorders Hollander, E. (1993). Obsessive-compulsive related dis- (pp. 209–234). New York: Plenum. orders. Washington, DC: American Psychiatric Medin, D. L., & Smith, E. E. (1984). Concepts and Press. concept formation. Annual Reviews of Psychology, Horne, P. J., & Lowe, C. F. (1996). On the origins 35, 113–138. of naming and other symbolic behavior. Journal Oxford University Press. (1971). The compact edition of the Experimental Analysis of Behavior, 65, 185– of the Oxford English dictionary.NewYork:Au- 242. thor. Jones, K. M., & Friman, P. C. (1997). Functional Pauli, P., Marquardt, C., Hartl, L., Nutzinger, D. O., analysis and treatment of entomophobia in a boy in Holzl, R., & Strain, F. (1991). Anxiety induced residential care: A preliminary case report. Paper by cardiac perceptions in patients with panic at- submitted for publication. tacks: A field study. Behaviour Research and Ther- Keedwell, P., & Snaith, R. P. (1996). What do anxiety apy, 29, 137–145. scales measure? Acta Psychiatrica Scandinavica, 93, Pavlov, I. P. (1927). Conditioned reflexes. Oxford: Ox- 177–180. ford University Press. Keijsers, G. P. J., Hoogduin, C. A. L., & Schaap, C. Peterson, C., & Seligman, M. (1984). Causal expla- P. (1994). Prognostic factors in the treatment of nation as a risk factor for depression: Theory and panic disorder with and without agoraphobia. Be- evidence. Psychological Review, 91, 347–374. havior Research and Therapy, 25, 689–708. Poling, A., Methot, L. L., & LeSage, M. G. (1995). Fundamentals of behavior analytic research.New Kohlenberg, R. J., Hayes, S. C., & Tsai, M. (1993). York: Plenum. Radical behavioral psychotherapy: Two contem- Rachman, S. (1977). The conditioning theory of fear porary examples. Clinical Psychology Review, 13, acquisition: A critical examination. Behavior Re- 579–592. search and Therapy, 15, 375–387. Kohlenberg, R. J., & Tsai, M. (1991). Functional an- Rachman, S. (1991). Neo-conditioning and the clas- alytic psychotherapy. New York: Plenum. sical theory of fear acquisition. Clinical Psychology Lattal, K. A. (1975). Reinforcement contingencies as Review, 11, 155–173. discriminative stimuli. Journal of the Experimental Rapee, R. M., Brown, T. A., Antony, M. M., & Bar- Analysis of Behavior, 23, 241–246. low, D. H. (1992). Response to hyperventilation Leigland, S. (1997). Is a new definition of verbal be- and inhalation of 5.5% carbon dioxide-enriched havior necessary in light of derived relational re- air across the DSM-III-R anxiety disorders. Jour- sponding? The Behavior Analyst, 20, 3–10. nal of Abnormal Psychology, 101, 538–552. Levitt, E. E. (1967). The psychology of anxiety. Indi- Rasmussen, S. A., & Eisen, J. L. (1992). The epi- anapolis: Bobbs-Merrill. demiology and differential diagnosis of obsessive Lindsley, O. R. (1991). From technical jargon to compulsive disorder. Journal of Clinical Psychiatry, plain English for application. Journal of Applied 53, 4–10. Behavior Analysis, 24, 449–458. Roche, B., & Barnes, D. (1997). A transformation of Lipkens, G., Hayes, S. C., & Hayes, L. (1993). Lon- respondently conditioned stimulus function in ac- gitudinal study of derived stimulus relations in an cordance with arbitrarily applicable relations. Jour- infant. Journal of Experimental Child Psychology, nal of the Experimental Analysis of Behavior, 67, 56, 201–239. 275–301. Margraf, J., Ehlers, A., & Roth, W. (1986). Sodium Rosch, E., & Mervis, C. B. (1975). Family resem- lactate infusions and panic attacks: A review and blance: Studies in the internal structure of cate- critique. Psychosomatic Medicine, 48, 23–51. gories. Cognitive Psychology, 7, 573–605. Margraf, J., Ehlers, A., & Roth, W. (1988). Mitral Ryle, G. (1949). The concept of mind. Chicago: Uni- valve prolapse and panic disorder: A review of versity of Chicago Press. their relationship. Psychosomatic Medicine, 50, 93– Sarbin, T. R. (1964). Anxiety: The reification of a 113. metaphor. Archives of General Psychiatry, 10, 630– May, R. (1950). The meaning of anxiety.NewYork: 638. The Ronald Press. Sarbin, T. R. (1968). Ontology recapitulates philol- McBurnett, K., Lahey, B. B., Frick, P. J., Risch, C., ogy: The mythic nature of anxiety. American Psy- Loeber, R., Hart, E. L., Christ, M. A. G., & Han- chologist, 23, 411–418. son, K. S. (1991). Anxiety, inhibition, and con- Sidman, M. (1960). Normal sources of pathological duct disorder in children: II. Relation to salivary behavior. Science, 132, 61–68. 156 PATRICK C. FRIMAN et al.

Sidman, M. (1971). Reading and auditory-visual alence and arbitrarily applicable relational re- equivalences. Journal of Speech and Hearing Re- sponding. Journal of the Experimental Analysis of search, 14, 5–13. Behavior, 56, 519–555. Sidman, M. (1994). Equivalence relations: A research Strachey, J. (Ed.). (1966). The complete introductory story. Boston: Authors Cooperative. lectures on psychoanalysis. New York: Norton. Sidman, M., & Tailby, W. (1982). Conditional dis- Strosahl, K. D., Hayes, S. C., Bergan, J., & Romano, crimination vs. matching to sample: An expansion P. (in press). Does field based training in behavior of the testing paradigm. Journal of the Experimen- therapy improve clinical effectiveness? Evidence tal Analysis of Behavior, 37, 5–22. from the acceptance and commitment therapy Skeat, W. R. (1963). A concise etymological dictionary training project. Behavior Therapy. of the English language. New York: Capricorn Taylor, S. (1995). Anxiety sensitivity: Theoretical per- Books. spectives and recent findings. Behavior Research Skinner, B. F. (1938). The behavior of organisms.New and Therapy, 33, 243–258. York: Appleton-Century-Crofts. Wickler, A. (1973). Conditioning of successive adap- Skinner, B. F. (1945). The operational analysis of psy- tive response to the initial effects of drugs. Con- chological terms. Psychological Review, 52, 270– ditioned Reflex, 8, 193–210. 277. Wittgenstein, L. (1958a). The blue and brown books. Skinner, B. F. (1953). Science and human behavior. New York: Harper & Row. New York: Free Press. Wittgenstein, L. (1958b). Philosophical investigations. Skinner, B. F. (1957). Verbal behavior.NewYork:Ap- New York: MacMillan. pleton-Century-Crofts. Wolpe, J. (1990). The practice of behavior therapy. Skinner, B. F. (1969). Contingencies of reinforcement: New York: Pergamon. A theoretical analysis. New York: Appleton-Cen- Wulfert, E., Greenway, D. E., Farkas, P., Hayes, S. C., tury-Crofts. & Dougher, M. J. (1994). Correlation between Skinner, B. F. (1974). About behaviorism. New York: a personality test for rigidity and rule-governed Knopf. insensitivity to operant contingencies. Journal of Skinner, B. F. (1989). The origins of cognitive Applied Behavior Analysis, 27, 659–671. thought. American Psychologist, 44, 13–18. Yalom, I. D. (1980). Existential psychotherapy. New Sofuoglu, S., Dogan, P., Besim, T., Basturk, M., & York: Basic Books. Tanrikulu, G. (1992). MAO-B activity in de- Received October 20, 1997 pression and anxiety disorders. Biological Psychia- Initial editorial decision November 6, 1997 try, 31, 634–638. Final acceptance November 21, 1997 Steele, D. L., & Hayes, S. C. (1991). Stimulus equiv- Action Editor, David P. Wacker