friman, hayes and wilson - why behavior analysts should study emotion-the example of anxiety

8/11/2019 Friman, Hayes and Wilson - Why Behavior Analysts Should Study Emotion-The Example of Anxiety

1/20

137

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1998, 31,137156 NUMBER1 (SPRING 1998)

WHY BEHAVIOR ANALYSTS SHOULD STUDYEMOTION: THE EXAMPLE OF ANXIETY

PATRICKC. FRIMANFATHER FLANAGANS BOYS HOME AND CREIGHTON UNIVERSITY SCHOOL OF MEDICINE

AND

STEVEN C. HAYES AND KELLYG. WILSONUNIVERSITY 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 thisdiscrepancy. 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 thatit is a simple matter to show that neutral stimuli can acquire discriminative functionsindirectly with no direct training; (b) that private events can readily acquire discriminativefunctions; (c) that anxiety disorders seem to occur with little apparent direct learning orthat the amount of direct learning is extraordinarily out of proportion with the amountof responding; and (d) that the primary function of anxious behavior is experientialavoidance. We conclude that the most interesting aspects of anxiety disorders may occuras a function of derived rather than direct relations between public events and overt andprivate responses with avoidance functions. Implicit in this conclusion and explicit in thepaper is the assertion that anxiety is a suitable subject for behavior-analytic study.

DESCRIPTORS: anxiety, emotion, avoidance, stimulus equivalence, relational frametheory

Anxiety is a topic of primary importancein most approaches to psychopathology.

Anxiety co-occurs so prevalently with psy-

chological distress that it, along with de-pression, has been described as the psycho-logical equivalent of fever (Carson, 1997).Major theoretical positions ranging fromFreudian psychoanalysis (May, 1950; Stra-chey, 1966) to existentialism (Yalom, 1980)describe anxiety as a central feature (some-times the central feature) of behavioral dis-

This research was supported by a grant to the first au-thor from the Carmel Hill Foundation and a grant to thesecond author from the National Institute on Drug Abuse

(DA08634-02). Special thanks are extended to Keith Al-len, whose questions led to this project, and to Joe Sprad-lin, Simon Dymond, Ray Miltenberger, Mike Handwerk,Jack Finney, Winifred Ju, Michael Rapoff, and Bill Cookfor comments on an earlier draft of the paper.

Address correspondence and requests for reprints to Pat-rick C. Friman, Youthcare Building, Boys Town, Nebraska68010 (E-mail: [email protected]).

orders. There are several peer-reviewed jour-nals devoted exclusively to the topic (e.g.,

Anxiety, Journal of Anxiety Disorders, Anxiety,Stress, & Coping). TheDiagnostic and Statis-tical Manual of Mental Disorders (4th ed.,DSM-IV; American Psychiatric Association,1994) lists 12 distinct anxiety diagnoses. Asearch for the term using the electronicpsychFirst database produced 6,425 refer-enced articles from 1993 to 1997. In con-trast, contemporary behavior analysts havepublished very little on anxiety, or indeed onthe scientific study of emotion in general. Inthis paper we discuss reasons for this reluc-tance, describe recent developments in theexperimental analysis of human behaviorthat make behavior-analytic study of anxiety(and emotion in general) more tenable, andbriefly examine some applied implications ofthese recent developments. The paper willconclude with a caveat pertaining to the


2/20

138 PATRICK C. FRIMAN et al.

termanxiety and a recommendation for in-creased research on the conditions that oc-casion its use.

REASONS FORLIMITEDBEHAVIOR-ANALYTIC RESEARCH

As a research focus, anxiety is perilous forbehavior analysts who wish to maintain fi-delity to their view of science. For example,even Freud thought the term was imprecise(Freud, 1917/1966). Precision refers to thenumber of ways that a specific phenomenonmay be construed verbally within a technicalanalysis; as the number goes down, precisiongoes up and vice versa (Biglan & Hayes,1995). Precise definition of terms is differ-

entially valued across various approaches toscience; it is fundamental to behavior anal-ysis (Poling, Methot, & LeSage, 1995). De-spite the enormous intellectual energy de-voted to anxiety since Freud, the number ofverbal constructions organized under theterm seems to have grown larger and moreambiguous rather than smaller and moreprecise (cf. Hallam, 1985; Keedwell &Snaith, 1996; Levitt, 1967).

One of the obstacles to precision is themetaphorical, idiomatic basis of the term

anxiety. Similar to other emotional terms,anxiety was inaugurated as an as if de-scriptive idiom or metaphor (e.g., it is as ifthere is stuff that controls thinking, Sarbin,1968, p. 413). Among its primary predeces-sors wasanguissefrom the Old French whichbecameanguishin Middle English andanx-iety in modern English (Sarbin, 1964,1968). Anguisse referred to choking sensa-tions in the throat, which, in its subsequentform, anxiety, was helpful in communicatingvague or excessive feelings associated withaversive events (Oxford University Press,1971, p. 95; Sarbin, 1964, 1968; Skeat,1963). The absurdity of the transition fromanxiety as metaphor to its current masquer-ade as technical term is revealed by posingthe possibility of a similar transition for oth-

er metaphors with related usages such asbutterflies in the stomach. The possibility ofbutterfly disorders, butterfly managementtechniques, butterfly scales, and debates on

the causal role of butterflies seems remote.Ano ther obstacle to behavi or-anal yti cstudy of anxiety is the illusion of its statusas an empirical phenomenon and the realityof its status as a mere concept or category(see Ryle, 1949). Categories are not empir-ical events, that is, they cannot be observeddirectly, and thus they are unappealing asresearch targets for behavior analysts. Yetconditions that govern category membershipare empirical events, and these have gener-ated abundant behavior-analytic study (e.g.,

Fields, Reeve, Adams, & Verhave, 1991;Herrnstein, 1984) that has profitably con-verged with research from other fields (e.g.,Medin & Smith, 1984; Rosch & Mervis,1975). A seemingly suitable topic for behav-ior-analytic study is thus the conditions thatoccasion anxiety as a categorical response.

An apparent obstacle to this potentially pro-ductive line of inquiry is the conventionalbut false assumption that anxiety, as a cate-gory, has an essential composition.

This essentialistic position on anxiety is

readily undermined with logical arguments(Wittgenstein, 1958a, 1958b). For example,a common definition of anxiety is physio-logical reactivity to events with uncertainbut potentially aversive outcomes. Walkingacross a street that is free of cars and sud-denly being overwhelmed by high-intensityphysiological responding (e.g., elevated heartrate, respiration, perspiration, and bloodpressure) is a well-accepted instance of anx-iety (e.g., agoraphobia; American Psychiatric

Association, 1994; Beck & Emery, 1985;Hallam, 1985). Walking across the samestreet and almost being hit by a car producesthe same physiology but is not an instanceof anxiety. It is an instance of fear. Both arerelated by physiology and avoidance and areseparated by the actuality of the threat (Beck


3/20

139ANXIETY

& Emery, 1985; Epstein, 1967; May, 1950).Being in the presence of a formerly reinforc-ing commodity after an extended period ofdeprivation (e.g., water, addictive substances,

sex) also produces much the same physiol-ogy but is not an instance of anxiety or fear.It is an instance of craving (Wickler, 1973;see also Pavlov, 1927, pp. 3537). It has nei-ther the uncertainty nor the avoidance. Thusphysiology cannot be the essence of anxiety,because it is part of many kinds of avoidance(e.g., of events with known and unknownthreat) and of many kinds of pursuit (e.g.,for sex, drugs, fluids, food). Similar cases arereadily made for the other elements said tocompose anxiety. For example, demonstrably

dangerous events can evoke the physiologyof fear but no avoidance and little uncer-tainty (e.g., contact sports, skydiving). Neu-tral events that are conditioned to evoke thephysiology of fear and avoidance (e.g., pho-bias) sometimes produce neither (Rachman,1977, 1991). Some events are avoided or es-caped (i.e., avoidance or escape responses arereinforced) but do not produce the physi-ology of fear, nor are they associated withdanger (e.g., alarm clocks, obnoxious per-sons).

Not surprisingly, given the persuasive log-ical counterarguments, the essentialistic po-sition on anxiety has generated little empir-ical support. The gold standard of psychi-atric diagnostic research is essentialistic; itinvolves identifying sensitive and specific bi-ological markers for psychiatric disorders(Hoes, 1986). Yet the abundant research onnumerous nominated markers, accompaniedby experimental preparations appropriate foranxiety (e.g., salivary cortisol, lactate infu-sion, carbon dioxide inhalation, mitral valveprolapse, monoamine oxidase levels), has yetto yield a sensitive and specific (i.e., essen-tial) marker (e.g., Margraf, Ehlers, & Roth,1986, 1988; McBurnett et al., 1991; Rapee,Brown, Antony, & Barlow, 1992; Sofuoglu,Dogan, Besim, Basturk, & Tanrikulu,

1992). In sum, anxiety appears to have noessential or nonreducible component, thepresence of which distinguishes participantfrom nonparticipant events. This conclusion

seems to set the stage for a behavioral anal-ysis of the various usages of the termanxiety.One more obstacle remains, however, and itis a theoretical one within behavior analysisitself: Skinners analysis of emotion.

Skinners analysis provided an opening forbehavior-analytic study of emotion and si-multaneously made stepping through itseem unnecessary. Consistent with our com-ments above, he criticized traditional waysof speaking about emotion. In fact, Skinnerscareer-long opposition to mentalism in sci-

ence could have employed anxiety as a text-book case. Mentalism commonly involvesexplaining behavior by appealing to inde-pendent variables that are inferred from thebehavior explained (Hayes & Brownstein,1986; Skinner, 1969; see also Wittgenstein,1958a, 1958b). For example, in their influ-ential book on anxiety and phobia, Beck andEmery (1985) state that anxiety disorders arecaused by an upset in the cognitive system(p. 86). As a case in point, they describe themost frequently occurring features of gen-

eralized anxiety disorder (GAD), some ofwhich involve upsets in the cognitive system(e.g., difficulty in concentrating, fear of los-ing control, etc.). The frequency of thesefeatures is then adduced as evidence of thecausal status of cognitive impairment. Yet ifthese cognitive upsets are an intrinsic part ofGAD, using them to explain it is circular,misleading, and mentalistic because the up-sets themselves remain unexplained.

Skinner went further than mere criticismof mentalistic use of terms such as anxiety,however; he also offered a coherent behav-ior-analytic alternative. In his landmark pa-per on the operational analysis of psycholog-ical terms, Skinner laid the philosophicaland theoretical foundation for a behavioralanalysis of emotion (Skinner, 1945). Con-


4/20


trary to caricatures of behaviorism that sug-gest a denial of private events such as emo-tion and cognition, radical behaviorismdoes not insist upon truth by agreement

and can therefore consider events takingplace in the private world within the skin(Skinner, 1974, p. 16). In contrast to manyemotion theorists, Skinner downplayed thescientific value of research on the form ofemotional responses and recommendedstudy of the contingencies that govern emo-tional talk instead (Skinner, 1945). To Skin-ner, the meaning of a term (e.g., anxiety)resides in the functional relationship be-tween its use and stimuli that are antecedentand consequent to the use. In other words,

understanding the meaning of the state-ment, Im anxious, requires knowledge ofthe context, both current and historical, thatoccasioned the utterance. Skinners argu-ments were elegant, plausible, and directlypertinent to emotion, but they were derivedentirely from an analysis of direct contingen-cies.

Although Skinners approach provided aneeded opening for scientific analysis ofemotion (and private events in general), hisemphasis on direct contingencies limited the

potential scope of research programs whenthey were conducted with fidelity to hisviews. Such programs would be simulta-neously viable and valid, yet not very pro-ductive. Understanding how this conclusion

was reached is an important step towards amore contemporary and potentially impor-tant approach to behavior-analytic study ofemotion.

The two fundamental goals of scientificexplanation in behavior analysis are predic-tion and control of the phenomenon understudy (Biglan & Hayes, 1995; Hayes &Brownstein, 1986; Skinner, 1969). To thelayman (and many, possibly most, psychol-ogists outside behavior analysis), emotionaltalk itself is satisfactory as an explanation forbehavioral events involving emotion. For ex-

ample, if a crying child complains that he orshe feels anxious and then runs away, mostparents (and other persons including psy-chologists) are likely to attribute the running

to anxious feelings (i.e., the child ranbecausehe or she felt anxious). This explanation,however, is predicated on a behaviorbehav-ior relationship and is thus incomplete. Feel-ing anxious is a behavioral event that cannotbe directly manipulated in a way that estab-lishes a functional relationship between itand running. Thus, although the role of feel-ing anxious as predictor can be readily es-tablished, its role as controlling variable can-not. To be complete in behavior analysis, anexplanation must ultimately appeal to ma-

nipulable environmental events.Skinner develops this logic using an ex-ample of a situation in which an electricshock has been reliably preceded by a bell.The bell comes to elicit reactions, primarilyin the autonomic nervous system, which arefelt as anxiety (Skinner, 1974, p. 61). Theindividuals behavior does not change be-cause he feels anxious; it changes because ofthe aversive contingencies which generatethe condition felt as anxiety. The change infeeling and the change in behavior have a

common cause (Skinner, 1974, pp. 6162).In this and other instances, Skinner arguesthat emotional feelings are co-occurringproducts of the same contingencies that pre-cipitate overt behaviors which the feelingsare said to cause (e.g., running) and thatthey have no functional significance, eitherin a theoretical analysis or the practical con-trol of behavior (Skinner, 1953, p. 181).Thus, although Skinner did establish the sci-entific plausibility of analyzing privateevents, he also suggested that such analyses

were theoretically and practically unneces-sary.

WHYEMOTION IS AVIABLEBEHAVIOR-ANALYTIC TOPIC NOW

The central theme of the previous sectionwas that anxiety is still not well understood.


5/20

141ANXIETY

Perhaps the rigor of behavior-analytic theoryand methods is needed for increased under-standing of this seemingly ubiquitous, yetmisunderstood, phenomenon. Although the

initial promise of Skinners radical approachto human behavior was a complete scienceof private and public events, study of theformer has lagged far behind study of thelatter. Now, however, the field of behavioranalysis is changing, and, correspondingly,the possibility of experimental analysis ofprivate events is increasing. The dangers forscience, discussed in the previous section re-main relatively intact. But the intellectualbarrier created by Skinners position is grad-ually giving way to empirical developments

in the analysis of human verbal behaviorthat, in turn, expand the possibilities for be-havior-analytic study of emotion.

Skinners analyses of schedules of rein-forcement benefitted from a wealth of ex-perimental evidence (e.g., Ferster & Skinner,1957; Skinner, 1938). In striking contrast,his analyses of verbal behavior were virtuallydevoid of experimental evidence (Skinner,1957) and thus did not benefit from the se-lective effects that data have on theorizing.In recent years, a large body of empirical

evidence from the experimental analysis ofhuman behavior has been published. Of par-ticular relevance to the analysis of emotionis the literature on derived relational re-sponding such as stimulus equivalence andthe transformation of stimulus functions(e.g., Barnes, 1994; Hayes & Hayes, 1989,1992; Hayes & Wilson, 1993; Sidman,1994; see also Horne & Lowe, 1996). Thisrapidly growing line of investigation mayfundamentally change the behavior-analyticview of verbal events and, with it, the anal-ysis of private events.

The research shows that humans readilylearn derived stimulus relations. If a lan-guage-able human, while in the presence ofone stimulus, A, learns to select an arbitrar-ily related stimulus, B, then this trained,

unidirectional relation will lead to a derived,bidirectional relation. That is, given B, A

will be chosen, without additional directtraining, even in children as young as 16

months (Lipkens, Hayes, & Hayes, 1993).By the age of 2 years, children learn evenmore remotely derived stimulusstimulus re-lations. For example, if A-B and A-C rela-tions are trained, then B-C and C-B rela-tions are derived (Devany, Hayes, & Nelson,1986; Lipkens et al., 1993). These relationshave been termed stimulus equivalence(Sid-man, 1971, 1994; Sidman & Tailby, 1982).

Recent research shows that many relationsother than equivalence (e.g., greater than,less than, opposition, difference, etc.) can belearned, applied arbitrarily to stimulusevents, and combined into networks of stim-ulus relations of almost unimaginable com-plexity (Dymond & Barnes, 1995, 1996;Steele & Hayes, 1991). In addition, otherprocesses, such as stimulus generalization,can combine with derived relational re-sponding to merge large and diverse classesof responding (Fields et al., 1991). Further-more, the psychological functions of the el-ements in these relational networks tend

to change or transform in accord with theunderlying derived stimulus relation. In thecase of equivalence relations, several dem-onstrations of this transformation are avail-able, and they include conditioned reinforc-ing functions (Hayes, Brownstein, Devany,Kohlenberg, & Shelby, 1987; Hayes, Koh-lenberg, & Hayes, 1991), discriminativefunctions of public (Hayes et al., 1987) andprivate (DeGrandpre, Bickel, & Higgins,1992) stimuli, elicited conditioned emotion-al responses (Dougher, Augustson, Mark-ham, Greenway, & Wulfert, 1994), extinc-tion functions (Dougher et al., 1994), andsexual responses (Roche & Barnes, 1997). Inthe case of relations other than equivalence(e.g., opposition), demonstrations of corre-sponding transformed stimulus functions are


6/20


also increasingly available (Dymond &Barnes, 1995; Roche & Barnes, 1997).

Studies documenting the transformationof stimulus functions propel a more contem-

porary behavior-analytic account of verbalbehavior and approach to emotion and otherprivate events. In addition, new behavioralterms might be needed (Hayes & Hayes,1992). For example, if a child is taught stim-ulus relations between three stimuli, A, B,and C, that bear no formal similarity to eachother, and then B is given a discriminativefunction through direct training (e.g., rein-forcement for waving in the presence of B),the function will obtain for C (i.e., wavingis now likely in the presence of C) without

direct training (Hayes et al., 1987). In thisexample, B is a discriminative stimulus butC is not, because a greater probability of re-inforcement for waving in its presence thanin its absence has not been established. Wav-ing in the presence of C is also not an in-stance of stimulus generalization, becausethere are no formal stimulus properties toaccount for the generalization. Nonetheless,C functions like a discriminative stimulus,but it is produced indirectly by the transfor-mation of stimulus functions through de-

rived stimulus relations, not by a history ofdirect contingencies (as in the conventionalSkinnerian account). Recent papers havesuggested that such functions are not justverbal; they may be a defining property ofverbal behavior (e.g., Barnes, 1994; Hayes &Hayes, 1992; Hayes & Wilson, 1994, 1996;Roche & Barnes, 1997). Thus C, in the pre-ceding example, might profitably be referredto as averbaldiscriminative stimulus (Hayes& Wilson, 1993). Although logically consis-tent, the argument for a modified nomen-clature has yet to recruit a consensus amongbehavior analysts (e.g., Leigland, 1997; J.Spradlin, personal communication, October23, 1997).

The comments above suggest the breadththat derived relational responding adds to

behavioral accounts of complex behavior, es-pecially when private events are involved. Asynthesis of just a few of the many relatedlines of research will underscore this point

and set the stage for application of derivedrelational concepts to anxiety disorders. Alarge, long-standing, cross-disciplinary lineof research shows that generalized respond-ing to stimuli with discriminative function(discriminanda) spreads to novel stimuli thatresemble the discriminanda (via stimulusgeneralization), resulting in large relational,polymorphous, or fuzzy categories of re-sponses (Fields et al., 1991; Herrnstein,1984; Medin & Smith, 1984; Rosch &Mervis, 1975; see also Wittgenstein, 1958b).

A more recent behavior-analytic line of re-search shows that events in relational classesspread not just with stimulus equivalence ef-fects but also with these stimulus generaliza-tion effects. For example, once StimulusEvents A and B are entered into a relationalclass, the events related discretely to A andto B via stimulus generalization can becomerelated to each other via stimulus equiva-lence (Fields et al., 1991). Thus, large rela-tional, fuzzy, or polymorphous categoriescreated via stimulus generalization can

merge with other large relational, fuzzy, orpolymorphous categories if a member fromone is entered into an equivalence relation

with a member from another. Another re-cent line of behavior-analytic research showsthat private and public events can becomepart of the same relational class (De-Grandpre et al., 1992). In other words, theselarge relational categories can (and usuallydo) include both public and private events.Lastly, a rapidly expanding line of researchshows that respondent (e.g., Dougher et al.,1994; Roche & Barnes, 1997) and other be-havioral functions (e.g., Hayes et al., 1987,1991) participate in relational networks andtransform the behavioral functions of theirconstituent responses. Collectively, theselines of research provide a preliminary basis


7/20

143ANXIETY

for penetrating the complexity of anxietydisorders.

Anxiety disorders seem irrational becausethey primarily involve high-rate avoidance

of, or escape from, events that have no readi-ly detectable direct relationship with punish-ment (e.g., open spaces, doorknobs, harm-less insects). Analysis of anxiety-like re-sponses in animals suggests a combinationof historically real and currently adventitiouscontingencies (Sidman, 1960, p. 66). Thus,even with a direct contingency analysis ofpertinent behaviors in organisms with muchsimpler response patterns than humans, anx-iety responses are puzzling (i.e., seem irra-tional). In the analysis of human anxiety re-sponses, the puzzles multiply in number andcomplexity because human response patternsare also strongly influenced by indirectrela-tionships between events and public and pri-vate responses to public and private events.

Of course, not all avoidance responding ispuzzling. Responses that are reinforced byavoidance of events with phylogenetic sig-nificance (e.g., fire, large fur-covered carni-vores, heights) readily surrender to analysis.So too do responses that are reinforced by

avoidance of seemingly nonsignificant eventswhen those events can be linked formally orfunctionally to aversive consequences. Butclinically significant anxiety appears to referto avoidance responses whose initiating con-ditions are direct but very remote and whoseperpetuating conditions are mostly derived.The life of the clinically anxious person maythus be influenced by iterations and reiter-ations of public and private events with re-active properties traceable to initiating con-ditions only through an almost fractal path-

way involving the processes of stimulus gen-eralization, derived relational responding,and transformation of stimulus function (cf.Hayes & Wilson, 1993, 1994; see also An-derson, Hawkins, & Scotti, 1997). It seemssafe to say that a true behavior analysis of

the concept of anxiety will require an ex-traordinary research effort.

Implications for the Analysis of Anxiety

A major implication of derived relationalresponding involves the potentially powerfuleffect it can have on self-discriminations: Itcan render them reactive (Dymond &Barnes, 1995, 1996). A related issue is thatthese effects are unlikely when nonverbal or-ganisms self-discriminate because they arenot nearly as prepotent for derived respond-ing as humans are. In fact, whether animalscan exhibit derived responding at all is acontested issue (Devany et al., 1986; Hayes,1989; Horne & Lowe, 1996). Even if re-

search ultimately unequivocally documentsderived relational responding in some non-humans, the argument here remains intact.That is, behavior resulting from derived re-lations requires an analysis that is somewhatdifferent from that offered by Skinner. Thus,

when anxiety-like responses (e.g., avoidance)in any organism (human or nonhuman) aretraceable to direct contingencies, Skinnersanalysis of emotion seems to be sufficient.That is, when only direct contingencies areinvolved, it follows that the change in feel-

ing and the change in behavior have a com-mon cause (Skinner, 1974, p. 62; see alsoSidman, 1960). When the anxiety-like re-sponses cannot be traced to direct contin-gencies and the organism is prepotent forderived relational responding (e.g., language-able persons), additional analyses are re-quired.

Consider two situations: In one, a hungryrat can obtain food immediately but the de-livery is accompanied by a small shock; inthe other, a small delay is required for thefood but no shock occurs. If the values areset properly, the rat will consume the foodimmediately (i.e., impulsively) and will beshocked. If the rat is trained to press onelever for food if it has been shocked andanother if it has not been shocked, it, in


8/20


effect, will have learned to report whether ithas been shocked (e.g., Lattal, 1975). Thedifference between a rat and a language-ablehuman is that the rats report (or self-dis-

crimination) is unlikely to have an effect onsubsequent similar (i.e., impulsive) respond-ing. That is, the report is unlikely to be bi-directionally related to the event being re-ported; for rats and most (perhaps all) non-human organisms, an eventreport relationdoes not entail a reportevent relation. Ab-sent this bidirectionality, there is no knownpathway for the report to influence subse-quent events similar to those reported. Sim-ilarly, there is no known pathway (absent ad-ditional experimental preparations) for the

rats report of shock to become aversive it-self. The report would not be bidirectionallyrelated to the event of shock, and thus it

would not share its functions. Functionally,the report is related to food, not to shock.Reports of punishing events are simply notaversive for organisms that are not prepotentfor derived relational responding.

As indicated above, the responses of lan-guage-able persons in similar circumstancesare different than those of simpler organismssuch as rats. For the human, the shock and

the report are likely to be bidirectionally re-lated and thus share functions; some of theeffects of the shock are likely to occur whenthe shock event is reported. In addition, dis-tinct from the simpler organisms, a personsreport of the behavior that led to the shockis likely to affect the probability of beingshocked in similar circumstances, becauseverbal (i.e., derived) reports are bidirectional(eventreport entails reportevent). Thus, apersons self-discriminations (or self-knowl-edge) can affect his or her behavior. To ap-preciate this point experientially, imaginebiting into a lemon. Some of the perceptualfunctions of biting occur merely as a func-tion of the thought, that is, the thoughts arereactive. Similar to thoughts about lemons,self-referential thoughts can be reactive. For

example, mere thoughts about engaging inpublic speaking can instigate autonomicarousal and self-limiting appraisals that arehighly correlated with reluctance. As another

example, mere thoughts about engaging insexually related behavior can instigate arous-al of a different sort that is correlated withapproach. A vast, albeit nonbehavioral, lit-erature documents the adverse influence ofpersistent self-deprecatory thoughts (Peter-son & Seligman, 1984). Skinner acknowl-edged the behavioral impact of self-knowl-edge: Self-knowledge has a special value tothe individual himself. A person who hasbeen made aware of himself is in a betterposition to predict and control his own be-

havior (Skinner, 1974, p. 35). However,Skinner did not supply an account of howthe behavioral effects occur. Such an accountis not readily obtained from an analysis ofdirect contingencies (refer to the behavior ofthe rat above). When indirect respondingthrough derived relations is added to theanalysis, however, a verbal basis for the be-havioral effects of self-knowledge is suggest-ed.

Human emotion is a rich source of ex-amples that suggest a verbal basis for many

important behavioral effects. Emotions areintegrated into loosely defined and labeledcategories. Yet, loose definition and labelingnotwithstanding, under the right circum-stances, merely saying or thinking about thelabel renders self-discrimination of the la-beled emotion a verbal (bidirectionally re-active) event (at least in part). To the extentthat this is true, it is not true that thechange in feeling and the change in behaviorhave a common cause (Skinner, 1974, p.62), because both nonverbal and verbal con-tingencies are mixed in the control of thebehavior and the feeling, and they may bemixed in different ways. For anxious lan-guage-able persons, reports of anxiety are notnecessarily mere statements whose sole func-tion is communication (for self or other)


9/20

145ANXIETY

about various contexts, bodily states, or be-havioral predispositions associated with anx-iety. The reports can also be reactive andthus generate untoward effects. That is, they

not only describe the persons behavior andcircumstances but can also alter the functionof the behavior and circumstances described.For some persons, anxiety is a potentially de-bilitating category whose members include avariety of functional altering verbal compo-nents (e.g., unpleasant memories, negativeself-evaluations, unfavorable social compari-sons, etc.) and related bodily states (e.g., au-tonomic arousal). In a colloquial sense, anx-iety is what anxiety does, and what it doesincludes what anxious persons say it does.

Thus, many public and private events areintegrated in the category of respondingknown as anxiety. A common observationabout anxiety episodes is the appearance ofsurplus responding. That is, the related re-sponses (e.g., arousal, avoidance, escape) arenot fully explained by the direct contingen-cies in the present nor by those in the ap-prehensible past. The derived relational ac-count seems to supply some of the missinginformation, however. As described previ-ously, through transformation of stimulus

functions some of the behavioral functionsof one class of stimuli can carry over to an-other class, and this other class then beginsto exhibit behavioral functions that are sim-ilar to those of the first class.

For example, if a person feels anxious anda punishing event occurs (e.g., social rejec-tion), the functions of the feeling and theevent may merge through bidirectional stim-ulus transformation. Circumstances similarto those that occasion the feeling, as well asthe verbal report of the feeling, may subse-quently exhibit punishing properties, andthe punishing event may become more anx-iety provoking. This merging of functioncan create clinically debilitating responsepatterns (e.g., arousal, avoidance, escape,etc.) especially if, through derived relational

responding, the feeling and the thought ofanxiety are strongly influenced by the prop-erties of punishment. Illustrative examplesare plentiful. Recounting a highly aversive

event often instigates some (often much) ofthe arousal and avoidance responding thatwas occasioned by the event itself. Merelythinking about the event can produce similareffects (Hayes, Strosahl, & Wilson, in press;Hayes & Wilson, 1993, 1994). Complicat-ing matters further is the expanding conflu-ence of functions that gather in anxious ep-isodes as circumstances change from episodeto episode and the relational class grows.Highly anxious persons generally do not ex-hibit the pertinent responses (e.g., arousal,

avoidance, escape, etc.) in the presence ofjust one thing (American Psychiatric Asso-ciation, 1994). Their responses occur in thepresence of so many things that they some-times behave as if they are afraid of theirown shadow.

Some of the complaints about anxious ep-isodes involve emotional subcategories andthe phenomena integrated under them. Thelabels for the categories and their associatedpublic and private events can, similar to anx-iety, produce clinical effects through direct

and derived relations. For example, a com-mon complaint about some forms of anxietyinvolves a sense of loss of control (e.g.,panic). Other forms involve embarrass-ment (e.g., social phobia). Some involvesocial humiliation (e.g., agoraphobia).There are numerous other examples. Thepoint here is not that these labels, or anxietyitself, should be accepted into the argot ofbehavior analysis. As indicated earlier, anxi-ety is a vaguely defined, largely metaphoricalterm and is unlikely to ever become a trulytechnical term. If anything, these other cat-egory labels are even more vague and met-aphorical. Rather, the point it is that (a) thelabels are important members of the lan-guage of the English-speaking culture, (b)they refer to important psychological phe-


10/20


nomena, and (c) through derived relationalresponding, they can have adverse influenc-es.

There are other important implications of

this verbal account of anxiety. For example,the account provides some support for talktherapy. If verbal events are functionally re-lated to behaviorally and clinically importantevents in a clients current and previous en-vironments, it follows that knowledge of therelationships could help the therapist andclient predict and influence important be-havioral outcomes. The clients self-reportsare likely to be informative on a number oflevels, including the direct communicationof relevant information and the exhibition

of clinically relevant psychological functions(Kohlenberg, Hayes, & Tsai, 1993; Kohlen-berg & Tsai, 1991; Wulfert, Greenway, Far-kas, Hayes, & Dougher, 1994). Events thatare presented verbally in clinical sessions(e.g., events presented in imagination) mayalso generate effects similar to those gener-ated by directly presented events. A well-known case in point is the imaginal exposureused in systematic desensitization (e.g., Wol-pe, 1990).

More generally, a verbal account of emo-

tion underscores the value (for clinical be-havior analysts) of attention paid to the cli-ents private events. Those events include po-tentially maladaptive verbal and nonverbalresponses that are the direct and indirect re-sults of socialization within a clients verbalcommunity. The English language containsmyriad emotional terms, in part because theverbal community at large trains its mem-bers to use the terms to discuss their behav-ioral histories and to respond to those his-tories in a socially acceptable manner.

The preceding comments constitute anargument for the theoretical importance ofa behavioral account of emotion. Toward theend of the argument, applied implications ofsuch an account began to emerge. This wasa natural progression. Behavior analysts are

pragmatists, and applied behavior analystsare especially so. To recruit agreement withour view of emotion as an important topicfor applied behavior analysts, it is essential

to show how study and treatment of clini-cally significant behavior can be improvedwhen an analysis of emotional talk is con-ducted.

To begin this task, we will argue that be-havior resulting in avoidance of or escapefrom negative emotional states, as verballyconstrued by an individual, is negatively re-inforced. Critical to the analysis is that some(perhaps most) of the aversive properties ofthe emotional states and reinforcing prop-erties of the avoidance are primarily derived

through arbitrarily applicable relational re-sponding. That is, emotional avoidance ap-pears to be an integral component of humanverbal behavior (Hayes, Wilson, Gifford,Follette, & Strosahl, 1996). Our aim inmaking the argument is to supply appliedbehavior analysts with theoretical and ap-plied reasons to step through the openingSkinner provided in 1945.

ANXIETYDISORDERS ANDEXPERIENTIAL AVOIDANCE

Conventional Diagnoses and BehavioralAlternatives

A wide variety of anxiety disorders areidentified as distinct entities in theDSMdi-agnostic system. Over time, the various re-visions of the DSMhave systematically in-creased the number of purportedly distinctdisorders. In the most recent half-step, fromthe third revised edition to the fourth edi-tion (American Psychiatric Association,1987, 1994), the number of diagnosableanxiety disorders increased from 9 to 12.

With no theory to restrain the increasingproliferation of disorders, and with profes-sional reinforcement for finer and finer dis-tinctions in their formal properties, an infi-nite number becomes possible (Carson,


11/20

147ANXIETY

1996, 1997; Follette, Houts, & Hayes,1992). These disorders are also often arbi-trary with respect to function, which is evi-dent in the following comparison of phobias

with sexual aversion. If a person is fearful ofand therefore persistently avoids snakes,shopping malls, blood, heights, social inter-actions, elevators, or germs, he or she is di-agnosed with an anxiety disorder. If the per-son is fearful of and therefore persistentlyavoids sex, however, he or she is diagnosed

with sexual aversion, which is not an anxietydisorder. This theoretically unguided diag-nostic proliferation and functionally arbi-trary diagnostic application could be re-duced substantially through use of a func-

tional-dimensional approach to diagnosticclassification.

The functional-dimensional approach hasrecently been proposed as a behaviorally sen-sible alternative to the DSM. The approachinvolves organizing diagnoses according tobehavioral processes (therefore functional),and these processes are seen as continua(therefore dimensional rather than categori-cal; Hayes et al., 1996). Experiential avoid-ance has been suggested as one functionaldimension that could integrate a variety of

diverse DSMdiagnoses.

Experiential Avoidance as aFunctional Dimension

Early in life, humans, and most other an-imals, learn a large and versatile repertoire ofstrategies (e.g., vigilance, withdrawal) foravoiding events (e.g., pain, danger). As hu-mans develop, their verbal repertoires quick-ly become quite elaborate, and, correspond-ingly, their responses to aversive events ex-hibit verbal properties. That is, the responses(e.g., bodily sensations, emotions, thoughts,memories, behavioral predispositions) occa-sioned by the events begin to participate inderived relations with the events. Through atransformation of function, the responsesthemselves can become aversive, resulting in

two categories of phenomena whose avoid-ance results in negative reinforcement: theevents and the responses to them (cf. Gif-ford, 1994; Hayes & Wilson, 1993; Hayes

et al., 1996). This outcome may be specificto language-able persons. As indicated pre-viously, simpler organisms (e.g., rats) do notexhibit bidirectional derived relational re-sponding, and thus there is no known path-

way for their responses to aversive events toalso become aversive. In other words, it maybe true that only language-able persons ex-hibit experiential avoidance or behavior

whose primary function is the elimination,minimization, or reduction in the form, fre-quency, or situational sensitivity of various

private events (Hayes et al., 1996). Relat-edly, it seems that all of the anxiety disorderscould be cogently classified as experientialavoidance disorders. Below is a brief reviewof a sample of DSM anxiety disorders; itspurpose is to show the unifying effect of thefunctional dimensional approach to diagnos-tic classification.

Obsessive compulsive disorder (OCD).OCD involves the presence of unacceptablethoughts (obsessions) and disruptive mal-adaptive responses (compulsions) that pro-

vide temporary avoidance or escape from thethoughts and the physiological responses as-sociated with the thoughts (Hollander,1993; Rasmussen & Eisen, 1992). Amongindividuals diagnosed with OCD, unaccept-able thoughts dominate conscious awareness.These thoughts involve themes such as self-contamination, doubts about whether onehas performed an important act (e.g., turnedout lights, left doors open), or the desire toengage in socially unacceptable behavior(e.g., profanity, aggression). When suchthoughts occur to persons diagnosed withOCD, they try to ignore them, distractthemselves, or develop elaborate rituals toavoid or escape from the thoughts. For ex-ample, if the persons thoughts involve un-clean hands, he or she is likely to systemat-


12/20


ically and repetitively wash them. If inter-rupted, the person is likely to become dis-tressed, believing the interruption hasinterfered with the quality of the washing,

which amplifies the presence of the thoughtabout having unclean hands. Therefore,washing begins again and the cycle contin-ues until the thought is absent or present ina much diminished form (e.g., Im almostclean).

Consistent with our experiential avoid-ance approach to anxiety disorders, the dis-ruptive washing has important functionalproperties: reducing, avoiding, or escapingprivate events (experience). Thoughts areprivate verbal behavior. OCD-type responses

are reinforced by avoidance of related, highlyunpleasant thoughts (e.g., My hands arenot clean) and the various other privateevents that occasion or follow these formu-lations.

Panic disorder with agoraphobia. Panic in-volves intense autonomic arousal that is notas related to actual events as it is to the re-action to those events; in a colloquial sense,it is fear of fear (Taylor, 1995). For example,persons with agoraphobia do not avoid pub-lic places (e.g., the mall) per se; they avoid

the possibility of what might happen if theyenter a public place. Thoughts of a panicattack in the public place, and of the con-stellation of embarrassing responses that

would become public if the attack were tooccur there, are the negative reinforcers forstaying home (e.g., Amering et al., 1997).This account does not fully explain panic,but it provides an advance when combined

with a contemporary behavioral view of ver-bal events.

Persons diagnosed with agoraphobia be-come extremely reactive to changes in theirphysiological state (Barlow, 1988). Small in-creases in heart rate may be interpreted ver-bally as catastrophic (Pauli et al., 1991). Thiscatastrophic interpretation has been theorizedto result in a positive feedback loop between

perceptions of physiological activity [that]can culminate in a panic attack (Pauli et al.,1991, p. 137; see also Clark, 1986; Ehlers,Margraf, Roth, Taylor, & Birbaumer, 1988;

Taylor, 1995). In this feedback loop, bodilystates and other reactions occasion a verbalformulation that links the present to an im-pending dire future (e.g., Im losing control.Im going crazy. I am about to die.). Unfor-tunately for those afflicted with panic, thebodily states occasioned by bona fide periland those that occur through derived rela-tional responding (i.e., their reactions to theirreactions) are virtually indistinguishable.Thus, panic can be very behaviorally disrup-tive. In addition, autonomic arousal (fear) oc-

casioned by the thought of panic is a strongernegative predictor of treatment outcome thanis frequency or intensity of panic itself (Keij-sers, Hoogduin, & Schaap, 1994; cf. Chamb-less & Gracely, 1988). In sum, panic disorder

with agoraphobia is cogently categorized asan experiential avoidance disorder.

Posttraumatic stress disorder (PTSD).PTSD is the most recent and the most wide-ly applied diagnostic category for trauma-re-lated emotional disturbance. Previous de-scriptors were more colloquial and specific

but appeared to refer to the same emotionalphenomenon (e.g., shell shock, rape traumasyndrome). PTSD involves direct exposureto a traumatic event and the subsequentemergence of three clusters of symptoms:reexperience (nightmares, flashbacks), avoid-ance in active (avoiding trauma-related stim-uli) and passive (numbing, disassociation)forms, and increased arousal (e.g., insomnia,hypervigilance, exaggerated startle responses;

American Psychiatric Association, 1994; Foa& Meadows, 1997; Foa & Riggs, 1995).The pertinence of the experiential avoidanceapproach to classifying PTSD is evident inits symptom clusters. The first cluster in-volves the unpleasant experience and reex-perience of the traumatic event, the avoid-ance or escape from which reinforces re-


13/20

149ANXIETY

sponses in the second cluster: numbing, ac-tive avoidance, and disassociation. The thirdcluster involves persistent arousal that has anapparently elicited basis, at least part of

which is due to formal similarities betweenthe traumagenic and current events. Thepervasiveness of the arousal, however, seemsto unduly tax respondent conditioning as itssole explanation. It seems likely that some(perhaps most) of the arousal is verbally pro-duced.

The hyperarousal of PTSD often extendsto conditions that bear no formal similarityto the original traumagenic settings. Situa-tions that are only verbally or metaphoricallyrelated, such as unknown or unpredictable

situations, may also generate hyperarousal.Although it is conceivable that such re-sponses are due solely to higher order con-ditioning (cf. Forsyth & Eifert, 1996a,1996b), experiential avoidance and derivedrelational responding contribute to a preciseaccount with considerably more scope. Forexample, the respondent account cannotreadily explain why avoidance would lead toincreases in hyperarousal (in fact, from a re-spondent perspective, decreases would be ex-pected). Consistent with an experiential

avoidance perspective, PTSD is better pre-dicted by the avoidance symptom cluster(e.g., numbing, avoidance, and dissociativesymptoms) than by the aversive stimuli (fearand horror) that were present at the precip-itating event (Foa & Riggs, 1995). From thisperspective, the reactivating role of theavoidance cluster stems from a derived re-ciprocal relationship between avoidance,guided by verbal formulations that increasesensitivity to emotional reactions, and per-ceived (imagined, verbally formulated) directconsequences that instigate more reactions.

The previous discussion was used to showvery briefly the application of the experien-tial avoidance approach to three prevalentanxiety disorders. Although we could readilyapply the approach to all 12 of the current

DSM anxiety disorders, the redundancy ofsuch an effort would likely place an unnec-essary burden on the reader. Before movingto treatment implications, we want to briefly

reiterate two central points. The first pointis that when analysis of anxiety disorders in-corporates derived and direct learning pat-terns, it encompasses vastly more of the rel-evant phenomena than a solely direct con-tingency analysis does, and it does so with-out resort to hypothetical constructs (i.e., itis strictly learning based).

The second point is that the functionalclassification of behavior disorders is fosteredby an analysis of verbal relations and thatthe anxiety disorders have a dominant, ver-

bally established function of experientialavoidance. As indicated above, a core con-stituent of anxiety is avoidance. What isneeded is a fuller understanding of all thephenomena that are avoided. Anxious per-sons are fearful not only of environmentalevents but also of their responses to thoseevents. A person with panic disorder doesnot just avoid public places; he or she avoidsthe full range of private behavior associated

with those places. Skinners contention thatemotion and overt behavior are controlled

by the same events is thus incorrect or atleast incomplete. A fuller understanding re-quires an analysis of the complex verbal con-tingencies that are involved in the humandisposition to categorize arbitrary events(e.g., a pounding heart) as negative emotionsand respond accordingly (Im feeling pan-icky, I have to get out). An analysis of directcontingencies might reveal the basis for thepersons avoidance of public places, but itcannot readily account for the avoidance ofhis or her thoughts and feelings about thoseplaces.

These points yield the central themes ofthis paper: A behavior analysis of humanemotion is not only possible but is ultimate-ly necessary for understanding and treatinganxiety disorders. To complete the analysis,


14/20


an examination of both direct and verbalcontingencies is needed.

Treatment Implications

The functional benefits from the study ofemotion are perhaps most apparent in thecontext of treatment. The dominant behav-ioral approach to anxiety disorders isstraightforward: Expose the anxious personsto the feared phenomenon repeatedly and al-low other behavioral processes (e.g., habitu-ation, positive reinforcement) to extinguishthe maladaptive avoidance response class. Al-though many investigators recognize thatprivate events may also be important, a com-mon belief is that extinguishing public reac-

tivity to the feared object will concomitantlyextinguish this private reactivity (e.g., Skin-ner, 1969, 1974). A wide variety of exposuretechniques have been developed that roughlyaccord with this view (e.g., systematic desen-sitization, response prevention).

If the present account is correct, however,that view may often be insufficient. To befully effective, exposure may frequently haveto include all, or at least more, of the eventsthat functionally occasion maladaptiveavoidance. Most anxious persons seek treat-

ment to master being in the presence of thefeared object or event while not thinkingabout or feeling fear. This means that theverbal aspect of fear is part of the fearedevent. To incorporate more of the phenom-enology of fear in treatment, knowledge ofthe feared events and of the verbal behavioroccasioned by the events is therefore needed.

There are data from several laboratoriesdeveloping exposure-based treatments foranxiety disorders that illuminate this prob-lem. An expanding body of evidence showsthat attention to avoided private events im-proves outcomes in exposure-based treat-ments (Barlow, 1988). For example, Craske,Street, and Barlow (1989) gave persons whohad been diagnosed with agoraphobia in-structional sets to either focus on feared so-

matic sensations or to engage in distractiontasks during exposure to feared settings. Re-sults showed that, although the distractiongroup exhibited greater improvement post-

treatment than did the focus group, the fo-cus group exhibited greater improvements atthe 6-month follow-up. Similar results havebeen shown in persons diagnosed with OCD(Foa & Kozak, 1986; Grayson, Foa, & Stek-etee, 1982).

Thus exposure is crucial for effective treat-ment of anxiety, not merely in the brutephysical sense (e.g., reinforce successivelycloser movements to a feared object) but alsoin the psychological sense, which is consid-erably more difficult. Experiential avoidance

is an integral part of language itself, psycho-logical avoidance is often readily availablethrough distraction or other means, and ex-posure-based procedures generate a signifi-cant rate of resistance or refusal (cf. Hayeset al., 1996; McCarthy & Foa, 1990). Thus,exposure-based treatment requires specialcomponents that undermine covert avoid-ance, maintain participation, and reduceavoidance of private verbal events. Absent adetailed analysis of the clients emotionalconstructions, effective programming of the

needed exposure will often be difficult andsometimes impossible.

Recently, some behavior analysts have de-veloped treatments whose primary compo-nent is effective exposure. For example, ac-ceptance and commitment therapy (ACT) isan exposure-based behavioral interventionthat deliberately focuses on psychological ex-posure and the weakening of avoidance ofprivate verbal events (Hayes et al., in press;Hayes & Wilson, 1993, 1994; Kohlenberget al., 1993). Consistent with the experien-tial avoidance perspective of anxiety, the

ACT perspective is that event-based unpleas-ant emotion is not the essence of the clientsdysfunction; it is the cluster of responses

whose function is to reduce, avoid, escape,or otherwise modify the unpleasantness.


15/20

151ANXIETY

ACT therefore strives to attenuate the rela-tion between the problematic responses andtheir avoidance function. Among othergoals, the method attempts (a) to experien-

tially demonstrate the futility of the clientsself-avoidance responses, (b) to redefine ex-periential control as the problem rather thanthe solution, (c) to reinforce the clients ef-forts to remain in the presence of the pre-viously avoided experiences, and (d) to re-duce the linkage between overt behavior andprivate events. In a general clinical setting,clients of therapists who have been trainedin these concepts and techniques improvedsignificantly more than did clients treated

with mainstream methods (Strosahl, Hayes,

Bergan, & Romano, in press).There is another reason that understand-

ing emotion may be important for treat-ment. Fears (anxieties) can spread and recedein idiosyncratic ways. That is, fear can gen-eralize to events that bear no formal similar-ity to the originally feared event and can di-minish in the presence of feared events that

were neither the focus of treatment nor for-mally similar to events that were. Solving theapparent puzzle requires understanding thederived relations between events. For exam-

ple, the panic-disordered persons fear of el-evators or small spaces may generalize to hisor her marriage, even though the two situ-ations share no formal properties. Yet bothsituations may occasion the verbal formula-tion, I feel trapped, resulting in the func-tions of trapped affecting the elevator andthe marriage in similar ways.

A recent study of a boy with insect phobiapartially illustrates these points (Jones & Fri-man, 1997). At the beginning of the study,the boy would not enter a room if he wasmerely told that an insect was present with-in. Classmates could instigate extreme dis-ruptive behavior in the boy by saying theyhad seen an insect (thus the wordinsectandactual insects appeared to be in an equiva-lence class and to have similar functions).

The highly successful extinction-based treat-ment involved reinforcing completion ofmath problems in the presence of live in-sects. The boys most salient fears were of

spiders and ladybugs, which were relativelyunavailable, so live crickets, which werereadily available at a local pet store, wereused instead. Eventually the boy completedhigh rates of math problems with insectspresent on his work table. Following treat-ment, the boys aversion to crickets (part oftreatment) and spiders and ladybugs (notpart of treatment) subsided to below clinicallevels. The reactive effects of classmatescomments about insects were also eliminat-ed. On the last day of the study, while he

was seated at his desk in class, the subjectsaw a spider near his leg, picked it up witha tissue, placed it in a wastebasket, and re-turned to his schoolwork.

Although the boys succe sses wereachieved with a conventional approach toinsect phobia, explaining them seems to re-quire an appeal to direct and derived contin-gencies. As described earlier, respondentfunctions (which are central to emotion)spread readily through equivalence classesand other forms of derived stimulus relations

(Dougher et al., 1994; Roche & Barnes,1997). In the study mentioned above, theeffects of extinction appeared to spread fromaccessible insects (crickets) to less accessibleinsects and related verbal taunts (neither of

which were the focus of treatment), suggest-ing that all were members of a fear-basedequivalence class. The spread of effects frominsect to insect may be attributed persua-sively to direct contingencies because of for-mal resemblance (Fields et al., 1991; Herrn-stein, 1984; Medin & Smith, 1984; Rosch

& Mervis, 1975). To account for the spreadof effects to classmates taunts, however, ver-bal analysis seems necessary.

A CAVEAT FORBEHAVIORANALYSIS

The first part of this article elaborated onthe vagueness of the term anxiety and the


16/20


phenomena to which it refers. Two caution-ary points need to made here, however. First,problems with the term should not deter be-havior analysts from study of an area of psy-

chology that is so vast and important. Sec-ond, the problems should also not inhibituse of the term in common social parlance.The problems merely indicate that the termis not sufficiently precise to be considered atechnical term.

Skinner advocated not just forbearancebut acceptance of ordinary mentalizedterms in everyday parlance: The words theyuse are part of a living language that can beused without embarrassment by cognitivepsychologists and behavior analysts alike in

their daily lives. But these words cannot beused in their science! (Skinner, 1989, p.18). This distinction needs to be stressed.Employing the term anxiety in everydaynonscientific life can help persons to knoweach other better, predict what the othermight do in given situations, and advancerelationships. Denying such use seems tohave no advantage; quite the contrary, denial

would almost certainly instigate a long listof disadvantages, with professional segrega-tion, frequent misunderstanding, and occa-

sional ridicule near the top of the list (cf.Bailey, 1991; Deitz & Arrington, 1983;Foxx, 1996).

In addition, words that are technically im-precise (e.g., anxiety, fear, embarrassment,shame, stress) and thus unsuitable for the lab-oratory may have great value for application,

just as some words that are technically pre-cise (e.g.,negative reinforcement, punishment)are exceptionally ill suited for the languageof application (Lindsley, 1991). Despite itsapparent technical opacity, the term

anxietydoes have well-established functional valuefor virtually all of the social-verbal commu-nity except behavior analysts. Thus, at thevery least, the term has an important role inclinical settings.

Behavior analysts might be tempted to

avoid using the word anxietyin articles per-taining to the phenomena associated withthe term and may avoid listing the term ina subject index. This type of linguistic con-

servatism has some clear disadvantages, animportant one of which is the difficulty thatpotentially interested nonbehaviorists wouldhave in finding the research. Expanding theinfluence of behavior-analytic research be-yond the boundaries of the field requirescommunications that cross disciplinary andideological boundaries. These communica-tions will sometimes involve use of vagueterms, such as anxiety, that appear to bevaluable virtually everywhere but the labo-ratory.

There is a final reason that behavior an-alysts should treat these terms seriously, evenif they are imprecise or mentalistic. Themost distorted lay concept may refer to dis-tinctions that are important within a tech-nical account. For example, the concept ofintention may underline the importance ofverbal temporal frames in behavioral regu-lation (Hayes & Wilson, 1993); the matter-spirit distinction may be understood as animportant and reasonably accurate distinc-

tion between verbal content and verbal con-text (Hayes, 1984). There are numerous oth-er examples. We can never know a priori ifa lay distinction or concept is nonsense untila technical account is attempted. Eschewinglay terms and the domains to which theyrefer merely because they are in conflict witha scientific account can limit scientific prog-ress. The scientific task is to understand thesource of control over lay terms and theirusenot their dictionary meaning and lit-eral conflict or agreement with preferredterms. Understanding lay terms often re-quires a detailed examination of the contentto which they refer. Some terms may referto nothing of scientific value, but others(e.g., emotion) may. In the latter case, anadequate behavioral account can sometimes


17/20

153ANXIETY

be aided by the cultural wisdom that is em-bodied in language traditions.

CONCLUDING REMARKS

According to poets, political pundits, andphilosophers, we seem to be living in an ageof anxiety. To date, behavior analysts havelargely avoided the study of anxiety (andemotion in general), whereas mainstreampsychologists have devoted enormous re-sources to it (but have avoided behavioranalysis). Admittedly,anxietyis a vague met-aphoric term for a diverse array of prevalentpsychological phenomena for which a widelyaccepted technical definition is not available(and we have not offered one here). This

vagueness is the primary reason for the be-havior-analytic avoidance of the topic. Im-precision of a term, however, is not a suffi-cient justification for such avoidance whenthe phenomenon to which it refers is so vastand so central to the psychology of humanbeings. In addition, relinquishing anxiety tomainstream psychology virtually guaranteesthat proper analyses of functional relationsbetween environment and behavior, the me-tier of behavior analysts, will not be con-ducted. The literatures on the experimental

analysis of verbal behavior and experientialavoidance provide numerous productiveleads for those who wish to pick up the trail.

While remaining ever vigilant of the philo-sophical dangers presented by study ofvaguely defined phenomena, behavior ana-lysts should end their avoidance and begintheir empirical and theoretical approach.

REFERENCES

American Psychiatric Association. (1987). Diagnostic

and statistical manual of mental disorders(3rd ed.rev.). Washington, DC: Author.

American Psychiatric Association. (1994). Diagnosticand statistical manual of mental disorders(4th ed.).

Washington, DC: Author.Amering, M., Katschnig, H., Berger, P., Windhaber,

J., Baischer, W., & Dantendorfer, K. (1997). Em-barrassment about the first panic attack predicts

agoraphobia in panic disorder patients. BehaviorResearch and Therapy, 35, 517521.

Anderson, C. M., Hawkins, R. P., & Scotti, J. R.(1997). Private events in behavior analysis: Con-ceptual basis and clinical relevance.Behavior Ther-apy, 28, 157179.

Bailey, J. S. (1991). Marketing behavior analysis re-quires different talk. Journal of Applied Behavior

Analysis, 24, 445448.Barlow, D. H. (1988). Anxiety and its disorders: Na-

ture and treatment of anxiety and panic.New York:Guilford Press.

Barnes, D. (1994). Stimulus equivalence and rela-tional frame theory. The Psychological Record, 44,91124.

Beck, A. T., & Emery, G. (1985). Anxiety disordersand phobias. New York: Basic Books.

Biglan, A., & Hayes, S. C. (1995). Should the be-havioral sciences become more pragmatic? Thecase for functional contextualism in research onhuman behavior. Applied and Preventive Psycholo-

gy: Current Scientific Perspectives, 5, 4757.Carson, R. C. (1996). Aristotle, Galileo, and the

DSM taxonomy: The case of schizophrenia.Jour-nal of Consulting and Clinical Psychology, 64,11331139.

Carson, R. C. (1997). Costly compromises: A cri-tique of the DSM. In S. Fisher & R. Greenberg(Eds.),Psychopharmacology scientifically reappraised(pp. 98114). New York: Wiley.

Chambless, D. L., & Gracely, E. J. (1988). Predictionof outcome following in vivo exposure treatmentof agoraphobia. In I. Hand & H. U. Wittchen(Eds.), Panic and phobias: Vol. 2. Treatment andvariables affecting outcome (pp. 209220). New

York: Springer.

Clark, D. M. (1986). A cognitive approach to panic.Behaviour Research and Therapy, 24, 461470.

Craske, M. G., Street, L., & Barlow, D. H. (1989).Instructions to focus upon or distract from inter-nal cues during exposure treatment of agorapho-bic avoidance.Behaviour Research and Therapy, 27,663672.

DeGrandpre, R. J., Bickel, W. K., & Higgins, S. T.(1992). Emergent equivalence relations betweeninteroceptive (drug) and exteroceptive (visual)stimuli.Journal of the Experimental Analysis of Be-havior, 58, 918.

Deitz, S. M., & Arrington, R. L. (1983). Factors con-fusing language use in the analysis of behavior.Behaviorism, 11, 117132.

Devany, J. M., Hayes, S. C., & Nelson, R. O. (1986).Equivalence class formation in language-able andlanguage-disabled children. Journal of the Experi-mental Analysis of Behavior, 46, 243257.

Dougher, M. J., Augustson, E., Markham, M. R.,Greenway, D. E., & Wulfert, E. (1994). Thetransfer of respondent eliciting and extinctionfunctions through stimulus equivalence classes.


18/20


Journal of the Experimental Analysis of Behavior,62, 331351.

Dymond, S., & Barnes, D. (1995). A transformationof self-discrimination response functions in accor-dance with the arbitrarily applicable relations ofsameness, more than, and less than. Journal of theExperimental Analysis of Behavior, 64, 163184.

Dymond, S., & Barnes, D. (1996). A transformationof self-discrimination response functions in accor-dance with the arbitrarily applicable relations ofsameness and opposition.The Psychological Record,46, 271300.

Ehlers, A., Margraf, J., Roth, W. T., Taylor, C. B., &Birbaumer, N. (1988). Anxiety induced by falseheart rate feedback in patients with panic disorder.Behavior Research and Therapy, 26, 111.

Epstein, S. (1967). Toward a unified theory of de-pression. In B. A. Maher (Ed.), Progress in exper-imental personality research (Vol. 4, pp. 189).New York: Academic Press.

Ferster, C., & Skinner, B. F. (1957). Schedules of re-

inforcement. New York: Appleton-Century-Crofts.Fields, L., Reeve, K. E., Adams, B. J., & Verhave, T.

(1991). Stimulus generalization and equivalenceclasses: A model for natural categories. Journal of the Experimental Analysis of Behavior, 55, 305312.

Foa, E. B., & Kozak, M. S. (1986). Emotional pro-cessing of fear: Exposure to corrective informa-tion.Psychological Bulletin, 99, 2035.

Foa, E. B., & Meadows, E. A. (1997). Psychosocialtreatments for posttraumatic stress disorder: Acritical review. Annual Review of Psychology, 48,449480.

Foa, E. B., & Riggs, D. S. (1995). Posttraumaticstress disorder following assault: Theoretical con-

siderations and empirical findings.Current Direc-tions in Psychological Science, 4, 6165.

Follette, W. C., Houts, A. C., & Hayes, S. C. (1992).Behavior therapy and the new medical model.Be-havioral Assessment, 14, 323343.

Forsyth, J. P., & Eifert, G. H. (1996a). The languageof feeling and the feeling of anxiety: Contribu-tions of the behaviorisms toward understandingthe function altering effects of language.The Psy-chological Record, 46, 607649.

Forsyth, J. P., & Eifert, G. H. (1996b). Systematicalarms in fear conditioning I: A reappraisal of

what is being conditioned.Behavior Therapy, 27,441462.

Foxx, R. M. (1996). Translating the covenant: The

behavior analyst as ambassador and translator.TheBehavior Analyst, 19, 147162.

Freud, S. (1966). Anxiety (lecture XXV). In J. Stra-chey (Ed.), The complete introductory lectures on

psychoanalysis (pp. 392411). New York: Norton.(Original work published 1917)

Gifford, E. V. (1994). Setting a course for change:The verbal context of acceptance. In S. C. Hayes,

N. S. Jacobson, V. M. Follette, & M. J. Dougher(Eds.),Acceptance and change: Content and contextin psychotherapy(pp. 218222). Reno, NV: Con-text Press.

Grayson, J. B., Foa, E. B., & Steketee, G. (1982).Habituation during exposure treatment: Distrac-tion versus attention focusing. Behaviour Researchand Therapy, 20, 323328.

Hallam, R. S. (1985). Anxiety. London: AcademicPress.

Hayes, S. C. (1984). Making sense of spirituality.Be-haviorism, 12, 99110.

Hayes, S. C. (1989). Nonhumans have not yet shownstimulus equivalence. Journal of the Experimental

Analysis of Behavior, 51, 385392.Hayes, S. C., & Brownstein, A. J. (1986). Mentalism,

behavior-behavior relations and a behavior analyt-ic view of the purposes of science. The Behavior

Analyst, 9, 175190.Hayes, S. C., Brownstein, A. J., Devany, J. M., Koh-

lenberg, B. S., & Shelby, J. (1987). Stimulus

equivalence and the symbolic control of behavior.Mexican Journal of Behavior Analysis, 13, 361374.

Hayes, S. C., & Hayes, L. J. (1989). The verbal ac-tion of the listener as a basis for rule-governance.In S. C. Hayes (Ed.),Rule-governed behavior: Cog-nition, contingencies, and instructional control(pp.153190). New York: Plenum.

Hayes, S. C., & Hayes, L. J. (1992). Verbal relationsand the evolution of behavior analysis. AmericanPsychologist, 47, 13831395.

Hayes, S. C., Kohlenberg, B. K., & Hayes, L. J.(1991). Transfer of consequential functionsthrough simple and conditional equivalence class-es.Journal of the Experimental Analysis of Behavior,

56, 119137.Hayes, S. C., Strosahl, K., & Wilson, K. G. (in press).

Acceptance and commitment therapy: Understandingand treating human suffering. New York: GuilfordPress.

Hayes, S. C., & Wilson, K. G. (1993). Some appliedimplications of a contemporary behavior-analyticaccount of verbal events.The Behavior Analyst, 16,283301.

Hayes, S. C., & Wilson, K. G. (1994). Acceptanceand commitment therapy: Undermining the ver-bal support for experiential avoidance.The Behav-ior Analyst, 17, 289303.

Hayes, S. C., & Wilson, K. G. (1996). Criticisms ofrelational frame theory: Implications for a behav-

ior analytic account of derived stimulus relations.The Psychological Record, 46, 221236.

Hayes, S. C., Wilson, K. G., Gifford, E. V., Follette,V. M., & Strosahl, K. (1996). Experiential avoid-ance and behavior disorders: A functional dimen-sional approach to diagnosis and treatment.Jour-nal of Consulting and Clinical Psychology, 64,11521168.


19/20

155ANXIETY

Herrnstein, R. J. (1984). Objects, categories, and dis-criminative stimuli. In H. L. Roitblatt, T. G. Bev-er, & H. S. Terrace (Eds.), Animal cognition(pp.233261). Hillsdale, NJ: Erlbaum.

Hoes, M. J. A. J. M. (1986). Biological markers inpsychiatry.Acta Psychiatrica Belgica, 86,220241.

Hollander, E. (1993). Obsessive-compulsive related dis-orders. Washington, DC: American PsychiatricPress.

Horne, P. J., & Lowe, C. F. (1996). On the originsof naming and other symbolic behavior. Journalof the Experimental Analysis of Behavior, 65,185242.

Jones, K. M., & Friman, P. C. (1997). Functionalanalysis and treatment of entomophobia in a boy inresidential care: A preliminary case report. Papersubmitted for publication.

Keedwell, P., & Snaith, R. P. (1996). What do anxietyscales measure?Acta Psychiatrica Scandinavica, 93,177180.

Keijsers, G. P. J., Hoogduin, C. A. L., & Schaap, C.P. (1994). Prognostic factors in the treatment ofpanic disorder with and without agoraphobia.Be-havior Research and Therapy, 25, 689708.

Kohlenberg, R. J., Hayes, S. C., & Tsai, M. (1993).Radical behavioral psychotherapy: Two contem-porary examples. Clinical Psychology Review, 13,579592.

Kohlenberg, R. J., & Tsai, M. (1991). Functional an-alytic psychotherapy. New York: Plenum.

Lattal, K. A. (1975). Reinforcement contingencies asdiscriminative stimuli.Journal of the Experimental

Analysis of Behavior, 23, 241246.Leigland, S. (1997). Is a new definition of verbal be-

havior necessary in light of derived relational re-

sponding?The Behavior Analyst, 20, 310.Levitt, E. E. (1967). The psychology of anxiety. Indi-anapolis: Bobbs-Merrill.

Lindsley, O. R. (1991). From technical jargon toplain English for application. Journal of AppliedBehavior Analysis, 24, 449458.

Lipkens, G., Hayes, S. C., & Hayes, L. (1993). Lon-gitudinal study of derived stimulus relations in aninfant. Journal of Experimental Child Psychology,56, 201239.

Margraf, J., Ehlers, A., & Roth, W. (1986). Sodiumlactate infusions and panic attacks: A review andcritique.Psychosomatic Medicine, 48, 2351.

Margraf, J., Ehlers, A., & Roth, W. (1988). Mitralvalve prolapse and panic disorder: A review of

their relationship.Psychosomatic Medicine, 50,93113.

May, R. (1950). The meaning of anxiety. New York:The Ronald Press.

McBurnett, K., Lahey, B. B., Frick, P. J., Risch, C.,Loeber, R., Hart, E. L., Christ, M. A. G., & Han-son, K. S. (1991). Anxiety, inhibition, and con-duct disorder in children: II. Relation to salivary

cortisol.Journal of the American Academy of Childand Adolescent Psychiatry, 30, 192196.

McCarthy, P. R., & Foa, E. B. (1990). Obsessive-compulsive disorder. In M. E. Thase, B. A. Edel-stein, & M. Hersen (Eds.),Handbook of outpatienttreatment of adults: Nonpsychotic mental disorders(pp. 209234). New York: Plenum.

Medin, D. L., & Smith, E. E. (1984). Concepts andconcept formation. Annual Reviews of Psychology,35, 113138.

Oxford University Press. (1971). The compact editionof the Oxford English dictionary. New York: Au-thor.

Pauli, P., Marquardt, C., Hartl, L., Nutzinger, D. O.,Holzl, R., & Strain, F. (1991). Anxiety inducedby cardiac perceptions in patients with panic at-tacks: A field study. Behaviour Research and Ther-apy, 29, 137145.

Pavlov, I. P. (1927). Conditioned reflexes. Oxford: Ox-ford University Press.

Peterson, C., & Seligman, M. (1984). Causal expla-

nation as a risk factor for depression: Theory andevidence.Psychological Review, 91, 347374.

Poling, A., Methot, L. L., & LeSage, M. G. (1995).Fundamentals of behavior analytic research. New

York: Plenum.Rachman, S. (1977). The conditioning theory of fear

acquisition: A critical examination. Behavior Re-search and Therapy, 15, 375387.

Rachman, S. (1991). Neo-conditioning and the clas-sical theory of fear acquisition.Clinical PsychologyReview, 11, 155173.

Rapee, R. M., Brown, T. A., Antony, M. M., & Bar-low, D. H. (1992). Response to hyperventilationand inhalation of 5.5% carbon dioxide-enrichedair across the DSM-III-R anxiety disorders. Jour-

nal of Abnormal Psychology, 101, 538552.Rasmussen, S. A., & Eisen, J. L. (1992). The epi-

demiology and differential diagnosis of obsessivecompulsive disorder.Journal of Clinical Psychiatry,53, 410.

Roche, B., & Barnes, D. (1997). A transformation ofrespondently conditioned stimulus function in ac-cordance with arbitrarily applicable relations.Jour-nal of the Experimental Analysis of Behavior, 67,275301.

Rosch, E., & Mervis, C. B. (1975). Family resem-blance: Studies in the internal structure of cate-gories.Cognitive Psychology, 7, 573605.

Ryle, G. (1949). The concept of mind. Chicago: Uni-versity of Chicago Press.

Sarbin, T. R. (1964). Anxiety: The reification of ametaphor.Archives of General Psychiatry, 10, 630638.

Sarbin, T. R. (1968). Ontology recapitulates philol-ogy: The mythic nature of anxiety.American Psy-chologist, 23, 411418.

Sidman, M. (1960). Normal sources of pathologicalbehavior. Science, 132, 6168.


20/20


Sidman, M. (1971). Reading and auditory-visualequivalences. Journal of Speech and Hearing Re-search, 14, 513.

Sidman, M. (1994). Equivalence relations: A researchstory. Boston: Authors Cooperative.

Sidman, M., & Tailby, W. (1982). Conditional dis-crimination vs. matching to sample: An expansionof the testing paradigm. Journal of the Experimen-tal Analysis of Behavior, 37, 522.

Skeat, W. R. (1963). A concise etymological dictionaryof the English language. New York: CapricornBooks.

Skinner, B. F. (1938). The behavior of organisms. NewYork: Appleton-Century-Crofts.

Skinner, B. F. (1945). The operational analysis of psy-chological terms. Psychological Review, 52, 270277.

Skinner, B. F. (1953). Science and human behavior.New York: Free Press.

Skinner, B. F. (1957). Verbal behavior. New York: Ap-pleton-Century-Crofts.

Skinner, B. F. (1969). Contingencies of reinforcement:A theoretical analysis. New York: Appleton-Cen-tury-Crofts.

Skinner, B. F. (1974). About behaviorism. New York:Knopf.

Skinner, B. F. (1989). The origins of cognitivethought.American Psychologist, 44, 1318.

Sofuoglu, S., Dogan, P., Besim, T., Basturk, M., &Tanrikulu, G. (1992). MAO-B activity in de-pression and anxiety disorders. Biological Psychia-try, 31, 634638.

Steele, D. L., & Hayes, S. C. (1991). Stimulus equiv-

alence and arbitrarily applicable relational re-sponding. Journal of the Experimental Analysis of Behavior, 56, 519555.

Strachey, J. (Ed.). (1966). The complete introductorylectures on psychoanalysis. New York: Norton.

Strosahl, K. D., Hayes, S. C., Bergan, J., & Romano,P. (in press). Does field based training in behaviortherapy improve clinical effectiveness? Evidencefrom the acceptance and commitment therapytraining project.Behavior Therapy.

Taylor, S. (1995). Anxiety sensitivity: Theoretical per-spectives and recent findings. Behavior Researchand Therapy, 33, 243258.

Wickler, A. (1973). Conditioning of successive adap-tive response to the initial effects of drugs. Con-ditioned Reflex, 8, 193210.

Wittgenstein, L. (1958a). The blue and brown books.New York: Harper & Row.

Wittgenstein, L. (1958b). Philosophical investigations.New York: MacMillan.

Wolpe, J. (1990). The practice of behavior therapy.

New York: Pergamon.Wulfert, E., Greenway, D. E., Farkas, P., Hayes, S. C.,

& Dougher, M. J. (1994). Correlation betweena personality test for rigidity and rule-governedinsensitivity to operant contingencies. Journal of

Applied Behavior Analysis, 27, 659671.Yalom, I. D. (1980). Existential psychotherapy. New

York: Basic Books.

Received October 20, 1997Initial editorial decision November 6, 1997Final acceptance November 21, 1997

Action Editor, David P. Wacker

friman, hayes and wilson - why behavior analysts should study emotion-the example of anxiety

Documents