Kelly G. Wilson Steven C. University

The Behavior Analyst 2000, 23, 25-43 No. 1 (Spring)

Why It Is Crucial to Understand Thinkingand Feeling: An Analysis and Application to

Drug AbuseKelly G. Wilson and Steven C. Hayes

University of Nevada

Behavior analysis has long accepted the legitimacy of the analysis of private events in a naturalscience of behavior. However, the topic has languished as a focus of empirical research in eitherapplied or basic arenas. We argue that recent empirical work examining the bidirectional nature ofverbal relations may shed light on the role of private events in complex human behavior. Skinnerargued that although it would be possible to analyze private events; we need not, because thoughtsand feelings were viewed as co-occuring products of the same contingencies that are responsiblefor changes in overt responses. However, the bidirectional transformation of stimulus function in-herent in verbal behavior changes the way that private events participate in complex behavioralepisodes for verbal organisms. We examine why we have reached such a conclusion, with specialemphasis on the role of self-awareness. Finally, we conclude with an application of our analysis tothe problem of substance abuse.Key words: cognition, emotion, private events, substance abuse, alcoholism, self-awareness, re-

lational frame theory, verbal behavior, derived stimulus relations, bidirectionality

A scientific analysis of human pri-vate events has been philosophicallyincluded in behavior analysis for over50 years (Skinner, 1945/1972), but ithas been largely excluded from empir-ical research within this tradition (Tay-lor & O'Reilly, 1997). At times the ex-clusion is based on methodologicalconcerns, very much in line with Wat-son's (1924) methodological rejectionof introspection. In this view, becauseinterobserver agreement is seeminglyimpossible with private events, theyought not to be included in scientificstudy (e.g., Lamal, 1998; Zuriff, 1988).At other times the exclusion is prag-matic, in line with Skinner (1953), whoargued successfully that an understand-ing of private events was scientificallylegitimate within behavioral psycholo-gy, but that it was unnecessary for ascientific understanding of overt activ-ity. An extensive empirical literatureon conditioned emotional responses

Preparation of this manuscript was supportedin part by Grant DA08634 from the NationalInstitute of Health, National Institute on DrugAbuse.

Address correspondence to Kelly G. Wilson,Department of Psychology, University of Ne-vada, Reno, Nevada 89557-0062.

exists in the classical conditioning lit-erature, but this literature ultimatelydoes not provide either an adequatemeans or a sufficient reason to studyongoing emotional responses in hu-mans. Thus, in behavior analysis theempirical analysis of thoughts and feel-ings is often viewed as either invalidor as legitimate and valid but largelyunnecessary (Friman, Hayes, & Wil-son, 1998; Friman, Wilson, & Hayes,1998). The purpose of this paper is tooffer a somewhat different analysis,making use of recent research in stim-ulus equivalence and related phenom-ena. We will then explore the appliedimplications of our analysis with par-ticular emphasis on application of theanalysis to drug abuse.

THE SOCIAL CONSTRUCTIONOF PRIVATE EVENTS

In a behavior-analytic approach, pri-vate events are, in an important sense,public events: "Self-knowledge is ofsocial origin. It is only when a person'sprivate world becomes important toothers that it is made important to him.It then enters into the control of thebehavior called knowing" (Skinner,1974, p. 31). In essence, Skinner's line

25

26 KELLY G. WILSON & STEVEN C. HAYES

of thinking argues that the social-ver-bal community shapes the identifica-tion of private events and the subse-quent stimulus control these events ex-ert. Skinner (1945/1972, p. 378) point-ed out that "differential reinforcementcannot be made contingent upon theproperty of privacy." He suggestedfour primary means by which the ver-bal community circumvents the prob-lem of privacy and shapes verbal re-sponses under the discriminative con-trol of private events, that is, self-knowledge:

1. Private stimuli are correlated withpublicly accessible stimuli. For exam-ple, a skinned knee or a bump on thehead is a publicly observable event thatis well correlated with pain as a privateevent. It is possible for the social-ver-bal community to teach a child to say"that hurts" in response to private painstimuli because people often can seewhat has happened to the child tocause pain, and thus can teach the childto make such a verbal report.

2. Publicly accessible responses arecorrelated with private stimuli. For ex-ample, flinching when a tooth istouched is well correlated with a tooth-ache. This allows the verbal commu-nity to shape talk about the privatelyexperienced painful stimulation.

3. Publicly shaped responses to pub-licly available stimuli may recede inmagnitude, and thus become private.An example of this might include re-hearsing lines of a play aloud, even-tually leading to rehearsing them in si-lent self-talk or "thought." Similarly,a child learns to suppress overt speech,especially in school, but may still en-gage in private forms of such speech(see also Skinner, 1957, p. 141).

4. Responses to public stimuli maybe metaphorically applied to privatelyfelt states. Skinner offers, as examples,terms such as agitated, ebullient, anddepressed. These terms originally re-ferred to events in the external worldwith which people interacted. To illus-trate, when a jar of liquid has somepowder placed in it and is agitated, thetwo may become mixed together and

move this way and that. They are un-settled. Similarly, when individuals ex-perience unusual events that "shakethem up," their physical state and theirbehavior may become less predictable.They may say that they feel unsettled.Skinner suggested that such metaphor-ic description or knowledge is partic-ularly imprecise, owing to the intrinsicimprecision of metaphor.

These four sources of control overknowledge of private events have with-stood the test of time within behavioranalysis, and the basic conceptualiza-tion seems behaviorally sound. ButSkinner added something else aboutthe role of private events: "Self-knowl-edge has a special value to the individ-ual himself. A person who has been'made aware of himself' is in a betterposition to predict and control his ownbehavior" (1974, p. 31). We believethat this claim provides a window onthe traditional behavioral account ofprivate events that reveals a neededaugmentation. Why would knowledgeof private events put an individual in abetter position to predict and controlhis or her own behavior? Asked in amore technical way, why would dis-criminated responses to one's own pri-vate responding exert behavior-regula-tory functions over any subsequent re-sponding?

THE IMPACT OFSELF-KNOWLEDGE

There seems to be little disagree-ment that human self-knowledge hasbehavioral functions. Consider a com-mon human scenario: A person pri-vately describes his previous impulsivealcohol consumption and its long-termcosts, and subsequently shows moreself-control. In this case, self-knowl-edge has been associated with somehealthy degree of insensitivity to short-term contingencies. We believe thatthese were the kinds of things thatSkinner was referring to in describingthe impact of self-knowledge. Theproblem is that these behavioral resultscannot come from self-knowledge as a

WHY ANALYZE THINKING AND FEELING 27

simple discriminated operant. Under-standing why a simple discriminatedoperant analysis of self-knowledge isunworkable will help us to explainwhy recent developments in the anal-ysis of verbal behavior can expand andimprove on traditional behavioral anal-yses of the role of private events.

Consider an operant experimentmodeled after the self-control scenario.A pigeon is given a choice between avery small amount of food immediate-ly or a large amount delayed a few sec-onds. In these circumstances, pigeonswill choose the small immediate rein-forcer (Rachlin & Green, 1972). Sup-pose we now teach the bird to "tell"which reinforcer was obtained, the im-mediate or the delayed, by reinforcingsome other response under the anteced-ent stimulus control of the delay be-tween its last key peck and the amountof access to the food hopper. The birdis performing an act of discriminationregarding its own behavior, a form ofself-knowledge (see Shimp, 1981,1982). However, we have no reason tosuppose that the bird will now choosethe large delayed reinforcer when re-introduced to the original choice. Thereport is occasioned by the choice, butin order for the report to change thefunctions of the discriminative stimuliin the choice situation, it would haveto be bidirectionally related to that sit-uation.' Simply providing a history ofreporting would not accomplish thistask. Similarly, respondent condition-ing appears to be a largely unidirec-tional conditioning process. Backwardconditioning rarely occurs and, even inthe limited conditions in which it ap-pears to, is extremely weak and tran-sitory (Hall, 1984; Mackintosh, 1974;see also Wilson & Blackledge, inpress, for extended discussion).We return now to the idea that "self-

knowledge has a special value to the

Bidirectionality refers to the fact that thetransformation of stimulus function seen amongmembers of a relational frame does not requireany particular temporal ordering, unlike classicalor operant conditioning.

individual himself" (Skinner, 1974, p.31). In cases of self-knowledge in non-verbal organisms, it seems unlikelythat there would be such special value.Further, we know of no empirical ex-ample in the animal literature thatwould contradict the example we havechosen.

The Implications of BidirectionalVerbal Processes for Self-Knowledge

Self-knowledge is very different fora verbally competent human than it isfor a nonverbal organism. The reasonis straightforward: Human verbal be-havior is known to be bidirectional inits relation to other stimuli. The sim-plest way to make this argument is toconsider the literature on stimulusequivalence. Many studies have dem-onstrated that when training is provid-ed sufficient to form an equivalenceclass, various psychological functionswill transfer among those stimuli, with-out regard to the sort of temporal or-dering required in typical classical andoperant conditioning paradigms. Func-tions demonstrated to transfer amongmembers of an equivalence class in hu-mans include conditioned reinforcingfunctions (Hayes, Brownstein, Devany,Kohlenberg, & Shelby, 1987; Hayes,Kohlenberg, & Hayes, 1991), discrim-inative functions (Dymond & Barnes,1995; Hayes et al., 1987), emotion-eliciting functions (Dougher, August-son, Markham, Greenway, & Wulfert,1994; Roche & Barnes, 1997), and ex-tinction functions (Dougher et al.,1994).

Stimulus equivalence and otherforms of derived stimulus relations areclearly involved in human verbal be-havior, regardless of one's interpreta-tion of this relationship (e.g., Hayes &Hayes, 1992; Home & Lowe, 1996).Empirically, we know that preparationsused in the study of derived stimulusrelations can be used directly to trainbasic verbal performances. For exam-ple, reading can be taught throughequivalence (e.g., Sidman, 1971) or ex-clusion (e.g., de Rose, de Souza, Ros-


sito, & de Rose, 1994) procedures. Inaddition, equivalence correlates withbasic verbal abilities in humans (e.g.,Devany, Hayes, & Nelson, 1986). In-deed, what we mean by such conceptsas a symbol or a word seems to requirebidirectional stimulus relations.

Derived stimulus relations occurwith humans at a very early age. If a16-month-old child learns to selectStimulus A when Stimulus B is pres-ent, he or she will then select (withoutexplicit experimental training) Stimu-lus B when Stimulus A is present (G.Lipkens, Hayes, & Hayes, 1993). Atleast by 23 months, human childrentaught to select Stimulus A in the pres-ence of Stimulus B, and to select Stim-ulus A in the presence of Stimulus C,will then select Stimulus B in the pres-ence of Stimulus C (Devany et al.,1986; G. Lipkens et al., 1993). Manyforms of arbitrarily applicable stimulusrelations (what we call relationalframes; Hayes & Barnes, 1998) can belearned and can combine into networksof great complexity (Dymond &Barnes, 1995; R. Lipkens, 1992; Roche& Barnes, 1997; Steele & Hayes,1991).

Skinner recognized the ubiquity ofbidirectionality in his descriptions ofverbal behavior. In one example, Skin-ner describes an individual' who hearsan electrician say "This is a Jones-plug." According to Skinner, "The ef-fect on the listener is not only to estab-lish Jones-plug as an appropriate tactbut to set up nonverbal behavior in re-sponse to similar stimuli, for example,behaving correctly when asked Pleasehand me a Jones-plug" (Skinner, 1957,p. 360). In Skinner's analysis, such bi-directionality was the result of a longhistory of the use of verbal operants.Although he offers detailed operantanalyses of the emergence of tacts andmands, he offers no technical analysisof the emergence or maintenance ofmore complex verbal operants thatclearly involve bidirectionality. The lit-erature on equivalence and its emer-gence in infancy did not exist in 1957.If it had, bidirectionality might have

played a more central role in Skinner'stechnical account of the formation ofthese verbal operants and their impactin areas such as self-knowledge.The key here for our present purpose

is that human verbal behavior func-tions bidirectionally. It is on that foun-dation that the behavioral impact ofself-knowledge stands. Suppose onelearns for the first time that anothername for lemon is betrang. Now spenda few moments imagining what itwould feel like to cut open a big, juicybetrang, and squeeze out all the be-trang juice into one's mouth. Somereaders of this article probably are nowsalivating, or feeling their teeth to beon edge, even though their experiencewith betrangs is limited to this para-graph. Such transfer of elicitationthrough equivalence classes has beenrepeatedly demonstrated in well-con-trolled laboratory conditions (Dougheret al., 1994; Roche & Barnes, 1997).Notice that the word lemon precededbetrang in the phrase another name forlemon is betrang. Classical condition-ing cannot account for the transfer offunction from lemon to betrang, be-cause the stimuli were introduced inthe wrong order. However, order wasnot important because, due to the bi-directional functioning of words, thephrase established an equivalence re-lation between the two. Some of thefunctions of lemon-probably them-selves dependent on an equivalence re-lation between lemons and actual lem-ons, which enabled the transfer ofsome of the functions of the actualevent to the word-then transferred tobetrang.We argue that this same process

makes self-knowledge both importantand useful on the one hand and oftenemotional and difficult on the other(Hayes & Gifford, 1997; Hayes & Wil-son, 1993; Wilson & Blackledge, inpress). It is useful because verbal re-ports can alter the impact of defectivecontingencies that may be described.Developmental studies have shown agradual transition in the impact of self-instructions in dealing with, for exam-


ple, temporal delay (Bentall & Lowe,1987). But in a different direction, cli-nicians are persistently confronted withthe emotional nature of self-knowl-edge. For example, persons who haveexperienced a traumatic event seem toreexperience the aversiveness of theevent in the report of it. Indeed, it isoften very difficult to get victims oftrauma to discuss traumatic events atall. We believe that this is so becausethe verbal report carries with it someof the functions of the original trauma.This same bidirectionality can helpheal trauma, however. If the person isable to discuss the original trauma free-ly and nondefensively, this kind of ver-bal exposure can change the emotionaland other behavioral functions of theactual stimuli associated with the trau-ma, such as riding in an automobile fora person who has been in a terrible ac-cident (see Pennebaker, 1997, for a re-view).Our larger point is this: Private

events such as "emotions" need to beconsidered in an account of complexhuman behavior because private eventsinvolve two sets of contingencies, bothverbally presented and directly experi-enced. These verbal contingencies canestablish a functional role for privateevents, based in particular on the bi-directionality of human language, thatwould simply be missed if these eventswere not considered.

PRIVATE EVENTS ASTHE LANGUAGE OF HISTORYWhy does the verbal community

spend as much time as it does teachingchildren to discriminate private events?Why do we wish to know what a childis thinking and feeling?

Feelings, Metaphor, and AnalogyWe argue that talk about feelings is

important to the verbal community be-cause these terms give access to the re-sponse implications of an individual'shistory that are idiosyncratic and dif-ficult to interpret. For example, the ver-bal community usually asks "Do you

feel lonely?" instead of "Have youbeen deprived of access to social inter-action?" because being in a position inwhich socializing is a valued activityis not a simple issue of social depri-vation. Sometimes social interactionscan be aversive. In such situations aperson may "feel" like being alone,but not feel lonely. Sometimes peoplesay they feel "alone in a crowd," andthus social deprivation may be func-tionally present but formally absent.When seeing a picture of dead pet, aperson may "feel lonely" even thoughthe loss occurred years ago. Seeingsomeone else who appears to havemany friends may make a personaware of a relative degree of socialpoverty, and the person may begin tofeel lonely without any change in thefrequency of social interactions. Thesedetails are complex and myriad, andthere is no well-agreed-upon way tocombine them. Issues of behavioralhistory are critical to social communi-cation, but the most important socialimplications of a behavioral historycan be communicated efficientlythrough the language of feeling andthinking. If a person says to you "I feellonely," it is likely that, whatever hishistory, your company would probablybe reinforcing to him.

Emotional talk is under the function-al control of a complex, fuzzy set ofevents including behavioral predispo-sitions, bodily sensations, contextualfeatures, and so on, and this talk seemsto be produced in part to alter the per-son's social environment. In that sense,human emotions are not specific thingsthat are to be discovered, contrary towhat most of psychology has assumed.Instead, emotions are constructed inthe process of the verbal communityteaching the child to tell others abouta set of historical facts and a currentcontext, information that is too com-plex and too cumbersome to recall indetail.

It is incorrect to think of verballyknown emotions in social terms, how-ever, without realizing the impact theyhave on the person feeling them. The


person feeling such complex emotionsas depression or anxiety does not ex-perience them as being socially con-structed under multiple sources ofstimulus control including a mix ofbodily states, historical facts, and cur-rent contextual cues, but as "really be-ing there." An illusion can be createdin which a person must deal with de-pression in much the same way that aperson must deal with a stone in his orher shoe or a bus rolling down thestreet towards him or her.

If emotions are fuzzy social-verbalconstructions, how are people taught todescribe (construct) them? Skinner'sfour methods cover the ground well,provided only that we augment his ac-count with the bidirectional and rela-tional nature of metaphor. Metaphorconsists of relating sets of relationsamong events. Through verbal meta-phor, a variety of subtle distinctionscan be made about private events.Complex sets of relations verbally con-structed in one domain in which eventsare quite public can be related to an-other, largely private domain in whichonly individual components are public-ly available. This can bring a set of re-lations to bear on material that is large-ly or wholly unavailable to the social-verbal community.

For example, suppose in therapy wesay "your relationship to anxiety islike being in a tug of war with a mon-ster. You don't need to win the war,you need to drop the rope." The met-aphor describes a set of relations in apublic domain, and gives a few linksto the private domain (e.g., the monsteris anxiety). Other events in the privatedomain need to be discovered or con-structed to fit the metaphor. It is notclear exactly what events correspond to"dropping the rope," for example. Inthe metaphor, a cessation of struggle isimplied. The person hearing the meta-phor may search for private responsesthat could individually or collectivelycorrespond to this aspect of the meta-phor in much the say way as the publicevents do. When the person later says"I've just learned to let go" the phrase

"let go" may refer to private actionsthat the public community would havea very difficult time describing, train-ing, or discriminating without the bi-directional nature of human language.

Talk about thoughts and feelings isunder multiple sources of control."Emotions" can be descriptions of pri-vate events, but what is described isoften a fuzzy set of events. For exam-ple, emotions are not merely bodilysensations, even though emotions mayinclude bodily sensations. Emotionaland cognitive talk are forms of dis-course that serve as predictions of fu-ture events, or can alter the social en-vironment through the persuasion ofothers or through an appeal to conven-tional reasons and explanations for be-havior. Private events conceived of thisway are not merely a coproduct of di-rect contingencies related to overt be-havior, but instead are the product ofboth direct and verbal contingencies.Their verbal nature often requires thatthey be understood as components ofcomplex instances of human behaviorin which direct contingencies do nottell the whole story, either pragmati-cally or descriptively, and must be aug-mented by the contingencies that op-erate in verbal behavior.

THE EXAMPLE OFDRUG DEPENDENCE

As a kind of extended example ofthese points, in this section we will ap-ply our analysis to a clinical problem.We have chosen drug dependence inpart because it presents such an enor-mous public health problem (Rice,Kelman, & Miller, 1991) and in partbecause some behavioral approaches todrug dependence illustrate the need forfurther understanding of the role ofthinking and feeling. We will developthis example gradually over severalsteps.

Traditional Behavioral Principles andDrug Dependence

Given the appropriate learning his-tory, stimuli that have no obvious for-


mal relation to a response can come toexert control over the response. Thus,a light in an operant chamber can cometo occasion lever pressing for food pel-lets, an example of discriminative con-trol. Also, pairings of some arbitrarilyselected stimulus with an establishedreinforcer can cause the arbitrary stim-ulus to become a conditioned reinforc-er. Further, reliable pairings of formallydifferent stimuli can also produce var-ious elicited responses, such as saliva-tion, which we know as classical con-ditioning.

These various direct conditioning ef-fects have been used to explain addic-tive behaviors, for example, drug crav-ing, or some forms of withdrawalsymptoms. Suppose a heroin addict ex-periences conditioned craving or with-drawal symptoms when he or she seesa car similar to the one driven by hisor her drug connection. A classicalconditioning analysis of this examplewould appeal to a history in whichdrugs were reliably paired with thepresence of the drug connection's car.This sort of effect can be reliably pro-duced in experiments with nonhumansubjects (e.g., Wickler & Pescor, 1967)and has been observed among detoxi-fied opiate addicts returning from treat-ment to their old neighborhoods(Wickler, 1977). Besides elicited re-sponding, a number of researchershave investigated discriminative (Bick-el & Kelly, 1988) and conditioned re-inforcing effects (Kelleher & Gold-berg, 1977) as they relate to addiction.One way to alter the effects of stim-

uli whose conditioning histories lead tothe maintenance of problematic druguse is to remove the stimuli them-selves. Some interventions, such astreatment communities, advise "relo-cation of the client to the residentialenvironment" in order to "avoid theusual settings and circumstances his-torically associated with drug abuse"(De Leon, 1988, p. 86; Washton,1988). The positive effects of reloca-tion can be seen among Vietnam-eraU.S. Army personnel. Those individu-als who began their drug use while

overseas had very low relapse to drugabuse upon returning home (Robins,Davis, & Goodwin, 1974). Animalstudies have also demonstrated this ef-fect. Rats that have been withdrawnfrom morphine show more rapid read-diction when they are returned to theenvironment where the addiction orig-inally occurred as opposed to a novelenvironment (Cushman, 1974).

Bickel and Kelly (1988) have sug-gested, though, that when drug-relatedstimuli are numerous and widely dis-tributed, removal from such stimulimay be impractical (DeGrandpre &Bickel, 1993; cf. Drummond, Cooper,& Glautier, 1990). Thus, other inter-vention strategies have sought to alterthe stimulus functions of existing en-vironmental cues as an alternative toremoving the actual stimuli.

Aversive counterconditioning hasbeen used to counteract the reinforcingproperties of various drug stimuli bypairing the stimuli with some aversiveevent, such as shock or chemically in-duced nausea (e.g., Cannon & Baker,1981). The goal of these treatments isto produce revulsion in the presence ofdrug-related stimuli, including the drugitself. Some evidence suggests thatsubjects who show greater conditionedaversion posttreatment had superioroutcomes at follow-up than did sub-jects with less aversion (Rimmele,Miller, & Dougher, 1989).

Besides counterconditioning, a vari-ety of research groups have pursuedprograms aimed at understanding theestablishment and reduction of variousforms of elicitation. Results of thiswork include an array of proceduresthat extinguish conditioned responses todrug cues (e.g., Dawe et al., 1993;McLellan, Childress, Ehrman, O'Brien,& Pashko, 1986; Monti et al., 1993;Rawe & Russell, 1980). Although sev-eral of these strategies have producedbetter outcomes than control conditions,the data are somewhat mixed (e.g.,Rawe & Russell, 1980). The lack ofclarity in the data is twofold. First,somewhat surprisingly, the treatmentsare not as powerful as animal models


might lead us to expect. Childress, Ehr-man, Rohsenow, Robbins, and O'Brien(1992) conclude that, for substanceabusers, "passive cue exposure (extinc-tion) effects can be demonstrated, butthe effects are modest" (p. 65; cf.Drummond et al., 1990).

Second, and even more perplexingthan the modest effects, the putativechange processes do not seem to fit theexisting data. A number of researchershave reported a failure of cue exposureto reduce significantly physiologicalresponses to drug stimuli (Childress,McLellan, Ehrman, & O'Brien, 1988;Drummond & Glautier, 1994). In stud-ies in which cue exposure does yieldimprovements in clinical outcomes, theprecise mechanisms of action are un-clear. These differences between whatanimal models of conditioning wouldlead us to expect and actual outcomeswith human subjects are not limited tosubstance abuse. Among phobics, forexample, changes in arousal levelsover the course of treatment do notpredict subsequent avoidance (Barlow,Leitenberg, Agras, & Wincze, 1969;Leitenberg, Agras, Butz, & Wincze,1971). As with extinction proceduresinvolving drug stimuli, the treatmentshave a positive effect, but change inconditioned arousal does not necessar-ily predict improvement, as it should ifthe processes are based entirely on di-rect conditioning. From a traditionalbehavioral viewpoint on emotion, thisseems problematic. If overt respondingand emotional reactions are coproductsof the same contingencies, we oughtnot to see major discrepancies betweenphysiological responses and overt be-havior, either among phobics or drugaddicts.The lack of a straightforward con-

nection between exposure, cue reactiv-ity, and subsequent relapse has led anumber of researchers to speculate onmediating variables such as improvedself-efficacy, the establishment of cop-ing strategies, or positive expectanciesabout the effects of the drug (e.g.,Drummond & Glautier, 1994; Marlatt,1990; Marlatt & Gordon, 1985; Monti,

Rohsenow, Abrams, & Binkoff, 1988).Other researchers have suggested thatconditioned responses may be sensitiveto emotional states. Childress et al.(1994), for example, suggest that be-sides actual drug stimuli, mood statessuch as depression and anger may alsoalter the functions of antecedent andconsequent stimuli. As evidence, theseresearchers found that induction of cer-tain mood states reliably increased opi-ate craving and withdrawal-like symp-toms upon exposure to drug cues.

Simple appeal to private events suchas emotions, expectancies, self-effica-cy, and the like is problematic to be-havior analysts for at least three rea-sons. First, we do not know the pro-cesses whereby these cognitive andemotional responses exert behavior-regulatory functions over subsequentovert responding. At best these are be-havior-behavior relations to be contex-tually interpreted, not accepted as orig-inating causal events (Hayes & Brown-stein, 1986). Behavior analysts rejectthe use of dependent variables to ex-plain dependent variables (especiallygiven the goal of behavioral influence,not mere prediction), and all forms ofresponding (including emotional andcognitive responding) are the depen-dent variables of behavior analysis(Biglan & Hayes, 1996).

Second, private events do not seemto affect overt responding in any sim-ple mechanical way (Hayes & Wilson,1995). Instead, the behavior-behaviorrelation appears to be under contextualcontrol. Arousal in the presence ofdrug stimuli is not readily distin-guished, at the level of physiology,from other sorts of arousal. Salivation,sweating, anxiety, and other such bodi-ly states are functionally differentthings for different persons in differentcontexts. Arousal on a roller coaster is"fun." Among drug addicts, however,high arousal may mean "I need to getto a Narcotics Anonymous meeting"for one person, whereas it might mean"treatment isn't working so I might aswell get high" for another. In either in-stance, this arousal is not "fun." In


terms of behavioral outcomes thearousal has markedly different func-tions. The roller coaster is approached(more tickets are purchased), whereasthe situations generating arousalamong the addicts are avoided (eitherby going to a meeting or by using).

Finally, most psychologists useemotion or thought to explain overt be-havior in an inherently mentalisticfashion. Among self-efficacy theorists,for example, thinking that one may beeffective at accomplishing a task issaid to cause one to be effective in ac-complishing that task (Bandura, 1995).These common theoretical problemsdo not mean, however, that a behav-ioral analysis of the role of privateevents must likewise be causal, mech-anistic, and mentalistic (Hayes &Brownstein, 1986; Hayes & Wilson,1995). Behavior analysts also need not,and ought not, ignore private events onthe basis of their being interesting, butirrelevant, epiphenomena (Forsyth &Eifert, 1996). Approaches exist that fitbetter within behavior analysis and thattake into consideration both the directeffects of contingencies and the indi-rect effects of verbal contingencies in-volved in cognition and emotion.

A Relational Stimulus FunctionView of Relapse

To illustrate the role of thoughts andfeelings in instances of relapse to drugdependence, we will begin by consid-ering a nonclinical example. We willuse the language of relational frametheory, but will not defend that usehere (for an exposition, see Hayes &Barnes, 1998; Hayes, Gifford, & Wil-son, 1996; Hayes & Wilson, 1993,1996). The strength and type of re-sponding occasioned by a stimulusevent that has its psychological func-tions as a result of relational respond-ing are due to both history and currentcontext. Consider the simple frame ofcoordination between the word car andactual cars. First, imagine hearing thewords "Visualize a car speeding to-ward you," spoken in a normal tone of

voice, while standing in your kitchen(see Context 1, Figure 1). In this ex-ample, the word visualize serves a dis-criminative function for the operant ofseeing in the absence of the thing seen(Skinner, 1957). In the context of theword visualize, the word car will makepsychologically present the visual fea-tures of an actual car. The words to-ward you would make present the vi-sual features of the front of a car. Theword speed would select for some ofthe stimulus features that distinguishstationary from speeding cars. Ourpsychological history with respect tothese verbal and nonverbal events arenot thoroughly compartmentalized bycontextual cues. We may do more than"see"" when asked to visualize. Be-cause we have a history in which see-ing speeding cars is accompanied bythe sounds that speeding cars make, wemay "hear" a roaring engine. If we areparticularly adept at imagining, wemay even experience some weak vis-ceral effects of seeing an approachingcar, such as slightly increased muscletension. It would be unlikely, however,to observe the emergence of any overtbehaviors, such as leaping about thekitchen in order to avoid the "speedingcar." In other words, the current non-imagined environment is still exertingconsiderable control over behavior.

Consider next the same frame of co-ordination in a different context. Imag-ine hearing the words "Watch out forthat car!" spoken in an urgent tone andhigh volume while standing in the mid-dle of an intersection (see Context 2,Figure 1). In this context, the tone andvolume of the speaker's voice, alongwith our physical location, would bringto bear a full array of psychologicalfunctions with respect to the word car.We might show many of the visceraland overt behaviors that would be oc-casioned by seeing an actual car. Forexample, we might experience in-creased muscle tension, respiration,and heart rate, and might leap from theroadway.

Finally, imagine hearing the words"Watch out for that car!" spoken again


FLAME -OFCOORDINArtIONI

"Car" < ,, __

Coniextual cues controllng which Psychologica tctios that arefuncftons of "car" are transforned transformed

Visual functions transformed andContext #1 pobiy"some *e4kE sformation

"Visualize a ca Event occurs i the kthen, voice piblome wea tsforionotvisderal functions (seeing thespeeding toward volume and tone are consistentIwipth car, perhaps someyou." conversation muscles no ove

(kitchen) itensis* Ofmuscles, no overtbeha.vi .._fun:ctions)

Visuial, visceral, and overt behavioralfiuntions tao , (activation of

a wide rietybodily stAtes,muscle tension, increased resiration,

Context fl Event occ in the street, volume increased heart rat; altered"Watch Cut for andtnie are consistent with wamings peipheral blod flw,runing

that car."I(intersction) ofenvfronmental dagers towad the roadsi, isuia scanning

ofthe environmet, many overtbehavioral fimctions present inavoiding an adtua speeding car

would be psych6logpally present)

Cognitive funcions transformed(e.g., thoughts about ars, such as"huh, there aren't -ay c here,"Event occurs in the kitchen, voice Qontext of iinc ous contexts leads

Context #3 volume and tone ae consistent with to transformation of cogtive"Watch out for warnings ofeenvironinenal cIan|crs, functions about-spak.er "Is he nuts?,"

that cr." (hst6ryand beaviora '~'~ few functions' of actual cars would beQdtchen) withrespect to this event in this pres no overt behairal

- ~~~~~~~~~~present,no overt behavimralcontext would be nonexistent) fti s s im..s,.,,, ........t, -fions, possibl-y somdei. Aiatedand transient visceral fimctions owing

to the volume and tone)

Figure 1. Contextual features controlling which stimulus functions of actual cars will be presentupon hearing the word car.

in an urgent and loud voice, but oc-curring in the kitchen (see Context 3,Figure 1). This example presents a mixof contexts with which we are unlikelyto have had any prior experience. Thetone of voice alone might cause some

highly transient occasioning of alarmreactions, such as muscle tensing andincreased vigilance. However, we areunlikely to have an elaborate behavior-al repertoire established in response tothese words in this context. In this in-


stance, the only psychological func-tions likely to be present as a result ofhearing the word car would bethoughts about cars, such as, "therearen't any cars here." We would notexpect to see any of the overt behav-iors seen in the previous example, andwe would expect only a few brief pri-vate responses.

The Context of Literality

The social-verbal community estab-lishes and maintains the verbal rela-tions involved in literal meaning. Insome contexts an individual respondsto an event that is part of a derivedstimulus relation with another event byemitting many of the responses wewould expect to see if the second eventwere actually present. We have termedthe social-verbal context that generatessuch behavior the context of literality(Hayes, Strosahl, & Wilson, 1999;Hayes & Wilson, 1994). The extent towhich we see the emergence of psy-chological function through these re-lations is a matter of degree, so wemay salivate when we hear the words"imagine the taste of a lemon," but wedo not eat the word lemon. And, asseen in the car examples above, whichpsychological functions transfer is un-der contextual control. All three of thecar examples involved the context ofliterality, in that when the listenerheard the word car, more than the sim-ple auditory functions of that wordwere present. However, in Contexts 1and 3, the listener's behavior was muchmore under the control of events otherthan the verbal stimulus car. In Con-text 2, by contrast, the contextual con-ditions conspire to bring the listener'sbehavior almost completely under thecontrol of the word car. It is as if thecar itself had been sensed (seen, heard,etc.).

Adaptive features of the context ofliterality. The context of literality doesnot produce all of the responses to anevent literally described by a verbalterm, but it does occasion many ofthem. As in the example provided

above, a verbally competent humancan respond to the verbal stimulus"Watch out for that car!" much as ifthey were responding to an actual car.In this situation, rapid responding isnecessary in order to avoid danger. Ifthere were actually a car speeding to-ward you, we would want a full arrayof potential responses immediatelyavailable. Verbal stimuli are effectivein part because of this transfer of stim-ulus functions from the events to whichwords are related, and the words them-selves, in this case, between the wordsspeeding car and the actual speedingcar. Skinner proposed the term rule-governed behavior precisely becausecertain human behaviors can come un-der the control of what Skinner calleddefective contingencies, that is, contin-gencies that would be unlikely to oc-casion adaptive behavior in and ofthemselves. Some examples of defec-tive contingencies include those thatare too remote (e.g., getting a PhD),small, or only cumulatively potent(e.g., the health consequences of smok-ing), or as in the current example ofthe oncoming car, not directly discrim-inable by the person to be affected.When the words are ones like "worklong and hard and you will get yourdoctorate," or "don't smoke, it willkill you," or "watch out for that car,"the more potent the functions exertedby the words, the more likely we willbehave effectively. Humans generate alot of words, though, and respondingto some of them as if they were "thereal thing" can get us into a lot of trou-ble.

The Context of Social Regulation andReason Giving

A second context that knits togetherbehavior-behavior relations betweencognition and emotion on the one handand more overt behaviors on the otheris the context of social regulation andreason giving. The verbal communityteaches us to talk about our histories inthe shorthand language of wants, wish-es, desires, dispositions, thoughts, and


memories. If "good reasons" are givenfor behavior, the social communitymay alter how a given instance of be-havior is treated. The verbal commu-nity demands a certain level of corre-spondence, however, between the lan-guage of thoughts and feelings and ex-pected patterns of overt behavior, andprovides consequences for the mainte-nance of this correspondence. If, forexample, someone explains their drugrelapse by saying, "I was very de-pressed," they will be thought to havesaid something sensible. Having a''good reason" for the relapse is likelyto generate some sympathy and to less-en socially imposed negative conse-quences. If, however, there are not oth-er overt signs of depression, the personexpressing such a reason may eventu-ally receive negative consequences forhaving "manipulated" the situation orhaving "lied."Of course, the correspondence be-

tween thoughts and emotions on theone hand and overt behavior on theother is not perfect. Thoughts and feel-ings are not literally the causes of be-havior. They participate as componentsof complex behavioral patterns that arein part maintained by socially mediatedcontingencies (e.g., social reinforce-ment for say-do correspondence).Thus, the correspondence betweenthoughts and feelings and overt actionis only partial.

The Example of Relapse toSubstance Abuse

The fact that stimuli can acquirepsychological functions as a result oftheir participation in relational respons-es can illuminate the role of thoughtsand feelings in relapse and also the ef-fects of a variety of treatments. To il-lustrate the role of relational stimulusfunction in relapse, consider five rela-tional frames that might be likely for adrug-dependent individual (see Figure2).Frame 1. Actual abstinence is likely

to be in a frame of coordination withthe word abstinent.

Frame 2. Abstinence (both actualand verbal) is likely to be in an "if ...then" relational frame with variousconsequences (both actual and verbal).Frame 3. Some of these conse-

quences (both actual and verbal) arelikely to be in a frame of coordinationwith the verbal label "having my lifetogether."Frame 4. Abstinence (both actual

and verbal) is likely to be in a frameof opposition with drug use (both ac-tual and verbal).Frame 5. Unbearable cravings (ver-

bal) are likely to be in an "if ... then"frame with eventual relapse (actual andverbal).

Such a set of stimulus relations es-tablished by the verbal community pro-vides a rich mix in which complextransformation of psychological func-tion may occur. For example, imaginean individual who, while at work, is ina context that occasions the thought"Wow, I've really quit using." This in-stance of verbally described abstinenceis in an "if ... then" frame with actualand verbally described reinforcers(Frame 2 above), and may begin to ac-quire reinforcing effects because ofthat relation. Just as a countdown stim-ulus that predicts the arrival of a rein-forcer will eventually itself come to bereinforcing, behaviors that support ab-stinence and abstinence thoughts can,through these derived relations, cometo have reinforcing functions. As a re-sult, an addict may work for the op-portunity to engage in behaviors thatmake abstaining (both actual and asverbally described) more likely. In ef-fect, it becomes reinforcing to think"I've really quit using," and extra ef-fort may be expended (e.g., getting aride to a Narcotics Anonymous meet-ing) to be in circumstances in whichsuch a verbalization is likely.

Likewise if actual drug use and ac-tual abstinence are in a frame of op-position (Frame 4), actual drug usewould be in an "if ... then" framewith the absence of these reinforcers.The formerly positive stimulus func-tions of drug use may be transformed


FRAME OF COQRDINATION

Actual Abstinence

I1F...TIJEN FRAME

FRM FR OPPOIflO

-~~~~~~~~Ata RAifrctulrsin

ActualAbstinen

"I can'tstandtheeI I

_:~~~FRM OFPRAM#3(OIN

(thought) Actu~~~~~~~~~~~~Alvetal Useng

'I ca -st-nd 3me

Figure 2. Examples of relational frames involving overt behaviors, stimulus events, thoughts,emotions, and bodily states as might be seen in a recovering drug addict.

into negative functions by the combi-nation of Frames 2 and 4 for those per-sons who experience the positive im-pact of actual and verbal abstinence.

This transformation of function can bequite adaptive, and may support the ad-dict in forgoing immediate reinforce-ment provided by drug use (and likely


present in thoughts of drug use) in lieuof the more distal and slowly accu-mulating reinforcers available for notusing. This clinical example is similarto those described in the analysis ofself-control provided at the beginningof this paper, and are implied by Skin-ner (1974) in his quotation about thevalue of self-knowledge.

Unfortunately, verbal events canalso support maladaptive behavior.This can happen in many ways. Verbalevents and actual events are not thesame thing, and the addict who isspeaking of successful abstinence mayalso be engaging in small behavioralsteps that will undermine abstinence,but initially not so dramatically as tochallenge the verbal construction ofsuccessful abstinence. For example, theperson who says "Wow, I've reallyquit using" may then begin to spendtime with old, using friends, thus in-creasing the risk of relapse though ex-posure to direct contingencies. Theperson may not have self-knowledge ofthese behavioral changes and theirrisks, either because the actions are notverbally categorized as behaviors thatwill undermine abstinence or, perverse-ly, because negative behaviors areframed positively (e.g., "My oldfriends are no longer a threat now thatI've quit," or "If I've really quit I needto stop running away from situationslike my old friends.").

Another kind of verbal support forrelapse comes when events interveneto alter a verbal relation that was sup-porting successful withdrawal fromdrug use. For example, suppose an ab-stinent drug addict is arrested for anold drug felony. This might occasionthoughts such as "I'm going to loseeverything I've worked for," or "Itdoesn't matter what I do or how hardI try." Such verbalizations could re-flect and support a diminution of the"if ... then" relation between absti-nence and reinforcement in Frame 2. Ifabstinence (in a verbal sense) no longerseemingly predicts positive outcomes,then the verbal supports for abstinencewill weaken.

An additional kind of verbal supportfor relapse occurs when verbal rela-tions seemingly explain or justify druguse. For example, the statement "Ican't stand these cravings" could tem-porarily augment the possible verballyconstructed reinforcers for abstinence,or could marshal some degree of socialsympathy for steps that make drug usemore likely.

It is worth noting that these trans-formations of stimulus function are notcaused by verbalizations, includingself-verbalizations. Rather, the verbali-zation reflects and instantiates a stim-ulus relation that emerges from a spe-cific history, occurs in particular envi-ronmental contexts, and has particularpsychological effects due to these his-torical and situational events.

Implications for UnderstandingTreatment Strategies

Given this analysis, several inter-ventions could be effective. One ap-proach would be to alter the stimulusfunctions of certain key members of re-lational networks. Both cue exposureand aversive conditioning provide ex-amples of this strategy. In fact, the ef-fectiveness of many cue exposure,aversive conditioning, and other expo-sure-based treatments is more under-standable given bidirectional transfor-mation of stimulus function (includingextinction functions; see Dougher etal., 1994). After all, treatments such ascovert sensitization (Rimmele et al.,1989) may rely solely on verbally con-structed images rather than on any di-rect exposure to alcohol or drugs.The use of antidipsotropic medica-

tions in the treatment of alcohol like-wise alters the stimulus functions ofthe drug, though again, the alterationof the functions of alcohol is likely tobe verbal, because antidipsotropics areoften effective even if the person neverdrinks alcohol while taking them. Thephysician puts the verbal event "alco-hol" in an "if ... then" frame with theverbal event "becoming deathly ill."The physician need not expose the pa-


tient to actual alcohol while describingthe effects of the combination, nordoes he or she need to have the personexperience actual illness. In all likeli-hood, the drug will be effective whenonly verbal "alcohol" and verbal "ill-ness" are presented. Transformation ofstimulus function takes care of the in-dividual's responding to the actualevents.Some of these treatments produce

change by intervening on the anteced-ent end of addictive behaviors. That is,they enter the relational network ver-bally or in the environment by alteringthe stimulus functions of the drug. Bycontrast, motivational interviewing(Miller & Rollnick, 1991) likely has itsmain effects by altering the stimulusfunctions on the consequential end ofthe addictive behavior contingency.Motivational interviewing has showngood effects in a recent clinical trialamong alcoholics (Project Match Re-search Group, 1997).

Motivational interviewing explicitlyforbids direct delivery of social pun-ishment of addictive behavior, but in-stead focuses on making psychologi-cally more present the relation betweenthe client's values and his or her ad-dictive behavior. The therapist does nottell the client what to value and howaddictive behavior is blocking thosevalues. Instead, the motivational inter-viewer tries to set conditions in whichthe client describes the valued goalsand the ineffectiveness of addictive be-havior in reaching those goals. Anyself-control, however, that emergesfrom discriminating contingencies thathave been operating necessarily re-quires bidirectional transformation ofstimulus function, as we have argued.A sensible behavioral approach,

based on the present analysis, is to at-tempt to alter the social-verbal contin-gencies that support the behavior-reg-ulatory functions of private events. Ourown approach, called Acceptance andCommitment Therapy (ACT; Hayes etal., 1999; Hayes & Wilson, 1993,1994, 1995), takes exactly that tack.ACT explicitly attacks not just the

products of literality (particular psy-chological functions present in re-sponse to a word) but also the contextof literality itself. That is, ACT at-tempts, through a variety of means, todisrupt the process of relational re-sponding and the transformation ofstimulus function by altering the con-texts in which these occur. For exam-ple, suppose a person has the thought"I can't stand these cravings." In anormal, literal context this thoughtmay have emotional functions (e.g.,agitation, upset) that could be allevi-ated by drug use. If, however, thethought "I can't stand these cravings"is said 100 times rapidly, these emo-tional functions subside, and the direct,auditory functions of the words them-selves become more salient. If the per-son visualized the words as if theywere written on a leaf, and watchedthem float by in a meditative practice,much the same thing might occur. ACTuses a variety of such techniques intherapy to manipulate the context ofliterality itself (in ACT these aretermed deliteralization or defusiontechniques).ACT has been shown to have wide-

ranging clinical effects (e.g., Strosahl,Hayes, Bergan, & Romano, 1998), anddata suggest that deliteralization is oneof the processes involved. For exam-ple, among depressed clients ACT pro-duces a rapid decrease in the literal be-lievability of supposedly depressogenicthoughts and a slower decrease in theiroccurrence and frequency, whereascognitive therapy shows the oppositepattern (Zettle & Hayes, 1986; Zettle& Raines, 1989). We argue that theseeffects add to accumulating evidencethat the impact of self-verbalizations iscontextually determined, as it shouldbe in a behavior-analytic model. Weare currently examining whether ACTaffects substance abuse in a large fed-erally funded trial, and preliminarydata are supportive (Wilson, Hayes,Gifford, & Chang-Judson, 1999). Con-trolled research has also shown thatACT aids in alleviating stress, anxiety,and depression (Bond & Bunce, in


press; Strosahl et al., 1998; Zettle &Raines, 1989), which are empiricalcorrelates of drug abuse.

CONCLUSION

In some respects this analysis is outof the ordinary for behavior-analyticliterature. For example, the above anal-ysis of relapse could be construed assuggesting that people relapse becauseof thoughts about relapse. What distin-guishes this account from a more tra-ditional mentalistic account is that ob-servable behavior (albeit sometimesobservable by only 1 subject) is beingrelated to independent variables foundin the environment, which is, in prin-ciple, manipulable. Behavior-behaviorrelations exist between thinking andfeeling (both of which, we argue, areverbal events for humans) and otherforms of behavior, but this relation isitself due to manipulable contextualvariables (Hayes & Brownstein, 1986;Hayes & Wilson, 1995).We have attempted to specify vari-

ous independent variables of whichthese behavior-behavior relations are afunction. For example, we have iden-tified the social context in which talkabout cognition and emotion alters so-cially mediated contingencies: the con-text of social regulation and reasongiving. We have tried to show that pri-vate events in humans are often verbalevents that depend upon a context ofliterality for their meaning and impact(i.e., a social-verbal community to es-tablish and maintain derived stimulusrelations).

Basic behavior-analytic laboratoriesare beginning to show how such verbalprocesses operate in clinically signifi-cant areas. For example, DeGrandpreand Bickel (1993) have shown that in-teroceptive stimuli resulting from drugingestion can participate in equiva-lence relations with arbitrary visualstimuli, and that discriminative controlover drug consumption can transferfrom drug-related to arbitrary visualstimuli via participation in equivalenceclasses (DeGrandpre, Bickel, & Hig-

gins, 1992). We also know that oldequivalence classes (and potentiallyold stimulus functions) may reemergeunder some conditions (Wilson &Hayes, 1996), and that attempts to sup-press thoughts and feelings about sub-stance-abuse-related words actuallymakes equivalence classes containingsubstance-abuse-related members moreresistant to disruption (Wilson, 1998).Little stands between such sets of find-ings and a functional account of a ver-bally constructed emotion in drug in-gestion, such as we have providedhere.

Understanding verbal process mayhelp us to make sense of some currenttreatments as well. For instance, cueexposure necessarily involves directexposure and subsequent extinction ofelicited responses to some, but not all,drug cues (e.g., Drummond et al.,1990; Monti et al., 1993). If some drugcues have their stimulus functionsthrough derived processes, not simplythrough direct training histories (e.g.,see DeGrandpre & Bickel, 1993;DeGrandpre et al., 1992), studies suchas Dougher et al. (1994), showingtransfer of extinction functions throughequivalence classes, provide somehope that these procedures could be ef-fective. However, findings by Pilgrimand Galizio (1990, 1995) suggest thatwe may also see changes in stimuluscontrol for some members of a classbut not others.The empirical analysis of thought

and emotion is entering a new era inbehavior analysis. Direct conditioningaccounts of emotion and thought couldonly go so far because of the verbalnature of human private events. De-rived stimulus relations provide bothan experimental avenue into the lion'sden of human thought and emotion anda signpost indicating why entering isimportant. Philosophically, behavioranalysis has been prepared to take thisstep for more than 50 years (Skinner,1945/1972). Empirically, it seems timeto proceed.


REFERENCES

Bandura, A. (1995). Comments on the crusadeagainst causal efficacy of human thought.Journal of Behavior Therapy and Experimen-tal Psychiatry, 26, 179-190.

Barlow, D. H., Leitenberg, H., Agras, W. S. &Wincze, J. P (1969). The transfer gap in de-sensitization: An analogue study. BehaviorResearch and Therapy, 7, 191-196.

Bentall, R. P., & Lowe, C. F (1987). The roleof verbal behavior in human learning. IH: In-structional effects in children. Journal of theExperimental Analysis of Behavior, 47, 177-190.

Bickel, W. K., & Kelly, T. H. (1988). The re-lationship of stimulus control to the treatmentof substance abuse. In B. A. Ray (Ed.), Learn-ing factors in substance abuse (pp. 122-140)(NIDA Monograph No. 84). Washington, DC:U.S. Government Printing Office.

Biglan, A., & Hayes, S. C. (1996). Should thebehavioral sciences become more pragmatic?The case for functional contextualism in re-search on human behavior. Applied and Pre-ventive Psychology: Current Scientific Per-spectives, 5, 47-57.

Bond, F W., & Bunce, D. (in press). Mediatorsof change in emotion-focused and problem-focused worksite stress management interven-tions. Journal of Occupational Health Psy-chology.

Cannon, D. S., & Baker, T. B. (1981). Emeticand electric shock alcohol aversion therapy:Assessment of conditioning. Journal of Con-sulting and Clinical Psychology, 49, 20-33.

Childress, A. R., Ehrman, R., McLellan, A. T.,MacRae, J., Natale, M., & O'Brien, C. P(1994). Can mood states trigger drug-relatedresponses in opiate abuse patients? Journal ofSubtance Abuse Treatment, 11, 17-23.

Childress, A. R., Ehrman, R., Rohsenow, D. J.,Robbins, S. J., & O'Brien, C. P. (1992). Clas-sically conditioned factors in drug depen-dence. In J. H. Lowinson, P. Ruiz, R. B. Mill-man, & J. G. Langrod (Eds.), Substanceabuse: A comprehensive textbook (2nd ed., pp.56-69). Baltimore: Williams & Wilkins.

Childress, A. R., McLellan, A. T., Ehrman, R.,& O'Brien, C. P (1988). Classically condi-tioned responses in opiod and cocaine depen-dence: A role in relapse? Learning factors insubstance abuse (NIDA Monograph No. 84).Washington, DC: U.S. Government PrintingOffice.

Cushman, P (1974). Detoxification of rehabili-tated methadone patients: Frequency and pre-dictors of long-term success. American Jour-nal of Drug and Alcohol Abuse, 1, 393-408.

Dawe, S., Powell, J., Richards, D., Gossop, M.,Marks, I., & Gray, J. A. (1993). Does post-withdrawal cue exposure improve outcome inopiate addiction? A controlled trial. Addiction,88, 1233-1245.

DeGrandpre, R. J., & Bickel, W. K. (1993).

Stimulus control and drug dependence. Psy-chological Record, 43, 651-666.

DeGrandpre, R. J., Bickel, W. K., & Higgins, S.T (1992). Emergent equivalence relations be-tween interoceptive (drug) and exteroceptive(visual) stimuli. Journal of the ExperimentalAnalysis of Behavior, 58, 9-18.

De Leon, G. (1988). The therapeutic commu-nity and behavioral science. Learning factorsin substance abuse (NIDA Monograph No.84). Washington, DC: U.S. Government Print-ing Office.

de Rose, J. C., de Souza, D. G., Rossito, A. L.,& de Rose, T. M. (1994). Stimulus equiva-lence and generalization in reading aftermatching to sample by exclusion. In S. C.Hayes & L. J. Hayes (Eds.), Understandingverbal relations (pp. 69-82). Reno, NV: Con-text Press.

Devany, J. M., Hayes, S. C., & Nelson, R. 0.(1986). Equivalence class formation in lan-guage-able and language-disabled children.Journal of the Experimental Analysis of Be-havior, 46, 243-257.

Dougher, M. J., Augustson, E., Markham, M. R.,Greenway, D. E., & Wulfert, E. (1994). Thetransfer of respondent eliciting and extinctionfunctions through stimulus equivalence clas-ses. Journal of the Experimental Analysis ofBehavior, 62, 331-351.

Drummond, D. C., Cooper, T., & Glautier, S. P(1990). Conditioned learning in alcohol de-pendence: Implications for cue-exposure treat-ment. British Journal of Addictions, 85, 725-743.

Drummond, D. C., & Glautier, S. (1994). Acontrolled trial of cue exposure treatment inalcohol dependence. Journal of Consultingand Clinical Psychology, 62, 809-817.

Dymond, S., & Barnes, D. (1995). A transfor-mation of self-discrimination response func-tions in accordance with the arbitrarily appli-cable relations of sameness, more than, andless than. Journal of the Experimental Anal-ysis of Behavior, 64, 163-184.

Forsyth, J. P., & Eifert, G. H. (1996). The lan-guage of feeling and the feeling of anxiety:Contributions of the behaviorisms toward un-derstanding the function-altering effects oflanguage. Psychological Record, 46, 607-649.

Friman, P. C., Hayes, S. C., & Wilson, K. G.(1998). Why behavior analysts should studyemotion: The example of anxiety. Journal ofApplied Behavior Analysis, 31, 137-156.

Friman, P. C., Wilson, K. G., & Hayes, S. C.(1998). Behavior analysis of private events ispossible, progressive, and nondualistic: A re-sponse to Lamal. Journal ofApplied BehaviorAnalysis, 31, 707-708.

Hall, J. E (1984). Backward conditioning inPavlovian type studies: Reevaluation and pre-sent status. Pavlovian Journal of BiologicalSciences, 19, 163-168.

Hayes, S. C., & Barnes, D. (1998). Analyzingderived stimulus relations requires more than


the concept of stimulus class. Journal of theExperimental Analysis of Behavior, 68, 235-244.

Hayes, S. C., & Brownstein, A. J. (1986). Men-talism, behavior-behavior relations and a be-havior analytic view of the purposes of sci-ence. The Behavior Analyst, 9, 175-190.

Hayes, S. C., Brownstein, A. J., Devany, J. M.,Kohlenberg, B. S., & Shelby, J. (1987). Stim-ulus equivalence and the symbolic control ofbehavior. Mexican Journal of Behavior Anal-ysis, 13, 361-374.

Hayes, S. C., & Gifford, E. V. (1997). The trou-ble with language: Experiential avoidance,rules, and the nature of verbal events. Psycho-logical Science, 8, 170-173.

Hayes, S. C., Gifford, E. V., & Wilson, K. G.(1996). Stimulus classes and stimulus rela-tions: Arbitrarily applicable relational re-sponding as an operant. In T. R. Zentall & PM. Smeets (Eds.), Stimulus class formation inhumans and animals (pp. 279-299). NewYork: Elsevier.

Hayes, S. C., & Hayes, L. J. (1992). Verbalrelations and the evolution of behavior anal-ysis. American Psychologist, 47, 1383-1395.

Hayes, S. C., Kohlenberg, B. K., & Hayes, L. J.(1991). Transfer of consequential functionsthrough simple and conditional equivalenceclasses. Journal of the Experimental Analysisof Behavior, 56, 119-137.

Hayes, S. C., Strosahl, K., & Wilson, K. G.(1999). Acceptance and commitment therapy:An experiential approach to behavior change.New York: Guilford.

Hayes, S. C., & Wilson, K. G. (1993). Someapplied implications of a contemporary be-havior-analytic account of verbal events. TheBehavior Analyst, 16, 283-301.

Hayes, S. C., & Wilson, K. G. (1994). Accep-tance and commitment therapy: Underminingthe verbal support for experiential avoidance.The Behavior Analyst, 17, 289-303.

Hayes, S. C., & Wilson, K. G. (1995). The roleof cognition in complex human behavior: Acontextualistic perspective. Journal of Behav-ior Therapy and Experimental Psychiatry, 26,241-248.

Hayes, S. C., & Wilson, K. G. (1996). Criti-cisms of relational frame theory: Implicationsfor a behavior analytic account of derivedstimulus relations. Psychological Record, 46,221-236.

Home, P. J., & Lowe, F C. (1996). On the or-igins of naming and other symbolic behavior.Journal of the Experimental Analysis of Be-havior, 65, 185-24 1.

Kelleher, R. T, & Goldberg, S. R. (1977).Fixed-interval responding under second-orderschedules of food presentation or cocaine in-jection. Journal of the Experimental Analysisof Behavior, 28, 221-231.

Lamal, P A. (1998). Advancing backwards.Journal of Applied Behavior Analysis, 31,705-706.

Leitenberg, H., Agras, W. S., Butz, R., & Win-

cze, J. P. (1971). Relationship between heartrate and behavioral change during treatmentof phobias. Journal of Abnormal Psychology,78, 59-68.

Lipkens, G., Hayes, S. C., & Hayes, L. J.(1993). Longitudinal study of derived stimu-lus relations in an infant. Journal of Exper-mental Child Psychology, 56, 201-239.

Lipkens, R. (1992). A behavior analysis ofcom-plex human functioning: Analogical reason-ing. Unpublished doctoral dissertation, Uni-versity of Nevada, Reno.

Mackintosh, N. J. (1974). The psychology ofanimal learning. New York: Academic Press.

Marlatt, G. A. (1990). Cue exposure and relapseprevention in the treatment of addictive be-haviors. Addictive Behaviors, 15, 395-399.

Marlatt, G. A., & Gordon, J. R. (Eds.). (1985).Relapse prevention: Maintenance strategies inthe treatment of addictive behaviors. NewYork: Brunner/Mazel.

McLellan, A. T., Childress, A. T., Ehrman, R.,O'Brien, C. P., & Pashko, S. (1986). Extin-guishing conditioning responses during opiatedependence treatment: Turning laboratoryfindings into clinical procedures. Journal ofSubstance Abuse, 3, 33-40.

Miller, W. R., & Rollnick, S. (1991). Motiva-tional interviewing: Preparing people tochange addictive behavior. New York: Guil-ford.

Monti, P M., Rohsenow, D. J., Abrams, D. B.,& Binkoff, J. A. (1988). Social learning ap-proaches to alcohol relapse: Selected illustra-tions and implications. In B. A. Ray (Ed.),Learning factors in substance abuse (pp. 141-159). Washington, DC: U.S. GovernmentPrinting Office.

Monti, P M., Rohsenow, D. J., Rubonis, A. V.,Niaura, R. S., Sirota, A. D., Colby, S. M., &Abrams, D. B. (1993). Alcohol cue reactivi-ty: Effects of detoxification and extended ex-posure. Journal of Studies on Alcohol, 54,235-245.

Pennebaker, J. W. (1997). Writing about emo-tional experiences as a therapeutic process.Psychological Science, 8, 162-166.

Pilgrim, C., & Galizio, M. (1990). Relations be-tween baseline contingencies and equivalenceprobe performances. Journal of the Experi-mental Analysis of Behavior, 54, 213-224.

Pilgrim, C., & Galizio, M. (1995). Reversal ofbaseline relations and stimulus equivalence: I.Adults. Journal of the Experimental Analysisof Behavior, 63, 225-238.

Project Match Research Group. (1997). Match-ing alcoholism treatments to client heteroge-neity: Project Match posttreatment drinkingoutcomes. Journal of Studies on Alcohol, 58,7-29.

Rachlin, H., & Green, L. (1972). Commitment,choice, and self-control. Journal of the Exper-imental Analysis of Behavior, 17, 15-22.

Rawe, M., & Russell, M. A. H. (1980). Rapidsmoking, cue exposure and support in the


modification of smoking. Behaviour Researchanid Thzerapy, 18, 363-372.

Rice, D. P, Kelman, S., & Miller, L. S. (1991).Economic costs of drug abuse. In W. S. Cart-wright & J. M. Kaple (Eds.), Economic costs,cost-effectiveness, finanacing, and community-based drug treatnent. Washington, DC: U.S.Government Printing Office.

Rimmele, C. T., Miller, W. R., & Dougher, M.J. (1989). Aversion therapies. In R. K. Hester& W. R. Miller (Eds.), Handbook of treatmentapproaches: Effective alternatives (pp. 128-140). New York: Pergamon Press.

Robins, L. N., Davis, D. H., & Goodwin, D. W(1974). Drug use by U.S. Army enlisted menin Vietnam: A followup upon their returnhome. American Journal ofEpidemiology, 99,235-249.

Roche, B., & Barnes, D. (1997). A transfor-mation of respondently conditioned stimulusfunction in accordance with arbitrarily appli-cable relations. Journal of the ExperimentalAnalysis of Behavior, 67, 275-301.

Shimp, C. P (1981). The local organization ofbehavior: Discrimination of and memory forsimple behavioral patterns. Journal of the Ex-perimental Analvsis ofBehavior, 36, 303-315.

Shimp, C. P (1982). On metaknowledge in thepigeon: An organism's knowledge about itsown behavior. Aniimal Learning & Behavior,10, 358-364.

Sidman, M. (1971). Reading and auditory-vi-sual equivalences. Journal of Speech andHearinig Research, 14, 5-13.

Skinner, B. F (1953). Science and human be-havior. New York: Macmillan.

Skinner, B. F ( 1957). Verbal behavior. NewYork: Appleton-Century-Crofts.

Skinner, B. F (1972). The operational analysisof psychological terms. In Cumulative record:A selection of papers. New York: Appleton-Century-Crofts. (Original work published1945)

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

Steele, D. L., & Hayes, S. C. (1991). Stimulusequivalence and arbitrarily applicable rela-tional responding. Journal of the Experimen-tal Analvsis of Behavior, 56, 519-555.

Strosahl, K. D., Hayes, S. C., Bergan, J., & Ro-mano, M. N. (1998). Does field based train-ing in behavior therapy improve clinical ef-fectiveness? Evidence from the Acceptance

and Commitment Therapy Training Project.Behavior Therapy, 29, 35-64.

Taylor, I., & O'Reilly, M. F (1997). Privateevents: A neglected and misunderstood con-struct in radical behaviourism. In K. Dillen-burger & M. F O'Reilly (Eds.), Advances inbehaviour analysis (pp. 23-45). Dublin, Ire-land: Appelbergs Bohtryckeri.

Washton, A. M. (1988). Preventing relapse tococaine. Journal of Clinical Psychiatry, 49,34-38.

Watson, J. B. (1924). Behaviorism. New York:Norton.

Wickler, W A. (1977). The search for the psy-che in drug dependence: A 35-year retrospec-tive survey. Journal of Nervous and MentalDisease, 165, 29-40.

Wickler, W. A., & Pescor, F T. (1967). Classicalconditioning of a morphine abstinence phe-nomenon, reinforcement of opioid drinkingbehavior and "relapse" in morphine addictedrats. Psychopharmacologia, 10, 255-284.

Wilson, K. G. (1998). Relational acquisition ofstimulus function in substance dependence: Apreliminary examination of drug versus non-drug related equivalence classes. Unpublisheddissertation, University of Nevada, Reno.

Wilson, K. G., & Blackledge, J. T. (in press).Recent developments in the behavioral anal-ysis of language: Making sense of clinicalphenomena. In M. J. Dougher (Ed.), Clinicalbehavior analysis. Reno, NV: Context Press.

Wilson, K. G., & Hayes, S. C. (1996). Resur-gence of derived stimulus relations. Journalof the Experimental Analysis of Behavior, 66,267-281.

Wilson, K. G., Hayes, S. C., Gifford, E. V., &Chang-Judson, G. (1999, May). The treat-ment of chronic, severe substance dependencewith Acceptance and Commitment Therapy.Paper presented at the annual meeting of theAssociation for Behavior Analysis, Chicago.

Zettle, R. D., & Hayes, S. C. (1986). Dysfunc-tional control by client verbal behavior: Thecontext of reason giving. The Analysis of Ver-bal Behavior, 4, 30-38.

Zettle, R. D., & Raines, J. C. (1989). Groupcognitive and contextual therapies in treat-ment of depression. Journal of Clinical Psy-chology, 45, 438-445.

Zuriff, G. E. (1988). Radical behaviorism andtheoretical entities. In A. C. Catania & S. Har-nad (Eds.), The selection of behavior: The op-erant behaviorism of B. F. Skinner: Commentsand consequences (pp. 216-217). New York:Cambridge University Press.

Date post:	07-Dec-2021
Category:	Documents
Upload:	others
View:	2 times
Download:	0 times

Kelly G. Wilson Steven C. University

Documents