The language of feeling and the feeling of anxiety: contributions of the behaviorisms toward understanding the function-altering effects of language.
The nature of language and its relation to emotional phenomena, however, has been one principal source of disagreement and controversy. The debate boils down to one question: Can one "feel" or know what it is to experience emotion without language or "cognitive" evaluative labels of that experience? Some have argued that emotion or affect is primary and essentially independent of language or cognition (Zajonc, 1980, 1984). Others, however, have stressed that what is felt depends on some cognitive or verbal appraisal of what is felt (Lazarus, 1982, 1984, 1991). Unfortunately, this "which comes first the chicken or the egg" debate has not moved cognitive-behavior therapy any closer to explaining the phenomena to which emotion refers. Rather, it has maintained artificial distinctions between language and cognition, on the one hand, and emotion or affective processes on the other (Rachman, 1981, 1984). Martin and Levey (1987) have attempted to resolve this debate by introducing the notion of evaluative responses (e.g., basic like/dislike-type responses), which are said to be subjective, immediate, preconscious, and largely unlearned or phylogenically based. Yet, it is unlikely that we are born knowing what our feelings are any more than we are born knowing the world and reality. Instead, we learn what our feelings are and this learning has a social-verbal experiential basis. The fact that the relation between language and emotion-affect is rarely discussed as a function of the social-verbal context suggests that we may have been asking the wrong questions about the right problem.
Instead of attempting to define what emotion is or is not, we should address the phenomena to which emotion refers in relation to the conditions that give rise to them (cf. Berscheid, 1990; Lazarus, 1984). In other words, we should be more concerned about the relation between language and feeling as an inseparable unit, how this relation is established, how it functions once learned, and how it can be modified or changed in therapy. In the late 1960s, the advent of cognitive and cognitive-behavioral theories and techniques within behavior therapy redirected the problem of emotion from simple conditioning to a problem of language, cognition, and meaning (Eifert, 1990). Thus, what one feels came to depend on what is said or appraised about what is felt. At the time, the dominant nonmediational S-R learning theories of emotion were not sufficiently developed to account for the language of feeling. Similarly, Skinner's (1938, 1974) statement that emotions are simply collateral by-products or epiphenomena of other behavior sealed the door on any interesting contributions from behavior analysis to the problem of language-emotion relations. In this vacuum, cognitive theorists have developed elaborate theoretical accounts of emotional phenomena (e.g., Beck & Emery, 1985; Lang, 1984, 1985; Lazarus, 1991). Yet, cognitive theorists have typically assumed the presence of underlying cognitive processes and their mediational relation to emotion and have obscured the more fundamental question of how people come to associate emotive functions with language and other private events.
The primary purpose of this paper is to show that both S-R learning theory and Skinner's treatment of emotion represent two behavioral accounts of emotion and not the behavioral account. We describe recent behavioral developments that explain the function of language and its role in the origin, maintenance, and change of emotional responding using anxiety-related phenomena as an example. Limiting the discussion to anxiety is arbitrary in that other emotional phenomena are also amenable to a behavioral analysis based on common behavioral principles and processes. Specifically, this paper (a) briefly explores the historical interest in the verbal basis of emotion; (b) describes and compares two dominant behavioral approaches for this understanding known as Paradigmatic Behaviorism (PB) and Radical Behaviorism (RB) and the relevant principles and treatments that derive from them; (c) provides a critical analysis of each approach; and (d) highlights some areas for potential synthesis and the implications for doing so. It should be pointed out from the outset that we will not attempt to define emotion, anxiety, or fear apart from the conceptual context and conditions under which those terms are used (RB and PB). Because emotion, and specifically, anxiety and fear lack any well established operationalization that clearly distinguishes between them, we will use such terms interchangeably as some psychophysiological set of functions that persons consider aversive and/or evaluate negatively. Finally, we hope that the addressed commonalities between respondent and operant views may lead to a renewed interest in behavior theory and further research and development of integrative and efficacious semantic behavior therapies.
Traditional Views: The Nature of Fear and Anxiety
Everyday language is replete with words referring to emotions, feelings, and other private experiences, and people often describe their behavior using these same terms. When we speak of emotions, we generally refer to the world of private experience, that is, emotions are about events occurring within the skin that are only directly available to an audience of one (Skinner, 1974). Whereas we may infer some emotional response in someone based on public observable accompaniments (e.g., trembling, blushing, facial expressions), we are unable to know in any exact sense what the other is feeling. The internal and subjective experience of emotions refers to the fact that emotions are, at the core, private psychological events which present a special problem of observation.
Although emotions have always played a central role in psychoanalytic and applied areas, academic psychology was reluctant to grant emotion any scientific status before the 1960s. This was due, in part, to the dominance of methodological behaviorism and logical positivism during this period (cf. Lazarus, 1991). The denial of emotion as a legitimate topic of scientific study took three forms (Hillman, 1960): the recommendation that psychologists (a) abandon the concept of emotion, (b) subsume emotion under other concepts (e.g., motivation), and (c) question the scientific utility of emotions as "things" that erroneously call for explanation. He quotes Meyer (1933) who wrote,
Why introduce into science an unneeded term, such as emotion, when there are already scientific terms for everything we have to describe? . . . I predict: the 'will' has virtually passed out of our scientific psychology today; the 'emotion' is bound to do the same. In 1950 American psychologists will smile at both of these terms as curiosities of the past. (p. 300)
Contrary to Meyer's prediction, psychology proceeded to embrace emotional phenomena as a legitimate area of study predominately addressing the question: What is emotion?
In view of the ubiquity of emotion, this seems a reasonable question, but it has been difficult to answer satisfactorily for at least two reasons. First, psychologists have assumed that when people speak of different feelings they are, in fact, speaking of different feelings. Using common linguistic categorical distinctions between emotions as a starting point, psychologists have borrowed nontechnical terms from lay language that reference feeling (e.g., anxiety, fear, joy, sorrow) and attempted to provide both conceptual and technical definitions of these terms. Theorists and researchers alike have sought an understanding of this private phenomenological world of emotional experience and characterized emotions differently based on hypothesized constructs (e.g., motivation, drives) and real variables and processes (e.g., physiology). Second, categorical distinctions of emotion invite a search for the underlying structure of feeling as something other than behavior. Everyday language underscores the difference between feelings and behavior (e.g., "I'm feeling down and don't want to go out tonight"), and laypersons commonly give feelings some causal role in their actions ("I don't fly in planes because I'm afraid"). Behavior therapists have also maintained this traditional view of emotion as something that people have, rather than what people do, by according emotional phenomena, such as anxiety or fear, some motivative or causal role.
Inadequacies of Behavior Therapy's Simple Conditioning View of Anxiety
Early behavior therapists approached the problem of emotion, and specifically fear, as one of simple Pavlovian or respondent conditioning. According to this view, any neutral stimulus (NS) paired with an unpleasant event or aversive unconditioned stimulus (UCS) that elicits a strong negative emotional unconditioned response (UCR) will later function as a conditioned stimulus (CS) that can elicit a negative conditioned response (CR; e.g., "anxiety") more or less similar to the original UCR. Several efficacious therapeutic techniques were derived from the respondent conditioning model such as systematic desensitization and in-vivo exposure (Eysenck, 1987; Wolpe, 1958). These techniques were rooted firmly in conditioning theories of Hull (1943), Pavlov (1927), Skinner (1938), and Watson (1930) and maximized the role of observable behavior while minimizing the role of cognitive mediation. However, several limitations of respondent conditioning theory became apparent with growing evidence suggesting that it could not alone account for some new data derived from laboratory investigations and clinical observations (see Eifert, 1990; Rachman, 1977; Seligman, 1971). For instance, simple respondent conditioning could not explain how fear is learned and unlearned in cases that did not involve direct conditioning experiences. Although several researchers (e.g., Eifert, 1990; Eysenck, 1987; Seligman, 1971) responded to these challenges by extending and modifying the basic respondent conditioning position, the more popular response was to abandon conditioning notions and behavior theory (e.g., Bandura, 1978, 1984; Meichenbaum, 1977). This abandonment was driven, in part, by practical concerns and clinical observations suggesting that human and animal learning differed and that this difference was attributable to humans' capacity for language and other symbolic thought processes (Staats & Eifert, 1990). Acknowledging the differences between infrahuman and human learning initially served as an embarrassment for learning theory rather than as a challenge to explore these implications systematically. Although the respondent conditioning paradigm was extended by introducing other concepts (e.g., incubation, preparedness), it was not developed in any systematic way to incorporate language-symbolic processes. Behavior therapists were equally frustrated with the failure of operant theory to account for language-emotive relations in terms useful for clinicians. Radical behaviorism, in rejecting both S-R learning theory and cognitive mediational accounts, had very little impact on the development of behavior therapy during this early period (cf. Hayes & Hayes, 1992).
Move to Cognitive Conceptualizations: Identifying the Right Problem But Adopting the Wrong Solution
As a result of growing dissatisfaction with nonmediational behavioral accounts and their eschewal of acknowledging the relation between language and emotional meaning, cognitive theories and constructs emerged to account for the relation between feeling, affect, and cognitive processes. For example, self-efficacy theory (Bandura, 1978, 1984), bio-informational theory (Lang, 1977, 1985) and cognitive models of anxiety (Beck & Emery, 1985) all place heavy emphasis on cognitive-mediational constructs (e.g., networks, nodes, expectancies, appraisals, and schemata) borrowed from information and computer science. These views have had great appeal and were quickly absorbed by the behavior therapy movement. They became known as cognitive-behavior theory and are currently the dominant approach for explaining emotional phenomena such as anxiety and depression.
Recognizing the importance of language and emotion in human affairs, cognitive psychologists have emphasized the inextricable relation between language and emotive functions, that is, how language-symbolic processes convey emotional meaning and how emotions depend on such processes. The principal objection of behaviorists to the cognitive view of emotions is not about the phenomena it refers to but about the explanations it offers for the phenomena. Typically, these explanations start with legitimate observations of language-behavior relations and then propose other hypothetical and mediational processes within the organism to account for these relations rather than emphasizing the conditions that produce them. As a result, we are left with circular and incomplete definitional accounts of behavior that rely on one hypothetical construct to explain another. Moreover, we are still left to ask how one comes to call such events emotional in some instances and something else in others. Behavior theorists, however, have generally recognized that "emotion" and its derivatives are not technical terms within psychology, but convenient summaries of classes of functional relations (cf. Eifert, 1990). In turn, behaviorists have felt no obligation to explain what emotion is apart from those conditions leading to the use of emotional descriptions (Zuriff, 1985). However, behavior theorists are obliged to explain the conditions that establish emotional meaning and how such language-emotion relations function once learned.
The Language Hypothesis and the Meaning of Anxiety
The availability and pervasiveness of language represents one significant difference between animal and human learning (Eifert, Forsyth, & Schauss, 1993). Language serves important symbolic functions by providing humans with emotional experiences without exposure to the actual physical stimuli or events that ordinarily elicit those responses (Staats & Eifert, 1990). Just as there is "knowing what to do" there is also "knowing what to feel." (Scruton, 1980). Both involve verbal understanding and some activity, and the relation between them. To account for these functions and how they are learned we must address the meaning of verbal events in the psychological sense and not in the literal sense. Many nonbehavioral theories and therapists erroneously accept the literal content of client verbalizations as if what clients say is in fact what is going on psychologically (e.g., I can't fly in a plane because I am anxious, or I can't speak in front of a group because I will panic). A behavioral approach to the psychological meaning of verbal-symbolic events entails different assumptions and strategies. Most notably, verbal relations are contextually sensitive and functionally defined (Hayes & Wilson, 1993). Thus, the meaning of anxiety in a psychological sense represents a complex act of relating largely arbitrary verbal-symbolic events with other events and psychological functions in and within a context.
This brings us to one of the unique and important features of the language of feeling (i.e., emotion words), that is, its referential or relational quality. For instance, words such as anxiety and fear either implicitly or explicitly establish relations with other events such as "I am anxious or afraid of . . . something, some event, or someone (including the self)." Simply to say, "anxious," "afraid," "love," or "depressed" without some relational referent would strip these words of their meaning (i.e., their functions). The relational quality of terms denoting emotions, in turn, can only be identified by certain descriptions. These descriptions are often tied to occurrences of behavior and events with a variety of stimulus functions (e.g., eliciting, evocative, reinforcing, punishing) and meanings (e.g., good, bad, pleasant, unpleasant, painful). In turn, people often describe their emotional experiences metaphorically in ways that others can understand (e.g., "When I feel anxious it's like a knife going through my chest" or "When I'm depressed . . . I feel empty"). These metaphorical extensions, however, have no real counterpart inside the person. Instead, they function to communicate the meaning of emotional experience by identifying and relating events with known stimulus functions.
Although this is a complex topic, the central point to be made here is that one cannot know what it means to feel X without first understanding what it is to feel X. To do this, however, requires verbal specification of "what is felt" (either privately or overtly) in order for it to be labeled as such. For instance, even saying to someone "I have a feeling" as vague as it may seem, meets this requirement. Second, the concept of emotion is, by definition, related to meaning and function. Nothing about the form of central nervous system responses (e.g., increased heart rate, electrodermal activity) requires them to stand for a particular emotional description (e.g., anxiety). Similarly, nothing inherent in the formal properties of words denoting emotional experience (e.g., anxiety, fear, anguish, terror, joy) requires it to stand in relation to a particular bodily or environmental referent. Thus, when we see an emotion on someone's face, we are establishing a relation based on a history of learning such relations (e.g., a grimace means fear). Depending on the cultural context, a variety of emotions may be fashioned from the same biological system; conversely, elements from several different systems may be incorporated into a single emotion. For the most part, however, biology only contributes elements to, and sets limits upon, the social construction of emotional behavior (Averill, 1980). Accordingly, the stimuli and events placed in a relation and the nature of that relation are determined by the social verbal community without regard for the formal properties of the stimuli involved (Hayes & Hayes, 1989). In other words, the relation between the "language of feeling" and the "feeling of feeling" is arbitrary in the sense that we learn to use certain words, and not others, to convey emotion. The remainder of the discussion describes and relates two behavioral traditions (respondent and operant) which have specifically and extensively addressed how these arbitrary relations are learned and function.
Unearthing the Social-Verbal Basis of Fear and Anxiety
The Respondent Tradition and Semantic Conditioning: Paradigmatic Behaviorism's Contributions
Overview and basic tenets. Paradigmatic behavioral theory (e.g., Staats, 1972; Staats & Eifert, 1990) states that the acquisition and maintenance of fear and anxiety do not require direct experience with painful or aversive events. Rather, the association of inappropriate or negative emotion-eliciting verbal-symbolic stimuli with certain stimulus events (e.g., situations, objects, images, or words) is sufficient for those objects or situations to acquire aversive functions. Central to the discussion of anxiety are two major tenets of PB. The first is that central nervous system (CNS) responses are at the core of anxiety problems (cf. Eifert, 1990; Staats & Eifert, 1990). These CNS responses are biologically derived and are most commonly called the alarm response (Barlow, 1988). However, this CNS response is not sufficient alone for emotion. For instance, a person may show comparable patterns of CNS activation while being chased in a game of tag and being chased in the city by a mugger; however, it is more likely that the latter case will be described as fear. The second tenet states that CNS responses require some negative verbal or symbolic evaluation to be called emotion. In turn, language-emotive functions can be acquired directly through aversive respondent conditioning or indirectly through semantic conditioning (Staats & Eifert, 1990).
Semantic conditioning and emotional meaning. The term semantic conditioning, first coined by Razran (1939), refers to the application of respondent conditioning principles to language, and specifically, the acquisition of meaning and language functions (see also Mowrer, 1954; Osgood, 1969). Thus, a word stimulus paired with an unconditioned stimulus (UCS) that elicits an emotional response will also come to elicit the emotional response. For example, the word snake (CS) may elicit some sensory visual response ([r.sub.sm]) after having been associated with a picture of a snake. Thereafter, the word snake will come to elicit part of the sensory response elicited by the picture of the snake (cf. Eifert, 1984). Similarly, if the snake picture can elicit a strong CNS response (e.g., pain or alarm), then the word snake will likely come to elicit part or all of this response ([r.sub.sm]). According to PB, it is through this process of simple or primary semantic conditioning that words acquire emotional meaning, that is, a word becomes meaningful when it comes through classical conditioning to elicit a conditioned response (Staats & Eifert, 1990). Hence, the emotional meaning of a word is functionally defined as conditioned response functions which are most commonly called evaluative (e.g., positive/negative, liked/disliked, good/bad; cf. Martin & Levey, 1987). Evidence for this kind of learning is commonplace such that words are not simply stimulus events, but denote events with emotive stimulus functions that have some psychological meaning. The establishment of some conditioned emotional or evaluative response using respondent conditioning procedures has been well documented in the literature using a variety of verbal and nonverbal conditioned and unconditioned stimuli (e.g., Maltzman, Raskin, Gould, & Johnson, 1965; Martin & Levey, 1978, 1987; Staats & Staats, 1957, 1959; Staats, Staats, & Crawford, 1962; Staats, Staats, & Heard, 1959). As a result, directly established stimulus functions of words can transfer to other words, symbols, and events indirectly via higher order conditioning.
Higher order conditioning and the transfer of emotion-eliciting functions. The principle of higher order conditioning states that when a stimulus, previously neutral with respect to a particular response, has acquired CS functions to elicit that response ([r.sub.sm]), the stimulus can now transfer the response to another neutral stimulus. This is possible, in part, because the CS functions as a higher order UCS in new learning situations. For instance, suppose a child has learned a direct association between the word HURT and sensations produced directly upon touching a hot stove, but has no experience with the word BITE. The word BITE at this point is a meaningless nonsense syllable. Thereafter, the child hears the word bite in relation to dogs. The child then asks a teacher, "what does BITE mean?," and the teacher responds "BITE means HURT." As a result, the child may not only respond to BITE as equivalent to HURT, but may also learn that DOGS BITE, meaning that they HURT, and may subsequently avoid dogs. Transfer of conditioned meaning responses to otherwise neutral stimulus events is explained, in large part, by transfer of implicit meaning responses ([r.sub.m]) between events (cf. Mowrer, 1954; Staats, 1963, 1975). For example, a previously neutral stimulus event (OMDX) is said to elicit some neutral meaning response ([r.sub.m1]) with respect to it. Subsequently, when OMDX is paired with another UCS event (e.g., shock) capable of eliciting some conditioned response ([r.sub.m2]), portions of this implicit response will likely transfer to the previously neutral [r.sub.m1] and change its meaning. Over the last 30 years, many studies have shown higher order conditioning and transfer of functions to new stimuli using language (e.g., Eifert, 1984, 1987; Hekmat, 1977; Maltzman, 1968; Osgood, 1969; Razran, 1961; Staats, 1963, 1975; Staats & Staats, 1957, 1958).
Affective, reinforcing, and directive functions of verbal or symbolic stimuli. The principles reviewed thus far have primarily emphasized the acquisition and transfer of emotional meaning responses based on respondent operations. In discussing the stimulus functions of language, however, both respondent and operant functions are mutually related. According to PB, once stimuli acquire emotion-eliciting respondent functions they will also (a) function as reinforcers and punishers and (b) evoke either approach or escape and avoidance behavior in operant learning (Staats, 1975). Similarly, stimuli that have acquired punishing or reinforcing functions through operant operations will transfer those functions to other stimuli present that may then also function as emotion-elicitors in respondent conditioning situations. This fundamental three-function learning principle can be summarized as follows: Once stimuli acquire affective-eliciting response functions on an unlearned or learned basis (A-function), they will also acquire both reinforcing (R-function) and directive or discriminative functions (D-function). Whereas the reinforcement value of any stimulus is defined by its emotion-eliciting functions that can occur either naturally or through previous conditioning (Staats, 1990), the directive functions of emotional stimuli are primarily shaped through experience. Evidence for the three-function learning principles has been reliably demonstrated in several laboratory studies using human subjects. For example, studies examining the R-function have shown that words that elicit positive meaning will strengthen a previous response whereas words that elicit negative meaning will suppress responding (e.g., Finley & Staats, 1967; Staats, Gross, Guay, & Carlson, 1973). Support for the D-function is largely based on studies showing that humans will approach stimuli that elicit positive emotional functions and will avoid stimuli that elicit negative emotional functions (Nunnally, Duchnowski, & Parker, 1965; Silverstein, 1973; Staats & Burns, 1982; Trapold & Winokur, 1967). There is also evidence that motivational operations (e.g., deprivation and satiation) can potentiate or depotentiate the A-R-D functions of language (Harms & Staats, 1978; Staats & Warren, 1974; for a summary, see Staats, 1975). This research on the acquisition and transfer of emotive functions via language has served as a basis for PB's concept of the verbal-emotional repertoire which represents a culmination of this learning.
The verbal-emotional repertoire. As emotional learning proceeds over one's lifetime in a cumulative and hierarchical way individuals develop a unique language-emotional repertoire consisting of many verbal-symbolic stimuli that elicit both positive and negative emotional responses which also have reinforcer and directive stimulus functions. The elements of the language-emotional repertoire can be elicited either directly or indirectly in new learning situations or they can indirectly function as a mediating response to those situations (Staats, 1994). Thus, behavior is viewed as a joint function of the person's repertoire and the current situation. Because the verbal-emotional repertoire serves as a basis for subsequent behavior and learning in new contexts, it is viewed as a basic cause or determinant of behavior and not simply as a dependent variable (Staats, 1975, 1994). In turn, individual differences in this repertoire determine why two individuals under similar circumstances behave differently. Persons who suffer from anxiety disorders will have likely developed complex repertoires of a variety of verbal and other symbolic events capable of eliciting negative affective responses which also function as punishers and direct escape and avoidance behavior (Staats, 1989; Staats & Eifert, 1990). For example, for some persons who are otherwise healthy, the sensations of a beating heart or chest pain may lead to a sequence of verbal and autonomic events that result in the belief that one is having a heart attack (Eifert, 1992). In this instance, a fast or irregular heart beat is not just a felt beating heart. Instead, it is an acquired and verbally mediated formulation of what it means to have a fast or irregular heart beat or chest pain (e.g., "I have heart disease" or "I am suffering from a heart attack"). Not only may the person respond to such sensations by rushing to an emergency room, but any other stimulus events (public or private) associated with this response may now acquire similar negative functions (e.g., exercise, smoking, working hard). According to Staats and Eifert (1990), it is because an emotional response has stimulus properties that it can further elicit overt motor behavior. In addition, persons can condition and recondition themselves in ways not explicitly taught by providing their own language-symbolic experiences that necessarily evoke various emotive-reinforcing-directive functions (Eifert, 1984, 1987). By doing so, however, persons also establish various stimulus functions to otherwise neutral stimulus events via higher order conditioning and transfer of emotional functions. Clinically, the problems clients present with are viewed as resulting from deficit and inappropriate verbal-emotional learning which serves as a basis for subsequent problems. Therefore, language-based therapeutic strategies explicitly target the emotive functions of language.
Language-Based Behavior Therapy
Basic therapeutic strategies: Altering the verbal emotion-eliciting functions of language. Central to many behavior therapies are attempts to alter or change emotional responses to stimuli or events (e.g., Eysenck, 1987; Rachman, 1977; Wolpe, 1958). Most often, this is accomplished by somehow getting clients to confront the feared objects and the aversive bodily events in a safe context, which facilitates corrective emotional learning and fear reduction. This is often achieved, or at least initiated, in therapy either directly or indirectly using language. In fact, it has been argued that language is one of the most powerful methods by which human behavior is controlled and changed (Eifert, 1987; Staats, 1972). If this is the case, then therapeutic strategies that systematically alter the emotion eliciting functions of a client's verbalizations may also produce changes in the reinforcing and directive functions of other verbal-symbolic events in the client's repertoire. Likewise, altering the reinforcing or directive functions of events may foster changes in their emotion-eliciting functions. For instance, many clients enter therapy with elaborate rules, reasons, and justifications for what they have done, why they do what they do, and what they believe needs to be done to overcome their problems (e.g., "I can't keep myself from getting anxious about speaking in front of people," or "I must not exert myself to avoid having a heart attack"). On the surface, these verbal utterances of "the problem" cannot cause any harm (e.g., being afraid that one might have a heart attack is not the same thing as having a heart attack). As clients struggle with their problems, however, they will often establish negative emotional functions with previously unrelated events (e.g., "To control my anxiety I will keep from going outside"). The thought of going outside may bring about the thought of anxiety and, along with it, other negative or aversive psychological functions. Thus, the goal of language-based therapies is to identify and alter the A-R-D functions of language based on the premise that if language exerts control over behavior, then altering the functions of language can produce changes in behavior. Language processes are therefore considered as both targets (dependent variable) and mediators (independent variable) for generalized therapeutic effects (cf. Lohr & Hamberger, 1990).
Emotional modification and the process of change: Semantic counterconditioning. Staats (1972) suggested that emotional problems such as excessive anxiety can be changed in therapy via semantic conditioning procedures. The process can be summarized as follows: Stimuli that elicit a negative emotional response can be altered directly by pairing the negative eliciting stimulus with a stimulus known to elicit a positive emotional response. Such language-based interventions work through the verbal-emotional repertoire by using words to elicit an emotional response in the client which will also produce changes in the emotional function of words or other events in relation to what is said. Consequently, collateral changes should also occur in the reinforcing and directive functions of emotion eliciting stimuli which may account, in part, for generalization of treatment gains beyond the therapeutic context. Several experimental and clinical studies have shown the effectiveness of this intervention technique in accelerating or inhibiting the process of extinction (Eifert, 1984; Eifert & Schermelleh, 1985; Staats, 1972). In addition, this technique is an operative in many well established behavioral treatments for anxiety-related problems (e.g., systematic desensitization, cue-controlled relaxation and imagery, imaginal flooding, and covert desensitization) which aim to change the emotion-eliciting functions of events. As Eifert (1987) points out, however, studies examining the clinical efficacy of purely language-based interventions (e.g., semantic and covert conditioning) have met with mixed success. In part, the mixed results of such interventions may represent a failure to tailor such procedures to a given client's presenting problems such as in idiographic functional assessment and treatment (cf. Eifert, Evans, & McKendrick, 1990; Hekmat, 1990). Studies that have matched treatments to presenting problems using an idiographic functional approach, especially to language-emotional relations, have shown more promising results (e.g., Eifert, in press; Hekmat, 1977, 1990).
The Limitations of Paradigmatic Behaviorism's Analysis of Emotion
PB has attempted to provide a systematic behavioral account of emotion by bridging both operant and respondent principles with verbal-emotive behavior at a time when neither behavior therapy, behavior analysis, nor psychology was interested in such relations. Despite PB's rich behavioral analysis, very few behavior therapists and applied behavior analysts have drawn upon PB. The question is why? Because of space limitations, we can only address some prominent limitations while also suggesting other possible considerations.
The utility of semantic conditioning based on mediational associationism. The respondent conditioning model upon which semantic conditioning is based suggests that conditioned functions to verbal or nonverbal events should invariably elicit some conditioned response irrespective of context (see Zuriff, 1985). This assumption, however, seems untenable given what we now know about contextual constraints on learning and behavior. For instance, one might react strongly to encountering a live snake in the woods, but react with curiosity to seeing a similar live snake caged at the zoo. Here, the snake stimulus is the same, but the context sets limits on the nature of the stimulus functions elicited. It is precisely because of these complexities that S-R analysis evolved into mediational S-R associationism. This view argues that processes occurring between stimulus and response account for the association of S-R and subsequent learning and behavior in different contexts. Similarly, PB explicitly includes mediational mechanisms (e.g., [R.sub.ms]) in its account of primary and higher order semantic conditioning (Staats, 1975; Staats & Staats, 1963); although these mechanisms are rarely made explicit in contemporary accounts (see Staats, 1994). The mediational perspective, however, outlived its usefulness when new mediational events were needed to account for each new instance of behavior that failed to conform to previous S-R associations (e.g., Cofer & Foley, 1942; see also Sidman, 1992, 1994). In other words, unlike directly observed stimulus relations, mediation implies events occurring between S and R that are not subject to direct observation or experimental manipulation. Mediational elements are at best inferences and at worst hypothetical constructs (cf. Zuriff, 1985). Without relating S and R to mediational elements and contextual variables, this paradigm is simplistic and too mechanistic to account for the flexibility of verbal behavior and the many stimulus functions for which language derives its many meanings.
The failure of higher order conditioning to account for the strength of the CR. PB explains how words acquire emotional functions without a history of direct conditioning by invoking the principle of higher order conditioning. Higher order conditioning principles escape from the criticism and restrictiveness of the requirement that meaning is acquired to each word solely by a direct association (i.e., pairing) with some positive or aversive UCS. The principle of higher order conditioning, however, raises another point of concern. A generally understood principle of higher order conditioning states that with the addition of each new neutral stimulus (NS1, NS2, NS3) paired with the original CS, we should also observe a decrement in the eliciting functions of the additional CSs to produce a CR of equal strength and size as the original CS. In human affairs, adventitious or intermittent reinforcing experiences could occur and thus strengthen an initially weak higher order relation; however, if this occurs, it would preclude the need for invoking the principle of higher order conditioning because such learning would represent direct semantic conditioning.
Failure of S-R learning to account for bidirectional semantic conditioning. As with respondent conditioning, semantic conditioning operations are based on unidirectional S-R relations such that a CS elicits some CR, but not the other way around (CR elicits a CS). For example, suppose we pair the word SNAKE with shock (UCS). After sufficient trials, the word SNAKE when presented alone will elicit some CR. What does not typically happen, however, is that when shock is first presented SNAKE will be elicited (backward conditioning). In language conditioning using only pairs of words, backward associations were quite commonly observed though they were not explicable by S-R learning theory or supported experimentally (cf. Jenkins, 1963). This finding of bidirectional equivalences between S-R posed a problem for S-R learning theory because researchers could not identify the stimulus and response terms a priori, that is, they were interchangeable. The question became, therefore, whether semantic S-R conditioning could account for bidirectionality? Although the answer to this question is fundamental to understanding the symbolic and emotional functions of language, it has not been adequately addressed within the S-R semantic conditioning paradigms (see Sidman, 1994). It is conceivable that humans have repeated experiences with both forward and backward conditioning experiences such that sometimes the CS precedes the UCR (S-R) and other times the UCR precedes the CS (R-S) thereby establishing a bidirectional relation between them. However, this extends the basic semantic conditioning to include other complex forms of S-S learning that are not typically contained in the basic S-R experimental paradigms; extensions that are neither inconsistent with nor sufficiently developed and considered by PB (cf. Jenkins, 1963; Jenkins & Palermo, 1964; Kjeldergaard, 1968).
The notion of the verbal-emotional repertoire is structural, mediational, and not explanatory. PB has taken the notion of history and repertoire in human learning to a different level by specifying how they are learned and function. This specification is not without its own problems, however. Two claims about PB's conception of repertoires are especially problematic. First, "it is hypothesized that while these basic behavioral repertoires [BBRs] are learned through prior experience, the repertoires also function in a causative and interactive manner with other BBRs and the stimulus situation in affecting subsequent behavior, experience, and learning of the individual" (Staats, 1990, p.103; italics added). Second, "explanation of the individual's behavior cannot be obtained only by knowledge of the current situation and behavior principles . . . explanation requires knowledge of the individual's BBRs" (Staats, 1994, p. 107, 108). Staats might better have replaced "cause" and explanation with "interpretation" in this context. Typically, behavioral accounts of causality require manipulation of environmental independent variables and observations of its effects on behavior (dependent variables) such as that used in a functional analysis (Schauss & Forsyth, 1994; Zuriff, 1985). To say that BBRs are causally related in any way to present behavior we would have to be able to manipulate BBRs independently of other variables that may be controlling the behavior in question. This is difficult, however, because the supporting evidence for BBRs is based largely on some observed behavior that is subsequently taken as evidence for some unobserved BBR. In line with some failures of S-R extensions to human learning, BBRs represent attempts to explain behavioral phenomena by inventing a mediational property of the organism responsible for the phenomena. As Palmer and Donahoe (1992) indicate, the hypothetical property is often invoked as an explanation for the phenomenon. For instance, BBRs have been implicated in different types of abnormal behavior as complex and fixed constellations of behavioral histories (e.g., Eifert, 1992; Heiby & Staats, 1990; Staats, 1989, 1990, 1994) which are explicitly, although not exclusively, given a causal or mediational role (Staats, 1994). Yet, a priori specification of mediating causal factors such as BBRs implies that the components of these repertoires are stable factors that all persons with certain behavior problems have. As such, they may detract from a functional analysis and limit further inquiry prematurely - although PB does argue that framework models must be tailored appropriately to the individual client (cf. Eifert, 1992).
Other general limitations and unresolved issues. A central tenet of PB is that emotional meaning is acquired via direct classical conditioning and/or indirectly though language experiences via higher order conditioning. We are still left to ask, what then is emotional meaning and what is transferring? Is it physiological responses, overt behaviors, some private symbolic activity, or a combination of these? The difficulties in specifying the nature and function of semantic meaning relations via respondent conditioning operations have been noted elsewhere (cf. Maltzman, 1968); however, they remain somewhat vague in PB's analysis. Whereas PB's statement of the interrelated functions of language (A-R-D) remains a significant and fertile area of further inquiry and clinical application - a point we take up again later - its contributions to language behavior therapy remain less clear (Staats, 1972). Although PB emphasizes verbal behavior and its modification in therapy, it does not add much beyond traditional behavioral techniques. Simply put, despite some new techniques summarized by Hekmat (1990) under the rubric semantic behavior therapy, PB's language behavior therapy is not really a new or different form of behavior therapy.
Finally, we do not mean to understate the important contributions that PB has made toward understanding human behavior, language, and emotional learning. Our intention is to point out some limitations that may indicate why PB has not been more widely adopted as a viable framework for behavior therapy. In particular, PB has largely maintained an adversarial relationship with behavior analysis, and in particular, applied behavior analysis (Staats, 1994), rather than attempting to draw upon its recent developments systematically. We will now discuss some of these developments as they relate to the problem of language and emotion. As will become apparent, behavior analytic accounts of verbal-emotive relations have also failed to draw upon the respondent or semantic conditioning literature. This lack of systematic integration has maintained artificial boundaries between both behaviorisms. In an attempt to rectify this situation, we will offer some points of synthesis between PB and applied behavior analysis.
The Operant Tradition and Recent Contributions From Clinical Behavior Analysis
Skinner's treatment of emotion and its social-verbal basis. Critics of radical behaviorism often wrongly state that Skinner denied the reality of private events such as feelings or emotions. Skinner's difficulty has never been with the reality of feelings but rather with what dualists and idealists have thought their deterministic nature or causal role to be. Throughout Skinner's writings, he makes the strategic and methodological assumption that emotions are not causes or determinants of behavior (Skinner, 1938). For instance, it does not help in the solution to a practical problem to be told that some feature of an individual's behavior is caused by anxiety or fear; we also need to be told how the fear and anxiety have been induced and how it may be altered. Skinner (1974) rightly points out that, "any therapeutic attempt to reduce the 'effects of anxiety' must operate upon the circumstances, not upon the intervening state," but his statement that "the middle term is of no functional significance, either in a theoretical analysis or in the practical control of behavior" is dubious (pp. 180-181). In our view, this dogmatic stance has seriously impeded an effective behavioral analysis of an important aspect of human behavior. Although in his later writings Skinner gradually moved toward an acknowledgment that emotional events can have stimulus properties (see Staats, 1988), his most important contribution was the recognition that emotional events are social-verbal constructions inextricably tied to social contingencies which entail an analysis of respondent and operant functions and verbal behavior (Skinner, 1957, 1984). Since the late 1970s, behavior analysts have used verbal behavior to address complex forms of human behavior that were previously understudied in behavior analysis (e.g., problem solving, thinking, emotions). Indeed, the functional analysis of verbal behavior has served as the cornerstone for behavior analytic research and theory about emotional behavior beginning with how people learn to label and describe their experience using language.
Overview and basic tenets: Verbal labeling and tacting. One of the fundamental functional units identified by Skinner (1957) is the tact. Tacts are distinguished from other verbal behavior, in part, because a response of a given form is more likely in the presence of some nonverbal stimulus event because of a history of generalized reinforcement provided by the verbal community beginning with simple examples. For example, if a child says "ball" because of a history of generalized reinforcement (e.g., "Good," "That's right," "Yes") for saying "ball" in the presence of such objects, the statement is a tact. Similar principles used to explain public tacts also apply to private ones or those occurring within the skin. For example, suppose a parent sees a child smiling, giggling, laughing and says "You're anxious, aren't you?" After many such occasions, the child's overt behavior that we would otherwise call happiness may come to mean "anxious." Thereafter, when the parent asks the child how she or he feels given similar overt accompaniments, the child may respond "I'm anxious." This example of tacting is unlikely in our verbal community. Yet, it illustrates the point that tacting a private event(s) is arbitrary in the same manner as tacting public events. Unlike public tacts, however, private tacts are necessarily inexact because the verbal community is unable to provide differential reinforcement for specific unseen private events in the same way as for observable events (Skinner, 1957). This is one reason why many different emotion words can be associated with similar bodily events and different bodily events may be associated with the same emotional descriptions. Several studies have also shown that response classes such as self-referred affect (e.g., I'm sad, I love, I'm anxious) can be increased, decreased, and made more specific as a function of reinforcement (e.g., Salzinger & Pisoni, 1958, 1960, 1961). Once language becomes associated with either public or private events with some psychological function, the words will also acquire similar functions and therefore meaning. However, an adequate functional account of verbal behavior requires knowledge of the person's history and the sources of control over present behavior (Hayes & Hayes, 1989). For instance, saying "I feel anxious" without a history of generalized reinforcement would not constitute private tacting according to Skinner's definition. The observation that people do name objects or events without a history of prior conditioning or generalized reinforcement (e.g., saying "dog" in the presence of a real dog after only seeing the word or a picture of a dog, or saying "I feel anxious" when seeing the word snake or a picture of a snake) presented notable difficulties for Skinner's account of verbal behavior. More recently, behavior analysts have addressed these difficulties using the phenomenon known as stimulus equivalence which accounts for relations not explicitly taught.
Stimulus equivalence. In general terms, stimulus-equivalence research involves teaching subjects to match comparison stimuli to sample stimuli (cf. Hayes & Hayes, 1989; Sidman, 1992, 1994). The stimuli are said to be equivalent if the relation between the samples and comparisons can be shown to have the properties of reflexivity, symmetry, and transitivity (Sidman & Tailby, 1983). For a relation to be reflexive, each stimulus must be matched to itself (e.g., given the word "SPEECH" as a stimulus, then the subject matches it to a comparison stimulus "SPEECH" from another array and vice versa). To show symmetry, the trained relation between two stimuli must be functionally reversible: When "given SPEECH, pick EMBARRASSMENT" is trained; "given EMBARRASSMENT, pick SPEECH" is likely to emerge without direct reinforcement (SPEECH=EMBARRASSMENT). Similarly, when "given SPEECH, pick ANXIETY" is trained, "pick ANXIETY, given SPEECH" emerges without reinforcement. Finally, for a relation to be transitive, a subject must be taught to match "SPEECH" to "EMBARRASSMENT," and then "EMBARRASSMENT" to "ANXIETY," and, without further training and a history of reinforcement for doing so the subject is likely to match "SPEECH to ANXIETY" and "ANXIETY to SPEECH." Thus, with equivalence, training two relations directly yields four new derived bidirectional relations not explicitly taught. Although there is no logical requirement that equivalence relations must be derived (Vaughan, 1989), the major interest in stimulus equivalence is precisely because these relations are derived and seem to correspond with language phenomena. Stimulus equivalence appears to represent a fundamental behavioral process that has only been convincingly shown in humans with verbal abilities using a variety of stimulus materials not all of which have been verbal (cf. Hayes & Hayes, 1989; Hayes & Hayes, 1992b; Sidman, 1994). The importance of stimulus equivalence for language resides in the fact that humans not only learn to respond to one event in terms of another, but also respond to stimulus events because of the stimulus functions of other events (cf. Sidman, 1994).
The transfer of functions through equivalence classes. Relations among stimuli are important because other functions can be moderated by those relations. With language, we are interested in words not as objects, but as events that have psychological functions (i.e., meaning) because they are placed in relations with other events. Sidman and Tailby (1983) assert that, by definition, the existence of a class of equivalent stimuli permits any variable that affects one member of the class to affect all members. As Dougher and Markham (1994) point out, functional and stimulus equivalence are similar in that both refer to a set of stimulus relations that can be identified by shared function and transfer of function. For instance, one member of an established equivalence class is selected and given some psychological function. Thereafter, other members of the class are tested to determine if they have acquired the function. If the function is acquired by other members of the class from which the stimulus was selected, but not by members of other classes, the function is said to have transferred through the equivalence class (cf. Dougher & Markham, 1994). This process, in turn, is relevant for emotional learning. For example, studies have shown that when shock is paired with one member of a class, electrodermal and avoidance responses transfer to other members of the class that were never previously reinforced with shock (e.g., Dougher, Augustson, Markham, Greenway, & Wulfert, 1994). A variety of other stimulus functions have also been reported to transfer through equivalence classes such as discriminative stimulus control (e.g., Lazar & Kotlarchyk, 1986), conditioned reinforcement and punishment (Greenway, Dougher, & Wulfurt, 1992), contextual control (Gatch & Osborne, 1989; Kohlenberg, Hayes, S. C., & Hayes, L. J., 1991), and consequential functions (Hayes, S. C., Kohlenberg, & Hayes, L. J., 1991; Kohlenberg et al., 1991). This research suggests how persons may develop strong emotional responses to events simply through language relations and how contextual cues can set constraints on what functions will transfer from one stimulus to the next. For example, suppose the word snake is paired with shock in the presence of a red light, but in the presence of a green light the shock is not forthcoming. As a result, the word snake and other stimuli in a class associated with this word may elicit respondents only in the presence of the red light and not in the presence of the green light. Thus, humans not only learn to respond to relations among events, but they also learn that such relational responding is dependent upon contextual or environmental circumstances. These two notions serve as the basis for relational frame theory.
Relational frame theory. Relational frame theory stipulates that the context (i.e., a frame) will select any given relational response (e.g., less than, greater than, equal to, if . . . then) between events and stimulus functions associated with those events (Hayes, 1994a). These relations can be arbitrary in that relations between two or more events do not necessarily depend on their formal stimulus properties. Instead, relations between events are established by contextual cues present which may include other words, their sequence, the nonverbal context, paralinguistic cues, and other events (cf. Hayes, 1994a; Hayes & Hayes, 1992a). For instance, an otherwise healthy person may learn a relation between chest pain and heart attack such that in the context of being alone, labeling sensations emanating from the chest as "pain" may also bring about other members of the class leading to a sequence of events by which the person believes that having chest pain equals having a heart attack. Yet, similar cardiac sensations produced during an exercise stress test in the safe context of a cardiology clinic may establish different stimulus relations with very different functions. The relational frame concept is different from the typical stimulus equivalence notions precisely because it places constraints of the types of relations and stimulus functions that are likely to emerge (see Hayes, 1991, for a technical analysis of the types of constrained relations). All verbal behavior, therefore, involves some contextually situated act of framing events relationally in terms of a variety of otherwise arbitrary relations. As such, verbal meaning can be understood as relations between events with a variety of psychological functions (e.g., reinforcing, eliciting, establishing, discriminative, and evocative) which depend on the contextual contingencies that establish and modify them (Hayes, 1994a; Hayes & Hayes, 1989, 1992b). For example, labeling one's bodily sensations as "anxiety" in one instance and "excitement" in another does not necessarily depend on the formal properties of what is felt (they may be the same). It does depend, however, on the application of a particular relational response between what is felt and what is said. It is precisely because humans are capable of contextually controlled relational responding that words can be flexibly applied, modified, and combined in new ways. Common examples of complex contextually controlled relational responding often takes the form of rules which can establish and maintain such relations.
Rules, rule following, and verbal-emotional relations. The term rule has been used by a variety of psychological theorists and means different things to each. In the traditional behavior analytic account, most psychologically significant behavior (i.e., that of the whole organism in and within a context) is contingency shaped. An important subset of this behavior, however, is rule governed. Behavior analysts have emphasized rules in the sense of governing events. According to Reese (1989), most rules can be formally identified by the specification of a relation between two or more events, or in Skinner's (1966, 1969) terms, they represent contingency-specifying stimuli. Hayes and his associates have defined rules simply as verbal antecedents that derive their functions from their participation in relational frames (e.g., Hayes & Wilson, 1994; Zettle & Hayes, 1982). Simply put, an antecedent verbal stimulus relation or rule is something that tells us what to do, when to do it, and what will happen when we do it. Thus, to understand a rule requires established relations between words as specified in the rule, their environmental referents, and stimulus functions which can be thought of as a culmination of complex sets of relational responding (e.g., stimulus equivalence, transfer of functions, and framing relationally). Rules are effective, in part, because they establish verbal relations with other aspects of the nonverbal world (e.g., objects or events). As a result, rule following is possible in entirely different contexts from that in which the rule was established. Rules also functionally establish previously unrelated stimulus functions (e.g., physiological respondents and operant avoidance and escape functions) with events specified in the rule (Hayes, Zettle, & Rosenfarb, 1989). One immediate and beneficial consequence of rule following is that it may bring us into contact with contingencies that would not have been contacted, or would have been otherwise contacted with deleterious effects (e.g., "If you have chest pain, go to the emergency room" or "Don't touch the stove when it is hot"). Alternatively, rule following may have deleterious effects when it results in an insensitivity to contingencies that might suggest otherwise. Such insensitivity is often seen in clients with anxiety or fear-related difficulties (e.g., "I can't speak in front of a group because I will have an anxiety attack" or "I can't fly in planes because I am too scared"). Clinically, many persons who come into therapy have established elaborate rules and stimulus relations that are largely ineffective. In fact, many behavior problems occur when ineffective verbal behavior gets in the way of effective action. If this is the case, the question arises, what can be done about it?
Language-Based Clinical Behavior Analytic Intervention Approaches
Clinical strategies: Changing the therapist-client social-verbal context. According to Kohlenberg, Tsai, and Dougher (1993), clinical behavior analysts seek to understand how the talking that goes on during the session can help the client with problems that occur outside the session in the client's daily life. The answer to this question is derived from the previous principles and is based on three assumptions: (a) Problematic behavior can be directly observed in session; (b) occurrences of problematic behavior in session account for generalization of treatment gains beyond therapy; and (c) many behavior problems are related to verbal behavior. Central to behavior analytic treatment of emotional dysfunction is that clients tend to assume that their thoughts and feelings have a causal role in their problem. For example, clients will often report that they cannot speak in front of a group because they are too anxious, or they cannot fly in a plane because they are too afraid. As a result of such rules, clients seek help from their therapist to get rid of the dysfunctional private event. According to Kohlenberg and colleagues (1993), the problem with attributing causal status to the feeling is that it may direct attention away from the factors that are responsible for both the inclinations to overt behavior as well as the private event. This leads to a corollary assumption made by clinical behavior analysts; that is, what the client thinks, feels, and does vary with environmental events which can be potentiated or depotentiated in the therapeutic context. Although clinical behavior analysts adopt a noncausal view of feelings, they are not ignored in the therapeutic process. Rather, it is because clients often develop a rigid verbal repertoire of interacting with the world that they are left with inflexible and ineffective ways of interacting with that world (Kohlenberg et al., 1993). In their book, Functional Analytic Psychotherapy (FAP), Kohlenberg and Tsai (1991) describe both operant and respondent behaviors relevant to expressing feelings and show how avoidance of feeling can lead to diminished contact with the environment. The FAP approach is generally designed to create a therapeutic environment that evokes, shapes, and reinforces feeling and its expression and acceptance. Thus, a primary clinical strategy is to alter the emotive functions of language (i.e., behavior-behavior relations), and not necessarily the feelings themselves, by establishing a new verbal context in therapy that changes the functions of these relations. This alternative, one that is different from eliminative or emotional control techniques in behavior therapy, is at the core of clinical behavior analytic strategies and is commonly called emotional acceptance.
Emotional acceptance as a process and not a technique. Acceptance-based strategies adopt a different view of private events and psychological health. Whereas nonacceptance generally refers to an unwillingness to experience unpleasant thoughts, feelings, and other private events believed to be the primary problem for the client, acceptance involves allowing thoughts and feelings to occur without any attempt to control them, that is, acceptance entails giving up or letting go of the struggle to change or to control (cf. Dougher, 1994; Hayes, 1994b; Jacobson, 1992). This involves experiencing events fully and without defense for what they are and not as what we say they are. In a technical sense, acceptance involves contacting automatic or direct stimulus functions of events without (a) acting to reduce or manipulate those functions and (b) acting because of their derived verbal functions. Private events are not viewed as the causes of suffering or the obstacles to change.
In contrast, most mainstream behavioral approaches view psychological problems differently, that is, negative or unpleasant thoughts and feelings are the problem, and therefore, must be either eliminated, controlled, or reduced (e.g., Barlow, 1988). Such treatments also maintain, either explicitly or implicitly, that to live a happy and successful life one must experience good feelings and thoughts and that bad or aversive ones are not healthy and are the cause of human suffering. By explicitly targeting these "unhealthy" private events in therapy, many behavior therapists are also assuming that the thoughts or feelings are the source of the problem. Otherwise, it would not make sense to target them for change. Similarly, clients often view their problems in the same terms and see the solution as one of better control. For instance, a panic patient may avoid malls, crowds, or leaving the house, and may self-medicare to prevent, reduce, and avoid experiencing unpleasant bodily sensations. With acceptance-based behavioral therapies, the emphasis is reversed (e.g., Ellis & Robb, 1994; Hayes, 1987, 1994b; Hayes, Jacobson, Follette, & Dougher, 1994; Kohlenberg & Tsai, 1991; Koerner, Jacobson, & Christensen, 1994; Linehah, 1994; Marlatt, 1994; Wulfurt, 1994). Instead of eliminating distressing thoughts or feelings, the focus is on altering the client's struggle for control, often verbally, with the goal of getting clients to behave despite what they think or feel. This is accomplished, in large part, by undermining, weakening, and altering the functions of verbal relations using paradoxical techniques.
Paradoxical verbal interventions and the modification of verbal-emotional functions. Although no unified theoretical statement exists to describe paradoxical interventions, most operate on the interpersonal social-verbal context from which symptoms develop and that maintains problems (cf. Hayes & Melancon, 1989). The term paradox derives from the Latin paradoxum, meaning something that is contrary to opinion or expectation. Logically, this translates into statements that are neither true nor false (e.g., "You must feel anxious in order not to feel anxious"). Clinically, paradoxical techniques commonly take the form of verbal injunctions (e.g., "If you don't want it, you've got it") and symptom prescriptions (e.g., "I want you to make yourself panic each time you leave your house") that, if taken literally, are nonsensical and contrary to what the social-verbal community dictates. Paradoxical techniques are used explicitly in therapy to break relational or functional equivalence classes with the unexpected effect of altering the maintaining functions of these patterns without necessarily changing their form (e.g., reason giving, control over private events, and avoidance of painful feelings). Paradoxical interventions rely on the patient's conceptualizations of the problem and their attempts at solving the problem. As Nardone and Watzlawick (1993) suggest, "clinical experience has shown that, ironically, it is often the patient's very attempts to solve the problem that, in fact, maintains it. The attempted solution becomes the true problem" (p. 51).
For example, anxiety or fear often cease to be mere words, but represent a series of other often involuntary events (i.e., the symptoms) and stimulus functions of what it means to have anxiety. Paradoxical techniques undermine this preexisting system of verbal constructions and reactions regarding reality by breaking the cycle of repetitive behavior that forms the "attempted solution" (cf. Nardone & Watzlavick, 1993). The therapist can deliberately create a paradox such that previously avoided and unwanted symptoms become voluntary. For instance, the therapist might say to a cardiophobic client with persistent chest pain who, despite repeated medical reassurance that she is otherwise healthy, continues to believe she has a heart condition, "Yes, I can see that you are very ill and getting worse . . . given all the negative medical tests you've been through, I'd bet that you are suffering from a rare condition with no known cure . . . it's hopeless." As a consequence, it is likely that reassurance and repeated visits to a cardiologist will no longer be an effective solution to the client's presumed problems (i.e., they will not work). Instead, the client's former solutions about their problems begin to be seen as problems in themselves (Hayes & Wilson, 1994). Paradoxical techniques place both the "symptoms" and language in a different context in which they lose their original function and meaning. Furthermore, they create a therapeutic double bind by implying that the only way to change is to remain unchanged (cf. Hayes & Wilson, 1994). Although outcome data on the clinical efficacy of paradoxical interventions is scant, such techniques are at the core of many effective behavioral (e.g., FAP, Kohlenberg & Tsai, 1991; and ACT, Hayes et al., 1994) and some nonbehavioral approaches (e.g., Bateson, Jackson, Haley, & Weakland, 1963; Nardone & Watzlavick, 1993; Selvini, Boscolo, Cecchin, & Prata, 1978).
Shortfalls of an Operant Analysis of Language-Emotion Relations
Recent advances within applied behavior analysis have taken issue with what has been the thorn in the side of behavior analysis, that is, its ignorance of emotion-behavior relations and complex human behavior. Reluctant to give emotion or other private events any special status, behavior analysis has approached the problem of emotion as one of verbal behavior. Describing the emotive functions of verbal relations has been a significant advancement over Skinner's relative neglect and inconsistent discussion of the topic. These advancements, which largely derived from basic experimental research, have also resulted in several new and exciting behavior analytic psychotherapies. As with PB, applied behavior analytic conceptualizations and treatments have gone largely unnoticed by mainstream behavior therapy. Again, the question is why?
The neglect and oversimplification of respondent verbal learning. Beginning with Skinner (1938, 1957), behavior analysts frequently distinguish between respondent and operant conditioning. Respondents have been viewed as simple elicited reflexive behaviors, whereas operants encompass more complex forms of learning emitted and shaped by consequences. Although Skinner (1957) mentioned respondent eliciting functions of language, many behavior analysts have failed to explore the processes involved systematically. Rather, behavior analysts have described verbal behavior and emotive functions largely using operant learning principles (Hayes, 1991; Hayes & Wilson, 1994) and have dismissed an entire literature showing similar effects via respondent conditioning operations (e.g., Eifert, 1987; Jenkins, 1963; Martin & Levey, 1987; Staats, 1975; Staats & Eifert, 1990). The omission of respondent conditioning in operant accounts of verbal relations stems largely from the difficulties and complexities in doing this kind of research with verbal humans (see Dougher et al., 1994, for one noteworthy exception). Using examples of unidirectional respondent conditioning in infrahumans, operant researchers have noted the failure of the respondent paradigm to account for verbal behavior, and especially, the emergence of bidirectional relations that are not explicitly taught and often delayed (cf. Hayes & Hayes, 1989). Relegating the domain of respondents to simple reflexive behavior involving unidirectional relations in animals does not accurately reflect the complex nature of this form of conditioning, especially verbal semantic conditioning with humans (see Staats, 1994). As Rescorla (1988) points out:
Describing Pavlovian conditioning as the endowing of a CS with the ability to evoke the same response as the US is a wholly inadequate characterization. Pavlovian conditioning is not the shifting of a response from one stimulus to another. Instead, conditioning involves the learning of relations among events that are complexly represented, a learning that can be exhibited in various ways. (p. 158)
In fact, behavior analytic criticisms of semantic conditioning have been of basic respondent conditioning research with infrahumans and not of semantic conditioning - an issue we will return to later. Semantic conditioning, however, heavily involves verbal (arbitrary relational) processes and thus a respondent preparation does not necessarily involve simply a respondent process. The distinction between procedure and process is important in interpreting the large literature on verbal semantic conditioning, which uses respondent conditioning procedures but seems to involve many processes. Likewise, operant procedures and preparations may involve respondent processes such as response generalization based on the acquisition and transfer of shared indirect functions (see Hayes, L. J., 1992, for a respondent analysis of stimulus equivalence). Respondent processes are often overlooked in the stimulus equivalence literature because operant preparations are not designed to test respondent processes. Staats (1975, 1994) has always argued against a sharp distinction between operant and respondent processes which can be derived from either preparation, and several behavior analysts have also begun to arrive at similar conclusions (e.g., Michael, 1993; Schlinger & Blakely, 1987, 1994).
Do operant conceptualizations of stimulus and functional equivalence answer an old respondent problem?. Beginning with Sidman, operant researchers typically lay claim to the notion of stimulus equivalence and the properties it shows (e.g., bidirectionality, transfer of function). The terms stimulus and functional equivalence, however, originated before 1939 and referred to an important phenomenon in the 1960s for S-R learning theorists interested in language (e.g., Hull, 1939; Jenkins, 1963). In fact, semantic conditioning research (e.g., Jenkins, 1963; McNeill, 1968; Osgood, 1969; Staats, 1957b, 1975) demonstrated several verbal behavioral processes and principles (e.g., transfer of function, stimulus equivalence, response equivalence, and bidirectional relations) that operant researchers have subsequently modified and claimed as their own (e.g., Hayes & Hayes, 1992a; Sidman, 1994). Because the "mechanisms" responsible for transfer of function and bidirectionality were not well known, researchers studying these phenomena using S-R learning paradigms typically appealed to mediational processes. Sidman, in contrast, gradually moved away from an appeal to mediational elements to nonmediational stimulus-stimulus associations (see Sidman, 1994, for a historical account). Since then, behavior analysts have extensively relied on Sidman's nonmediational account of equivalence as the working model of complex relational responding in verbal humans. However, many are still unsure about what it is they are studying with equivalence (Hayes, L. J., 1992). Sidman (1994) has argued that equivalence is a fundamental behavioral process only found in humans and others suggest that equivalence depends on language (cf. Hayes & Hayes, 1992a). What then is the basic process and what is it that we are studying with equivalence (cf. Hayes, L. J., 1992)? Given that equivalence relations can emerge from either an S-R or S-S paradigm, one could argue that Sidman and others simply reframed the terms of the equation to fit an operant framework while leaving the phenomena of equivalence unexplained. It is also unclear whether functional classes alone will simply yield stimulus equivalence or vice versa, and if the two represent different processes (Dougher & Markham, 1994). Although much is known about the type of history needed to establish equivalence relations, research has only recently begun to specify how those relations can be modified or changed once established (e.g., Pilgram & Galizio, 1990, 1995). Knowing how to change equivalence classes reliably is critical from an intervention perspective given that equivalence has been applied as a working model and explanation of how language-emotive functions operate. Using stimulus equivalence to explain verbally mediated behavior problems in clients with unknown and complex histories of relational responding may leave some to ask whether the kinds of histories necessary to establish stimulus equivalence in the laboratory represent similar processes as applied to a given client's relational behavior, and, if so, how do we know?
The need to account for verbal repertoires and history. Terms such as history and repertoire are frequently used as explanations or partial explanations in applied behavior analytic accounts of verbal-emotional behavior. History and repertoires derive their importance, in part, because present behavior is said to depend on past behavior in similar or related contexts. In nonlaboratory contexts where the history or repertoires are not known, these terms can be vague, in part, due to the complexities involved in specifying such events in humans. Such specification requires systematic longitudinal research to address the relation between history, repertoires, and present behavior (e.g., how do repertoires develop and function?). Although this kind of research tends to be intensive and time consuming, it is necessary to give repertoire and history any meaning in clinical behavior analytic accounts of human behavior. Unfortunately, the tendency has been to use "repertoire" in general and vague ways. For instance, in Verbal Behavior, Skinner (1957) was quite vague in discussing the importance of verbal repertoires.
We observe that a speaker possesses a verbal repertoire in the sense that responses of various forms appear in his behavior from time to time in relation to identifiable conditions. A repertoire, as a collection of verbal operants, describes the potential behavior of a speaker. To ask where a verbal operant is when a response is not in the course of being emitted is like asking where one's knee jerk is when the physician is not tapping the patellar tendon . . . A repertoire of verbal behavior is a convenient construct. We are concerned here not only with the fact that specific forms of verbal behavior are observed but that they are observed under specific circumstances. These controlling circumstances add a dynamic character to "repertoire." (p. 22)
Although history and repertoire can be used as convenient summaries of behavioral classes, they are vague when they fail to specify the contingencies or functional relations such that a given instance could be classified clearly in terms of those relations. When such specification is available, perhaps repertoire is superfluous. By naming a repertoire as this or that, there is always the risk that others may then treat a repertoire as something that exists apart from descriptions of consequential effects and functional relations. For example, in their discussion of clients with emotional difficulties Hayes and Wilson (1994) invoke "emotional avoidance repertoires" and Kohlenberg and Tsai (1994) refer to clients as possessing "inadequate repertoires." Unfortunately, used in this way, such terms are no better or worse than cognitivists invoking constructs such as schemata, appraisals, nodes, and networks (Poppen, 1989). Unless repertoires can be identified and differentiated from one another in terms of both the histories that give rise to them and their different functions in a given instance (see Staats, 1994, for such a specification) they are of little explanatory value.
The vague and unspecified nature of rule governance. The notion of rule governance remains equally vague and researchers remain divided about what rule governance is, how it is learned, and how it functions in human affairs. One explanation of rule governance is that we tend to follow rules because previous behavior in response to similar verbal stimuli has been reinforced (Skinner, 1969). As Reese (1989) notes, this is an abstract definition because it requires a definition of "following rules" as an operant which can include almost anything. Others have noted that rule-governed behavior is difficult to define because there is no set of common features that may define rule-governed behavior as a functionally distinct category (cf. Cerutti, 1989). Definitional attempts, however, have often resulted in classification of rules as "antecedent verbal stimuli" (e.g., Catania, Matthews, & Shimoff, 1990; Hayes, 1986) or as a verbal description of a contingency (Braam & Malott, 1990). Yet, as Burns and Staats (1991) note, many "rules" do not describe contingencies (see also Schlinger & Blakely, 1994). It is also likely that rules may emerge as consequences of other verbal and nonverbal events via their participation in relational frames, thereby broadening the classification beyond antecedent stimuli. Distinguishing between rule-governed behavior and contingency-shaped behavior is also difficult and many are unsure how to separate the two. Relatedly, behavior analysis has largely left unspecified the kind of history and skills necessary for rule-following. Conceivably, many rules may be understood but not followed because a person lacks the requisite motor skills to follow a rule. In sum, the way in which rules function in human behavior is not well understood.
Other general limitations and unresolved issues. Many applied behavior analysts use terms such as emotion and feelings freely. Yet, the paucity of basic research and integration of both respondent and operant learning concerning human verbal behavior has placed behavior analytic accounts of emotional phenomena in a weakened and largely interpretive position. This is especially true when considering acceptance-based behavior analytic therapies that deal with feeling, thinking, and emoting. Whereas the notion of acceptance has intuitive appeal when understood as verbal relational responding, it is equally difficult to pin down. As Hass (1994) argues, envisioning what a scientific study of acceptance would entail is difficult. From a practical and treatment outcome standpoint, identifying the types of derived relational responding that would increase acceptance would be useful. Are there verbal contexts that enhance acceptance? Can acceptance be instructed? The remainder of this paper brings together the strengths and limitations of both respondent and operant accounts of verbal-emotional relations by suggesting some possible areas for synthesis amid what appears as irreconcilable differences.
A Critical Appraisal of the Two Behaviorisms: Controversies, Contributions, and the Possibility of Synthesis
Distinguishing Multiple Function-Altering Processes From Preparations
Controversy 1: Preparation and process distinctions are no longer viable. As indicated, Skinner viewed respondent and operant conditioning as entirely separate processes (Skinner, 1938, 1974) and he considered emotions as unimportant for a science of behavior (Skinner, 1984). This is not surprising given that Skinner's theory was based largely upon basic research with nonverbal organisms. Historically, behavior analysts have maintained the operant and respondent distinction because of "(a) the absence of an immediate environmental precursor that could be said to elicit the behavior in question; (b) the sensitivity of the behavior to its consequences; and (c) the possibility that distinct response systems are involved in the two types of behavior" (Hineline, 1986, p. 55). Schlinger and Blakely (1994), however, argue against such a distinction on several grounds. First, the terms evoke and elicit do not distinguish operant and respondent behavior because both are defined solely in terms of their temporal controlling relation to subsequent behavior. Rescorla (1988) also points out that respondent conditioning does not require continuity between some CS and UCS, but rather the evocative functions of any given CS are dependent on the base rate of the UCS. Similar base-rate relations are necessary to establish events as reinforcers or punishers in operant conditioning. Second, distinguishing between operant and respondent behavior on the basis that the former is controlled by consequent events and the latter by antecedent events is inaccurate. In fact, both can be evoked (elicited) by antecedent events and shaped by relevant consequences. Thus, Schlinger and Blakely (1994) conclude that the more appropriate distinction between operant and respondent relations is between the types of function-altering histories responsible for the evocative relations that characterize both sets of relations.
The respondent-operant bridge 1: Multiple interrelated functions of language. PB was the first to point out that respondent and operant processes are often difficult to separate and that both respondent or operant preparations can endow stimuli with three interrelated functions. More recently, behavior analysts have broadened the range of stimulus functions that operate in learning preparations by describing the function-altering and establishing operations that give rise to them. Function altering refers to environmental operations that alter the behavioral functions of other stimuli (Schlinger, 1993). Such operations may include, but are not limited to, those typically called respondent, operant, those that produce conditioned reinforcers and punishers, matching to sample, and other verbal operations (e.g., observational and verbal learning, semantic generalization, and stimulus equivalence). As Schlinger (1993) points out, respondent preparations that establish a previous neutral stimulus as a CS, also establish the CS to function as a conditioned reinforcer or punisher. PB makes a similar observation and adds that stimulus events established as reinforcers or punishers in operant preparations will function as CSs to evoke respondents in respondent preparations. Thus, any preparation that results in the transfer of stimulus function meets the description of function altering.
As for emotional learning, the most interesting implication of the function altering notion is that verbal events derived from either a respondent or operant preparation can mimic the effects of both operant and respondent processes. Schlinger (1993) points out that verbal function-altering stimuli can alter several behavioral functions, including (a) the evocative functions of stimuli that mimic the evocative effects of discriminative and motivational events, (b) the evocative functions of stimuli that mimic respondent conditional stimuli, and (c) the reinforcing or punishing functions of stimuli that mimic the same effects that result from nonverbal procedures" (see also Staats, 1994). Thus, making a distinction between respondent and operant functions at the level of process can be difficult, especially in the area of verbal-emotive behavior. However, preparations are important to establish process and semantic conditioning research with its reliance on higher order conditioning has a difficult time accounting for arbitrarily derived stimulus relations and transfer that emerge without direct higher-order pairings. Although stimulus generalization may seem a likely alternative to higher-order conditioning, it can be dispensed with when considering language-emotive functions.
The respondent-operant bridge 2: Transfer of function via stimulus and response generalization. According to the classical use of stimulus generalization, when a response conditioned to one stimulus transfers to another, the degree of transfer varies directly along some gradient of stimulus similarity (e.g., shades of gray). Yet, the classical and technical requirements of stimulus generalization do not typically apply to language, in part, because words often do not fall on a gradient of stimulus similarity and yet somehow come to share the same meaning or stimulus functions. For example, words denoting snakes such as COBRA, SIDEWINDER, and GARTER cannot be placed along a gradient of stimulus similarity but may elicit functionally equivalent responses. Semantic generalization, in contrast, accounts for this apparent limitation by suggesting that the necessary similarities that account for the generalization effect lie in the meaning or conditioned stimulus functions rather than in objective physical or topographical characteristics of the eliciting stimuli (Osgood, 1969). The stimuli are similar only to the extent that they have previously been conditioned to the same or a similar response. For instance, suppose we pair the word PINCH with shock and on a separate occasion we pair the word PAIN with shock. Let us also assume that after several trials both words can elicit a CR. Later, we present the word PAIN as a sample and ask the subject to chose a word with a similar meaning from an additional array of words that includes the word PINCH. It is likely that the subject will choose PINCH as a comparison when given the sample word PAIN and vice versa. Variants of this simple example have been shown in hundreds of studies and show that bidirectional semantic relations can emerge based on respondent preparations without a history of higher-order pairings (see Osgood, 1969, for a review). Thus, words may come to mean the same thing simply because they share similar response functions. What these studies suggest is that semantic generalization is based largely on functional or response equivalence (i.e., shared meaning or psychological functions) rather than stimulus similarity (Razran, 1939); although much of this research relied on mediational elements and is subject to the same criticisms of mediational learning theory noted earlier. Alternatively, stimulus equivalence research has provided some of the best nonmediational preparations to account for transfer of function and emergent relations not explicitly taught or directly trained. It is unclear to us why PB has not attempted to include stimulus equivalence preparations in its account of verbal-emotive functions systematically. This omission has left PB unable to account for derived relational responding which is a ubiquitous aspect of human verbal behavior.
Emoting as a Predisposition to Act or a Cause
Controversy 2: Schisms at the level of causal function of emotion. A related issue concerning emotional phenomena concerns the effects that such events have in influencing subsequent behavior. For instance, how does the experience of aversive bodily events influence present behavior and sensitivity to other events? PB specifically affords emotion a causal role in its analysis, whereas RB does not. Both positions are primarily strategic, in part, because each is consistent with their respective goals and levels of analysis. PB explicitly recognized both importance of emotional phenomena in human behavior and behaviorisms' cursory and ill-defined analysis of this topic as simple conditioning (see Staats, 1963, 1975). Thus, PB attempts to bring emotional phenomena within an experimental and primarily behavioristic analysis by specifying the behavioral histories responsible for emotional behavior that explicitly includes language behavior (cf. Staats & Eifert, 1990). In so doing, PB argues that private events that we otherwise call emotional can and do have stimulus properties that control other behavior - an observation consistent with the vast literature on emotional behavior and operant work on drug discrimination in animals. Behavior analysts, in contrast, typically make the strategic assumption that causal events are of the same type as permitting control (i.e., control resides in the environment and behavior is a dependent variable with no causal status). According to Skinner (1974), "in the end we find ourselves dealing with two events - the emotional behavior and the manipulable environmental conditions of which that behavior is a function - which comprise the proper subject matter for the study of emotion" (p. 180). Yet, Skinner's frequent reference to emotions in his writings gives an unclear picture about their causal status. For example, in Science and Human Behavior, Skinner writes "interoceptive stimuli (e.g., the sensations that arise from the respiratory, circulatory systems) are the principal stimuli to which one reacts in feeling and emotion" (p. 261, emphasis ours). Later, in Verbal Behavior, he writes "emotional stimuli do not elicit responses, they establish dispositions to behave" (p. 158). Thus, one is left to ask whether private physiological events can have stimulus properties that control human behavior?
Respondent-operant bridge 3: Accounting for emotive events as established predispositions to act. One behavior analytic principle that appears relevant to emotive events influencing subsequent behavior is known as an establishing operation. According to Michael (1993), an establishing operation (EO) refers to an environmental event, operation, or stimulus condition that affects the organism by momentarily altering (a) the reinforcing effectiveness of other events and (b) the frequency of occurrence of that part of the organism's repertoire relevant to those events and consequences. The former operation is called reinforcer establishing whereas the latter is termed evocative (i.e., eliciting in respondent terms). Michael argues that most events, operations (e.g., deprivation and satiation) or stimuli that alter the reinforcing effectiveness of other events also alter the momentary frequency of occurrence of any behavior followed by those other events. Similarly, others have argued that operant reinforcement preparations also alter the evocative function of stimulus events (e.g., Schlinger & Blakey, 1994; see also Hayes et al., 1989, for a related discussion). Although behavior analysts use different terms than PB's A-R-D theory, they are essentially describing similar processes. Both views suggest that when anxiety or fear-related behavior is evoked, the functions of other events in the environment may change such that previously established reinforcers no longer function in that way.
We might also ask how anxiety or fear becomes established in the first place. Most clients with an anxiety problem do not report doing an emotion, that is, emitting emotional behavior as an operant. Rather, emotions such as anxiety or fear may be elicited or evoked to predispose a person for action and thus may be established by respondent processes. In this view, "e-motion" literally means to elicit or evoke motion (action) and suggests that emotive events, at least initially, may often involve respondents which can be evoked either interoceptively or exteroceptively and shaped by consequences. Whereas PB has included the stimulus and evocative properties of emotive events in its account, clinical behavior analysis has largely ignored this issue and quite possibly for good reasons. Foremost, any functional account of verbal-emotive behavior must act upon the environmental events that either potentiate or depotentiate the act of feeling. Therefore, to say that private events evoke or cause behavior would place behavior analysts in a weak position because those events cannot be changed directly without first acting upon the environment. In this view, operant behavior can produce UCRs that, in turn, also function as an establishing stimulus for operant behavior (Hayes et al., 1989). Because the stimulus properties of the UCR also function verbally, it is easy to see how those evocative stimulus properties can transfer through relational classes as a function of the conditions that produce them. PB, in contrast, has taken what is an essentially useful account of the multiple stimulus functions of language-emotive relations and has come dangerously close to reifying it in its theoretical development. Although this issue is far from settled, it does appear that both PB and RB are addressing similar function-altering processes and could proceed in a pragmatically useful and cooperative way if each approach was willing to incorporate the other's useful contributions systematically.
A Clinical Example of Synthesis: Phobic Fear
Both PB and RB argue and provide evidence for the contention that individuals do not require direct painful or traumatic conditioning experiences to develop fears, anxiety, or phobias. Instead, it is the association of words or other symbolic stimuli that belong in a relational class with other stimuli with aversive evocative functions in a particular context that is sufficient. In other words, aspects of the nonverbal world come to function as verbal stimuli via language-symbolic processes. With this is mind, two questions arise: (a) Are we able to use the principles derived from both perspectives to account for phobic fear acquisition?, and (b) how would a behavior therapist use these principles together to affect behavioral change in treatment?
According to PB, the respondent association of negative-emotion eliciting verbal stimuli with certain objects or events is the basis of emotional meaning (cf. Eifert, 1990). Behavior analysis adds to this by introducing contextual control. Thus, the eliciting or evocative functions of words or other verbal and nonverbal stimulus events will depend on the circumstances that select those functions. With phobic anxiety, those functions will largely be affective (i.e., eliciting or evocative), punishing, and directive (i.e., escape and avoidance behavior). Thus, when a particular context selects a particular stimulus, PB predicts that three additional functions also occur. Thereafter, other stimuli in that context may also acquire similar functions to the extent that they are evoked by the context, that is, the physical environment/stimulus or the verbal labeling thereof, which are functionally identical (e.g., "[thinking about] a snake crawling into my sleeping bag while camping freaks me out"). Furthermore, PB suggests that persons can condition and recondition themselves by providing their own verbal-symbolic stimuli via higher-order conditioning. Behavior analytic approaches also agree that this may occur; however, how, when, and why people recondition themselves depends on the context that selects for those relations. Behavior analysis suggests that new relations, or in this case new phobic stimuli, can emerge in ways not explicitly taught via stimulus equivalence and transfer of functions. For example, if a person learns that PAIN goes with SPIDER and PAIN goes with SNAKE, the same person may also learn that SPIDER and SNAKE go together without explicit training. Furthermore, if PAIN has acquired a variety of aversive functions (e.g., bruises, bites, stings, burns), then it is likely that SPIDER and SNAKE will also acquire these functions without explicit training via transfer of function. At this point, however, PB predicts that SPIDER and SNAKE will also acquire eliciting, punishing, and avoidance or escape functions. The multiple stimulus functions of verbal-emotive events has not been formally addressed in clinical behavior analytic accounts, although, this may change (e.g., function-altering, establishing operations).
Turning to treatment, both PB and RB rely heavily on semantic-based interventions and each approach attempts to alter phobic avoidance differently. Whereas PB often targets the stimulus functions of language directly via semantic counterconditioning, behavior analytic therapies target the functions of language indirectly using paradoxical techniques. In both cases, an attempt is made to undermine emotional avoidance and control over the occurrence of painful or unwanted bodily sensations. As indicated, the labeling of bodily arousal as aversive or unpleasant is an example of verbal learning and subject to the same principles that control other behavior. In turn, phobic clients will do everything they can to escape or avoid confrontation with, or "thinking" about, the feared stimulus. In the process, they will also produce the very "feelings" that they want to avoid and may inadvertently establish functions associated with unpleasant bodily sensations with other stimulus events. Applied behavior analytic approaches are reluctant to acknowledge that the stimulus properties of bodily events can control behavior; however, this is exactly what we are led to believe given their analysis. PB has specified how private events can have stimulus properties and evoke behavior (e.g., Staats & Eifert, 1990). Language-based behavior therapies, such as semantic counterconditioning and the systematic use of emotionally relevant and behavior-directive self-statements (cf. Eifert, 1987; Hekmat, 1990; Staats, 1972), are aimed at changing the client's overt behavior indirectly by altering the A-R-D functions of verbal events in the client's repertoire.
Although not explicitly stated, one goal of behavior analytic verbal techniques is to undermine a client's verbal-emotional repertoire of reason-giving and control over emotional events by establishing a new social-verbal context in therapy that does not fit with the client's existing verbal-emotional repertoire. We have argued that verbal interventions do change the verbal-emotional repertoire in a way that enables clients to generalize treatment gains in a social-verbal community that still upholds reason-giving, control, and emotional avoidance as desirable (cf. Forsyth & Eifert, in press). Traditional language-based behavior therapy interventions have been successful for several problems and with many clients (Eifert, 1990), but these techniques aim to gain control over or eliminate aversive bodily sensations and responses. As a consequence, they may play into the very social-verbal system that clinical behavior analytic approaches consider the source of the primary problem (Hayes, 1987; Hayes, 1994b). This problem could account for some treatment failures with traditional sematic behavior therapies, and behavior analytic psychotherapies may offer an intriguing alternative in those cases.
Therapists who carefully use reframing or paradoxical verbal interventions may rapidly and simultaneously alter the affective, reinforcing, and directive functions of both verbal and other stimuli in the client's natural environment. Alessi (1992) suggests that reframing and paradoxical directives might be the result of verbal analog conditioning based on semantic conditioning principles. We need to explore systematically the possibility of combining traditional language-based behavioral interventions with paradoxical acceptance-based interventions. Will semantic counterconditioning facilitate acceptance? Will using acceptance at the beginning of therapy make subsequent applications of other semantic verbal interventions more successful? More appropriate interventions may be planned by selecting key stimuli from the problem situation, changing their stimulus functions, and inserting them within the therapist-client verbal interaction. This would represent an exciting merger of PB's language-based therapeutic techniques with acceptance-based behavior analytic treatments.
Summary and Conclusions
The ubiquity of emotion and emotional phenomena in basic and applied areas within psychology can neither be overstated nor its role in human behavior understated. Yet, the exact nature of emotional phenomena and its role in human behavior is not well understood by much of psychology including behavior therapy. A central reason for this is that researchers and clinicians alike have assumed that emotions exist and that they are things to be discovered rather than what people do. This has resulted in a central preoccupation with the many response dimensions of emotion (e.g., cognitive, physiological, and overt behavioral) in an attempt to address "what emotion is or is not." Lazarus (1984) argued that this is the fundamental question faced by emotion theorists; however, it may equally have been the wrong question contributing to an abandonment of conditioning approaches and a weakened commitment to learning theory.
This article began by asking a different and perhaps more fundamental question about emotional phenomena: How do people come to know their feelings? The answer to this question entailed an analysis of language and its functions and conditions that establish these relations. This is a fundamental question about meaning which, for any behavioral analysis, dictates a consideration of how terms are used and the conditions that lead to their use (Ryle, 1949; Wittgenstein, 1967). Both PB and RB take the perspective that language is at the core of how we come to know our feelings. The corollary of this perspective is that feelings are necessarily related to language. Thus, when we speak of feelings such as anxiety or fear we are using language, just as when we evaluate a "feeling" we are basing this evaluation on language or other symbolic processes. Both relations are taught, modified, and changed by experience within a particular social-verbal community that ultimately produces them. Thus, in keeping with the functionalistic behavioral tradition, both approaches explicitly avoid categorizing or defining different emotions apart from the conditions that lead to verbal-emotive relations. The advantages of this noncategorical approach to emotional phenomena allows for flexibility and carefully crafted functional analysis of individual client problems and treatments appropriately tailored to such problems (cf. Eifert et al., 1990).
More behavior analytic work needs to be done concerning emotional phenomena and the behaviorisms have only begun to scratch the surface. Artificially separating the often related functions derived from either preparation has contributed to an "either or" account of how language-emotive relations develop and function. To take advantage of the contributions of the behaviorisms, however, calls for cooperation and synthesis. Cooperation could begin by reexamining the viability of the view that respondent and operant verbal preparations involve entirely different processes. PB could benefit by taking a closer look at the nonmediational stimulus equivalence research and the contextual variables shown to potentiate and depotentiate the stimulus functions of language relations. Likewise, RB may find value in the respondent semantic conditioning work, and particularly, PB's account of the multiple interrelated functions of language and how they relate to verbal-emotional repertoires in influencing subsequent learning. Such an analysis will entail examination of the verbal, and not necessarily veridical, relation between language and feelings and how these relations are established, modified, and changed as a function of the conditions that lead to them. This cooperative approach may not address what emotion is or is not, but it may bring cognitive-behavior therapy closer to addressing a more fundamental question of what emotion "is for."
We thank William Dube and an anonymous reviewer for their helpful comments. Correspondence concerning this article should be addressed to John P. Forsyth, Department of Psychology, West Virginia University, Box 6040, Morgantown, WV 26506-6040.
ALESSI, G. (1992). Models of proximate and ultimate causation in psychology. The American Psychologist, 47, 1359-1370.
AVERILL, J. R. (1980). Emotion and anxiety: Sociocultural, biological, and psychological determinants. In A. O. Rorty (Ed.), Explaining emotions (pp. 37-72). Berkeley: University of California Press.
BANDURA, A. (1978). Reflections on self-efficacy. Advances in Behaviour Research and Therapy, 1, 237-269.
BANDURA, A. (1984). Recycling misconceptions of perceived self-efficacy. Cognitive Therapy and Research, 8, 231-255.
BARLOW, D. H. (1988). Anxiety and its disorders: The nature and treatment of anxiety and panic. New York: Guilford Press.
BATESON, G., JACKSON, D. D., HALEY, J., & WEAKLAND, J. H. (1963). A note on the double-bind-1962. Family Processes, 2, 154-161.
BECK, A. T., & EMERY, G. (1985). Anxiety disorders and phobias: A cognitive perspective. New York: Basic Books.
BERSCHEID, E. (1990). Contemporary vocabularies of emotion. In B. S. Moore & A. M. Isen (Eds.), Affect and social behavior (pp. 22-38). New York: Cambridge University Press.
BRAAM, C., & MALOTT, R. W. (1990). "I'll do it when the snow melts": The effects of deadlines and delayed outcomes on rule-governed behavior in preschool children. The Analysis of Verbal Behavior, 8, 67-76.
BURNS, G. L., & STAATS, A. W. (1991). Rule-governed behavior: Unifying radical and paradigmatic behaviorism. The Analysis of Verbal Behavior, 9, 127-143.
CATANIA, A. C., MATTHEWS, B. A., & SHIMOFF, E. (1990). Properties of rule-governed behavior and their implications. In D. E. Blackman & H. Lejeune (Eds.), Behavior analysis in theory and practice: Contributions and controversies (pp. 215-230). Hove, England: Erlbaum.
CERUTTI, D. T. (1989). Discrimination theory of rule-governed behavior. Journal of the Experimental Analysis of Behavior, 51, 259-276.
COFER, C. H., & FOLEY, J. P. (1942). Mediated generalization and the interpretation of verbal behavior. I. prolegomena. Psychological Review, 49, 513-540.
DOUGHER, M. J. (1994). The act of acceptance. In S. C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and change: Content and context in psychotherapy (pp. 37-45). Reno, NV: Context Press.
DOUGHER, M. J., AUGUSTSON, E., MARKHAM, M. R., GREENWAY, D. E., & WULFERT, E. (1994). The transfer of respondent eliciting and extinction functions through stimulus equivalence classes. Journal of the Experimental Analysis of Behavior, 62, 331-352.
DOUGHER, M. J., & MARKHAM, M. R. (1994). Stimulus equivalence, functional equivalence, and the transfer of function. In S.C. Hayes, L. J. Hayes, M. Sato, & K. Ono (Eds.), Behavior analysis of language and cognition (pp. 71-90). Reno, NV: Context Press.
EIFERT, G. H. (1984). The effects of language conditioning on various aspects of anxiety. Behaviour Research and Therapy, 22, 13-21.
EIFERT, G. H. (1987). Language conditioning: Clinical issues and applications in behavior therapy. In H. J. Eysenck & I. M. Martin (Eds.), Theoretical foundations of behavior therapy (pp.167-193). New York: Plenum.
EIFERT, G. H. (1990). The acquisition and treatment of phobic anxiety: A paradigmatic behavioral perspective. In G. H. Eifert & I. M. Evans (Eds.), Unifying behavior therapy: Contributions of paradigmatic behaviorism (pp. 173-200). New York: Springer.
EIFERT, G. H. (1992). Cardiophobia: A paradigmatic behavioral model of heart-focused anxiety and non-angina chest pain. Behaviour Research and Therapy, 30, 329-345.
EIFERT, G. H. (in press). Why we need to match treatments to clients and not to labels: At least some of the time. Journal of Behavior Therapy and Experimental Psychiatry.
EIFERT, G. H., EVANS, I. M., & MCKENDRICK, V. (1990). Matching treatments to client problems not diagnostic labels: A case for paradigmatic behavior therapy. Journal of Behavior Therapy and Experimental Psychiatry 21, 163-172.
EIFERT, G. H., FORSYTH, J. P., & SCHAUSS, S. L. (1993). Unifying the field: Developing an integrative paradigm for behavior therapy. Journal of Behavior Therapy and Experimental Psychiatry, 24, 107-118.
EIFERT, G. H., & SCHERMELLEH, K. (1985). Language conditioning, emotional instructions, and cognitions in conditioned responses to fear-relevant and fear-irrelevant stimuli. Journal of Behavior Therapy and Experimental Psychiatry, 16, 101-110.
ELLIS, A., & ROBB, H. (1994). Acceptance in rational-emotive therapy. In S. C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and change: Content and context in psychotherapy (pp. 91-102). Reno, NV: Context Press.
EYSENCK, H. J. (1987). Behavior therapy. In H. J. Eysenck & I. Martin (Eds.), Theoretical foundations of behavior therapy (pp. 3-34). New York: Plenum.
FINLEY, J. R., & STAATS, A. W. (1967). Evaluative meaning words as reinforcing stimuli. Journal of Verbal Learning and Verbal Behavior, 6, 193-197.
FORSYTH, J. P., & EIFERT, G. H. (in press). "Cleaning-up cognition" in triple response fear assessment using an individualized functional-analytic approach. Journal of Behavior Therapy and Experimental Psychiatry.
GATCH, M. B., & OSBORNE, J. G. (1989). Transfer of contextual stimulus function via equivalence class development. Journal of the Experimental Analysis of Behavior, 51, 369-378.
GREENWAY, D. E., DOUGHER, M. J., & WULFURT, E. (1992, May). Transfer of consequential functions via stimulus equivalence: The role of generalization. Paper presented at the 18th annual meeting of the Association for Behavior Analysis, San Francisco.
HARMS, J. Y., & STAATS, A. W. (1978). Food deprivation and conditioned reinforcing value of food words: Interaction of Pavlovian and instrumental conditioning. Bulletin of the Psychonomic Society, 12, 294-296.
HASS, J. R. (1994). The elusive nature of acceptance. In S. C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and chance: Content and context in psychotherapy (pp. 33-35). Reno, NV: Context Press.
HAYES, L. J. (1992). Equivalence as process. In S. C. Hayes & L. J. Hayes (Eds.), Understanding verbal-relations (pp. 97-108). Reno, NV: Context Press.
HAYES, S. C. (1986). The case of the silent dog-verbal reports and the analysis of rules. A review of Ericson and Simon's Protocol Analysis: Verbal Reports as Data. Journal of the Experimental Analysis of Behavior, 45, 351-363.
HAYES, S. C. (1987). A contextual approach to therapeutic change. In N. S. Jacobson (Ed.), Psychotherapists in clinical practice: Cognitive and behavioral perspectives (pp. 327-387). New York: Guilford.
HAYES, S. C. (1991). A relational control theory of stimulus equivalence. In L. J. Hayes & P. N. Chase (Eds.), Dialogues on verbal behavior (pp. 19-40). Reno, NV: Context Press.
HAYES, S. C. (1994a). Relational frame theory: A functional approach to verbal events. In S. C. Hayes, L. J. Hayes, M. Sato, & K. Ono (Eds.), Behavior analysis of language and cognition (pp. 930). Reno, NV: Context Press.
HAYES, S. C. (1994b). Content, context, and the types of psychological acceptance. In S. C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and change: Content and context in psychotherapy (pp. 13-32). Reno, NV: Context Press.
HAYES, S. C., & HAYES, L. J. (1989). The verbal action of the listener as a basis for rule-governance. In S.C. Hayes (Ed.), Rule-governed behavior (pp. 153-190). New York: Plenum.
HAYES, S. C., & HAYES, L. J. (1992a). Some clinical implications of contextualistic behaviorism: The example of cognition. Behavior Therapy, 23, 225-249.
HAYES, S. C., & HAYES, L. J. (1992b). Verbal relations and the evolution of behavior analysis. American Psychologist, 47, 1383-1395.
HAYES, S. C., JACOBSON, N. S., FOLLETTE, V. M., & DOUGHER, M. J. (Eds.) (1994). Acceptance and change: Content and context in Psychotherapy. Reno, NV: Context Press.
HAYES, S. C., KOHLENBERG, B. S., & HAYES, L. J. (1991). The transfer of specific and general consequential functions through simple and conditional equivalence classes. Journal of the Experimental Analysis of Behavior, 56, 119-137.
HAYES, S. C., & MELANCON, S. M. (1989). Comprehensive distancing, paradox, and the treatment of emotional avoidance. In M. Ascher (Ed.), Paradoxical procedures in psychotherapy. New York: Guilford Press.
HAYES, S. C., & WILSON, K. G. (1993). Some applied implications of a contemporary behavior-analytic account of verbal events. The Behavior Analyst, 16, 283-301.
HAYES, S. C., & WILSON, K. G. (1994). Acceptance and commitment therapy: Altering the verbal support for experiential avoidance. The Behavior Analyst, 17, 289-303.
HAYES, S. C., ZETTLE, R. D., & ROSENFARB, I. (1989). Rule-following. In S.C. Hayes (Ed.), Rule-governed behavior (pp. 191-220). New York: Plenum.
HEIBY, E. M., & STAATS, A. W. (1990). Depression: Classification, explanation, and treatment. In G. H. Eifert & I. M. Evans (Eds.), Unifying behavior therapy: Contributions of paradigmatic behaviorism (pp. 220-246). New York: Springer.
HEKMAT, H. (1977). Semantic behavior therapy: Unidimensional or multidimensional? Behavior Therapy, 8, 805-809.
HEKMAT, H. (1990). Semantic behavior therapy of anxiety disorders: An integrative approach. In G. H. Eifert & I. M. Evans (Eds.), Unifying behavior therapy: Contributions of paradigmatic behaviorism (pp. 201 219). New York: Springer.
HILLMAN, J. (1960). Emotion: A comprehensive phenomenology of theories and their meanings for therapy. London: Routledge & Kegan Paul.
HINELINE, P. N. (1986). Retuning the operant-respondent distinction. In T. Thompson & M. Zeiler (Eds.), Analysis and integration of behavioral units (pp. 55-79). Hillsdale, NJ: Erlbaum.
HULL, C. L. (1939). The problem of stimulus equivalence in behavior theory. Psychological Review, 46, 9-30.
HULL, C. L. (1943). Principles of behavior. New York: Appleton-Century-Crofts.
JACOBSON, N. S. (1992). Behavioral couple therapy: A new beginning. Behavior Therapy, 23, 493-506.
JAMES, C. T., & HAKES, D. T. (1965). Mediated transfer in a four-stage, stimulus equivalence paradigm. Journal of Verbal Learning and Verbal Behavior, 4, 89-93.
JENKINS, J. J. (1963). Mediated associations: Paradigms and situations. In C. N. Corer & B. S. Musgrave (Eds.), Verbal behavior and learning: Problems and processes (pp. 210-245). New York: McGraw Hill.
JENKINS, J. J., & PALERMO, D. S. (1964). Mediation processes and the acquisition of linguistic structure. In U. Bellugi & R. Brown (Eds.), The acquisition of language: Monographs of the society for research or child development, 29, 141-169.
KJELDERGAARD, P. M. (1968). Transfer and mediation in verbal learning. In T. R. Dixon & D. L. Horton (Eds.), Verbal behavior and general behavior theory (pp. 67-96). New Jersey: Prentice-Hall.
KOERNER, K., JACOBSON, N. S., & CHRISTENSEN, A. (1994). Emotional acceptance in integrative behavioral couple therapy. In S.C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and change: Content and context in psychotherapy (pp. 109-118). Reno, NV: Context Press.
KOHLENBERG, B. S., HAYES, S. C., & HAYES, L. J. (1991). The transfer of contextual control over equivalence classes through equivalence classes: A possible model of social stereotyping. Journal of the Experimental Analysis of Behavior, 56, 505-518.
KOHLENBERG, B. S., & TSAI, M. (1991). Functional analytic psychotherapy. New York: Plenum.
KOHLENBERG, R. J., & TSAI, M. (1994). Improving cognitive therapy for depression with functional analytic psychotherapy: Theory and case study. The Behavior Analyst, 17, 305-319.
KOHLENBERG, B. S., TSAI, M., & DOUGHER, M. J. (1993). The dimensions of clinical behavior analysis. The Behavior Analyst, 16, 271-282.
LANG, P. J. (1977). Imagery in therapy: An information processing analysis of fear. Behavior Therapy, 8, 862-886.
LANG, P. J. (1984). Cognition in emotion: Concept and action. In C. Izard, J. Kagan, & R. Zajonc (Eds.), Emotions, cognition, and behavior. New York: Cambridge University Press.
LANG, P. J. (1985). The cognitive psychophysiology of emotion: Fear and anxiety. In A. H. Tuma & J. Maser (Eds.), Anxiety and the anxiety disorders. Hillsdale, NJ: Erlbaum.
LAZAR, R. M., & KOTLARCHYK, B. J. (1986). Second-order control of sequence-class equivalences in children. Behavioral Processes, 13, 205-215.
LAZARUS, R. S. (1982). Thoughts on the relations between emotion and cognition. American Psychologist, 37, 1019-1024.
LAZARUS, R. S. (1984). On the primacy of cognition. American Psychologist, 39, 124-129.
LAZARUS, R. S. (1991). Cognition and motivation in emotion. American Psychologist, 46, 352-367.
LINEHAN, M. M. (1994). Acceptance and change: The central dialectic in psychotherapy. In S. C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and change: Content and context in psychotherapy (pp. 73-86). Reno, NV: Context Press.
LOHR, J. M., & HAMBERGER, L. K. (1990). Verbal, emotional, and imagery repertoires in the regulation of dysfunctional behavior: An integrative conceptual framework for cognitive-behavioral disorders and interventions. In G. H. Eifert & I. M. Evans (Eds.), Unifying behavior therapy: Contributions of paradigmatic behaviorism (pp. 153-172). New York: Springer.
MALTZMAN, I. (1968). Theoretical conceptions of semantic conditioning and generalization. In T. R. Dixon & D. L. Horton (Eds.), Verbal behavior and general behavior theory (pp. 291-339). New Jersey: Prentice Hall.
MALTZMAN, I., RASKIN, D. C., GOULD, J., & JOHNSON, O. (June, 1965). Individual differences in the orienting reflex and semantic conditioning and generalization under different UCS intensities. Paper presented at the annual meeting of the Western Psychological Association, Honolulu.
MARLATT, G. A. (1994). Addiction, mindfulness, and acceptance. In S. C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and change: Content and context in psychotherapy (pp. 175-197). Reno, NV: Context Press.
MARTIN, I. M., & LEVEY, A. B. (1978). Evaluative conditioning. Advances in Behaviour Research and Therapy, 1, 57-102.
MARTIN, I. M., & LEVEY, A. B. (1987). Knowledge, action, and control. In H. J. Eysenck & I. M. Martin (Eds.), Theoretical foundations of behavior therapy (pp. 133-152). New York: Plenum.
MCNEILL, D. (1968). On theories of language acquisition. In T. R. Dixon & D. L. Horton (Eds.), Verbal behavior and general behavior theory (pp. 406-420). New Jersey: Prentice Hall.
MEICHENBAUM, D. (1977). Cognitive-behavior modification: An integrative approach. New York: Plenum Press.
MEYER, M. F. (1933). That whale among the fishes-theory of emotions. Psychological Review, 40, 292-300.
MICHAEL, J. (1983). Evocative and repertoire-altering effects of an environmental event. The Analysis of Verbal Behavior, 2, 19-21.
MICHAEL, J. (1993). Establishing operations. The Behavior Analyst, 16, 191-206.
MOWRER, O. H. (1954). The psychologist looks at language. American Psychologist, 9, 660-694.
NARDONE, G., & WATZLAWICK, P. (1993). Clinical practice, processes, and procedures. In G. Nardone & P. Watzlawick (Eds.), The art of change (pp. 45-72). San Francisco, CA: Jossey-Bass.
NUNNALLY, J. C., DUCHNOWSKI, A. J., & PARKER, R. K. (1965). Association of neutral words with rewards. Journal of Personality and Social Psychology, 1,270-274.
OSGOOD, C. E. (1969). The nature and measurement of meaning. In J. G. Snider & C. E. Osgood (Eds.), Semantic differential technique: A sourcebook (pp. 3-41). New York: Aldine-Atherton.
PALMER, D. C., & DONAHOE, J. W. (1992). Essentialism and selectionism in cognitive science and behavior analysis. American Psychologist, 47, 1344-1358.
PAVLOV, I. P. (1927). Conditioned reflexes: An investigation of the activity of the cerebral cortex. New York: Dover.
PILGRAM, C., & GALIZIO, M. (1990). Relations between baseline contingencies and equivalence probe performances. Journal of the Experimental Analysis of Behavior, 54, 213-224.
PILGRAM, C., & GALIZIO, M. (1995). Reversal of baseline relations and stimulus equivalence: I. Adults. Journal of the Experimental Analysis of Behavior, 63, 225-238.
POPPEN, R. L. (1989). Some clinical implications of rule-governed behavior. In S. C. Hayes (Ed.), Rule-governed behavior: Cognition, contingencies, and instructional control (pp. 325-357). New York: Plenum.
RACHMAN, S. (1977). The conditioning theory of fear acquisition. Behaviour Research and Therapy, 15, 375-387.
RACHMAN, S. (1981). The primacy of affect: Some theoretical implications. Behaviour Research and Therapy, 19, 279-290.
RACHMAN, S. (1984). A reassessment of the "primacy of affect." Cognitive Therapy and Research, 8, 579-584.
RAZRAN, G. (1939). A quantitative study of meaning by conditioned salivary technique (semantic conditioning). Science, 90, 89-91.
RAZRAN, G. H. S. (1961). The observable unconscious and the inferable conscious in current Soviet psychology: Interoceptive conditioning, semantic conditioning, and the orienting reflex. Psychological Review, 68, 81-147.
REESE, H. W. (1989). Rules and rule-governance: Cognitive and behavioristic views. In S.C. Hayes (Ed.), Rule-governed behavior (pp. 3-84). New York: Plenum.
RESCORLA, R. A. (1988). Pavlovian conditioning: It's not what you think it is. American Psychologist, 43, 151-160.
RYLE, G. (1949). The concept of mind. New York: Barnes and Noble.
SALZlNGER, K., & PISONI, S. (1958). Reinforcement of affect responses in schizophrenics during the clinical interview. Journal of Abnormal and Social Psychology, 57, 84-90.
SALZINGER, K., & PISONI, S. (1960). Reinforcement of verbal affect responses of normal subjects during the interview. Journal of Abnormal and Social Psychology, 60, 127-130.
SALZINGER, K., & PISONI, S. (1961). Some parameters of the conditioning of verbal affect responses in schizophrenic patients. Journal of Abnormal and Social Psychology, 63, 511-516.
SCHAUSS, S. L., & FORSYTH, J. P. (1994). Functional analysis and functional interpretation: One is not the other. The Naturalist, 1, 32-34.
SCHLINGER, H. D. (1993). Separating the discriminative and function altering effects of verbal stimuli. The Behavior Analyst, 16, 9-23.
SCHLINGER, H. D., & BLAKELY, E. (1987). Function-altering effects of contingency-specifying stimuli. The Behavior Analyst, 10, 41-45.
SCHLINGER, H. D., & BLAKELY, E. (1994). A descriptive taxonomy of environmental operations and its implications for behavior analysis. The Behavior Analyst, 17, 43-57.
SCRUTON, R. (1980). Emotion, practical knowledge and common culture. In A. O. Rorty (Ed.), Explaining emotions (pp. 519-536). Berkeley: University of California Press.
SELIGMAN, M. E. P. (1971). Phobias and preparedness. Behavior Therapy, 2, 307-320.
SELVINI, P. M., BOSCOLO, L., CECCHIN, G., & PRATA, G. (1978). Paradox and counter-paradox: A new model in the therapy of the family in schizophrenic transaction. New York: Aronson.
SIDMAN, M. (1992). Equivalence relations: Some basic considerations. In S. C. Hayes & L. J. Hayes (Eds.), Understanding verbal relations (pp. 15-27). Reno, NV: Context Press.
SIDMAN, M. (1994). Equivalence relations and behavior: A research story. Boston, MA: Authors Cooperative.
SIDMAN, M., & TAILBY, W. (1983). Conditional discrimination vs. matching to sample: An extension of the testing paradigm. Journal of the Experimental Analysis of Behavior, 37, 23-44.
SILVERSTEIN, A. (1973). Acquired pleasantness and conditioned incentives in verbal learning. In D. E. Berlyne & K. B. Madsen (Eds.), Pleasure, reward, preference (pp. 177-226). New York: Academic Press.
SKINNER, B. F. (1938). The behavior of organisms. New York: Appleton.
SKINNER, B. F. (1953). Science and human behavior. New York: MacMillan.
SKINNER, B. F. (1966). An operant analysis of problem solving. In B. Kleinmuntz (Ed.), Problem solving: Research, method, and theory (pp. 225-257). New York: Wiley.
SKINNER, B. F. (1969). Contingencies of reinforcement: A theoretical analysis. Englewood Cliffs, NJ: Prentice-Hall.
SKINNER, B. F. (1957). Verbal behavior. New York: Appleton-Century-Crofts.
SKINNER, B. F. (1974). About behaviorism. New York: Knopf.
SKINNER, B. F. (1984). The operational analysis of psychological terms. Behavioral and Brain Sciences, 7, 547-581.
STAATS, A. W. (1957a). Learning theory and "opposite speech." Journal of Abnormal and Social Psychology, 55, 268-269.
STAATS, A. W. (1957b). Verbal and instrumental response hierarchies and their relationship to problem solving. American Journal of Psychology, 70, 442-446.
STAATS, A. W. (1963). Complex human behavior. (with contributions by C. K. Staats). New York: Holt, Rinehart & Winston.
STAATS, A. W. (1972). Language behavior therapy: A derivative of social behaviorism. Behavior Therapy, 3, 165-192.
STAATS, A. W. (1975). Social behaviorism. London: Erwin-Dorsey Press.
STAATS, A. W. (1988). Skinner's theory and the emotion-behavior relationship: Incipient change with major implications. American Psychologist, 43, 747-748.
STAATS, A. W. (1989). Anxiety disorders. In A. W. Staats (Ed.), Personality and abnormal behavior: A paradigmatic behaviorism approach. Unpublished manuscript (Available from A. W. Staats, Dept. of Psychology, University of Hawaii, Honolulu, HI 96882).
STAATS, A. W. (1990). Paradigmatic behavior therapy: A unified framework for theory, research, and practice. In G. H. Eifert & I. M. Evans (Eds.), Unifying behavior therapy: Contributions of paradigmatic behaviorism (pp. 14-54). New York: Springer.
STAATS, A. W. (1994). Psychological behaviorism and behaviorizing psychology. The Behavior Analyst, 17, 93-114.
STAATS, A. W., & BURNS, G. L. (1982). Personality specification and interaction theory. Journal of Personality and Social Psychology, 43, 873-881.
STAATS, A. W., & EIFERT, G. H. (1990). The paradigmatic behaviorism theory of emotions: Basis for unification. Clinical Psychology Review, 10, 539-566.
STAATS, A. W., GROSS, M. C., GUAY, P. F., & CARLSON, C. C. (1973). Personality and social systems and attitude-reinforcer-discriminative theory: Interest (attitude) formation, function, and measurement. Journal of Personality and Social Psychology, 26, 251-261.
STAATS, C. K., & STAATS, A. W. (1957). Meaning established by classical conditioning. Journal of Experimental Psychology, 54, 74-80.
STAATS, A. W., & STAATS, C. K. (1958). Attitudes established by classical conditioning. Journal of Abnormal and Social Psychology, 57, 37-40.
STAATS, A. W., & STAATS, C. K. (1959). Effect of number of trials on the language conditioning of meaning. Journal of General Psychology, 61, 211-223.
STAATS, A. W., & STAATS, C. K. (1963).
STAATS, A. W., STAATS, C. K., & CRAWFORD, H. L. (1962). First-order conditioning of a GSR and the parallel conditioning of meaning. Journal of General Psychology, 67, 159-167.
STAATS, A. W., STAATS, C. K., & HEARD, W. G. (1959). Language conditioning of meaning to meaning using a semantic generalization paradigm. Journal of Experimental Psychology, 57, 187-192.
STAATS, A. W., & WARREN, R. D. (1974). Motivation and three-function learning: Deprivation-satiation and approach-avoidance to food words. Journal of Experimental Psychology, 103, 1191-1199.
TRAPOLD, M. A., & WlNOKUR, S. W. (1967). Transfer from classical conditioning and extinction to acquisition, extinction, and stimulus generalization of a positively reinforced instrumental response. Journal of Experimental Psychology, 73, 517-525.
VAUGHAN, M. E. (1989). Rule-governed behavior in behavior analysis: A theoretical and experimental history. In S.C. Hayes (Ed.), Rule-governed behavior (pp. 97-118). New York: Plenum.
WATSON, J. B. (1930). Behaviorism. New York: Norton.
WlTTGENSTEIN, L. (1967). Philosophical investigations. Oxford, England: Oxford University Press.
WOLPE, J. (1958). Psychotherapy by reciprocal inhibition. Stanford, CA: Stanford University Press.
WULFERT, E. (1994). Acceptance in the treatment of alcoholism: A comparison of alcoholics anonymous and social learning theory. In S.C. Hayes, N. S. Jacobson, V. M. Follette, & M. J. Dougher (Eds.), Acceptance and change: Content and context in psychotherapy (pp. 203-217). Reno, NV: Context Press.
ZAJONC, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151-175.
ZAJONC, R. B. (1984). On the primacy of affect. American Psychologist, 39, 117-123.
ZETTLE, R. D., & HAYES, S. C. (1982). Rule-governed behavior: A potential theoretical framework for cognitive-behavior therapy. In P. C. Kendall (Ed.), Advances in cognitive-behavioral research and therapy (Vol. 1, pp. 73-118). New York: Academic Press.
ZURIFF, G. E. (1985). Behaviorism: A conceptual reconstruction. New York: Columbia University Press.