Está en la página 1de 14

Rosenfarb, Irwin S.(1992). A behavior analytic interpretation of the therapeutic relationship. Psychological Record, Vol 42(3), pp. 341-354.

A BEHAVIOR ANALYTIC INTERPRETATION OF THE THERAPEUTIC RELATIONSHIP


Contents
1. 2. 3. 4. 5. Interpersonal Experiences and Psychopathology The Therapeutic Relationship as a Shaping Process The Behavior to be Shaped Generalization to the Natural Environment: 1. The Use of Natural Reinforcers Generalization to the Natural Environment: II. Change Through a ContingencyShaping Process 6. Research on the Therapeutic Relationship 7. Summary and Conclusions 8. References Section: THE THERAPEUTIC RELEATIONSHIP The paper examines, from a behavior analytic perspective, the mechanisms through which change occurs within the context of the therapeutic relationship in individual psychotherapy. The analysis focuses upon the therapist's shaping of the client's behavior through subtle nonverbal cues and explicit verbal analyses. In the relationship, therapists modify behavior that has created difficulties for the client in the natural environment, and clinical change is dependent upon the extent to which those in the natural environment will reinforce functionally similar response classes. Ferster's distinction between natural and arbitrary reinforcement and Skinner's distinction between contingency-shaped and rule-governed behavior are also used to describe the manner through which change occurs. The goal of the paper is to stimulate behavior therapists to look more closely at the therapeutic relationship as a mechanism of clinical change and to help those of other perspectives see how their effectiveness within the relationship can be understood using the principles of learning. In the early days of behavior therapy, relationship variables were often seen as ancillary to more clearly developed mechanisms of change (cf. Eysenck, 1960; Wolpe, 1954). Relationship variables were difficult to specify objectively. Furthermore, the therapeutic relationship was often associated with psychoanalysis and other "less scientific" types of therapy. In these early days, the principles of learning were rarely utilized to understand the processes through which the therapeutic relationship itself could lead to therapeutic changes. As behavior therapy has developed, therapists have begun to view clinical change as mediated, at least in part, by the relationship between the client and therapist (Wilson & Evans, 1977). Yet, in spite of this realization, little has been done to integrate the principles developed in the experimental laboratory with an understanding of the processes of change within the therapeutic relationship. Instead, when behavior therapists

have investigated the therapeutic relationship, they have often compared specific relationship techniques, such as empathy, warmth, or unconditional positive regard, to specific therapeutic techniques, such as systematic desensitization or modeling, to see which factor was more important in accounting for clinical change (see Morris & Magrath, 1983, and Sweet, 1984, for reviews). Behavior therapists, however, have done little to investigate how the relationship itself influences therapeutic change (Emmelkamp, 1986). Although much therapeutic change occurs through the use of specific behavioral techniques, the therapeutic relationship is often an important source of clinical outcome. Moreover, the relationship provides the framework through which specific therapeutic techniques are implemented. It also seems important to understand the processes of change within the therapeutic relationship so that the relationship itself does not become reified and used as an explanation for behavior change. A behavioral analysis, it seems, needs to specify the critical interpersonal ingredients of change within the therapeutic relationship and relate those changes to changes made in other interpersonal relationships and to changes made through other specific behavioral techniques. The purposes of this paper are thus two-fold: to behaviorally interpret changes that occur within the context of the therapeutic relationship, and to integrate this analysis with a broader, functional analytic approach to learning. Interpersonal Experiences and Psychopathology Clients come to therapy because their behavior has been ineffective in the natural environment. Frequently, interpersonal relationships are problematic. Client behavior may either lack positive social consequences or may lead to negative social consequences. Most psychotherapies thus focus upon these difficulties in social relationships (Goldiamond & Dyrud, 1968; Kohlenberg & Tsai, 1987). Many kinds of early experiences with significant others may have played a role in the development of these interpersonal difficulties. Clients may not have been positively reinforced for significant social behavior early in life (Bowlby, 1988). Those who have not been taught the important nuances of social behavior early in development are at a clear disadvantage in subsequent interpersonal interactions. Ferster (1973) has discussed the importance of these early interpersonal experiences in the development of prosocial behavior: The child who fails to come under the control of the mother's behavior is progressively left behind in his development of interpersonal behaviors, and whole sectors of interpersonal reactivity are not available to him as a means of commerce with the external world, much along the lines of the classical connotations of the fixation of a personality at a particular stage of development. (p. 863) Clients may have also had a history of excessive punishment early in life. Punishment not only hinders the development of new skills but leads to the tendency to withdraw from potentially punishing situations (Azrin & Holz, 1966). Those who have been sexually abused as children, for example, not only suffer the sequelae of such severe negative interpersonal interactions, but such a history may also prevent individuals from seeking out opportunities where positive interpersonal experiences may become operative. Such

punishment may lead to the failure to learn those important nuances of adult interpersonal relationships developed only by interacting with members of the opposite sex. Because of a client's early learning history and subsequent interpersonal failures, a therapist within the therapeutic relationship may need to develop prosocial behaviors that have never been developed in the past. In the relationship, the lack of positive social reinforcement and the history of excessive punishment may need to be reversed. Therapists may need to encourage positive social behavior that has never been encouraged, and they may need to avoid punishing positive social behavior that others have previously punished (cf. Skinner, 1953, p. 371). Therapists, it seems, need to react differently toward the client within the therapeutic relationship than the way others have reacted to the client in the past (Alexander & French, 1946; Appelbaum, 1978; Beier & Young, 1984; Ferster, 1979a; Gibbons, 1985; Wolf, 1966). As therapists begin to react differently toward the client than the way others have reacted in the past, therapists may begin to shape new, more positive behavioral repertoires for clients. Moreover, as these adaptive repertoires are developed, clients may begin to emit the same functional behaviors in the natural environment in the presence of functionally similar stimuli. If others in the natural environment also reinforce the same response classes that therapists reinforce, then changes that occur within the context of the therapeutic relationship will generalize to the natural environment. The Therapeutic Relationship as a Shaping Process Therapists begin to shape behavior within the therapeutic relationship by modifying their own interpersonal behavior in reaction to the client's behavior. Therapist interpersonal feedback cues are used to shape new client responses. often, subtle therapeutic cues serve to reinforce selected aspects of client behavior. A therapist's turn of the head, a change in eye contact, or a change in voice tone may all reinforce selected client behavior (Ferster, 1979b). one therapist, for example, may lean forward in her chair whenever a client begins to discuss interpersonal difficulties with his mother. Another therapist may begin to nod his head as clients begin to discuss such material. A third may maintain more eye contact. In all three cases, each therapist's behavior may be serving as both a reinforcing stimulus for previous client behavior and as a discriminative stimulus for the further discussion of such relevant material. Therapeutic change within the relationship, however, occurs primarily because of the verbal interchange between the therapist and client (Ferster, 1979b; Hamilton, 1988). Therapists use their own verbal behavior to modify client verbal behavior within the session. Numerous types of verbal analyses are used to modify behavior and research has as of yet failed to clearly delineate the specific types of verbal analyses that are most efficacious (Greenberg, 1986; Hamilton, 1988; Mahrer, 1988). An especially potent form of behavior change within the relationship occurs when therapists attempt to interpret client behavior within the session. Some therapists may point out discrepancies between client verbal and nonverbal behavior. other therapists may attempt to relate client behavior within the session to client behavior with significant others early in life. Still others attempt to compare behavior in therapy to client behavior outside of therapy. Each therapist, it seems, has his or her own unique way of effecting change, both verbally and nonverbally, and because almost all complex behavior is

multiply determined, any number of therapist reactions may prove beneficial (Ferster, 1979b). Client reactions within the relationship also often determine the specific therapeutic strategies used to shape behavior. Some clients, for example, respond best to interventions focusing on client behavior toward the therapist. other clients react more positively to verbal encouragement of changes made outside of therapy. Still other clients respond favorably to therapists' attempts to connect client behavior within the session to similar behavior emitted with significant others. Just as the therapist shapes the client's behavior within the relationship, so the client often subtly shapes the specific therapeutic strategy used. The Behavior to be Shaped At the beginning of therapy, therapists often use the relationship to determine the client's specific interpersonal difficulties. Beier and Young (1984,p. 129) provide several examples of how the first therapeutic hour often provides clues to important clinical behavior. one client, for example, may excessively flatter the therapist by telling him how much her friend has complemented his work. A second client may attempt to impress the therapist with his knowledge of psychological terminology. A third may answer questions in short, direct sentences, leaving the therapist with the impression that he will do little to volunteer personal information. Finally, a fourth client may appear on the verge of leaving the therapist's office throughout the entire first session, giving the impression that she often leaves or avoids emotionally difficult encounters. In all cases, the therapist may begin to hypothesize about the manner in which clients relate to significant persons in their lives. Shapiro (1989) has pointed out that the way clients describe themselves and others often provides clues to important clinical material. one client, Shapiro noted, discussed a confrontation with her boyfriend as if she were giving "court testimony." She described every detail of the interaction as if she were making a case and reviewing the evidence. Another client was "very prideful" in therapy. He had difficulty admitting weaknesses and often talked about the weather or the latest football scores. In the first case, the therapeutic material became the client's need to justify her own behavior and in the second case the relationship focused upon the client's forced self-confidence and his artificiality. Often, nonverbal cues are indicative of important clinical material. Beier and Young (1984,pp. 253-254) provide the following interchange between client and therapist: CLIENT: I guess I'm just the type of guy that was meant to be a loner, damn it (deep sigh). THERAPIST: ...when you sigh and look so sad I get the feeling that I'd better leave you alone in your misery, that I should walk on eggshells and not get too chummy or I might hurt you even more... CLIENT: I feel like a loner, I feel that even you don't care about me... THERAPIST: I wonder if other folks need to pass this test, too?

In this example, not only is the therapist pointing out the client's nonverbal messages, but he or she is also attempting to modify the way the client relates to the therapist. Moreover, the therapist is trying to make the connection between the client's way of relating to the therapist and his manner of relating to others in the natural environment. Many therapists focus on the discrepancy between client verbal and nonverbal behavior within the session. Shapiro (1989) provided an example of a housewife who came to therapy complaining of feeling sexually inadequate and incompetent as a wife. From further discussions, it became clear that the client was merely mimicking her husband's descriptions of her behavior. Shapiro asserted that he noticed a lack of conviction in the client's voice. Her voice sounded "stilted or rote, perhaps like the voice of a child reciting a lesson" (Shapiro, 1989, p. 45). The therapeutic relationship thus focused not on the client's feelings of inadequacy and incompetence, but on attempts to reinforce behaviors indicative of the client's own wishes and desires. The relationship provided an opportunity for the client to develop assertive behavior, independent from her husband's evaluation. At times, therapists use their own behavior as a clue to important client behavior within the relationship. Yalom (1989,p. 95), for example, used his own boredom as a stimulus to begin to confront a client to make changes within the relationship. The client dealt with personally intimate information by either giggling or by entertaining the therapist. Yalom's intervention consisted, in part, of simply making the client aware of when she was acting superficial and when she was revealing personally relevant information. The client's behavior quickly came under the control of these contingencies and the client began discussing serious and painful personal issues. In the following example, Shapiro (1989) showed how almost any interaction within the relationship may be used as an attempt for behavior modification by the therapist. A tense, "driven" man came to therapy in an agitated state and started berating himself for getting upset: What's the matter with me! I was just looking for a parking space and I spotted one and some guy cut in front of me. You know me! I had to pull along side of him and give him a blast and let him know what I thought of him! I couldn't let it go! I can't let anything go! The wear and tear I cause myself! I can't take anything in stride! (p. 87). Instead of discussing the client's tendency to not take things `'in stride,,' Shapiro points out that the important behavior expressed within the relationship was the client's selfpunishment. Thus, as with almost any discussion, although the client referred to events taking place elsewhere, there were components of his behavioral repertoire that could be modified directly within the therapeutic relationship. Therapists' absences due to vacations, illnesses, or a pregnancy may provide an avenue to important clinical material (Gibbons, 1985; Kohlenberg & Tsai, 1987). Ultimately, termination caused by a therapists's leaving may hold important clues to the way the client has dealt with the termination of other important relationships in life. Maladaptive behavior, similar to that expressed when significant others have left the client in the past, may begin to appear. Within the therapeutic relationship, however, the therapist may attempt to modify and change those maladaptive responses and thus may help the client learn to cope with significant loss.

Generalization to the Natural Environment: 1. The Use of Natural Reinforcers The analysis has focused upon the ways in which therapists modify client behavior within the therapeutic relationship. Ferster (1967) noted that this therapeutic shaping is an example of a natural reinforcement process. Natural reinforcers occur with little planning or awareness. They are intimately tied, in a moment-by-moment way, to variations in a person's behavior. When natural reinforcers are in effect, an "exquisitely fine interplay" (Ferster, 1979a, p. 289) between the behavior and its consequence is apparent. Arbitrary reinforcers lack this fine-grained natural relation to behavior. When arbitrary reinforcers are used in therapy, the behaviors reinforced are often different operants than those reinforced in the client's natural environment. The distinction between natural and arbitrary reinforcement, however, is often subtle. Ferster (1972) provided the following example to illustrate the "delicacy" of the functional distinction between natural and arbitrary reinforcers. A therapist told a client that she became depressed when angry. This intervention was arbitrary, Ferster notes, because the client's ability to observe her depression and its antecedent condition was lost when the client's relationship with her therapist ended. Social control is arbitrary when the listener adjusts his or her reaction specifically to the behavior required from the speaker (Ferster, 1972, emphasis added). Therapists in the relationship use natural reinforcers when their behavior is under the control of the client's behavior. Therapists, however, may react arbitrarily when their behavior is under the control of contingencies outside of the relationship. A therapist, for example, who says, "Good boy!" to a child, not because the child has deserved the praise, but because the therapist's supervisor has said that praise is important, has reacted arbitrarily to the child's behavior. Arbitrary reinforcers often speak more for the therapist's needs than for those of the client (Ferster, 1967). The use of natural reinforcers within the therapeutic relationship helps to ensure that changes made within the therapeutic relationship will generalize to the natural environment. The most potent natural reinforcer within the relationship and the one most similar to the type of reinforcement offered in the natural environment is social reinforcement offered by the therapist (see Deitz, 1989, for a discussion of the "natural" nature of social reinforcement). As Skinner (1982) has stated, "By contriving relatively unambiguous social contingencies, the therapist builds a repertoire that will be naturally effective in the client's daily life" (p. 5) Arbitrary reinforcers, though, may often serve as useful transition procedures to ensure that behavior will be emitted that will be positively reinforced in the natural environment. At times, there is no natural therapeutic intervention that may create behavior that will be reinforced by others. Arbitrary interventions may then be required to evoke the desired behavior. Therapeutic contracts or homework assignments, for example, although arbitrary in nature, are often critical in the development of behavioral repertoires that will be positively reinforced by others in the natural community. Generalization to the Natural Environment: II. Change Through a Contingency-Shaping Process Generalization to the natural environment is also enhanced because of the gradual manner in which the therapist shapes new client behavior within the relationship. Skinner (1969)

added an important dimension to his analysis of human behavior when he distinguished between rule-governed and contingency-shaped behavior. Briefly, Skinner stated that contingency-shaped behavior is behavior emitted because of certain past consequences, whereas rule-governed behavior is behavior emitted because individuals are following a stated rule. Individuals, for example, may learn to play billiards "intuitively" by being shaped by the consequences of their play. They become more likely to repeat shots that went in the pocket and less likely to repeat shots that were missed. Alternately, individuals can learn to play billiards by calculating the angles, masses, distances, and fractions of each shot. Such individuals have not been shaped by the consequences of their play. Instead, they have learned rules about how to play the game. Although contingency-shaped and rule-governed behavior may look identical, they are actually very different behaviors functionally. "In the first case, the individual feels the rightness of the force and direction with which the ball is struck, while in the second case, he feels the rightness of his calculations, but not of the shot itself" (Skinner, 1969, p. 166). The human operant literature has empirically analyzed the distinction between rulegoverned and contingency-shaped behavior and has generally shown that whereas rulegoverned behavior may at times become "insensitive" to changing contingencies, contingency-shaped behavior tends to be more sensitive to subtle changes in the contingencies and may be more likely to adapt to changing contingencies (see Hayes, 1989, for a review). When therapists modify their own interpersonal cues to shape behavior within the therapeutic relationship, a contingency-shaping process is in effect. The client has made contact with the subtle, social consequences of his or her behavior. Thus, it may be that when the client is now faced with similar social contingencies in the natural environment, he or she may be likely to be sensitive to the subtle, social reactions of others and to adapt when those contingencies change. Generalization to the natural environment is enhanced when therapists use natural reinforcers within the relationship and when clients make contact, through a contingency-shaping process, with the subtle social consequences of their behavior. Just as with arbitrary interventions, though, at times it may be critical for therapists to use rules or instructions to affect clinical change. When clients are faced with life-threatening situations, for example, therapists may need to intervene directly to instruct clients on the behavioral changes they need to make. At other times, therapists may wish to override the natural contingencies (Skinner, 1982). Many positive behavioral repertoires will never be developed unless instructions are given to ensure that the behavior is emitted. Moreover, therapeutic rules or instructions may often be necessary to help clients make contact with the long-term consequences of their behavior, instead of being controlled by the shortterm, more immediate contingencies (cf. Rachlin, 1974). Research on the Therapeutic Relationship This paper has focused upon attempts by therapists to modify behavior within the therapeutic relationship that will lead to positive changes outside of therapy. Changes made within the therapeutic relationship, however, are difficult to predict. Clinical outcome is often dependent upon moment-by-moment changes in the relationship (cf. Rice & Greenberg, 1984) and the contingencies controlling this contingency-shaping process may be subtle. Research therefore may never be able to isolate all the relevant factors affecting this shaping process (cf. Skinner, 1969, p. 171). Yet, research may help us look at important therapeutic interactions that are associated with therapeutic change.

Early research in verbal operant conditioning, for example, showed that verbal behavior could be brought under the control of the contingencies of reinforcement. Such research also suggested that verbal behavior in psychotherapy was modified by the therapist in the same way as was other operant behavior. The literature showed that the verbal units to be reinforced could include such variables as the type of memory reported (Quay, 1959), the type of affect displayed (Salzinger & Pisoni, 1958), and whether or not statements referred to the self (Adams & Hoffman, 1960). Moreover, such research showed that the type of therapist factors that could serve as reinforcers could range from nonverbal cues (Greenspoon, 1954) to psychoanalytic interpretations (Timmons, Noblin, Adams, & Butler, 1965). Other research along these lines attempted to analyze traditional psychotherapy transcripts and look for the contingencies controlling client verbal behavior. Both Murray (1956) and Truax (1966), for example, analyzed therapy sessions of Carl Rogers to determine the contingencies controlling the client's behavior. Both studies found that Rogers tended to react positively to certain classes of verbal behavior (those suggesting client independence, for example) and negatively to other classes (such as dependency on others). over time, those behaviors that were approved of increased in frequency and those disapproved of decreased. Although it was clear from both these studies and from the verbal operant conditioning literature that verbal changes in psychotherapy could be understood using the principles of learning, it also seemed that these studies failed to adequately replicate the subtle contingencies operating within the therapeutic interaction (Krasner, 1971). Recent statistical developments have helped to uncover the subtle therapist and client interactions associated with clinical change. Sequential analytic techniques (cf. Gottman, 1982), for example, have at least two distinct advantages over traditional linear models in attempting to understand the therapeutic relationship. First, the analysis assesses momentby-moment changes within the relationship. Understanding the relationship in such a molecular fashion seems critical in the attempt to uncover positive therapeutic interchanges (Mahrer, 1988). Second, the sequential analysis research methodology tends to be inductive in nature, without preconceived therapeutic biases (Hill, 1990). The analysis simply assesses how changes in certain classes of therapist behaviors are associated with changes in certain classes of client responses, and how changes in client behaviors lead to changes in therapist responses. The analysis attempts to map out sequences of therapeutic interactions and to associate those interactions with clinical outcome. Such an inductive research strategy seems an important and necessary way to explore a new field of inquiry (cf. Skinner, 1950). This research thus may help us begin to specify the shaping that occurs between therapists and clients. Research from other theoretical perspectives may also be helpful in beginning to develop an inductive approach to understanding change within the therapeutic relationship. Various researchers working within a client-centered perspective, for example, have explored the concept of "experiencing" (see Klein, Matieu-Coughlan, & Keisler, 1986, for a review). Briefly, this research has indicated that when clients discuss thoughts, feelings, and behavior in the present tense, rather than referring to thoughts, feelings, or behavior occurring elsewhere, such therapeutic interactions are positively correlated with behavioral changes made outside of therapy.

Other researchers, working within a psychodynamic perspective, have attempted to formulate the client's main relationship patterns with the therapist and with significant others (cf. Luborsky & Crits-Christoph, 1989). Known as the Core Conflictual Relationship Theme (CCRT) Luborsky and colleagues have attempted to quantify the client's needs or desires in significant relationships, the way others respond to the client in such relationships, and the way the client typically responds in such situations. Using the CCRT, Luborsky, Crits-Christoph, and Mellon (1986) found that relationship patterns reported with significant others bore a striking resemblance to the relationship displayed within the therapeutic relationship. As yet, though, data relating changes in relationship patterns to behavioral measures of client progress are limited. Luborsky and colleagues have also attempted to examine the factors surrounding shifts in mood states within the therapeutic relationship. In one study (Luborsky, Singer, Harte, Crits-Christoph, & Cohen, 1984), therapy tapes of one client, Mr. Q, were examined to determine the context in which the client either become more or less depressed. Results indicated that the client became more depressed (as measured by voice quality and statements made) when discussing feelings of helplessness within the relationship, and he became less depressed when the therapist made supportive, reassuring comments. Such a research strategy, known as the symptom-context method, highlights an inductive approach to the study of the therapeutic relationship and may be adapted by behavior analysts to explore factors surrounding changes in client behavior within the relationship. From a behavior analytic perspective, Hayes and his colleagues have focused on developing a contingency-shaping approach to social skills training. Although most social skills training programs are based upon attempts to identify and train specific component skills (cf. Bellack & Hersen, 1979), this work has attempted to use a gradual, contingency-shaping process to teach social skills. In the first study, Azrin and Hayes (1984) developed a contingency-shaped treatment to teach college-aged males to discriminate nonverbal indicants of interest among females. Students were shown videotapes of a female interacting with an unseen male and were asked to indicate on a Likert-type scale how interested they thought the female was in her partner. Those who received feedback on the accuracy of their ratings (compared to the females' actual ratings) were better able to discriminate cues of interest in females and improved more on role-play measures of social skill than did students who did not receive such feedback. Rosenfarb, Hayes, and Linehan (1989) integrated this type of contingency-shaped feedback into a treatment package for adults with significant social skills deficits. Clients participated in an eight-session individual role-playing training program. After each roleplay, some clients received "experiential feedback" from their therapist: Their therapist rated their social skills on a Likert-type scale. Results indicated that clients receiving such contingency-shaped feedback from their therapist improved significantly more on both role-play and self-report measures of social skill than did clients who were given instructions on specific behaviors to modify during role-playing. The results of both studies thus indicate that it is not necessary to teach social skills by isolating the specific components to modify. Social skills may be taught through a gradual, contingencyshaping process. The work by Hayes and his colleagues has thus attempted to experimentally examine and validate critical components of the therapeutic relationship by developing specific treatment packages containing the hypothesized ingredients of change. For psychotherapy

research to flourish, it seems important to not only specify through correlational analyses those therapeutic interactions associated with clinical change, but it also seems necessary to experimentally test whether those hypothesized factors are indeed critical to outcome. Summary and Conclusions The present paper attempts, within a behavior analytic framework, to understand changes that occur within the context of the therapeutic relationship in psychotherapy. The interpretation focuses upon the gradual shaping of the client's behavior. The therapist observes both verbal and nonverbal patterns in the client's behavior and attempts to modify those behavioral repertoires that have created interpersonal difficulties for the client. The client's behavior changes because of changes in the therapist's interpersonal feedback cues. If functionally similar behavioral repertoires are also reinforced outside of therapy, then changes made within the relationship will generalize to the natural environment. The goals of the paper are two-fold: to help stimulate behavior therapists to look more closely at factors within the therapeutic relationship as important processes of clinical change, and to help therapists of other perspectives see how their effectiveness within the therapeutic relationship can be understood using the principles of learning. Behavior therapy research has been hampered, in general, by the lack of critical explorations on the effectiveness of the therapeutic relationship. Behavior therapy research has concentrated on the development of therapeutic techniques to the exclusion of the exploration of processes of change within the relationship. Behavior analysis, however, seems well suited to understand such relationship processes. It is hoped this paper will help break down the barriers that have led behavior therapists to investigate the procedures of psychotherapy while tending to ignore the processes of change within the therapeutic relationship. References ADAMS, J. S., & HOFFMAN, B. (1960). The frequency of self-reference statements as a function of generalized reinforcement. Journal of Abnormal and Social Psychology, 60, 384-389. ALEXANDER, F., & FRENCH, T. M. (1946). Psychoanalytic therapy: Principles and application. New York: Ronald. APPELBAUM, S. A. (1978). Pathways to change in psychoanalytic therapy. Bulletin of the Menninger Clinic, 42, 239-251. AZRIN, N. H., & HOLZ, W. C. (1966). Punishment. In W. K. Honig (Ed.), operant behavior (pp. 380-447). New York: Appleton-Century-Crofts. AZRIN, R. D., & HAYES, S. C. (1984). The discrimination of interest within a heterosexual interaction: Training, generalization, and the effects on social skill. Behavior Therapy, 15,173-184. BEIER, E. G., & YOUNG, D. M. (1984). The silent language of psychotherapy (2nd ed.). New York: Aldine.

BELLACK, A. S., & HERSEN, M. (Eds.). (1979). Research and practice in social skills training. New York: Plenum. BOWLBY, J. (1988). Developmental psychiatry comes of age. American Journal of Psychiatry, 145, 1-10. DEITZ, S. M. (1989). What is unnatural about "extrinsic reinforcement"? The Behavior Analyst, 12, 255. EMMELKAMP, P. M. G. (1986). Behavior therapy with adults. In S. Garfield & A. Bergin (Eds.), Handbook of psychotherapy and behavior change (3rd ea., pp. 385-442). New York: Wiley. EYSENCK, H. J. (1960). Behavior therapy and the neuroses. oxford: Pergamon. FERSTER, C. B. (1967). Arbitrary and natural reinforcement. The Psychological Record, 17, 341 -347. FERSTER, C. B. (1972). Clinical reinforcement. Seminars in Psychiatry, 4, 101-111. FERSTER, C. B. (1973). A functional analysis of depression. American Psychologist, 28, 857-870. FERSTER, C. B. (1979a). Psychotherapy from the standpoint of a behaviorist. In J. D. Keehn (Ed.), Psychopathology in animals (pp. 279-303). New York: Academic. FERSTER, C. B. (1979b). A laboratory model of psychotherapy: The boundary between clinical practice and experimental psychology. In P. Sjoden, S. Bates, & W. Dockens (Eds.), Trends in behavior therapy (pp. 23-38). New York: Academic. GIBBONS, J. D. (1985). A radical behavioral and psychoanalytic analysis of a case presentation. Paper presented at the annual meeting of the Association for Behavior Analysis, Columbus, OH. GOLDIAMOND, I., & DYRUD, J. E. (1968). Some applications and implications of behavioral analysis for psychotherapy. In J. Schlien (Ed.), Research in psychotherapy, Vol. III (pp. 54-89). Washington, DC: APA. GOTTMAN, J. M. (1982). Time-series analysis. New York: Cambridge. GREENBERG, L. S. (1986). Change process research. Journal of Consulting and Clinical Psychology, 54, 4-9. GREENSPOON, J. (1954). The effects of two nonverbal stimuli on the frequency of members of two verbal response classes. American Psychologist, 9, 384. HAMILTON, S. A. (1988). Behavioral formulations of verbal behavior in psychotherapy. Clinical Psychology Review, 8, 181-193. HAYES, S. C. (Ed.). (1989). Rule-governed behavior. New York: Plenum.

HILL, C. E. (1990). Exploratory in-session process research in individual psychotherapy: A review. Journal of Consulting and Clinical Psychology, 58, 288-294. KLEIN, M. H., MATIEU-COUGHLAN, P., & KEISLER, D. J. (1986). The experiencing scales. In L. S. Greenberg & W. M. Pinsoff (Eds.), The psychotherapeutic process (pp. 21-72). New York: Guilford. KOHLENBERG, R. J., & TSAI, M. (1987). Functional analytic psychotherapy. In N. Jacobson (Ed.), Cognitive and behavior therapists in clinical practice (pp. 388-443). New York: Guilford. KRASNER, L. (1971). The operant approach in behavior therapy. In A. E. Bergin & S. L. Garfield (Eds.), Handbook of psychotherapy and behavior change (pp. 612-652). New York: Wiley. LUBORSKY, L., & CRITS-CHRISTOPH, P. (1989). Understanding transference: The CCRT method. New York: Basic. LUBORSKY, L., CRITS-CHRISTOPH, P., & MELLON, J. (1986). Advent of objective measures of the transference concept. Journal of Consulting and Clinical Psychology, 54, 39-47. LUBORSKY, L., SINGER, B., HARTE, J., CRITS-CHRISTOPH, P., & COHEN, M. (1984). Shifts in depressive state during psychotherapy: Which concepts of depression fit the context of Mr. Q's shifts? In L. N. Rice & L. S. Greenberg (Eds.), Patterns of change (pp. 157-193). New York: Guilford. MAHRER, A. R. (1988). Discovery-oriented psychotherapy research: Rationale, aims, and methods. American Psychologist, 43, 694-702. MORRIS, R. J., & MAGRATH, K. H. (1983). The therapeutic relationship in behavior therapy. In M. J. Lambert (Ed.), Psychotherapy and patient relationships (pp. 154-189). Homewood, IL: Dow Jones-lrwin. MURRAY, E. J. (1956). A content-analysis method for studying psychotherapy. Psychological Monographs, 70(13), whole no. 420. QUAY, H. (1959). The effect of verbal reinforcement on the recall of early memories. Journal of Abnormal and Social Psychology, 59, 254-257. RACHLIN, H. (1974). Self-control. Behaviorism, 2, 94-107. RICE, L. N., & GREENBERG, L. S. (Eds.). (1984). Patterns of change. New York: Guilford. ROSENFARB, I. S., HAYES, S. C., & LINEMAN, M. M. (1989). Instructions and experiential feedback in the treatment of social skills deficits in adults. Psychotherapy, 26, 242-251.

SALZINGER, K., & PISONI, S. (1958). Reinforcement of affect responses of schizophrenics during the clinical interview. Journal of Abnormal and Social Psychology, 57, 84-90. SHAPIRO, D. (1989). Psychotherapy of neurotic character. New York: Basic. SKINNER, B. F. (1950). Are theories of learning necessary? Psychological Review, 57,193-216. SKINNER, B. F. (1953). Science and human behavior. New York: Free Press. SKINNER, B. F. (1969). Contingencies of reinforcement. New York: Appleton-CenturyCrofts. SKINNER, B. F. (1982). Contrived reinforcement. The Behavior Analyst, 5, 3-8. SWEET, A. A. (1984). The therapeutic relationship in behavior therapy. Clinical Psychology Review, 4, 253-272. TlMMONS, E. O, NOBLIN, c. D., ADAMS, H. E., & BUTLER, J. R. (1965). operant conditioning with schizophrenics comparing verbal reinforcers vs. psychoanalytic interpretations: Differential extinction effects. Journal of Personality and Social Psychology, 1, 373-377. TRUAX, C. B. (1966). Reinforcement and nonreinforcement in Rogerian psychotherapy. Journal of Abnormal Psychology, 71, 1-9. WILSON, G. T., & EVANS, I. (1977). The therapist-client relationship in behavior therapy. In A. S. Gurman & A. M. Razin (Eds.), Effective psychotherapy (pp. 554-565). New York: Pergamon. WOLF, E. (1966). Learning theory and psychoanalysis. British Journal of Medical Psychology, 39, 1-10. WOLPE, J. (1954). Reciprocal inhibition as the main basis of psychotherapeutic effects. Archives of Neurology and Psychiatry, 72, 205-226. YALOM, I. D. (1989). Love's executioner and other tales of psychotherapy. New York: Basic. The author thanks Steve Hayes, Joe Haas, Phil Lewis, Barry Burkhart, Michael Dougher, F. Dudley McGlynn, and Suzanne Brannon for their thoughtful comments on earlier drafts of the manuscript. Preparation of this manuscript was supported in part by National Institute of Mental Health Grant MH14584. Address reprint requests to Irwin S. Rosenfarb, Department of Psychology, UCLA Family Project, 1285 Franz Hall, Los Angeles, CA 90024. ~~~~~~~~ IRWIN S. ROSENFARB Auburn University and University of California, Los Angeles