Вы находитесь на странице: 1из 25



Some Clinical Implications of Contextualistic Behaviorism:

The Example of Cognition





University of Nevada

Radical behaviorism can be interpreted either mechanistically or contextualistically. The present article describes the essenceof the contextualistic interpretation and con- trasts it with other positions common in behavior therapy. Contemporary behavior analysis, contextually viewed, is applied to the problem of cognition. Derived stimulus relations are argued to be at the core of cognition, and that the relation betweencogni- tion and other kinds of behavior is contextually situated. New therapeutic strategies emerge from a reexamination of the philosophy and theory of contextualistic behavior analysis.

While the growth of behavior therapy has been one of the success stories in applied psychologyover the last 25 years, its intellectual development seems less dramatic. The present article examines the contribution contextualistic behaviorism might make to the further progress of behavior therapy.

A Brief History of Intellectual Trends in Behavior Therapy

Early behavior therapy was committed both to the application of clearly specified and replicable techniques, made available by well-designed and sys- tematic experimental research, and to learning theory (Eysenck, 1972). For example, Franks and Wilson (1974) argued that the common element in be- havior therapies was an adherence to "operationally defined learning theory and conformity to well established experimental paradigms" (p. 7). Media- tional learning theory was embraced: "One can study inferred events or processes and remain a behaviorist as long as these events or processes have measurable and operational referents" (Franks & Wilson, 1974, p. 7). 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. Very few clinicians were trained in the position. On the con- trary, "Methodological behaviorism is much more characteristic of contem-

Requests for reprints should be sent to: Steven C. Hayes, Department of Psychology, Univer- sity of Nevada, Reno, NV 89557-0062.



Copyright 1992by Associationfor Advancementof BehaviorTherapy

All rights of reproduction in any form reserved.



porary behavior modifiers than is radical behaviorism" (Mahoney, Kazdin, & Lesswing, 1974, p. 15). Thus, in early behavior therapy, behavioral tech- nology was most closely linked to experimental methodology, mediational learning theory, and methodological behavioral philosophy. As behavior therapy developed, cognitive techniques and concepts were in- creasingly added to its core. Early cognitive mediational accounts of behavior change (e.g., Bandura, 1969) soon blossomed into the cognitive therapy move- ment (e.g., Mahoney, 1974; Meichenbaum, 1977). Despite protests from some founders that these ideas had been present all along (e.g., Wolpe, 1980), the very vitality of the movement suggested that many therapists felt otherwise. The cognitive movement sought not just the addition of a range of verbal psy- chotherapies to the technical armamentarium of behavior therapy, but also the addition of modern cognitive mediational constructs to mediational learning theory concepts already in place. With the cognitive wave fully assimilated both technologically and theoret- ically, behavior therapy in the 1980s settled down to an expansion into new areas (e.g., behavioral medicine, behavioral assessment), new populations or problems (e.g., AIDS, spousal violence), and new leadership roles (e.g., helping to develop DSM-III and IV, editing important mainstream journals, par- ticipating in the direction of federal funding). The recent expansion of be- havior therapy, in other words, has been largely in the areas of technology and political influence.

Radical Behaviorism: An Alternative Tradition

More recently, some behavior therapists have been considering the value of radical behaviorism as a set of coordinating assumptions for the develop- ment of behavior therapy (e.g., Jacobson, 1991a, 1991b; Kohlenberg & Tsai, 1987). This interest in radical behaviorism is not so much an interest in radical behaviorism as it had always been but in a radical behaviorism that has evolved out of this original position. In recent years, two major subtypes within rad- ical behavioral thinking have been identified, and the distinction between these types and traditional behavioral philosophy can now be made more clearly. In addition, a set of new findings has emerged in basic behavior analysis and the significance of these findings for the analysis of verbal humans' behavior is becoming clearer. Some discussion of the history of these developments will be helpful in understanding their relevance to contemporary behavior therapy. We will consider the philosophical developments first, then the findings and their implications.

Systemic Foundations of Behaviorism

Behaviorism was originally a movement against consciousness as the sub- ject matter of psychology and introspection as the method of its investigation (Watson, 1924, pp. 2-5). Watson (1924, p. 14) claimed behavior as the subject matter of psychology and defined it by its form: behavior was muscle move- ments and glandular secretions. From his perspective, all activities of the or- ganism could be reduced to these events (a kind of metaphysical behaviorism); and even if mental or other non-movement activities existed, they could not



constitute the subject matter of a scientific psychology because public agree- ment as to their occurrence was impossible (a kind of methodological be- haviorism). Thus for Watson, scientific legitimacy was an issue of public ob- servability. Skinner deviated considerably from these views, Skinner distinguished the subjective/objective dichotomy (which he thought to be of fundamental scientific importance) from the private/public dichotomy (which he thought was not fundamental). Skinner (1945) defined scientific observations as those under the control of a certain kind of contingency. Only when an observation was controlled by particular stimulus events (largely those of a nonverbal sort) and a general history of reinforcement for speaking under the control of those events, as opposed to control by audience factors, states of reinforceability, and so on, was the observation scientifically valid. As such, observations could be private and objective (scientifically legitimate) or public and subjective (scientificallyillegitimate), depending upon the contingencies controlling the observations. Public subjective observations are readily demonstrated. In discussing this point in classes we show the class the following words on the blackboard for about one-half second and then cover them:

Paris in the the spring The class is asked to write down what they read. Virtually all will write "Paris in the spring." If interrater reliability were being calculated, it would be very high, and if that were the metric for objectivitywe could be confident the words indeed were "Paris in the spring." But these observations were subjective in the sense that they were controlled by individual histories with particular words - "Paris in the spring" is familiar, "the, the" is unfamiliar, and so on- rather than by what was observed and a history of reinforcement for control by observed events. In Skinner's view, the observations of the class are scientifically illegitimate, even though highly "reliable." This is the sense in which "radical behaviorism" is radical or "to the root":

Even its core concepts and observations are defined in terms of contingencies, specifically, those bearing on the behavior of the scientist. Skinner rejected methodological behaviorism because he did not believe that public agreement provided assurance of proper contingency control. As in the example above, it is easy to find instances where whole groups of observers are similarly influenced by motivational states and other subjective conditions. In solving this problem by way of contingency analysis, Skinner opened up behaviorism to the very thing Watson was trying to eliminate" Introspec- tive observations of private events. For example, Skinner said that radical be- haviorism "does not insist upon truth by agreement and can therefore con- sider events taking place in the private world within the skin. It does not call these events unobservable" (Skinner, 1974, p. 16). Radical behaviorism is not interpretable as extreme behaviorism for this reason, despite claims to the con- trary (e.g., Mahoney, 1989). In a fundamental sense, radical behaviorism is not part of the tradition of "behaviorism" at all because all psychological ac-




tivities that are contacted in a scientifically valid manner are subject to anal- ysis. In this way at least, radical behaviorism rejects both methodological be- haviorism and Watsonian metaphysical behaviorism. Skinnerians did not move rapidly to investigations of thinking and feeling for another reason, however. Skinner felt that an understanding of private events was not necessary for a scientific understanding of overt activity (Skinner, 1953). We will analyze why he came to this conclusion later and will argue that there are reasons to believe that he was mistaken.

Inconsistencies in Skinner's position. Radical behaviorism was a position

virtually defined by the writings of B. E Skinner, and Skinner often failed to be precise about his core assumptions, or wrote about them in contradic- tory ways (Parrott, 1983, 1986; Hayes, Hayes, & Reese, 1988). The definition of behavior is a case in point. In 1938 Skinner defined behavior both mechan- ically and interactively- one definition following the other in successive para- graphs. The mechanical definition, in the tradition of Watson, treats behavior as an organismic phenomenon, taking place in settings made up of things and events conceptualized as existing independently of the behavior itself. In ac- cord with this view, Skinner defined behavior as "the movement of an organism or of its parts in a frame of reference provided by the organism itself or by various external objects or fields of force" (Skinner, 1938, p. 6). When be- havior is defined as movement in this way, changes in the conditions of its occurrence or "frame of reference" may change the occurrence of behavior or its probability but they do not change behavior per se. For example, raising a hand to get attention and raising a hand to stretch a muscle is the same be- havior when behavior is defined by movement. The two episodes differ, of course, because the "causal determinants" of the behavior differ, but the be- havior is a raising of the hand in either case. In 1938 Skinner also defined behavior as an interaction between an organism and its environment, the implication being that behavior and its environment constitute a unitary phenomenon. In Skinner's words, behavior is "the func- tioning of an organism which is engaged in acting upon or having commerce with the outside world" (1938, p. 6). From this perspective, behavior does change its nature as its circumstances of occurrence change because the behavior and those circumstances are two aspects of a single event. In other words, getting attention is a different behavior than stretching regardless of the similar move- ments involved. The 1938 book is the only place where Skinner attempted a definition of "behavior" and the two definitions given differ in a philosophically fundamental way. As a result, some Skinnerians have defined behavior much as Watson did, as movement in a frame of reference (e.g., Johnston & Pennypacker, 1981). Conversely, some radical behaviorists view behavior as an event of a whole organism interacting in and with a context (e.g., Hayes et al., 1988; Morris,


Similar problems occur in other areas of Skinner's work including his view of causality, the relation between psychology and biology, the nature of stimuli, and other topics. In each case, Skinner's analysis is either ambiguous or con- tradictory at a fundamental level. For example, Skinner's position sometimes has a linear, causal quality, as when he claims that: "We have discovered more



about how the living organism works and are better able to see its machine- like properties" (Skinner, 1953, p. 47). At other times, he rejects the causal logic of reductionism (Skinner, 1938, p. 418-432. See Hayes et al., 1988 for

a more detailed analysis of Skinner's views). Because Skinner embraced philosophically incompatible views, it is not pos- sible to articulate the radical behavioral position. There appear to be two dis- tinct philosophical systems operating under the banner of radical behaviorism. Understanding the nature of this division is important for our purposes since the clinical implications of the two versions of radical behaviorism are quite different.

The Metaphilosophy of Stephen Pepper

The division within contemporary radical behaviorism can be understood from the point of view of the philosophical categories constructed by Stephen C. Pepper. Pepper (1942) delineated four kinds of philosophical system or world views on the basis of what he called their "root metaphors" and their truth criteria, namely: Formism (e.g., Plato); Organicism (e.g., Hegel), Mech- anism (e.g., S-R learning theory) and Contextualism (e.g., James). These world views, he argued, were orthogonal to each other because their assumptions differed so greatly that meaningful discourse among them was impossible. The inconsistencies in Skinner's position may be understood as an unwitting at- tempt to operate on the basis of two sets of incompatible assumptions, those characteristic of mechanism on one hand, and of contextualism on the other. Pepper's idea was that humans philosophize on the basis of certain common- sense models, and that the understanding achieved in this manner is then metaphorically applied to the world. Different philosophies operate on the basis of different models or root metaphors. The root metaphor of mecha- nism is the machine. A machine (such as a lever) consists of discrete parts

(e.g., a fulcrum and lever), a relation among these parts (e.g., the lever must sit atop the fulcrum), and forces to make the parts operate (e.g., pressing down on one end of the lever produces a precisely predictable force at the other end). If we wished to understand a machine, we would need to disassemble

it and identify the parts, relations, and forces that constitute it and its opera-

tion. Note also that when the machine is disassembled, the parts remain un- changed despite their independence from the rest of the machine. In other words, a spark plug is a spark plug whether screwed into a cylinder or sitting on the kitchen table. When applied to the subject matter of behavioral psychology, a mechanical metaphor takes stimuli, responses, cognitions, and other parts of a psycho- logical event to be discrete parts, related to each other by "mechanisms," and animated by forces (e.g., information, drives). The existence of such parts in the world is assumed: our job as theorists is simply to find ways to "take the cover oft" (literally, to dis-cover them) so that they can be seen. The parts are further assumed to retain their nature when isolated from the whole. Accord- ingly, mechanists often make use of research preparations that isolate hypothe-

sized components so that they may be studied out of context (e.g., sensation is studied as a means to understand perception). The goal of mechanistic research is the development of a model of the ma-



chinery that is assumed to exist. If such a model is shown to correspond to


range of relevant observations (especially if it is predictively verified) then


is said to be true. Hypothetico-deductive theorizing is a classic example of

this strategy, and mechanistic psychologies gravitate toward it. The archetype of mechanistic psychology is S-R learning theory, and its descendent, information processing. In modern mechanical theorizing, the computer is usually taken to be the base model. Entities such as "short term memory stores" are believed to exist and to be discoverable. Elaborate models of mental machinery are developed and carefully subjected to empirical test. Contextualistic philosophizing is quite different. In contextualism, the root

metaphor is the historically situated action, alive and in the present, such as "going to the store" or "making dinner." Actions such as these are whole units involving an action in and with a context. In the world of common-sense, it

is not possible to separate "going to the store" into distinct units. For example,

the fact that a person is walking to the store does not mean that the action

is walking, while the home that was left behind or the store that is approached

are separate. "Going to the store" is all of these working together. Further, even this occurs in a context (e.g., "needing something from the store," or "having money to buy things," or "knowing where the store is" or "being an organism that eats"). Thus, the event constituting the focus of analysis from

a contextualistic standpoint is abstracted from an ever widening circle of pos- sible events. The most trivial act may lead to a concern with the whole uni- verse. What keeps analysis from being overwhelmed by the need to become

ever more inclusive is that analysis is taken to be an activity that itself has

a context and a purpose. Thus, analysis need be taken only to the point at

which its purpose is achieved. Insofar as a way of speaking achieves its pur- pose, it is "true." Theorizing is thus taken to be an act of construction, not discovery, because there are no grounds to say that a given analysis is the cor- rect analysis. As the act-in-context is applied to psychological theorizing, note first that there are no fundamental "parts." The parts identified in the act of analysis are held to exist as features of the analysis not as aspects of the world. As such, the contextual theorist may be legitimately interested in anything from grasping a rattle to adopting a life-style. Further, the parts that are identified are identified in concert with a context such that if context changes so does the event. Walking to the store is a different action than walking the plank, even if the muscles contract in the same sequence. In short, a contextualistic perspective mitigates against the use of topographical or formal criteria as defining characteristics of behavior. An example may be helpful in this regard. Suppose a behaviorist watches an animal in a cage and notes that the animal is scratching at the door and biting the latch. If asked "what is the operant here?" an appeal to these re- sponse forms alone will not supply the answer. What must be considered are the historical and current contexts that would organize these forms into the interactive unit called an operant. One analyst might conclude that the an- imal is engaging in "when-confined-get-out-of-the-cage-behavior," another "when-not-stimulated-produce-stimulation-behavior." These descriptions or-



ganize the same topographies into completely different events, not in the sense that they are influenced by different forces, but in the sense of qualitatively different interactions with different contextual participants. Both of the operants described above might also be organized into larger or smaller units. For example, rearing up to access parts of the environment (as when the animal does so to sniff at an opening) might itself be understood as an operant with a distinct history. Indeed, an infinity of operant construc- tions are simultaneously possible, and there are no grounds in contextualistic thinking on which to argue that one or the other of them is "the correct one" save the extent to which a given construction serves the purposes of the ana- lyst. Contextualistic truth, in other words, is that which accomplishes the analysts' goals. Pepper's categories are useful in the present context not because behaviorists self-consciously consider themselves to be mechanists or contextualists, but because they point to fundamental differences in the assumptions and premises of groups of behavior analysts and therapists. We and others have argued else- where that radical behaviorism is, in essence, contextualistic (Hayes et al., 1988; Hayes, 1987; Morris, 1988). It would be more accurate to say, however, that in the hands of some, radical behaviorism appears contextualistic, while in the hands of others, it appears mechanistic. It is not our intention to suggest that either mechanism or contextualism is correct, the other wrong. Evaluated on their own terms, both mechanism and contextualism are coherent, viable positions. Our point is, rather, that Skinner combined the two in an incoherent fashion; one consequence of which has been to render the term "radical behaviorism" useless as a category name. Radical behaviorism encompasses both mechanistic and contextualistic types of behaviorism. The two positions are incompatible, and the differences cannot be resolved by way of a compromise. In the remainder of the article we will, therefore, restrict our analysis to only one of these interpretations of radical behaviorism- the contextualistic position-and will develop the clinical im- plications of this position as it bears on cognition. First we will describe in more detail the core conceptions of contextualistic behaviorism.

Contextualistic Behaviorism

Though several types of contextualistic behaviorism are possible, we will restrict our analysis to a type defined by four characteristics: a) a focus on the psychological level of analysis, b) a commitment to contextually delimited constructs, c) a pragmatic view of truth, and d) an interest in the joint goals of description, prediction, control, and interpretation achieved with precision, scope and depth, and based upon verifiable experience.

PsychologicalLevel of Analysis

Contextualistic behaviorism defines the psychological level of analysis as the study of whole organisms interacting in and with a context. At this level of analysis behavior cannot be separated from context, and parts of the or- ganism cannot be separated from the whole. Because the unit is considered



to be fundamentally indivisible and interactive, contextualists hold that psy- chology can never be explained by events at other levels of analysis. For ex- ample, a psychological act-in-context cannot be explained by appeal to ac- tions of various parts of the organism involved in the interaction (e.g., its brain, glands, etc.). This is not to say that contextualistic behaviorism denies the legitimacy of other levels of analysis (biology, anthropology). On the con- trary, these other levels are recognized as legitimate in their own right and are further assumed to provide a context for psychological analyses.

Contextually Delimited Constructs

The unit of analysis in contextualistic behaviorism is an interactive whole. All constructs in a contextualistic system are contextually delimited. An act alone and cut off from a context is not considered a psychological act at all. Legs moving is not the same action as walking to the store.

Pragmatic Truth Criterion

If the unit is interactive, all abstractions of distinct features are themselves

arbitrary acts occurring in and with a context. In other words, the act of anal- ysis is the situated action of the analyst, not the discovery of truth. As with all "acts-in-context" we can organize actions on the basis of their purpose:

in behavioral terms, operants are organized by their consequences. This means that selecting among certain scientific constructs over others can only be justified on the basis of utility--the achievement of some end that the analyst

is seeking in doing the analysis. Skinner was quite clear on this point, claiming

that scientific knowledge "is a corpus of rules for effective action, and there

is a special sense in which it could be 'true' if it yields the most effective action

(A) proposition is 'true' to the extent that with its help the listener


responds effectively to the situation it describes" (Skinner, 1974, p. 235).

Goals of Analysis

Unlike all other world views, the truth criterion of contextualism is subor- dinated to something else: The goals of the analysis. "Serious analysis for [the

If from one

texture you wish to get to another, then analysis has an end, and a direction,


contextualist] is always either directly or indirectly practical

, enterprise becomes important in reference to the end" (Pepper, 1942, pp.

250-251). From a contextualistic perspective, any event or any feature of an event is conceptualized as an event-in-context. This conceptualization presents

and some strands have relevancy to this end and others do not, and


problem when the context of an event comes under consideration, because


then becomes the event-in-context with the result that a contextualistic anal-

ysis of anything can devolve into an appreciation of the whole, about which nothing can be said. A word about the whole holds itself apart from the whole, and thus a true appreciation of the whole is mute (L. J. Hayes, in press).

What saves contextualism from silence is a goal. The act of the scientist

is an act in context. It is a purposive act (not causally but descriptively). Anal-

ysis ends not with a discovery of the truth, but with the production of verbal constructions that help achieve a goal. In contextualistic behaviorism, "truth"



is that achievement. Any goal may be embraced, but the analytic practices

useful in terms of one goal may not be useful for another (Hayes & Brown- stein, 1986). Thus, contextualistic psychologies may differ widely depending on their goals.

No justification can be given for the selection of these goals because justification necessarily appeals to truth or correctness, and for a contextu- alist, truth cannot be evaluated except in terms of a goal. Goals are therefore

a priori events that can be articulated but not justified. If a scientist seeks,

for example, a personal experience of coherence and eschews any interest in prediction and control, this cannot be criticized as illegitimate. Many prag- matists, starting with James himself, have failed to appreciate the necessary but a priori nature of goals for contextualistic analyses. As a result they have

either been vague about their goals or have dogmatically asserted that their goals were the only ones possible. Skinner, for example, very clearly adopted

a pragmatic view of truth (see the quote at the end of the last section), but

he often seemed to be arguing that prediction and control were the goals of science, not just his goals. Contextualistic behaviorism, as here defined, is a psychological variety of what has been termed "functional contextualism" (Biglan & Hayes, 1991). "Functional contextualism" was coined to distinguish positions having an in- strumental character from more descriptive forms of contextualism that seek simply an understanding of participants in an interaction (e.g., Rosnow & Georgoudi, 1986). The behavior analytic form of contextualistic behaviorism has as its goals the description, prediction, control, and interpretation of or- ganismic interactions in and with a context. It seeks empirically-based anal- yses that achieve all of these goals jointly (not any one in isolation) with preci- sion (a restricted set of constructs apply to any particular event), scope (a wide number of events can be analyzed with these constructs), and depth (analytic constructs at the psychological level cohere with those at other levels). Causality. The interest in control does not mean that contextualistic be- haviorists embrace an ontological view of causality. A holistic perspective cannot take the position that there really are "causes."For one event to "cause" another at least two independent events are necessary. All contexts are in prin- ciple relevant to any event, and there are no truly independent events (Hayes, 1989). Causality is instead taken to be a useful way of speaking about the achievement of goals when certain contextual features are assumed. For example, for a contextualist, saying that a spark caused the explosion may be a useful way of speaking because it points to ways of avoiding such results: avoid sparks. Fuel, oxygen, and a certain base temperature were also necessary features for the explosion, and these might have been described as "causes." The operation of these features in turn depend upon other contex- tual features (e.g., in the early stages of the big bang substances interacted in different ways), that could also be called "causes." The selection of any one "cause" depends upon the other features being assumed, and the pragmatic effect of the description. For instance, in most human environments oxygen is present, and the spark is a more important focus of causal constructions. In other cases (as when welding in a vacuum) the spark is assumed and explo-



sions are more likely to be explained by the loss of the vacuum and the resul- tant presence of oxygen. This concludes our discussion of philosophical developments in behavior analysis. It is to contextualistic behaviorism that we are seeing some contem- porary behavior therapists turn their attention in recent years. We suspect that contextualism represents a more palatable philosophical position for clinical workers than did traditional mechanistic behaviorism. This aside for the mo- ment, recall our earlier suggestion that the evolution in the movement known as behavior analysis was occurring on more than one front. The other notable difference between behavior analysis as originally conceived and the new wave of this movement has to do with a set of unusual findings that are drawing into question some long held assumptions and premises. We turn now to these developments.

Stimulus Equivalence

When humans are taught a series of related conditional discriminations, the stimuli that enter into those discriminations often become connected to one another in ways not explicitly trained. The phenomena involved are typi- cally investigated in a matching to sample format. For example, given the pres- ence of a particular unfamiliar visual form (the sample- call it A1), the person

Equivalence Relation

















\4k' A

















. /








A diagrammaticpresentation of the trained and derived relations in a typical stimulus

equivalencenetwork.The solidarrowsindicatetrained relations; the dotted arrowsindicate de- rived relations.

FIG. 1.



will be taught to choose another particular unfamiliar visual form from an array of three such forms (the comparisons-call them B1, B2, and B3). We could say that the person learns "given A1 pick BI." The person is then taught

to select another unfamiliar visual form from another array of forms, given

the same sample, or "given A1 pick CI." With this kind of training, it is likely that without additional training, the person will select A1 from an array of comparisons, given B1 or given C1 as samples. The person is also likely to select B1 given C1 as a sample, and C1 given B1 as a sample (e.g., Sidman,

1971; Sidman, Cresson, & Willson-Morris, 1974), This set of phenomena is called "stimulus equivalence" (see Figure 1). An equivalence class is said to exist if the stimuli in the class show the three defining relations of reflexivity, symmetry, transitivity, and combinations of these (Sidman & Tailby, 1982). This definition has some notable difficulties, but we will adopt it as a point of departure. In matching-to-sample proce- dures, reflexivity is identity matching. For example, given A the person picks

A from an array. Symmetry refers to the functional reversibility of the condi-

tional discrimination: The trained discrimination "given A1 pick BI" leads

to the derived discrimination "given B1 pick AI." This reversibility must be

demonstrated in the absence of direct reinforcement to be considered sym- metry (Sidman, Rauzin, Lazar, Cunningham, Tailby, & Carrigan, 1982). To demonstrate transitivity, at least three stimuli are required. If after the dis- criminations "given A1 pick BI" and "given B1 pick CI" have been taught,

"given A1 pick CI" emerges without additional training, transitivity has been demonstrated. The derived relations "given C1 pick AI" is usually termed simply an equivalence relation and is viewed as a combination of symmetry and tran- sitivity (Fields, Verhave, & Fath, 1984).

A major source of interest in stimulus equivalence is the apparent corre-

spondence between the stimulus equivalence phenomenon and language phenomena. In naming tasks, symmetry and transitivity between written words, spoken words, pictures, and objects is commonplace. Several studies on stimulus equivalence have used naming-like preparations using auditory and visual stimuli (e.g., Dixon & Spradlin, 1976; Sidman, 1971; Sidman & Tailby, 1982; Sidman, Kirk, & Willson-Morris, 1985; Spradlin & Dixon, 1976). The

equivalence phenomenon may provide a new avenue for the empirical investi- gation of language. For many, it has become a kind of working model of semantic relations.

If stimulus equivalence is a preliminary model of verbal stimulation, one

would expect to see it emerge readily in humans, but not so readily or perhaps

not at all in non-humans. This has turned out to be the case. Stimulus equiva- lence has been shown with a wide variety of human subjects using a wide va- riety of stimulus materials (Dixon, 1977; Dixon & Spradlin, 1976; Gast, Van- Biervlet, & Spradlin, 1979; Hayes, Tilley, & Hayes, 1988; Mackay & Sidman, 1984; Sidman, 1971, Sidman et al., 1974; Spradlin, Cotter, & Baxley, 1973; Sidman & Tailby, 1982; Spradlin & Dixon, 1976; VanBiervlet, 1977; Wulfert

& Hayes, 1988) even with children as young as two years old (Devany, Hayes,

& Nelson, 1986). It has not been shown with non-human organisms, however

(D'Amato, Salmon, Loukas, & Tomie, 1985; Kendall, 1983; Sidman et al., 1982;


nnws ~ HAVES

Lipkens, Kop, Matthijs, 1988. As for Mclntire, Cleary, & Thompson, 1987 and Vaughan, 1988, see Hayes, 1989a). Furthermore, children without spon- taneous productive use of signs or speech also show little equivalence (Devany et al. 1986). To understand why stimulus equivalence is having a major impact on be- havioral researchers one should note that is not readily predicted from a three term contingency formulation. If an organism learns that the probability of reinforcement for selecting B1 is greater in the presence of A1 than in the pres- ence of, say, A2, this does not imply that the probability of reinforcement for selecting A1 is greater in the presence of B1 than in its absence. The development of theoretical accounts of stimulus equivalence is still in its infancy in behavior analysis. Some researchers view stimulus equivalence as a primitive, unanalyzable into component processes (e.g., Sidman 1986; 1990). Another alternative has appealed to the historical development of rela- tional responding (Hayes, 1991a; Hayes, & Hayes, 1989). The latter view holds that the action of relating two arbitrary stimuli itself has a history. For ex- ample, with enough instances of directly trained symmetrical responding in a given context, derived symmetrical responding may emerge with respect to novel stimuli in that context. Such relational derivation is arbitrarily applicable in the sense that it need not be based on formal properties of the relata, and may be brought to bear on stimuli by virtue of contextual cues to do so. In this view, stimulus equivalence is viewed as a special case of arbitrarily ap- plicable relational responding (Hayes & Hayes, 1989). In both analyses, how- ever, equivalence is a case of something new. Neo-behavioral analyses of this kind of phenomenon have existed for most of this century. The existing behavioral interpretations, however, either have appealed to derived relations between stimuli and responding to explain de- rived relations between stimuli (Hull, 1934), or have relied on processes such as classical conditioning to explain the findings (e.g., Staats & Staats, 1957). The former approach begs the question-the latter imposes process limita- tions that do not fit the actual experimental preparations used in the equiva- lence literature.

Implications of Contextualistic Behaviorism: The Example of Cognition

To show where these philosophical developments and new findings may lead, we will consider the topic of cognition. The role of cognitive events in psy- chopathology has been among the most discussed topics in behavior therapy over the last decade. There are arguments about whether or not cognition plays a controlling role in organism-environment interactions, arguments about the nature of cognition or the nature of disturbed cognition, and arguments over the role of cognitive change in the amelioration of psychopathology. In this paper we will consider the nature of cognition, the nature of cognitive con- trol, and the clinical implications of this analysis.

The Nature of Cognition

Cognition and the whole organism. From the view of contextualistic be- haviorism, cognition is an activity of a whole organism interacting in and with



the world. In other words, it is psychological activity.

activity of the brain, nor is it activity taking place in a nonspacio-temporal "mental" world. Brain activity occurs concomitant to all actions of whole or- ganisms, but as a matter of definition the actions of parts of the organism are not psychological actions. It would be a rare psychologist who suggested that legs walk to the store or genitals make love, yet the idea that brains think seems to be widely and uncritically accepted. Since contextualistic behaviorism uses the term "behavior" for psycholog- ical interactions, it is proper to say that cognition is behavior. This is not enough, however. What kind of behavior is it?

Cognition as a kind of knowing. Cognition is a kind of knowing. The word "know" in English is a very interesting one, because it has two entirely distinct etymological roots: the Latin words "gnoscere,"meaning "know by the senses; perception" and "scire," meaning "know by the mind; reason." The accident of the similar pronunciation of these Latin words, combined with their related meanings, lead in English to their collapse into a single verb: to know. Thus knowing is commonly said to have two aspects: apprehending and compre- hending. Originally, these were simply two different kinds of knowing. Historically, cognitive psychologyand behavior analysis each studied a single kind of knowing. Behavior analysis studied knowing by direct contact with the world (e.g., direct principles of contingency control), while cognitive psy- chology studied knowing by indirect, logical means (e.g., principles of cogni- tive mediation). It might be assumed that a more complete account could come from their combination, but the contextualistic behaviorist can find little of value in most of cognitive psychology, even if it is admitted that there are two kinds of knowing. The philosophy and goals differ so dramatically that the analytic constructions in one have little utility for the other.

It is specifically not the

In the hands of information processors,

cognitive psychologyis mechanistic.

The computer serves as the ultimate machine model. Hypothetico-deductive

theories are developed that purport to identify the true parts of this machine via predictions of its functioning. The experimental preparations are for the most part designed to isolate hypothesized cognitive components, rather than to examine cognition as situated action. (The "talk aloud while thinking"

methods described by Ericsson & Simon,

tion, but the exception proves the point. These methods were first used by John Watson in 1920and have been readily appreciated by modern behavior analysts, e.g., Hayes, 1986; Wulfert, Dougher, & Greenway, 1991). Context is consid-

ered mechanically if at all. Cognitive processes are taken to be "synonymous with brain processes" (Ellis & Hunt, 1983, p. 11), and thus reducible to the isolated action of a part of the organism. No sense of the whole organism interacting with the world is sustained. Rather than attempt to combine the two fields, a more philosophically sen- sible approach to greater completeness is for contextualistic behavior analysis to develop analyses for both kinds of knowing. What, behaviorally, might it mean to know the world indirectly, not through direct experience or con- tact? This is where stimulus equivalence comes in.

Stimulus equivalence and the second kind of knowing. If an event "A" has

1984 provide an interesting excep-




a psychological function, and that event is in an equivalence relation with an- other event "B", under certain conditions "B" may acquire a new psycholog- ical function based on the function of "A" and the relation between "A" and "B." Suppose a child is trained that the written word DOG is called "dog" and that the word goes with actual dogs. We may say that the child has had two relations directly trained: DOG ~ dog and DOG ~ "dog." Later the child plays with a dog for the first time and enjo2cs it. We may say that dogs have directly acquired various joyous functions by virtue of the play. Now, upon hearing his mother say "dogs" from another room the child may smile and go to the other room even though a dog is not visible and the child has no direct history of reinforcement for any of these activities in response to the word "dog." The child knows something about the word "dogs," but it is an indirect form of knowing. This kind of effect has been shown in many studies. Transfer of discrimina- tive functions have been shown across equivalence relations in simple equiva- lence classes (Hayes, Brownstein, Devany, Kohlenberg, & Shelby, 1987; Koh- lenberg, Hayes, & Hayes, 1991), and conditional equivalence classes (Wulfert

& Hayes, 1988), and across symmetrically related stimuli (e.g., Catania, Horne,

& Lowe, 1989; De Rose, Mcllvane, Dube, Galpin, & Stoddard, 1988; Lazar,

1977; Lazar & Kotlarchyk, 1986; Gatch & Osborne, 1989). Transfer of conse- quential functions has also been shown (Hayes et al., 1987; Hayes, Kohlen- berg, & Hayes, 1991). Because this second kind of knowing occurs readily in verbal humans but with difficulty or not at all in non-humans or non-verbal humans, it is easy to see how behavior analysis-with its commitment to the animal behavior tradition-and cognitive psychology have had so little to say to one another. From a modern behavior analytic viewpoint, cognition may be defined as the derivation of arbitrarily applicable stimulus relations. "Logic" and "reason" refer to the derivation of such relations according to certain conventional rules. Because these rules are themselves the product of (or an instantiation of) ar- bitrarily applicable derived stimulus relations it is not correct to say that logic or reason produce cognitive interactions-rather they are such interactions (Hayes, 1989b).

Cognitive Control

We come now to the key issue of the relation between certain cognitive events-we will use the words "thoughts"-and other forms of activity such as emotion or overt behavior. In the cognitive tradition the link is mechanical. Thus, the focus of therapy is on the nature or occurrence of particular thoughts. If a person is thinking an "irrational" thought and then feeling anxious, the focus of therapy would be on the presence of this thought or the sources of its "irrational" form; that the thought leads to anxiety would be explained on the basis of these same dimensions of presence and form. Based on conceptualizations such as these, therapists can either: a) estab- lish a new thought (e.g., the therapist may help the person relate "should" and "irrational" for the first time); or b) make an old thought more or less likely (e.g., asking a client to think "self-reinforcing" thoughts whenever a partic-



ular situation has been faced). What is important about these approaches is that the thought-emotion relation itself is not at issue, much as the movement of one end of a lever is easily explained by the pressure on the other end. The two are mechanically linked. The behavioral objection to this kind of theorizing is pragmatic. Skinner expanded the objection to such a degree that he saw little use in the analysis of emotions or cognitions. The nature of his objection, and the reasons for his excessive expansion of the point, deserve a more extended discussion.

Cognition and thepragmatic purposes of contextualistic behaviorism. The

conceptualization of the relationship between thoughts and other activities has assumed a variety of forms within behaviorism and behavior therapy (see Figure 2). The Watsonian behaviorists essentially tried to eliminate non-motor or glandular behavior from consideration, either because the existence of such was in question or because an adequate method for their scientific study was unavailable. A few early behavior therapists encouraged a focus on overt be-

havior only, on the assumption that thoughts (and feelings, etc.) would change on their own. The current mainstream position in behavior therapy is that

a change in thoughts will produce changes in overt activity. Traditional as-

sociationistic forms of behavior therapy and cognitive therapy do not differ in this regard. The idea that relaxation will encourage a phobic to approach

is philosophically the same as the idea that rational thoughts will do likewise.

Contextualistic behaviorism takes a different view. It has an interest in the joint goals of description, prediction, control, and interpretation. The inclu- sion of control puts certain constraints on the kinds of analyses that are useful for the scientist, because only statements that point to events external to the behavior of the individual organisms being studied can directly lead to con-

trol. Scientific rules are rules for scientists and their consumers, not rules for the world. Since therapists are-and can only ever be-in another organisms's environment, rules for modifying behavior must start there. Skinner said it this way: "In practice, all these ways of changing a man's mind reduce to manipulating his environment, verbal or otherwise" (Skinner, 1969, p. 239). Of course, analyses of human action that do not seek control as a goal need not be constrained in this way (Hayes & Brownstein, 1986), but in the world of therapy such a goal is assumed-by the client at least. Thus, control as a goal forces a contextual focus to psychological analyses. Mediational analyses of all kinds must be rejected when the supposed media- tional elements are in principle non-manipulable (cf. Watkins, 1990). When thinking is related to other psychological activities, a contextual behaviorist has two things to point out: First, both forms of activity are the dependent variables of psychology and both occur in and with a context. A claim that

a thought controls overt behavior is at best the description of a behavior-

behavior relation. Second, a behavior-behavior relation is itself an event-in- context. The structure of the contextualist's point is the same whether one is speaking of the behavior effects of thinking, emoting, sensing, and so on. These are all conceived of as behavior-behavior relations (where "behavior" is defined as psychological activity), and such a relation itself is viewed as contextually




Conceptualizing the Relation Between Thinking and Other Forms of Behaving





Overt Behavior




Overt Behavior




Overt Behavior




FIG. 2,


Four ways of conceptualizing the relationship between thinking and overt acting.

situated. Any behavior-behavior relation is inherently incomplete as an expla- nation if one's goals include control because the direct application of this knowledge is impossible. Behavior cannot be directly manipulated- only con- textual variables have this quality. If causality is a way of speaking about how to accomplish such goals, behavior-behavior relations can never be "causal" for the contextual behaviorist.



For example, the formulation "John does poorly because he thinks irra-

tional thoughts" must be rejected as a causal formulation, not because there

is no such thing as thinking or because there can be no relation between thinking

and other activity but because it is impossible to act directly given such a for- mulation. To intervene, the clinician must do something (e.g., talk to the client). Whatever is done will be the context for the client's future action. Therefore, when measured against a pragmatic criterion, causal constructions must point

to ways of altering context, not to the effect of action cut off from context.

A more directlyuseful formulation would specifythe contextual features that

participate in doing poorly, thinking irrationally, and in the relationship be- tween them.

If one sought only prediction, thinking could be considered "causal" by

a contextualist, but it would be a rare therapist who sought only prediction.

If, conversely, one is a mechanist, it is quite legitimate to view thoughts as causes, because causality for a mechanist is a description the operation of the machine based on its parts, relations, and forces. Skinner took the pragmatic point one step further. He argued that feelings, cognitions, and so on were not important to the understanding of behavior because the conditions that gave rise to feelings (etc.) were isomorphic with those that gave rise to associated overt behavior. For example, while recog- nizing that aversive events occasion both anxiety and avoidance, Skinner claimed that "the middle term is of no functional significance, either in a the- oretical analysis or in the practical control of behavior" (1953, p. 181). No new contextual features emerge in the analysis of feeling or thoughts. In effect, Skinner's radical behaviorism gave scientific legitimacyto cognition and emo- tion with one hand, but took away their practical importance with the other (Parrott, 1986). There is a flaw in Skinner's reasoning that is revealed in the stimulus equiva- lence phenomenon. The histories that give rise to the derivation of stimulus relations and the transfer of control seen in equivalence are not the histories that establish a particular instance of a trained stimulus relation. In an organism without derived stimulus relations of this kind-without the second kind of knowing- Skinner is entirely correct. If all psychological history is either direct or indirect based upon formal properties of events (e.g., stimulus generalization), an analysis of overt behavior can always ignore the phenomena of emotion and cognition. A rat who is afraid and avoids some particular circumstance does not avoid it because of fear. Rather, both the fear and the avoidance are due to a direct and identical history with that cir- cumstance (of shock, for example). There is, as a result, no need to study fear as a means of understanding avoidance. In verbal organisms the situation is different. The conditions that gave rise to fear and avoidance, and the fear and avoidance themselves, may all enter into networks of derived stimulus relations. The social/verbal history of the person participates in the overt behavior. Verbal organisms have a history with both kinds of knowing. A thought-overt behavior relation involves both, and analyzing the thought leads to a contextual analysis of a key feature of the total event. For example, a verbal adult has learned many verbal rules about



fear and what it means to be fearful. In situations viewed as fearful by the person (such as feeling closed in at a mall), this verbal history is part of the total event. No amount of understanding of a person's direct history with malls is sufficient to understand the role of these indirect forms of knowing. For this reason, understanding the behavior-behavior relations between thinking and overt action, emotion and overt action, and so on, can be crucial to un- derstanding a given instance of behavior. We earlier defined contextualistic behaviorism on the basis of its view of the psychological level of analysis, a commitment to contextually delimited constructs, a pragmatic view of truth, and on the basis of its goals. The anal- ysis of cognition has these same features. Cognition is viewed as an interac- tion of a whole organism in and with a context. The analysis of cognitive con- trol is explicitly pragmatic, and is guided by the goal of controlling the phenomenon. Many common analyses are rejected as a result.

Implications for Cognition and Therapy

The contextualistic behavior conceptualization leads to an entirely different range of therapeutic approaches. Instead of focusing on the presence or ab- sence of cognitive events, contextual variables can be manipulated to poten- tiate or depotentiate the thinking-acting (behavior-behavior) relation itself. Whereas cognitive therapy has focused on a change in the form of private events, contextual therapies focus on changing their function without neces- sarily changing their form. When an individual derives a stimulus relation, this action occurs in a con- text established by the social/verbal community. Basic research on stimulus relation derivation has shown contextual control over both the form of rela- tions derived (e.g., Steele & Hayes, 1991; Wulfert & Hayes, 1988) and the transfer of functions through these relations (e.g., Hayes, Kohlenberg, & Hayes, 1991; Hayes, Thompson, & Hayes, 1989). Thus, the impact of derived relations is itself contextually sensitive. Suppose an agoraphobic thinks "I am going to humiliate myself in this mall," then starts to feel panicky, and leaves the mall. A contextualistic behaviorist would not explain the panic or escape on the basis of the form of the thought per se, but on the contexts giving rise to the thought-emotion or thought-overt behavior relation. These contexts include a) those that give meaning to the content of the thought, or the context of literal meaning; b) those that estab- lish the valence of humiliation, or the context of evaluation; c) those that allow the person to "explain" and "justify" behavior on the basis of the private events, or the context of reason-giving; and d) those that establish the goal of avoiding

"undesirable" private events, or the context of emotional and cognitive control.

Instead of trying to change the thought or the emotion, the therapist can change these contexts in an attempt to change the behavior-behavior relation itself (Hayes, 1987; Hayes, Kohlenberg, & Melancon, 1989; Hayes & Melancon, 1989). We will give a brief example of contextual strategies for depotentiating a behavior-behavior relation in each of these four cases. Some of the examples are trivial, but they have been selected for the ease with which the point can be made.



a) The context of literal meaning depends upon socially conventional equiva-

lence classes and other derived stimulus relations. Altering that context re- quires altering the "meaningful" use of language, not by providing new equiva- lence classes (e.g., new thoughts or new beliefs), but by using language in ways that are confusing, nonsensical, arational, or experiential. To take one example originally used by Titchner in demonstrating his context theory of meaning,

(Lundin, 1990), if the person said the word "humiliation" over and over again very fast for perhaps 5 minutes, the word would temporarily loose all rela- tional functions, retaining only its auditory functions. This would not have lasting impact in itself, but would help the client see that literal meaning is a contextually-controlled phenomenon, and the issue may not be the form of the thinking but the contexts that determine the impact of thinking.

Similarly in Acceptance and Commitment Therapy (ACT) the beginning

client is gravely warned by the therapist not to believe a word the therapist is saying (Hayes, 1987). This kind of statement attacks literal meaning, be- cause if it is itself believed then it must not be believed. With enough attacks of this kind of literal meaning, the stimulus function of verbal events may begin to depend less on their literal (conventional) meaning and more on their role as a guide to effectiveaction. Many thoughts, no matter how "reasonable," have proven themselves to be useless guides to action. For example, the obsessive-compulsive has repeatedly experienced the futility of effortfullytrying to eliminate undesirable thoughts. In this context, we would argue that the obsessive-compulsive is trying to follow a completely reason- able rule: "I don't like thinking X; therefore do not think X." Unfortunately, "do not think X" is itself a thought about X and is thus doomed to fail. The

client has repeatedly experienced the worthlessness of this thought as a guide to action, and yet its literal reasonableness is so great that it is hard to abandon. Rather than argue logically with an eminently reasonable thought, an alter- native is to attack the hegemony of literal meaning itself. b) The context of evaluation is based on conventional verbal agreements about what is bad and good. When combined with literal meaning, evalua- tions present themselves not as actions, but as actions of the world: we say "that is bad" not "that is, and I evaluate it as bad." Such a verbally-established illusion structures our environment in a powerful way because it hides the in- direct and arbitrary nature of evaluation and places it in the world of things. If this very process is a problem, it does no good for a clinician to attempt to alter evaluative statements directly (e.g., "you shouldn't say should") be- cause the clinician is using the selfsame context of evaluation. The context of evaluation is challenged by refusing to order events evaluatively. To take one example, also from ACT, a therapist might take any report of any emo- tion, thought, memory, and so on, regardless of its conventional valence, and respond to it positively but non-differentially. This should challenge the con- text of evaluation because the therapist's openness would not depend upon the evaluative content. Example: Client -"I am worried I'll humiliate myself"; Therapist - "Wonderful. That's really neat. Tell me about that."

c) The context of reason-giving is established by the tendency of the verbal

community to support actions if a sufficiently good explanation is given for



them. Often reasons for one action devolve into the presence of other actions (e.g., "I couldn't go because I was afraid" or "I did it because I was worried about you."). When these verbal linkages are supported by listeners, the behavior-behavior relation itself is also supported because the presence of one behavior signals altered contingencies surrounding the other. For example, a slight illness that would never lead to cancelling an impor- tant engagement might lead to cancelling an undesirable appointment. The cancellation can be done "with a clear conscience" and a ready and "honest" excuse is available, even to oneself. Reason-giving thus tends to glue behaviors together in conventional ways. A variety of strategies might challenge reason- giving. It could be challenged directly. For example, if a client is asked "why" repeatedly, the shallowness of reason-giving quickly emerges: Client-"I decided I had to leave, even though it didn't make sense," Therapist-"why did you decide that?", Client-"well, I just wanted to go," Therapist - "why did you want that?", Client-"I don't know! Why are you asking me this?" Usually within about four "whys" the verbal structure of reasonableness begins to fray, to be replaced by irritation at the "unreasonable" behavior of the ther- apist. Reason-giving could be challenged more indirectly simply by treating reason-giving as a thing to be noticed rather than an explanation to be taken literally. Example: Client-"I just had to leave. I had no choice." Therapist- "That's an interesting thought. You ought to thank your mind for that one. What else has your mind had to say?" d) The context of emotional control is a core issue for many clients. It is common for clients to come into therapy with a list of undesirable thoughts, feelings, memories, and bodily sensations that seemingly need to be removed, altered, or avoided, and indeed we as therapists name most of the disorders we treat and the treatments themselves in the same way ("anxiety disorders," "anxiety management," and so on). Emotional control glues behaviors together because usually implicit in the effort to change feeling X is that if it doesn't change some other form of activity cannot occur (e.g., "I can't travel! I'm too anxious!"). The context of emotional control is often challenged in acceptance- oriented approaches through deliberate exercises that create the private events that are being avoided. Although emotional control has been massively sup- ported in mainstream behavior therapy, the recent interest in acceptance-based approaches in behavior therapy shows signs of possible changes to come (e.g., Hayes, 1991b; Jacobson, 1991b). In ACT numerous metaphors and exercises try to make evident the pervasiveness and uselessness of emotional control efforts. Often efforts are made to take the client's own words and turn them from emotional control declarations into emotional exposure declarations. For example, a depressed client might say "I'll never do well until I feel better" to which the therapist might reply "I think that's right. You won't do well until you do a better job of feeling what is here to be felt-until you feel better." Later on, every time "feel better" is used by the client to explain why some feeling has to change, "feel better" can be brought up to shift the issue. Contextual strategies often seem confusing or paradoxical toclients because they fail to address content in the normal way. Contextual strategies neither



agree with nor disagree with the content of clients' self-verbalizations. Rather, they place these activities into a different social/verbal context. For example, in the context of evaluation and emotional control, "negative thoughts" must be changed. These contexts are so ubiquitous that their presence is hardly no- ticed. A client complaining of being terribly worried or terribly anxious im- plicitly expects a therapist to agree that worry or anxiety is bad and must be removed. If the therapist does not behave accordingly, the client will often be extremely confused- as if the natural order of things is being challenged. In the different social/verbal context of deliteralization and emotional accep- tance, thoughts evaluated as negative may simply be noticed, with no efforts either to change them or to behave in accord with them. There is no necessary reason that worry or anxiety or sadness or any private event must be removed if their disruptive effects are dependent upon particular contexts. A second alternative exists: change the contexts. In principle, there is no reason that the direct cognitive strategies and the contextual strategies could not be combined. For example, a cognitive ther- apist might arrange conditions that potentiate a thought-emotion relation in order to provide more evidence for the need to modify thoughts themselves. Ellis (1962) has long combined cognitive control with acceptance strategies, as has Barlow and others (e.g., Barlow & Craske, 1989; Barlow, Craske, Cerny, & Klosko, 1989). For our purposes it is more important to see that purely con- textual forms of therapy are also possible, in which cognitions are dealt with continuouslyand deliberately, but without any interest in creating new thoughts or modifying the probability of old ones (e.g., Hayes, 1987). The reason this is important for the topic is that it shows how unusual and counter-intuitive therapy alternatives flow from contextualistic behavioral philosophy. The data on the impact of such strategies are limited, but those that are available show both that these strategies can make an impact on a variety of disorders (Biglan, 1990; Hayes, 1987; Zettle, 1984; Zettle & Raines, 1989) and that they work through different psychological processes than cognitive strategies (Khor- akiwala, 1990; McCurry, 1991; Zettle & Hayes, 1986; Zettle & Raines, 1989).


Contextualisticbehaviorism is a subtle position, with assumptions that differ radically from normal ways of thinking about the world. We have considered the topic of cognition in this article, but we might instead have analyzed the role of sense of self, of consciousness, or the therapeutic relationship, or dozens of other areas where the therapeutic implications of this approach are novel and non-obvious. The therapeutic value of ideas is ultimately an empirical matter, but it is important to have a range of ideas to test. Behavior therapy has been relying on common sense, associationism, and cognitive theory for many years. Perhaps a renewed and enlightened examination of contemporary behaviorism will provide an alternative source of technical innovation and theoretical analysis.





Bandura, A. (1969). Principles of behavior modification. New York: Holt, Rinehart, & Winston. Barlow, D. H., Craske, M. G., Cerny, J. A., & Klosko, J. S. (1989). Behavioral treatment of panic

disorder. Behavior Therapy, 20, 261-282.

Barlow, D. H., & Craske, M. G. (1989). Mastery of your anxiety andpanic. New York: Graywind Publishing Company. Biglan, A. (1990). A contextual approach to treating family distress. In G. Singer & L. lrvin (Eds.),

Supporting the family: Enabling a positive adjustment to children with disabilities. Balti-

more: Paul H. Brookes.

Biglan, A., & Hayes, S. C. (1991). Should the behavioral sciences become more pragmatic? The casefor functional contextualism in research on human behavior. Unpublished manuscript.

Brewer, W. F. (1974). There is no convincing evidence for operant or classical conditioning in adult humans. In W. B. Weimer & D. S. Palermo (Eds.), Cognition and the symbolic processes. Hillsdale, NJ: Erlbaum.

Catania, A. C., Horne, P., & Lowe, C. E (1989). Transfer of function across members of an equiva-

lence class. The Analysis of Verbal Behavior,

7, 99-110.

D'Amato, M. R., Salmon, D. P., Loukas, E., & Tomie, A. (1985). Symmetry and transitivity of conditional relations in monkeys (Cebus apella) and pigeons (Columba livia). Journal of

the Experimental Analysis of Behavior, 44, 35-47.

De Rose, J. T., Mcllvane, W. J., Dube, W. V., Galpin, V. C., & Stoddard, L. T. (1988). Emergent simple discrimination established by indirect relation to differential consequences. Journal

of the Experimental Analysis of Behavior, 50, 1-20.

Devany, J. M., Hayes, S. C., & Nelson, R. O. (1986). Equivalence class formation in language-

able and language-disabled children. Journal of the Experimental Analysis of Behavior, 46,


Dixon, L. S. (1977). The nature of control by spoken words over visual stimulus selection. Journal

of the Experimental Analysis of Behavior, 27, 433-442.

Dixon, M. H., & Spradlin, J. E. (1976). Establishing stimulus equivalences among retarded adoles-

cents. Journal of Experimental Child Psychology, Ellis, A. (1962). Reason and emotion in psychotherapy.

Ellis, H. C., & Hunt, R. R. (1983). Fundamentals of human memory and cognition (3rd ed.).

Dubuque: Brown. Ericsson, K. A., & Simon, H. A. (1984). Protocol analysis: Verbal reports as data. Cambridge, MA: MIT Press. Eysenck, H. J. (1972). Behavior therapy is behavioristic. Behavior Therapy, 3, 609-613. Fields, L., Verhave, T., & Fath, S. (1984). Stimulus equivalence and transitive associations: A

methodological analysis. Journal of the Experimental Analysis of Behavior, 42, 143-157. Franks, C. M., & Wilson, G. T. (1974). Annual review of behavior therapy." Theory and practice.

New York: Brunner/Mazel. Gast, D., VanBiervlet, A., & Spradlin, J. E. (1979). Teaching number-word equivalences: A study

21, 144-164. Secaucus, New Jersey: Stuart.

of transfer. American Journal of Mental Deficiency, 83, 524-527.

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.

Hayes, L. J. (1989). Philosophical implications of the interbehavioral field. The Interbehaviorist, 16, 23-27. Hayes, L. J. (in press). Telling the truth. In S. C. Hayes & L. J. Hayes (Eds.), Understanding verbal relations. Reno, NV: Context Press. Hayes, L. J., Thompson, S., & Hayes, S. C. (1989). Stimulus equivalence and rule following.

Journal of the Experimental Analysis of Behavior, 52, 275-291.

Hayes, L. J., Tilley, K., & Hayes, S. C. (1988). Extending equivalence class membership to gusta-

tory stimuli. The Psychological Record, 38, 473-482.



Hayes, S. C. (1986). The case of the silent dog: Verbal reports and the analysis of rules. A review

of K. Anders Ericsson and Herbert A. Simon, 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. Jacobson (Ed.), Psy-

chotherapists in clinicalpractice: Cognitive and behavioralperspectives (pp. 327-387). New

York: Guilford. Hayes, S. C. (1989a). Nonhumans have not yet shown stimulus equivalence. Journal of the Ex-

perimental Analysis of Behavior, 51, 385-392. Hayes, S. C. (Ed.). (1986b). Rule-governed behavior." Cognition, contingencies, and instructional

control. New York: Plenum. Hayes, S. C. (1991a). 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. (November, 1991b). While we weren't watching: Some non-obvious clinical implica- tions of advances in basic behavior analysis. Invited address presented at the meeting of

the Association for Advancement of Behavior Therapy, New York. Hayes, S. C., & Brownstein, A. J. (1986). Mentalism, behavior-behavior relations and a behavior analytic view of the purposes of science. The Behavior Analyst, 9, 175-190. Hayes, S. C., Brownstein, A. J., Devany, J. M. Kohlenberg, B. S., & Shelby, J. (1987). Stimulus equivalence and the symbolic control of behavior. Mexican Journal of Behavior Analysis, 13, 361-374. 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: Cognition, contingencies, and instructional

control. (pp. 153-190). New York: Plenum. Hayes, S. C., Hayes, L. J., & Reese, H. W, (1988). Finding the philosophical core: A review of

Stephen C. Pepper's WorldHypotheses. Journal of the Experimental Analysis of Behavior,

50, 97-111. Hayes, S. C., Kohlenberg, B. S., & Hayes, L. J. (1991). Transfer of consequential functions through simple and conditional equivalence classes. Journal of the Experimental Analysis of Be-

havior, 56, 119-137.

Hayes, S. C., Kohlenberg, B. S., & Melancon, S. M. (1989). Avoiding and altering rule control as a strategy of clinical treatment. In Hayes, S. C. (Ed.), Rule-governed behavior: Cogni-

tion, contingencies, and instructional control (pp. 359-385). New York: Plenum.

Hayes, S. C., & Melancon, S. M. (1989). Comprehensive distancing, paradox, and the treatment of emotional avoidance. In M. Ascher (Ed.), Paradoxicalprocedures in psychotherapy (pp. 184-218). New York: Guilford. Hull, C. L. (1934). The concept of the habit-family hierarchy and maze learning: Part I. Psycho-

logical Review, 41, 33-52.

Jacobson, N. S. (1991a). To be or not to be behavioral when working with couples: What does

it mean. Journal of Family Psychology, 4, 436-445.

Jacobson, N. S. (November, 1991b). Acceptance and change. Presidential address delivered to the meeting of the Association for Advancement of Behavior Therapy, New York.

Johnston, J. M., & Pennypacker, H. S. (1981). Strategies and tactics of human behavioral re-

search. Hillsdale, NJ: Erlbaum. Kendall, S. B. (1983). Tests for mediated transfer in pigeons. The PsychologicalRecord, 33, 245-256.

Khorakiwala, D. (1991). An analysis of the process of client change in a contextual approach

to therapy. Unpublished doctoral dissertation. University of Nevada, Reno, NV. 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, R. J., & Tsai, M. (1987). Functional analytic psychotherapy. In N. Jacobson (Ed.),



Psychotherapists in clinical practice." Cognitive and behavioral perspectives

(pp. 388-443).

New York: Guilford. Lazar, R. (1977). Extending sequence-class membership with matching-to-sample. Journal of

the Experimental Analysis of Behavior, 27, 381-392.

Lazar, R. M., & Kotlarchyk, B. J. (1986). Second-order control of sequence-class equivalences

in children. Behavioural Processes, 13, 205-215.

Lipkens, R., Kop, P. E M., & Matthijs, W. (1988). A test of symmetry and transitivity in the conditional discrimination performances of pigeons. Journal of the Experimental Analysis

of Behavior, 49, 395-409.

Lundin, R. W. (1990). Theories and systems of psychology (4th ed.). New York: Heath.

Mackay, H. A., & Sidman, M. (1984). Teaching new behaviors via equivalence relations. In

P. Brooks, R. Sperber, & C. McCauley (Eds.), Learning and cognition in the mentally retarded

(pp. 493-523). Hillsdale, NJ: Erlbaum. Mahoney, M. J. (1974). Cognition and behavior modification. Cambridge, MA: Ballinger. Mahoney, M. J. (1989). Scientific psychology and radical behaviorism: Important distinctions based in scientism and objectivism. American Psychologist, 44, 1372-1377.

Mahoney, M. J., Kazdin, A. E., & Lesswing, N. J. (1974). Behavior modification: Delusion or deliverance? In C. M. Franks & G. T. Wilson (Eds.), Annual review of behavior therapy:

Theory and practice (pp. 12-40). New York: Brunner/Mazel.

McCurry, S. M. (1992). Client metaphor use in a contextual form of therapy. Unpublished doc-

toral dissertation. University of Nevada, Reno, NV. Mclntire, K. D, Cleary, J., & Thompson, T. (1987). Conditional relations by monkeys: Reflex-

ivity, symmetry, and transitivity. Journal of the Experimental Analysis of Behavior, 47,


Meichenbaum, D. H. (1977). Cognitive-behavior York: Plenum.

Morris, E. K. (1988). Contextualism: The world view of behavior analysis. Journal of Experimental


An integrative approach. New

Child Psychology,

46, 289-323.

Parrott, L. J. (1983). Systemic foundations for the concept of private events: A critique. In N. M.

Smith, P. T. Mountjoy & D. Ruben (Eds.), Radical reassessment in psychology." The interbe-

havioral alternative.

Parrott, L. J. (2986). The role of postulation in the analysis of inapparent events. In H. W. Reese

New York: University of America Press.

& L. J. Parrott (Eds.), Behavior science: Philosophical,

vances (pp. 35-60). Hillsdale, NJ: Erlbaum.

methodological and empirical ad-

Pepper, S. C. (1942). World hypotheses:

A study in evidence. Berkeley: University of California


Rosnow, R. L., & Georgoudi, M. (Eds.). (1986). Contextualism and understanding

science. New York: Praeger. Sidman, M. (1971). Reading and auditory-visual equivalences. Journal of Speech and Hearing

in behavioral

Research, 14, 5-13.

Sidman, M. (1986). Functional analyses of emergent verbal classes. In T. Thompson & M. D.

Zeiler (Eds.), Analysis and integration of behavioral units (pp. 213-245). Hillsdale, New Jersey:


Sidman, M. (2990). Equivalence relations: Where do they come from? In D. E. Blackman &

H. Lejeune (Eds.), Behaviour analysis in theory and practice: Contributions and controver-

sies (pp. 93-114). Hillsdale, NJ: Erlbaum. Sidman, M., Cresson, O., & Willson-Morris, M. (1974). Acquisition of matching-to-sample via

mediated transfer. Journal of the Experimental Analysis of Behavior, 22, 261-273.

Sidman, M., Kirk, B., & Willson-Morris, M. (1985). Six-member stimulus classes generated by

conditional-discrimination procedures. Journal of the Experimental Analysis of Behavior,

43, 21-42. Sidman, M., Rauzin, R., Lazar, R., Cunningham, S., Tailby, W., & Carrigan, P. (1982). A search



for symmetry in the conditional discriminations of rhesus monkeys, baboons and children.

Journal of the Experimental Analysis of Behavior, 37, 23-44.

Sidman, M., & Tailby, W. (1982). Conditional discrimination versus matching to sample: An ex-

pansion of the testing paradigm. Journal of the ExperimentalA nalysis of Behavior, 3 7, 5-22.

Skinner, B. E

Skinner, B. E (1945).

52, 270-276. Skinner, B. E (1953). Science and human behavior. New York: The Free Press.

Skinner, B. E (1969). Contingenciesof reinforcement:A theoreticalanalysis.New York: Appleton-

Century-Crofts. Skinner, B. E (1974). About behaviorism. New York: Knopf. Spradlin, J. E., Cotter, V. W., & Baxley, N. (1973). Establishing a conditional discrimination without direct training: A study of transfer with retarded adolescents. American Journal

(1938). Behavior of organisms. New York: Appleton-Century-Crofts.

The operational analysis of psychological terms. Psychological Review,

of Mental Deficiency, 77, 556-566.

Spradlin, J. E., & Dixon, M. (1976). Establishing a conditional discrimination without direct training: Stimulus classes and labels. American Journal of Mental Deficiency, 80, 555-561. Staats, C. K., & Staats, A. W. (1957). Meaning established by classical conditioning. Journal

of Experimental Psychology, 54, 74-80.

Steele, D. L., & Hayes, S. C. (1991). Stimulus equivalence and arbitrarily applicable relational

responding. Journal of the Experimental Analysis of Behavior, 56, 519-555.

VanBiervlet, A. (1977). Establishing words and objects as functionally equivalent through manual

sign training. American Journal of Mental Deficiency, 82, 178-186.

Vaughan, W. (1988). Formation of equivalence sets in pigeons. Journal of Experimental Psy-

chology: Animal Behavior Processes, 14, 36-42.

Watkins, M. J. (1990). Mediationism and the obfuscation of memory. American Psychologist, 45, 328-335. Watson, J. B. (1920). Is thinking merely the action of language mechanisms? British Journal

of Psychology, 11, 87-104.

Watson, J. B. (1924). Behaviorism. New York: Norton. Wolpe, J. (1980). Cognitive behavior: A reply to three commentaries. American Psychologist, 35, 112-114. Wulfert, E., & Hayes, S. C. (1988). The transfer of conditional sequencing through conditional

equivalence classes. Journal of the Experimental Analysis of Behavior, 50, 125-144.

Wulfert, E., Dougher, M. J., & Greenway, D. E. (1991). Protocol analysis of the correspondence of verbal behavior and equivalence class formation. Journal of the Experimental Analysis

of Behavior, 56, 489-504. Zettle, R. D. (1984). Cognitive therapy of depression: A conceptual and empirical analysis of

component andprocessissues. Unpublished doctoral dissertation available from the University of North Carolina at Greensboro. Zettle, R. D., & Hayes, S. C. (1986). Dysfunctional control by client verbal behavior: The context

of reason giving. The Analysis of VerbalBehavior, 4, 30-38.

Zettle, R. D., & Raines, J. C. (1989). Group cognitive and contextual therapies in treatment of


Journal of Clinical Psychology, 45, 438-445.

RECEIVED: September 6, 1991 ACCEPTED: January 17, 1992