Beyond Skinner? a Review of Relational Frame Theory: a Post-Skinnerian Account Ofhuman Language and Cognition by Hayes, Barnes-Holmes, and Roche

The Analysis of Verbal Behavior 2003, 19, 19-27

Beyond Skinner? A Review of Relational Frame Theory: A Post-Skinnerian Account ofHuman Language and Cognition by Hayes, Barnes-Holmes, and Roche

J. Grayson Osborne, Utah State University In their book, Relational Frame Theory: A Post-Skinnerian Account ofHuman Language and Cognition (2001), Hayes, Barnes-Holmes and Roche challenge behavior analysts to put aside Skinner and Verbal Behavior in favor ofrelational frame theory's approach to human language and cognition. However, when viewed from the contexts ofbehavior analysis, the principles ofbehavior analysis, and the principles ofthe founder of behavior analysis, Relational Frame Theory fits squarely in the Skinnerian, behavior analytic tradition. As with Verbal Behavior, Relational Frame Theory and its theses may be thought of as logical and empirical extensions of that which precedes them.

Hayes, Barnes-Holmes, & Roche (2001) The Off-Putting Parts present a provocative account of human language and cognition in terms ofrelational frame Adversarial tone. The authors set an theory (RFT).' They mean their account to suc- adversarial tone almost immediately in the Pref- ceed where, they claim, Skinner's book, Ver- ace. They state that, iftheir account is correct, bal Behavior (VB), and its premises failed; "... many of the most prominent Skinnerian failed, that is, as a generally accepted account ideas about human complexity must be put of the definition, development, and operation aside or modified virtually beyond recognition ofhuman language and cognition. To some be- (p. xii)," and "... it is now time for behavior havior analysts, Skinner's purported failure in analysis to abandon many ofthe specific theo- this area will come as a surprise. To others, it retical formulations of its historical leader in will not. To still other behavior analysts, RFT the domain of complex human behavior ... (p. will be off-putting, less through its theses, than xii)." These sentences suggest a broad putting through its tone. The scope of RFT also will aside. If there is such, then the sentences are be seen as grand. It is, by necessity, as it pur- justified. Ifnot, they would seem to overreach, ports to explain all complex human behavior unnecessarily. This review will attempt to ex- (see, e.g., Hayes, Strosahl, & Wilson, 1999; amine just how much putting aside there re- Roche, Barnes-Holmes, Barnes-Holmes, ally is. Stewart & O'Hora, 2002). However, while the Writing style. The authors state their style "... authors attempt to separate themselves from will be a bit loose for scientific writing some Skinner-but not from behavior analysis-it sections will be lightly referenced, some pro- is the thesis ofthis review that they do not suc- cedures will be vaguely described, and so on ceed in the former, and, indeed, appear to be (p.1)." Without a formal presentation, schol- proceeding, in RFT, much in the manner of ars from outside behavior analysis will not be their illustrious predecessor. convinced in the absence ofthe details that are loosely omitted, and the presence of the sections that are lightly referenced, and the pro- I Throughout this review, RFT is used to refer to cedures that are vaguely described; whereas the book under review, and RFT, the theory that scholars from inside the area may consider the underpins the book. Sometimes the choice of one book unhelpful because it should be more for- over the other seems arbitrary and it probably is. mal. The book does read as though it has been loosely edited, e.g., the use of"I" or "My" when the authors are multiple (e.g., pp. 37,123,129, Address correspondence to J. Grayson Osborne, 130) and a later tact of this condition with in- Deptartment of Psychology, Utah State University, structions to the reader to ignore this usage at 2810 Old Main Hill, Logan, UT 84322-2810 (email: that place (p. 242). Some of this can be for- graysono(coe.usu.edu). given if it makes a section more understand- 19 20 J. GRAYSON OSBORNE able. However, to approach RFT with a rea- witness their subtitle-there are interesting par- sonably good knowledge of behavior analy- allels that they cannot avoid. First, they pro- sis, in general, and some knowledge of RFT, pose a theory which is less than formal, much in particular, is, still, to encounter occasions in as most of behavior analytic theory, that the the book when references to support assertions members of the Society for the Quantitative are desired, ifonly so that someone from a dif- Analysis of Behavior (SQAB) have not at- ferent perspective would not consider them tended to, is less than formal. Second, essen- empty. The same is true for more experimental tially, they propose an extension of what is detail about procedures that have been forgot- known in behavior analysis at this time, much ten or are vaguely recalled by the reader but as Skinnerproposed VB as an extension ofwhat with which someone from a different perspec- was known in behavior analysis in his time. tive might not be familiar at all. There may be Third, as RFT is an extension of behavior no winners here. analysis, it honors the philosophical founda- If you are reading this review, the book is tions thereof-as did VB. Among other descrip- probably written for you as well as me, that is, tors, behavior analysis is contextualistic, rela- persons sufficiently experienced with behav- tional (non-mechanistic), non-reductionistic, ior analysis to understand that, to the argu- post hoc, and inductive. Ditto RFT. Fourth, VB ments, the omitted references and, to the ex- provided an alternative way of thinking about periments, the omitted details, are basically a larger area of human behavior than that pro- unimportant. But how many are there of us? vided either within the broader extent of ex- The answer is: fewer and fewer.2 The way that perimental psychology at the time or outside the book is written suggests it is pointed to- psychology. Ditto RFT, and the proposed RFT ward a small, uniformly and familiarly trained, alternative to VB is one that will be considered audience (i.e., we happy few). and researched ultimately within the context If the book is noticed by those outside be- ofbehavior analysis, as was-to the extent that havior analysis will they respond in the absence it has been-VB. Fifth, as with VB, in RFT of an understanding ofwhat behavior analysis much ofthe authors' exposition is conjectural- is, as Chomsky responded to Skinner? Will the -as it must be-in the absence of supportive reasons be the same, that is, the failure of the research. RFT, too, is an extension from what authors to make the reader familiar enough with is known to that which isn't. Sixth, dealing as behavior analysis that the new constructs of they do with human language and its appurte- the book logically follow from what is currently nances necessitates almost complete overlap known and what has been proposed in the with the subject matter of VB, except, as men- past?3 This may be but one (unfortunate) char- tioned above, that applications to education, acteristic RFT shares with VB, about which, social processes, psychotherapy, and religion perhaps, little could have been done and about are primarily covered not in VB but elsewhere which there is more to say below. (e.g., Skinner, 1953; 1968), while most ofthese subjects are covered in RFT. Seventh, the tone Following in the Footsteps: Parallels with of the authors of RFT is assured. They exude Verbal Behavior and Skinner the same zeal that Skinner-who, to his critics, also sounded assured-did. The authors of As much as the authors of RFT appear to RFT feel they are theoretically correct, occa- want to separate themselves from Skinner- sional caveats to the contrary notwithstanding. Their early data suggest the validity ofthe ex- ' Here I speak not of the growth of the Associa- tensions they propose-as the early data in be- tion for Behavior Analysis but of the seemingly havior analysis did for Skinner. The authors of dwindling number of persons being trained in the RFT exhibit the same confidence that a young experimental analysis ofbehavior as may be inferred generation of behavior modifiers did some 40 from the lack of subscription growth ofthe Journal ofthe Experimental Analysis ofBehavior years ago. Since many of us were historically 3I grant that providing sufficient background evi- guilty of this same confidence, we should be dence ofthe vitality and validity ofbehavior analy- understanding. Finally, the authors of RFT at- sis may be nearly impossible to accomplish in a tempt from a few basic principles to explain single book that does not have that goal as its prin- much about complex human behavior. This is cipal focus. squarely in-not separate from-the broad be- BEYOND SKINNER? 21 havior analytic tradition (e.g., Staats & Staats, porally post-Skinner, this retention renders 1963; Staats, 1968). As Sidman (1994) has separation from Skinnerian ideas-which are written in reference to RFT, "[There is] ... much inextricably bound together with behavior that is congenial in this attempt to formulate a analysis' foundations-moot. While not spe- general theory ... (p. 556)." cifically rejected in RFT, and presumably, until shown otherwise, these principles are also What Prominent Ideas Are Put Aside? ofcontinuing importance to the explanation of human behavior. Let us divide Skinner's ideas grossly into two Skinner's ideas that became prominent did categories: those generated by and from the so outside the field ofpsychology and had most findings ofbehavior analysis and its attendant to do with the core idea that simple extension philosophy and those that comprise behavior to human behavior ofthe aforementioned foun- analysis itself. An incomplete list of Skinner's dational principles was possible. The ideas and ideas that comprise behavior analysis itself- their underlying philosophy (radical behavior- there is neither an attempt nor a need to be ex- ism) became prominent because of their im- haustive here-could begin arbitrarily by para- plications with respect to human nature in phrasing some of the chapter titles of Behav- Western society and culture. Skinner formally ior of Organisms (Skinner,1938): (operant) put some ofthose ideas on human cultural be- conditioning and extinction; reinforcement; havior in Science andHuman Behavior (Skin- periodic reconditioning, i.e., schedules of re- ner, 1953)-behavior in groups, education, inforcement; the discrimination of a stimulus, economics, and religion that are pretty straight- i.e., the development of stimulus control; the forward examples of these conceptual exten- differentiation of a response, i.e., shaping; and sions-as a context for VB, which he had al- drive, i.e., establishing operations. Among oth- ready largely written (Skinner, 1957). At the ers, these foundational constructs of behavior level they were presented, the cultural exten- analysis have been fleshed, tweaked, honed, sions were simple. In fact, it was their simplic- and modified over the years (e.g., establishing ity that was a focus of attention of the critics. operation for drive) by several generations of There is no question these extended ideas were behavior analytic researchers and theorists. and are prominent, if only in the reaction they Except in the details and the size ofthe picture provoked when the non-psychological world they paint-micro or macro-there is little that became aware of them and their implications remains controversial about them except for human nature. Take for example, the furor among those same researchers and theorists. that followed the publication ofBeyond Free- By themselves, our perspectives on these ideas, dom and Dignity (Skinner, 1971), Skinner's most ofwhich are also foundational to experi- treatise on agency (e.g., Wheeler, 1973). Ar- mental psychology in general, are not preva- guably, Beyond Freedom and Dignity was lent. Witness the fact that we can't even make Skinner's most controversial book outside of ourselves, as behavior analysts, effectively psychology. The current generation has prob- heard on the rewards versus reinforcement is- ably never heard of it. Indeed it would be sur- sue. Indeed, a book to promote this outcome prising ifcurrent behavior analyst graduate stu- has been written (Cameron & Pierce, 2002), dents read it. It is certainly not on one proposed, yet there are no guarantees it will be read and essential reading list (Saville, Beal, & Buskist, acted upon by the psychology community be- 2002). Again, the authors ofRFT suggest put- yond behavior analysis. Presumably, there is ting aside neither the application ofthese foun- no one in behavior analysis interested in put- dational ideas to human behavior nor any of ting these foundational ideas aside. They have the philosophical tenets that comprise radical carried us far and they will carry us farther. behaviorism. Thus, in terms ofthe foundational In fairness, the authors ofRFT don't appear principles ofbehavior analysis, their extended to suggest putting any of these foundational application to human behavior, and the tenets ideas aside. Ifthat is the case, how much sepa- of radical behaviorism, the authors ofRFT re- ration is there from Skinner? Insofar as it is main squarely Skinnerian. possible to infer from RFT, it would appear that Are the ideas in VB prominent? By promi- the authors retain all the basic principles of nent ideas to be put aside, the authors ofRFT behavior analysis. While their writings are tem- mean only those of Skinner that pertain to hu- 22 J. GRAYSON OSBORNE man language and cognition. First, are qua researcher). To whom, then, are the ideas Skinner's ideas in VB actually prominent? Have in VB prominent? If you are reading this, it is they, for example, led to research? About 20 probably, but not necessarily, you. It certainly years ago, "Not much," was the answer to the is me, but it is me only in the sense that noth- latter (Knapp, 1980; McPherson, Bonem, ing has come along to supplant the ideas in VB. Green, & Osborne, 1984). The authors ofRFT Hayes et al. would like to change that and they currently assert, not much, also. However, it could have done so without setting up VB and was additionally pointed out by MacPherson its author first in order to then knock them et al., at that time, that VB was a highly cited down. book. Unfortunately, most ofthe citations were What, then, is actuallyput aside in VB? The by its critics, most ofwhom were not behavior actual putting aside is neither methodological, analysts and most ofwhom were not even psy- nor philosophical, nor analytic. Indeed, in RFT, chologists (McPherson et al., 1984). VB was behavior analysis per se remains untouched in notorious outside behavior analysis, but its large measure. What is put aside is Skinner's ideas were then and are now anything but definition ofverbal behavior. In one sense this prominent. Speculatively, if we gave a quiz is a rather fundamental putting aside: it rel- today to the members of the International As- egates most of what has been examined in be- sociation ofBehaviorAnalysis that asked those havior analysis, both experimental and applied, members to identify (not define, mind you) to the realm of non-verbal behavior, where Skinner's six basic verbal operants described most of us would agree that non-human ani- in VB more than halfwould not be able to pro- mal experimental analysis already was anyway. vide a complete list. (Don't peek; take the chal- The exceptions-large ones-are in human lenge before looking at the footnote.)4 Indeed, experimental analysis (other than those deal- in how many graduate behavior analysis pro- ing with relational responding of a particular grams is VB even read, notwithstanding its high kind) and in applied behavior analysis. placement on the essential reading list (Saville But what does this change of definition re- et al., 2002)? Whither prominence ofits ideas? ally change? As with the progress of any sci- Second, does the prominence ofthe ideas to ence, the data already validly gathered and which the authors refer lie with the critics of analyzed do not all ofa sudden become invalid. VB? I have not looked at what has happened to They are, however, subject to reinterpretation, VB since 1984, the publication date of our but that process occurs on a continuing basis graduate students' brief foray into citation as a part of normal science, no matter the ad- analysis, but it is hard to believe that, having vent of new theory. Further, it is not clear that pretty much uniformly rejected VB and its ideas the extant technology of non-human animal many years ago, the critics would consider experimental analysis or applied behavior these rejected ideas prominent now. These crit- analysis will change with RFT/RFT. It will still ics would also do considerably worse on the be necessary to construct and apply differen- six-verbal-operant quiz than our own col- tial reinforcement contingencies to get one leagues and students most of whom have at class ofbehavior to increase and another class least heard of mands and tacts, if not echoics, of behavior to decrease, regardless ofwhether textuals, autoclitics, and intraverbals. Even one redefines the behavior being changed as Chomsky's review wasn't directed at the con- nonverbal or not. In other words, no paradigm tents of VB as much as it was directed at what shift lurks in RFT. VB stood for-an empirical approach to language Why redefine verbal behavior? The redefi- development with its implied behaviorism un- nition of verbal behavior is a necessary first derlying (he erroneously assumed) an S-R psy- step according to the authors of RFT because chology. Skinner's definition of verbal behavior is es- All of this is sad, since it seems as though sentially all encompassing, and, as such, may the authors of RFT have erected a strawman account in some measure for the lack of (one of several, e.g., Willard Day, philosopher progress in the study of verbal behavior. The Skinnerian definition of verbal behavior is so broad that all non-human animals are said to 4Mands, echoics, textuals, intraverbals, tacts, and be emitting verbal behavior ifthey behave with autoclitics. respect to contingencies arranged by humans BEYOND SKINNER? 23

(Skinner, 1957; p. 108). A literal interpretation behavior analytic cloth but results in a some- of Skinner's definition makes most of the ex- what different weave. perimental literature ofbehavior analysis to this All of the foregoing is not to say that new point the study of verbal behavior, a position principles ofbehavior are not presented in RFT that would not be defended by any contempo- on the human side ofthis divide. They are. But rary behavior analyst. Acceptance of a redefi- they, too, are a result of behavior analytic re- nition of verbal behavior further implies the search all-be-they ofthe last, temporally, post- necessary re-interpretation of the data of hu- Skinner years. man experimental analysis, which is not much of either a putting aside or a step since that What Is Changed by What Is Put Aside? endeavor has been going on since Harold Weiner showed that nurses didn't behave ex- The verbal operant and relationalframe as actly as pigeons and rats did on schedules of equivalent. Before reading RFT, I believed that reinforcement. The redefinition of verbal be- I understood what constituted a relational havior in RFTmakes verbal behavior relational frame. Reading the book changed that, so the behavior (specifically, arbitrarily applicable, book was very instructional. Prior to reading relational behavior) and that act has some large it, a relational frame appeared to be the con- implications. text within which an operant occurred. That It's the species, stupid The attempted sepa- context, broadly construed, was set by the fifth ration of RFT from Skinner is a species thing. term in a five-term contingency (Sidman, The redefinition of verbal behavior in RFT 1986). The frame was the contextual stimulus makes such behavior uniquely human-for the that implied, "In my presence, responses that time being. On the other hand, as discussed relate stimuli in a particular way (e.g., greater above, while not formally stated in RFT, all of than, equivalence, etc.) will be reinforced."5 I the known principles ofbehavior appear to still learned from RFT that a relational frame as apply to human and non-human animals with noun (p. 34)-my incomplete understanding respect to nonverbal behavior and, maybe with of it-is no different than relational frame as respect to human verbal behavior, too. The lat- verb. As verb? As action? Yes, and ter appears to depend on what one concludes it definitionally as operant. Indeed, the only kind is that verbal humans do. According to RFT, of operant that sentient, normal, verbal human verbal behavior, qua arbitrarily applicable, re- beings emit. What the authors ofRFTpropose lational responding, is all that humans do. is that humans, when they behave, are relating It is at this level, that of continuity or dis- stimuli and that is all (!) they do. Humans are continuity between Homo sapiens and all other predisposed to acquire this ability to relate animals, that RFT wishes to divide from Skin- stimuli through experience from birth with lan- ner, and this is potentially a huge divide. On guage. No language, no relating. Behavior ana- one half of the divide the standard definition lysts have probably used the phrase relational ofbehavior is good for pigeons, rats, and sun- responding to denote the act of relating but dry other species (but not necessarily, on oc- didn't realize at the same time that they were casion, for sea lions and an African grey par- speaking of relationalframing. rot), while on the other half a modified defini- This is a huge conceptual leap, but it fol- tion ofbehavior is good only (at this time) for lows generally from what is known about the Homo sapiens. field's failed (or controversial) attempts to pro- But, to reiterate, on both sides ofthis divide duce relational responding at least where the foundational principles of behavior analy- equivalence is the relation and nonhuman ani- sis are unmodified; there is only a change in a mals or pre-verbal and nonverbal retarded chil- definition. This modified definition of verbal dren are the subjects. It is a bold induction from behavior owes much to behavior analysis in- extant data, including the exceptional sched- cluding the context(s) that produced it and the ule ofreinforcement performances ofhumans. training of most who redefined it. Yet, more than occasionally it feels as though RFTs pur- I In fact, some of the authors themselves are still veyors think that the definition as modified and using phrases such as "... the response participate(s) its ramifications have been spun from whole in a relational frame ... (Barnes-Holmes, Barnes- cloth. Arguably, the redefinition is spun from Holmes, & Cullinan, 2000; p. 76)." 24 J. GRAYSON OSBORNE

While it is a bold induction for our field, it has additional inference on the part ofexperiment- already been made by others in related fields, ers seems to be required over that of non-rela- e.g., when children move out of the sensory- tional responding qua operant. An observer can motor stage (Piaget & Inhelder, 1964) and see or have equipment record the (relational) when they evidence a cognitive processing shift response of a human just as s/he can observe (Nelson, 1977). In this sense, with RFT, maybe or have equipment record (non-relational) le- behavior analysis is just catching up with de- ver pressing by the rat. With evidence ofa suf- velopmental and cognitive psychology. ficient number of instances in either case it is However, this induction may reach too far. inferred that the individual responses comprise It seems to ignore that verbal humans, under an operant related to a particular consequence. circumstances where rules appear to be non- But with humans the authors of RFT propose functional, behave under the actual conse- that each response relates at least two stimuli quences ofan actual schedule ofreinforcement (more when the construct, stimulus network, (e.g., Subject H in Mathews, Catania, & is included). That is, the operant is inferred but, Shimoff, 1985). Consider also instances where in addition, the stimuli to which the human a verbal human is conditioned without aware- operant relates must also be inferred because ness (e.g., Rosenthal & Baer, 1969; 1970). The the actual relating ofthe stimuli is no more di- authors ofRFT are aware ofthe need to exam- rectly observable than the operant is. Inferences ine behavior that is and isn't affected by self- relating the stimuli to the human operant also talk and self-rules and propose methods for its result post hoc from the observation of mul- determination, but suggest that such study will tiple instances, all correlated with a particular uncover "... self-rules that are not expressed consequence. Conceptually, this seems little (p. 136)," not that it will uncover cases in which different than inferring an operant without its human behavior might occur in the absence of purportedly related stimuli. Both are difficult rules. A question: If human behavior absent philosophically and theoretically, but, perhaps, rules can be found, is it no longer relational less so empirically. Moreover, inferring con- behavior? That is, from the perspective ofRFT trolling stimuli has been the business of be- can and do sentient, verbally behaving human havior analysis since the time of Watson. beings respond non relationally? Skinner Presumably, then, until it is otherwise dis- seemed to imply this distinction in his defini- covered, everything that is currently known tions ofcontingency-shaped and rule-governed about the operant for non-human animals also behavior in which he proposed that, culturally applies to this modified version ofthe operant and historically, contingency-shaped behavior for humans. For instance, we need now to think preceded rule-governed behavior (Skinner, that reinforcement increases the likelihood of 1969), but the distinction may not be a part of a particular kind of relational responding in RFT. If it is not, this is not a terminal problem. humans, and extinction presumably makes the Continued exploration of when (and if) non- particular kind ofrelational responding that was relational reinforcement contingencies apply to being reinforced less likely to occur. Also, func- human behavior can now be re-contextualized tional stimuli make a certain type ofrelational in terms of RFT and would seem worth rein- responding more or less likely to occur (e.g., vigorated exploration. Osborne & Koppel, 2001), and establishing Changes in the definition ofthe verbal oper- operations may enable some (relational) operant. The change in the definition ofverbal be- ants and disable others. This seems a lot less havior, in turn, necessitates a change in the difficult to incorporate in one's thinking than, definition ofthe verbal operant. While the defi- for example, the shifts in thinking that were nitional nature ofthe operant in RFTis changed necessary many years ago to understand the for humans, none of its other (Skinnerian, be- autoshaping process and how it was superim- havioral analytic) characteristics appear to be. posed on and occasionally conflicted with dis- The operant is still an empirical construct, an criminative operant conditioning. Behavior abstraction once removed from observable analysis incorporated the autoshaping phenom- data. It is still inferred post hoc as a class of enon then and eventually, ifthe data bear fruit, responses all ofwhich correlate with the same should incorporate this new definition of the consequence. operant for humans. Indeed, most behavior For relational responding qua operant, an analysts who are working with relational op- BEYOND SKINNER? 25

erants probably wouldn't give the foregoing behaviorists and this is but another example of esoteric matters much thought since they don't their following in his footsteps? seem a part of the normal proceedings in be- The combination of the two properties-re- havior analysis. Indeed, it does not seem likely sponse and stimulus-in the relational frame that general experimental practices in behav- (verb) and the relational frame (noun) brings ior analysis will change as a result of the re- the same kind of confusion with it. It seems definition of the verbal operant for humans. too convenient. It feels simultaneous, with both Responses as stimuli. What seems not context (noun) and behavior (verb) occurring worked out very well yet in RFT and what has at the same time, rather than linearly as in our always been troublesome .for behavior analy- more usual logic. One wants to ask, "When is sis in general is the notion that behavior also it context?" "When is it behavior?" "When we has stimulus properties so that responses can say it is?" For many behavior analysts, in this also be stimuli. This is difficult to talk about decision between response and stimulus, the because it seems when we do as though we are role of the organism will loom large, which it cake having and cake eating concurrently. usually does when we lack the data to show When is a response a stimulus? When we say that it doesn't. Typically where we lack the data, it is. When isn't it a stimulus? You know the we can and do resort to reinforcement history answer. It is fairly easy to conceptualize re- as context to conceptually resolve this diffi- sponses as stimuli when the behavior of an- culty. RFT seems little different in this. other organism is what one is responding to Skinner and relational responding. There is and Skinner did so in VB in his definition of also the implication in RFTthat Skinner didn't verbal behavior as the consequential result of really understand relational responding. This the interaction between speaker and listener. It owes to the gift ofhindsight. The phrase, rela- is less easy when it is your own behavior that, tional responding, now rolls off our tongues it is said, you are (almost always, conceptu- so easily it almost seems as though it has been ally) responding to. In these cases the inferred a part of our lexicon and our thinking forever. stimuli and responses always seem to be co- But it hasn't. Before it became comfortable to vert and, thus, can seem to be convenient fic- use this phrase, we had difficulty saying what tions. They assist our behavior analytic expla- was meant when we attempted to talk of rela- nations of perception, and remembering, and tional responding. We worried about what the awareness, and consciousness (Skinner, 1974). relations were we were referring to (Sidman, In fact, they help wherever a verbal report is 1994). required to conceptually complete an act in Consider the longer view. As reported in which the Rs follow the Ss in an orderly fash- RFT, Skinner was aware ofrelational respond- ion. In other words we most often propose Ss ing in the sense ofthe operant being evidenced to follow Rs and Rs to precede subsequent Ss by responses that were controlled by relations to maintain the integrity of an S-R logic that, between stimuli (p. 24). However, these rela- at one level-the mechanistic-Skinner de- tions were formal in the sense ofphysics (e.g., cried, for example, for its implications of man larger than, or intermediate, in terms of size). as automaton, but at another level-the con- And both RFTtheorists and Skinner accept that ceptual-he embraced, for examples, the S-R animals, as well as humans, are capable ofthis chain and the discrimination of a discrimina- kind of relational responding. For RFT theo- tion (Skinner, 1974). Perhaps we can continue rists, this type of relational responding is non- to do this with impunity because the external verbal as it would be for most behavior ana- validity ofthese two constructs, where the parts lysts, but possibly not for Skinner given the of a chain and a double discrimination are liberal interpretation of his definition of ver- clearly overt, permits us to generalize the pro- bal behavior. In retrospect this is the leading cess of creating intervening Ss and Rs where edge of the change in the definition of the op- the same parts are covert (see also Zuriff, 1985; erant. But since it includes non-human animals, pp. 78-80). The authors of RFT lean hard on there must be another change by RFT implicit this logic to construct their extensions of RFT in the redefinition of the verbal operant that to psychotherapy and to the construct of self. keeps the animal-human divide open. But what else would we have them do since What the RFT redefined verbal operant re- the precedent exists with Skinner and other ally is. The final step in the redefinition of the 26 J. GRAYSON OSBORNE

verbal operant is to understand that stimuli that (verbal). Major portions of VB itself, therefore, are not formally related by physics can still be would seem to be not put aside very far. Fur- formally related in terms oftheir control of an ther, Barnes-Holmes et al.'s analysis of operant. In essence, stimuli that aren't dimen- Skinner's constructs in VB implies an RFT an- sionally alike can function similarly, that is, swer for a question raised above as to whether they can constitute a functional stimulus class. there may be some human behavior that is non- The same response may be produced by the relational. written phrase, Take out the garbage, the spo- ken phrase, "Take out the garbage," or a video The Other Constructs ofRFT clip ofsomeone taking out the garbage. Such a stimulus class fits well within the Skinnerian In the present review there is neither the idea of the generic stimulus class (Skinner, space nor the inclination to deal specifically 1935). The difference for RFT lies in the fact with the other constructs of RFT such as de- that verbal responding relates these non-for- rived relational responding, the entailments, mally related stimuli. When it does, it is said and function transformation. All have empiri- in RFT-speak to be "arbitrarily applicable" re- cal roots in behavior analysis. Some have been sponding. It is this "arbitrary applicable" re- dealt with elsewhere (Sidman, 1994). They are sponding that is uniquely human, and, by defi- post-Skinnerian in the sense that they were nition, verbal-the second step in the redefi- being researched when Skinner was no longer nition process. It is in this sense, and only in active in the laboratory. As far as this reviewer this sense, thatRFTis post-Skinnerian in a way knows, Skinner had little to say about them. that is more than temporally post. Neverthe- RFT's conceptual expansion ofthese constructs less, in defining verbal behavior this way, RFT from their narrowly focused beginnings in continues to utilize the concept ofthe stimulus equivalence class research is very useful. The class and the response class and both of those theoretical development of these constructs constructs, at their origin, are Skinnerian.6 In within RFT exemplifies the orderly develop- this last sense, even the redefinition ofthe ver- ment ofthe experimental analysis ofbehavior, bal operant by RFT is clearly a Skinnerian ex- more specifically the orderly development of tension, and, clearly, Skinner lives on in RFT human experimental analysis, rather than any and RFT. paradigmatic shift. The implied scope of the RFT-proposed changes is assuredly not that of VB Constructs Retained Newton to Einstein. Coming full circle, some of RF7s authors seem to suggest that VB's basic constructs can SUMMARY be retained too, but only as they constitute nonverbal operants (Barnes-Holmes et al., 2000). In essence, in this reviewer's opinion, there's While this certainly isn't a synthesis in any a new VB on the block, one largely grown out sense ofthat word ofthe constructs of VB with of the old. Little of behavior analysis and of those ofRFT, it suggests theoretically that RFT Skinner's contributions to behavior analysis is is broadly inclusive even of the constructs of changed thereby. Further, while much ofRFT/ VB at one level (non-verbal) if not at another RFT is post-Skinnerian temporally, little of it is, in any sense, beyond Skinner. To the contrary, RFT/RFT flows from the inextricable 6I desperately want to stay away from the brou- contexts of Skinner and behavior analysis. haha over whether equivalence can be defined in- There is simply no escaping that. dependently of the notion of stimulus class and the related and seemingly unresolved issues ofwhether stimulus class and stimulus relations can co-exist, REFERENCES since I am not sure that I can contribute much to those discussions at this time. All I am arguing here is that RFT includes the foundational behavior ana- Barnes-Holmes, D., Barnes-Holmes, Y., & lytic construct, stimulus class. It does (see Barnes, Cullinan, V. (2000). Relational frame theory Healy, & Hayes, 2000; p. 162) and, in so doing, is and Skinner's Verbal Behavior: A possible both Skinnerian and behavior analytic. synthesis. The BehaviorAnalyst, 23, 69-84. BEYOND SKINNER? 27

Barnes, D., Healy, O., & Hayes, S. C. (2000). perimenter by a more aware subject: the Relational frame theory and the relational double agent effect. Psychological Review, evaluation procedure: Approaching human 76, 425-432. language as derived relational responding. Rosenfeld, H. M., & Baer, D. M. (1970). Un- In J. C. Leslie & D. Blackman (Eds.), Ex- biased and unnoticed verbal conditioning: perimental and applied analysis ofhuman The double agent robot procedure. Journal behavior (pp. 149-180). Reno, NV: Con- of the Experimental Analysis ofBehavior, text Press. 14, 99-105. Cameron, J., & Pierce, W. D. (2002). Rewards Saville, B. K., Beal, S. A., & Buskist, W. and intrinsic motivation. Westport CT: (2002). Essential readings for graduate stu- Bergin & Garvey. dents in behavior analysis: A survey of the Hayes, S. C., Barnes-Holmes, D., & Roche, B. JEAB and JABA Boards ofEditors. The Be- (2001). Relational frame theory: A post- havior Analyst, 25, 29-35. Skinnerian account ofhuman language and Sidman, M. (1986). Functional analysis of cognition. New York: Kluwer, Academic/ emergent verbal classes. In T. Thompson & Plenum Publishers. M. D. Zeiler (Eds.), Analysis and integra- Hayes, S. C., Strosahl, K. D., & Wilson, K. G. tion of behavioral units. Hillsdale, NJ: (1999). Acceptance and commitment Earlbaum. therapy. New York: Guilford. Sidman, M. (1994). Equivalence relations and Knapp, T. (1980). Beyond Verbal Behavior. behavior: A research story. Boston: Authors' Behaviorism, 8, 187-194. Cooperative. Mathews, B. A., Catania, A. C., & Shimoff, E. Skinner, B. F. (1935). The generic nature of (1985). Effects of uninstructed verbal be- the concepts ofstimulus and response. Jour- havior on nonverbal responding: contin- nal ofGeneral Psychology, 12, 40-65. gency descriptions versus performance de- Skinner, B. F. (1938). Behavior oforganisms. scriptions. Journal of the Experimental New York: Appleton-Century-Crofts. Analysis ofBehavior, 43, 155-164. Skinner, B. F. (1953). Science and human be- McPherson, A., Bonem, M., Green, G., & havior. Toronto: Collier-MacMillan. Osborne, J. G. (1984). A citation analysis of Skinner, B. F. (1957). Verbal behavior New the impact on research of Skinner's Verbal York: Appleton-Century-Crofis. Behavior The Behavior Analyst, 7, 157- Skinner, B. F. (1968). The technology ofteach- 167. ing. New York: Meredith. Nelson, K. (1977). The syntagmatic-para- Skinner, B. F. (1969). Contingencies of rein- digmatic shift revisited: A review of re- forcement: A theoretical analysis. New search and theory. Psychological Bul- York: Appleton-Century-Crofts. letin, 84, 93-116. Skinner, B. F. (1971). Beyondfreedom anddig- Osborne, J. G., & Koppel, L. (2001). Acquisi- nity. New York: Knopf. tion, generalization, and contextual control Skinner, B. F. (1974). About behaviorism. New of taxonomic and thematic relational re- York: Random House. sponding. The Psychological Record, 51, Staats, A. W. (1968). Learning, language, and 185-205. cognition. New York: Holt, Rinehart, & Piaget, J. & Inhelder, B. (1964). The early Winston. growth of logic in the child. London: Staats, A. W., & Staats, C. K. (1963). Complex Rutledge & Kegan Paul. human behavior: A systematic extension of Roche, B., Barnes-Holmes, Y, Barnes-Holmes, learning principles. New York: Holt, D., Stewart, I., & O'Hora, D. (2002). Rela- Rinehart, & Winston. tional frame theory: A new paradigm for the Wheeler, H. (Ed.) (1973). Beyond thepunitive analysis of social behavior. The Behavior society. San Fransisco: Freeman. Analyst, 25, 75-91. Zurriff, G. E. (1985). Behaviorism: A concep- Rosenfeld, H. M., & Baer, D. M. (1969). Un- tual reconstruction. New York: Columbia noticed verbal conditioning ofan aware ex- University Press.