Emergent Behaviorism

Emergent Behaviorism Author(s): Peter R. Killeen Source: Behaviorism, Vol. 12, No. 2 (Fall, 1984), pp. 25-40 Published by: Cambridge Center for Behavioral Studies (CCBS) Stable URL: http://www.jstor.org/stable/27759042 . Accessed: 26/06/2013 20:51 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Cambridge Center for Behavioral Studies (CCBS) is collaborating with JSTOR to digitize, preserve and extend access to Behaviorism. http://www.jstor.org This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions Behaviorism, Fall 1984, Vol. 12, Number 2 EMERGENT BEHAVIORISM Peter R. Killeen Arizona State University I conclude that his only valid objections In this article I examine Skinner's objections to mentalism. ? that mentalism might afford the "specious explanations" that are incomplete, circular, or explanations faulty in other ways. Unfortunately, the mere adoption of behavioristic terminology does not solve that problem. It the nature of "private events," while providing no protection from specious explanations. I argue that camouflages covert states and events are causally effective, and may be sufficiently different in their nature to deserve a name ABSTRACT: concern not force a dualistic metaphysics: Such a distinction can be call such events"mental"does by an "emergent behaviorism." Emergent behaviorism would make explicit use of theories. It easily assimilated would be inductive and pragmatic, and would evaluate hypothetical constructs in terms of their utility in clarifying and solving the outstanding problems of the discipline. other than "behavior."To Skinner vacillated in his view on the nature and causal efficacy of events that occur inside the skin of an organism. In his reply to Blanshard he stated that "ideas, motives, and feelings have no part in determining conduct, and therefore no part in explaining it" (Blanshard & Skinner, 1967, p. 325), and several years later "the mental laws of physiological of William James, the mental psychologists like Wundt, the stream of consciousness apparatus of Sigmund Freud have no useful place in the understanding of human behavior" (Skinner, 1972, p. 19). Why then do thoughts and feelings seem to cause behavior? "They usually occur in just the place that would be occupied by a cause.... For example, we often feel a state of deprivation or emotion before we act in an appropriate way. If we say something to ourselves before saying italoud, what we say aloud seems the expression of an inner thought" (1972, p. 19). For some reason Skinner never considered that through classical conditioning, such precursors to action might come to elicit it. Not everywhere does Skinner deny the efficacy of private events: "It is particularly helpful to describe behavior which fails to satisfy contingencies, as inT letgo too soon' or 'I struck too hard.' Even fragmentary descriptions of contingencies speed the acquisition of effective terminal behavior, help to maintain the behavior over a period of time, and reinstate itwhen forgotten" (1957, p. 445). How can he admit that "Verbal behavior has practical effects upon the speaker as a listener"? (1957, p. 440), yet deny the causal efficacy of ideas, thoughts, and feelings? Part of the reason is that like Bertrand Russell, whom he admired, Skinner preferred not to use the term "cause" (all mentions of it in the index of and Human Behavior refer to perjorative discussions) and he found it especially seem to have given problematic in referring to intermediate links of causal chains. "Effects" a reason must that he be that science of behavior less trouble. Another part of the recognized Science I thank G. Graham, R. J. Herrnstein, and P. Prins for comments on this paper. A number of the issues raised here have also been discussed by Sober (1983). This article will be published in B. F. Skinner: Consensus and Controversy, 1986, edited by Drs. Sohan and Celia Modgil, under the Falmer International Master-Minds Challenged series, Falmer Press, Sussex, England. The book will contain 24 chapters in a paired-debate from Professors B. F. Skinner, Willard Day, and Daniel Denett. Barcombe format, among 25 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions them contributions KILLEEN is prior to a science of logic, and he respected the laws of conditioning more than the "laws" of valid inferences. He wished to bring his verbal behavior under the control of his data, even if that required some inconsistencies given our current state of knowledge. For what we witness is not a slow shift in position as Skinner's perspective changed. There is no monotonie relation between the degree of causal efficacy given covert events and the year in which that opinion was published. Many of Skinner's objections to "Mental Way Stations" are to ad hoc and untestable "explanations" loosely inferred from causal introspection, not to the possibility of "practical effects" deriving from covert behavior. These and other objections are clearly stated in his article "Behaviorism at 50" (1963), where he notes that Mental Way Stations (MWS) are often accepted as terminal data, and thereby abort the search for environmental causes of the MWS, or, going in the other direction, the existence of itsputative consequences. Even when the MWS is anchored to both prior stimuli and consequent responses, "various things at the way station which alter the relation between terminal events.... Mental states happen alter one another.... Dissonant cognitions...will not be reflected in behavior if the subject can, 'persuade himself that one condition was actually different...."(It is leftambiguous as to whether he believes that such interactions occur, or whether they are precipitated out as one of the unclean and unnecessary byproducts of a cognitive vocablulary). "The effects still provide a formidable stronghold for mentalistic theories designed to bridge the gap between dependent and independent variables" (p. 240). With what engines does he assault that stronghold? The methodological objections are: to the predilection for unfinished causal sequences, b) thatMWS "burdens a science of a) behavior with all the problems raised by the limitations and inaccuracies of self-descriptive repertoires", and c) "Perhaps the most serious objection.... Responses which seem to be describing intervening states alone may embrace behavioral effects" (p. 241). That is, inferences about MWS may be based in part on the observation of our subsequent overt behavior. As we shall see, these are weak objections. What rule will Skinner establish once the stronghold is breached? He notes that "These disturbances in simple causal linkages between environment and behavior can be formulated and studied experimentally as interactions among variables; but the possibility has not been fully exploited" (p. 240). He deals briefly with multiple causation in Science and Human Behavior (1953), and much more extensively throughout his monumental Verbal Behavior (1957). The product was a series of astute observations, but they have not been adopted by many students as an effective approach to these issues. In part this is because of the enormous difficulty of the subject; in part because in Verbal Behavior, Skinner "put the reader through a set of exercises for the express purpose of strengthening a particular verbal repertoire.... I have been trying to get the reader to behave verbally as I behave"(p. 455). He succeeded in that attempt for many of us; what he did not provide was an experimental analysis of complex cases such as verbal behavior. Itwas all talk, often brilliant talk, and he leftus talking, often somewhat less brilliantly. His citations are primarily to intellectuals and men of letters, and this is themodel he provides us, not that of an experimenting scientist. OBJECTIONS TO MENTALISM Random House defines behavior as "the aggregate of observable responses of an organism to internal and external stimuli." This was the domain of Skinner's early work, as reported in The Behavior of Organisms (1938). But Skinner was always eager to address the most sophisticated aspects of human behavior. To do that, he could either devise new analytical tools, or he could adont those of others, or he could assert that no new tools were 26 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions EMERGENT BEHAVIORISM needed. In attempting to extrapolate the laws of conditioning to humans, Skinner clearly took the last course. Itwas an optimistic gambit; the combination of a few simple principles can easily generate complex outcomes inany field of investigation. His principles were based on "movements of an organism or of its parts in a frame of reference provided by the organism itselfor by various external objects or fields of force"( 1938, p. 6). His independent variables were reinforcement contingencies defined in terms of motivational operations defined in terms of deprivation operations, terms of both physical parameters and the nature of the behavior Skinner's "generic," or functional, definition of stimuli and responses schedule parameters, and stimuli defined in that they controlled.1 provided a trapdoor to escape from the problems that would otherwise be posed by attention, elaboration, and other mental processes. Unfortunately, itwas a route that he seldom took, preferring to study behavior where physical definitions of stimuli and responses were usually good enough. This early history of reinforcement shaped Skinner's approach to the analysis of human behavior. Skinner recognized that "mental states alter one another," and repudiated them for that reason.2 But the omission of such states left him with an inadequate vocabulary, which he then expanded bymoving some stimuli inside the organism ("private stimuli"), and by treating all other aspects of mental states as responses. Seeing became behavior, and imagination became "seeing without the thing seen."3 But these are assertions, not demonstrated facts. They may serve as the axioms of a parsimonious behavioral system, and that is largely how Skinner used them. But they cannot then also be used as arguments against other systems, or against behavioral systems with augmented axioms, such as the assumption that covert events are sufficiently different from overt ones to deserve separate treatment as a separate category of events. The competition between such systems is in terms to those who of explanatory power, not in terms of whose axioms are least objectionable don't hold them. Let us, nonetheless, review Skinner's objections to the inclusion of mental events in his (and anyone else's) system. Keat (1972) summarizes them: For Skinner, mentalism is to be rejected because? (1) It lacks "explanatory power." (2) It involves the employment (3) It tends to invoke (4) It distracts of "theories." "homunculi" our attention (5) It involves a dualistic regarding man or "inner agents," as an "autonomous agent." from the study of behavior. ontology of the "mental" and the "physical." (p.55) To these we may add those mentioned above and not included here: The limitations of a self-descriptive repertoire, and the contamination of descriptions of internal states by discriminations based on overt behavior. What of these? Explanatory Power Skinner notes that after we have explained a response in terms of mental states or activities or feelings, we still need to explain themental state. But there is nothing wrong with that. Experimental analysis of one of the links in a causal chain should not necessarily be faulted because itdoes not include the previous ones; analysis must inevitably stop at some 27 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions KILLEEN point short of the Ultimate Cause. We might infer from a person's pale face, hand over stomach, and low moan that he is in pain, and thus seek aid for him. We do not attempt to move his hand, nor silence themoans, nor rouge the face: The inference of an internal state of distress ismore likely to be useful to him. Nor do we yet need to inferthat itwas something he ate, or that itwas the flu that is going around, or that he was punched. Such determinations will certainly help, but we can take immediate and effective action based on our inference alone. Are we more likely to be helpful ifwe know that he ate strange food, but deny that he is indistress? Similarly, there are many things thatmight make us angry, and knowledge of that internal state will be much more effective in explaining subsequent behavior than knowledge of the (often seemingly innocuous) stimuli that preceded it.We might get by without telling others of our needs, our goals, our ideas and ideals, and merely be viewed as a "private" person; were we to convey in their place our reinforcement histories and immediate stimulus context, unflavored by our subjective evaluations of them, we'd be condemned either as a tedious obscurant, or, if taken seriously, as an unreflecting pawn of our history. Theories Skinner has stated that "whether experimental psychologists like itor not, experimental psychology is properly and inevitably committed to the construction of a theory of behavior. A theory is essential to the scientific understanding of behavior as a subject matter" (1969, pp.vii-viii). What he objects to are particular types of theories: "any explanation of an observed fact which appeals to events taking place somewhere else, at some other level of observation, described in different terms, and measured, ifat all, in different dimensions" (p.vii). But what's wrong with such theories? Although we should not try to model our scientific discipline too closely on others, for our problems are different,we can evaluate this general statement by asking whether the condemned types of explanation are used in other fields. Any reasonable interpretation of Skinner's words shows that theyare not only used, but that they constitute some of the most powerful and elegant theories that we know! Newton's theory of universal gravitation explained the tides on the earth by events "taking place somewhere else"; about as elsewhere as one can get. Skinner should be the last to want to posit some hypothetical stuff that bridges the spatial gap in order to avoid action at a distance. It is not clear how to interpret "at some other level of observation," other than as observations that employ instruments to aid our senses. But thenwe must include the germ theory of disease, a theory greatly abetted by themicroscope. Should we give that up for fear of a behaviorist's censure, and hope that holistic medicine will be ready for us when we need it? in different terms" is often a strength of theories. Theories are a map, or "Described us by providing a coherent frame of reference within which to view the serve and metaphor, with the rules to develop specific models of it. Ifevents are not recondite, phenomena, along we seldom think of them in terms of theories. I don't seek theoretical explanation forwhy my dog barks at cats. But I do generate hypotheses about why he howls at railroad whistles. And those involve theories that use "different terms"?speculation about evolutionary constraints and wolves howling to gather for a hunt. In general, a failed description elicits another attempt, in other terms ("Let me put itthisway"). We all know thatmany successful theories involve descriptions in other terms: Chemists explain molar physical properties of substances in terms of internal states, such as the valence of atoms, polarization of molecules, and so on. 28 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions KILLEEN ifat all, in different dimensions". Differences in the dimensions used to "Measured, describe a phenomenon may indicate a basic and insurmountable incommensurability, or theymay indicate an arbitrary historical decision or a matter of style or convenience. We may record motion and shape in Cartesian or polar coordinates; we may quantify sound in terms of their energy over time, or in terms of their power spectrum. In this latter case, it is the insight provided by Fourier's Law that reveals the equivalence between the two representations. It is this realization of the unity of phenomena behind the apparently different dimensional frames that provides some of themost potent rewards for scientists and inspiration for theiraudience. What more beautiful evidence of this could one want than waves that symbol of 20th century physics, Einstein's equation relating energy to mass and the speed of light? "Perhaps," you say, "Skinner was merely trying to assert that things like emotions don't have dimensions that would support conversion to the dimensions of things like running speed." They don't currently, although some people "Get so mad they could spit," thus In a similar manner atoms sometimes emit providing latency and ballistic measurements. particles, whose latency and trajectory can tell us how "excited" theywere. Of course, one need not assume that excitation is nothing more than the propensity to emit an electron, nor that the level of excitation should be characterized only in terms of one type of cloud chamber tracing. A similar analysis holds for emotions and other types of inner causes. It may be true that good measures of emotional states and their strengths are not yet available. However, we should not elevate this current deficiency into a permanent dogma, or we shall never have them. Let us instead look to theory to suggest new types of dimensions and new ways of relating them. Horn uneuli Skinner objects tomentalism because it leads us to invoke homunculi. But itneedn't do that, nor is the doing of that necessarily bad. Part of Skinner's objection here is the argument, for he does not want us to view humans as unfinished-causal-sequence "autonomous agents" (Skinner, 1974). As suggested above, one can defer the analysis of in the causal chain without denying their existence. In fact, one can be a links prior or "self determinist and still believe that it is useful to treat humans as autonomous, governing" (Audi, 1976). The governors (e.g., traits and values) may have been aligned by nature and nurture, but the individual's interactions with his or her current environment may be treated more effectively by a theory of internal rules and set-points than by untested reference to histories of reinforcement. Not that such behavior is uncaused, but that more might be gotten out of focusing on the internal, self-reflective dialogue as cause than by focusing on bells, whistles, and cumulative records. In fact, histories of reinforcement are often reconstructed through verbal reports of individuals who are less than disinterested observers, and they thus suffer some of the same liabilities as introspective reports. "History of reinforcement" is a critical construct in Skinner's theory, but has received little critical analysis. In study of human behavior, it often serves as generic explanation, of the same order of "inner man." Skinner's problem with homunculi is that he never took then seriously enough. What he seems to mean by them are small creatures with all the capacities of larger ones, who live ex machina. Clearly, there would be a problem inside and make the decisions?homo packing them in. All themore a problem since theywould need their own homunculi. And they theirs. The implication is that an infinitenumber would be needed, thus forestalling any specification of themachinery thatmight actually get the job done, and thereby reducing this 29 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions EMERGENT BEHAVIORISM type of explanation to absurdity. Skinner is objecting to the hypostatization of a solution, in themanner ofMoli?re's doctor who explained the soporific properties of opium as being due to their virtus dormitiva: "The function of the innerman is to provide an explanation which will not be explained in turn" (Skinner, 1971, p. 14). If this is its only function, then it is certainly counterproductive. But not all infinite series are regresses, nor do all of them carry us infinitelyfar from the starting point. If enough tasks are accomplished by each homunuculus before itpasses its output to the next higher demon, then the job will get done with a finite number of levels; the series will converge, and itmay converge quickly. But why bother with such a hierarchy? Simon ( 1973) provides themost eloquent answer, and Dawkins (1976) discusses hierarchies ingreater detail. Just as computer programs based on subroutines are easier towrite, to troubleshoot, and tomodify, behavioral programs are more effective when arranged hierarchically (Dennett, 1978). Cognitive psychologists analyze the role of "scripts," "sch?mas," and "prototypical actions" in the control of behavior (Abelson, 1981); behaviorists speak of higher-order stimulus control, but seldom attempt to explain that in turn. The computer subroutine metaphor is useful, not only because such programming shares features with behavioral programming, but because of the way in which people characterize the actions at each level. As Dennett (1978) has observed, a single line of code is discussed mechanistically. So might be a single subroutine. But for sufficiently large systems, whether they control petroleum refineries or play chess, programmers revert to functional ("in order to") and intentional ("it is attempting to") discourse. (Of course, this could be folly, responsible in part for the high costs of gasoline and our inability to write a grand master chess program!) Similarly, for reflexive or automatized behavior, it is natural to speak mechanically, in terms of short chains of cause and effect. But why believe, against the everyday evidence afforded by the ordinary verbal behavior of even themost sophisticated radical behaviorists, that the emergent language of function and intention is not a superior mode of discourse concerning complex human behavior? Distraction from the Study of Behavior an organism isobserved Patently a behaviorist's complaint. But since behavior is "what an to be another organism may be doing many doing" (Skinner, 1938, p. 6), organism by be fit for scientific study. Ifwe as that don't behavior, yet may qualify interesting things broaden the definition by annexing private events, ithas to be demonstrated why "response than "image." If produced private stimulus" is a better label for the object of your mind's eye the answer is that we don't have a "copy" in our head, then we must ask clarification of the nature of "private stimuli"; either they carry themeaning of copy/image, or an important images, and our ability to inspect them?will be leftunaccounted for. phenomenon?our Behaviorism has helped us to focus on the environmental context of actions, and that has been all to the good. It isall too easy to inventunverifiable "mental"causes even when the behaviorism precipitating environmental events are obvious to any who would look. Radical should not scientists it is like excellent major But, something discipline, discipline. provides in. The world is bountiful with substantive issues, to which behaviorism can contribute It is by focusing on these issues that the important perspectives and methodologies. will be and improved. We should study behavior, but we should methodologies perspectives is a theoretical also study what goes on inside organisms. To call that "behavior" in most Skinner boxes; labelling it commitment; it certainly isn't what gets measured "behavior" is a least a metaphorical extension, ifnot a magical mand. Perhaps other terms 30 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions KILLEEN for itwould be better. "The decision as towhether to limit psychological theory to stimulus response relationships...or to admit non-behavioral theoretical termswould seem to depend (Zuriff, 1976, p. 51). only on the heuristic values of the two approaches" Dualism Not a simple issue. But neither is itone that is solved by claiming that "my toothache isas physical as my typewriter" (Skinner, 1972, p. 384). How can that be? Is not a toothache in in different terms, and that limbo of "some other level of observation, described measured...in different dimensions," along with theories and other embarrassments? Skinner's verbal behavior here appears to be more under the control of formal properties (alliteration) than physical properties. We may wish to avoid the long, usually dense, sometimes tedious, philosophical analyses of the mind/body issue by such a fiat. But the difference between toothaches and typewriters remains. Now we must invent some new dimension along which to differentiate them.Whatever we label that dimension, itwill have two ends (or at least directions), and thus support a new charge of dualism. " Mental" need not be equated with angels, uncaused causes, and other superstitions. It is a convenient rubric for many of the phenomena that led most of us into the field of psychology, and which remain fit subjects for study, despite our years of neglect. The act of faith that such phenomena are best treated as behavior threatens to overextend our primary the clock construct, and to jeopardize it in the eyes of critics even where it is secure?when strikes thirteen, the previous twelve become suspect. Natsoulas (1983) concluded that was not in a "Skinner has not got the problem of conscious content right. Moreover...he position to get this problem right because of the impoverished conceptual apparatus...given the philosophy of psychology to which he stands committed" (p. 19). Can we not countenance a variety of constructs as requisite to the variety of events in our purview? I suggest thatwe can, and thatwe can do so without loss of our soul, and even without the loss of our identity as behaviorists. Limitations of Self-Descriptive Repertoires, and Their Confounding by Overt Behavior objections may be treated together. The first is based on two accurate observations; We are not "wired" to be able to observe all of our covert processes. And, we rely on a verbal community to shape our verbal repertoire, but that has only restricted access to our internal events, and is thus able to afford only imperfect tutelage. For both reasons, we These often rely on observations of our overt in our behavior attempts to clarify our internal events or states. Thus, we might offer "I'm hungry" as an explanation for a second helping, although that utterance isbased more on the public stimulus of refilling the plate than on any internal stimulus. This is often a useful thing to do. Of course, not infrequently such post-hoc explanations are mere fabrications, social conventions that are difficult to disapprove, and which thus function to stifle further probes (or to facilitate further interactions; Goffman, 1971). They are "lines", and if theywork well on our audience, we may even be taken in by them ourselves (Nisbett & Wilson, 1977; Sabini & Silver, 1981). Of course, "there are good reasons why a speaker should turn his verbal behavior upon himself," and establishing stimulus control by internal stimuli, calibrated by the observation of public stimuli (and by such helpful labels as "I guess Iwas hungrier than I thought"), isone reason. Such statement should not be taken as absolute assertions of causality so much as 31 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions EMERGENT BEHAVIORISM hypotheses that may lead to self-knowledge, however uncertainly. They also may lead to frames of reference that exert continuing control over our subsequent overt behavior (Bern, 1972). Because our introspections can be unreliable does not mean that they, and the private events that they label, do not function as links in a causal chain. The private events may control overt behavior even when not introspected. Introspection may supply labels that amplify the control of private events?even when the labels are inaccurate, and even when they are incorrect, and thus block control by the stimuli on which they are based. In their review of attitude and attitude change, Cialdini, Petty and Cacioppo (1981) note that "no longer are researchers questioning ifattitudes predict behaviors, theyare investigating when attitudes predict behaviors.... have] found that attitudes had causal [Researchers predominance over behaviors" (p.366). In fact, Skinner acknowledges the existence of inner causes. Zuriff (1979) lists ten inner causes invoked in the writings of Skinner, and suggests that Skinner's objection is not to inner causes, per se, but to the kinds of theories that both postulate them and that are also incomplete, ad hoc, not measurable, and so on (cf. Creel, 1980). Apparently, it is not mentalism that is a problem, just bad mentalism. In summary, inner causes need not be introspected to serve as causes. Our fallibility in overt describing them, and our attempts to improve those descriptions by calibration with behavior, should provoke interested, even intense, study of the relation between inner and outer causes. It is Skinner's inconsistency in stating his position on these matters that has scared off young researchers, for no matter what theydo, theyare liable to be condemned by a mentalist if they study inner causes, as a methodological their sophisticated peers?as behaviorist if they ignore them. HYPOTHESES, THEORIES, EMPIRICS have often gotten poorly sorted in Skinner's metaphysics. Hypotheses are scientific guesses; both behaviorists and mentalists make them. They may be hard to notice in Skinner's writings because of the certitude with which he states them, but ask yourself, wherever he makes an assertion, "What are the data underlying that statement?" You will find many unsupported. For example, in discussing the process by which a community answer questions such as "Do you intend shapes us to observe our own behavior in order to not involve mediating processes" such as "do to go?" he emphasizes that such contingencies not a properly documented (See the interchange feelings (1972, p. 19). If this is fact, it is between Zajonc(1980, 1984) and Lazarus (1982, 1984) on this issue). If it is an opinion, it is unfair to dress it in the italics of a fact. By avoiding calling his ideas hypotheses, Skinner These avoids having to test them. Because theydefine Theories are ways of organizing bodies of data and empirical laws. and expanded, modified the nature of appropriate data, and because they are continually come after our data cannot as useful or useless. But theories theyare not so much true or false must grow together with the collection is complete, as Skinner suggests they should; they turn in so that may be modified by the they data, so that theymay guide their collection, and evidence. that some a theory is rightly regarded with some respect, it is inevitable tomake, hard often is distinction scientists try to elevate their speculations into theories (the not have however, Most speculations, since, at their inception, new theories are speculative. to called be and thorough enough, worked through existing data, are not broad enough for such a shortcut to the top motivated much of disdain a that I believe righteous theories). Because 32 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions KILLEEN Skinner's (1975) disparagement of theory. But the theoretical road is not itself easy. Theoreticians may labor harder over data than those who collected them, and may understand them more profoundly. Good theory is to be admired. It guides research. It summarizes and simplifies manifolds of data. It is the lietmotif that provides continuity and harmony to empiricism. The Insidious Nature of Covert Hypotheses Skinner was correct in noting that premature hypothesis formation and testing may preclude a more open-minded and fruitful exploration of phenomena. But even more dangerous is a premature closure that isattained by not treating one's insights as hypotheses, and therefore in need of testing. Thinking of a speculation as a hypotheses helps us to disengage from it if it turns out wrong. If we think it as an assumption or take it to be its invalidation would threaten thewhole system, and we are more likely to cling tenaciously to it,and to become doctrinaire. Or worse, bellicose. Witness the recent review of Gibson's work by Costali (1984), who questions the existence of cognitive structures, and seeks "an alternative scheme that does not merely question the solutions put forward by cognitive psychologists but converts their very problems from implicit to conspicuous nonsense." Cognitive psychologists are not innocent of confusion and error, both of the types that Skinner has identified, and others as well (Sampson, 1981). But denigration such as Costall's ismore likely to hurt the behaviorist movement than the cognitivist, for itplaces us in the roles of confounder, critic and saboteur, in all less seemly and less productive postures than that of originator. Science must be self-correcting, but since so much is provisional and since all systems contain flaws, the demonstration of a superior approach is much more effective in bringing about correction than is fault-finding. Let us agree with axiomatic, (1981) that "it is premature to depict cognitive accounts as inherently flawed and unworthy of the serious, dispassionate attention of radical behaviorists" (p. 167). Wessells THE POSSIBILITY OF EMPIRICAL MENTALISM success was not due to the fact that itwas atheoretical (itwas not?it was merely inchoately theoretical), nor that itwas anti-mentalistic (spitting into that wind has Its success was due to its kept us from studying some of themost intriguing phenomena). not get too far removed that discourse its and insistence for data, respect empiricism?its from it.The Queen told Alice that she could believe as many as six impossible things before breakfast; when analysis is stacked onto more than one or two tenuous assumptions, behaviorists lose their taste for themeal. This isnot to say that theory is thereby denied them. Anyone current with the discussions of maximizing versus matching in the analysis of Behaviorism's choice, of economic behaviorism, of optimal foraging, of behavioral momentum, will recognize that the name of the primary journal may soon have to be changed to the Journal of theExperimental -andTheoretical Analysis of Behavior. I think this is excellent progress. and The study of mind has proceeded from the opposite direction?speculative some a word. not With four-letter that it involves is its That practice, problem, unempirical. it is possible to say "mind" without a shudder of revulsion, and even to consider that itmight be preferable to "covert behavior," the first term of which has the synonyms "clandestine," "surreptitious," and "furtive," suggesting the overthrow of elected governments and the same experimental mining of harbors. The second term seduces us into believing that the events at some other for will do behavior overt of in the served have that study techniques level of observation, described in different terms, and measured in different dimensions. They may, but why limit ourselves? 33 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions EMERGENT BEHAVIORISM "The objection to inner states is not that they do not exist, but that theyare not relevant ina functional analysis"(Skinner, 1953, p. 35). "The major behaviorist objection to cognitive or mental explanations is not that they are not objective, or that they are based on the probability of investigation of ...but rather that they...decrease introspection, "Mentalistic ideas are so seductive that one is variables" (Branch, 1982, p. 373). manipulable of being led by them down the garden path of introspection and can afford to entertain the seductress" a behaviorist minded only tough mysticism....Perhaps Is that all that Skinner and other radical in Catania, 1982, p. 375). (Pavio, quoted behaviorists have been objecting to then, a dalliance with unproductive ideas? (Do not dismiss that objection by noting that ideas, as collateral by-products, cannot themselves in in danger any case produce behavior. It is the texts inwhich you read them, and what you say about them to yourself and to others, that has the behaviorists worried, not ideas themselves). But what if this concern ismisplaced? What if invocation of inner states and mental explanations turns out to be useful? What ifSkinner was rightwhen he said that "part of human progress has been the improvement of our description of these things [internal states]"(1964, p. 106)? We know the importance of chains of conditioned reinforcers: "Unless the gap between behavior and the ultimate reinforcer is bridged with a sequence of conditioned reinforcers, other behavior will occur and receive the full force of reinforcement" (Skinner, 1958, p. 95). about the gap between stimulus and response? How do humans mediate long gaps there, if not by talking to themselves, visualizing objects in familiar locations, and performing other mental gymnastics? Behaviorists accept such feats, preferring only thatwe call them behavior: seeing behavior, choice behavior, imagining behavior. But that qualification must take some of the fun out of it, for few behaviorists study those covert chains that mediate so much of human behavior. What Other researchers, less certain about the irrelevance of internal states, have attained insights about them that even behaviorists must applaud, if that be only briefly before they attempt to "translate" them, or with Costali, "reduce them to explicit nonsense." Posner (1982) has outlined the rich interaction ofmodels and data in thedevelopment of attentional theory. E. R. John, a critic of the behavioristic minimization of research on mental phenomena, has traced stimuli into the brain bymeasuring the evoked responses to rhythmic stimuli, tagged by its own rhythmicity. Among his findings: localization of the areas responding to the stimuli, and their spread during conditioning; upon introduction to the experimental chamber the rhythmic brain activity commences, "as if the animal were rehearsing the experience of themeaningful stimulus"; activization of that evoked response when the event is expected but absent; in differential conditioning, errors preceded by the rhythm appropriate to the incorrect (and not displayed) stimulus (John, 1976). The elegant work of Shepard and his students on mental transformations should by now be well known to all psychologists. In an accessible introduction to his work, Shepard (1978) notes the important role that imagery has played in the physical and life sciences. Shepard and Podgorny (1978) document and demonstrate many of the pervasive parallels between In his paper "Psychophysical Complementarity" cognitive and perceptual processes. he that argues (Shepard, 1981) evolutionary constraints have shaped the nature of our and and that these achieve a rapid "mesh" with the processes, perceptual cognitive environment via certain predispositions that guide our inferences in uncertain situations. arguments, perhaps, but he makes exceptional use of them. This view is Unexceptional parallel with that of Land (1978), a most practical scientist, who argues that animals evolved in a "polar partnership" with theworld around them,with an inner order related to an outer order; related, Shepard would say, as a key to a lock. Many of Shepard's critical papers are 34 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions KILLEEN in Shepard and Cooper discussed (1982). There one finds scores of demonstrations such as: The time taken to decide whether two figures are the same depends not only on how many degrees one must mentally rotate them to get them aligned, itdepends also on whether one rotates them the short way or the long way; when the subject is given a head-start on rotating an image before a second figure is presented, if it is presented when the collected and subject should have reached that orientation with their image, reaction time is uniformly short and independent of that orientation; subjects can rotate an image of a concrete object, but not of a frame of reference; rotation is either continuous or, ifnot, the size of thediscrete steps must be less than 30 degrees. Other transformations, such as dilation, contraction, translation, and translation of attention to different parts of the image, are also reported (see also the work of Kosslyn; e.g., 1980). The wealth of creative experiments and systematic analyses thwarts simple summary. More than Skinner ever did, Shepard and his associates are treating seeing as behavior. They seem not at all hampered by their resolution that the "thing" to which "seeing without the thing seen" refers isan image, and that subjects need not do without it. To the contrary. But behaviorists have never been very much interested in perception (Kantor, 1970); perhaps we should leave imagery to those who have a head start on us, both experimentally and conceptually (for the latter, see Block, 1981). Our interests have been pragmatic, concerning "manipulate variables of which behavior is a function"?Behavior Modification. Of course even there imagery will sometimes intrude. Covert sensitization and desensitization involve imagery, but except for the inconvenience, those procedures could be effected overtly. And, presumably, covert behavior "behaves" in the same fashion as overt behavior. But perhaps not always: Meichenbaum (1974) reviews studies where the covert processes do not respond to contingencies in the same fashion as overt behavior. Given theuntoward overt responses of some animals to contingencies of reinforcement, and their recalcitrance to ordinary conditioning procedures (Breland & Breland, 1966),we should not be surprised to find that there are special constraints on cognitive conditioning. But, with few exceptions (e.g., Catania, Matthews, & Shimoff, 1982;McGuigan, 1978;Neuringer, 1981), behaviorists have not been looking. "If those espousing a so-called radical behavioristic approach to cognition are to convert thosewhom they label cognitivists, then theywill have to do so with experimentation and data" (Wasserman, 1983). If,and when, we assay that shiftfrom rhetoric to research,we may be infor some surprises (Longstreth, 1971). But, finally, does not the study of mental phenomena, in cognitive terms, force us into a dualistic metaphysics? No, itdoes not. As Bunge (1980) notes, the thesis that dualism does not force us to adopt eliminative or vulgar materialism...i.e., rejecting psychophysical mind and brain are identical, that there is no mind.... Psychobiology i.e., the thesis that suggests...emergentism, is an emergent property possessed only by animals endowed with an extremely complex and plastic mentality nervous system.... [This] is not saying thatminds constitute a level of their own, and this simply because disembodied [One] can hold that themental (or even embodied) minds, but only...minding animals... relative to the merely physical without reifying the former, (pp.216-217) Emergent materialism. What there are no is emergent is the difference between this and mentalism? difference boils down to this: whereas mentalism has (simple) terms alone, materialism gropes for (usually complex) explanations possibly social circumstances as well. (Bunge, 1980, p. 214) The for everything mental in mental explanations of the mental in terms of brain processes and It appears, then, that what I have been calling empirical mentalism Bunge would call emergent materialism. Call itby either name, it is time for "tough minded behaviorists" to entertain the notion. It is time for them to become "emergent behaviorists." 35 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions EMERGENT BEHAVIORISM PRINCIPLES OF EMERGENT BEHAVIORISM What would Emergent Behaviorism be like? It would not be a "methodological behaviorism" that draws a boundary for its subject coextensive with the skin. Nor would itbe a cognitive psychology that, at itsworst, draws the same boundary but stays inside of it.Nor would it be a radical behaviorism that has been short-sheeted by its own parsimony, contorting its basic theoretical terms to cover areas that are out of their reach. It would be behavioral. Our roots are in the study of animal behavior, and in the rejection of introspective mental models for its explanation. An impressionistic theory of human motivation provides a poor foundation for a theory of animal behavior. But, conversely, a theory of the behavior of non-verbal animals provides a problematic basis for the study of human behavior, including that covert behavior called mental activity. Our behaviorism would be emergent because it recognizes the causal relevance of mental states, and thus the utility of having theoretical terms within the system to refer to those states. Itwould be inductive. But its induction would be Peirce's(Rescher, 1978), not Skinner's on observed based It the of would involve 1954). generation hypotheses (Verplanck, on based those the of modification and empirical tests hypotheses regularities ("abduction"), concern as events "at another Insofar these level," this may ("retroduction"). hypotheses constitutes a rejection of Skinner's positivism. Like Skinner's behaviorism, itwould be pragmatic (Platt, 1973). Itwould recognize that "the rationality and progressiveness of a theory are most closely linked?not with its confirmation or itsfalsification?but ratherwith itsproblem solving effectiveness"(Laudan, 1977, p. 5). The same is true of research traditions; those that are successful are those that lead to the solution of an increasing range of empirical and conceptual problems. Emergent behaviorism would result from a cumulation of successful theories and the problem domains that they have addressed?not by fiat, nor by aping the traditions of other sciences, nor by certain proscribing techniques on doctrinal grounds. This may seem unexceptional, but I believe that the commitment of op?rant behaviorists to description rather than problem solving has thwarted such development.5 Emergent behaviorism would be prudent with its resources, but not parsimonious. Parsimony is a better criterion for choosing competing theories than for generating new ones. The conceptual stinginess that it entails impoverishes exploration and thus stifles progress. It puts all the the emphasis on cost, rather than cost-effectiveness. New constructs should be viewed as new parameters in a regression equation; they involve a real cost, for they decrease the degrees of freedom in the data/theory system. But, if the data are sufficiently strong and numerous, such constructs may greatly improve our ability to comprehend them. Gratuitous and ad hoc explanations are to be rejected, whether they are mentalistic or behavioral, simply because they don't pay their way (Rescher, 1978). They dilute the explanatory power of the theory by introducing a "parameter" that can account for only a small domain of the data under purview. In sum, emergent behaviorism would focus on behavior, not on para-scientific theories about it such as those derrogated by Skinner in the first lines of this paper. Its goals are to develop successful theories of behavior, and it recognizes that those theories may involve mental terms. Its methods would include hypothesis generation and testing. It would be pragmatic, not parsimonious, asking firstwhat we can do, and only second, what itwill cost. Finally, because itopens the door to new freedom in theory construction, requiring only that the products pay theirway, it should restore the excitement that some in our field have been searching for at the back door of cognitivism. 36 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions KILLEEN REFERENCES status of the script concept. American R. P. (1981). Psychological Abeison, Audi, R. (1976). B. F. Skinner on freedom, dignity, and Bern, D. J. (1972). Self-perception B. & Skinner, B. F. (1967). The 27, 317-337. Block, N. (1981). Blanshard, N. (1982). Misrepresentating Branch, M. Breland, K., & Bunge, M. Breland, M. MA: Catania, A. C, with nonverbal The MIT 163-186. 4, 1-62. 6, debate. Philosophy and Phenomenological Row. and Brain Sciences, The Behavioral 5, 372-373. A psychobiological New York: approach. The Behavioral Pergamon. and Brain Sciences, 5, 374-375. Matthews, B. A., & Shimoff, E. (1982). Instructed versus shaped human verbal behavior, responding. Journal of the Experimental Analysis of Behavior, 38, 233-248. Cialdini, R. B., Petty, R. E., & Cacioppo, 357404. A. P. (1984). Are theories of perception necessary? A review of Gibson's Perception". Journal of the Experimental Analysis of Behavior, 41, 109-115. (1980). Radical epiphenomenalism: B. F. Skinner's account "The Ecological of private events. Behaviorism, 8, 31-53. R. (1976). Hierarchical organization: a candidate principle for ethnology. In P. P. G. Batesonand (Eds), Growing Points in Ethology (pp. 7-54). New York: Cambridge University Press. D. C. to Visual Approach R. A. Hinde Dawkins, Dennet, interactions J. T. ( 1981). Attitude and attitude change. Annual Review of Psychology, 32, Costali, Creel, R. Behaviorism, Psychology, New York: MacMillan. Behavior. (1982). Antimisrepresentationalism. 36, 715-729. Press. behaviorism. problem: Psychologist, of behavior. Social problem of consciousness?a New York: Harper & (1966). Animal (1980). The mind-body A. C. Catania, Imagery. Cambridge, (1967). Theory of Motivation. Bolles, R. C. in Experimental theory. Advances Research, the explanation (1978) Brainstorms. New York: Bradford Books. Goff man, E. (1971). Relations in public: Microstudies (1976). A model of consciousness. regulation (pp. 1-50). New York: Plenum. John, E. R. order. New York: of the public and D. In G. E. Schwartz Basic Books. (Eds.), Consciousness Shapiro and Kantor, J. R. ( 1970). An analysis of the experimental anaylsis of behavior (TE AB). Journal of the Experimental 13, 101-108. self Analysis of Behavior, Keat, R. (1972). A critical examination of B. F. Skinner's objections to mentalism. I, 53-70. Behaviorism, Killeen, P. R. (1981). Learning as causal inference. In M. L. Commons & J. A. Nevin (Eds.), Quantitative analysis of behavior (Vol. I): Discriminative properties of reinforcement schedules (pp. 89-112). Cambridge, MA: Ballinger. Kosslyn, S. M. (1980). MA: Image and mind. Cambridge, Harvard Land, E. H. (1978) Our "polar partnership" with the world around Loudan, L. (1977). Progress and itsproblems: Toward University Press. us. Harvard Magazine, January/February, 23-26. a theory of scientific growth. Berkeley: University of California Press. Lazarus, R. S. (1982). Thoughts Lazarus, R. S. (1984). On on the relations between emotion and cognition. American Psychologist. the primacy of cognition. American Psychologist, 39, Longstreth, L. E. (1971) A cognitive interpretation of secondary reinforcement. In J. K. Cole on motivation (pp. 33-80). Lincoln: University of Nebraska Press. Lowe, C. F. (1983). Radical behaviorism and human psychology. behavior (pp. 71-93). New York: John Wiley & Sons. McGuigan, Hall. Meichenbaum, Michael, Natsoulas, D. (1974). Cognitive behavior modification. J. (1980). Untitled presentation. Behaviorism, T. (1983). Perhaps In G. C. L. Davey Principles of covert behavior. F. J. (1978). Cognitive psychophysiology: 8, the most difficult problem Morristown, 37,1019-1024. 124-129. NJ: General (Ed.) Nebraska (Ed.), Animal models Press. 161-163. faced by behaviorism. Behaviorism, 37 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions of human Cliffs, NJ: Prentice Englewood Learning symposium II, 1-26. EMERGENT BEHAVIORISM A call for change. Behaviorism, Neuringer, A. J. (1981). Self-experimentation: T. D. Platt, J. R. (1973). The Francisco: Freeman. Skinnerian I. (1982). Cumulative Posner, M. development Rescher, N. (1978). Peirce's philosophy of attentional the punitive (Ed), Beyond theory. American society (pp. 22-56). San Psychologist, 168-179. 37, IN: University Press. of science. Notre Dame, in the pigeon. Journal of the Experimental Reynolds, G. S. (1961). Attention Sabini, J.& Silver M. In H. Wheeler revolution. 9, 79-94. than we know: Verbal reports on mental processes. Psychological (1977). Telling more Nisbett, R. E., & Wilson, Review, 84, 231-259. of Behavior, 4, 203-208. Analysis ( 1981). Introspection and causal accounts. Journal of Personality and Social Psychology, 40,171 179. psychology as ideology. American Sampson, E. E. (1981). Cognitive Shepard, R. N. (1978). The mental Shepard, R. N. organization Shepard, image. American Psychologist, In M. Kubovy Psychophysical complementarity. (pp. 279-341). Hillsdale, NJ: Lawrence Erlbaum Associates. R. N., & Cooper, L. A. (1982). Mental R. N., & Zare, S. (1983). Path-guided Sidman, M. (1958). By-products Sidman, M. (1979). Remarks. Simon, H. A. & J. R. (1973). The organization Pomerantz images and their transformations. Cambridge, (Eds.), MA: Perceptual The MIT Press. that resemble perceptual processes. InW. K. Estes (Ed.), 189-237). Hillsdale, NJ: Lawrence Erlbaum Associates. apparent motion. Science, 220, 632-634. of aversive control. Journal of the Experimental Behaviorism, 36, 730-743. 125-137. (1981). Shepard, R. N., & Podgorny, P. (1978). Cognitive processes Handbook of learning and cognitive processes (vol. 5) (pp. Shepard, Psychologist, 33, Analysis of Behavior, I, 256-280. 7, 123-126. of complex systems. In H. H. Pattee (Ed.), Hierarchy theory. Skinner, B. F. (1938). The behavior of organisms. New York: Appleton-Century. Skinner, B. F. (1953). Science Skinner, B. F. (1957). and human behavior. New York: Free Press. Verbal behavior. New York: Appleton-Century-Crofts. Skinner, B. F. (1958). Reinforcement today. American at 50. Science, Skinner, B. F. (1963). Behaviorism 13, 94-99. Psychologist, 140, 951-958. Skinner, B. F. (1964). Comments at symposium. In T. W. Wann (Ed.), Behaviorism bases for modern psychology (pp. 97-108). Chicago: The University Press. Skinner, B. F. (1969). Contingencies of reinforcement: A and phenomenology: Contrasting theoretical analysis. New York: Appleton-Century-Crofts. Skinner, B. F. (1971). Beyond freedom and dignity. New York: Knopf. and behaviorism. Skinner, B. F. (1972). Humanism Skinner, B. F. (1974). About Skinner, B. F. (1975). The behaviorism. The Humanist, New York: Knopf, steep and thorny way July/August, 18-20. 1974. to a science of behavior. American Psychologist, 30, 42-49. Sober, E. (1983) Mentalism and behaviorism in comparative psychology. In D. W. Rajecki (Ed.), Comparing Studying man studying animals (pp. 113-142). Hillsdale, NJ: Lawrence Erlbaum Associates. to action: The allocation Staddon, J. E. R. (1980). Limits Staddon, J. E. R. (1983). Adaptive of individual behavior. New York: Academic behavior and learning. New York: Cambridge behavior: Press. University Press. P. E. Meehl, W. S. Verplanck, W. K. Verplanck, W. S. (1954). Burrhus F. Skinner. In S. Koch, K. MacCorquodale, Estes, C. G. Mueller, Jr.,& W. N. Schoenfeld, Modern learning theory (pp. 267-316). New York: Appleton-Century Crofts. Wasserman, E. A. (1983). Is cognitive psychology behavioral? Psychological Wessells, M. G. (1981). A critique of Skinner's views on the explanatory 9, 153-170. Zajonc, R. B. (1984). On the primacy of affect. American Psychologist, Record, 33, 6-11. inadequacy of cognitive theories. Behaviorism, 39, 117-123. 38 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions KILLEEN Zuriff, G. E. (1976). Stimulus Zuriff, G. E. (1979). Ten equivalence, grammar, and inner causes. Behaviorism, internal structure. Behaviorism, 4, 43-52. 7, 1-8. NOTES and some are ?Some of these constructs are operationally defined (contingencies and motivational operations) 1000 schedule that is ineffective inmaintaining functionally defined (stimuli and responses). Thus, a Fixed Ratio behavior remains a Fixed Ratio 1000 schedule, but a "stimulus" that is ineffective in controlling behavior is not a stimulus. This is a curious mix of types. Certainly motivational constructs can be defined functionally (see, e.g., Bolles, 1967). So can reinforcement contingencies, although that ismore difficult and requires a feedback/control language Skinner and subsequent 1980, 1983). And stimuli and responses can be defined operationally?as (Staddon Experimental Analysts of Behavior routinely reported experiments using lights of some color, counting lever-presses of some force and so on. Unless we designed explicit experiments (e.g., as did Reynolds [1961]), we couldn't know that responses of a slightly different topography were under the control of different aspects of the same stimulus. No matter admired, such analyses are usually undertaken only when the experimenter runs into trouble (see, e.g., 1958). Thus while the op?rant in theory isan act (i.e., it is defined functionally), in practice it is usually defined of the cue, the operand urn, and the feeder. operationally as the closure of three switches in sequence?those how much Sidman, 2Without such interaction, there would be little need for the constructs. In cases where behavior becomes habitual or automatized, and thereforemore resistant tomodification, themore complex control provided by internal events such as self-instruction has given way to simpler chains of behavior; guidance by mental events vanishes, and with it the need for corresponding theoretical terms. But there are many occasions, themost characteristically human, where such control is present: the ability to guide, speculate, calculate and evaluate. Skinner discusses these different sources of control in his important article on rule-governed vs. contingency-shaped him in this venture to terra cognita. behavior (1969). Unfortunately, few behaviorists followed 3This arresting assertion is like a Zen koan, involving the nice contradiction that transitive verbs need not have predicates. Just why Skinner should deny that "private stimuli"can be "seen" is unclear. He denies the name "image" to an ensemble of them for fear of our seduction by the belief that there is a physical copy present in the brain. However, experiments to be cited below suggest that functionally the best way to treat such private events may be as images, because of themany properties they share with the referents of that term in its natural language use. that optimize learning in animals also satisfy Hume's conditions for inferring causality?contiguity in space, precedence in time, and constancy of conjunction (Killeen, 1981). But these are transcended and modified by mental activity, just as Korte's laws for perceived motion of flashing lights are modified by placing an arched "path" above the lights (Shepard & Zare, 1983). 4The conditions introduction of the construct "establishing stimulus* 5One of the few such elaborations I have seen recently isMichael's (1980; see also Sidman, 1979). I believe that itwas Skinner's contempt for (other people's) theory that lefthis students unprepared to develop his own. 39 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions 40 This content downloaded from 129.219.247.33 on Wed, 26 Jun 2013 20:51:14 PM All use subject to JSTOR Terms and Conditions