Goldiamond, Israel

Concepts of social power and its allocation are currently being applied to many social issues. In a related manner, questions are being increasingly raised with regard to the constitutional and human rights of prisoners, mental patients, and other subjects of institutional control.3 It is only to be expected, given this intellectual and social climate, that behavior modification procedures used in institutions (and elsewhere) should come under scrutiny. These procedures, needless to say, have not been singled out for such examination, since their examination is part of a larger inspection. Nevertheless, the use of terms such as “control,” “social control,” “conditioning” and the explicit relation of the procedures to a conceptual system derived from the animal laboratory seem to make them targets of choice, as well as of opportunity. Any reader is aware of the heat which these terms have generated when used simply as part of a conceptual system and applied in the abstract to human behavior. There is nothing new about it. What is new is the joining of the old philosophical arguments to the current use of applied behavior analysis. This confluence is compounded by the application of the term “behavior modification” to a variety of questionable institutional practices whose proponents refer to the conceptual system and its application as justification. A renewed challenge to the conceptual system and its applications is now under way. It threatens to affect opportunities for basic research as well as for the development of socially useful approaches and instruments. It is not, accordingly, a trivial matter of concern only to students and practitioners of behavior modification.

An example of the heat generated by abstract application is given by the reviews of Skinner’s (1971) book. With few exceptions, these were characterized by misunderstandings and less charitable distortions of human experimentation, behaviorism, and conditioning. No work, of course, is beyond criticism, but the reviews often told more of the postures of their authors than of the book reviewed. On a more practical level, Wexler (1973) has recently reviewed some relations between behavior modification and the law, with special emphasis on patients in mental hospitals. He has reflected some serious questions which have been raised about these procedures. Among these questions is the extent to which token economies, the potency of whose reinforcers may rely upon their deprivation, also deprive patients of their constitutional or human right to them. Wexler is not unmindful of the fact that elimination of such procedures may raise therapeutic problems. These questions and others he has noted have disturbed many members of the community of applied and other behavior analysts. These are genuine issues and Wexler has raised them with skill and understanding. His research has been thorough and his scholarship in both law and behavior analysis impressive. If the article is disturbing, it is nevertheless a relief to find a discussion of this nature free of the distortions, misunderstandings, or cant which have characterized too many other discussions. His review is, accordingly, an impressive and important contribution to the field and should be read for its major points, since only those germane to the present discussion will be considered here. The field of behavior analysis is fortunate to have come under scrutiny by a legal scholar who understands it.

One focus of Wexler’s discussion, and the legal concern it reflects, is upon the constitutional issues (cf. Sen. Ervin, 1973) raised by agents of institutions who intentionally apply explicit behavior control techniques to what is, in essence, a captive population. This population, subject as it is to total institutional control, can be legally seen as under coercion and thereby deprived of constitutional rights to freedom to assent or dissent. The issue becomes critical when the arena for assent or dissent is submission to behavioral control procedures which may shape and control the direction of future assent, dissent and, indeed, choice itself. Philosophical counterarguments that all choice is so controlled are beside the point when viewed in terms of the stark constitutional problem involved. In question form it might be stated: because a person has been classified as a patient, has an institution or its agents been authorized to deprive him of rights to assent or dissent, especially in those areas where the issue is to accept or not accept the implantation of the institutional value system over one’s own? Our political system is characterized by a well-justified suspicion of the potential for damage when powers are concentrated and has attempted to separate and diffuse powers, to institute checks and balances, along with other safeguards including due process of law. To what extent are there such checks in an institution, and has due process been extended to specify which behaviors are within the institution’s purview and which are reserved to the patient? What institutional safe-guards other than personal integrity and the sloppiness of the system exist against potential abuse? And if behavior modification is providing the powerful means of efficient control some of its adherents (and opponents, as well) claim for it, the mitigating slack is removed.

These questions are not idle ones. In my various trips around the country I have seen and received first-hand reports from various mental institutions. Some reports have been characterized by ingenious solutions which I have found helpful. But in one institution patients were sleeping on a bare floor because they were not participating in the institutionally decreed behaviors and thereby not getting the tokens by which they could purchase bed-space. What are the limits on deployment of consequences? In another, a drug abuser was diagnosed, upon commitment, as an “inadequate personality.” It was decided to help her become more adequate by attaching consequences to shape her out of her Southern accent into a more businesslike Midwestern one. What are the limits on behaviors under purview? In yet another, the hospitalization of a patient for depression was related to the disintegration of supports outside: his wife had initiated divorce action, his partner had absconded with the assets of the firm, and his friends and adult children had deserted him. It was decided to alleviate his depression by reinforcing with attention activities (such as ping pong) specified on a posted list and to extinguish his legitimate references to impending doom by ignoring them. What are the limits on statement of contingencies? In all three cases the institutional agents were not trained in behavior modification, but the programs were approved as such by the institution.4 In other cases, better trained people have been involved. Needless to say, professional inadequacy is not limited to any particular psychological orientation, but the issue is an important one in any discussion of ethics, legality, or human relations.

These three examples are attributable to inadequate training. The incompetence demonstrated parallels the cocktail party interpretation of dreams by someone who has read a psychoanalytic book on dream interpretation. It appears to be inevitable that abuses will have occurred, given the avidity with which behavior modification has been sought after and given the rapidity with which some of its procedures have spread. Given the existence of such a market, books and manuals for varying audiences have proliferated, and many of these make claims not supported by the present evidence or are less than adequate in other ways.5 The public’s lack of sophistication in this area also extends to identification of the field itself, so that brain surgery in order to change behavior is identified as a behavior modification procedure. The neurosurgeons, of course, might not so identify their own procedures, but practitioners of punitive and other controls often so identify their procedures. Experimental and applied analysts of behavior have reason to be concerned, as people, regarding the possibilities for individual damage or ineffectiveness which may ensue; as citizens, regarding the constitutional issues involved; and as professionals whose discipline and opportunities may be affected.6

Since constitutionality has been raised as an issue, we shall open our discussion with an examination of the constitution as a guide for a discussion of ethical and legal issues raised by applied behavior analysis. The arguments that will be developed are that its safeguards provide an excellent guide for program development of an effective application of behavior analysis to problems of social concern and that the violation of these rights can be counterproductive to the patient, to the aims of institutional agents whose incentives are therapeutic, and to the therapeutic aims of the society which sponsors the patient-therapist (programmer, teacher, etc.) relation. Such violation may, however, serve other ends which, in our society at least, are not considered consonant with the social contract assumed to underlie its polity.


Behavioral contracting is a proclaimed feature of many programs in behavior modification. The contracting may be explicit, as when a professional and patient agree, in writing, on outcomes. The explicit agreement may be verbal. When the negotiations are between two consenting adults, the rationale is familiar to practitioners of the better-established psychotherapies. The contributions of applied behavior analysis do not simply lie in the explicitness of goals or outcomes (called “targets”) and the necessary negotiations, but in the products of the added requirement that the procedures directed toward these also be explicit, in parallel with the procedural requirements of programed instruction. On the other hand, the contracting may be implicit, as when someone buys a programed text in calculus in the hope of being able to apply its concepts and procedures to mathematical problems upon completion. Needless to say, there is no record of successful suit for breach of contract against a publisher if the outcome specified in the title is not attained, nor am I familiar with any in psychotherapy. However, the reader is undoubtedly aware of the controversy over accountability in education. The discussions are being extended to psychotherapy, and explicit societal consequences may in the future be applied to this area as well. At present, however, the psychotherapeutic contract, as the term is variously used in these helping professions, is not legally binding ( for an opinion on the legal and other implications of the contractual relationship between physician and patient, see Fletcher, 1972).7 Accordingly, the term “contract,” while lacking the full legal sanctions typical of commercial transactions, does share with these a quid-pro-quo relationship between consenting adults such that, if one party behaves in one way, the other party will behave in another way. The outcome agreed upon by both may thereby be attained-that is, both parties will “work toward” its attainment in recognizable ways.8

The Constitution may be viewed as a statement of governmental organization and powers, or as a contract between a federal government on the one hand and its constituent states and people on the other. So considered, all legislators and judicial and executive officers “both of the United States and of the several States” are specifically required to promise to support the terms of this contract (Article VI). Although there are no specific parallel requirements enumerated for another party to the contract, namely, the people, the historical context of the Constitution suggests their scope. Rousseau’s The Social Contract (1762) relates the legitimacy of government to consent of the governed, in contrast to an alternative approach9 of compliance to rules for obedience decreed unilaterally by a higher source. The Declaration of Independence explicitly states that governments “(derive) their just Powers from the Consent of the Governed.” Further, “Governments are instituted among Men” in order to “Secure these Rights,” namely, “life, liberty, and the pursuit of happiness.” Where these are abrogated, it is the people’s “Right, it is their Duty, to throw off such Government, and to provide new Guards for their future Security.” The converse is that the contractual obligations on the people, when a government is established with their consent, is to support that government in all the ways that are negotiated.

The Constitution may also be considered as a program contract, as the term is used in programed instruction. The fulfillment of such contracts requires specification of (1) targets, or explicitly stated outcomes, (2) current repertoire10 which is relevant to the outcome, (3) the steps which will mediate between current repertoire and target repertoire (which can be developed) and (4) a system of consequences explicitly contingent on advancement through the required progressions, and which maintain such behavior.

The outcome of the Constitutional program is stated, as in all good program-contracts, at the very beginning of the document: “We the People of the United States, in order to” followed by the seven outcomes. All seven are stated positively, e.g., “establish justice,” not “eliminate injustice”; “secure the blessings of liberty,” not “undo the curse of tyranny.” The sense of the outcome is conveyed not only by the wording but by the possible alternatives excluded, for example, “establish the one true Church and thereby propagate the one true Faith.” The Preamble is not stated as explicitly as a therapeutic contractor might desire, but its goals are considerably broader and more ambitious. They are also intended for “our posterity”; therefore, the time limitations which typically govern other contractual statements of objectives do not hold. The current repertoires relevant to these ends are partly available in the social context of the time (which includes English history and common law and religious and Biblical traditions11) and are also specified in the starting structure and powers enumerated in the Constitution (e.g., Article I, Section 8, for Congressional Powers), with a significant provision to be noted shortly. The procedures for program change are included in the foregoing, as well as contract renegotiation and amendment in Article V. The Supreme Court and usage, of course, have provided additional procedures. Consequences to maintain adherence to the contract are also articulated. Some involve aversive control, but the majority of maintaining consequences are relatable to reinforcers, e.g., procedures for growth through adding new states.

A property of the contract which is critical to our argument is that the powers assigned to one of the contracting parties, namely, the federal government, are limited to those explicitly stated in the contract, as proposed for amendment two years after signing and as amended (1791): “The powers not delegated to the United States by the Constitution … are reserved to the States respectively, or to the people.” With regard to one of the other contracting parties, the powers not explicitly “prohibited by it [the contract] to the States” are similarly reserved to the States and the people. Stated otherwise, one contracting party has only the powers explicitly specified by the contract. The other party has all other powers except those explicitly withdrawn. This, of course, is the exact opposite of a central authority in which all powers reside, except those delegated or granted to other authorities. The Constitution, accordingly, provides for a limited system-limited, in essence, only to those procedures specified. This allocation of powers is designed to produce and maintain only those positive outcomes which are similarly specified. It does not provide for a total system. In political terms, it is not the blueprint for a totalitarian state. It stands thereby in sharp contrast to the total systems developed in mental hospitals and other institutions. While it has been the violation of specific constitutional rights by therapeutic (or related research) agents that is the current concern, it should be noted that such violation is occurring in the institutional or social context of systems whose political assumptions are diametrically opposed to those underlying the Constitution. These total institutions are characterized not only, as noted, by sharp deviation from certain assumptions basic to the society which sponsors them, but also by their assumption, under social sponsorship, of corrective or therapeutic goals. Possibly, the political antinomy noted of constitutional and institutional procedures resides in the assumptions underlying these goals. Accordingly, we shall explore the relationship of the antinomy to therapeutic goals, as currently defined.


Goffman (1961) notes in his discussion of “total institutions”: “A basic social arrangement in modern society is that the individual tends to sleep, play, and work in different places, with different co-participants, under different authorities, and without an over-all rational plan. The central feature of total institutions can be described as a breakdown of the barriers ordinarily separating these three spheres of life” (1962, pp. 5-6). The major characteristics Goffman notes are that all aspects are conducted in the same place and under a single authority. They are tightly scheduled under an explicit system of rules and controls and are under a single rational plan which derives from institutional aims; in some institutions, members are usually “required to do the same thing together.”

Nontotal institutions, such as factories, may also have lunchrooms and recreational facilities, and scheduled lunch and recreation periods, but Goffman notes that “the ordinary line of authority does not extend to them.”

The term “total” is self-assigned by some institutions. For example, the Patuxent Institution, set up as a therapeutic prison, describes itself as a “total-treatment facility” (Goldfarb, 1974).

Stanton and Schwartz (1954) consider the mental hospital as a “total social institution” and as “a place where ordinary civil liberties are called ‘privileges'” (p. 244). Further, in the mental hospital they investigated, “seriously disturbing conflicts about delegation of authority, about ‘authoritarianism,’ about freedom . . . were not only frequent at the hospital among both patients and staff-they were almost the rule” (p. 244).

It would appear that some of the legal questions which Wexler raises with regard to token economies derive, to a considerable extent, from their existence within total institutions. The token economist operating as their agent need not consider himself singled out for persecution. However, with reference to Goffman’s three spheres, namely, sleep, play, and work, the token economist may deliberately make their availability or method of delivery contingent upon behavior in accord with institutional aims, that is, withhold or present them as reinforcers. This makes him conspicuous and vulnerable to legal scrutiny.

The Constitution, as was noted earlier, sets up a limited government rather than a totalitarian one, and civil rights are to be considered in this context. Whence the total power wielded by total institutions?

A person may “be deprived of life, liberty, or property” with due process of law and may not without it; this limitation is imposed on the federal government (Fifth Amendment) and upon the States (Fourteenth). Sentences and fines which attach liberty or property typically specify, within limits, duration of attachment of liberty, and type and amount of property (usually financial) which is forfeit. Whence the lack of specificity, the unpredictability and pervasiveness of deprivations by total institutions, which are often governmental agents and which seem to be exceptions from the Constitutional concept of limited government?12

A part of the answer, in my opinion, lies in the frequent vagueness of definitions of mental illness, in the frequent unpredictability of the exact topography of the disturbing behaviors or in the time and place of their occurrence and therefore, in the consequent ascription of those behaviors to an underlying and pervasive pathological state, of which they are considered manifestations or symptoms. These formulations affect the therapeutic practice for which the institution is set up. Where the disturbance is episodic, since its exact timing and location can not be predicted, surveillance is required whenever and wherever the disturbance can occur. Since the exact topography of the disturbance and the degree to which it will disturb others can also not be specified, isolation from others seems prescribed. Whether the disturbance is episodic or continuous, the behaver is considered as a patient who must be treated, and his pathology must be eliminated or brought under control so that we can predict that future disturbance will not recur. It would seem that a heavy burden of striving toward omnipotence and omniscience (including clairvoyance) is laid upon the institution. Temptations to dictate to the patient, to attempt unilaterally to substitute one’s own aims for the patient’s or otherwise to assume parental roles (punitive or indulgent or farsighted) are dangled before the institutional agent.13

The total and long-term responsibility described, the roles assumed in accord, and the ideology of pervasive illness which rationalizes them may be related, in part at least, to requirements imposed by the larger social system. Some of these are evident in the immediate attacks upon an institution and the mental health system, followed by outcries for a “thorough” investigation, when it is discovered that the perpetrator of a series of crimes was once a mental patient. The system is then under fire for not having foreseen the future, that is, for having released the patient before the illness was totally extirpated. The system may then be threatened with enforced personnel changes and with a diminution in funding. It is the already low state of funding, the system argues in defense, that has made it custodial rather than curative; this renders it powerless.

To this social attack is added the current attack by proponents of civil liberties that even these powers be diminished. One outcome of total institutional control that has been noted is institutionalization. This describes the acquisition of new patterns of behavior which accord with institutional requirements. These patterns are then cited by the custodial system as evidence of the very mental illness for whose treatment the patients were committed. These patterns are considered antitherapeutic and as discontinuous with those required outside. Patients have been regarded as generally powerless against their total controllers. Indeed Kesey’s (1962) stirring novel, and its adaptation as a play, depicts patients who attempt to maintain their human dignity against the total power of institutional agents. The ascription of such powerlessness has been part of an assault on institutions by social scientists, among others.

Although they agree that the current approach is degrading, Braginsky, Braginsky, and Ring (1969) present a view diametrically opposed to the notion that institutionalization is a product of patient powerlessness against an all-powerful institution. Rather, they argue that the patients’ patterns include “impression management,” namely, that patients adroitly manage the impressions of them held by the staff in order to “achieve outcomes congruent with their primary motivations” (p. 46) and that there is considerable “continuity between the patients’ life style outside the hospital community and what they develop within the hospital community itself” (p. 46). Stated succinctly, patients utilize the mental illness model to attain their (sensible) ends and are the opposite of powerless. The authors’ thesis can be translated into clear opérant terms: staying in the hospital is a reinforcer whose requirement for delivery is behavior defined as sick. Patients in one of their studies varied their behaviors in accord with experimentally-instigated changes in definition. Braginsky, Braginsky, and Ring propose institutional solutions considerably different from the present total institutions.

In any case, whether the patient is viewed as being manipulated, or as reacting to manipulation, or as manipulating, the mental institution as a total institution is under unsympathetic scrutiny. When one adds to this the attack on total institutions in general, the dismantling propensities of economy-minded state administrations, the recent legal opinions noted in Wexler’s review, and the growing issue of civil liberties, it would seem evident that there is strong pressure to diminish whatever powers such institutions possess. For these days, at least, to paraphrase W. S. Gilbert, “An institution’s lot is not an ‘appy one.” Nor, it would seem, will be the lot of those of its agents who institute programs which capitalize on its totalinstitutional properties.

Whether or not the newer approaches and ideologies will replace the mental illness ideologies will, of course, be a function of the extent to which the newer social contingencies, which the newer ideologies rationalize, replace the less recent ones rationalized by the present models. Ideologies can exist for a long time without having much social impact and models can exist for a long time without having much scientific impact, but the “idea whose time has come,” that is, which begins to have social or scientific impact and to exert an influence over behavior which has hitherto been lacking, derives its onset of power from changes in social or scientific contingencies. These exert new behavioral requirements, and the ideologies and models which rationalize these behaviors and the contingencies of which they are a part now seem invincible (and even causative).14

Mental hospitals have been charged with dehumanization and ineffectiveness for some time, but there is a question as to whether the current eagerness to dismantle them in several states stems simply from the chord of humanitarian concern over these patients and their families which the charges strike, or from the rationalization thereby provided for reallocation of state funds to redress other human problems, which impose stronger requirements. The oil crisis has hastened our realization that the availability of unlimited energy, which even underlay design trends in urban architecture, has been an illusion. We shall have to alter our ways to live within energy limits. Similarly, we shall have to learn to live with limits on support for social programs. I shall defer consideration of the precipitating crises to a later discussion of models which are responsive both to the crises and the constitutional and ethical issues of concern.

As part of the effort to assign priorities, attempts are now being made to apply cost-benefit analyses to total systems as compared to other systems. This, of course, requires development of other systems, as well as refinement of measures of outcomes, costs, and benefits. Possibly, we are in for a period of reallocation of resources to set up alternatives which provide an opportunity for comparative shopping. Setting up alternatives implies dissatisfaction with present enterprises, and we should accordingly expect continued scrutiny, much of it unsympathetic, of the present enterprise. On the other hand, comparing the alternatives to the present enterprise implies the existence and support of the present enterprise, and we should accordingly expect it and its related rationale to continue for some time, although with diminished support and self-confidence.

Societal pressure for institutional clairvoyance, it was noted, dovetails nicely with that model of mental illness which is centered on a therapeutic mission (therapeuein, Greek, to cure) in which the therapist is a social agent who (a) is clairvoyant and produces irreversible results, and (b) contracts this outcome with society rather than with the patient. We have noted that the first requirement supplies part of the rationale for the total institution, whose design conflicts with the more limited constitutional system which sanctions it. The second requirement conflicts with the contractual nature of the larger constitutional system, in which the contracted outcome is between the government and the governed rather than being an agreement which imposes an outcome on a third party who has not entered into the negotiations, as in the alleged Mafia usage.

Both requirements have been ascribed to the medical model. With regard to the first requirement, the permanent and ubiquitous cure desired for mental illness parallels the definition of successful cure in other branches of illness. After successful appendectomy, for instance, the surgeon can be clairvoyant. Appendicitis will recur never and nowhere. Similarly, certain types of immunization have long-term effects. However, one can pose a different model of treatment. We would not judge, for instance, the quality of an internist’s treatment of pneumonia by the nonrecurrence of pneumonia. As a matter of fact, respiratory ailments may now be more likely. Societal pressure on mental health professionals, for some reason, implies assignment of the appendicitis model of illness rather than the pneumonia model. One of the reasons for this social choice may be the fact that many professionals have accepted it for themselves, if not as describing the present state of their art, then as describing its desired state in the future.15 An ubiquitous outcome is considered desirable not only by those who derive this illness model from personality models but also by such opponents of both personality and illness models as behaviorists and behavior therapists. Rather than ascribing desired ubiquitousness to relief from illness, they ascribe it to a process of “stimulus generalization.” If this eludes the present state of the art, it is a desired goal. For instance, I have a letter on my desk which asks me, regarding one of our programs, for “follow-up data and/or evidence for generalization outside of the treatment setting.” (My answer, I hope, will be satisfactorily responsive to the intent of the question rather than to its form.)

The source of that model of illness to which the social demand of professional clairvoyance and its related solution of total institutionalization are ascribed is said to be the medical model, for which various alternatives have been proposed. However, I have already noted two caveats. First, the social demand and accompanying solution are not confined to the medical model. Behavioral models can serve, as well as the various approaches which rationalize our present prison system. All of these have in common the definition of successful intervention by nonrecurrence of the presenting problem. Second, this definition also excludes those branches of medicine which deal with practices other than “ectomies” (or immunizations, etc.), namely, the vast field of complaints whose successful treatment and alleviation are not defined by nonrecurrence. Stated otherwise, both by inclusion of other models and by exclusion of much of medical practice, the medical model is not the culprit.

If guaranteed nonrecurrence does not characterize medical practice in general, neither does legally enforced exposure to treatment. For example, written consent is required for surgery.16 The contract is between the two parties involved. In practically every section of the hospital except the psychiatric ward, a patient can decline a given form of treatment, can refuse medication, and can leave the hospital AMA (against medical advice)-even when it is thought that his doing so endangers his life and limb. It is interesting that when departure AMA does constitute a threat to life or limb, a psychiatrist may be sent for and may provide the loophole in the otherwise constitutionally-concordant staff behavior.17

If adherence to a medical model and to a model of mental illness underlies total institutional control and the impositional nature of its treatment, such models are exceptions in the fields of medicine and medical treatment. Mental illness may be a myth, as Szasz (1961) states (more appropriately, it is a term applied to a class of models), but total control and its concomitants, including divorce from personal responsibility, are not logically implied by the medical model. Further, they may also be derived from other models. These include behavioral, educational, psychological, and social, among other alternatives suggested. Each has been paired with the medical model as the opposite end of a dimension in a juxtaposition suggested by health-disease or their various synonyms, with the implied dichotomy being innovation-status quo.

Some of the insistence on this dichotomy, both by proponents and opponents, is undoubtedly related to such political and economic considerations as to who will set policy, be professionally responsible, and collect payments. In this social context, the medical model of mental illness rationalizes control and payment to medical practitioners. Other models of disturbing behavior rationalize control and payment to members of other disciplines. I can not guess at the extent to which this very real professional conflict underlies the conflict in ideologies, but my discussion will not be concerned with this conflict. Rather, it will be limited to the relation between models and the stimulus control they ultimately exert over those behavioral outcomes to which the implicit social contract between society and its helping professions is addressed.


The present section will compare two orientations toward treatment which, I believe, can profitably be applied to present practice and research. I hope thereby to make explicit certain assumptions and procedures whose present implicitness creates problems in comparison and analysis. The term for one of the orientations is a new one. In the section thereafter, I shall present a model which derives from this orientation and is thoroughly consistent with (a) the constitutional requirements of mutual contracting and limitation of power, (b) other ethical obligations which the Constitution exemplifies, (c) the therapeutic needs of the patient (or other consumer), and (d) the investigative and analytic requirements of behavior analysis. As a matter of fact, by using this model the needs of the investigator and client can best be met through meeting constitutional requirements and ethical obligations. The more general orientation, into which the specific model fits, is shared by many other approaches and models and is not exclusive to it. I believe the presentation may make explicit a direction toward which the field has been moving. Hopefully, it will hasten the process.

The orientation to be proposed is a constructional one. This is defined as an orientation whose solution to problems is the construction of repertoires (or their reinstatement or transfer to new situations) rather than the elimination of repertoires. Help is often sought because of the distress or suffering that certain repertoires, or their absence, entail. The prevalent approach at present focuses on the alleviation or the elimination of the distress through a variety of means which can include chemotherapy, psychotherapy, or behavior therapy. I shall designate these approaches as pathologically oriented (pathos, Greek, suffering, feeling). Such approaches often consider the problem in terms of a pathology which-regardless of how it was established, or developed, or is maintained-is to be eliminated. Presented with the same problem of distress and suffering, one can orient in a different direction. The focus here is on the production of desirables through means which directly increase available options or extend social repertoires, rather than indirectly doing so as a by-product of an eliminative procedure. Such approaches are constructionally oriented; they build repertoires.

The fact that the outcomes are described differently is not simply a matter of verbal redefinition. The differences that can result become clearest when considered in terms of the four elements of a program, previously noted.

1. Outcomes or targets: – Although similar outcomes may be produced by the two orientations when viewed in terms of distress alleviated, the outcomes of the two approaches are not necessarily similar when viewed in terms of repertoires established. For example, in a series of treatment sessions one can progressively decrease stuttering and thereby increase the ratio of fluent words to total utterances. One can also progressively instate and extend a specific fluency pattern which consists of well-junctured speech and thereby increase the ratio of fluent words to total (and decrease stuttering). Viewed in terms of elimination of stuttering or increase in fluency (the alternate statements can simply be verbal redefinition), the outcomes may be similar. However, viewed in terms of patterns established, the outcomes may be quite different. And the training procedures and other program elements must also differ. This raises questions about outcome comparison.

2. Current usable (relevant) repertoires: – Where the outcomes, in terms of repertoires to be established, differ, the search for what is currently relevant must be oriented differently. For example, one can focus on (and try to describe) what is wrong, or is lacking, in order to correct it. In the other case, since one is trying to construct new repertoires, one must focus on what repertoires are available, are present, and are effective. Accordingly, different data bases are required. Where there is overlap in the data bases, they can be interpreted differently. For instance, one can consider the presenting symptoms as among the pathologies to be overcome or eliminated; they can be considered as indicators of a pathology to be specified. On the other hand, the presenting symptoms can be considered as among the entry repertoires available for construction or program guidance; they can be considered as successful instruments which produce reasonable outcomes to be specified and harnessed. For example, a pervasive cockroach phobia can be interpreted as an unreasonable fear which is so crippling to the wife that she cannot move from room to room unaided. On the other hand, it can be interpreted as highly successful instrumental behavior which dramatically forces the husband to provide the legitimate attention which he had hitherto withheld and deprived her of.18 The program thereby initiated is to teach him to be responsive to her legitimate needs and to teach her to express these in ways which get across to him more readily.19

3. Sequence of change procedures: – Given different target outcomes and different starting points selected for their relevance to the outcome, the mediating procedures which convert entry repertoire to target repertoire must also differ. The data which are considered as designating progress will differ, as must assessment of therapeutic effectiveness. In the phobia case just cited, although the phobia may progressively diminish, the graphs will be of increasing communication. In the case of a severely regressed schizophrenic woman, the change procedures have involved instatement of a multitude of specific repertoires, some in sequence, and some concurrent.

4. Maintaining consequences: – The contingencies of which each of the steps in a program is a component may also differ in pathologically and constructionally oriented programs. The consequences in one case may be progressive relief, diminution of aversive control, or gradual progression to such relief. In the other case, they may be explicit reinforcement of units in a progression, or gradual progression toward the repertoire to be established. In the latter case, assessment concentrates on reinforcers in the natural environment. These reinforcers can be those which have hitherto been disrupting behavior. For example, a mother considers herself at a loss in rearing her son. His obnoxious behavior continually enrages her, and both his misbehavior and her rage are increasing. She reports that she is a complete failure. Our analysis is that she is a complete success. She has successfully shaped escalating misbehavior by ignoring it when it was mildly disturbing and acting only when it had exceeded the previous limit to her tolerance. This suggests that her attentiveness is a powerful reinforcer. His main way of getting it now is by infuriating her. She is to use this reinforcer to maintain progression through a different kind of program she will apply.20

The symptom whose elimination is the target of a pathological approach may not only be considered as a currently usable repertoire (the cockroach phobia mentioned) but also as an important guide to critical reinforcers. For example, an obsessional patient talked rapidly and almost without stop about emanations attacking her thoughts; her eyes were piercing and she was agitated throughout. She had been an inpatient on and off and an outpatient for 20 years. Her black and purple costume made her immediately recognizable at the emergency room which immediately sent her to Psychiatry. She was supported by a small pension and lived alone and friendless in a small rented room. In the event that she were “cured,” what could she find to occupy her all day? At the present, she came to the hospital and met all kinds of different and bright people who cared. She belonged – she had community. If we were crazy enough to think we could “cure” her, she was not crazy enough to be “cured.” Such elimination had been the thrust of the various preceding therapeutic efforts, which had made little progress. We told her that regardless of her behavior, she was always welcome: she was a permanent part of hospital records and was provided access to them. Community was a critical reinforcer and the intervention strategy opened with this provision while developing other contingencies.21

Before continuing the presentation, I must stress that I am not thereby distinguishing between, say, psychoanalysis and behavior analysis. Psychoanalytic therapy contains constructional procedures, and applied behavior analysis contains eliminative procedures. The distinction I am making is between eliminative procedures deriving from a pathological orientation and constructional procedures deriving from a constructional orientation. These cut across different schools and models. I shall return to this issue shortly.

The issue of sets: – Successful elimination of a pattern related to distress can alleviate that distress and suffice unto itself. Medicine, among other disciplines, supplies abundant examples. Where a transient disturbance makes extant personal repertoires or environmental resources temporarily unavailable, direct elimination or control of that disturbance will also be effective. Where, however, the solution requires establishment of repertoires, an eliminative approach presents problems. This relates to the fact that the set of elements which do not bear on a problem is usually more extensive than the set of elements which do.22 A simple example, drawn from a discussion with the obsessive patient (who talked incessantly about things going wrong) may illustrate the point. The correct answer, I noted to her, to the calendric question, What is today?, is March 26, 1974. Period. The number of incorrect answers is limitless. It makes possible (as I noted) extended conversation. I can state: It is not March 15, 44 B.C. I thereby display my knowledge of history. Or, I can be highly original and creative: It is not September 57, 2074. Barring such hidden agendas in my answer, I can try continually to exclude elements in the hope of isolating the right one or try to develop some logical rule to guide such eliminations and save time. Learning what to do using this approach can be slow and painful, as can trying to abstract rules which guide appropriate behavior. Costello (1974) reports that when she attaches a time-out consequence to repetitive word stutters this particular pattern is attenuated but she must attach time-out to repetitive syllables, etc. Costello is consistent, but where reinforcement is or was occasional, our knowledge of laboratory contingencies teaches us that both punishment and extinction can produce perseveration of behavior. The ineffective pattern may therefore not be eliminated, and an element from the set of effective responses may not appear.

There are, of course, contingencies other than those given, but one effect of focusing on the larger set of problems to eliminate has been the slowed pace of development of intervention procedures and formulations relevant to obtaining satisfaction. It is impressive to note the contrast found in patient work-ups between the extensive and detailed reporting of present illness and its history and the skimpy and global suggestions for treatment. Indeed, the contrast found in the professional’s report often very accurately reflects an identical contrast in the patient’s self-report. This will be long on affliction and short and succinct on treatment (You help me). The parallelism in the two reports is not accidental. Nor is it derived solely from mutual manipulation based on the current identification of the helping mental health professions with the elimination of pathology. For the patient, since presentation of pathology is a necessary condition for admission, he will therefore so present himself; for the therapist, since treatment of pathology is his repertoire, he will therefore solicit and shape it. These operants undoubtedly enter on many occasions, but other contingencies may contribute to the continual recurrence of this parallel.

One strength of the parallel derives from the fact that it is found not only between professional’s presentation of patient and patient’s presentation of self, but also between patient’s seeking help from others and patient’s seeking help from self. Stated otherwise, the patient does not come for treatment when things are going well not simply because he then lacks the ticket for admission, so to speak, as might be inferred from a simple operant analysis, but for the same reason that he will typically not bother to analyze what is going on in his own life when things are going well, either. It may be argued that it is human nature to define problems only when things are going wrong, that is, when the crisis is upon us. This observation can be explained in more mundane contingency terms, namely, that when things are described as going right, certain referent behavior-reinforcement contingencies are in effect. Analysis of these is not only costly on its own, but by displacing the referent behaviors, it thereby disrupts their contingencies. Accordingly, the doubly costly behavior of analysis is typically assumed only when the referent contingencies change so that the referent behaviors or consequences become excessively costly. Stated in common language, the patient is hurting. He may then take time to analyze and reflect, but characteristically, what he will then reflect on is what went wrong, and what and how it should be avoided or eliminated. The likelihood of constructional solution is then remote. In parallel manner, it can be shown that the patient will seek outside help under similar conditions, often after repeated failure of self-analysis of the type reported.

Accordingly, the consumer of the remedial service as well as the delivery agent may both focus on the distress involved and its alleviation. The patient seeks help because things are going wrong and the therapist is a member of a helping profession. More than the mutual shaping described earlier goes on: both are governed by similar ideologies. The personal history of the consumer as well as the professional history of the delivery agent may then become the observation of problems viewed as distressful or cases in distress, of ways in which things have gone wrong, of cries for help which went unheeded, of the often arduous and dubiously successful nature of the remedial undertaking. With Muller (1953) they can conclude that “the tragic sense is the profoundest sense of our common humanity,” and that our acceptance of this sense provides the hope that we might thereby “be freed from the vanity of grandiose hopes as of petty concerns. We might learn that ‘ripeness is all,’ and that is enough” (p. 374).

Parallelism may, however, be developed in another manner. With regard to individual analysis, it is when things are going right that one might try to analyze the relevant contingencies and thereafter attempt to replicate them, and then observe the conditions under which replication is successful, attempt to institute them, and so on. The helping professional might help the patient do so. Of course, people are currently doing this, professionals are currently doing this and often, indeed, ask other professionals for such constructional advice to solve a problem. However, compilations of do’s do not tend to produce professional acclaim, nor are theories which organize such concepts and which direct data acquisition as prevalent as are those with a pathological orientation (such writings do contain constructional elements and syntheses). The constructional approaches tend to be regarded as cook-books, and their authors as technicians or popularizers. Nevertheless, they do strike a tremendously popular chord; witness the success of Dale Carnegie books and programs and other How to books. What I am proposing is that academic thinkers orient their presently highly developed analytic and research repertoires to fill this gap. This will require shifts in orientations and expectations on many sides. This was exemplified by the comment of a patient when asked, after listing his grievances and deficiencies, to now devote at least equal time to his strengths and assets: “I didn’t know I was coming for a job interview.” The comment also illustrates the fact that explicitly in other areas, and implicitly in the area of concern, we are already constructional. I suggest we start to make explicit and systematize what we are already doing. At the present time, the tremendous gap between academic theory and research, on the one hand, and practice, on the other, is related to the recurrent threat to tear apart at least one professional organization.

The social ambience of the pathological orientation: – A pathological orientation is congruent not only with the pessimistic views noted but is also fed by an optimistic source. The example of medical science has indicated that certain types of distress at least can be alleviated, eliminated, or prevented by alleviation, elimination, or prevention of an underlying pathology. The consequences of ignoring pathology or its early indicators (for example, cancer) are well-known, as are the consequences of deferring treatment until the crisis is upon the patient. The pathological orientation, accordingly, derives strength not only from this optimistic medical tradition, but also from the pessimistic humanist source noted, from ethical imperatives on a professional to provide what a patient seeks for a patient who defines his outcome in terms of distress, and from other human experience as reported first hand and refined by our literary tradition. The pathological orientation has profoundly affected our literate culture.23 Its widespread acceptance is implied by W.H. Auden’s characterization of our period as The Age of Anxiety.

The pathological orientation has also profoundly affected our social institutions, the activity and training of relevant professionals, and the related scientific cultures and traditions.

Complex social institutions have developed in response to pressures to alleviate the distress which behavior patterns can produce. The solutions have classically been couched in terms such as alleviation, elimination, or prevention of distress, protection of the individual or social system against the patterns, and so on. More recently, efforts have been made to redefine the patterns as acceptable. The distress is then attributed to the effects of social reactions to the patterns.

Where the distress is individual, it may be conceptualized and defined experientially or in terms more observable by others. Examples are terms such as anxiety and depression.24 Individual distress may also be conceptualized physiologically and, indeed, the ease with which experiential distress can be altered chemically requires no elaboration. Or distress can be conceptualized as societal and defined through observables which are considered their indicators. Examples are terms such as anomie and alienation.

The various frameworks of distress have influenced each other considerably, but the heavier conceptual traffic has been from the individual approach to the social, in our society at least.25 We are thereby bequeathed with the concept of social pathology and its accoutrements, e.g., social health, social disease, social cure. These concepts, borrowed from individual pathology, have been used to explain and absolve that pathology, e.g., it’s the society that’s sick, not him. Indeed, those schizophrenics whom one framework would regard as mentally ill and deviating from normal, another would regard as superior in perception and insight to the sick society from which they have wisely withdrawn. There has also been considerable conceptual traffic between physiologically and psychologically defined frameworks of pathology. One direction is exemplified by the effort to find organic and genetic determinants of pathology. The reverse flow is exemplified by psychosomatic medicine. Nor has the controlled animal conditioning laboratory been disregarded. Examples are the production and analysis of animal neurosis, experimental neurosis, and comparative psychopathology.

Each of these approaches, singly or in combination, has generated considerable research and theory. These have been devoted to the conceptual, methodological, and practical problems involved. The approaches and institutions involved are among the major contributors to the prevalent acceptance of the pathological orientation.

The social contingencies which the eliminative approach rationalizes are beyond the scope of this discussion. The relation between academic socialization in the social sciences and the reinforcement given to finding inadequacies in the work of others need not be elaborated: “critical” comment is raised to a high level with regard to academic theory and research as well as the social scene.26 The existence of more options in the set of solution-irrelevant elements than in the other set facilitates, of course, the finding of inadequacies. Any investigation must have omitted from consideration more variables than it included. And focus on the larger set also makes possible, as was noted in the example of the obsessive patient, extended discourse, research, and publication.27 None of the foregoing should be interpreted to mean that extended discourse, research, and publication are unnecessary. The need for them is more critical than ever.

In all events, the pathological orientation is by now so prevalent that it is generally considered as self-evident, as though conceptualized among “the laws of the Persians and Medes, never to be revoked” (Esther, 1:19). Indeed, some of the patterns involved have at other times not been considered pathological, and may not be so in some other cultures. In our society and our time they are so considered, and the term mental illness (emotionally disturbed, etc.) can be considered as a metaphor implying treatment by means and agencies used for other illness. Indeed, the diagnoses, or classifications, are based mainly on patterns considered undesirable and therefore to be eliminated, controlled or overcome. Thus hallucinations, delusions, and thought disorders enter into the diagnosis of schizophrenia and labeling as a schizophrenic. Crippling fears enter into the phobic disorders. And so on. Persons may be viewed as disturbing others by the face they present and what they do-as well as by what they do not do. And, as has been indicated, learning how they may present themselves or behave in manners satisfying to themselves and others is not readily come by through focus on what is currently wrong. Given this discrepancy, solution may be elusive, and ascription of the patterns to a pervasive pathology whose outbreaks are unpredictable makes sense. As was noted earlier, vagueness of definition of pathology enters into the rationale for total institutions and the antinomy they pose for the larger constitutional system. A constructional model may deal with some of these issues otherwise and arrive at different conclusions, which do not always imply the antinomy, in some cases. This will be considered in greater detail in the next section

The conditions for choice of a constructional approach, a pathological one, or their combination, have not yet been systematically explored. Given the possibility that they may yield different cost benefit ratios, depending on the conditions (in which term I subsume the type of problem), the current preponderance of one approach over the other may have produced unnecessarily high cost-benefit ratios. These may contribute to (or at least provide arguments for) current challenges to the present system. Among these is the assault on the mental health professions being made not only by administrations on federal and state levels, but also by groups to their political left, by those concerned with civil liberties, as well as by such consumer-oriented groups as Nader’s. This suggests a third parallel to the eliminative self-evaluations and therapist evaluations noted. This is an eliminative societal evaluation of professional practices. The recurrent emergence of new forms of psychotherapy, the continual reformulations of classic forms and approaches, and the apparent market for simplistic restatements of these and of behavior modification may also derive from dissatisfaction with the cost-benefit ratios presently obtaining. In those cases where the alternatives proposed are simplistic, the high cost of the present system may have been attributed, in part, to the complexity of its formulations. And recently being added to these challenges are those posed by the constitutional critics. On the one hand, institutions are being ordered by courts to provide the treatment which is their rationale, and on the other, their funds are cut, as are funds for professional training.

The constructional-pathological pairing: – Although much of what is now going on is new, the general problems have been discussed for some time. Indeed, professionals themselves have been among the sharpest critics of their own systems. In an effort to pin down good guys and bad guys, or otherwise to dichotomize different approaches, various paired terms have been offered. Often, as was noted earlier, these vary in the term which they oppose to medical, but they share this common term as the discriminandum.

The pairing I am suggesting is constructional-pathological. The terms may be considered to be apposed terms, since similarities can be described (they both relieve distress). However, they call for different data bases, which can be expressed in terms of different outcomes, starting points and diagnostic formulations, and are not comparable. The extent to which they represent antipodes or antinomies, or are orthogonal can not be assessed at present. If they are orthogonal, then the implication is that we must describe problems, their solution, their history, and their understanding in terms of intersecting coordinates, one of which can have a zero value. The assumption of orthogonality seems to present the fewest difficulties at present.

The pairing is not congruent with many others currently proposed. The term that everybody’s concept would most like to be paired against, medical model, is not congruent with a pathological orientation. Seated as I am in a wheelchair, neither I nor my physicians are interested in eliminating my paraplegia. We would like to program walking. And we wish that neurology would more rapidly learn how to grow nerves. Nor is medical opposed to environmental. Much of current cancer research is concerned with environmental variables. These may be viewed as producing pathology. On the other hand, they may maintain organic controls over regulated cell growth. Terms which are frequently paired with medical, such as preventive, psychological, social, educational, behavioral (cf. Cowen, 1973) can contain pathological as well as constructional approaches, as in preventing illness or maintaining health, eliminating defensiveness or creating self-esteem, overcoming discrimination or establishing fair employment, stamping out illiteracy or teaching Russian, eliminating homosexuality or establishing cross-sexual relations. Terms such as diagnosis-treatment have been proposed. However, assessment can be for constructional purposes and treatment for pathological purposes. Furthermore, one can diagnose a problem in terms of what needs to be done. Within the psychological domain, the mentalist-behaviorist, or psychodynamic-behaviorist dichotomies are not applicable since each contains both pathological and constructional formulations and procedures. Finally, within the behaviorist domain, the therapy-modification or respondent-operant dichotomies present the same problem, as does the term program. Within his framework, Wolpe has pioneered both desensitization and assertive procedures, and Lang is computer-programing the former. Consequential control over behavior can be aversive as well as reinforcing.28 It has been noted that constructional and pathological orientations are comparable when viewed from the outcome requirement of the latter, namely, elimination of a pathology or alleviation of distress. Viewed from the outcome requirement of the former, namely, construction of an outcome, they differ. The data bases of the two differ, as do the procedures utilized. This creates problems in translation of concepts and transfer of procedures across orientations. Accordingly, given two models, each of which combines both constructional and pathological elements, it would seem that communication between the two models is facilitated when comparable elements are related-for instance, the constructional elements of one model and constructional elements of the other. How a practitioner operating in one model constructs a repertoire may be useful in the construction of a similar repertoire by a practitioner operating in another model. However, how one eliminates a pathological element may not be useful in the instatement of a constructional element-within the same model, or across models.

Accordingly, translation and interchange may be facilitated by attention to constructional and pathological elements in models. This would also hold for patterns of a given person or institution. It is fashionable to denounce mental hospitals as producing institutionalization, or the exact opposite of what the implicit social contract calls for, and to decry their incompetence. How do we make them competent? Such analysis seems difficult, and the changeover seems costly: more staff, etc. Furthermore, such attack generates counteraggression by the institution and professional agents involved. The likelihood of change is lessened. If however, the institution is viewed as competently programing the undesirable outcomes observed, different consequences may ensue. Stated in constructional terms, the procedures used by the institution are competently establishing the outcomes identifiable by the patients’ repertoires. The program is an implicit one-it would require extreme cynicism to regard it as explicit. The analytic task then becomes one of making the program explicit, and harnessing the competent procedures to produce different outcomes, which are in accord with the social contract, as was noted in the earlier discussion of individual cases. Just as the institution can learn from its own effectiveness-if it is viewed constructionally-so, I believe, can professionals using different models learn from each other if they view professional behavior constructionally.

A specific model for research and intervention will be presented, which is constructionally oriented and falls within the operant-behaviorist tradition. Working details of the model, that is, its specific applications to contingency change, are not presented but are reserved for other publication-our main concern here is conceptual and ethical. Because the model follows the operant tradition, it is concerned with the development of validatable procedures with individual patients treated on a long-term basis (or groups so treated). For the same reason, the procedures are not validated only at the end of the program, but as they move along, in terms of fine-grain relations. Operant laboratory research has not only been used to shape and establish patterns, but also to investigate functional relations-that is, lawfulness. Stated otherwise, it can be used to contribute to knowledge. Indeed, this is its major objective-for us. For the patient, it is attainment of personally-desirable goals.

Because the model is constructional, I believe that its procedures can be transferred and used by professionals working within other models, where these procedures are examined for constructional relevance. In developing the model, we have insisted that our aim is not to develop a new therapy which supplants others, but is to make explicit what goes on in psychotherapy or treatment, however named. I believe that for the present, at least, our major contributions lie in this area. We do not intend to supplant other models not out of any great sense of humility but because we choose to view the professional work of others from a constructional orientation. Thus, we can very selfishly learn from their successes.

The model to be proposed has one other important feature. It accords completely with the constitutional requirements met by most of medical practice. These include contracting with the patient on outcomes and procedures we both consider worthwhile, rather than considering the patient as a third party who is to meet ends defined by two other parties, using procedures they set. The contract also requires explicit statement of our areas of concern, and requires our being limited to them. All others are reserved to the patient. The total institution, it will be recalled, differs in these two important respects. Our stance was not taken simply to accord with constitutional requirements. It derives from the analytic and therapeutic necessities imposed by the model.


While the experimental operant laboratory has been cited as the major source for the procedural and conceptual requirements of clinical and other areas of social intervention, it may be more instructive to consider the contributions of programed instruction. Stated in oversimplified terms, we can view the therapist not as a reinforcement machine, but as a program consultant, namely, a teacher or guide who tries to be explicit.

Programed instruction (p.i.) derives, of course, from the experimental operant laboratory. The four basic elements of a good instructional program are found in the behavior-shaping or stimulus-fading procedures of the animal laboratory. Indeed, p.i. has served to articulate procedures used to bring animal behavior to some researchable level. While suggesting new procedures for the laboratory,29 p.i. did not get off the ground floor until this laboratory technology was applied to it. Fittingly, the first programed text, the Holland and Skinner (1961) program, expounded the analysis which was derived from such laboratories. We can state, using the same criteria whereby we infer these from human behavior, that we can teach animals to be creative, to abstract, to conceptualize, to think, to develop and apply insight to solve new problems. Indeed, such demonstrations serve to remind us that the human equivalents may also be products of similar programing in the past, and may suggest explicit procedures to remedy deficits. However, our repertoires as programers in the laboratory are quite limited when compared to our repertoires as functioning members of social systems. The abstractions we program in the laboratory are far less complex than the abstractions and types of control found outside. These involve the use of language and of other socially-established repertoires which p.i. utilizes and taps. And as I tell my classes in the introductory lecture on experimental analysis, when I discuss applications and cite p.i.: “Yes, Virginia, behaviorists can teach people new concepts and orientations and ways to think, and their basic research includes these areas.” Similarly, in programing self-control, we can teach people new insights into the solution of their own problems, can investigate the means whereby the outcomes are produced, and can make reasonably shrewd inferences about the development of their initial presenting problems. The insight-therapy behavior-therapy dichotomy is a false one, and the treatment-understanding dichotomy can also be a false one.

Although the p.i. model can be closer in some areas to the social programs discussed than the laboratory model, it is not congruent with them. Whether this is because the social areas will require models of their own, or because they are not as explicitly developed, can not be answered at present. With the exception of a few programs, for example, the Foxx and Azrin (1973) and Azrin and Foxx (1974) programs to establish toileting, and our own fluency program and its related programs,30 most clinically-oriented programs have not explicitly specified and standardized each of the steps between entry and target repertoires in the manner of p.i. Rather, in this respect, they more closely resemble the hand-shaping procedures of the laboratory. Here, although the target is known and the general sequence can be specified, the procedures are not so explicit that they can be automated, and there is considerable room for variation and invention, depending on a variety of factors, including the outcome desired. Laboratory procedures also include those involved in auto-shaping, which is suggestive of clinical self-solution.

Nevertheless, the p.i. model is close enough for my present purposes, which bear upon constitutional issues. Before going into these and other extensions, the main differences will be noted between a model dictated by the requirements of p.i. and the pathological model which presently prevails. I should like to reiterate, for reasons already noted, that the issue is not the superiority of one model over the other, nor of acceptance or rejection of one at the expense of the other. Both models (and others, as well) have been developed by sensible people who are sensitive to la condition humaine. The model to be suggested is offered because it is closer to the procedures common to p.i. and the associated laboratory-based conceptual system. The introductory sketch will be organized according to the four program elements mentioned. It will extend the specific requirements of this model to the general comments made in the parallel comparison of the con-structional and pathological orientations.

1. Target or outcome: – Being explicit about the target or outcome is, of course, the first order of business in p.i. The target is usually suggested by the title of a programed text, for example, Neuroanatomy (Sidman and Sidman, 1967), or Really understanding concepts (Markle and Tiemann, 1971). Indeed, examination of Hendershot’s (1968, 1973) encyclopedic catalogue of programs in print is highly instructive: each title typically refers to that repertoire which the program is intended to establish, or construct.

With rare exceptions, the title does not refer to the overcoming of a deficit, nor the elimination of troublesome patterns. Possibly the student is buying Neuroanatomy because he wishes to overcome his ignorance in the field, or because he wishes to eliminate the crippling anxiety he currently feels over keeping up with the course. This may keep him from cracking the standard text, or he may be overwhelmed each time he opens it and this puts him further behind, etc. These presenting complaints, either of deficits to be overcome (inadequacy, ignorance) or of disturbances to be eliminated (anxiety, bibliophobia), are usually genuine. They may be profitable ways for the professional (and patient) to conceptualize the problem, but an alternate constructional model is suggested by p.i.

2. Entry behavior, or current relevant repertoire: – The constructional property of the outcome in p.i. dictates what the starting point should be, namely, those successful repertoires which the purchaser already brings with him. The program will be built on these. In a programed text, these are stated in an introductory comment which typically follows the title page. This informs the reader (or instructor) that the ensuing program presumes mastery of a specified prerequisite, since it will start out at that point. A set of criteria may then be provided to see if his background meets these requirements, or those of a less advanced or even more advanced text. As we apply this to personal and interpersonal problems, we attempt to ascertain the relatable skills in the client’s repertoire. And this will include a past history-of successful patterns and solutions.31

3. Sequence of change steps: – This sequence, of course, is the printed text of the programed book. In a linear programed text, each successive step (frame) is a miniature program containing the elements of the larger program in which it is embedded. The attained target of the preceding step is the current repertoire. The step itself consists of a behavioral requirement which either differs somewhat from the requirement at the preceding step (shaping), or is identical to the preceding requirement but under different stimulus control (fading), or both. This linear model, conceptually elegant and useful as it is in a text, is often too simple for the requirements of a social program. Indeed, even in texts, the program may branch, recycle, may provide options, or be open to original contributions or other unexpected developments and capitalize on them. The reader is referred to the p.s.i. movement (personalized systems of instruction, Keller, 1968; Sherman, 1974), for other extensions. Differences between these extensions need not concern us here. What is of concern is that the p.i. (or p.s.i.) program tends throughout to be constructional. When deficits and patterns considered inappropriate occur, the student may be referred, depending on the type of program (linear, branching, etc.), to an earlier point in the program, to another unit, or to another source to construct the desired repertoire.

4. Progression-maintaining consequences: – The opening words of the Holland and Skinner (1961) program introduce a quotation from Thorndike and Gates, some 30 years earlier: “If, by a miracle of mechanical ingenuity, a book could be so arranged that only to him who had done what was directed on page one would page two become visible, and so on, much that now requires personal instructions could be managed by print” (p. v). The teaching machine, of course, so arranges things admirably. Where no consequences other than presentation of the next unit are provided, what maintains progression? Why bother? In a well-defined program, successive delivery of successive steps constitutes progression toward the outcome, and such delivery (viewed as a stimulus) or such progression (viewed behaviorally) may be considered as a maintaining consequence for advancement-providing the program outcome itself serves this reinforcing function.32 Progressive mastery of a course or of the psychotherapeutic outcomes itself becomes reinforcing, and no tokens, points, M&M’s, or other extrinsic reinforcers are then needed. Need I point out that behavior analysis does indeed utilize intrinsic reinforcement?

It is when the program outcome does not (or can not) on its own serve this function that extrinsic consequences find use, either in addition to or instead of the intrinsic consequences. Where additional consequences are employed, they may derive their reinforcing property through linkage to a potent back-up, as when money, tokens, or points can be exchanged for luxuries or necessities; or they may in themselves be potent, as when M&M’s, cigarettes, or food are used (subject, of course to the conditions which make other consequences potent, e.g., deprivation). They may also substitute for the intrinsic consequence, as when scientific publication is maintained by promotion or avoidance of dismissal. It is customary in many quarters to lament such contingencies, and they are considered less desirable than the (intrinsic) program-specific consequences noted. However, they are often necessary. For example, the production of steel is vital to the well-being of practically every citizen, including the steel-worker. Yet for how long will steel production be maintained by interest alone? Possibly in besieged Leningrad, or in the battle of Britain, but not otherwise.

In a programed text, the student himself controls delivery of the reinforcer, and a true contingency relation does not hold.33 This poses conceptual and procedural problems, but these are not critical to the present discussion. The difference between the programed text in the student’s hands and the machine (or tutor) program for the same material resides in the agency which defines the response required for delivery of reinforcement. In the case of the latter, definition is independent of the subject, whereas in the case of the former, it is not. On the basis of this distinction, it should be noted that constructional clinical practice can be closer to the teaching machine and laboratory contingency than it is to the textbook type of contingency. In a social program where, say, the presenting complaint is the behavior of a problem child and the parents are the clients, the reinforcing events are the child’s changed behaviors at home. The child is the defining agent. If he reciprocates their changed behaviors toward him, he may thereby reinforce their changed behaviors and analyses, which are the targets of the sessions. In a self-management program where, say, the presenting complaint is prolonged scratching which produces and aggravates skin lacerations, skin healing is contingent on target behavior, and the skin, so to speak, defines this requirement. Both examples, the social program and the self-management program, define true contingencies.

The more conventional pathological approach may, of course, also produce changes in others which have the effects desired as well as self-management. Indeed, the relation of the changes produced to the patient’s behavior, problems, and insights often form a considerable part of the discussion. However, they are not as systematically articulated nor as systematically woven into the program as they are in the p.i.-laboratory model. The locus of maintenance of patient progress may (therefore) be assigned to therapist, patient, or interactional variables, such as transference, rapport, or other “therapeutic relationships.”


The model which has just been presented rationalizes, I believe, much of what is currently going on in applied behavior analysis (as distinguished from what is being called behavior modification or behavior therapy).

One of the major contributions of the experimental analysis of behavior has been its explicit formulation of procedures for construction of repertoires and their maintenance, along with a functional analysis which has provided means for further development. It was these successes that suggested application to human problems, where the issue might be expressed in terms of construction of repertoires. Such early successes, and the seemingly simple means by which they were attained, undoubtedly led to the mushrooming of the field. Indeed, until very recently, the popular press tended to associate B. F. Skinner with spectacular results in training animals, for example, teaching pigeons to play ping-pong. This involved the construction of a hitherto unheard of repertoire, hence it was newsworthy. Also newsworthy was the ability of a psychologist to do so well in training.

The rarity persists. As Hilgard and Bower (1966) noted: “It is not wise to dismiss [‘the animal stunts’] as merely signs of cleverness on the part of the trainers. These practical demonstrations serve as important empirical supports for certain aspects of the system – a kind of support very much needed for learning theories, and notably lacking thus far. No other learning theorist has been able to train an animal before an audience in a prompt and predictable manner . . . [thereby] epitomizing the principles of his theory . . . . [Other] demonstrations have usually relied upon exhibiting the results of earlier training. By contrast, Skinner’s pigeons can be brought before a class and taught various tricks before the eyes of the audience” (p. 144, emphasis mine – I.G.). At that time (1966) they noted that “most striking results” had not only been obtained in “animal training” but also in “programed instruction.”34 Among the commonalities of these fields is the construction of repertoires, and the deployment of positive reinforcement in this process. Certainly, no text that relied on shock elimination to maintain progression would sell. Since that edition of the book, as the reader is aware, further extensions of Skinner’s systematic approach have been made. Examples are behavior modification, behavior pharmacology, biofeedback, and social analysis.

The thrust of the quotation still holds. We are highly skilled in developing new patterns of behavior and teaching new understandings (p.i.). We are skilled in doing so in an explicit and precise way that enables us to learn what was at work. We thereby increase our own constructional repertoires and our understanding. I submit that we might learn from this. Control of behavior by punishment, by threat, by blackmail, or by other coercive means is as old as culture and may add comparatively little to behavior technology. What has been learned is how to deploy advances in physical technology for these age-old practices.

In the present section, I shall indicate how we have applied the constructional model described to one particular setting, namely our own service-research unit and its requirements. These requirements may differ for other units. Accordingly, the particular instruments I shall discuss are not presented for their universal applicability, but rather as illustrations of how one application of the constructional model meets constitutional and ethical requirements. Hopefully, it may spur planning along these lines elsewhere.

The instruments to be described derive from the requirements of a university system of hospitals and clinics35 with a strong research reputation, both basic and applied. The institution has therefore, in many areas, pioneered new methods of successful intervention for patients and trained other professionals. In the process of providing such services, we have attempted to be continually explicit about the repertoires we were applying, the repertoires the patients were applying, the patients’ relation to the resources available to them in their more typical ecology, what changes were required, and the functional relations between these. The research purpose is evident, but I regard these as fulfilling therapeutic requirements as well. If the patient can use a similar research approach, he may better assess the contingencies of his life-and teach us. Our research aim is furthered. We both hope to gain insight into the contingencies which govern his repertoires, how to change the contingencies, and how to assess them. Special instruments had to be developed for these purposes. Their development and assessment were governed by the constructional rationale described in the preceding section.

The Constructional Questionnaire: – The initial interview after acceptance to our services is guided by a questionnaire developed to obtain data for each of the four program elements described: targets, current relevant repertoires, change procedures (often strategy at this stage), and available and potential supports or reinforcers for maintenance through the program and thereafter. The questionnaire is presented in the appendix to eliminate the digression necessary to describe it in detail. My focus in using the questionnaire is on ascertaining the critical reinforcer, namely, what the patient is after, which I regard as presently pertinent for three reasons.

First, if we can find out what the patient is after, and if we agree to help get it (we need not agree; this will be elaborated in the discussion of contracts), progression toward this goal will serve as the program reinforcer. Extrinsic reinforcement in the form of tokens, points, etc., is then not necessary, and concentration can be on mutually agreed-upon goals, the means for whose attainment can be as clearly relevant to the social contract as are the ends. The progression requires record-keeping vital to (both) our interests. The patient may then readily assent to other requirements we both agree on. Renegotiation is expected, and coercion is absent.

Second, in my discussions with others, I have been repeatedly asked questions such as: “But how can you tell what the patient is really after?” T submit that in most cases this is readily evident-if one asks the right questions or observes appropriately.

Third, if we try to ascertain what he is after constructionally, we can more readily bring to bear those constructional, procedural, and analytic skills which are, at present, our important repertoires. Anyone can eliminate behavior if he sets himself to it. People have stopped stuttering overnight on impelling occasions. Switching to appropriate juncturing is not something anyone can do readily. We teach this very competently. Anyone can go on a diet and lose weight. Developing satisfying eating patterns is a different story. A constructional target must harness constructional data. Our graphs will therefore be of acquisition, maintenance, and related conditions. This is something any operant laboratory psychologist should feel at home with.

The first series of questions deals with outcomes. The first question in the series is direct: If we were successful, what would the outcome be for you? Typically, this is answered as relief from described distress, or elimination of a problem, but given the prevailing culture, we regard this as a sign of responsiveness to it. Patients classified as psychotic, inpatient or outpatient, may give answers similarly responsive: “I’d be the Virgin Mary”-a ten-year inmate of a state hospital, classified as paranoid-schizophrenic. My retort (puzzled) : “Gee, I don’t presume you’d have been dead these two thousand years. Must be something about the Virgin Mary that sends you?” Answer: “You’re darn right. Mother and Child. Mother and Child.” The patient had no children and her husband was a pimp who sold her services. She is tied to him for a variety of good reasons, and wants nothing more than a normal family with a husband as constant as the Father of the infant Jesus. Psychotic wish? (This was an outcome we could not help provide, given the available resources.) This information was readily inferred from the rest of her responses which resembled those of anyone else.36

The second question in the series attempts to redefine the outcome (if it has not been so defined) in observable and constructional terms. We have found the Martian observer, through whose eyes the outcome is seen, to be highly useful. (Reasonable substitutes can serve.) What does he observe which can be punched on IBM cards and analyzed by the computer? Prompts and corrections are given (“I’d be happier”-“The Martian can’t observe comparisons nor read your mind. How does it show?”-“I wouldn’t yell at my wife”-“How can he observe what you don’t do?”).

The second series solicits what is going well or is to be excluded and those changes which can be by-products of the program changes. The first question sets an immediate limit upon the program consultant. One smoker stated that he knew that relations with his wife were critical. He stipulated that this was not to be considered. Most clients report receptiveness to any area deemed relevant. The second question is set up for hidden agendas, among other data. The answer can also indicate potential reinforcers. (An obesity problem: “I’d be playing basketball again.”)

The fourth series solicits repertoires presently (or previously) available and possible stimulus control. As one patient, who had entered in a deeply depressed state, commented: “Do all your patients leave feeling this euphoric?”

The fifth series, on consequences, has been useful in fluency and related problems for research purposes. By and large, it has thus far not been needed to ascertain critical consequences.

Administration of the questionnaire may take from one to three hours, and more than one sitting. Only rarely is further information needed for the first stage. Although in most cases, it is remarkably simple to pinpoint the critical reinforcer, being able to help the patient obtain it is an entirely different story.37

With both outpatients and hospitalized patients, the symptom can be considered as a neon arrow pointing to the critical consequences-if it is considered a positively reinforced operant (see Goldiamond, 1969 for other suggestions on identifying the critical consequences). By the time the patient applies for relief, or is hospitalized, this operant can have become very costly, albeit still reinforced. If it is eliminated, and if the critical reinforcer thereby also becomes unobtainable, we may get “spontaneous recovery” of the operant or other operants may become established, some of which may be less desirable.38 Accordingly, we must specify the operants to be substituted by the program for the symptoms which are currently part of the contingency. Delivery of the critical consequences is thereby assured. The operants substituted are derived by agreement and are operants which, unlike the symptoms, may not be punished by others, but may actually be socially reinforced. For example, a young man with a shoe fetish of long standing also collected the unusable stockings (with runs and holes) as well as brassieres of his fiancee’s friends. He had been at the bottom of his class and possessed few job skills. In short, he had no place in the sun. The fetishes gave him a certain eclat. The therapy program consisted of the construction of job skills, among others. He was soon promoted and became an employee valued by others. The fetish and its related behaviors (which could jeopardize him legally) disappeared, without programed use of eliminative procedures. Social standing was established.39

Program Recommendations and Analysis: – The protocol is analyzed for a patient write-up. This serves the same function as the more conventional pathologically-oriented work-up, but is constructional. For reasons similar to those given earlier, its outline is presented in the appendix. Three comments will be made about the write-up.

First, it reverses the space ratio between pathological and constructional analysis typically obtaining in the patient work-up. It is very long on constructional elements, and short on pathology.

Second, where pathology is presented, it is presented as a strength, that is, a sensible operant. “X is so competent a librarian, that she has been promoted rapidly. She has also been elected as president of her local professional association. Her competence extends as well to …. Accordingly, she is so put upon by urgent requests from others that she has had little time to tend to areas of greater interest to her. She takes her work home with her. During the last six months she has begun to develop uncontrollable tremors, which she reports as anxiety attacks. As a consequence, demands on her are being relaxed”-but by means which jeopardize her interests.40

Third, history is important, but the history is of operants which have been shaped or otherwise developed. These obtain reinforcing consequences otherwise not available. A stutterer, for example, traced the background of his pattern to the stress he suffered during childhood; his parents were increasingly at each other’s throats, and their marriage broke up when he was seven. Specific onset was at five, when there was a dinner guest who stuttered. He recalled intentionally imitating the dinner guest. He discovered one day that the pattern controlled his speech-it was now involuntary. “What did your mother do when you started stuttering?””She was beside herself with fury.” One can argue tel père, tel fils, or, on the other hand, visualize a period when his parents’ attention is turned on each other, and withdrawn from him. The dinner guest is obviously listened to by the parents. The child imitates. Initial smiles are replaced, as the pattern progresses, by the full attention accompanying fury. This is not an ideal way to deliver attention, nor an ideal way to obtain it, but given the deprivation and circumstances, they serve. Data supporting interpretation of his pattern as an operant can be obtained from his history, discussed with him, and used to analyze and change the current contingencies of which his (expensive) pattern is a part in his speaking ecology (the laboratory program proceeds somewhat differently). Accordingly, the history of the operants is interwoven into the work-up.

Change procedures follow the initial statement. The tentative outcomes, stated explicitly, introduce the possible program. For the librarian, the outcomes might involve an explicit reordering of priorities. Learning how to turn down or defer low-order requests is a way which maintains the good relations she values (teaching her to turn things down in a way less costly than her symptom), as is revising her work arrangements so that she can do her work while on the job.

The remaining sections of the work-up follow the p.i. outline discussed in the model.

In some cases, one can present this work-up to the patient for his approval: Is this what you would like to see entered into your record? Do you have any objections or changes ? In one recent case, the patient had the write-up duplicated and presented it to his friends when they asked why he was seeing a shrink.41

Compared to the ease with which we can develop a coherent account using the more conventional pathologically-oriented work-up, we find that this task is often difficult and time-consuming.

The Contract: – The write-up forms the basis for a contract between patient and therapist. It suggests to the therapist what goals he might offer as possibilities. The patient may have other ideas, in which case negotiations may ensue. In all events, eventually a contract will be developed. The contract can be explicit, either written or verbal, or implicit, but the same rationale applies in all cases. Presented in the appendix is a form for a written contract to be negotiated between patient and programer. This lists the outcomes toward which the program is directed. Their establishment constitutes satisfactory termination of the sessions. It also lists other requirements. Contracts may be renegotiated upon call. I shall confine my discussion to two issues related to constitutional and ethical issues.

First, the fact that both parties must agree indicates that the therapist must be a consenting party. We have on occasion declined to enter into an agreement. This has occurred when we considered the outcome the patient was after to be illegal, unethical or dubious, non therapeutic, beyond our capabilities, or something we could not live with.42 An example of an outcome desired by a patient, but which we did not consider worthwhile, and the resolution, is given by a college student with a spider phobia. The products of the phobia had been, progressively, weekly meetings for two years with an assistant professor of psychology, who was considered one of the most eligible bachelors on campus (he kept the relation professional), sessions with an associate professor in student health (similar relations), and she now came to me. In classic terms, she would be described as an hysteric, but we considered this as an opérant which had been shaped by male attention. It had been so effective as to preclude the shaping of other ways of a maid with a man-her own peers. We felt that in the limited time available, programing should be directed toward attaining the latter end. After discussion, this outcome was agreed upon, and she was assigned a graduate student therapistfemale. The resolution exemplified is that involved in other successful negotiation. If the two parties do not agree on outcomes, what outcomes can be found which are mutually agreeable? In a very few cases, negotiation has been protracted.

Second, with whom and about whom is the contract? The potential for abuse is less, I believe, when the objects of the two prepositions italicized are the same rather than when they differ. The concern of the contracting parties is most properly with each other. When the United States and Mexico changed their boundary so that Comanche lands became American, the Comanches protested that they had not been a party to disposition of their lands, in possible contrast to the tribes with whom William Penn negotiated. I have seen commercially available forms, designated as “behavioral contracts.” These allow a mother to list for posting the behavior she requires of her child at stipulated times. Possibly, the listing makes a mother’s desires explicit to her child, and possibly this facilitates communication. Both of these may be desirable, but this hardly deserves being called a contract. I am not arguing against explicitness or against parental responsibility. My argument is a terminological one, not in order to have tidy semantics, but because words can govern other behaviors.

In our programs, an explicit contract is signed, and we are always a signatory. The other signatory is the “party of the second part”-whoever gets the services for which they have contracted. Marital contracting was not invented by behavior modification, but when the terms were agreed upon, the assenting partners to this contract were also in a contractual client-professional relation with their lawyers or marriage broker. As was noted earlier, when parents apply for a change of their child’s behavior, we contract with the parents to change the parents’ behavior. If we are successful, the parents will learn how to obtain increased delivery of reinforcers from the dispensing agent, their child. There is a practical as well as an ethical purpose to this. It has been argued that the parents are in far greater contact with the child than the (out-patient) therapist can be, that such contact will extend beyond the therapy sessions, and therefore that it is more parsimonious to train them to be the change agents for their child. This argument makes some sense, but I believe it should be extended one step. Each parent is in far greater contact with himself than anyone else can hope to be. Therefore, each parent should be trained to be the change agent for his own behavior. Where the reinforcing agent for such change is another person, say, their child, they must learn to “read” him. They must learn how to increase his dispositions to respond favorably to them, which occurs most readily when he gains thereby, as well. Learning to “read” under one set of conditions can carry over to other “books.” Typically, it does.

Of course, you may object, it may be easy for me to take this stance. After all, the patients we deal with are (a) more or less rational, are (b) outpatients, and (c) come of their own volition, that is, they are hurting and want relief.

The first two objections can be disposed of readily. Some of our patients have been extremely disturbed. One ambulatory schizophrenic had been an outpatient since childhood and an inpatient for one-third of his adult life. A major revision we have had to make for such patients is in the written requirements for contractual relations and for logs. While the system has not yet been fully applied to inpatients, I have interviewed, on a regular basis, patients in a state institution. Their responses are encouraging in that they make the same kind of good sense as do the outpatients with whom we have worked. We shall shortly be starting projects with other populations.

The third objection is a far more serious one. A partial translation is: What if the gains from the disturbing patterns outweigh their costs? A stuttering case exemplifies one outcome. A college student of 19 was so severe a stutterer that his face during speech became grotesquely distorted. Within one month on our program, he was fluent and facial symptoms43 disappeared. He then flew home, where he was to be part of our self-control program. He now became eligible for the draft, having been exempted because of his severe stuttering. The stuttering was reinstated. Shortly thereafter, another college student entered the program and was informed by an aunt, midway through the program: “Frank, you’re going to be eligible for the draft if you keep this up.” His comment was: “Auntie, I won’t be kept out by this means. I’ll take my chances like anyone else.”44 The two students were members of two different youth subcultures. To change the patterns of the former student would have required our changing his cultural affiliations. This was obviously beyond our capabilities and it was also not what he had come for. We learned from these experiences to ascertain prospective patients’ dispositions in this regard before starting.45

With outpatients, the answer to the question is straightforward and is given by the nature of a contractual relation between two consenting adults. How one deals with institutionalized subjects about whom a contract is signed by two other parties (institution and agent) will be deferred to the section explicitly devoted to constitutional issues.

Records and Intervention: – Intervention typically centers around records kept by the client and by us. These records vary with the nature of the outcome, and are prescribed in the patient write-up and contract.

The Weekly Program Worksheet, presented in the appendix, is filled out by the program consultant every week, during the session, and one copy is taken by the client. The subgoal lists each of the targets for the following week, stated constructionally.46 Each subgoal is numbered, and the same number codes the corresponding current pattern to be used as a base and the corresponding guide for change (program guide)-consequences to be harnessed may be listed here.

If the program is effective, the subgoals listed for next week will appear in the records of that next week. They should be in force. What maintains recording? At least two possibilities are suggested. One is a program-progress payoff. As the entries are discussed, there may be changes in outcome, and record-keeping is reinforced. The other involves maintenance, by the client, of the programer’s behavior. In different terminologies, keeping records is a demand characteristic, or part of a transference relation, or an operant which delivers therapist approval, etc. Indeed, we do have evidence bearing directly on this issue showing that the number of entries in one patient’s log in each of eight weeks was a direct function of the number of supportive comments the therapist had made the preceding week.47 If the number of entries can be a function of therapist requirements, can not their contents and forms also be so governed? How does this affect the validity or independence of the data?

A simple answer is that the verbal behaviors of a patient or subject do not cease to be operants, governed by all the variables involved in operant behavior, when the person becomes a patient or an experimental subject. When the psychoanalyst says that a patient has gained insight into his own problems, what he is describing is the patient’s analysis of his own behaviors and their determinants using analytic concepts (properly) in the process. Many of our patients also gain insight-into the contingencies governing their behavior. Patients under psychoanalysis can change with or without insight, and so can ours. In all cases, we tend to get out what we have put in. It is important that the influence of such variables be considered, and this has been at least one important service provided by the concept of transference. In all events, the pay-off is the cost-benefit ratio of the changes in the referent patterns in their referent ecologies. And a system which provides for continual explicit evaluation can facilitate this.

Presumably, validity of records, that is, their honesty, is also so maintained. However difficult it may be to check on the validity of certain experiences whose report the therapist reinforces, entries describing contingencies can be spot-checked-and we do.

Finally, it should not be assumed that there is no discussion of affect, or emotion. Most logs require such notation under a column headed “Comments.” We consider emotions neither as caused by behavior, in the James-Lange tradition, or as causing behavior, in the more classic tradition. We consider them as contingency-related. Often they serve to indicate important contingencies which have been omitted. A record which reports a particular pattern and its immediate reinforcement along with the comment “felt miserable” obviously requires closer scrutiny.

Initially, in accord with the pathological demand characteristics of our culture, the entries under Comments are of the distressful emotions. The contingencies reported in the adjacent columns are typically in keeping. Extinction, high cost, and punishment contingencies usually accompany reports of anger and fear, in accord with the laboratory literature on the emotional effects of such contingencies. Occasionally, atypical entries appear: a homosexual masturbated and a clinically obese patient stuffed himself after the occurrence of transactions describable as extinction and high cost contingencies. In all cases, affect is related to the contingencies and is used to teach the patient to uncover such contingencies in their inception and before they become controlling. Thus, the blushing of a woman increased until her face turned purple, at which point the others noted that their conversation embarrassed her and changed the topic. She was told: Your skin is more sensitive to the embarrassing trend of a conversation than your ears are. Heed it. When you start feeling hot, stop, look, and listen, and start changing the direction of the conversation then.48 A contingency analysis of emotions does not attempt to eliminate those emotions considered undesirable, disruptive, or distressful. It attempts to sensitize people to those emotions so they can be utilized to analyze and control the contingencies relevant to them and thereby to control these emotions.

In this stage, the patient uses the distressful affect to change the relevant contingencies. At another stage, he sets up the conditions so that these contingencies do not occur. He may also work toward setting up the contingencies related to the more desirable emotions-those we call pleasant and “constructive.”

The results and procedures deriving from our research and delivery system will be published elsewhere. The application of constructional behavior analysis to the social problems which we have faced has generated surprises for us in terms of the directions which the solutions required us to take. The experimental analysis of behavior is a new field, and its extensions to these complex directions even newer. Where the territory is uncharted, it would be surprising if we were not surprised. Stated in another manner, which is relevant to the next section, future discoveries in such new fields will confirm their old habit of disconfirming the predictions made by those who have not yet explored them. There is one sense in which I have not been surprised. I had always regarded behavior analysis as an orientation which is usable in the analysis of complex problems. With regard to the simpler problems with which it had earlier contact, it was useful only to the extent that it contributed precision and explicitness. And this I find still to be the case.


When I was a graduate student, the history of our field was summarized in a poem which may be familiar to most readers:

Alas, poor psychology, sad is her Fate!

First she lost her Soul, and then she lost her Mind,

And then she lost Consciousness.

Now all that’s left to her is her Behavior

And the less said of that, the better!

The poem portrays psychology’s divorce from certain philosophic concepts (and alliance or mésalliance with others). In like manner, the behavioral divorce from certain psychotherapy concepts (and alliance or mésalliance with others), has taken various forms. Among the earliest of these was (conditioned) reflex therapy (Bekhterev). Dollard and Miller (1950) attempted to consider psychotherapy from the vantage of classical learning theory. The application of its principles to clinical intervention was later designated as behavior therapy, with major conceptual strands from Pavlovian approaches, classical learning theory, the experimental analysis of behavior, and social learning theory, among others.49 The major contributor associated with the first two is, of course, Wolpe. These conceptual systems were extended by others to embrace aversion therapy through classical associative linkage of aversive stimuli with imagery or with other representations, although behavior-punishment contingencies might also be used. When procedures and formulations from operant laboratories, associated with Skinner, were extended to nonclinical as well as clinical areas, the question of a distinguishing name arose.50 This has been generally accepted by adherents as applied behavior analysis, for reasons noted earlier.51 The social learning approach is associated with Bandura, and among its conceptual origins is the earlier laboratory research by Tolman (e.g., cognitive maps, vicarious learning). The areas named have been included between the covers of Bandura’s scholarly Principles of Behavior Modification (1969). Krasner (1970) has proposed extension of this term from procedures relatable to learning and conditioning laboratories to applications from the more general field of experimental psychology. This hasty thumbnail sketch omits and slights many developments, as well as many controversies, both procedural and conceptual, between the families subsumed under the title. However, it serves to indicate the nature of the definitional problem. In contrast to psychology’s shrinking definition, behavior modification’s has been expanding.

Despite the difficulty of describing its exact shape, one should have no difficulty in describing what behavior modification is not. Psychosurgery, for example, is not one of the chapter headings in a textbook on experimental psychology, nor are learning psychologists licensed in its use. By the same token, many of the other practices currently ascribed to behavior modification do not fall within its domain.

Popular confusion stems partly from the fact that “behavior can be changed, or modified” by a variety of techniques, including drugs, hypnosis, aversive therapy, rewards and punishments, implanted electrodes, and psychosurgery.52 Since all of these can modify behavior, the popular press then labels them as behavior modification techniques. However, it should be noted that behavior can be changed or modified by psychoanalysis, Gestalt therapy, primal screams, lectures, books, jobs, religion. By the same logic, these must also be included in the definition of behavior modification techniques. Like the frog in Aesop’s fable, the definition has become so inflated, it has burst.

What partially underlies this particular confusion is a failure to distinguish between dependent and independent variables53 (or effects and causes) on the one hand, and control and analysis, on the other.

A dependent variable may be a function of a variety of different independent variables, and the same effect may be produced by a variety of different causes. The direction of motion by a sphere may be a function of gravitational forces, of remote control by radio, of control by the navigators within, or of other variables or causes. To designate all of these, therefore, as directional techniques, and therefore to assume some similarity between them other than the trivial observation noted is questionable. To designate all of them by the properties of one, e.g., attractional techniques, or of more than one, e.g., guidance techniques, suggests either confusion or sloganeering for ulterior purposes (“Travel with Interplanetary: our navigators are as dependable as gravity.”).

Because one can analyze the data by the same conceptual system does not mean that the same (conceptually derivative) control system is involved. Because the movements of the sphere and the flying of a kite may both be comprehended by the same scientific discipline, physics, does not make the boy flying the kite an engineer or other kind of applied physicist-although the navigator may well be an engineer. And the boy needs know no physics to fly or construct his kite. Indeed, when he makes a new one, he may strain the predictive and analytic knowledge of his physicist father.54

Accordingly, with regard to the first source of confusion, namely, dependent and independent variables, just because some procedure can be used to modify behavior does not make it a behavior modification technique. Psychoanalytic therapy and behavior therapy can both be used to modify behavior, but since they employ different conceptual schemes which harness their independent variables differently, they are not both behavior modification techniques. Behavior modification refers only to that body of procedures and conceptual systems derivable from experimental psychology or experimental learning theory. The reader is referred to Bandura (1969) for what these might be and to the sharply divergent approaches presented. Stances taken and procedures deployed by one school may not therefore legitimately be used to designate the stances and procedures of another.55 To take a more familiar example-because psychoanalytic and nondirective therapies both appear in a book called Psychotherapy similarly does not legitimize designating the stances and procedures of one by the other. In this article I am taking a stance for applied behavior analysis, which is one particular orientation and approach, and is only one among the range of approaches appearing between Bandura’s covers.

With regard to the second source of confusion, namely analysis and control, just because some sets of procedures can be analyzed in operant terms does not make them behavior modification procedures. These refer to the explicit and systematic application of procedures derived from the conceptual systems noted. Thus, the piano teacher may modify behavior, but, to date, behavior modification is not used for this purpose. Prisons have used isolation and solitary confinement (“the hole”) as punishment (“correction”). They have restored privileges (such as exist) contingent on behavior they have considered desirable. The restoration has been progressive, contingent on progressive change in behavior, or total, contingent on reversal of behavior. They have rewarded favored prisoners and made them trusties. All of this before the various behavior theories were heard of. What is often new is the justification of these procedures and the objections to them-in the name of behavior modification. The fact that the procedures can be analyzed in operant terms is irrelevant to the inappropriateness of the designation. Many other kinds of analyses can also be made, including psychoanalytic-indeed, such an interpretation has been given events in a concentration camp. Similarly, industry has made extended use of incentive systems. It rewards workers for their efforts with tokens (“money”) which are exchangeable for commodities and services. To hold Ayllon and Azrin accountable is ludicrous. Indeed, as Parsons (1974), an industrial consultant, notes in his significant (and long overdue) reassessment of the Hawthorne effect, “behavior modification” techniques “have not included worker performance in industry” (p. 929).

Yet another source of public confusion derives from psychologists themselves. Some have been subject to the confusions noted and have perpetuated them. Others have proceeded in a more original way. We have heard of acid freaks, of Jesus freaks, of Guru freaks. These are people for whom, if the Apocalypse is not imminent, salvation is, and it is immediately attainable by engaging in the practices of the group. We have tended to equate these movements with youth, but such visions have not been confined only to them (the acid movement was led by a psychologist over 30). Nor have the movements been confined to those explicitly concerned with altered states of consciousness. Such movements may also be oriented to the imminent solution of pressing practical problems. We might even consider the possibility of behavior modification freaks. Thus (the emphases to be made are mine – I.G.), “I believe that the day has come when we can combine sensory deprivation with drugs, hypnosis and astute manipulation of reward and punishment to gain almost absolute control over an individual’s behavior,” writes McConnell (1970, p. 74). If the public adds a few more terms to the combination, who can blame it? Further: “We should reshape our society so that we all would be trained from birth to want to do what society wants us to do. We have the techniques now to do it.”

On a less imminent note McConnell continues: “I foresee the day when we could convert the worst criminal into a decent respectable citizen in the matter of a few months-or perhaps even less time than that…. For misdemeanors or minor offenses we would administer brief, painless punishment, sufficient to stamp out the antisocial behavior. We’d assume that a felony was clear evidence that the criminal had somehow acquired full-blown social neurosis and needed to be cured, not punished. We’d send him to a rehabilitation center where he’d undergo positive brainwashing until we were quite sure he had become a law-abiding citizen who would not again commit an antisocial act. We’d probably have to restructure his entire personality.” McConnell is calling for total institutions-with a vengeance. His statements accord with my earlier analysis; namely, the “complete control over . . . (the) environment” he calls for is relatable (a) to the requirements of clairvoyant prediction of no further disruption after release, (b) which, in turn, is related to a model requiring elimination of an underlying pathology, (c) whose unpredictability, in turn, is related to a vague definition of that pathology in terms such as “neurosis” and the “entire personality.”

One need not evaluate the scientific evidence for the assertions made, nor their ethics, to question the tenability of the propositions. McConnell states that “the legal and moral issues raised by such procedures are frighteningly complex, of course.” They may be frightening, but it is questionable that they are “frighteningly complex.” They are manifestly simple: the procedures are clearly unconstitutional.56

Rather than describing such procedures as the shining product of the architects and engineers of the Brave New World that McConnell foresees, one can more simply describe them as an unimaginative extension of the deficiencies of the present system and a regression to an earlier era when “profane tongues were treated . . . by squeezing them in a cleft stick for as long as an hour” (Schwitzgebel, 1972), to mention one of the less drastic means then used. Indeed, society’s definition of legally treatable deviance (called “neurosis” by McConnell) has included Quakers: “A statute of 1657 (by Massachusetts) prohibited their entry into the colony, and provided that for the first offense a male Quaker could have one ear cut off (p. 268, emphasis mine – I.G.).57 Entry was criminal behavior to be extirpated; the passage and repeal of the XVIIIth amendment created and eliminated a class of criminals numbering in the millions; behavior which is criminal in South Lake Tahoe ceases to be so simply by moving both feet into Stateline (I am referring to the gambling laws of California and Nevada, of course). The size of the brain-washing caseloads would be matched only by the size of the unbrain-washing caseloads (as people crossed borders or changed the laws) in the system proposed!

I have devoted this much space to McConnell’s article not because of the standing of its author in the fields of science fiction (he has published here) or behavior modification, but because of the standing of his article as a target. The inflammatory nature of its title (“Criminals can be brainwashed-now”) and the inflammatory content (extensions are made to mental health) have met with equally inflamed responses. Although the article is probably the most widely excerpted article in its field-in the popular press-and although it is cited as a source for defining the procedures, rationale, and goals of behavior modification, or at least of a significant part of its practitioners, it does not speak for the field. Nor, I imagine, does it speak for a significant part of its practitioners. If this were the case, the field would indeed deserve the calumny being dispensed by leafleteers, and the bitter criticism being expressed by reasonable people.58 The consequences of the tone and message of the article and its reception have not only been its wide attribution, but also a concern translated into public policy. The concern is genuine, and taps into social contingencies which vary with the group involved.

The article refers to the creation of a Brave New World. In an earlier article, I referred to that work: “Brave New World depicts a society which is far beyond what we possess today, but not beyond the bounds of reason. A level of technology this advanced will contain stimuli that at present we can not even conceive of. . . . If our behavior [then] does not conform to the new stimuli but stays the same way it is today, we will become extinct more rapidly than did the dinosaurs . . . Inasmuch as I can not predict what future stimuli will be created, I can not predict what behavioral or societal developments will occur. I would suggest, however, that we keep our eyes open and try to understand what is going on, especially in the scientific community” (Goldiamond, 1965). McConnell is projecting all the defects of the prison system of today upon the events of the future. As I also noted: “The notion that behavioral technology will mean a prison state or manipulation of behavior on a total scale ignores some of the more recent developments in the experimental analysis of behavior and in self-control. When one starts to apply experimental analysis to practical problems, the procedures which develop in practice differ considerably from those which may be projected from a theoretical understanding” (Ibid).59

Indeed, to cite but three deviations from the expectations of McConnell’s article, Cohen and Filipczak (1971) set up a reformatory environment which increased the options available to delinquents and set as their targets limited and clearly defined set of repertoires (academic achievement). And Fairweather et al. (1969) set up a contingency system within a mental institution, with limited and clearly defined targets relatable to the patients’ requirements for normal life outside the hospital, and developed mutually supportive groups. And Keehn (Keehn, et al.) surmised that the critical consequence for skid-row alcoholism was skid-row community. The requirement for membership was alcoholism. Community was provided on a rented farm, with the members democratically planning goals and programs for each other. Membership was contingent upon meeting such goals and meeting Keehn’s contractual responsibilities of detoxification. The community began to develop services for neighbors and was on the way to becoming self-supporting when the project terminated.60

A carefully planned long-term project which provides measures enabling continual evaluation is one in which the possibilities for incremental knowledge are optimized-to the extent that one deviates from ages-old coercive procedures (which have little to teach us that we do not already know) and moves in a direction of mutual consent between contracting parties. This is a constitutional direction.


In discussing constitutional and ethical issues, I shall be guided by the four programing elements already noted, namely (a) outcome, (b) entry repertoire, (c) change pr