Orne, M. T. Perspectives in biofeedback: Ten years ago, today, and … . In L. White & B. Tursky (Eds.),Clinical biofeedback: Efficacy and mechanisms. New York: Guilford Press, 1982. Pp.422-437.



It is an exceedingly difficult task to discuss meaningfully a considerable number of highly diverse chapters, and as these chapters are incisive and some of them deal with their topics in a definitive fashion, it becomes presumptuous even to try to deal with them in a summary manner. Fortunately, each section has had individual discussions; further, Miller's keynote address not only serves to introduce the chapters that follow, but can also be reread fruitfully as the discussion. Both John Lacey and I felt we should address some of the more overall issues that emerge from this volume and are reflected in the field in general. In particular, it seems worthwhile to look at how we got to be here. This volume grew out of a research symposium on clinical biofeedback. Many of us were at a conference some years ago (October 20-22, 1969) in Santa Monica, California, where the term "biofeedback" became institutionalized.


At this conference, there were discussions about whether the phenomenon was to be called "operant conditioning," "self-regulation," or "biofeedback." Each of these terms focused on different aspects, and each had vocal proponents among the 142 individuals who attended the organizational meeting.

A strong contingent of scientists argued for using the concept of "operant conditioning," since much of the technology relevant to biofeedback has been derived from the technology used in operant conditioning. Further, the work of Miller (1969) and DiCara (DiCara & Miller, 1968) had shown that dramatic visceral changes

Martin T. Orne. Department of Psychiatry, University of Pennsylvania, Philadelphia, Pennsylvania; Unit for Experimental Psychiatry, The Institute of Pennsylvania Hospital, Philadelphia, Pennsylvania.




could be brought about with operant techniques, even in curarized animals. Similarly, the early work of D. Shapiro, Tursky, and Schwartz (1970) with blood pressure, and that of D. Shapiro, Crider, and Tursky (1964) with the electrodermal response, was presented within the operant-conditioning paradigm. This was equally true for the work of Engel (1972) in treating arrhythmias, and for Kamiya's research (1969) with alpha feedback training. Thus, there were a number of distinguished investigators, some of who had worked with animals and others who had worked with humans, who felt that "operant conditioning of visceral responses" was the appropriate terminology and would lead future investigators to think about the phenomenon in the proper fashion.

A number of participants, however, felt that the most important aspect of the new technologies was that they helped individuals to learn how to control bodily functions that were previously viewed as automatic or reflexive in nature. These participants objected to the term "operant conditioning" because it had overtones suggesting that individuals were somehow passive; whereas it was obvious that whatever mechanisms of learning were involved in gaining control over such parameters as blood pressure, heart rate, or brain waves also involved the individual's active participation and ultimately resulted in these individuals gaining increased mastery over themselves. Particularly since the mechanisms over which the individuals would seek to gain control were generally involved with reestablishing an appropriate homeostasis, the concept of "self-regulation" seemed to emphasize what, to many participants of the meeting, was the key factor.

Finally, yet another group argued that the notion of "feedback," already well established in engineering and neurophysiology, was the process central to learning control over involuntary physiological processes. The proponents of this view recognized that motivation was important, but cogently argued that, in the operant conditioning of human subjects at least, it was impossible to present the reinforcer without simultaneously presenting information. Since the motivation of an individual with frightening cardiac symptoms to control these symptoms is likely to be greater than any motivation created by some trivial monetary reward, it seemed best to speak about the effect of information on gaining control. Though such a situation could be conceptualized in operant terms without much difficulty (by acknowledging that the motivation to control the cardiac symptoms made certain information reinforcing), the majority of the participants were eager to adopt a term that lacked the psychological overtones of passivity associated with "conditioning."

It is also true that the studies that began carefully to address the relative importance of information as opposed to motivational factors for the process of acquiring control over physiological functions remained to be done; in part, this issue is only now being addressed in a basic way in some work such as that discussed by Brener (see Chapter 2). Further, the kinds of processes studied under the rubric of "biofeedback" typically involve a preexisting ability of the individual to assert volitional



control in one direction with a lack of such an ability in the other. For example, in the presence of light, it is very easy to block alpha; but it requires some training to learn to ignore the visual inputs and to generate alpha freely. Again, it is relatively easy for individuals to produce electrodermal responses on demand, whereas it is more difficult to block the production of electrodermal responses. Consequently, the usual operant-research strategy of comparing the ability to increase and decrease a response could easily be misleading in what has become known as biofeedback research. For a variety of reasons, then, the term "biofeedback" was deemed desirable. Since it was a new term, it did not have any previous associations to contend with; it sounded both novel and exciting, and it implied not only a new area of research, but perhaps even a new era of research.


It is difficult to imagine how recently the term "biofeedback" came to denote a field of investigation. It is worthwhile, however, to remember the excitement and enthusiasm that characterized the Santa Monica meeting only a little over 10 years ago, largely due to the expectation of the participants that they were helping to usher in a new area of research and clinical application. The emerging field of biofeedback was unusual in that it brought together psychologists, physiologists, engineers, neurobiologists, physicians, psychophysiologists, and enthusiastic laymen; all of these people shared the hope that man could learn to self-regulate not only blood pressure, heart rate, or brain waves, but also levels of activation and states of anxiety, as well as the hope that it would be possible ultimately to replace many of the drugs in current usage, including the ubiquitous tranquilizers, simply by training people to self-regulate. These did not seem far-out hopes, not only because we were meeting in California, but because it really looked as though there were data to suggest that rapid learning of psychophysiological self-control, aided by appropriate electronic devices, was a plausible eventuality.

The meeting out of which this volume grew was quite different. The organizers carefully selected participants to represent the major areas of work in biofeedback over the past several years, including a number of people who were at that first meeting. Much basic work has been done in the field since then. Perhaps almost as interesting as the issues that are discussed in this volume are some of the issues that are conspicuously absent.

At the Santa Monica meeting, perhaps over half of the attendees were there because of an interest in alpha feedback training, a phenomenon that seemed to provide a bridge between Eastern traditions of meditative discipline and Western know-how. Great enthusiasm was engendered by the publication in such august scientific journals as Time, Life, and Newsweek of promises of instant samadi, which



would allow people to gain in a few sessions benefits of meditation that would normally take a lifetime to acquire. Much of that promise appears to have been illusory, and it is interesting to note that no one in this volume discusses the clinical application of alpha feedback training.

Equally striking is the difference between the scientific issues that preoccupied basic science participants 10 years ago and those prominent today. The burning question then was that of whether operant conditioning of physiological parameters was directly possible, or whether some form of mediation inevitably occurred. While the reasons for discrepancies between early and later empirical findings in this area have yet to be clarified, the issue as such is no longer a vital one for the understanding of clinical biofeedback. Strikingly, the two major factors that led to the organization of the Biofeedback Research Society -- the work by DiCara and Miller (1968) showing huge physiological effects with the operant conditioning of curarized animals, which provided the emerging field with much-needed scientific stature, and the work on alpha feedback training, which was widely accepted as having been shown to have salutary effects on psychological states -- no longer form a significant basis for today's studies of biofeedback. Indeed, even the very concept of "biofeedback," which seemed so intuitively sensible, has been effectively challenged by Bernard Tursky's elegant discussion (see Chapter 5). While I believe that Tursky is undoubtedly right in his conceptual critique, the concept of "biofeedback" is likely to be with us for a long time to come. The research that has been prompted by the notion of "biofeedback" can stand on its own; and if some of the data that originally piqued our interest turned out to be different from what was once anticipated, investigators have learned enough that is new and worthwhile to justify continuing concern with these matters.

Having commented upon the appropriate absence of a discussion of alpha feedback training in the context of a symposium on clinical biofeedback, it is difficult to leave the topic without pointing out that even here some key issues have yet to be resolved. Paskewitz and I (Paskewitz & Orne, 1973) reported that feedback training to increase alpha density is easily achieved in the presence of light, but very difficult (if not impossible) to demonstrate with dark-adapted subjects in a totally dark environment. Further, contrary to our expectations, dark-adapted subjects exposed to the threat of shock and even exposed to actual electric shock failed to show a drop in alpha density, though they gave clear evidence of increased arousal behaviorally, by retrospective inquiry, and by dramatic increases in heart rate and spontaneous electrodetmal responses (Orne & Paskewitz, 1974).

Though these findings clearly document that alpha density changes need not accompany changes in level of arousal, further analysis of these data demonstrates that while most subjects showed no changes in alpha density in response to shock or threat of shock, some subjects showed decreases in alpha density (in line with the generally accepted view), while some subjects showed an increase in alpha density



associated with increased heart rate and spontaneous galvanic skin responses. In fact, the highest increment in alpha density (observed in a totally darkened room) was associated with intense arousal in a subject whose response to increased arousal is sharply increased alpha density. The observation that there are individual differences in alpha dynamics in relation to arousal is congruent with observations by Travis, Kondo, and Knott (1975) and others. (For a detailed discussion of these issues, see Orne & Wilson, 1978.)

I am trying to suggest that while alpha feedback training is unlikely to be a particularly useful clinical procedure, for selected subjects it could potentially be highly effective, while for some other selected subjects, training to decrease alpha might have the desired relaxing consequence. These comments are not intended to tout alpha feedback training; rather, since some of our work has been instrumental in suggesting the lack of a physiological mechanism that would justify the widespread use of alpha feedback training, they are intended to point out that even here the data are not all in. There may yet be a potentially useful phenomenon in this area, provided the basic mechanism can be elucidated and the rational application based on an understanding of the underlying processes.


As scientists, one would like to first understand the basic mechanism underlying any technique that is to be applied clinically. There is no substitute for such an understanding. The remarkable effectiveness of modern medicine owes much to such a systematic scientific approach. Traditional medicine has, however, also been an empirical art. A great many drugs were used centuries before their pharmacological mechanisms were fully appreciated. While it is considered essential today that any therapeutic approach provide a reasonable rationale for its effectiveness, there is often a wide gap between that rationale and clinical application. Behavior therapy, for example, justifies its existence by asserting that its procedures are related to findings of laboratory research. However, as London (1972) has pointed out, the link is often tenuous and even nonexistent, though the assertion has important heuristic value. Similarly, the rationale for biofeedback is sufficiently plausible and well-accepted to provide an umbrella for a wide range of therapeutic maneuvers that seem to be effective.

Since in many instances the systematic documentation of the mechanisms by which presumed effects occur is lacking, it becomes a central question to ask whether the therapeutic effects are specific or nonspecific effects. What do we mean by a "specific effect?" This has been incisively discussed by Jasper Brener from a conceptual point of view (see Chapter 2), and has been touched upon by others as well.

While, from a theoretical viewpoint, it is very important whehter or not bio-



feedback makes it possible for individuals to learn control over visceral or quasivisceral responses in a way that they could not do otherwise, this does not distinguish between whether such learning takes place primarily by way of an efferent or an afferent stimuli, or whether it is fed forward or fed backward. Further, from the point of view of application at least, the pathway is far less of an issue than the question of whether it is a specific effect.

I would like to distinguish here between important conceptual scientific questions such as, "Are we really dealing with feedback effects?" or "How is the effect mediated?" which are not key problems relating to the application of the phenomenon, and the question of whether there is a specific effect, which is central to clinical application. I believe that these types of issues tend to be confounded, and I intend to focus here on the issue of specificity.

When asking the question, "What is a nonspecific factor, as opposed to a specific factor?" one of the major problems is that what is nonspecific today becomes specific tomorrow. In other words, what is nonspecific relates very much to the present state of ignorance, and the task of medical science has always been to make whatever is known into more and more specific knowledge. To take a well-known example, the discovery of digitalis was made by a witch who picked foxglove at the new moon and mixed it with bats' wings and similar good things, which resulted in a concoction that effectively treated dropsy. Dropsy was believed at the time to be an illness, and was only later recognized as a symptom, now called edema. Of course, the concoction only effectively treated edema resulting from cardiac causes. There was the problem, however, of the specific as opposed to the nonspecific component of that concoction. It took some time to learn that it was not essential that foxglove be picked at the new moon; neither were the bats' wings necessary; and after many years of work, scientists now know that there is a specific effect of digitalis, which is contained in foxglove. It appears an easy question to resolve, but at the time that people had this concoction and asked, "Does it work?" they had to ask first, "Is the effect of this concoction specific or nonspecific?"

The first thing to establish is whether there is a specific effect of the concoction in its totality -- bats' wings and all -- and to do that, investigators must have an appropriate kind of control for the concoction as a whole. For example, early researchers might have tried to determine whether an equally foul-smelling concoction presented by the same witch is as effective. It might have turned out that it was important that the witch did not know which concoction she was giving the patient, and so on. Only after it was shown that there is a real effect on dropsy specific to this concoction could it have made sense to begin the analytical work to take this mixture apart and determine whether the new moon was important, whether the bats' wings were important, and whether the old woman with her convictions was needed or whether a doctor might be just as effective. These kinds of questions are secondary to having first established whether there is a specific effect worthy of inquiry.



It appears to me that many of the problems of researchers reside in both undercontrol and overcontrol. In other words, control procedures are often applied too soon, before it has been established that a "concoction" works as a totality. The first question must be whether there is a phenomenon worth investigating. Only then does it become sensible to engage in ever more careful control procedures, to dissect and analyze the aspects of the total procedure in which the phenomenon resides. Consequently, what is specific and nonspecific depends upon where investigators are in this sequence of research. In the early stage, they ask, "Is there a specific effect due to the totality of the concoction?" Later on, they ask: "Is there a specific effect due to the bats' wings?" "Is there a specific effect due to the foxglove being picked at the new moon?" "Is there a specific effect due to the kind of person who gives the concoction?" and so on. Much disagreement can be avoided if researchers are careful not to confuse the issue of where they are in a line of investigation with what they are trying to do.

The basic question of whether there is a specific effect due to the overall procedure has yet to be answered for much of the biofeedback literature. For example, in the biofeedback of epilepsy, it was first essential to show that there is a specific phenomenon that is not simply an attention variable, that is not simply a matter of being nice to these people whom everybody ignores as "crocks" -- in other words, to show that something is happening that is a specific effect above and beyond the general nonspecific effects. To answer this deceptively simple question is often an exceedingly difficult and complex problem. However, until this is done, it is often difficult to justify going on to dissect the process itself and to see what is there that accounts for the specific effect. Until the mechanism is understood, no one has a true understanding of the phenomenon. At the same time, it is appropriate, given the present stage of development of biofeedback, to establish first that there is a specific effect. Sterman's research (see Chapter 12) has gone far to show such an effect and is now appropriately exploring the specific mechanisms involved.


Much of the discussion of those of us oriented toward doing basic laboratory research must appropriately focus on mechanisms. Those of us who are more directly concerned with the application often feel that while they ultimately need to study mechanisms and that this is the appropriate way of going about it, it is first necessary to establish that the phenomena are real. How does an investigator go about that? It seems appropriate here to address the question of what constitutes an appropriate control. This question has been raised in several of the papers, and I do not intend to review the matter in a generic sense here, except to point out that an appropriate control can never be specified unless researchers have a theory that they seek to test.



There is no such thing as specifying an appropriate control in the abstract. A control becomes possible only after investigators are willing to specify, for the moment at least, what they consider to be the active ingredient they are studying. Unless such an active ingredient is specified, they cannot control for it; by the same token, as soon as they specify the active ingredient to be investigated, they implicitly specify what constitutes some of the meaningful controls.

In this context, I intend to talk about the placebo controls, because they happen to be an area of my particular interest. A very interesting comment has been made on how to compare the effect of biofeedback on headache with that of a placebo control (see Cox & Hobbs, Chapter 13); the effect of biofeedback on headache was compared with the effect of a placebo (in this case, a pill without an active ingredient) in a controlled study of headache. On the face of it, this is a very interesting and sensible control. Unfortunately, it ignores what a placebo is intended to do. In psychopharmacology, particularly, a placebo is intended to be a control that conveniently embodies all aspects of pill-taking behavior, with the patient's expectancies and beliefs and the physician's attitudes. This seems to be conveniently and easily done by giving an identical pill without an active ingredient.

It should be emphasized, however, that the effects of drugs as such are almost never studied. There are some studies -- very few -- such as that by Lyerly, Ross, Krugman, and Clyde (1964), in which amphetamines and barbiturates respectively were slipped into subjects' orange juice without their knowledge; they were then asked to rate how they felt. The investigators also gave these subjects the same drugs as pills without indicating what they were. It turned out that the effects were quite different when subjects knowingly took a drug from what they were when the subjects had ingested the drug without their knowledge. The mood changes were considerably larger when associated with pill-taking behavior, probably because they could now attribute them to a pill and acknowledge them comfortably.

Normally, the effect of a pill containing an active pharmacological ingredient is compared with that of a pill without an active pharmacological ingredient (placebo). Note that investigators do not study the pharmacological ingredient as such; they only study the effect of taking a pill with or without it. Another way of putting it is that a drug effect includes the placebo effect plus an active pharmacological effect. It is worth keeping in mind that any treatment procedure has placebo components which have powerful effects on their own.

The placebo, then, is a way of attempting to control for the totality of the nonpharmacological effects associated with receiving drug therapy. It is this concept that has generalized to the notion of the placebo effect of psychotherapy, as, indeed, there are profound placebo effects of psychotherapy (A. P. Shapiro, 1959). Similarly, there are placebo effects of biofeedback. Any time a treatment receives a great deal of publicity, and particularly if it involves elaborate equipment, investigators can anticipate profound placebo effects.



When researchers begin to examine placebo effects with care, it becomes clear that if a therapist knows which patient is receiving the active drug and which is receiving the placebo, a larger effect with the active ingredient almost invariably results. There is a simple reason for that. Doctors giving a new pill generally expect some effect, and they also are typically quite concerned about patients' reactions. If a patient comes to a doctor and says, "Doctor, I have a ringing in my ears, " and the individual is on the active drug, the physician will say, "Well now, come, I want to examine you." After a thorough physical, the doctor is likely to say, "Here's my home number. Call me if anything strange happens. " On the other hand, if the patient who is on placebo comes in saying exactly the same thing, the physician pats this patient on the back and says, "Don't worry, my friend. It will go away. " Clearly, these people are treated radically differently, and it is therefore not surprising that it is important for therapists to be blind as to whether they are giving drug or placebo.

As an example, I recall the first clinical trials of Marsilid. You may remember this was the first effective, widely used antidepressant, an MAO-inhibitor that was eventually taken off the market because of liver complications associated with its use. In the original drug trials, two patients came down with glomerulonephritis. They were, of course, immediately taken off drugs. It was only much later that it was learned that both of these patients happened to be on placebo. (If, by chance, both of those patients had been on the drug, it might never have been released.)

I participated in the discussion of a placebo biofeedback experiment that is discussed elsewhere in this volume (see the round-table discussion of Cox & Hobbs and Adams, Brantley, & Thompson), where I asked, "Did you know which individuals were receiving the active ingredient?" The reason I asked was because the patient who showed side effects was on the active treatment and therapy was, of course, discontinued. I would have preferred that the principal investigator hadn't known which patients were receiving the active treatment, because he might then also have discontinued treatment of one or two people on the placebo therapy who were complaining of side effects; this virtually never happens as long as the responsible physician knows that the patient is on placebo. It is for this reason that not only the therapist who is in contact with the patient, but also the physician who has overall medical responsibility, must be blind. Indeed, it can be argued that especially that individual must be blind, because he or she is going to make the decision as to whether a patient needs a little reassurance and can continue, or whether that patient should be taken off the experimental therapy. It is crucial that a physician not know whether the patient is or is not on placebo at that point, because differential treatment vitiates the entire double-blind procedure.

Another issue I take up briefly in the same discussion is that it is not uncommon for technicians to be blind but for the patient not to be blind. In one of the first studies comparing Thorazine to placebo, the medical staff, nurses, and attendants



were carefully kept blind. However, when I asked some patients whether they were getting drugs, they had no difficulty telling me whether they were or were not, because those on drug had a dry mouth, and patients do talk to each other. As one patient aptly pointed out, "I may be crazy, but I am not dumb."

How the patient may be kept blind in a placebo study on biofeedback is a huge problem. For example, the control most typically used is a yoked control; that is, one subject is run with biofeedback, and a second subject, who is "yoked" to the first subject, receives the feedback signal that has been displayed for the first subject instead of being given feedback; thus the amount of "reinforcement" given to these two subjects is matched. In the same fashion, each subject is paired or yoked to another, one receiving true feedback and the other receiving false feedback. In our early alpha studies, we found that even though subjects could not increase alpha density in a dark room, they were aware (typically within 10 minutes of trials) of when they were not receiving accurate feedback. They felt that there was something different. They couldn't spell it out, but they knew that something was wrong and they responded differently. There is little question that a simple yoked control is not appropriate with virtually any of the biofeedback procedures, because it is an inactive placebo control. It is closely analogous to the realization of the patient who is getting a placebo and doesn't have a dry mouth that he or she is not getting the active drug. If subjects get a hint that they are not getting feedback, then they will simply not respond.

The notion of what constitutes an adequate control in such a situation is a very tricky point. By the way, Lubar (see Chapter 11) has added a very interesting and very useful and appropriate aspect in his procedure by including a spot that remains responsive to muscle control. That is an excellent way of creating an active placebo. It is one of the very few studies I know of that has used a procedure analogous to an active placebo with biofeedback.

There are some other ways of attempting to create an active placebo in biofeedback research. One of them is to start out by giving subjects actual biofeedback and letting them experiment with it, and then gradually to degrade the signal until eventually they are presented with the same information that had been given to yoked controls. The problem here is that, as I mentioned earlier, the physiological response to biofeedback training is rarely symmetrical. That is, subjects typically only need to learn control in one direction; they inherently have control in the other direction. For example, with alpha feedback training, subjects could always block alpha with fairly good success, simply by attempting to see, even in a dark room. Again, with EMG feedback, subjects can always increase muscle tension; the training is needed to teach them to decrease muscle tension, and so on. Since subjects have volitional control in one direction, it is an easy matter for them to determine whether they have any control, simply by trying to block alpha or to increase muscle tension. Therefore, if researchers gradually degrade the signal, they are always faced



with the possibility that subjects may at any time decide to test their control and thus may readily find out that it is nonexistent.

An alternative that I find ingenious was used in a study by Otis, McCormick, and Lukas (1974). Employing EMG feedback, they set up contingencies in a way that when subjects tensed their muscles, they obtained correct feedback; but when they relaxed, they obtained random feedback. In this fashion, the subjects had control -- that is, the unilateral kind of control that people always have and don't need to learn -- but they could be given false yoked feedback for relaxation as long as the circuit was arranged to override the yoked control and give correct feedback for tensing muscles when a certain EMG level was exceeded.

The control employed by Otis et al. is a potentially useful approach to creating a viable, active placebo for feedback training. The subject does have control over one direction of change, but it is not the direction that matters. I would hope that this ingenious and useful solution to some of the problems of control associated with biofeedback will be more widely used.1

Not only is it necessary for a placebo to be active in order to provide similar expectancies for the treatment and control groups; it is also important to consider the expectancies of the therapist in assessing the effects of a placebo. It is not merely that the therapist must be blind as to which patient is receiving an active treatment; it is important to know something about the therapist's beliefs concerning the active treatment.

Consider, for example, Blanchard's report earlier in this volume that, with migraine, placebo typically had a 16.5 % effectiveness (see Blanchard, Table 2, "Comments on the Chapters by Beatty and by Surwit"). There is a tendency to take an observation of this kind, derived from one study, and to use it as a rule of thumb for an estimate of nonspecific effects in other treatment contexts of the same condition. A secondary analysis, carried out by Fred Evans in our laboratory, shows the difficulties with such an approach. He reviewed a whole series of placebo studies (Evans, 1974) that treated the placebo as if it were a new analgesic drug. He pointed out:

The efficacy of an unknown analgesic can be determined by comparing it with a strong analgesic, such as morphine, and perhaps to a weaker analgesic, such as aspirin. What could be called the index of drug efficiency can be expressed as the ratio: Reduction in pain with unknown active drug/Reduction in pain with known analgesic. For example, comparing a mild analgesic such as codeine with morphine would yield a relative low index, indicating it is less effective than morphine. Comparing codeine with aspirin would presumably yield an index greater than 1.00, indicating that codeine is more effective than aspirin. (p. 294)

1 Editors' note: An alternative methodology for insuring double-blind procedures with yoked subjects has also been noted earlier by D. Shapiro (Cohen, Graham, Fotopoulos, & Cook, 1977; see Chapter 3).



Using this index, he then compared the effectiveness of placebo with a strong and a mild analgesic. He observed that the index of placebo efficiency compared to a standard dose of aspirin was .54, indicating that placebo is 54% as effective as aspirin (averaging 10 studies). If placebo is a relatively weak drug compared to aspirin, what could be expected of its efficiency when compared to a standard dose of morphine? Surprisingly, contrary to expectation, the index of efficiency was not much lower; rather, it turned out to be .56 compared to morphine. The same drug (placebo) now suddenly seems quite powerful, having some 56% of the effectiveness of morphine. When Evans compared placebo with Darvon, it again showed an efficiency index of .56. In each instance, placebo was compared blind to a different drug, but its actual capacity to relieve pain was far greater when it was being compared to morphine than when it was being compared to aspirin. Thus, the efficiency index turned out to be identical when placebo was compared against a powerful, a mild, or a moderate analgesic.

The analysis by Evans elegantly demonstrates that the effectiveness of a placebo varies with physicians' expectations about the effectiveness of the drug they are administering. The relative effectiveness is essentially identical, regardless of whether it is being compared to a powerful or a weak analgesic. This means that the strength of the placebo effect is a function of the physicians' expectations concerning the strength of the therapeutic effect of the active drug with which it is being compared. To speak of a fixed magnitude of a placebo effect ignores this little-known but startling attribute of placebos.

Though troubling at first, these observations actually make a great deal of sense. Since the placebo effect assesses the totality of nonspecific factors, the expectations of therapists and those around them about therapeutic outcome should either maximize or minimize the nonspecific components. Unfortunately, this also means that the notion of somehow simply measuring the magnitude of placebo effects in the abstract is unlikely to be helpful; instead, the placebo component must be assessed in the specific therapeutic context in which the presumed active treatment is being given.

One additional point concerning placebo effects is worth considering. Typically, investigators think of these effects only in terms of positive therapeutic consequences. However, just as the expectations of therapists and patients concerning a positive outcome may help to bring about such a result, so negative expectations by therapist and patient may serve to interfere with specific therapeutic effects. In a classic study by Lasagna, Tetreault, and Fallis (1962), it was found that the 14% of patients who had reacted positively to each of several placebo administrations showed mean pain relief of 95 % to morphine. On the other hand, from among the 21 % of patients who had failed to respond to placebo, the administration of morphine resulted in only 54 % of pain relief. It seems reasonable to view their results as showing



that, just as positive nonspecific factors may serve to potentiate therapeutic effects, negative nonspecific factors may counteract them. It is impressive that these effects can be powerful enough to interfere with the analgesic effects of morphine; and it should also be kept in mind that in addition to patient and therapist expectancies prior to treatment, the manner in which a treatment is carried out, the apparent competence of therapists and their technicians, the appropriateness of the setting, and a myriad of related factors may serve to create a profound nonspecific therapeutic effect under an appropriate set of circumstances. By the same token, however, the same kind of nonspecific factors may also serve to defeat a specific therapeutic effect. Some confusing findings concerning the clinical application of biofeedback are very likely to be related to these issues.

Much of the discussion in this volume has focused on the problems of control of nonspecific factors, which are typically subsumed under the catch phrase of "placebo effects." The need for studies to control for these factors relates to the present level of knowledge in the field of biofeedback. As knowledge increases, it becomes possible to specify more and more factors that initially were lumped together under the rubric of nonspecific effects. In some areas of medicine, as, for example, in the assessment of antibiotics, there is relatively little need for placebo controls. Enough is known so that dose-response curves and comparisons with other active drugs generally suffice. The better investigators are able to specify the relevant factors, the smaller and less important the area of nonspecific factors becomes. It is for this reason that today's nonspecific factors may tomorrow become identified as a series of specific factors.

The nonspecific effects of psychotherapy provide an example of this kind. There is little doubt that psychotherapy has placebo components. However, some components that are considered placebo effects may in fact be quite specific. While no study is yet available, I would like to share some observations still in progress. A phenomenon that seems particularly intriguing is the "honeymoon" of psychotherapy, which is generally thought of as a combination of transference and placebo effect. However, it is well known that troubled individuals who are given an appointment for some 2 weeks after the time at which they seek help generally become much calmer during this time; when they are seen and a therapist explores the recent past, they report having felt better since the appointment was made.

At first sight, this may be interpreted as an obvious placebo effect, since no treatment has taken place. On further examination, however, it turns out that a number of important reality changes have taken place in such patients' lives since they not only made the decision to seek help, but made an appointment with a therapist. When these patients are asked, "Tell me, with whom did you discuss going into psychotherapy?" it turns out that they have typically discussed it with spouses and other significant others, who know that the patient has obtained an appointment to see a therapist; and while the relationship may have been very tenuous and



the spouse quite angry with the patient, the knowledge that the patient is going to seek treatment significantly alters the spouse's behavior. Typically, a spouse is relieved that a patient is seeking help and that there is a prospect for change; equally, the spouse will worry about what the patient will say to the therapist, and it is characteristically important to the spouse that he or she is presented as a good, desirable person. Consequently, the way the patient is treated will have changed during the time preceding the first visit. Assume for the moment that such an effect is common and real. It is clearly due to psychotherapy. It is clearly not a function of the specific treatment by the therapist; and in that sense, it is a nonspecific effect. By the same token, this effect can be identified, isolated, and measured; thus, it becomes a specific effect that can be taken into account in its own right.

This example illustrates that, when analyzed, something that looks to all the world like a nonspecific effect is likely to have a number of very specific, identifiable components. As long as these specific components are not understood, it is necessary to use a broad placebo control; but once a phenomenon has been established, investigators can begin to focus on and acquire a better understanding of many of the specific components that were initially subsumed under the general rubric of "placebo effects." As investigators are able to do this, they gain specific control over them.


I now want to say a few words about biofeedback in the context of this discussion. It is clear that some specific effects due to biofeedback have been documented. However, they are far more limited than had once been thought. It is made clear in this volume, for example, that in the treatment of neuromuscular disorders, stroke, and the whole area of rehabilitation, biofeedback technologies may be extremely useful in what appear to be specific ways. The treatment of epilepsy by means of biofeedback also appears to be a specific effect. While it has not been talked about much, the work on fecal incontinence also appears to be in the category of a specific effect. On the other hand, biofeedback therapy of hypertension is a more complex matter. Some level of learned volitional control over blood pressure within a relatively brief time can be readily shown. However, the magnitude of that effect, except under very special circumstances, tends to be quite trivial from a clinical point of view. Biofeedback has achieved some clinical results in this area, but when these are compared with results of such therapies as relaxation training, the latter are as significant as the former.

While biofeedback at one time was thought of as a very specific intervention, it has become increasingly clear that biofeedback therapy in actual practice, particularly as applied to stress-related disorders, involves changes of a kind that biofeedback



as such is a relatively minor, though potentially effective, adjunctive treatment. However, it is not the treatment. Rather, as has been discussed in this volume, treatment for these disorders involves changes in life style; and biofeedback may help both the patient and the therapist obtain some useful information to help bring about the necessary changes.

In a variety of studies, the therapeutic effect of biofeedback has been compared to that of relaxation therapy, self-hypnosis, or meditation; typically, there has been little difference among them. In thinking about the failure to find differences among these various approaches, investigators must not forget that the findings were that each of these techniques showed considerable therapeutic effect. Rather than focusing on the lack of differences in a negative sense, it may be appropriate to ask why relaxation training and related techniques have been overlooked for many years, at least in this country. From my point of view at least, one of the major contributions that biofeedback has made is to reawaken our interest in the importance of a number of potentially significant therapeutic tools. Though the overall results are similar with these various procedures, it does not seem likely that each of these techniques is equally effective with the same individual. Perhaps it is time research was directed at determining which therapeutic approach is most effective for which kind of patient under what circumstances.

Looking back on the development of biofeedback, it appears that the technique is a specific treatment only in a very few instances; on the other hand, in the most common uses of biofeedback today -- in stress-related disturbances, as an adjunct to various forms of psychotherapy, and even as a technique of self-exploration -- there is little doubt that therapists are dealing with a myriad of factors, of which the appropriate use of electronic aids plays a relatively minor role. Nonetheless, the potential for measuring change can provide an important source of information to both therapist and patient; and as the therapeutic community develops more sophistication vis-a-vis biofeedback, relaxation training, and related techniques, investigators may anticipate that these methods will ultimately be incorporated into both behaviorally and psychotherapeutically-oriented treatment programs in an effective and integrated manner.



DiCara, L. V., & Miller, N. E. Instrumental learning of vasomotor responses by rats: Learning to respond differentially in the two ears. Science, 1968, 159, 1485-1486.

Engel, B. T. Operant conditioning of cardiac function: A status report. Psychophysiology, 1972, 9, 161-177.

Evans, F. J. The placebo response in pain reduction. In J. J. Bonica (Ed.), Advances in neurology (Vol. 4, Pain). New York: Raven Press, 1974.

Kamiya, J. Operant control of the EEG alpha rhythm and some of its reported effects on consciousness. In C. T. Tart (Ed.), Altered states of consciourness: A book of readings. New York: Wiley, 1969.



Lasagna, L., Tetreault, L., & Fallis, N. E. Analgesic drugs and experimental ischemic pain. Federal Proceedings, 1962, 21, 326.

London, P. The end of ideology in behavior modification. American Psychologist, 1972, 27, 913-920.

Lyerly, S., Ross, S., Krugman, A., & Clyde, D. Drugs and placebos: The effects of instructions upon performance and mood under amphetamine sulphate and chloral hydrate. Journal of Abnormal and Social Psychology, 1964, 68, 321-327.

Miller, N. E. Learning of visceral and glandular responses. Science, 1969, 163, 434-445.

Orne, M. T., & Paskewitz, D. A. Aversive situational effects on alpha feedback training. Science, 1974, 186,458-460.

Orne, M. T., & Wilson, S. K. On the nature of alpha feedback training. In G. E. Schwartz & D. Shapiro (Eds. ), Consciousness and self-regulation: Advances in research and theory (Vol. 2). New York: Plenum, 1978.

Otis, L. S., McCormick, N. L., & Lukas, J. S. Voluntary control of tension headaches. Proceedings of the Biofeedback Research Society: Fifth Annual Meeting, 1974.

Paskewitz, D. A., & Orne, M. T. Visual effects on alpha feedback training. Science, 1973,181, 360-363.

Shapiro, A. K. The placebo effect in the history of medical treatment: Implications for psychiatry. American Journal of Psychiatry, 1959, 116, 73-78.

Shapiro, D., Crider, A. B., & Tursky, B. Differentiation of an autonomic response through operant reinforcement. Psychonomic Science, 1964, 1, 147-148.

Shapiro, D., Tursky, B., & Schwartz, G. E. Differentiation of heart rate and systolic blood pressure in man by operant conditioning. Psychosomatic Medicine, 1970, 32, 417-423.

Travis, T. A., Kondo, C. Y., & Knott, J. R. Subjective aspects of alpha enhancement. British Journal of Psychiatry, 1975, 127, 122-126.

The preceding paper is a reproduction of the following book chapter (Orne, M. T. Perspectives in biofeedback: Ten years ago, today, and … . In L. White & B. Tursky (Eds.),Clinical biofeedback: Efficacy and mechanisms. New York: Guilford Press, 1982. Pp.422-437.). It is reproduced here with the kind permission of The Guilford Press.