Orne, M. T. Implications for psychotherapy derived from current research on the nature of hypnosis. American Journal of Psychiatry, 1962, 118, 1097-1103.




In this paper we propose to discuss the implications for psychotherapy of some of the hypotheses that have derived from our research on hypnosis. We have been concerned with the essential nature of the hypnotic process rather than its therapeutic application. The problems encountered appear to have a close parallel to those seen in the study of psychotherapy. Further, here too we are dealing with a significant interpersonal relationship which alters the boundaries of consciousness.


The behavior of a subject when hypnotized would seem easy to describe. Usually his eyes will be closed, he will appear passive, if requested to perform an action it will be done slowly, his voice will be low and often childlike and he will tend to respond only to cues from the hynotist. This description however does not hold in an historical perspective. Mesmer (1) traditionally has been given credit for re-discovering hypnosis. At his seances or "baquets" patients would have an hysterical seizure, or fit, subsequently lapsing into a sound sleep. This pattern of behavior was induced solely by silent passes and the structure of the situation. They would awaken relieved of their symptoms without any verbal suggestions to this effect. However, in a sample of several thousand subjects we have never observed behavior of this type. Coue (2 ), familiar as the originator of the phrase, "Every day in every way, I am feeling better and better," certainly employed hypnotic techniques, yet his patients did not close their eyes, appear in a trance, or go to sleep. These are only two of many examples of the variability which has been associated historically with the behavior of a subject in hypnosis.

Naturally the hypnotist may, by appropriate cues, modify the behavior of the subject in hypnosis. However, we are describing here how the subject acts on being hypnotized in the absence of specific instructions. Today we tend to see the very standard pattern of behavior which we have described. What would account for such variability in the past and such consistency today ? It seemed reasonable to hypothesize that the behavior the subject will manifest upon going into hypnosis is a direct function of his conception of how a hypnotized subject behaves. In view of the widespread dissemination of knowledge through mass media of communication and the relatively uniform current beliefs about hypnotic behavior, it would be easy to understand why subjects today will seem to behave similarly, and it would appear equally plausible that Mesmer's and Coue's subjects did not behave in this way.

We decided to test empirically the hypothesis that subjects' behavior in hypnosis is a function of their prior knowledge about the role of the hypnotized subject (3) . In order to do this, we needed an item of behavior which could plausibly be associated with hypnosis but in fact had never been observed occurring spontaneously. Such an item of behavior is hard to come by, since a very wide range of hypnotic behavior has already been reported in the literature. We finally hit upon catalepsy of the dominant hand. Catalepsy has frequently been observed in hypnosis; however, when it occurs, it occurs in all limbs. It never occurs in one hand, while the other remains flaccid. However, catalepsy of the dominant hand sounds somewhat scientific; it evokes in the college students, whom we planned to use as subjects, vague memories

1 Read at the 117th annual meeting of The American Psychiatric Association, Chicago, Ill., May 8-12, 1961.

2 The research on which this paper is based was supported in part by Public Health Service Research Grant M-3369, National Institute of Mental Health.

3 Massachusetts Mental Health Center and Harvard Medical School, Boston, Mass.




of stutterers, tests of hand dominance, and so on.

First I gave a lecture on hypnosis to a group of university students in an introductory psychology class. Included was a demonstration with several of the class serving as volunteers. Unknown to the class, these subjects had previously been hypnotized and had been told to manifest catalepsy of the dominant hand. Along with other well-known phenomena of hypnosis, this characteristic was casually pointed out as typical. At a later time, members of this class were hypnotized and tested for catalepsy and for the first time we were able to observe a new behavioral characteristic of hypnosis --namely, catalepsy of the dominant hand without any specific suggestion. We view this as a demonstration of the effect of prior experience or knowledge on hypnotic behavior. Later we performed a more rigorous experiment with matched sections of the same class. The two sections received identical lectures, but one lecture included the new item and the other omitted it. Volunteers from the class, with both sections randomly mixed, were then hypnotized by a hypnotist who did not know which subjects belonged to which section. Catalepsy of the dominant hand was found to occur only in the section which had received the lecture demonstrating it as a typical characteristic.

The implications of this type of study are rather striking. It seems to suggest that we have no uncontaminated data about the behavior which is characteristic of hypnosis per se. Since it is practically impossible to find "naive" subjects, there seems little prospect of obtaining such data in the future, short of cross-cultural and historical studies. Probably most, if not all, of the behavioral characteristics of hypnosis can be understood in terms of the subject's previous knowledge and the cues transmitted during the process of induction. It is entirely possible to conceive of the typical hypnotic trance for the most part as an historically developed artifact occurring along with a process, the essential behavioral manifestations of which are little known. The basic process, without the gross behavior which is so variable, might be called the essence of hypnosis. This is the real focus of our research interest.

The problem of recognizing which elements ascribed to hypnosis are artifactual or epiphenomenal is extremely difficult. The nature of hypnosis is such that any expectation the hypnotist entertains may unwittingly be communicated to the subject who then acts in a way that demonstrates the validity of the expectation. Thus, we have the potentiality for the occurrence of self-fulfilling prophecies without the investigator becoming aware of his own role therein. It is necessary to get outside of the immediate interaction into its context in order to recognize these variables.

In experimental research with hypnosis this gross variability of the phenomenon persists. Here, too, the subject will behave in accordance with his perception of the experimenter's expectations. It is generally recognized that unwittingly detailed and accurate communications may take place in the form of implicit or non-verbal cues. Clearly the subject does not respond merely to the verbal suggestions but rather to the totality of the situation from which he actively attempts to ascertain the behavior which is desired. In an experimental situation the subject may further derive considerable information about the experimenter's wishes from the procedure of the experiment itself. The extent to which the experimental procedure communicates the hypotheses of the experimenter will depend upon the subject's previous knowledge and sophistication. However, some knowledge will inevitably be communicated.

The totality of cues which communicate the hypotheses or wishes of the hypnotist (including implicit and non-verbal cues from the experimenter and cues provided by the experimental procedure) we have termed the demand characteristics of the experimental situation. What demand characteristics are perceived will vary with the subject's prior knowledge, and different demand characteristics may be perceived by different subject populations in the same experiment. These will, under some circumstances, be the major determinant of the subject's behavior. For example, if we test the Babinski reflex of a medical student, then regress him to the age of 3 months and test the Babinski again, our


1099 Martin T. Orne

expectation of plantar extension rather than flexion is communicated by the procedure in the light of the subject's knowledge.

Experimental findings which are a function of the demand characteristics may or may not be replicable depending upon whether the replication provides the same cues. But certainly they do not permit us, to make valid inferences about real life situations. It is clear that unless the demand characteristics of an experiment are understood, ecologically valid(4) inference is difficult. Much of our work therefore has been concerned with the development of techniques which permit the study of demand characteristics. Three major approaches are possible : 1. To manipulate the demand characteristics purposefully; 2. To study what demand characteristics were perceived by inquiring into the subject's perception of the purpose of the experiment ; 3. To study the effect of the demand characteristics by including a group subjected to a dummy treatment which omits the experimental or independent variable such as hypnosis. The behavior of this group can then be accounted for in terms of the demand characteristics rather than in terms of the independent variable.

We have discussed these problems of control in detail elsewhere. I would like merely to point out that the common practice of telling the subject a fictitious but plausible reason for an experiment is an example of manipulating the demand characteristics. This is effective only insofar as the subject actually believes what he is told. It is incumbent on the experimenter in each instance, to determine what were the subject's actual beliefs in the situation.

The procedure of inquiring into the subject's perception of an experiment after it is over has inherent factors which make obtaining a valid report difficult. It is common knowledge among our subject population that they are supposed to be ignorant of those aspects of an experiment which are not specifically explained to them. They are aware that too much information about a study would disqualify their performance as subjects. This eventuality runs directly counter to their motives for participating in psychological studies. It would vitiate their efforts to contribute to science and research and make their investment of time meaningless. They are therefore motivated to respond with "I don't know" to questions about their perceptions of the purpose of the experimental tasks. The experimenter on the other hand is equally motivated not to obtain such information, as he no more than the subject relishes the thought of wasting his time and does not wish to exclude a subject's performance. As a result he may all too easily accept the initial "I don't know" and the interlocking motivations of subject and experimenter will thus lead to a pact of ignorance. If the experimenter does not accept the initial denial of knowledge by the subject but acts upon the hypothesis that the subject may know more than he is telling (much as we would in the therapeutic situation) he will find that most subjects will be able to verbalize very specific hypotheses about the experiment which may or may not coincide with those of the experimenter. In elicting the subject's beliefs about the experiment it is desirable to have the inquiry performed by a member of the research group other than the experimenter. In several studies the subjects' beliefs about the experimenter's hypotheses proved to be better predictors of what they did in the experiment than their reports of what they thought they had done.

Finally, we have used extensively a procedure which attempts to maintain the demand characteristics of the situation, to maximize the subject's response to them, but to eliminate the variable to which the experimental result is usually ascribed. One example of such a procedure involves the use of simulating subjects. Subjects who had failed to enter hypnosis during repeated sessions were told that they were to simulate entering hypnosis for the hypnotist. They were further informed that the hypnotist would know that some subjects were trying to simulate but not which; that if he discovered that the subject was simulating he would terminate the experiment; but that successful simulation was possible. Under these circumstances subjects are able to behave in a manner difficult and ofttimes impossible to distinguish from, hypnosis without, however, the subjective alteration in experience which distinguishes the hypnotic state. Simulators are not detectable by



such tests as pinching or faradic stimulation and are highly motivated to carry through their role. They must of necessity base their behavior on the demand characteristics of the situation rather than let it be determined by hypnosis.

The hypnotist in this type of design has to be ignorant of the true status of the subject in order that the subtle non-verbal communications remain constant for both groups of subjects. The real-simulator design is intimately related to the blind design in certain types of psycho-pharmacological experiments. The use of simulating subjects makes possible the extension of the blind design to experiments where we cannot conceal from the subject whether he received the experimental or placebo (dummy) treatment. At the same time the hypnotist cannot discriminate between groups of subjects. Such design then enables us to decide between explanations of individual items of behavior of real subjects as due to the experimental independent variable or alternately as due to the intervening variables of the demand characteristics of the experiment. Such items of behavior that are produced by simulating as well as real subjects may be explained in terms of the demand characteristics of the experiment -- or they may, in the case of real subjects, be produced by another mechanism. In these instances, further research is necessary to prove which of these alternative explanations is responsible for the effect.


In the development of these techniques it became clear that the experimental situation in and of itself is a peculiarly powerful one. Simulating subjects are extremely motivated. They will endure a high level of painful stimulation, undertake arduous, embarrassing, and boring tasks. They will also match the apparent transcendence of normal capabilities observed in hypnotic subjects. This is probably due to a unique combination of motives which makes it possible for the subject to gratify his wishes to contribute to research and to further knowledge, while at the same time putting one over on the experimenter, thus deprecating an authority figure.

The psychological experiment, even without simulating subjects, is a peculiar situation. In the course of our methodological investigations we became intrigued with the social psychology of the experiment. By mere agreement to participate in a study the subject grants the legitimacy of a remarkably wide range of requests. The category of potential behavior which can be demanded of a subject during his participation in a psychological experiment is so broad that it is limited only by the social constraints on the experimenter. For example, if a student asks a peer to do 10 push-ups as a favor, he will be asked, "Why?" If the same student secures the agreement of the same fellow student to take part in an experiment and then tells him to do 10 push-ups, he will be asked, "Where?" In other words, the willingness to do a favor does not in and of itself legitimize the request to do 10 push-ups without further explanation. However, participation in an experiment causes the subject to assume a legitimate purpose without further explanation being required. We have tried to find absurd "experimental" tasks which subjects would refuse to perform and have been unable to do so. Only such tasks would presumably meet with refusal that conflict with the basic value system of the subject. By its very nature, this hypothesis cannot easily be tested in the laboratory.

It is clear that once the subject has given his consent to participate in an experiment he voluntarily puts himself under the control of the experimenter, The potential range of behavioral control is extremely broad, at least for the usual college student volunteer. This surrender to control I would interpret as deriving from the subject's involvement with, and high regard for, science and experimentation. The process of selection of volunteers by itself tends to guarantee such a value orientation. The authority of the experimenter then derives from his association with science. The control is sufficiently legitimized by the declaration, "This is an experiment," which is accepted as valid by our subject population. This aspect of the experimental situation in non-hypnotic experiments, which, incidentally are equally subject to effects of demand characteristics, must be taken into account when we wish to infer to real life situations.


1101 Martin T. Orne

Many of the considerations which apply to hypnosis and the experimental situation may be also applicable to the doctor-patient relationship. Here, too, is a peculiar authority relationship wherein the patient makes certain assumptions about the positive motivation of the doctor. Furthermore, the patient places himself under the care of the physician much the same way as the subject under the control of the experimenter. In most doctor-patient relationships, it is potentially possible to elicit a wide range of behavior from the patient by the statement,"This is a test." Few patients insist on knowing the logic of the test prior to complying with instructions.

Our knowledge about hypnosis is derived almost exclusively from studies conducted in clinical, experimental or quasi-experimental settings. Hypnosis is often seen as an extremely potent technique of behavioral control; however, because of the degree of behavioral control which is already inherent in the settings in which hypnosis is studied we have been unable to demonstrate any increment of control over the hypnotized subject from the amount of control which can be exercised by the physician or experimenter over the unhypnotized subject. At this point the question has not been answered, whether hypnosis increases the hypnotist's control over the subject beyond that which is already inherent in the situation prior to the induction of trance.


It is not clear what the essential qualities of the hypnotic state are, or what, if anything, psychodynamically relevant is changed by the existence of this state. There are, however, certain constant qualities of social interaction that characterize hypnosis. Thus, the behavior of the hypnotist changes as dramatically as the behavior of the subject. Many aspects of hypnosis can best be conceptualized as a folie a deux (a set of complementary role expectations about an unreal definition of the situation). Thus, the subject acts as though he were unable to resist the suggestions of the hypnotist and the hypnotist acts as though he were all-powerful. By the same token not only does the subject experience the perception of a suggested hallucination but the hypnotist also acts as though the subject were in fact seeing the hallucination. Particularly striking is the behavior of the hypnotist with a subject in age regression. His speech becomes altered from that customary in addressing an adult to that typically used when addressing a child. In fact, if the hypnotist does not do this and fails to obtain the phenomenon from the subject, one tends to say his technique is faulty. Poor technique in other words, on close inspection, turns out to be the failure of the hypnotist to complement the role which he had assigned to the subject.

We have been filming hypnotic sessions in order to analyze the interaction in more detail. In some of this research we have used both real and simulating subjects with the hypnotist being unaware of their true status. In one instance, the hypnotist had become convinced on meeting the subject that he was a simulator, whereas in fact the subject was able to enter deep hypnosis easily and quickly. In the interaction that ensued, the subject failed to enter deep trance and became quite hostile to the hypnotist, who, while giving suggestions as usual, failed to play convincingly a complementary role.

Perhaps one of the major problems in the psychiatric use of hypnosis is the great difficulty of employing a technique which demands that the therapist enter into a folie a deux with the patient. One requirement of successful hypnosis is that the patient should be able to ascribe magical powers to the therapist. It is necessary for the therapist employing hypnosis to enter into this relationship, act out and participate in the folie a deux, while maintaining sufficient objectivity to recognize that he does not acquire the power the patient ascribes to him. This is perhaps one of the explanations of why thoroughly competent psychiatrists and analysts reporting on isolated uses of hypnosis, usually during their war time experiences, describe attempts at treatment which violate their own very excellent knowledge of psychodynamics. Somehow they seem to get caught in the interaction process and seduced into attempting to compel the patient to change in a manner which is blatantly incompatible with the particu-



lar patient's personality. Obviously such an endeavor will lead to failure. In every instance of hypnotherapy where we have had the opportunity of exploring what actually took place, it has been clear that the basic psychodynamic mechanisms operating are in no way altered or suspended because the patient enters hypnosis.

Another aspect of hypnotherapy, which may be responsible for the apparent changes in the defensive organization of the patient in hypnosis, may be the role expectations inherent in the use of hypnosis both by the patient and the therapist. Thus, in the case of symptom removal, the therapist is seen as an omnipotent figure by the patient and participates with the patient in this aspect of the folie a deux as well. The therapist takes responsibility for the patient's relinquishing of the symptoms and the patient is able to mobilize a greater degree of flexibility by identifying with the omnipotent therapist. Again in the case of obtaining material otherwise unavailable to consciousness, the therapist assumes responsibility for the patient's verbalizations; the patient feels, "because I'm in hypnosis, I'm not responsible for what I say." In some instances the function of hypnosis may be to legitimize a change in behavior which the patient wishes to undertake but cannot without an appropriate excuse. As has been pointed out by others, we can ascribe a similar function to the psychoanalytic couch. Not only is the couch historically related to hypnosis, but it may also be related in a structural sense, in that it symbolizes an alteration in the situation, which clearly delineates when a patient can feel without responsibility for his verbalized thoughts. The getting up from the couch is somewhat analogous to the hypnotist's waking up the patient when the shared expectations are that he is again to behave socially, i.e., censor his speech as in everyday life. These procedures may be seen as redefining behavior which normally is regarded as deviant into variant behavior legitimized, and called for, by the situation.

Some of the implicit shared expectations about therapy, closely analogous to the demand characteristics in experimental situations, may play a major part in determining the course of psychotherapy. For example, it is felt by the advocates of hypoanalysis that the use of hypnosis makes possible more rapid progress in treatment (5, 6). Assuming that the rate of progress is in fact more rapid, it might well be due not to the actual use of hypnosis but rather to the firmly held belief by both patient and therapist that the use of hypnosis speeds up therapy. It is entirely possible that certain shared expectations about the length of therapy govern in large part the rate of progress. In recent years the length of treatment has tended to increase, no doubt in a large measure due to changing orientation about goals. Perhaps some of this lengthening may also be due, however, to certain shared expectations about how long treatment should take. It is an interesting but difficult empirical issue raised tangentially in the literature on termination dates to what extent shared expectations about the length of treatment will affect the rate of progress in treatment.

Not only may certain shared expectations affect the length of treatment but also their presence or absence may determine much of the outcome of treatment. We would like to draw an analogy between the behavior of the hypnotized subject and that of the patient in therapy. The behavior in response to hypnosis is, as we have pointed out, largely 'dependent' upon the knowledge about the role of hypnotized subjects that the subject brings with him, while subsequent behavior depends upon cues from the hypnotist and the subject's interpretation of these cues in the light of his previous experience. The development of the folie a deux of hypnosis is facilitated by the fact that subjects know a good deal about the role that they are expected to play and that the hypnotist is explicit about the behavior he wishes to elicit. The therapeutic situation is different for many patients in that they know little about the patient's role in treatment and we as therapists are loathe to communicate the behavior both in and between sessions that is expected of them. Despite our reluctance to give the patient information about his role, it is vital for him to somehow learn enough about this role to make the therapeutic interaction possible. In the complex interaction which constitutes the therapeutic process the patient's ability


1103 Martin T. Orne

to behave appropriately will largely determine whether he is perceived by the therapist as a good or a poor patient for treatment. The therapist's perception of the patient will in turn in large part determine the subsequent course of treatment. In some ways this may be analogous to a hypnotist attempting to hypnotize a good subject whom he believes to be simulating, and failing to obtain an adequate trance. This analysis would seem consistent with the report of Redlich and Hollingshead (7 ) that educated middle and upper class patients receive psychotherapy while uneducated, lower class patients receive EST. Not only does the psychotherapy patient come from the same social class as his therapist but he also has the greatest amount of knowledge about the role the patient is expected to play in treatment.

We might hypothesize that this anticipatory socialization for the role of patient is the crucial variable. With increasing mental health education we might expect an ever larger proportion of patients to be able to play this role. It could become quite standardized much in the same way as has the response from any member of our culture to any technique of hypnotic trance induction. While this may be desirable, it would make extremely difficult our ever discovering what is really essential to the process of psychotherapy. The historical confusion about hypnosis would find a parallel here. As in hypnosis, the behavior of patients in psychotherapy has varied widely. The same method of therapy has led to different results in the hands of different therapists, while dissimilar therapeutic efforts frequently have led to the same results. Perhaps here too we need to concern ourselves with separating the constant invariant essence of the therapeutic process from those variable aspects that are artifactual and epiphenomenal.

The recent reports by the British of 15-minute therapeutic sessions at intervals leading to dramatic results might be evaluated from the perspective we have discussed. Just as we have omitted in our group of simulators the crucial variable of hypnosis while maintaining the demand characteristics of the situation, the British psychiatrists using few, brief, intermittent sessions have omitted variables which we believe to be necessary for the therapeutic process, namely, long interviews, several times a week, over a prolonged period. They have, however, included the demand characteristic of the therapeutic situation. A comparison of the results obtained could be regarded as a quasi-experiment to distinguish between the effect of psychotherapy and the effect of the patient's perception that he is in psychotherapy.


We have applied to psychotherapy some hypotheses that arose from our current research on hypnosis. Whether or not there exists a valid analogy between the therapeutic process and the hypnotic state, the specific hypotheses may be potentially fruitful in our inquiries about the nature of the therapeutic process. Any conclusions about their validity will of course have to await concrete results of research. Some of the methodological considerations and tools we have developed to deal with problems in the study of hypnosis may be useful in testing similar hypotheses about psychotherapy.


1. Boring, E. J. : A History of Experimental Psychology. 2nd Ed. New York and London Appleton-Century-Crofts, 1950.

2. Coue, E.: Self Mastery Through Conscious Autosuggestion. New York : Amer. Library Serv., 1922, p. 83.

3. Orne, M. T.: J. Abnorm. Soc. Psychol., 58: 277, May 1959.

4. Brunswik, E.: Systematic and Representative Design of Psychological Experiments. University of California Syllabus Series No. 304, 1947.

5. Brenman, M., and Gill, M. M.: Hypnotherapy. New York : International Universities Press, 1947.

6. Wolberg, L. R.: Hypnoanalysis. New York : Grune & Stratton, 1945.

7. Hollingshead, A. B., and Redlich, F. C. Social Class and Mental Illness : A Community Study. New York : John Wiley & Sons, 1958.

The preceding paper is a reproduction of the following article (Orne, M. T. Implications for psychotherapy derived from current research on the nature of hypnosis, American Journal of Psychiatry, 1962, 118, 1097-1103.). It is reprinted with permission from the American Journal of Psychiatry, Copyright 1962. American Psychiatric Association.