Director, Studies in Hypnosis Project, Massachusetts Mental Health Center, Boston, Mass.
Instructor in Psychiatry, Harvard Medical School.
Investigators doing research with hypnosis need not wait for events under study to occur naturally; they can induce particular phenomena at will in the laboratory, rapidly, repetitively, and easily. States can be induced that mimic a wide range of naturally occurring psychopathology. The hypnotized subject can cross the boundary of awareness or consciousness with relative ease. Material may be made inaccessible to consciousness, or normally inaccessible material may be made available to consciousness.
A number of competent investigators have used hypnosis to study psychologic processes 22. A number have rejected it. Many of the classic studies on the subject were carried out by individuals who, after investigations of varying duration, turned to other avenues of inquiry.
What specifically is it about hypnosis that so disappoints the investigator? It appears to offer its practitioners omnipotence. Many students become disillusioned when they learn that the power of the hypnotist is more apparent than real. Such reactive rejection prevents workers from learning that the subject-hypnotist relationship need not differ essentially from the more usual doctor-patient one.
Contradictory findings are frequently reported. There have been striking differences of opinion between clinicians investigating hyp-
Schilder, P.: MEDICAL PSYCHOLOGY, New York, International Universities Press, 1953.
Weitzenhoffer, A. M.: HYPNOTISM: AN OBJECTIVE STUDY IN SUGGESTIBILITY, New York, Wiley, 1953.
676
677
nosis in patients and experimental psychologists working in the laboratory. The experimentalist blames the clinician's inadequate methodology, while the clinician complains of the experimentalist's inadequate hypnotic technique.
Contradictory findings in hypnotic research can be attributed to the essential qualities of the state itself. The hypnotized subject is characterized by his ready response to cues (to "suggestion") by the hypnotist. He is motivated to please the hypnotist to the best of his ability. He responds to explicit suggestions and to implicit ones as well. The hypnotist conducting the experiment, since he inevitably has a hypothesis to test, may unwittingly convey his expectations to the subject. The subject obliges the hypnotist by manifesting the appropriate behavior. The hypnotist will then have created the very phenomenon which he believes to be external to himself.
No less astute an observer than Charcot reported that rubbing the top of the subject's head caused his subject to pass from sleep to somnambulism. Hypotheses that are self-fulfilling prophecies are by no means alien to psychiatric research.
Effect of Demand Characteristics
The totality of cues which convey to a subject the purpose of an experiment (of which he is presumably unaware) can be termed the demand characteristics of the experimental situation. These are cues that are not merely inherent in the instructions; they may be conveyed by the experimental design itself. By what the subject is asked to do, he may be given to understand what the investigator expects him to do. This concept of demand characteristics, drawn from Lewin's work, implies that we are dealing with social perceptions, and like all perceptions these are dependent upon the past experience of the perceiver.
For example, asking a college student to participate in a sensory deprivation study usually implies that he is expected to show alterations in perception, perhaps even to hallucinate. Most students have heard about such studies and have some idea about results already elicited. A request to participate in a sensory deprivation study may therefore have the demand characteristic of altering perceptions.
678
Different subjects may perceive different demand characteristics, given the same experimental situation. The demand characteristics of a particular experiment with a college undergraduate population may differ from those perceived by a National Guard population. The effect of demand characteristics may play a role in any experimental framework where the subject is highly motivated to please the experimenter. The demand quality itself derives from the subject's motivation.
The subject's reasons for participating in an experiment warrant consideration. The pay of experimental subjects is insufficient to justify the time and effort involved or to explain their willingness to endure the, at times, considerable discomfort that may be a necessary part of the experiment. The student subject feels that he is contributing to the advance of science and knowledge. If he is to contribute maximally, the experiment must "work"; i.e., the experiment must demonstrate what the professor thinks it should. Hence the demand quality of demand characteristics.
In attempting to study the effect of demand characteristics, we have used a post-experimental inquiry to learn from the subject what he conceives to be the purpose of the experimental situation. He is usually loathe to disclose this, since he knows the rules of the research game well enough to realize that too much knowledge invalidates his performance. To him, this means that he has wasted his time. Therefore, he has some interest in not knowing too much. And the investigator likewise realizes that if the subject knows too much, he must discard the performance. Therefore, he has a stake in not pushing too hard. As a result, a community of interests frequently exists between subject and researcher, leading to a collaborative pact of ignorance.
When subjects are asked, "What do you think this experiment is about?," they usually respond, "I don't know." We do not accept this answer, but push the inquiry further. "I know you don't know," subjects are told, "but you must have some idea about it. What do you think? Take a guess." At this, subjects may suddenly verbalize lucid hypotheses, clearly demonstrating that they are by no means naive and that they know exactly what the investigator is attempting.
679
Investigative Procedures
I am constantly amazed at how sophisticated college undergraduates actually are. We use six questions in our experiments, rephrasing them as indicated: (1) "What do you think this experiment is about?" (2) "What do you think this experiment is trying to prove (to demonstrate) ?" (3) "What do you think we expect to find?" (What is my hypothesis?) (4) "What would you predict our results to be?" (5) "What do you think others have done?" (What is your hypothesis?) (6) "What do you think you did in the experimental situation and what do you think your behavior demonstrated?"
We found we could predict back to the subject's behavior not from what he thought he did during the experiment but from what he perceived our expectations to be. It may be argued that the subject generalizes to the experimenter's hypothesis from his own behavior in the experiment, and that as a result demand characteristics that have influenced the behavior of the subject are actually not elicited by a post-experimental inquiry of this kind but that the subject merely verbalizes an hypothesis based on what he did. For methodological reasons, therefore, in order crudely to examine demand characteristics that may be built into the experimental procedure, the pre-experimental inquiry was devised.23 Subjects are drawn from the same population as the experimental subjects but are never run in the experimental procedure proper. The total procedure to which the experimental subject is exposed is described in detail. The investigator then asks, "If we did this experiment with you, what do you think it would be about? What would we be trying to demonstrate? What would you do?" and so forth.
This investigative procedure can be used with experiments which employ pre and post-testing with some intervening experimental manipulation. Subjects are sentient beings whose thoughts markedly affect their subjective experiences, rather than passive objects to be manipulated experimentally, and controls such as these help to clarify the effect that the thinking of the subject may have on the results obtained.
680
During one pilot study, pre and post-tests used in early sensory deprivation experiments were utilized. The same set of experimental conditions was described to two groups of six subjects each. For one group the experimental purpose was expressed as a study on sensory deprivation; to the second group it was referred to as a study of time perception. In comparing the pre and post-experimental test results, a significant difference was found (P <.001) between the two groups. This suggests, even with these small samples, that calling an experiment a sensory deprivation study introduces significant demand characteristics. Though this type of procedure does not constitute proof, it permits the investigator to recognize potential sources of effect other than that which he is trying to study. Within a few hours, he can obtain a good first approximation of the level of sophistication of his subjects and the effect this may have on the experimental data to be gathered.
Behavior in Hypnosis
The state Mesmer induced was not hypnosis as we know it today, but rather a kind of hysteric seizure followed by sleep with subsequent relief from symptoms. I have never seen this, except occasionally in the Pentecostal Churches during healing services. The current description of hypnosis, as most of us think of it today, dates to the Marquis de Puysegur (1751-1825) 24. Yet the trance that Rosen has described in the preceding pages differs from this. His hypnotized subject need not seem to be asleep. He may look awake. As an end-point, he may be unable to bend his arm while, nevertheless, his eyes are wide open. Descriptive behavioral characteristics obviously do not define hypnosis with any degree of adequateness.
How the subject will behave in hypnosis depends upon his knowledge of hypnosis. This can be readily shown. In a demonstration-lecture on hypnosis to a freshman class in psychology we used two subjects who, while previously hypnotized without the knowledge of the class, had been given a suggestion that during hypnosis they would develop catalepsy of the dominant hand.
681
During the subsequent demonstration-lecture, there was therefore catalepsy of the dominant hand. During the lecture, it was stated that this is characteristic of the hypnotic state. The lecture was otherwise accurate. Some members of the class who were then hypnotized, spontaneously developed catalepsy of the dominant hand. The term "catalepsy of the dominant hand," with its vague tie-in to stuttering, sounds scientific enough to be plausible.
These experiments suggest that behavior associated with hypnosis may be largely an historical accident. If specific behavior becomes widely publicized as characteristic of hypnosis, subject and hypnotist alike may view it as typical. It will then tend to occur when hypnosis is induced. Since neither naive subjects nor naive hypnotists can be found, it becomes an interesting though unanswerable question as to just what behavioral characteristics serve as criteria for hypnosis if both subject and hypnotist are without pre-conceptions. At this particular historical moment it seems most appropriate to view hypnosis, as proposed by White25 and as described by Rosen26, as a kind of behavior consensually agreed upon between subject and hypnotist.
Control Groups
One of the major problems posed when studying the effects of hypnosis therefore becomes that of setting up an adequate control group. Hypnosis should be compared with the waking state. Subjects are used as their own controls or different waking subjects are utilized. But this experimental situation means that different groups are exposed to different demand characteristics. Even rigorously standardized instructions will not exclude subtle cues which may be effective despite their subtlety. By analogy with double-blind and even triple-blind pharmacologic studies, experiments were designed so that the hypnotist would not know whether or not he was dealing with a hypnotized subject. This required a group of subjects who could successfully simulate the hypnotic state, so that whatever subtle cues were conveyed by the experimenter would nevertheless be the same for both hypnotized and simulating subjects. The experimenter did not know which were which. He there-
26 Rosen, H.: HYPNOTHERAPY IN CLINICAL PSYCHIATRY, Julian Press, New York,1953.
682
fore behaved in a similar fashion towards all. Thus, differences between these two groups could not be attributed to demand characteristics communicated by the experimenter.
Most theorists today agree that hypnosis leads to increased motivation by the subject to comply with the requests of the hypnotist. White27 and Sarbin 28 have pointed out that the subject in hypnosis wishes to play the role of the hypnotized subject. While high motivation and role-playing undoubtedly are factors of the hypnotic state, they are by no means unique to it.
Our interest in hypnosis lies largely in those aspects of the trance which can not be accounted for in terms of role-playing and motivation. In order to throw these aspects -- often viewed as dissociation, alteration of consciousness, cortical inhibition, etc. -- into clear relief, it is desirable to have control subjects who show high motivation and role-playing but not the other more difficult-to-define factors present in the trance state. If subjects could be highly motivated to simulate hypnosis, they would play the role of the hypnotized subject and at the same time have a high level of motivation; but insofar as they would not enter trance, they would not experience the changes of consciousness found in hypnotized subjects.
Simulation of Hypnosis
One frequently reads that it is difficult if not impossible to simulate being hypnotized. Subjects when asked to simulate are unsuccessful. They smile frequently, hesitate, and ask, "Am I doing a good job?" This is not adequate. In our experiments the subject was told that the hypnotist would not know if he was simulating or not. This made a tremendous difference. He was thus given a challenge. He was then able to simulate, and to simulate well. As long as the subject knows the hypnotist knows, he has no strong motive to simulate well; the moment he knows the hypnotist does not know, he tries to do a good job -- and does it.
Accordingly, one of my associates would determine the subject's hypnotizability. If he failed repeatedly to enter the trance,
28 Sarbin, T. R.: "Contributions to Role-Taking Theory: I. Hypnotic Behavior," Psychological Review, Vol. 57, 1950, pp. 255-270.
683
he would then be given the following instructions: "We'd like you to try to fool Dr. O. He knows some people are trying to, but he doesn't know who. Other people, not knowing more than you about hypnosis, have been able to fool him. If he catches on that you are faking, he will stop the experiment. Unless he stops the experiment he has not caught on. Fake through the experiment and fake when he asks you questions about it. Only when I return to the room are you to tell the truth to both of us."
In point of fact, I did not know who the simulators were, and different investigators were of course used in the experimenter's role. In this situation, the college undergraduate was asked to help contribute to research, to the progress of science, by making a fool of one of his professors (a hypnotist at that) whom he viewed as an omnipotent figure. Our subjects did a magnificent job of simulating. A large majority could not be detected. If the observer knows in advance who the simulators are, he picks up a dozen cues to show that a particular subject is simulating. He can not understand how any one can fail to spot him. But if he does not know, he does not pick up the cues. These subjects look as though they are hypnotized, and they tolerate situations that one would a priori expect only a hypnotized subject to be able to tolerate.
We are presently engaged in trying to investigate systematically what these cues are, whether they can be learned, if practice makes for improvement, etc.
It has been stated in the literature that a simulating subject can be detected if anesthesia is induced and a painful stimulus presented. This is not so. With some subjects anesthesia was induced and the usual things were done, like pinching, and forcibly flexing and squeezing the terminal phalanx of the little finger. Neither real nor fake subjects reacted. We also tried a small inductorium. Neither group reacted even to the highest level. One of my associates presented electric shock from 10-850 milliwatts, a quite considerable jolt, to the leg, and found to his surprise that simulators tolerated appreciably higher levels of shock than did hypnotized subjects. In this particular situation therefore the simulating college student subject is extremely highly motivated.
It has been suggested that these simulators are actually in hypnosis. This would mean that they were hypnotized by mere instruction. They were not. They could not go into hypnosis despite
684
repeated attempts in our experimental situation. Neither did they report any of the subjective experiences of hypnosis.
Hypnosis as a Subjective Experience
We are tending more and more to define hypnosis in terms of subjective experience. Alterations in this approach have become our prime focus of interest. They are at the very core of the phenomenon.
But how can we get at it? There are several ways. For one, we use inquiries. I do one post-experimentally, then an associate walks the subject to the elevator and perhaps gives him a lift home, all the while casually asking questions. Sometimes the roommate of a subject will report the subject's statements about a session and his experiences during it. In this way, one can get at aspects of the experience not obvious at first glance.
Another perhaps more fruitful approach involves the use of suggestions when the investigator knows what the subject is experiencing, although the subject does not realize this. A series of suggestions is given the subject: "You can not bend your arm. You can not take your hands apart. The room is growing warm. The room is growing cold. The metal ball you're holding is growing hot." The metal ball is a chemical gadget and if turned upside down does get hot. However, the experimenter handles it as if it were still cold. When the ball become hot, the subject tends to believe that his perception of the ball has altered in much the same way as his eyes have become heavy.
Such procedures afford the possibility of developing a subjective scale of experience, because the subject does not know that the investigator knows what he has felt. He is therefore asked, "How did it feel when you could not bend your arm?" ("Very real.") "How did the room feel when it got cold?" ("Real cold.") "How did the ball feel?" ("As though it had a little heater in it.") "Did the arm feel more real than the room getting cold?" ("Yes, the arm felt more real.") "Did the feeling of the arm getting stiffer and stiffer feel more real than the ball getting hot?" ("The arm being stiff was more real.") The subject does not state -- to us, to his roommate, to anyon -- that he had caught on. He has not. We can begin to get some feeling about how real these subjective events actually seem to be. By judicious question-
685
ing we can begin to approximate a scale of the subjective events. We know the reality quality of how the ball feels.
A major controversial issue has revolved about whether an individual in hypnosis can transcend his normal volitional capacities. Great care must be taken when evaluating experimental data. For example, it has frequently been reported that physical capacity increases in hypnosis. The experimental procedure generally used to demonstrate this "fact" consists of comparing a subject's performance of some task in hypnosis with his performance of the same task during the waking state. Some studies appear methodologically sound: they use the subject as his own control for fatigue, and employ identical instructions in both conditions. But the instructions are identical only for the experimenter.
For the subject, their meaning varies as his state differs. Since hypnosis increases a subject's motivation to follow the hypnotist's suggestions, those who make this comparison, in effect, are comparing a subject's performance in response to a simple request with one in response to a highly motivating suggestion. We have been able to show that subjects, properly motivated while awake, can exceed their hypnotic performance on such tasks as holding a kilogram weight at arm's length or pulling an ergograph. We see, then, not increased physical capacity in hypnosis but increased performance to be explained in terms of motivational differences.
Stage hypnotists often place a subject across two chairs, letting him be supported only at his shoulders and toes, and then have some member of the audience sit on his midriff. This illustrates the pitfalls in interpreting data on hypnosis. Every one of our simulators was able to perform it successfully. Anyone can do this in the waking state if he wishes to. It appears so dramatic that one is persuaded it must be due to hypnosis, yet it need have nothing to do with it. It is much too easy to attribute to hypnosis effects that may have nothing to do with it.
Trance Logic -- a Characteristic of Hypnosis
So far similarities between real and faking subjects have been discussed. There is a significant difference. We term it "trance logic" and view it as a "real," though not necessarily universal, aspect of hypnosis. Some hypnotized subjects show it; no simu-
686
lator ever does. To illustrate: let us assume I suggest to a hypnotized subject that he hallucinate seeing Dr. B. in an unoccupied chair. The subject may describe Dr. B. as sitting there, point to the exact place where he sees him, and yet add, "You know, I can see the chair right through him." This mixing of the real with the hallucinated occurs only with actual trance subjects. Simulators, given the same suggestion, will say that they see Dr. B. or that they do not see him, but never that they see through him.
There are other comparable kinds of behavior. Trance logic manifests itself in various ways. In an age regression study, for instance, one subject acted, talked, and wrote like a six-year old, yet correctly spelled all words in sentences like "I am conducting an experiment which will assess my psychological capacities." A simulator would not be so naive; he would attempt to talk and to write like a six-year old. But logical incongruity does not bother the hypnotized subject. It is a central feature of hypnosis.
Let the subject hallucinate seeing Dr. B. seated on an empty chair. Have him interact with this hallucinated figure. Then ask the subject to turn about and direct his gaze toward where Dr. B. is actually sitting. This is a type of situation described by Erickson. The simulator when asked "Who is sitting there?" responds to the hallucinated figure and talks to it, so that it seems the real thing. If he is directed toward the actual Dr. B., and asked who that may be, he responds, "No one is there," or "I don't know what that is." Simulators who were asked at the end of the experiment to state why they reacted in this way characteristically explained. "You told me he was sitting in that empty chair, so I went along with it, figuring that a hypnotized subject would see him there and talk to him. When you pointed to where Dr. B. actually was, I said, 'No one was there' (or 'I did not know who it was') "because you told me to see him in the empty chair and after all there is only one Dr. B."
This statement by the simulator is true and logical, but in the same situation a subject in deep hypnosis tends to react differently. Consider again the situation where it has been suggested that an hallucinated Dr. B. will be seen seated on the empty chair, after which the subject's gaze is directed toward the actual Dr. B. The hypnotized subject will look at him with a surprised ex-
687
pression, do a "double take," look back at the empty chair, see the hallucinated Dr. B. there, perhaps look back and forth a couple of times in a puzzled manner, and then perhaps explain, "Why, there are two of them," or "I did not know they were twins," or make some other similar and very superficial explanation. The hypnotized subject is not bothered by the sudden appearance of two identical individuals. If he concludes that one is hallucinated and is then asked which, he may gaze first at the hallucination, then at the actual person, and finally reach the correct conclusion. If asked how, he may explain, "My thinking would make the hallucination do things. So I thought he ought to shrug his shoulders. One did and the other didn't." Interestingly enough, the subjective reality of the hallucination may be preserved throughout this experience. Subjects with complete recall during inquiry will discuss the experience as one in which both images held the same reality value for them.
DISCUSSION
Dr. Paul E. Huston: If you requested a hypnotized subject to simulate not being hypnotized, would he then, when looking at Dr. B., see through him? If so, how could you ever know when anyone is hypnotized?
Dr. Martin T. Orne: If a subject is asked to simulate not being hypnotized, his behavior will depend on how he is given the suggestion. If he construes it to mean that he is to act as though completely awake, he will then refuse to hallucinate.
However, if the suggestion means to him that he is to look awake, act awake, and seem awake but still manifest all hypnotic phenomena, he will hallucinate much as he would without the specific suggestion. One hypnotized subject, for instance, delivered a lecture -- and his audience did not know he was hypnotized. He was instructed to lecture not only to the audience actually present but, as well, to an individual whom he was to hallucinate as sitting in the first row. The fact that he addressed himself to the hallucinated individual as well as to the students actually present, suggests that he was in hypnosis.
So far as knowing whether a particular subject is ever hypno-
688
tized, this to me seems a somewhat philosophical issue. I view the phenomenon of hypnosis as a subjective experience.
Dr. Huston: However, there is a difference between being a hypnotized subject simulating being awake, and a non-hypnotized subject simulating a trance.
Dr. Orne: If what we have termed "trance logic" is present, this appears characteristic of hypnosis. It is a strange mixture of reality and perception with induced hallucinations. It seems related to the concept of primary process. It cannot be attributed to the expectations of the hypnotized subject since the simulator does not show it. We prefer to call it "trance logic" to avoid theoretical argument while still investigating it.
While our approach in studying hypnosis has clarified some aspects of the phenomenon, it has obscured others. The ability of simulators to duplicate the behavior of hypnotized subjects does not mean that they experience similar subjective events. The use of simulators as controls has primary relevance to so-called transcendence of normal volitional capacity. Some basic alterations occur in the thought processes during hypnosis. These alterations, of which the subject has no previous knowledge, differentiate "reals" from simulators. But hypnosis as such tends to motivate the subject to respond to the requests of his hypnotist. Therefore, the use of a highly motivated control group like the simulators tends to obscure rather than facilitate an understanding of the course of this increased motivation.
A number of basic issues can be explored with the techniques described. These include: physiological responses to hypnosis, increased memory functioning, the possibility of inducing antisocial or self-destructive behavior in hypnosis, etc. Entirely different techniques are required to determine why some subjects enter hypnosis while others do not; and study of this may get us some leverage on the problem of where the apparent increase in motivation seen in hypnosis is derived. As of now, we can not begin to answer a variety of core issues -- the effect of the situation, the effect of the hypnotist, the effect of early training in hypnosis, the persistence of post-hypnotic behavior and its nature. All of these issues are relatively unexplored.
689
We have chosen to be somewhat pedestrian in our efforts, realizing that our initial task consists of isolating the artifacts of hypnosis from that which needs to be explained. Much theorizing has been premature, and much that has been theorized does not apply to the "real" aspects of the phenomenon. Only when the artifact has been separated from the essence -- which may be a long time hence -- shall we be in a position to develop a meaningful theory of hypnosis.