Hypnosis, Motivation, and the Ecological Validity of the Psychological Experiment 1

MARTIN T. ORNE 2

Institute of the Pennsylvania Hospital and University of Pennsylvania

INTRODUCTION

It is the hope of understanding the sources of human behavior that has always been the basis of psychology. Regardless of whether attention momentarily turned to other species, to simpler mechanisms, or to apparently remote mathematical concerns, the ultimate object of interest has always been man. The very scientists who objectified psychology and examined simple processes of behavior in animals revealed an intense and abiding concern for improving the human condition. Pavlov, for example, devoted many years to the study of psychiatric treatment techniques; Watson invented the first behavior therapy for phobias; Hull helped found the Department of Behavioral Science at Yale which merged anthropology, psychiatry, and psychology; and Skinner wrote Walden II.

1. The research from our laboratory which is reported on in this paper was supported in part by Contract #DA-49-193-MD-2647 from the U.S. Army Medical Research and Development Command, Grant #AF-AFOSR-707-67 from the Air Force Office of Scientific Research, Contract #Nonr 4731(00) from the Office of Naval Research, and by a grant from the Institute for Experimental Psychiatry.

2. I would like to express appreciation to my colleagues, Harvey D. Cohen, Kenneth R. Graham, and David A. Paskewitz, for helpful comments in the preparation of this paper. I am particularly grateful to A. Gordon Hammer, Frederick J. Evans, Emily C. Orne, and David Rosenhan for their detailed criticisms and many incisive suggestions. Appreciation is also due Karen Ostergren for her editorial comments.

187

188 Nebraska Symposium on Motivation, 1970

Some kinds of research can be carried out only with animals, but other questions which address themselves to human experience and complex human behavior can be asked only by studying human subjects. Not surprisingly, then, psychology's focus of interest has inevitably returned to man. In this discussion we will not be dealing with animal research. This paper will concern itself solely with experimental studies of human behavior and experience. Experimental studies of human behavior typically have considerable face validity. Generalizations from laboratory findings appear to have intuitive merit so that both the investigator and his scientific public are inclined to make the inferential leap to domains of behavior and experience beyond the laboratory. Such a leap is a complicated one, often not warranted on the basis of current evidence.

Consider the physical sciences. Assuming that conditions are kept constant, phenomena observed in the laboratory will tend to obtain outside the laboratory environs, but even here, under some circumstances, as Heisenberg has pointed out in atomic physics, the procedures involved in observing an event may so distort it as to make prediction impossible. This type of difficulty is a far more serious issue in psychology. Human behavior and experience may be distressingly modified when subjects are aware that they are being studied. The subject's participation in the psychological experiment may have significant motivational and perceptual consequences that can alter his behavior sufficiently to make inference to other contexts hazardous and, under some circumstances, misleading.

In most instances the purpose of carrying out an experimental study is the hope of answering significant questions about enduring human attributes, motivations, and behaviors. However, the psychological experiment is, by its very nature, what Garfinkel (1967) has called episodic. This means that it is regarded by everyone concerned as isolated from the rest of an individual's experience. It is implicitly understood that at the conclusion of the experiment the episode will be concluded and the individual will be basically unchanged. Fears expressed, actions undertaken, emotions felt in the context of an experiment are experienced as specific to that situation and intended not to carry over beyond it. In this regard it is analogous to other episodic events such as playing a game or having a good cry in response to a sad movie. Yet we hope to use observations obtained

189 Hypnosis, Motivation, and Ecological Validity

in episodic situations to draw meaningful inferences about enduring motivations of individuals as they manifest themselves in nonepisodic contexts.

It is possible that some mental processes are more validly scrutinized in the laboratory than others. Sleep, for example, though perhaps modified somewhat by the situation, will, on the whole, be the same process within the laboratory as outside of it; but the tendency to rouse in response to stimuli is likely to be affected, and the content of dream reports even more so.

Hypnosis, with its potential for bringing about a meaningful dyadic relationship in a remarkably short time and permitting apparently profound modifications of a subject's experience, would appear particularly amenable to laboratory research. Here, hopefully, is a sufficiently robust phenomenon such that its effects ought to swamp any differences the experimental situation may introduce. Yet our experiences in studying hypnotic phenomena have served to focus attention on the peculiar nature of the psychological experiment. I do not think it is a mere coincidence that observations on the nature of the psychological experiment grew out of studies of hypnosis. Hypnosis is a caricature of a meaningful dyadic relationship and, as such, emphasizes and throws into relief components that, while present in a wide range of other situations, would not as easily be recognized elsewhere.

In this paper I intend to discuss some of our attempts to understand hypnosis and the implications of our studies for the understanding of the special nature of the experimental situation. I will, then, outline a number of procedures I have called quasi controls that are designed to help the investigator evaluate the potential impact of the experimental situation itself on the data that are obtained. I will also briefly illustrate the use of these methods in other areas of research. Moreover, I will try to discuss the particular problems and limitations of quasi controls and the difference between these and conventional controls (see Orne, 1969).

Once the implications of the subject's recognition that he is the object of study are recognized, it becomes tempting to explain away rather than to try to understand broad areas of research findings. It seems appropriate that hypnosis, which first uniquely demonstrated these problems, should also provide an unusually

190 Nebraska Symposium on Motivation, 1970

compelling example of the limitations of the explanatory power of these ideas. For these reasons I have reviewed the evidence for and against the view that much of hypnosis can be understood by an exclusively motivational analysis. Thus, it has been widely accepted that the hypnotized subject has an increased motivation to please the hypnotist, a view that I myself formerly strongly supported (Orne, 1959b). This concept makes such sufficient intuitive sense that it had not been previously challenged, nor had it been tested empirically. When subjected to scrutiny, however, it becomes clear that it cannot account for the behavior of hypnotized individuals. Finally, I hope to outline constructive implications for psychological research that may be drawn from this work and the directions to which they point in future studies of human behavior and experience.

THE NATURE OF HYPNOSIS

Hypnosis is unquestionably a fascinating phenomenon, yet sooner or later most investigators become troubled by aspects of it that are not immediately apparent. With considerable distress they learn that their first-hand observations of hypnotic behavior seem not to be consistent with descriptions in the literature. Even more troubling, the more one reads the literature, the more confused the published picture becomes. It soon becomes evident that when the material is reviewed historically, very different behaviors have characterized hypnosis at different times.

Mesmer, for example, is considered the "father of modern hypnosis" (Binet & Féré, 1888; Boring, 1950), yet the behavior that occurred during his treatments does not at all resemble what is seen today. In his clinic, individuals would sit around an oaken tub filled with magnets, holding on to metal rods (to conduct the magnetic fluid). The surroundings were impressively carpeted and draped, music was played in the background, incense burned -- all designed to heighten expectation. Eventually, Mesmer, in magician's robes, would come down and, without uttering a word, lay his hands on one patient and then another. Patients would, one by one, appear to have a seizure, a hysteric fit, from which they went to sleep, only to awaken minutes or hours later without symptoms.

191 Hypnosis, Motivation, and Ecological Validity

We do not see hysteric seizures followed by sleep as part of hypnosis today. Nor is hypnosis pursued nonverbally today. Rather today we speak to our patients and they respond peacefully, as it were, and without hysterics. Similar discordances between descriptions of behavior generally recognized to be hypnosis today and those of an earlier period can be found in the work of Coué (1922), not to mention Charcot (1886), who, for instance, indicates that if the top of the subject's head is rubbed he will pass from the stage of sleep to the stage of somnambulism.

It seemed to me important to determine how a phenomenon which everyone agrees is the same may be characterized by such variant behaviors. On the basis of White's (1941) theoretical framework, an experiment was developed to test the hypothesis that subjects in hypnosis behave in whatever manner they believe characteristic of hypnotized individuals (Orne, 1959b). For this purpose it was desirable to find an item of behavior that had not previously been associated with hypnosis and yet might seem plausible to the average college undergraduate. Such an item, by the way, was difficult to come by in view of the broad range of behaviors that, at one time or another, have been attributed to hypnosis. After some search, however, I came up with catalepsy of the dominant hand.3 While catalepsy itself is believed by some authorities to be an invariant accompaniment of hypnosis, when it occurs it always occurs in the entire body -- both hands, both feet, even the trunk. It had never been reported to occur in one hand while the other remained flaccid. Nevertheless, catalepsy of the dominant hand might sound plausibly scientific to an undergraduate psychology student, vaguely reminding him of learning about crossed hemispheres, stuttering, etc.

Two matched sections of a large college class were given a lecture on the nature of hypnosis which included a demonstration. Three "volunteers" were selected from the class and hypnotized. (Unbeknownst to the group, these individuals had been previously hypnotized and given a posthypnotic suggestion that the next time they entered hypnosis they would manifest catalepsy of the dominant hand.) Hypnosis was induced with the volunteers, and the

192 Nebraska Symposium on Motivation, 1970

classic hypnotic phenomena were then demonstrated. Casually catalepsy was tested and, in passing, it was mentioned that catalepsy of the dominant hand was a typical hypnotic behavior. Two of the subjects happened to be right-handed and one left-handed, illustrating the point. In all other respects the lecture was accurate and the subjects were actually deeply hypnotized. The parallel class received the identical lecture and demonstration with the same three subjects. The only difference was that catalepsy was not tested and not commented upon.

Approximately one month later subjects from both classes were invited to come to the laboratory to participate in an experiment. When they were hypnotized we observed a new characteristic of the hypnotic trance -- catalepsy of the dominant hand -- associated exclusively, of course, with attending the appropriate lecture!

This experiment, trivial from one point of view, has, nevertheless, broad implications. It would appear that in hypnosis we are dealing with a chameleon. In other words, the subject's hypnotic behavior is influenced by whatever behaviors he believes to be characteristic of being hypnotized.

Today there seems to be considerable consistency in hypnotic behavior. I once asked a large number of individuals at random to describe hypnosis, and found that, despite initial disclaimers to the contrary, they invariably were able to describe the behaviors displayed by hypnotized people. The general views about the phenomenon can apparently be traced back to the descriptions given by Thomas Mann (1931) in Mario and the Magician and by George Du Maurier (1895) in Trilby. Although, of course, not everyone has read these two particular works, an incredibly large number of fictionalized accounts take their inspiration from them. We have been unable to find adults in our culture who do not have at least a rudimentary acquaintance with the kind of behavior which characterizes hypnosis.

The subject's prior knowledge about hypnosis is, of course, only one part of his knowledge of what constitutes appropriate behavior. Cues from the hypnotist, on an ongoing basis, also serve to define how a subject should behave. For this reason it is almost impossible to determine what precisely constitute the core phenomena of hypnosis. Working with children did not avoid the problem.

193 Hypnosis, Motivation, and Ecological Validity

since even some six-year-olds appeared to have a reasonably accurate concept of what constitutes hypnotic behavior.

It seemed another way around this problem would be to ask anthropologists to work with a native population. One might reasonably assume that African Bushmen would not have read the slick magazines or been influenced by movies or television. These individuals would be truly naive. It sounded encouraging when a former student described hypnotized Bushmen as behaving very similarly to college undergraduates he had seen in hypnosis. The encouragement was short-lived, however, when it became clear what had occurred. The hypnotist, working through an interpreter, had proceeded to induce hypnosis. How was he to know, however, whether an individual was hypnotized? In order to recognize the state and report on it in some detail, he had very carefully, but without awareness, shaped the behavior until it resembled that with which he was familiar. This behavior pattern rapidly then became stable and reliably associated with hypnosis. In other words, while natives had not heard of hypnosis, the hypnotist had, and he then induced the kind of behavior with which he was familiar. Of course these observations can tell little about what the intrinsic behavioral characteristics of hypnosis are.

It should be noted that the hypnotist was proceeding in a rigorously empirical fashion to study hypnosis by hypnotizing naive subjects and observing their responses. He was able to note the differences in their behavior in hypnosis and outside of hypnosis. Unfortunately, the observed behavior was a function more of his preconceptions than of hypnosis. For example, had the hypnotist believed that catalepsy was an invariant accompaniment of hypnosis (a view, in fact, currently held by several outstanding clinicians), he would have been able to report that, in his experience, this is true not only with American subjects but even with naive natives. Thus we seem to be dealing with a phenomenon whose behavioral manifestations include whatever the hypnotist or the subject believes them to be. Two hypnotists with differing views can go out into the real world and validate their views by careful experiments without necessarily being aware that the behavior of their subjects mirrors their own preconceptions. It is not surprising, then, that the literature of this field has been characterized by an inability to

194 Nebraska Symposium on Motivation, 1970

replicate and acrimonious disputes about the inaccuracies of other people's research.

Demand Characteristics in Hypnotic Research

When I became aware of the extent to which subjects' expectations and subtle cues from the hypnotist may affect the behavior of the hypnotized individual, there seemed to be no reason to assume that this would hold only for the general behavior which characterizes hypnosis; it should also affect those specific behaviors that are considered data in an experimental context using hypnosis. Subjects should respond not only to specific verbal suggestions but also to the context in which these suggestions are made. The behavior that is expected and desired of them in an experiment would, of course, be communicated only partially by the specific suggestions; it would also be communicated by subtle cues which the subject might have gathered about the experiment beforehand, cues given by the experimenter in the course of the study, and, perhaps most importantly, by the experimental procedure itself. The sum total of these cues which would communicate the purpose of the research, the hypothesis that the investigator hoped to demonstrate, the kind of behavior he expected to find, I termed the demand characteristics of the experimental situation.

I was first impressed with the effect of such variables while carrying out pilot work to elucidate the nature of hypnotic age regression. Controversy has long existed about the extent to which hypermnesia can reliably be induced by age regression. In previous work I had observed that in addition to increased memory one also finds increased confabulation (Orne, 1951). This observation had earlier been made by Stalnaker and Riddle (1932). It seemed reasonable to use the well-known Carmichael, Hogan, and Walter (1932) effect in serial recall as a means of studying this process. In their classic study it had been shown that labeling ambiguous figures would, with serial recall, yield a progressive change in reproductions in accordance with the labels. Thus the subject would be shown the original labeled ambiguous figures and then asked to draw six serial reproductions of these figures; he would then be age-regressed back to the time when he observed the original picture, and told to copy his hallucination. This should establish a

195 Hypnosis, Motivation, and Ecological Validity

clear mnemonic drift in the direction of the label and away from the original figure. It would then be possible to discover whether age regression led to increased accuracy of recall or, instead, to increased confabulation.

The seven drawings could be randomized and then given to "blind" judges with the instruction to order them in the sequence in which they were made. Accuracy of recall would cause judges to place the seventh picture (the copy of the hallucination) early in the series, whereas an absence of increased memory or an increase in confabulation would leave the picture in seventh place, perhaps even cause judges to ask if some drawings were missing.

One subject showed the Carmichael, Hogan, and Walter effect to a striking degree. Thus when he reproduced an ambiguous picture that had been labeled "canoe," he was not satisfied to add pointed canoelike tips, but by the second time had added a paddle, and by the third and fourth drawings he had managed to fill in someone sitting and paddling, with his girl friend sitting in the bow, eventually adding a guitar, picnic basket, and other things which seemed to belong to the scene. Somehow the effect was working too well, and on a hunch the subject was told, "That's fine but now I want you to draw the design as you originally saw it." Very agreeably he accurately reproduced the original picture. When asked why he had elaborated the drawings so dramatically he explained that if the label said canoe, this would clearly be what was wanted, and if he was asked to reproduce it several times at half-hour intervals, naturally he would be expected to improve the resemblance to a canoe. Why else would he be asked to do this?

The subject's observation was extremely instructive. Incidentally, I have been unable to replicate the Carmichael, Hogan, Walter study despite several efforts to do so, and wonder about the extent to which similar factors may have determined the original results.

Subsequently, becoming more interested in this kind of effect, I replicated the Ashley, Harper, and Runyon (1951) study which had been planned to reconcile a controversy about the effect of economic status in children on perceiving coin size. At the time there was considerable controversy about the Bruner-Goodman effect (1947), which was that poor children perceive coins as larger than do rich children. This effect, while repeatedly replicated in

196 Nebraska Symposium on Motivation, 1970

some laboratories, could not be found by others. Ashley, Harper, and Runyon had argued that the difficulty in showing a clear effect was the interindividual variability and what ideally would be desirable were perfectly matched subjects who differed only in the extent to which they perceived themselves to be rich or poor. Further, they argued, this could readily be obtained by hypnosis. Thus, the same subject could be tested in his normal state and in hypnotically induced rich and poor states. One would thereby have perfectly matched "groups" of subjects, differing only in the extent to which they believed themselves to be rich or poor. Using this technique, Ashley, Harper, and Runyon obtained data clearly supporting the Bruner-Goodman hypothesis.

A pilot study for the replication was carried out with a few subjects (Orne, 1959b) and it was clear that Ashley, Harper, and Runyon described a powerful effect indeed. In brief, the experimental procedure was to hypnotize the subject and obtain his coin size estimates, then to induce amnesia for the individual's past life and to provide him with a different life history. With the "poor" suggestion he was told that he came from a large family, that life had been hard on him, that while he managed to get by, there was never quite enough to eat, and that he had had to work extremely hard from an early age contributing to the family; it was possible for him to go on in school only by obtaining complete scholarships, and even then he worked extra to try to send whatever money he could home, and so on. When these suggestions were given, the subject's whole behavior changed. He became extremely intense in his efforts, seeming to care a great deal about the procedure, and when asked to make estimates by matching the size of a light spot of variable diameter to that of particular coins he would grossly overestimate. Again amnesia was induced for this experience and a new life history was provided, indicating that the student had come from a very wealthy family, that he could remember in his childhood how the chauffeur would take him to school in the Rolls, that he had lived in a big house, that his friends had envied him, that he had gone to the best schools and belonged to the best clubs, that he was somewhat annoyed because his monthly allowance had been cut to $2,000, which was barely enough to squeeze by on, and so forth. With these instructions the subject behaved very differently,

197 Hypnosis, Motivation, and Ecological Validity

slouching and leaning back in the chair, showing a certain amount of contempt for the situation, apparently not caring about what he was doing, while judging the size of the coins considerably more accurately, with perhaps a slight tendency to underestimate.

There was one additional procedure which had been carried out by Ashley, Harper, and Runyon, which was to give each subject a metal slug four times with statements that it was made of lead, silver, gold, and platinum, respectively, and then to ask for size estimates for those identically sized disks. Again a lawful change in estimates of the disk size was shown, particularly under the poor condition, depending upon the value of the presumed composition of the disc.

When subjects were awakened and memory was induced for the experience, they were asked what the experiment was about. Typically they said they didn't know, weren't sure, and so forth. When pressed, without having been given cues as to what was expected, they suddenly began to explain that they supposed that I expected money to mean more to them if they were poor than when they were rich and that, therefore, it ought to seem bigger in the poor condition, whereas a rich boy would most likely see it as smaller.

It should be emphasized that nothing which had been said, as such, would have communicated why the experiment was being done. The suggestion per se would not have helped. However, the procedure itself was a dead giveaway. From the point of view of the subjects, they were being asked to make coin size estimates three times, the only difference being suggestions concerning the extent to which they saw themselves as rich or poor. If this was not sufficient to get the idea across, there was the matter of being given identically sized disks and asked to make coin size estimates, after being informed that the disks were composed of different metals. Certainly if the experimenter went to the trouble of telling them what these disks were made of, he anticipated differences in their responses, and the only parameter that was available for such differences to occur was that of perceived size. Therefore, it was the procedure rather than anything which the subjects were told that communicated the experimenter's expectations.

One subject, however, had a different idea. He came to an alternative, plausible hypothesis, which was that the poor students

198 Nebraska Symposium on Motivation, 1970

would work harder and be more accurate in their estimates, whereas rich students would not care and their estimates would therefore go all over the lot. When the data were subsequently examined, it turned out that the subjects who formulated the Bruner-Goodman hypothesis yielded data which fully supported this hypothesis. The only aberrant subject was the one with the different hypothesis, and in his case the estimates made while poor were more accurate and showed less variability than those he made while rich. These data indicated that it would be vital to take into account the subject's perception of what was expected -- what I called the demand characteristics of the experiment -- in order to understand his behavior.

In a more systematic way another group of subjects was run in the Ashley, Harper, and Runyon design and careful post-inquiries were carried out. These inquiries were judged by independent judges, and a high degree of association could again be demonstrated between the perceived hypothesis of the experiment and the subject's performance.

It should be emphasized that the effect which demand characteristics have on behavior in such a situation ought not to be conceptualized as voluntary compliance with what the subject believes is expected of him. While this may in some instances occur, the more typical situation is one in which the individual's experience is changed by the situation. Thus Mesmer's patients did not voluntarily choose to behave as if they had a hysteric seizure; rather such a seizure resulted from the pressures of the total situation. If an appropriate interview could have been carried out with Mesmer's patients, it is likely one would have elicited the information that a seizure of this kind characteristically "happens" in the course of treatment.

Unfortunately, it was not possible with a procedure of this kind to determine whether subjects were actually responding to the demand characteristics of the situation or whether they were responding in line with the original hypothesis and subsequently, by reflecting upon their responses, deducing what had been expected. Accordingly, I wanted to devise a technique to explore what subjects might be perceiving in the situation without, however, first giving them the opportunity of responding in hypnosis. It could

199 Hypnosis, Motivation, and Ecological Validity

not then be argued that the subject's response was predicated on the subjective experience of being poor or rich.

It was crucial in running a comparison group of this kind that it be exposed to the identical set of cues provided for the hypnotic group, since I had observed that hypnotists characteristically treated hypnotized subjects differently from waking individuals. One way to try to solve this problem was to ask subjects to simulate. But the literature was very explicit in indicating that subjects are unable to simulate successfully, and when I first applied the procedure it became clear that this was true. Thus, when I asked a subject to simulate hypnosis he would begin somewhat reluctantly and, after a few minutes, look at me somewhat sheepishly, smile, and ask, "Am I doing all right?" He might then proceed to simulate a few more moments, only to interrupt again or in some other way clearly indicate his ambiguous status. It would have been obvious to any discerning layman that these subjects were not hypnotized.

On reflection, however, it became clear why this was the case. Subjects knew that I knew they were simulating. They were being asked to carry out an essentially impossible task, namely, to simulate hypnosis knowing full well that I was aware of their actual status. There was no way in which they could "win."

A relatively minor variation of this procedure altered the situation in significant ways. In it, one of my associates instructed the subject, who had repeatedly been unable to enter hypnosis despite wishing to do so, as follows: "Your task today will be somewhat different. You are going to try to fool Dr. Orne and make him think you are actually going into deep hypnosis. He knows that he will be working with highly hypnotizable individuals who are able to enter deep trance but that some subjects will try to simulate. He will not know which is which, however. I can't tell you how to do this task. Do the best you can based on whatever knowledge you have of hypnosis. However, if Dr. Orne catches on to the fact that you are simulating he will stop the experiment immediately. So don't think you have given yourself away, for as long as he continues with the experiment, you will have been successful in fooling him. This is a difficult task but we have found some intelligent subjects are able to do it successfully. Good luck!" Given this simple instruction, untrained, unhypnotizable individuals proved capable of

200 Nebraska Symposium on Motivation, 1970

behaving in ways which were exceedingly difficult to distinguish from the behavior of deeply hypnotized subjects. Without special test procedures experienced hypnotists were unable to discriminate these individuals at better than chance levels.4

The simulating subjects provided a way of evaluating in behavioral terms the kind of demand characteristics implicit in the experimental situation. We now had a group of subjects who were maximally responsive to the demand characteristics. When the Ashley, Harper, Runyon experiment was repeated with highly hypnotizable and simulating subjects run blind, we observed identical results in both groups.

It should be emphasized that the failure to find differences between subjects in the Ashley, Harper, Runyon design who are deeply hypnotized and those who are simulating tells us absolutely nothing about the Bruner-Goodman effect. It merely indicates that a highly plausible alternative hypothesis exists to explain the Ashley, Harper, Runyon finding: namely, that the response is a

201 Hypnosis, Motivation, and Ecological Validity

function of the demand characteristics of the situation rather than the direct effect of the hypnotic suggestion to be "poor" or "rich."

Compliance as an Inadequate Explanatory Mechanism

With the recognition that many findings about hypnosis can be explained by the subject's response to the demand characteristics of the situation, it is inevitable that attempts should be made to explain the entirety of the phenomenon on the basis of such mechanisms. In an early attempt to conceptualize hypnosis (Orne, 1959b) I built upon the formulations of Hull (1933), White (1941), and Sarbin (1950) and proposed that hypnosis is characterized by (1) a desire on the part of the subject to play the role of the hypnotized subject, (2) an increase in suggestibility which was translated as an increase in motivation to conform to the wishes of the hypnotist, and (3) a less well defined aspect which I regarded as the essence of hypnosis, in the context of which increased motivation to please and a tendency to play the role of the hypnotized subject could be seen as epiphenomena. In order to understand the essential characteristics of hypnosis it would, of course, be necessary to determine how much of the phenomenon could be explained in terms of role playing and an increased motivation of the subject to please the hypnotist.

The close similarity between an increased motivation of the subject to please the hypnotist and a conceptualization such as an increased responsivity to the demand characteristics of the situation is not accidental. I assumed that the increased motivation to please the hypnotist was an intrinsic attribute of hypnosis. Perhaps because the formulation intuitively seemed sound or because it was congruent with the Zeitgeist, it was not challenged despite the absence of empirical evaluation. Indeed, it was perhaps inevitable that others would conclude that, because simulating subjects could mimic much of the behavior of the hypnotized individual, one could explain away the actual phenomenon. Not only has there been a tendency to assume that all hypnotic phenomena can be understood in terms of demand characteristics, but also demand characteristic effects have erroneously been equated with compliance.

Relatively recently, influenced by the work of London and Fuhrer (1961) and Rosenhan and London (1963a), I have begun to

202 Nebraska Symposium on Motivation, 1970

examine empirical evidence bearing on the hypothesis that hypnosis increases the subject's motivation to please the hypnotist. It seems appropriate to review this work in the context of this paper, partly because it deals directly with motivational issues, but even more so because it clearly demonstrates the danger of explaining away rather than understanding a phenomenon. Certainly it shows that an uncritical -- and, in my view, inappropriate -- use of the concept of demand characteristics cannot adequately account for a complex phenomenon such as hypnosis.

There are three ways in which the concept of motivation to please the hypnotist or to comply with his wishes has been used to account for hypnotic phenomena: first, by way of the view that an individual who has been hypnotized by an experimenter is then more motivated to follow his instructions than if he had not been hypnotized; second, by way of the view that the tendency for an individual to enter hypnosis in the laboratory is merely a specific example of the general tendency to comply with requests of the experimenter (in other words, those individuals who enter hypnosis during an induction procedure are the very ones who initially already had a greater motivation to please the experimenter than those who do not respond in this way) ; and, finally, by way of the view that posthypnotic phenomena do not represent a special class of events but can simply be understood in terms of the subject's continued intense wish to please the hypnotist.

The Effect of Inducing Hypnosis on the Subject's Wish to Please the Hypnotist

On the basic question of whether the induction of hypnosis increases the motivation of the subject to comply with the wishes of the hypnotist it is difficult to obtain clear-cut findings. The widely held conviction that this is true is based on casual observations that hypnotized subjects seem to do whatever is requested of them. For this reason I normally begin my lectures to students by asking one individual to remove his right shoe, another to take off' his necktie, still another to give me his wallet, and still another to exchange his coat with his neighbor. After obtaining compliance with a number of utterly ridiculous requests of this kind, I ask the students whether they have been hypnotized, and upon receiving a

203 Hypnosis, Motivation, and Ecological Validity

negative answer, I am able to point out that if they had previously gone through an induction process and had been asked to carry out the identical actions, observers would have said, "See, this demonstrates how a hypnotized individual is controlled by the hypnotist." Obviously this kind of control was already inherent in the student-teacher relationship; however, because the full amount of control one person might exert over another is normally neither exercised nor measured, it appears that hypnosis causes a significant increment in compliance. In a variety of ways we have tried to find behaviors which subjects would not carry out in the context of an experimental situation. We have been unable to do so. Thus, it has not been possible to demonstrate a unique increment in control due to hypnosis. (It was in the context of this question that the experiments in serial addition which will be referred to later were carried out.)

It is possible to phrase the question concerning the hypnotic subject's endurance thus: Will an individual in hypnosis perform a difficult task longer than he will out of hypnosis? Several studies have shown increased performance during hypnosis if one does not specially motivate the waking state (see Hull, 1933). Here the question of appropriate controls is exceedingly complex. This problem was traditionally phrased in terms of whether it is possible for hypnosis to cause an individual to transcend his normal volitional capabilities. Several studies have shown that it is relatively easy with appropriate motivation for nonhypnotized subjects to exceed their previous hypnotic performance (Orne, 1954; Barber & Calverley, 1964; Levitt & Brady, 1964). There can be no doubt that under these conditions the kinds of instructions used during hypnosis are quite different from those used in the waking state or for a control group. Furthermore, if subjects think that the experimenter is attempting to demonstrate increased performance during hypnosis they may easily provide him with supporting data, not by increasing their hypnotic performance but by decreasing their waking performance (see, for example, Zamansky, Scharf, & Brightbill, 1964).

One ingenious solution, eliminating differences in instructions, has been used in the study by London and Fuhrer (1961). These workers used identical hypnotic procedures for very good and very poor hypnotic subjects, both groups having been convincingly told that their hypnotic performance was adequate for the purposes

204 Nebraska Symposium on Motivation, 1970

of the experiment. With a variety of physical tasks the authors report the startling finding that unsusceptible subjects showed a significantly greater increment in performance in response to hypnotic instructions than good hypnotic subjects. Certainly these findings do not support the general belief that hypnosis increases the motivation of the hypnotized individual to carry out whatever kinds of tasks are suggested. Evans and Orne (1965) have replicated these studies and, while their data did not reach significance, they showed trends in the direction reported previously. In any case, the results, again, do not support the view that there is an increase in performance with hypnosis.

As a subsidiary question in this section we may also ask whether hypnosis, without any specific posthypnotic suggestions, has the effect of enhancing a subject's wish to please the hypnotist after the trance is terminated. This is of particular interest to clinicians, for it has often been said that the induction of hypnosis alters the relationship between doctor and patient in such a way that the patient becomes more prone to accept, uncritically, requests by the doctor and, in some ways, becomes more dependent upon him.

We have carried out a study which bears upon this question. It makes use of a postcard-sending procedure that has also been employed in other studies to be reported later in this paper. It is therefore first of all necessary to describe this experimental procedure. (This technique was tested and then utilized to study posthypnotic behavior by Esther Damaser [1964] in our laboratory.)

The Effect of Laboratory Instructions on Behavior over a Period of Weeks

A technique was needed which would measure the extent to which individuals would persist in carrying out requests without any further reinforcement beyond knowing that their behavior was of interest and importance to the research study, and without knowing or understanding how that behavior would be used. We decided upon a task which would yield data over as long a period as desired and which would not necessitate the subject's returning to the laboratory or being given any feedback whatsoever (Orne, 1963). Particularly well suited to these criteria is the request to mail one postcard each day to a post office box. The subject can be

205 Hypnosis, Motivation, and Ecological Validity

given any desired number of cards and appropriate instructions. The receipt of the subject's card provides quantifiable, objective data about the effectiveness of instructions in inducing individuals to carry out a trivial but tedious daily task.

The first study, concerned mainly with evaluating the procedure, was designed to investigate whether or not a single experimental manipulation would visibly affect the rate of postcard sending, or whether individual variability was so great that the task would prove to be useless for experimental purposes. We postulated that the primary variable involved would be the manner in which the task was defined rather than monetary reward as such; in particular, that if subjects agreed to perform the task and were paid in advance for so doing, noncompliance would evoke guilt.

Thirty-four subjects solicited through student employment services were randomly assigned to four groups.5 All subjects were given a paper-and-pencil test and then one interview. During this interview they were told that another part of the experiment involved their doing something every day for eight weeks, that this something could be done at home and that it didn't take much time. The subjects were asked if they would like to participate in this part of the experiment. Three subjects refused, saying that they had enough to do without burdening themselves further; all the other subjects agreed to do this "something." Subjects were told that the reason for getting their commitment before explaining the task was that we did not wish them to discuss the experiment, and we felt it less likely that subjects who were actually taking part in the study would discuss it. After the subject had agreed to do this undefined something every day for eight weeks, he was given 56 preaddressed postcards and instructed to mail one every day.

Experimental groups differed as follows:

Group A. Subjects were paid $2.50 and were told that this sum represented payment for the entire experiment.

Group B. Subjects were paid $8.10 and were told that this sum represented payment for the entire experiment.

Group C. Subjects were paid $2.50 and were told that this sum was payment for participation in the paper-and-pencil test.

206 Nebraska Symposium on Motivation, 1970

They were further told that they would be paid 10¢ for every postcard that they mailed, and that they would be paid this sum after the eight weeks had passed. They were told that they would be paid for only one postcard each day during the 56 days.

Group D. Subjects were paid $2.50 and told that this sum represented payment for participation in the paper-and pencil test. They were then also paid 10¢ per postcard and told that they were being paid in advance because we were confident that they would mail all of the postcards. Thus each of these subjects was given a total of $8.10, the same as Groups B and C, but of course the significance of this money differed for the three groups.

In analyzing the data, the number of postcards sent was corrected in such a fashion that subjects could get credit for only one postcard each day. As predicted, Group D performed differently from the other three groups in having a higher mean and a smaller variance (F = 8.243; .01 < p < .02). None of the other three groups differed significantly from each other either in their means or in their variances.

207 Hypnosis, Motivation, and Ecological Validity

This finding, incidentally, confirmed our belief that subjects' motivation to undertake experimental tasks is not essentially a function of payment within the range of rewards usually given in laboratory research. However, the importance of the finding here is that it demonstrates that postcard sending over time is a behavioral task sensitive enough to be used as a measure of motivation.

The Effect of Hypnosis on Responses to Subsequent Requests

We are now prepared to return to the study (Orne, 1963) concerned with the effects of hypnotic trance on the subsequent relationship between the participants. It utilized 53 volunteer subjects for hypnotic research who were given 100 postcards with the request to mail one daily, much in the same fashion as previously described. If the successful induction of hypnosis does somehow make the experimenter more important to the subject, then the request to send postcards should be more effective with subjects who responded by going into hypnosis than with those subjects who failed to do so.

The 53 volunteer subjects were first administered an abbreviated version of the group adaptation (Shor & E. Orne, 1962) of the Stanford Hypnotic Susceptibility Scale, Form A (Weitzenhoffer & Hilgard, 1959). Subsequently they were individually given a modification of a preliminary version of the Stanford Hypnotic Susceptibility Scale, Form C (Weitzenhoffer & Hilgard, 1962), adapted to permit the administration of perceptual items and allowing an 11-point scale of hypnotic depth to be used. At the conclusion of the hypnotic part, subjects were asked by the same hypnotist-experimenter, in the same fashion as in the previous experiment, whether they would be willing to participate in something else. One subject did not agree to participate in the postcard tasks. All others were given 100 self-addressed postcards. The correlation was computed between postcard sending and hypnotizability.6 It was found to be -.11, obviously insignificant, and in a direction opposite to that required by the "compliance hypothesis." We feel compelled to conclude that the assumption of stronger compliance motivation on the part of successfully hypnotized subjects is not justified.

208 Nebraska Symposium on Motivation, 1970

The Relationship between the Subject's Willingness to Please the Experimenter prior to Hypnosis and His Subsequent Response to Hypnotic Suggestions

Over the years we have worked with a large number of subjects. For methodological reasons we have been interested in the extremes of the distribution -- the highly susceptible and largely insusceptible individuals. One would expect that if good subjects are particularly motivated to please the experimenter they would be more reliable in keeping appointments and arriving punctually. Yet Emily Orne noticed in several studies which used insusceptible subjects and very good somnambulists that the insusceptible individuals were consistently more reliable and punctual. This was true despite the fact that everyone in the laboratory is exceedingly careful to treat subjects with consideration and to communicate our appreciation for their participation. This is particularly true with highly hypnotizable individuals, since they represent a relatively small proportion of the population and are essential for the research. (In studies which require subjects to come back more than once, the attrition rate of good somnambulists is especially striking.) For some time we have been studying arrival times in the laboratory. Evans and Emily Orne have been able to show consistent significant negative correlations between punctuality and the ability to respond to hypnosis. (These correlations also hold particularly well for later sessions.)

We have also done a pilot study with a small number of subjects (N = 17) evaluating the relationship between postcard sending and hypnotizability when the request is made before the question of hypnosis is raised with the subject. With a small sample of this kind no significant trends emerged. If there is a trend, it is unlikely to be of any significant magnitude. Neither of these results justifies the assumption that a stronger motivation to comply is characteristic of individuals who subsequently show that they are able to enter hypnosis.

The data most relevant to the relationship between effort and hypnotizability, where hypnotizability was measured afterward, were obtained by Rosenhan and London (1963a), investigating the baseline performance of subjects on physical endurance as measured by a hand dynamometer and weight holding, before the possibility that the subjects were to be hypnotized was mentioned. Significant

209 Hypnosis, Motivation, and Ecological Validity

differences between susceptible and insusceptible subjects emerged, indicating a superior performance on the part of the insusceptible subjects. To the extent that we may assume no difference in physical capacity correlated with hypnotizability, these data suggest that subjects who prove to be good hypnotic subjects before the induction of hypnosis are, if anything, less motivated to follow the instructions of the experimenter than individuals who subsequently show themselves to be insusceptible. In a subsequent paper, the same authors (Rosenhan & London, 1963b) found an opposite trend in baseline responses related to the learning of nonsense material and suggest that baseline differences may be task specific. While further work is required on this issue, the indications are that those who become good somnambulists are not characterized by the general tendency to be highly motivated to carry out instructions.

Some data are available concerning baseline performances of subjects who know that hypnotic experiments will be conducted. Under these circumstances London and Fuhrer (1961) report that insusceptible subjects performed better on a variety of performance tasks. Shor (1959), during an experiment which required subjects to choose the highest level of electric shock that they would be able to tolerate, found good hypnotic subjects stopping at a considerably lower level of intensity than insusceptible subjects. Care must be taken in the interpretation of these kinds of data, however, as Zamansky, Scharf, and Brightbill (1964) have demonstrated that the relatively poor baseline performance of good hypnotic subjects may be a function of their wish at some level to have their hypnosis performance appear particularly dramatic.

POSTHYPNOTIC PHENOMENA

If one wishes to explain hypnosis as simply a function of the subject's wish to please the hypnotist, posthypnotic suggestion is of particular interest. It is the phenomenon which is usually associated with the notion of increased behavioral compliance on the part of the subject and, more than any other, implies an increased control over the individual's behavior. Yet here too, as we shall see, data do not support such an explanation.

Two studies were performed by Hull's group (1933) to measure

210 Nebraska Symposium on Motivation, 1970

the persistence of posthypnotic behavior. Kellogg (1929) suggested to subjects that they would, while reading a book of sonnets, breathe twice as fast as usual on even-numbered pages and twice as slowly as usual on odd-numbered pages. Subjects were tested immediately after waking, on the following day, and then at intervals of one to two weeks and, in some cases, at longer intervals of up to 99 days. A control group of subjects was not hypnotized but given the same instructions. Kellogg's conclusions were as follows:

The performance of the waking control subjects shows no loss in the power of the suggestion with time. In all cases, at the last time of testing, nine or ten weeks after the instructions were given, the response is greater than at the first test. The performances of the trance subjects, on the other hand, show a loss in the force of suggestion with time, the ratio obtained at the last test being in all cases lower than at the first waking test. [p. 507]

Because Kellogg's experiment involved practice effects from repetition of the response, Patten (1930) investigated the persistence of posthypnotic suggestion over time, using each subject only once. They were instructed to press their right forefinger down when the names of animals appeared while they were looking for words which repeated themselves on the list revolving on a memory drum. Subjects were tested at time intervals varying from 0 to 33 days after the posthypnotic suggestion had been given. A waking control group was employed which was given the same instructions. The conclusions Patten draws are:

The magnitude of the response of the amnesic subjects decreases with time. The magnitude of the response of the conscious subjects decreases somewhat with time.... The reactions of the amnesic subjects are not as large nor as persistent as those of the conscious control subjects and they exhibit a somewhat faster decrement with time. [p. 325]

Damaser (1964) in our laboratory conducted a study of posthypnotic behavior utilizing the sending of postcards, which is a task rather well suited to an investigation of this phenomenon. It was felt important to control for possible personality and treatment differences between good and poor hypnotic subjects. For this reason a somewhat different design from that of the earlier study was used. From a large number of subjects, a group of very good hypnotic subjects and a group of subjects who were able to achieve only

211 Hypnosis, Motivation, and Ecological Validity

medium hypnotic depth were selected. After the decision was made that a subject was adequate for the deep group, he was randomly assigned to one of three categories. The first group was given a posthypnotic suggestion in hypnosis that they would send one postcard each day, and they then were awakened with suggestions of amnesia. The second group was given the identical posthypnotic suggestion and then awakened with amnesia, but in the waking state was also requested to send one postcard each day. The third group was deeply hypnotized but not given any posthypnotic suggestion except amnesia; however, after being awakened each subject was asked to send a postcard every day. Thus of the three groups, one had "posthypnotic suggestion alone," the second had "posthypnotic suggestion plus a waking request," while the third had "waking request alone." These groups had been carefully equated for hypnotizability and had been exposed to the same amount of interaction, both hypnotic and waking, with the experimenter. The same design was carried through with the medium hypnotic group; here, however, the amnesia instructions for posthypnotic behavior were not given. The results are shown in Table 2.

212 Nebraska Symposium on Motivation, 1970

Because of technical difficulties the sample sizes are too small to draw definitive conclusions. However, it will be noted that the deep and the medium "posthypnotic suggestion alone" groups sent considerably fewer cards than the other two groups. This finding is consistent with those reported by Kellogg (1929) and Patten (1930).

It must be pointed out that great care is needed in interpreting the observation that a simple request appears to elicit behavior from laboratory subjects more reliably over a longer period of time than a posthypnotic suggestion. From the point of view of the subject the task may be entirely different. Thus, the individual who has been given a posthypnotic suggestion tends to perceive his role in the experiment as complying as long as he feels compelled to do so; some subjects even interpret their role as trying to resist the suggestion. Characteristically, subjects do not feel it appropriate to push themselves to carry out suggestions, as they would perceive this to be cheating. This makes sense if the subject correctly appraises the purpose of the experiment as an attempt to measure the response to a posthypnotic suggestion. On the other hand, a subject complying with a waking request who has committed himself to the task views it as his obligation to carry it out as long as he possibly can. We should, therefore, not be surprised by the fact that the latter group of individuals carry out behaviors more consistently and over a longer period. It is for this reason that Damaser (1964) included the group which combined waking request and posthypnotic suggestion. We had hoped somehow to potentiate the waking request by means of posthypnotic suggestion. While results based on such small samples are presently equivocal, they do not point in that direction. If anything, posthypnotic suggestion interfered with the waking request, though the differences are not statistically significant.

Because of the different ways in which subjects interpret a waking request and posthypnotic suggestion, it is difficult to design an experiment with appropriate controls. Perhaps we need to be satisfied with recognizing that posthypnotic suggestion is a different phenomenon from the waking request. Certainly the consistently poorer performance of good hypnotic subjects should convince those of our colleagues who deny the existence of posthypnotic suggestion that what happens is at any rate different from simple compliance. And it should caution those of us who would assume

213 Hypnosis, Motivation, and Ecological Validity

that hypnosis is a uniquely effective tool for the control of behavior.

However, from our present viewpoint another aspect of the data is perhaps more relevant. If we assume that hypnosis increases the motivation of the individual, certainly postcard sending in response to a waking request should be more frequent by good hypnotic subjects than by medium hypnotic subjects. However, if we examine our data on this, we find that the medium subjects sent considerably more postcards than the deep subjects a difference which would reach significance had it been predicted. Thus, here again we are compelled to dismiss the simple formulation that hypnotic phenomena are invariably associated with high motivation on the part of the good hypnotic subject.

Does the Subject Carry Out a Posthypnotic Suggestion in Response to Interpersonal Needs or to Intrapsychic Needs?

Two recent studies were designed to help clarify the mechanism of posthypnotic behavior and tend to show that the response cannot be explained in terms of pleasing the hypnotist. In one study by Nace and Orne (1970) it was reasoned that an overt response to a posthypnotic suggestion requires that the need to carry out the behavior be sufficiently strong to result in action; however, there may be forces opposing this action. If one considers those subjects who fail to respond to a posthypnotic suggestion, in at least some a strong residual action tendency must exist which is nearly sufficient to lead to overt behavior. In an appropriate situation, the effects of such an action tendency should be demonstrable.

Subjects were seen in three experimental sessions in the laboratory. All the experimental rooms contained pencil boxes with a standard variety of different-colored pencils. On each occasion subjects were given one of the Stanford Hypnotic Susceptibility Scales (Weitzenhoffer & Hilgard, 1959, 1962) and were required to carry out other tasks, including filling out questionnaires and receipt forms for payment. The latter activities provided opportunities for unobtrusive observation of choice of writing instrument, four times in the first session, three in the second, and two in the third. In the third session, Form C of the Stanford Hypnotic Susceptibility Scale (Weitzenhoffer & Hilgard, 1962) was used, but to it the following

214 Nebraska Symposium on Motivation, 1970

posthypnotic suggestion item was added: "When you wake up and I take off my glasses you will feel a compelling desire to play with the blue pencil." After the trance the posthypnotic suggestion was tested, and the subject's experiences during the session were discussed with him without any particular reference to the posthypnotic suggestion. The experimenter then indicated to the subject that before finishing, it was necessary for him to fill out another questionnaire, asked him to make himself comfortable behind the desk, and left the room. Later the secretary observed which writing instrument was chosen to sign the receipt. Thus, in the third session both opportunities for observation of choice of writing instrument came after the cue for the carrying out of the posthypnotic suggestion.

The question was whether subjects who had received the posthypnotic suggestion but had not carried it out would now tend to choose the blue pencil for filling out the questionnaire. The task required the subjects to use some writing implement, and therefore provided them an opportunity to discharge, in the absence of the hypnotist, an action tendency that had previously been insufficiently strong to be expressed in behavior. The results, set out in Tables 3 and 4, were strikingly lawful. The very best hypnotic subjects, of course, carried out the posthypnotic suggestion. The essentially unsusceptible individuals did not carry out the posthypnotic suggestion, nor did they tend especially to use the blue pencil in filling out the questionnaire. However, those individuals who responded to a good many of the hypnotic items but did not enter a very deep trance were the ones who subsequently used the blue pencil for the

215 Hypnosis, Motivation, and Ecological Validity

questionnaire. These data strongly support the view that a posthypnotic suggestion which is not carried out will, in some subjects, nonetheless lead to an action tendency which, under appropriate circumstances, will manifest itself in overt behavior. The use of the blue pencil in filling out the questionnaire can be seen as a relatively easy posthypnotic test item. Of relevance to our discussion here is that the use of the blue pencil occurred not in the presence but in the absence of the experimenter, apparently serving an intrapsychic need, perhaps being analogous to a Zeigarnik effect (1927). It would seem that the posthypnotic suggestion certainly increased the likelihood of the use of the blue pencil, but that this effect was not the same thing as an increased tendency to please the hypnotist.

Using a very different approach, Orne, Sheehan, and Evans (1968) investigated the effect of a posthypnotic suggestion on behavior occurring outside the experimental context. Fisher (1954), in a now classic study, hypnotized subjects and gave them the posthypnotic suggestion that whenever they heard the word "psychology" they would scratch the right ear. On their awakening, he tested the suggestion and found they responded. Shortly thereafter one of his colleagues entered the room, and a "bull session" ensued, implicitly terminating the experiment. During this session the word "psychology" came up, and 9 of the 12 subjects failed to respond. When the colleague eventually left, Fisher turned to the subjects in such a way as to imply that the experiment was still in progress, and used the word "psychology" conspicuously, thereby eliciting the suggested ear scratching in all subjects. These findings have supported the view that posthypnotic behavior can best be

216 Nebraska Symposium on Motivation, 1970

understood as the result of the subject's providing what the hypnotist desires of him. There are two issues subtly intertwined in this experiment, however: (1) whether the subject will carry out a posthypnotic suggestion outside the experimental context if he is not specifically required to do so, and (2) whether the subject's interpretation of an ambiguous suggestion can subsequently be modified by the hypnotist.

Consider the suggestion: "Whenever you hear the word ‘psychology’ you will scratch your right ear." It would seem extremely unlikely that the suggestion was intended or taken to mean that henceforth and for the rest of his life the subject was to respond by scratching his right ear whenever he heard the word. Rather, most subjects would correctly interpret it to mean "As long as this experiment continues, you will scratch your right ear whenever you hear the word ‘psychology.’" If subjects interpreted it in this fashion, their behavior was merely an appropriate reflection of how they perceived an ambiguous instruction, and would have no bearing upon the more basic issue of whether an unambiguous posthypnotic suggestion would be carried out outside an experimental setting.

It is to this latter issue that the Orne, Sheehan, and Evans (1968) study was addressed. Subjects, whose hypnotizability had carefully been established in earlier sessions, were solicited for an experiment which involved two sessions on succeeding days. This requirement was discussed with them prior to their coming to the laboratory. On the first day they were given a series of personality tests and deep hypnosis was induced. At that time they were given the posthypnotic suggestion that for the next 48 hours whenever they heard the word "experiment" they would run their hand over their head. It should be noted that this suggestion has a specific time limit which is legitimized by the fact that the individual will be returning to the laboratory on the succeeding day. The session was completed, the subject sent home. The next day when he came to the laboratory and the investigator came to pick him up in the waiting room, as they walked down the corridor the latter remarked casually, "I appreciate your making today's ‘experiment.’" This, however, was not the crucial test. Subjects are reimbursed for their experimental participation in the main office, which is some distance

217 Hypnosis, Motivation, and Ecological Validity

from the experimental suite, but through which they must pass in order to come and go. This is a routine procedure to which subjects had earlier become accustomed. As the secretary paid the subject on the first day, in a perfunctory fashion she asked him to sign the payment receipt for the "experiment." The next day when the subject arrived in the waiting room, he was met by the receptionist, who asked whether he was there for the psychological or the physiological "experiment." These, then, were the crucial tests outside the experimental context.

A crucial issue bearing on the interpretation of the findings was whether the secretary and receptionist were indeed perceived to be outside the experimental setting or whether the subject recognized them to be really agents of the experimenter. For these reasons, a special group of quasi controls simulating hypnosis was also run. The hypnotist was blind to the actual status of the subjects. By virtue of their instructions, simulating subjects (a comparison group which we will discuss at greater length later) are particularly prone to be suspicious of the experimental situation. As a result, they provided a very hard test of whether the hypnotic subjects perceived what happened in the main office and the waiting room as part of the experiment or not.

The results were clear. Deeply hypnotized subjects characteristically responded to the posthypnotic suggestion outside the experimental context, whereas simulators characteristically did not. The greater tendency to respond outside the experiment on the part of the deeply hypnotized subject was matched by an equally interesting difference within the experimental situation. Thus, simulators responded more consistently to the posthypnotic suggestion when the word "experiment" was used by the hypnotist than deeply hypnotized subjects, who occasionally seemed unaware of the use of the word when it was sufficiently well embedded in the context.

These findings indicate that the posthypnotic phenomenon cannot be explained by a strong motivation on the part of the hypnotized subject to please the hypnotist. Rather, the posthypnotic suggestion appears to set up a temporary compulsion for the subject to respond independently of whether the hypnotist is present or even aware of the response.

218 Nebraska Symposium on Motivation, 1970

INCONSISTENT NONEXPERIMENTAL OBSERVATIONS

Throughout this discussion we have examined data selected because they raise questions about the assumption that hypnosis necessarily increases the motivation of subjects to carry out the request of the hypnotist. The most relevant area where one would find evidence to support this view is the clinical situation. It is the impression of all therapists with whom I have discussed the matter that the induction of hypnosis during treatment seems to intensify the transference relationship and thereby more rapidly makes the therapist a significant figure for the patient. Gill and Brenman (1959) have discussed this issue at length in their classic text.

Equally relevant are the clinical data on the removal of symptoms. These data, in fact, provide the only clear-cut situations where individuals can be shown to undertake behavior in response to a request in hypnosis, which they appear unwilling or unable to undertake on the basis of a waking request. Unfortunately, truly controlled data are lacking, and a study is needed to investigate in a rigorous fashion the extent to which the induction of hypnosis facilitates symptom removal. One observation has often been made, namely, that the success of therapeutic suggestions does not depend upon the depth of hypnosis (as traditionally measured) achieved by the patient. Thus, sometimes individuals who would, by a laboratory definition, be considered unsusceptible show a dramatic therapeutic response. This point is interesting in the light of London and Fuhrer's (1961) findings that hypnotic procedures affect the performance of individuals who appear unsusceptible to hypnosis as measured by standard tests.

Another type of evidence, best exemplified by Pattie's (1935) paper on uniocular blindness, shows that some subjects will go to incredible lengths to succeed in following a suggestion. Thus, one of his subjects spent hours in the library learning about optics in order to be able to maintain the appearance of uniocular blindness. Similarly, subjects will, in response to a suggested burn, produce localized scratch marks and will, if given the opportunity, often actually burn themselves surreptitiously with a cigarette, thus raising a blister. It would be possible to go on in this way indefinitely, listing anecdotal data which suggest the intensity with

219 Hypnosis, Motivation, and Ecological Validity

which some subjects attempt to meet the demands placed upon them by a hypnotic suggestion. Indeed, it is this kind of evidence which causes us to assume so easily that an increase in motivation is a concomitant of hypnosis.

There are, however, differences between the clinical situation and the experimental one. In the former, suggestions are given that have intense personal reference, whereas in the latter the tasks are obviously peripheral to the interests of the subject. Further, most anecdotal evidence, such as that presented by Pattie, is based upon work with subjects with whom the experimenter has had prolonged and intense contact. These situations, though in a laboratory context, approximate clinical situations. Perhaps it will be necessary to distinguish between what happens as the result of largely impersonal induction of hypnosis in the laboratory and the phenomenon we see in clinical practice based on a highly meaningful, intensely personal relationship. In the laboratory little or no contact with the experimenter is necessary for the induction of hypnosis. Even a tape recording is fully adequate as an induction stimulus. Hypnosis takes place largely as a function of factors (such as attitude, aptitude, personality, etc.) which the subject brings to the experiment. This is, of course, quite distinct from the situation in clinical practice, where there usually exists a significant personal relationship, as well as the patient's expectations of help. It is quite possible that there is an interaction between the induction of the hypnotic phenomenon, its effect on the motivation of the individual, and these two types of settings.

To recapitulate briefly, the main points that have been made so far are:

1. Using human beings as subjects in the context of episodic relationships (in hypnotic studies and other psychological experiments) is a complex matter. Generalizing from episodic laboratory behavior to everyday life is hazardous.

2. This became particularly apparent in our early studies on hypnosis which indicated (a) that the very heterogeneous behavioral events associated with hypnosis may all be viewed as behaviors that meet "the expectations of the hypnotist," and (b) that the effects of demand characteristics can determine the

220 Nebraska Symposium on Motivation, 1970

kinds of data that are obtained in studying mental processes, where hypnosis is used as a means to induce those processes.

3. The whole question of the interaction of the motives of experimental subjects with their changing perceptions of the total experimental context needs to be scrutinized.

4. This kind of examination of the hypnotic situation reveals that the phenomena cannot, in fact, be fully explained in terms of strong or increased motivation to comply with the demands of the hypnotist, though, no doubt, some compliance occurs in the course of interpersonal relations such as therapy or in experimental situations where compliance is inevitably maximized.

We shall now turn to a closer look at the operation of demand characteristics in experimental situations other than hypnosis.

THE EXPERIMENTAL SITUATION AS A SPECIAL FORM OF INTERACTION

Our model for experiments comes from the physical sciences. There, in order to establish the effect of a particular independent variable, one compares objects exposed to a particular value of the variable with control objects which are not, all other conditions being held constant. When the object of study becomes an active, sentient human being, his awareness of the process of study may interact with the parameters under study. This is obvious in hypnotic research, but the problem is present in varying degrees in all behavioral research involving man.

THE MOTIVATION OF THE EXPERIMENTAL SUBJECT

Far from being passive responders, as assumed by the conceptual model of the traditional experiment, subjects are active participants in a very special form of social psychological interaction which we call the psychological experiment. A subject does not miraculously arrive in the experimenter's laboratory, but he is brought there by specific incentives. He is somehow solicited. Even if participation is made obligatory in the context of a course, a modicum of cooperation must in some way be involved, and more often than not we are

221 Hypnosis, Motivation, and Ecological Validity

dealing with individuals who have had the option of participating and have volunteered. While we tend to use financial rewards to secure subject participation in our research, the level of monetary reward usually involved is hardly sufficient to compensate the individual for his time, much less for the tedium, discomfort, or even pain that may be involved. With the exception of the occasional "professional subject," payment (within the range of monetary reward usually employed) appears secondary.

Elsewhere (Orne 1959a, 1962, 1970), I have emphasized that a great many subjects volunteer to participate in research in the hope of helping science, contributing to human welfare, and so forth. Certainly there is also a wide range of other motives that bring the subject to the laboratory. These include not only curiosity and an interest in learning about experiments but also, not infrequently, hopes of obtaining some form of personal help by participating. Whatever brings the subject, it is my conviction that he soon acquires a stake in the research. Seeing that he is already there and already participating, he is likely to want his participation to be useful. While no doubt subjects are concerned about what psychologists think of them and how they evaluate them -- a motive that has been singled out, emphasized recently by Milton Rosenberg (1969), and termed "evaluation apprehension" -- I have been equally impressed by the extent of the subject's concern to have his data be useful to the investigator. For example, in a psychophysiological study, equipment failure can in no way be construed as reflecting upon the subject's capabilities, yet it is invariably resented. In one way or another subjects tend to experience great concern about the success of the study as a whole and hope that their performances will contribute to it.

In the context of research subjects are willing to tolerate discomfort and even pain, so long as it seems relevant and necessary to the research task at hand. However, this tolerance is limited to discomfort that is perceived as essential for the research. For example, subjects will accept with relatively little annoyance repeated venipunctures in order to permit the investigator to draw several blood samples at different times in the course of an investigation. However, they greatly resent being stuck twice in an experiment requiring a single venipuncture because the investigator is unable

222 Nebraska Symposium on Motivation, 1970

to find the vein the first time. Whereas in the former case the discomfort associated with blood samples being taken was necessary for the purpose of the experiment, in the latter instance the additional discomfort of the second puncture is related to the experimenter's ineptitude rather than the needs of the research.

From the subject's point of view, the only way in which his participation can have real meaning is for the experiment to be important and successful. The greater the investment of time, effort, and discomfort on the part of the subject, the more he is compelled to care about its outcome and assume that it is important. Anyone who has carried out research where human subjects are exposed to considerable stress in the course of an experiment but are otherwise treated with consideration (and where the situation is so arranged that no subject perceives himself to have failed in the situation) can attest to the importance his subjects ascribe to the research. In fact, the more stress and discomfort there is associated with the study, the more likely it will be for the subject to think of the research as vitally important. It should not surprise us that the more investment the individual makes, the more he will be committed to assume his efforts are not in vain.

It is consistent with this that we have found it quite impossible to design an experiment which the subjects perceive as meaningless. For example, some years ago we tried to create a meaningless experiment. The subject was placed in a room with a pile of several thousand pages of serial additions in front of him. Each page took about 10 minutes to complete. The experimenter told him to begin working and after completing the first sheet to go on to the next and so forth, and to work as accurately and as rapidly as he could. He then asked for the subject's watch and left him with the words, "Just keep working; I will be back eventually." We found it necessary to stop subjects, since after several hours they came out merely to go to the bathroom and wanted to return to the room to continue their work.

When it became clear that this task was not seen as meaningless, it was modified and subjects were given the same pile of papers for serial addition, together with a pile of several hundred instruction cards. They were told to complete a sheet of paper and then to pick up the top card, which would tell them how to proceed. They were

223 Hypnosis, Motivation, and Ecological Validity

again asked for their watches by the investigator, who told them to just keep following the instructions, and promised to return eventually. After the subject completed the first sheet he picked up the top card, which read: "Tear up the sheet you have just completed into a minimum of 32 pieces and throw it into the wastebasket. Then go on to the next sheet and work as accurately and rapidly as you can. Once you have completed the next sheet, pick up the next card, which will give you instructions as to what to do next." All the cards were identical.

Again subjects persisted in working on this task for hours. We reasoned that perhaps they thought that sooner or later they would run across a card which would be different and tell them that they had somehow won, so we added yet another variation. This time the subjects were again placed in the room and given a stack of papers to work on. However, instead of a stack of instruction cards, there were now merely three instruction cards. They were told, as before, to pick up the top card for their instructions, which would be self-explanatory. Their watches were taken from them and they were told that the investigator would eventually return. This time the card said: "Tear up the page you have just completed into a minimum of 32 pieces and throw it in the wastebasket. Go on to the next page and continue to work as accurately and rapidly as you can. Place this card on the bottom of the pile. When you have completed the page, pick up the card at the top of the pile and follow its instructions." Under these circumstances the subjects knew by the time they had completed the third sheet that there were no surprises. All cards would be the same. Yet they continued to work at a high rate with remarkably few errors.

We then reasoned that because there was a one-way screen in the room, subjects suspected they were being watched. We eliminated the one-way screen. By mounting a microphone in the table, we were able to monitor the scratching of the pencil on the sheets of paper and thereby obtain an accurate measure of rate. We also marked the back of the pages with invisible ink and were therefore also able, by random sampling even from small pieces, to obtain an adequate measure of accuracy. The results were not appreciably different from those of the earlier studies. The interesting part of the research, however, was the fact that not a single subject doubted

224 Nebraska Symposium on Motivation, 1970

that we were somehow able, in ways he was unable to discern, to obtain an accurate record of his performance. No subject assumed that the experiment was meaningless. Indeed, they found it interesting and were convinced that there was some very important reason why the study was being carried out. Some subjects explained it as a frustration-tolerance task, some thought it was an ingenious way to determine the amount of inner-directedness, and others advanced various different hypotheses. The fact was that all subjects appropriately imputed meaning to the experiment, though the specific hypotheses were idiosyncratic. Parenthetically, this and similar work suggest that a James-Lange type of notion may apply to the recognition of meaning in a life situation. The meaning may not, contrary to widely held current beliefs, reside out in the environment but rather may be in large measure a function of involvement in a task. Once effort is invested, the individual will find meaning.

The extent to which a subject invests in the outcome of the experiment will, of course, be a function of many factors. Thus if he has to go to some trouble to volunteer and the experiment is tedious or stressful, the investment will tend to be higher. If the experiment involves little effort and participation makes no demands, he may care less about its outcome. He may even be inclined to see if he can throw sand into the wheels of progress. The subject's perceptions of the real situation will also be a factor. To the extent that the experimenter seems competent, it will be easier for the subject to cling to the notion that the experiment has a significant purpose. To the extent that the experimenter seems confused, unprofessional, or lackadaisical, the subject may well come to believe that the experimenter himself is merely satisfying some course requirement. To the extent, however, that the subject is involved in the study and is committed to its outcome, he will want it to work.

HOW DEMAND CHARACTERISTICS FUNCTION

The remarkable tendency of some subjects to help an investigator by validating an implicitly communicated hypothesis forcibly drew my attention to one aspect of the potential importance of demand characteristics. Certainly if subjects could show such extreme

225 Hypnosis, Motivation, and Ecological Validity

compliance, an investigator must concern himself with how the individual perceives what the investigator seeks to prove. However, I soon realized that in most situations conscious compliance of subjects was a rare phenomenon. The kind of behavior I observed when trying to replicate the Carmichael, Hogan, and Walter (1932) experiment involved the subject's deliberate falsification of his actual memory to produce memories that he thought I hoped to see. However, except in rare instances, subjects conceive their role in the experiment as having to be honest in reporting their actual experiences, and therefore, willful distortions of this kind are rare. As I have pointed out elsewhere, subtle cues which are not fully recognized by the subjects are more likely to alter their behavior than obvious ones. The latter tend to obtrude themselves upon the individual's awareness and demand a response which would be tantamount to falsifying data wittingly in order to please the experimenter. This tends to be recognized as a kind of cheating, which would negate the value of the experiment. (A similar effect was observed by Rosenthal [1963] when he tried to pay subjects more for producing good, that is, biased, data; under these circumstances he noted less bias.)

It was never my impression, except in rare cases, that the mechanisms by which demand characteristics affect subjects' behavior were those of willful or conscious compliance, though it would appear that I have not been sufficiently explicit about this point. I see the subject's behavior and experience modified in subtle and complex ways by the operational implementation of the experimental task. Much as a green light on the road communicates directly the appropriate behavior to a driver, and just as a chair in a waiting room calls forth "be seated" behavior, so the experimenter's intentions and hypotheses are communicated through his operations. From the subject's point of view, these operations serve to define the appropriate behavior, the "right" behavior, the way one acts. The subject is not being compliant in any useful sense of that word. Rather, he is behaving in ways that, unthinkingly, he perceives as correct and appropriate.

Consider the experiment on unilateral catalepsy (Orne, 1959b). The experimental manipulation caused subjects to believe that unilateral catalepsy was characteristic of the behavior of hypnotized

226 Nebraska Symposium on Motivation, 1970

individuals. Later, when they were hypnotized, they manifested this item of behavior because, as far as they were concerned, it was part of being hypnotized. They validated the hypothesis of the experimenter -- not explicitly to please him but simply because that was hypnosis from their point of view!

With increasing awareness of the part that subtle cues may play in experimental situations, I have tended to expand the concept to include all aspects of the individual's motivated perception of the experimental situation. It has become increasingly clear that the experimental situation for the subject depends upon what he perceives to be the relevant variables. These may or may not correspond to what the experimenter intended. Thus subjects who vary in sophistication, when run in identical experimental procedures, may be participating in what, from their points of view, are entirely different experiments! Unless it is possible for the investigator to ascertain in detail how the subject perceives the situation, it is impossible for him to know whether a given set of stimuli or a particular manipulation has produced the intended effect. An approach of this kind recognizes the need to study, in an experiment, the effects of specifiable conditions on human behavior; but it also recognizes that the conditions must be specified not only as they are seen by the investigator but also as they are seen by the subject. How the conditions are seen by the subject will, however, depend upon a myriad of subtle cues, many of which are rarely if ever reported when the experiment is published.

The concept of demand characteristics derives from the work of Lewin (1929), the term being a literal translation of Aufforderungs-charakter, usually referred to as a valence in the English literature. It shares the ambiguity of the Lewinian concept concerning its locus. It is an attribute of the experimental situation while at the same time it is the consequence of the subject's perception of the organization of the experimental situation. Its elucidation in any instance requires not only a knowledge of the cues objectively present in the experimental situation but also an understanding of the subject's perception of the situation. 7

227 Hypnosis, Motivation, and Ecological Validity

THE CUES WHICH HELP DETERMINE THE DEMAND CHARACTERISTICS OF THE EXPERIMENTAL SITUATION

The sum total of cues which communicate the purposes and intent of the experiment, particularly regarding those aspects about which the subject is not informed (or believes himself not to have been informed) constitutes the demand characteristics. What the subject actually perceives is a function of his attention, motivation, and his interpretation of the cues based on his prior knowledge, experience, suspiciousness, and such other factors as his intelligence.

The subject's perception of the experiment will, of course, depend in large part upon the instructions. These will be believed, however, only to the extent that they are congruent with the scuttlebutt concerning the research, the behavior of the experimenter, and the experimental procedure itself. The congruence among these different sets of cues will in large part determine the plausibility of the instructions. This is, of course, particularly true of deception research. Unfortunately, subjects have learned to distrust psychologists, and for this reason the experimental procedure often serves as a more compelling set of cues than the instructions. Here, as in other aspects of our experience, actions speak louder than words (see, for example, Silverman, Shulman, & Wiesenthal, 1970; Stricker, Messick, & Jackson, 1967).

Very significant information, which is generally ignored and rarely reported by investigators, comes to the subject from the manner in which he is solicited, the place in which the experiment is conducted, the person of the experimenter, and his mien and dress. One very potent source of cues that is rarely controlled but which may dramatically affect the subject's perception is the secretary or

by Rosenthal (1966) may frequently be confounding variables. These effects depend in large part on the experimenter outcome expectations. They can become significant determinants of data by causing subtle but systematic differences in (1) the treatment of subjects, (2) the selection of cases, (3) the observation of data, (4) the recording of data, and (5) the systematic analysis of data. To the extent that bias effects cause subtle changes in the way the experimenter treats different groups they may alter the demand characteristics for those groups. In some studies, therefore, demand characteristics may be one of the important ways in which experimenter bias can be mediated. Conceptually, however, the two processes are very different. Experimenter effects are rooted in the motives of the experimenter, while demand characteristic effects depend upon the perception of the subject.

228 Nebraska Symposium on Motivation, 1970

research assistant with whom the subject speaks in order to arrange an appointment. It is essential to talk to the subject in order to arrange an appointment, and some form of reaction to questions, which are almost inevitable, must be made. Monitoring telephone conversations of this kind is an eye-opening experience. The general instruction "Don't tell the subject anything about the experiment" is well-nigh impossible to follow. Even if the research assistant strives not to answer the questions, the degree of discomfort with which she refuses to answer certain questions is in and of itself a communication. More likely than not, however, a good deal of information is transmitted to the subject as a quid pro quo in the process of having him accommodate to the schedule of appointments. If an investigator really tries to control this source of random variation, then he must spell out in detail the kind of answers that are appropriate for any conceivable question that might be asked. Fortunately, these tend to be reasonably finite and, provided the experimenter makes himself available to whoever does the scheduling in the early part of the study, he can get to know the kind of information that is communicated in this fashion.

A similar caution, of course, should apply to anyone other than the investigator who meets the subject, though, of course, he himself will need to think through that information may legitimately be communicated. Even experienced investigators will often be unaware of the highly variable amount of information that they make available to their subjects in informal contact before the beginning of the study.

One final source of information that is rarely controlled but is extremely important in many studies is the technician. Especially in psychophysiological studies where electrodes must be attached to the subject, the technician or the investigator (if he combines the roles) must interact with the subject over a considerable period of time. Silence is extremely oppressive and almost impossible to maintain. Questions about the experiment are all but inevitable, and considerable care is required if one hopes to have a fairly standard amount of information communicated.

The importance of the secretary and the technician as sources of cues is seen to be particularly striking once it is recognized that the subject appropriately can trust information from these sources far

229 Hypnosis, Motivation, and Ecological Validity

more than that coming from the investigator. He recognizes that these individuals are less likely to try to misrepresent the purpose of an experiment, and he is inclined to believe that they will have no reason to withhold or alter information whereas the investigator may well feel compelled to do so as part of the experiment.

THE RULES OF THE GAME

Most volunteer subjects participating in a psychological experiment recognize it as an elaborate ritual with rules that are reasonably well understood by them as well as by most experimenters. For example, subjects recognize that once they agree to participate in a psychological experiment they cannot be told the precise purpose of the study, since this may influence their behavior. They realize full well that it may be necessary for the investigator to deceive them as part of the experiment. It is this widespread recognition that is often particularly troublesome and that results in the type of comment made to me by a subject that "psychologists always lie." This awareness has implications for deception studies which make them troublesome, not only because of the moral problems but because of methodological issues to which we will return later.

It is this recognition of the rules of the game of the experiment which prevents the subject from being overly compliant in the sense of willfully falsifying data. The rules subsume the episodic nature of the experiment. One consequence of this is the willingness of subjects to take abuse in the course of an experiment. Some studies expose subjects to situations that make even the listener feel uncomfortable. One would expect subjects in such situations to become angry and hostile. Peculiarly enough, I know of no instance where a subject became sufficiently outraged to assault an investigator -- though at times one feels such behavior might have been justified. Complaints about treatment are very unusual and even direct verbal chastisement of an investigator is rare. Furthermore, subjects assume that during the course of psychological experiments adequate precautions for their safety and welfare are taken. They recognize the constraints of reality upon the investigator and the consequences that would result if any irreversible serious harm were caused in the course of the investigation.

230 Nebraska Symposium on Motivation, 1970

Agreement to participate in an experiment involves a far more complex social contract than is generally appreciated. The nature of this contract will at times become a recognizable determinant of the subject's behavior.

THE POSTEXPERIMENTAL INTERVIEW

Even a postexperimental inquiry, designed inter alia to check on demand characteristics, will not be free of those same demand characteristics. As I have indicated, subjects tend to be concerned that their data be useful to the investigator; thus they may feel that if they concede awareness of aspects of the study about which they were not informed (such as a deception), they would compromise the utility of their findings. For such reasons, even when the investigator asks what the study was about he may well be told "I don't know" or given some similar noncommittal answer.

Unfortunately, the needs of the investigator in this tend to parallel those of the subject: he is no more anxious to learn that the subject has seen through his deception than the subject is to tell him. Certainly, having to discard the data from a particular subject -- particularly if he has yielded the right kind of data -- is less than desirable. He, even more than the subject, has a stake in the outcome of the research. The peculiar interdigitation of the needs of the experimenter and the subject tends to result in what may best be described as a "pact of ignorance" (Orne, 1962) -- an unspoken agreement between investigator and subject that they shall not look too deeply into the subject's awareness of the situation.

The above considerations suggest that it is essential that a postexperimental interview be conducted in a manner analogous to a clinical interview. The experimenter must elicit the subject's perception of the experiment. Often this is sought by asking the following questions: What is the experiment about? What do you feel the experiment is trying to demonstrate? What do you think is my hypothesis in the experiment? How do you think you performed in the experiment? How do you think others performed in the experiment? Have you any comment which may help to clarify

231 Hypnosis, Motivation, and Ecological Validity

things not apparent in the experimental situation? To ask questions of this kind by way of a questionnaire at the end of an experiment very rarely succeeds in eliciting a clear picture of what the subject thought about the experiment. He is most likely to respond in a way that will protect the integrity of the experimental performance.

On the other hand, in an interview it is possible for the investigator to convey that the experiment is truly over and that he now wants to understand how the subject perceived the situation. In other words, he changes the role relationship between subject and experimenter. In many ways during a postexperimental interview the setting can be one where the subject becomes a coinvestigator, collaborating with the experimenter to help him understand what was really happening. In the context of such an interview it is possible to push the subject to elaborate his perceptions of the situation without providing him any substantive information. It will not be possible, of course, to determine the extent to which the subject's insights into the experimental procedure actually played a role in his performance during the experiment. His understanding may well have emerged only during his reflection on the situation with the experimenter.

As useful as postexperimental inquiry information may be, it is subject to major limitations. Not only is it difficult to prevent subtle biases from intruding, but also it is impossible to be certain how the insights that are elicited have affected prior behavior. One thing is certain -- experimenters being human, it is far easier to conduct exhaustive inquiries designed to elicit a subject's true perception of the situation during pilot research, before the investigator has a major investment in the particular experimental design, than it is during the actual study itself.

The inquiry procedure is a technique which obviously can be carried out only with human subjects. In one sense it may legitimately be considered a control procedure in that it helps to clarify the meaning of the data observed during the experiment. In another sense, however, it is a very special kind of control, quite dissimilar from the usual sort. Instead of employing a group of subjects who are not exposed to the experimental variable under study to help clarify its effect, it utilizes the active cognitive processes of the subject, which normally confound the data, as a means to clarify them.

232 Nebraska Symposium on Motivation, 1970

THE CONCEPT OF QUASI CONTROL

The purpose of an interview after an experiment is to help the investigator understand how the subject perceived the specific stimuli so that he is better able to interpret the behavioral data. With the major emphasis on behavior that has characterized psychology, at least until very recently, it has been fashionable to deny the relevance of such information. This should not be taken to mean, however, that the behavior has not been interpreted. Quite the contrary! Conventionally the investigator has operationally defined concepts such as stress, incentive, frustration, and so forth in terms of some manipulable experimental condition and then gone on to study the consequent behavior in order to make inferences about the basic psychological processes under investigation. In other words, instead of determining from the subject how he perceived the situation in order to interpret his behavior, the investigator has decided for the subject what the situation was supposed to mean, and, assuming that he was correct in his interpretation, has drawn inferences from the behavior. In many instances this procedure works because, intuitively putting himself in the subject's place, the investigator correctly surmises how the subject perceives the situation. We might say that this experimental model has functioned because the investigator has unwittingly taken advantage of information about what may have been going on in the subject's experience; he may have even talked to subjects to find out, even though he probably writes about the experiment as if it had been carried out quite independently of such information.

A particularly dramatic example of this kind was seen when Richard Solomon and his associates (Turner & Solomon, 1962) began to study traumatic avoidance conditioning with human subjects. Subjects were given a warning signal 10 seconds before a very painful 15-second shock was delivered to the ankle. A variety of manipulanda were employed. If the subjects moved the manipulandum appropriately while being shocked they could terminate the shock. If the appropriate response was given after the warning signal and before the shock was delivered, they could avoid it entirely. These procedures are closely analogous to the shuttlebox avoidance conditioning paradigm used with dogs.

233 Hypnosis, Motivation, and Ecological Validity

Subjects were run by Lucy Turner with minimal instructions in the hope of approximating the situation of the animals. The investigators observed that many subjects showed little or no learning over several sessions, accepting instead a large number of very painful electric shocks. When these subjects were asked about the experiment, it became clear that they thought the study was a test of their capacity to tolerate painful electric shock. Given such a set, the idea that the shocks could be or should be avoided had simply not occurred to them. On the other hand, subjects who were told that they could do something to avoid the shocks had relatively little difficulty in discerning and learning the appropriate response.

In these as well as many other studies, not giving the subject explicit verbal information about the task in no way prevented them from forming definite hypotheses about the study and behaving accordingly. It never does. In this particular case, a preexperimental procedure used to set shock level -- which "required" individuals to tolerate as high a shock as they could -- may have inadvertently served to create the specific set which prevented the acquisition of avoidance behavior. Thus a change of experimental procedure might also have been designed effectively to communicate that avoidance was expected. The basic point remains, however, that the behavior cannot be interpreted without understanding how the experiment is perceived.

One can easily imagine how one might have varied a large number of parameters other than instructions and shown their effects upon the subject's ability to learn to avoid shock. One might eventually have inferred a number of mechanisms or even motives affecting the avoidance learning. This is precisely what we are forced to do when we are working with nonhuman species. Fortunately, with human subjects a relatively simple procedure -- the inquiry -- designed to elicit their perception of the experiment allows the investigator to understand what is probably occurring. Frequently one finds that the feature of an experiment which is intended by the investigator to represent one set of circumstances outside the laboratory is seen by the subjects in the laboratory as representing an entirely different set of circumstances. Under those conditions the inference drawn from the experimental data is bound to be incorrect.

234 Nebraska Symposium on Motivation, 1970

The inquiry procedure, then, is the simplest technique to find out about the active cognitive processes that the subject introduces into the experimental situation. Its purpose is to help the investigator understand (not measure) how those processes may have affected the results. The inquiry is only one of a family of such techniques that constitute a special kind of control procedure. They are riot controls in the classic sense. Rather they focus upon the elucidation of the possible interaction of the subject's cognitive processes and his experimental behavior.

For some time I have tried to find an appropriate term for these techniques. They all share the property of telling us little about the actual variables under investigation but a great deal about the procedures used to study them. For this reason the term "procedural control" has been suggested. It would be appropriate since they characteristically lead to modifications in the experimental procedure. The term "active control" is appealing since it emphasizes the extent to which the subject is required to participate actively rather than respond passively. Another significant aspect is that a proper inquiry procedure alters the role relationship between subject and experimenter. Instead of being an object of study the subject becomes a collaborator, a colleague even, helping to solve the difficult problem of understanding why he has behaved in some particular fashion. The term "collaborative control" suggests itself. Again, the techniques all are designed to clarify the perceptions of the subject and the context of the experimental situation as he perceives it. This idea could be expressed in the term "contextual control." These procedures address themselves to what may be considered the validity of the experimental procedure. They ask whether the operational definition by which a construct is translated into an experimental condition is appropriate. They ask whether a deception in an experiment has been successful. They shed light on the type of extraexperimental situation to which the findings of a given experimental procedure may be appropriately extended. The term "inferential control" would emphasize a basic feature of such techniques, their helping to establish the extent to which the inferential leap from one context to another is justified.

Since each of the suggested terms grasps a particular aspect of the procedure and none of them is fully descriptive, I have opted

235 Hypnosis, Motivation, and Ecological Validity

for the term "quasi control." This is a term which emphasizes that the concept is not of a true control, and it therefore is less likely to bring about confusion with other meanings. As I shall try to illustrate, quasi-control procedures allow inference about the adequacy of an experimental procedure, not about either the independent or the dependent variables under investigation. Thus they are crucial in helping the investigator design a better study which will permit more definitive and valid answers to his questions.

INQUIRY DATA AS THE BASIS FOR MANIPULATING DEMAND CHARACTERISTICS

A psychophysiological study by Gustafson and Orne (1965) illustrates the use of inquiry data to elucidate the demand characteristics which might explain confusing experimental findings. The example is useful because it illustrates how, once this situation was understood, it became possible to deliberately manipulate demand characteristics and treat them as experimental variables in their own right. The results of this explicit manipulation enabled us to understand an experimental result which otherwise appeared contrary to findings in nonexperimental situations.

A number of studies in recent years have investigated the detection of deception (more popularly known as lie detection) using the galvanic skin response (GSR) as the dependent variable. In one such study, Ellson, Davis, Saltzman, and Burke (1952) reported a very curious finding. Their experiment dealt with the effect which knowledge of results can have on the GSR. After the first trial, some subjects were told that their lies had been detected, while others were told the opposite. This produced striking results on the second trial: those who believed that they had been found out became harder to detect the second time, while those who thought they had deceived the polygraph on Trial 1 became easier to detect on Trial 2. This finding, if generalizable to the field, would have considerable practical implications. Traditionally, interrogators using field lie detectors go to great lengths to show the suspect that the device works by "catching" the suspect, as it were. If the results of Ellson et al. were generalizable to the field situation, the very procedure which the interrogators use would actually defeat

236 Nebraska Symposium on Motivation, 1970

the purpose for which it was intended by making subsequent lies of the suspect even harder to detect.

Because the finding of Ellson et al. runs counter to traditional practical experience, it seemed plausible to assume that additional variables might be involved in the experimental situation. The study by Ellson et al. was therefore replicated by Gustafson and Orne (unpublished study, 1962) with equivocal results. Postexperimental interviews with subjects revealed that many college students apparently believe that the lie detector works with normal individuals and that only habitual liars could deceive a polygraph. Given these beliefs, it was important to the student volunteers that they be detected. In that respect the situation of the experimental subjects differs markedly from that of the suspect being interrogated in a real-life situation. Fortunately, with the information about what most experimental subjects believe, it is possible to manipulate those beliefs and thereby change the demand characteristics of the Ellson et al. study.

Two groups of subjects were given different information about the effectiveness of the lie detector. One group was given information congruent with the widely held belief and told: "This is a detection of deception experiment. We are trying to see how well the lie detector works. As you know, it is not possible to detect lying in the case of psychopathic personalities or habitual liars. We want you to try your very best to fool the lie detector during this experiment. Good luck." These instructions tried to maximize the kind of demand characteristics which might have been functioning in the Ellson et al. study, and it was assumed that the subjects would want to be detected in order to prove that they were not habitual liars. The other group was given information which prior work (Gustafson & Orne, 1963) had shown to be plausible and motivating; they were told, "This is a lie detection study, and while it is extremely difficult to fool the lie detector, highly intelligent, emotionally stable, and mature individuals are able to do so." The demand characteristics in this case were designed to maximize the wish to deceive.

From that point on, the two groups were treated identically. They drew a card from an apparently randomized deck; the card had a number on it which they were to keep secret. All the possible

237 Hypnosis, Motivation and Ecological Validity

NOTE: Ss were not given information about the success of detection until after the trial on which these data are based.

A multiple chi-square contingency analysis (Sutcliffe, 1957) was used to analyze the departures from expected frequencies in the entire table. Neither the chi-square components for each variable alone, nor the interaction between variables, were significant.

From L. A. Gustafson and M. T. Orne, "Effects of perceived role and role success on the detection of deception," Journal of Applied Psychology, 1965, 49, 412-417. Copyright 1965 by the American Psychological Association and reproduced by permission.

numbers were then presented by a prerecorded tape while a polygraph recorded the subjects' GSR responses. On the first such trial, the "detection ratios" -- that is, the relative magnitudes of the critical GSR responses -- in the two groups were not significantly different. When the first trial was over, the experimenter gave half the subjects in each group the impression that they had been detected, by telling them what their number had been. (The experimenter had independent access to this information.) The other half were given the impression that they had fooled the polygraph, the experimenter reporting an incorrect number to them. A table of random numbers was used to determine, independent of his actual GSR, which kind of feedback each subject received.

A second detection trial with a new number was then given. The dramatic effects of the feedback in interaction with the original instructions are visible in Table 6. Two kinds of subjects now gave large GSRs to the critical number: those who had wanted to be detected but yet had not been detected, and also those who had hoped to deceive and yet had not deceived. (This latter group is analogous to the field situation.) On the other hand, subjects whose hopes had been confirmed now responded less and thus became

238 Nebraska Symposium on Motivation, 1970

NOTE: A multiple chi-square contingency analysis here shows that neither information given nor motivation (n Detect vs. n Deceive) has significant effects by itself. The relevant chi-square values, calculated from partitioned subtables, are .25 (p > .95) and .00 respectively (df = 1). However, successful detection does depend significantly on the interaction between information and motivation x2 = 30.94; p < .001; df = 1).

From L. A. Gustafson and M. T. Orne, "Effects of perceived role and role success on the detection of deception," Journal of Applied Psychology, 1965, 49, 412-417. Copyright 1965 by the American Psychological Association and reproduced by permission.

harder to detect, regardless of what their hopes had been. Those who had wanted to be detected, and indeed had been detected, behaved physiologically like those who had wanted to deceive and indeed had deceived.

This effect is an extremely powerful but also an exceedingly subtle one. The differential pretreatment of groups is not apparent on the first trial. Only on the second trial do the manipulated demand characteristics produce clear-cut differential results, in interaction with the independent variable of feedback. Furthermore, we are dealing with a dependent measure which is often erroneously assumed to be outside of volitional control, namely, a physiological response -- in this instance, the GSR. This study serves as a link toward resolving the discrepancy between the laboratory findings of Ellson et al. (1952) and the experience of interrogators using the lie detector in real life.

It proved possible in this experiment to use simple variations in instructions as a means of varying the demand characteristics. The success of the manipulation may be ascribed to the fact that the instructions themselves reflected views that emerged from interview data, and both sets of instructions were congruent with the experi-

239 Hypnosis, Motivation, and Ecological Validity

mental procedure. Only if instructions are plausible -- a function of their congruence with the subjects' past knowledge as well as with the experimental procedure -- will they be a reliable way of altering the demand characteristics. In this instance the instructions were not designed to manipulate the subjects' attitude directly; rather they were designed to provide differential background information relevant to the experiment. This background information was designed to provide very different contexts for the subjects' performance within the experiment. We believe this approach was effective because it altered the subjects' perception of the experimental situation, which is the basis of the demand characteristics in any experiment. It is relevant that the differential instructions in no way told subjects to behave differently. Obviously, subjects in an experiment will tend to do what they are told to do -- that is the implicit contract of the situation -- and to demonstrate this would prove little. Our effort here was to create the kind of context which might differentiate the laboratory from the field situation and which might explain differential results in these two settings. Plausible verbal instructions were one way of accomplishing this end. (Also see Cataldo, Silverman, & Brown, 1967; Kroger, 1967; Page & Lumia, 1968; Silverman, 1968.)

Unless verbal instructions are very carefully designed and pretested they may well fail to achieve such an end. It can be extremely difficult to predict how, if at all, demand characteristics are altered by instructions, and frequently more subtle aspects of the experimental setting and the experimental procedure may become more potent determinants of how the study is perceived.

THE NONEXPERIMENT, OR PREEXPERIMENTAL INQUIRY

One of the major logical problems with postexperimental inquiry data is the difficulty of determining whether the subject's behavior was a function of how he perceived the experiment or whether his perceptions of the experiment serve post hoc to rationalize what he actually did. In part at least, this problem is made more acute because, as an experiment progresses, additional information is made available to the subject and he may become aware of the purpose of the study or of some deception in the later

240 Nebraska Symposium on Motivation, 1970

stages of an experiment. In that case his awareness will not have affected his earlier performance but may well have affected his later performance. (In many studies where there are strong order effects, different orders of presentation will yield different amounts of information at various points in the experiment.)

One approach to this problem has been suggested by Richard Solomon (personal communication). One would need to run a large number of subjects, only a small percentage of whom would actually complete the study. The others would be used as "sacrifice" groups. The experiment would be stopped for different groups at different junctures in the procedure and an inquiry carried out perhaps for one group shortly after the beginning, for another perhaps halfway through, for a third perhaps three-quarters of the way through, and so on. In this way it would be possible to establish the subjects' perceptions at different stages of an experimental procedure. Some of the subjects at least would be evaluated before they had had much opportunity to give behavioral responses, so that one could get an idea of the kind of cues which are made available to the subject by his own behavior. Unfortunately, such a technique is extremely time-consuming and costly.

An alternative procedure does not ask about the demand characteristics inherent in an experimental procedure at various points; rather it tests the overall impact much in the way inquiry does, but it does so without allowing the subject to experience his own response to the independent variable. This technique for uncovering the demand characteristics of a given experimental design is the preinquiry, or "nonexperiment." This procedure was independently proposed by Riecken (1962). A group of persons drawn from the same population as the actual experimental subjects will be selected from are asked to imagine that they are subjects themselves. They are shown the equipment that is to be used and the room in which the experiment is to be conducted. The procedure is explained in such a way as to provide them with information equivalent to that which would be available to the actual experimental subjects. However, they are not exposed to the experimental treatment; it is only explained.

In a nonexperiment on a certain drug, for example, the participant would be told that subjects are given a pill. He would be

241 Hypnosis, Motivation, and Ecological Validity

shown the actual pill, read the instructions destined for the experimental subjects, and then asked to produce data as if he had been subjected to the experimental treatment, that is, had taken the pill. If there are pretests in the study he would be given the pretest instructions first, take the pretests, then shown the pill and given the pill instructions, but not allowed to take it. He would then be told that an actual subject would have to wait a half hour for the pill to take effect. He would then be given the actual posttests, or asked to fill out the rating scales, or requested to carry out any behavior appropriate to the actual proposed experiment.

The nonexperiment yields data similar in quality to inquiry material, but it is obtained in the same form as an actual subject's data. Direct comparison of nonexperimental data and experimental data is therefore possible, but caution is essential. If these two kinds of data are identical, it shows only that the subject population in the actual experiment could have guessed what was expected of them. It cannot tell us that such guesses were the real determinants of their behavior.

Kelman (1965) has suggested that such a technique might appropriately be used as a social psychological tool to obviate the need for deception studies. The economy of such procedures is appealing and working in a situation where subjects become quasi collaborators instead of objects to be manipulated is more satisfying to many of us. Nonetheless it would seem dangerous to draw inferences to the actual situation in real life from results obtained in this fashion. In fact, when subjects in preinquiry experiments perform exactly as subjects do in actual experimental situations, it becomes impossible to know how much their performances is due to the independent variables and how much to the demand characteristics of the experimental situation. (See also Freedman, 1969, in this regard.)

In most psychological studies when one is investigating the subject's best possible performance in response to different physical or psychological stimuli, there is relatively little concern for the kind of problems introduced by demand characteristics. The need to take these factors into account becomes far more pronounced when investigating the effects of various interventions such as drugs, psychotherapy, hypnosis, sensory deprivation, conditioning of

242 Nebraska Symposium on Motivation, 1970

physiological responses, and so forth on performance or experiential parameters. Here the possibility that the subject's response may inadvertently be determined by the demand characteristics rather than the process itself must certainly be considered. Equally subject to these influences are studies where attitude changes rather than performance changes are explored. The investigator's intuitive recognition that a subject's perceptions of an experiment and of its meaning are very likely to affect the nature of his responses is probably one of the main reasons why deception studies have become so popular in the investigation of attitude change.

PROBLEMS OF INFERENCE

There is in the literature a very good illustration of the value of the nonexperiment. One of the major attractions of Festinger's (1957) cognitive dissonance theory has been that it makes predictions possible which appear to be counterexpectational; that is, it is commonly thought that the predictions made on the basis of intuitive common sense are wrong whereas those made on the basis of dissonance theory are both unexpected and validated by experimental findings. Bem (1967) has shown in an elegant application of the techniques which I have called the preinquiry experiment that subjects to whom the situation in typical cognitive dissonance studies is described in detail without being really placed in the situation are able to produce data closely resembling those observed. Bem appropriately questions the assertion that the dissonance theory allows counterexpectational predictions. His data effectively make the cognitive dissonance studies which his own nonexperiments replicate far less compelling than they were thought to be by showing that subjects could have figured out the way others respond. It should not be concluded, however, that because Bem was able to show that subjects not exposed to the situation could second-guess the experiment, further empirical studies are unnecessary. On the contrary, his findings show only that the avowed claims of these studies to counterexpectational results were not, in fact, justified, and that a more stringent test is required in future experiments that aim to demonstrate such effects.

In the same way, of course, I cannot agree with Kelman's sug-

243 Hypnosis, Motivation, and Ecological Validity

gestion that we can build a reliable social psychology on data obtained from subjects' role playing. Certainly it is true that subjects can often correctly role play the kind of behavior that would occur in actual situations. When we are dealing with such a situation and we do the actual experiment, the findings tend to strike us as relatively trivial. This is particularly true because we have, at best, validated our intuitive common sense. However, we are also left with a nagging doubt as to whether the subjects in the actual experiment were not responsive merely to the demand characteristics in the situation rather than to the independent variables themselves. Only when we succeed in setting up an experiment where the results are counterexpectational in the sense that a preinquiry yields findings different from those obtained from subjects in the actual situation can we be relatively sure that these findings represent the real effects of the experimental treatment.

It appears to me then that nonexperiments, or role-playing experiments, which are close relatives, are in essence first cousins of the simple postexperimental inquiry. The data they yield are based upon the active use of the subject's cognitive processes. Such nonexperiments require the subject to think through the situation and figure out how other subjects would behave. Certainly this is an extremely useful procedure to clarify findings obtained in actual experiments, but it does not supplant the need for substantive research. Subjects simply do not always know how they would respond when confronted with the actual situation.

THE PREINQUIRY TO HELP CLARIFY THE DEMAND CHARACTERISTICS

When it is possible to produce the same experimental results despite extreme variations in experimental procedures, or when identical experimental procedures carried out in different laboratories yield radically different results, the probability of demand characteristic effects must be seriously considered. An area of investigation characterized in this way was the early studies on sensory deprivation. The initial findings attracted wide attention because they not only had great theoretical significance for psychology but seemed to have practical implications for the space program

244 Nebraska Symposium on Motivation, 1970

as well. A review of the literature indicated that dramatic hallucinatory effects and other perceptual changes were typically observed after the subject had been in the experiment approximately two-thirds of the total time; however, it seemed to matter relatively little whether the total time was three weeks, two weeks, three days, two days, twenty-four hours, or eight hours. Clearly, factors other than physical conditions would have to account for such discrepancies. As a first quasi control we interviewed subjects who had participated in such studies (unpublished study). It became clear that they had been aware of the kind of behavior that was expected of them. Next a preinquiry was carried out, and from participants who were guessing how they might respond if they were in a sensory deprivation situation, we obtained data remarkably like those of the actual studies (Stare, Brown, & Orne, 1959). We were then in a position to design an actual experiment in which the demand characteristics of sensory deprivation were the independent variables (Orne & Scheibe, 1964).

One group of subjects were run in a "meaning deprivation" study which included the accouterments of sensory deprivation research but omitted the condition itself. They were required to undergo a physical examination, provide a short medical history, sign a release form, were "assured" of the safety of the procedure by the presence of an emergency tray containing various syringes and emergency drugs, and were taken to a well-lighted cubicle, provided food and water, and given an optional task. After taking a number of pretests, the subjects were told that if they heard, saw, smelled, or experienced anything strange they were to report it through the microphone in the room. They were again reassured and told that if they could not stand the situation any longer or became discomforted they merely had to press the red "panic button" in order to obtain immediate release.

They were then subjected to four hours of isolation in the experimental cubicle and given posttests. The control subjects were told that they were controls for a sensory deprivation study and put in the same objective conditions as the experimental subjects. Table 7 summarizes the findings, which indicate that manipulation of the demand characteristics by itself can produce many findings that have previously been ascribed to the sensory deprivation condition.

245 Hypnosis, Motivation, and Ecological Validity

246 Nebraska Symposium on Motivation, 1970

Of course, neither the quasi controls nor the experimental manipulation of the demand characteristics sheds light on the actual effects of the condition of sensory deprivation. They do show that demand characteristics may produce effects similar to those ascribed to sensory deprivation.

SIMULATORS

The use of the subjects' active cognitive processes to figure out what constitutes "appropriate" behavior can be carried one step further to provide yet another technique for uncovering demand characteristics, i.e., the use of simulators. Subjects are asked to pretend that they are being affected by an experimental treatment which they do not actually receive or to which they are immune. Note that this is merely an extension of the nonexperiment; but, whereas in that procedure the investigator is aware of which subject is which and is likely to treat them quite differently, the simulating technique permits subjects to be run by an experimenter who is in fact unaware of their actual status. As I have already pointed out, for this technique to be effective it is essential that the subject be aware that the experimenter is actually blind as well as that the experimenter actually be blind. The use of a blind experimenter makes certain that any effects due to experimenter bias are held constant across the actual group and the simulating group. We use this technique extensively in the study of hypnosis, and an example of it has already been described. Fortunately, it is possible for unhypnotized subjects to deceive an experimenter by acting as though they have been hypnotized. Obviously, it is essential that these subjects have no special training regarding the variables being studied so that they have no more information than that actually available to the real subjects. The simulating subjects have to guess what real subjects do in an experimental situation in response to instructions administered by a particular experimenter.

This design permits us to separate experimenter bias effects from demand characteristic effects. In addition to his other functions, the experimenter may be asked to judge whether each subject is "real" or simulating. In our work, this judgment tends to be random and unrelated to the true status of the subject; nevertheless, we have

247 Hypnosis, Motivation, and Ecological Validity

frequently observed differences between the behaviors of subjects contingent upon whether the experimenter has judged them to be hypnotized or just simulating. Such differences are likely to be due to differential treatment and bias, whereas differences between actual hypnotized subjects and actual simulators are likely to be due to hypnosis itself.

I would like to emphasize that the simulating group is a specialtreatment group. It is not a true control group in the classical sense, though it appears in some ways to be very similar. It is easy to confuse the simulating group with a true control group, since we have a blind experimenter treating two groups in identical ways and the data are being given in the same form by both groups; but they are different. Actually, the simulator is a very close relative of the role player in the nonexperiment. As in the postexperimental inquiry, we are still asking subjects to collaborate with us and to "psych out" how other subjects exposed to the actual treatment might behave. Consequently the inferences from the results of such a quasi-control group must be drawn with extreme caution. If we do not observe differences between deeply hypnotized subjects and simulators, this is not evidence that hypnosis is only a response to the demand characteristics of the situation. It merely indicates that our procedure has been unable to prove that something more is involved.

The problem here is the same as we have discussed with the preinquiry and the nonexperiment. In all probability there are many real effects of hypnosis which can be mimicked successfully by simulators. Thus the failure to find differences says a good deal about the effectiveness of the experimental procedure for studying hypnosis, but little about hypnosis itself. This is much the same situation as that with a patient who successfully malingers by simulating low back pain. His ability to persuade an orthopedic surgeon that his simulated pain is real cannot be taken as evidence that other patients do not have low back pain. It does not establish the nonexistence of the syndrome, but highlights the fallibility of the procedures used to establish the diagnosis.

Only when we are able to demonstrate differences between the behavior of real and simulating subjects can we feel that the experiment is compelling in demonstrating that a given effect is due to the presence of the independent variable. In the case of hypnotic

248 Nebraska Symposium on Motivation, 1970

experiments, the limitations of the procedure are such that these differences will either have to involve transcendence of unhypnotized performances or turn out to be counterexpectational in nature.

AN EXAMPLE OF THE USE OF SIMULATORS AS QUASI-CONTROLS

The problem of inference with quasi controls is illustrated in several studies (Evans, 1966; Orne & Evans, 1966) which we have carried out to investigate what happens if the investigator disappears after hypnosis has been induced. This is not an easy question to resolve. The hypnotist's disappearance must be managed in such a way as to seem plausible, truly accidental, and in accord with the implicit agreement between subject and hypnotist that the latter is responsible for the welfare of the former during the course of the experiment. Such a situation was finally created in a study requiring two sessions with subjects previously trained to enter hypnosis readily. It was explained to them that in order to standardize the procedure all instructions, including the induction and termination of hypnosis, would be carried out by tape recording.

The experimenter's task became essentially that of a technician turning on the tape recorder, applying electrodes, presenting experimental materials, etc. He did not say anything throughout the study, since every item of instruction was given by means of the tape recorder. Each subject came for two such sessions. During the second session, while the subject was deeply hypnotized and tapping his foot in rhythm with hallucinated music, the tape recorder came to a grinding halt, the light on the desk went out, and the experimenter tried in vain to reactivate the machine by flicking the switch. He then muttered under his breath that the fuse must be blown and dashed from the room. The subject's behavior was observed through a one-way screen throughout the experiment in order to discover what happened when the experimenter left the room. Would hypnosis be terminated immediately once the subject was alone, as some theories would predict, or would it take a period of time for the subject to pull himself out of hypnosis, as one would expect if the condition were some special altered state?

Despite the "cuteness" of the deception procedure of the simulated power failure, there was no certainty that subjects would

249 Hypnosis, Motivation, and Ecological Validity

accept it as such. For this reason quasi controls were run. In this particular situation simulating subjects were the most appropriate kind of quasi-control, and they were included without the hypnotist being aware which subjects were in fact deeply hypnotizable and which were unhypnotizable individuals instructed to simulate. The findings were clear-cut: five out of the six deeply hypnotized subjects did not rouse immediately when the hypnotist left. They required varying periods of time to pull themselves out of the state. When they did, they appeared somewhat disoriented and confused. However, five out of the six simulating subjects also continued to simulate hypnosis after the experimenter left the room. In fact, they did "better" than the real subjects and feigned trance until the technician returned half an hour later! During the subsequent postexperimental inquiry it became clear that the simulators caught on to the fact that the power failure was contrived and that they behaved as they thought a hypnotized subject would if there had been a true power failure. The simulating subjects quite cogently pointed out that having been instructed to fool the experimenter, they were on guard lest they themselves be fooled. In contrast, only one of the hypnotized subjects suspected that the power failure was contrived.

Because the quasi-control subjects are a special-treatment group, their purpose is not primarily to provide comparison objects for actual subjects; rather, it is to clarify the adequacy of the experimental procedure. Since the simulators caught on to the fact that the power failure was contrived, it was possible that at some level the hypnotized subjects also had responded to the same cues, even though they did not verbalize such awareness. Therefore, despite statistically significant differences between these two groups, the results did not permit any conclusions to be drawn about the nature of the hypnotic phenomenon. Rather, the behavior of the quasi controls demonstrated an inadequacy of the deception procedure and indicated that the behavior of the hypnotized subjects could not be confidently interpreted at face value.

Having found the deception manipulation wanting, we reexamined in detail the cues which might have allowed the subjects to become aware that the power failure was a planned part of the experiment and that his behavior would actually be observed after

250 Nebraska Symposium on Motivation, 1970

it. The kind of nitty-gritty details which most investigators fail to publish and which, even if they wished to, most editors would delete for the sake of brevity, are crucial in this regard. In the study just described, to facilitate observation after the "power failure," the overhead lights remained on and only the bright desk lamp went out as the tape recorder stopped; an obvious one-way mirror to which we hoped (in vain) that the subject would have adapted remained in the room; and, further, a small polygraph continued to run in the experimental room. So we now redesigned the study with considerable attention to such details. The one-way screen was eliminated and replaced by another less well known and carefully hidden means of observation. When the power failure occurred all the lights in the room were extinguished. Just enough light was filtered into the room through the window curtains from daylight -- or from a strategically placed parking light outside at night -- to make observation possible. Furthermore, physiological recordings were ostensibly made by means of a large Ampex instrumentation tape recorder and a small polygraph. Both of these machines stopped, along with the lights and the tape recorder, at the power failure. (Actually the wires from the subject were run through to another room at some distance in order to permit continuous recording of relevant data.) When the experimenter left the room the hall lights were out. Every effort was made by scrupulous attention to such details to have the power failure appear genuine.

With these revised conditions simulating and hypnotized subjects were run in another experiment by a blind experimenter. This time five out of six simulating subjects stopped simulating the moment the technician left the room, and indulged in exploratory or other private behavior, clearly indicating that they felt themselves to be unobserved. On the other hand, the hypnotized group behaved much as they had in the earlier experiment, pulling themselves out of the state only after a considerable period of time had elapsed. This behavior was in line with the data provided during an inquiry, which indicated that both real and simulating subjects accepted the power failure as genuine.

Significant differences were again obtained, but, in contrast to the findings of the first study, those from the second study allow meaningful inference. The behavior of the quasi-controls clearly

251 Hypnosis, Motivation, and Ecological Validity

indicates that the power failure was accepted as genuine, and therefore it is legitimate to accept the behavior of the hypnotized individuals at face value.8

DEMAND CHARACTERISTICS IN PSYCHOLOGICAL EXPERIMENTS

In the effort to understand hypnosis, the need to concern oneself with the effect of subtle cues which help to define the subject's appropriate response became abundantly clear. In other areas of inquiry -- most notably psychopharmacology -- similar effects have long been recognized under the heading of placebo response. These effects, complex yet exceedingly powerful, are ubiquitous. They are

These techniques are true quasi-controls and therefore do not permit any definitive inference to be drawn about the dependent variables. Certainly it would be inappropriate to conclude from such findings that the uninformed group must of necessity have been aware of the deception. As is the case with all quasi controls, they serve only to clarify flaws in the experimental procedure. In this instance they would show that the particular experimental and inquiry techniques employed cannot discriminate between informed and uninformed subjects. This could be either because the deception was so transparent that even uninformed subjects were aware of the true state of affairs or because an inadequate experimental technique was utilized that was insensitive to true differences between subjects who differed in their awareness of deception. The latter possibility, which implies that the same behavior could result from different mental processes in these two groups, cannot be excluded. The Golding-Lichtenstein study, for example, is an effective critique of the techniques that have been used to evaluate subject awareness and indicates the need for improved inquiry methods. The experiment provides a useful paradigm for testing the effectiveness of new inquiry procedures in assessing awareness. However, beyond showing that the evidence of nonawareness based on this inquiry technique is inconclusive, the experiment sheds no light on the extent to which uninformed subjects run with the Valins procedure were, in fact, aware of false feedback.

252 Nebraska Symposium on Motivation, 1970

quickly summarized in the clinical bromide that suggests one use a new drug while it still works.

In industrial psychology "placebo effects" have been recognized since the classic Hawthorne (Roethlisberger & Dickson, 1939) studies, and the investigators' enthusiasm has accounted for initial dramatic successes in a wide range of psychotherapeutic interventions.

It should be emphasized that the phenomena we are discussing here all stem from the individual's recognition that his behavior is under close scrutiny and that changes of some kind are expected. These are not effects of experimenter bias in the sense that the observations of the investigator are distorted or even that he treats the control and experimental groups differently in some subtle fashion. The focus of these effects lies in the perception of the subject more than the behavior of the experimenter. As long as the subject perceives himself to be under investigation, effects of this kind cannot be eliminated.

It is perhaps an irony that in areas of applied research which attempt to solve problems considered by many as mundane, effects of this kind are recognized and taken into account, while in much research carried out in the laboratory they have been systematically ignored.

Demand characteristics, the experimental analogue of the placebo effect of psychopharmacology or the Hawthorne effect of industrial psychology, cannot be controlled in the classic sense; that is, they cannot be eliminated by simply holding the effect constant or by allowing for it in some statistical sense. Rather, they are phenomena in their own right.

Perhaps the problem is best illustrated in psychopharmacology. Studying a drug by comparing it with the effect of placebo does not allow the investigator to determine the effect of the drug alone. Rather, he studies the placebo response plus drug effect as compared to the placebo response without drug effect. Such a design does not permit correct inference about the effect of drugs independent of the subject's awareness that he is taking medication, as has been demonstrated by Ross, Krugman, Lyerly, and Clyde (1962). Since psychopharmacology -- in contrast to most laboratory research -- is concerned to draw inferences about the effect of drugs on

253 Hypnosis, Motivation, and Ecological Validity

individuals who know they are taking medication, this issue can usually be ignored. Even here, however, a drug effect may function in interaction with the placebo effect, which would not necessarily become apparent in a double-blind study. For example, Fisher, Cole, Rickels, and Uhlenhuth (1964) showed that when meprobamate was compared with a placebo in a double-blind study and the medication was administered with scientific objectivity by the physician, there were no differences between drug and placebo; however, in the same double-blind design, when medication was administered with enthusiasm, conveying the expectation of improvement, meprobamate was clearly superior to the placebo. It would appear that, under some circumstances at least, real drug effects require appropriate "demand characteristics" before they become apparent, or, put in another way, real drug effects may be blocked by situational factors.

Experimental studies with human subjects rarely have utilized independent variables that are as readily manipulated as drugs, nor do we have the advantage of a control procedure even approaching the adequacy of the placebo, yet we are all too ready to extend conclusions from a context where a subject knows his behavior is under scrutiny to contexts where this is not the case. Though this is a problem of little moment to the psychopharmacologist, it is of great moment to the psychologist.

These difficulties are illustrated particularly well by the dramatic and highly original work of Milgram (1963, 1965) in obedience behavior. These studies are important because they address themselves to significant issues and, in contrast to most social psychological research, make an effort to address the question of ecological validity.

In all of the variations of Milgram's research, subjects are told that they are participating in an experiment to study the effect of punishment on memory. Two subjects, one of whom is a confederate, participate together. Each subject draws straws with the other and "by chance" the true subject ends up in the teacher's role. The other subject (the confederate) is strapped into an awesome-looking chair and electrodes are attached to him. Both subjects are always shown the shock generator which permits the administration of shocks from a very low voltage up to 450 volts.

254 Nebraska Symposium on Motivation, 1970

The machine is made all the more awesome by labeling the high end of the scale in red letters, "Danger -- Severe Shock -- XXX."

Before the beginning of the experiment a shock described as 45 volts is administered to the true subject, presumably so that he knows what the shocks are like. This, incidentally, is an extremely uncomfortable shock. The teacher's role is then explained to the subject -- how to present a paired-associate learning task to the learner, how he is to determine the duration of the punishment, and how a higher level of shock must be administered for each failure. Initially the learner is able to answer correctly two out of three times, but he rapidly begins to fail more and more frequently, requiring the subject to administer ever higher levels of shock to the victim: "Starting with 75 volts the learner begins to grunt and moan. At 150 volts he demands to be let out of the experiment. At 180 volts he cries out that he can no longer stand the pain. At 300 volts he refuses to provide any more answers to the memory test, insisting that he is no longer a participant in the experiment and must be freed" (Milgram, 1965, p. 60).

The experimenter remains calm throughout this procedure and instructs the subject to continue. If the teacher says that the learner is no longer responding and therefore there is no point in going on, he is told, "Treat the absence of an answer as equivalent to a wrong answer and follow the usual shock procedure." If the subject does not wish to do so, he is told, "You have no other choice, you must go on." Regardless of the degree of agitation shown by the subject or the anguished screams of the victim, the experimenter remains imperturbable and demands that the subject continue.

Milgram asked 45 psychiatrists what percentage of subjects they thought would continue to the top of the scale in this experiment, and on the average the view was that no more than one in a thousand would persist. However, in a series of studies which examined the effect on obedience of varied parameters such as the relationship between subjects (friends versus strangers), the proximity to the subject (in the next room, in the same room, the teacher being required to hold the learner's hand on the shock electrodes), and the legitimacy of the setting (Yale University versus run-down quarters in Bridgeport), a disturbingly high percentage of subjects nonetheless continued to shock the learners up to the very top of the scale.

255 Hypnosis, Motivation, and Ecological Validity

Finally, in one particularly dramatic study the learner inquired at the beginning of the study whether the fact that he had some heart trouble would be a problem. During the study the learner began to complain about chest pain and eventually appeared to pass out, with the strong implication that he had had a heart attack. Even under these circumstances a goodly number of subjects continued to administer shocks as directed to the top of the scale.

Milgram concludes that the willingness of subjects to administer shocks in this experiment indicates how easily obedience can be exacted from individuals by a legitimizing context. He implies that a similar mechanism operates in nonexperimental situations where violent and dangerous actions may be demanded and elicited from individuals, and that perhaps these studies help to explain the kind of behavior which occurred in concentration camps.

An entirely different way of viewing Milgram's studies is possible. The Milgram studies are a special instance of deception research, and in all such experiments it is vital that the investigator determine whether it is the subject or he, himself, who is deceived. Elsewhere (Orne & Holland, 1968) we have discussed the cues which indicate to the subject that the situation is basically safe. Of particular importance is the behavior of the experimenter, who sits by without being perturbed despite the moans, groans, and cries of anguish. While one might assume that the experimenter is merely unfeeling, it becomes more difficult to understand that he fails to be disturbed when the subject appears to pass out. The subject cannot help but be aware of the experimenter's culpability should anything happen to one of his subjects.

The superfluous nature of the subject's actual role is another aspect that could well raise doubts in the subject. For he does nothing that could not equally well be done by the investigator himself. Again, the necessarily awesome shock generator which is labeled "Danger -- Severe Shock -- XXX" may well provide cues that the situation is one more of role playing than of actual physical danger for the victim. In short, the labeling paradoxically indicates too much apparent danger for the situation to appear plausible.

For reasons that have been discussed earlier, it is exceedingly difficult to determine the extent to which subjects may have or may not have at some level perceived the many paradoxes in this

256 Nebraska Symposium on Motivation, 1970

situation and recognized that it must, of necessity, be completely safe for the apparent victim. The problems of interpretation are compounded by the fact that both subjects who accept the experiment as totally real and subjects who see through the experiment, recognizing that their own behavior is under investigation, are the ones likely to stop, whereas it is the subjects who are partially aware of the situation who are the ones most likely to be obedient. Therefore, awareness of deception is not only difficult to determine but, even if known, is not a simple correlate of obedience.

The problems of inference from such a study are illustrated by a different series of studies which were designed to investigate whether deeply hypnotized subjects can be compelled to carry out self-destructive or antisocial actions. Some years ago, Rowland (1939) showed that deeply hypnotized subjects could be compelled to pick up a poisonous snake with their bare hands, remove with their bare fingers a penny dissolving in a beaker of fuming nitric acid, and finally throw the acid at a research assistant. He also showed that these same subjects in the waking state, as well as other subjects, when asked whether they would carry out these actions, recoiled with horror and said emphatically, "No." This study was replicated by Young (1952) with essentially parallel findings. We (Orne & Evans, 1965) repeated the procedure yet a third time, adding, however, a quasi-control group of simulating subjects. Subjects were run blind. We were able to validate the observations of Rowland and Young that five out of six deeply hypnotized subjects could be compelled to carry out all three of these dangerous, self-destructive, and antisocial actions. However, six out of six simulators also carried out these behaviors.

Once we noted that simulators would carry out the behaviors, we ran subjects who were merely co-opted from the corridors and agreed to participate in a study, and we found that they were also ready to carry out the behaviors. Compliance turned out to be a function of the expectation that the experimenter conveyed in explaining the requirements. If he clearly expected compliance, subjects complied; if he asked whether subjects would carry out these actions, they said no.

Postexperimental inquiries revealed that the distinctions were not so much ones of the command quality of the way the request

257 Hypnosis, Motivation, and Ecological Validity

was made as of its cue quality. If one asks a subject whether he is willing to carry out such a behavior, the appropriate response is, of course, no. He would be foolish to do so. On the other hand, if it is clearly communicated that the experimenter expects him to carry out the actions, it is implicitly but nonetheless effectively communicated that, regardless of appearances, it is safe to proceed. As the experiment was being described, you, no doubt, realized that all kinds of safety precautions must have been taken to prevent the snake from biting and the acid from hurting either the subject or the research assistant. The awareness that I could not afford to have anyone hurt in my laboratory is not limited to this audience. It is shared by our subjects. Their appropriate recognition that the situation was, in the final analysis, totally safe, was a function of the kind of cues communicated in the experimental situation despite the fact that only one subject had the slightest inkling of how one could carry out these actions without either being hurt or hurting someone else.

The inference drawn from the Milgram studies was that, outside of an experimental situation, it would be relatively easy to cause individuals to maim or even kill for some presumed greater good, and it is tempting, of course, to point to the concentration camp situations or even the present war as examples of such eventualities. If one wished to draw conclusions from the Orne and Evans (1965) study, one would have to say that individuals would be prepared not only to hurt others but also to risk serious injury or death themselves. There is no doubt that under some conditions men will behave in precisely such a manner. However, it seems to me that the resemblance of these conditions to the experimental context is phenotypic rather than genotypic. Unfortunately, no evidence is available which allows one to determine the extent to which the Milgram experimental situation was perceived as "real" or the extent to which subjects were convinced that it was basically safe. In order to draw inference beyond the laboratory such information would be crucial. Ultimately, I believe, other kinds of data will be required to shed light on these issues. I do know that in the Orne and Evans study subjects did not for a moment expect that they would be seriously harmed or that they would harm anyone else, despite appearances to the contrary. Perhaps it is the recognition of the inadequacy of

258 Nebraska Symposium on Motivation, 1970

the experimental situation to yield answers to questions about antisocial behavior under hypnosis which makes me worry about the meaning of the Milgram experiments.

THE EXPERIMENT AS A TECHNIQUE OF INVESTIGATION

In this discussion I have focused upon my growing awareness of the complexity of experimental research with human subjects. When first confronted with the recognition of the difficulties inherent in working with one's fellow man as an experimental subject, one is tempted to discard the psychological experiment as a means of acquiring reliable knowledge; yet I certainly would not intend such a conclusion and would view it as most unfortunate. The experimental method remains the most effective single technique of exploring a broad range of problems. I would be equally troubled to see these comments taken as justification for replacing rigorous research with casual observation or for a return to armchair speculation. Indeed, it is precisely because of my interest in utilizing the experimental approach as a means of studying psychological mechanisms that I have felt compelled to concern myself with the psychology of the psychological experiment and the motivation of both the subject and the experimenter.9

The problems we encounter in psychological experiments may be somewhat more complex but not different in kind from those encountered in many sciences that seek to bridge the gap between laboratory research on a relatively isolated system and the function of that system in the complexity of the real world. No physiologist would assume that observations in vitro may cavalierly be assumed to represent the actual mechanism in vivo. On the contrary, he recognizes that an enzyme system studied in the test tube must then be studied in the liver slide, then in the liver itself, and finally in the total organism. While some of these steps may be telescoped, the translations from studies of relatively simple environments to ones of increasing complexity are required.

Indeed these problems are not confined to studies of living tissue. In aerodynamics, for example, the wind tunnel has become a crucial instrument. Without the opportunity to test variations in

259 Hypnosis, Motivation, and Ecological Validity

design under the controlled conditions of the wind tunnel, where a multitude of complex measurements can be obtained relatively easily and cheaply, modern plane design would not be possible. But it is not possible to take the observations in the wind tunnel and use them uncritically in the construction of an actual airplane. Rather, it is essential to include a series of conversion factors to match Reynold's numbers if one hopes to build a craft that will actually fly. Perhaps because of the staggering complexity inherent in all psychological research we have heretofore avoided concerning ourselves with the development of the conversion factors necessary to allow us to draw inference across situations. Egon Brunswik's (1947) concept of the ecological validity of research represents a rare but incisive effort at emphasizing this issue in the study of perception.

In my view the single most important variable which must be taken into account in psychological research with man is his recognition that he is under observation, that he is the object of study. Appropriately, Webb, Campbell, Schwartz, and Sechrest (1966) have emphasized the virtue of unobtrusive measures as one way in which this difficulty may be circumvented. Certainly, to the extent that we observe behaviors when subjects are unaware that they are being observed, we will eliminate difficulties.

The deception study is another, albeit methodologically difficult, approach to the problems that arise when the subject knows he is being studied; field studies are still a third. Even the judicious use of anecdotal material may, under some circumstances, serve important purposes.

While each of these approaches has considerable face validity and, with appropriate caution, may be extremely useful and important, it must be remembered that these techniques are often so involved, difficult, and complex that one could not ever hope to carry out with them the systematic work of exploring the effect of a given parameter on human behavior and experience. We cannot, for this reason, afford to give up the ordinary psychological experiment as a means of study. It is the only way parametric data can reasonably be obtained. On the other hand, neither can we afford to ignore the effects that are inevitably introduced when an active, thinking individual participates in a psychological experiment.

260 Nebraska Symposium on Motivation, 1970

In recognizing the importance of demand characteristics and their effects in psychological experimentation, it is all too easy to go to the opposite extreme and assume that they account for most, if not all, findings. Such a nihilistic view is at least as naive as that which denies the potential importance of these factors. We cannot afford merely to speculate on their importance in any given set of studies, and it is for this reason that I emphasize quasi-controls as one means by which at least a rough assessment of their effect may be obtained. I would hope that other, more sophisticated techniques will ultimately be devised which will permit the development of systematic conversion factors to allow the translation of the findings from the laboratory to other situations. Finally, we will need to recognize the necessity of demonstrating that psychological mechanisms which have been studied in the laboratory hold in other settings. To do so we will have to recognize that laboratory observations are not of necessity better data -- they are only data which are better controlled in a very clear sense but inevitably less well controlled in yet another sense. Ultimately, we seek congruence between the systematic findings obtained in laboratory experiments and those obtained in studies using unobtrusive measures, observations gathered from field studies, experiments in nature, systematized clinical observations, and even ordinary common-sense experience. In such a context, applied research may teach as much about basic psychological mechanisms as laboratory findings may clarify phenomena in the real world.

One would hope that these and other efforts clarifying the methodological issues of experimental research will ultimately lead to a psychology which is oriented to understanding and predicting significant human experience and behavior in extraexperimental as well in as experimental settings.

REFERENCES

Aronson, E., & Carlsmith, J. M. Experimentation in social psychology. In G. Lindzey & E. Aronson (Eds.), The Handbook of Social Psychology. Vol. 2. Research Methods. Reading, Mass.: Addison-Wesley, 1968.

Ashley, W. R., Harper, R. S., & Runyon, Dale L. The perceived size of coins in normal and hypnotically induced economic states. American Journal of Psychology, 1951, 64, 564-572.

261 Hypnosis, Motivation, and Ecological Validity

Barber, T. X., & Calverley, D. S. Toward a theory of "hypnotic" behavior: Enhancement of strength and endurance. Canadian Journal of Psychology, 1964, 18, 156-167.

Bern, D. J. Self-perception: An alternative interpretation of cognitive dissonance phenomena. Psychological Review, 1967, 74, 183-200.

Binet, A., & Féré, C. Animal magnetism. (Trans.) New York: Appleton, 1888.

Boring, E. J. A history of experimental psychology. (2nd ed.) New York & London: Appleton-Century-Crofts, 1950.

Bruner, J. S., & Goodman, C. C. Value and need as organizing factors in perception. Journal of Abnormal and Social Psychology, 1947, 42, 33-44.

Brunswik, E. Systematic and representative design of psychological experiments with results in physical and social perception. (Syllabus Series, No. 304) Berkeley: University of California Press, 1947.

Carmichael, L., Hogan, H. P., & Walter, A. An experimental study of the effect of language on the reproduction of visually perceived forms. Journal of Experimental Psychology, 1932, 15, 73-86.

Cataldo, J. F., Silverman, I., & Brown, J. M. Demand characteristics associated with semantic differential ratings of nouns and verbs. Educational and Psychological Measurement, 1967, 27, 83-87.

Charcot, J. M. Oeuvres complètes. Paris: Aux Bureaux du Progrès Médical, 1886. 9 vols.

Coué, E. Self mastery through conscious autosuggestion. New York American Library Service, 1922. P. 83.

Damaser, Esther Experimental study of long-term posthypnotic suggestion. Unpublished doctoral dissertation, Harvard University, 1964.

Damaser, Esther C., Shor, R. E., & Orne, M. T. Physiological effects during hypnotically requested emotions. Psychosomatic Medicine, 1963, 25, 334-343.

Du Maurier, G. Trilby. New York: Harper, 1895.

Ellson, D. G., Davis, R. C., Saltzman, I. J., & Burke, C. J. A report on research on detection of deception. Indiana University, Department of Psychology, 1952, Contract Nonr 60nr-18011, Office of Naval Research.

Evans, F. J. The case of the disappearing hypnotist. Paper presented at the meeting of the American Psychological Association, New York, September 1966.

Evans, F. J., & Orne, M. T. Motivation, performance, and hypnosis. International Journal of Clinical and Experimental Hypnosis, 1965, 13, 103-116.

Festinger, L. A theory of cognitive dissonance. New York: Row & Peterson, 1957.

Fisher, R. M. An empirical examination of the dissociation hypothesis of hypnosis. Unpublished bachelor (honors) dissertation, Harvard University, 1960.

Fisher, S. The role of expectancy in the performance of posthypnotic behavior. Journal of Abnormal and Social Psychology, 1954, 49, 503-507.

Fisher, S., Cole, J. O., Rickels, K., & Uhlenhuth, E. H. Drug-set interaction: The effect of expectations on drug response in out-patients. In P. B. Bradley, F. Flugel, and P. Hoch (Eds.), Neuropsychopharmacology. Vol. 3. New York: Elsevier, 1964. Pp. 149-156.

Freedman, J. L. Role playing: Psychology by consensus. Journal of Personality and Social Psychology, 1969, 13, 107-114.

262 Nebraska Symposium on Motivation, 1970

Garfinkel, H. Studies in ethonomethodology. Englewood Cliffs, N.J.: Prentice-Hall, 1967.

Gill, M. M., & Brenman, Margaret. Hypnosis and related states. New York: International Universities Press, 1959.

Golding, S. L., & Lichtenstein, E. Confession of awareness and prior knowledge of deception as a function of interview set and approval motivation. Journal of Personality and Social Psychology, 1970, 14, 213-223.

Gustafson, L. A., & Orne, M. T. Effects of heightened motivation on the detection of deception. Journal of Applied Psychology, 1963, 47, 408-411.

Gustafson, L. A., & Orne, M. T. Effects of perceived role and role success on the detection of deception. Journal of Applied Psychology, 1965, 49, 412-417.

Holland, C. H. Sources of variance in the experimental investigation of behavioral obedience. Unpublished doctoral dissertation, University of Connecticut, 1967.

Hull, C. L. Hypnosis and suggestibility. New York & London: Appleton-Century, 1933.

Kellogg, E. R. Duration and effects of post-hypnotic suggestions. Journal of Experimental Psychology, 1929, 12, 502-514.

Kelman, H. C. The human use of human subjects: The problems of deception in social-psychological experiments. Paper presented at the meeting of the American Psychological Association, Chicago, September 1965.

Kroger, R. O. The effects of role demands and test-cue properties upon personalitytest performance. Journal of Consulting Psychology, 1967, 31, 304-312.

Levitt, E. E., & Brady, J. P. Muscular endurance under hypnosis and in the motivated waking state. International Journal of Clinical and Experimental Hypnosis, 1964, 12, 21-27.

Levy, L. H. Awareness, learning, and the beneficent subject as expert witness. Journal of Personality and Social Psychology, 1967, 6, 365-370.

Lewin, K. Die Entwicklung der experimentellen Willenspsychologie and die Psychotherapie. Leipzig: Verlag Von S. Hirzel, 1929.

London, P., & Fuhrer, M. Hypnosis, motivation, and performance. Journal of Personality, 1961, 29, 321-333.

Mann, T. Mario and the magician. Lowe-Porter (Trans.) New York: Knopf, 1931.

Milgram, S. Behavioral study of obedience. Journal of Abnormal and Social Psychology, 1963, 67, 371-378.

Milgram, S. Some conditions of obedience and disobedience to authority. Human Relations, 1965, 18, 57-76.

Mills, T. M. A sleeper variable in small groups research: The experimenter. Pacific Sociological Review, 1962, 5, 21-28.

Nace, E. P., & Orne, M. T. Fate of an uncompleted posthypnotic suggestion. Journal of Abnormal Psychology, 1970, 75, 278-285.

O'Connell, D. N., Shor, R. E., & Orne, M. T. Hypnotic age regression: An empirical and methodological analysis. Journal of Abnormal Psychology, in press.

Orne, M. T. The mechanisms of hypnotic age regression: An experimental study. Journal of Abnormal and Social Psychology, 1951, 46, 213-225.

Orne, M. T. Hypnosis, hypnotherapy, and medical practice. Tufts Medical Journal, 1953, 21, 14-25.

263 Hypnosis, Motivation, and Ecological Validity

Orne, M.T. Die Leistungsfähigkeit in Hypnosis und im Wachzustand. Psychologische Rundschau, 1954, 5, 291-297.

Orne, M. T. The demand characteristics of an experimental design and their implications. Paper presented at the meeting of the American Psychological Association, Cincinnati, September 1959. (a)

Orne, M. T. The nature of hypnosis: Artifact and essence. Journal of Abnormal and Social Psychology, 1959, 58, 277-299. (b)

Orne, M. T. On the social psychology of the psychological experiment: With particular reference to demand characteristics and their implications. American Psychologist, 1962, 17, 776-783.

Orne, M. T. The nature of the hypnotic phenomenon: Recent empirical studies. Paper presented at the meeting of the American Psychological Association, Philadelphia, September 1963.

Orne, M.T. Demand characteristics and the concept of quasi-controls. In R. Rosenthal & R. Rosnow (Eds.), Artifact in behavioral research. New York: Academic Press, 1969. Pp. 143-179.

Orne, M. T. Hypnosis, state or role? Paper presented at the 5th International Congress on Hypnosis and Psychosomatic Medicine, Mainz, Germany, April 1970.

Orne, M. T., & Evans, F. J. Social control in the psychological experiment: Antisocial behavior and hypnosis. Journal of Personality and Social Psychology,1965, 1, 189-200.

Orne, M. T., & Evans, F. J. Inadvertent termination of hypnosis with hypnotized and simulating subjects. International Journal of Clinical and Experimental Hypnosis, 1966, 14, 61-78.

Orne, M. T., & Holland, C. H. On the ecological validity of laboratory deceptions. International Journal of Psychiatry, 1968, 6, 282-293.

Orne, M. T., & Scheibe, K. E. The contribution of nondeprivation factors in the production of sensory deprivation effects: The psychology of the "panic button." Journal of Abnormal and Social Psychology, 1964, 68, 3-12.

Orne, M. T., Sheehan, P. W., & Evans, F. J. Occurrence of posthypnotic behavior outside the experimental setting. Journal of Personality and Social Psychology, 1968, 9, 189-196.

Page, M. M., & Lumia, A. R. Cooperation with demand characteristics and the bimodal distribution of verbal conditioning data. Psychonomic Science, 1968, 12, 243-244.

Patten, E. F. The duration of post-hypnotic suggestions. Journal of Abnormal and Social Psychology, 1930, 25, 319-334.

Pattie, F. A. A report of attempts to produce uniocular blindness by hypnotic suggestion. British Journal of Medical Psychology, 1935, 15, 230-241.

Riecken, H. W. A program for research on experiments in social psychology. In N. F. Washburne (Ed.), Decisions, values and groups. Vol. 2. New York: Pergamon Press, 1962. Pp. 28-42.

Roethlisberger, F. J., & Dickson, W. J. Management and the worker. Cambridge, Mass.: Harvard University Press, 1939.

Rosenberg, M. J. The conditions and consequences of evaluation apprehension. In R. Rosenthal & R. L. Rosnow (Eds.), Artifact in behavioral research. New York: Academic Press, 1969. Pp. 279-349.

Rosenhan, D., & London, P. Hypnosis: Expectation, susceptibility and perform ance. Journal of Abnormal and Social Psychology, 1963, 66, 77-81. (a)

Rosenhan, D., & London, P. Hypnosis in the unhypnotizable: A study in rote learning. Journal of Experimental Psychology, 1963, 65, 30-34. (b)

264 Nebraska Symposium on Motivation, 1970

Rosenthal, R. On the social psychology of the psychological experiment: The experimenter's hypothesis as unintended determinant of experimental results. American Scientist, 1963, 51, 268-283.

Rosenthal, R. Experimenter effects in behavioral research. New York: Appleton-Century Crofts, 1966:

Ross, S., Krugman, A. D., Lyerly, S. B., & Clyde, D. J. Drugs and placebos: A model design. Psychological Reports, 1962, 10, 383-392.

Rowland, L. W. Will hypnotized persons try to harm themselves or others? Journal of Abnormal and Social Psychology, 1939, 34, 114-117.

Sarbin, T. R. Contributions to role taking theory: I. Hypnotic behavior. Psychological Review, 1950, 57, 255-270.

Shor, R. E. Hypnosis and the concept of the generalized reality-orientation. American Journal of Psychotherapy, 1959, 13, 582-602.

Sher, R. E. Physiological effects of painful stimulation during hypnotic analgesia under conditions designed to minimize anxiety. International Journal of Clinical and Experimental Hypnosis, 1962, 10, 183-202.

Shor, R. E., & Orne, Emily C. The Harvard Group Scale of Hypnotic Susceptibility, Form A: An adaptation for group administration with self-report scoring of the Stanford Hypnotic Susceptibility Scale, Form A. Palo Alto, Calif.: Consulting Psychologists Press, 1962.

Silverman, I. Role-related behavior of subjects in laboratory studies of attitude change. Journal of Personality and Social Psychology, 1968, 8, 343-348.

Silverman, I., Shulman, A. D., & Wiesenthal, D. L. Effects of deceiving and debriefing psychological subjects on performance in later experiments. Journal of Personality and Social Psychology, 1970, 14, 203-212.

Stalnaker, J. M., & Riddle, E. E. The effect of hypnosis on long-delayed recall. Journal of General Psychology, 1932, 6, 429-440.

Stare, F., Brown, J., & Orne, M. T. Demand characteristics in sensory deprivation studies. Unpublished seminar paper, Massachusetts Mental Health Center and Harvard University, 1959.

Stricker, L. J., Messick, S., & Jackson, D. N. Suspicion of deception: Implications for conformity research. Journal of Personality and Social Psychology, 1967, 5, 379 389.

Sutcliffe, J. P. A general method of analysis of frequency data for multiple classification designs. Psychological Bulletin, 1957, 54, 134-137.

Turner, L. H., & Solomon, R. L. Human traumatic avoidance learning: Theory and experiments on the operant-respondent distinction and failures to learn. Psychological Monographs, 1962, 76 (Whole No. 559), 1-32.

Valins, S. Cognitive effects of false heart-rate feedback. Journal of Personality and Social Psychology, 1966, 4, 400-408.

Webb, E. J., Campbell, D. T., Schwartz, R. D., & Sechrest, L. Unobtrusive measures: Nonreactive research in the social sciences. Chicago: Rand McNally, 1966.

Weitzenhoffer, A. M., & Hilgard, E. R. Stanford Hypnotic Susceptibility Scale, Forms A and B. Palo Alto, Calif.: Consulting Psychologists Press, 1959.

Weitzenhoffer, A. M., & Hilgard, E. R. Stanford Hypnotic Susceptibility Scale, Form C. Palo Alto, Calif.: Consulting Psychologists Press, 1962.

265 Hypnosis, Motivation, and Ecological Validity

White, R. W. A preface to a theory of hypnotism. Journal of Abnormal and Social Psychology, 1941, 36, 477-506.

Young, P. C. Antisocial uses of hypnosis. In L. M. LeCron (Ed.), Experimental hypnosis. New York: Macmillan, 1952. Pp. 376-409.

Zamansky, H. S., Scharf, B., & Brightbill, R. The effect of expectancy for hypno sis on prehypnotic performance. Journal of Personality, 1964, 32, 236-248.

Zeigarnik, B. Über das Behalten von erledigten and unerledigten Handlungen. Psychologische Forschung, 1927, 9, 1-85.

Table 5 (p. 237) (from Gustafson, L.A., & Orne, M.T. Effects of perceived role and role success on the detection of deception. Journal of Applied Psychology, 1965, 49, 412-417.) is reproduced here with the kind permission of the American Psychological Association © 1965. No further reproduction or distribution of this table is permitted without written permission of the publisher.

Table 6 (p. 238) (from Gustafson, L.A., & Orne, M.T. Effects of perceived role and role success on the detection of deception. Journal of Applied Psychology, 1965, 49, 412-417.) is reproduced here with the kind permission of the American Psychological Association © 1965. No further reproduction or distribution of this table is permitted without written permission of the publisher.

Table 7 (p. 245) (from Orne, M.T., & Scheibe, K.E. The contribution of nondeprivation factors in the production of sensory deprivation effects: The psychology of the “panic button.” Journal of Abnormal and Social Psychology, 1964, 68, 3-12.) is reproduced here with the kind permission of the American Psychological Association ©1964. No further reproduction or distribution of this table is permitted without written permission of the publisher.