Redalyc.Hypnosis and Placebos: Response Expectancy As a Mediator of Suggestion Effects

Anales de Psicología ISSN: 0212-9728 [email protected] Universidad de Murcia España

Kirsch, Irving Hypnosis and Placebos: Response Expectancy as a Mediator of Suggestion Effects. Anales de Psicología, vol. 15, núm. 1, 1999, pp. 99-110 Universidad de Murcia Murcia, España

Disponible en: http://www.redalyc.org/articulo.oa?id=16715108

Cómo citar el artículo Número completo Sistema de Información Científica Más información del artículo Red de Revistas Científicas de América Latina, el Caribe, España y Portugal Página de la revista en redalyc.org Proyecto académico sin fines de lucro, desarrollado bajo la iniciativa de acceso abierto anales de psicología © Copyright 1999: Servicio de Publicaciones de la Universidad de Murcia 1999, vol . 15, nº 1, 99-110 Murcia (España). ISSN: 0212-9728 Hypnosis and Placebos: Response Expectancy as a Mediator of Suggestion Effects

Irving Kirsch

University of Connecticut

Titulo: Hipnosis y placebos: la expectativa de respuesta Abstract: In this paper, data are reviewed indicating that como mediador de los efectos de la sugestión hypnotic and placebo effects share a common mechanism: Resumen: En este artículo se revisan datos que sugieren response expectancy. Placebos and hypnotic procedures que los efectos de la hipnosis y de los placebos comparten may have therapeutic effects because they change the un mismo mecanismo: la expectativa de respuesta. Los client's expectations. Both placebo and hypnosis are placebos y los procedimientos hipnóticos tienen efectos viewed as instances of the broader phenomenon of terapéuticos porque modifican las expectativas de los suggestion. While other variables may play important roles clientes. Se conciben ambos como ejemplos del fenómeno in the response to suggestions, it remains to be known más amplio de la sugestión. Otras variables pueden jugar un whether they affect responses by changing expectations or papel importante en la respuesta a las sugestiones, pero aún have a direct impact on them. The potential disdvantages of falta por saber si afectan a dichas respuestas a través de una using the idea of trance in therapy are also discussed. modificación de las expectativas o tienen un impacto Key words: Hypnosis, placebo, suggestion, response expec- directo sobre las respuestas. Se discuten también los efectos tancy negativos de la utilización de la idea del trance en la terapia. Palabras clave: Hipnosis, placebo, sugestión, expectativa de respuesta

Hypnosis is a procedure in which a person gestions contain challenges. Amnesia sugges- designated as hypnotist suggests changes in ex- tions, for example, require that the person not perience to a person designated subject (Kirsch, recall particular information. These hypnotic 1994; Kirsch & Lynn, 1995). A representative phenomena have been characterized as the range of hypnotic suggestions are sampled in "domain of hypnosis" (Hilgard, 1973). An standardized scales of suggestibility, most of adequate theory of hypnosis should be able to which are highly reliable and substantially cor- account for the full range of behavior con- related with each other (Council, in press). tained in this domain. Suggestions on these scales are generally There are many books and journal articles thought to be of three basic types. Ideomotor focusing on the topic of hypnosis and hypno- suggestions are suggestions that a particular ac- tizability. In contrast, despite notable excep- tion, such as an arm rising in the air (arm levi- tions, such as the work by Amigó (in press) tation), will occur automatically, without and Capafons (in press) at the University of awareness of volitional effort. Challenge sugges- Valencia, the broader subject of suggestion has tions are suggestions that the person cannot been relatively neglected. There are historical perform an act that is normally under volun- reasons for this state of affairs. The effects of tary control, such as bending an arm (arm ri- mesmerism and hypnosis have seemed so un- gidity). Cognitive suggestions are suggestions for usual that many observers dismissed them as various cognitive or perceptual distortions, fraudulent, and others attributed them to some such as selective amnesia, pain reduction, and special condition or state. It seemed impossi- hallucinations. Note that some cognitive sug- ble for suggestion to have such extraordinary effects, without there being some kind of special state creating them. * Dirección para correspondencia / Address for corres- pondence: Irving Kirsch. University of Connecticut. Storrs, The results of research on hypnosis in the CT 06269-1020. United States of America. twentieth century have clearly invalidated that E-Mail: [email protected] conclusion. One of the first things that was

- 99 - 100 I. Kirsch learned through experimental research on most clearly indicates the need for a change in hypnosis was that all behaviors seen in hypno- focus from the topic of hypnosis to the topic sis can also be obtained without hypnosis. of suggestion. Hilgard's data indicated that the Clark Hull (1933, p. 391) wrote: effect of hypnotic inductions on suggestibility The only thing which characterizes hypnosis as such were generally quite small. A person who re- and which gives any justification for calling it a "state" sponds to six of the twelve suggestions con- is its generalized hypersuggestibility. That is, an increase in suggestibility takes place upon entering the tained in a typical hypnotic suggestibility scale hypnotic trance. The difference between the hypnotic is likely to respond to five of them even with- and normal state is therefore quantitative rather than out the induction of hypnosis (Kirsch, 1997). qualitative. No phenomenon whatever can be produced in hypnosis that cannot be produced to lesser Some of the effects of hypnosis may not degrees by suggestions given in the normal waking be suggested by the hypnotist. The best known condition. the essence of hypnosis lies in the fact of of these is so called spontaneous amnesia, change in suggestibility [emphasis in the original]. which in past centuries was a hallmark of hyp- These data suggest that suggestion, rather than nosis, but currently is virtually nonexistent. Al- hypnosis, is the fundamental phenomenon on though not directly suggested by the hypnotist, which we should focus. these phenomena are far from spontaneous. This conclusion is reinforced by two more Instead, they too are products of suggestion-- recent sources of data. One is the work of in this case, suggestions that are transmitted by nonstate theorists, and in particular, the ex- the culture. Young and Cooper (1972) demon- perimental work of T. X. Barber and his fol- strated this quite nicely. They told one group lowers. It was Barber (1969) who most clearly of subjects that hypnotized people experience demonstrated that all of the observed effects spontaneous amnesia, and they told a second of hypnosis, including the increase in suggesti- group that hypnotized subjects do not experi- bility that was observed following a hypnotic ence spontaneous amnesia. The subjects were induction, could be duplicated by nonhypnotic later tested for their beliefs about hypnosis. procedures. Thus, hypersuggestibility, the sin- Forty-eight percent of those in the first group gle remaining justification that Clark Hull agreed, with the statement "If I were to be could find for retaining the trance concept, hypnotized I would not remember what had can be produced by nonhypnotic methods. happened after I woke up (Young & Cooper, Importantly, the Barber Suggestibility Scale 1972). In contrast, only 15% of the second differed from previous scales in that it in- group expected to experience spontaneous cluded a measure of subjective or experiential amnesia. When later hypnotized and tested for response to suggestion, as well as a measure of so called "spontaneous" amnesia, 37% of the behavioral response. Thus, Barber was able to subjects in the first group displayed it, com- demonstrate that thesubjective responses to pared to only 10% of the in the second group. suggestion did not depend on the induction of Thus, the occurrence of amnesia was hardly a hypnotic trance. This, of course, cast doubt "spontaneous." This is further confirmed by on the concept of hypnotic trance, doubt another interesting finding in these data. which has been reinforced by the failure to Across both groups, 75% of the subjects who find any physiological or even self-report expected amnesia experienced it, whereas markers of the hypothesized state. It turns out none of those who did not expect amnesia ex- that the most prominent experts cannot distin- perienced it. This illustrates a central theme of guish the self-reports of hypnotized subjects this paper: the mediating role of expectancy in from those of subjects who have not been the ability of suggestion to elicit a response. hypnotized (Kirsch, Mobayed, Council, & What makes hypnosis interesting is peo- Kenny, 1992). ple's responses to such suggestions as analge- Paradoxically, it is the work of E. R. Hil- sia, amnesia, age regression, nonvolitional gard (1965) on suggestibility as a trait that movements, and positive and negative halluci- anales de psicología, 1999, 15(1)

Hypnosis and Placebos: Response Expectancy as a Mediator of Suggestion Effects 101 nations. Without phenomena such as these, nonvolitionally, rather than emitted intention- there would be little to interest scholars or lay ally. Often this distinction is clear in the phras- persons in hypnosis. So if these responses do ing of the suggestion. "Raise your hand" is an not require hypnosis for their production, and instruction, whereas "Your hand is getting if the effect of hypnosis is merely to enhance lighter and beginning to rise" is a suggestion. them to a slight degree, then clearly, the focus The first calls for an intentional act; the sec- of investigation should be on the broader topic ond for a nonvolitional response. Similarly, of suggestion and its effects. "Take this medication" is an instruction. In Broadening the focus of attention from contrast, "It will help you sleep" is a sugges- hypnosis to suggestion also has the advantage tion, because in suggests to the person that of adding the very important topic of placebo taking the pill will automatically induce sleep. effects. Placebos are typically viewed as arti- Sometimes, the suggestive nature of a facts to be controlled in treatment outcome re- communication is not apparent from its lin- search. However, the documented effects of guistic character. In fact, suggestions need not placebos on pain, asthma, tension, anxiety, de- be linguistic utterances at all. Suggestive in- pression, blood pressure, heart rate, sexual formation may be conveyed by the size, shape, arousal, skin conditions, nausea, vomiting, gas- and color of a pill, for example, or by the be- tric motility, and angina (reviewed in Kirsch, havior of a model. Furthermore, the suggestive 1990) reveals it to be a very important phe- nature of words depend on more than the nomenon, well worthy of study in its own words themselves. The command "Sleep!" for right. example, can be interpreted as either an in- There are, of course, both similarities and struction or a suggestion, depending on the differences between hypnotic and placebo context in which it is delivered. When given to phenomena. Whereas responses to the kind of a child who is still awake at 11 p.m., it is suggestions typically used in hypnosis are no- clearly a command; when given to a volunteer toriously trait-like, individual differences in re- at a stage hypnosis performance, it is a sugges- sponse to placebos appear to be very unreli- tion. What is important, then, is the meaning able. There does not appear to be a "placebo of the communication, as it is understood by reactor," comparable to the "hypnotizable sub- the person to whom it has been directed. ject." Still, the stimulus for hypnotic and pla- Of course, most hypnotic subjects want to cebo responses are suggestions, and hypnotic experience suggested phenomena, and they in- inductions are so nonspecific that they might tentionally do their best to bring them about. well be regarded as placebos. In fact, placebo However, the experience of automaticity or substances have been used successfully as nonvolition is part of the communicated sug- hypnotic inductions (Baker & Kirsch, 1993; gestion, so that simple behavioral compliance Glass & Barber, 1961). Most important, as will would not be experienced as a successful re- be shown in this paper, the effects of hypnotic sponse. and placebo suggestions are mediated by re- Statements are often interpreted as sugges- sponse expectancy (Kirsch, 1985, 1990). But tions, rather than commands, because the re- first, it may be worthwhile to define what is sponse is not experienced as being under di- meant by the term suggestion. rect volitional control. Most people cannot in- tentionally reduce pain, see things that are not What is a Suggestion? there, sleep, or forget, in the same direct way that they can raise or lower their arms. If they A suggestion is a communication indicating can accomplish these responses at all, it is by that an individual will experience a particular doing other things that are under volitional response. It differs from an instruction or control. For example, they may think about command in that the response is to occur other things to distract themselves from the

anales de psicología, 1999, 15(1)

102 I. Kirsch pain, or they might try to imagine the sug- pectancy theory, I thought it important to dis- gested hallucination. Their behavior is much tinguish between these stimulus expectancies and like that of an actor who thinks of sad experi- response expectancies (Kirsch, 1985). Response ences in order to produce tears. The thoughts expectancies are expectancies for the occur- of the sad experiences are voluntary acts rence of nonvolitional responses, such as pain, aimed at producing the nonvolitional response alertness, fear, sadness, and joy. Like expected of tears. Still, it is the contextually determined stimuli, expected responses are valued out- meaning of the communication that differenti- comes, and as such they are determinants of ates a suggestion from an instruction or com- voluntary behavior. We may drink coffee in mand. The stage director who tells an actor to the morning to help wake us up, and we may cry is giving an instruction; whereas the physi- avoid it in the evening, if we think it will keep cian who says "Crying is a side effect of this us from getting to sleep. However, response medication" and the hypnotist who says "You expectancies seem to have a property that are beginning to cry" are giving suggestions. stimulus expectancies do not have: They are Finally, it is important to note that whether self-confirming. When people expect to feel or not some verbal or nonverbal stimulus is a alert, they often do feel alert; and when people suggestion does not depend on the response of expect to stay awake, they may find themselves the subject. Suggestions are suggestions, even unable to sleep. when subjects do not respond to them. Like stimulus expectancies, response expectancies are derived from direct and vicari- Suggestions, Expectancies, and Non- ous experience. Experience with active drugs volitional Responses and medications, for example, leads us to expect those drugs and medications to have par- A suggestion is a particular type of stimulus. It ticular effects. Similarly, being told that a drug is a stimulus that conveys information that a has a particular effect or observing its effect nonvolitional response will occur. Sometimes on others can produce an expectation of that the response occurs, sometimes it does not. So effect when the drug is ingested. However, the question is, what is it that determines the data indicate that direct experience is more response to a suggestion? In my work on hyp- powerful than vicarious experience in shaping nosis and placebo effects, I have focused on response expectancies (Wickless & Kirsch, the role of expectancy as a mediating variable. 1989). Response expectancies are somewhat Accepting a suggestion means coming to ex- similar to self-efficacy expectations, and in pect that the suggested event will occur. some circumstances the two constructs over- That expectancies are determinants of be- lap considerably (Kirsch, 1985). For example, havior is a central tenant of cognitive- expected anxiety (a response expectancy) is the behavioral learning theory (Tolman, 1955) so- primary determinant of self-efficacy for ap- cial learning theory (Bandura, 1977; Rotter, proaching a feared object or situation 1954; Mischel, 1973), behavioral decision the- (Schoenberger, in press),and a hypnotic re- ory (Edwards, 1954), achievement motivation sponse expectancy is closely related to the be- theory (Atkinson, 1957; Heckhausan, 1977), lief that one is capable of experiencing a sug- and the theory of reasoned action (Ajzen & gested effect. Unlike self-efficacy expectations, Fishbein, 1980). In all of these theories, behav- however, response expectancies are anticipa- ior is postulated to be a multiplicative function tions of outcomes that will occur and are only of expected outcomes and their values. partly based on evaluations of one’s ability. In these theories, expected outcomes are The response expectancy that coffee will make generally thought of as external events, such as one alert, for example, is likely to be inde- food, water, money, school grades, recogni- pendent of self-efficacy judgments. Also, be- tion, and affection. In developing response ex- cause the experience of automaticity is central anales de psicología, 1999, 15(1)

Hypnosis and Placebos: Response Expectancy as a Mediator of Suggestion Effects 103 to hypnosis, expectancies for responding are placebo injections are more effective that pla- more properly termed response expectancies. cebo pills (Traut & Passarelli, 1957). Also, pla- The self-confirming nature of response ex- cebo effects are readily obtained in relatively pectancies indicate that they might mediate the sterile, nonclinical, experimental settings. effects of suggestion. The logic behind this Second, placebos produce both positive conclusion is as follows. First, the suggestive and negative effects, and they do so in the nature of a communication is defined by the same people. Furthermore, the specific nature recipient's interpretation that a nonvolitional of the effect (i.e., whether the placebo affects response is expected to occur. Second, re- gastric motility, sexual arousal, pain percep- sponse expectancy is defined as an expectancy tion, etc.) depends on the information avail- of a nonvolitional response (Kirsch, 1985), in able to the recipient. Placebo analgesics, for contrast to intention, which has been defined as example, have very different effects than pla- the expectancy of a voluntary behavior (Ajzen cebo tranquilizers. These characteristics of & Fishbein, 1980). Thus, a connection be- placebo effects cannot be explained by faith, tween response expectancy and suggestion is trust, hope, or any other hypothesis based on implied by the definitions of these terms. Per- the valence of placebo-induced expectancies. haps we are hard-wired in such a way that ex- Third, recent data from my laboratory in- pecting a subjective experience produces that dicate that placebo pain reduction cannot be experience, in the same way that deciding explained by mechanisms like anxiety reduc- (termed intending in reasoned action theory) to tion or endorphin release (Montgomery & emit a voluntary act (e.g., lifting one's arm) Kirsch, 1996). These are global mechanisms produces that act. The two clearest examples that would affect the entire body. We obtained of this phenomenon are placebo effects and a placebo effect by administering the placebo hypnosis. in the guise of a local anesthetic and applying a pain stimulus to treated and untreated parts of Placebo Effects the body. Because the pain stimulus was ap- plied simultaneously to both the treated (by Although the mechanisms by which placebos placebo) and untreated locations, the differ- produce their effects have not yet been estab- ences in reported pain could not have been lished, the data exclude some hypotheses and due to any global changes in sensitivity, per- indicate some parameters within which any ception, or affect. After all, one cannot be anx- successful theory must fit. Most importantly, ious in one hand and calm in the other. explanations of placebo effects must account Taken together, these data demonstrate for their specificity. Data indicating the highly that placebo effects are specific to the infor- specific nature of placebo effects indicate that mation with which the placebo is administered. they cannot be explained fully by such global This specificity makes theories based on global factors as rapport, trust, faith, hope, anxiety mechanisms implausible. An adequate theory reduction, or endorphin release. of placebo responding must be able to account First, it is clear that placebo effects are not for the specific nature of the effects. entirely due to the quality of the doc- tor/patient relationship. Although such rela- Classical Conditioning tionship factors as touch have been demonstrated to affect some therapeutic outcomes Classical conditioning is one of the most (Whitcher & Fisher, 1979), there are ample frequently proffered theoretical explanations data indicating that the effects of placebos de- of placebo effects (Turkkan, 1989; Wick- pend on their information value. With rela- ramasekera, 1980). According to conditioning tionship factors held constant, different place- accounts of placebo effects, effective medical bos produce different results. For example, treatments are conditioning trials, during

anales de psicología, 1999, 15(1)

104 I. Kirsch which the vehicles (pills, capsules, etc.) in stand the occurrence of salivation following a which active medication (the US) is delivered stimulus that signals food. The conditional become conditional stimuli (CSs), thereby ac- stimulus elicits the expectation of food, the quiring the capability of evoking the effects of thought of food causes the organism to sali- medication as conditional responses (CRs). vate. Similarly, because of their previous asso- The classical conditioning model has the ad- ciation with active medication, pills elicit ex- vantage of being able to account for the speci- pectations of particular changes. But why ficityof placebo reactions. It predicts side ef- should the expectation produce the expected fects as well as therapeutic effects, and it is effect? consistent with the observation that placebos Unlike contemporary accounts of classical evoke the same responses as the active drugs conditioning, traditional stimulus substitution they are replacing. However, there are two models seem to explain the occurrence of the problems with conditioning explanations of placebo responses. However, leaving aside the placebo phenomena. First, they are based on data that led to the replacement of that model an outdated account of classical conditioning. by contemporary cognitive models, data from Second, there are abundant disconfirming data. many placebo studies are inconsistent with Traditional accounts of classical condition- traditional conditioning models: First, with ing suggest that parings of the CS with the US tranquilizers as UCs, conditioning trials lead to the automatic evocation of URs follow- weaken the placebo response instead of ing presentation of the US. In contrast, con- strengthening it (Meath, Feldberg, Rosenthal, temporary theorists (e.g., Rescorla, 1988) view & Frank, 1956; Pihl & Altman, 1971; Rickels, classical conditioning as a means by which in- Lipman, & Raab, 1966; Segal & Shapiro, 1959; formation is acquired. Conditioning trials en- Zukin, Arnold, & Kessler, 1959). Also, there is dow the CS with information value, so that it an inverse relation between the strength of the becomes a cue for the occurrence of the US. US (i.e., the tranquilizer) and the magnitude of As a result, the CR may not be the same re- the placebo effect that is presumed to be the sponse as the unconditional response (UR). In- CR (Rickels et al., 1966). Second, placebo ef- stead, the CR is an anticipatory response that fects can resist extinction over periods as great prepares the organism for the onset of the US. as two years or more (Boissel, et al., 1986; Also, pairings of conditional and unconditional Coryell & Noyes, 1988; Traut & Passarelli, stimuli result in conditioning only under those 1957). Third, when people expect effects that circumstances in which they impart informa- are different from those produced chemically tion value to the CS. One way of interpreting by the drug, the placebo effect is consistent this is that the effect of conditioning trials on with the expectation rather than with the UR behavior is mediated by expectancies, so that if (Fillmore & Vogel-Sprott, 1992; Hull & Bond, expectancy change is blocked, so too is the ef- 1986; Kirsch & Weixel, 1988). Finally, under fect on behavior. some circumstances, placebos can produce ef- This contemporary understanding of con- fects that are stronger than those of the active ditioning phenomena can explain how place- drugs that are presumed to be the USs bos (the CS) come to generate internal repre- (Frankenhaeuser, Post, Hagdahl, & Wrangsjö, sentations of active medications (the US), and 1964; Ikemi & Nakagawa, 1962; Lyerly, Ross, since responses have stimulus value, it is not Krugman, & Clyde, 1964; Ross, Krugman, much of a stretch to invoke this model to ac- Lyerly, & Clyde, 1962; Wolf, 1950). All of count for the acquisition of response expec- these data are inconsistent with explanations of tancies as a function of conditioning trials placebo effects based on simple Pavlovian (Kirsch, 1985). However, this leaves the oc- models. currence of the placebo response (the presum- edCR) unexplained. It is not difficult to under- anales de psicología, 1999, 15(1)

Hypnosis and Placebos: Response Expectancy as a Mediator of Suggestion Effects 105

Psychophysiological Explanation and Notwithstanding the importance of such data, the Hypothesis of Unmediated Expec- they do not constitute an explanation. Instead, tancy Effects they are phenomena in need of explanation. How is it that placebos enhance endorphin re- Faith, hope, rapport, and anxiety reduction lease (if and when that does in fact occur)? are psychological intervening variables that The problem that is illustrated by the en- have been hypothesized to mediate the rela- dorphin release example is that which occurs tion between expectancy and expected re- when any psychological variable (such as sug- sponse. The data I reviewed earlier in this pa- gestion) is hypothesized to produce a physio- per indicate that none of them are necessary logical effect. Following the conventions pro- for evocation of placebo effects. It is possible posed by Michael Hyland (1985), discovering that there are no intervening psychological an apparent cause of a physical effect ought to variables between a response expectancy and trigger a search for the physiological substrates an expected response (Kirsch 1985). In this of the psychological event. Hyland proposed a sense, the relation between expectancy and re- principle of mind/brain complementarity, sponse may be the same as that between inten- based on Bohr's principle of wave/particle tion and intended response. The relation be- complementarity in physics. Briefly stated, it tween an intention and a voluntary response suggests that mind states and their correspond- has been characterized as “immediate” (Ajzen ing brain states are complementary descrip- & Fishbein, 1980), meaning that it is not medi- tions of the same underlying event, a position ated by any psychological intervening variables that Hyland and I have shown to be a logically between the intention and the response. In- necessary consequence of virtually all monist deed, recent work in cognitive and social psy- mind-body philosophies (Kirsch & Hyland, chology indicate that previously formed inten- 1987). If mind states and body states are de- tions may activate behavior automatically, with scriptions of the same underlying event, it is no need to focus attention on the intended re- technically improper to state that one causes sponse at the time that it is initiated (Bargh, J. the other. Instead, psychophysiological expla- A., & Gollwitzer, 1994; Kirsch & Lynn, 1997, nation requires establishing three sorts of rela- 1998). Similarly, the relationship between sad- tions between variables. These are: (1) causal ness the perception of loss is presumed to be relations between mind states and other mind immediate, in this sense of the term. One may states, (2) causal relations between physiologi- ask, for example, "Why are you sad?" and the cal states and other physiological states, and (3) answer may be "Because my mother just died." identity relations between mind states and There seems no need to ask, "Why did that their corresponding physiological states. make you sad?" I have hypothesized that the It is here that the data on the specificity of relation between response expectancies and placebo effects becomes particularly useful. It nonvolitional responses may have this same suggests that we need to look for physiological immediate quality. If this is the case, further substrates of very specific psychological states. explanation requires consideration of physio- Even the psychological construct of expec- logical rather than psychological variables. tancy is likely to be too broad. Instead, we One of the pitfalls to avoid in psycho- need to establish the physiological correlatives physiological theorizing is that of treating a of specific expectancies, such as expectations physiological variable as a mediating variable of alterations in arousal, pain sensitivity, nau- between a psychological variable and a de- sea, and so on. pendent variable. For example, a finding that endorphin release may be required for certain kinds of pain reduction tempts one to consider it an explanation of placebo pain reduction.

anales de psicología, 1999, 15(1)

106 I. Kirsch

Hypnosis amnesia (Young & Cooper, 1972), the ability to resist suggestions (Lynn, Nash, Rhue, Hypnotic Inductions as Nondeceptive Frauman, & Sweeney, 1984; Spanos,Cobb, & Placebos Gorasssini, 1985), the ability to breach suggested amnesia (Silva & Kirsch, 1987), and the What is a hypnotic induction? Charcot induced nature of "hidden observer" reports (Spanos & hypnosis by clanging gongs, flashing lights, ap- Hewitt, 1980). In sum, hypnotic inductions are plying pressure to subjects' heads. Braid as nonspecific as placebos, but hypnotic and thought that eye-fixation was necessary, but placebo-induced experiences and behaviors are the Spiegels have subjects roll their eyes, and as specific as the expectancies that mediate many hypnotists merely ask subjects to close their occurrence. them. Most contemporary inductions include The seminal study of McGlashan, Evans, suggestions for relaxation, but increased alert- and Orne (1969) has been interpreted as indi- ness can be suggested instead, and relaxation cating that hypnosis and placebo effects are can be prevented by having subjects pedal a not related. That study purported to demon- stationary bicycle. The only common ingredi- strate that hypnotic analgesia was more effec- ent to these inductions is the label hypnosis. As tive than a placebo among highly responsive Sheehan and Perry (1976, p. 72) noted, "it is subjects, thus indicating that there was more to not the procedural conditions per se that are hypnosis than expectancy effects. As impor- important but whether or not the subject per- tant as that study was, however, there was a fa- ceives them as part of a context that is 'appro- tal flaw in its design. One of the essential priate' for displaying hypnotic behavior." characteristics of placebo controlled investiga- When the effect of administering a drug is tions is that the placebo match the treatment found to be independent of the its specific infor which it is serving as a control. The impor- gredients, the drug is deemed to be a placebo. tance of this requirement derives from the fact Similarly, hypnotic inductions must be expec- that different placebos have different effects. tancy manipulations, akin to placebos, because In a review of double-blind drug studies, for their effects on suggestibility are independent example, Evans (1974) concluded that placebo of any specific component or ingredient. In morphine was considerably more effective fact, it is possible to produce all of the sugges- than placebo Darvon, which in turn was more tive effects of hypnosis by giving subjects pla- effective than placebo aspirin. In the cebos and telling them that the medication McGlashan et al. study, the placebo was pre- produces a hypnotic state (Baker & Kirsch, sented as an analgesic and administered in 1993; Glass & Barber, 1961). If hypnosis is an Darvon capsules. It stands to reason that sub- altered state or condition, then it is a state that jects who have experienced hypnotically- is produced by placebos. This establishes a induced hallucinations and amnesia during the clear association between hypnosis and pla- selection procedure would expect greater pain cebo effects. relief from hypnosis than from Darvon. Recall that placebo effects are highly spe- In a more recent study, a colleague and I cific, in that the nature of the effect depends compared hypnotic analgesia to to the effects on the information that is presented to sub- of two different placebos. One of these was jects. A similar specificity is found in hypnotic described to subjects as a "pain-reducing anal- responses. Subjects' responses during hypnosis gesic." The other was described as a "hypnotic depend on their expectancies of how a hyp- drug" that "increases suggestibility." We repli- notic subject should respond. These expectan- cated the superiority of hypnosis to placebo cies have been shown to affect the qualitative when the placebo was presented as a pain- experience of trance (Henry, 1985), spontane- relieving drug. However, placebo and hypnosis ous arm catalepsy (Orne, 1959), spontaneous were equally effective when the placebo was anales de psicología, 1999, 15(1)

Hypnosis and Placebos: Response Expectancy as a Mediator of Suggestion Effects 107 presented as a drug that induces hypnosis These data provide strong evidence for a (Baker & Kirsch, 1993). Furthermore, in both causal relation between expectancy and hypno- the placebo condition and the hypnosis condi- tizability, but they still leave some variance in tion, expectancy was significantly correlated responsiveness unexplained. It is possible that with pain reduction, whereas the association expectancy is the sole proximal determinant of between hypnotizability and pain reduction hypnotizability and that the residual variance is was only marginal. a result of measurement error. Conversely, the unexplained variance may be due to a talent or Individual Differences in Responsive- personality characteristic, the nature of which ness is yet to be established.

Expectancy determines the circumstances Placebos, Hypnosis and "Sensory Sug- under which a good hypnotic subject experi- gestibility" ences and displays hypnotic phenomena. It also determines the kind of phenomena that There is an important difference between pla- good subjects experience and display. But cebos and hypnosis that is worth noting for what determines the degree to which a subject practical reasons. The administration of place- responds? Does expectancy produce hypnotic bos entails deception. Hypnosis does not. Phy- experiences only in susceptible subjects, or is it sicians and psychotherapists are justifiably re- also one of the determinants of hypnotic sus- luctant to use placebos for this reason. Be- ceptibility? cause the use of hypnosis does not require de- Expectancy is one of the few stable corre- ception, it can be used as a Nondeceptive lates of hypnotizability (Kirsch & Council, means of exploiting the therapeutic potential 1992). Although early studies indicated that of suggestion. For example, practitioners who these correlations were only moderate, much use hypnosis to enhance cognitive-behavior higher correlations, some as high as .71 and psychotherapy inform their clients that hypno- .84, have been reported in more recent studies sis does not involve going into a trance, but is (Council et al., 1983; Council et al., 1986; instead a method of helping them become Johnson et al., 1989; Kirsch, 1991). Still, corre- deeply involved in creating new experiences lation does not establish causality. It is possible (Kirsch, 1993). that expectancy is an epiphenomenon rather The deceptive nature of placebos makes than a cause of responsiveness. More convinc- that phenomena similar to the tests from ing evidence of causality is provided by studies which the concept of "secondary suggestibil- in which manipulated expectancies produced ity" (Eysenck & Furneaux, 1945) was derived. changes in responsiveness. Kirsch, Council, In fact, the labels deceptive and nondeceptive sug- and Mobayed (1987) demonstrated that altered gestion might be preferable to direct and indirect, expectancies can account for more variance so as to avoid confusion with the very differ- than trait hypnotizability (i.e., pre-manipulation ent concepts of direct and indirect suggestions responsiveness) in subsequent hypnotic sug- used by the Ericksonians. Gheorghiu's (1989) gestibility. In another study (Wickless & "indirect-direct" approach to measuring "sen- Kirsch, 1989), the effect of an expectancy ma- sory suggestibility" occupies an intermediate nipulation was so strong that 73% of the sub- ground between the nondeceptive suggestions jects scored in the high range of responsive- used in hypnosis and the deceptive procedures ness (9-12) on form C of the Stanford Scale used in earlier attempts to measure sensory and the remaining 27% scored in the moderate suggestibility. range (5-8). Not one subject scored in the low There is, however, a dimension on which range (0-4). placebo and hypnotic suggestions differ from those used to measure sensory suggestibility.

anales de psicología, 1999, 15(1)

108 I. Kirsch

Both placebo and hypnotic suggestions elicit affect response by means of their impact on response expectancies. They are suggestions expectancy, or are there other mediating vari- that changes will occur within the individual. ables that have direct (immediate) effects on In contrast, tests of secondary or sensory sug- response? gestibility are more closely related to stimulus Hypnotic procedures have evolved as if expectancies. They are suggestions that the their creators were aware of the importance of stimulus will change. The placebo literature shaping subjects' response. Skilled practitio- demonstrates clearly that response expectan- ners are permissive. They present and respect cies are self-confirming. Stimulus expectancies choices, often in the form of therapeutic dou- may be less so, perhaps because internal states ble-binds, so that either choice promotes im- are more ambiguous than external stimuli. In provement. They prevent failure by beginning any case, suggesting that a person has become with easy tasks that the client is almost certain less sensitive to pain is not the same thing as to accomplish, and they define tasks so that suggesting that the pain stimulus has been re- failure is impossible. They evaluate perform- duced in intensity. The degree to which they ances at any level as indications of success, elicit comparable results and the correlations and they structure expectations so that even between them are worthy topics for future in- small improvements are seen as significant be- vestigation. ginnings. They are alert to random fluxuations and capitilize on those that occur in a desired Conclusions direction. They also prepare clients for set- backs by labeling them in advance as inevita- The data presented in this paper indicate that ble, temporary, and useful learning opportuni- hypnotic responses and placebo effects share a ties. These practices, which have evolved from common mechanism, that of response expec- clinical and experimental hypnosis, can be tancy. They share a common mechanism be- used to maximize expectancy effects in non- cause they are subsets of a broader phenome- hypnotic psychotherapy as well. non: the phenomenon of suggestion. Like pla- There is one mistake, however, that most cebos, hypnosis produces therapeutic effects clinical hypnotists continue to make, and that by changing client's expectancies. But unlike is the continued use of the altered state con- placebos, hypnosis does not require deception cept and terminology. The idea of going into a in order to be effective. Whereas placebos are trance scares many clients and inhibits them presented deceptively as pharmacological from experiencing hypnotic effects. In con- treatments, hypnosis is presented honestly as a trast, debunking the altered state myth and psychological procedure. Furthermore, hon- presenting hypnosis from a cognitive- estly informing clients about what has been behavioral perspective enhances subsequent learned through research about the nature of responsiveness. hypnosis may reduce resistance and increase Not only does trance terminology scare responsiveness to hypnotic interventions. clients away from the potential benefits of Expectancy is not the only variable mediat- hypnosis, it also scares many professionals ing the effects of suggestion on behavior, but needlessly. There are hospital administrators it is certainly a critical variable. Among the who prohibit the use of hypnosis in their insti- questions that remain to be answered are the tutions and therapists who are reluctant to put following: What other variables can be shown their clients into an altered state, fearing that to influence response to suggestion, and what the clients might get stuck in that state. De- are the relations between those variables and spite the involvement of such prominent fig- expectancy? Specifically, is expectancy the fi- ures as Binet, Freud, Hull, and Hilgard, hyp- nal link in the causal chain between suggestion nosis remains stigmatized as a mysterious, and response, so that other mediating variables quasi-mystical procedure. The trance concept anales de psicología, 1999, 15(1)

Hypnosis and Placebos: Response Expectancy as a Mediator of Suggestion Effects 109 surely deserves at least some of the blame for this unfortunate state of affairs.

References

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and Eysenck, H. J., & Furneaux, W. D. (1945). Primary and sec- predicting social behavior. Englewood Cliffs, NJ: Prentice- ondary suggestibility: An experimental and statistical Hall. study. Journal of Experimental Psychology, 35, 485-503. Amigó, S. (in press). Self-Regulation Therapy: Suggestion Fillmore, M., & Vogel-Sprott, M. (1992). Expected effect of Without Hypnosis. In I. Kirsch, A. Capafons, E. caffeine on motor performance predicts the type of re- Cardeña, & S. Amigó (Eds.), Clinical hypnosis and self- sponse to placebo. Psychopharmacology, 106, 209-214. regulation: Cognitive-behavioral perspectives. Washington, Frankenhaeuser, M., Post, B., Hagdahl, R., & Wrangsjö, B. DC: American Psychological Association. (1964). Effects of a depressant drug as modified ex- Atkinson, J. W. (1957). Motivational determinants of risk- perimentally-induced expectation. Perceptual and motor taking behavior. Psychological Review, 46, 359-372. skills, 18, 513-522. Baker, S. L., & Kirsch, I. (1993). Hypnotic and Placebo An- Gheorghiu, V. A. (1989). The development of research on algesia: Order Effects and the Placebo Label. Contempo- suggestibility: Critical considerations. In V. A. Gheor- rary Hypnosis, 10, 117-126. ghiu, P. Netter, H. J. Eysenck, & R. Rosenthal (Eds.), Bandura, A. (1977). Social learning theory. Englewood Cliffs, Suggestion and suggestibility: Theory and Research (pp. 3-55). NJ: Prentice Hall. Berlin: Springer-Verlag. Barber, T. X. (1969). Hypnosis: A scientific approach. New Glass, L. B., and Barber, T. X. (1961). A note on hypnotic York: Van Nostrand Reinhold. behavior, the definition of the situation, and the pla- Bargh, J. A., & Gollwitzer, P. M. (1994). Environmental cebo effect. Journal of Nervous and Mental Diseases, 132, control of goal-directed action: Automatic and strate- 539-541. gic contingencies between situations and behavior. Ne- Heckhausen, H. (1977). Achievement motivation and its braska symposium on motivation, 41, 71-124. constructs: A cognitive model. Motivation and Emotion, Boissel, J. P., Philippon, A. M., Gauthier, E., Schbath, J., 1, 283-329. Destors, J. M., & the B.I.S. Research Group. (1986). Henry, D. (1985). Subjects' expectancies and subjective ex- Time course of long-term placebo therapy effects in perience of hypnosis. Unpublished doctoral disserta- angina pectoris. European Heart journal, 7, 1030-1036. tion, University of Connecticut. Capafons, A. (in press). Applications of Emotional Hilgard, E. R. (1965). Hypnotic susceptibility. New York: Har- Self-regulation Therapy. In I. Kirsch, A. Capafons, E. court, Brace & World. Cardeña, & S. Amigó (Eds.), Clinical hypnosis and self- Hilgard, E. R. (1973). The domain of hypnosis: With some regulation: Cognitive-behavioral perspectives. Washington, comments on alternate paradigms. American Psychologist, DC: American Psychological Association. 28, 972-982. Coryell, W., & Noyes, R. (1988). Placebo response in panic Hull, C. L. (1933). Hypnosis and suggestibility: An experimental disorder. American Journal of Psychiatry, 145, 1138-1140. approach. New York: Appleton-Century Crofts. Council, J. R. (in press). Measures of hypnotic responding. Hull, J. G., & Bond, C. F. (1986). Social and behavioral con- In I. Kirsch, A. Capafons, A. Cardeña, S., & Amigó sequences of alcohol consumption and expectancy: A (Eds.), Clinical hypnosis and self-regulation: Cognitive- meta-analysis. Psychological Bulletin, 99, 347-360. behavioral perspectives. Washington, DC: American Psy- Hyland, M. E. (1985). Do person variables exist in different chological Association. ways? American Psychologist, 40, 1003-1010. Council, J. R., Kirsch, I., & Hafner, L. P. (1986). Expec- Ikemi, Y., & Nakagawa, S. (1962). A psychosomatic study of tancy versus absorption in the prediction of hypnotic contagious dermatitis. Kyoshu Journal of Medical Science, responding. Journal of Personality and Social Psychology, 50, 13, 335-350. 182-189. Johnston, J. C., Chajkowaski, J., DuBreuil, S. C., & Spanos, Council, J., Kirsch, I., Vickery, A. R., and Carlson, D. N. P. (1989). The effects of manipulated expectancies (1983). "Trance" vs. "skill" hypnotic inductions: The on behavioural and subjective indices of hypnotisabil- effects of credibility, expectancy, and experimenter ity. Australian Journal of Clinical and Experimental Hypnosis, modeling. Journal of Consulting and Clinical Psychology, 51, 17, 121-130. 432-440. Kirsch, I. (1985). Response expectancy as a determinant of Edwards, W. (1954). The theory of decision making. Psycho- experience and behavior. American Psychologist, 40, 1189- logical Bulletin, 51, 380-417. 1202. Evans, F. J. (1989). The independence of suggetibility, pla- Kirsch, I. (1990). Changing expectations: A key to effective psycho- cebo response, and hypnotizability. In V. A. Gheor- therapy. Pacific Grove, CA: Brooks/ Cole. ghiu, P. Netter, H. J. Eysenck, & R. Rosenthal (Eds.), Kirsch, I. (1991). The social learning theory of hypnosis. In Suggestion and suggestibility: Theory and Research (pp. 145- S. J. Lynn & J. Rhue (Eds.), Theories of hypnosis: Current 154). Berlin: Springer-Verlag. models and perspectives (pp. 439-466). New York: Guilford Evans, F.J. (1974). The placebo response in pain reduction. Press. Advances in Neurology, 4, 289-296. Kirsch, I. (1993). Cognitive-Behavioral Hypnotherapy. In J. W. Rhue, S. J. Lynn, & I. Kirsch (Eds.), Handbook of

anales de psicología, 1999, 15(1)

110 I. Kirsch

clinical hypnosis (pp. 151-171). Washington, DC: Ameri- Rescorla, R. A. (1988). Pavlovian conditioning: It's not what can Psychological Association. you think it is. American Psychologist, 43, 151-160. Kirsch, I. (1994). APA definition and description of hypno- Rickels, K., Lipman, R., & Raab, E. (1966). Previous medi- sis: Defining hypnosis for the public. Contemporary Hyp- cation, duration of illness, and placebo response. The nosis, 11, 142-143. Journal of Nervous and Mental Disease, 142, 548-554. Kirsch, I. (1997). Suggestibility or Hypnosis: What Do Our Ross, S., Krugman, A. D., Lyerly, S. B., & Clyde, D. J. Scales Really Measure? International Journal of Clinical and (1962). Drugs and placebos: A model design. Psychologi- Experimental Hypnosis, 45, 212-225. cal Reports, 10, 383-392. Kirsch, I., & Council, J. R. (1992). Situational and personal- Rotter, J. B. (1954). Social learning and clinical psychology. ity correlates of suggestibility. In E. Fromm and M. Englewood Cliffs, NJ: Prentice-Hall. Nash (Eds.), Contemporary hypnosis research (pp. 267-292). Schoenberger, N. E. (in press). Expectancy and fear. In I. New York: Guilford. Kirsch (ed.), Expectancy, experience, and behavor. Washing- Kirsch, I., & Hyland, M. E. (1987). How thoughts affect the ton, DC: American Psychological Association. body: A metatheoretical framework. Journal of Mind and Segal, M., & Shapiro, K. L. (1959). A clinical comparison Behavior, 8, 417-434. study of the effects of resperine and placebo on anxi- Kirsch, I., & Lynn, S. J. (1995). The altered state of hypno- ety. Archives of Neurology and Psychiatry, 81, 392-398. sis: Changes in the theoretical landscape. American Psy- Sheehan, P. W., & Perry, C. W. (1976). Methodologies of hypno- chologist, 50, 846-858. sis. Hillsdale, NJ: Lawrence Erlbaum Associates. Kirsch, I., & Lynn, S. J. (1997). Hypnotic involuntariness Silva, C. E., & Kirsch, I. (1987). Breaching amnesia by ma- and the automaticity of everyday life. American Journal of nipulating expectancy. Journal of Abnormal Psychology, 96, Clinical Hypnosis, 40, 329-348. 325-329. Kirsch, I., & Lynn, S.J. (1998). Social-cognitive alternatives Spanos, N. P., & Hewitt, E. C. (1980). The hidden observer to dissociation theories of hypnosis. Review of General in hypnotic analgesia: Discovery or experimental crea- Psychology, 2, 66-80. tion? Journal of Personality and Social Psychology, 39, Kirsch, I., & Weixel, L. J. (1988). Double-Blind Versus De- 1201-1214. ceptive Administration of a Placebo. Behavioral Neuro- Spanos, N.P., Cobb, P.C., & Gorassini, D. (1985). Failing to science, 102, 319-323. resist hypnotic test suggestions: A strategy for self- Kirsch, I., Council, J. R., & Mobayed, C. (1987). Imagery presenting as deeply hypnotized. Psychiatry, 48, 282 - and response expectancy as determinants of hypnotic 292. behavior. British Journal of Experimental and Clinical Hyp- Tolman, E. C. (1932). Purposive behavior in animals and nosis, 4, 25-31. men. New York: Appleton-Century-Crofts. Kirsch, I., Mobayed, C. P., Council, J. R., & Kenny, D. A. Traut, E. F., & Passarelli, E. W. (1957) Placebos in the (1992). Expert judgments of hypnosis from subjective treatment of rheumatoid arthritis and other rheumatic state reports. Journal of Abnormal Psychology, 101, 657- conditions. Annals of the Rheumatic Diseases, 16, 18-22. 662. Turkkan, J. S. (1989). Classical conditioning: The new he- Lyerly, S. B., Ross, S., Krugman, A. D., & Clyde, D. J. gemony. Behavioral and Brain Sciences, 12, 121-179. (1964). Drugs and placebos: The effects of instructions Whitcher, S. J., & Fisher, J. D. (1979). Multidimensional re- upon performance and mood under amphetamine sul- action to interpersonal touch. Journal of Personality and phate and chloral hydrate. Journal of Abnormal and Social Social Psychology, 37, 87-96. Psychology, 68, 321-327. Wickless, C., & Kirsch, I. (1989). The effects of verbal and Lynn, S. J., Nash, M. R., Rhue, J. W., Frauman, D. C., & experiential expectancy manipulations on hypnotic Sweeney, C. A. (1984). Nonvolition, Expectancies, and susceptibility. Journal of Personality and Social Psychology, Hypnotic Rapport. Journal of Abnormal Psychology, 93, 57, 762-768. 295-303. Wickramasekera, I. (1980). A conditioned response model McGlashan, T. H., Evans, F. J., & Orne, M. T. (1969). The of the placebo effect: Predictions from the model. Bio- nature of hypnotic analgesia and placebo response to feedback and Self-Regulation, 5, 5-18. experimental pain. Psychosomatic Medicine, 31, 227-246. Wolf, S. (1950). Effects of suggestion and conditioning on Meath, J. A., Feldberg, T. M., Rosenthal, D., & Frank, J. D. the action of chemical agents in human subjects--the (1956). Comparison of reserpine and placebo in treat- pharmacology of placebos. Journal of Clinical Investigation, ment of psychiatric outpatients. A. M. A. Archives of 29, 100-109. Neurology and Psychiatry, 76, 207-214. Young, J., & Cooper, L. M. (1972). Hypnotic recall amnesia Mischel, W. (1973). Toward a cognitive social reconceptu- as a function of manipulated expectancy. Proceedings of alization of personality. Psychological Review, 80, 252-283. the 80th Annual Convention of the American Psychological As- Montgomery, G., & Kirsch, I. (1996). Mechanisms of Pla- sociation, 7, 857-858. cebo Pain Reduction: An Empirical Investigation. Psy- Zukin, P., Arnold, D. G., & Kessler, C. (1959). Comparative chological Science, 7, 174-176. effects of phenaglycodol and meprobromate on anxi- Orne, M. T. (1959). The nature of hypnosis: Artifact and ety reactions. Journal of Nervous and Mental Disease, 129, essence. Journal of Abnormal Psychology, 58, 277-299. 193-195. Pihl, R. O., & Altman, J. (1971). An experimental analysis of the placebo effect. Journal of Clinical Pharmacology, 11, (Artículo recibido: 6-5-98; aceptado 1-3-99) 91-95. (Article received: 6-5-98; acepted 1-3-99) anales de psicología, 1999, 15(1)