Home Curriculum Vitae Publications Conference Reports Forthcoming Extramural Colloquia Expert Testimony Teaching Healthcare The Human Ecology of Memory Research Archive Publications and Reports Rants

Foxes and Hedgehogs and Hypnosis,

Oh My!


John F. Kihlstrom

University of California, Berkeley



My title refers not just to a line from the Wizard of Oz, but to a famous essay by Isaiah (pronounced eye-ZIE-uh) Berlin (1909-1997),1 which I think of often in connection with hypnosis.

Slide2.JPG (57254 bytes)I am referring, of course, to "The Hedgehog and the Fox" (1953), which in turn takes its title from a fragment of verse by the 7th-century (B.C.) Greek poet Archiolus: "The fox knows many things, but the hedgehog knows one big thing".


At one level, this aphorism reflects the difference between the fox, with lots of resources at its command, and the hedgehog, with a only single, but highly effective, defense against them. But there’s probably more to it than that, as Berlin immediately makes clear in his essay:

Taken figuratively, the words can be made to yield a sense in which they mark one of the deepest differences which divide writers and thinkers, and, it may be, human beings in general. For there exists a great chasm between those, on one side, who relate everything to a single central vision, one system, less or more coherent or articulate, in terms of which they understand, think and feel – a single, universal, organising principle in terms of which alone all that they are and say has significance – and, on the other side, those who pursue many ends, often unrelated and even contradictory, connected, if at all, only in some de facto way, for some psychological or physiological cause, related to no moral or aesthetic principle. These last lead lives, perform acts and entertain ideas that are centrifugal rather than centripetal; their thought is scattered or diffused, moving on many levels, seizing upon the essence of a vast variety of experiences and objects for what they are in themselves, without, consciously or unconsciously, seeking to fit them into, or exclude them from, any one unchanging, all-embracing, sometimes self-contradictory and incomplete, at times fanatical, unitary inner vision (p. 436-437.

Shakespeare was a fox, Berlin asserts, as were Herodotus, Aristotle, Montaigne, Erasmus, Moliere, Goethe, Pushkin, Balzac, and Joyce. On the hedgehog side are Dante, along with Plato, Lucretius, Pascal, Hegel, Dostoevsky, Nietzsche, Ibsen, and Proust. Good company, whichever side you happen to find yourself on. But each side parts company with the other over the matter of whether there is only one truth, in the singular, or many truths, in the plural. For Berlin, the age of Enlightenment, for all its many virtues, is the age of hedgehogs, because of its view that, as he put it (in his essay on "The Divorce between the Sciences and the Humanities", 1974, p. 326), "every genuine question has one true answer…" – true for everyone, in every place, at every time; that the process leading to this truth is publicly accessible; and that genuinely true answers do not contradict each other, resulting in a seamless web which holds everything together. He, himself, preferred a more liberal, pluralistic (and, for that matter, Romantic) stance, which abjures the quest for a grand synthesis and asserts instead that, in the clash of ideas, it does not have to be the case that one idea is true and the other is false.

In some ways, "The Hedgehog and the Fox" is related to Berlin’s other canonical essay, "Two Concepts of Liberty" (1957). In this, his inaugural lecture as professor of political and social theory at Oxford, he drew an important distinction between positive liberty, in which societal restrictions are placed on individual freedom in order to achieve a higher good – a good that is held to be good for everyone, everywhere, at all times -- and negative liberty, in which individuals are permitted to pursue their own individual visions of the good, and to make their own mistakes in the process. In a sense, negative liberty is for foxes, while positive liberty is for hedgehogs.

Positive liberty has its appeal, Berlin argued, as indicated by the attraction of various utopian schemes of social organization; but it also lays the ground for totalitarianism, however benign that totalitarianism may appear to be. Again, in Berlin’s words, this time from "The Pursuit of the Ideal" (1988):

Utopias have their value – nothing so wonderfully expands the imaginative horizons of human potentialities – but as guides to conduct they can prove literally fatal…. The possibility of a final solution -- even if we forget the terrible sense that these words acquired in Hitler’s day – turns out to be an illusion; and a very dangerous one. For if one really believes that such a solution is possible, then surely no cost would be too high to obtain it: to make mankind just and happy and creative and harmonious for ever – what could be too high a price to pay for that? To make such an omelette, there is surely no limit to the number of eggs that should be broken – that was the faith of Lenin, of Trotsky, of Mao, for all I know of Pol Pot (1997, p. 13).

According to Berlin, Tolstoy was a fox who believed in being a hedgehog. And according to some commentators, Berlin himself shared Tolstoy’s problem. He was a fox by nature, but he also clearly felt that the world would be better if everyone were a fox – an intellectual position verging on hedgehogism, and which sometimes made him reluctant to take a stand on the political and social issues of his time, such as labor strife or the Cold War. He understood that the natural sciences are by their very nature monistic, even if the social sciences and humanities must reflect the pluralism of the world that they seek to understand. He fully realized that justice requires a balance between positive and negative liberty.

This tension between pluralism and monism is also characteristic of hypnosis. Hypnosis is a complicated phenomenon: in my view, an altered state of consciousness, involving imaginative experiences associated with subjective conviction bordering on delusion and experienced involuntariness bordering on compulsion, which takes place in the context of a particular social interaction between hypnotist and subject, itself embedded in a wider sociocultural matrix of understanding about mind and behavior. So hypnosis deserves to be approached from diverse points of view. Even so, the pull of monism is extremely strong, and throughout its history hypnosis theory and research has been characterized by controversies between various theoretical points of view, each of which claims to know the "one big thing" about hypnosis, or at least to know the "one big way" to study it.

Slide8.JPG (46356 bytes)We can see this tension at several different levels. For example, J.P. Sutcliffe famously noted how easy it is for investigators to fall into a stance of either credulity or skepticism, and tried to find a third way to approach hypnosis (Sutcliffe, 1960, 1961).


Slide9.JPG (110342 bytes)Later, Ronald Shor picked up the theme, discussing the "clannish loyalties and polemics" that force investigators to choose between the Charybdis of assured optimism and the Scylla of disciplined objectivity -- or was it the other way around? -- and discussed how important it is to achieve a situation where the investigator maintains methodological rigor without losing the positive catalyst which is critical to producing hypnosis in the first place (Shor, 1979). The point is that we wouldn't have these problems if hypnosis were not a multifaceted pheomenon requiring explanation at multiple levels.

Slide10.JPG (85628 bytes)One topic within hypnosis which illustrates the tension between monism and pluralism is the question of the internal structure of the scales developed to measure hypnotizability. Although these scales typically represent individual differences in hypnotizability in terms of a single sum score, factor analyses indicate that they have a somewhat complex internal structure. At the first level, factor analysis typically distinguishes between direct, challenge, and cognitive suggestions. Direct suggestions are for the facilitation of motor response, as when subjects are instructed to hold out their arms and told that a magnetic force is pulling their hands together. Challenge suggestions are for the inhibition of motor response, as when subjects are asked to interlock their fingers and told that they cannot unclasp them. Cognitive suggestions involve alterations in perception and memory, such as hallucinations, posthypnotic suggestion, and posthypnotic amnesia.

Slide11.JPG (66249 bytes)Jack Hilgard documented this structure in separate analyses of the Stanford Form A and Form C, reported in his 1965 monograph, Hypnotic Susceptibility (Hilgard, 1965). 


Slide12.JPG (64907 bytes)More recently, Nick Spanos and his colleagues confirmed this structure in an analysis of the 24 items comprising the Harvard Group Scale and the Stanford Form C (Spanos, D'Eon, Pawlak, & Mah, 1989-1990). Discovery of the multidimensional nature of hypnotizability led to the development of the Stanford Profile Scales of Hypnotic Susceptibility, available in two parallel forms, which permit assessment of subjects' strengths and weaknesses on various subcategories of suggestions falling within the domain of hypnosis.

Although the factorial complexity of hypnosis has been widely accepted among hypnosis researchers, assigning a content interpretation to the factors is made difficult by the fact that the factors also differ in terms of the difficulty levels of their constituent items. Thus, as Hilgard originally noted, direct suggestions are relatively easy, while challenge and cognitive suggestions are considerably more difficult. In fact, Coe and Sarbin argued strongly that the factor structure of hypnotizability was an artifact of item difficulty, and that, appearances to the contrary notwithstanding, a "single role-relevant skill" (p. 1) ran through the matrix of interitem correlations (Coe & Sarbin, 1971). Spanos and his colleagues made a similar argument almost 20 years later (Spanos et al., 1989-1990), and Kirsch and his colleagues have also taken a position along these lines (Kirsch, Silva, Comey, & Reed, 1995). The factorial complexity of hypnotizability is good for the fox, but a single-process interpretation of this apparent complexity is what the hedgehog has in mind.

It has not been easy to separate difficulty level from content, because the difficulty levels of the various scale items are fixed by their standardized scoring procedures. The obvious solution is to adjust the scoring criteria so as to equate the mean difficulty levels of direct, challenge, and cognitive suggestions. However, investigators have been reluctant to take this step because it would effectively destroy the standardized scoring procedure, and compromise the value of the scales as instruments for subject selection.

Slide13.JPG (97139 bytes)However, Tellegen and Atkinson offered a promising solution (Tellegen & Atkinson, 1976). Employing a shortened version of the Harvard Group Scale, these investigators evaluated subjects' hypnotizability in terms of both the standard dichotomous behavioral criteria, and a four-point scale of subjective experience. By adjusting the dichotomous scoring criterion for each item on the subjective scales, they were able to equate the difficulty levels of direct and challenge suggestions.

Slide14.JPG (47620 bytes)Principal factor analysis, with varimax rotation, yielded a two-factor solution. One of these factors was defined by four challenge suggestions, the other by three direct suggestions.  Tellegen and Atkinson concluded that the factor structure of hypnotizability was not an artifact of item difficulty, and that a content interpretation of the factors was appropriate (Tellegen & Atkinson, 1976). Specifically, they proposed that there are several dimensions of hypnotizability, analogous to Thurstone's primary mental abilities. However, these hypnotizability factors are themselves positively intercorrelated, so that assessing hypnotizability in terms of a single score remains justifiable. However, it should be noted that this conclusion rests on an analysis of subjects' self-reports of subjective experience, not their overt behavioral response to test suggestions. 

Slide15.JPG (83970 bytes)More recently, Martha Glisky, Susan McGovern, and I replicated and extended the work of Tellegen and Atkinson, by developing an entirely new scale of hypnotic susceptibility permitting behavioral responses to suggestions to be scored along a continuous, rather than dichotomous, scale. In our study, 100 college undergraduates, none of whom had any previous experience with hypnosis, received an individual administration of a 16-item scale which we called the Arizona Motor Scale of Hypnotizability (AMSH), so named because it was constructed at the University of Arizona and focuses on direct and challenge suggestions within the ideomotor domain.2

The Arizona Motor Scale consists of an induction of hypnosis (by eye closure) taken almost verbatim from the SHSS:C, followed by a series of representative hypnotic suggestions. Most of these suggestions were drawn from the Stanford or Harvard scales; others were devised anew, taking the established suggestions as models. Although the focus of Arizona Motor Scale was on direct and challenge suggestions, three items representing the cognitive factor (mosquito hallucination, posthypnotic suggestion, and posthypnotic amnesia) were also included in the scale, so that scores resembling those of the Harvard Group Scale:A or the Stanford Form A could be derived for purposes of selecting subject for other studies being conducted in the laboratory. Eliminating the first item (Eye Closure), which is administered during the induction procedure, and the three cognitive items, left a pool of six direct suggestions and six challenge suggestions for analysis.

Slide16.JPG (54874 bytes)In contrast to the dichotomous scoring of the Stanford and Harvard scales, the Motor Scale permits subjects' behavioral responses to each direct and challenge suggestion to be scored on a continuous nine-point scale. This was achieved by dividing the text for each suggestion into nine roughly equal segments. For each item, the experimenter recorded the segment in which the subject met an objective behavioral criterion of response. For example, in Item #2, Hand Lowering, a subject whose outstretched hand lowered at least 6 inches during the first segment of text received a score of 8; one whose hand had not fallen that distance, even after a 10-second interval of silence had elapsed at the end of the suggestion, received a score of 0. In Item #3, Arm Immobilization, a subject whose arm lifted more than 2 inches during the first segment of text received a score of 0, while one whose arm had not lifted this amount even after a 20-second silent interval at the end of the suggestion, received a score of 8. The reversed scoring of direct and challenge suggestions is due to differences in the nature of these two types of suggestions: subjects pass direct suggestions by displaying the observed response (e.g., by their outstretched arms falling quickly after the suggestion is given), while challenge suggestions are passed by failing to do so (e.g., by failing to lift the arm even after a long interval has passed).

Slide17.JPG (53723 bytes)Examining the distribution of scores for each direct and challenge suggestion of the AMSH, it was apparent, that, on the whole, the direct suggestions were easier than the challenge ones. For each item, therefore, we tried to select a cutpoint that would yield approximately a 50% rate of passing each suggestion. This wasn’t possible in all cases (some direct suggestions were too easy, while some challenge suggestions were too hard), so we opted instead to adjust scores in such a way as to yoke the pass rate of each direct suggestion to a challenge suggestion.

Slide18.JPG (55848 bytes)In the end, it was possible to establish cutpoints that closely equated the direct and challenge suggestions in terms of both mean pass percent and range. The average dichotomous scores for the six direct suggestions and the six challenge suggestions were essentially identical.

Slide19.JPG (80705 bytes)A principal-components analysis of the six direct and six challenge suggestions, thus rescored dichotomously, yielded two factors. Factor 1 consisted of the six direct suggestions, while Factor 2 consisted of the six challenge suggestions. Thus, at least so far as the distinction between direct and challenge suggestions is concerned, the factor structure of hypnotizability is not an artifact of item difficulty, because it continues to appear even when the two classes of items have been scored in such a way as to equate their difficulty levels. The heterogeneity of hypnosis is real, and it has to be accounted for by our theories. There is something qualitatively different between responding to direct suggestions for the facilitation of motor behavior, such as hand lowering, and challenge suggestions for the inhibition of motor behavior, such as arm immobilization.

Slide20.JPG (73606 bytes)Of course, this conclusion strictly applies only to ideomotor suggestions, as suggestions of a more cognitive type, such as hallucinations and amnesia, were left out of the analysis. However, there is evidence that factorial heterogeneity applies to the cognitive as well as the motor items. The evidence for this comes from the Stanford Profile Scales themselves, which focus on the cognitive items (scores for the ideomotor items, grouped into a single "motor pool", are derived from earlier testing on the Stanford Forms A and C). In Hypnotic Susceptibility, Jack Hilgard reported a somewhat impressionistic analysis of the Profile Scales which yielded a number of different profiles: some subjects were good at positive hallucinations but not negative hallucinations, others the reverse (Hilgard, 1965).

Slide21.JPG (66291 bytes)In 1979, Heather Brenneman, then a graduate student in my laboratory at Harvard, reported a re-analysis of the Profile Scales standardization data, employing more rigorous quantitative techniques of cluster analysis. Her analysis yielded 12 different profiles of hypnotizability, including a number that form contrasting pairs. One group of subjects is particularly good at both negative hallucinations and amnesia and other posthypnotic phenomena, while another group falls down on these same items; another pair of groups diverges on dreams and regressions and the positive hallucinations. But in the present context, what is important is that Brenneman equated the various subscales for difficulty before submitting them to cluster analysis. Thus, the clusters were not artifacts of item difficulty. Instead, they reflected genuine qualitative differences between different types of items, and the underlying abilities they represent.

A single-factor theory can’t explain a multifaceted phenomenon, but single-factor theories are what we mostly get in hypnosis. Partly, I think, this is a legitimate response to the complexity of the phenomenon itself. Hypnosis is complex, and we probably need different theories to account for different aspects of the phenomenon, and different investigators are going to pursue different lines of inquiry. But that’s thinking like a fox. If you’re thinking like a hedgehog, then what you want is one big theory, a universal theory that will encompass everything. And that’s where we get problems: when a theory which offers a perfectly good explanation of one aspect of hypnosis is overgeneralized to hypnosis in its entirety.

Slide22.JPG (87040 bytes)The problem is exacerbated by our embrace, at least at the level of rhetoric, of the logic of falsificationism popularized by Karl Popper (1902-1994). According to this principle, scientists are not supposed to seek evidence that is consistent with their theories, but rather they are supposed to seek evidence that is inconsistent with them. Popper’s reasoning here is of course correct, from a strictly logical point of view, because no amount of positive evidence that a hypothesis is true can guarantee that tomorrow we might not discover an exception that shows it to be false; but in practice hardly anybody actually does this. Rather, what seems to happen is that, when there are two competing theories, proponents of one view attempt to falsify the other, and when that attempt succeeds, claim that their own theory wins by default.

Slide23.JPG (52437 bytes)As an example, almost 20 years ago I reported that subjects experiencing posthypnotic amnesia will be unable to recall a list of words memorized during hypnosis, but did not impair their ability to use these words as free associates and category instances (Kihlstrom, 1980). In fact, the unremembered learning experience caused a priming effect, such that words from the memorized list were more likely to appear as free associates or category instances than they would have been, had the learning experience not occurred at all. The experiment was subsequently cited as a convincing demonstration of the distinction between episodic and semantic memory (Tulving, 1983), and later came to be recognized as one of the earliest demonstrations of a dissociation between explicit and implicit expressions of episodic memory (Schacter, 1987).

In performing the experiment, I have to admit that I didn’t have any theory of hypnosis in mind. In fact, I have never considered myself to be a theorist of hypnosis; my interest in hypnosis -- other than the fact that hypnosis research is intrinsically more interesting for both experimenter and subject than most other psychological research -- has always been for what it can tell us about other things, such as cognition, personality, psychopathology, and consciousness. My purpose was mostly descriptive: following recent advances in memory theory, I was interested in seeing how posthypnotic amnesia would affect different classes of memory tasks. As it happens, in that paper I announced a revision in my previous theory of posthypnotic amnesia, which entailed a disruption in the search process postulated by classical two-stage theories of memory retrieval. Rather, I speculated that posthypnotic amnesia entailed a dissociation between contextual and other features of memory traces, impairing performance on tests which required information about episodic context, but sparing performance on tests which did not. I even used the term "dissociation" – not to describe hypnosis, but rather in a manner which was entirely compatible with the usage of that term as it was evolving within cognitive psychology more generally – as when we would now say that posthypnotic amnesia dissociates explicit from implicit memory, impairing the former but sparing the latter (unfortunately, the explicit-implicit distinction had not yet been introduced in memory theory).

Slide24.JPG (99697 bytes)A couple of years later, Nick Spanos and his colleagues responded with a "re-evaluation" of my claim that posthypnotic amnesia produces impairments on episodic, but not semantic memory tasks; they weren’t interested in spared priming, which was the really interesting discovery in the 1980 paper (Spanos, Radtke, & Dubreuil, 1982). One condition essentially replicated my procedure, with a suggestion that their subjects would be unable to remember the words that they learned while they were hypnotized. In a comparison condition, however, Spanos et al. gave an "alternative" suggestion that subjects would be unable to think of the words in any way.

Slide25.JPG (57389 bytes)When Spanos et al. replicated my procedure, they replicated my results (they even obtained significant priming on the free association test). However, the alternative suggestion abolished the priming effect, and even produced a little negative priming. As a result, they concluded that "there is no intrinsic relationship between posthypnotic amnesia and episodic memory tasks. The dissociation between episodic and semantic memory… is dependent on the task requirements conveyed to subjects…." This conclusion, in turn, was held to be consistent with their view that hypnotic behavior represents a strategic social enactment, rather than any alteration in consciousness.

Here’s an excellent example, I think, of experimenters so bound to their own theoretical point of view, and so eager to discredit the findings of their theoretical competitors, that they miss the point of their own experiment. The simple fact of the matter is that their "alternative suggestion" is not for posthypnotic amnesia at all. Rather, it entails what neurologists and neuropsychologists call agnosia – a failure of semantic memory. In fact, hypnotic agnosia has been observed before – there are items representing it on the Stanford Profile Scales (Hilgard, 1965), and some in this room will remember Fred Evans’ report, in the 1970s, of an experiment in which he suggested to subjects that the number "6" would have no meaning for them (Evans, 1972). The fact that hypnotic suggestions for agnosia can impair performance on tests of semantic and procedural memory is interesting, and it is sad that the effect has gone completely unstudied for almost 20 years. But this experimental result cannot impeach dissociations between episodic and semantic memory, or between explicit and implicit expressions of episodic memory, because it concerns a different kind of suggestion entirely.

This episode from the history of hypnosis research illustrates, I think, the particular virtues of the fox, as opposed to those of the hedgehog. Spanos’s single-minded devotion to a single conceptual framework, by which everything about hypnosis was to be explained, effectively blinded him both to an interesting phenomenon waiting to be (re)discovered and explored, and to the possibility of expanding his framework so that it might do better justice to the phenomenon being explained – in this case, the very complex phenomenon of hypnosis. As I noted earlier, hypnosis involves cognitive changes that take place in a particular interpersonal context. You cannot ignore either of these factors, either the cognitive ones or the social ones, if you wish to understand what hypnosis is all about. Spanos’s insight, that the subject’s interpretation of the hypnotist’s suggestion is critical to how the subject will respond to the suggestion, is completely valid – it is what Martin Orne had in mind when he constantly reminded us that we needed to understand experiments from the subject’s point of view (Orne, 1973). And for this reason, Spanos et al. were quite right to conclude, as they did, that hypnosis "cannot be adequately understood without giving due consideration to subjects’ understandings of the context in which they are tested" (Spanos et al., 1982, p. 572).

But explanation doesn’t end there: we still have to understand the cognitive mechanisms by which that understanding is translated into the subjectively compelling experiences of hypnosis. Once the subject has understood that the suggestion targets words as items in a memorized list, rather than words as entries in the mental lexicon, then other processes take over. These processes are cognitive in nature, and a theory that focuses on interpersonal processes has no conceptual vocabulary to describe them – nor, for that matter, any methodological apparatus to investigate them. If you want to understand posthypnotic amnesia, you’ve got to do research on memory, and that research has to take account of the principles by which memory operates.

Slide28.JPG (59351 bytes)That’s why Fred Evans and I looked at organizational strategies in posthypnotic amnesia (Evans & Kihlstrom, 1973; Kihlstrom & Evans, 1979). At the time we began our work, the dominant theory of memory held that retrieval was facilitated by organizational strategies. Posthypnotic amnesia, being reversible, is a disruption of retrieval, so it made sense to look for evidence of disorganization, and we found it.

Slide29.JPG (65050 bytes)Moreover, a set of studies conducted with Leanne Wilson showed that the disorganization was specific to the temporal relations among items in memory, not their semantic relations (Kihlstrom & Wilson, 1984; Kihlstrom & Wilson, 1988; Wilson & Kihlstrom, 1986). Later, Spanos and his colleagues also studied organization during hypnotic amnesia, and interpreted the effects from a social-psychological point of view (e.g., Spanos & Bodorik, 1977). However, the original hypothesis of disorganized retrieval, and the later hypothesis that the disorganization would be specific to the temporal relations among items, would never have arisen from a social-psychological theory of hypnosis.

Slide30.JPG (48140 bytes)To take another example, there is no way, from a purely self-presentational point of view, to understand the fact that posthypnotic amnesia impairs recognition less than it impairs recall (Kihlstrom, 1985, November). It won’t do to say that this dissociation (because that it what it is) reflects the subject’s understanding of posthypnotic amnesia and the context in which it is tested, because the suggestion is perfectly clear: the subjects are not supposed to remember the things they did or experienced while they were hypnotized, until the hypnotist says the words, ‘Now you can remember everything". Simulators understand this clearly, which is why they perform poorly on both recall and recognition, in contrast to real hypnotic subjects, who recognize more than they recall -- as we have known since the mid-1960s (Barber & Calverley, 1966; Williamsen, Johnson, & Eriksen, 1965), and as Spanos and his colleagues were reminded when they did the comparison (Spanos, James, & Degroot, 1990).

Why, then, do amnesic subjects recognize hypnotic experiences that they cannot recall? The answer, I think, lies in a cognitive theory, proposed by George Mandler, that recognition is a judgment that can be mediated two quite different processes: retrieval, which is closely related to explicit memory, and familiarity, which is closely related to implicit memory (Mandler, 1980). The theory predicts that where amnesia impairs explicit memory but spares implicit memory, subjects can use a priming-based feeling of familiarity to make relatively accurate recognition judgments. And this is exactly what happens in posthypnotic amnesia. Moreover, the theory only makes this prediction in circumstances where subjects are actively trying to remember what happened to them – an attitudinal stance which contradicts the notion that subjects are actively trying to present themselves as amnesic. Moreover, the theory makes this prediction in advance – it does not have to be elaborated post-hoc to accommodate an unexpected observation.

The point of this is not that Spanos was wrong and I am right. The point is that Spanos and I were investigating different aspects of hypnosis: he was interested in how subjects interpret suggestions, I was interested in how the principles of memory mediate subjects’ response to these suggestions. There never was any necessary incompatibility between cognitive and social-psychological approaches to hypnosis, and it’s a mistake to believe there was, or to act as if there were. There’s plenty of hypnosis to go around for everyone, and everyone can make a positive contribution, so long as nobody makes a claim that their theory is universal and sufficient.

History repeats itself, so Hegel argued.3  More recently, Irving Kirsch and Steve Lynn (1995, 1998b) have proposed a new "social-cognitive" theory of hypnosis (Kirsch & Lynn, 1995, 1998a, 1998c), which differs in some respects from the "sociocognitive" account of hypnosis proposed earlier by Spanos (Spanos, 1991). According to the theory, responses to hypnotic suggestions are generated automatically from response expectancies developed in the hypnotic context. In various papers, they have mustered the evidence supporting their view, especially with respect to the correlates of hypnotizability, and applied their theory to the experience of involuntariness that lies at the core of hypnosis. That’s fine so far as it goes, but they went further by attempting to discredit various versions of the "neodissociation" theory of hypnosis proposed by Hilgard, and elaborated in various respects by Ken Bowers and Erik Woody, and by myself (Kirsch & Lynn, 1998b, 1998c). Here again, I think we see the Popperian legacy: if one theory is to be true, all competing theories must be wrong. And here, again, I think that the Popperian legacy is being misapplied. There are problems with neodissociation theory, to be sure, but it is not necessarily the case that any failure of neodissociation theory counts as evidence in favor of the social-cognitive alternative.

In fact, both theoretical approaches have virtues, and both have their problems. Kirsch and Lynn may not care for the hidden observer paradigm on which Hilgard based much of his argument; but, as I pointed out (Kihlstrom, 1998), dissociations are ubiquitous in hypnosis. They occur between explicit and implicit memory, in posthypnotic amnesia and posthypnotic suggestion; and they occur between explicit and implicit perception, in hypnotic analgesia, the sensory anesthesias, and the negative hallucinations. Neodissociation theory doesn’t attempt to explain all of hypnosis: it merely seeks to draw attention to the dissociations which occur within hypnosis, and to attempt to understand them in terms of cognitive structure and process. These phenomena can’t be ignored, and they can’t be explained solely in terms of automatic responses to suggestions or response expectancies. Something has to go in the middle, after formulation of the expectation, and before the response to suggestion occurs. And that is where neodissociation theory focuses its attention – asking questions about cognitive architecture that social-psychological theories set aside. Neodissociation theory may be unable to explain dissociations between explicit and implicit memory, and between explicit and implicit perception, but at least it acknowledges them, and at least it tries.

Slide34.JPG (76310 bytes)On the social-cognitive side, I don’t think that anyone would argue that expectancies are irrelevant to hypnosis, but it remains to be demonstrated whether expectations, taken by themselves, are all that powerful predictors of hypnotic response. More than 15 years ago, Ron Shor and I assessed the impact of expectations on hypnotizability by administering to subjects, prior to their taking the Harvard Group Scale, a questionnaire which described each suggestion on the scale and asked them to predict their response to it (Shor, Pistole, Easton, & Kihlstrom, 1984). The correlation between predicted and actual total scale scores was only r = .34 – higher than the typical correlation between hypnotizability and absorption, to be sure, but still leaving the vast bulk of variance unexplained.

Slide35.JPG (54058 bytes)To my knowledge, the highest correlation yet obtained between naturally occurring (i.e., not experimentally manipulated) expectancies and actual behavioral response, r = .55, was in a study by Council and his colleagues, which measured subjects’ expectancies after they had experienced the response to induction (Council, Kirsch, & Hafner, 1986). I’m sure that the subjects had expectancies before the induction, but I am also certain that these expectancies were revised on the basis of the actual induction experience – by the speed with which eye closure occurred, for example. So, there must be at least part of hypnotic response that is independent of expectations; or, put another way, expectations are at least as much a function of behavior as behavior is a function of expectations.

Slide36.JPG (78829 bytes)It has to be said, too, that the concept of automaticity is highly problematic. In the first place, any notion that response expectations automatically generate responses to hypnotic suggestions has to describe the nature of the automatic processes which intervene between expectancy and experience, and this is the domain of cognitive psychology. So there’s no escaping the details of internal, cognitive mechanisms – the same sorts of internal cognitive mechanisms which apparently worry Kirsch and Lynn when it comes to dissociations (Kirsch & Lynn, 1998a, 1998b). Moreover, cognitive psychology offers a technical definition of automaticity: according to traditional formulations, automatic responses are inevitably evoked by appropriate stimuli, and their execution consumes no attentional resources. But neither of these statements is true of hypnotic responses.

Slide37.JPG (60819 bytes)Consider, as just one example, the prototypical example of hypnotic automaticity: posthypnotic suggestion. Posthypnotic behaviors seem to interrupt the normal flow of the subject’s experience, thought, and action, occurring involuntarily and outside awareness. But appearances are deceiving. In 1987, Spanos and his colleagues showed that posthypnotic responding did not occur outside the experimental setting in which the posthypnotic suggestion was originally administered (Spanos, Menary, Brett, Cross, & Ahmed, 1987). The ability of the cue to evoke the posthypnotic response depends on the context in which the cue is administered – or, put another way, on the subject’s interpretation of the cue. All by itself, this violates a fundamental feature of automaticity, which is that automatic responses are inevitably executed upon presentation of the appropriate stimulus.

Slide38.JPG (53670 bytes)Moreover, in a doctoral dissertation completed in my laboratory, Irene Tobis (nee Hoyt) found that execution of a posthypnotic suggestion interfered with other ongoing activities, and vice-versa, even when there was no conflict between them (Hoyt & Kihlstrom, 1986, August). This finding violates another fundamental feature of automaticity, which is that automatic responses do not interfere with other responses, because their execution consumes no cognitive resources.

Now, it’s true that cognitive psychology offers a revisionist view of automaticity, which loosens the requirements of context and resource independence. But this is a memory-based view of automaticity, and it holds that automatized responses become automatized by virtue of extensive practice: hypnotic subjects just don’t have enough opportunity to practice, between the time the posthypnotic suggestion is administered and the time it is given. So the experience of involuntariness, as powerful as it is for the subject, appears to be illusory – just as neodissociation theory, or at least one form of it, says it is (Kihlstrom, 1992).

Interestingly, another form of neodissociation theory, proposed by Woody and Bowers (Woody & Bowers, 1994), holds that hypnotic involuntariness is by no means illusory, and reflects a disruption of executive control similar to that found in neurological patients with frontal-lobe damage . Moreover, Spanos proposed a version of social-cognitive theory which holds that hypnotic involuntariness is illusory, by virtue of the subject’s misattributions concerning the causes of his or her behavior. So it’s not the case that experimental findings on automaticity necessarily strike blows for one side, and against the other. Spanos could have cited Hoyt’s experiment as evidence in favor of his theory, and against that of Woody and Bowers. By the same token, Hoyt’s finding that posthypnotic response is not automatic, at least in the technical sense of the term, is problematic for both Woody and Bowers, who are dissociation theorists, and for Kirsch and Lynn, who are not. The point is that each theory needs to be tested in its own terms. And when we do this, we will find that each theory corrects itself, just as my findings about spared implicit memory caused me to abandon my original theory that posthypnotic amnesia resulted from a disruption of the search process in memory retrieval (Kihlstrom, 1985).

If it should happen that each theory has a piece of the action, this should not surprise or distress us, because in the final analysis hypnosis is a complex phenomenon, and a variety of different theories can shed light on it. The hypnotic subject responds to suggestions from the hypnotist, and so it is critical to understand how the subject interprets these suggestions, what the subject expects is going to happen, what the subject’s attitudes toward the whole thing are, and the like. But these responses entail profound changes in mental state, and so it is also critical to understand how internal cognitive structures and processes operate to produce hypnotic responses.

There is more than enough hypnosis to go around, and everyone can get a piece of the action. We are more likely to learn about how hypnosis works if we work cooperatively to approach the phenomenon with an attitude of open inquiry, than if we constantly talk past each other in an atmosphere of competitive hypothesis testing.

"The fox knows many things, but the hedgehog knows one big thing." A nice turn of phrase, but as Berlin himself cautioned us:

Of course, like all over-simple classifications of this type, the dichotomy becomes, if pressed, artificial, scholastic and ultimately absurd. But if it is not an aid to serious criticism, neither should it be rejected as being merely superficial or frivolous; like all distinctions which embody any degree of truth, it offers a point of view from which to look and compare, a starting-point for genuine investigation (p. 437).

Slide40.JPG (81775 bytes)Of course, this caution did not prevent Berlin from providing us with a list of hedgehogs and foxes in politics and literature. I have assembled a similar list of hedgehogs and foxes in hypnosis -- but (somewhat after the matter of Pierre Fermat, perhaps) I find that time does not permit me to present this list today. Except for one: a pioneer in the early revival of scientific hypnosis, Robert W. White, who died earlier this year (White, 1941). White was a student of Henry Murray at Harvard, Murray’s successor as director of the Harvard Psychological Clinic, and mentor to Martin Orne. He posed for the original of the "hypnosis card" (14BM) of the TAT, and published several groundbreaking research studies, including an experiment on hypnotic hypermnesia that is still cited today, and an important theoretical paper. White was a fox who understood that hypnosis is a complex phenomenon, involving both cognitive and social processes. There was no doubt in White’s mind that hypnosis was "an altered state of the person". But there was also no doubt in his mind that "the hypnotized person is… a human being who hears and understands and who tries to behave in the different ways which are proposed to him."

In order to achieve an adequate understanding of hypnosis, it is not enough that we know one big thing. We must know many things, either individually or collectively. As hypnosis enters the 21st century, let us give each other the negative freedom to investigate what interests us about hypnosis. In this way, each of us will work out a part of the puzzle, a puzzle that has intrigued psychology from the time of James, and Freud, and Pavlov, to now.


Author Note

Invited address presented at the annual meeting of the American Psychological Association, San Francisco, August 2001. An earlier version was presented at the 50th annual meeting of the Society for Clinical and Experimental Hypnosis, 1999.  Click here to read some prefatory remarks specific to that occasion.  The point of view represented in this paper is based on research supported by Grant #MH-35856 from the National Institute of Mental Health.



1Berlin’s was not exactly a household name, even for an academic philosopher: when, sometime during World War II, Winston Churchill scribbled a memo indicating that he wanted to invite "I. Berlin" to dinner, he found himself sitting next to Irving Berlin instead. (The error wasn’t discovered until Churchill turned to his guest and asked him to name the most important piece he had written lately: the answer was "White Christmas".) But Berlin was an extremely important figure in both philosophy and literature. At his death, one obituary wrote that he knew everybody worth knowing in the 20th century. An argument could be made that the British tradition of ordinary-language philosophy was established during the 1930s in his rooms at All Souls College. He altered our traditional views of the Enlightenment and of Romanticism. When he was knighted, his nomination was not for any particular achievement, but simply "for talking". Interestingly, Berlin left us no "great book" -- just (just!) a mass of essays (the most important collected by Henry Hardy in The Proper Study of Mankind: An Anthology of Essays, 1997; see also Michael Ignatieff’s Isaiah Berlin: A Life, 1998). Two of those essays have entered the canon of English literature, and one of them is especially relevant to us today.  Return to text.

2In reporting some of the results of this research to you today, however, I want to make clear that we developed this new scale for specific research purposes, and we do not intend that it be substituted for the traditional measures like the Harvard and Stanford Scales. These instruments can use some tweaking around the edges, to take account of developments which have occurred in the 40 years since their introduction – in our lab, for example, we have modified their language a little bit, and added scales of subjective success and experienced involuntariness to the Harvard Scale. But still and all, the Stanford and Harvard scales put the study and application of hypnosis on a firm empirical base, and made it possible for one laboratory to attempt to replicate the work of another. This would not be possible if every laboratory used its own idiosyncratic measure of hypnotizability. The Stanford and Harvard scales have become institutionalized, now, and the costs of abandoning them at this point strike me as just too high.  Return to text.

3As paraphrased by Karl Marx, who added added, rather unkindly, "the first time as tragedy, the second time as farce" ("The 18th Brumaire of Louis Bonaparte", 1852).  Return to text.



Barber, T. X., & Calverley, D. S. (1966). Toward a theory of "hypnotic" behavior: Experimental analyses of suggested amnesia. Journal of Abnormal Psychology, 71, 95-107.

Coe, W. C., & Sarbin, T. R. (1971). An alternative interpretation to the multiple composition of hypnotic scales: A single role-relevant skill. Journal of Personality and Social Psychology, 18, 1-8.

Council, J. R., Kirsch, I., & Hafner, L. P. (1986). Expectancy versus absorption in the prediction of hypnotic responding. Journal of Personality and Social Psychology, 50, 182-189.

Evans, F. J. (1972). Posthypnotic amnesia and the temporary disruption of retrieval processes. APA, Honolulu, September.

Evans, F. J., & Kihlstrom, J. F. (1973). Posthypnotic Amnesia as Disrupted Retrieval. Journal of Abnormal Psychology, 82(2), 317-323.

Hilgard, E. R. (1965). Hypnotic susceptibility. New York: Harcourt, Brace, & World.

Hoyt, I. P., & Kihlstrom, J. F. (1986, August). Posthypnotic suggestion and waking instruction. Paper presented at the 94th annual meeting of the American Psychological Association, Washington.

Kihlstrom, J. F. (1980). Posthypnotic amnesia for recently learned material: Interactions with "episodic" and "semantic" memory. Cognitive Psychology, 12(2), 227-251.

Kihlstrom, J. F. (1985). Posthypnotic amnesia and the dissociation of memory. Psychology of Learning and Motivation (G.H. Bower, Ed.), 19, 131-178.

Kihlstrom, J. F. (1985, November). Free recall, cued recall, and recognition in posthypnotic amnesia. Paper presented at the 26th annual meeting of the Psychonomic Society, Boston.

Kihlstrom, J. F. (1992). Dissociation and dissociations: A comment on consciousness and cognition. Consciousness & Cognition, 1, 47-53.

Kihlstrom, J. F. (1998). Dissociations and dissociation theory in hypnosis: Comment on Kirsch and Lynn (1998). Psychological Bulletin, 123(2), 186-191.

Kihlstrom, J. F., & Evans, F. J. (1979). Memory retrieval processes in posthypnotic amnesia. In J. F. Kihlstrom & F. J. Evans (Eds.), Functional disorders of memory (pp. 179-218). Hillsdale, N.J.: Erlbaum.

Kihlstrom, J. F., & Wilson, L. (1984). Temporal organization of recall during posthypnotic amnesia. Journal of Abnormal Psychology, 93(2), 200-208.

Kihlstrom, J. F., & Wilson, L. (1988). Rejoinder to Spanos, Bertrand, and Perlini. Journal of Abnormal Psychology, 97, 381-383.

Kirsch, I., & Lynn, S. J. (1995). Altered state of hypnosis: Changes in the theoretical landscape. American Psychologist, 50(10), 846-858.

Kirsch, I., & Lynn, S. J. (1998a). Dissociating the wheat from the chaff in theories of hypnosis: Reply to Kihlstrom (1998) and Woody and Sadler (1998). Psychological Bulletin, 123(2), 198-202.

Kirsch, I., & Lynn, S. J. (1998b). Dissociation theories of hypnosis. Psychological Bulletin, 123(1), 100-115.

Kirsch, I., & Lynn, S. J. (1998c). Social-cognitive alternatives to dissociation theories of hypnotic involuntariness. Review of General Psychology, 2(1), 66-80.

Kirsch, I., Silva, C. E., Comey, G., & Reed, S. (1995). A spectral analysis of cognitive and personality variables in hypnosis: Empirical disconfirmation of the two-factor model of hypnotic responding. Journal of Personality & Social Psychology, 69, 167-175.

Mandler, G. (1980). Recognizing: The judgment of previous occurrence. Psychological Review, 87, 252-271.

Orne, M. T. (1973). Communication by the total experimental situation: Why it is important, how it is evaluated, and its significance for the ecological validity of findings. In P. Pliner & L. Krames & T. Alloway (Eds.), Communication and affect (pp. 157-191). New York: Academic.

Schacter, D. L. (1987). Implicit memory: History and current status. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 501-518.

Shor, R. E. (1979). The fundamental problem in hypnosis research as viewed from historic perspectives. In E. Fromm & R. E. Shor (Eds.), Hypnosis: Developments in research and new perspectives. New York: Aldine.

Shor, R. E., Pistole, D. D., Easton, R. D., & Kihlstrom, J. F. (1984). Relation of predicted to actual hypnotic responsiveness, with special reference to posthypnotic amnesia. International Journal of Clinical & Experimental Hypnosis, 32(4), 376-387.

Spanos, N. P. (1991). A sociocognitive approach to hypnosis, Theories of hypnosis: Current models and perspectives. (pp. 324-361). New York, NY, US: The Guilford Press.

Spanos, N. P., & Bodorik, H. L. (1977). Suggested amnesia and disorganized recall in hypnotic and task motivated subjects. Journal of Abnormal Psychology, 295-305.

Spanos, N. P., D'Eon, J. L., Pawlak, A. E., & Mah, C. D. (1989-1990). A multivariate study of hypnotic susceptibilty. Imagination, Cognition & Personality, 9(1), 33-48.

Spanos, N. P., James, B., & Degroot, H. P. (1990). Detection of simulated hypnotic amnesia. Journal of Abnormal Psychology, 99, 179-182.

Spanos, N. P., Menary, E., Brett, P. J., Cross, W., & Ahmed, Q. (1987). Failure of posthypnotic responding to occur outside the experimental setting. Journal of Abnormal Psychology, 96, 52-57.

Spanos, N. P., Radtke, H. L., & Dubreuil, D. L. (1982). Episodic and semantic memory in posthypnotic amnesia: A reevaluation. Journal of Personality and Social Psychology, 43, 565-573.

Sutcliffe, J. P. (1960). "Credulous" and "skeptical" views of hypnotic phenomena: A review of certain evidence and methodology. International Journal fo Clinical and Experimental Hypnosis, 8, 73-101.

Sutcliffe, J. P. (1961). "Credulous" and "skeptical" views of hypnotic phenomena: Experiments in esthesia, hallucination, and delusion. Journal of Abnormal & Social Psychology, 62, 189-200.

Tellegen, A., & Atkinson, G. (1976). Complexity and measurement of hypnotic susceptibility: A comment on Coe and Sarbin's alternative interpretation. Journal of Personality and Social Psychology, 33, 142-148.

Tulving, E. (1983). Elements of episodic memory. Oxford, U.K.: Oxford University Press.

White, R. W. (1941). A preface to the theory of hypnotism. Journal of Abnormal & Social Psychology, 36, 477-505.

Williamsen, J. A., Johnson, H. J., & Eriksen, C. W. (1965). Some characteristics of posthypnotic amnesia. Journal of Abnormal Psychology, 70, 123-131.

Wilson, L., & Kihlstrom, J. F. (1986). Subjective and categorical organization of recall during posthypnotic amnesia. Journal of Abnormal Psychology, 95(3), 264-273.

Woody, E. Z., & Bowers, K. S. (1994). A frontal assault on dissociated control, Dissociation: Clinical and theoretical perspectives. (pp. 52-79). New York, NY, US: The Guilford Press.


This page last revised 04/08/10 02:58:37 PM.