ArticlePDF Available

The Altered State of Hypnosis: Changes in the Theoretical Landscape



Presentations of theories of hypnosis in scholarly and introductory texts portray the field as dominated by two warring camps, variously referred to as state and nonstate or as special process and social psychological. Current issues and theories in the hypnosis literature are examined. In the process, we seek to dispel the myth that hypnosis theorists can be grouped into two camps. Although there is considerable controversy about the nature of hypnosis, no issues separate all so-called special process theorists from all social psychological theorists. Instead, virtually all substantive differences between theorists cut across this apparent distinction. Furthermore, the positions taken on many of the important issues dividing the field can no longer be portrayed as simple dichotomies, such as state vs nonstate or trait vs situation. Positions on these issues can more accurately be described as points on a continuum. We conclude by drawing attention to specific questions and issues that remain unresolved. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
The Altered State of Hypnosis
Changes in the
Irving Kirsch
Steven Jay Lynn
University of Connecticut
Ohio University
Presentations of theories of hypnosis in scholarly and in-
troductory texts portray the field as dominated by two
warring camps, variously referred to as state and nonstate
or as special process and social psychological. Current
issues and theories in the hypnosis literature are exam-
In the process, we seek to dispel the myth that hyp-
nosis theorists can be grouped into two camps. Although
there is considerable controversy about the nature of
nosis, no issues separate all so-called special process the-
orists from all social psychological theorists.
tually all substantive differences between theorists cut
across this apparent distinction. Furthermore, the posi-
tions taken on many of the important issues dividing the
field can no longer be portrayed as simple dichotomies,
such as state versus nonstate or trait versus situation. Po-
sitions on these issues can more accurately be described
as points on a continuum. We conclude by drawing at-
tention to specific questions and issues that remain un-
Does hypnosis produce an out-of-the-ordinary, trance-like, "al-
tered" state of consciousness? Psychologists are divided on this
question. Some say yes, others say no. In general, state theories
maintain that hypnosis induces a unique state of conscious-
... In contrast, nonstate theories generally maintain that
hypnosis is not a distinct physiological state. . . . The skeptics
differ in their emphasis, but all agree that the impressive effects
of hypnosis stem from the power of social influence, not from
a trance-like state of altered consciousness. (Kassin, 1995, pp.
his quotation, taken from a current introductory
psychology text, is typical of descriptions of hyp-
nosis theories. Similar presentations can be found
in other texts, and Kassin's (1995) presentation of hyp-
nosis is distinguished only because it is clearer and more
sophisticated than many. Nevertheless, it is wrong. The-
ories of hypnosis can no longer be dichotomized into two
warring camps, and few researchers give the idea of trance
the kind of centrality implied in Kassin's description (cf.
E. R. Hilgard, 1975).
Interest in hypnosis has increased dramatically in
recent decades, moving it into the orbit of mainstream
psychology. Substantial numbers of mental health profes-
sionals use hypnosis regularly (Kraft & Rudolfa, 1982;
Rhue, Lynn, & Kirsch, 1993), and researchers have de-
voted considerable attention to the topic (Lynn & Rhue,
1991a). Clinical research indicates that hypnosis is more
than a faddish addition to the clinician's grab bag of psy-
chotherapeutic tools. Meta-analyses have demonstrated
that the addition of hypnosis to cognitive-behavioral and
psychodynamic treatments substantially enhances their
efficacy (Kirsch, Montgomery, & Sapirstein, 1995; Smith,
& Miller, 1980).
The increasing acceptance of hypnosis as an adjunct
to treatment makes an understanding of the basic phe-
nomena of hypnosis more important than ever. Although
the lay public still holds certain misconceptions about
hypnosis, scientific research and careful clinical obser-
vation over the past few decades have resulted in an ac-
curate understanding and an emerging consensus among
scientists about the basic phenomena of hypnosis. E. R.
Hilgard (1973a) specified a domain of hypnosis that char-
acterizes the type of phenomena that are included within
(e.g., muscular movements, sensory distortions, halluci-
nations, posthypnotic amnesia, and hypnotic dreams) and
lie outside of
faked or shammed responses to sug-
gestions) hypnosis. More recently, the American Psycho-
logical Association's Di /ision of Psychological Hypnosis
(1993) adopted a definition of hypnosis as a procedure
wherein changes in sensations, perceptions, thoughts,
feelings, or behavior are suggested. This definition has
been endorsed by scholars representing a broad range of
theoretical orientations (Chaves, 1994; Fromm, Hilgard,
& Kihlstrom, 1994; see also Kirsch, 1994, for a list of
authors of the definition).
In addition, it is now known that (a) the ability to
experience hypnotic phenomena does not indicate gul-
libility or weakness; (b) hypnosis is not related to sleep;
(c) hypnotic responsiveness depends more on the efforts
and abilities of the person hypnotized than on the skill
of the hypnotist; (d) participants retain the ability to con-
trol their behavior during hypnosis, and they are aware
of their surroundings and can monitor events outside of
the framework of suggestions during hypnosis; (e) spon-
Edilor's note. Ernest R. Hilgard served as action editor for this article.
Author's note. We thank Mark Alicke, Robert Nadon, Thurman Mott,
Auke Tellegen, Steven Jones, Jonathan Neufeld, and Judith Pintar for
their helpful comments on the manuscript.
Correspondence concerning this article should be addressed to Irving
Kirsch, Department of Psychology, U-20, University of Connecticut,
406 Babbidge Road, Storrs, CT 06269-1020; or to Steven Jay Lynn,
Department of Psychology, Ohio University, Athens, OH
October 1995 American Psychologist
Copyright 1995 by the American Psychological Association. Inc. OOO3-O66X/95/$2.OO
Vol. 50. No. 10. 846-858
Irving Kirsch
taneous posthypnotic amnesia is relatively rare; (f) sug-
gestions can be responded to with or without hypnosis,
and the function of an hypnotic induction is merely to
increase suggestibility to a minor degree; (g) hypnosis is
not a dangerous procedure when practiced by qualified
clinicians and researchers; (h) most hypnotized persons
are neither faking nor merely complying with suggestions;
(i) hypnosis does not increase the accuracy of memory;
and (j) hypnosis does not foster a literal re-experiencing
of childhood events (E. R. Hilgard, 1965; Kirsch, Silva,
Carone, Johnston, & Simon, 1989; Rhue et al., 1993;
Nash, 1987).
When the state of hypnosis theory was last reviewed
in this journal (Spanos & Barber, 1974), a theoretical
consensus was emerging based on the idea that focused
attention, absorption, and involved imagining were cen-
tral to hypnotic responding (E. R. Hilgard, 1975; J. R.
Hilgard, 1970; Spiegel & Spiegel, 1978; Tellegen & At-
kinson, 1974). Unfortunately, the emergent consensus was
soon shattered by data indicating that participants could
respond to suggestion while imagining conflicting events
(Bartis & Zamansky, 1990; Kirsch, Council, & Mobayed,
Spanos, Weekes, & DeGroh, 1984; Zamansky,
Zamansky & Clark, 1986). Furthermore, although
it has long been widely acknowledged that imagination
plays a role in hypnotic responding, the relation between
measures of imaginative traits (e.g., absorption) and hyp-
notic responsiveness is quite small (reviewed in Kirsch &
Council, 1992). A vigorous debate resumed as these data
led theorists to focus on other hypothesized processes,
such as dissociation (E. R. Hilgard, 1986), response ex-
pectancy (Kirsch, 1990), and tacit interpretations of
gestions (Spanos, 1991).
The perception that theories of hypnosis can be
grouped into two opposing factions was once accurate.
During the 1960s and 1970s, these factions were referred
to as the state camp and the nonstate camp. In the 1980s,
the divisions remained the same, but the labels changed
from state and nonstate to special process and social psy-
chological. However, hypnosis researchers have changed
their theories considerably (e.g., E. R. Hilgard, 1986;
Spanos, 1991), and new theories of hypnosis have been
proposed by a new generation of scholars (see Lynn &
As a result, what was once real has become
a myth. The notion of warring camps is now outdated.
In this article we elaborate on these themes in re-
lation to a number of current, empirically-based theories
of hypnosis. In the process, we seek to dispel the myth
of the warring camps. We do not mean to imply that a
consensus about the nature of hypnosis has been achieved.
To the contrary, disparate theories continue to vie for
empirical support (see Lynn & Rhue, 1991a), and the-
orists remain as contentious as ever. Nevertheless, there
are no issues separating all so-called special process the-
orists from all social psychological theorists. Instead, vir-
tually all substantive differences between theorists cut
across this apparent distinction. Furthermore, the posi-
tions taken on many of the important issues dividing the
field can no longer be portrayed as simple dichotomies,
such as state versus nonstate or trait versus context. Po-
sitions on these issues can more accurately be described
as points on a continuum.
The Fate of the State Debate
Thirty years
most hypnosis researchers believed "that
there is an hypnotic state which fundamentally differs
from the waking state" (Bowers, 1966, p. 277). This hy-
pothesized state was assumed to be the "essence" of
1959) and, in fact, its definition. It was almost
universally agreed that hypersuggestibility was a charac-
teristic of the hypnotic state and that responses to sug-
gestions were facilitated by other hypothesized charac-
teristics of the trance state.
The state debate began when Sarbin (1950) and
T. X. Barber (1969) rejected the idea that hypnotic re-
sponses were due to an altered state of consciousness.
Barber's operational approach to hypnosis was derived
from logical positivism and neobehaviorism, whereas
Sarbin's position was based on social psychological role
theory. Nevertheless, Barber and Sarbin both maintained
that all of the phenomena of hypnosis, including behav-
ioral responses to suggestion, subjective responses to sug-
gestion, and even the subjective experience of a trance
could be accounted for without postulating any
special state or condition. From their perspective, hyp-
notic behaviors are like other complex social behaviors:
They are a product of such factors as ability, attitude,
expectancy, attribution, and interpretation of the
situation. During the 1960s and 1970s, the altered state
issue was acknowledged to be the most contentious issue
in the field (Sheehan & Perry, 1976).
The dissolution of the camps began with E. R. Hil-
gard's (1969) suggestion that the term state could be in-
terpreted as a descriptive classification rather than as an
October 1995 American Psychologist
Steven Jay
explanatory construct. As described by Kihlstrom (1985,
405), "the term 'state' is construed only as a kind of
shorthand, with no causal properties or denning features
associated with it." Similarly, according to Perry (1992),
in applying the term altered state of consciousness, "all
that is meant is that in hypnosis, many individuals report
experiencing subjective alterations" (p. 240). However,
the fact that subjective alterations like relaxation and in-
voluntariness are reported during hypnosis has always
been acknowledged by nonstate theorists (e.g., T. X. Bar-
ber, 1969) who are quick to add that these alterations are
also reported by unhypnotized control participants.
Kihlstrom noted correctly that if the term state is con-
strued in this weak metaphorical sense, then "the question
of whether hypnosis is a special state of consciousness
disappears as a substantive issue" (p. 405). It becomes a
hair-splitting question of preferences in terminology.
Former state theorists retained the term altered state as
a descriptive label for the various behaviors and self-re-
ported experiences that were observed following a hyp-
notic induction. Nonstate theorists rejected the term as
misleading, fearing that most readers would think it re-
ferred to the original definition mentioned earlier.
But the state debate did not disappear. Many people
in the field of hypnosis did not accept the weakened def-
inition of the term state. Instead, they retained the tra-
ditional notion of trance as an altered state that is distinct
from the ordinary waking state and that is instrumental
in producing hypnotic responses. Tart (1983) was es-
pecially critical of weakened versions of the state con-
struct. He defined altered states as denoting "major al-
terations in both the content and pattern of functioning
of consciousness" and cautioned, "The major pattern
connoted by 'state' should not be trivialized by using the
word 'state' to refer to any change in condition" (p. 19).
Although the state debate has not disappeared, it
has altered considerably. Instead of two opposing camps,
there is now a continuum of positions on the issue. At
one end of this continuum are scholars who espouse the
concept of hypnotic state in its strongest possible form,
as a condition that is fundamentally different from normal
waking consciousness and from other altered states, such
as daydreaming and relaxation. These include most of
the followers of the late Milton Erickson and contem-
porary psychoanalytic theorists. According to Erickson
(1941/1980), literalness (e.g., answering "no" to the
question "Do you mind telling me your name?") and
spontaneous (i.e., unsuggested) catalepsy and amnesia are
behavioral markers of the hypnotic trance state. Psycho-
analytic theorists regard easier access to the unconscious,
a shift toward primary-process thinking, greater ego re-
ceptivity, and regression of the ego as defining character-
istics of the trance state (Fromm, 1992; Nash, 1991). The
newest addition to unambiguous state theories is the the-
ory of dissociated control, in which responses to sugges-
tions are due to a state that is functionally equivalent to
one produced by frontal lobe lesions (Woody & Bowers,
At the other end of the continuum are theorists who
use the term state to describe hypnotic phenomena but
deny that it explains or causes those phenomena in any
way (E. R. Hilgard, 1969; Kihlstrom, 1985); those who
acknowledge allegiance to the state construct but then
ignore it entirely in their theories of hypnotic responding
(McConkey, 1991; Sheehan, 1991); and those who ex-
plicitly reject the hypnotic state construct as inaccurate
and misleading (Coe & Sarbin,
Dixon & Laurence,
Kirsch, 1991; Kirsch & Council, 1989; Lynn &
Rhue, 1991b; Nadon, Laurence, & Perry, 1991; Spanos,
1991).' Surprisingly, this latter group now includes
some prominent Ericksonian clinicians (Zeig & Rennick,
1991) who maintain that the concept of trance has little
explanatory value and claim that it only distracts from
their preferred emphasis on hypnosis as an interpersonal
Between the two extreme positions on the state issue,
there is a concept of trance as an altered state, one that
is not unique to hypnosis. For example, Cheek and
LeCron (1968) wrote
At times everyone daydreams. This definitely is a trance state
. . . Whenever we become absorbed in what we are doing, we
slip into hypnosis. . . . Any strong emotion may produce hyp-
A state of shock certainly is similar to a trance state,
In a similar vein, Walters and Havens (1993) noted,
These latter theorists do not reject the idea that there are altered
states of consciousness, such as those produced by various drugs. Nor
do they reject the hypothesis that hypnotic behaviors may reflect im-
portant changes in subjective experience. What they do reject is the idea
that those experiences can be explained as a consequence of an hypnotic
state that is fundamentally different from normal waking consciousness.
Because this concept of hypnosis as a fundamentally different state is
so prevalent among both professionals and laypeople, they also deem
state terminology misleading, even when used in a purely descriptive
sense (Kirsch, 1992).
October 1995 American Psychologist
The term trance refers to the daydreamy state of mind often
experienced by people when they meditate, listen to music, at-
tend a long lecture, drive on a freeway at night, undergo a hyp-
notic induction, or wait patiently in a therapist's waiting room
. .
Whether elicited by a hypnotist or induced by a
a trance is a trance, (pp. 23-24)
The problem with these definitions of trance is that they
are too inclusive, loose, and imprecise to be subjected to
empirical study. Without some means of determining
whether a person is in trance, it is impossible to test any
hypotheses about the effects of trance.
The hypnotic state construct proposed by Spiegel
and Spiegel (1978) is similarly inclusive, embracing such
naturally occurring phenomena as daydreaming, intense
concentration, absorption, infatuation, placebo effects,
and dissociative disorders. Some precision is added,
however, by the identification of intense focal concen-
tration with diminished peripheral awareness as the es-
sential feature of hypnosis (cf. Shor, 1962). According
to Spiegel and Spiegel, the trance state "necessitates the
elimination of distracting or irrelevant stimuli. . .We
not only pay attention to our given task, we also ward
off distractions" (p. 22). Except for the use of the term
trance state, this view is identical to that of a perspective
temporarily adopted by many theorists (see Spanos &
Barber, 1974) but abandoned in the light of disconfirm-
ing evidence.
Edmonston (1991) has provided a very precise def-
inition of the hypnotic trance state as a state that is iden-
tical to a particular nonhypnotic state. Although he has
amassed a great deal of evidence to support the hypothesis
that hypnosis and relaxation are functionally equivalent,
his hypothesis is disconfrmed by the finding that various
"alert" inductions, including those in which relaxation
is inhibited by vigorous exercise, are as effective as tra-
ditional relaxation inductions for enhancing suggestibility
(Banyai, 1991).
In summary, there are many different positions on
the altered state hypothesis. Some theorists reject the
concept outright. Others accept it nominally. A third
group characterizes a number of naturally occurring
phenomena that happen to everyone on a daily basis as
hypnosis. Still others regard these naturally occurring
phenomena as part of a continuum leading to, but also
different from, hypnosis. Contrary to premature obitu-
the altered state controversy is alive and well. But
it does not divide the hypnosis community into two op-
posing camps. Although some theorists still endorse the
concept of trance as an explanatory mechanism (see
Kirsch, 1993), the positions taken by most researchers in
what used to be the two camps are now very similar (cf.
E. R. Hilgard, 1987; Kihlstrom, 1985; Kirsch, 1992;
Perry, 1992).
There are two sets of data that have led to a con-
vergence of opinion on the state issue among most re-
searchers. The first is the very modest effect of hypnotic
induction on suggestibility. Most people are almost as
responsive to so-called "waking suggestions" as they are
to the same suggestions given in a hypnotic context
R. Hilgard, 1965). The second is the consistent failure
to find any reliable markers of the hypothesized state.
Erickson's (1941 /1980) hypothesized behavioral markers
(literalism, catalepsy, and amnesia) have either failed to
distinguish hypnotized from nonhypnotized participants
(Green et al., 1990) or have been shown to be products
of participants' perceptions of the hypnotic role (Orne,
Young & Cooper, 1972). Similarly, no physiological
or even self-report markers of a hypnotic trance have been
found (Dixon & Laurence, 1992; Kirsch, Mobayed,
Council, & Kenny, 1992).
On the basis of the conceptual changes described in
this section, Kassin's (1995) characterization of the altered
state issue might be restated as follows: Does hypnosis
produce an out-of-the-ordinary, trancelike, "altered" state
of consciousness? This was once a central question that
divided hypnosis theorists into two opposing camps.
However, researchers have established that one's response
to hypnosis depends more on the abilities, beliefs, and
interpretations of the person hypnotized than on the use
of a hypnotic induction (E. R. Hilgard, 1965; Kirsch,
Spanos, 1991). Hypnotic inductions increase sug-
gestibility, but for most people the increase is small. A
person who responds to 6 out of 12 suggestions without
an induction (12 is the number of suggestions on the most
frequently used scales of hypnotic susceptibility) might
respond to 7 after an induction. Furthermore, similar
increases in response may be produced by nonhypnotic
procedures (T. X. Barber, 1969). As a result, when the
term hypnotic state is used by researchers today, it is usu-
ally in a descriptive sense to denote the subjective changes
that hypnotized persons report experiencing. It is not used
to explain those changes. Instead, most hypnosis re-
searchers agree that the impressive effects of hypnosis stem
from social influence and personal abilities, not from a
trancelike state of altered consciousness.
The Trait Debate That Was Not There
There is ample support for the hypothesis that hypnotic
responsiveness is a traitlike, aptitudinal capacity of the
person: Different measures of hypnotizability are mod-
erately to highly intercorrelated, typically in excess of .60
(Bowers, 1983), and a test-retest correlation of .71 has
been reported following a retest interval of
years (Pic-
cione, Hilgard, & Zimbardo, 1989). Despite these data,
the hypothesis that hypnotizability is a stable trait remains
We are continually amazed at the number of students
who come away from their initial immersion in the hyp-
nosis literature thinking that there is a debate between
trait theorists and state theorists. As people who are better
acquainted with the literature know, the trait hypothesis
and the state hypothesis are not incompatible, and many
scholars subscribe to both. Equally mistaken is the iden-
tification of the trait hypothesis with the so-called state
or special process camp. In fact, the two issues are or-
thogonal, and a person's position on one of them does
October 1995 American Psychologist
not predict his or her position on the other. For example,
many Ericksonians who believe that hypnosis is an altered
state reject the idea that hypnotizability is a stable trait.
J. Barber (1991), for example, maintained that the hyp-
notic capacity of supposedly low individuals could be un-
locked by tailoring hypnotic procedures to the individual's
unique personality and by counteracting inhibitions
against experiencing hypnosis.
The idea that the trait hypothesis is somehow anti-
thetical to nonstate theories is also an error, as shown by
the following historical sketch. In his initial formulation
of a social psychological model of
Sarbin (1950)
hypothesized that individual differences in hypnotic re-
sponding were due to differences in the talent or capacity
for role involvement. Sarbin and Coe (1972) conceptu-
alized the tendency to become involved in imagining as
a skill that people possess to different degrees. Two years
later, T. X. Barber, Spanos, & Chaves (1974) embraced
the constructs of absorption (Tellegen & Atkinson, 1974)
and imaginative involvement (J. R. Hilgard, 1970) as ex-
planations of differences in responsiveness. From these
constructs, Wilson and T. X. Barber
1983) devel-
oped the virtually identical construct of fantasy proneness,
further researched by Lynn and Rhue (1988).
There are at least three positions that can be iden-
tified with respect to the trait issue. One is that hypnotic
responding does not require any particular skills or abil-
and that when the proper procedures are used, al-
most anyone can become highly responsive to suggestion.
This is the basis for the emphasis on complex, individ-
ualized inductions and suggestions that are taught in ad-
vanced Ericksonian workshops.
A second position is that hypnotic responding re-
quires at least some stable imaginative propensities or
other cognitive abilities but that those skills are not suf-
ficient for the production of suggested responses. In ad-
dition, the person must interpret suggestions correctly
(Gorassini & Spanos, 1986) or be sufficiently convinced
that they are capable of responding (Wickless & Kirsch,
Data supporting this position (see Spanos, 1991,
for a review) indicate that most, but not all, participants
can be taught to respond as untrained high hypnotizables.
The minority who remain unresponsive are thought to
be unmotivated or to lack the requisite basic abilities
(Spanos, 1991). Most hypnotizability training programs
have been developed from a cognitive-behavioral per-
spective, but at least one is rooted in neodissociation the-
ory (Barabasz & Barabasz, 1989).
Finally, there are those who believe that responses
to the most difficult suggestions, such as suggestions for
amnesia and for hallucinations, require a relatively rare
aptitude that cannot be taught. From this perspective, the
effect of training is to increase compliance, rather than
enhance suggestibility (Bates, 1990).
Similar to the state issue, the trait issue can most
accurately be portrayed as a continuum, rather than a
dichotomy. Compared with the state issue, however, it is
a somewhat truncated continuum. Most scholars recog-
nize a role for stable individual differences and a role for
contextual variables in determining hypnotic response.
Furthermore, most stress the importance of considering
the interaction between these variables (Dixon & Laur-
1992, p. 57; Lynn & Rhue, 1991; Nadon et al.,
Spanos, 1991, pp. 329-330). Where they differ is
on the relative emphasis that should be placed on these
two types of variables.
Dissociation and the Special Process
Shortly after rejecting trance as an explanatory construct,
E. R. Hilgard (1973b) proposed his influential neodis-
sociation theory. It was then that state and nonstate were
abandoned as labels for the opposing camps and were
replaced by the labels special process and social psycho-
logical. The term special process was coined by Spanos
(1982) as a label for trance theories and neodissociation
theories. What these two types of theories share, according
to Spanos, is the hypothesis that hypnotic behavior is
qualitatively different from nonhypnotic behavior, either
because it is produced by a trance state or because it is
produced by dissociation. In contrast to special process
theories, cognitive-behavioral theories explain hypnotic
behavior in terms of the same variables that account for
nonhypnotic behavior.
It is amazing that this distinction has been accepted
so widely. According to neodissociation theory, dissocia-
tions include such normal, mundane phenomena of ev-
eryday life as dichotic listening, distraction, parapraxes,
involved imagining, and dreaming. In other words, there
is nothing very special about dissociation. "Daily life,"
wrote E. R. Hilgard (1973b, p. 406), "is full of many
small dissociations." Conversely, almost all socio-cogni-
tive theorists in the field recognize that hypnosis is special
enough to consitute an important field of study (see,
for an important exception). The
kinds of suggestions made during hypnosis can produce
profound changes in experience, and accumulating data
indicate that hypnosis can enhance the effects of psycho-
therapy substantially (Kirsch et al., 1995).
E. R. Hilgard's (1994) neodissociation theory and
the socio-cognitive theory of Spanos (1991) are generally
considered rival theories. Although each is portrayed by
its detractors in simple terms, both of them are, in fact,
quite intricate, and despite our immersion in the hypnosis
literature for many years, we continue to find new sub-
tleties in them. Most recently, we have enjoyed discovering
their hidden similarities.
Neodissociation Account of "Nonvolitional"
E. R. Hilgard's (1977, 1986, 1994) neodissociation model
is based on three assumptions: (a) There is a central con-
trol system—"executive ego"—that performs planning
and monitoring functions; (b) beneath the central control
system there are relatively automonous subordinate cog-
October 1995 American Psychologist
nitive-behavioral systems; and (c) the subsystems of con-
trol are arranged hierarchically.
Dissociation is denned by E.
Hilgard (1977,
2) as a division of consciousness, in which attentive effort
and planning are carried out without awareness. Hilgard
believed that hypnotic phenomena could be viewed
dissociative in nature whereby hypnotic suggestions were
hypothesized to result in a dissociation between executive
and monitoring functions that were otherwise integrated.
Hypnotic suggestions were thought to modify the normal
hierarchical relationships among various systems and
subsystems of central cognitive control, resulting in func-
tional changes. Although most executive functions recede
during hypnosis, the monitoring and observing functions
remain active but lack normal criticality. This dissociation
in cognitive functioning, in conjunction with the existence
of amnesia
an amnesia-like barrier between the dis-
sociated parts, explains the experience of involuntariness
or automaticity that accompanies many hypnotic
The neodissociation model of how ideomotor and
challenge responses are produced by participants is por-
Figure 1, which is an amalgam of two figures
used by E. R. Hilgard (1986) to illustrate neodissociation
theory. As shown, hypnotic suggestions are hypothesized
to act upon the central control structure (or executive
to create
communication barrier that
segment of itself from conscious awareness.
A control substructure for some ordinary movement, such
as lifting or bending an arm, is depicted below the central
control structure. Because the hypnotist cannot directly
participants' actions,
central control
executive ego must move
movement of the arm. Thus, the movement or its inhi-
is an
intentional act (cf. Kihlstrom, 1992). The
participant experiences
it as
involuntary, because
it is
controled by that portion of the ego that has been sepa-
rated from awareness.
The Socio-Cognitive Account of "Nonvolitional"
Sarbin (1989) and Spanos (1986) also considered hypnotic
responses to be intentional acts rather than involuntary
happenings. This continues to mislead even the most so-
phisticated readers into thinking that hypnotic responses
are viewed as simple deception. But this has never been
their position. Instead, Sarbin and Spanos have consis-
tently maintained that at least some participants delude
themselves into believing that their intentional acts are
We once found this difficult to understand. How can
people delude themselves into believing that their own
intentional acts are involuntary? Do they enact the re-
sponse and then promptly forget that they have done so?
These apparent multiple spontaneous occurrences
permanent amnesia seemed very unlikely until we real-
ized that we made
mistake to assume that participants
were aware of the intentionality of their acts, at least while
they were performing them.
fact, Spanos (1991) has
maintained that at least some of the intentions of hypnotic
participants are "tacit." In other words, participants do
not forget their intentions; they were never aware of them
in the first place.
The Neodissociation Account of "Nonvolitional" Responding
Central Control Structure
"Executive Ego"
•Communication Barrierl
Available to Unavailable to
Consciousness I Consciousness
October 1995
American Psychologist
is reasonable to ask how an intention
can be unconscious. As Bowers (1992) suggested, an "un-
conscious intention" appears to be an oxymoron. As
turns out, there is one other theory of hypnosis that rests
on the idea of unconscious intentions: E.
(1986) neodissociation theory. The voluntary nature of
dissociative responses was most clearly articulated
Kihlstrom (1992), who cited Shor's (1979) explanation
Although the hypnotic subject may look as if
is no longer in
control of his own volitional activities—for example, he may
behave as if
is unable to bend his hypnotically stiffened el-
only because at some deeper
within the boundaries of
he is actively, deliber-
ately, voluntarily keeping his elbow stiff while simultaneously
orchestrating for himself
illusion that he is really trying his
best to bend it. (p. 124)
The socio-cognitive account of ideomotor and challenge
responses is displayed in Figure 2. Although Spanos did
not use terms like "executive ego," "central control
structure," or "communication barrier," his model of the
production of ideomotor and challenge responses is re-
markably similar to E. R. Hilgard's. In both theories, the
participant intentionally initiates or inhibits the response,
but somehow manages
keep that intention out
There are, of course, some differences between the
socio-cognitive and neodissociation explanations of
motor and challenge suggestions. According to E. R. Hil-
gard (1986), the ability to intentionally initiate an action
without awareness
division of consciousness into two parts. Although the
"hidden" part of consciousness is not available to the rest
of consciousness, it is nevertheless a part of consciousness.
In contrast, the socio-cognitive model does not require a
division of consciousness into two parts. Instead, it relies
on the hypothesis, most clearly articulated in a now classic
article by Nisbett and Wilson (1977), that people may
not be introspectively aware of the cognitive processes
that mediate the effect of stimuli on behavior. According
to their article,
When people attempt to report on their cognitive processes,
that is, on the processes mediating the effects of a stimulus
response, they do not do so on the basis of any true
introspection. Instead, their reports are based on
implicit causal theories, or judgments about the extent
particular stimulus is
plausible cause of a given
response. This suggests that though people may not be able
to observe directly their cognitive processes, they will some-
times be able to report accurately about them. Accurate re-
ports will occur when influential stimuli are salient and are
plausible causes of the responses they produce, and will not
occur when stimuli are not salient or are not plausible causes.
(Nisbett & Wilson, 1977, p. 231)
Extending this to hypnotic responses, reports of involun-
tariness would be based on such salient stimulus factors
as the passive wording of the suggestion and the perceived
the hypnotic participant (Kirsch, 1985; Lynn,
Rhue, & Weekes, 1990; Spanos, 1986).
Note that there is no dispute about whether partic-
ipants' reports of involuntariness are accurate. Neodis-
sociation and socio-cognitive theorists agree that the re-
ports are not accurate. The responses are in fact inten-
tional, and participants are mistaken when they conclude
otherwise. The difference between neodissociation and
socio-cognitive models of ideomotor responding concerns
the cause of this misattribution that hypnotic participants
This distinction may seem subtle, and one is justified
in asking whether it is meaningful. The question is, does
it have any empirical consequences? We think
E. R. Hilgard (1973b) introduced the concept of the hid-
The Socio-Cognitive Account of "Nonvolitional" Responding
Executive Functions
October 1995
American Psychologist
Nonvolitonal Responding as Described in Dissociated Control Theory
Central Control Structure
den observer
describe the phenomenon by which
person registers and stores information in memory, with-
out the person being aware that the information has been
processed. In
typical study, high hypnotizable partici-
pants are able to recover concealed experiences or mem-
ories during hypnotic suggestions when they are informed
that they possess
hidden part that can be contacted by
the hypnotist with
prearranged cue.
E. R. Hilgard (1973b) described one of his partici-
pant's behavior in response to so-called "hidden observer"
suggestions as follows:
All the while that she was insisting verbally that she felt no pain
in hypnotic analgesia, the dissociated part of herself was re-
porting through automatic writing that she felt the pain just as
in the normal nonhypnotic state, (p. 398)
In Hilgard's model, the hidden part of consciousness is
there all along, experiencing pain, for example, despite
analgesia suggestions. It is for this reason that the hidden
part is open to communication through hidden observer
suggestions. The function of hidden observer instructions
is to open this normally hidden part of consciousness to
communication. With respect to ideomotor suggestions,
the hidden part
consciousness knowingly initiates
movements in response to ideomotor suggestions. To our
knowledge, hidden observers have not been asked to re-
port on their experience of ideomotor responses, but neo-
dissociation theory should predict that they would report
their movements as volitional.
In socio-cognitive models, executive functions that
operate outside of awareness cannot be communicated
with, because they do not constitute
secondary con-
In fact, Spanos and Hewitt (1980) have argued
that the hidden observer is an "experimental creation".
In a series of studies on the hidden observer phenomena,
Spanos and his associates (Spanos, Gwynn, & Stam, 1983;
Hewitt, 1980; Spanos, Radtke,
1984) have shown that hidden observer reports are as-
sociated with the nature and explicitness of the instruc-
tions that participants receive about the hidden observer.
Hence, according
Spanos, socio-cognitive variables
(e.g., beliefs, expectations, or interpretations of instruc-
rather than dissociated executive functions, are
the determinants of hidden observer responding.
Theory of Dissociated Control
Bowers and his colleagues (Bowers, 1992; Woody & Bow-
1994) have recently advanced a revised neodissocia-
tion theory that they refer to as
theory of dissociated
control. Bowers's account of ideomotor responding is de-
Figure 3. Whereas E.
Hilgard (1994) and
Spanos (1991) used different terminology to state similar
things, E. R. Hilgard (1994) and Bowers (1992) used sim-
ilar terminology but espoused substantially different the-
Bowers relocated the influence
E. R. Hilgard's model, the influence of the hypnotist's
is on
the executive ego,
central control
structure (see Figure 1). In contrast, Bowers hypothesized
that the hypnotist's words directly activate subsystems of
control. The induction of hypnosis modifies the central
control structure, not by creating a division of conscious-
ness but by weakening executive control over the subsys-
It is this weakening of executive controls that allows
the subsystems to be invoked directly by suggestion.
Bowers' (1992) new theory makes the idea of two
opposing camps more outmoded than ever. In some re-
spects, neodissociation theory and socio-cognitive theory
are closer to each other than either of them are to Bowers'
theory of dissociated control. For example, according to
both E. R. Hilgard (1994) and Spanos (1991), the hyp-
notic response is an intentional act, although the partic-
ipant may not be aware of this. However, according to
Bowers, the response is truly involuntary: It is produced
See also Kirsch and Lynn's (1995) recent elaboration of the socio-
cognitive model.
October 1995
American Psychologist
by the hypnotist's suggestion and is not cognitively me-
diated. Bowers' new theory also differs from both neo-
dissociation and socio-cognitive theory with respect to
the altered state issue. The theory of dissociated control
implies a strong version of the altered state hypothesis.
Hypnosis is hypothesized to be a state wherein executive
control over subsystems are weakened, thus permitting
their direct activation. In fact, hypnotized participants
are hypothesized to be similar to patients with frontal
lobe disorders (Woody & Bowers, 1994). In contrast, Spa-
nos (1982, 1991) has consistently rejected the altered state
hypothesis, and E. R. Hilgard
1973a) has denied
any causal relation between hypnotic state and response
to suggestion. On the other hand, both Bowers and Spanos
have rejected the idea of a division of the ego into two
components separated by a communication barrier, and
neither have put much stock in the so-called hidden ob-
server. So in some ways, the theory of dissociated control
resembles socio-cognitive theory more than neodissocia-
tion theory.
There is, however, one dimension along which socio-
cognitive theories can be distinguished from alternative
paradigms. These theories are particularly parsimonious,
requiring neither altered state nor dissociation as explan-
atory constructs.
Current Questions and Controversies
Although hypnosis theories can no longer be reduced into
two opposing camps, theoretical consensus is far from
achieved. Competing accounts of reported involuntari-
ness are still evolving (Kirsch & Lynn, 1995; Woody &
Bowers, 1994), and the data are insufficient for a definitive
resolution. In addition, the following questions about
hypnosis remain unresolved and continue to drive theo-
retically relevant research:
there a uniquely hypnotic state that serves
as a background or gives rise to the offered sub-
jective experiences produced by suggestion?
Having failed to find reliable markers of trance after 50
years of careful research, most researchers have concluded
that this hypothesis has outlived its usefulness. Neverthe-
most clinicians still believe that hypnosis is a unique
state with causal properities (see Kirsch, 1993), as do some
influential theorists (e.g., Fromm, 1992; Nash, 1991;
Spiegel & Spiegel,
Woody & Bowers, 1994). Because
the state hypothesis is an existential rather than a universal
proposition, it cannot be disproved, but can, in principle,
be verified. As E. R. Hilgard (1973a) noted, "the fact that
no signs are now present, or that none are likely to be
found, does not deny the possibility that some subtle in-
dicators will be eventually uncovered" (p. 978).
What makes responsiveness to hypnosis so
stable? The failure to identify reliable personality cor-
relates accounting for more than a small proportion of
the variance in hypnotizability seems at odds with the
idea of a trait, the reliability of which rivals that of IQ.
Although relatively high levels of predictive power have
been shown for response expectancy (Johnston,
kowaski, DuBreuil, & Spanos, 1989; Kirsch,
& Council, 1992; Kirsch, Silva, Comey, & Reed, 1995),
multiple predictors, including imaginative ability and ab-
sorption, may prove capable of accounting for additional
variance (see Nadon, Laurence, & Perry, 1987).
To what extent can hypnotic responsiveness
be modified? This issue has important clinical as well
as theoretical implications. Hypnotizability may have
clinical relevance in diminishing patients' response to
pain (see Chaves, 1989; Stam, 1989) and in treating dis-
orders and conditions that have an involuntary compo-
nent (Perry, Gelfand, & Marcovitch, 1979; Wadden &
Anderton, 1982). Although hypnotizability can be mod-
ified in the laboratory, it will be important to demonstrate
that training gains can be translated into positive clinical
outcomes in the real world. Moreover, learning more
about the ability to modify hypnotic responsiveness will
broaden understanding of the stable or trait-like properties
of hypnosis (Lynn & Rhue, 1991).
What is the role of cognitive strategies in
hypnotic involuntariness and responding? Most
socio-cognitive theorists have viewed cognitive strategies
as central to hypnotic responding (see, however, Silva &
Kirsch, 1992) and have attempted to determine the spe-
cific strategies whereby particular responses are produced.
Although neodissociation theorists have not emphasized
cognitive strategies, they have acknowledged their im-
portance (E. R. Hilgard, 1986, p. 230; Kihlstrom, 1992).
In contrast, Bowers (1992) views cognitive strategies as
epiphenomena, and Kirsch (1985, 1991) has hypothesized
that the effects of these strategies may be mediated by
response expectancy. Establishing the role of cognitive
strategies is rendered particularly difficult by the widely
held belief that at least some of these strategies may op-
erate outside of awareness (Kihlstrom, 1992; Lynn et al.,
Shor, 1979; Spanos, 1991).
Does hypnosis produce fundamental shifts in
information processing? Socio-cognitive theorists
take issue with claims that hypnotized persons process
information in ways that are fundamentally different from
nonhypnotized persons. If verified, these claims would
justify adopting a strong version of the altered state hy-
pothesis. Findings that hypnotized participants process
information more holistically (Crawford, 1981), with rel-
atively little effort (see Bowers & Davidson, 1991), and
with greater verbal automaticity (Dixon & Laurence,
1992) than nonhypnotized participants warrant greater
attention and careful evaluation. It is particularly im-
portant to establish whether the differences are specific
to the induction of hypnosis or if they are correlates of
hypnotizability or products of hypnotizability by task in-
teractions. Data of the latter sort are very important with
respect to the trait issue, but they do not substantiate the
hypothesis of hypnosis as a state.
How can we better understand the unique
subjective experience of the participant? A task
of hypnosis theories is to explain not only variations in
hypnotic responses, but also variations in hypnotic ex-
October 1995 American Psychologist
periences. Many researchers have neglected to capture
the richness and complexity of participants' experiences
and their attempts to respond to suggestions in creative
and at times highly individualistic ways (see McConkey,
Sheehan, 1991). Descriptive phenomenological
methods, including clinical videotaped interviews (Shee-
han & McConkey, 1982), could help unravel the relations
between and among contextual factors,