FUNCTIONAL BRAIN IMAGING AND THE INDUCTION

OF TRAUMATIC RECALL:

A Cross-Correlational Review Between

Neuroimaging and Hypnosis

ERIC VERMETTEN1,2

University Medical Center and Central Military Hospital, Utrecht, The Netherlands

J. DOUGLAS BREMNER

Emory University and Emory Hospital, Atlanta, Georgia, USA

Abstract: The behavioral and psychophysiological alterations during

recall in patients with trauma disorders often resemble phenomena

that are seen in hypnosis. In studies of emotional recall as well as in

neuroimaging studies of hypnotic processes similar brain structures

are involved: thalamus, hippocampus, amygdala, medical prefrontal

cortex, anterior cingulate cortex. This paper focuses on cross-correla-

tions in traumatic recall and hypnotic responses and reviews correla-

tions between the involvement of brain structures in traumatic recall

and processes that are involved in hypnotic responsiveness. To further

improve uniformity of results of brain imaging speciﬁcally for trau-

matic recall studies, attention is needed for standardization of hyp-

notic variables, isolation of the emotional process of interest (state),

and assessment of trait-related differences.

In the last 10 years, there has been a rapid increase in our under-

standing of the brain processes that are involved in processing of

traumatic events (see Stern & Silbersweig, 2001). Much of this research

is related to the processing of stress, memory, and emotion (see reviews

of Armony & LeDoux, 1997; Baddeley et al., 2000; Bremner, Krystal,

Southwick, & Charney, 1995; Bremner & Narayan, 1998; Cahill, 2000;

LeDoux, 1993; McGaugh, Cahill, & Roozendaal, 1996; Nijenhuis,

The International Journal of Clinical

and Experimental Hypnosis

2004, Vol. 52, No. 3, pp. 280–312

Manuscript submitted November 02, 2002; ﬁnal revision received October 24, 2003.

This study was supported NIMH 1R01MH56120-01A1, and a Veterans Administra-

tion Career Development Award to Dr. Bremner.

Address correspondence to Eric Vermetten, MD, PhD, Department Psychiatry,

University Medical Center Utrecht, Utrecht University, Heidelberglaan 100, 3584 XC

Utrecht, The Netherlands. E-mail: e.vermetten@azu.nl

0020-7144/04/5203-280$16.00 # The International Journal of Clinical and Experimental Hypnosis

280

Van der Hart, & Steele, 2002; Packard & Cahill, 2001; Phillips, Drevets,

Rauch, & Lane, 2003a, 2003b; and the special issues of the International

Journal Clinical and Experimental Hypnosis, April and July 2003). Among

the factors that contributed to this increase are the availability of high-

quality functional brain imaging techniques, cross-fertilization of

different disciplines (e.g., cognitive and developmental psychology,

nuclear medicine, pharmacology, molecular biology, psychiatry), and

the increasing speciﬁcity of induction procedures for traumatic recall in

study protocols. (For a description of the characteristics of two of the

main neuroimaging methods, functional magnetic resonance imaging

(fMRI) and positron emission tomography (PET), see the Appendix).

Innovative experimental designs in the ﬁeld of neuroimaging have

increased our understanding of basic processes of memory storage and

emotion processing, lesion studies have pointed to brain regions that

are critically involved, and clinical studies in a variety of patient

populations have revealed a neural circuitry that is involved in differ-

ent disorders. These developments have also further contributed to our

current understanding of brain processes involved in pain, the phe-

nomenology of consciousness, and emotional processing in general and

have led to an understanding of regional patterns of activation and

deactivation.

Still, little is known about this neural circuitry that underlies (altera-

tion of) perceptual processing in patients with psychopathology in

which emotive processing is challenged in relation to exposure to

traumatic events. This alteration of perceptual processing can be

challenged by using visual, acoustic, or other sensory stimuli, or with

personalized narratives that induce recall of traumatic events. The

pattern of metabolic changes in the brain can be measured and corre-

lated with the subjective emotional response. This emotive process may

be paralleled by a set of involuntary/automatic processes that occur:

effects in heart rate, perceptual and emotional alteration, time distor-

tion, and analgesia (Hull, 2002), upon which patients may be selected

for assessment of their regional blood ﬂow patterns (Lanius et al., 2002).

These studies have started to appear in scientiﬁc journals in the last 8

years but are still scarce.

Despite promising study results, the ﬁeld of hypnosis has not fully

used the momentum that arose from these developments. Several

imaging studies in healthy populations have demonstrated differences

in the neural circuitry that is involved in response patterns across

hypnotic states, e.g., alterations of pain affect and pain modulation

(Faymondville et al., 2003; Rainville, Duncan, Price, Carrier, & Bushnell,

1997), alteration of visual processing (Kosslyn, Thompson, Costantini-

Ferrando, Alpert, & Spiegel, 2000), or hypnotic alteration of acoustic

perception (Szechtman, Woody, Bowers, & Nahmias, 1998). Most of

these studies have used high and low hypnotizable subjects to gain

NEUROIMAGING AND HYPNOSIS 281

insight in the neural mechanisms of perceptual alteration by measur-

ing alteration in regional brain blood ﬂow. From these studies, it

appears that high hypnotizables are capable of modifying their brain

metabolism in response to a speciﬁc set of instructions to alter affect,

pain, or other experiences and have pointed out that subjects can

differentially alter (block or stimulate) certain perceptual functions,

e.g., ‘‘taking the color out of a picture’’ that is presented in front of

them. To a considerable extent, high hypnotizables are capable of

modifying the circuitry with which their brains process stimuli. To

date, few of these studies have used the cumulative power of combin-

ing these knowledge-based resources in neuroimaging studies in

patient populations.

It has been a decade since studies by Stutmann and Bliss (1985),

Spiegel, Hunt, and Dondershine (1988), and Frischholz, Lipman, Braun,

and Sachs (1992) empirically conﬁrmed Janet’s early notions (1889) that

there is an overlap between the phenomena that are typically related to

hypnosis and the phenomena occurring in emotional recall in post-

traumatic stress disorder (PTSD). These patients have demonstrated

enhanced susceptibility to ‘‘hypnotic’’ situations, which traumatic re-

call can be considered to be. Hypnotic induction can mobilize a wide

spectrum of responses, varying from increased anxiety to ﬂashbacks

that can occur with or without feelings of detachment to other dis-

sociative experiences, such as numbing or freezing, feelings of in-

voluntariness, and loss of self-agency.’’ Moreover, classic hypnotic

responses such as time distortion, analgesia, and derealization can

occur along with these memories. The content of the emotion is also

widespread and can change rapidly depending on the focus of atten-

tion: e.g., anger, shame, guilt, or disgust. These responses can have

bimodal effects, such as enhanced attention versus lowering of atten-

tion or out of body experiences versus detailed focus on details, and can

also be reﬂected on the level of psychophysiological alteration, e.g.,

increased versus decreased heart rate. Although these may be related to

hypnotic virtuosity, this has not been studied yet.

Within a general framework of identiﬁcation, production, and reg-

ulation of emotional recall (see Phillips et al., 2003), hypnotic response

patterns are related to the involvement of different brain correlates

(Lanius et al., 2002). We pose that insight in these hypnotic response

patterns needs to be taken into account when analyzing brain correlates

of traumatic recall in trauma disorders, e.g., in PTSD but also in

dissociative identity disorder (DID) and borderline personality disor-

der (BPD). Moreover, hypnotic paradigms can provide additional

information regarding the involvement of involuntary mechanisms

in traumatic recall. In addition, we feel that by cross-correlating the

phenomenology and neurophysiology of traumatic recall and hypnosis

similarities in parameters, results can be found that improve our

282 ERIC VERMETTEN AND J. DOUGLAS BREMNER

understanding of hypnosis and basic elements of consciousness and

emotion. To explore this relation, we will review the imaging results in

these studies.

TRAUMA AS A HYPNOTIZING AGENT

It is a known fact that traumatic stress can mobilize responses that

have hypnotic features. These can be seen in a variety of situations,

e.g., in the battered and abused child who creates an invisible identity

so as not feel the pain and humiliation (identity alteration, amnesia,

R. Loewenstein, personal communication, November 2000), in jour-

nalists when watching an execution as an eyewitness (dissociation;

Freinkel, Koopman, & Spiegel, 1995), in survivors of the Estonia ferry

disaster who attempted to rescue other survivors (numbing; Ericksson

& Lundin, 1996), in people who witnessed victims jumping from the

World Trade Center on September 11, 2001 (verbal inhibition, Spiegel,

personal communication, 2001), or the responses in orphaned

Rawandan children (stupor). Traumatic experiences can mobilize hyp-

notic responses that resemble the hypnotic state during which intense

absorption in the hypnotic focal experience (Tellegen & Atkinson,

1974) can be achieved by means of a dissociation of experience

(Hilgard, 1977; Spiegel et al., 1988, p. 301). It was Janet who described

the splitting of consciousness that occurrs in response to traumatic

stress and the consequences of trauma on memory and identity. Janet

described a constellation of symptoms that we now categorize as PTSD

or dissociative disorders, including dissociative amnesia and fugue,

with a central assumption that different aspects of the traumatic

experience are not actively available to consciousness, although they

may have an inﬂuence on behavior (Loewenstein, 1993; Spiegel &

Carden˜ a, 1991).

The psychological processes that were captured as core components

of the hypnotic experience as described by Spiegel (1997) are: (a)

absorption; (b) dissociation, and (c) automaticity. These three factors have

been postulated because they explain the phenomena best in a hypnotic

situation (Spiegel & Carden˜ a, 1991). Hypnosis has been best deﬁned as

an altered state of consciousness. Recently, a new deﬁnition of hypnosis

has been coined in which the use of the word hypnosis as part of the

hypnotic situation is not necessary for the induction nor description of

the state of hypnosis (A. Barabasz, personal communication, September

2003).

Absorption is deﬁned as a narrowing and intensiﬁcation of attention,

a disposition for having episodes of single total attention that fully

engage one’s representational resources (Tellegen & Atkinson, 1974).

Physiological arousal can produce this narrowing of attention, which is

directed more to central aspects of the traumatic experience than to

NEUROIMAGING AND HYPNOSIS 283

peripheral aspects (Christianson, 1992). Narrowing of attention can be

functional in that all attention can be devoted to essential threat stimuli

and defensive concerns. Dissociation can be described as a kind of

divided or parallel access to awareness where several systems may

have some independence. It refers to a compartmentalization of ex-

perience, which can be considered complementary to absorption. The

term refers both to its origins, i.e., the splitting of consciousness that

may occur during trauma, and to the subsequent process of associating

or assigning experiences, as they occur over and over in time, to speciﬁc

states of consciousness, ego centers, or affective states (Crabtree, 1992).

Dissociation can also be part of an autohypnotic process (e.g., ‘‘I am

invisible; I have no feelings’’), which is applied to reduce the perception

of pain and the personal implications of trauma (Van der Kolk &

Van der Hart, 1989). Automaticity may be deﬁned as the tendency to au-

tomatically develop a belief in a suggested reality or the nonvolitional

transformation of a suggested idea to a suggested effect (Van Der Hart

& Van Der Kolk, 1991). Hypnotic automaticity reﬂects an altered sense

of self-agency consistent with a modiﬁcation of the property of mineness

of self-generated intentions and voluntary actions. The involuntariness

is captured in this description as well, representing the recognition of

one’s own volition and capacity to act (P. Rainville, personal commu-

nication, September 25, 2002; cf. Krystal, 1988). A diagram illustrating

the shared neurophysiology of hypnosis with the neurophysiology of

traumatic recall situations in highly hypnotizable subjects is illustrated

in Figure 1.

HYPNOTIC SUSCEPTIBILITY IN TRAUMA-RELATED

PSYCHOPATHOLOGY

A central theme in trauma-related psychopathology is that physical,

emotional, or sexual trauma can play a major role in the shift of this

control function manifesting psychological dysfunctions and/or bodily

or somatic problems (Van Der Kolk et al., 1996). This can be viewed as a

Figure 1. Hypothesized similarities between hypnosis and traumatic recall. This diagram

may be especially true for high hypnotizable subjects, which is usually the case in

patients with PTSD. (Adapted from P. Rainville, personal communication, 2002).

284 ERIC VERMETTEN AND J. DOUGLAS BREMNER

disembodied process with an emphasis on the information processing

analysis of attention mechanisms but also as a state of engagement of

the body-self in the interaction with an object of consciousness, with

emphasis on the biological substrate for the representation of self

(Damasio, 1999) and the property of selfhood (Metzinger, 2000). This

disembodiment can also be seen as a disengagement strategy that

serves a natural defensive function (Gilbert, 2000).

Hypnotizability has been described as the fundamental capacity to

experience dissociation in a structured setting. It underlies the ability to

enter trance; it involves the ability to segregate and idiosyncratically

encode experience into separate psychological or psychobiological

processes (Janet, 1898). Like dissociation, hypnotizability can be related

to a lack of agency or control versus loss of control over psychological

and sometimes also physical functions. It is a dispositional term that

points to its manifestation under certain circumstances, e.g., hypnotic

induction. The critical alteration in these processes occurs in what

Damasio called ‘‘feeling of knowing,’’ which is a fundamental aspect of

self-reﬂective consciousness that can be separated in hypnosis (p. 280,

1999). Self-representation is a derivative of this fundamental function

of consciousness. It is thought that in hypnosis, and also in traumatic

situations, these representations can be disrupted or processed in

separate streams of information. Self-representation is a hierarchically

organized function with activity in some ﬁrst-order maps in the brain

that are necessary (but not sufﬁcient) for higher-order representation of

self (e.g., autobiographical self), regulation of cognition and behavior,

and other more extended forms of consciousness.

From these notions, hypnotic capacity can be considered to be both

a liability and an asset; from the perspective of a defense strategy, it

serves a protective purpose (e.g., not remembering or not feeling),

however it can also become maladaptive and lead to dysfunctions

(e.g., time gaps, estrangement from inner feelings, ﬂashbacks) and

(psycho)pathology, like PTSD and dissociative or other trauma spec-

trum disorders. The disposition itself does not change but can be

considered ‘‘sensitized.’’ The symptoms of the dissociative and post-

traumatic states have been hypothesized to ﬁt in a diathesis-stress

model that views pathological dissociation as originating from an

interaction between innate hypnotizability and traumatic experience

(Butler, Duran, Jasiukaitis, Koopman, & Spiegel, 1996). If traumatic

experiences involve a hypnotic process or induce a hypnotic state,

then we should expect traumatized patients to show higher hypnotiz-

ability, in particular while still suffering from their trauma-induced

disorder. One would expect that they have higher scores on classical

hypnotizability scales than other psychiatric patient groups and

healthy or trauma controls. Indeed, several studies supported the

hypothesis that trauma-spectrum-disorder patients demonstrate

NEUROIMAGING AND HYPNOSIS 285

higher scores on classic hypnotic susceptibility scales than other psy-

chiatric patient groups and normal control subjects (Frischholz et al.,

1992; Spiegel et al., 1988; Stutman & Bliss, 1985). Their attention and

arousal systems are altered, rendering them prone to entering hyp-

notic states, with a relative decoupling between irrelevant external

events and mental (emotional) states during hypnotic states. It is not

the experience of trauma; it is the psychopathology that accounts for

the difference in hypnotic susceptibility. What happens with their

hypnotic susceptibility after successful treatment is largely unknown.

Although Janet observed that recovered patients became less hypno-

tizable (Janet, 1898), this ﬁnding still awaits testing in systematic

research.

RECALL OF TRAUMATIC MEMORIES

The ﬁeld of trauma spectrum disorders (consisting of PTSD, dis-

sociative disorders, (DD)), and perhaps also borderline personality

disorder (BPD, see Schmahl, McGlashan, & Bremner, 2002) has re-

ceived a great deal of interest in brain imaging studies. The phenom-

enology of theses disorders is at the heart of the interface between

memory and emotion.

Reexperiencing, Traumatic Recall, Flashbacks, and Flashbulb Memories

One of the most intriguing aspects of trauma disorders is the

reexperiencing phenomena. Numerous labels and descriptions have

been applied to this phenomenon (vanOyen Witvliet, 1997). In earlier

days traumatic recall was also described as ‘ﬂashback’, the reliving of

the traumatic event with strong emotional involvement (Frankel, 1994).

Flashback can lead to sleeping problems, irritability, feeling worse with

traumatic reminders, and secondary avoidance. For a long time ﬂash-

backs were assumed to lack a recognizable neurophysiological corre-

late – therefore they were thought to be at least as likely to be the

product of imagination as it is of memory (Frankel, 1994). However, in a

recent study in 62 PTSD patients comparing ﬂashbacks with ordinary

autobiographical memory performance on cognitive tasks demon-

strated that ﬂashback periods were associated with a speciﬁc decre-

ment in visuospatial processing, not speciﬁc with decrements on a

verbal processing task. Flashback periods were found to be associated

with increases in a wide range of autonomic and motor behaviors

(Hellawell & Brewin, 2002).

Flashbacks share a phenomenology with what has been described

by Brown and Kulik in 1977 as ﬂashbulb memory, to refer to the vivid

recollections that humans may have of events considered to be of

particular signiﬁcance to the individual. These memories were de-

scribed as having a photographic quality and as being accompanied

286 ERIC VERMETTEN AND J. DOUGLAS BREMNER

by a strong apparel of contextual information (weather, background

music, clothes worn, etc.) pertaining to the time and place where the

event was ﬁrst known. From a memory point of view we now know

that these memories are not perfectly accurate and are subject to

decay, but what does not necessarily decay is their capacity to evoke

emotions similar to the ones felt upon when ﬁrst exposed (Conway

et al., 1994). We feel that ﬂashbulb memories are formed by the

activity of evolutionary old brain mechanism evolved to capture

emotional and cognitive information relevant to the survival of the

individual. In the modern neuroimaging era some of the original

assumptions made by Brown and Kulik have since been challenged,

but the phenomenon in question has remained an important area of

research (Davidson & Glisky, 2002; Sierra & Berrios, 1999). The ex-

periences share clinical features such as involuntary paroxysmal

repetition, sensory vividness, and a capacity to trigger emotions like

anxiety, shame, or anger.

We prefer to use the term ‘traumatic recall’. This can be deﬁned as

imaginary (or virtual) re-exposure to a traumatic event in which the

person experienced, witnessed, or was confronted by death or serious

injury to self or others, and responded with intense fear, helplessness,

or horror, in which a re-experience of similar emotional responses

occur. They usually differ from usual/normal (autobiographical) mem-

ories in their emotional involvement (Van Der Kolk & Van Der Hart,

1991). Their nature is that a recall of the helplessness and uncontrol-

lability of the situation at that time, co-occurs with narrowing of the

attention so that ‘it feels like being back there’ (i.e., when and where the

traumatic event occurred). There can be a sense of loss of control or of

self-agency (‘‘That’s not who I am’’ or ‘‘It is not me to whom that

happened’’). There can be an autonomic response (such as tachycardia,

tachypnea, and diaphoresis) that can induce a feeling of panic (‘‘I’m not

going to make it’’). The recall may be activated by a variety of trauma-

related stimuli, thoughts about the trauma, the context, information

about the trauma, and trauma-related images, sounds, or smells, all

factors of which the person does not have to be aware. Veterans can

reveal this effect potently when they are exposed to darkness and

demonstrate augmented startle reﬂexes (Grillon, Morgan, Southwick,

Davis, & Charney, 1996).

Storage and Retrieval of Traumatic Memories

With long-term storage, memories are shifted from hippocampus

to neocortical areas, where the sensory impressions take place (Kim &

Fanselow, 1992; Phillips & LeDoux, 1992). This shift in the process of

memory storage to the cortex may represent a shift from conscious

representational memory (explicit memory) to unconscious memory

processes (episodic and implicit memory) that indirectly affect

NEUROIMAGING AND HYPNOSIS 287

behavior (Wallenstein, Eichenbaum, & Hasselmo, 1998). The cogni-

tive neuroscience perspective (see Brewin, 2001) favors a dual re-

presentational model of traumatic memories that proposes separate

memory systems underlying vivid reexperiencing versus ordinary

autobiographical memories of trauma. These two can be separated in

hippocampally-dependent and non-hippocampally-dependent forms

of memory, and are differentially affected by extreme stress. Within

this system, the strength of traumatic memories relates, in part, to

the degree to which certain neuromodulatory systems, particularly

catecholamines and glucocorticoids, are activated by the traumatic

experience (Cahill, 1997; Hasselmo, 1995). Both the quantity of re-

lease of these stress hormones, and the functional availability of the

target brain areas (e.g. hippocampus) modulate the encoding of

memories of the stressful event; ineffectiveness of the system may be

responsible for breakdown in the stream of events and changes in

the central and peripheral processing of the events. This can lead to

the wide spectrum of memory symptoms, ranging from hypermne-

sia, amnesia, deﬁcits in declarative memory, delayed recall of abuse,

and other memory alterations or distortions in trauma disorder

patients.

It should be kept in mind that traumatic memories are not ﬁxed or

indelible, but can change over time. Enhanced memory for the gist of

emotional events seems to be a dominant theme. What is encoded

depends on what was perceived, and what is encoded determines

what will be retrieved (Tulving & Thomson, 1973). Neuroimaging

may shed a light on the retrieval aspect of memory and its emotional

involvement by investigating brain processes that are occurring dur-

ing traumatic recall (Baddeley et al., 2000; Bremner, Krystal, Charney,

& Southwick, 1996; Sara, 2000; Zola, 1998). In PTSD patients ‘traumatic

cues’, such as a particular sight or sound reminiscent of the original

traumatic event, typically can induce a cascade of anxiety and fear-

related symptoms, sometimes without conscious recall of the original

traumatic event. This traumatic stimulus may not always be easy to

identify; it may be that through exposure to a movie, a smell, or more

subtle, a gesture or voice, a memory is metaphorically ‘reawakened’ –

traumatic memories can remain indelible for years or decades and

can be recalled by a variety of stimuli and stressors. A model of

extinction to explain this does not seem to qualify in these cases; a

better model seems to be the failure of successful inhibition of

traumatic memories.

Traumatic recall may not always be processed in an integrated mode

of consciousness. This may be a discontinuous experience with amnesic

gaps. Zimbardo, LaBerge, and Butler (1993) compared the emotional,

cognitive, and physiological responses of subjects experiencing induced

physiological arousal with and without awareness of the source of their

288 ERIC VERMETTEN AND J. DOUGLAS BREMNER

arousal. When subjects received posthypnotic suggestions for arousal

(increases in heart and respiration rate) with and without amnesia for

its source only hypnotizable subjects were expected to differ between

conditions. Indeed, for the hypnotizable subjects, unexplained arousal

produced signiﬁcant and dramatic effects when compared with

explained arousal, including misattributions (Zimbardo et al., 1993).

These experiments demonstrated that ‘discontinuous experiences’ can

contribute to the development of psychopathological symptoms in

normal persons. But recall can also be hypnotically blocked, e.g. by

posthypnotic suggestion. Here a disruption of retrieval like in post-

hypnotic amnesia or posthypnotic suggestion refers to a subjects difﬁ-

culty in remembering, after hypnosis. This is not a state-dependent

memory, but it does involve a disruption of retrieval processes similar

to the functional amnesias observed in clinical dissociative disorders. In

a situation like this implicit memory, however, is largely spared, and

may underlie subjects’ ability to recognize events that they cannot

recall (Kihlstrom, 1997).

INDUCTION PROCEDURES OF TRAUMATIC RECALL

Recall of traumatic events in imaging studies is usually embedded

in a so-called ‘‘activation paradigm’’ of re-experiencing traumatic

events. In this paradigm, the patient is asked to brieﬂy (for 1 to 2

minutes) recall a memory that is induced by a personal narrative,

smell, picture, or sound with different traumatic load (traumatic vs.

neutral). For the purpose of this paper, we focus on the recall induced

through emotional or cognitive induction. For the purpose of this

paper, we focus on the recall induced through emotional or cognitive

induction.

Traumatic Recall Through Emotional and Cognitive Induction

Typically in a traumatic-script procedure, the patient writes a

narrative of his or her two most traumatic personal events some days

before the scanning. Usually two neutral texts are made at that time for

the no-activation scan. This text is edited for length (30–40 seconds) and

content. The script is audio taped or can be read during the scan

procedure. The script can then be presented in ﬁrst or second person,

usually present tense. Immediately before each scan the participant is

instructed to ‘‘close your eyes, listen carefully to the audiotape or voice

and imagine the described events as vividly as possible, as if you were

actually participating in the event again’’ (cf. Lanius et al., 2001; Osuch

et al., 2001; Shin et al., 2000). The participant is then usually scanned 6

to 12 times with a 10-minute interval between scans. When the patient

is lying in the scanner, and the radioactive ligand is administered

intravenously, a trauma script (prepared by a participating patient)

NEUROIMAGING AND HYPNOSIS 289

similar to the one below (B. Elzinga, personal communication, July

2000) can be read:

Listen carefully to the script, and try to imagine as vividly as possible

the experience:

My mom is taking a shower. Dave comes up to me in the living room, where I

am standing. He is whispering in my ear, ‘‘I would prefer to kiss your private

part.’’ I think he is saying that as he presses my breast. Soon his hands sweep

down to my private area and he begins to massage it. His touch is not

welcoming; his pressing my breast hurt me and so does his touching my private

area. I am confused and afraid. Mom can come out of the bathroom any minute.

I want to tell him ‘‘stop,’’ but I don’t. It seems as if I can’t find my voice.

Eventually, I make gestures that imply I don’t want any more touching. He

eventually stops, after calling my name a couple of times. I am relieved, and I

seek some quiet corner of the apartment, just as my mom comes out of the

shower.

Now, continue to imagine the experience from the beginning to the end,

until I ask you to stop.

When applying the model of induction of emotional memories in a

trauma population, some points need to be considered:

(1) A prerequisite for successful implementation of a recall paradigm and

completion of the task in neuroimaging research is the ability of

the participating subjects to have reasonable control over their emo-

tional response in recalling traumatic events. In a PET paradigm, they

need to be able to return to a normal state within approximately 10

minutes. Subjects may – even though they are informed and have given

informed consent – become tearful, panicked, and emotionally over-

whelmed during the recall and feel an urge to suppress these responses.

Sometimes this fails and leads to termination of the scan (Osuch et al.,

2001).

(2) Extreme stress, high or low arousal, and fatigue are distinct psycholog-

ical factors that can separately and interactively affect how information

is processed – rendering it especially inﬂuential because it is not sub-

mitted to critical reality testing in a calm, relaxed, and rested state. This is

what Bowers described as a situation in which type II unconscious

inﬂuences occur. These describe how information is processed outside

normal awareness, initiative, and volition, speaking of dissociated experi-

ence and dissociated control as two complementary aspects of hypnotic

responsiveness (Bowers, 1973). Low-level monitoring of the process

when exposed to traumatic slides and sounds and calling this to a halt

will typically occur in the trauma-control subject; the situation is dif-

ferent in patients with PTSD. Their dissociated experience refers to the

fact that the (induced) state of affairs seems to occur nonvolitionally, –

which means here that the effort involved is not well presented in

conscious experience. These observations contribute to a framework

in which brain correlates of traumatic recall can be understood as

dissociated control. Upon asking subjects to voluntarily start recalling

290 ERIC VERMETTEN AND J. DOUGLAS BREMNER

a situation (‘‘Now continue to imagine the experience from the begin-

ning’’), some will anticipate becoming stressed and voluntarily control

the situation, and some will become upset and may not be able to stop

recalling (involuntary response).

(3) An important aspect in recall inductions is the content of intrusions.

Research studies suggest that they are not random fragments of the

experience. Typically, they represent stimuli that were present shortly

before the moments with the largest emotional impact (Ehlers et al.,

2002). They need not be sensory per se. Reynolds and Brewin de-

scribed elaborations of the original experience as the most intrusive,

linked to preoccupations with appraisals of the trauma and its se-

quelae, rather than presenting trauma memories (Reynolds & Brewin,

1998). This needs to be taken into account when preparing a narrative

script.

(4) Of importance in the induction of traumatic recall for brain imaging

studies is the theme of general versus speciﬁc induction of trauma-

related memories. Typically, in a general paradigm a standardized set

of images or words is presented, and the response pattern in the target

population can be calculated by averaging the responses. In a trauma-

speciﬁc paradigm, an individual induction is prepared before the brain

imaging procedure. In this paradigm, the surprise effect of the induc-

tions is somewhat diminished since the subject will recognize his or

her speciﬁc elements. Ehlers provides examples of the speciﬁc (sen-

sory) nature of the traumatic events from which it appears that

traumatic triggers are speciﬁc for both nature and content of the

trauma-related stimulus. In designing an experiment using olfaction

as a trauma-related cue in combat-related PTSD, we were to choose a

traumatic smell that could either be speciﬁc for each person or a smell

that all veterans reported as a trigger for traumatic memories. All

veterans had been exposed to diesel during their combat experience,

and diesel was present throughout the war. This smell therefore

seemed to qualify as both a generic and speciﬁc trauma-related smell

in the population (Vermetten, Schmahl, Southwick, & Bremner, 2003).

The same can be applied to trauma-related words and other types of

sensory stimulation.

(5) Laboratory studies have demonstrated that central cues of a traumatic

event are usually well remembered, whereas memory for peripheral

details is poor (Christianson, 1992). The narrowing of attention is often

used as an explanation for this ﬁnding. High anxiety and arousal are

thought to focus the attention on central aspects, such as the weapon

used, and hinder a full processing of the situation. It is thought that

changes in the perfusion of limbic brain structures that coincide with the

high arousal and/or anxiety, such as the amygdala and the hippocam-

pus, can lead to fragmented memories and personality fragmentation

(Spiegel, 1989; Van Der Kolk, Burbridge, & Suzuki, 1997). Narratives

should be written according to these notions.

(6) In all imaging studies in traumatic recall, the patient anticipates the

presentation of trauma (-related) material, and some researchers have

performed a dry run with the patient. Then the subject is not ‘‘cold’’ to

NEUROIMAGING AND HYPNOSIS 291

the trauma cue. It needs to be taken into account that this may dampen

the activation of the brain when exposed to the challenge.

(7) Last, in addition to the ﬁrst observation of this section, many clinicians

have described a ‘‘dissociative’’ or ‘‘hypnotic’’ blocking of perceptual

aspects as an adaptive response to trauma. Pain in recall can be blocked,

time processing can be distorted, or processing of the perception of

emotions like threat cannot be adequately processed. Patients may

dissociate during the experience and unless this is assessed at each

between-scan interval (to assess whether this is a positive or negative

phenomenon, see Lanius et al., 2002; Nijenhuis et al., 2002) it may

explain a difference in participant responding. In case patients do

dissociate, a systematic procedure needs to be administered to help

reorient them to the common environment and enable them to continue

with the scanning procedure reliably. In PET protocols, this is especially

important since the production of radioactive material is delivered in a

time-wise manner, and typically each interscan interval is set to 10

minutes.

FUNCTIONAL BRAIN IMAGING RESULTS IN TRAUMATIC

RECALL IN TRAUMA DISORDERS

To date, 12 imaging studies that used a symptom provocation

paradigm in PTSD have been published. Seven studies used PET

(Bremner, Narayan, et al., 1999; Bremner, Staib, et al., 1999; Osuch

et al., 2001; Pissiota et al., 2002; Rauch et al., 1996; Shin et al., 1997, 1999),

three used fMRI (Lanius et al., 2001, 2002; Rauch et al., 2000), and two

used SPECT as imaging technique (Liberzon et al., 1999; Zubieta et al.,

1999). The design, patient population, induction method, measure of

recall, psychophysiological coregistration, and changes in brain me-

tabolism are tabulated in Table 1. These studies have used various chal-

lenge models, exposing the subject – at varying levels of complexity –

to perceptual stimulations that range from exposing patients to slides

and sounds, smells of trauma-related experiences, to reading narrative

scripts, to the administration of pharmacologic agents like yohimbine

(see reviews by Bremner, 2002; Hull, 2002). Reexperiencing of traumatic

events typically coincides with heightened attention, lack of awareness

for the surroundings, and loss of perception of time. At the same time,

emotions of fear, shame, disgust, anger, and sadness, may occur and

sometimes coincide with dissociation, freezing, and other psychophy-

siological arousal phenomena (Nijenhuis et al., 1998).

The ﬁrst PET studies in traumatic recall used combat slides and

sounds and script-driven imagery in PTSD patients. The results sug-

gested that symptoms associated with traumatic recall were mediated

by the limbic and paralimbic systems within the right hemisphere.

Activation of visual cortex corresponded to the visual component of

PTSD reexperiencing phenomena (Rauch et al., 1996). When generating

Note. TC ¼ trauma controls, HC ¼ healthy controls, HR ¼ heart rate, GSR ¼ Galvanic Skin Response, SUDS ¼ Subjective Units of Distress,

PAG ¼ periaquaductal gray, ri ¼ right, le ¼ left, act ¼ n accumbens, VAS ¼ visual analog scale, CADSS ¼ Clinician Administrated Dissociative Symptom

Scale, STAIS ¼ State-Trait Anxiety Inventory Trait Test, MVA ¼ Motor Vehicle Accident, VVIQ ¼ vividness visual imagery questionnaire.

mental images of combat-related pictures, increased regional cerebral

blood ﬂow (rCBF) in the ventral anterior cingulate cortex (ACC) and

right amygdala was seen; when viewing combat pictures, subjects with

PTSD showed decreased rCBF in Broca’s area (Shin et al., 1997). These

ﬁrst PET studies of traumatic recall in PTSD have since led to a rapid

increase in similar studies modifying the experimental condition and/

or study population.

There is overlap but also considerable diversity in various traumatic

recall studies. The ACC, middle and superior temporal, middle frontal,

right orbitofrontal, occipital, hippocampal, parahippocampal, anterior

temporal, and inferior frontal cortices have all been implicated in

different studies, demonstrating either increases or decreases in perfu-

sion depending on the study conditions and sample population

(Phillips et al., 2003a, 2003b). In general, in comparison to trauma-

control subjects, these studies reveal an exaggerated response activa-

tion in the right (Rauch et al., 1996; Shin et al., 1997) or left (Liberzon

et al., 1999) amygdala, and in the sensorimotor cortex (Bremner,

Narayan, et al., 1999; Shin et al., 1997) and attenuated responses within

the medial prefrontal cortex (mPFC) (Bremner, Narayan, et al., 1999;

Shin et al., 1999) in patients with PTSD. In line with this, imaging

studies of normal autobiographical memory (i.e., no emotional activa-

tion) in healthy subjects compared to memory-control tasks have

pointed to mPFC and (left) hippocampus that are just particularly

responsive to such memories (Conway et al., 1999); other studies point

to right frontal cortices, medial parietal cortex, and cerebellum (Nyberg,

Forkstam, Petersson, Cabeza, & Ingvar, 2002).

Current studies support a model of PTSD in which (a) the amygdala

is hyperresponsive to threat-related stimuli, and (b) interconnected

areas may provide insufﬁcient ‘‘top-down’’ inhibition by mPFC and

ACC of amygdala response. This relative dysfunction of mPFC and

ACC is thought to lower the threshold of amygdala response to fearful

stimuli and is central to symptom mediation (Pitman, Shin, & Rauch,

2001; Villarreal & King, 2001). Thus, dysfunction of the mPFC areas

may provide a neural correlate of a failure of extinction of fearful

stimuli in PTSD.

Recall induction of emotion speciﬁcally activated the ACC. This

brain structure is critically involved in cognitive induction of emo-

tional responses and processes attention, executive functions, and

semantic and episodic memory. ACC activation represents a normal

brain response to traumatic stimuli that serves to inhibit feelings of

fearfulness when there is no true threat. Failure of activation in this

area and/or decreased blood ﬂow in the adjacent subcallosal gyrus

(area 25) may lead to increased fearfulness that is not appropriate for

the context, facilitating exaggerated emotional and behavioral re-

sponses (hyperarousal) to conditioned stimuli (Hamner, Lorberbaum,

296 ERIC VERMETTEN AND J. DOUGLAS BREMNER

& George, 1999). Posterior cingulate cortex (PCC) and motor cortex

and anterolateral prefrontal cortex are also known to modulate

emotion and fear responsiveness (Bremner, 2002). PCC plays an

important role in visuospatial processing and is therefore an impor-

tant component in the preparation for coping with a physical threat.

PCC also has functional connections with the hippocampus and

adjacent cortex.

In a meta-analysis of PET and fMRI studies of general emotional

activation reviewing 43 PET and 12 fMRI activation studies spanning

almost a decade of research, Phan, Wager, Taylor, and Liberzon (2002)

describe brain areas that are involved in emotion induction with

cognitive demand, typical paradigms of the recall of autobiographical

elements or visual imagery: