E D I T E D
BY
RICHARD R.
BOOTZIN,
J O H N F.
KIHLSTROM
AND
DANIEL 1.
SCHACTER
UNIVERSITY
OF
ARIZONA
Contributors ix
Foreword X
Preface xii
PART ONE: GENERAL OVERVIEWS
1 The Neurocognition of Sleep Mentation: Rapid 1
Eye Movements, Visual Imagery, and Dreaming
John Antrobus
vii
6 Behavioral Responses During Sleep 77
Frederick J. Evans
Index 203
viii
CONTRIBUTORS
John Antrobus, The City College of the City Universityof New York
Pietro Badia, Bowling Green State University
Michael H . Bonnet, Long Beach VeteransAdministration Hospital and University of
California,Irvine
Rosalind Cartwright, Rush-Presbyterian-St. Luke's Hospital
David F. Dinges, University of Pennsylvania
Eric Eich, The Universityof British Columbia
Frederick J . Evans, Universityof Medicine and Dentistry of New Jersey
Les Goldmann, Reed College
John Harsh, Universityof Southern Mississippi
J . Allan Hobson, Harvard Medical School
Milton Kramer, Sleep Disorders Center of Greater Cincinnati
Marta Kutas, Universityof California,San Diego
Stephen LaBerge, Stanford University
Wallace B. Mendelson, State Universityof New York at Stony Brook
ix
FOREWORD
xi
PREFACE
xii
nately, the methodology in these studies is generally clinical apd hermeneutic,
without the careful controls characteristic of experimental psychology.
Nevertheless, important questions about sleep persist that must be ad-
dressed at a purely psychological level: How much information from the envi-
ronment can be processed by the sleeper? What sorts of mental activities are
associated with sleepwalkingand sleeptalking?Are dreams meaningful, and if
so why aren’t they better remembered? What are the possibilities for learning
during sleep? Such questions have been asked by psychologists and others,
but only rarely have they been addressed with the concepts, principles, and
methods of contemporary cognitive psychology. One factor contributing to
this situation is the isolation of psychoanalysis, with its interest in dreams,
from mainstream scientific psychology. Another is the phenomenology of
sleep itself: The sleeper typically appears oblivious to his or her environment,
and the occurrence of sleep is generally inferred from this lack of responsive-
ness and the absence of conscious awareness of events during sleep. The com-
mon identification of cognition with consciousness leaves the impression that
little or no mental activity occurs during sleep.
Of course, studies in which sleepers are awakened at various points in
the night’s sleep typically yield reports of thoughts, images, and dreams, but
these are rarely remembered spontaneously in the morning. Some individuals
engage in episodes of sleepwalking and sleeptalking, which can involve rela-
tively complex speech acts and behavioral sequences-but, again, these are
rarely if ever remembered. One explanation for this universally experienced
memory deficit is that the cortical centers that mediate complex mental pro-
cessing are disengaged or deactivated during sleep, with the result that the
sleeper does not attend to environmental and mental events occurring during
periods of sleep. Because these events are not noticed, they are not processed
in a manner that encodes accessible traces of them in memory.
Thus, the most commonly accepted explanation of postsleep amnesia is
in terms of consolidation failure. For example, it has been proposed that the
low level of cortical arousal characteristic of sleep effectively prevents the
sleeper from performingthe cognitive operationsnecessary to encode memory
traces of sleep events that are accessible in the subsequent waking state. Ac-
cording to this view, dreams and other sleep events are remembered only when
the sleeper awakens during them, permitting retrieval from short-term mem-
ory. If the sleeper awakens shortly after a dream has occurred, residual infor-
mation retrieved from short-term memory may serve as a cue to the retrieval
of a highly degraded long-term memory trace. If retrieval is delayed until all
traces of the dream have decayed or been displaced from short-term memory,
the long-term traces will be virtually inaccessible.
As intuitively appealing as this explanation is, recent empirical work on
sleep, as well as theoretical advances in the study of cognition, call it into
account. For example, the idea of consolidation failure assumes a rigid distinc-
tion between short-term and long-term memory that is not supported by the
xiii
current literature. Moreover, cognitive theorists have recently begun to distin-
guish between effortful processes, which require intention and consume atten-
tional resources, and automatic processes, which do not. More important, the
domain of automatic processing has expanded to include rather complex
mental activities-the kind that ordinarily would be expected to leave residual
traces in permanent memory. Finally, studies of both brain-damaged patients
and intact subjects support a concept of implicit perception and memory, in
which current and past events may influence ongoing experience, thought,
and action even though the individual lacks awareness (concurrent or retro-
spective) of the events themselves.
Recent research on sleep processes bears directly on these theoretical
ideas. While sleepers are ordinarily considered unable to engage in complex,
intentional cognitive activities during sleep, some evidence tends to contradict
this assumption. For example, in the phenomenon of lucid dreaming, selected
subjects report that they become aware of the fact that they are dreaming and
are able to consciously direct the contents of the dream, while remaining
asleep. This claim is difficult to verify objectively, for obvious reasons. But a
series of highly provocative and apparently well-controlled studies has shown
that subjects can make discriminative responses to verbal suggestions deliv-
ered while they are asleep. Although the possibility of sleep learning had been
firmly rejected in a widely influential review published in 1955, studies of
memory of events occurring during surgical anesthesia strongly suggest that
sleep learning may be possible, provided that its success is measured in terms
of implicit rather than explicit memory. These preliminary findings, which
have emerged from a variety of laboratories employing rather different para-
digms, suggest that cognitive activity of considerable complexity may be possi-
ble during sleep, provided that it is mediated by automated procedures and
assessed by measures that do not require awareness of the events on the part
of the subject.
Despite the research possibilities offered by current advances in theory,
most texts on sleep, whether intended for undergraduate use or scholarly refer-
ence, pay little attention to cognitive processes, aside from our ubiquitous fail-
ure to remember our dreams. A major exception to this generalization is The
Mind in Sleep (Arkin, Antrobus, & Ellman, 1978), which provided a compre-
hensive account of research on cognitive processes during sleep since the dis-
covery of the EEG correlates of dreaming. The volume focuses on such topics
as the difference between stage REM and stage NREM mental activity, factors
affecting dream recall, the effects of presleep and intrasleep stimulation, REM
deprivation, sleeptalking, and night terrors. Unfortunately, the book is now
out of date, especially with respect to current thinking in cognitive psychology;
although a new edition has been promised, it is not yet available. Another
primary source is a recent monograph, Dreaming: A Cognitive-Psychological
Analysis (Foulkes, 1985). As comprehersive as the Arkin et al. anthology is,
its contributors have almost nothing to say about the problem that concerned
Freud most: the meaning of dreams. Foulkes’ book approaches this problem
xiv
from the perspective of cognitive psychology, particularly psycholinguistics
and cognitive development. His analysis clearly indicates that sleep research-
ers and cognitive psychologists have much of interest to say to each other.
In 1986, the Neurosciences Institute of Rockefeller University hosted a
small workshop on the remembering and forgetting of dreams, which testifies
to an increased interest in sleep among neurobiologists and psychiatrists; but
cognitive psychologists were decidedly underrepresented at this meeting. A
further impetus to dialog is the recently published report of an ad-hoc commit-
tee of the National Academy of Sciencesto the Army Research Institute, con-
cerning psychological techniques to enhance human performance (Enhancing
Human Performance:Issues, Theories, and Techniques, 1988). While gener-
ally critical of most of the techniques that have been proposed for this purpose,
the Committee strongly urged that further attention be devoted to the phe-
nomenon of learning during sleep. We expect that this suggestion will lead
to a greatly improved environment for research on all aspects of cognition
in sleep.
The present volume represents the Proceedings of the Arizona Confer-
ence on Sleep and Cognition, held in Tucson January 19-22, 1989. A princi-
pal concern of the conference was the implications of recent work on implicit
memory and other aspects of information-processingoutside of awareness for
studies of cognitive processes during sleep, and the role of the sleep laboratory
as a vehicle for studying various aspects of information processing outside of
awareness, in the absence of the active deployment of attention. To this end,
selected investigators in the area of sleep who have an interest in cognitive
processes were brought together with their counterparts in the area of cogni-
tion who have an interest in sleep. The goal of the conferencewas to create an
>
environment in which representativesof these two quite different areas would
meet and exchange ideas in a spirit of open inquiry and constructive criticism
and advance the study of cognition during sleep beyond the stage of specula-
tive thought. We hope that one outcome of this meeting, and publication of
the Proceedings, will be a new agenda for research on sleep that will carry
investigat rs of both camps well into the 2 1st century.
Part ne of the volume contains two papers in which mental activity
during sle p is used as a link between neuroscience and cognitive science.
Antrobus provides a sweeping review of the literature on eye movements and
image in dreaming, with particular reference to cortical activation during
sle . imilarly, Hobson provides an update of the Hobson-McCarley Activa-
tion-Synthesis model of dreaming.
Part Two includes a number of papers on various aspects of mental ac-
tivity during sleep. Kutas offers an authoritative review of the literature on
event-related (evoked) potentials during sleep. Harsh and Badia describe a
series of experiments on the transfer to sleep of conditioned responses ac-
quired during the normal waking state. Badia continues this discussion, focus-
ing on the acquisition of new conditioned responses during sleep. Evans
contributes an overview and update of his extremely provocative studies of
xv
response to verbal suggestions during sleep, and the relation of this phenome-
non to hypnosis and other dissociative states. Eich provides a new look, origi-
nally prepared for the 1988 National Academy of Sciences and Army Re-
search Institute study of techniques for enhancing human performance, at the
possibilities for sleep learning. LaBerge summarizes his programmatic re-
search on lucid dreaming, in which the sleeper becomes aware of the fact that
he or she is dreaming, and exercises some control over the content of the
dream. Finally, Goldmann explores the parallels between information pro-
cessing during sleep and during adequate surgical anesthesia.
Part Three focusses on cognition in the hypnogogic and hypnopompic
states occumng in the transition from waking to sleeping and back again.
Mendelson reviews the physiological and cognitive effects of drugs used in
treating insomnia. Bonnet examines the factors affectingsubjective reports of
sleep onset-which, he notes, almost never agree with physiological measures.
Dinges summarizes his research on performance impairments in the period
after awakening and illustrates his argument with a dramatic example of hyp-
nopompic reverie.
Part 4 returns to the question of dreams, this time in a clinical as opposed
to an experimental context, with an emphasis on the effects of stress on
dreams. Cartwright describes a new study of the dreams of women going
through divorce and explores the implications of her findings in relation to
the question of the meaning of dreams. Kramer summarizes his ongoing work
on nightmares in Vietnam veterans suffering posttraumatic stress disorder.
The Arizona Conference on Sleep and Cognition was supported in part
by the Science Directorate of the American Psychological Association. We are
grateful to Alan G. Kraut, Virginia E. Holt, and Barbara Calkins of the Science
Directorate for their efforts. Matching funds were provided by various units
of the University of Arizona (Department of Psychology, Lee Sechrest, Head;
Committee on Cognitive Science, Menill Garret, Director; College of Nurs-
ing, L. Claire Parsons, Dean; and the Sleep Disorders Center of the Arizona
Health Sciences Center, Stuart Quan, Director) and by the UpJohn Company
(James W. Battles, Medical Sciences Liaison). Final preparation of the Pro-
ceedings was supported in part by Grant #MH-35856 from the National Insti-
tute of Mental Health.
We are grateful to a number of individuals for their efforts in behalf of
the conference. Our colleagues Larry E. Beutler, L. Claire Parsons, and Stuart
F. Quan served as chairs for the conference sessions. Also at the University of
Arizona, Carol Cantor, Lucy Canter Kihlstrom, and Martha L. Glisky pro-,
vided organizational assistance-as did Karen Shoun of the Doubletree Hotel.
Elizabeth C. Tang of Communications: Visual designed the stunning confer-
ence poster. Barbara Calkins represented the APA Science Directorate at the
Conference; Lynne Lamberg, a science writer with a special interest in sleep,
grilled us during breaks and after-hours and forced us to express our ideas in
comprehensible English. Professors Robert A. Bjork (UCLA), David Koulack
xvi
(Manitoba), and Jonathan Winson (Rockefeller)made vigorous and incisive
contributions to the conference sessions. Finally, we thank those investigators
who enriched the conference by contributing summaries of their research to
the poster sessions: Andrew Brylowski (University of Texas Southwestern
Medical Center), Mindy Engle-Friedman (Baruch College, CUNY), Irene P.
Hoyt (University of Wisconsin), Robert Nadon (University of Pennsylvania),
Kan Paller (Yale School of Medicine), and Jim Wood (University of Arizona).
Richard R. Bootzin
John F. Kihlstrom
Daniel L. Schacter
xvii
3NO I t l V d
CHAPTER 1
THE NEUROCOGNITION
OF SLEEP MENTATION:
RAPID EYE MOVEMENTS, VISUAL IMAGERY,
AND DREAMING
JOHN ANTROBUS
activation within REM sleep so that the awakened dreamer is unable to recall
the relevant looking cues, or perhaps the frontal eye fields do not influence
EM direction until they are sufficiently activated by medial reticular forma-
tion (MRF) excitation beyond some minimal level. Note that it is assumed
that the visual dream imagery is produced regardless of whether the frontal
eye fields are sufficiently active to modify EM direction.
The Herman et al. (198 1) study provides the strongest evidence that
there is a relation between EM pattern and the visual imagery of REM sleep.
However, the evidence is consistent with both the looking hypothesis, which
implies that the cortex is the origin of EM control in REM sleep, and the
activation-synthesis model (Hobson & McCarley, 1977), which proposes that
REMs are controlled by activity that originates in the pons in the form of PGO
spike trains. According to the activation-synthesis model, the PGO activity is
carried to the occipital and frontal cortices, where the cortex synthesizes the
information about REM pattern to create a visual scenario that is consistent
with it.
Evidence that the pontine origin of PGO activity is out of cortical reach
has come from demonstrations of EMS in REM sleep in decorticate prepara-
tions (Jeannerod, 1966; Jeannerod & Mouret, 1962; Mergner & Pompeiano,
1981). Herman (in press) countered this interpretation of the decorticate data
by showing that the EM frequency is reduced by 80%,the bursting characteris-
tic is eliminated, and the EM patterns are stereotyped and repetitive. This
observation is perhaps the strongest evidence for the cortical control of REMs
and against the activation-synthesis position that subcortical units determine
the spatial and temporal patterns of REMs.
A second test of the looking response models of REM sleep EMS asks
whether the velocities of the EMSare similar to those of looking responses in
the waking state. The test is complicated by several factors: (a) the resting
muscle tonus is different in the two states, (b) the eyes are covered by the
lids during sleep, (c) waking saccadic movement is generally initiated by an
extrafoveal external visual stimulus, and (d) waking PGO activity tends to be
masked by the larger electromyogram activity in unrestrained animals. In two
comparisons of waking and REM EMS, Herman and colleagues (Herman et
al., 1981 ;Herman et al., 1983)found that REM EMSare quite similar to wak-
ing saccades with the eyes closed. The left-right patterns of EMS matched the
predictions of judges from the verbal reports of the subjects best when judges
took account of the imagined head movements of the dreamer. The temporal
distribution of EMS in REM sleep is, therefore, more similar to the waking
EM distribution when the head is unrestrained than when the head is held in
a fixed position. Herman and colleagues concluded, therefore, that the oculo-
motor system in REM sleep coordinates both eye and head movement.
Aserinsky, Lynch, Mack, TzankoE, and Hum (1985), however, argued
that, in REM sleep, the eye not only moves more slowly but also evidences a
“disengagement of the usual velocity-amplitude relationship seen in waking
movements” (p. 9). In the waking state, the angular velocity of the eye is pro-
Neurocognition 7
ACTIVATION-SYNTHESIS MODEL
The Activation-Synthesis model (Hobson & McCarley, 1977) consists of a
neurophysiologicalactivation model (McCarley & Hobson, 1979, which was
recently revised (Hobson, Lydic, & Baghdoyan, 1986), followed by a cognitive
synthesis process by which the cortex combines the neural information from
various activated subcortical locations to form dreamlike mentation. Their
“working sketch” (Hobson & McCarley, 1977, p. 1340)of the model assumes
that the “forebrain is tonically activated, probably via the midbrain RF that
is also responsible for its activation during waking. Thus the forebrain is made
ready to process information.”
They proposed that the stimuli from which dreams are constructed origi-
nate not in the “cognitive areas of the cortex” (Hobson & McCarley, 1977, p.
1347)but in a noncognitive or reflex process in the pontine brain stem, namely
PGO activity. Referring to the close association in time of PGO spikes and
REMs during Stage 1 REM sleep, they suggested that this process indirectly
provides spatial information that can be used in the construction of dream
imagery. Because the direction and temporal pattern of waking EMSand ves-
tibular activity provides some form of spatial information, they suggested that
the EMSand vestibular activity of REM sleep, even though they are randomly
driven by PGO spikes, might initiate similar forms of information processing
that might subsequently be synthesized into segments of a dream. This posi-
tion differs from that of the ASCIT model (Antrobus, 1990) in which these
brain stem generators provide only nonspecific activation to widespread areas
of neural networks in the association cortex and, thereby, determine the infor-
mation processing rate rather than more specific spatial information.
According to Hobson and McCarley (1 977, p. 1347), the process of
transforming this information into a dream is “likened to a computer search-
ing its addresses for key words”; “best fits to the relatively inchoate and incom-
plete data provided by the primary stimuli are called from memory, the access
to which is facilitated during dreaming sleep” by the tonically activated fore-
brain: “In other words, the forebrain may be making the best of a bad job
from the relatively noisy signals sent up to it from the brain stem.” The two
metaphors, of synthesis and of making the best of a bad job, might be restated
today in the more precise parallel distributed processes (PDP) vector models
8 JOHN ANTROBUS
tialling out of TRC, it will be only a minute part of the REM/NREM difference.
Knowledge of TRC can correctly sort 92.5% of REM/NREM pairs; knowledge
of bizarreness scores may add another 0.5-1 .O% at most.
of the information that flows through the system must be expanded to include
all the feedforward and feedback pathways that become active when the sys-
tem is not externally driven as well as an account of how these pathways are
state (waking, REM, NREM) dependent. This opens up a variety of new possi-
bilities concerning the relation between REMs and visual imagery. Spontane-
ous independent activity might occur in spatial, visual movement and higher
processing modules but might be transmitted to neighboring modules only
when the system is sufficiently activated by the MRF. At that point, a winner-
takes-all effect may allow one module to dominate another. For example, a
visual module may produce an event that higher cognitive modules are
obliged to accommodate to, or the higher cognitive module may inhibit in-
compatible productions in the lower visual modules. None of this visual
dreaming production need involve the oculomotor system unless the frontal
eye fields and the brain stem oculomotor systems are sufficiently well-
activated.
Because this kind of information is not accessible to introspection in
waking or in sleep, one must examine the patterns of neural activity in the
structures putatively responsible for the relevant component cognitive pro-
cesses in order to reduce such speculations to a reasonable model. However,
detailed evidence on the neural circuits that control EMScan only come from
animal studies, and this necessarily precludes examination of the relation of
neural activity to visual imagery in different components of this complex
system.
The general strategy pursued here is to identify the pathways that enable
EM control in waking perception and to then determine how closely this ac-
tion pattern operates in REM sleep. In waking perception, the decision to
move the eyes is distributed over many cortical and subcortical centers and is
based on a wide variety of information classes such as visual, auditory, and
eye and head position and velocity as well as on higher cognitive judgments
and values. Within the cortex, much of the perceptual and cognitive informa-
tion is funneled through the frontal eye fields that putatively decide where in
space to move the eyes and decide to communicate this, typically, to the supe-
rior colliculus, which further refines the decision and passes the command to
the oculomotor system that programs the EMS. However, the fine tuning of
the motor execution is camed out in a sideways loop into the seventh vermal
lobule of the cerebellum (Noda, Murakami, Yamada, & Aso, 1988), which
takes account of eye position and velocity relative to head and body position.
Before we examine some of the components that contribute to EMS in
waking and sleep, it is important to note that the activation of all components
is dependent on MRF activation. This activation, largely pontine in origin,
not only activates the cortical areas that produce both visual imagery and EM
decisions but also drives the final leg of the oculomotor system, namely, the
oculomotor neurons. At the most general level then, the co-occurrence of vi-
sual imagery and EMSin REM sleep is dependent on MRF activation. Within
REM periods, the occurrence of REM may also be due to a higher level of
MRF activation than that which exists during periods of ocular quiescence,
12 JOHN ANTROBUS
cerebellum to the brainstem, we must consider the possibility that this bursting
activity in REM sleep is triggered by cerebellar processes. This hypothesis is
also supported by the finding that discharge in the tegmental reticular nucleus
of Bechterew, which has cerebellar connections, is coherent with both PGO
waves and REMs and is also coupled with EM during waking (McCarley,Nel-
son, Hobson, & Strassman, 1981).
The role of the cerebellum in coordinating EM and head movement with
spatial search is well-known, and the cerebellum is generally thought to fine
tune the general EM control commands initiated in the frontal eye fields of
the cerebral cortex (Bruce & Goldberg, 1985). Further support for this conjec-
ture is provided by Hobson and McCarley’s (1972) demonstration that the
Purkinje cells of the cerebellum are more active in REM than in either NREM
sleep or waking and that, within REM sleep, they are more active during
REMs than ocular quiescence. Activation of the cerebellum in conjunction
with REMs is not evidence, of course, that the cerebellum is processing cortical
input, but it does demonstrate the operation ofone further link between cortex
and the oculomotor system during REM sleep EMS.
Although the major pathway from frontal eye fields to the brain stem
includes the superior colliculus, little is known about the relation between
collicular and cerebellar unit activity. The McCarley et al. (198 1) study shows
that it does not have direct input to the PGO units.
If the precise route from cerebral cortex to the burst cells of McCarley et
al. ( 1978)in REM sleep is to be identified with precision, it will require exten-
sive prior tagging in the waking state of individual neurons in the frontal eye
fields-cerebellum-burst cell sequence. Individual adjacent units, such as hold
units and fire units (particularly in the frontal eye fields), often serve quite
different functions in waking perception (Bruce & Goldberg, 1985), so that
measures of lead time and coherence between different nuclei or fields are
of little value if the comparisons are based on random samples of units in
each group.
The absence of high frequency and high amplitude burst cells in the cere-
bral cortex is not, in itself, evidence that the pontine burst cells are free of
cortical control. As described later, the oculomotor neurons have unique en-
ergy requirements, and only an inefficient brain design would place these cells
in the cerebral cortex surrounded by dense associative neuronal pathways and
distant from the oculomotor neurons. A better design would place the com-
plex decision making in the cortex and would place the neurons with high
volley rates, which it modulates, close to the oculomotor neurons, namely, in
the brain stem. The location of the burst cells and the input to the oculomotor
neurons in the MRF seem to fit this organization.
larly their relation to the visual imagery of dreaming sleep. The oculomotor
neurons always fire in bursts, proportional in duration to the projected size of
the saccade. But many neurons show a bursting pattern only when they are not
processing or transmitting information, and this is particularly true of sleep,
especially NREM sleep and the transitions from either waking or NREM to
REM sleep (Greene, Haas, & McCarley, 1986). Many of the rhythmic neural
patterns such as electroencephalogram (EEG) alpha and spindles occur when
the eyes close or in sleep states. These bursting patterns are generally termi-
nated by activation from the MRF (McCarley, Benoit, & Barrionuevo, 1983).
The bursting and rhythmic patterns within individual units can be determined
from the electrochemical properties of the units, but their functions are not
understood and may differ from structure to structure (Llinas, 1989).
It is the pattern of firing across a large set of neurons that putatively
carries the information processed by a nucleus or field. However, almost noth-
ing is known about how the shifts in spike patterns within a given unit affect
the pattern of a network of units within a given structure. Perhaps one excep-
tion to this pessimistic picture is the bursting patterns that lead to recruitment
of neighboring cells and lead to widespread rhythmic activity. In this case,
surely, the neurons carry no cognitive information other than the message that
the structure is in an off state.
The information carried by the pattern of PGO spikes is more difficult
to decode. The pontine leg shows a bursting pattern (Greene et al., 1986; Nel-
son et al., 1983), but its input to the lateral geniculate does not (Fourment &
Hirsch, 1980), even though other geniculate units do show bursting patterns
during REM sleep. Monaco et al. ( 1984) found that the primary spike in occip-
ital PGO waves is correlated with EM direction and amplitude in REM sleep
but not in waking.
In conclusion, it is clear that the information available about the EM
production system in REM sleep is, by itself, insufficient to support any strong
model regarding EM activity and the cognitive process of visual imagery pro-
duction. The evidence does not clearly support the cortical origin of EM con-
trol proposed by Herman et al. ( 198 I), but neither does it support the Hobson
and McCarley ( 1977) position that the REMs are totally free of cognitive con-
trol. Finally, the uninterpretability of cortical PGO bursting and high fre-
quency spiking does not support the activation-synthesis hypothesis that the
spikes interrupt ongoing cognitive processes to produce bizarre mentation.
Even the EM information identified in the occipital cortex by Monaco et al.
(1984) must remain suspect inasmuch as it is not seen in the waking state.
This latter point suggests one recommendation for all future research on
this issue, namely, what are the appropriate waking controls for the study of
EMSin REM sleep? If dreaming, as imaged perception, provides the context
for an EM, then the waking control condition must approximate the imagined
perceptual event. Because experimenters have no access to each subject’s
dreams, they must at the very least employ a variety of illumination levels and
visual events as well as head position and movement conditions. Obviously, if
16 JOHN ANTROBUS
ral control mechanisms for the two properties and that it may be worthwhile,
in future research, to distinguish among various properties of visual imagery.
Despite Rechtschaffen’s support for the phasic REM-visual imagery
model, the precise relation between the clarity of visual imagery, on the one
hand, and the REM sleep PGO and subcortical and cortical pattern of activa-
tion, on the other, is unclear. Although the PGO-associated occipital response
(Nelson et al., 1983) may include no external visual pattern information, it
may produce a general depolarization of pattern receptor cell units in areas
16, 17, and 18 so that higher order hypercomplex cells of the peristriate cortex
can be activated to produce visual patterns that are consonant with the cogni-
tive sequences generated in the associative cortex. Thus, mentation may be
produced throughout REM sleep but may generate more visual detail during
PGO-vestibular-related activation of the occipital cortex.
Although the role of the vestibular nuclei in the modification of sleep
mentation has been discussed only in the context of EMS, these nuclei may
provide a trigger for dreams of falling or floating. Vestibular activity, when
accompanied by a sporadic drop in the Stage 1 REM inhibition of alpha motor
neurons, may account for the dreamer’s experience of waking following a large
body jerk that is accompanied by a dreamed episode of being off balance. The
studies of the relation between REM sleep EMS and sleep mentation reports
by Roffwarg, Herman, and colleagues cited earlier indirectly support this
model. A significantly greater correlation between EM direction and the re-
ported directional shift of the dreamer’s gaze is observed when head move-
ments are considered rather than ignored. Experimental evidence for this
notion could be obtained in human sleep by means of pneumocaloric stimula-
tion of the vestibular system.
CONCLUSION
This chapter has reviewed evidence in support of the position that the PGO-
EM link is part of a feedforward system for the control of waking saccades.
Although waking saccades may be elicited by either visual or auditory input,
the executive decision to emit a saccade during waking and REM sleep appears
to be made in the cortex. Although this system is down throughout most of
sleep, the activation of the cortex, vestibular, and oculomotor system by the
MRF during portions of stage 1 REM is sufficient for the system to output
P W and REM periodically.The cognitive role in this cerebral cortex-cerebel-
lum-PW-EM sequence is demonstrated by the association of EMs in REM
sleep with the clarity and direction of visual imagery and with the total amount
of information reported. Eye movements and other phasic activity in REM
sleep are therefore interpreted as evidence of a temporary increment in the
activation of the cortex, including the visual association structures that gener-
ate the imaginal characteristics of dreaming.
The lawfulness of this control system is evidence against the interpreta-
tion of PGO activity as a random process. Therefore, it is evidence against
both Hobson and McCarley’s ( 1977) proposal that PGO is a random process
Neurocognition 21
that may, among other things, account for the bizarreness of sleep mentation
and Crick and Mitchison’s ( 1 986) proposal that PGO activity is a random
process that frees gridlocked neural nets during REM sleep.
References
Antrobus, J. S. (1983). REM and NREM sleep reports: Comparison of word frequen-
cies by cognitive classes. Psychophysiology, 20, 562-568.
Antrobus, J. ( 1986).Cortical and cognitive activation, perceptual thresholds, and sleep
mentation. Journal ofMind and Behavior, 7, 193-2 10.
Antrobus, J. ( 1990).Cognitive and cortical activation and sleep mentation. Manuscript
submitted for publication.
Antrobus, J., Hartwig, P., Rosa, D., Reinsel, R., & Fein, G. (1987). Brightness and
clarity of REM and NREM imagery: Photo response scale. Sleep Research, 16,
240.
Aserinsky, E., & Kleitman, N. (1953). Regularly occurring periods of ocular motility
and concomitant phenomena during sleep. Science, 118,361-375.
Aserinsky, E., Lynch, J. A., Mack, M. E., Tzankoff, S. P., & Hum, E. (1985). Compari-
son of eye motion in wakefulness and REM sleep. Psychophysiology, 22, 1-10.
Breitmeyer, B. C. (1986). Eye movements and visual pattern perception. In E. C.
Schwab & H. C. Nusbaum (Eds.), Pattern recognition by humans and machines
(pp. 65-87). New York Academic Press.
Bruce, C. J., & Goldberg, M. E. (1985). Primate frontal eye fields: I. Neurons discharg-
ing before saccades. Journal of Neurophysiology, 53,603-635.
Buttner, U., Buttner-Ennever, J. A., &Hem, V. (1977). Vertical eye movement related
unit activity in the rostra1 mesencephalic reticular formation of the alert monkey.
Brain Research, 130, 239-252.
Cohen, B., Feldman, H., & Diamond, S. P. (1969). Effects of eye movement, brain
stem stimulation, and alertness on transmission through lateral geniculate body
of monkey. Journal of Neurophysiology, 32, 583-594.
Crick, F., & Mitchison, G. (1983). The function ofdream sleep. Nature, 304, 11 1-1 14.
Crick, F., & Mitchison, G. (1986). REM sleep and neural nets. Journal of Mind and
Behavior, 7, 229-250.
Fein, G., Raz, J., Brown, F. F., & Memn, E. L. (1988). Common reference coherence
data is confounded by power and phase effect. Electroencephalographand Clinical
Neurophysiology, 69, 58 1-584.
Foulkes, D., & Pope, R. (1973). Primary visual experience and secondary cognitive
elaboration in stage REM: A modest confirmation and extension. Perceptual and
Motor Skills,3 7, 107- 1 18.
Fourment, A., & Hirsch, J. C. (1980). Synaptic potentials in cat’s lateral geniculate
neurons during natural sleep with special reference to paradoxical sleep. Neurosci-
ence Letters, 16, 149-154.
Goff, W. R., Williamson, P. D., Van Gilder, J. C., Allison, T., & Fisher, T. C. (1980).
Neural origins of long latency evoked potentials recorded from the depth and
cortical surface of the brain in man. In J. E. Desmedt (Ed.), Progress in clinical
neurophysiology (Vol. 7, pp. 126- 145). Basel, Switzerland: Karger.
Greene, R. W., Haas, H. L., & McCarley, R. W. (1986). A low threshold calcium spike
mediates firing pattern alternations in pontine reticular neurons. Science, 234,
738-740.
Hem, V., Buttner-Ennever, J. A., & Hepp, K. (1982). The primate oculomotor system:
I. Motoneurons. A synthesis of anatomical, physiological, and clinical data. Hu-
man Neurobiology, I , 77-85.
22 JOHN ANTROBUS
Henn, V., Hepp, K., & Buttner-Ennever,J. A. ( 1 982). The primate oculomotor system:
11. Premotor systems: A synthesis of anatomical, physiological, and clinical data.
Human Neurobiology, I, 87-95.
Herman, J. H. (1984). Experimental investigations of the psychophysiology of REM
sleep including the question of lateralization. Research Communications in Psy-
chiatry and Behavior, 9, 53-15.
Herman, J. ( 1989). Transmutative and reproductive properties of dreams: Evidence
for cortical modulation of brain stem generators. In J. A. Antrobus and M. Bertini
(Eds.), The neuropsychology of dreaming sleep. Hillsdale, NJ: Erlbaum.
Herman, J. H., Barker, D. R., & R o h a r g , H. P. (1983). Similarity of eye movement
characteristicsin REM sleep and the awake state. Psychophysiology, 20, 537-543.
Herman, J. H., Rofiarg, H. P., Taylor, M. E., Boys, R. M., Steigman, K. B., &Barker,
D. R. (1981). Saccadic velocity in REM sleep dreaming, normal visual activity,
and total darkness. Psychophysiology, 18, 188.
Hobson, J. A., Lydic, R., & Baghdoyan, H. A. ( 1 986). Evolving concepts of sleep cycle
generation: From brain centers to neuronal populations. Behavioral and Brain
Sciences, 9, 37 1-448.
Hobson, J. A., & McCarley, R. W. ( 1972). Spontaneous discharge rates of cat cerebellar
Purkinje cells in sleep and waking. Electroencephalograph and Clinical Neuro-
physiology, 33,451-469.
Hobson, J. A., & McCarley, R. W. (1977). The brain as a dream state generator: An
activation-synthesis hypothesis of the dream process.American Journal of Psychi-
atry, 134, 1335-1348.
Ito, M. ( 1 974). The control mechanisms of the cerebral motor systems. In F. 0.Schmitt
& F. G. Worden (Eds.), The neurosciences: Third study program (pp. 293-303).
Cambridge, MA: MIT Press.
Jeannerod, M. ( 1966). The phasic phenomena of paradoxical sleep. Revue Lyonnaise
du Medicine, 15, 27-44.
Jeannerod, M., & Mouret, J. ( 1 962). Etude des mouvements oculaires observes chez
I’homme au cours de la veille et du sommeil [The study of eye movements in man
observed during the course of waking and sleep]. Society de Biologie de Lyon, 25,
1407-14 10.
Kaufman, L., Okada, Y., Tripp, J., & Weinberg, H. (1982). Evoked neuromagnetic
fields. Annals of the New York Academy of Sciences, 388, 191-2 13.
Kondo, T. (1 988). Late REM activation and sleep mentation. Unpublished master’s
thesis, The City College of New York.
Kornhuber, H. H. (1 974). Cerebral cortex, cerebellum, and basal ganglia: An introduc-
tion to their motor function. In F. 0. Schmitt & F. G. Worden (Eds.), The neuro-
sciences: Third study program (pp. 267-280). Cambridge, MA: MIT Press.
Kosslyn, S. M. (1 988). Aspects of a cognitive neuroscience of mental imagery. Science,
240, 162 I - 1626.
Kurtzberg, D., & Vaughan, H. G. (1982). Topographic analysis of human cortical po-
tentials preceding self-initiated and visually triggered saccades. Brain Research,
243, 1-9.
Ladd, G. ( I 892). Contributions to the psychology of visual dreams. Mind, I, 299-304.
Llinas, R. R. (1989). The intrinsic electrophysiologicalproperties of mammalian neu-
rons: Insights into central nervous system function. Science, 242, 1654-1 664.
McCarley, R. W., Benoit, O., & Bamonuevo, G. (1983). Lateral geniculate nucleus
unitary discharge in sleep and waking: State- and rate-specific aspects. Journal of
Neurophysiology, 50,798-8 18.
McCarley, R. W., & Hobson, A. (1975). Neuronal excitability modulation over the
sleep cycle: A structural and mathematical model. Science, 189, 58-60.
McCarley, R. W., Nelson, J. P., & Hobson, J. A. (1978). Ponto-geniculo-occipital
(PGO)burst neurons: Correlativeevidence for neuronal generators of PGO waves.
Science, 201,269-272.
McCarley, R. W., Nelson, J. P., Hobson, J. A., & Strassman, A. ( 198 1). A cross-correlo-
Neurocognition 23
J. ALLAN HOBSON
Three factors that are of relevance to the level, source, and mode of informa-
tion processing by the brain-mind state can be quantitatively estimated from
neurophysiological data. They are (a) activation (Factor A), which estimates
the electrical energy level of the information processing system from the rate
of discharge of midbrain reticular formation neurons; (b) input source (Factor
I), which estimates the provenance of data to be processed by the system from
the ratio of external to internal stimulus strength values; and (c) mode of pro-
cessing (Factor M), which estimates the way that data will be treated by the
system by calculating the ratio of cholinergic to aminergic neuronal discharge.
The model is called AIM as both a shorthand eponym combining the
first letters of each of the three factors and as a method of conveyingthe aspira-
tions of a still evolving concept. Since each of the three factors of AIM can be
defined in cognitive terms and can be measured as a neurobiological variable,
the model is both conceptually promising and empirically anchored. The ulti-
mate goal is to unify the brain and mind in a new way. This unity is implied
by the hybrid term bruin-mind,implying a functional identity of the two
domains.
When we mathematically combine Factors A, I, and M in various ways,
both explanatory and predictive principles emerge that render the model heu-
ristically valuable whatever its ultimate verity may be when it is corrected or
elaborated in the light of new data. The key to the success of the concept-
and its major limitation-is its focus on the global states of waking, sleeping,
25
26 J. ALLAN HOBSON
and dreaming and its focus on the very superficial formal aspects of cognition
during those states while ignoring, for the time being, many fascinating cogni-
tive details that also characterize and differentiate those states.
Subjectivity
Reports by human subjects of their internal mental states, despite wide vari-
ability and notorious unreliability in special cases, have both consensual and
face validity that cannot be denied. The global level of the model accommo-
dates this problem nicely by concentrating on the most unequivocal and uni-
versal aspects of mental states, their formal properties. Thus, if 80 of 100
subjects report visually vivid, constantly animated, emotionally charged, nar-
ratively coherent but bizarrely organized mental experience on experimental
AIM Model 27
awakening from REM sleep, I tend to believe that most of the 80 subjects, and
perhaps even all 100 subjects, actually had such an experience.
Isomorphism
The correlation of these formal mental state variables with brain states as they
can now be measured in humans justifies the assumption of an isomorphic
relation between the consensually validated subjective experiences and the
objectively recordable states of the brain. This correlation may be even
stronger than is now supposed given the uncertainties of accurate reports of
subjective data out of sleep and the relative superficiality of the currently re-
cordable electroencephalogram (EEG) parameters of brain activity. Isornor-
phic is defined as a similarity of form in the two domains of brain and cogni-
tion (McCarley & Hobson, 1977). Thus, for example, if memory is poor for
dreams, we can assume that the biological substrate of memory is impaired.
If, conversely, dream vision is sharp and intense, we can assume that the visual
brain is strongly activated in a way that simulates wake-state activation and
stimulation.
Animal Models
To overcome the problem posed by the inaccessibility of the human brain to
deep neurobiological analysis, we have recourse to animal models whose
brains can be studied at the level of regions, nuclei, neuronal populations,
individual neurons, membranes, and even molecular neurotransmitters (Hob-
son, Lydic, & Baghdoyan, 1986). As payment for the rich data that are thus
obtainable, we accept the uncertain assumption that all mammalian brains,
or at least those of the human, monkey, and cat, share common mechanisms
of state control so that we can build our model up from subhuman data
sources. This strategy makes no assumptions about whether the animals stud-
ied experience consciousness as we humans know it. I will now briefly discuss
the evidence that encourages such assumptions and provides the substantive
ground for the AIM model.
Brain
Muscle
Brain
Figure 2 A comparison of REM sleep electrographic features in man and cat, based
on a time-lapse photographic study by Theodore Spagna.
As the NREM sleep state evolves toward REM sleep in the cat (see Figure
3), we notice an increase in SI with the evolution of the pontine-geniculate-
occipital (PGO) waves (Callaway, Lydic, Baghdoyan, & Hobson, 1987). The
increase in SI parallels the resurgence of activation, but the cross-correlation
is no more linear than the relation of activation to SE at sleep onset; rather, it
is similarly and strikingly exponential. It is also significant that the thalamo-
cortical oscillation of NREM sleep subsides inversely with the increment in
PGO wave frequency. At a later and critical point in the transition from
NREM to REM, the PGO waves are no longer the isolated, large-amplitude
single spikes and wave complexes but become clustered into groups of 6-10
smaller waves (see Figure 4).That the PGO waves are genuinely sensory stim-
uli of entirely internal provenance is made clear by the discovery that during
REM sleep eye movement direction is encoded in the left-to-right amplitude
difference of the waves in the two lateral geniculate bodies and occipital corti-
ces (Callaway et al., 1987; Monaco, Baghdoyan, Nelson, & Hobson, 1984).
This discovery further justifies the assumption that these internally generated
sensory stimuli actually arise on the motor side in keeping with two important
concepts: First, that motor commands arising in the upper brain are abundant
in REM sleep but are quenched by inhibition of the anterior horn cells in the
30 J. ALLAN HOBSON
c-(
spinal cord and, second, that REM sleep is characterized by a marked increase
in the excitability of internal communication systems of the brain.
We are thus justified in postulating that in REM sleep there is not only
the diminished access to external data by the brain-mind that is typical of all
sleep but also an increase in the strength of internally generated data. It seems
likely, moreover, that these PGO waves are a reflection of the disinhibition of
the efferent copy or corollary discharge systems of the brain by which sensori-
motor integration has been conceptualized since the time of the German neu-
rophysiologist, Hermann von Helmholtz (for a discussion, see Hobson,
1988a). Although PGO waves are not easily recorded from human subjects,
recent evidence supports the assumption that they do exist in our species and
that they also signal the direction of eye movement from brainstem motor
centers to the cerebral cortex (McCarley & Ito, 1983; McCarley, Winkelman,
&Durn, 1983).
In the shift from waking through NREM sleep to REM, the brain-mind
has thus been deactivated (Factor A) with a concurrent decrease in the access
of external data to the system and then reactivated with a concurrent increase
in internal stimulus strength. We can therefore say with confidence that the
cognitive differences between waking and REM sleep could not possibly de-
pend only (if at all) on a changed level of activation. Rather, it is clear that
both the waking and REM sleep states differ from NREM sleep and that they
differ in equal degrees on this dimension. This difference parallels the cogni-
tive downshift in NREM sleep that was pointed out by Antrobus (Chapter 1,
by J. Antrobus in this volume). Input source, however, has shifted radically
from external (in waking) to internal (in REM sleep), whereas NREM sleep
----
AIM Model 31
EOG-
I -
0 Se/ectiviv Gradient
C lnterpenetrotion Grodient
10 I ( F T G ) 3 I (DRN)
OExcitotory *Inhibitory
Figure 5 Schematic illustration of the hypothesized relation between physiological
selectivity and anatomical interpenetration of REM-on (left column) and REM-off
(right column) cells whose reciprocal interconnections are modeled in Factor A.
I
I
Midbrain
I
I
'
I
RF
Pons I LC
II p
I
Medulla I
I
I
I
I
I Atmia andrtlusce Twitches I
I
Spinal I
Cord I
I-----I- ---- l - - - l H
0 Inhibitory OExcitatory 5%
Ret N
*
4
@
1-
LC
RN
PN
01 I I
w S D W S D
Behoviord Sfafe
(4 (B)
b Aminergic
A/C Ratio
I I I
Waking NREM REM
Sleep Sleep
(C)
Figure 7 Behavioral state and neuronal discharge rate. (A) D-on cells. (B) D-off
cells. (C) State-dependentchanges in aminergic to cholinergic ratio.
AIM Model 35
Wake
100
10 -
S@
1-
0.1-
Figure 8 At sleep onset, a sudden and precipitous drop in the value of S could be
caused by the sudden drop in the value of SE (with eye closure) after a slow decline in
activation.
Figure 9 Three dimensional state space defined by the values for brain activation
(A), input source and strength (I), and mode of processing (M).
logical signs of REM sleep are still present. It has been difficult for many sleep
scientists to understand what physiological mechanisms could allow such
wake state motor behavior as voluntary movement to occur during REM sleep
because, by definition, motor output is blocked in REM. The AIM model
allows us to consider another possibility: that lucid dreaming is a unique and
specific state of the brain-mind. Because it has some features of waking and
some features of REM, it is a hybrid state. Where would lucid dreaming fall
in the state space?
It seems unlikely that a change in activation could account for the results
because the level of cognitive activation is, if anything, higher than in REM
sleep or in waking. Rather, one suspects a change in input source, with a tem-
porary suppression of internally generated signals, allowing cortical com-
mands to dominate the oculomotor apparatus in a wake-state fashion. Put
another way, the usually weak top-down forces gain control of the usually
strongbottom-up forces that are battling for control of the brain-mind in REM
sleep. We might imagine Factor I as increasing toward waking values (ie.,
38 J. ALLAN HOBSON
toward the back ofthe state space, though still keeping to the right side because
activation remains high). Since the REM-PGO generator network in the brain
stem is activated via aminergic disinhibition, one way that such an effect could
be enhanced is by turning on the locus coeruleus and raising the value of MA,
which would put a brake on the cholinergic REM sleep generator. This would
have the effect of moving AIM upward in the state space. The net effect of
raising Factors I and M toward waking levels, while conserving a high level of
Factor A, would be to locate lucid dreaming on the right-hand wall of the state
space in the midpath of the rapid reset trajectory of the system when REM
suddenly gives way to waking at the end of each REM cycle.
This view predicts (a) that lucid dreamers will frequently awaken from
REM sleep once dream consciousness is achieved and (b) that lucidity will be
easiest to induce at times in the night when the system is likely to be changing
from REM to waking. This might be true during the brief period of descending
to Stage 1 episodes at sleep onset but will be even more likely at the end of
REM periods late in the night. This hypothesis could be tested in longitudinal
home-based studies of experienced lucid dreamers.
REVERIE
Dramatic state dissociations occur on arousal from NREM sleep, especially
in sleep-deprived subjects (Chapter 12, by D. F. Dinges in this volume). Al-
though they are electrically awake (that is, showing signs of high levels of acti-
vation) and capable of input-output processing of the wake state type (that is,
also having high levels of input source), such individuals appear to be unable
to keep a cognitive test task in mind, and hence they perform poorly. They
may even confabulate and produce garbled language only loosely rooted in
the task structure imposed by the experimenter. Thus, we are again dealing
with a paradox: Features of wake are hybridized with features of NREM sleep.
Variations of this functionally significant anomaly occur naturally in
children with night terrors and in adults (such as physicians and other night-
shift workers) who find themselves unable to perform complex cognitive tasks
when suddenly aroused from NREM sleep by duty calls. Broughton (1 968)
has dubbed these phenomena disorders of arousal. Dinges (Chapter 12, by D.
F. Dinges in this volume) has postulated that “sleep inertia” may mediate this
phenomenon because, despite their best intentions, the subjects feel irresist-
ibly pulled back down into sleep.
What could this inertial element of the brain-mind be, according to
AIM? Factors A and I are probably not the culprits because the EEG is fully
activated and wakelike input-output processing is possible. It seems rather that
it is attention, the ability to hold a concept or task in mind and focus on
it, that is defective. This would result if Factor M remained effectively low,
so that the cortex, while electrically activated and capable of input-output
transactions, remained biochemically demodulated. How could such an
event occur?
Suppose that, early in the night, the serotonergic and noradrenergic sys-
tems of the brain have been operating at half-speed for 1 to 3 hours. And
AIM Model 39
CONCLUSlON
On the assumption that the states of waking and sleep effect cognition via
neurobiological mechanisms, a three-factor model is proposed. Reflecting the
electrical energy level of the brain is Factor A (for activation), which has a
well-establishedcognitive analogue in network modeling parlance. The prove-
nance of the information to be processed is assessed by Factor I (for input
source), which measures the relative access and strength of external versus
internal stimuli or data. Factor I also relates to motor activity via the corollary
discharge or efferent copy concept. The most speculative and original compo-
nent of the model is Factor M (for modulation), the estimated ratio of aminer-
gic to cholinergic neurotransmitter concentration, which measures the mode
of information processing manifested by the brain-mind. By mediating atten-
tion, learning, and memory processes, Factor M alters the capacity of the acti-
vated brain-mind to focus and perform intentional analytic computations of
its data.
References
Anderson, J. (1985). Cognitive psychology and its implications (2nd ed.). New York:
Freeman.
Black, I. B., Adler, J. E., Dreyfus, C. F., Friedman, W. F., LaGamma, E. F., & Roach,
A. H. (1987). Biochemistry ofinformation storage in the nervous system. Science,
236, 1263-1268.
Broughton, R. (1968). Sleep disorders: Disorders of arousal. Science, 159, 1070-1078.
40 J. ALLAN HOBSON
Callaway, C. W., Lydic, R., Baghdoyan, H. A., & Hobson, J. A. (1987). Ponto-genicu-
lo-occipital waves: Spontaneous visual system activation occurring in REM sleep.
Cellular and Molecular Neurobiology, 7, 105- 149.
Flicker, C., McCarley, R. W., & Hobson, J. A. (198 1). Aminergic neurons: State con-
trol and plasticity in three model systems. Cellular and Molecular Neurobiology,
I, 123-166.
Hess, W. R. (1954). The diencephalic sleep centre. In E. D. Adrian, F. Bremer, &
H. H. Jasper (Eds.), Brain mechanisms and consciousness (pp. 1 17-1 36). Oxford,
England: Blackwell.
Hobson, J. A. (1988a). The dreaming brain. New York: Basic Books.
Hobson, J. A. ( 1988b). Homeostasis and heteroplasticity: Functional significance of
behavioral state sequences. In R. Lydic (Ed.), Clinical physiology of sleep (pp.
199-220). Bethesda, MD: American Physiological Society.
Hobson, J. A. (1989). Sleep. San Francisco: Scientific Amencan Library.
Hobson, J. A. (in press). A new model of brain-mind state: Activation level, input
source and mode of processing (AIM). In J. Antrobus & M. Bertini (Eds.), The
mind-brain in dreamingsleep: Activation andasymmetry. Hillsdale, NJ: Erlbaum.
Hobson, J. A., Lydic, R.,& Baghdoyan, H. A. (1986). Evolving concepts of sleep cycle
generation: From brain centers to neuronal populations. Behavioral and Brain
Sciences, 9, 37 1-448.
Hobson, J. A., & Steriade, M. (1986). Neuronal basis of behavioral state control. In V.
Mountcastle & F. E. Bloom (Eds.), Handbook ofphysiology: The nervous system
(Vol. 4, pp. 70 1-823). Bethesda, MD. American Physiological Society.
Laborit, H. (1986). L’Inhibition de l’action [The inhibition of action] (2nd ed). Mon-
treal, Canada: Masson.
Mamelak, A. N., & Hobson, J. A. (1989). Dream bizarreness as the cognitive correlate
of altered neuronal behavior in REM sleep. Journal of Cognitive Neuroscience, 1,
201-222.
McCarley, R. W., & Hobson, J. A. (1977). The neurobiological origins of psychoana-
lytic theory. American Journal of Psychiatry, 134, 121 1- 1221.
McCarley, R. W., & Ito, K. (1983). Intracellular evidence linking pontine reticular
formation neurons to PGO wave generation. Brain Research, 280, 343-348.
McCarley, R. W., Winkelman, J. W., & D u e , F. H. (1983). Human cerebral potentials
associated with REM sleep rapid eye movement: Links to PGO waves and waking
potentials. Brain Research, 274, 359-364.
Monaco, A. P., Baghdoyan, H. A., Nelson, J. P., & Hobson, J. A. (1984). Cortical PGO
wave amplitude and eye movement direction are correlated in REM sleep but not
in waking. Archives of Italian Biology, 122, 2 13-223.
CHAPTER 3
MARTA KUTAS
I extend sincere thanks to C. Van Petten for her helpful comments on this manuscript.
Research efforts were supported by a Research Scientist Development Award from the National
Institute of Mental Health (MH00322) and by a grant from the National Institute ofchild Health
and Human Development (HD226 14).
ERP Studies 45
is a clinical tool that provides an index of the integrity of the sensory pathways
in the central nervous system. For diagnostic purposes, the most informative
EP components are those that are consistently elicited by specific stimulus
parameters in the face of fluctuations in arousal or attention. Studies over the
past 20 years have indicated that the auditory brain evoked potential (ABEP),
also known as the auditory brain stem response (ABR) or the brainstem
evoked response (BER), providesjust such a response. The ABR consists of a
series of small vertex-positive waves occurring within the first 10 ms after an
abrupt stimulus (Campbell, Picton, Wolfe, Baribeau-Braun, & Braun, 1981;
Chiappa, Gladstone, & Young 1979; Stockard, Stockard, & Sharbrough,
1978). Peak 5 of the response is usually recognizable to within 20 dB of the
behavioral auditory threshold (Galambos & Hecox, 1978; Picton, Woods,
Baribeau-Braun, & Healy, 1977; Picton, Stapells, & Campbell, 1981). While
there is considerable intersubject variability in component amplitudes, the la-
tencies of the BER peaks are remarkably stable across replications, provided
that the subject’s age, gender, and body temperature as well as the physical
properties of the eliciting stimulus and recording parameters are held constant
(e.g., Moore, 1983). Such reliability, in fact, has sanctioned the development
of a normative data base against which deviant responses can be judged. Ab-
normalities in the timing of various components relative to norms signal a
possible cause for concern in either audiological or neurological domains.
With a few exceptions that have been difficult to replicate (Lukas, 1980, 198l),
it has been demonstrated that the first five waves of the BER are insensitive to
attentional manipulations in an awake adult (Picton & Hillyard, 1974; Wol-
do@ Hansen, & Hillyard, 1987; Woods & Hillyard, 1978). Thus, ABR mea-
surement has become a routine means of objective audiometry, especially for
subjects who are unwilling or unable to undergo traditional methods of exami-
nation (e.g., neonates, young children, demented individuals).
The insensitivity of the BER to arousal levels and attentiveness in awake
subjects suggests that the BER should be the same regardless of whether the
subject is awake or asleep. The literature on this question is, however, contra-
dictory. For instance, Amadeo and Shagass ( 1973) reported latency shifts as
large as 0.25 ms in different stages of sleep. In contrast, Hellekson, Allen, Gree-
ley, Emery, and Reeves (1979) found no differences in the latency of any ERP
components between waking and light naps. Osterhammel, Shallop, and Ter-
kildsen (1985) noted only minor changes in the latency of Peak 5 for only
some subjects and only for low intensity stimuli. Campbell and Bartoli ( 1986)
suggested that such inconsistencies in the literature may be due to a number
of factors including (a) failure to control the stimulus input (that is, using loud-
speakers rather than headphones), (b) failure to control for covariation in
core temperature during the night (Jones, Stockard, & Weidner, 1980; Mar-
shall & Donchin, 198l; Stockard, Sharbrough, & Tinker, 1978), and (c) small
sample size.
Campbell and Bartoli (1986) evaluated the BER at different intensities
and rates of stimulus presentation during wakefulness, Stages 2 and 4, and
46 MARTA KUTAS
REM sleep while maintaining precise control over the stimulus input (with an
ear-mold hearing aid device) and monitoring temperature in 9 female subjects
( 18-25 years old). They reported that regardless of (a) the stage of sleep, (b)
the time of night, (c) the rate of stimulus presentation, and (d) the intensity of
the stimulus, sleep did not appear to have a significant effect on any of the
components of the click-evoked ABR. They concluded “that auditory infor-
mation is transmitted without alteration from the periphery through the brain-
stem relay centres during sleep” (p. 146). Before we accept this conclusion, we
must remember that the BER is a sensitive index of the timing operations of
the auditory system but is silent about the processing of the contents of the
auditory signal.
Components of the auditory evoked response occumng between 10 and
100 ms have also been called into service for clinical hearing assessments.
These components, referred to as middle latency responses (MLR), come in
the form of transient or steady-stateresponses. Transient responses are evoked
whenever stimuli are presented at slow rates (lO/s or less), whereas steady-
state responses arise when stimuli are presented at such a high repetition rate
that the responses to successive stimuli overlap (e.g., Stapells, Linden, Suffield,
Hamel, & Picton, 1984). The auditory MLR has received considerable atten-
tion in audiometry because of the possibility that it might allow determination
of frequency-specific auditory thresholds. And indeed, threshold estimates
from MLRs correlate well with behavioral thresholds, with the estimates from
the steady-state response being less variable than those derived from transient
responses (Galambos, Makeig, & Talmachoff, 1981; Goldstein & Rodman,
1967; Klein, 1983; Mendel et al., 1975; Shallop & Osterhammel, 1983; Sta-
pells, 1984). Moreover, the effects of stimulus rate on the amplitude of the
steady-state MLR are quite similar during wakefulness and sleep. For exam-
ple, although the amplitude of the response is smaller during sleep, the maxi-
mum amplitude is still recorded at stimulus rates between 30-50 tones/s. The
amplitude of the response increases as the intensity of the stimulus increases.
A similar stimulus-intensity/response-amplitudefunction occurs at all stages
of sleep, although the slope of the suprathreshold intensity-amplitude relation
is lower during sleep than waking. Likewise, a similar intensity phase (i.e.,
latency) relation occurs in all stages of sleep and wakefulness, with the phase
of the response decreasing as the suprathreshold intensity of the stimulus in-
creases. Thus, while the steady-stateMLR is sensitive to changes between sleep
and wakefulness, it is little affected by transitions across the different sleep
stages (e.g., Linden, Campbell, Hamel, & Picton, 1985). As such, it holds
promise as a viable technique for generating electrophysiological audiograms
at different frequencies during sleep.
' This N2 and P3 should not be mistaken for the N2 and P3 components that are referred
to in the ERP studies of human information processing in awake adults.
48 MARTA KUTAS
ner & Yingling, 1977). Moreover, Marczynski (1978) argued that it is the cho-
linergic component of the ARAS that is involved in the genesis of surface
negative potentials of this type.
Studies demonstrating that CNVs can be elicited without a motor re-
sponse requirement have shifted the emphasis from response priming to mobi-
lization of effort for anticipated activity, wherein the amplitude of the CNV
indexes the total allowance of effort devoted to preparatory activity (for re-
views see Hillyard, 1973; Tecce, 1972). Most investigationsthus have empha-
sized the relation between the CNV and intensive (e.g., alertness, arousal, con-
centration, effort, expectancy) rather than selective aspects of attention. And
to some extent this seems reasonable. Generally, attentiveness to the eliciting
stimuli yields larger CNVs than distraction, more difficult tasks yield larger
CNVs than easy tasks, and anticipation of noxious or informative stimuli
yields larger CNVs than benign or uninformative stimuli (for a review, see
Donchin, Ritter, & McCallum, 1978). However, the relation is by no means
a simple one: Too difficult a task, too stressful a task, or too many simulta-
neous task requirements often yield a CNV of a diminished amplitude. De-
spite this complex function between CNV amplitude, attentional resources,
and arousal (e.g., Tecce & Hamilton, 1973), it is possible to determine the
relation under specified circumstances during waking and to contrast it with
that obtained under the same conditions during sleep.
One such study was conducted by Salamy, Lester, and Jones (1975). In
this experiment, 3 subjects were asked to respond (i.e., press a microswitch
taped to their right hand) to white noise forewarned by a tone occumng 1 s
ahead on 3 consecutive nights. Recordingswere obtained prior to sleep, during
REM sleep, and immediately after morning awakening. The results suggested
that, although the subjects responded correctly 65% of the time in REM sleep,
a CNV either did not develop or was dramatically attenuated during REM
sleep. However, there are aspects of the data that warrant caution. For in-
stance, it would be informative in interpreting the ERPs to compare the wave-
forms associated with a response against those elicited on no-response trials.
Given the limited number of trials per average (n = 16), however, it is unlikely
that the data could have been sorted on the basis of behavior (e.g., by reaction
time). The noise level of the data presented, especially in the REM condition,
indicates that caution is well-advised. Moreover, as the authors noted but dis-
missed, the reaction times during REM were significantly slower (five- to six-
fold) than when the subjects were awake. While the relation between CNV
amplitude and reaction time is typically not linear, very prolonged reaction
times tend to be associated with very small CNVs. Finally, the recordings were
restricted to two electrode locations, vertex (Cz) and the left occiput, each
referred to the left mastoid. Given that the latter part of a CNV elicited by a
simple, right-handed response typically has a slight contralateral predomi-
nance over the motor strip, recordings from the central scalp locations and
over the left hemisphere might provide a more precise test of the hypothesis.
In any case, a replication of this type of study as well as recordings of CNV in
ERP Studies 49
would be less costly to pursue. If the CNV requires active engagement of atten-
tional mechanisms and this is not possible during sleep, then no CNVs will be
expected. However, we must question whether the concept of attention can
be invoked in the same breath as sleep. If the CNV is truly an indicator of
CNS excitability, one might expect differences in CNV generation and mainte-
nance between REM and NREM sleep. While it would not be surprising to
find that CNVs are smaller during sleep than during waking, it is theoretically
crucial to determine whether there are any significant differences in CNV am-
plitude, morphology, and scalp distribution during the different stages of sleep.
The critical role of muscarinic cholinergic mechanisms both in REM sleep
(e.g., Hobson, 1988)and in the generation of slow negative potentials such as
the CNV (Marczynski, 1978; Pirch, Corbus, Rigdon, & Lyness, 1986) suggest
that, if CNVs are to be recorded during sleep, they should vary across the
stages, especially between REM and NREM. In addition, if the proposed in-
volvements of REM sleep and CNV amplitude in some aspect of memory
have some validity, one might expect differential effects of sleep stages on
CNV generation and amplitude. Certainly, any study of these issues should
attempt to gather the relevant ERP data from the same subjects during the
different sleep stages under the same experimental conditions (difficult as
this may be).
To date, intuition and evidence lead me to argue that the so-called cogni-
tive ERP components require at least a modicum of attention for their elicita-
tion. The most likely candidate as an exception to this rule is the MMN.’
The MMN was first identified and was subsequently studied by Naatanen and
associates (Naatanen, 1986; Naatanen & Gaillard, 1983; Naatanen, Gaillard,
& Mantysalo, 1978; Naatanen, Sams, & Alho, 1986). The MMN is a negative
component beginning around 100 ms after a stimulus change in a repetitive
homogeneous stimulus stream, even when the change is not consciously per-
ceived. That is, if subjects are asked to attend to one ear in a dichotic listening
task and to ignore stimuli in the other ear, tones in the unattended ear that
deviate from the ongoing sequence of tones elicit an MMN. The larger or more
obtrusive the deviation and the less frequently it occurs, the larger the MMN.
Naatanen has proposed that the MMN reflects the activity of an automatic,
preperceptual neuronal mismatch process; namely, the neurophysiological
basis of echoic memory in audition. The MMN has been found to be modality
specific and is presumed to be independent of and insensitive to attentional
manipulations.
The latter presumption was tested in a situation considered to provide
“the most extreme case of attention to environmental stimuli, namely sleep”
(Paavilainen et al., 1987, p. 247). Seven subjects who were instructed to wake
up early in the morning took part in an experiment that same evening, during
which they listened to tones as they read a book; gradually, they fell asleep.
However, I could easily argue that it does not properly belong in the arsenal of cognitive
components.
ERP Studies 51
References
Amadeo, M., & Shagass, C. ( I 973). Brief latency click-evoked potentials during waking
and sleep in man. Psychophysiology, 10,244-250,
Anch, M. ( 1 977). The auditory evoked brain response during adult human sleep. Wak-
ing and Sleeping, 1, 189- 194.
Bell, I., &Campbell, K. (1 988). In T. W. Picton (Ed.), Human event-related potentials.
EEG handbook (Vol. 3, p. 364). Amsterdam: Elsevier.
Besson, M., Fischler, I., Boaz, T. L., & Raney, G. (1989). Effects ofautomaticsemantic
priming on explicit and implicit memory tests. Manuscript submitted for publica-
tion.
Bruneau, N., Martineau, J., Ragazzoni, A., & Roux, S. (1980). Event-related slow po-
tentials evoked during ankle jerk conditioning in wakefulness and NREM sleep.
Electroencephalography and Clinical Neurophysiology, 49,93- 101 .
Buchsbaum, M., Gillin, J. C., & Pfefferbaum, A. (1 975). Effect of sleep stage and stimu-
lus intensity on auditory average evoked responses. Psychophysiology, 12, 707-
712.
Campbell, K. B., & Bartoli, E. A. (1986). Human auditory evoked potentials during
natural sleep: The early components. Electroencephalographyand Clinical Neuro-
physiology, 65, 142-149.
Campbell, K. B., Picton, T. W., Wolfe, R. G., Baribeau-Braun, J., & Braun, C. (198 I).
Clinical evoked potential studies: 11. Auditory brainstem responses. Sensus, I, 9-
19.
Chiappa, K. H., Gladstone, K. J., & Young, R. R. (1979). Brainstem auditory evoked
responses: Studies of waveform variation in 50 normal human subjects. Archives
ofNeurology, 36, 8 1-87.
Coles, M., & Gratton, G. (1986). Cognitive psychophysiology and the study of states
and processes. In E. R. J. Hockey, A. W. K. Gaillard, & M. G. H. Coles (Eds.),
Energetics and human information processing. Dordrecht, The Netherlands: Mar-
tinus Nijhof.
Coles, M. G. H., Gratton, G., Bashore, T. R., Eriksen, C. W., & Donchin, E. (1985). A
psychophysiological investigation of the continuous flow model of human infor-
mation processing. Journal of Experimental Psychology: Human Perception and
Performance, 11, 529-553.
Csepe, V., Karmos, G., & Molnar, M. (1987). Evoked potentials correlates of stimulus
deviance during wakefulness and sleep in cat-animal model of mismatch negativ-
ity. Electroencephalographyand Clinical Neurophysiology, 66, 57 1-578.
de Jong, R., Wierda, M., Mulder, G., & Mulder, L. J. M. (1988). Use ofpartial stimulus
information in response processing. Journal of Experimental Psychology: Human
Perception and Performance, 14, 682-692.
Donchin, E., Ritter, W., & McCallum, W. C. (1978). Cognitive psychophysiology:The
endogenous components of the ERP. In E. Callaway, P. Tueting, & S. H. Koslow
(Eds.), Event-related brain potentials in man (pp. 349-44 1). New York: Academic
Press.
Gaillard, A. W. K., & Ritter, W. (Eds.). (1983). Tutorials in event related potential
research:Endogenous components. Amsterdam: Elsevier/North-Holland.
Galambos, R., & Hecox, K. E. (1978). Clinical applications of the auditory brainstem
response. The Otolaryngologic Clinics of North America, I I , 709-722.
Galambos, R. S., Makeig, S., & Talmachoff, P. J. (1981). A 40-Hz auditory potential
recorded from the human scalp. Proceedings of the National Academy of Sciences,
78,2643-2647.
Goldstein, R., & Rodman, L. (1967). Early components of averaged evoked responses
to rapidly repeated auditory stimuli. Journal of Speech and Hearing Research, 10,
607-705.
Granda, A. M., & Hammack, J. T. (1961). Operant behavior during sleep. Science,
133, 1485-1486.
Gratton, G., Coles, M. G. H., Sirevaag, E., Eriksen, C. W., & Donchin, E. (1988). Pre-
and poststimulus activation of response channels: A psychophysiologicalanalysis.
Journal of Experimental Psychology: Human Perception and Performance, 14,
33 1-344.
Hellekson, C., Allen, A., Greeley, H., Emery, S., & Reeves, A. (1979). Comparison
of inter-wave latencies of brain stem auditory evoked responses in narcoleptics,
primary insomniacs and normal controls. Electroencephalography and Clinical
Neurophysiology, 47, 742-744.
Hillyard, S. A. (1973). The CNV and human behavior. Electroencephalography and
Clinical Neurophysiology, Suppl. 33. Amsterdam: Elsevier.
Hillyard, S. A., & Picton, T. W. (1987). Electrophysiology of cognition. In F. Plum
(Ed.), Handbook ofphysiology, Section I : The nervous system: Vol5. Higherfunc-
tions of the nervous system, Part 2 (pp. 5 19-584). American Physiological Society,
pp. 5 19-584.
Hobson, J. A. (1988). The dreaming brain. New York: Basic Books.
Holender, D. ( 1986). Semantic activation without conscious identification. Behavioral
and Brain Sciences, 9, 1-66.
Johnson, R., Jr. (1988). The amplitude of the P300 component of the event-related
potential: Review and synthesis. In P. K. Ackles, J. R. Jennings, & M. G. H. Coles
(Eds.), Advances in Psychophysiology (Vol. 3, pp. 69-138). Greenwich, C T JAI
Press.
Johnson, R., Jr., Rohrbaugh, J. W., & Parasuraman, R. (Eds.). (1987). Current trends
in event-related potential research, Electroencephalography and Clinical Neuro-
physiology, Suppl. 40. Amsterdam: Elsevier.
Jones, T. A., Stockard, J. J., & Weidner, W. J. (1980). The effects of temperature and
acute alcohol intoxication on the brain stem auditory evoked potential in the cat.
Electroencephalographyand Clinical Neurophysiology, 49,23-30.
Kevanishvili, Z., & von Specht, H. (1979). Human slow auditory evoked potentials
during natural and drug-induced sleep. Electroencephalographyand Clinical Neu-
rophysiology, 47,280-288.
Klein, A. J. (1983). Properties of the brain-stem slow wave component: 1. Latency,
amplitude, and threshold sensitivity.Archives of Otolaryngology, 109, 6- 12.
Kornhuber, H. H., & Deecke, L. (1965). Hirnpotentialanderungen bei Wilkurbeweg-
ungen und passiven Bewegungen des Menschen: Bereitschaftspotential und re-
afferente Potentiale [Changes in brain potential during voluntary and passive
ERP Studies 55
Paavilainen, P., Cammann, R., Alho, K., Reinikainen, K., Sams, M., & Naatanen, R.
(1987). Event-related potentials to pitch change in an auditory stimulus sequence
during sleep. In R. Johnson, Jr., J. W. Rohrbaugh, & R. Parasuraman (Eds.), Cur-
rent trends in event-related potential research (EEG Suppl. 40). Amsterdam: Else-
vier/North Holland.
Picton, T. W., & Hillyard, S. A. (1974). Human auditory evoked potentials: 11. Effects
of attention. Electroencephalographyand Clinical Neurophysiology, 36, 19 1-200.
Picton, T. W., Hillyard, S. A., Krausz, H. I., & Galambos, R. (1974). Human auditory
evoked potentials: I. Evaluation of components. Electroencephalography and
Clinical Neurophysiology, 36, 179- 190.
Picton, T. W., Stapells, D. R., & Campbell, K. B. (198 1). Auditory evoked potentials
from the human cochlea and brainstem. Journal of Otolaryngology, 10, 1-41.
Picton, T. W., Woods, D. L., Baribeau-Braun, J., & Healey, T. M. J. (1977). Evoked
potential audiometry. Journal of Otolaryngology, 6, 90- 1 19.
Pirch, J. H., Corbus, M. J., Rigdon, G. C., & Lyness, W. H. (1986). Generation of
cortical event-related slow potentials in the rat involves nucleus basalis cholinergic
innervation. Electroencephalographyand Clinical Neurophysiology, 63,464-475.
Rechtschaffen, A., & Kales, A. (1968). A manual of standardized terminology, tech-
niques, and scoring system for sleep stages of human subjects. Washington, DC:
U.S. Government Printing Office.
Regan, D. (1972). Evoked potentials in psychology, sensory physiology and clinical
medicine. London: Chapman & Hall.
Regan, D. ( 1989). Human brain electrophysiology: Evokedpotentials and evoked mag-
neticfields in science and medicine. Amsterdam: Elsevier/North Holland.
Rohrbaugh, J. W., & Gaillard, A. W. K. (1983). Sensory and motor aspects of the
contingent negative variation. In A. W. K. Gaillard & W. Ritter (Eds.), Tutorials
in event related potential research: Endogenous components (pp. 269-3 10). Am-
sterdam: Elsevier/North-Holland.
Rugg, M. D. (1985). The effects of semantic priming and word repetition on event-
related potentials. Psychophysiology, 22,642-647.
Rugg, M. D. (1987). Dissociation of semantic priming, word and non-word repetition
by event-related potentials. Quarterly Journal of Experimental Psychology, 3YA,
123-148.
R u g , M. D., Furda, J., & Lorist, M. (1988). The effects of task on the modulation of
event-related potentials by word repetition. Psychophysiology, 25, 55-63.
R u g , M. D., & Nagy, M. E. (1987). Lexical contribution to non-word repetition
effects: Evidence from event-related potentials. Memory and Cognition, 15, 473-
481.
Salamy, J., Lester, J., &Jones, K. ( 1975). REM sleep and contingent negative variation
development. Electroencephalography and Clinical Neurophysiology, 39, 20 1-
204.
Shallop, J. K., & Osterhammel, P. A. (1983). A comparative study of measurements
of SN- 10 and the 40/sec middle latency responses in newborns. Scandinavian Au-
diology, 12,91-95.
Skinner, J. E., & Yingling, C. D. (1977). Central gating mechanisms that regulate
event-related potentials and behavior. In J. Desmedt (Ed.), Progress in clinical
neurophysiology: Attention, voluntary contraction, and event-related cerebral po-
tentials (Vol. l, pp. 30-69). Basel, Switzerland: Karger.
Stapells, D. R. ( 1984). Studies in evoked potential audiometry. Unpublished doctoral
dissertation, University of Ottawa, Canada.
Stapells, D. R., Linden, R. D., Suffield, J. B., Hamel, G., & Picton, T. W. (1984).
Human auditory steady state potentials. Ear and Hearing, 5, 105- 1 13.
Stockard, J. J., Sharbrough, F. W., & Tinker, J. A. (1978). Effects of hypothermia on
the human brainstem auditory response. Annals of Neurology, 3, 368-370.
Stockard, J. J., Stockard, J. E., & Sharbrough, F. W. (1978). Nonpathologic factors
influencing brainstem auditory evoked potentials. American Journal of EEG
Technology, 18, 177-209.
ERP Studies 57
The transfer of learning from the wake to the sleep state has been the subject
of several studies involving conditioned discrimination (e.g., Beh & Barratt,
1 965) and operant control procedures (e.g., Williams, Morlock, & Morlock,
1966). These studies show that stimulus control transfers from waking to
sleep. The degree to which learned behavior can be controlled, however, has
been found to be greatly reduced by sleep. One of the goals of this research was
to investigate factors related to the level of control exerted by stimuli presented
during sleep. An additional goal was to provide further information about the
effects of stimulus control on sleep. Addressing these issues may increase our
understanding of brain organization in sleep and may also lead to procedures
for controlling sleep-related behavioral disorders.
Figure 1 Mean response probability and mean response latency for responses dur-
ing non-REM (NREM) and REM sleep under different test conditions.
avoidance, in which subjects were told to respond to terminate the tone (65
dB of 30-s maximum duration) and that failure to do so would result in a
louder (90 dB) tone; and (d) incremental escape/avoidance, in which subjects
were told to respond to terminate the tone that was initially 45 dB but, in the
absence of a response, increased in 10 dB increments every 10 s up to 95 dB.
The results of these studies for the first night of testing are shown in
Figure 1. Note first the data for the three procedures that involved responding
to a fixed intensity (65 dB) tone. In the instructions only group, the percentage
of responding varied across sleep stages, with responses during sleep Stages 3
and 4 occumng relatively infrequently (see top of Figure 1). Having subjects
respond by terminating the tones (escape group) or avoiding the presentation
of louder tones (escape/avoidance group) resulted in an increased likelihood
of responding. Responding occurred more quickly in the escape/avoidance
group (see bottom of Figure 1). Thus, with these three procedures, responding
to a 65-dB tone varied as a function of the consequences of the responding.
These findings are consistent with those of earlier studies concerning rein-
forcement for responding in sleep (cf. Williams et al., 1966).
The incremental escape/avoidance procedure resulted in the greatest
control over behavior. Responding occurred on 90-100% of the trials in each
stage of sleep. Note that the average response latency with the incremental
procedure was 20-25 s during Stages 2,3, and 4 and during REM sleep. This
means that the tone intensity had increased from 45 dB to 65 dB at the time
of responding, which was the decibel level of the fixed-intensity tone under
the other three procedures. Presumably, the greater behavioral control with
60 JOHN HARSH AND PIETRO BADIA
the incremental procedure, at the same average decibel level as the other pro-
cedures, is attributable to fluctuations in response thresholds during all stages
of sleep.
The data from the four experimental procedures demonstrate that reli-
able control of behavior in all stages of sleep can be obtained with the selective
use of reinforcement contingencies. Significantly, the control was obtained
with stimuli presented at a high rate (every 4-5 min). Although the data in
Figure 1 are from only the first night of testing, reliable responding with the
incremental procedure has been maintained across 10 consecutive test nights
(Badia, Harsh, & Balkin, 1986). Because the incremental procedure proved to
be the most successful in maintaining responding, it was used in subsequent
studies.
EFFECTS ON SLEEP
Earlier investigators concerned with behavioral control in sleep studied the
electroencephalographic (EEG) record at the time of test stimuli to assess
whether subjects had responded while they were asleep or following arousal
from sleep (e.g., Williams et al., 1966). Using EEG alpha as evidence of awak-
ening, it has generally been concluded that subjects are able to respond during
sleep. Responding without alpha activity has frequently been observed in our
studies. However, we approached the question of whether sleep was disturbed
by the behavioral control procedure more generally and included subjective
sleep reports, physiological measures of variables related to sleep, and mea-
sures of daytime sleepiness.
Subject Reports
The morning reports we obtained in early experiments (e.g., Badia et al., 1984;
Badia, Harsh, Balkin, ORourke, & Burton, 1985; Harsh et al., 1987)assessed
subjects’ perceptions of sleep quality and also assessed recall of stimulus/re-
sponse events that occurred in sleep. In these experiments, subjects reliably
responded to stimuli presented on an average of once every 4-8 min. When
asked to report on the quality of their sleep, subjects typically reported that
their sleep was not disturbed or was only minimally disturbed. When asked
how many tones were presented during the night, subjects typically reported
that only 6- 10 tones (up to 100 were actually presented) had been presented.
Thus, although behavior was reliably controlled by frequently presented stim-
uli, there was little awareness of the stimulus presentations and no or minimal
sleep disruption was attributed to the behavioral control procedure.
The results were different when subject reports were obtained by waking
subjectsjust after tone presentations. As part of a study of the relation between
cognitive activity and response variables (see Burton, Harsh, & Badia, 1988),
subjects were awakened 30 s after tones were presented during Stage 2 and
REM sleep. The stimulus presentation procedure in this experiment was tai-
lored to provide reports following both response and no response trials. Addi-
tionally, the tones were presented relatively infrequently (a variable interval
Stimulus Control 61
Figure 2 Percentage of time spent awake in non-REM (NREM) sleep stages and
in REM sleep under control condition and 4-min and 1-min interstimulus interval
experimental conditions.
45-min schedule). As with the morning reports, the results support the conclu-
sion that subjects are capable of responding without the perception of waking.
That is, subjects responded without reporting awakenings on an appreciable
number of trials. However, following approximately 50% of the responses,
subjects reported that they had been awakened by the tones. This contrasts
with the morning recall of less than 10% of the tones. Thus, subject reports
obtained more closely in time to the stimulus/response events were more
likely to indicate perceived wakefulness.
Physiological Variables
Electrophysiological data were recorded in each of our experiments. Thirty-
second epochs of sleep records were scored following conventional guidelines
(Rechtschaffen & Kales, 1968). The results obtained across experiments pro-
vide a consistent picture of the effects of behavioral control on sleep structure.
Thus, when subjects are tested with an average interstimulus interval of 4 rnin
or more, the only effect on structure is a reduction in Stage 4 sleep (Badia et
al., 1984; Badia et al., 1985; Harsh et al., 1987). The observation of minimal
change in sleep structure under these test conditions is in accord with the sub-
jects’ morning reports (described previously) of minimal sleep disturbance.
Data obtained by Magee, Harsh, and Badia (1987) indicate that inter-
stimulus intervals shorter than 4 rnin can result in severe sleep disruption.
Magee et al. presented stimuli after every 4 rnin or after every 1 rnin of sleep.
As in previous studies, under the 4-min condition, sleep structure was mini-
mally disturbed. However, when stimuli were presented after every 1 rnin of
sleep, sleep structure was profoundly altered. Figure 2, which is based on data
62 JOHN HARSH AND PIETRO BADIA
Figure 3 The percentage of trials followed by awakenings, stage shifts, brief changes,
and no changes in the sleep record.
from the Badia et al. (1985), Harsh et al. ( 1987), and Magee et al. (1987) stud-
ies, illustrates this outcome. It appears that, with young adults, sleep structure
is minimally disturbed with interstimulus intervals of 4 min or more but is
greatly disturbed with an interstimulus interval of 1 min. It should be noted
that although older adults are as behaviorally responsive, their sleep is much
more vulnerable to disruption (Harsh, Purvis, Badia, & Magee, 1989).
The recordings at the time of and just after tone presentations were also
examined for evidence of brief electrophysiological changes attributable to
stimulus presentation and/or response emission. Figure 3 presents data on
the brief electrophysiological changes. Included is the percentage of tones and
responses accompanied by awakenings (as scored in 30-s epochs), shifts in
sleep stage, alpha events (<15-sbursts), EEG speeding (< 15-sincrease in EEG
frequency), and increases in electromyograph (EMG) recording. Also pre-
sented is the percentage of trials accompanied by none of the changes (no
change). The data were obtained using the incremental procedure, with tones
occurring every 4 min on the average. Tone intensity began at 45 dB and
increased in 10 dB increments every 10 s up to 95 dB. It can be seen that
awakenings occurred only infrequently, and short bursts of alpha, which are
sometimes taken as a sign of brief awakening or arousal, occurred during or
followingonly 20%of the tones. Thus, responding occurred during polygraph-
ically defined sleep on the large majority of trials.
Although awakenings may have been infrequent, some changes were
found on all but 16%of the trials. The significance of these changes in relation
to the sleep/wake continuum and to overall sleep quality is unclear. It is inter-
esting in this regard that, when tones begin at a higher dB level and increase
more rapidly (see Magee et al., 1987), there is a much greater probability of
brief electrophysiological changes; however, as long as the interstimulus inter-
val is not less than 4 min, there are minimal effects on sleep structure and
daytime sleepiness.
Stimulus Control 63
Daytime Sleepiness
One index of sleep quality is level of daytime sleepiness. Badia et al. (1985)
assessed daytime sleepiness under three test conditions. Under one condition,
stimuli were presented with an average interstimulus interval of 8 min. Under
a second condition, the average interstimulus interval was 4 min. The third
condition was a control condition during which stimuli were not presented
and subjects slept without disturbance. The nighttime test conditions were
continued for 4 consecutive nights, and daytime sleepiness was assessed in
morning and afternoon naps. No evidence was found of differential daytime
sleepiness.
Magee et al. (1987) added to these data. Subjects were tested for 1 night
under two experimental conditions, that is, stimuli were presented after every
4 min of sleep or after every 1 min of sleep. Daytime sleepiness was not in-
creased under the 4-min condition relative to the control condition, replicat-
ing the findings of Badia et al. ( 1985). Thus, presenting signal stimuli as fre-
quently as once every 4 min of sleep is not sufficient to disrupt the process
underlying sleep’s function of minimizing daytime sleepiness. This finding is
consistent with the subjects’ morning reports (described earlier) that their sleep
had been normal or nearly normal. The finding also fits well with the observa-
tion that sleep structure is not substantially altered under these conditions. In
contrast, under the 1-min condition, subjects were profoundly sleepy. The
sleepiness data from this condition are consistent with the finding of a greatly
altered sleep structure. This outcome reinforces the conclusion that there are
limits on the ability to respond to stimuli without disrupting fundamental
processes.
i\
;"
6 10
I
6
- 10 6
T E N T H S OF REM CYCLE
10 6 10
Figure 4 Mean response latencies across consecutive tenths of REM cycles. Means
are based on subjects ( N = 6) with four identifiable REY/NREM cycles during the night.
CONCLUSlON
The findings of this research suggest the followingconclusions. First, with rein-
forcement for responding, control of learned behavior can be maintained reli-
ably by stimuli presented during sleep. Second, when stimuli are presented 4
min or more apart, behavioral control results in little or no change in sleep
structure, in daytime sleepiness, or in perceptions of sleep quality. Neither
perceived wakefulness nor wakefulness as it is scored on the sleep record are
necessary for responding, although stimulus/response events typically result
in brief EEG or EMG change. Third, within-subject, within-night variance in
responsiveness is complexly related to time of night, sleep stage, and REM/
NREM Cycle.
References
Badia, P., Harsh, J., & Balkin, T. (1986). Behavioral control over sleeping respiration
for 10 consecutive nights. Psychophysiology,23,409-4 1 1.
Badia, P., Harsh, J., Balkin, T., Cantrell, P., Klempert, A., O'Rourke, D., & Schoen,
L. (1984). Behavioral control of respiration in sleep. Psychophysiology,21, 494-
500.
66 JOHN HARSH AND PIETRO BADlA
Badia, P., Harsh, J., Balkin, T., ORourke, D., & Burton, S. (1985). Behavioral control
of respiration in sleep and sleepiness due to signal-induced sleep fragmentation.
Psychophysiology, 22, 5 17-524.
Beh, H. C., & Barratt, P.E. H. (1965). Discrimination and conditioning during sleep
as indicated by the electroencephalogram.Science, 147, 1470- 1471.
Burton, S., Harsh, P., & Badia, P. (1988). Cognitive activity in sleep and responsiveness
to external stimuli. Sleep, 11, 6 1-68.
Harsh, J., Badia, P., ORourke, D., & Burton, S. (1983). K-complexes and sleep spin-
dles in relation to behavioral responses elicited during sleep. Psychophysiology,
20,444.
Harsh, J., Badia, P., ORourke, D., Burton, S., Revis, C., & Magee, J. (1987). Factors
related to behavioral control by stimuli presented during sleep. Psychophysiology,
24,528-534.
Harsh, J., Purvis, P., Badia, P., & Magee, J. (in press). Behavioral responsiveness in
sleeping older adults. Biological Psychology.
Harsh, J., Stone, P., Leiker, M., & Badia, P. (1989). Ultradian rhythms in responsive-
ness to stimuli presented during sleep. Manuscript submitted for publication.
Lammers, W., Carpenter, G., Harsh, J., & Badia, P. (1989, June). Responsivity in sleep,
temperature, and time of night. Paper presented at the annual meeting of the Asso-
ciation of Professional Sleep Societies, Washington, DC.
Magee, J., Harsh, J., & Badia, P. (1987). Effects of experimentally induced sleep frag-
mentation on sleep and sleepiness. Psychophysiology, 24, 528-534.
Rechtschaffen, A., & Kales, A. (Eds.) (1968). A manual of standardized terminology,
techniques, and scoring system for sleep stages of human subjects. Los Angeles:
Brain Information Service/Brain Research Institute, University of California at
Los Angeles.
Williams, H. L., Morlock, H. C., & Morlock, J. V. (1966). Instrumental behavior dur-
ing sleep. Psychophysiology, 2,208-2 16.
CHAPTER 3
MEMORIES IN SLEEP:
OLD AND NEW
PIETRO BADIA
the two, the sleep-wake procedure is the one most commonly used. Two vari-
ants of the latter are used. Some researchers test participants when they
awaken in the morning after presenting material repeatedly through the night,
with the focus on the formation of long-term memory (i.e., learning in sleep).
Other researchers awaken the participants immediately or shortly after pre-
senting the material and focus on the duration of short-term memory.
There is a third procedure that relates to the transfer of training and
retrieval issues and not to the formation of new memories. Researchers using
this procedure present information in the waking state and then test for reten-
tion in the sleeping state. With this wake-sleep procedure, the focus is usually
on information processing (i.e., on retrieval in sleep of material learned in the
waking state). The wake-sleep procedure does not assess the formation of new
memories in sleep. It is important to distinguish studies focusing on the re-
trieval of old memories in sleep (information processing) from those focusing
on the formation of new memories in sleep. There is no doubt that informa-
tion processing occurs in sleep; there is some doubt about the formation of
new memories in sleep.
spond when you hear the tone”) that they learned while in the awake state.
They were able to process the information and to respond to the tones, but
they were not able to form the new memory of having detected the tones and
having responded to them.
You may have noted that such findings show a marked similarity to
anterograde amnesia, that is, old experiences can be retrieved but new experi-
ences are not remembered. For instance, this deficit closely resembles the
memory deficit present in people with hippocampal damage. Indeed, later I
will suggest that the basis for “amnesia” in sleep results directly from a hippo-
campal deficit. Let me return to the literature on memory formation and sum-
marize the findings without giving the details of each study.
To an extent, we are all familiar with the basic findings concerning mem-
ory formation (learning) in sleep. This literature is quite consistent. When the
material to be learned or remembered is factual (names, numbers, events,
etc.), the evidence for learning or for memory in sleep is systematically nega-
tive. The conclusions drawn about 30 years ago by Emmons and Simon (1956)
and Simon and Emmons (1956) have been confirmed, elaborated, and
strengthened, with few exceptions (Aarons, 1976; Eich, 1989). In essence, the
conclusions concerning memory formation of factual material relate to the
occurrence of an electroencephalographic (EEG) waking pattern or alpha. In
brief, the probability of learning in sleep is directly related to the duration of
alpha emitted at the time the material is presented.’ The exact durations of
alpha necessary for the formation of long-term memory vary slightly and de-
pend on the delay interval of testing. In general, it is thought that alpha dura-
tions less than 5-10 s at the time material is presented will not permit even
weak learning to occur (Koukkou & Lehmann, 1968). In addition, studies
have indicated that the longer the duration of alpha, the more accurate and
longer the memory (Lehmann & Koukkou, 1974; Shimizu et al., 1977). I
should note, however, that when short-term memory is of primary interest
and subjects are awakened immediately after the information is presented, the
findings are less consistent. Some researchers report short-term memory in the
absence of alpha (Oltman et al., 1977)and others do not (Shimizu et al., 1977).
The generalization regarding alpha responding applies to the formation
of new memories in sleep and does not apply to accessing old memories in
sleep. Subjects can retrieve material already stored in the long-term memory
and process the information in short-term (working) memory in the absence
of alpha.
How have these findings on memory formation in sleep been inter-
preted? One interpretation focuses on state dependency and retrieval pro-
cesses. The argument is that memory is state dependent; if memory occurs in
the state of sleep, it cannot be retrieved in the state of awake. In the previous
example, a retrieval view would argue that the new memory of responding to
the tones in sleep does exist; it simply cannot be retrieved in the waking state.
The state-dependency hypothesis has received considerable support in assess-
ments of drug and nondrug states. However, I should note that the transfer of
memory from wake to sleep, and not from sleep to wake, suggests that state
dependency, if it is operating, is asymmetrical. In addition, other data that I
will note shortly are incompatible with the state dependency view.
One could reject the state-dependency and retrieval view and, as noted,
argue that new memories in sleep are not formed. We could then theorize
about why they are not formed. For example, it could be argued that new
memories cannot be formed in sleep because the duration of short-term mem-
ory is too brief to permit rehearsal. Without rehearsal, transfer to long-term
memory and consolidation cannot occur.
Or one could argue that the link between short-term and long-term
memory is severed in sleep. In a sense, the argument is that the brain in sleep
is organized in a way that does not permit the formation of memories. Further-
more, the functional state of this organization is partially revealed by the pre-
vailing EEG pattern. For memories to be formed, alpha responding (an EEG
awake pattern) must be present during the presentation of the material to be
learned (short-term memory trace). I will discuss this later. Let me now turn
to some positive findings concerning learning in sleep. These findings involve
procedures that require less cognitive loading.
In brief, studies have reported that classical conditioning and habitua-
tion (for humans and nonhumans) occur in the sleeping state and that the
effects carry over to the waking state. These studies demonstrate that learning,
in the absence of alpha, is possible. Interpreting both positive and negative
instances of learning in sleep thus becomes a special challenge for emerging
theories.
CONDITIONING IN SLEEP
Several well-conducted studies have used classical conditioning in humans to
demonstrate learning (memory) in sleep. An elegant study composed of sev-
eral experiments was reported by Beh and Barratt (1965). In their first experi-
ment, they showed that classical conditioning in the waking state transfers to
the sleeping state, a finding that others have since documented. As noted, such
findings relate to the transfer of learning from waking to sleep and not to mem-
ory formation per se. Beh and Barratt’s second study is of greater theoretical
interest and relates directly to memory formation in sleep. They presented
Memories in Sleep 71
interpreting the data on learning and memory in sleep. I should note that the
distinction between declarative and procedural knowledge is also supported
by a rich literature involving brain lesioned animals (Isaacson, 1982). The
distinction first appeared in the literature on artificial intelligence and was later
applied to biological memory by Squire (1987). Other memory distinctions
also exist, but they have not achieved the same biological anchoring (see
Sherry & Schacter, 1987, for an excellent analysis on the topic).
Generally, the distinction between declarative and procedural knowl-
edge is clear. Declarative knowledge is explicit, everyday remembering of fac-
tual content, sentences, words, numbers, events, names, and so forth. The
information is acquired quickly and is verbalized with ease. The acquisition
of declarative knowledge is severely impaired in patients with anterograde am-
nesia and is dependent on the integrity of the medial temporal region.
In contrast, procedural knowledge is implicit, is acquired more slowly, is
not easily verbalized, and is closely tied to procedure. It includes skill learning,
perceptual learning, classical conditioning, habituation, sensitization, and
priming. By its nature, procedural knowledge does not depend on one brain
structure or region and is little, if at all, impaired in patients with anterograde
amnesia.
References
Aarons, L. ( 1976). Sleep assisted instruction. Psychological Bulletin, 83, 1-40.
Badia, P., Harsh, J., Balkin, T., ORourke, D., & Burton, S. (1985). Behavioral control
of respiration in sleep and sleepiness due to signal-induced sleep fragmentation.
Psychophysiology, 22, 5 17-524.
Beh, H. C., & Barratt, P. E. H. (1965). Discrimination and conditioning during sleep
as indicated by the electroencephalogram. Science, 147, 1470- 147 1 .
Berger, T. W., & Orr, W. B. (1983). Hippocampectomy selectivelydisrupts discrimina-
tion reversal learning of the rabbit nictitating membrane response. Behavioral
Brain Research, 8,49-68.
Bonnet, M. (1982). Performance during sleep. In W. Webb (Ed.), Biological rhythms,
sleep and performance (pp. 205-237). New York: Wiley.
Eich, E. (1 989). Learning during sleep. Paper presented at the Arizona Conference on
Sleep and Cognition, Tucson, AZ.
Emmons, W. H., & Simon, C. W. (1956). The non-recall of material presented during
sleep. American Journal of Psychology, 6, 76-8 1 .
Evans, F. J., Gustafson, L. A., OConnell, D. N., Ome, M. T., & Shor, R. E. (1970).
Verbally induced behavioral responses during sleep. Journal of Nervous and Men-
tal Disease, 148,467-476.
Firth, H. (1973). Habituation during sleep. Psychophysiology, 10,43-5 1.
Hobson, J. A., & Schmajuk, N. A. (1988). Brain state and plasticity: An integration of
the reciprocal interaction model of sleep cycle oscillation with attentional models
of hippocampal function. Archives Italiennes de Biologie, 126, 209-224.
Home, J. A. (1988). Why we sleep. New York Oxford University Press.
Isaacson, R. L. (1982). The limbic system. New York: Plenum Press.
Johnson, L. C., Townsend, R. E., & Wilson, M. R. (1 975). Habituation during sleeping
and waking. Psychophysiology, 12, 574-584.
76 PIETRO BADIA
Koukkou, M., & Lehmann, D. (1968). EEG and memory storage in sleep experiments
with humans. Electroencephalography and Clinical Neurophysiology, 25, 455-
462.
Lehmann, D., & Koukkou, M. (1974). Computer analysis of EEG wakefulness-sleep
patterns during learning of novel and familiar sentences. Electroencephalography
and Clinical Neurophysiology, 37, 73-84.
Maho, C., & Bloch, V. (1983). Acquisition of classical conditioning during REM sleep
in rats. Sleep Research, 12, 160.
McDonald, D. G., & Carpenter, F. A. (1975). Habituation of the orienting response in
sleep. Psychophysiology, 12,618-623.
McDonald, D. H., Schicht, W. W., Frazier, R. E., Shallenberger, H. D., & Edwards,
D. J. (1975). Studies of information processing in sleep. Psychophysiology, 12,
624-629.
Oltman, P. K., Goodenough, D. R., Koulack, D., Maclin, E., Schroeder, H. R., &
Flannagan, M. J. (1977). Short-term memory during Stage 2 sleep. Psychophysiol-
Ogy, 14,439-444.
Rickert, E. J., Lorden, J. F., Dawson, R., Jr., Smyly, E., & Callahan, M. F. (1979).
Stimulus processing and stimulus election in rats with hippocampal lesions. Be-
havioral Neural Biology, 27, 454-465.
Sherry, D. F., & Schacter, D. L. (1987). The evolution of multiple memory systems.
Psychological Review, 94,439-454.
Shimizu, A., Takehashi, H., Sumitsiyi, N., Tanaka, M., Yoshida, I., & Kaneko, Z.
(1977). Memory retention of stimulations during REM and NREM stages of
sleep. Electroencephalography and Clinical Neurophysiology, 43, 658-665.
Simon, C. W., & Emmons, W. H. (1956). Responses to material presented during
various levels of sleep. Journal of Experimental Psychology, 51,89-97.
Squire, L. R. (1987). Memory and brain. New York: Oxford University Press.
Winson, J. (1986). Behaviorally dependent neuronal gating in the hippocampus. In
R. L. Isaacson & K. H. Pribram (Eds.), The hippocampus (Vol. 4,pp. 77-92). New
York Plenum Press.
CHAPTER 6
BEHAVIORAL RESPONSES
DURING SLEEP
FREDERICK J. EVANS
Cognitive activity during sleep was virtually ignored prior to the discovery of
the link between REM sleep and dreaming (Aserinsky & Kleitman, 1953).
Freud had rejected hypnotic recall and abreaction in favor of the dream as the
royal road to the unconscious. For Freud (1922), the fact that the dream was
a product of sleep was quite incidental; indeed, it was a nuisance given the
difficulty of eliciting dream recall in some patients.
Historically,cognitive activity during sleep was the basis of the first form
of psychotherapy. In the fourth century BC, on the small Turkish island of
Cos, the Aesculapian temples were the site of a systematic method of healing
centered around suggestions of cure and well-being administered by priests to
the sleeping ill. Collison (1988) likened the procedures of the Aesculapians to
hypnosis rather than sleep. Certainly, an association between hypnosis and
sleep has a long tradition. However, the electroencephalograph(EEG) of hyp-
nosis reveals a waking rather than a sleep state (Evans, 1979b). Nevertheless,
this fascination with possible behavioral and subjective similarities between
sleeping and hypnotic behavior along with our conviction that cognitive activ-
ity during sleep is richer than mere dreaming led a group of us to conduct the
studies to be described.
sleeping partners. Loud noises that are repetitive or familiar are less likely to
wake the sleeping person than soft, strange, novel, or meaningful noises.
Wake/sleep phenomena have been studied in the laboratory. For example,
some people claim that they awaken regularly at a preselected time. Oswald
( 1962)found that, when appropriate waking instructions were given, sleeping
subjects woke up only when specified names of friends were spoken. Condi-
tioned responses and discrimination between auditory stimuli may be elicited
during sleep if a waking response tendency has been established (Beh & Bar-
ratt, 1965; Granda & Hammack, 1961; McDonald, 1966; Weinberg, 1966;
Williams, Morlock, & Morlock, 1966; Zung & Wilson, 1961). Sleep/wake
phenomena, especially sleep learning, has been studied extensively in Russia
and in a few controlled studies in the United States (Simon & Emmons, 1956).
A controversial claim is that hypnoticlike suggestions can be given suc-
cessfully during sleep (Bernheim, 1889; Bertrand, 1826; Fresacher, 1951; Gill
& Brenman, 1959; Schilder & Kauders, 1927). Barber ( 1956)whispered hyp-
noticlike suggestions to 22 subjects who were asleep in their rooms. Some
subjects responded to the suggestions; physiological and EEG criteria were
not used to monitor sleep. Using a sleep/sleep paradigm, Cobb et al. (1965)
documented cognitive activity behaviorally in a pilot EEG study with proce-
dures similar to the one described next.
Procedure
For 2 nights, 19 male student nurses slept in a standard sleep laboratory using
EEG monitoring techniques. During on-line visual diagnosis of alpha-free
REM sleep, suggestions were presented verbally to subjects. Typical sugges-
tions were “Whenever I say the word itch, your nose will feel itchy until you
scratch it”; “Whenever I say the word pillow, your pillow will feel uncomfort-
able until you move it.” The suggestions were tested by saying the cue word
(itch or pillow) once. An attempt was made to test each cue word on at least
two separate occasions during the same REM period in which the suggestion
was given, during all subsequent REM periods that night, and during REM
periods of the second night. The suggestion itself was not readministered on
the second night. Suggestionswere not repeated after their initial presentation.
However, two new suggestions were presented each night whenever possible.
Suggestions were administered and cues were tested only during REM
sleep. At least 120 s of alpha-free REM sleep were required between cue-word
presentations. It was conservatively assumed that visually detected alpha indi-
cates arousal. During the sleep period, the presence of even minimal alpha
(three cycles) was signaled by the technician using a small light outside the line
of sight of the subject. Only subjects who displayed an alpha density exceeding
40% during an eyes closed, waking-rest trial were included in the study.
The subject’s behavior was observed by the experimenter, who was in
the same room. When the subject awakened in the morning, his memory for
the session was tested both directly (during an interview) and indirectly (by
response to cue words in the context of a word association test). Any behav-
Sleep Suggestion 79
ioral or physiological response to the critical cue word was observed. When the
subject awakened after Night 2, a more detailed inquiry was made to evaluate
memory for the sleep events.
The subject was not told before either session that suggestions would be
given, but he was told that sleep cycles were being studied.
Response Rate
During the 2 nights, 416 cue words were presented during sleep, and 89 correct
responses were observed. On the average, the 19 subjects responded to a mean
of 2 1.2%of all cue words; the highest response rate by a subject was 48%.
Response Latency
A successful response tendency was mobilized slowly. The average response
latency was 32 s. Latency increased as the temporal dissociation between the
administration of the suggestion and the cue word increased. For example,
latencies for immediate and cany-over responses were 19 s and 59 s, respec-
tively (p < .001). In contrast, a similar suggestion given during hypnosis or
tested posthypnoticallyproduces a response latency of only a few seconds.
Replication Study
A subsequent study (Perry, Evans, O’Connell, Orne, & Orne, 1978)confirmed
these characteristics, although a lower response rate ( 14%vs. 18%) and even
longer response latencies were found. Three additional findings confirmed
that the motor behavior during sleep was not random but was directly elicited
by the suggestions and their associated cue words: (a) Correct videotaped re-
sponses were discriminated from random body movement behavior by blind
raters; (b) interspersed dummy cue words not associated with suggestionsthat
Sleep Suggestion 81
had been administered did not elicit behavioral responses appropriate to any
suggestion; and (c) if a cue word was presented before the suggestion had been
given, it did not elicit the specific behavioral response that would later be asso-
ciated with the suggestion.
Response Frequency
Hypnotizable subjects were found to awaken following stimulation less fre-
quently than insusceptiblesubjects. Consequently, they slept longer, and more
cues were tested. The correlations between response-rate percentage (which
effectively controls the difference in the frequency of cue administrations) and
both HGSHS:A ( r = .42) and SHSS:C ( r = .39) were of borderline significance
(p < .lo, two-tailed).
with the more hypnotizable subjects in an attempt to reverse the amnesia for
the sleep experiences (Evans, 1979b).
Some subjects responded much more frequently than they had during
the original 2 nights. While the induction of hypnosis and the hypnotic experi-
ence gained since the original nights may have contributed to these increases,
the evidence indicates that neither hypnotic depth nor the interpersonal vari-
ables could account for all of the increases.
Specific presleep suggestions aimed at increasing sleep responsivity may
have been helpful. However, hypnotic and/or set influences did not merely
affect new suggestions. Many responses were obtained to cue words associated
with suggestions that had been neither repeated nor apparently recalled since
the first night 5 months before.
Similar problems of interpretation recurred when attempts were made
to utilize hypnosis, either before or after the sleep session, to obtain recall of
the sleep events. Hypnosis helped, but again the results were not a direct func-
tion of hypnotic depth. Some hitherto unrecalled old suggestionswere recalled
with hypnotic techniques other than regression. This result may indicate that
the techniques originally used to probe morning recall were insufficiently sen-
sitive. On the other hand, the possible timeless or contextless effects of hypno-
sis have been documented by the work on source amnesia (Evans, 1979a; Ev-
ans & Thorn, 1966).
and 4, subjects often recognized letters from these stages. The incidence of
guessing, that is, of incorrectly recalled words of letters, was virtually zero. No
control subject (without a presleep set) recalled any words correctly.
A secondary result is important theoretically. Although none of the
words were presented simultaneously with alpha activity, whenever words pre-
sented during the REM stage were subsequently recalled, transient slower fre-
quency alpha (10.25 Hz vs. 9.64 Hz, p < .01) had been evoked within 30 s
after the presentation of the stimuli during sleep.
A third important finding of this study confirms some of the Russian
claims. Of the 9 subjects, 7 had been administered the HGSHS:A and the SHSS:
c . The correlations between the total (all stages) recall of words and the HGSHS:
A and the SHSS:C were .69 and .42, respectively. The respective rank corre-
lations with REM stage recall were .4 1 and .49.
It would seem that under optimal conditions in the laboratory, sleep
learning of relatively easy material can occur with subsequent waking recall.
The practical significance of these results is uncertain. Hypnosis may play a
role in further exploration of the theoretically exciting but practically limited
phenomenon of sleep learning.
SUMMARY
These studies have documented that meaningful behavioral suggestions can
be administered and responded to during sleep, without necessarily evoking
arousal from sleep. Although these sleep-induced responses are based on both
a subjective experience and a related behavioral response, the repertoire of
cognitive activity shown by some subjects is somewhat primitive, as revealed
by the slow response latency or the simplistic word association material suc-
cessfully learned during unintenupted sleep. The sleep-induced behavior was
not remembered using typical recall paradigms in the subsequent waking state
but, rather, remained available for recovery on later nights, even up to 5
months later, sometimes in a dramatic and transformed fashion. This amnesia
has some similarities to another induced amnesia, posthypnotic amnesia.
Posthypnotic amnesia is a temporary failure to utilize temporal sequence cues
as an organizational strategy (Evans & Kihlstrom, 1973). However, posthyp-
notic amnesia differs from the waking state amnesia of the sleep-induced be-
havior in an important way: Posthypnotic amnesia is easily reversed on cue,
but the sleep-induced response to suggestion is not.
The limit of sleep-induced suggestion is not known because there is no
way to evaluate whether the failure to respond is intentional or is simply due
to a failure of stimulus registration. We have no way of knowing if a subject
actually hears the test cue word (or suggestions) or the target words for sleep
learning unless (a) there is arousal, which disqualifies the material, or (b) there
is a response. The response rates reported must be seen as lower bounds.
The presence of the dissociative sleep response behavior is closely related
to other forms of dissociative behavior, especially responsivity to hypnosis.
The highly hypnotizable person is able to fall asleep easily, to take naps (Evans,
86 FREDERICK J. EVANS
References
Aserinsky, E., & Kleitman, N. (1953). Regularly occurring periods of eye modality,
and concomitant phenomena, during sleep. Science, 118,213-309.
Barber, T. X. (1956). Comparison of suggestibility during “light sleep” and hypnosis.
Science, 124, 405.
Beh, H. C., & Barratt, P. E. H. (1965). Discrimination and conditioning during sleep
as indicated by the electroencephalogram. Science, 147, 1470-147 1.
Bernheim, H. (1889). Suggestive therapeutics:A treatise on the nature and uses of hyp-
nosis. New York: Putnam.
Bertrand, A. J. F. ( 1826). Du magnetisme animal en France et des jugement qu’en ont
portes les societes savantes [On animal magnetism in France and the judgements
of the scientific societies]. Paris: Balliere.
Cobb, J. C., Evans, F. J., Gustafson, L. A., OConnell, D. N., Ome, M. T., & Shor,
E. R. ( 1965). Specificmotor response during sleep to sleepadministered meaning-
ful suggestion: An exploratory investigation. Perceptual and Motor Skills, 20,
629-636.
Collison, D. R. (1988, August). A visit to Cos: In search of origins and transitions.
Presidential address at the 1 I th Triennial Congress of Hypnosis and Psychoso-
matic Medicine, The Hague.
Evans, F. J. ( 1977). The subjectivecharacteristics of sleep efficiency. Journal ofAbnor-
ma1 Psychology, 86, 561-564.
Evans, F. J. ( 1979a). Contextual forgetting: Posthypnotic source amnesia. Journal of
Abnormal Psychology, 88,556-563.
Sleep Suggestion 87
Evans, F. J. (1979b). Hypnosis and sleep: Techniques for exploring cognitive activity
during sleep. In E. Fromm & R. E. Shor (Eds.), Hypnosis: Research developments
andperspective (rev. ed., pp. 139- 183). Chicago: Aldine.
Evans, F. J. (1989). The hypnotizable subject, the hypnotizable patient: Studies in dis-
sociative processes. In A. F. Bennett & K. M. McConkey (Eds.), Cognition in
individual and social processes (Vol. 3). Amsterdam, Holland: Elsevier Science
Publishers, B.V.
Evans, F. J., & Kihlstrom, J. F. (1973). Posthypnotic amnesia as disrupted retrieval.
Journal ofAbnorma1 Psychology, 82,317-323.
Evans, F. J., Gustafson, L. A., OConnell, D. N., Ome, M. T., & Shor, R. E. (1969).
Sleep-induced behavioral response: Relationship to susceptibilityto hypnosis and
laboratory sleep patterns. Journal of Nervous and Mental Disease, 148,467-476.
Evans, F. J., Gustafson, L. A., O’Connell, D. N., Ome, M. T., & Shor, R. E. (1970).
Verbally induced behavioral responses during sleep. Journal of Nervous and Men-
talDisease, 150, 171-187.
Evans, F. J., & Thorn, W. A. F. (1966). Two types of posthypnotic amnesia: Recall
amnesia and source amnesia. International Journal of Clinical and Experimental
Hypnosis, 14, 162-179.
Fresacher, L. A. (195 1). A way into the hypnotic state. British Journal ofMedica1 Hyp-
notism, 3, 12-13.
Freud, S. ( 1922). Introductory lectures on psycho-analysis (J. Riviere, Trans.). London:
Allen & Unwin.
Gill, M. M., & Brenman, M. (1959). Hypnosis and related states: Psychoanalytic stud-
ies in regression. New York: International Universities Press.
Granda, A. M., & Hammack, J. T. (1961). Operant behavior during sleep. Science,
133, 1485-1486.
Hoskovec, J. ( 1966).Hypnopedia in the Soviet Union: A critical review of recent major
experiments. International Journal of Clinical and Experimental Hypnosis, 14,
308-3 15.
McDonald, D. G. ( 1966, January). Conditional and unconditional automatic responses
during sleep (Report No. 65-28, Ad. 48 1520). San Diego, CA: U.S. Navy Medical
Neuropsychiatric Research Unit.
Oswald, I. (1962). Sleeping and waking: Physiology and psychology. New York: El-
sevier.
Perry, C. W., Evans, F. J., OConnell, D. N., Ome, M. T., & Orne, E. C. (1978). Behav-
ioral response to verbal stimuli administered and tested during REM sleep: A
further investigation. Waking and Sleeping, 2, 3 17-329.
Rubin, F. (Ed.). (1968). Current research in hypnopedia. New York: Elsevier.
Schilder, P., & Kauders, 0. (1927). Hypnosis. In S. Rothenberg (Trans.), Nervous and
Mental Disease Monograph Series (No. 46).
Shor, R. E., & Orne, E. C. ( 1962). The Harvard Group Scale of Hypnotic Susceptibility,
Form A. Palo Alto, CA: Consulting Psychologists Press.
Simon, C. W., & Emmons, W. H. (1956). EEG, consciousness,and sleep. Science, 124,
1066- 1069.
Weinberg, H. (1966). Evidence suggesting the acquisition of a simple discrimination
during sleep. Canadian Journal of Psychology, 20, 1- 1 1.
Weitzenhoffer, A. M., & Hilgard, E. R. (1962). The Stanford Hypnotic Susceptibility
Scale, Form C . Palo Alto, CA: Consulting Psychologists Press.
Williams, H. L., Morlock, H. C., & Morlock, J. V. (1966). Instrumental behavior dur-
ing sleep. Psychophysiology, 2, 208-2 16.
Zung, W. W. K., & Wilson, W. P. (1961). Response to auditory stimulation during
sleep: Discrimination and arousal as studied with electroencephalography. Ar-
chives of General Psychology, 4, 548-552.
CHAPTER 7
Is it possible for people to register and retain what is said in their presence
while they sleep? If it is possible, is the learning that takes place during sleep
efficient enough to be of practical as well as theoretical significance?These are
the questions of chief concern in this chapter. To address these issues, research
dealing with a number of factors that may have an important influence on
sleep learning is summarized in the second section of the chapter, while in
the third section, some tentative conclusions concerning the possibility and
practicality of learning during sleep are offered, and prospects for future re-
search are outlined. Much of the material covered in both of these sections
has been culled from a remarkably thorough and trenchant review of the sleep
learning literature by Aarons ( 1976), which I strongly recommend to inter-
ested readers.
As will become apparent in the course of this discussion, solid facts about
sleep learning are scarce, and only one of the variables to be considered-the
level of electroencephalographic(EEG) activation that accompanies or follows
the presentation of a to-be-learned or target item-has been researched in a
rigorous manner. Though the present dearth of reliable data is unfortunate, it
is also understandable.For many years followingthe publication ofthe carefully
controlled EEG experimentsby Emmons and Simon ( 1956; Simon & Emmons,
1956), sleep learning was a dead issue. These investigators demonstrated that
verbal information presented during sleep was irretrievable upon awakening
unless presentation coincided with alpha activity, an EEG indicator of arousal
or wakefulness. Their negative results, in combination with a critical commen-
tary (Simon & Emmons, 1955) on the positive results that had been obtained
by others (e.g., Fox & Robbins, 1952; Leuba & Bateman, 1952), caused most
researchers in the United States and other Western countries to abandon the
idea that people may be able to learn while they sleep.
88
Learning During Sleep 89
In more recent times, however, there has been a modest revival of inter-
est in the possibility of sleep learning, owing to three developments. First, a
number of studies have shown that during slow wave (alpha free) sleep, sub-
jects are able to make complex discriminations between repetitive auditory
signals (e.g., Oswald, Taylor, & Treisman, 1960) and to perform, when cued
with appropriate sensory stimuli, motor responses that they had learned while
awake (e.g., Okuma, Nakamura, Hayashi, & Fujimori, 1966). One implica-
tion of these and related results (see Koulack & Goodenough, 1976; Lehmann
& Koukkou, 1974) is that, even during deep sleep, short-term storage of new
information is possible, as is access to old information in long-term memory.
Second, evidence from several sources (see Firth, 1973; Goodenough, 1978)
suggests that habituation or conditioning of various physiological responses,
such as heart rate and galvanic skin response (GSR), can occur during sleep,
albeit at a slower rate than occurs during wakefulness. Because both habitua-
tion and conditioning represent forms of learning, this evidence implies that
the inability to remember information presented during sleep may be attribut-
able not to problems in storing the information but, rather, to a failure to
retrieve the information on waking (Koukkou & Lehmann, 1983; Koulack &
Goodenough, 1976).Third, there have been numerous reports from the Soviet
Union and other Eastern European nations of success in demonstrating sleep
learning (see Hoskovec, 1966; Rubin, 1968, 1971). Though there can be no
doubt that learning is dramatically impaired during sleep (see Goodenough,
1978), these reports recommend a reappraisal of the conclusion that sleep
learning is impossible and raise a number of interesting questions concerning
the conditions under which learning may occur. It is to these conditions that
I now turn.
Sleep Factors
This article was prepared with the aid of the National Academy of Sciences (American)
and the Natural Sciences and Engineering Research Council (Canadian), and it profited from the
insightful comments of Jennifer Campbell and Damn Lehman.
90 ERIC ElCH
performance and both the level and the duration of EEG wakefulness or acti-
vation patterns that follow item input. Evidence of this sort has been supplied
by a number of studies (e.g., Jus & Jus, 1972; Koukkou & Lehmann, 1968;
Lehmann & Koukkou, 1974; Oltman et al., 1977), one of which is described
next for the purpose of illustration.
In the study by Koukkou and Lehmann (1 968), short sentences were
auditorily presented to subjects during slow wave (Stage 2 or 3) sleep, and the
duration of the EEG activation (alpha) pattern produced by the presentation
of each sentence was measured. Upon awakening, the subjects completed a
test of nominally noncued or “spontaneous” recall, which was succeeded by
a test of old/new sentence-recognition memory.
The results showed that the duration of EEG activation that followed
the presentation of a given sentence was quite short (M = 9 s) for sentences
that were neither recalled nor recognized, was significantly longer (M = 26 s)
for sentences that were recognized but not recalled, and was longer still (M =
165 s) for sentences that were spontaneously recalled verbatim (Koukkou &
Lehmann, 1968, Table IIB). Thus, the postsleep recollection of sentences pre-
sented during slow wave sleep was systematically related to the duration of
EEG activation that occurred after presentation. (In later work, Lehmann and
Koukkou, 1974, demonstrated an analogous correlation between memory
performance and the level [i.e., EEG wave frequency] of postpresentation acti-
vation.) The fact that intermediate durations of activation were associated
with successful recognition but with unsuccessful recall suggests that recogni-
tion may be a more sensitive measure of memory for sleep-presented material
than is spontaneous recall, a point to which I will return later.
In an effort to provide a theoretical rationale for their results, Koukkou
and Lehmann ( 1968) proposed that the duration (and level; see Lehmann &
Koukkou, 1974)of EEG activation that occurs after the presentation of a tar-
get item reflects the time available for the long-term storage of that item. This
proposal is reminiscent of the consolidation interpretation of sleep-learning
problems put forth by Hebb (1949), who theorized that there are two distinct
forms of memorial representation: a short-term store in the form of reverberat-
ing neural circuits and a long-term store involving the development of more
permanent neural “knobs.” According to this account, it is the transformation
or consolidation of information from a short- to a long-term representation
that is curtailed by the absence of EEG activation.
Several observations are compatible with the consolidation account (see
Goodenough, 1978; Lehmann & Koukkou, 1974). For example, somnambu-
lists can carry out complex motor actions and can respond appropriately to
sensory input during very deep (Stage 4) sleep, but they cannot recall their
actions and responses once they awaken (Jacobson, Lales, Lehmann, & Zwei-
zig, 1965). Also, the apparent accuracy of dream recall is high if sleepers are
awakened during Stage 1 periods of REM sleep (a stage characterized by a
fairly active EEG), but without sleep interruption, dream recall decreases with
increased time spent in slow wave sleep after the end of the REM period (De-
Learning During Sleep 91
ment & Kleitman, 1957). And a numerical stimulus presented during deep
sleep that is not followed by appreciable EEG activation can be recalled if the
subject is intentionally and rapidly awakened before the short-term trace of
the stimulus ceases to exist (Oltman et al., 1977).
Though much of the difficulty in recalling events that take place during
sleep may reflect the impaired consolidation or long-term storage of these
events, the possibility that recall difficulties may be due to deficient retrieval
should not be overlooked. Within the last 20 years, several retrieval-based ac-
counts of sleep-learning problems have been advanced (see Foulkes, 1966;
Goodenough, 1978).One of the more recent of these, the functional state-shift
hypothesis of Koukkou and Lehmann (1983), is framed around the concept
of state-specijic memory: the idea that what has been learned in a particular
state of mind or brain is best remembered in that state (see Eich, 1989; Over-
ton, 1984). According to Koukkou and Lehmann (1983), the forgetting of
events that transpire during sleep (whether internally generated dreams or ex-
ternally presented items) is a function of the magnitude of the difference be-
tween the functional (EEG defined) states in which storage and retrieval of
the events take place. Their hypothesis accords well with a number of diverse
findings, including the aforementioned fact that if a transient period of wake-
fulness (as indicated by an increase in EEG activation) occurs soon after the
presentation of a target item, then the subsequent recall of that item will be
possible during full wakefulness. In addition, the state-shift hypothesis cames
the intriguing implication that information acquired during sleep may be ac-
cessible for retrieval during later occasions of sleep, though not during inter-
vening periods of wakefulness. Evidence pertinent to this implication will be
examined shortly. But first, I would like to make one other point concerning
the correlation between EEG activation and memory performance.
As noted earlier, a number of Soviet and Eastern European studies have
reported success in producing reliable, sometimes robust, sleep learning
effects. In these studies, presentations of the material to be learned are not
regulated according to particular EEG patterns (as is customary in Western
studies) but are timed to correspond with sleep onset, initial sleep, and early
morning sleep, which are optimal times for eliciting EEG activations with al-
pha waves (Aarons, 1976). Thus it is possible, indeed probable, that partici-
pants in these studies are not really asleep when presentation occurs but in-
stead are in a rather drowsy, but nonetheless conscious, state. Is it any wonder,
then, that so-called sleep learning is possible under such circumstances? The
obvious answer is no, but there is more to it than that. Unlike their Western
counterparts, Eastern researchers generally do not find the question of whether
subjects are really asleep during presentation of the learning material to be
important or meaningful. Their primary concern is not with the theoretical
possibility of learning during deep sleep, but with the practical purpose it
serves to present learning material to superficially sleeping subjects. This is
one of several salient differences (others will be discussed in due course) that
distinguishesthe prototypical Western study of sleep learning from the proto-
92 ERIC ElCH
Sleep-SpecificMemory
In 19 10, Prince speculated that many people have difficulty remembering
their dreams not because they do not want to remember, as Freud (1900/1953)
and other psychodynamically oriented theorists of the day were claiming, but
rather because they cannot remember, due to the mismatch between the states
of natural sleep and ordinary wakefulness. Intuitively, Prince’s idea seems
plausible, and so does Koukkou and Lehmann’s (1983) hypothesis that fail-
ures of waking memory for experimentally devised materials (such as sen-
tences) presented during sleep are attributable to the shift from sleeping to
waking states. Plausibility is one thing, however, and proof is quite another.
What empirical evidence is there to support the proposition that memory for
events occumng during sleep is specific to the sleep state?
To my knowledge, only one study-described briefly by Evans, Gustaf-
son, OConnell, Orne, and Shor ( 1966) and more elaborately by Evans ( 1972)
and Evans et al. (1969, 1970)-has sought to secure such evidence.
In this study, 18 student nurses slept in a laboratory for 2 or 3 nights.
During the first night, suggestions ( e g , “Whenever I say the word itch, your
nose will feel itchy until you scratch it”) were auditorily presented to subjects
while they were in alpha-free Stage 1 sleep. The suggestions were then tested
immediately by saying a cue word (e.g., itch) and observing the subjects’ be-
havioral responses. Of the 18 subjects tested, 1 1 were able to perform the sug-
gested responses while remaining in Stage 1 sleep. After the subjects awakened,
they did not remember the verbally presented suggestions, nor did they re-
member responding to them. In addition, when presented with the same cue
words that had elicited an appropriate response during sleep, the subjects
did not respond behaviorally when awake. Thus, the subjects appeared to
have a dense waking amnesia for events that had occurred during the prior
night’s sleep.
That the absence of waking memory reflected transient amnesia rather
than permanent forgetting is implied by the observation that, of the 1 1 subjects
who had responded to cue words during the first night, 7 responded to the
same cues during the second night. Thus, in the majority of cases, successful
second-night responding to cue words during sleep occurred even though (a)
the suggestions themselves were not readministered and (b) the subjects had
no intervening waking recollection of the suggestions or of their responses
during sleep.
After an interval of approximately 5 months, 7 subjects were retested on
a third sleep night. None of these subjects remembered the events of either
earlier evening, and 5 of the 7 had responded on both prior nights to the cue
words of the initial night. These 5 subjects responded, while in Stage 1 sleep,
to cue words from the first night’s sleep, despite the fact that the suggestions
had neither been readministered nor recalled in the intervening months.
Learning During Sleep 93
Item Factors
clear empirical evidence to back their claim. Bliznitchenko (1968; also cited
in Aarons, 1976), a pioneer in applied Soviet research on sleep learning, ar-
gued that repeated item presentations in the same sequence are a prerequisite
to improvements in learning during sleep, but he too supplied no solid sup-
porting data.
With respect to the spacing of item presentations, an early experiment
by Coyne, which is described in detail by Simon and Emmons (1955), indi-
cated that distributed repetitions of the material to be learned (number-word
pairs) produced better sleep learning than did massed repetitions. However,
Coyne’s results are hopelessly confounded by the fact that the distributed repe-
titions occurred during the period just preceding wakefulness (typically a light,
drowsy state), while the massed repetitions occurred during deeper and possi-
bly less receptive stages of sleep (see Simon & Emmons, 1955).
The conflicting results revealed by several studies involving nonsense
syllables, common words, simple sentences, and even Chinese-English paired
associates prompted Simon and Emmons ( 1955)to conclude that meaningful-
ness is not a critical determinant of sleep learning. However, as Aarons ( 1976)
commented, in no sleep learning study have the semantic or denotative di-
mensions of the learning material been systematically manipulated. Aarons
also remarked that, apart from whatever role that semantic meaningfulness
may play in sleep learning, the personal meaningfulness or affective signifi-
cance of the learning material may be important. Indeed, hand movements
and electrographic (K complex) responses occur more frequently when sub-
jects are presented, during deep sleep, with their own rather than with some-
one else’s name (Oswald et al., 1960), and emotionally toned words, in con-
trast to neutral items (e.g., dumb, sin vs. drum, sit), provoke more pronounced
eye movements and cardiovascular changes when presented during sleep than
during wakefulness (Minard, Loiselle, Ingledue, & Duatlich, 1968).Given that
personally or affectively meaningful material is more apt to be registered dur-
ing sleep, it would be interesting to know whether such material is also more
likely to be retained upon awakening.
Earlier I pointed out that a salient difference between Western and East-
ern studies of sleep learning is that in the former, material is presented only
during EEG-defined sleep, whereas in the latter, presentation occurs at the
beginning and the end of the normal sleep cycle. Another significant difference
concerns the learner’s familiarity with the material. In most Western studies,
subjects do not know what it is they will hear while they sleep, and they usually
participate in only a single sleep-learningsession. In contrast, Eastern investi-
gators have developed a “hypnopaedic tutorial system” (see Rubin, 1970) in
which the presentation of material during superficial sleep is coordinated with
ongoing audiovisual presleep and postsleep instruction that lasts for several
weeks. Though it has been claimed that this system acceleratesthe learning of
telegraphy, foreign-language vocabulary, and other types of practical knowl-
edge (see Aarons, 1976; Bliznitchenko, 1968; Rubin, 1971), the absence of
appropriate controls makes it impossible to determine how much of the learn-
96 ERIC ElCH
ing is attributable to waking instruction alone. Still, the possibility exists that
the presentation of learning material during sleep improves performance in
the waking state, provided that the material is familiar to the learners prior to
its presentation.
Preliminary support for this possibility has been provided by Tilley
( 1979). Subjects in his experiment examined a set of 20 pictures of common
objects at bedtime. Later that night, a tape-recorded list of 10 words-the
names of half of the pictures in the original set-was presented 10 times during
either Stage 2 or REM sleep. The following morning, the subjects were tested
for free recall and recognition of the complete set of 20 picture names.
In comparison with items that had been studied at bedtime only, those
that had been presented both before and during sleep were significantly better
recalled and recognized. Curiously, the beneficial effect of repetition during
sleep was much more evident in the morning retention of items that had been
repeated during Stage 2 sleep than those that had been repeated during REM.
This finding is curious in that REM, which is characterized by low-amplitude
EEG activity and the periodic appearance of alpha frequencies, would seem
to be more conducive to the processing of incoming information than would
non-REM (in this case, Stage 2) sleep. At any rate, Tilley’s (1979) results are
clearly consistent with the Soviet claim that presleep learning can be strength-
ened or reinforced through within-sleep repetition. It is equally clear, however,
that Tilley’s results need to be replicated and, if possible, extended to other
types of learning materials and retention tasks.
Task Factors
Subject factors
Age
Several authors have speculated about whether the ability to learn during sleep
is dependent on age, but no one has provided any telling data. Svyadoshch
(1962/1968), for example, employed subjects ranging in age from 10 to 60
years in a series of studies concerning the reproduction of stories presented
during sleep. Though Svyadoshch asserted that the majority of his subjects,
regardless of their age, demonstrated a high level of text reproduction (arbi-
trarily defined as 66% or more of the story material), he did not provide a
breakdown of reproduction scores by age group. Svyadoshch also offered no
hard numbers to support a second, seemingly contradictory assertion concern-
ing the relation between sleep learning and age: specifically, that the ability to
assimilate speech during sleep can be acquired artificially by means of sugges-
tions delivered in the context of either deep hypnosis or ordinary wakefulness
and that children and adolescents,being more suggestible by nature than older
adults, are especially adept at developing sleep-learning abilities.
Interestingly, the idea that an optimal period for learning how to learn
during sleep may arise at an early age also occurred to Aarons ( 1976), but for
different and more defensible reasons. These include the observation that (a)
100 ERIC ElCH
even as early as 3 days after birth, the human voice and its fundamental fre-
quency are more effective than other sounds in eliciting behavioral and physi-
ological reactions during alert, relaxed, and somnolent states (Hutt, Hutt, Len-
ard, Bernuth, & Muntjewerff, 1968); (b) children appear to acquire second
languages more readily than do adults, which suggests a greater facility in pho-
netic processing during wakefulness that could conceivably carry over to sleep;
and (c) in comparison with older children, younger children devote more at-
tention to and are more likely to remember auditorily rather than visually
presented information (Hallahan, Kauffman, & Ball, 1974). Though the fore-
going observations are compatible with the developmental hypothesis ad-
vanced by Aarons (1976), more direct evidence is needed.
Health
Given that (a) between 5% and 10% of otherwise healthy medical students
suffer from chronic sleep disturbances that range from mild to moderate in
severity (Johns, Gay, Goodyear, & Masterton, 197I), (b)’emotionalstress dis-
rupts the natural sleep cycles of men and women alike (Breger, Hunter, &
Lane, 197I), and (c) both mentation and physiological processes during sleep
are influenced by menstruation in women (Sheldrake & Cormack, 1974), the
need to screen subjects for sleeplearning research on the basis of specific cri-
teria related to their physical and psychological health seems clear. Yet apart
from the research of Zukhar’ et al. (196511968)in which people with histories
of sleep disturbance were specifically excluded from participation, health-re-
lated variables have not been taken into account in prior studies of sleep learn-
ing. Instead, researchers have simply assumed that their subjects are in gener-
ally good health and have normal hearing. As Aarons (1976) remarked,
information on personal health and sleep habits would help investigators de-
termine the suitability of a particular person to a particular sleep-learning in-
tervention, and it is therefore hoped that the gathering of such information
will become a standard practice in future studies of sleep learning.
nosis, at other times it refers to a strong waking set that is induced in the
subjects to convince them that sleep learning is a bona fide phenomenon, and
on still other occasions it connotes both of these senses. In addition, the evi-
dence these researchers have presented to support their position cannot be
regarded as compelling.
Consider, for example, the work of Kulikov (1 964/1968). Subjects in his
studies included 21 grade school and 15 college students, all of whom were
highly susceptibleto hypnosis (as tested by the method of hand gripping). The
subjects were separated into three groups of 12, each composed of 7 children
and 5 adults; random assignment of subjects to groups was not specifically
noted by the author.
Subjectsin the first group were repeatedly presented during natural sleep
with a narrative (a Tolstoy story for the children; a description of nervous
system functions for the adults) and were tested for recall of the text when they
awoke. These subjects were not, as Kulikov (1964/1968) put it, “prepared”
for sleep learning; that is, they had received no specific suggestions for assimi-
lation and retention of the text prior to its presentation. Kulikov did not spec-
ify the number of times the text was presented, the precise form of the recall
test (i.e., whether it was spontaneous or prompted), or the duration of the
retention interval. Further, it is not clear from Kulikov’s account exactly when
the text was presented; on procedure, he remarked only that the text was pre-
sented, via tape recorder, at a volume that was below the threshold of hearing
in the waking state and that sleep was monitored by activity records (assessing
the absence of motor movements) and pneumographic tracings (assessing the
absence of marked respiratory reaction). Be that as it may, Kulikov found that
only 1 of the 12 subjects in this group had any waking recollection of the text,
and the 1 exceptional subject was a boy who had taken part in previous studies
in which hypnopaedic suggestions had been delivered.
Kulikov began testing of subjects in the second group by establishing
contact with them while they slept. After the subjects had been sleeping for 1
or 2 hours, tape-recorded suggestions were presented (e.g., “You are sleeping
peacefully, do not wake up” and “Your breathing is becoming deeper and
deeper”). Having made contact with the sleeping subjects in this manner, the
suggestion was given, “Now you will hear a story, listen to what is said, try
and memorize it as much as possible, you will remember this all your life, and
whenever wanted you will be able to relate it.” The text was then presented
(an unspecified number of times) and was followed by additional suggestions
to remember the text and to sleep soundly.
The impact these suggestions had on the subjects’ waking-recall perfor-
mance appears to have been profound. Among the 12 subjects in the second
group who had been prepared with a suggested set to learn while asleep, the
percentage of idea units contained in the text that were recalled averaged 64%
and ranged from 47% to 87%;there was no appreciable difference in the per-
formance of the children and the adults. These figures are remarkably similar
to those yielded by the third group of subjects, who were awake at the time of
text presentation (mean recall of 66%, range from 44% to 80%).
Learning During Sleep 103
Three main findings emerged from the recognition test. First, subjects
who had not received the set to learn during sleep recognized none of the
target words from any sleep stage. Second, subjects who had received the set
recognized, on the average, .28, .lo, and .OO of the words that had been pre-
sented during Stages REM, 2, and 4, respectively; none of these subjects ever
claimed to recognize a word that was not a true target. Thus, only those words
that had been presented to suggested-set subjects during REM sleep were rec-
ognized at a better-than-chance level. Third, among the suggested-set subjects,
those who had a relatively high level of hypnotic susceptibility (as indexed by
the Stanford Hypnotic SusceptibilityScale, among other instruments) tended
to recognize more REM stage targets than did subjects who had a relatively
low level (r = .49).
Viewed as a whole, the results of Evans’ (1 972) experiment seem to
square with the Soviet position that sleep learning is possible in hypnotically
susceptible subjects in whom a strong set to learn has been established. As
such, Evans’ results illuminate a number of interesting issues for future re-
search. Consider first the concept of a suggested set. Intuitively, it seems rea-
sonable to suppose that the induction of a set will increase the subjects’ moti-
vation to learn while they sleep. If motivation is indeed one of the keys to
successful sleep learning, then the odds of observing significant sleep-learning
effects should be improved by offering subjects a substantial monetary reward
for good retention performance (e.g., Levy et al., 1972), by ensuring that the
material to be learned during sleep is pertinent to the subjects’ personal needs
or educational goals (e.g., Balkhasov, 1965/1968),or by restricting the subject
sample to individuals who have a strong interest in the research (e.g., Svya-
doshch, 1962/ 1968).
Examination of the role of hypnotic susceptibility in sleep learning, (re-
viewed by Hilgard, 1979)indicates that high hypnotizables are able to process
information outside of conscious awareness more effectively and completely
than are low hypnotizables. A striking example of this “splitting” of conscious-
ness, a process termed dissociation,is shown when a person discovers that he
or she is reacting, in an apparently automatic or involuntary manner, to a
suggestion implanted previously under hypnosis. Owing to their greater dis-
sociative abilities, high hypnotizables may be able to attend and process in-
coming information selectively, without consciousness awareness, after they
have fallen asleep. Lacking this ability, low hypnotizables have to awaken to
process similar information and are therefore incapable of learning while they
sleep. Though this hypothesis is as speculative as it is sketchy, it does seem to
fit with the findings that, in comparison with low hypnotizables, high hypno-
tizables are (a) less likely to awaken either spontaneously or following verbal
stimulation (Evans, 1972), (b) more likely to respond to behavioral sugges-
tions administered during sleep (Evans et al., 1966, 1969), and (c) more adept
at changing their dream experiencesto conform with specific presleep instruc-
tions (Belicki & Bowers, 1982). These findings, in addition to the others men-
tioned earlier in this section, suggest that the relations among hypnotizability,
Learning During Sleep 105
dissociability, and sleep learning comprise an inviting target for future re-
search.
CONCLUSlON
Whether it is possible and practical for people to learn while they sleep is a
question to which Western and Eastern researchers have given different an-
swers. Little, if any, learning has been revealed in most Western studies,
wherein novel verbal material has been presented to randomly selected sub-
jects during a single session of EEG-defined sleep. The learning that has mate-
rialized in these studies has frequently been found to be correlated with both
the duration and the level of EEG wakefulness patterns that coincide with or
closely follow presentation of the learning material. In contrast, evidence of
substantial sleep learning has emerged in numerous Eastern studies, wherein
familiar material was presented to suggestible subjects who had received a
strong presleep instruction set to learn and were willing to participate in a
lengthy training regimen. No attempt has been made in these studies to input
information during deep stages of sleep; instead, presentations have been
timed to correspond with sleep onset, initial sleep, and early morning sleep,
periods in which significant EEG activations are likely to occur. Any improve-
ments in performance obtained under these conditions would thus appear to
reflect a composite of wake and sleep experience and not pure, unadulterated
“sleep learning.”
Although it appears clear that information presented during sleep that is
not accompanied by EEG activation is not retained upon awakening, it would
be interesting to know, for theoretical as well as for applied reasons, whether
there is any substance to the Soviet claim that substantial improvements in
learning can be achieved by a systematic program of combined wake/sleep
instruction. It would also be informative to discover whether such improve-
ments are dependent (a) on the learner’s age, health, capacity to acquire
knowledge in the waking state, susceptibility to hypnosis, and motivation to
learn; (b) on the nature of the learning materials (e.g., whether they are affec-
tively intoned or personally insignificant)and the methods of material presen-
tation (e.g., air- vs. bone-conducted transmission); and (c) on the means by
which retention of the material is measured (e.g., whether the test taps implicit
vs. explicit memory). These are among the many issues that remain to be
settled in future research aimed at investigating both the possibility and the
practicality of learning during sleep.
References
Aarons, L. ( 1976). Sleepassisted instruction. Psychological Bulletin, 83, 1-40.
Adams, J. A. (1980). Learning and memory: An introduction. Homewood, IL: Dorsey
Press.
106 ERIC ElCH
Arkin, A. M., & Antrobus, J. S. ( 1978). The effects of external stimuli applied prior to
and during sleep on sleep experiences. In A. M. Arkin, J. S. Antrobus, & S. J.
Ellman (Eds.), The mind in sleep: Psychology and psychophysiology (pp. 351-
39 1). Hillsdale, NJ: Erlbaum.
Baddeley, A. D. ( 1976). The psychology of memory. New York: Basic Books.
Balkhasov, I. (1968). The rapid teaching of a foreign language by lessons heard during
sleep. In F. Rubin (Ed.), Current research in hypnopaedia (A. Scott, Trans.) (pp.
160-163). New York: Elsevier. (Originally published in 1965)
Bartlett, J. C., & Santrock, J. W. (1979). Affect-dependent episodic memory in young
children. Child Development, 50, 5 13-5 18.
Belicki, K., & Bowers, P. (1982). The role of demand characteristics and hypnotic
ability in dream change following a presleep instruction. Journal of Abnormal
Psychology, 91,426-432.
Bliznitchenko, L. ( 1968). Hypnopaedia and its practice in the USSR. In F. Rubin (Ed.),
Current research in hypnopaedia (pp. 202-209). New York: Elsevier.
Breger, L., Hunter, I., & Lane, R. W. (1971). Theeffectofstress on dreams. New York:
International Universities Press.
Bruce, D. J., Evans, C. R., Fenwick, P. B. C., & Spencer, V. ( 1970). Effects of presenting
novel verbal material during slow-wave sleep. Nature, 225, 873-874.
Dement, W. C., & Meitman, N. (1957). The relation of eye movements during sleep
to dream activity: An objective method for the study of dreaming. Journal of Ex-
perimental Psychology, 53, 339-346.
Eich, E. (1984). Memory for unattended events: Remembering with and without
awareness. Memory and Cognition, 12, 105- 1 1 I .
Eich, E. (1989). Theoretical issues in state dependent memory. In H. L. Roediger 111
& F. I. M. Craik (Eds.), Varieties of memory and consciousness:Essays in honor
of Endel Tulving (pp. 33 1-354). Hillsdale, NJ: Erlbaum.
Eich, J. E., Weingartner, H., Stillman, R. C., & Gillin, J. C. (1975). State-dependent
accessibility of retrieval cues in the retention of a categorized list. Journal of Verbal
Learning and Verbal Behavior, 14,408-4 17.
Elliott, C. R. (1968). Extracts from an experimental study of the retention of auditory
material presented during sleep. In F. Rubin (Ed.), Current research in hypnopae-
dia (pp. 6-27). New York: Elsevier. (Original work published 1947)
Emmons, W. H., & Simon, C. W. (1956). The non-recall of material presented during
sleep. American Journal of Psychology, 69, 76-8 1.
Evans, F. J. (1972). Hypnosis and sleep: Techniques for exploring cognitive activity
during sleep. In E. Fromm & R. E. Shor (Eds.), Hypnosis: Research developments
andperspectives (pp. 43-83). Chicago: AldineJAtherton.
Evans, F. J., Gustafson, L. A., OConnell, D. N., Orne, M. T., & Shor, R. E. (1966).
Response during sleep with intervening amnesia. Science, 152, 666-667.
Evans, F. J., Gustafson, L. A., OConnell, D. N., Orne, M. T., & Shor, R. E. (1969).
Sleepinduced behavioral response: Relationship to susceptibility to hypnosis and
laboratory sleep patterns. Journal of Nervous and Mental Disease, 148,467-476.
Evans, F. J., Gustafson, L. A., OConnell, D. N., Orne, M. T., & Shor, R. E. (1970).
Verbally induced behavioral responses during sleep. Journal of Nervous and Men-
talDisease, 150, 171-187.
Firth, H. (1973). Habituation during sleep. Psychophysiology, 10,43-5 1.
Foulkes, D. (1966). Thepsychology of sleep. New York: Scribners.
Fox, B. H., & Robbins, J. S. (1952). The retention of material presented during sleep.
Journal of Experimental Psychology, 43.75-79.
Freud, S. (1953). The interpretation of dreams. In J. Strachey (Ed. and Trans.), The
standard edition of the complete psychological works of Sigmund Freud (Vol. 4-
5 ) . London: Hogarth Press. (Original work published 1900)
Goodenough, D. R. (1978). Dream recall: History and current status of the field. In
A. M. Arkin, J. S. Antrobus, & S. J. Ellman (Eds.), The mind in sleep: Psychology
and psychophysiology (pp. I 13- 140). Hillsdale, NJ: Erlbaum.
Learning During Sleep 107
Goodwin, D. W., Powell, B., Bremer, D., Hoine, H., & Stem, J. (1969). Alcohol and
recall: State dependent effects in man. Science, 163, 1358- 1360.
Hallahan, D. P., Kauffman, J. M., & Ball, D. W. (1974). Developmental trends in
recall of central and incidental auditory material. Journal of Experimental Child
Psychology, 17,409-42 1.
Hebb, D. 0.( 1949). The organization of behavior: A neuropsychological theory. New
York: Wiley.
Hilgard, E. R. ( 1979). Divided consciousness in hypnosis: The implications of the hid-
den observer. In E. Fromm & R. Shor (Eds.), Hypnosis: Developments in research
and new perspectives. New York: Aldine.
Hoskovec, J. (1966). Hypnopaedia in the Soviet Union: A critical review of recent
major experiments. International Journal of Clinical and Experimental Hypnosis,
14,308-315.
Hutt, S. J., Hutt, C., Lenard, H. G., Bernuth, H., & Muntjewefi, W. J. (1968). Audi-
tory responsivity in the human neonate. Nature, 218, 888-890.
Itil, T. M., Menon, G. N., Bozak, M., & Sangor, A. (1982). The effects of oxiracetam
(ISF 2522) in patients with organic brain syndrome. Drug Development Research,
2,447-46 I .
Jacobson, A., Lales, A., Lehmann, D., & Zweizig, J. (1965). Somnambulism: All night
electroencephalographic studies. Science, 148, 975-977.
Jacoby, L. L. (1982). Knowing and remembering: Some parallels in the behavior of
Korsakoff patients and controls. In L. S. Cermak (Ed.), Memory and amnesia.
Hillsdale, NJ: Erlbaum.
Jacoby, L. L., & Dallas, M. (198 1). On the relationship between autobiographical
memory and perceptual learning. Journal of Experimental Psychology: General,
110,306-340.
Jacoby, L. L., & Witherspoon, D. (1982). Remembering with and without awareness.
Canadian Journal of Psychology, 36, 300-324.
Johns, M. W., Gay, T. J. A., Goodyear, M. D. E., & Masterton, J. P. (1971). Sleep
habits of healthy young adults: Use of sleep questionnaire. British Journal of Pre-
ventative and Social Medicine, 25,236-24 I .
Johnson, M. K., Kahan, T. L., & Raye, C. L. (1984). Dreams and reality monitoring.
Journal of Experimental Psychology: General, 113, 329-344.
Jus, K., & Jus, A. (1972). Experimental studies on memory disturbances in pathologi-
cal and physiological conditions. International Journal of Psychobiology, 2, 205-
208.
Kintsch, W. (1970). Models of free recall and recognition. In D. A. Norman (Ed.),
Models of human memory (pp. 33 1-373). New York: Academic Press.
Koukkou, M., & Lehmann, D. (1968). EEG and memory storage in sleep experiments
with humans. Electroencephalography and Clinical Neurophysiology, 25, 455-
462.
Koukkou, M., & Lehmann, D. (1983). Dreaming: The functional state-shift hypothe-
sis. British Journal of Psychiatry, 142, 221-23 I .
Koulack, D., & Goodenough, D. R. ( 1976). Dream recall and dream recall failure: An
arousal-retrieval model. Psychological Bulletin, 83.97 5-984.
Kulikov, V. N. (1968). The question of hypnopaedia. In F. Rubin (Ed.), Current re-
search in hypnopaedia (A. Scott, Trans.) (pp. 132-144). New York: Elsevier.
(Original work published 1964)
Lehmann, D., & Koukkou, M. (1974). Computer analysis of EEG wakefulness-sleep
patterns during learning of novel and familiar sentences. Electroencephalography
and Clinical Neurophysiology, 37, 73-84.
Leuba, C., & Bateman, D. ( 1 952). Learning during sleep. American Journal ofPsycho1-
Ogy, 65,301-302.
Levy, C. M., Collidge, F. L., & Stabb, L. V. (1972). Paired associate learning during
EEG-defined sleep: A preliminary study. Australian Journal of Psychology, 24,
219-225.
108 ERIC EICH
Minard, J., Loiselle, R., Ingledue, E., & Duatlich, D. ( 1 968). Discriminative electrooc-
ulogram deflections (EGDs) and heart rate (HR) pauses elicited during main-
tained sleep by stimulus significance. Psychophysiology, 5,232.
Okuma, T., Nakamura, K., Hayashi, A., & Fujimori, M. (1966). Psychophysiological
study on the depth of sleep in normal human subjects. Electroencephalography
and Clinical Neurophysiology, 21, 140- 147.
Oltman, P. K., Goodenough, D. R., Koulack, D., M a c h , E., Schroeder, H. R., &
Flanagan, M. J. (1977). Short-term memory during Stage 2 sleep. Psychophysiol-
ogy, 14, 439-444.
Oswald, I., Taylor, A. M., & Treisman, M. ( 1960). Discriminative responses to stimula-
tion during human sleep. Brain, 83. 440-453.
Overton, D. A. (1984). State dependent learning and drug discriminations. In L. L.
Iverson, S. D. Iverson, & S. H. Snyder (Eds.), Handbook ofpsychopharmacology
(Vol. 18, pp. 59-127). New York: Plenum Press.
Peters, R., & McGee, R. ( 1982). Cigarette smoking and state-dependent memory. Psy-
chopharmacology, 76, 232-235.
Prince, M. ( 1 9 10). The mechanism and interpretation of dreams. Journal ofAbnormal
P.yychology, 19, 137- 195.
Richardson-Klavehn, A., & Bjork, R. A. ( 1988). Measures of memory. Annual Review
(?fPsvcholog.v,39, 475-543.
Rubin. F. ( 1968). Current research in hypnopuedia. New York: Elsevier.
Rubin, F. ( 1970). Learning and sleep. Nature, 226,477.
Rubin, F. ( I97 I). Learning and sleep. Bristol, England: Wright.
Schacter, D. L. ( 1987). Implicit memory: History and current status. Journal ofExperi-
mental Psychology: Learning, Memory, and Cognition, 13, 501-5 18.
Schacter, D. L., & Kihlstrom, J. F. (1989). Functional amnesia. In F. Boller & J. Graf-
man (Eds.), Handbook qfneuropsychology. Amsterdam: Elsevier.
Schacter, D. L., & Tulving, E. (1982). Memory, amnesia, and the episodic/semantic
distinction. In R. L. Isaacson & N. E. Spear (Eds.), The expression ofknowledge.
New York: Plenum Press.
Sheldrake, P., & Cormack, M. (1974). Dream recall and the menstrual cycle. Journal
qfPsychosomatic Research, 18,347-350.
Simon, C. W., & Emmons, W. H. (1955). Learning during sleep? Psychological Bulle-
tin, 52. 328-342.
Simon, C. W., & Emmons, W. H. (1956). Responses to material presented during
various levels of sleep. Journal of Experimental Psychology, 5 I , 89-97.
Svyadoshch, A. M. (1968). The assimilation and memorization of speech during natu-
ral sleep. In F. Rubin (Ed.), Current research in hypnopaedia (A. Scott, Trans.)
(pp. 9 1 - I 17). New York: Elsevier. (Original work published 1962)
Tilley, A. J. ( 1979). Sleep learning during Stage 2 and REM sleep. Biological Psychol-
OR^, 9. 155- I6 1.
Watkins, M. J. ( 1979). Engrams as cuegrams and forgetting as cue overload: A cueing
approach to the structure of memory. In C. R. Puff (Ed.), Memory organization
and structure. New York: Academic Press.
Zavalova, N. D., Zukhar, V. P., & Petrov, Y. A. (1968). The question of hypnopaedia
(preliminary communication). In F. Rubin (Ed.), Current research in hypnopae-
dia (A. Scott, Trans.) (pp. 145-151). New York: Elsevier. (Original work pub-
lished 1964)
Zukhar’, V. P., Kaplan, Y. Y., Maksimov, Y. A., & Pushkina, 1. P. (1968). A collective
experiment on hypnopaedia. In F. Rubin (Ed.), Current research in hypnopaedia
(A. Scott, Trans.) (pp. 152- 159). New York: Elsevier. (Original work published
1965)
CHAPTER 8
LUCID DREAMING:
PSYCHOPHYSIOLOGICAL STUDIES OF
CONSCIOUSNESS DURING REM SLEEP
STEPHEN LABERGE
Although we are usually not explicitly aware that we are dreaming while we are
dreaming, at times a remarkable exception occurs, and we become conscious
enough to realize that fact. “Lucid” dreamers (the term derives from van
Eeden, 1913) report being able to remember the circumstances of waking life
freely, to think clearly, and to act deliberately upon reflection, all while experi-
encing a dream world that seems vividly real (Green, 1968; LaBerge, 1985;
Gackenbach & LaBerge, 1988). This contrasts with the usual characterization
of dreams as states that typically evince no reflective awareness or true volition
(Rechtschaffen, 1978).
Lucid dreaming is normally a rare experience. Though most people re-
port having had a lucid dream at least once in their lives, only about 20% of
the population reports having lucid dreams once a month or more (Snyder &
Gackenbach, 1988).
Although most people have experienced lucid dreams, some theoreti-
cians have considered them impossible and even absurd (e.g., Malcolm, 1959).
In the absence of empirical evidence, most sleep researchers have been in-
clined to accept Hartmann’s (1975) “impression” that lucid dreams are “not
typical parts of dreaming thought, but rather brief arousals” (p. 74; see also
Berger, 1977). Schwartz and Lefebvre (1973) noted that frequent transitory
arousals are common during REM sleep and proposed that these “microawak-
enings” are the physiological basis for lucid dream reports. Although no one
has found any evidence for this mechanism, their proposal has been the pre-
dominant opinion (cf. Foulkes, 1974) until the last few years.
109
110 STEPHEN LABERGE
I am grateful to Ronald Hertel, the Institute for Human Development, the Mericos Foun-
dation, and the Reality Foundation for financial support and to Lynne Levitan for editorial
assistance.
Lucid Dreaming 111
nights in the Hull University sleep laboratory while the former monitored the
polygraph. Worsley reported signaling in 8 lucid dreams, all of which were
described by Hearne ( 1978)as having occurred during REM sleep.
However, demonstrations that the signaling of lucid dreams occurs dur-
ing REM sleep raise another kind of question: What exactly do we mean by
the assertion that lucid dreamers are “asleep?” Perhaps these “dreamers” are
not really dreamers, as some have argued in the last century; or perhaps this
“sleep” is not really sleep, as some have argued in this century. How do we
know that lucid dreamers are really asleep when they signal? If we consider
perception of the external world as a criterion of being awake (to the external
world), we can conclude that they are actually asleep (to the external world)
because, although they know they are in the laboratory, this knowledge is a
matter of memory, not perception. Upon awakening, lucid dreamers report
total immersion in the dream world and no sensory contact with the exter-
nal world.
One might object that lucid dreamers are simply not attending to the
environment; rather than being asleep, perhaps they are merely absorbed in
their private fantasy worlds as are those, for example, who are deeply im-
mersed in a novel or daydream. However, according to the reports of lucid
dreamers (LaBerge, 1980b, 1985), if they deliberately attempt to feel the bed-
covers they know they are sleeping in or try to hear the ticking of the clock
they know is beside their bed, they fail to feel or hear anything except what
they find in their dream worlds. Lucid dreamers are conscious of the absence
of sensory input from the external world; therefore, on empirical grounds,
they conclude that they are asleep.
Conversely, if subjects claim to have been awake while showing physio-
logical signs of sleep (or vice versa), we might have cause to doubt their subjec-
tive reports. However, when the subjective accounts and objective physiologi-
cal measures are in clear agreement (as they are here), it is embarrassingly
awkward to assert (as some critics have done) that subjects who report being
certain that they were asleep while showing physiological indications of
unequivocal sleep were actually awake (cf. LaBerge, Nagel, Dement, & Zar-
cone, 1981).
Some critics have suggested that demand characteristics might account
for our results. It is true that.our subjects were under demand to have, signal,
and report lucid dreams, but could demand alone account for their actions if
they were not lucid in the first place? If they merely unconsciously signaled,
we would have found REM periods with signals without subsequent reports
of lucidity, but we did not. If they merely reported having signaled without
actually having done so, we would have found reports without signals, which
we did not. Further, by this account, where would the reported and observed
signals have come from?
The evidence is clear: Lucid dreaming is an experiential and physiologi-
cal reality; though perhaps paradoxical, it is clearly a phenomenon of sleep.
112 STEPHEN LnBERGE
1.0-
0.0 -
1.0-
0.0 -
-l.O-, , , , , , , , , ,: , , , , , , , , , ,
-5 -4 -3 -2 - 1 0 1 2 3 4 5
MINUTES
Figure 1 Grand mean z-scores and standard errors for REM density (EM), respira-
tion rate (RR), heart rate (HR), and scalp skin potential responses (SP) during the 5
min before the onset of lucidity (black bars) and the 5 min after the onset of lucidity
(white bars). Epochs are 30 s in length and the dotted line represents the signaled onset
oflucidity. Sample sizes vary with variable and epoch, but all values are averaged across
lucid dreams and subjects.
Given the finding that lucid dreams reliably occur during activated (pha-
sic) REM, measures of central nervous system activation, such as eye move-
ment density, should contribute something to the pattern of lucid dream dis-
tribution. Because it has been observed that eye movement density starts at a
low level at the beginning of REM periods and increases until it reaches a
peak after approximately 5-7 min (Aserinsky, 197I), we (LaBergeet al., 1986)
hypothesized that lucid dream probability should follow a parallel develop-
ment. Accordingly, we found that mean eye movement density correlated pos-
itively and significantly with lucid dream probability ( r = .66, p < .Ol).
Lucid dreams have most commonly been reported to occur late in the
sleep cycle (Green, 1968). LaBerge et al. (1986) tested this hypothesis by first
determining for each of their 12 subjects the time of night that divided their
total REM time into two equal parts. All but 1 of the subjects had more lucid
dreams in the second half of their REM time than in the first half (binomial
114 STEPHEN LABERGE
:
'Ii
Lucid Dreaming 115
test, p < .O 1). For the combined sample, relative lucidity probability was calcu-
lated for REM Periods 1-6 of the night by dividing the total number of lucid
dreams observed in a given REM period by the corresponding total time in
the REM stage for the same REM period. A regression analysis clearly demon-
strated that relative lucidity probability was a linear function of ordinal REM
period number ( r = .98, p < .0001).
Lucid dreams are initiated in two distinct ways. Subjects usually report
having been in the midst of a dream when a bizarre occurrence causes suffi-
cient reflection to yield the realization that they are dreaming. Less frequently,
subjects report having briefly awakened from a dream and then, falling back
asleep, directly entering the dream with no (or very little) break in conscious-
ness (Green, 1968; LaBerge, 1985). Here is an example of a wake-initiated
lucid dream:
I was lying awake in bed late in the morning listening to the sound of
running water in the adjoining bathroom. Presently an image of the
ocean appeared, dim at first like my usual waking imagery. But its
vividness rapidly increased while, at the same time, the sound of running
water diminished; the intensity of the internal image and external sound
seemed to alter inversely (as if one changed a stereo balance control from
one channel to the other). In a few seconds, I found myself at the seashore
standing between my mother and a girl who seemed somehow familiar. I
could no longer hear the sound of the bath water, but only the roar of the
dream sea. (LaBerge, 1980b, p. 85)
Note that the subject was continuously conscious during the transition
from wakefulness to sleep. This fact suggests that Foulkes (1985) was overstat-
ing the case by claiming that it is “a necessary part of the experience we call
‘sleep’ that we lose a directive and reflective self. You can’t fall asleep, or be
asleep, if your waking self is still regulating and reflecting upon your conscious
mental state” (p. 42).
Because lucid dreams initiated in these two ways ought to differ physio-
logically in at least one respect (i.e., an awakening preceding one but not the
other), the SVLDs were dichotomously classified as either wake-initiated
(WILD) or dream-initiated (DILD), depending on whether or not the reports
>
3I
i
i
5
Lucid Dreaming 117
ing (i.e., the onset of the lucid dream), when she began sexual activity (in the
dream), and when she experienced orgasm.
The subject reported a lucid dream in which she camed out the experi-
mental task exactly as agreed upon. Data analysis revealed a significant corre-
spondence between her subjective report and all but one of the autonomic
measures; during the 15-s orgasm epoch, mean levels for VEMG activity,
VPA, SCL, and respiration rate reached their highest values and were signifi-
cantly elevated compared with means for other REM epochs. Contrary to ex-
pectation, heart rate increased only slightly and nonsignificantly.
of consciousness than the few that are currently distinguished (e.g., dreaming,
sleeping, waking, etc.).
It may be helpful to consider lucidity from a cognitive-developmental
perspective. According to Piaget ( 1926), children pass through three stages of
understanding of the concept dream. In the first stage, they believe that dreams
take place in the same external world as all other experiences. In the second
stage, children treat dreams as if they were partially external and partially in-
ternal. This transitional stage gives way to the third stage in which children
recognize the dream is entirely internal in nature, a purely mental experience.
These foregoing developmental stages refer to how children think about
dreams when they are awake. While asleep and dreaming, children, and also
adults, tend to remain at the first stage, implicitly assuming that the dream
events are external reality. Out-of-body experiences,with a contradictory mix-
ture of material and mental (external and internal), may provide examples of
the second stage (LaBerge, Levitan, Brylowski, & Dement, 1988). In the fully
lucid dream, the dreamer attains the third stage, realizing that the dream world
is distinct from the physical world.
Foulkes (1982, 1985) has emphasized the idea that the growth of the
mind, whether dreaming or awake, shows parallel degrees of development:
“there are ‘stages’ of dream development which individual children reliably
pass through one after the other, and that the precise age at which they reach
a new stage is at least partially predictable from independent measures of their
waking mental development” (1985, p. 137).
In this view, lucid dreaming represents what ought to be a normal ability
in adults. If this is correct, why are lucid dreams so rare, especially in cases
such as nightmares, where lucidity would be extremely helpful and rewarding?
A possible answer may be found by comparing lucid dreaming with another
cognitive skill, language acquisition. All normal adults speak and understand
at least one language. But how many would do so if they were never taught?
Unfortunately, in this culture, with few exceptions, we are not taught to
dream.
LaBerge ( 1 980c) demonstrated that lucid dreaming is a learnable skill
and that there are a variety of techniques available for inducing lucid dreams
(LaBerge, 1985; Price & Cohen, 1988). M e r g e and colleagues have experi-
mented with methods for helping dreamers realize that they are dreaming by
means of external cues applied during REM sleep, which, if incorporated into
dreams, can remind dreamers that they are dreaming (Merge, 1980b). They
have tested a variety of stimuli, including tape recordings of the phrase “this
is a dream” (LaBerge, Owens, Nagel, & Dement, 1981), conditioned tactile
stimuli (Rich, 1985), olfactory stimuli (LaBerge, Brylowski, & Levitan, un-
published data, 1986), and light (LaBerge, Levitan, Rich, & Dement, 1988).
The most promising results so far have resulted from light stimuli.
The psychophysiological studies reviewed here all support the following
picture: During REM dreaming, the events we experience (or seem to experi-
ence) are the results of patterns of CNS activity that produce in turn effects on
our autonomic nervous systems (ANS) and bodies, which are to some extent
Lucid Dreaming 123
modified by the specific conditions of active sleep but are still homomorphic
to the effects that would occur if we were actually to experience the corre-
sponding events while awake.
This conclusion may need further qualification and explanation. Al-
though the events we appear to perceive in dreams are illusory, our feelings in
response to dream content are real. Indeed, most of the events we experience
in dreams are real; when we experience feelings like anxiety or ecstasy in
dreams, we really do feel anxious or ecstatic at the time. When we think in
dreams, we really do think (whether clearly or not is another matter). If we
think in our dreams that Monday comes before Sunday, it is not the case, as
some philosophers (e.g., Malcolm, 1959) assert, that we have only dreamed
we thought; we may have thought incorrectly (to the usual way of thinking),
but we thought nonetheless.
If we vividly imagined a detailed sequence of movements, like walking
around the room, it is probable that motor areas of the brain would be acti-
vated in the same pattern that is activated in actual walking. However, they
would presumably be less activated than when walking. Otherwise, what
would prevent us from actually walking when we imagined doing so?
In REM sleep, a spinal paralysis causes the muscles of locomotion and
vocalization to fail to completely execute the action orders programmed by
the brain. Thus, in REM, unlike the waking state, nothing impedes the brain
from issuing sequences of motor commands at normal levels of activation,
and this probably contributesto the experienced reality of dreamed action.
As for the afferent side of the equation, a great deal of evidence suggests
that imagery uses the same neural systems as perception in the corresponding
sensory mode (e.g., see Farah, 1988; Finke, 1980). In this view, the essential
difference between a perception and a corresponding image is how the identi-
cal neural system acquires sufficient activation to produce a conscious experi-
ence. In the case ofperception, neural excitation (and the resultant experience)
is generated by external input, driving activation of the particular schema to-
be-perceived in a largely bottom-up process. In the case of imagining (likewise,
hallucinating or dreaming), the experienced image is generated internally by
topdown processes activating the appropriate neural network (schema).
Imaginations and perceptions are normally distinguishable by the fact
that images are usually much less vivid than perceptions. Normally, percep-
tions seem real and images seem imaginary. How real something appears de-
pends mainly on its relative vividness, and experienced vividness is probably
a function of the intensity of neural activation. Thus, we may conjecture that
images usually involve a lesser degree of neural activation than the corre-
sponding perceptions, and this results in a lesser degree of experiential reality
for imagination. At least two factors contribute to this state of affairs: One is
that, while we are awake, sensory input produces much higher levels of activa-
tion than imaginary input. Imagination interfereswith perception in the same
modality (Perky, 1910; Segal, 1971), and we may suppose the reverse is true
as well. Another more speculative factor favoring perceptual processes over
imagination in the waking state is the existence of a neural system to inhibit
124 STEPHEN LABERGE
References
Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA: Harvard Uni-
versity Press.
Antrobus, J. S. (1986). Dreaming: Cortical activation and perceptual thresholds. Jour-
nal of Mind and Behavior, 7, 193-2 12.
Antrobus, J. S., Antrobus, J. S., & Fisher, C. (1965). Discrimination of dreaming and
nondreaming sleep. Archives of General Psychiatry, 12, 395-401.
Antrobus, J. S., Dement, W., & Fisher, C. (1964). Patterns of dreaming and dream
recall: An EEG study. Journal of Abnormal and Social Psychology, 69, 244-252.
Arkin, A., Antrobus, J., & Ellman, S. (Eds.). (1978). The mind in sleep. Hillsdale, NJ:
Erlbaum.
Aserinsky, E. (197 1). Rapid eye movement density and pattern in the sleep of young
adults. Psychophysiology, 8, 36 1-375.
Aserinsky, E., & Kleitman, N. (1953). Regularly occumng periods of eye motility and
concomitant phenomena during sleep. Science, 118,273-274.
Berger, R. ( 1977). Psyclosis: The circularity of experience. San Francisco: Freeman.
Brylowski, A., Levitan, L., & LaBerge, S. (1989). H-reflex suppression and autonomic
activation during lucid REM sleep: A case study. Sleep, 12, 374-378.
Dane, J. (1984). An empirical evaluation of two techniquesfor lucid dream induction.
Unpublished doctoral dissertation, Georgia State University.
Dement, W., & Kleitman, N. (1957). Cyclic variations in EEG during sleep and their
relation to eye movements, body motility, and dreaming. Electroencephalography
and Clinical Neurophysiology, 9, 673-690.
Farah, M. J. (1988). Is visual imagery really visual? Overlooked evidence from neuro-
physiology. Psychological Review, 95, 307-3 17.
Fenwick, P., Schatzman, M., Worsley, A., Adams, J., Stone, S., & Baker, A. (1984).
Lucid dreaming: Correspondence between dreamed and actual events in one sub-
ject during REM sleep. Biological Psychology, 18,243-252.
Finke, R. A. (1980). Levels of equivalence in imagery and perception. Psychological
Review, 87, 1 13- 132.
Foulkes, D. (1974). [Review of Schwartz and Lefebvre ( 1973)]. Sleep Research, 3, 1 13.
Foulkes, D. (1981). Dreams and dream research. In W. Koella (Ed.), Sleep 1980 (pp.
246-257). Basel, Switzerland: Karger.
Lucid Dreaming 125
In this chapter, I will (a) draw comparisons between sleep and general anesthe-
sia and (b) present evidence for cognitive processing under general anesthesia.
To those familiar with general anesthesia, it may be puzzling that this topic is
included in a conference devoted to sleep and cognition. Although anesthetists
and surgeons speak of putting the patient “to sleep,” the two states are worlds
apart, both physiologicallyand psychologically.
Before noting these differences, let me explain why cognitive processing
under general anesthesia may be of interest to this audience. First, because it
is difficult, if not impossible, to know whether a sleeper or an anesthetized
patient is phenomenally aware, research in each area provides ideas and meth-
ods that are of mutual benefit. Second, many authors have referred (I believe
mistakenly) to instances of the recall of perioperative events as dreams. Al-
though these instances may or may not be dreams, it is possible that they are
influenced in a manner similar to dream recall. And finally, increasing evi-
dence shows that patients under general anesthesia continue to hear, process,
and respond to auditory input. Responses to this input bear comparison to
those displayed by sleeping subjects.
Orienting
The orienting response can be seen as behavioral, perceptual, or physiological
efferent activity that can be elicited by a particular stimulus and occurs in close
130 LES GOLDMANN
Recognition
Several researchers have examined the recognition of words or phrases played
under anesthesia, with differing results. Dubovsky and Trustman (1976) gave
obstetric patients letter-word pairs to remember under anesthesia (e.g., “G is
for game”). The experimental group did not do significantly better than the
control group on simple correct pairings. However, in a more sophisticated
study, Millar and Watkinson ( 1983) found that when the “false alarm” rate
was taken into account, significant differences occurred on recognition of the
Anesthesia 131
target words. Loftus, Schooler, Loftus, and Glauber (1985) conducted a sim-
ilar (if slightly less rigorous) experiment with only a single, highly knowledge-
able and motivated subject (one of the authors). No significant results were
found on three recognition tests. As an afterthought, the authors tested one of
the doctors present at the operation and found that her recognition scores also
did not differ from chance. The authors concluded that the results “still leave
open the possibility that information presented under anesthesia leaves some
lasting impression that cannot be revealed in tests of retention that require
remembering to be deliberate or intentional.”
Goldmann (1986) gave two groups of patients a preoperative “general
information” questionnaire. This included questions such as “What is the
blood pressure of an octopus?’ The experimental group listened to the an-
swers during anesthesia. Both groups received suggestionsfor an easy recovery
as well as a suggestion of the patient’s own choosing. None of the patients
recalled hearing the answers, but the experimental group performed signifi-
cantly better on a postoperative recognition test. As with the source amnesia
experienced with hypnosis, patients acquired new knowledge but were uncer-
tain about where or how they had learned it. Unlike hypnotic source amnesia,
neither hypnosis nor a posthypnotic cue aided the patients in recalling how
they had learned the new facts.
Recall
Reports of recall in the anesthetic literature are rare. Retrospective studies
have often reported an incidence of recall around 2% (Hutchinson, 1960).
Recall has been associated with the absence of premedication and volatile an-
esthetics, light anesthesia, equipment failures, poor anesthetic technique
(Breckenridge& Aitkenhead, 1983), and highly salient events. For many rea-
sons, cases have been difficult to verify. As with dreaming, patients may feel
as if something happened but have no specific recall of the event.
If evidence of dreaming were based solely on recall, we might conclude
that dreams are rare. However, most people believe that we dream far more
than we can remember and that the failure to recall does not suggest that
dreaming did not occur. The failure is consideredto be one of retrieval. Several
factors associated with dream recall are when and how a subject is awakened;
the setting of the awakening; the motivation, personality, and cognitive style
ofthe subject;the demand characteristicsofthe interview; the rapport between
subject and interviewer; and the characteristicsof the dream itself. The charac-
teristics of the dream include aspects of dream content that relate to salience
and intrapsychic conflict.
If factors that influence the recall of dreams also influence the recall of
awareness under anesthesia, then perioperative events will most likely be re-
called if the patient is awakened quickly after the operation, after hearing
something highly salient (that will not cause intrapsychic conflict) during the
operation, achieves good rapport with the interviewer, and is both motivated
and interested in reporting.
132 LES GOLDMANN
Interviewer: Does anything come to mind that was said to you or about
you in theater [the operating room]?
Patient: No. I am just thinking about my family.
These messages were designed not to stress the patient. Because the patient’s
hearing was not completely occluded, intraoperative events that were of
greater salience may have overshadowed the prerecorded messages.
Finally, in a study designed to maximize recall, the experimenter was
not blind to the intraoperative stimulus. All patients were women, and all were
good hypnotic subjects. The same suggestion was used throughout: “When
you awaken from your anesthetic, you will believe for a moment that you
have green hair.” The statements of 3 patients are suggestive of the taped mes-
sage. Only the excerpt from the first interview occurred under hypnosis.
Patient A
Interviewer: Anything at all. Anything come to mind?
Patient: Well, I dreamt of green strands.
Patient B
Interviewer: Does anything come to mind when you think about your
operation?
Patient: No.
Interviewer: What about now. Anything at all?
Patient: I keep thinking about green things.
Patient C
Interviewer: How are you this morning?
Patient: Pretty good.
Interviewer: Everything went well?
Patient: Great. I’m leaving this morning. I can’t wait to wash my hair.
Even the use of an informed interviewer,good hypnotic subjects, and hypnosis
does not supply incontrovertible evidence for recall. The connections, how-
ever, are intuitive and plausible.
CONCLUSlON
It is clear from the previous evidence that recollection of perioperative events
decreases as the intentionality of retrieval strategies increases. Eich ( 1984)
wrote that “events that occur while a person is asleep, anesthetized, or selec-
tively attending to other ongoing events . . . are rarely revealed in tests of
retention that require remembering to be deliberate or intentional.” Similar
findings have been found with Korsakoff patients (Graf, Squire, & Mandler,
1984). Evidence for learning can be found where otherwise it might not have
been when methods such as spelling, puzzle completion, or word recognition
are used. These methods supply cues and contextual elements. Hypnosis may
enhance remembering by reinstating contextual elements, by allowing the pa-
tient to supply particular affective cues, and by improving rapport. However,
134 LES GOLDMANN
even with the aid of hypnosis, only a tiny fraction of the events that occur
under anesthesia may be available for recall, although they may perhaps be
remembered in another way. It is possible that cognitive processing under gen-
eral anesthesia occurs outside conscious awareness.
Evidence from subliminal perception, blindsight, and dichotic listening
as well as research with amnesic patients (Dixon, 1981;Eich, 1984;Marcel,
1983)suggests that information processing occurs both consciously and non-
consciously. Nonconsciously processed events are not available for recall, al-
though they may influence behavior (Mattis & Kovner, 1984,p. l 18). Conse-
quently, a careless remark by a surgeon may result in a longer hospital stay
or slower healing. Conversely, a patient's recovery may be improved by the
judicious use of suggestions. As with hypnosis, the suggestion may not be re-
called, but its influence on the patient's subsequent behavior may nevertheless
be important and valuable as a tool in the aid of healing.
References
Bennett, H. L., Davis, H. S., &Giannini, J. A. (1985). Nonverbal response to intraoper-
ative conversation. British Journal ofAnaesthesia, 57, 174- 179.
Blacher, R. S. (1975). On awakening paralyzed during surgery: A syndrome of trau-
matic neurosis. Journal of the American Medical Association, 234,67-68.
Bonke, B., Schmitz, P. I. M., Verhage, F., & Zwaveling, A. (1986). Clinical study of
so-called unconscious perception during general anaesthesia. British Journal of
Anaesthesia, 58, 957-964.
Bonnett, 0. T. (1966). Effects of positive suggestions on surgical patients. Pacific Jour-
nal of Medicine and Surgery, 297-300.
Breckenridge, J. L., & Aitkenhead, A. R. (1983). Awareness during anaesthesia: A re-
view. Annals of the Royal College of Surgeons, 65,93-96.
Brice, D. D., Hetherington, R.R., & Utting, J. E. (1970). A simple study of awareness
and dreaming during anaesthesia. British Journal ofAnaesthesia, 42, 535-54 1.
Cheek, D. B. (1962). Importance of recognizing that surgical patients behave as though
hypnotized. American Journal of Clinical Hypnosis, 4,227-236.
Cheek, D. B. (1964). Surgical memory and reaction to careless conversation. American
Journal of Clinical Hypnosis, 6,237-240.
Cheek, D. B. (1979, November). Awareness of meaningful sounds under general anes-
thesia: Considerations and a review of the literature 1959-1979. Paper presented
at the Annual Scientific Meeting of the American Society of Clinical Hypnosis.
Crawford, J. S. (197 1). Awareness during operative obstetrics during general anaesthe-
sia. British Journal ofAnaesthesia, 43, 179- 182.
Dixon, N. ( 198 1). Preconsciousprocessing. New York: Wiley.
Dubovsky, E. L., & Trustman, R. (1976). Absence of recall after general anesthesia.
Anesthesia and Analgesia, 55, 696-67 1.
Eich, E. (1984). Memory for unattended events: Remembering with and without
awareness. Memory and Cognition, 12(2), 105-1 1 1.
Eisele, V., Weinreich, A., & Bartle, S. (1976). Perioperative awareness and recall. Anes-
thesia and Analgesia, 55, 5 13-5 18.
Evans, C., & Richardson, P. H. (1988). Improved recovery and reduced postoperative
stay after therapeutic suggestions during general anaesthesia. Lancet II: 49 1-493.
Anesthesia 135
INSOMNIA:
THE PATIENT AND THE PILL
WALLACE B. MENDELSON
In one sense, we have learned a great deal about patients with chronically
disturbed sleep; in another sense, we know very little. A major advance in the
last decade has been the growing recognition that poor sleep is often a symp-
tom of a series of diagnosable, treatable conditions. Complaints of poor qual-
ity sleep may result, for instance, from sleep apnea, nocturnal myoclonus,
depressive illness, and many other causes (Mendelson, 1987). On the other
hand, there remains a core of patients with chronically disturbed sleep in
whom none of these disorders are found. In this article, I shall use the term
insomniacs to refer to these unhappy individuals. I would like to suggest that
the study of insomniacs leads to knowledge of how hypnotics act and that,
conversely, the study of hypnotics provides insight into the nature of insom-
nia. In summary, I will argue that the clinical effects of hypnotics cannot be
well-understood if they are assessed only in terms of enhancing sleep defined
by the electroencephalogram (EEG). On the contrary, hypnotic efficacy may
involve alterations in the patient’s perception of his or her state of conscious-
ness. Similarly, the development of tolerance may be influenced not only by
pharmacologic issues, but also by a learned process. Before we look at these
issues, however, let us briefly review some general thoughts about insomnia.
Insomnia is a complaint, a symptom and not an illness. In the Associa-
tion of Sleep Disorder Centers nosology (1979), a variety of causes of poor
sleep are listed. After eliminating the many physiological (e.g., sleep apnea),
psychiatric (e.g., depression) and chronobiologic (e.g., phase lag) problems, at
least two major categories remain. The first of these is subjective disorder of
initiating and maintaining sleep (DIMS) without objective findings. Although
the exact terminology for this condition will probably change in forthcoming
139
140 WALLACE B. MENDELSON
revisions, the thought remains the same, that is, that many insomniacs com-
plain strongly about poor sleep while showing minimal disturbances during
polysomnography. Thus, there can be a substantial disparity between patients’
subjective reports of their sleep and objective measures from the same night.
A second form of insomnia is persistent psychophysiological DIMS.
In these patients, internal cues (ruminative worries about not sleeping) and
external cues (the bedroom environment) lead to poor sleep. Their EEG mea-
sures of sleep may be relatively normal or disturbed. In either event, however,
patients’ morning estimates of their sleep are relatively consistent with the
EEG findings. In clinical situations, the distinction between subjective and
persistent psychophysiologicalinsomnia is often unclear, because conditioned
elements may be superimposed on the subjective disorder. Nonetheless, the
basic principle that insomnia may involve a dissociation of subjective experi-
ence from objective measures, as well as conditioned responses, sets the back-
ground for the discussion of hypnotics. Let us now look at sleep studies of
insomnia and at data on possible physiological differences between insomni-
acs and control subjects.
ses of sleep EEGs in insomniacs have not confirmed a decrease in either slow
wave or total EEG power (Mendelson, in press). In summary, EEG sleep stud-
ies of insomnia have found relatively few changes in the sleep of insomniacs.
This may reflect the measures that we use; perhaps more subtle measures
would reveal differences. On the other hand, this may resemble the Sherlock
Holmes story in which a mystery was solved on the basis of the absence of a
dog’s bark. The important point might be the absence of substantial differ-
ences between the EEG sleep of insomniacs and normal subjects.
Insomniacs traditionally complain not only of disturbed sleep but also
of daytime fatigue or sleepiness.An objective measure of daytime wakefulness
may be obtained from the Multiple Sleep Latency Test (MSLT), which records
the mean latency to sleep during four or five 20-min periods in which a patient
is placed in bed in a darkened room (Mendelson, 1987). Interestingly, MSLT
studies in both young and older insomniacs have shown normal daytime
wakefulness (Dement, Seidel, & Carskadon, 1982, 1984; Mendelson et al.,
1984, 1986; Stepanski, Zorick, Roehrs, Young, & Roth, 1988). There are at
least two possible interpretations of these observations. First, given the rela-
tively mild sleep disturbance shown in EEG studies, the normal MSLTs may
confirm that insomniacs are not really sleep deprived. And second, insomni-
acs may indeed suffer from some degree of chronic mild sleep deprivation, but
they may respond differently from normals in terms of how this affects their
daytime wakefulness. At least two observations support this latter view. One
comes from a study by Stepanski, Lamphere, Badia, Zorick, and Roth (1984),
which assessed the relation of nocturnal sleep disturbance and daytime wake-
fulness in patients with sleep apnea, nocturnal myoclonus, and insomnia. In
the first two groups, the number of nocturnal arousals correlated with de-
creased daytime wakefulness. In normal control subjects (who had few arous-
als), there was no significant relation. In the insomniacs, however, the number
of nocturnal arousals was positively associated with increased daytime wake-
fulness. The type of arousal was also critical: Very brief arousals (increased
EEG frequency and electromyograph [EMG] amplitude) were associated
with daytime wakefulness, whereas more substantial awakenings were asso-
ciated with sleepiness. A later study by Stepanski et al. (1988) also found
that shorter sleep times in insomniacs correlated with increased daytime
wakefulness.
A second line of evidence suggesting that insomniacs may respond to
sleep deprivation differently than normal subjects comes from Bonnet ( 1986),
who compared recovery sleep after 64 hours of total sleep loss in older insom-
niacs and control subjects. In both groups, sleep efficiency was very high
(about 97%)on the first recovery night. In the normal subjects, values returned
toward baseline by the second recovery night, whereas for the insomniacs this
did not take place until the fourth night. Thus, the issue of whether insom-
niacs are not sleep deprived or whether they suffer from some form of mild
chronic sleep deprivation to which they respond abnormally remains to
be settled.
142 WALLACE 8. MENDELSON
issue remains open, but it seems possible that slow wave sleep is not necessarily
experienced subjectively as the deepest or most restful sleep.
Studies of the effects of hypnotics on total sleep mirror the data on base-
line sleep among insomniacs: Just as the deficit in total sleep in insomniacs
is small, the enhancement of sleep by hypnotics is also modest. Gillin and
Mendelson ( 198 l), in a review of hypnotic efficacy studies, found that the
mean increase in total sleep time was 35 min. An often-cited study of fluraze-
pam (which at the time was the most widely prescribed hypnotic) reported an
increase in total sleep of only 6-8% over 1 month (Kales, Kales, Bixler, &
Scharf, 1979, with sleep latency significantly decreased only during Nights
1 1 - 1 3. Thirty milligrams of flurazepam given to insomniacs over 5 weeks en-
hanced total sleep by a mean of 2 1 min (Mitler, Seidel, Van Den Hoed, Green-
blatt, & Dement, 1984). Roehrs, Zorick, Kaffeman, Sicklesteel, and Roth
(1982) reported an increase in total sleep of 29 min in a study using 15 mg of
flurazepam. In spite of these relatively modest improvements in EEG-defined
sleep, approximately 20 million prescriptions are written for hypnotics every
year, and another 3%or 4%ofthe population takes over-the-counter sedatives.
Thus, it seems that many people feel they are helped by these agents. Studies
of subjective responses to hypnotics bear this out. Rickels (1983) found that,
in six of seven studies of flurazepam, patients reported that the drug was more
beneficial than a placebo. Significantlyimproved quality of sleep was also sub-
jectively described in five of the seven studies. One possible interpretation is
that insomniacs suffer from a long-term form of sleep deprivation and that
the modest increase of total sleep from hypnotics corrects this problem. An
alternative view is that the subjective benefit of hypnotics is very dispropor-
tionate to the small hypnotic-induced improvement in sleep (in much the
same way that the subjective distress of insomniacs is disproportionately
greater than their modest decrements in sleep). If the latter is the case, several
interesting hypotheses need to be considered:
mobile accidents (Betts & Birtle, 1983; Binnie, 1983;Linnoila, 1978). Indeed,
it could be argued that no clinical hypnotic exists that does not produce cogni-
tive alterations. Although cognitive changes have traditionally been viewed as
side effects, they may be a daytime manifestation of the therapeutic effect, that
is, alterations in the patients’ experiences of whether they are awake or asleep.
A number of studies have shown that benzodiazepinesalter auditory arousal
during sleep. Preliminary evidence suggests that, during forced awakenings 10
min after the first sleep spindle, insomniacs are more likely to report having
been asleep after receiving flurazepam than placebo (Mendelson et al., 1987).
Certainly this area appears to be ripe for exploration.
TOLERANCE TO HYPNOTICS
We have previously described how conditioned responses might result in the
complaint of insomnia, either in persistent psychophysiologicalDIMS or as a
complication to subjective insomnia. Conditioned responses may also play a
role in clinical tolerance to hypnotics. Just as the efficacy of hypnotics must
be measured both physiologically and subjectively,tolerance may be a combi-
nation of traditional pharmacologicissues (i.e., pharmacokineticsand dynam-
ics) and learned responses. Classical conditioning theory could be applied to
suggest that environmental cues associated with drug administration may re-
sult in conditioned responses that are opposite to the unconditioned effect of
the drug (Siegel, 1975; Spielman, Caruso, & Glovinsky, 1987). These “com-
pensatory” responses, which contrast with those resembling the uncondi-
tioned drug response, have been described in a number of studies. King, Bou-
ton, and Musty (1987), for instance, gave the sedative benzodiazepine
midazolam to rats at 48-hr intervals until tolerance was established. When
they gave saline to the rats in the same environment used for drug injection,
animals tested 2 min after administration became hyperactive. Similar com-
pensatory responses have been demonstratedfor motor behavior after chlordi-
azepoxide (File, 1982) and for the hypothermic effect of ethanol (Melchior
& Tabakoff, 1985). These observations have a number of interesting clinical
implications. As Spielman et al. (1987) pointed out, the conditioning theory
of tolerance predicts that the occasional administration of placebo (which is
in effect an extinction trial) would be superior to continuous drug administra-
tion in maintaining efficacy. Occasional placebos might also be more benefi-
cial than drug holidays. The clinical utility of this principle has not yet been
established. However, it exemplifiesonce again the complexity of the relation
between the patient and the pill.
References
Association of Sleep Disorder Centers. ( 1979). Diagnostic classification of sleep and
arousal disorders. Sleep, 2, 1 - 137.
146 WALLACE B. MENDELSON
The ability to measure the onset of sleep reliably is clearly a centraI component
of modern sleep research. Because behavioral quiescence, muscular relax-
ation, alpha drop, respiratory decrease, and auditory threshold increase gener-
ally show strong agreement, sleep onset may be considered to be well-defined.
Unfortunately, subjective reports of sleep onset frequently do not agree well
with physiological measures. It is not unusual for people to report having been
awake when they are awakened from sleep (Anch, McCoy, & Somerest, 1980;
Campbell & Webb, 198 1;Rechtschaffen, 1968). Some patients, usually those
who are depressed, claim to have had no sleep during nights in which physio-
logical indicators suggest that they had several hours of sleep. One of the most
persistent areas of disagreement exists between objective and subjective re-
ports (from patients with reported difficulty falling asleep) of the length of time
it initially takes to fall asleep at night. At least nine studies have documented
median electroencephalograph(EEG) sleep onset latencies of 51 min com-
pared with subjective median estimates of 82 min in such patients (Carskadon
et al., 1976; Frankel, Coursey, Buchbinder, & Snyder, 1976; Hauri & Cohen,
1977; Mendelson, Garnett, Gillin, & Weingartner, 1984; Monroe, 1967;
Nease, Monroe, Bonnet, & Kramer, 198 1;Olmstead, Hauri, Percy, & Hellek-
son, 1980; Spielman et al., 1980; Tietz, Zorick, Roth, & Kaffeman, 1980).
I 1 I I I I I 1
2 4 8 16 32 64 128 256
TIME AFTER SLEEP ONSET (MINUTES)
timed EEG intervals from the appearance of the first sleep spindle rather than
from the point of alpha drop because alpha may decrease for reasons other
than sleep onset (like opening the eyes) and because a substantial proportion
of normal young adults do not produce significant alpha at any time. In the
initial study (Bonnet & Moore, 1982), 12 subjects spent 5 nights each in the
laboratory. The subjectswere awakened at specified intervals ranging from the
first sleep spindle and 1 min after the first spindle to 140 min after the first
spindle. When the subjects were awakened, they were asked if they had been
asleep at that point. If they answered no, they were asked how long it had been
since the technician had told them good night. If they answered yes, they were
asked how long it had taken to fall asleep and how long they had been asleep.
Figure 1 plots the percentage of subjectswho reported themselves to be asleep
as a function of how long they had been asleep when awakened. It can be seen
from the figure that, whereas only 40% of the subjects reported being asleep 1
min after the first sleep spindle, 100%of the subjects reported being asleep 25
min after the first sleep spindle. By classical perceptual criteria, the threshold
of sleep onset (i.e., the point at which 50% of the subjects reported being
This work was supported by the Veterans Administrationand by the Sleepwake Disorders
Research Institute. The author would like to acknowledge the many sleepless nights of Sarah
E. Moore and Ralph Downey 111, whose respective master’s thesis and doctoral dissertation are
described at length in this chapter.
150 MICHAEL H. BONNET
.
.. .
301
0
U
20
l o t
I I 1 I I I I I
2 4 8 16 32 64 128 256
TIME AFTER SLEEP ONSET (MINUTES)
Figure 2 Plot of auditory arousal threshold level in the awake state and after various
intervals of sleep. (The open circle corresponds to the initial spindle awakening. Used
with permission from Bonnet & Moore, 1982.)
asleep) would fall between 4 min and 8 min. Subjects were always awakened
by an audiometer using an ascending method of limits procedure. Figure 2
plots auditory arousal thresholds for the same awakenings displayed in Figure
1. It can be seen that auditory thresholds reached 70% of their stable Stage 2
sleep value after 1 min of sleep. The data reinforce the notion that physiologi-
cal changes such as muscle relaxation and auditory threshold increase occur
rapidly at sleep onset but that cognitive-perceptual changes occur more slowly
and that little subjective certainty of sleep occurs for at least the first 8 min of
sleep, even in normal young adults. These young adults could clearly differen-
tiate longer periods of sleep, and many studies have shown that normal young
adults can estimate their nocturnal sleep latencies with relative accuracy
(Baekeland & Hoy, 197 1; Frankel et al., 1976; Monroe, 1967).
W
I-
la60 1 a Insomniac
0N o r m a l
<
z
I-
u)
w
w
2
!-
V
W
7
m
?
W
2
I-
V
w
-2
m
3
u)
CONDITION
resulting values were used in an analysis of variance (ANOVA) with terms for
group (normal vs. insomniac), place of estimate (bedroom vs. nonbedroom),
and time of test. For the short intervals and long intervals, no significanteffects
were found for group, place, or time of day. For oral temperature, there was
no significant effect for group (F< 1.OO),but there was the expected effect for
time of day, F(4,80) = 15.18, p < .01. As in the first study, it was found that
the insomniacs scored significantly higher on the State Anxiety scale of the
Spielberger State-Trait Anxiety Inventory (Spielberger, Gorusch, and Lu-
shene, 1968). The significant effect was the main effect, F( 1,20) = 8.22, p <
.O1. The anxiety ANOVA effects for place of estimate, for time of day, and for
interactions were not significant. In the afternoon nap attempt, both insom-
niac and normal sleepers estimated relatively well. The insomniacs slightly
overestimated the interval, and the normal sleepers estimated the interval al-
most correctly. No significant difference was shown in the subjective to objec-
tive sleep latency ratio for normal and insomniac sleepers, t(22) = 0.72. As
expected from the other data, no napassociated temperature differenceswere
found, but anxiety again differentiatedthe groups, t(22) = 2.54,p < .01.
Sufficient observations of time estimates, oral temperature measures,
and anxiety scores were available so that Spearman rank-order correlations
could be calculated for each subject across the day. The correlations were then
154 MICHAEL H. BONNET
mates. Sleep latency estimates from this last night were used to determine
changes in perceived sleep latency. Subjects who received the control condi-
tion first received the training condition at least 1 week later. Subjects who
received the treatment condition first could not be scheduled for a control
condition (because they had already been trained to identify the cognitive sur-
round of sleep onset).
To date, 4 subjects have completed both the control and the treatment
conditions, and 1 subject has completed the treatment condition only. The
data indicate that insomniacs can learn to differentiate the three sleep condi-
tions. Subjects improved their correct sleep state differentiation from 46% on
the control week to 93% on treatment weeks. The 93% correct on treatment
weeks implies that a very good discrimination can be made. Of course, the
more important issue is the subjective to objective sleep latency ratio. This
ratio was also decreased. The mean ratio was 4.0 after the final laboratory
night in the control condition and 1.8 after the final laboratory night in the
treatment condition, F( 1,4) = 13.26, p < .05. Although this seems to imply
that patients had greatly improved estimates, the subjective to objective ratio
on the baseline night directly preceding treatment was 2.1, and that was not
significantly different from the 1.8 value following treatment. The numbers
should stabilize with a greater number of subjects.
Of additional interest, subjects were asked to estimate the probability
with which they thought they would be able to fall asleep in less than 30 min
on a given night across the course of the study. Those estimates ranged from
18-28% prior to the treatment night. On the final laboratory night, the esti-
mate was 59%, F( 1,4) = 22.62, p < .01. This finding corresponds to reports
from subjects that they learned to feel more in control of their sleep behavior.
One subject had a long-term ( 5 month) follow-up. That subject reported that
sleep remained improved and that, despite some bad nights, nights were not
all bad as they previously had been. When the subject slept in the sleep labora-
tory once again, she could still identify the three conditions with 85% accu-
racy. In addition, at 5-month follow-up, that subject had a subjective to objec-
tive latency ratio of 1.5 compared with a ratio of 3.3 at the end of the
experiment proper.
SUMMARY
If you ask yourself how long it took to fall asleep last night, you will probably
be faced with a definition based primarily on absence (i.e., “I do not remember
much happening after I went to bed, so I probably fell asleep quickly”). It is
difficult to make accurate retrospective judgments of periods without clear
markers, and it is surprising how well most people can make such judgments.
Our studies have shown that when normal subjects are given clear markers
and immediate judgments, there is still a 4-8 min period around sleep onset
when cognition is changing and determination of sleep state is uncertain. In
most people, this error has little impact. In patients with insomnia, errors in
defining sleep state are much greater, and such errors in perception may form
Sleep Onset 157
the basis for the long-term use of hypnotics or other medication. Thus, the
problem of sleep onset in patients with insomnia is real.
The current work was based on a small and select group of insomniacs
who appeared to suffer primarily from a problem of misperception of sleep
onset. These patients almost certainly had a problem linked to events that
occurred during their sleep onset, and the present data suggest that those
events were probably cognitive. It is likely that a cognitively based treatment,
perhaps the one described here, can be of benefit. However, some cautions are
in order. There is no evidence that the results described herein will apply to
patients with major depression or with a long history of medication use. It is
also unlikely that a cognitive treatment program will be helpful for patients
with disorders that significantly fragment sleep, because each brief arousal
may bring a return of conscious events that may become mixed with the sleep
onset definition. Rather, it is hoped that individuals can begin to apply objec-
tive behavioral methods to diagnose and treat some forms of insomnia and
that these methods will serve as a starting point for the development of behav-
ioral treatments for more difficult insomnias.
References
Anch, A. M., McCoy, J. S., & Somerest,J. S. (1980). The relationship between EEG
alpha activity and perceived awakening. Sleep Research, 9, 22.
Aschoff, J., Geidke,H., Poppel, E., & Wever, R. ( I 972). The influence of sleep interrug
tion and sleep deprivation on circadian rhythms in human performance.In W. P.
Colquhoun (Ed.), Aspects of human efficiency (pp. 135- 150). London: English
University Press.
Baekland, F., & Hoy, P. (197 1). Reported vs. recorded sleep characteristics.Archives
of General Psychiatry, 24, 548-55 1.
Bonnet, M. H., & Moore, S. E. (1982). The threshold of sleep: Perception of sleep as a
functionof time asleep and auditory threshold. Sleep, 5,267-276.
Campbell, S. S., & Webb, W. B. (198 1). The perception of wakefulness within sleep.
Sleep, 4, 177- 183.
Carskadon, M. A., Dement, W. C., Mitler, M. M., Guilleminault,C., Zarcone, V. P.,
& Spiegel, R. (1 976). Self-reportsvs. sleep laboratory findings in 122 drug-free
subjects with complaints of chronic insomnia. American Journal of Psychiatry,
133, 1382-1388.
Downey, R. D., &Bonnet,M. H. (1989, June). Sleep-wakediscrimination in subjective
insomnia improves as a function of sleep onset feedback. Paper presented at the
meeting of the Association of Professional Sleep Societies, Washington, DC.
Falk, J. L., & Bindra, D. (1954). Judgments of time as a function of serial position and
stress. Journal of Experimental Psychology, 47,279-284.
Frankel, B., Coursey, R., Buchbinder, R., & Snyder, F. (1976). Recorded and reported
sleep in chronic primary insomnia. Archives of General Psychiatry, 33, 6 15-623.
Hauri, P., & Cohen, S. (1977). The treatment of insomnia with biofeedback: Final
report of Study 1 . Sleep Research, 6, 136.
Johns, M. W., Masterton, J. P., & Bruce, D. W. (1971). Relationship between sleep
habits, adrenocorticalactivityand personality.Psychosomatic Medicine, 33,499-
508.
158 MICHAEL H. BONNET
Langer, J., Wapner, S., & Werner, H. ( 1 96 1). The effect of danger upon the experience
of time. American Journal of Psychology, 74, 94-97.
Masao, K. (1978). Performance and time of day: Diurnal variations in physiological
and psychological tests. Journal of Child Development, 14, 16-24.
Mendelson, W. B., Garnett, D., Gillin, J. C., & Weingartner, H. (1984). The experience
of insomnia and daytime and nighttime functioning. Psychiatry Research, 12,
235-250.
Monroe, L. J. ( 1967). Psychological and physiological differences between good and
poor sleepers. Journal ofAbnorma1 Psychology, 72,255-264.
Moore, S. E. (198 1). Estimates of sleep latency and of other temporal intervals in insom-
niac and normal sleepers. Unpublished master's thesis, University of Cincinnati.
Moore, S. E., Bonnet, M. H., & Warm, J. S. (1982). Time estimation in normal and
insomniac subjects. Sleep Research, 'I 1, 16 1.
Moore, S. E., Bonnet, M. H., Warm, J. S., & Kramer, M. (1980). Estimates of sleep
latency in the morning and at sleep onset in insomniac and normal subjects. Sleep
Research, 10, 219.
Nease, N. F., Monroe, S. E., Bonnet, M. H., & Kramer, M. (198 1). Perception of sleep
latency across nights and conditions in insomniacs. Sleep Research, 10, 220.
Olmstead, E., Hauri, P., Percy, L., & Hellekson, C. (1 980). Subjective vs. objective
estimation of sleep onset in normal sleepers and in insomniacs. Sleep Research,
9,216.
Pfaff, D. ( 1968). Effects of temperature and time of day on time judgments. Journal of
Experimental Psychology, 76,4 19-422.
Rechtschaffen, A. (1968). Polygraphic aspects of insomnia. In H. Gastaut, E. Lugaresi,
G. Berti, G. Ceroni, & G. Coccagna (Eds.), The abnormalities of sleep in man.
Bologna, Italy: Gaggi.
Spielberger, C. D., Gorusch, R. L., & Lushene (1968). The state-trait anxiety inven-
tory. Preliminary test manual for form X . Tallahassee, FL:Florida State Univer-
sity.
Spielman, A. J., Tannenbaum, R., Adler, J., Saskin, P., Pollak, C. P., RofFwarg, H. P.,
& Weitzman, E. D. (1980). A comparison between objective and subjective sleep
latency in patients with chronic insomnia. Sleep Research, 9,225.
Tietz, E. I., Zorick, F. J., Roth, T., & Kaffeman, M. E. (1980). Objective and subjective
sleep parameters in documented and undocumented insomniacs. Sleep Research.
9, 222.
CHAPTER 1 2
The research and substantive evaluation upon which this manuscript is based were s u p
ported in part by Office of Naval Research Contract N00014-80-C-0380, in part by National
Institute of Mental Health Grants MH-19156 and MH-44193, and in part by a grant from the
Institute for Experimental Psychiatry Research Foundation. I am grateful to Kelly A. Gillen for
help in preparing the manuscript.
Hypnopompic Cognitive Performance 161
Table 1
STUDIES OF PERFORMANCE UPON AWAKENING
Study Task
Table 1 (continued)
Study Task
the DST immediately upon awakening from sleep. Awakening was done audi-
torily by a telephone, which rang continuously for 1 min and, if not answered,
rang on and off every 2 s for another minute. If the subject still did not respond,
his or her name was spoken over the intercom until a verbal response was
elicited. After answering the phone, the subject was asked how much time had
elapsed since he or she had last spoken with the experimenter (immediately
prenap) and was asked to estimate his or her sleepiness on a 10-point analogue
scale. Following this, the subject was instructed to hang up the phone, lie back
down (in the dark), and perform the DST. The polygraph was kept running
throughout awakening performance, which was audio tape-recorded. After
electrode removal, DST performance was again assessed repeatedly.
In one major study (Dinges, 1986; Dinges, Ome, Whitehouse, & Ome,
1987; Dinges, Whitehouse, Ome, & Ome, 1988), the amount of continuous
wakefulness (sleep loss) prior to a 2-hr nap opportunity was varied from 6 hr
to 52 hr. This had the effect of producing marked differencesin nap sleep-stage
infrastructure and amount, which permitted an assessment of hypnopompic
cognition as a function of varying sleep depths or intensities. At the most ex-
treme intensity, sleep inertia was profound, and intrusions of hypnopompic
reverie occurred during the DST at awakening. The following example illus-
trates the power of our experimental protocol to produce a dramatic hypno-
pompic condition that is characterized by social interaction with simultaneous
performance impairment, hypnopompic reverie, misjudgment of sleepiness,
and a dissociation between the electroencephalogram (EEG) and behavior.
Table 2
DIALOGUE BETWEEN SUBJECT AND EXPERIMENTER DURING
PRESLEEP PERFORMANCE OF DESCENDING SUBTRACTION TASK
AND UPON AWAKENING FROM SLEEP
~~ ~
Postsleep Performance
Table 2 (contlnued)
Minute Speaker Content
130 Subject: 648 64. . ah 63. . ummm 500 and ah. . (sigh)
Experimenter: Continue!
Subject: 500 and um . . let’s see. . um . .696 685 640, ah,
um. . . 632 637 637
Experimenter: Continue!
132 Subject: What if people ran faster than normal people run h o m e t h a n
the normal person runs faster than the square root of two
times. . . (mumbles incoherently). . . and normal quote-
unquote people take up two derivatives of normal people in
skin brackets. . .
[l correct, 4 errors]
133 Experimenter: That’s good.
Note: After the presleep test, the subject, who had been awake for 52 hr, took a 2-hr nap. AAer
the nap, the subject did not respond to the telephone bell but picked up the phone when his name
was called.
which the DST was repeatedly performed. The script in Table 2 begins as the
subject is lying on a bed, in a dark quiet room, just before being allowed to
sleep. His presleep (postsleep-loss)DST performance for 3 min of 75 correct
answers and 2 errors was below his presleep-loss levels of 82-99 correct an-
swers and 0-2 errors; this net loss in number completed is characteristicof the
cognitive slowing on subject-paced tasks that typically occurs with sleep loss
(Dinges, 1989b).
Followingthis performance, as is typical for intensely sleepy persons, the
subject fell asleep (Stage I ) within 30 s of being permitted to do so and had his
first epoch of slow wave sleep (SWS)4 min later (Dinges, 1986).He remained
asleep for the entire 2-hr period. He had no REM sleep but accumulated 82.5
min (69%of total sleep time) of SWS (70 min of Stage 4 sleep),which is nearly
as much SWS as healthy young adults of his age acquire in an average 8-hr
nocturnal sleep (cf. R. L. Williams, Karacan, & Hursch, 1974)! He was in
Stage 3 sleep at the time of the awakening bell. Although he did not answer
the awakening telephone call until his name was spoken, which indicates an
intense sleep depth, he interacted immediately thereafter, affirming that he
was awake, estimating the time since he had last spoken to the experimenter,
and providing a rating of his sleepiness.
Curiously, he estimated his sleepiness at a score of 6 at awakening, indi-
cating that he was less sleepy than he had been before the nap (rating of 8).
This was a clear dissociationbetween his self-report and his DST performance,
which was far worse at awakening than it was after 52 hr of wakefulness, just
prior to the nap. We have analyzed enough data to show that this hypnopom-
pic misjudgment of sleepinessis common in subjectswhose sleep has followed
166 DAVID F. DINGES
a sustained period of wakefulness beyond 18 hr, but that it does not occur if
the nap is taken before sleep deprivation (Dinges, 1988). The intensity of the
sleep inertia seems to make it difficult to estimate how sleepy one feels. This
finding is consistent with Sewitch’s (1984) report that the length of continuous
NREM sleep affects the normal sleeper’s perception of having been awake or
asleep, such that awakenings from SWS or after prolonged periods of NREM
sleep most often result in subjects being least able to identify accurately
whether they had been awake or asleep.
Following his time and sleepiness estimates and his interaction with the
experimenter over the phone, the subject was asked to perform the DST. He
was totally unable to execute subtractions correctly (only his repeat of the
starting number was correct; his four subtraction errors are in italics in Table
2), and he could not remember where he was in the sequence, despite two
prods to continue from the experimenter.
He also had considerable difficulty preventing hypnopompic reverie
from intruding into his oral output. The spoken reverie that occurred 2 min
into the awakening DST performance (1 32 min in Table 2) began after the
experimenter prompted him a second time to continue. It is noteworthy that
the apparent evocation of this spoken reverie by the experimenter is consistent
with the observations of Broughton (1 968, 1982) that parasomnic episodes
such as sleep walking and enuresis can be triggered by external stimuli and,
therefore, can be considered disorders of arousal or of partial arousal. In fact,
the awakening performance behavior of this subject is reminiscent of the con-
fusional arousals from SWS described by Broughton (1968), which he sug-
gested were due to “impaired cerebral responsiveness or of functional deaffer-
entation” (p. 1074). The actual reverie of our subject at 2-min postawakening
is illustrative of the semicoherent material we have observed in the awakening
reverie of other subjects. Neologistic phrases such as “normal people in skin
brackets” are referred to, along with phenomenological material appropriate
to the individual such as “the square root of two times” (this subject was a
math major and, indeed, the reverie has the quality of a statement of a mathe-
matical problem).
Interestingly, the subject indicated that he had dreamed during the nap;
this response was made in a questionnaire booklet he completed while elec-
trodes were being removed, 15 rnin after awakening. No dream content was
asked for and, consequently, it is unknown whether the dream related to his
hypnopompic reverie. This was not the only reverie he experienced, however.
Despite coherent social interaction and the completion of performance tasks,
he experienced other reverie intrusions into his cognitive performance oral
output at 6 min and 35 min postawakening. Although reverie was less dra-
matic in many other subjects to the extent that it lasted only a few seconds
rather than a full minute, it was apparent in most who had undergone inten-
sive sleep loss in the form of one- to five-word intrusions into oral performance
output. Such intrusions were most common in the hypnopompic period.
Hypnopompic Cognitive Performance 167
Figure 7 Vertex (Cz-AIA2) EEG recordings from a healthy young adult with eyes
closed lying on a bed in a dark quiet bedroom, taken at eight different times. Record
(A) after 6 hr ofwakefulness-instructed to lie awake quietly. (B) after 52 hr ofwakeful-
ness-minute 1 after being instructed to lie awake quietly. (C) 2 rnin after B-still
attempting to-stay awake. (D) 1 rnin after C and after being instructed to go to sleep.
(E) 2 min after D (asleep by polysomnographic criteria). (F) after 2-hr nap-minute 1
of DST performance at awakening (see text). (G) 2 rnin after F-during reverie of final
minute on DST (see Table 2). (H) 2 rnin after completion of DST at awakening-
instructed to remain awake, but S fell asleep.
from the sleepdeprivation period show increased slow wave activity. Of par-
ticular interest is the comparison between Records C and G. Although no
EEG recording was made during the presleep DST performance, Record C (at
52 hr and 2 rnin of sleep loss) was obtained 2 rnin into the 3-min presleep
wake, eyes-closed baseline (i.e., 1 rnin before the subject was allowed to sleep).
Record G was obtained 2 rnin into the postsleep awakening DST perfor-
mance, when the reverie at the bottom of Table 2 was elicited. There is no
obvious difference between these records, and they are clearly different from
the high-amplitude, slow-frequencywaveforms apparent in the sleep records
at 1 rnin and 2 rnin of nap onset (Records D-E) and when the subject fell
asleep after completing the awakening DST (Record H). Despite the lack of
EEG differences between Records C and G, differences at these times in DST
performance and cognitive coherency are profound (Table 2).
tude of sleep inertia effects during the hypnopompic period. The simplest way
to manipulate sleep pressure is through prior wakefulness. But which physio-
logical process underlies sleep pressure and correlates with manifestations of
sleep depth, especially pressure for SWS? Although biochemical changes must
underlie any such process, and candidates have been proposed (e.g., see Chap-
ter 2 by Hobson in this volume), there is reason to suggest that body core
temperature and, presumably, brain metabolic activity may be an essential
link in the hypnopompic process. Body core temperature shows a circadian
rhythm, but sleep (especially the first SWS period of the night) also has the
(evoked) effect of thermal down-regulation (Gillberg & Akerstedt, 1982).
There is evidence that this effect may be enhanced whenever the pressure for
sleep is increased. Aschoff, Giedke, Poppel, and Wever ( 1972) observed that
“after two days without sleep, one night of uninterrupted sleep results in an
exaggerated drop in rectal temperature” (p. 144). A significant covariation
between oral temperature and performance upon awakening has been re-
ported (Rosa & Bonnet, 1985).But the relation between SWS and temperature
may go both ways (Sewitch, 1987). Experimentally induced lowering of rectal
temperature at sleep onset has been associated with an increase in Stage 4 sleep
and a lengthening of the first NREM/REM cycle (Sewitch, Kittrell, Kupfer, &
Reynolds, 1986). Passive afternoon body heating has been observed to result
in an enhanced drop in rectal temperature during the first 2-4 hr of subse-
quent nocturnal sleep (Horne & Staff, 1983).
One possible mechanism underlying the increasingly impaired perfor-
mance at awakening of subjects who have experienced intense pressure for
sleep is a decline in cerebral metabolism resulting from thermal down-regula-
tion exacerbated by sleep pressure. The purpose of such down-regulation is
unclear, although speculation is easy (e.g., the increased need for protein syn-
thesis, which is favored during sleep, especially during periods of lowest basal
metabolic level). Whatever its purpose, such basal drops may have a longer
time course than the EEG manifestation of specificsleep stages and may make
coherent cortical function impossible until metabolic or specific biochemical
activity has been increased through the passage of time (e.g., circadian varia-
tion), a change in sleep stage (e.g., accumulation of REM sleep), or increased
physical activity at awakening (e.g., getting out of bed)-all of which are inter-
correlated such that REM sleep is more likely to occur as body core tempera-
ture is rising, and awakenings are more likely to occur from REM sleep
(Dinges, 1989a). Unfortunately, despite many studies of hypnopompic perfor-
mance, none have assessed body core temperature as a covariant of sleep iner-
tia beyond looking at circadian variability.
The hypothesized widespread metabolic decline in cortical activity that
would covary with pressure (and therefore with depth) of sleep would probably
make it exceedingly difficult for a person to perform well if aroused abruptly
from sleep when basal metabolic levels are low. Virtually every type of perfor-
mance, especially all cognitive performances involving memory and atten-
tion, would be adversely affected. There is no reason to believe, however, that
Hypnopompic Cognitive Performance 171
the subject could not interact socially at some simple level such as name ac-
knowledgment; indeed, this is what appeared to be the case with our subject.
On the other hand, when performance is demanded, reverie can intrude, espe-
cially in response to a stimulus (e.g., the experimenter says “continue”). The
nature of such semicoherent reverie may be determined by the extent to which
the “functional deafferentation” suggested by Broughton ( 1968) has taken
place. The reverie would, therefore, be the result of insufficient neural metabo-
lism for coherent mentation. The inhibition needed for directed cognition
would be lacking.
REVERIE
of reverie and the nature of cognition in general may well be the same. What
happens to this process as REM sleep (and circadian time) accumulate re-
mains to be determined.
References
Akerstedt, T., & Gillberg, M. (1 979). Effects of sleep deprivation on memory and sleep
latencies in connection with repeated awakenings from sleep. Psychophysiology,
16,49-52.
Aschoff, J., Giedke, H., Poppel, E., & Wever, R. (1972).The influence of sleep intermp
tion and sleep deprivation on circadian rhythms in human performance. In W.P.
Colquhoun (Ed.), Aspects of human efficiency (pp. 135- 150). London: English
Universities Press.
Hypnopompic Cognitive Performance 173
Fort, A., & Mills, J. N. (1972). Influence of sleep, lack of sleep and circadian rhythm
on short psychometric tests. In W. P. Colquhoun (Ed.), Aspects of human efi-
ciency (pp. 1 15- 127). London: English University Press.
Gastaut, H., & Broughton, R. J. (1965). A clinical and polygraphic study of episodic
phenomena during sleep. In J. Wortis (Ed.), Recent advances in biologicalpsychia-
try (Vol7, pp. 197-22 1). New York: Plenum Press.
Gillberg, M., & Akerstedt, T. (1982). Body temperature and sleep at different times of
day. Sleep, 5, 378-388.
Goodenough, D. R., Lewis, H. B., Shapiro, A., Jaret, L., & Sleser, I. (1965). Dream
report following abrupt awakening from different kinds of sleep. Journal of Per-
sonality and Social Psychology, 2, 170- 179.
Grosvenor, A., & Lack, L. C. (1984). The effect of sleep before or after learning on
memory. Sleep, 7, 155-167.
Hartman, B. O., & Langdon, D. E. ( 1965).A second study on performance upon sudden
awakening (School of Aerospace Medicine Report No. TR-65-6 I). Brooks AFB,
TX: USAF.
Hartman, B. O., Langdon, D. E., & McKenzie, R. E. (1965). A third study on perfor-
mance upon sudden awakening (School of Aerospace Medicine Report No. TR-
65-63). Brooks, TX: U.S. Air Force.
Haslam, D. R. (1982). Sleep loss, recovery sleep, and military performance. Ergonom-
ics, 25, 163- 178.
Horne, J. A., & Staff, L. H. E. (1983). Exercise and sleep: Body-heating effects. Sleep,
6, 36-46.
Hume, K. I., & Mills, J. N. (1977). Rhythms of REM and slow wave sleep in subjects
living on abnormal time schedules. Waking and Sleeping, 3,29 1-296.
Jeanneret, P. R., & Webb, W. B. (1963). Strength of grip on arousal from full night’s
sleep. Perceptual and Motor Skills, 17, 759-76 1.
Kleitman, N. ( 1963). Sleep and wakefulness. Chicago: University of Chicago Press.
Koulack, D., & Schultz, K. J. (1974). Task performance after awakenings from differ-
ent sleep stages. Perceptual and Motor Skills, 39, 792-794.
Langdon, D. E., & Hartman, B. 0. (1961). Performance upon sudden awakening
(School of Aerospace Medicine Report No. 62-17). Brooks, TX: U.S. Air Force.
Lavie, P., & Giora, Z. (1973). Spiral aftereffect durations following awakening from
REM and non-REM sleep. Perception and Psychophysics, 14, 19-20.
Lavie, P., & Sutter, D. (1975). Differential responding to the beta movement following
awakening from REM and NONREM sleep. American Journal of Psychology, 88,
595-603.
Lubin, A., Hord, D., Tracy, M. L., &Johnson, L. C. (1976). Effects ofexercise, bedrest
and napping on performance decrement during 40 hours. Psychophysiology, 13,
334-339.
Mavromatis, A. (1987). Hypnagogia. London: Routledge & Kegan Paul.
Naitoh, P., Muzet, A., Johnson, L. C., & Moses, L. (1973). Body movements during
sleep after sleep loss. Psychophysiology, 10, 363-368.
Okuma, T., Nakamura, K., Hayashi, A., & Fujimori, M. (1966). Psychophysiological
study on the depth of sleep in normal human subjects. Electroencephalography
and Clinical Neurophysiology, 21, 140-147.
Patrick, G. T. W., & Gilbert, J. A. (1896). On the effects of loss of sleep. Psychology
Review, 3, 469-483.
Pritchett, T. P. (1969). An investigation of sudden arousal from rest: A study of im-
paired performance on an addition task (Doctoral dissertation, University of Ken-
tucky, 1964). Dissertation Abstracts International, 30, 2934B. (University Micro-
films No. 69-20,443)
Rechtschaffen, A., Hauri, P., & Zeitlin, M. (1966). Auditory awakening thresholds in
REM and NREM sleep stages. Perceptual and Motor Skills, 22,927-942.
Rosa, R. R., & Bonnet, M. H. (1985). Sleep stages, auditory arousal threshold, and
body temperature as predictors of behavior upon awakening. International Jour-
nal ofNeuroscience, 27, 73-83.
Hypnopompic Cognitive Performance 175
Rosa, R. R., Bonnet, M. H., & Warm, J. S. (1983). Recovery of performance during
sleep following sleep deprivation. Psychophysiology, 20, 152- 159.
Schacter, D. L. (1976). The hypnagogic state. A critical review ofthe literature. Psycho-
logical Bulletin, 83, 452-48 1.
Scott, J. (1969). Performance after abrupt arousal from sleep: Comparison of a simple
motor, a visual-perceptual, and a cognitive task. Proceedings of the 77th Annual
Convention of the American Psychological Association, 4,225-226.
Scott, J., & Snyder, F. (1968). “Critical Reactivity” (Pieron) after abrupt awakenings
in relation to EEG stages of sleep. Psychophysiology, 4, 370.
Seminara, J. L., & Shavelson, R. J. (1969). Effectiveness of space crew performance
subsequent to sudden sleep arousal. Aerospace Medicine, 40,723-727.
Sewitch, D. E. (1984). NREM sleep continuity and the sense of having slept in normal
sleepers. Sleep, 7, 147- 154.
Sewitch, D. E. (1987). Slow wave sleep deficiency insomnia: A problem of thermo-
downregulation at sleep onset. Psychophysiology, 24,200-2 16.
Sewitch, D. E., Kittrell, E. M. W., Kupfer, D. J., & Reynolds, C. F. (1986). Body tem-
perature and sleep architecture in response to a mild cold stress in women. Physi-
ology and Behavior, 36,95 1-957.
Stones, M. J. (1977). Memory performance after arousal from different sleep stages.
British Journal of Psychology, 68, 177- 181.
Tebbs, R. B., & Foulkes, D. ( 1966). Strength of grip following different stages of sleep.
Perceptual and Motor Skills, 23, 827-834.
Webb, W. B., & Agnew, Jr., H. (1964). Reaction time and serial response efficiency on
arousal from sleep. Perceptual and Motor Skills, 18, 783-784.
Webb, W. B., & Agnew, Jr., H. (1967). Sleep cycling within 24-hr periods. Journal of
Experimental Psychology, 74, 158-1 60.
Webb, W. B., & Agnew, Jr., H. (197 1). Stage 4 sleep: Influence of time course variables.
Science, 174, 1354-1357.
Wilkinson, R. T., & Stretton, M. (197 1). Performance after awakening at different
times of night. Psychonomic Science, 23,283-285.
Williams, H. L., Hammack, J. T., Daly, R. L., Dement, W. C., & Lubin, A. (1964).
Responses to auditory stimulation, sleep loss, and the EEG stages of sleep. Electro-
encephalography and Clinical Neurophysiology, 16,269-278.
Williams, H. L., Morlock, Jr., H. C., & Morlock, J. V. (1966). Instrumental behavior
during sleep. Psychophysiology, 2,208-2 16.
Williams, R. L., Karacan, I., & Hursch, C. J. (1974). EEG of human sleep: Clinical
applications. New York: Wiley.
CHAPTER 13
eye movement density of the epoch preceding the awakening. Low eye move-
ment density leads to no recall or to recall that is “thoughtlike.” High density
is related to more storylike content (Cartwright, Stephenson, Kravitz, & East-
man, 1987). The REM reports also become more dreamlike as the night pro-
gresses, and because eye movement density also increases with time of night,
these two measures can be somewhat confounded.
Generally, the progression across the night shows movement from short,
reality-oriented thought in the first half of the night, when more NREM sleep
is present, to longer, perceptually vivid, nonreality tested experiences in the
second half of the night, when REM sleep is proportionately higher.
The field has been troubled not only by how to grasp dream content
experimentally but also by how to define dreaming itselE Should the definition
focus on its psychological properties whenever and wherever it occurs (e.g., at
sleep onset and in NREM sleep) or on only the mental content that accompa-
nies REM periods? This question remains debatable. In my own work, I prefer
to sample REM periods and to define dreaming as the content of these periods,
which may be scaled as more or less dreamlike psychologically. This decision
was made because so much theorizing about dreaming starts from the proper-
ties of REM sleep, which is highly activated and closely associated with the
PGO spiking that, in man, appears to be related to the REMs that, in turn, are
associated with the dreamlike quality of mentation. Not that dreaming only
takes place in REM periods: Several good studies have shown that dreaming
occurs in NREM periods (Brown & Cartwright, 1978). However, I chose to
examine the REM stage because dreaming is most often associated with REM.
In fact, dreaming probably occurs cyclically throughout the 24-hr cycle, but it
is only sustained long enough to produce the narrative continuity necessary
for these experiences to be recalled when a highly aroused brain state persists
in the absence of demands for attention from external sources (Koulack &
Goodenough, 1976).
Table 1
CURRENT MODELS OF DREAM MEANING AND FUNCTION
Crick-Mitchison Hobson-McCarley Breger Freud
Variable (1983) (1977) (1969) (1955)
The theories of dream meaning and function are many, but they can be
grouped into four rough categories (see Table 1).
The Crick and Mitchison (1983) theory relates dreaming to previous
waking but views content as accidental and useless material to be purged to
make room in memory for better data. The Hobson and McCarley (1977)
theory, at least as originally published, postulates that the images originate
randomly from the pons but grants that associated meaning is then synthe-
sized by the cortex to express the current concerns camed over from waking
into sleep. Breger is one of many modem theorists who believe dream meaning
depends on the structure of the individual’s memory networks, particularly
those related to affect, and believe the function of dreaming is assimilative,
“integrating contemporary material into the adaptive solutions worked out in
the past” (Breger, Hunter, 8z Lane, 1971, p. 188). This position relates dreams
both to the immediate and recent presleep experience and to related memory
networks from the past. It implies no immediate impact on subsequent waking
behavior unless the material becomes conscious and is worked on to accom-
modate a change in outlook or behavior on the basis of an insight from a
dream. In this view, dreams have meaning and function but exert no effect on
postsleep states or traits without further integrative work. These theorists are
thus closer to the Freudians, who see the work of dreams as the gratification
of hidden impulses. Dream work is regularly done at night; only with a lot of
additional work during waking, to make the unconscious conscious, will this
effect any change in the waking personality.
Breger’s information-processing hypothesis has been taken one step fur-
ther by Greenberg, Pillard, and Pearlman (1972), who see the assimilation
work of dreams as adaptive for waking functioning in the original stress-pro-
ducing situation. According to these authors, behavior after a night of dream-
ing should be more productive (or at least less distressed) and should contrib-
ute to better coping in the situation that originally provoked anxiety. Jones
(1962), also associated with this position, took one more theoretical step to
relate dreams to postsleep waking by suggestingthat dreams preserve and pro-
tect the waking self-image. In other words, Jones believed that dreams serve
not just a stress-related adaptive function but a full-timejob.
To go beyond the descriptive level and test some of these theories has
been enormously difficult. The controlled studies have largely been failures.
Even the dream deprivation studies have run afoul because of the amount of
182 ROSALIND CARTWRIGHT
individual variability that occurs in studies with small sample sizes. The mean-
ing of the situation to the subject, including both the experiment itself and the
target stimulus; the subjects’ long-term traits and coping styles; and the sub-
jects’ own transformation language as it relates to what is in the memory net-
work of like experiences must be taken into account before much headway
can be made in understanding dream meaning and function.
How then can we best proceed? One good method is to take advantage
of anything promising from past work and to build better designs to clarify
helpful insights. I focused on work that has tested the assimilative function as
that starting point, particularly on the two studies reported by Breger, Hunter,
and Lane in “The effect of stress on dreams” (197 l), which represents a land-
mark in this area. Breger et al. took advantage of the laboratory to collect
dreams through sleep monitoring and, thus, avoided all the problems associ-
ated with home-recalled dreams. However, they used not experimenter-de-
signed manipulations to produce affect but genuine experiments in nature.
One consisted of presleep group therapy sessions and the other consisted of
dream collections prior to and following elective surgery. The studies were not
tightly controlled, but they are important because they took real-life events
that have tremendous personal meaning and traced their impact on dream
content. The studies were conducted in 1964- 1966, more than 20 years ago,
and they stand today as the best in the field.
The thinking is clear: External input is replaced in REM sleep by inter-
nally generated images, bits of memories generated by systems activated from
the presleep period, which becomes increasingly distant in the course of the
night. By looking at these images, we can make inferences about the memory
systems. To do this, we put the system under stress, generate good strong
affect, and watch the work of assimilation take place through dreams. This
can best be done by finding a distressing event that will threaten the currently
organized self-structure in some basic way so that the work of dreams and
their relevance to the specific emotional turmoil can be highlighted and exam-
ined. For this reason, I undertook the study of the dreams of people undergo-
ing divorce. Divorce is a stressful event. No matter how strongly one wishes
to end a bad relationship, no one gets out scot-free. One partner may retaliate
and cause so much trouble that both partners become stressed beyond mea-
sure for a considerable time before they reach a new reorganization as self-
respecting singles. Divorce involves a lot of loss, not only the loss of love but
also the loss of physical security (a home), of economic security (a standard of
living), of a social network of friends and shared relatives, and (most of all) of
self-esteem.
Although bereavement is credited with creating more stress on most re-
cent events scales, many divorcing people have told me that life would be so
much easier if only their spouse would “drop dead.” Divorce provides a good
natural stressor to work with. There is also a large pool of subjects to draw
from (there were 1.5 million divorces last year). The distress lasts a longtime-
most divorces take 1 to 2 years because the courts are crowded-and people
Network Model 183
vary in how well they cope. Some people are initially upset but then take back
their investment in the shared life and reach a new internal reorganization.
Others proceed from the state of stress to focus on the loss, shame, and guilt
of failure and become clinically depressed. This gives us some ability to stratify
a sample into those who are coping well with the stress of a major life change
and those who are not. Depression rates are high among those who are divorc-
ing (approximately 37%according to recent work).
What does it take to get through this life event successfully?Time, good
friends, good genes, good luck, and a good dream system was my hypothesis.
We know that stress disrupts sleep and that stress that culminates in the
hopelessness of depression can create a type of sleep disruption that affects
REM sleep. The work by Kupfer (1976) and Reynolds, Shaw, Newton, Cable,
and Kupfer (1983), which has been replicated by Rush, Giles, Rofiarg, and
Parker (1982); by Gillin and Borbely (1985); and by Cartwright (1983), shows
major affective disorder to have some usual (although not universal) effects on
the first REM period of the night and some additional effects on the distribu-
tion of REM across the night (see Table 2).
If REM sleep is disrupted, what happens to dreams? Do they malfunc-
tion in some way and fail to assimilate affective experience into memory net-
works? Is that why depressed patients feel at their worst first thing in the morn-
ing? If dreaming has an assimilative function, can we see this at work in
subjects who are going through divorce without becoming depressed? If good
copers under stress show productive dreaming, is there a threshold of affect
above which dreaming is overloaded and breaks down? Since we have meth-
ods of indexing depression in both the waking and sleep states (high eye move-
ment density and reduced REM latency), we have measures to help identify
those with good and poor dream functioning.
Table 2
DIFFERENCE BETWEEN REM SLEEP IN DEPRESSED
AND NORMAL PERSONS
Variable Normal Group Depressed Group
First REM
Latency 80-1 20 min <65 min
Eye movement density 4.5 >1.5 per min
Duration 8-1 2 min 20 min
All night
depression. The study is still in progress. The target sample contains 300 peo-
ple ( 150 men and 150 women) who are examined at the time of marital separa-
tion when the decision to divorce from a first marriage has been made. All are
unmedicated volunteers who have agreed to return in 1 year for a follow-up.
So far, 170 subjects have been seen (85 men, 85 women). Of these, 87 (40
men, 47 women) have returned for follow-up. Of the original sample, 42 sub-
jects ( 19 men and 23 women) were currently depressed according to the con-
joint criterion of meeting a Schedule of Affective Disorders and Schizophrenia
(SADS) (Endicott & Spitzer, 1978) Research Diagnostic Criteria (RDC)
(Spitzer, Endicott, & Robbins, 1978) diagnosis and yielding a Beck score
above 15. On SADS RDC criteria alone, the figures were much higher, with
47% of the men and 5 1% of the women classed as currently in an episode of
major affective disorder. Most people get over this. At follow-up, only 7.5% of
the men and 6.3% of the women remained depressed.
The original evaluation was used to select 90 subjects (45 men and 45
women) for the sleep laboratory study to examine the effects of divorce on the
sleep and dreams of 30 depressed and 15 nondepressed subjects of each sex
at the initiation of the break-up and a year later. To date, 54 subjects have
been studied for 3 nights, and 36 subjects have been restudied at a 1-year
follow-up.
The study of sleep showed that 18 subjects had short REM latencies (less
than 65 min when averaging Nights 2 and 3), and 13 subjects had high eye
movement density at the first REM stage (greater than 1.5 on the Kupfer
scale), but only 3 subjects had both markers. Both markers were relatively
stable over a 1-year period. The product-moment correlation for REM latency
was .53 and for eye movement density was .55. The factor most heavily con-
tributing to this stability was family history of depression. For those with fam-
ily history positive (n = 21), the correlation between the two REM latency
measures (at the breakup point and 1 year later) was .73. For those without a
Network Model 185
LOW
B€CK D€PRESSfON SCORES
Figure 1 Eye movement density in first and second half of night in relation to Beck
depression scores.
family history for affective disorder (n = 13), the correlation was .06. Not all
subjects who showed the markers of depression, however, were currently de-
pressed on their own self-reports. Twelve of 18 subjects with short REM laten-
cies had Beck scores above 145, and 8 of 1 3 subjects with high eye movement
density had Beck scores above 15. For those with a SADS diagnosis and self-
reported current depression, short REM latency was associated with higher
dreamlike fantasy ratings (on a 1-5 scale) and higher affect scores (1 = no
efect. 2 = mild,and 3 = strong) for the dream collected from the first REM
than those with normal REM latencies. In contrast, high eye movement den-
sity was associated with lower than normal dreamlike fantasy scores and neu-
tral affect in the first REM period (see Table 4).
Both sleep markers of depression were stable over 1 year, but they
seemed to represent two different subgroups. Short REM latency subjects had
earlier, more affective and storylike dreams, whereas high eye movement den-
sity subjects showed poor dream formation and reduced dream affect. The
numbers are too small for formal testing of an interaction effect at this time,
but the presence of an affective disorder seems to heighten the differences in
dreaming that are represented in these sleep groups.
Things are never as straightforward as one would wish. The so-called
depression markers are not statelike but rather are more traitlike. None of the
3 subjects who showed both markers were currently depressed, but all 3 had
past divorce-related depression episodes by RDC. The two markers seem to
represent two different sleep sub-groups: Those with early REM are affect re-
186 ROSALIND CARTWRIGHT
sponsive in terms of generating more and better fantasies, and those with
higher than normal phasic events at the beginning of the night have disrupted
dreaming. This means we need a large sample size to examine how stable these
findings are over time and what happens to these groups psychologically.
What does the sequential dream content look like in the laboratory for
those subjects who are coping well and those who are not? What theory do
they match, if any? Are Hobson and McCarley (1977) right in stating that,
because REM is turned on by the unthinking pons, the images are not selected
to tell a story but rather give the cortex the work of trying to interpret them as
best it can in the light of its own agenda of emotional concerns? Or are images
selected from a memory network that has been stimulated in response to some
particular type and level of affect aroused in connection with a recent experi-
ence, and do they relate the issues to specific memory components?
My best guess leads me to accept the latter premise. Because dreams
from the beginning to the end of the night often have repeated images, mem-
ory networks seem to be activated not randomly but in response to some per-
sistent issues. Dreams look like creative constructions, drawing on memory
systems that are activated by persistent emotional concerns. Memories are
encoded as unique bits but are organized into networks of related materials
and are decoded according to our needs, without our conscious attention or
direction. Emotional connections are made via earlier memories. These are
then lit up selectively when the brain is activated in REM sleep. Dream mean-
ing relates to the status of our ongoing needs, and dreams function to assimi-
Network Model 187
Table 3
DIFFERENCES IN DREAM CHARACTERISTICS BETWEEN
DEPRESSED AND NONDEPRESSED WOMEN (N = 29)
Dream Characteristic Depressed Group NondepressedGroup
late new data and reorganize related memories. Once the network is stimu-
lated in the first REM period, it is held in a state of readiness, and other parts
of it are more likely to be stimulated the next time REM occurs when older
elements of the same network are stimulated (see Figure 3).
Which network comes into play is governed by the type and level of
affect carried over from the day. When all is well with our world, the level is
low, and different networks may be stimulated in the same night. When we
are in an unstable state and affect runs high, the system may jam or fail to
work at its best. This appears to happen with higher then normal eye move-
ment density. Then dreams do not seem to progress across the night.
Dreaming appears to be a part of our normal information processing,
which relates waking experiences to past emotional experiences to past emo-
Table 4
DREAM CONTENT SCORES FOR DEPRESSED AND NONDEPRESSED
SUBJECTS
Dreamlike Fantasy' Affe c t b
Short Short
REM High Normal REM High Normal
Group Latency EMD REM Latency EMD REM
Depressed
M 3.40 2.38 3.00 1.90 1.oo 1.60
SD 1.07 0.92 0.71 0.88 0.76 0.55
Nondepressed
M 2.67 2.20 2.70 1.oo 1.40 1.60
SD 0.82 1.10 0.82 0.63 0.89 0.84
Note: EMD = eye movement density. For short REM latency, n = 18; for high EMD, n = 13;
for normal REM, n = 19.
Measured on a 1-5 scale, where 1 = no recall and 5 = well-developed dream. Measured
on a 1-3 scale, where 1 = neutraland 3 = strong.
188 ROSALIND CARTWRIGHT
References
Beck, A. (1967). Depression. New York: Harper & Row.
Berger, R. (1963). Experimental modification of dream content by meaningful verbal
stimuli. British Journal of Psychiatry, 109, 722-740.
Breger, L. (1969). Dream function: An information processing model. In L. Breger
(Ed.), Clinical-cognitive psychology (pp. 182-227). New Jersey: Prentice-Hall.
Breger, L., Hunter, I., & Lane, R. ( 1971). The effect of stress on dreams. Psychological
Issues Monograph, 27(7, Suppl. 3).
Brown, J., & Cartwright, R. (1978). Locating NREM dreaming through instrumental
responses. Psychophysiology, 15, 35-39.
Cartwright, R. (1983). Rapid eye movement sleep characteristics during and after
mood-disturbing events. Archives of General Psychiatry, 40, 197-20 1.
Cartwright, R. ( 1986). Affect and dream work from an information processing point
of view. Journal of Mind and Behavior, 7,4 1 1-427.
Cartwright, R., Stephenson, K., Kravitz, H., & Eastman, C. (1987). Life events effects
on REM sleep. Sleep Research, 16, 267.
Crick, F., & Mitchison, G. (1983). The function of dreaming sleep. Nature, 304, 11 1-
114.
Network Model 189
Endicott, J., & Spitzer, R. (1978). A diagnostic interview: The schedule for affective
disorders and schizophrenia. Archives of General Psychiatry, 35, 837-844.
Foulkes, D. ( 1 966). Thepsychology of sleep. New York: Scribners.
Freud, S. (1955). The Interpretation of Dreams. New York: Basic Books. (Original
work published 1900.)
Gillin, J., & Borbely, A. (1985) Sleep: A neurobiologicalwindow in affective disorders.
Trends in Neuroscience, 8, 537-542.
Greenberg, R., Pillard, R., & Pearlman, C. (1972). The effect of dream (stage REM)
deprivation on adaption of stress. Psychosomatic Medicine, 34, 257-262.
Hobson, A., & McCarley, R. (1977). The brain as a dream state generator: An activa-
tion-synthesis hypothesis of the dream process. American Journal of Psychiatry,
134, 1335-1348.
Jones, R. (1962). Ego synthesis in dreams. Cambridge, MA: Schenkman.
Koulack, D., & Goodenough, D. (1976). Dream recall and dream recall failure: An
arousal-retrieval model. Psychological Bulletin, 83,975-984.
Kupfer, D. (1976). REM latency: A psychobiologic marker for primary depressive ill-
ness. Biological Psychiatry, 11, 159- 174.
Monroe, L., Rechtschaffen, A., Foulkes, D., & Jensen, J. (1965). Discriminability of
REM and NREM reports. Journal of Personality and Social Psychology, 2, 456-
460.
Reynolds, C., Shaw, D., Newton, T., Cable, P., & Kupfer, D. (1983). EEG sleep in
outpatients with generalized anxiety: A preliminary comparison with depressed
outpatients. Psychiatric Research, 8, 8 1-89.
Rush, A., Giles, D., Roliivarg, H., &Parker, R. (1982).Sleep EEG and dexamethasone
suppression test findings in outpatients with unipolar and major depressive disor-
ders. Biological Psychiatry, 17, 327-34 1 .
Spitzer, R., Endicott, J., & Robbins, E. (1978). Research diagnostic criteria: Rationale
and reliability. Archives General Psychiatry, 35, 773-782.
Trenholme, I., Cartwright, R., & Greenberg, G. (1984). Dream dimension differences
during a life change. Psychiatric Research, 12, 35-45.
Webb, W., & Cartwright, R. (1978). Sleep and dreams. Annual Review of Psychology,
29,223-252.
Zimmerman, W. ( 1970). Sleep mentation and auditory awakening thresholds. Psycho-
physiology, 6 , 540-549.
CHAPTER 1 4
NIGHTMARES (DREAM
DISTURBANCES) IN POSTTRAUMATIC
STRESS DISORDER:
IMPLICATIONS FOR A THEORY OF DREAMING
MILTON KRAMER
Sleep that comes on quickly at an appropriate time and place, proceeds with-
out incident to terminate in a smooth awakening at a desirable time, and
leaves us feeling refreshed is what most of us seek from Somnos, the god of
sleep. Dreaming, Freud ( 1900) suggested, maintains the sleeping process by
binding the affective stimuli that might otherwise disrupt the continuity of the
process. How then are we to understand the disturbing dream that seems to
disrupt the process it was designed to protect? Is the disturbing dream, as
Freud ( 1920) suggested, a failure in an otherwise effective mechanism? Does
the dream lead to disruption because it cannot bind the emotion that surges
up periodically during the night?
Certainly this view of the sleepprotective function of dreaming is conso-
nant with two observations about REM sleep that are generally well-estab-
lished (Anch, Bowman, Mitler, & Walsh, 1988). First, if dreams maintain
sleep, and if the longer one sleeps the more likely one is to wake up after the
initial settling down process, then the distribution of REM sleep across the
usual sleep period is appropriate: It is positively accelerated, with more REM
occurring later in the night. This progression contrasts with that of deep sleep
(Stages 3-4), which decelerates (i.e., less occurs) as the night goes on. Second,
a period of dreaming sleep (REM sleep) often ends in a brief arousal. However,
sleepers generally do not remember this brief arousal. The importance of the
arousal for understanding the disturbing dream, which leads the sleeper to
awaken and recall a troublesome event, is that the nucleus for an awakening
190
Posttraumatic Stress Disorder 191
exists in the nature of the dreaming process itself: A brief, unrecalled arousal
is available to be converted into a troublesome awakening.
An example of a disturbing dream will concretize some of the problems
the experience presents and will call attention to some of the factors that pre-
dispose one to disturbed dreaming and that stimulate its appearance. The fol-
lowing dream was reported by a 26-year-old man who was a patient in a Veter-
ans Administration Hospital.
I dreamed I was back in Vietnam and I had thrown a grenade into one of
the Vietnamese huts. I went inside and there was one of the babies blown
up all over the inside of the hut. I woke up and was temfied, nauseated
and crying.
This dream meets our criteria for a disturbing dream: Troubling contents lead
to an awakening associated with negative affect and the recall of a prior trou-
bling dream.
Most dreams have no accompanying feeling. However, when a feeling
tone is recalled, a negative feeling is more common than a positive one (Hall
& Van de Castle, 1966). The romantic notion of the “land of dreamy dreams”
is not borne out by the experience of dreaming.
In fact, in a study (Taub, Kramer, Arand, &Jacobs, 1978)that compared
recalled dreams, recalled nightmares, and the concept of nightmares, the con-
cept of nightmares was reported to be more intense than the experience of the
nightmare. In addition, the nightmare and the recalled dream were reported
to have different contents. The disturbing dream is troublesome, does not
live up to its advance billing, and is clearly distinguishable from the ordi-
nary dream.
Although dream meaning is not the topic of this chapter, I will offer
an interpretation of this dream of infanticide that I arrived at without any
knowledge of the dreamer. The interpretation conforms with an analysis sub-
sequently provided by the dreamer’s therapist. (A more detailed examination
of this dream appears in Kramer & Roth, 1977.)
It was my impression that the dreamer felt he had committed a sanc-
tioned, aggressive (destructive) act, at a distance, against a fragile, possibly
pregnant woman and had caused her (or perhaps her unborn child) to die. The
dreamer did not want to be held responsible for the act (which he felt was
disgusting).
What do we know about the dreamer from the report given by his thera-
pist? He had been in Vietnam. He felt very guilty about some of his behavior
there. In his personal life, he had assaulted his second wife and had been ac-
cused of causing his first wife to abort by having intercourse late in the preg-
nancy and rupturing her membranes. He was the oldest of nine children in a
rural family. His father was a vagrant, violent alcoholic with whom he often
fought in defense of his mother. He had to assist on one occasion in delivering
one of his siblings because the doctor could not come and they could not get
his mother to the hospital. During his treatment, he reported a dream similar
192 MILTON KRAMER
to the one already described in which the baby was his daughter. The first
dream was described in the second session, and the dream about his daughter
was described in the sixth session.
Although one could discuss at length the meaning of this disturbing
dream and could raise a number of questions about predisposing factors, prox-
imal stimuli, the selection of content, internal organization,and so forth, these
speculations are not our current focus. We must focus inbtead on the awaken-
ing with negative affect that the dreamer reports.
The REM dream is often terminated by a brief arousal. The intensity of
the dream experience shows a systematic development across the REM period
(Kramer, Roth, & Czaya, 1975). The periodicity of the development of dream
content intensity is similar to the periodicity of the density of eye movements
seen during REM sleep. The intensity of the dream increases rapidly, peaking
at 10-20 min, then declines and starts up again. This rise and fall of emotion
across the dream period is consonant with the notion that there is a surge of
emotion during REM sleep, which the dream contains or attempts to contain.
The rise and fall of emotion, anxiety, fright, and so forth parallels physiological
activity, as reflected in eye movement density, which suggests a possible psy-
chophysiological parallel. Does the terminal arousal become an awakening
because the intensity surge exceeds the integrative capacity of the dream expe-
rience? This question must be explored further.
If an emotional surge exceeds the integrative capacity of the dream expe-
rience and results in a dream disturbance, is the surge related to the content
of the dream, to the preoccupation with and recall of dreams by the dreamer,
to the emotional condition (psychopathology)of the dreamer (i.e., predisposi-
tional issues), or to the responsivity of the dreamer to the experience (i.e.,
failure to bind or integrate an experience). Kramer, Schoen, and Kinney
(1984a) examined these issues and found that subjects who reported two or
more disturbed dream experiences per week had the same content categories
and dream frequencies as vivid dreamers. The vivid dreamers also reported
much higher dream recall in general than did the disturbed dreamers (89%
vs. 54%).
The disturbed dreamers, however, were more troubled people both from
a trait and a state point of view. Their scores were higher on all Minnesota
Multiphasic Personality Inventory (MMPI) scales and on the emotionally
based scales of the Cornell Medical Index, and they reported a greater number
of psychiatric hospitalizations. Their daily frequency of disturbed dreams ap-
peared to be related to current feelings (e.g., anger, sensitivity, and general
emotion). Of greatest interest was the fact that the disturbed dreamers showed
a greater degree of emotional responsivity to the dream experience than did
the vivid dreamers, and more of them had been described by others as acting
frightened in their sleep.
Dream disturbance sufferers, selected for having bad dreams, are more
psychologically troubled, recall fewer normal dreams, and are more respon-
sive to their dreams than are vivid dreamers. The increased responsivity sug-
PosttraumaticStress Disorder 193
gests an integrative failure that may be affected by the altered emotional state
(increased psychopathology) of disturbed dreamers.
The work of Kramer, Schoen, and Kinney (1984a) would lead to the
prediction that disturbed dreamers would be more responsive to stimuli (e.g.,
the dream experience)than normal subjects. Indeed, that was the case. Above-
threshold tones were presented during sleep to a group of disturbed dreamers
and a group of control subjects (Kinney & Kramer, 1985). Control subjects
responded in 52% of the trials, but disturbed dreamers responded in 93% of
the trials. Related to this was a stimulus identification in 40% of the control
subjects but in only 8% of the disturbed dreamers.
The dreaming process is linked to a process of repeated arousal. Arousal
followed by awakening in disturbed dreamers appears to result from increased
responsivity to the dream experience,which is related to the psychophysiologi-
cal alterations associated with being emotionally troubled. In chronic sufferers
from dream disturbance (e.g., patients with chronic delayed posttraumatic
stress disorder), the arousal threshold is increased (Schoen, Kramer, & Kin-
ney, 1984), which suggests that their emotional focus may be internal (i.e., on
the dream experience) and supports the theory that patients are more respon-
sive to the negative dream experience.
There is reason to believe that the sleep process is related to alterations
in subjective state, particularly affect, mood, or feelings. If the sleep process
has been successful, it is generally agreed that one feels down, tired, and sleepy
before sleep and feels refreshed after sleep. This view of sleep relates, in a gen-
eral way, to the theory that dreaming is related to emotional alterations during
sleep and that bad dreams are related to (and often the consequence of) altered
feeling states during the day (Kramer, Schoen, & Kinney, 1984a).
To examine the relation between sleep and subjective states, we assessed
mood, as indexed by an adjective check list, before and after sleep (Kramer &
Roth, 1972; Roth, Kramer, & Roehrs, 1976). Mood at night, before sleep, is
generally experienced more intensely;that is, one is more unhappy, aggressive,
friendly, clear thinking, or anxious prior to going to sleep at night than in the
morning. Further, moods experienced at night are more variable than those
experienced in the morning. Mood, therefore, generally decreases from night
to morning and becomes less variable.
As mood changed systematically from night to morning, the relation of
sleep physiology (Kramer, Roehrs, & Roth, 1976)and dream content (Kramer
& Roth, 1980)to mood change across the night was studied. The decrease in
unhappiness from night to morning was found to be related to whom and (to
a lesser extent) to what one dreams about, and changes in sleepiness and clear
thinking across the night were related to the amount of non-REM (NREM)
sleep the dreamer had experienced. These observations were utilized in formu-
lating the mood regulatory theory of sleep (Kramer, in press). From the per-
spective of the dream, the theory suggests that the dream subserves a selective,
affective regulatory function (Kramer & Roth, 1978). The dream may be
194 MILTON KRAMER
Sequential Interpretations
1. The thematic pattern focuses on the subject’s wish to depend on or
cling to the doctor and hospital even if she has to claim illness, but
the cuteness reveals a seductive motive as well.
2. The subject expresses, vis-A-vis the wife of a friend, the feared and
expected rejection, which she tries to minimize.
3. This dream can be seen as a turning point because the victory in the
tennis game appears to discharge effectively the tension generated by
the conflict between the wish to be close to the doctor/experimenter
for care and love (the disturbing motive) and the feared rebuff of
abandonment (reactive motive). The subject switches to a successful
competition that she wins with a partner and thus finds a successful
solution. The inference is that whatever she was struggling with has
been conquered.
4. The successful victory with a partner in the previous dream permits
the subject to deny her need for the doctor, which she admits she
once had. The bandage provides evidence that the need continues.
5. Recognition of her continuing repressed need for the doctor for care
and love causes the patient to intensify her rejection of him by being
critical of the doctor’s qualifications and by expressing her wish to
get even or to hurt him. It is the familiar double assertion: “Not only
do I not need you, but you’re no good anyway.’’
The sequential pattern in this series expresses a dependent/sexual longing to-
ward the experimenter/doctor, which leads to a feared (expected) rejection by
the wife/mother in the second dream. The subject masters the conflict in the
third dream by an aggressive victory with her own partner. The fourth dream
reveals a rejection of the previous need, although a recognition is present that
the need still exists. In the last dream of the night, a more intense rejection in
the form of an attack on a doctor serves to deny the need further.
obtained from a female subject (on the third dream night) after an REM-pe-
riod awakening (Kramer et al., 1964).
Repetitive Dreams
I. Somebody was lost. It was a dog and they were trying to find out
where it lived. It wasn’t my dog, though. I wasn’t lost. This person
who was lost was always fumbling around leading everybody else
around because he didn’t know what he was doing. Some boy, I
think. Somehow we had telephone numbers trying to find the right
one. It was supposed to be that little boy that was lost.
2. They filled up the car. There wasn’t enough room, unless I went
back with the people we went back with before. I could go back with
someone else. The place we were going was an orphanage
someplace, some house, a place like that.
3. I was dreaming about visiting, I think it was some EEG laboratory,
or something like that where the mothers could leave their children,
and they could go shopping. I doubt whether they could, there
wouldn’t be enough room for all of these people.
Repetitive Interpretations
1. The boy (association = experimenter) is seen as lost and misleading
others. Not the subject but the boy is lost. There is concern in the
dream that she will be misled by the experimenter because of his
inexperience, so she tries to call home.
2. In a setting of abandonment (orphanage),the subject hopes
anxiously that there will be room for her to return home. She has a
ride home, but the car is crowded.
3. The subject is in the laboratory temporarily and her mother is going
to return for her (implied). There isn’t enough room. In all the
dreams of the night, the subject deals with her fear of being
abandoned and her method of recontacting her family: calling on
the phone, riding in a car, or being picked up.
These patterns are elaborations of the general thesis that dreams may
subserve a problem-solving function (French & Fromm, 1964). Through this
effort at emotional problem solving, the dream may succeed or fail to contain
the emotional surge that leads to the arousal/awakening that is the hallmark
of disturbed dreaming. Difficulty in emotional problem solving in waking life
is certainly more characteristic of the psychopathologically disturbed subjects
who have been found to be more responsive to arousal/awakening after a
REM period and who report more bad dreams.
Sleep is generally a successful process, although there is much concern
expressed about a disturbed night’s sleep. Given that the process is generally
successful, one would expect to see the progressive-sequential pattern occur
Posttraumatic Stress Disorder 197
more often than the repetitive-traumatic pattern: In the two laboratory sub-
jects, that was the case (50%vs. 32%;Kramer et al., 1964).
If the progressive-sequential pattern is more common, one ought to see
dream content change across the REM periods of the night. This change in-
deed was found when the dreams were analyzed of 22 subjects who slept for
20 consecutive nights in the laboratory and described dream content from
each REM period of the night (Kramer, McQuarrie, & Bonnet, 198 1). Word
count and dream content showed systematic and statistically significant
change across the night. Three character variables in particular showed an
increase and decrease across the night, in keeping with the inverted U-shaped
curve that the progressive-sequential pattern describes. And the character vari-
able was best correlated with change on the dimension of unhappy mood
across the night (Kramer & Roth, 1980).
A number of studies have shown evidence that dreams are a regular and
orderly process (Kramer, 1982). The dreams of the same person on one night
are distinguishable from those on another night. This supports the notion that
the dreams of the night are indeed focused on a single topic, whether they are
handled in a progressive-sequential or a repetitive-traumatic manner. Dreams
in individuals are differentiable,but the dreams of one night are highly corre-
lated with the dreams of the next night. Dreams, therefore, are unique to the
individual dreamer.
The emotional preoccupations or current concerns of the dreamer, as-
sessed from Thematic Apperception Test (TAT) stories (Kramer, Roth, &
Palmer, 1976) and presleep and postsleep verbal samples (Kramer, Roth, Ar-
and, & Bonnet, 198 I), have been shown to be significantly correlated with the
content of a night's dreams. The dream content is linked thematically
(Kramer, Moshiri, & Scharf, 1982) both to the current concerns of the subject
before sleep and to the emotional preoccupation of the dreamer the subse-
quent morning. The notion of emotionally relevant material being processed
by the dreamer during his or her dreams is supported by these observations.
Thus, dream content is related across the waking-sleeping-wakingcontinuum.
What occurs at times of emotional upheaval, when disturbing dreams
occur with increased frequency, such as those that occur in chronic delayed
posttraumatic stress disorder (CDPTSD)? There may be a process of reactiva-
tion of previously troubling life experiences. This reactivation may be a re-
sponse to a breakdown in a current emotionally significant relationship (e.g.,
marriage), and the previously troubling experiences may become a metaphor
to express feelings such as hostility and rage stirred by the breakdown. This
previous experience is then linked to earlier childhood experiences.
This rather traditional psychodynamic view is supported by a study of
the sleep (Kramer & Kinney, 1988), dreams (Kramer, Schoen, & Kinney,
1984b), and psychological states (Kinney & Kramer, 1985) of CDPTSD pa-
tients. The sleep and dreams of Vietnam combat veterans with disturbed
dreaming were explored. They were then compared with combat-matched
control subjects without CDPTSD and with normal nonveterans.
198 MILTON KRAMER
The subject was surprised that a significant portion of his dreams focused
on his family. We learned later that, followingthe experience in the laboratory,
he went back and sought out his wife to see if there could be a reconcilia-
tion. His therapeutic work was rejuvenated by work on both past and current
problems.
The dreams of this subject reflect his use of the Vietnam situation in a
metaphorical sense to reflect his difficulties (Le., his difficulties in changing).
His hostility, which goes back to childhood and is directed toward his brother,
is manifested again in Vietnam, but it affects both a brother in arms and his
own brother from the past. His act of sibling rivalry results in a disruption of
his relationship with his parents. He is ambivalent about this disruption and
flees from them but also wishes to remain in contact with them. He cannot
resolve this tension. His plight leads him to be potentially out of control, per-
haps because of the intensity of his emotion. This potential loss of control is
reflected in his current situation with his wife. It is the breakdown with his
wife that stimulated his current difficulty, which is focused on Vietnam.
It is apparent from this case that the Vietnam experience can become a
metaphor for chronic interpersonal difficulty with peers and parental figures,
both current and past. The wish to dominate the peer leads to competition
with destructive consequences. It serves to alienate the subject from the par-
ents (and parental support figures) and potentially leads to a loss of control.
This example illustrates three levels of problems in the Vietnam veteran.
A breakdown in current life situations echoes and stirs up similar problems
from the past. Vietnam then serves as a metaphor to express the difficulties.
Those aspects of Vietnam that are in focus currently are those that have some
meaning as determined by events in the subject’s past, prior to Vietnam. The
Vietnam experience is a factor, but not the only factor, that dominates the life
of the Vietnam veteran.
The current precipitant is the new trauma in CDPTSD. In the illustrative
case, the current “universal stressor” is marital disruption. These stressful
events may occur in predisposed individuals. The Vietnam combat control
group may be a high-risk group for developing CDPTSD. The sleep of the con-
trol group (Kramer & Kinney, 1988)was disturbed when compared with that
of the normal group: 87% sleep efficiency, eight awakenings per night, more
light sleep, less REM sleep, and decreased deep sleep. Furthermore, their
dream recall was strangely low both overall (45%) and in REM sleep (50%)
(Kramer, Schoen, & Kinney, 1984b).
It may be that these combat veterans without dream disturbances are
vulnerable to developing CDPTSD in response to a current stress. This develop-
ment would facilitate the return of dreams about Vietnam, and those dreams
would also be expressive of the subject’s current problems. Disturbing dreams
and troubled awakenings are reinforced (perpetuated) by nonspecific, inter-
mittent stimuli (apparently meaningless noise) at night.
The experience of the Vietnam veteran with CDPTSD may be placed in a
more traditional psychodynamic perspective. The Vietnam experience re-
PosttraumaticStress Disorder 20 1
flected but did not completely determine their later lives. The patients with
CDPTSD had a reactivation of memories of Vietnam brought about by some
breakdown in their current life equilibrium. The case illustration involved a
disruption in a marital relationship. The current preoccupations of the patient
echoed both his Vietnam experience and his childhood experience. This form
of illness is not particularly unusual. Once the dynamic is started, psychophys-
iological factors (i.e., alterations in sleeping and dreaming) may serve to keep
it alive.
Theories of dreaming need to address dreams that go on automatically,
outside of awareness, and those that enter awareness and have the potential
for a more direct effect on the dreamer’s consciousness (Kramer, 198 1). One
needs both an assimilative and an accommodative theory of dreaming, with
the former being reductive and the later being potentially transforming. The
selective-affective mood regulatory theory of dreaming (Kramer & Roth,
1978; Kramer, in press) is an example of a n assimilative theory. The experi-
ence of the disturbing dream opens the possibility for a n extension of this
assimilative view to encompass some degree of transformation and accommo-
dation,
References
Anch, A. M., Bowman, C. P., Mitler, M. M., & Walsh, J. K. (1988).Sleep: A scientific
perspective. Englewood Cliffs, NJ: Prentice-Hall.
French, T., & Fromm, E. (1964).Dream interpretation. New York: Basic Books.
Freud, S. (1900). The interpretation of dreams. In J. Strachey (Ed. and Trans.), The
standard edition of the complete psychological works of Sigmund Freud (Vols. 4-
5 ) . London: Hogarth Press.
Freud, S. (1920). Beyond the pleasure principle. In J. Strachey (Ed. and Trans.), The
standard edition of the complete psychological works of Sigmund Freud (Vol. 18).
London: Hogarth Press.
Hall, C., & Van de Castle, R. ( 1966). The content analysis of dreams. New York Apple-
ton.
Kinney, L., & Kramer, M. (1984).Sleep and sleep responsivity in disturbed dreamers.
Sleep Research, 13, 102.
Kinney, L., & Kramer, M. (1985). Personality differences among disturbed dreamers,
“sick” controls and normal veterans. Sleep Research, 14, 130.
Kramer, M. (1981). The function of psychological dreaming: A preliminary analysis.
In W. P. Koella (Ed.), Sleep 1980 (pp. 182- 185). Basel, Switzerland:Karger.
Kramer, M. (1982). The Psychology of the dream: Art or science. Psychiatric Journal
of the University of Ottawa, 7, 87- 100.
Kramer, M. (in press). The mood regulatory function of sleep. In A. Moffit &
M. Kramer (Eds.), Thefunction of dreaming. Albany, NY: SUNY Press.
Kramer, M., & Kinney, L. (1988). Sleep patterns in trauma victims with disturbed
dreaming. Psychiatric Journal of the University of Ottawa. 13, 12- 16.
Kramer, M., McQuame, E., & Bonnet, M. (1981). Problem solving in dreaming: An
empiricaltest. In W. P. Koella (Ed.), Sleep 1980 (pp. 174- 178).Basel, Switzerland
Karger.
202 MILTON KRAMER
Kramer, M., Moshiri, A., & Scharf, M. (1982). The organization of mental content in
and between the waking and dream state. Sleep Research, 11. 106.
Kramer, M., Roehrs, T., & Roth, T. (1976). Mood change and the physiology ofsleep.
Comprehensive Psychiatry, 17, 161-165.
Kramer, M., & Roth, T. (1972). The mood-regulatory function of sleep. In W. P.
Koella & P. Levin (Eds.), Sleep 1972 (pp. 563-57). Basel, Switzerland: Karger.
Kramer, M., & Roth, T. (1977). Dream translation, Israel Annual of Psychiatry, 15,
336-3561.
Kramer, M., & Roth, T. (1978). The dream as selective affective modulator. Acad.
Form, 22, 13.
Kramer, M., & Roth, T. (1980). The relationship of dream content to night-morning
mood change. In L. Popoviciv, B. Asgian, & G. Badiv (Eds.), Sleep 1978 (pp. 62 1-
624). Basel, Switzerland: Karger.
Kramer, M., Roth, T., Arand, D., & Bonnet, M. ( 198 1). Waking and dreaming menta-
tion: A test of their interrelationship, Neuroscience Letters, 22,83-86.
Kramer, M., Roth, T., & Czaya, J. (1975). Dream development within a REM period.
In Sleep I974 (pp. 406-408). Basel, Switzerland: Karger.
Kramer, M., Roth, T., & Palmer, T. (1976). The psychological nature of the REM
dream report and T.A.T. stories. Psychiatric Journal of the University of Ottawa,
I , 128-135.
Kramer, M., Schoen, L. S., & Kinney, L. (1984a). Psychological and behavioral fea-
tures of disturbed dreamers. Psychiatric Journal of the University of Ottawa, 9,
102- 106.
Kramer, M., Schoen, L. S., & Kinney, L. (1984b). The dream experience in dream-
disturbed Vietnam veterans. In B. A. Van der Kolk (Ed.). Washington, DC:Amer-
ican Psychiatric Press.
Kramer, M., Schoen, L. S., & Kinney, L. (1987). Nightmares in Vietnam veterans.
Journal of the American Academy of Psychoanalysis, IS,67-8 1.
Kramer, M., Whitman, R., Baldridge, B., & Lansky, L. (1964). Patterns of dreaming:
The interrelationship of the dreams of a night. Journal of Nervous and Mental
Disorders, 139,426-439.
Roth, R., Kramer, M., & Roehrs, T. (1976). Mood before and after sleep. Psychiatric
Journal of the University of Ottawa, 1, 123-127.
Shoen, L. S., Kramer, M., & Kinney, L. (1984). Auditory thresholds in the dream
disturbed. Sleep Research, 13, 102.
Taub, J., Kramer, M., Arand, D., & Jacobs, G. (1978) Nightmare dreams and night-
mare confabulations, Comprehensive Psychiatry, 19,285-29 1.
INDEX
Across States Connectionist Model of Imagery reaction time. See Reaction time
and Thought (ASCIT), 5,7,9 time estimation. See Time perception
Activation, Input Source, Modulation (AIM), vigilance, 163
25-40 Cornell Medical Index, 192
Alzheimer’s disease, 100-101
Amnesia, 80,83,85,98, 134, 160 Demand characteristics, 1 1 1, 152
anterograde, 69,72-74, 144 Depression, 184-1 88
waking, 85,92,97 Dichotic listening, 134
Anesthesia, 127-135 Disorders of Initiating and Maintaining Sleep
awareness under, 128 (DIMS). See Insomnia
cognitive processing under, 128-1 33 Dream deprivation, 181- 182
Animal models, 27 Dreams, 179- 189
Anxiety, 152-1 54 Activation-Synthesis model, 7- 10, 17
Artificial intelligence, 73 and affect, I9 1- I97
Auditory item presentation bizarreness, 8- 10, 17-20
air conduction, 94 Crickand Mitchison theory, 17-18, 181
bone conduction, 94 and depression, 183-1 88
Atonia, 33 Freud’s theory of, 181, 190
Awakenings, 62,155, 168-169 function of, 181-183, 186-188
lucid dreams, 36-38,109-126
Beck Depression Inventory, 183-1 86 network model, 179- I89
Behavioral response NREM dreams, 180
after anesthesia, 130 recall of, 90,96, I66
in sleep, 77-87 repetitive dreams, 194- 197
response latency, 64-65,80 reports, 160, 179-180, 198-200
Blind sight, 134 and stress, 182-202
Brain metabolism, 170-1 7 1 combat, 198-201
divorce, 182-188
Cerebellum, 14, 16-17,20,73
Circadian rhythm, 151-152, 161,170, 172 Electroencephalogram(EEG)
Classical conditioning, 70-75, 128 alpha, 60,62,69-72,78-79,83-85,88-
conditioned stimulus, 7 I 92, 120,149
in sleep, 70-72,78,89 K-complexes,49,7 1,95
of tolerance to hypnotics, 145 NREM sleep. See NREM sleep
unconditioned stimulus, 7 1 REM latency, 183- 188
Cognitive performance tasks REM sleep. See REM sleep
clock reversal, 163 sleep efficiency, 140
letter cancellation, 163 sleep onset latency. See Sleep onset
logical reasoning, 162 latency
memory tasks, 162 sleep spindles, 63, 149, 151
mental arithmetic, 161-168 slow wave sleep, 140, 142, 144, 165- I 72
203
204 INDEX