http://cdp.sagepub.com/ScienceCurrent Directions in Psychological http://cdp.sagepub.com/content/18/2/84The online version of this article can be… [626971]

http://cdp.sagepub.com/ScienceCurrent Directions in Psychological
http://cdp.sagepub.com/content/18/2/84The online version of this article can be found at: DOI: 10.1111/j.1467-8721.2009.01614.x 2009 18: 84 Current Directions in Psychological ScienceRoss Levin and Tore NielsenDreamingNightmares, Bad Dreams, and Emotion Dysregulation: A Review and New Neurocognitive Model of  
Published by:
http://www.sagepublications.com
On behalf of: 
  Association for Psychological Science
can be found at: Current Directions in Psychological Science Additional services and information for    
  http://cdp.sagepub.com/cgi/alerts Email Alerts:  
http://cdp.sagepub.com/subscriptions Subscriptions:  
http://www.sagepub.com/journalsReprints.nav Reprints:  
http://www.sagepub.com/journalsPermissions.nav Permissions:  
What is This? – Apr 1, 2009 Version of Record >>
at St Petersburg State University on December 9, 2013 cdp.sagepub.com Downloaded from at St Petersburg State University on December 9, 2013 cdp.sagepub.com Downloaded from

Nightmares, Bad Dreams, and
Emotion Dysregulation
A Review and New Neurocognitive Model of Dreaming
Ross Levin1and Tore Nielsen2
1Ferkauf Graduate School of Psychology, Yeshiva University, and2Dream & Nightmare Laboratory, Sacre ´-Coeur Hospital,
Montreal
ABSTRACT— Nightmares—vivid, emotionally dysphoric
dreams—are quite common and are associated with a
broad range of psychiatric conditions. However, the origin
of such dreams remains largely unexplained, and therehave been no attempts to reconcile repetitive traumatic
nightmares with nontraumatic nightmares, dysphoric
dreams that do not awaken the dreamer, or with morenormative dreams. Based on recent research in cognitive
neuroscience, sleep physiology, fear conditioning, and
emotional-memory regulation, we propose a multilevelneurocognitive model that unites waking and sleeping as a
conceptual framework for understanding a wide spectrum
of disturbed dreaming. We propose that normal dreamingserves a fear-extinction function and that nightmares
reflect failures in emotion regulation. We further suggest
that nightmares occur as a result of two processes that weterm affect load—a consequence of daily variations in
emotional pressures—and affect distress—a disposition to
experience events with high levels of negative emotionalreactivity.
KEYWORDS— nightmares; dreaming; emotion regulation
Nightmares—vivid and highly emotionally dysphoric dreams
that awaken the individual from sleep—are among the most
commonly experienced sleep disorders (for extensive reviews of
this literature, see Levin & Nielsen, 2007; Nielsen & Levin,
2007). While fear and terror are the predominant emotions as-
sociated with nightmares, other emotions such as rage are not
uncommon (Zadra, Pilon, & Donderi, 2006). Like most dreams,
nightmares typically occur during rapid-eye-movement (REM)
sleep.Nightmares typically imply nocturnal awakening (Levin &
Nielsen, 2007), whereas bad dreams are usually defined as
negatively toned dreams that do not awaken the dreamer (Levin
& Nielsen, 2007; Zadra & Donderi, 2000; Zadra et al., 2006).
Despite phenomenological similarities between nightmares and
bad dreams, it remains unknown whether they are two qualita-
tively distinct phenomena or a single phenomenon varying in
intensity. We suggest that bad dreams involve similar processesand merely differ in how effective (or ineffective) they are in
regulating shifting surges of current affect levels, a process we
refer to as regulating affect load (see below for further discus-
sion). Accordingly, we use the term disturbed dreaming (DD)
when referring to both nightmares and bad dreams.
PREVALENCE AND DEMOGRAPHIC
CHARACTERISTICS OF DD
Occasional episodes of DD are ubiquitous in the general pop-
ulation. Epidemiological studies indicate that about 85% of
adults report experiencing at least one nightmare within the
previous year (Levin, 1994), with about 2 to 6% of respondents
reporting weekly nightmares. Furthermore, nightmare incidence
is reported at significantly higher rates in younger adults starting
at age 14, in women after age 14, and in clinical populations
(Levin, 1994; Nielsen, Stenstrom, & Levin, 2006). As night-
mares are rarely reported spontaneously as clinical problems or
inquired about in routine health screenings, true prevalencerates are likely higher. In addition, retrospective reporting sig-
nificantly underestimates true DD prevalence and incidence
rates (Zadra & Donderi, 2000).
Perhaps the most robust finding in the DD literature is the
strong association between DD frequency and waking psycho-pathology (e.g., Berquier & Ashton, 1992; Blagrove, Farmer,
& Williams, 2004; Hartmann, Russ, Oldfield, Sivan, & Cooper,
1987; Levin & Fireman, 2002; Levin & Nielsen, 2007; Nielsen,
Laberge, Tremblay, Vitaro, & Montplaisir, 2000). Because mostAddress correspondence to Ross Levin, Ferkauf Graduate School of
Psychology, Albert Einstein College of Medicine, 145 Rousso Bld.,1165 Morris Park Ave., Bronx, NY 10461; rlevin@yu.edu.CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE
84 V olume 18—Number 2 Copyright r2009 Association for Psychological Science at St Petersburg State University on December 9, 2013 cdp.sagepub.com Downloaded from

of these clinical disorders are marked by considerable waking
emotional distress, their association with nightmares suggests
that nightmare production is related to a personality style
characterized by intense reactive emotional distress (Belicki,
1992; Blagrove et al., 2004; Levin & Fireman, 2002; Levin &
Nielsen, 2007; Nielsen et al., 2000). Furthermore, it has long
been noted that DDs are often precipitated by stressful lifeevents (Berquier & Ashton, 1992; Hartmann et al., 1987). DDs
are most commonly associated with trauma exposure and post-
traumatic stress disorder (PTSD), and there is a strong link
between trauma exposure and subsequent DD (e.g., Mellman,
David, Kulick-Bell, Hebding, & Nolan, 1995; Woodward,
Arsenault, Murray, & Bliwise, 2000).
Further evidence for a link between increased stress and DD
comes from a landmark prospective study by Wood, Bootzin,Rosenhan, Nolen-Hoeksema, and Jourden (1992), who found
nightmare incidence to be twice as high immediately after the
1989 San Francisco earthquake in two San Francisco Bay-area
groups than in an Arizona sample, despite equal baselinefrequencies. Importantly, these differences were dose-response
specific to proximity to the earthquake epicenter—those who
were closer had more nightmares.
THE AMPHAC/AND NEUROCOGNITIVE MODEL OF
DISTURBED DREAMING
Despite the proliferation of recent experimental work on DD,nightmare pathogenesis remains largely unexplained. Current
work by us (Levin & Nielsen, 2007; Nielsen & Levin, 2007)
incorporating recent advances in cognitive neuroscience, sleep
neurophysiology, and fear conditioning—particularly in relation
to PTSD and sociocognitive-based diathesis (i.e., vulnerability)–stress models of psychopathology—supports a multilevel model
of dream function and nightmare production that unites neural
and cognitive processes in both waking and sleeping. The
neurophysiological branch of this model is termed the AMPHAC
network, after its presumed underlying neurophysiologicalcenters: the amygdala (A), the medial prefrontal cortex (MP), the
hippocampus (H), and the anterior cingulate cortex (AC). The
cognitive branch is termed the affect network dysfunction (AND)
model. Together, the two branches integrate explanatory con-
cepts at both a neural level (i.e., a cohesive and interconnected
network of limbic and forebrain regions underlying emotional
expression and representation) and a cognitive level (i.e., a
dream-production system that transforms fear memories into
dream and nightmare imagery). Disruption of processes at these
levels can account for a variety of features associated with
nightmare imagery (lack of emotional control, bizarre features,
or replay of traumatic memories).
The AMPHAC/AND model stipulates that DD results from
dysfunction in a network of affective processes that, duringnormal dreaming, are presumed to serve the adaptive function of
fear-memory extinction. Indeed, the underlying neurophysiolo-gy and biochemistry of REM sleep appears to be primed to
activate these very systems. At the cognitive level, dreaming is
proposed to facilitate fear-memory extinction by three processes:
memory-element activation, memory-element recombination,
and emotional expression.
The first process refers to the increased availability of a wide
range of memory elements during dreaming. For example, it haslong been noted that, with the exception of trauma memories,
dreams often do not represent coherent episodic memories; the
deconstruction of memories into isolated elements or basic units
is considered by most dream and sleep researchers to be a car-
dinal phenomenal feature of dreaming. The second process,
memory-element recombination, is largely responsible for the
continuous assembly of isolated memory units into a constantand phenomenologically coherent flow of dream imagery. We
propose that this organization occurs during dreaming: New
image contexts are produced for highly emotionally arousing
memorial elements. We propose that these new memorial com-
ponents are rendered into virtual simulations that maximizetheir impact on limbic structures, in a manner functionally
identical to that which occurs during waking. Limbic structures
respond more readily to perceptual stimuli than to imaginal
stimuli. The new representations are then recombined to intro-
duce contextual elements that are incompatible with existing
fear memories, thus facilitating emotional processing by pro-
viding novel contexts for fear that reinforce the development ofnew extinction memories. The reality mimesis endemic to dream
phenomenology (i.e., that dreams feel real and are experienced
as waking perception, not simply as hallucination) ensures that
fear memories are processed in a medium similar to that in which
they were first encoded, thus facilitating emotion regulation.
We consider the third process, emotional expression, to be
a necessary step in dreaming’s fear-extinction function, as itmaximizes the involvement of neural structures—primarily but
not limited to those of the limbic system—to further ensure the
adequate deployment of attentional resources in order to down-
regulate negative emotional arousal.
We suggest that engagement of these fear-extinction processes
may be the default function of REM sleep, with dreaming rep-
resenting the experienced result of these mechanisms. Repre-
sentation of specific memorial components in dream content is
then determined by ongoing daytime demands on the emotional-
memory system—in other words, we dream about what we are
emotionally preoccupied by in waking.
We use the term affect load (AL) to refer to the ongoing ac-
cumulation of stressful and emotional negative events that im-
pinge on an individual’s capacity to effectively regulate emotion.
AL, in our model, is a state (i.e., transitory) factor considered
to be a primary determinant of DD incidence. Thus, as AL
increases, so does the probability of DD. In contrast, affectdistress (AD), defined as a dispositional tendency to experience
heightened distress in response to emotional stimuli, is proposed
to be a major determinant of whether DDs will become clinical
Volume 18—Number 2 85Ross Levin and Tore Nielsen
at St Petersburg State University on December 9, 2013 cdp.sagepub.com Downloaded from

waking problems such as anxiety or fear. AD is akin to the
negative-affect dimension recently proposed for distress-based
disorders, in that all such disorders involve heightened
emotional activation. Individuals high in AD are particularly
reactive to both fearful and disturbing visual stimuli, and they
report creating more vivid images than do those low in AD,
suggesting that reality mimesis greatly facilitates emotionactivation.
At the neural level, the fear-extinction function is supported
by a network of limbic, paralimbic, and prefrontal regions that
constitute the control center for emotion expression and regu-
lation during both sleeping and waking. At the broadest level,
the amygdala is the control center for AL and is strongly im-
plicated in fear conditioning. The medial prefrontal cortexserves as the mediator of extinction by regulating impulsive
emotional expression via selective gating within the amygdala.
The hippocampus plays a crucial role in the encoding and
consolidation of episodic memories, as well the representation of
stimuli in novel contexts—a crucial mechanism for emotionprocessing. Last, the anterior cingulate mediates AD; this region
has been implicated in pain distress, social exclusion, and
separation anxiety and in processing negative emotional stimuli.
Taken together, the cognitive and neural explanatory levels
constitute an emotion network within which disruptions produce
increasing DD, beginning with occasional bad dreams and
proceeding to mildly distressing idiopathic nightmares and,finally, to repetitive and highly disturbing nightmares. Occa-
sional bad dreams and nightmares without much accompanying
distress the following day often occur in response to increasing
levels of AL and usually remain isolated incidents. However, in
vulnerable individuals primed for selective emotional reactivity(i.e., those with high AD), these dreams may serve as activators
for previously encoded fear-memory structures and lead to en-
hanced waking distress—and, subsequently, to more frequent
and disturbing nightmares. Thus, we suggest that individuals
high in AD utilize encoding biases to selectively scan their
dream imagery for threats and may experience their nightmares
as more threatening and distressing than individuals low in AD,leading to a preponderance of false alarms of impending danger.
Thus, for these individuals, nightmares may well be likened
to the same false-alarm responses that have been noted to occur
in panic disorder.
STRENGTHS OF THE MODEL
The AMPHAC/AND model is consistent with current literature
from cognitive neuroscience, sleep physiology, and fear condi-
tioning. Furthermore, the fact that the model unites waking and
sleeping processes renders it highly amenable to empirical
investigation, in that emotion-regulation processes should be
reflected in convergence across the waking–dreaming contin-uum. One of the central components of the model is that while ALis proposed to directly affect the incidence of DD, it is the AD
component that is responsible for waking dysregulation of
emotions and the connection to psychopathology. Thus, AD is
presumed to mediate the commonly observed relationship be-
tween nightmare incidence and waking distress-based psycho-
pathology; current work being conducted in our laboratory is
directly testing these assumptions.
EMPIRICAL EVIDENCE FOR THE MODEL
While broadly speculative at this stage, empirical evidence from
the neurophysiology of sleep and dreaming and the affective-
neuroscience literature are consistent with these formulations.
For example, the work of Foa (Foa & Kozak, 1986), Lang (Lang,
Davis, & Ohman, 2000), and LeDoux (2000) on fear-memorystructures and fear conditioning highlights fear’s automaticity,
its disproportionate emphasis on response elements (‘‘running
away from a monster’’), and its resistance to extinction. Research
has demonstrated that frequent nightmares are associated with a
number of personality characteristics (heightened imagery in-volvement, fantasy proneness, psychological absorption, and
increased emotional activation to internal states) that are
broadly consistent with our AD component (Levin & Nielsen,
2007). Further, imagery vividness is associated with increased
fear activation, heightened memorial clarity for perceived neg-
ative events, and increased difficulty monitoring the sources of
threats. In the recurrent nightmares of PTSD, fear-memory ele-ments may be globally activated in a highly coherent manner,
producing nightmares that consistently reproduce past fearful
experiences.
Support for the crucial role of AD in mediating the relation
between nightmare frequency and subsequent psychopathology
comes from studies by Belicki (1992) and Levin and Fireman
(2002) demonstrating that DD frequency is largely independent
from waking psychopathology when AD is controlled for, a
finding subsequently confirmed by at least three independent
investigations.
Empirical support for the role of AL in the generation of DD
is abundant. Heightened life stress is associated with increasedoverall dream recall and with DD in particular, and at least
three studies have demonstrated that individuals who have
frequent nightmares report that major distressing life events
frequently precipitate their nightmares. That nightmares are a
ubiquitous feature of trauma exposure also underscores this
point.
On the neural level, there is ample evidence of anatomical
connections between the four designated brain regions, and all
have been implicated in emotional expression and regulation.
Further, these brain regions are associated with both state and
trait differences in emotional responding and in distress-based
emotional disorders, particularly PTSD. Last and perhaps mostcrucial, imaging studies in both animal and human samples have
86
V olume 18—Number 2Nightmares, Dreaming, and Emotion Regulation
at St Petersburg State University on December 9, 2013 cdp.sagepub.com Downloaded from

found that activity in all four brain regions increases in REM
sleep above levels seen in wakefulness or non-REM sleep. Thus,
the network is a vital component of normal dreaming and is likely
influential in shaping emotional imagery during dreaming (see
Nofzinger, 2004, for a review of brain-imaging studies and REM
sleep, and McGaugh, 2004, for a review on the neural under-
pinning of heightened emotional processing).
FUTURE DIRECTIONS
Study of the neurophysiology of dreaming is still in its infancy,and any models explaining dreaming (as opposed to REM sleep)
are likely to remain speculative for some time. As other brain
components are likely to be integral in generating and shapingdream imagery, our neural model is not meant to be all-inclusive.
Thus, while we believe that the anterior limbic system is central
to nightmares, by no means do we believe that it is the sole seat of
dreaming.
Similarly, despite our emphasis here on fear extinction, that
should not be taken as the sole function of dreaming. While otherestablished dreaming models purport similar functions (i.e.,
threat detection, memory consolidation, mood regulation), the
question of dream function has befuddled brain scientists and
philosophers alike for some time and is not likely to be answered
soon. In addition, the proposition that DD serves an ongoing fear-
extinction function in individuals low in AD has not beendirectly subjected to empirical inquiry and remains an important
area for future investigation. In addition, our model does not
directly address the question of adaptive versus nonadaptive
fears in an evolutionary context, although we presume that fear
extinction is highly adaptive despite its predilection for exces-sive false positives (e.g., nightmares, panic attacks).
For these reasons, our proposed model is meant to serve as a
heuristic to generate further research into these mechanisms.
For example, as activated fear memory structures are presumed
to have an organizing (albeit costly) effect on dream content,
empirical investigation of the organizational coherence of both
the nightmares and normal dreams of individuals with frequent
nightmares would help to elucidate the mechanisms. Similarly,
if fear memories are responsible for the nonconscious detection
of threat, it would be informative to investigate whether
individuals high in AD perform similarly to individuals with
anxiety disorders or PTSD on an affective backward-masking
paradigm or the emotional color-word Stroop test. It would also
be interesting to determine if individuals with high AD who havenightmares demonstrate more readily conditioned fear re-
sponses while awake than do those with low AD and nightmares.
Finally, prospective research tracking relations among mood,
stress, and perceived coping effectiveness both before and after
nightmares would be invaluable in determining how nightmaresoriginate.Recommended Reading
Levin, R., & Fireman, G. (2002). (See References). A representative
study of recent empirical research on nightmares.
Levin, R., & Nielsen, T.A. (2007). (See References). A comprehensive
and state-of-the-art review of dream and nightmare pathogenesis,discussing the AMPHAC branch of the model in considerablygreater detail than the current paper.
Nielsen, T.A., & Levin, R. (2007). (See References). This paper
discusses the AND branch of the model in greater detail than thecurrent paper.
REFERENCES
Belicki, K. (1992). Nightmare frequency versus nightmare distress:
Relations to psychopathology and cognitive style. Journal of
Abnormal Psychology ,101, 592–597.
Berquier, A., & Ashton, R. (1992). Characteristics of the frequent
nightmare sufferer. Journal of Abnormal Psychology ,101, 246–
250.
Blagrove, M., Farmer, L., & Williams, E. (2004). The relationship of
nightmare frequency and nightmare distress to well-being. Journal
of Sleep Research ,13, 129–136.
Foa, E.B., & Kozak, M.J. (1986). Emotional processing of fear:
Exposure to corrective information. Psychological Bulletin ,99,
20–35.
Hartmann, E., Russ, D., Oldfield, M., Sivan, I., & Cooper, S. (1987).
Who has nightmares? The personality of the lifelong nightmare
sufferer. Archives of General Psychiatry ,44, 49–56.
Lang, P.J., Davis, M., & Ohman, O. (2000). Fear and anxiety: Animal
models and human cognitive psychophysiology. Journal of Affec-
tive Disorders ,61, 137–159.
LeDoux, J.E. (2000). Emotion circuits in the brain. Annual Review of
Neuroscience ,23, 155–184.
Levin, R. (1994). Sleep and dreaming characteristics of frequent
nightmare subjects in a university population. Dreaming ,4, 127–
137.
Levin, R., & Fireman, G. (2002). Nightmare prevalence, nightmare
distress, and self-reported psychological disturbance. Sleep, 25,
205–212.
Levin, R., & Nielsen, T.A. (2007). Disturbed dreaming, posttraumatic
stress disorder, and affect distress: A review and neurocognitive
model. Psychological Bulletin ,133, 482–528.
McGaugh, J.L. (2004). The amygdala modulates the consolidation
of memories of emotionally arousing experiences. Annual Review
of Neuroscience ,27, 1–28.
Mellman, T.A., David, D., Kulick-Bell, R., Hebding, J., & Nolan, B.
(1995). Sleep disturbance and its relationship to psychiatricmorbidity after Hurricane Andrew. Amerian Journal of Psychiatry ,
152, 1659–1663.
Nielsen, T.A., Laberge, L., Tremblay, R., Vitaro, F., & Montplaisir, J.
(2000). Development of disturbing dreams during ado-lescence and their relationship to anxiety symptoms. Sleep, 23,
727–736.
Nielsen, T.A., & Levin, R. (2007). Nightmares: A new neurocognitive
model. Sleep Medicine Reviews ,11,
295–310.
Volume 18—Number 2 87Ross Levin and Tore Nielsen
at St Petersburg State University on December 9, 2013 cdp.sagepub.com Downloaded from

Nielsen, T.A., Stenstrom, P., & Levin, R. (2006). Nightmare frequency
by age, gender and 9/11: Findings from an Internet questionnaire.
Dreaming ,16, 145–158.
Nofzinger, E.A. (2004). What can neuroimaging findings tell us about
sleep disorders? Sleep Medicine ,5(Suppl. 1), S16–S22.
Wood, J.M., Bootzin, R.R., Rosenhan, D., Nolen-Hoeksema, S., &
Jourden, F. (1992). Effects of the 1989 San Francisco earthquakeon frequency and content of nightmares. Journal of Abnormal
Psychology ,101, 219–224.Woodward, S.H., Arsenault, N.J., Murray, C., & Bliwise, D.L. (2000).
Laboratory sleep correlates of nightmare complaint in PTSDinpatients. Biological Psychiatry ,48, 1081–1087.
Zadra, A., & Donderi, D.C. (2000). Nightmares and bad dreams: Their
prevalence and relationship to well-being. Journal of Abnormal
Psychology ,109, 273–281.
Zadra, A., Pilon, M., & Donderi, D. (2006). V ariety and intensity of
emotions in nightmares and bad dreams. Journal of Nervous and
Mental Disease, 194, 249–254.
88 V olume 18—Number 2Nightmares, Dreaming, and Emotion Regulation
at St Petersburg State University on December 9, 2013 cdp.sagepub.com Downloaded from