International Journal of Psychophysiology 46 (2002) 227–241

Probing awareness during sleep with the auditory odd-ball

paradigm
Kimberly A. Cote*
Psychology Department, Sleep Research Laboratory, Brock University, St. Catharines, Ont., Canada L2S 3A1

Received 7 July 2002; received in revised form 13 August 2002; accepted 3 September 2002

Abstract

During the waking state, a late component of the auditory event-related potential, P300, is elicited when subjects
detect a rare ‘target’ stimulus. It is usually not elicited when subjects either ignore or fail to detect the stimulus. The
presence of P300 is therefore thought to reflect conscious processing of the stimulus. Since P300 has been shown to
be an attention-dependent cognitive component in wakefulness, one might suppose that it would be absent during
sleep—a time in which information processing of external stimuli is commonly thought to be inhibited. This review
examines the presence or absence of P300 in studies employing auditory odd-ball paradigms in sleep. Research to
date indicates that P300 can be recorded during the transition to sleep and then reappears in REM sleep. Stimuli that
are rare and intrusive are more likely to elicit the classic parietal P300 in REM sleep. There is, however, little or no
positivity at frontal sites. This is consistent with brain imaging studies that show frontal deactivation is characteristic
of REM sleep. These findings indicate that while sleepers may be able to detect stimulus deviance in stage 1 and
REM, the frontal contribution to consciousness may be lost. In non-REM sleep, a later positive wave at 450 ms does
not vary according to experimental manipulation in the same way as the waking and REM P300s. This non-REM
sleep-related positivity may therefore underlie mechanism distinct from the waking P300.
䊚 2002 Elsevier Science B.V. All rights reserved.

Keywords: Sleep; REM sleep; Event-related potential; P300; Consciousness

1. Introduction ‘«a recurring state of existence characterized by 1.

The nature of sleep has often been described on
the basis of observable behaviours in the sleeper.
These behaviours include: reduced movement,
diminished response to external stimuli, a stereo-
typical sleep posture, and the ability to awaken
(i.e. reversibility distinguishing it from coma and
death) (Flanigan, 1972). Anch et al. (1988)
describe sleep as:
*Tel.: q1-905-688-5550x4806; fax: q1-905-688-6922.
reductions in awareness of and interaction with the environ-
ment, 2. lowered motility and muscular activity, and 3.
partial or complete abeyance of voluntary behavior and
consciousness (p. 2).’
Similarly, Carskadon and Dement (2000) define
sleep according to the extent to which information
is processed:
‘«sleep is a reversible behavioral state of perceptual
disengagement from and unresponsiveness to the environ-
ment (p. 15).’

0167-8760/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 8 7 6 0 Ž 0 2 . 0 0 1 1 4 - 9
228 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
Most definitions of sleep refer to inhibition of
processing stimuli from the external environment,
suggesting that there is little or no consciousness
during sleep. Indeed, sleep and wakefulness are
often described as unconscious and conscious
states. Perhaps sleep is commonly thought of as
an unconscious state, where awareness of stimuli
is blocked, because of reflection upon subjective
experience in which memory for external events
during sleep is quite impaired upon subsequent
waking recall. Broughton proposed that ‘retrograde
amnesia’ upon awakening may prohibit accurate
recall of internal mental events (e.g. dreams)
(Broughton, 1973). Similarly, subjective report
cannot be used as a reliable way to understand
information processing of external events during
sleep. As will become apparent in the present
review, when appropriate methods are employed
which allow the experimenter to access the mental
state of the sleeper, a very different picture of
awareness during sleep emerges. Event-related
potential (ERPs) offer a probe of awareness during
sleep.
2. Measurement of sleep
The description of different ‘stages’ of sleep
arose from observations of variability in arousal
thresholds and EEG characteristics across the
night. Rechtschaffen and Kales (1968) edited a
standardized manual for sleep scoring methods
that remains the primary guide for stage classifi-
cation today. Stage 1 is characterized by reduction
in the relative amount of alpha frequencies (8–12
Hz) and the appearance of slower theta waves (4–
7 Hz) compared to wakefulness. This light stage
of sleep is usually accompanied by slow rolling
eye movements, which distinguish it from wake-
fulness and REM sleep. Many researchers consider
stage 1 to be a transition period between the
waking and the sleeping states, rather than a stage
of sleep per se (Johnson et al., 1976; Ogilvie et
al., 1991). Johnson et al. (1976) initially proposed
that sleep was more closely tied to stage 2 and the
appearance of phasic events—K-complexes and
spindles. They observed that subjects were able to
detect external signals within stage 1 sleep. In
stage 2, subjects were no longer able to respond
to these stimuli, and K-complexes and spindles
occupied the EEG tracing. In this definition, sleep
occurs when subjects are no longer conscious of
the external environment (i.e. no longer signaling
awareness with a response). Stages 3 and 4 sleep
are referred to as slow-wave sleep (SWS) because
of the predominance of large amplitude delta
(0.25–3 Hz, )75 mV) waves in the EEG. SWS
is often also referred to as ‘deep’ sleep because
arousal thresholds are lower than in any other
stage of sleep (Rechtschaffen et al., 1966).
Based on physiologic criteria, Carskadon and
Dement (2000) summarize non-REM sleep as ‘«a
relatively inactive yet actively regulating brain in
a moveable body’, and describe REM sleep as
‘«a highly activated brain in a paralyzed body’.
Non-REM sleep consists of stages 1 through 4.
The transition from non-REM to REM sleep is
marked by reduction in EMG activity and by the
presence of binocularly synchronous rapid eye
movements. REM sleep appears approximately
every 90 min throughout the night with each period
being longer in duration. The electrophysiologic
characteristics of REM sleep are more similar to
wakefulness and stage 1 than to non-REM stages
of sleep (for review, see Kahn et al., 1997). During
REM, wakefulness, and stage 1, the EEG consists
of high-frequency and low-amplitude activity.
Despite this similarity, a unique feature of REM is
the presence of muscle atonia. REM sleep is
alternatively referred to as ‘paradoxical sleep’
because although the EEG appears like wakeful-
ness, it is extremely difficult to awaken the sleeper.
3. Cognition and consciousness during sleep
The study of differences between sleep and
wakefulness, as well as variations within sleep, is
referred to as ‘state-dependent’ aspects of con-
sciousness (Kahn et al., 1997), and is considered
to be within the domain of cognitive neuroscience.
Information from the external environment is obvi-
ously processed throughout all stages of sleep, as
evidenced by the fact that the sleeper will reliably
be aroused by increasingly intense stimuli or the
familiar sound of an alarm clock. Moreover, the
sudden call of one’s own name will often quickly
arouse the sleeper. Determining the extent of such
 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
awareness during sleep (e.g. stimulus meaning or
conscious processing); however, is difficult to
ascertain.
A major problem with the study of information
processing during sleep is that the experimenter
does not have access to the mental state of the
sleeping subject. During the waking state, the
researcher can ask subjects to make overt responses
by providing subjective reports of their mental
activity in order to determine the extent of cogni-
tive processing. Subjective reports cannot be com-
municated in sleep. The experimenter could ask
the subject to signal awareness, for example, by
pressing a button. Again, this would be difficult
during sleep. Muscle tonus may be so low that the
sleeper is essentially paralyzed. If the sleeper fails
to overtly respond to the external stimulus, does
this mean they are not conscious of the stimulus
or rather that they are conscious but cannot execute
the motor response? Or, perhaps the sleeper has
detected the stimulus but has forgotten the instruc-
tion to respond which was given in the waking
state. Researchers may awaken the sleeper and ask
for reports of prior mental activity. The inability
to report awareness of prior stimulus presentation
may be due to a failure of memory storage or
retrieval. As in the waking state, the interpretation
of a failure to respond is equivocal.
In spite of these difficulties, evidence has accu-
mulated to suggest that some form of awareness
or consciousness may persist during sleep. In non-
REM sleep, the so-called ‘disorders of arousal’
such as sleepwalking and sleeptalking provide
evidence of a type of consciousness within sleep
(for review, see Broughton, 1982). During sleep-
walking, there is obviously some degree of aware-
ness of the external environment since the person
typically navigates through their home successful-
ly. The behaviours carried out during sleepwalking
may seem quite purposeful, including eating,
coherent conversations, driving a car, violence, and
even homicide (Guilleminault et al., 1995).
Although there is complete amnesia of the expe-
rience upon awakening, the failure of memory
does not provide evidence that the sleepwalker
was unaware or unconscious of their physical
actions at the time of the somnambulism.
229
There is some evidence for consciousness during
REM sleep. Support for this comes from studies
of incorporation of stimuli into dream content,
lucid dreaming, arousal thresholds, and the brain’s
response to external, relevant stimuli (evoked
potentials). The incorporation of external stimuli
into dream content during REM sleep suggests
that consciousness may be possible in sleep.
Dement and Wolpert (1958) provided early reports
that externally presented stimuli could be reliably
incorporated into dreams during EEG-defined
REM sleep. Their report has been subsequently
validated (for review, see Burton et al., 1988),
indicating that incorporation of external stimuli
into dream content are more favourable during
REM sleep. Of course, the sleeper need not nec-
essarily be conscious of the external stimulus in
order to incorporate it into a dream.
It is not easy to dismiss evidence of conscious-
ness provided by lucid dreamers. A lucid dream is
one in which the sleeper becomes aware of the
fact that they are dreaming (LaBerge and Levitan,
1995). They are typically recorded during REM
sleep. In the laboratory, it has been shown that
sleepers are able to communicate their ‘lucidity’
to the experimenter. For the most part, muscles
are largely inhibited during REM sleep. An excep-
tion is the muscle controlling eye movements. In
order to communicate lucidity, the subject employs
a pre-established pattern of eye movements. Thus,
sleepers are able to not only become conscious of
the fact that they are dreaming, but also to con-
sciously signal to the experimenter that they are
dreaming. It is possible that the lucid dream is
triggered because of a brief arousal from REM to
wakefulness. In support of this, Ogilvie et al.
(1982) reported that lucid dreaming was associated
with increased alpha EEG activity, a neurophysiol-
ogical marker typically associated with relaxed
wakefulness. Even if one accepts that lucid dream-
ers are indeed conscious of their mental activity
within sleep, only a small proportion of the pop-
ulation are able to do so.
Arousal thresholds provide an indirect measure
of consciousness. In the usual arousal threshold
study, subjects are presented with auditory stimuli
at different times of the night. Stimulus intensity
is gradually increased until the subject awakens.
230 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
This arousal threshold is used to infer depth of
sleep across the night. Auditory arousal thresholds
have been shown to vary according to sleep stage,
time of night, salience of the stimulus, and stimu-
lus intensity (Williams et al., 1964; Rechtschaffen
et al., 1966; Busby et al., 1994). Arousal thresholds
are highest in SWS relative to both stage 2 and
REM. As the night progresses, there is a general
decrease in arousal threshold, indicating that sleep-
ers can be more easily awakened by stimuli in the
early morning hours (Williams et al., 1964). Stage
2 and REM sleep occur more frequently in the
second-half of the night, however, whereas SWS
(for which the arousal threshold is high) occurs
mainly in the first-half of the night. Meaningful
stimuli such as the sleeper’s name or a baby’s cry
require less intensity to evoke an arousal from any
stage of sleep (Oswald et al., 1960). These elicited
awakenings indicate that detection of meaningful
stimuli is maintained during sleep.
Another way to investigate consciousness during
sleep is through the use of ERPs. They are changes
in electrical activity of the nervous system that are
‘evoked’ by an external physical stimulus or ‘emit-
ted’ by an internal psychological event (Picton et
al., 1995). These psychological events include
alteration of the subject’s level of attention or
arousal, decision-making, memory comparison, or
relevance of the stimulus. ERPs offer a method to
probe the extent of information processing in sleep
independently of the subject’s overt response (for
review, see Campbell et al., 1992; Bastuji and
´
Garcıa-Larrea, 1999; Atienza et al., 2001).
4. The P300 component of the auditory ERP
The P3 component of the ERP is most often
elicited in the ‘odd-ball’ task. In this paradigm,
the subject is presented with a train of frequently
occurring ‘standard’ stimuli. At odd and random
times, the stimulus is changed. The subject may
be asked to detect this ‘target’ usually by keeping
a running mental count or by button pressing upon
its presentation. If the rare target is detected, P3,
a positive is elicited (Sutton et al., 1965; Donchin,
1981; Picton, 1992; Verleger, 1997). When the
target is easy to detect (i.e. deviance between the
standard and target is large), P3 peaks at approxi-
mately 300 ms (thus the label ‘P300’). If the
subject ignores the stimuli, P300 is not elicited
upon presentation of the rare stimulus (Duncan-
Johnson and Donchin, 1977). Moreover, even if
the subject is actively engaged in the signal detec-
tion task, but fails to detect the target, P300 is not
elicited. Thus, the presence of a P300 has been
said to reflect stimulus discrimination or conscious
detection of the target stimulus (Picton, 1992).
P300 is thought to be elicited when controlled
processing is engaged (i.e. slow, effortful, and
conscious processing); rather than automatic proc-
essing which is fast and unchecked.
The absence of a P300 is more ambiguous.
Subjects can make detections yet a P300 will not
always be elicited. For example, if subjects are
asked to push one button upon detection of the
frequently presented ‘standard’ stimulus and anoth-
er button upon detection of the infrequently pre-
sented target, a P300 will be elicited following the
target. No P300 (or only a very small amplitude
P300) will be visible following the standard, even
though the subject clearly signalled their awareness
of it. Thus, while the presence of a P300 can be
used as positive evidence of consciousness, its
absence cannot be used to infer unconsciousness.
Although P300 is generally not elicited in inatten-
tive subjects, there is an exception to this rule. If
the rare stimulus is sufficiently loud as to ‘intrude’
into consciousness, a P300 may be elicited (Put-
nam and Roth, 1990). An infrequently presented
stimulus that is intrusive, novel, or salient in
meaning will likely force the attention of the
subject, despite no specific task instruction.
Since the first report of this late positive com-
ponent (Sutton et al., 1965), the P300 has been
widely studied and the conditions required to elicit
it have been extensively outlined (Donchin, 1981;
Picton, 1992; Verleger, 1997). Vaughan and Ritter
(1970) first described the scalp distribution of the
P300 as maximum in amplitude at mid-parietal
regions. The scalp distribution of the P300 has
been shown to be somewhat variable depending
on a number of factors, including modality of
presentation and task demands (Johnson, 1993).
In light of this widespread variability, the source
generators of the P300 have been speculated to be
multi-focal (Picton, 1992), perhaps dependent on
 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
the specific task and modality of stimulus presen-
tation. The amplitude of P300 is inversely propor-
tional to the probability of the occurrence of the
target (Duncan-Johnson and Donchin, 1977). The
functional significance of the P300 wave remains
disputed (see review by Verleger, 1988 and com-
mentary). A commonly cited function of P300 is
that of contextual updating of working memory
(Donchin, 1981; Donchin and Coles, 1988). The
memory for the frequently occurring stimulus is
well formed. The memory for the rare stimulus is
relatively less well formed. Upon presentation of
the rare stimulus, its memory representation needs
to be revised or updated. More generally, it seems
that the presence of P300 indicates that information
has been consciously processed. Following this
level of information processing, responses includ-
ing memory updating, decision-making, and
behaviours may take place.
According to Donchin and Coles (1988), a true
P300 must meet certain criteria: maximum over
parietal–central areas of the scalp; peak at approx-
imately 300 ms (if target detection is easy); and
vary inversely in amplitude as a function of stim-
ulus probability. A number of other late positivities
can also be recorded to rare stimuli. These may be
distinguished by their response to experimental
manipulation and scalp topography. An earlier P3
peaking at approximately 250 ms has a prominent
frontal distribution (Squires et al., 1975). It can
be elicited by highly novel and unexpected stimuli
even if the subject is not actively attending to the
stimulus. This ‘P3a’ is thus considered to reflect
an automatic detection of novelty. The P300 (also
called P3b to distinguish from P3a), peaking later
at approximately 350 ms, has the more typical
parieto-central distribution. Unlike P3a, the ampli-
tude of P3b or P300 is highly dependent on the
level of attention of the subject. Researchers have
also described a P3-like wave that is elicited by
novel environmental sounds (e.g. animal sounds
or music) (Fabiani and Friedman, 1995; Spencer
et al., 1999). It is similar to the P3a in that it has
a more anterior scalp topography than the P3b. Its
latency (350 ms) is too late to be considered a
P3a. It has thus been labeled the ‘novel-P3’.
231
5. Can P300 be recorded in sleep?
If P300 cannot be elicited in awake but inatten-
tive subjects, it is perhaps unlikely that it can be
elicited in sleeping subjects. In earlier reviews,
Campbell et al. (1992) and Harsh (1994) both
claimed that there was little evidence that P300
´
could be elicited during sleep. Bastuji and Garcıa-
Larrea (1999) note that detection of stimulus
deviance and salience remains possible in stage 2
sleep as revealed by variation in the K-complex
with manipulation of stimuli parameters. They
conclude that the mechanisms underlying deviance
detection in REM sleep seem to be more similar
to those in the waking state, as indicated by the
presence of a P300-like waveform. Atienza et al.
(2001) point to evidence suggesting that P3a
(automatic orienting response) may be possible in
sleep. They indicate that studies reporting P3b or
P300 in stage 2 differ in amplitude, latency and
topography from the waking P300. Despite these
careful reviews of the literature, a controversy
remains as to whether a ‘true’ P300, one that
underlies the same mechanisms as the waking
endogenous P300, may be elicited in any stage of
sleep.
What evidence should we be willing to accept
that a positive wave evoked by auditory stimuli in
sleep represents the P300? For a positive peak in
sleep to be considered functionally equivalent to
the waking P300, it should have the following
characteristics: a target-specific, parietal maxi-
mum, positive peak at approximately 300 ms,
whose amplitude varies inversely with the proba-
bility of target presentation. If these criteria are
not met, a positive peak in sleep would be consid-
ered suspect as a candidate P300. The auditory
odd-ball paradigm is ideal to investigate the pres-
ence of P300 in sleep for a number of reasons.
First of all, the characteristics of P300 in the
waking state have been well described using such
stimulus parameters. For instance, the 2-stimulus
paradigm is necessary to show that the candidate
P300 varies in amplitude with changing probability
of the deviant stimulus. While a single stimulus
may elicit a P300 when it occurs infrequently, the
odd-ball paradigm permits evaluation of stimulus
discrimination capabilities. Lastly, the use of audi-
232 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
tory stimuli assures constancy of stimulus input to
the inner ear in spite of variation of head position
and body movement (when the acoustic stimuli
are delivered through earphone inserts).
Early studies employing pitch odd-ball para-
digms during non-REM sleep reported a late pos-
itive wave following the deviant stimulus
(Wesensten and Badia, 1988; Nielsen-Bohlman et
al., 1991; Van Sweden et al., 1994; Winter et al.,
1995). Refer to Table 1 for an overview of studies
employing auditory odd-ball paradigms during
sleep. The delayed latency of these positive waves,
between 400 and 800 ms, raises suspicion as to
their functional similarity with the waking P300.
The late positivity recorded in non-REM sleep is
more consistent with components of the evoked
K-complex (Bastien and Campbell, 1992). Rare
deviant stimuli will also elicit the K-complex in
non-REM sleep.
Salisbury et al. (1992) and Salisbury and
Squires (1993) reasoned that an intrusive stimulus
would be most likely to elicit a P300 in sleep.
They therefore investigated the effects of an inten-
sity odd-ball paradigm in non-REM sleep. P300
was not elicited in either wake or sleep states
when a rare low intensity stimulus (40 dB nHL)
was employed; however, a late positive wave was
elicited in stage 2 sleep when a louder deviant (60
dB nHL) was employed. There were a number of
methodological and interpretive problems with this
study. In the modified odd-ball paradigm, the
deviant stimulus changed in both intensity and
pitch. Thus, the effects cannot be said to be due
to the intensity of the rare stimulus. Furthermore,
the small amplitude positivity (f2 mV) peaked at
approximately 400 ms, but could only be detected
when employing peak-to-peak (N2–P3) measure-
ment because of the overlapping negativity of the
K-complex. To what extent this positive wave was
related to the K-complex cannot be determined
from their analysis. To demonstrate that a positive
wave in sleep is independent of the K-complex,
trials in which a K-complex is elicited need to be
averaged separately from those in which it is not.
This is usually done by visual inspection of the
continuous EEG record. If the positive wave is
independent of the K-complex, it should still occur
in the non-K-complex trials. The influence of the
K-complex can also be minimized with amplitude
rejection criteria set to remove trials exceeding
"100 mV in stage 2 sleep. In a subsequent paper,
Salisbury and Squires (1993) report on two sepa-
rate experiments run to investigate the effects of
pitch and intensity deviance on P300 in non-REM
sleep. Once again, peak-to-peak measurement
methods were employed and K-complexes were
not removed. In the first experiment, a late positive
peak that authors label ‘P3’ was identified in stage
2 sleep. It did not vary in amplitude or latency
across five different conditions where pitch devi-
ance was manipulated (Table 1). The candidate
P300 had an unusual occipital scalp distribution.
In the second experiment, four different intensity
deviants were delivered in separate conditions. The
latency of the P3, (stated as 388 ms in their text
but appearing closer to 450 ms in the grand
average waveforms shown), did not differ between
stages 2 and SWS, nor did it differ across the four
intensity deviant conditions. The amplitude of the
very small positive-going wave was again unusual,
being measured as a negative potential at all
electrode sites. It was maximum (less negative) at
Oz.
In contrast to the above studies, many investi-
gators have failed to record P300 in any stage of
sleep. Paavilainen et al. (1987) did not find evi-
dence of P300 activity in non-REM sleep, even
though a pitch odd-ball task was employed. They
suggested that the absence of P300 could have
been due to the small magnitude of stimulus
deviance (only 50 Hz). Loewy et al. (1996) used
a similar pitch odd-ball task. In different condi-
tions, a ‘small’ (1050 Hz) or ‘large’ (2000 Hz)
deviant was presented infrequently. There was still
no evidence of a late positive wave. Similarly,
Nordby et al. (1996) found no P300 in sleep when
pitch stimuli were presented at 800 and 1200 Hz
in an odd-ball paradigm.
There have been recent reports of more con-
vincing P300-like waves in REM sleep when pitch
odd-ball paradigms were employed (Niiyama et
al., 1994; Bastuji et al., 1995). Both studies report-
ed small amplitude P300s (-4 mV) during REM
sleep. Neither investigation observed P300 activity
in non-REM sleep. In addition to recording ERPs
during REM sleep, these studies had two further
Table 1
Review of studies employing auditory odd-ball paradigms in sleep
a b
Frequent stimuli Infrequent stimuli Sleep stage Properties of candidate P300
Paavilainen et al. (1987) 1000 Hz (90%); 75 dB; 1050 Hz Wake Passive: P250–300 ms (P3a)
50 ms; ISI 510 ms Stage 2 No P300
REM Not reported
Wesensten and Badia (1988) 1000 Hz (70 or 50%); 45 dB; In separate conditions: Wake Active: P300
50 ms; ISI 1 s Night 1: 1400 Hz (30%) Stage 2 P700

K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241

Night 2: 1400 Hz (50%) REM P700
Nielsen-Bohlman et al. (1991) 1000 Hz (80%); 50 dB; 1500 Hz (10%) Wake Passive: P300 (only to novel stimulus)
50 ms; ISI 1 s Complex novel tones (10%) Stage 2 P420 (larger to novel stimulus)
(novel tones: 44 dB; 300 ms) REM Not reported
Salisbury et al. (1992) ISI 1.2 s; In separate conditions: Wake Passive: P300
Soft: 3 kHz, 60 dB (90%) Soft: 250 Hz; 40 dB Stage 2 Loud deviant: P400
Loud: 250 Hz, 40 dB (90%) Loud: 3 KHz; 60 dB REM Not reported
Salisbury and Squires (1993) Experiment 1: In separate conditions: Wake Not reported
1000 Hz (90%); 50 dB; 1.5, 2.0, 2.5, 3.0, 3.5 KHz Non-REM P404 (no difference across conditions)
100 ms; ISI 1.9 s REM Not reported
Experiment 2: In separate conditions: Wake Not reported
2000 Hz (90%); 50 dB; 55, 60, 65, 70 dB Non-REM P388 (no difference across conditions)
100 ms; ISI 1.9 s REM Not reported
Van Sweden et al. (1994) 1000 Hz (86%); 2000 Hz Wake Passive: no P300
100 ms Stage 2 P800
REM P500
Niiyama et al. (1994) 1000 Hz (80%); 60 dB; 2000 Hz Wake Active: P351
100 ms; ISI 1.5–2.5 s Stage 2 No P300 (no P300, KCs removed)
REM P361
Bastuji et al. (1995) 1000 Hz (90%); 70 dB; 2000 Hz Wake Active: P344
60 ms; ISI 1.2 s (5% variance) Stage 2 No P300 (P450)
REM P455
Winter et al. (1995) 1000 Hz (80%); 65 dB; 1200 Hz (10%); 2000 Hz (10%) Wake Active: P300; passive: P3a
50 ms; ISI 1 s Drowsy: P430
Stage 2 No P300 (smaller P430 to large deviant)
REM Not reported
Loewy et al. (1996) 1000 Hz (80%); 80 dB; In separate conditions: Wake Passive: no P300
55 ms; ISI 600 ms 1050 Hz (20%) Stage 2 No P300
2000 Hz (20%) REM No P300
Nordby et al. (1996) 800 Hz (80% or 20%); 75 dB; 1200 Hz (20% or 80%) Wake Passive: no P300

233
50 ms; ISI 1.5 s Non-REM No P300
 234
Table 1 (Continued)
a b
Frequent stimuli Infrequent stimuli Sleep stage Properties of candidate P300
REM No P300
Sallinen et al. (1996) 1000 Hz (97%); 50 dB; 1100 Hz (1.5%); 2000 Hz (1.5%) Wake Passive: P250–P550
50 ms; ISI 625 ms Stage 2 Not reported
REM P450–P550
Cote and Campbell (1999a) 70 dB (95%); 1000 Hz; 90 dB Wake Not reported
50 ms, ISI 1.5 s Stage 2 No P300 (P446 KCs removed)

K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241

REM P321
Cote and Campbell (1999b) 1000 Hz; 0, 60, 80, 100 dB; Equiprobable delivery Wake Passive: P373 (100 dB)
50 ms; ISI 2.0 s Stage 2 No P300 (KCs present)
REM P363
Cote et al. (2001) 80 dB; 1000 Hz; In separate conditions: Wake Active: P300 in all conditions
50 ms; ISI 1.0–2.0 s Pitch: 2000 Hz (05, 10, 20%) Passive: P315 (100 dB, 5% condition)
Intensity: 100 dB (05, 10, 20%) Stage 2 No P300 (P420)
REM Intensity (5%): P365
Note: This list does not include all studies in which odd-ball paradigms were employed (e.g. some studies of K-complexes or MMN used stimulus parameters or
sweep times such that P300 could not be ascertained).
a
Column 2 lists the stimulus parameters for the frequent stimulus (pitch, intensity, probability, duration, and inter-stimulus interval). In column 3, the stimulus
parameters for the deviant stimulus are the same as those listed for the frequent stimulus unless otherwise stated.
b
Active waking indicates a behavioural response was required to target stimuli; passive waking indicates no response was required.
 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241 235

strengths over prior studies: both topographic

recordings and waking data collection for appro-
priate comparison. During the waking state, sub-
jects were asked to detect rare, deviant stimuli.
Neither study included a waking condition in
which subjects ignored the stimulus. An ignore
condition would provide a more appropriate com-
parison to sleep since subjects also do not respond
while asleep (and presumably also ignore the
stimulus). The REM P300 reported by Niiyama et
al. is similar to the waking-P300 in polarity and
latency; however, it was only apparent in the
difference wave, having a much reduced amplitude
and an unusual occipital scalp distribution. The
positivity reported by Bastuji et al. had the expect-
ed scalp distribution, but was considerably delayed
in latency and reduced in amplitude. Other claims
of P300 in REM sleep, specifically following pitch
deviants, have been made; however, the latency
was quite delayed, peaking at approximately 500 Fig. 1. ERPs following rare auditory stimuli in the presence
ms (Van Sweden et al., 1994; Sallinen et al., and absence of K-complexes. The dark solid line represents
1996). The REM-related positivity reported by trials on which K-complexes were evoked by auditory stimuli
(KCq). The lighter solid line represents trials on which no K-
Niiyama et al. and Bastuji et al. may be a candidate complexes were apparent (KCy). A late positivity, P450, can
P300; however, the experimental manipulations be observed in both KCq and KCy trials at the occipital site.
that would allow for comparison with a wake- It is masked by the large negativity associated with the K-
P300 (i.e. waking-ignore condition, varying prob- complex (KCq trials) at frontal sites. Note: positive is an
ability) were not carried out. upward deflection in the waveform in this and all other figures.
(Figure is re-published from Cote and Campbell (1999a).)
Based on the suggestion by Salisbury et al.
(1992) that a loud stimulus may be more likely to
elicit a P300, and following from recent studies ent when K-complexes were absent. This non-
depicting P300 in REM sleep, a series of studies REM P450 was however maximum over
were carried out using odd-ball paradigms where occipito-parietal areas of the scalp (Fig. 1).
the rare stimulus varied in intensity rather than Because of the delayed latency and unique scalp
pitch (Cote and Campbell, 1999a,b; Cote et al., topography, the non-REM P450 was not consid-
2001). In addition, attempts were made to deter- ered to be a candidate P300.
mine if P300 in sleep varied according to manip- Cote and Campbell (1999b) again examined the
ulations in the same manner as the endogenous role of stimulus intensity in eliciting a P300 during
ERP recorded in wakefulness. Although P300 is REM sleep. Auditory tone pips having an intensity
typically only evoked when the subject attends to of either 0, 60, 80 or 100 dB SPL were delivered
the stimulus, it may also be elicited in an awake, at random with equal probability. During the wak-
inattentive subject if the stimulus is sufficiently ing state, when subjects read a book, the loud 100-
intrusive. An odd-ball task was therefore employed dB stimulus elicited short (P3a) and long latency
to determine if high intensity stimuli would elicit (P300) positive waves (peaking at 293 and 373
the P300 during sleep (Cote and Campbell, ms, respectively). In stage 2 non-REM sleep, a
1999a). A large amplitude parieto-central P300, P300 could not be observed in stage 2 sleep
peaking at 321 ms, was apparent in REM sleep regardless of the level of stimulus intensity. During
following loud deviant stimuli. In stage 2 sleep, a REM sleep, a P300 (latency 363 ms) was elicited
later positive wave, peaking at 446 ms, was appar- by the 100-dB stimulus. Contrary to Sallinen et al.
236 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
(1996), there were no differences in P300 ampli-
tude between phasic and tonic REM. The earlier
positive peak (i.e. P3a) at frontal sites was not
apparent in REM. A P300 was elicited to the
loudest of equiprobable tones without any specific
instruction to respond to a target. In this paradigm,
it is likely that the loudest stimulus (i.e. most
salient) will automatically become an implicit
target, forcing the subject to attend to it
preferentially.
In attempt to resolve reports of pitch-elicited
P300s (Niiyama et al., 1994; Bastuji et al., 1995)
and intensity-elicited P300s in REM sleep (Cote
and Campbell, 1999a,b), a subsequent study exam-
ined the effects of both types of deviants in the
same subjects during a single night (Cote et al.,
2001). Probability of deviant occurrence was also
manipulated since it is known to affect the ampli-
tude of a true P300. Subjects were randomly
assigned to one of three deviant probability groups,
in which the deviant occurred on 20, 10 or 5% of
trials. During wakefulness, subjects were asked to
either attend to (count the deviants), or ignore
(read a book) the stimuli. A large amplitude P300
was apparent following both pitch and intensity
deviants during the waking attend condition. It
was maximum at parietal sites and was also appar-
ent at fronto-central sites. Its amplitude decreased
with increasing probability of deviant presentation.
In the waking-ignore condition, a slightly smaller
P300 was evident to the intensity deviant. No P300
was evident to any of the pitch deviants in the
waking-ignore condition. Refer to Fig. 2 for ERPs
in wakefulness for pitch and intensity odd-ball
paradigms. During stage 2 sleep, a small amplitude
positive wave was visible following pitch and
intensity deviants. Similar to many of the previous
investigations, its latency was prolonged and its
scalp distribution posterior compared to the waking
P300. During REM sleep, an attenuated positivity
having the usual P300 latency and parieto-central
scalp distribution was visible to the very rare
intensity deviant (Fig. 3). However, the frontal
dispersion of P300 that was observed in the waking
state was absent during REM sleep. In contrast to
studies by Niiyama et al. (1994), Bastuji et al.
(1995), the pitch deviant failed to elicit a late
positivity in this study. A possible explanation for
this discrepancy is that the sleeper may perceive
pitch deviants to be relatively less salient in com-
parison to loud deviants when delivered to the
same subjects during a single night.
6. General discussion
The extent of information processing and con-
sciousness during sleep is of interest to scientists
and philosophers. Some may argue that awareness
of external events is not possible, particularly
during REM sleep when mental resources are
focused toward internal dream activity. Others may
hold the view that consciousness must be possible
in order for the sleeper to be aware of their dreams,
as in the case of lucid dreaming. A major meth-
odological problem in this area of research is how
to define and measure consciousness. Conscious-
ness may be measured in a number of ways—on
the basis of the EEG, subjects could be said to be
awake or asleep (conscious or unconscious). On
the basis of the subject’s response to externally
presented targets, they could be said to be con-
sciously aware of the stimuli. Finally, on the basis
of P300, subjects could also be said to be aware
of the external stimulus, even in the absence of a
behavioural response. The P300 is especially crit-
ical to defining consciousness within sleep, a
period in which behavioural responses are rare.
The presence of a P300 provides neurophysiolog-
ical evidence for high-level processing (i.e. stim-
ulus discrimination or perhaps conscious
processing). Its absence is ambiguous.
In the waking state, the endogenous ERP com-
ponent, P300, may be identified as a target-specif-
ic, parietal maximum positive wave, peaking at
approximately 300 ms (in healthy adults when the
discrimination task is easy). A true P300 will vary
in amplitude inversely with the probability of
target stimulus presentation (Duncan-Johnson and
Donchin, 1977). P300 is typically elicited only if
the subject is attentive to the stimuli and detects
the relevant stimulus. An exception to this rule is
stimuli that are very loud will elicit a P300 in
subjects whose attention is directed elsewhere
(Putnam and Roth, 1990), possibly because the
loud stimulus intrudes into consciousness. In this
case, the loud stimulus becomes an implicit target
 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241 237

Fig. 2. ERPs following rare (5%) pitch and intensity deviants in waking-attend and waking-ignore conditions. (Note: solid line
represents deviant stimuli; dotted line represents standard stimuli.) P300 is apparent following the loud stimulus even when subjects
are instructed to ignore it. No P300 can be seen following rare pitch deviants in the waking-ignore condition. (Figures are re-
published from Cote et al. (2001).)
238 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
Fig. 3. ERPs following rare (5%), loud (95 dB) pitch vs. inten-
sity deviants in REM sleep. (Note: solid line represents inten-
sity odd-ball; dotted line represents pitch odd-ball.) A parietal
maximum positive peak occurs following the intensity deviants
only. (Figure is re-published from Cote et al. (2001).)
where the subject is forced to attend to it as the
more salient stimulus, regardless of whether or not
there is instruction to do so.
Early sleep studies considered a late positive
wave in non-REM sleep, peaking at approximately
700–900 ms, to be a candidate P300 (refer to
Table 1). These late positivites were most likely
in fact components of the auditory evoked K-
complex (see other reviews in the issue for addi-
tional information related to the evoked
K-complex). Other studies have reported small
amplitude positive waves approximately 400–450
ms in stage 2 sleep, following presentation of a
rare, deviant stimulus. The similarity of this non-
REM P450 to the attention-dependent P300 of
wakefulness remains to be shown through experi-
mental manipulation. The non-REM P450 differs
from the waking P300 in its latency, target-speci-
ficity, and scalp topography. It is thus generated
by distinct brain mechanisms and likely represents
functions altogether different from the attention-
dependent P300 in wakefulness. Hull and Harsh
(2001) argue that P450 is a state-dependent wave-
form distinct from the waking P300, largely
because of its reduced amplitude and insensitivity
to task relevance.
The positivity recorded in REM sleep is a more
convincing candidate P300, although there is some
variability in response. Small amplitude positive
waves have been recorded following pitch deviants
(Niiyama et al., 1994; Bastuji et al., 1995), how-
ever, their latency and scalp distribution were not
similar across studies. Furthermore, a REM P300
could only be identified in 75% of subjects in
Niiyama et al. and 73% of responses (across all
subjects) in Bastuji et al. Still other studies
employing similar pitch odd-ball paradigms failed
to discern a positive wave in the expected latency
range (Table 1). Studies employing intensity odd-
ball paradigms to investigate P300 in REM sleep
have recorded positivities that conform more to
the waking P300 (i.e. parietal maximum, target-
specific, -400 ms). It is perhaps the relevance of
these loud deviants that triggers the P300. This is
supported by recent studies employing name stim-
uli to investigate P300 in REM sleep (reviewed
elsewhere in this issue).
These auditory odd-ball studies indicate that the
subject appears to have some awareness of the
external environment during REM sleep, in a
manner similar to that during wakefulness, as
indicated by the consistent parietal P300. It is
perhaps not surprising that P300 is specific to
REM sleep. The similarities between waking and
REM states have led some to view REM as a
period of wakefulness within sleep, rather than a
different type of sleep (Horne, 1988). In addition,
wakefulness and REM have been shown to be
similar in EEG frequency and amplitude charac-
teristics, as well as other ERP components (e.g.
N1, P2, MMN). The presence of P300 in REM
sleep may indicate that the mechanisms underlying
its function in wakefulness are also engaged during
sleep! That is, active contextual updating of work-
ing memory is taking place in sleep. Alternatively,
the presence of P300 in sleep may reveal that
P300 does not actually reflect such high-level
cognitive processing when evoked in REM sleep;
rather, P300 may reflect a type of automatic or
pre-conscious processing. There is some evidence
that P300 might more accurately reflect a type of
pre-conscious processing. In the case of a neuro-
 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
logical patient with cortical blindness (i.e. damage
to the occipital cortex leading to a ‘blindsight’),
while engaged in a visual odd-ball task, the patient
will fail to signal detection of a rare stimulus, yet
a P300 will still be elicited (Shefrin et al., 1988).
In this circumstance, although evidence of con-
sciousness is not provided by the behavioural
response, P300 is still apparent. The P300 observed
in REM sleep, lacking frontal activity, may be
automatically generated to the relevant stimulus.
What might be the purpose of a stage of sleep
that is characterized by enhanced information proc-
essing capabilities, allowing the sleeper to monitor
the environment? In the case of REM sleep, this
monitoring ability is periodic (i.e. 4–5 REM per-
iods per night), and increases in duration as sleep
progresses (i.e. more monitoring towards early
morning hours). This evidence of heightened
awareness during sleep is reminiscent of a sentinel
theory of REM sleep function. The sentinel theory
proposed that REM sleep serves to periodically
prepare the brain for arousal in the event that the
sleeper might need to respond to threatening stim-
uli (e.g. predators) in the environment (Snyder,
1966). Contrary to the predictions of such a theory,
species that are considered to be more prey than
predator do not have more REM sleep. Neverthe-
less, ERP studies during sleep employing odd-ball
paradigms indicate that REM is a period of
enhanced stimulus processing capabilities. This
appears to be the case for both internal (i.e.
dreams) and external (i.e. stimuli) processing.
Perhaps the purpose of such processing capabilities
during REM sleep is not specifically to protect
against threatening prey, but more generally to
provide sensory stimulation to the brain during a
long bout of sleep. Alternatively, this enhanced
processing may represent a by-product of other
functions of REM sleep.
P300 recorded during stage 1 and REM differs
in its topographic characteristics from that typically
elicited in the waking state. It is either attenuated
or absent at frontal sites. P300 at frontal sites has
been shown to be absent in stage 1 when subjects
fail to respond (Cote et al., 2002), and also in
REM sleep (Bastuji et al., 1995; Cote and Camp-
bell, 1999b; Cote et al., 2001) The absence of a
frontal P300 may be due to ‘functional deafferen-
239
tation’ of the forebrain during sleep. In support of
this, neuroimaging studies of sleep show that REM
sleep is a period of reduced frontal activation
(Hobson et al., 1998; Maquet, 2000). It may be
that frontal contributions to consciousness are
much reduced during REM. Perhaps this explains
the bizarreness and lack of organization character-
istic of REM dreams. The parietal P300 may
indicate that the deviant stimulus is detected;
whereas, the frontal lobes may contribute to the
‘experience’ of consciousness. For instance, loud
stimuli will elicit a startle reflex in waking sub-
jects. It is possible that the subject may not
experience these loud stimuli as ‘startling’ during
REM sleep. This may explain why aspects of the
startle reflex, such as blinking, and changes in
muscle tonus do not follow presentation of the
loud stimulus during REM. The absence of affect
and experience might explain another phenome-
non. Very loud and longer duration stimuli will
normally awaken the sleeper. The short duration
stimuli typically used in sleep studies do not elicit
awakenings. It is possible that awakening from
sleep requires prior frontal activation and arousal.
The contribution of the frontal lobes to conscious-
ness may be further investigated using a stimulus
novelty paradigm during sleep onset and through-
out the various stages of sleep. Fabiani and Fried-
man (1995) showed that novel auditory stimuli
elicit a frontal P3 component.
Individual differences and variability in response
to external stimuli during sleep may explain dif-
ferences in P300 response between and across
studies. Recently, Voss and Harsh (1998) exam-
ined how a subject’s personality characteristics
might be related to information processing at sleep
onset. Individual differences in coping styles were
categorized as Monitors (i.e. information seeking),
and Blunters (i.e. information avoiding). Monitors
were more responsive to stimuli (i.e. behavioural
response) in wakefulness and stage 1. In addition,
the Monitors had a larger P300 response, and a
smaller N350 response. Voss and Harsh (1998)
suggest that the sleep-related N350 response may
reflect blocking of the stimuli. Other personality
characteristics might also influence the attention-
dependent, endogenous P300 component of the
ERP. For instance, it is possible that introverts
240 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
may exhibit P300 in REM at lower thresholds than
extraverts. Individual differences in the sensory
processing of the stimulus (components -50 ms)
may explain differences in the subsequent long-
latency ERPs and the P300 response in particular.
The intra-subject variability in P300 response may
also be further investigated. For example, there
may be night-to-night variability due to habitua-
tion, or fluctuations in many diurnal factors (e.g.
caffeine intake, mood, stress, daytime sleepiness).
In summary, studies employing auditory odd-
ball paradigms demonstrate that P300 may be
recorded during REM sleep, especially following
more salient stimuli such as loud, rare tones. There
is no evidence for either a pitch-elicited or an
intensity-elicited P300 in non-REM stage 2 sleep.
The presence of P300 in REM sleep indicates that
it is a period of enhanced information processing
relative to non-REM sleep. The extent of this
information processing appears to be similar to
that of the waking state. Future research should
focus on the stimulus parameters and individual
differences that contribute to the expression of
consciousness during sleep.
References
Anch, M.A., Browman, C.P., Mitler, M.M., Walsh, J.K., 1988.
Sleep: A Scientific Perspective. Prentice-Hall, Englewood
Cliffs.
Atienza, M., Cantero, J.L., Escera, C., 2001. Auditory infor-
mation processing during human sleep as revealed by event-
related brain potentials. Clin. Neurophysiol. 112,
2031–2045.
Bastien, C., Campbell, K., 1992. The evoked K-complex, all-
or-none phenomenon? Sleep 15, 236–245.
´
Bastuji, H., Garcıa-Larrea, L., 1999. Evoked potentials as a
tool for the investigation of human sleep. Sleep Med. Rev.
3, 23–45.
´
Bastuji, H., Garcıa-Larrea, ` F., 1995.
L., Franc, C., Mauguiere,
Brain processing of stimulus deviance during slow-wave
and paradoxical sleep: a study of human auditory evoked
responses using the odd-ball paradigm. J. Clin. Neurophy-
siol. 12, 155–167.
Broughton, R., 1973. Confusional sleep disorders: interrela-
tionships with memory consolidation and retrieval in sleep.
In: Boag, T.J., Campbell, D. (Eds.), A Triune Concept of
the Brain and Behavior. University of Toronto Press,
Toronto.
Broughton, R., 1982. Human consciousness and sleepywaking
rhythms: a review and some neuropsychological considera-
tions. J. Clin. Neuropsychol. 4, 193–218.
Burton, S.A., Harsh, J.R., Badia, P., 1988. Cognitive activity
in sleep and responsiveness to external stimuli. Sleep 11,
61–68.
Busby, K.A., Mercier, L., Pivik, R.T., 1994. Ontogenetic
variations in auditory arousal threshold during sleep. Psy-
chophysiology 31, 182–188.
Campbell, K., Bell, I., Bastien, C., 1992. Evoked potential
measures of information processing during natural sleep. In:
Broughton, R.J., Ogilvie, R.D. (Eds.), Sleep, Arousal, and
Performance. Birkhauser, Boston.
Carskadon, M.A., Dement, W.C., 2000. Normal human sleep:
an overview. In: Kryger, M.H., Roth, T., Dement, W.C.
(Eds.), Principles and Practice of Sleep Medicine. W.B.
Saunders Company, Philadelphia.
Cote, K.A., Campbell, K.B., 1999a. P300 to high intensity
stimuli during REM sleep. Clin. Neurophysiol. 110,
1345–1350.
Cote, K.A., Campbell, K.B., 1999b. The effects of varying
stimulus intensity on P300 in REM sleep. Neuroreport 10,
2313–2318.
Cote, K.A., de Lugt, D.R., Campbell, K.A., 2002. Changes in
scalp topography of event-related potentials and behavioral
responses during the sleep onset period. Psychophysiology
39, 29–37.
Cote, K.A., Etienne, L., Campbell, K.B., 2001. Neurophysiol-
ogical evidence for the detection of external stimuli during
sleep. Sleep 24, 791–803.
Dement, W., Wolpert, E.A., 1958. The relation of eye move-
ments, body motility, and external stimuli to dream content.
J. Exp. Psychol. 55, 543–553.
Donchin, E., 1981. Surprise! «surprise!? Psychophysiology
18, 493–513.
Donchin, E., Coles, M.G.H., 1988. Precommentary: is the
P300 component a manifestation of context updating?
Behav. Brain Sci. 11, 355–425.
Duncan-Johnson, C.C., Donchin, E., 1977. On quantifying
surprise: the variation of event-related potentials with sub-
jective probability. Psychophysiology 14, 456–467.
Fabiani, M., Friedman, D., 1995. Changes in brain activity
patterns in aging: the novelty oddball. Psychophysiology
32, 579–594.
Flanigan, W.F., 1972. Behavioral states and electroencephalo-
gram of reptiles. In: Chase, M.H. (Ed.), The Sleeping Brain.
BISyBRI, UCLA, Los Angeles.
Guilleminault, R.J., Kushida, C., Leger, D., 1995. Forensic
sleep medicine and nocturnal wandering. Sleep 9, 721–723.
Harsh, J.R., 1994. Event-related potential changes during the
wake-to-sleep transition. In: Ogilvie, R.D., Harsh, J.R.
(Eds.), Sleep Onset. APA, Washington, DC.
Hobson, J.A., Pace-Schott, E.F., Stickgold, R., Kahn, D., 1998.
To dream or not to dream? Relevant data from new neuroim-
aging and electrophysiological studies. Curr. Opin. Neuro-
biol. 8, 239–244.
Horne, J.A., 1988. REM sleep. In: Horne, J.A. (Ed.), Why
We Sleep. Oxford University Press, New York.
 K.A. Cote / International Journal of Psychophysiology 46 (2002) 227–241
Hull, J., Harsh, J.R., 2001. P300 and sleep-related positive
waveforms (P220, P450, and P900) have different deter-
minants. J. Sleep Res. 10, 9–17.
Johnson, L.C., Hanson, K., Bickford, R.G., 1976. Effect of
flurazepam on sleep spindles and K-complexes. Electroen-
cephalo. Clin. Neurophysiol. 40, 67–77.
Johnson Jr., R., 1993. On the neural generators of the P300
component of the event-related potential. Psychophysiology
30, 90–97.
Kahn, D., Pace-Schott, E.F., Hobson, J.A., 1997. Conscious-
ness in waking and dreaming: the roles of neuronal oscilla-
tion and neuromodulation in determining similarities and
differences. Neuroscience 78, 13–38.
LaBerge, S., Levitan, L., 1995. Validity established of
DreamLight cues for eliciting lucid dreaming. Dreaming 5,
159–168.
Loewy, D.H., Campbell, K.B., Bastien, C., 1996. The mismatch
negativity to frequency deviant stimuli during natural sleep.
Electroencephalo. Clin. Neurophysiol. 98, 493–501.
Maquet, P., 2000. Functional neuroimaging of normal human
sleep by positron emission tomography. J. Sleep Res. 9,
207–231.
Nielsen-Bohlman, L., Knight, R.T., Woods, D.L., Woodward,
K., 1991. Differential auditory processing continues during
sleep. Electroencephalo. Clin. Neurophysiol. 79, 281–290.
Niiyama, Y., Fujiwara, R., Satoh, N., Hishikawa, Y., 1994.
Endogenous components of the event-related potential
appearing during NREM stage 1 and REM sleep in man.
Int. J. Psychophysiol. 17, 165–174.
Nordby, H., Hugdahl, K., Stickgold, R., Bronnick, K.S., Hob-
son, J.A., 1996. Event-related potentials (ERPs) to deviant
stimuli during sleep and waking. Neuroreport 7, 1082–1086.
Ogilvie, R.D., Hunt, H.T., Tyson, P.D., Lucescu, M.L., Jenkins,
D.B., 1982. Lucid dreaming and alpha activity: a preliminary
report. Perceptual Motor Skills 55, 795–808.
Ogilvie, R.D., Simons, I.A., Kuderian, R.H., MacDonald, T.,
Rustenburg, J., 1991. Behavioral, event-related potential,
and EEGyFFT changes at sleep onset. Psychophysiology
28, 54–64.
Oswald, I., Taylor, A.M., Treisman, M., 1960. Discriminative
response to stimulation during human sleep. Brain 83,
440–453.
Paavilainen, P., Cammann, R., Alho, K., Reinikainen, K.,
¨¨ ¨
Sams, M., Naatanen, R., 1987. Event-related potential to
pitch change in an auditory stimulus sequence during sleep.
In: Johnson, R., Rohrbaugh, J.W., Parasuraman, R. (Eds.),
Current Trends in Event-Related Potential Research. Elsev-
ier, Amsterdam, pp. 246–255 EEG Suppl. 40.
Picton, T.W., 1992. The P300 wave of the human event-related
potential. J. Clin. Neurophysiol. 9, 456–479.
Picton, T.W., Lins, O.G., Scherg, M., 1995. The recording and
analysis of event-related potentials. In: Boller, F., Grafman,
J. (Eds.), Handbook of Neuropsychology. Elsevier Science
B.V., Amsterdam, vol. 10. pp. 3–73.
Putnam, L.E., Roth, W.T., 1990. Effects of stimulus repetition,
duration, and rise time on startle blink and automatically
elicited P300. Psychophysiology 27, 275–297.
241
Rechtschaffen, A., Hauri, P., Zeitlin, M., 1966. Auditory
awakening thresholds in REM and NREM sleep stages.
Perceptual Motor Skills 22, 927–942.
Rechtschaffen, A., Kales, A., 1968. A Manual of Standardized
Terminology: Techniques and Scoring System for Sleep
Stages of Human Subjects. BISyBRI, UCLA, Los Angeles.
Salisbury, D., Squires, N.K., 1993. Response properties of
long-latency event-related potentials evoked during NREM
sleep. J. Sleep Res. 2, 232–240.
Salisbury, D., Squires, N.K., Ibel, S., Maloney, T., 1992.
Auditory event-related potentials during stage 2 NREM
sleep in humans. J. Sleep Res. 1, 251–257.
Sallinen, M., Kaartinen, J., Lyytinen, H., 1996. Processing of
auditory stimuli during tonic and phasic periods of REM
sleep as revealed by event-related brain potentials. J. Sleep
Res. 5, 220–228.
Shefrin, S.L., Goodin, D.S., Aminoff, M.J., 1988. Visual
evoked potentials in the investigation of ‘blindsight’. Neu-
rology 38, 104–109.
Snyder, F., 1966. Toward an evolutionary theory of dreaming.
Am. J. Psychiat. 123, 121–136.
Spencer, K.M., Dien, J., Donchin, E., 1999. A componential
analysis of the ERP elicited by novel events using a dense
electrode array. Psychophysiology 36, 409–414.
Squires, N.K., Squires, K.C., Hillyard, S.A., 1975. Two variants
of long-latency positive waves evoked by unpredictable
auditory stimuli in man. Electroencephalo. Clin. Neurophy-
siol. 38, 387–401.
Sutton, S., Braren, M., Zubin, J., John, E.R., 1965. Evoked
potential correlates of stimulus uncertainty. Science 150,
1187–1188.
Van Sweden, B., Van Dijk, J.G., Caekebeke, J.F.V., 1994.
Auditory information processing in sleep: late cortical poten-
tials in an odd-ball paradigm. Neuropsychobiology 29,
152–156.
Vaughan, H.G., Ritter, W., 1970. The sources of auditory
evoked responses recorded from the human scalp. Electroen-
cephalo. Clin. Neurophysiol. 28, 360–367.
Verleger, R., 1988. Event-related potentials and cognition: a
critique of the context updating hypothesis and an alternative
interpretation of P3. Brain Behav. Res. 11, 343–427.
Verleger, R., 1997. On the utility of P3 latency as an index of
mental chronometry. Psychophysiology 34, 131–156.
Voss, U., Harsh, J., 1998. Information processing and coping
styles during the wakeysleep transition. J. Sleep Res. 7,
225–232.
Wesensten, N.J., Badia, P., 1988. The P300 component in
sleep. Physiol. Behav. 44, 215–220.
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. Electroencephalo. Clin.
Neurophysiol. 16, 269–279.
Winter, O., Kok, A., Kenemans, J.L., Elton, M., 1995. Auditory
event-related potentials to deviant stimuli during drowsiness
and stage 2 sleep. Electroencephalo. Clin. Neurophysiol. 96,
398–412.