When sleep is perceived as wakefulness: an experimental study on state perception during physiological sleep.

Abstract

While electrophysiologically measured sleep and perception of sleep generally concur, various studies have shown this is not always the case. The objective of the present study was to assess the perception of actual state during sleep by the technique of planned awakenings and interviewing subjects on the preawakening state. Sixty-eight (43 females, 25 males) young (mean age: 24.1, SD 5.1 years) normal sleeping subjects were deliberately awakened out of consolidated sleep, either stage 2 (S2), or REM sleep, during the first night in a non-clinical sleep laboratory. While the preawakening state was experienced as sleep in 48 cases (70.6%), it was experienced as wakefulness in 20 cases (29.4%). The percentage of awake judgements was somewhat, but not significantly, higher for awakenings out of S2 (38.2%), to REM sleep (20.6%). The proportion of mismatches between electrophysiologically defined sleep and state judgements was time-dependent with more awake judgements for REM sleep in the second half of the sleep period (41.7%) than in the first one (17.4%). Those subjects who made an awake judgement more frequently had a feeling of being aware of the situation and their surroundings than those who made a sleep judgement (80% versus 33%). Awareness during sleep may be a cognitive style, which favours mismatches between state perception and electrophysiologically defined sleep. Sleep periods with concordant or discordant state judgements did not differ in electrophysiologically defined sleep onset latency, sleep efficiency, or sleep state distribution.

Full text

Page 1

When sleep is perceived as wakefulness: an experimental study on
state perception during physiological sleep
DOREEN WEIGAND, LARS MICHAEL and HARTMUT SCHULZ
Department of Educational Science and Psychology, Free University Berlin, Berlin, Germany
Accepted in revised form 22 August 2007; received 8 December 2006
SUMMARY
While electrophysiologically measured sleep and perception of sleep generally concur,
various studies have shown this is not always the case. The objective of the present
study was to assess the perception of actual state during sleep by the technique of
planned awakenings and interviewing subjects on the preawakening state. Sixty-eight
(43 females, 25 males) young (mean age: 24.1, SD 5.1 years) normal sleeping subjects
were deliberately awakened out of consolidated sleep, either stage 2 (S2), or REM sleep,
during the first night in a non-clinical sleep laboratory. While the preawakening state
was experienced as sleep in 48 cases (70.6%), it was experienced as wakefulness in 20
cases (29.4%). The percentage of awake judgements was somewhat, but not
significantly, higher for awakenings out of S2 (38.2%), to REM sleep (20.6%). The
proportion of mismatches between electrophysiologically defined sleep and state
judgements was time-dependent with more awake judgements for REM sleep in the
second half of the sleep period (41.7%) than in the first one (17.4%). Those subjects
who made an awake judgement more frequently had a feeling of being aware of the
situation and their surroundings than those who made a sleep judgement (80% versus
33%). Awareness during sleep may be a cognitive style, which favours mismatches
between state perception and electrophysiologically defined sleep. Sleep periods with
concordant or discordant state judgements did not differ in electrophysiologically
defined sleep onset latency, sleep efficiency, or sleep state distribution.
keywords
awakening, awareness, cognitive activity, sleep perception
INTRODUCTION
For decades, the electroencephalogram (EEG) has been used
as a valid method to discriminate between the two behavioural
states of sleep and wakefulness. Based on the fact that
electrophysiological patterns correlate quite well with sub-
jective state perception (Lindsley, 1952), it has become
common to assume a kind of isomorphism between electro-
physiologically measured sleep and subjectively experienced
sleep. Any discrepancy between perceived and measured sleep
was declared to be sleep state misperception, implying the
inability of the subject to correctly perceive a given state. Sleep
state misperception even became a diagnostic category. It was
termed sleep state misperception in the first version of the
International Classification of Sleep Disorders (ICSD, 1990)
and paradoxical insomnia in the revised version (ICSD-2,
2005). As a consequence of the definition of sleep by
operationalized electrophysiological criteria, the role of cog-
nitive processes for state perception during sleep has been
widely neglected.
Although striking discrepancies between physiological or
pharmacologically induced sleep and state perception have
been observed occasionally (Fischgold and Schwartz, 1961, p.
214 and p. 366), systematic studies of this phenomenon are
quite rare. Various earlier studies (Alster and Lavie, 1989;
Anch et al., 1982; Campbell and Webb, 1981; Knab and Engel,
1988; Langford et al., 1972) compared objective and subjective
sleep data by assessing the ability of subjects to signal
spontaneous awakenings during night sleep by pressing a
button. In all of these studies, it was shown that a number of
electrophysiologically defined arousals was either not per-
ceived, or at least not signalled by the subjects. This was
Correspondence: Doreen Weigand, Georg-August-Universita ¨ t Go ¨ ttin-
gen, Georg-Elias-Mu ¨ ller-Institut fu ¨ r Psychologie, Gosslerstraße 14,
37073 Go ¨ ttingen, Germany. Tel.: +49-551-39-3665; Fax +49-551-39-
3544; e-mail: weigand@psych.uni-goettingen.de
J. Sleep Res. (2007) 16, 346–353
346
? 2007 European Sleep Research Society

Page 2

especially true for NREM sleep, while in REM sleep sponta-
neous arousals were more often perceived and signalled. From
these results, it was concluded that sleep perception is more
accurate in REM sleep than in other sleep stages. Studies
which used the technique of deliberate awakenings to measure
sleep perception (Amrhein and Schulz, 2000; Bonnet and
Moore, 1982; Foulkes, 1962; Mendelson, 1995a,b, 1998;
Mercer et al., 2002; Rotenberg, 1993; Sewitch, 1984a,b)
confirmed the dependency of state judgement from the actual
sleep state. In normal sleeping subjects the proportion of those
who judged the state before an arousing tone signal as awake
was in the range of 19% (Amrhein and Schulz, 2000) to 27%
(Sewitch, 1984b). This rate was higher for NREM sleep and
ranged between 40% (Mercer et al., 2002) and 55% (Sewitch,
1984b). Sewitch (1984b) described sleep ⁄wake perception as a
process of discrimination that involves cognitive interpreta-
tions of physiological and psychological data. According to
this assumption, uncertain decisions occur when inconsistent
aspects of experience have to be taken into consideration, or
when perceptions are missing that would lead to a certain
decision. Mercer et al. (2002) extended this approach by
comparing sleep ⁄wake perception of normal sleepers and
patients with insomnia. The data were analysed by a statistical
technique based on signal detection theory. Awake judgements
after having been awoken from sleep were consistently higher
for insomniac patients than for normal controls, and they
differed between sleep states. For control subjects, deliberate
awakenings resulted in 39% wake judgements in S2 sleep and
19% in REM sleep. The corresponding values were clearly
higher for insomniac patients with 80% for S2 and 62% for
REM sleep.
Beside the sleep state, other internal events and cognitive
experiences, which may influence the state judgement, were
explored. Gibson et al. (1982) observed that most subjects used
a few criteria to discriminate between states. The perceived
depth of sleep was the most frequently used criterion, although
it did not differ between correct and incorrect sleep ⁄wake
judgements. Three cognitive criteria were associated with
correct sleep ⁄wake estimations, namely ?control of thoughts?,
the ?awareness of surroundings?, and a ?time element? (Gibson
et al., 1982). The ongoing mental activity before planned
awakenings was further evaluated in a study by Foulkes (1962)
who showed that thoughts are more image-like just before
awakenings from REM sleep compared with NREM sleep.
Other studies (Foulkes and Vogel, 1965; Vogel et al., 1966)
confirmed that subjects experienced that they were able to
control mental activity during NREM sleep, whereas mental
activity typically ?just happened? during REM sleep. Addition-
ally, the extent of contact to the external situation differed
across sleep stages, with a higher degree of orientation in space
and time during NREM than during REM sleep.
The aim of the present study was to further evaluate those
factors upon which the state judgement after a deliberate
awakening from sleep is dependent. In contrast to earlier
studies, the number of planned awakenings was low. While the
first 27 subjects were awakened twice, the next 41 subjects were
awakened only once. All subjects were awakened in their first
laboratory night. The reason for reducing the number of
awakenings from two to one was that, in some cases, it was
difficult to identify a second segment of uninterrupted sleep, as
defined below, to waken subjects a second time and interview
them. The data from the two parts of the study were combined
to create a large enough sample for statistical analysis. To
avoid interaction effects, in cases with two awakenings, only
the first one was included in the analysis.
METHODS
Subjects
The sample consisted of 68 young persons (43 females and 25
males; mean age 24.1 years, standard deviation 5.1 years)
without sleep complaints, most of whom were psychology
students at the Free University of Berlin, recruited by
advertisement. These volunteer subjects did not receive pay-
ment but were informed about their sleep patterns in the
laboratory and were given a printout of their polysomno-
graphic sleep profile as a bonus for participation. Before study
participation, all volunteers were interviewed about their sleep
habits. Only those subjects who evaluated their sleep subjec-
tively as ?good? or ?very good? and who did not report any
difficulties of falling or staying asleep were included. In
addition, subjects were informed about the time schedule and
technical routine of the laboratory to adapt their sleep rhythm
as necessary. Subjects were also asked to avoid stimulants like
coffee or tea in the late afternoon and evening hours before the
sleep study. During polysomnographic sleep recording on the
study night, none of the subjects showed indications of a sleep
disorder. Thus, all subjects who participated in the study were
able to be included in the data analysis.
Procedures
Subjects arrived at 21:30 hours in the sleep laboratory, where
they were instructed about the procedure, in particular that
they would be awakened and interviewed once (or twice)
during the night. They were made familiar with the 10-item
questionnaire of the nocturnal interview (Table 1). Subjects
were informed that the tone signal, indicating the beginning of
the interview, could occur at any time, independent of whether
the subject was asleep or awake. After placing the electrodes,
subjects went to bed at about 23:00 hours. Before turning off
the light they read the questions of the interview again. Forty-
one subjects were awakened once and 27 subjects were
awakened twice during the night by a standardized audio
signal with an acoustic pressure of 70 dB (at a distance of 1 m
from the head) and a frequency of 400 Hz. Immediately, after
the awakening signal the experimenter entered from the
adjacent recording room and completed the interview, while
the subject stayed lying in bed. The room remained darkened,
apart for a little light shining through the connecting door.
Afterwards subjects could sleep until the next morning (or to
When sleep is perceived as wakefulness 347
? 2007 European Sleep Research Society, J. Sleep Res., 16, 346–353

Page 3

the second awakening). Subjects were awakened and inter-
viewed during their stay on the first night in sleep laboratory.
Potential sleep disturbances by the first night effect (Agnew
et al., 1966) were tolerated in favour of logistic advantages of
single night recordings in a student-run laboratory.
Polysomnography
The polysomnographic data were recorded with a commercial
digital measurement system (Sagura-2000 for Windows). The
registration with standard filter settings included four EEGs
(F3-A2, C3-A2, C4-A1, O2-A1), two horizontal EOGs, chin
EMG, and ECG. Respiration was controlled with a chest belt,
and oxygen saturation of the blood and pulse were measured
with an oximeter, which was placed on a finger of the non-
dominant hand. Sleep scoring was performed according to the
rules of Rechtschaffen and Kales (1968).
Awakening conditions
Deliberate awakenings were performed in stable S2, and in
REM sleep. To guarantee consolidated preawakening sleep
phases, awakenings had to be preceded either by a minimum of
7.5 min of uninterrupted REM sleep or 15 min of un-
interrupted S2 sleep. The respective number of 30 s epochs
before awakening had to be continuously assigned to the same
sleep stage, either S2 or REM, and had to be free of arousals,
which were detected and rated visually online. Planned
awakenings did not take place before the end of the first
NREM–REM sleep cycle. The sequence of awakening from S2
or REM sleep was randomized within (in case of two
awakenings) and between subjects. All awakenings and inter-
views were performed by DW and LM.
Nocturnal interview
The subjects had been familiarized with the 10 questions (Q1–
Q10) of the interview (Table 1) before the start of the sleep
recording in the evening. Any ambiguities were clarified with
the experimenter, if necessary. The questions were adapted
from an earlier study in our laboratory (Amrhein and Schulz,
2000), with only slight changes of the wording. The perception
of the preawakening state was assessed using the first three
questions. Q1 requested a state judgement (sleep or awake), Q2
asked about the certainty of the state judgement (absolutely
sure – fairly sure – not sure), and Q3 about the quality of the
preawakening state (deep sleep – light sleep – transitional state
– awake but sleepy – wide awake). Q4 demanded an estimate
of the duration of prior wakefulness if the actual state was
rated as awake, and Q5–Q9 asked about different cognitive
aspects in relation to the state judgement. Q5, an open
question, asked about those internal perceptions which led to
the sleep ⁄awake judgement. Q6 asked about three aspects of
mental activity, namely distinctness, image or thought-like-
ness, and controllability, while Q7–Q9 assessed formal aspects
of the transition from the pre- to the postawakening state. Q7
asked whether a state transition was perceived or not, Q8
about the speed of awakening, in case of a sleep judgement,
and Q9 asked about any awareness of the surrounding, or
orientation before the awakening. The last question (Q10)
asked for a time estimate since lights out (or since the last
awakening, in case of two awakenings).
Statistics
Statistical analysis was performed with the chi-square test for
nominal data (sleep ⁄awake judgement, items referring to
mental content, perceived transition and awareness of the
surroundings), and the t-test for independent samples for
metric data (sleep latencies, duration of awakening, time
estimates of the subjects). Two aspects of the temporal position
of the deliberate awakenings were examined, (a) its position
within a given NREM–REM sleep cycle, and (b) its position in
the whole sleep phase.
Position within the NREM–REM cycle
For awakenings out of REM sleep, the number of REM sleep
epochs before the awakening was counted and compared with
the duration of a neighbouring, undisturbed REM sleep phase.
This undisturbed REM sleep phase could be located either
Table 1 Questions of the interview
1. Were you awake or did you sleep before you heard
the tone signal?
2. How sure are you about your decision?
a) Absolutely sure
b) Fairly sure
c) Not sure
3a. If ?sleep? was answered to question 1:
Before you heard the tone signal, were you...
a) ...in deep sleep,
b) ...in light sleep, or
c) ...in a transient state between awake and sleep?
3b. If ?awake? was answered to question 1:
Before you heard the tone signal, were you...
a) ...wide awake,
b) ...awake but sleepy, or
c) ...in a transient state between awake and sleep?
4. If ?awake? was answered to question 1:
How long had you been awake before you heard the signal?
5. On which impressions is your judgement of having been
awake or asleep based?
6. Was there something on your mind before being awakened?
If yes:
a) Was it rather clear and distinct or rather vague and blurred?
b) Was it rather image-like or rather thought-like?
c) Could you control it yourself or did it just happen?
7. Did you notice a transition from the sleep to wakefulness?
If yes:
Can you describe it further?
8. How quickly did you become awake?
9. Before you heard the signal were you aware of where you
were and where the things around you were located?
10. Approximately how much time passed since the light was turned
off (or: since the last awakening) until now?
348D. Weigand et al.
? 2007 European Sleep Research Society, J. Sleep Res., 16, 346–353

Page 4

before or after the cycle with the experimental awakening,
depending on whether the deliberate awakening occurred early
or late in sleep. In the same way, the duration of NREM sleep
before awakenings out of stage 2 were compared with the
duration of uninterrupted NREM sleep phases.
Position within the whole sleep phase
The potential influence of the duration of prior sleep on the
state judgement was also tested. To this end, the distribution of
intervals between lights off and the start of the interview was
established separately for S2 and REM sleep awakenings, and
the median was determined. Interviews before the median were
designated ?early?, and those after the median ?late?.
RESULTS
State judgements
Sixty-eight young subjects (43 women and 25 men) with a
mean age of 24.1 years (SD: 5.1 years) participated in the
study. Half of them (n = 34) were awakened from sleep
stage 2, the other half from REM sleep. While for 48
subjects (70.6%) the estimated state before deliberate awak-
ening was concordant with electrophysiologically measured
sleep, 20 subjects (29.4%) gave a discrepant state rating.
They assumed that they were awake before the arousing
signal occurred. Ninety percent of the subjects with a
mismatch between perceived and measured preawakening
state were either fairly sure (n = 12) or absolutely sure
(n = 6) that their state estimate was correct, while only two
were unsure. In contrast to this, subjects with a concordant
state judgement had fewer ?sure? ratings (64.6% sure versus
35.4% unsure). About half of the subjects with a concordant
state judgement had perceived the preawakening state as
deep sleep (47.9%), while 29.2% rated the state as light
sleep, and 22.9% as a transitional state between sleep and
wakefulness. Subjects with discordant state judgement rated
the preawakening state either as a transitional state (45%),
as awake but sleepy (45%) or, in two cases (10%), as wide
awake (Table 2).
Although the proportion of awake judgements was higher
for awakenings out of S2 (13 ⁄34 = 38.2%) than for REM
sleep (7 ⁄34 = 20.6%), the difference was not significant. The
two sleep states also did not differ significantly in the level of
certainty of subjective judgements (Table 2).
Table 2Characteristics of state judgements as either awake or asleep*
Questions Response categories
S2 sleepREM sleep
Awake Asleep AwakeAsleep
Q 1: Were you awake or did you sleep before you heard the tone signal?
Q2: How sure are you about your decision?
13
4
8
1
21
4
8
9
7
2
4
27
9
10
8
Absolutely sure
Fairly sure
Not sure1
Q3: Before you heard the tone signal, were you ...?
(Depending on the response to Q1, Q 3 was worded differently; see Table 2)
Deep sleep
Light sleep
Transient state
Awake but sleepy
Wide awake
–
–
7
4
2
8–
–
2
5
–
15
4
8
–
–
10
3
–
–
Q6: Was there something on your mind before being awakened? Yes
No
9
4
13
8
5
2
24
3
If yes, Q6a: Was it rather clear and distinct or rather vague and blurred? Clear
Vague
No response
2
6
1
6
4
3
3
2
–
19
5
–
If yes, Q6b: Was it rather image-like or rather thought-like? Image-like
Thought-like
No response
5
3
1
3
8
2
4
1
–
20
4
–
If yes, Q6c: Could you control it yourself or did it just happen? Control
No control
No response
0
7
2
2
9
2
3
2
–
6
18
–
If asleep, Q7: Did you notice a transition from sleep to wakefulness?Yes
No
No response
10
10
1
4
9
–
2
5
–
13
13
1
Q9: Before you heard the signal were you aware of where
you were and where the things around you were located?
Yes
No
10
3
86
1
8
1319
*There were not enough usable answers to question 4 to allow statistical analysis.
When sleep is perceived as wakefulness349
? 2007 European Sleep Research Society, J. Sleep Res., 16, 346–353

Page 5

Characteristics of state judgements
The additional questions of the interview gave some insight
into the processes, which may have led subjects to rate the state
immediately prior to hearing the tone signal either as asleep or
awake. Two-thirds of the subjects reported that there was
something on their mind when they were cued, with a slightly
higher proportion in REM than in S2 sleep (85.3% versus
64.7% respectively; v2= 3.8, P < 0.05). The rate of mental
activity was similar for asleep and awake state judgements
(77.0% versus 70.0% respectively).
In those subjects, who had perceived having had any
thoughts, the quality of thought differed between conditions.
The mental content was more clear, when aroused out of REM
sleep, compared with S2 sleep (75.9% versus 44.4% respec-
tively; v2= 4.7, P < 0.05), more image-like (REM: 82.8%
versus S2: 42.1%; v2= 8.5, P < 0.05), but with a similar low
level of controllability in both sleep states (REM: 31.0%
versus S2: 11.1%). Of these three mental content variables,
only clarity co-varied significantly with the state judgement.
Clarity of mental content was more frequently reported when
the state was perceived as asleep (73.5%) compared with
awake (38.5%; v2= 5.0, P < 0.05), while there was no
difference for the features image-like (65.7% for asleep versus
69.2% for awake), and controllability (22.9% for asleep versus
25.0% for awake).
Awareness of the actual situation, when the tone was
sounded (Question 9) differed across judgements with a
significantly higher rate of awareness when the state was rated
awakethanasleep(80.0%
v2= 12.3, P < 0.05). Awareness ratings, on the other hand,
did not differ across sleep states S2 and REM (52.9% versus
41.2% respectively).
versus33.3%respectively;
The temporal position of the interviews
Position within the NREM–REM cycle
The mean duration (±SD) of REM sleep prior to an
awakening was 11.8 min (±6.9) and 21.9 min (±10.7) for
uninterrupted REM sleep. The respective mean durations for
NREM sleep were 25.6 min (±16.9) and 66.2 min (±20.8).
The comparison shows that planned awakenings out of REM
sleep started about half way through a REM sleep phase, while
planned awakenings out of S2 sleep took place mostly before
the middle of a NREM sleep phase.
Position within the whole sleep phase
To control for the time of night effect, interviews were
separated into ?early? and ?late? ones, according to their
position in the sleep phase. Early and late were defined in
relation to the median of the time points of all interviews as
described above. The medians were computed separately for
S2 and REM sleep and were very similar with 239 min (range:
86–430) for S2 and 241 min (range: 136–380) for REM sleep.
For S2 sleep, the state judgements did not differ significantly
between early and late awakenings. Awake judgements were
35.3% for early and 41.1% for late interviews. In contrast, for
REM sleep, the proportion of awake judgments was low
(5.9%) for early interviews and significantly higher (35.3%) for
late interviews (v2= 4.50, P < 0.05; Table 3).
Polysomnographic sleep measures
Sleep quality for sleep and wake judgements after deliberate
awakening was also analysed. There were no significant
differences in sleep latency, sleep efficiency, or in the amount
of the different sleep stages (Table 4). Finally, subjects with
concordant or discordant state judgements also did not differ
in sex or age.
DISCUSSION
The results suggest that an appreciable amount of time asleep
is not perceived as sleep but as wakefulness, even by those
without a sleep complaint. The absolute number of mis-
matches between electrophysiologically measured sleep and
state judgements may be overestimated in our study, as
deliberate awakenings were induced in a first laboratory night
Table 3 Distribution of sleep and wake judgements in relation to the
temporal position of the interview in the sleep phase
The interview
started from:
Early* Late*
v2-test
(P-value)
AwakeAsleepAwakeAsleep
S2 sleep
REM sleep
6
1
11
16
7
6
10
11
0.12 (NS)
4.50 (<0.05)
*?Early? and ?late? were defined by the temporal position of the
interview relative to the median time point of all interviews. The
medians were computed separately for S2 and REM sleep. They were
239 min for S2 and 241 min for REM sleep.
Table 4 Polysomnographic sleep parameters (mean ± SD)
State judgement
AwakeSleep
No. subjects*
Sleep latency (min)?
Sleep efficiency (%)?
Wake (%)
NREM stage 1 (%)
NREM stage 2 (%)
NREM stage 3 (%)
NREM stage 4 (%)
REM sleep (%)
18
24.0 ± 20.2
88.8 ± 9.1
14.7 ± 9.1
4.9 ± 2.6
46.6 ± 4.8
8.5 ± 5.0
9.2 ± 6.2
16.7 ± 6.7
42
19.6 ± 13.1
89.1 ± 8.8
14.4 ± 8.5
6.1 ± 3.8
46.3 ± 7.2
8.6 ± 2.8
8.1 ± 5.6
15.8 ± 4.1
*Caused by a technical defect 16 sets of digitally stored polysomno-
graphic data were lost.
?Sleep latency was defined as the time interval between lights off and
the first epoch of S2 sleep.
?Sleep efficiency was defined as the ratio of total sleep time (TST),
divided by total time in bed (TBT) ·100. Wake time, which was caused
by the interview, plus the time to resume sleep after the interview, was
eliminated.
350D. Weigand et al.
? 2007 European Sleep Research Society, J. Sleep Res., 16, 346–353

Page 6

without prior adaptation. Thus, a first night effect (FNE) with
increased arousal has to be taken into account (Agnew et al.,
1966; Lorenzo and Barbanoj, 2002). In spite of this study
design limitation, the results concur with those of other studies
which had used deliberate awakenings to gain information on
state perception in sleep (Amrhein and Schulz, 2000; Mendel-
son, 1995b; Mercer et al., 2002; Sewitch, 1984a,b). All studies,
including the present one, come to the conclusion that state
perception deviates in a substantial number of cases from
measured sleep, as defined by electrophysiological criteria.
While this observation is not new (Borkovec et al., 1981,
Coates et al., 1987; Frankel et al., 1976), it has been largely
neglected. As a consequence, cognitive processes during sleep
and their contribution to state perception are poorly under-
stood.
Although the percentage of awake judgements did not differ
significantly between S2 and REM sleep, the tendency of more
awake judgements after awakenings out of S2 sleep was in line
with the results from earlier studies (Mendelson, 1995b; Mercer
et al., 2002; Sewitch, 1984a,b). The more pronounced differ-
ence in earlier studies may be due to a higher rate of planned
awakenings in these studies. A greater number of awakenings
per night may influence the perception of the current sleep state
in two ways. First, multiple awakenings result in a fragmen-
tation of sleep, which may increase the general level of arousal
and,thus, causeincreased frequency of awake judgements, even
if electrophysiological recordings indicate sleep. Second, multi-
ple awakenings offer a chance to compare among the different
preawakening states and therefore may contribute to discrim-
ination learning. This may strengthen any experiential differ-
ences in the perception of S2 and REM sleep.
Major correlates for the mismatches between state judge-
ment and measured sleep were (i) the degree of awareness of
the sleeper for the surrounding, and (ii) the distance between
sleep onset and the planned awakening.
Awareness of the sleeper for the situation and the surrounding
Those subjects who felt themselves oriented in their surround-
ings before they were awakened by the acoustic signal had a
high probability of perceiving the preawakening state as
wakefulness. If the loss of consciousness is a pivotal criterion
for sleep (De Manace ´ ı¨ne, 1896; Massimini et al., 2005), it is
conceivable that the opposite, namely the inner experience of
being aware of one?s situation and surroundings favours the
perception that the actual state is wakefulness and not sleep.
Moreover, in contrast to an outside observer who would use
behavioural criteria to make a sleep ⁄wake judgement, the
sleeping brain does not have access to such information, and
therefore behavioural criteria cannot be used by the sleeper as
a basis for critical control and contingent correction of the
subjective state perception. But, why is there a dissociation
between state judgement and electroencephalographic (EEG)
criteria of sleep?
The present data show that the same EEG state, which
was perceived as sleep in 70–80% of subjects studied, was
perceived 20–30% as wakefulness in interviews. The rela-
tionship between EEG patterns and state perception seems
to be asymmetric. While it would be highly improbable to
judge a given state as sleep when the EEG indicates
wakefulness (cf. Mercer et al., 2002, their Tab. 2), a
dissociation between EEG-defined states and state judgement
is not rare in sleep, as the data show. One main reason for
such a discrepancy may be that state perception is more
closely related to mental activity than to the EEG. The
answers to question 6 of the interview show that mental
activity is abundant during sleep. Preawakening mental
activitywas reportedby 65%
awakened out of S2 sleep, and even higher, namely 85%,
after REM sleep awakenings.
Beside the quantity, also the quality of mental activity may
contribute to state judgement. Mental activity in REM sleep
was rated as more image-like and clearer than in S2 sleep.
Mental activity in REM sleep, which is generally more dream-
like (Hobson et al., 2000), contrasts sharply to cognitive
processes when awake and, thus, induces a sleep judgement.
The case is different for cognitive activity in S2 sleep, which
was rated more frequently as thought-like, and thus is closer to
wakeful cognitive activity. The similarity in structure and
content of mental activity between wakefulness and some
states of sleep, in combination with the loss of behavioural
control, may produce the perception of being awake, even
when the EEG indicates sleep. It would be of interest whether
high-density EEG mapping or brain imaging techniques would
show any difference between sleep states which are concordant
or discordant with state judgement. Balkin et al. (2002) have
studied local brain blood flow changes during the process of
reorientation after awakening. Whether such changes would
differ systematically also in respect to subjective state judge-
ment is still an open question.
of subjects whowere
The time of the planned awakening
A descriptive analysis of the positions of the deliberate
awakenings out of REM and S2 sleep in a given NREM–
REM cycle showed that they most often occurred in the
middle of the respective sleep phase, or earlier. Thus, state
ratings during the interviews were probably not much influ-
enced by approaching sleep phase transitions.
While time of night did not have much effect on state
judgement when subjects were aroused from S2 sleep, there
was a significant effect for REM sleep interviews. The
proportion of awake judgements in this sleep state was low,
when subjects were awakened early in the sleep phase, but high
when this occurred late in the sleep phase. The observed
interaction between REM sleep and time of night replicates the
same finding by Mendelson (1998). In addition, Mendelson
(1998) found a significant interaction for S2 sleep with a high
rate of awake judgements early in sleep, and a low rate late in
sleep. That this interaction was not observed for S2 in this
study is probably due to the different timing of the awaken-
ings. While the early awakenings in Mendelson?s study were
When sleep is perceived as wakefulness 351
? 2007 European Sleep Research Society, J. Sleep Res., 16, 346–353

Page 7

placed in the first NREM sleep phase (?10 min after first
spindle?), S2 awakenings in the present study spared the first
NREM sleep phase. This assumption is also compatible with a
study by Rotenberg (1993) who found a decrease in awake
judgements from the first to the second NREM–REM sleep
cycle.
Semantics of sleep–wake discrimination
From the nocturnal interviews it appeared that subjects
frequently had difficulties describing the preawakening state,
especially when awakened from S2 sleep. The internal expe-
rience of this state appears to be vague and diffuse in many
cases. In the present study, it was also observed that subjects
rethought their answers and requested that they be allowed to
revise an answer to a given question more frequently after
awakenings from S2 than from REM sleep. This suggests that
giving a state description, or making a discrimination, may be
more difficult than the semantic categories of the interview
would suggest.
One way to overcome this problem would be to use a fuzzy
response format instead of a point estimate (Gehrman et al.,
2002) to document cognitive activity and state perception.
From the present data set, it became clear that in an
appreciable number of awakenings out of electrophysiologi-
cally defined sleep, the subjects experienced the state as
transient, something between sleep and awake (cf. Tab. 2).
Thus, different response formats for sleep questionnaires or
interviews should be applied in future studies.
The technique of deliberate awakenings
The technique of deliberate awakenings has a few shortcom-
ings, which may have influenced the results. First, if the
subjects are informed about the procedure before going to bed,
expectancy of the awakening(s) may increase arousal during
sleep. Second, memory for sleep-related cognitive activity is
notoriously poor. This may be related to a general difficulty
transferring material from one mental state into another one,
as in Koukkou and Lehmann?s (1983) suggestion that forget-
ting dreams is a function of the magnitude of the difference
between states during storage and recall. Third, sleep inertia
after a deliberate awakening (A˚kerstedt et al., 2002) may
likewise contribute to the difficulty in giving a coherent report
of the preawakening cognitive activity. Finally, one may
assume that the different methodological peculiarities are not
fixed but vary, depending on the state or stage where the
awakening occurs, and the duration of prior sleep. This makes
systematic control of the methodologically induced error
variance even more difficult.
In summary, there appears to be an asymmetrical relation-
ship between objectively measured sleep ⁄wake states, on the
one hand, and state judgement on the other. While the
correlation between electrophysiologically defined wakefulness
and state perception is perfect in normal subjects, the
correlation is less perfect for the sleep state. Depending on
cognitive activity, the actual sleep stage, and the duration of
prior sleep, portions of electrophysiologically defined sleep are
perceived as wakefulness. Even if this discrepancy can be
reduced by positive feedback (Downey et al., 1989; Sewitch,
1984b), its existence should remind us that electrophysiological
criteria to define sleep are valid descriptors of the physiological
state, but do not consider cognitive activity, which largely
contributes to state perception. A final point to discuss is the
proper design for studying state perception in sleep.
Between versus within subjects designs
If there are stable individual differences in state perception, a
within subjects design would be probably more appropriate
than a between subjects design, as we have applied in the
present study. A within subjects study could be performed
either with multiple awakening within a single night, or over
a series of nights with a single or only few awakenings per
night. Another alternative would be to determine one or
more factors, which may be correlated with state perception,
and to compare groups of subjects who score either high or
low on those factors. Our group is just developing a
questionnaire to assess one such candidate factor, namely
orientation or awareness during sleep (Rothkirch et al.,
2006).
ACKNOWLEDGEMENT
We thank Catherine N. Oliver for linguistic support.
REFERENCES
Agnew, H. W., Jr, Webb, W. B. and Williams, R. L. The first night
effect: an EEG study of sleep. Psychophysiology, 1966, 2: 263–266.
A˚kerstedt, T., Billiard, M., Bonnet, M., Ficca, G., Garma, L.,
Mariotti, M., Salzarulo, P. and Schulz, H. Awakening from sleep.
Sleep Med. Rev., 2002, 4: 267–286.
Alster, J. and Lavie, P.. Spontaneous awakenings within sleep in
normal and fragmented sleep. In J. Horne (Ed.) Sleep ?88.
Proceedings of the Ninth European Congress on Sleep Research,
Jerusalem, September 1988. Gustav Fischer Verlag, Stuttgart, New
York, 1989, 193.
Amrhein, C. and Schulz, H. Selbstberichte nach dem Wecken aus dem
Schlaf – ein Beitrag zur Wahrnehmung des Schlafes. Somnologie,
2000, 4: 61–67.
Anch, A. M., Salamy, J. G., McCoy, G. F. and Somerset, J. S.
Behaviorally signalled awakenings in relationship to duration of
alpha activity. Psychophysiology, 1982, 19: 528–530.
Balkin, T. J., Braun, A. R., Wesensten, N. J., Jeffries, K. and Varga,
M. The process of awakening: a PET study of regional brain activity
patterns mediating the re-establishment of alertness and conscious-
ness. Brain, 2002, 125: 2308–2319.
Bonnet, M. H. and Moore, S. E. The threshold of sleep: perception of
sleep as a function of sleep time and auditory threshold. Sleep, 1982,
5: 267–276.
Borkovec, T. D., Lane, T. W. and Van Oot, P. H. Phenomenology of
sleep among insomniacs and good sleepers: wakefulness experience
when cortically asleep. J. Abnorm. Psychol., 1981, 90: 607–609.
Campbell, S. S. and Webb, W. B. The perception of wakefulness
within sleep. Sleep, 1981, 4: 177–183.
352D. Weigand et al.
? 2007 European Sleep Research Society, J. Sleep Res., 16, 346–353

Page 8

Coates, T. J., Killen, J. D., Silverman, S., George, J., Marchini, E.,
Hamilton, S. and Thoresen, C. E. Cognitive activity, sleep distur-
bance, and stage specific differences between recorded and reported
sleep. Psychophysiology, 1987, 20: 243–250.
De Manace ´ ı¨ne, M. Le Sommeil. Masson, Paris, 1896.
Downey, R., Bonnet, M. H., DiMatteo, M. R., Tochtrop, M. and
Dexter, J. R. Sleep–wake discrimination 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, 1989. Sleep Res., 1989, 18: 225.
Fischgold, H. and Schwartz, B. A. A clinical, electroencephalographic
and polygraphic study of sleep in the human adult. In: G. E. W.
Wolstenholme and M. O?Connor (Eds) The Nature of Sleep.
Churchill, London, 1961: 209–231.
Foulkes, W. D. Dream reports from different stages of sleep. J.
Abnorm. Soc. Psychol., 1962, 65: 14–25.
Foulkes, D. and Vogel, G. Mental activity at sleep onset. J. Abnorm.
Psychol., 1965, 70: 231–243.
Frankel, B. L., Coursey, R. D., Buchbinder, R. and Snyder, F.
Recorded and reported sleep in chronic primary insomnia. Arch.
Gen. Psychiat., 1976, 33: 615–623.
Gehrman, P., Matt, G. E., Turingan, M., Dinh, Q. and Ancoli-Israel,
S. Towards an understanding of self-reports of sleep. J. Sleep Res.,
2002, 11: 229–236.
Gibson, E., Perry, F., Redington, D. and Kamiya, J. Discrimination of
sleep onset stages: behavioral responses and verbal reports. Percept.
Mot. Skills, 1982, 55: 1023–1037.
Hobson, J. A., Pace-Schott, E. F. and Stickgold, R. Dreaming and the
brain: Toward a cognitive neuroscience of conscious states. Behav.
Brain Sci., 2000, 23: 793–842.
ICSD. International Classification of Sleep Disorders: Diagnostic and
Coding Manual. Diagnostic Classification Steering Committee,
Chairman M. J. Thorpy. American Sleep Disorders Association,
Rochester, Minnesota, 1990.
ICSD-2. International Classification of Sleep Disorders, 2nd ed.
Diagnostic and Coding Manual. American Academy of Sleep
Medicine, Westchester, Illinois, 2005.
Knab, B. and Engel, R. R. Perception of waking and sleeping: possible
implications for the evaluation of insomnia. Sleep, 1988, 11: 265–
272.
Koukkou, M. and Lehmann, D. Dreaming: the functional state-shift
hypothesis. A neuropsychophysiological model. Br. J. Psychiat.,
1983, 142: 221–231.
Langford, G. W., Meddis, R. and Pearson, A. J. Spontaneous
arousals from sleep in human subjects. Psychonom. Sci., 1972, 28:
228–230.
Lindsley, D. B. Psychological phenomena and the electroencephalo-
gram. EEG Clin. Neurophysiol., 1952, 4: 443–456.
Lorenzo, J. L. and Barbanoj, M. J. Variability of sleep parameters
across multiple laboratory sessions in healthy young subjects: the
?very first night effect?. Psychophysiology, 2002, 39: 409–413.
Massimini, M., Ferrarelli, F., Huber, R., Esser, S. K., Singh, H. and
Tononi, G. Breakdown of cortical effective connectivity during
sleep. Science, 2005, 309: 2228–2232.
Mendelson, W. B. Pharmacological alterations of the perception of
being awake or asleep. Sleep, 1993, 16: 641–646.
Mendelson, W. B. Effects of Flurazepam and Zolpidem on the
perception of sleep in insomniacs. Sleep, 1995a, 18: 92–96.
Mendelson, W. B. Effects of Flurazepam and Zolpidem on the
perception of sleep in normal volunteers. Sleep, 1995b, 18: 88–91.
Mendelson, W. B. Effects of time of night and sleep stage on
perception of sleep in subjects with sleep state misperception.
Psychobiology, 1998, 26: 73–78.
Mercer, J. D., Bootzin, R. R. and Lack, L. C. Insomniacs? perception
of wake instead of sleep. Sleep, 2002, 25: 564–571.
Rechtschaffen, A. and Kales, A. (Eds) A Manual of Standardized
Terminology, Techniques and Scoring System for Sleep Stages of
Human Subjects. BIS ⁄BRI, UCLA, Los Angeles, 1968.
Rotenberg, V. S. The estimation of sleep quality in different stages and
cycles of sleep. J. Sleep Res., 1993, 2: 17–20.
Rothkirch, M., Weigand, D. and Schulz, H.. Entwicklung eines
Fragebogens zur Messung der Orientiertheit im Schlaf. Somnologie,
2006, 10: 43 (abstract).
Sewitch, D. E. NREM sleep continuity and the sense of having slept in
normal sleepers. Sleep, 1984a, 7: 147–154.
Sewitch, D. E. The perceptual uncertainty of having slept: the inability
to discriminate electroencephalographic sleep from wakefulness.
Psychophysiology, 1984b, 21: 243–259.
Vogel, G., Foulkes, D. and Trosman, H. Ego functions and dreaming
during sleep onset. Arch. Gen. Psychiat., 1966, 14: 238–248.
When sleep is perceived as wakefulness353
? 2007 European Sleep Research Society, J. Sleep Res., 16, 346–353