Sleep, 4(2): 177-183
© 1981 Raven Press, New York
The Perception of Wakefulness
Within Sleep
Scott S. Campbell and Wilse B. Webb
Department of Psychology, University of Florida, Gainesville, Florida
Summary: We studied the relationship between electroencephalographic
(EEG)-defined awakenings and subjects' awareness of them at the time of their
occurrence in 40 men (mean age, 55.3 years), The subjects used a push-button
device to signal wakefulness during an evening's sleep in the laboratory; continuous EEG and electro-oculograms were also made. Results were as follows
(after eliminating data for 12 nonresponders): 52% of the responses occurred
during EEG-defined wakefulness (stage 0), 12% in a sleep stage that was followed immediately by stage 0, and 36% in a sleep stage not followed by stage O.
However, analysis of the last finding indicated that 73% of these responses
were associated with at least 4-6 sec of preceding alpha activity. Thus, signals
in the absence of EEG criteria were actually less than 10% of the total recorded. Signals of wakefulness were closely associated with the onset of EEGdefined wakefulness (and 84% of the signals occurred within 4 sec to 1 min of
the EEG period), We conclude that self-signaled arousals deserve consideration in clinical analyses in which reports of arousals are a part of the reported
sleep disturbance. Key Words: Sleep-Perceived wakefulness-Electroencephalogram.
Wakefulness, in the sleep literature, is defined electroencephalographic ally as
"alpha activity and/or low voltage, mixed frequency EEG ... usually, but not
necessarily, accompanied by a relatively high tonic EMG, and often REMs and
eye blinks" (Rechtschaffen and Kales, 1968). There has been little effort to relate
this EEG definition to the subjective state of wakefulness. This is puzzling in light
of ample clinical and common laboratory knowledge that there is only a loose
linkage between recall of awakenings and the presence of EEG-defined awakenings.
The studies previously concerned with EEG measures and wakefulness are few
and suggest limitations ofthe EEG as a measure of wakefulness. This is evident in
several studies which have examined the relationships of EEG measures and the
Accepted for publication February 1980.
Address correspondence to S. S. Campbell at Box 52, Department of Psychology, University of
Florida, Gainesville, Florida 32611.
Address reprint requests to W. B. Webb at Department of Psychology, University of Florida,
Gainesville, Florida 32611.
177
178
S. S. CAMPBELL AND W. B. WEBB
transition from wakefulness to sleep at sleep onset (Loomis et at., 1937; Kamiya,
1961; Foulkes and Vogel, 1965; Walker in Agnew and Webb, 1972a). It may be
generally concluded from these studies that, at least at sleep onset, the relationship between EEG-defined sleep and the subjective presence of sleep is neither
simple nor unequivocal. Foulkes and Vogel, for example, intensively examined
"mental activity at sleep onset." Subjects were asked, after arousal from various
stages at sleep onset, what they were experiencing "just before I called you."
Subsequently, subjects were asked to rate their "state of consciousness" on a 5point scale (0-4): awake and alert, awake but drowsy, drifting off to sleep, in
light sleep, in deep sleep. The median ofthe subject means for arousals from stage
1 was 2.00 (range, 1.25-2.67) and from stage 2 was 2.25 (1.67-2.67). It is to be
noted that these means barely reach or exceed "drifting off to sleep" and the
ranges include" awake but drowsy."
Analogous results were reported by Walker (Agnew and Webb, 1972a). Subjects were placed in bed in dark, sound-attenuated rooms for 45 min while their
EEGs were recorded. At the end of the period they were asked ifthey had gone to
sleep. Forty-four percent who reached EEG levels of stage lor 2 said "no"; 2 of
13 subjects reaching stage 3-4 replied "no."
We have been studying the sleep of older subjects. Two facts are apparent.
EEG-defined awakenings (stage 0) occur frequently and for extended periods;
there may be large discrepancies between these objectively defined events and the
sUbjective recall of such events on morning arousal. We could find no studies
specifically concerned with EEG-defined awakenings and subjects' awareness of
them at the time of their occurrence. The present study was undertaken to
explore this relationship. Simply, subjects were instructed to signal periods of
wakefulness by the behavioral response of pressing a button, while being electroencephalographically recorded during one night of laboratory sleep.
METHODS
Forty male subjects between the ages of 50 and 60 years (mean, 55.3) were
selected from a population of career service employees and salaried professionals
at the University of Florida after questionnaire responses indicated normal sleep
patterns, routine medical histories, and the absence of recent physical ailments.
For each session, two subjects reported to the laboratory approximately 1Y2 hr
prior to their normal bedtimes (about 11 p.m.) for electrode placement. Standard
placement sites and application procedures were employed and have been described elsewhere (Williams et aI., 1964). In addition, a push-button device was
taped into the preferred hand of each subject in a manner amenable to operation
by slight pressure applied by the thumb. Subjects were instructed to depress the
button each time they became aware of being awake within the sleep period.
At the completion of the presleep regimen, subjects retired to separate,
soundproof, temperature-controlled rooms. Continuous EEG and electro-oculographic (EOG) recordings were obtained for each subject for one night using a
Grass model VI electrophysiograph. Alpha activity (8-13 Hz) was recorded from
occipital leads (P3-T5). Activation of the push-button device was also registered
Sleep. Vol. 4, No.2, /98/
PERCEPTION OF WAKEFULNESS
179
on a separate channel of each subject's record. Each record was independently
scored in 1 min epochs by two technicians using the Agnew and Webb (l972b)
sleep stage scoring criteria.
RESULTS
An initial review of our data revealed a distinct population of subjects who had
very few signals of wakefulness or who failed to respond to extended periods of
wakefulness (3 min or more). These "nonresponses" could have been due to
either a failure or an inability to comply with instructions or to an absence of
subjectively experienced periods of awakening. Since our interest lay in the relationship between the EEG state and signal responses, we eliminated these nonrespondents from further analyses. The population eliminated and their response
levels are displayed in Table 1 with comparisons to the "experimental" group.
The remaining group of 28 subjects gave 242 signals of wakefulness. In our
initial analyses of the data, 52% of the responses occurred in stage 0, 12% in a
sleep stage which was immediately followed by stage 0, and 36% in a sleep stage
not followed by stage O.
A closer examination of these data revealed that the 12% of the responses
occurring in a sleep stage followed by stage 0 were artifactually classified as within
stage responses due to the 1 min scoring criterion used. With one exception
(which could not be scored due to muscle artifacts), all of these signals were
preceded by at least 6 sec of alpha (stage 0) and followed by 1 min or more of stage
O. These responses are subsequently treated as stage 0 signals.
When EEG-defined arousals were signaled, they were virtually immediate relative to the appearance of stage O. As noted, 12% occurred "within stages" after 6
sec or more alpha activity. Of the remaining stage 0 responses, those which
occurred within stage 0 minute scoring epochs, 94% occurred within the first
minute. The remainder occurred within 2 min.
Paradoxically, 36% of the signals indicating wakefulness were within unambiguous sleep stages and were not followed by a stage O. This paradox is reduced
by a closer examination of the scoring epochs involved. Of these 88 within-stage
responses, 73% were associated with at least 4-6 sec of preceding alpha activity.
This finding further emphasizes the immediacy of perceived wakefulness relative
to the EEG criteria of wakefulness. It further reduces signals of wakefulness in the
absence of EEG criteria to 24 signals, or slightly less than 10% of the 242 signals.
U sing the presence of EEG indications of wakefulness events, there were 211
episodes of stage 0 periods that were 1 min or longer. Fifty-seven (27%) of these
were not signaled. We could determine only two characteristics of the unsignaled
responses that were noteworthy. There was a "time of night" (Table 2) effect,
since 30% of the unreported episodes occurred within the first hour of sleep. We
examined the possibility that these may be "amnesic" type arousals from stage 3
or 4, but found only 2 of the 17 "nonresponses" preceded by these stages. Most of
the unreported arousals were of short duration and were within sustained sleep.
Eighty-four percent of these episodes of stage 0 were less than 2 min long. The
mean duration was 1.88 min (range, 1-9 min).
Sleep, Vol. 4, No.2, 1981
-
'""
00
c
~
~
.....
~
.'"
-
TABLE 1. Response characteristics of subjects eliminated from data base and
cumulative statistic vs. "experimental" cumulative statistics
~
Subject
Total responses
(n)
Response to
stage 0
Nonresponses to
stage 0
Responses to
stage 0 > 3 min"
Nonresponses to
stage 0 > 3 min
528
568
577
585
594
603
610
625
641
648
676
671
I
I
I
5
0
0
6
I
I
0
3
0
2
0
4
0
0
0
0
0
0
0
0
0
0
<"'l
I
~
I
9
13
4
7
18
12
6
10
2
4
6
5
0.83
(9.4% of
total stage 0)
8.0
(90.6% of
total stage 0)
0.17
(6.3% of total
stage 0 > 3 min)
2.5
(93.7% of total
stage 0 > 3 min)
4.46
(68.7% of
total stage 0)
2.04
(32.3% of
total stage 0)
1.07
(88.2% of total
stage 0 > 3 min)
0.14
of total
stage 0 > 3 min)
I
4
2
1
2
0
5
I
0
3
6
0
0
4
0
1
3
I
0
0
I
VJ
~
~
~
tTl
t-<
t-<
~
~
tl
~
tXl
Mean
Experimental
mean
1.67
8.60
" Excluding bathroom and final wakings.
I
(II. 8%
~
tXl
tXl
181
PERCEPTION OF WAKEFULNESS
TABLE 2. Distribution of reported and unreported stage 0 episodes
by time of night (adjusted for latency)
0-135 min
136-240 min
241-345 min
346 min-waking
Total
Reported
33
35
25
32
125
Unreported
25
10
14
8
57
Episode
In addition to the failures to signal unambiguous periods of EEG "wakefulness," an additional group of un signaled events must be acknowledged. We noted
above that signals that were presumably occurring in the absence of EEG arousals
were, in fact, associated with brief periods (4-6 sec) of alpha. For each of these 64
episodes we randomly selected a 1 min epoch, matched for subject and the stage
of signaling, in which signaling was not present. We found that 22% of these
selected episodes had alpha periods of longer than 4 sec.
One final analysis was undertaken. The stages associated with the signaled
arousals were determined. Shown in Table 3 are signaled wakings that were
associated with subsequent stage 0 (64% as designated above); the stages in which
a signal occurred but was not followed by sustained stage 0 are also presented.
The" actual" percentage of stages either preceding or accompanying the signaled
arousals, as well as the "expected," are shown. The "expected" distribution
displayed by the subjects across the night is adjusted for the exclusion of stage 0
[i.e., the five stages (1-4 + REM) comprise 100% of the night]. Responses are
clearly weighted toward stage 1 and stage REM; responses for stages 3 and 4 in
instances of unambiguous arousal are less frequent. The differential weighting of
REM is not present in the case of "within-stage" responses.
DISCUSSION
We have been concerned with two definitions of wakefulness during sleep;
electroencephalographically defined wakefulness (stage 0) and a behavioral signal
by the subject that he is awake. We have examined the relationship between these
two criteria in a group of subjects who demonstrably could perform the required
signaling.
Subjects' signaling of wakefulness were clearly associated with the presence of
an EEG sign of wakefulness. Sixty-four percent of the signals were associated
with the onset of an EEG epoch of stage O. In addition, 26% of the signals
TABLE 3. Distribution of responses in stages immediately preceding subjectively
acknowledged stage 0 events and within stages in which wakefulness was signaled
and not associated with stage 0
Preceding stage
Actual %
Expected %
2
3
11.7
51.3
6.4
53.2
Within stage
4
REM
3.2
4.5
29.2
13.6
60.2
6.1
14.7
19.5
6.4
53.2
2
4
REM
3.4
2.3
20.5
6.1
14.7
19.5
3
Sleep, Vol. 4, No.2, 1981
182
S. S. CAMPBELL AND W.B. WEBB
occurring apparently within uninterrupted sleep were immediately preceded by
brief episodes of EEG wakefulness. Only 10% of the signals of wakefulness were
not associated with an EEG indication of wakefulness.
The signals of wakefulness were closely associated with the time of occurrence
of the onset of EEG arousal. Eighty-four percent of the signals occurred within 4
sec to 1 min of the EEG-defined period. The awareness of wakefulness, as signaled, is not a slow and gradual process.
On the other hand, signals of wakefulness did not occur in the presence of some
periods of EEG arousal. Twenty-seven percent of the EEG-defined periods of
wakefulness of at least 1 min in duration were not signaled. In addition, an estimated 20% of brief periods of EEG arousal which the subjects had shown some
responsivity to, by signals within stages, were not signaled.
Clearly, the two measures of wakefulness are related. Signals of being awake
are preponderantly associated with EEG-defined periods of wakefulness. These
signals have short latencies and may be associated with very brief EEG episodes.
On the other hand, there are substantial periods of EEG arousal which are not
signaled. In addition to un signaled brief episodes which the subjects, on other
occasions, showed a capability of responding to, 27% of episodes lasting 1 min or
more were not signaled.
Some of the discontinuity between the measures may be the result of methodology. Some signals of wakefulness may have been "false positives" due to the
use of a single button push response. However, the device was designed to
minimize such responses, and we believe that the preponderance of such signals
were indications of wakefulness independent of EEG signaled wakefulness. The
absence of signals in response to EEG wakefulness could have been due to an
inability to respond or a failure to follow instructions due to forgetfulness. It is to
be recalled, however, that these were selected subjects, all of whom demonstrated
successful responding several times during the night. It seems likely, in view of
the highly efficient signaled responses, that some significant portion of the failures
to signal wakefulness represents periods in which wakefulness was not perceived.
The relationship of perceived wakefulness to sleep stage is not a simple one.
Our subjects signaled arousals from all stages of sleep, although more often from
stages 1 and REM and less frequently from stages 3 and 4.
Minimally, these data indicate that self-signaled arousals deserve consideration
in clinical analyses where reports of arousals are a part of the reported sleep
disturbance. Additional studies should clarify the meaning of wakefulness within
sleep.
ACKNOWLEDGMENT
This work was supported by National Institute on Aging Grant 5ROl AG
00805-02.
REFERENCES
Agnew H and Webb WB. Measurement of sleep onset by EEG criteria. Am J EEG Techno/
12:127-134.1972a.
Sleep. Vol. 4. No.2, 1981
PERCEPTION OF WAKEFULNESS
183
Agnew H and Webb WB. Sleep stage scoring. Journal Supplement Abstract Service of APA. manuscript #293. May 1972b.
Foulkes D and Vogel G. Mental activity at sleep onset. J Abnorm Psychol 70:231-243, 1965.
Kamiya J. Behavioral, subjective and physiological aspects of drowsiness and sleep. In: DW Fiske and
SR Maddi (Eds), Functions of Varied Experience. Dorsey Press, Homewood, Ill, 1961.
Loomis A, Harvey E, and Hobart G. Cerebral states during sleep, as studied by human brain potentials. J Exp Psychol 21: 127-144, 1937.
Rechtschaffen A and Kales A (Eds). A Manual of Standardized Terminology, Techniques and Scoring
System for Sleep Stages of Human Subjects. Brain Information Service/Brain Research Institute,
University of California at Los Angeles.
Walker P. Sleep onset as a function of auditory stimulation rates, response requirements, and novelty
of the environment. University of Florida PhD dissertation, 1972 (unpublished).
Williams R, Agnew H, and Webb WB. Sleep patterns in young adults: An EEG study. EEG Clin
Neurophysiol 17:376-381, 1964.
Sleep, Vol. 4. No.2. 1981