J. Sleep Res. (2002) 11, 53±59
Sleep- and circadian-dependent modulation of REM density
SAT BIR S. KHALSA1, DEIRDRE A. CONROY1, JEANNE F. DUFFY1,
C H A R L E S A . C Z E I S L E R 1 and D E R K - J A N D I J K 1 , 2
1
Division of Sleep Medicine, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA
Centre for Chronobiology, University of Surrey, Guildford, UK
2
Accepted in revised form 1 October 2001; received 19 April 2001
SUMMARY
Rapid eye movement (REM) density, a measure of the frequency of rapid eye
movements during REM sleep, is known to increase over the course of the sleep
episode. However, the circadian modulation of REM density has not been thoroughly
evaluated. Data from a forced desynchrony protocol, in which 20 consecutive sleep
opportunities were systematically scheduled over the entire circadian cycle, were
analysed. The REM density was evaluated from polysomnographically recorded REM
sleep episodes, and analyzed as a function of time in the sleep opportunity and as a
function of phase in the circadian cycle. The REM density showed a robust increase
over the course of the sleep episode. This sleep-dependent increase was observed
regardless of circadian phase, because data analyzed from di€erent thirds of the
circadian cycle exhibited a similar pattern. The REM density did not show a signi®cant
circadian-dependent modulation for data from the entire sleep opportunity. However,
analysis of circadian modulation from separate thirds of the sleep opportunity revealed
a signi®cant circadian modulation in the last third of the sleep episode. Maximum REM
densities were observed when the last third of the sleep episode coincided with the wakemaintenance zone, i.e. 8±10 h before the crest of the circadian rhythm of REM sleep
propensity. These results con®rm the dominant sleep-dependent modulation of REM
density, and indicate that the density of REMs is greatest when sleep pressure is low,
such as in the latter part of the sleep episode, at which time the circadian modulation of
REM density is also appreciable.
KEYWORDS
forced desynchrony, rapid eye movements, REM sleep
INTRODUCTION
It is now well known that rapid eye movement (REM) density,
which is a measure of the frequency of REMs during REM
sleep, increases over the course of the sleep episode in
successive REM sleep episodes (Aserinsky 1969, 1973; Benoit
et al. 1974; Benson and Zarcone 1993; Castaldo and Krynicki
1974; Cohen 1975; Feinberg 1974; Feinberg et al. 1980; Ficca
et al. 1999; Foster et al. 1976; Geisler et al. 1987; Kobayashi
et al. 1980; Takahashi and Atsumi 1997; Zimmerman et al.
1980). Protocols utilizing extended sleep episodes have suggested that REM density increases with prolonged sleep
Correspondence: Sat Bir S. Khalsa PhD, Division of Sleep Medicine, Department of Medicine, Brigham and Women's Hospital,
Harvard Medical School, 221 Longwood Avenue, Boston, MA
02115, USA. Tel.: +1 617 732 7994; fax: +1 240 269 6205; e-mail:
[email protected]
Ó 2002 European Sleep Research Society
duration and eventually saturates at a maximal level (Aserinsky 1969, 1973; Feinberg et al. 1980; Wehr et al. 1993). The
hypothesis that REM density is an index of sleep satiety has
been proposed (Aserinsky 1969). Accordingly, an inverse
relationship between the depth of sleep and REM density
(Feinberg et al. 1988; Lucidi et al. 1996) has been observed.
Recovery sleep following partial or total sleep deprivation is
associated with a reduction in REM density proportional to
the degree of sleep loss (Aeschbach et al. 1996; Feinberg et al.
1987, 1988; Lucidi et al. 1996; Reynolds et al. 1986, 1987;
Travis et al. 1991). Autonomic activation has been associated
with bursts of rapid eye movements during REM sleep (Spreng
et al. 1968), although in another study no correlation was
reported between REM density and autonomic variables
(Vgontzas et al. 1997). Further evidence of the hypothesized
relationship between REM density and arousal comes from a
recent study showing that REM density was higher just before
53
54
S. S. Khalsa et al.
transitions to wakefulness than before transitions back to sleep
(Barbato et al. 1994). Furthermore, a systematic reduction in
the duration of prior wakefulness enhances REM density
(Feinberg et al. 1987). The inverse relationship between
Process S (representing homeostatic sleep drive) and REM
density has lead to the hypothesis that Process S exerts an
inhibitory in¯uence on REM density (Kupfer et al. 1984).
The REM density patterns observed in extended sleep
episode and sleep deprivation (SD) studies have suggested that
sleep-dependent REM density modulation is apparently independent of circadian phase. In free-running laboratory conditions, during which time subjects self-select sleep and wake
times in the absence of time cues and at atypical circadian
phases, the time course of REM density appears to behave as it
does under normal conditions, with a progressive increase in
successive REM sleep episodes. However, the normal temporal
expression of REM sleep with its progressively increasing
episode duration over the course of the sleep episode is no
longer observed under these conditions (Zimmerman et al.
1980). In this study, we evaluate the sleep- and circadiandependent modulation of REM density using the forced
desynchrony protocol, which is capable of distinguishing
sleep- and circadian-dependent modulation (Dijk and Czeisler
1994, 1995; Dijk et al. 1997, 1999a; Wei et al. 1999).
MATERIALS AND METHODS
Data for this analysis were acquired from ®ve young men aged
21±30. Prior to study, all subjects were entrained to a regular
sleep±wake schedule, were screened for any current or past
medical or psychiatric history, and were drug-free at the time
of study. Subjects underwent a 31±34-day in-laboratory study
which included a constant routine protocol prior to and
following 20 cycles of a 28-h day forced desynchrony protocol.
The forced desynchrony segment consisted of an activity±rest
schedule with 18 h 40 min of scheduled wakefulness in dim
light (10±20 lux) and 9 h 20 min of scheduled sleep in
darkness (<0.03 lux), during which time subjects were con®ned to bed and instructed to attempt sleep. During the forced
desynchrony protocol the timing of the 20 sleep opportunities
occurred at di€erent circadian phases, because the imposed
day-length was outside of the range of circadian entrainment
in dim light. Complete descriptions and details of the protocol
and methods have been published previously (Czeisler et al.
1999; Dijk and Czeisler 1994, 1995; Dijk et al. 1997, 1999a;
Wei et al. 1999).
Hourly blood samples collected during intermittent segments throughout the study were assayed for plasma melatonin concentration using a radioimmunoassay with a sensitivity
of 5 pg mL)1 (Pharmasan Laboratories, Osceola, WI, USA).
Circadian phase of the melatonin rhythm was evaluated by
®tting a dual-harmonic regression model to the data from the
entire study (Czeisler et al. 1999). The circadian period and
phase of the ®tted maximum of the melatonin rhythm from
this analysis were used to assign a circadian phase to each sleep
opportunity of the forced desynchrony protocol.
Polysomnography during the forced desynchrony protocol
was performed using a Nicolet Ultrasom Sun 386I sleep
system. The recording montage included central and occipital electroencephalogram (EEG) (referenced to mastoids),
right and left electrooculogram (EOG), chin electromyogram
(EMG), and electrocardiogram (ECG). The EOG recordings
were AC-coupled with a time constant of 0.3 s. Sleep stage
was scored in 30-s epochs by trained polysomnographic
technicians using standard criteria (Rechtscha€en and Kales
1968). Rapid eye movements during REM sleep epochs were
scored visually by displaying each epoch in 15-s segments on
a CRT screen, and the total number of REMs was counted
for each 30 s REM epoch, thereby yielding values of REM
density (number of eye movements per 30 s). Each eye
movement which was detectable above the background noise
exhibited a rapid time course, and appeared simultaneously
on both right and left EOG channels was counted regardless
of the amplitude of the eye movement (Aserinsky 1969,
1973; Hong et al. 1995). Rapidly cycling left and right
REMs, and stepwise saccades in the same direction of gaze
were all counted as separate eye movements. Slow eye
movements, and the passive capacitive signal decay to
baseline following REM excursions were not counted (Boukadoum and Ktonas 1986). All REM density values from
three subjects were scored by DAC, and from the other two
subjects by SBSK.
All REM sleep epochs in all 20 sleep opportunities for three
subjects were evaluated. In one subject, three sleep recordings,
and a segment of another were unavailable for analysis. In
another subject, one sleep recording was unavailable, two sleep
recordings included only one eye movement channel, but were
scored and included in the analysis, and segments of data from
another sleep recording were unavailable.
For each subject, the REM density values for all REM
epochs in all sleep recordings were binned with respect to
circadian phase (using circadian time referenced to the
melatonin acrophase) in one analysis, and with respect to
time since the start of the sleep opportunity (lights out) in a
separate analysis. Five circadian bins were used (72 degrees
per bin), and 5 bins were used for the 540-min sleep
opportunity (112 min per bin). Z-scores were calculated for
each subject by subtracting the average of all REM density
values (from all of the sleep recordings) from each individual
REM density value and then dividing this by the standard
deviation of all REM values. All of the Z-scores within each
circadian or sleep bin were averaged for each subject
individually. Finally, the circadian and sleep bins were
averaged across all ®ve subjects to yield plots of REM density
as a function of circadian phase and as a function of time in
the sleep opportunity. To evaluate circadian- and sleepdependent modulation over di€erent sleep segments and
circadian segments, respectively, the binning procedure described above was applied to data speci®cally drawn from
thirds of the sleep opportunity and from thirds of the
circadian cycle. Statistical signi®cance of circadian- and
sleep-dependent modulation was determined using a repeated
Ó 2002 European Sleep Research Society, J. Sleep Res., 11, 53±59
REM density modulation
measures analysis of variance (ANOVA) with a Huyn±Feldt
correction (e-values reported).
RESULTS
From the total of 96 sleep recordings analyzed from the ®ve
subjects, a total of 16,817 30-s REM sleep epochs were scored
for REMs, with the number of REM sleep epochs per subject
ranging from 2798 to 3872. The distribution of the number of
REM counts per epoch revealed an exponential-type decrease,
with the epochs showing 0, 1 or 2 eye movements accounting
for 29% of all epochs. Epochs with 10 or fewer REMs
accounted for 65% of all epochs, epochs with 11±20 REMs
accounted for 19% of all epochs, and epochs with more than
30 REMs accounted for 16% of all epochs.
The pro®le of average REM density as a function of time in
the sleep opportunity is shown in Fig. 1. The lowest REM
density occurred in the ®rst ®fth of the sleep opportunity,
increasing over the course of the sleep opportunity, with the
®nal ®fth of the sleep opportunity showing the highest REM
density. For each subject, this same general trend applied, with
the lowest REM density in the ®rst ®fth of the sleep
opportunity, an increase over the course of the sleep opportunity, and the highest REM density in the last (four subjects)
or penultimate ®fth (one subject) of the sleep opportunity.
ANOVA con®rmed that this sleep-dependent modulation was
signi®cant (F4,16 ˆ 47.3, P < 0.00001, e ˆ 1.0), and a regression analysis through the temporal pro®le was linear
(P < 0.05) with an r-value of 0.96.
To evaluate whether this sleep-dependent increase applied to
di€erent circadian phases, the entire data set from each subject
was subdivided into three circadian segments (300±60 degrees,
60±180 degrees and 180±300 degrees, where 0 degrees represents the ®tted maximum of the melatonin rhythm). The data
within each third of the circadian cycle were then binned into
Figure 1. Sleep-dependent modulation of rapid eye movement (REM)
density. Average Z-scores of REM density (‹SEM) are plotted as a
function of time in the sleep opportunity. A robust sleep-dependent
increase in REM density is apparent. Data analyzed and presented in
all ®gures are from analysis of a total of 96 sleep episodes from ®ve
subjects.
Ó 2002 European Sleep Research Society, J. Sleep Res., 11, 53±59
55
®fths of the sleep opportunity, averaged within each bin,
and then averaged across subjects. The resulting averages are
plotted in Fig. 2. In all three circadian segments, the lowest
REM density was in the ®rst ®fth of the sleep opportunity with a
general increase in REM density over the course of the sleep
opportunity and a maximum REM density in the ®nal ®fth of
the sleep opportunity. Regressions through the temporal pro®les
in each circadian segment were linear (P < 0.05) with r-values
of 0.78 for the 300±60 degree segment, 0.83 for the 60±180
degree segment, and 0.99 for the 180±300 degree segment.
A two-way ANOVA revealed a trend for a main e€ect of
circadian phase (F2,8 ˆ 3.8, P ˆ 0.07, e ˆ 0.78) and a signi®cant main e€ect for time within the sleep episode
(F4,16 ˆ 47.3, P < 0.0001, e ˆ 1.0). Post-hoc Tukey HSD tests
revealed that REM density was signi®cantly lower in the ®rst
®fth of the sleep opportunity compared with all others. In
addition, REM density during the second ®fth of the sleep
opportunity was signi®cantly lower than that during the fourth
®fth of the sleep opportunity and REM densities during the
second and third ®fths of the sleep opportunity were signi®cantly lower than that for the last ®fth of the sleep opportunity.
The pro®le of average REM density as a function of
circadian phase is shown in Fig. 3. Although average REM
density appears to be higher between 216 and 360 degrees and
lower between 0 and 216 degrees, the overall circadian
modulation in Fig. 3 was not statistically signi®cant
(F4,16 ˆ 1.4, P ˆ 0.27, e ˆ 1.0).
To evaluate the existence of any dependence of circadian
modulation on the time in the sleep opportunity, the entire
data set from each subject was subdivided into three segments
of the sleep opportunity (0±180 min, 180±360 min, and 360±
540 min of the scheduled sleep opportunity). The data from
each third of the sleep opportunity were then binned into ®fths
of the circadian cycle, and averaged within each of these ®ve
Figure 2. Sleep-dependent modulation of rapid eye movement (REM)
density evaluated for di€erent segments of the circadian cycle. Average
Z-scores of REM density (‹SEM) for consecutive ®fths of the sleep
opportunity from di€erent thirds of the circadian cycle are plotted.
The sleep-dependent increase in REM density is apparent regardless of
circadian phase.
56
S. S. Khalsa et al.
Figure 3. Circadian-dependent modulation of rapid eye movement
(REM) density. Average Z-scores of REM density (‹SEM) are
double-plotted as a function of circadian phase. There is no signi®cant
circadian- dependent modulation.
bins and then averaged across subjects. The resulting averages
are double-plotted in Fig. 4 for each third of the sleep
opportunity. The general pattern observed in the overall
circadian modulation of REM density in Fig. 3 is re¯ected in
the patterns for each third of the sleep opportunity, with
higher REM density between 216 and 360 degrees and lower
REM density after 0 degrees. This modulation appears
smallest in the ®rst third of the sleep opportunity (Fig. 4, top
panel), larger in the second third of the sleep opportunity
(Fig. 4, middle panel) and strongest in the ®nal third of the
sleep opportunity (Fig. 4, bottom panel). Subdivision of the
sleep opportunity into ®rst and second halves, and analysis for
circadian modulation revealed a similar trend. The circadian
modulation was not statistically signi®cant in the ®rst half of
the sleep opportunity, but was statistically signi®cant by oneway ANOVA in the second half (F4,16 ˆ 3.2, P < 0.05,
e ˆ 1.0), with the highest REM density in the bin centred at
252 degrees. This was the same circadian bin that showed the
highest REM density in the last third of the sleep opportunity
(Fig. 4, bottom panel).
A two-way ANOVA for circadian phase across thirds of the
sleep episode showed signi®cant main e€ects for circadian
phase (F4,16 ˆ 3.0, P ˆ 0.051, e ˆ 0.91) and time of the sleep
episode (F2,8 ˆ 121.9, P < 0.0001, e ˆ 0.67), and the interaction term was not signi®cant (F8,32 ˆ 1.58, P ˆ 0.17). The
REM density was signi®cantly higher in the fourth ®fth of the
sleep opportunity than in the ®rst. The REM density was
lowest between 300 and 60 degrees, and signi®cantly increased
between 60 and 180 degrees and between 180 and 300 degrees.
DISCUSSION
The results of this study con®rm and extend those of previous
studies, that REM density increases as a function of time in the
sleep episode (Aserinsky 1969, 1973; Benoit et al. 1974; Benson
Figure 4. Circadian-dependent modulation of rapid eye movement
(REM) density evaluated for di€erent segments of the sleep opportunity. Each panel shows average Z-scores of REM density (‹SEM)
double-plotted as a function of circadian phase from di€erent thirds of
the sleep opportunity. The circadian-dependent modulation is only
signi®cant in the last third of the sleep opportunity (bottom panel)
where the highest REM density is apparent during the wakemaintenance zone.
and Zarcone 1993; Castaldo and Krynicki 1974; Cohen 1975;
Feinberg 1974; Feinberg et al. 1980; Ficca et al. 1999; Foster
et al. 1976; Geisler et al. 1987; Kobayashi et al. 1980; Takahashi and Atsumi 1997; Zimmerman et al. 1980). As suggested
by earlier studies in which sleep episodes occurred at atypical
circadian phases (Foret et al. 1973; Zimmerman et al. 1980),
Ó 2002 European Sleep Research Society, J. Sleep Res., 11, 53±59
REM density modulation
this relationship persists regardless of the circadian phase
during which the sleep episode occurs (Fig. 2).
Circadian modulation of REM density has been suggested
in two previous reports (Kobayashi et al. 1980; Krynicki
1975). In a protocol employing sleep opportunities presented
following wakefulness lasting either 16, 26, 30 or 34 h,
Kobayashi et al. (1980) reported that REM density was low
after 16 h, highest after 26 h and then decreased progressively
after 30 and 34 h of wakefulness. They interpreted this increase
and subsequent decline as evidence of a circadian modulation.
However, the data from that preliminary report are at odds
with other sleep deprivation studies of REM density, which
universally report a decreasing REM density with increasing
prior wakefulness (Aeschbach et al. 1996; Antonioli et al.
1981; Feinberg et al. 1987, 1988; Lucidi et al. 1996; Travis
et al. 1991). On the basis of spectral analysis of the timing of
rapid eye movements in REM sleep, Krynicki (1975) reported
detection of a `24-h rhythm in one of the least-squares variance
spectra' in addition to ultradian components. However, few
details were presented in support of this observation. Furthermore, it is not clear how a 24-h component could be
determined from the 8 h nocturnal sleep episodes used in that
study. They also argued that their observed decrease in REM
frequency during wakefulness just prior to sleep onset,
followed by an increase in eye movements after sleep onset,
also suggests a circadian rhythm. However, it is not clear that
comparison of eye movements from wakefulness with those
from REM sleep is justi®able.
Although the occurrence of rapid eye movements is an
essential characteristic of REM sleep, and REM sleep exhibits
a strong circadian modulation, the results of the present study
suggest that the modulation of REM density may be largely
independent from the mechanisms that regulate REM sleep
itself. Data acquired under both free-running and forced
desynchrony protocols have revealed that REM sleep propensity is highest near the time of the core body temperature
minimum (Czeisler et al. 1980a, b; Dijk et al. 1997, 1999a; Dijk
and Czeisler 1995; Wyatt et al. 1999), whereas the peak REM
density occurs much earlier in the circadian cycle (see below).
There is ample other evidence supporting the independence of
REM sleep and REM density from a number of studies using
di€erent experimental protocols, and noting signi®cant di€erences in behavioral characteristics between these two processes. For example, in an extended sleep protocol it was observed
that there was no signi®cant correlation between REM
percentage and REM density (Aserinsky 1969). The REM
percentage has been shown to increase in response to selective
REM deprivation, whereas REM density decreases and then
remains stable on subsequent nights (Antonioli et al. 1981).
Even if sleep deprivation is not selective, recovery REM sleep
is enhanced, whereas REM density decreases (Lucidi et al.
1996). Developmental di€erences between REM sleep cycles
and REM density have been noted (Petre-Quadens and De Lee
1974). Finally, the temporal expression of REM density and
REM sleep cycles have been shown to di€er within a sleep
episode (Benoit et al. 1974), or under circumstances in which
Ó 2002 European Sleep Research Society, J. Sleep Res., 11, 53±59
57
sleep episodes occur at atypical circadian phases (Foret et al.
1973; Zimmerman et al. 1980).
The hypothesis that REM sleep timing is modulated by
di€erent physiological mechanisms than is REM density,
together with the strong evidence indicating that REM sleep
propensity is driven by the circadian clock, has led to the
suggestion that REM density may not be modulated by the
circadian clock (Zimmerman et al. 1980). To e€ectively
evaluate circadian REM density modulation, it is necessary
to use a protocol in which sleep is scheduled at a variety of
di€erent circadian phases (Zimmerman et al. 1980). This
condition is ful®lled with the forced desynchrony protocol,
which has the additional advantage over previous research
protocols in that wake durations are held constant throughout
the protocol, thereby maintaining homeostatic sleep drive
relatively stable prior to sleep opportunities. Circadian modulation of REM density is much weaker than that of sleep
satiety/sleep need. Furthermore, the circadian modulation is
also dependent upon prior sleep history, in that a signi®cant
modulation was only observed at high REM density levels, late
in the sleep episode. Such an interactive component has also
been reported for sleep propensity, sleep structure and
neurobehavioral characteristics, which also show an increasing
circadian modulation with time since the start of the wake or
sleep episode (Dijk and Czeisler 1995; Jewett et al. 1996; Wyatt
et al. 1999). (See also Dijk et al. 1999b for further discussion
on interactions).
The circadian pattern of REM density in this study is
consistent with the observed circadian modulation of sleep
propensity. The REM density levels were relatively elevated at
270 degrees (approximately 240 degrees with circadian phase
referenced to the core body temperature minimum at 0 degrees;
Figs 3 and 4), which is approximately 6 h prior to the middle
of the sleep episode and 2 h prior to the start of the sleep
episode on a typical nocturnal sleep schedule. This circadian
phase has been associated with elevated wakefulness, long
sleep latency, low sleep eciency and elevated neurobehavioral
performance as determined in forced desynchrony studies
(Dijk et al. 1999a; Dijk and Czeisler 1994, 1995; Wyatt et al.
1999), spontaneous desynchrony studies (Czeisler et al.
1980a, b; Strogatz et al. 1987) and in an ultra-short sleep±
wake schedule condition (Lavie 1986). It has been referred to
as the `forbidden zone for sleep' (Lavie 1986) and the `wake
maintenance zone' (Strogatz et al. 1987). The REM density
levels are relatively depressed at the bin centered at 36 degrees
(Figs 3 and 4), which occurs about 1.5 h before wake time on a
typical sleep±wake schedule. This is close to the core body
temperature minimum, which occurs about 2.5 h prior to wake
time, and is associated with a signi®cant reduction of wakefulness and performance, an elevated sleep propensity and an
elevated REM sleep propensity (Czeisler et al. 1980a, b; Dijk
et al. 1997, 1999a; Dijk and Czeisler 1994, 1995; Wyatt et al.
1999). Two possibilities that could account for the circadian
modulation of REM density include a direct modulation by
the circadian pacemaker, and an indirect pathway, in which
the circadian pacemaker in¯uences REM density indirectly
58
S. S. Khalsa et al.
through its modulation of sleep propensity. Additional forced
desynchrony studies of the circadian modulation of REM
density with larger sample sizes would allow for a ®ner
resolution of the temporal pro®le of this modulation, which
may help distinguish between these possibilities.
ACKNOWLEDGEMENTS
The experimental studies were supported in part by NIA grant
P01-AG09975 and were performed in a General Clinical
Research Center supported by grant M01-RR02635 from the
National Center for Research Resources. SBSK was supported
by NIMH grant RO1-MH45130 and NHLBI Senior NRSA
Fellowship F33-HL09588. JFD was supported by NHLBI
grant T32-HL07901. We thank Dr S. Hurwitz and Dr K.P.
Wright for assistance with statistical analysis.
REFERENCES
Aeschbach, D., Cajochen, C., Landolt, H.-P. and BorbeÂly, A. A.
Homeostatic sleep regulation in habitual short sleepers and long
sleepers. Am. J. Physiol., 1996, 270: R41±R53.
Antonioli, M., Solano, L., Torre, A., Violani, C., Costa, M. and Bertini,
M. Independence of REM density from other REM sleep parameters
before and after REM deprivation. Sleep, 1981, 4: 221±225.
Aserinsky, E. Relationship of rapid eye movement density to the prior
accumulation of sleep and wakefulness. Psychophysiology, 1973,
1994, 10: 545±558.
Aserinsky, E. The maximal capacity for sleep. Rapid eye movement
density as an index of sleep satiety. Biol. Psychiatry., 1969, 1: 147±159.
Barbato, G., Barker, C., Bender, C., Giesen, H. A. and Wehr, T. A.
Extended sleep in humans in 14 hour nights (LD 10: 14):
relationship between REM density and spontaneous awakening.
Electroenceph. Clin. Neurophysiol., 1994, 90: 291±297.
Benoit, O., Parot, S. and Garma, L. Evolution during the night of
REM sleep in man. Electroenceph. Clin. Neurophysiol., 1974, 36:
245±251.
Benson, K. L. and Zarcone, P. Jr. Rapid eye movement sleep eye
movements in schizophrenia and depression. Arch. General Psychiatry., 1993, 50: 474±482.
Boukadoum, A. M. and Ktonas, P. Y. EOG-based recording and
automated detection of sleep rapid eye movements: a critical review
and some recommendations. Psychophysiology, 1986, 23: 598±611.
Castaldo, V. and Krynicki, V. Sleep and eye movement patterns in two
groups of retardates. Biol. Psychiatry., 1974, 9: 231±244.
Cohen, D. B. Eye movements during REM sleep: the in¯uence of
personality and presleep conditions. J. Pers. Soc. Psychol., 1975, 32:
1090±1093.
Czeisler, C. A., Du€y, J. F., Shanahan, T. L., Brown, E. N.,
Mitchell, J. F., Rimmer, D. W., Ronda, J. M., Silva, E. J., Allan,
J. S., Emens, J. S., Dijk, D.-J. and Kronauer, R. E. Stability,
precision, and near-24-hour period of the human circadian pacemaker. Science, 1999, 284: 2177±2181.
Czeisler, C. A., Weitzman, E. D., Moore-Ede, M. C., Zimmerman, J. C.
and Knauer, R. S. Human sleep: its duration and organization
depend on its circadian phase. Science, 1980b, 210: 1264±1267.
Czeisler, C. A., Zimmerman, J. C., Ronda, J. M., Moore-Ede, M. C. and
Weitzman, E. D. Timing of REM sleep is coupled to the circadian
rhythm of body temperature in man. Sleep, 1980a, 2: 329±346.
Dijk, D.-J. and Czeisler, C. A. Contribution of the circadian
pacemaker and the sleep homeostat to sleep propensity, sleep
structure, electroencephalographic slow waves, and sleep spindle
activity in humans. J. Neurosci., 1995, 15: 3526±3538.
Dijk, D.-J. and Czeisler, C. A. Paradoxical timing of the circadian
rhythm of sleep propensity serves to consolidate sleep and wakefulness in humans. Neurosci. Lett., 1994, 166: 63±68.
Dijk, D.-J., Du€y, J. F., Riel, E., Shanahan, T. L. and Czeisler, C. A.
Ageing and the circadian and homeostatic regulation of human
sleep during forced desynchrony of rest, melatonin and temperature
rhythms. J. Physiol. (Lond.), 1999a, 516: 611±627.
Dijk, D.-J., Jewett, M. E., Czeisler, C. A. and Kronauer, R. E. Reply
to technical note: nonlinear interactions between circadian and
homeostatic processes. Models or metrics? J. Biol. Rhythms., 1999b,
14: 604±605.
Dijk, D.-J., Shanahan, T. L., Du€y, J. F., Ronda, J. M. and
Czeisler, C. A. Variation of electroencephalographic activity during
non-rapid eye movement and rapid eye movement sleep with phase
of circadian melatonin rhythm in humans. J. Physiol. (Lond.), 1997,
505: 851±858.
Feinberg, I., Baker, T., Leder, R. and March, J. D. Response of delta
(0±3 Hz) EEG and eye movement density to a night with 100
minutes of sleep. Sleep, 1988, 11: 473±487.
Feinberg, I. Changes in sleep cycle patterns with age. J. Psychiatr.
Res., 1974, 10: 283±306.
Feinberg, I., Fein, G. and Floyd, T. C. EEG patterns during and
following extended sleep in young adults. Electroenceph. Clin.
Neurophysiol., 1980, 50: 467±476.
Feinberg, I., Floyd, T. C. and March, J. D. E€ects of sleep loss on delta
(0.3±3Hz) EEG and eye movement density: new observations and
hypotheses. Electroenceph. Clin. Neurophysiol., 1987, 67: 217±221.
Ficca, G., Gori, S., Ktonas, P., Quattrini, C., Trammell, J. and
Salzarulo, P. The organization of rapid eye movement activity
during rapid eye movement sleep is impaired in the elderly.
Neurosci. Lett., 1999, 275: 219±221.
Foret, J., Lantin, G. and Benoit, O. Irregular hours of sleep and the
organization of paradoxical sleep in man. In: W. P. Koella and
P. Levin (Eds) Sleep. Proceedings of the First European Congress
on Sleep Research. Karger, Basel, Switzerland, 1973: 439±442.
Foster, F. G., Kupfer, D. J., Coble, P. and McPartland, R. J. Rapid
eye movement sleep density: an objective indicator in severe
medical±depressive syndromes. Arch. General Psychiatry., 1976,
33: 1119±1123.
Geisler, P., Meier-Ewert, K. and Matsubayshi, K. Rapid eye movements, muscle twitches and sawtooth waves in the sleep of
narcoleptic patients and controls. Electroenceph. Clin. Neurophysiol., 1987, 67: 499±507.
Hong, C. C., Gillin, J. C., Dow, B. M., Wu, J. and Buchsbaum, M. S.
Localized and lateralized cerebral glucose metabolism associated
with eye movements during REM sleep and wakefulness: a positron
emission tomography (PET) study. Sleep, 1995, 18: 570±580.
Jewett, M. E., Dijk, D.-J., Kronauer, R. E., Boivin, D. B. and
Czeisler, C. A. Homeostatic and circadian components of subjective
alertness interact in a non-additive manner during 48 hours of sleep
deprivation. J. Sleep Res., 1996, 5 (Suppl. 1): 101.
Kobayashi, T., Endo, S., Saito, Y. and Tsuji, Y. Is there a circadian
rhythm in REM activity? Sleep Res., 1980, 9: 273.
Krynicki, V. Time trends and periodic cycles in REM sleep eye
movements. Electroenceph. Clin. Neurophysiol., 1975, 39: 507±513.
Kupfer, D. J., Ulrich, R. F., Coble, P. A., Jarrett, D. B., Grochocinski,
V., Doman, J., Matthews, G. and BorbeÂly, A. A. Application of
automated REM and slow wave sleep analysis. II. Testing the
assumptions of the two-process model of sleep regulation in normal
and depressed subjects. Psychiatry Res., 1984, 13: 335±343.
Lavie, P. Ultrashort sleep-waking schedule III. `Gates' and `forbidden
zones' for sleep. Electroenceph. Clin. Neurophysiol., 1986, 63:
414±425.
Lucidi, F., Devoto, A., Violani, C., De Gennaro, L., Mastracci, P. and
Bertini, M. Rapid eye movements density as a measure of sleep
need: REM density decreases linearly with the reduction of
Ó 2002 European Sleep Research Society, J. Sleep Res., 11, 53±59
REM density modulation
prior sleep duration. Electroenceph. Clin. Neurophysiol., 1996, 99:
556±561.
Petre-Quadens, O. and De Lee, C. Eye movement frequencies and
related paradoxical sleep cycles: developmental changes. Chronobiologia, 1974, 1: 348±355.
Rechtscha€en, A. and Kales, A. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human
Subjects. U.S. Government Printing Oce, Washington, DC, 1968.
Reynolds, C. F. III, Kupfer, D. J., Hoch, C. C., Houck, P. R., Stack, J.
A., Berman, S. R., Campbell, P. I. and Zimmer, B. Sleep
deprivation as a probe in the elderly. Arch. General Psychiatry,
1987, 44: 982±990.
Reynolds, C. F. III, Kupfer, D. J., Hoch, C. C., Stack, J. A.,
Houck, P. R. and Berman, S. R. Sleep deprivation in healthy elderly
men and women: e€ects on mood and on sleep during recovery.
Sleep, 1986, 9: 492±501.
Spreng, L. F., Johnson, L. C. and Lubin, A. Autonomic correlates of
eye movement bursts during stage REM sleep. Psychophysiol., 1968,
4: 311±323.
Strogatz, S. H., Kronauer, R. E. and Czeisler, C. A. Circadian
pacemaker interferes with sleep onset at speci®c times each day: role
in insomnia. Am. J. Physiol., 1987, 253: R172±R178.
Takahashi, K. and Atsumi, Y. Precise measurement of individual rapid
eye movements in REM sleep of humans. Sleep, 1997, 20: 743±752.
Ó 2002 European Sleep Research Society, J. Sleep Res., 11, 53±59
59
Travis, F., Maloney, T., Means, M., March, J. D. and Feinberg, I.
Acute deprivation of the terminal four hours of sleep does not
increase delta (0±3-Hz) electroencephalograms: a replication. Sleep,
1991, 14: 320±324.
Vgontzas, A. N., Bixler, E. O., Papanicolaou, D. A., Kales, A.,
Stratakis, C. A., Vela-Bueno, A., Gold, P. W. and Chrousos, G. P.
Rapid eye movement sleep correlates with the overall activities of
the hypothalamic-pituitary-adrenal axis and sympathetic system in
healthy humans. J. Clin. Endocrinol. Metab., 1997, 82: 3278±3280.
Wehr, T. A., Moul, D. E., Barbato, G., Giesen, H. A., Seidel, J. A.,
Barker, C. and Bender, C. Conservation of photoperiod-responsive
mechanisms in humans. Am. J. Physiol., 1993, 265: R846±R857.
Wei, H. G., Riel, E., Czeisler, C. A. and Dijk, D.-J. Attenuated
amplitude of circadian and sleep-dependent modulation of electroencephalographic sleep spindle characteristics in elderly human
subjects. Neurosci. Lett., 1999, 260: 29±32.
Wyatt, J. K., Ritz-De Cecco, A., Czeisler, C. A. and Dijk, D.-J.
Circadian temperature and melatonin rhythms, sleep, and neurobehavioral function in humans living on a 20-h day. Am. J. Physiol.,
1999, 277: R1152±R1163.
Zimmerman, J. C., Czeisler, C. A., Laxminarayan, S., Knauer, R. S.
and Weitzman, E. D. REM density is dissociated from REM
sleep timing during free-running sleep episodes. Sleep, 1980, 2:
409±415.