Theme : Dreams

SCN Controlled Circadian Arousal and the Afternoon "Nap Zone"

Roger J. Broughton
Division of Neurology, University of Ottawa and Ottawa Hospital (General Campus), Ottawa, K1H 8L6, Canada
Abstract
This paper outlines a conceptual model for the regulation of the circasemidian sleep propensity process with emphasis on a possible mechanism of the afternoon "nap zone". It is proposed that the afternoon nap zone is due to increasing sleep propensity after morning wakening (Borbély's Process-S) being overwhelmed by a light-sensitive SCN-dependent circadian arousal process of the type discovered by Edgar et al., (1993) and currently being identified in its pathways and neurochemistry by Jouvet and colleagues. It is maintained that this arousal process is reflected in the circadian core body temperature pattern, and that under normal entrained conditions the latter does not resemble a sine-wave or skewed sine-wave. Rather it is very asymmetrical in time and somewhat asymmetrical in amplitude. Cosinor type analyses which enforce symmetry in time and amplitude are therefore ill suited to adequately curve-fit the empirical data. The shape of the circadian arousal system was clarified by meta-analyses of data from three laboratories for three conditions: the normal entrained state, the constant routine, and temporal isolation. Under normal entrained conditions for about one-third of the circadian day core body temperature, and therefore the assumed intensity of the circadian arousal system, is below the mesor with the nadir being at about 0500h; and for about two-thirds of the circadian day it is above the mesor with the acrophase on average being at about 2100h. For modeling purposes, the homeostatic process (Process-S) employed the actual data of the Zurich laboratories for night sleep, but altered the equation for the daytime period to ensure an exponential increase after wake-up. Combining these modified processes indicated that the nap zone could be explained, as predicted, by an increasing homeostatic pressure for sleep across the daytime being reversed by the circadian arousal process. This 2-process combination predicted quite well the shape of the entire circasemidian sleep/wake propensity process and can explain the presence of morning sleep inertia without requiring a third process. It would appear that the circadian arousal process can be modified in phase and in amplitude by a number of normal and pathological conditions.

Current Claim: Altering the shape of the circadian core temperature curve from a sine-wave or skewed sine-wave to that of data empirically observed in the normal entrained state, equating this curve with the newly identified circadian arousal process, and combining it with a homeostatic sleep function, predicts quite accurately the daily circasemidian (biphasic) sleep/wake propensity curve.

Activate the ShortNotes by clicking on this link. Your notes will be stored in this area and automatically retrieved upon your next visit.
The circasemidian wake-sleep pattern

When following the common pattern of habitual monophasic night sleep, we typically experience a transitory period of increased daytime sleepiness around the time of the mid-afternoon followed by a period of heightened arousal which then rapidly diminishes shortly before the onset of the next night of sleep. This so-called "afternoon nap zone" can be expressed variously as a temporary period of poorer performance (Blake, 1967); the last nap given up in ontogenetic development (Webb and Dinges, 1989); the return of an afternoon nap after retirement and in the elderly, when social constraints are reduced (Webb and Dinges, 1989); napping in students (Dinges et al., 1980); siesta taking both in cultures which limit night sleep amount (Broughton, 1983) and in so-called primitive tribes studied by combined anthropological and polysomnographic techniques (Quadens, 1990); the period of both maximum sleepiness and (more or less) irresistible sleep attacks in sleep disordered patients with excessive daytime sleepiness (Richardson et al., 1978; Broughton, 1989a); and an increased rate both of accidents attributed to sleepiness and of death from all causes (Mitler et al., 1988). In fact, despite its varying inter-individual and intra-individual variability, the afternoon nap zone represents the most intense predictable fluctuation of level of sleepiness/alertness experienced during the normal waking daytime period.

Possible mechanisms

Numerous possible mechanisms have been suggested for this secondary transient period of facilitated sleep. The initial proposal (Broughton, 1975) was that the major (typically night-time) sleep period together with the afternoon nap zone reflect an endogenous brain rhythm providing a twice a day increase in sleep propensity, i.e., that sleep/wake regulation shows a circasemidian (circa, about; semi, a half; dias, a day), as well as a circadian, pattern. It was also noted in this paper that the major and minor sleep periods are about 180° apart. An endogenous origin for the 2/day sleep propensity phenomenon was later proven by its presence when subjects slept ad libitum under conditions of temporal isolation (Zulley and Campbell, 1985). It was further suggested (Broughton, 1975) both that the pattern might reflect a 12-hour rhythm for slow wave sleep (SWS) which typefies the first third of night sleep and afternoon naps of sufficient duration to permit its appearance; and that the circa-12-hour periodicity might represent a superharmonic of the fundamental 24-hour circadian rhythm linked to the solar day, a suggestion also made by Aschoff and Gerkema (1985).

Three later, and not mutually incompatible, specific proposals were that: (1) in phylogenetic evolution there is preferential selection of biological rhythms whose periodicities show simple fixed integer ratios (of 2:1 and 3:1) thereby allowing them to be consistently phase related to each other and to the solar day with consequent energy efficiences (Broughton, 1985) whereas, for example, rhythms at 24, 17, 7 and 5 hours cannot; (2) that the 2/day periodicity is not actually one of sleep but rather one of arousal with sleep being facilitated and/or simply permitted to meet the needed daily quota during the intercalated periods of lowered arousal (Broughton, 1989b); and (3) that the nap zone reflects the combined effects of increasing sleep pressure across the daytime and an opposing circadian system (Broughton, 1994; Webb, 1994). In Broughton (1994) it was further specified that the afternoon nap zone would be created by the increase in sleep propensity due to accumulating wakefulness after morning wake-up being overwhelmed and reversed by an active circadian arousal system of the type shown to exist in primates (squirrel monkeys) by Edgar et al. (1993) rather than by Process-C as defined and tested in respect to the afternoon nap zone by Borbély et al. (1989). A schematic representation of this conceptual model (Broughton, 1994) is provided in Figure 1.

Despite individual differences (Lack and Lushington, 1996), the remarkable consistency of the 2-day sleep propensity curve was revealed at the 1991 Zurich sleep/wake modeling workshop in a meta-analysis of experimental results accomplished by superimposing the sleep propensity curves around the 24 hours from four protocols with quite different demands on the sleep regulatory system(s) (Broughton and Mullington, 1992). These four experimental paradigms ranged from: hourly assessments of speed of falling asleep (sleep latency in min) across a habitual entrained night of sleep followed by daytime wakefulness (Richardson et al., 1982); similar sleep latency measures (Carskadon, personal communication) during the so-called "constant routine" introduced by Mills et al. (1978) in which controls are made for the main factors which can alter ("mask") the shape of circadian rhythms including sleep, light/dark cycles, rest/activity levels and large meals; the percentage (%) of subjects asleep across the 24-hours during temporal isolation with ad libitum sleep (Zulley and Campbell, 1985); and sleep amount (min) during attempts to sleep in 7-min periods every 20-min around the clock in Lavie's (1986) so-called 7:13 ultrashort sleep schedule. Therefore, whether sleep is fragmented into brief 7-min maximum naps around the 24 hours, follows its endogenous tendency without time constraints or is measured as sleep latency under very different conditions, the circasemidian sleep/wake propensity curve remains robust and essentially the same, at least across groups of young adult subjects.

The Borbély-Daan-Beersma 2-process model

The Borbély-Daan-Beersma 2-process model of sleep/wake regulation combines a homeostasis sleep process (Process-S) with a circadian process (Process-C) (Borbély, 1982; Daan and Beersma, 1983; Daan et al., 1984). Sleep homeostasis is said to be reflected in the quantified EEG (Q-EEG) during sleep measured as spectral power in the lower frequency bands (slow wave activity, SWA) which was initially proposed as comprising 0.5-2.5 Hz (Borbély et al., 1981) and later modified to 0.75-4.5 Hz (Brunner et al., 1990). Process-S is said to increase exponentially with accumulating wakefulness across the daytime and decrease exponentially across the night sleep period, and also to be increased by sleep deprivation and, in homeostatic fashion, be dissipated by recovery sleep. Increased power in the selected SWA frequency band is considered a measure of "sleep intensity" (Borbély, 1982). The model, as proposed, does not specify predictions about Q-EEG changes in the daytime waking EEG when Process-S is accumulating in exponential fashion; but one could reasonably expect that a similar SWA increase might occur.

Process-C in this model is believed to derive from a single light-sensitive SCN oscillator and be manifested in the daily pattern of the core body temperature with increased sleep propensity occurring at times of lowered body temperature. The shape of the circadian process is considered to be either a pure sine-wave (Borbély, 1982) or a somewhat skewed sine-wave (Daan et al., 1984). Curve-fitting of this circadian process has usually been done either by single cosinor analysis for a 24-hour periodicity or a two-cosinor analysis for combined 24-hour and 12-hour periodicities.

Some weaknesses of the current 2-process model

As currently defined, this model unfortunately does not explain a number of empirical experimental findings. Concerning Process-S, Q-EEG analyses repeated across the waking daytime have not uniformly shown a progressive increase in delta power. Both the studies of Cacot et al. (1995) and of Sterman (personal communication) show a temporary afternoon period of increased spectral power in most or all frequency bands followed by a reduction in the late afternoon-early evening period with again an increase prior to evening sleep onset. Moreover, our laboratory has not been able to confirm the expectation that sleep deprivation leads mainly or exclusively to an increase in the spectral power in the delta range of the waking EEG. In fact we have found little, if any, change in this band but rather a strong increase in both absolute and relative power in the theta-1 (4-6 Hz) and theta-2 (6-8 Hz) frequency bands (Yan et al., 1998). As the model specifies SWA changes in the sleep EEG, it is not unexpected that the large majority of the published Q-EEG studies related to the 2-process model have exclusively examined the nocturnal and diurnal sleep EEG (e.g., Knowles et al., 1986; Brunner et al., 1990; Dijk et al., 1991) rather than the awake EEG, although such studies are beginning to appear (Cajochen et al., 1998).

Concerning Process-C, the assumption that an increase in sleep propensity necessarily correlates closely with periods of lowered core body temperature is also open to serious challenge. Both the afternoon nap zone under entrained conditions (Dinges et al., 1980) and the equivalent minor sleep period in the temporal isolation studies of Campbell and Zulley (1989) occur somewhat before or near the time of maximum, rather than minimum, core body temperature; and they are very distant in circadian time from the daily temperature nadir which is, of course, associated with the major sleep period. Similarly, under the free-running conditions of temporal isolation in which napping is not permitted, sleep and circadian temperature may become dissociated in the phenomenon called "internal desynchronisation" (Aschoff et al., 1967; Wever, 1979). Therefore the periods of heightened sleep propensity and the actual timing of both naps and the major sleep period are not always in close association with periods of low core body temperature.

The model also does not predict the occasional appearance of so-called bicircadian days consisting of a 48-hour wake/sleep rhythm with about 32 hours of wakefulness followed by 16 hours of sleep under conditions of more or less prolonged temporal isolation (Chouvet et al., 1974; Honma and Honma, 1988). The phenomenon has also been reported under the conditions of constant low illumination and the unique performance demands experienced in polar treks, for example, by the two members of the unassisted (no dogs, no food drop offs) Weber-Malakov North Pole expedition (Broughton et al., 1994). The small amount of core body temperature data which exists during the bicircadian pattern suggests that it lengthens in period (Colin et al., 1968).

Another striking feature of the laboratory data in the normal entrained state, whether masked or unmasked, is the regular finding that the daily core body temperature fluctuation is far from sinusoidal in shape. This is a characteristic remarked on by Vokac and Vokac (1987) and Broughton et al. (1993). Under normal entrained conditions the lower portion of the curve (i.e., that below the mesor or daily average level), which consists of decreasing temperature to the nadir and back up to the mesor average level, is some 40-50% shorter than the period of higher core body temperature above the mesor (see also Figure 2). It has been shown (Broughton et al., 1993) that cosinor analysis and other derived single sine-wave approaches will distort this asymmetrical shape in predictable ways (including a false delay of the fitted nadir). On the other hand, curve-fitting approaches which do not assume a specific shape provide a more accurate description of the real data. This improvement includes a more precise definition of the time of the temperature nadir as an index of circadian phase and a simultaneous statistical explanation of a substantially higher percent of the variance of the data (Broughton et al., 1993; Lack and Lushington, 1996).

A polynomial least squares regression curve fit is one approach which in general gives a more accurate description of the actual data than that derived by single or double cosinor-based analyses and without the wave-shape assumptions of the latter (Broughton et al., 1993; Lack and Lushington, 1996). When applied using 10 regressions to circadian temperature data of 10 randomly chosen subjects participating in another project (Broughton et al., 1993), polynomial regression explained a range of the variance across subjects of 69-95% (mean 88.1%) compared to only 20-83% (mean 57.9%) for cosinor analysis. When used to fit the meta-analysed circasemidian sleep/wake distribution pattern reported by Broughton and Mullington (1992), a ninth-order polynomial regression explained 99.5% of the variance. The equation of this curve is provided elsewhere (Broughton, 1994). The software applied (TableCurve, Jandel Scientific, Corte Madre, CA) fits 221 equations including ones with a very large variety of approaches with varying or no assumptions, and included cosinor analysis. Polynomial regression gave the best curve fit of all 221 algorithms.

The 2-process model, as currently defined, also does not accurately predict the now well documented circasemidian pattern of sleep/wake distribution around the 24-hours. There is mention in early studies of appearance of ultradian components (Daan et al., 1984) but these are not fully defined and their shape does not approximate that of the circasemidian sleep propensity curve. An explicit attempt by Borbély et al. (1989) to predict the timing of the afternoon nap zone using the model's most commonly employed skewed-sinusoidal function for Process-C was sufficiently inaccurate that a repeat effort was made further altering the shape of the circadian process. Even when this was done, comparison of the predicted sleep/wake propensity curve with real laboratory data indicated that the model had become acceptably accurate only for the timing of the afternoon nap zone but remained highly inaccurate for sleep/wake status at all other times of day (Broughton and Mullington, 1992; Fig. 1; also Fig. 7). The main reason for this poor predictive power appeared to stem from the insistence that in the normal entrained state the circadian process is either a pure sine-wave or a somewhat skewed sine-wave.

In an attempt to define modeling parameters which would better predict the empirical laboratory results for the entrained state, the assumption can be made that the associated circadian core body temperature curve directly reflects the circadian arousal process identified by Edgar et al. (1993). This system, which has its primary site with SCN afferents in the anterior hypothalamus (Lin et al., 1996), appears to be activated by the new stimulant modafinil (Bastugi and Jouvet, 1988; Billiard et al., 1994; Broughton et al., 1997). The SCN/hypothalamic/lower brainstem/cortical GABA-ergic connections are currently being identified by Michel Jouvet and his collaborators. This recently discovered arousal system appears to be the biological basis for the daily periodicity of wakefulness and of activity peaks across the animal kingdom and is therefore of major importance.

To better clarify the shape(s) of Process-C it is instructive to make comparisons of typical circadian core body temperature patterns under normally entrained, constant routine, and free-running conditions using the meta-analysis approach previously employed to define the pattern of the circasemidian sleep/wake distribution (Broughton and Mullington, 1992).

Activate the ShortNotes by clicking on this link. Your notes will be stored in this area and automatically retrieved upon your next visit.
In order to better define the most characteristic shape of the circadian temperature fluctuation, typical results under the most common experimental conditions expressed as group means for similar age groups were chosen for meta-analysis. As the analysis was to be performed in sidereal (clock) time, the results were selected and expressed in clock time, as opposed to circadian phase. Published graphic results were digitally scanned and a file created for manipulation. Inevitably, in published studies, the time axis is not constant either for hour of onset or for page spacing. The data were therefore considered to repeat themselves across subsequent 24-hour periods, as is normally done in double and triple plotting; and then were adjusted ("wrapped") within the file around the 24-hours to begin at the same hour. Similarly, the ordinate is regularly scaled differently in different publications. Consequently, it was necessary to rescale it for direct superimposition of results across the selected studies. No significant distortions arose from such data re-alignments. This horizontal and vertical scaling process was repeated for three published representative studies of 24-hour core body temperature variations from each of the normal entrained state, constant routine, and temporal isolation (with napping not permitted and permitted).

For the normal entrained state, data came from Aschoff (1981), Minors and Waterhouse (1984) and Czeisler et al. (1986), the latter consisting of the baseline data before a constant routine was applied. Core body temperature results under the condition of a constant routine used the data of Minors and Waterhouse (1984), Monk et al. (1992) and Dijk et al. (1992). This technique removes the masking effects of sleep (by involving one night of total sleep deprivation); variations in ambient light and temperature (by using constant levels of low illumination and of environmental temperature); activity levels (by constant bed rest); and large meals (by eating small snacks every hour around the 24 hrs). The effects of temporal isolation ("free-running") with napping both not permitted (Czeisler et al., 1980; Strogatz et al., 1987) and permitted (Campbell and Zulley, 1989) were similarly analysed.

For each condition, to better characterize the circadian temperature patterns, first, the mean average of the data shown for the individual conditions was calculated and then curve-fit by appropriate algorithms similar to the approach performed in earlier studies for sleep/wake data (Broughton and Mullington, 1992). This curve-fit employed the polynomial regression software inherent in Excel (Microsoft Systems, Seattle, WA) with the criteria both of using at least the 5 regressions needed to fit a bimodal distribution (Lack and Lushington, 1996) and the lowest number sufficient to explain at least 98% of the variance in the data. The curve-fit formula for the 3-experiment averaged data under each condition is provided at the top of each figure.

In order to test the main hypothesis, i.e., that in the normal entrained state the biphasic 24-hour sleep/wake propensity curve is due to a Process-S type function being reversed by a circadian arousal process (Broughton, 1994), the newly characterized circadian temperature process was combined with a modified Process-S to attempt to predict the empirical laboratory sleep/wake data.

To use the data describing the assumed circadian arousal process (based on body temperature data) in an attempt to predict sleep/wake propensity status in sidereal time, it was assumed that the pattern selected should be that derived under the conditions of the prediction. For example, to make predictions under the usual real-life conditions which include more or less regular hours of a major sleep period, regular awakening either by environmental Zeitgebers or by an alarm clock, a typical light/dark cycle, usual activity levels and normal meals, the circadian arousal process to be employed should be that which is present under such normal entrained conditions and which uses sidereal time for which such predictions would be desired.

Concerning Process-S during night sleep, the shape defined by Borbély and Achermann (1992) was employed using their formula values. However, their values for the daytime wake period, when plotted, indicated that the portion of rapid exponential increase occurred soon after daytime wake onset rather than in the evening. Moreover, their assessment of Process-S in the daytime involves quantification of delta power during day sleep whereas, as noted above, there is evidence that theta, but not delta power increases in the waking EEG with increasing levels of sleepiness. A different pattern was therefore created with an exponentially increasing value across the day. Process-C (derived from experimental data under the entrained condition) and Process-S (derived from experimental data only for the night period and with an assumed pattern in the daytime) were then superimposed in sidereal time. As a first approximation, it was assumed that their effects on sleep propensity are opposing (Edgar et al., 1993) and without any complex interactions.

Finally, the predicted curve was superimposed upon the meta-analysis of the empirical laboratory sleep/wake data (from Broughton and Mullington, 1992; Fig. 1) and the prediction of Borbély et al. (1989).

Activate the ShortNotes by clicking on this link. Your notes will be stored in this area and automatically retrieved upon your next visit.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

The circadian core body temperature curves from the three studies of the normal entrained state in young adults are superimposed in Figure 2. The figure confirms that under normal entrained conditions the circadian distribution of core body temperature is quite asymmetrical across time as the portion below the mesor (group-average level across the 24-hour period) consistently lasts about one-third of the sidereal day and the period of increased temperature above the mesor lasts about two-thirds of the day. There is also a lesser degree of amplitude asymmetry, with the portion below the mesor being of somewhat greater amplitude than the portion above.

Results under the condition of a constant routine are shown in Figure 3. This technique removes the masking effects of sleep (by involving one night of total sleep deprivation); variations in ambient light and temperature (by using constant levels of low illumination and of environmental temperature); activity levels (by constant bed rest); and large meals (by eating small snacks every hour around the 24 hrs). Again the curves from different studies are very similar and the 3-experiment mean and its curve-fit formula are provided. Under the condition of a constant routine the circadian core body temperature can be seen to become somewhat more symmetrical both in time (across the 24-hour day) and in amplitude.

The effects of temporal isolation ("free-running") with napping both not permitted (Czeisler et al., 1980; Strogatz et al., 1987) and permitted (Campbell and Zulley, 1989) are similarly shown in Figure 4. Under these conditions, core body temperature is essentially sinusoidal in sidereal time with the experimental sleep data (not shown) indicating that the onset of the major sleep period occurs most frequently around the time of the temperature nadir. The naps, however, peak around the time of temperature maximum with a superimposed, and apparently nap-sleep evoked, small decrease. This can be seen as the transient decrease in temperature at around 1600h in the data of Campbell and Zulley (1989). Again, data from different studies are almost superimposable.

Forced desychrony involves an attempt to impose a sleep/wake schedule ("circadian day") outside of the normal entrainment range of the human circadian system in order to distinguish the patterns of the circadian temperature and the homeostatic sleep processes (Dijk et al., 1992; Dijk and Czeisler, 1995). When plotted in sidereal time (Dijk et al., 1992; Fig. 2), the shape of the circadian process is asymmetrical and quite similar to that of the normal entrained state whereas, when plotted by circadian phase, it is quite symmetrical and sinusoidal.

For the normal entrained condition a sixth order polynomial regression was sufficient to explain 98.7% of the variance (r2=0.9867) in the 3-experiment averaged data. The specific equation is shown in Figure 2. This fitted curve of the circadian core body temperature variation under entrained conditions confirmed the asymmetry in time of approximately 1:2 (8 hrs below the mesor and 16 hrs above) already evident in the raw data and the lesser amplitude asymmetry (greater for the portion below the mesor). For operational reasons it was decided to consider this function as representing a direct efferential expression of the status of the circadian arousal system under conditions of normal entrainment and with normal levels of masking.

For the constant routine a fifth-order polynomial regression explained 98.5% of the variance. The formula is given in Figure 3. It was assumed that this curve reflects the circadian arousal process under conditions involving unmasking from the effects of sleep, activity levels, the light/dark cycle, and large meals combined with the superimposed effects of one night of total sleep deprivation. The curve-fit, like the mean group data, indicated that during a constant routine the circadian arousal process had become somewhat more sinusoidal and symmetrical than in the normal entrained state.

Data from the temporal isolation conditions which either did not permit or permitted napping had 99.3% of the data explained by sixth-order polynomial regression. It is evident that under free-running conditions the core body temperature's fluctuation in sidereal time became much more sinusoidal. Again it was assumed that the resultant curve reflects the circadian arousal process under these highly particular living conditions. Figure 5 shows the three curve-fits superimposed for direct comparison.

The two processes are presented in Figure 6 with increasing sleep propensity plotted upwards. Therefore the circadian arousal process (fitted curve for empirical core body temperature data in Fig. 5) is plotted inversely to the normal procedure, whereas Process-S is plotted in the usual fashion. The superimposition clearly shows that during the normal night-sleep-period arousal (Process-C) is low with a minimum at around 0500h, arousal starts to increase at around 0500h, before the usual time of morning wake-up which averages at around 0700h and then increases across the daytime with a maximum around 2100h. Process-S, on the other hand, decreases exponentially during the night, reaches its minimum at around 0700h, and increases exponentially across the daytime. It can be seen, as predicted (Broughton, 1994), that the increasing daytime Process-S is crossed by the increasing arousal process at the usual time of the afternoon nap zone of around 1400-1500h. Later in the day a combination of high Process-S and rapidly decreasing arousal precedes sleep onset as an apparent "sleep gate" (Lavie, 1986) and leads to sleep initiation.

The predicted curve was then superimposed upon the meta-analysis of the empirical laboratory sleep/wake data (from Broughton and Mullington, 1992; Fig. 1) and the prediction of Borbély et al. (1989). It can be seen (Figure 7) that the first two curves are quite similar. By comparison, the prediction of Borbély et al. (1989) using the usual skewed-sinusoidal circadian process is a considerably weaker prediction of the empirical laboratory data.

Activate the ShortNotes by clicking on this link. Your notes will be stored in this area and automatically retrieved upon your next visit.
The Borbély-Daan-Beersma 2-process model has made a major theoretical contribution to our understanding of sleep/wake regulation. It clearly identifies and distinguishes the two main processes involved; and the model's postulate of a single circadian oscillator (the paired SCN nuclei) appears much closer to the evidence, at least for human data, than are models which propose two, three, or more, independent or coupled oscillators. The circadian component has been characterized over the past 15 years by many workers, but especially in the elegant studies of the Munich, Birmingham, Groningen and Boston groups. Yet, when applied to the consensus 24-hour distribution of wake/sleep patterns (Broughton and Mullington, 1992), the model does not accurately predict empirical laboratory data characterizing the circasemidian sleep/wake propensity curve.

Employing the fitted shape of empirical circadian temperature data, and abandoning the belief that the circadian process in the normal entrained living conditions must be a sine-wave or skewed sine-wave, produces a much closer fit to the 24-hour sleep/wake propensity pattern defined by laboratory data. This prediction is arguably as close as one can reasonably expect for data arising from different laboratories in different countries using differing technologies. The predicted 24-hour sleep/wake propensity process by this approach clearly demarcates both the nocturnal major sleep period and the afternoon nap zone, as well as the morning and especially the much stronger late afternoon/early evening periods of low sleep propensity that Lavie (1986) has called "forbidden zones for sleep" and Strogatz (1986) refers to as "wake maintenance zones". As in Broughton and Mullington (1992), the magnitude of the curve-fitted afternoon period of increased sleep propensity ("nap zone") is about 25% of that of the major sleep period. This afternoon "nap zone" represents the greatest predictable variation in daytime sleepiness/alertness levels.

Perhaps more impressive than the "nap zone" is the immediately subsequent wake maintenance zone which is the most robust component of the daytime data in the meta-analysis of Broughton and Mullington (1992) and which for most people represents the period of maximum daily sustained alertness. Even after sleep deprivation, sleep only rarely occurs during this period of heightened alertness. In short, the daily 24-hour sleep/wake propensity curve truly consists of two periods of heightened alertness with low sleep probability and two intercalated periods of reduced arousal with increased sleep propensity.

There is recent strong evidence for a circadian arousal process which is controlled by the SCN. Edgar et al. (1993) showed in the squirrel monkey that SCN destruction not only leads to loss of the circadian periodicity of activity/inactivity, and therefore of wake/sleep patterns, but also to an increase in sleep amount, thus necessitating involvement of an active arousal process. Similarly, it has also been shown that bright light stimulation in man not only phase sets the circadian pacemaker, but also has an alerting effect (Dawson and Campbell, 1991). While performing c-fos studies of the effects of the new stimulant modafinil, Lin et al. (1996) noted that the substance's major effect was mainly circumscribed to an area of the anterior hypothalamus which has inputs from the SCN. This discovery has led to identification of a new GABA-ergic wake maintaining system whose connections appear to include SCN-anterior hypothalamus, periaquaductal grey matter in the pons and medulla, and rostral projections to the forebrain (Jouvet, personal communication). This system would represent the mechanism of the daily awakening from sleep and the succeeding wake period during which alertness is further modulated by a group of other arousal systems (including the cholinergic reticular, serotonergic raphé, dopaminergic extrapyramidal, noradrenergic locus coeruleus and histaminergic posterior hypothalamic projection systems) in a hierarchy yet to be determined. The arousal generated by this light-sensitive system has superimposed upon it a number of lower amplitude circa-3-4 hour and 90-120 min ultradian rhythms of alertness level which are beyond the scope of the current modeling effort, as is explanation of the phenomenon of bicircadian days.

The results reported here indicate that under normal entrained conditions the circadian arousal system expressed in sidereal hours is neither sinusoidal or quasi-sinusoidal. Rather it is very asymmetrical in time and somewhat so in amplitude. This lack of a sinusoidal shape for core body temperature was remarked on by Vokac and Vokac (1987) and Broughton et al. (1993). The data confirm the limitations of employing single or double cosinor fits in situations where the pattern of the process does not at all approximate a sine-wave function (Broughton et al., 1993). The three meta-analyses also show the problems inherent in making the assumption that the endogenous circadian process must always have the same shape under all conditions, as for sidereal time do the forced desynchronization results of Dijk and colleagues (Dijk et al., 1992; Dijk and Czeisler, 1995). It seems self-evident that one should not use circadian functions derived under one condition to predict sleep/wake status under another, and that, if one wishes to predict in clock time, one should use clock time as opposed to circadian phase.

The use of the true efferential shape of the circadian arousal process as Process-C rather than a sine-wave or skewed sine-wave, in combination with an altered Process-S for the daytime period, has been found to predict a much closer approximation to the shape of empirical laboratory data of the circasemidian sleep/wake propensity curve as meta-analysed by Broughton and Mullington (1992) than does the normally used skewed sine-wave approach (Borbély et al. 1989). If one follows the two processes across the nychthemeron for the normal entrained state (Figure 6), a number of features become clear. The night sleep period is characterized by high but decreasing Process-S and low circadian arousal. Circadian arousal begins to increase before morning wake-up. Morning sleep inertia may be explained as the period when Process-S has not yet been reduced to its minimum by (sufficient) sleep duration and circadian arousal is still relatively low. A similar conclusion was made by Webb (1994). Such a mechanism is more parsimonious adding a third process to unaltered descriptions of Process-S and Process-C, as proposed by Folkard and Åkerstedt (1992).

Across the morning period arousal levels increase and Process-S remains low giving rise to the morning wake maintenance zone of Strogatz (1986). The afternoon nap zone can then be explained as an over-riding of increasing Process-S by the increasing and more powerful circadian arousal process. Experimental support for this mechanism is provided in the study of Krupa et al. (1998) in which phase advance and phase delay of the circadian temperature process by bright light therapy led to a parallel advance and delay of the afternoon peak of worst performance on a complex choice reaction time task. After the nap zone period, circadian arousal continues to increase strongly creating the late afternoon-early evening wake maintenance zone. This is followed in turn by increasing homeostatic pressure for sleep and declining arousal levels leading to a "sleep gate" (Lavie, 1986) for sleep onset. Sleep would then become highly probable, when a threshold of the combined effects of these two processes is reached. These findings have led to a modified conceptual model shown in Figure 8.

This conceptual model has been developed, as have all others, using absolute levels only. Yet many biological processes are sensitive to rate of change rather than to absolute levels. There is some evidence that rate of change may well play a role in probability of sleep/wake status. For example, Campbell and Broughton (1994) found that self selected sleep onset of the major sleep period followed soon after the time of the most rapid rate of change of decreasing core body temperature (i.e., the peak in the first derivative of the data).

It is beyond the breadth of this article to describe quantitative, as opposed to descriptive, processes, as was the case for the original (Borbély, 1982) paper for which quantified descriptors of Process-C and time constants for Process-S were added later (Daan et al., 1984). Such quantification and predictions of sleep/wake status following a change from the normal entrained state to other conditions will be considered in subsequent papers.

The analyses performed here have further clarified the shape of the circasemidian sleep propensity curve and a possible mechanism of the afternoon nap zone. It is of interest to consider why this phenomenon is not always expressed as napping behavior. Although its overall magnitude in the entrained condition, documented earlier in Broughton and Mullington (1992) and further analysed here, is some 25-30% of that of the major sleep period, the nap zone often passes relatively unnoticed other than as a tendency towards increased yawning, poorer concentration and further effects other than overt sleep. In fact, unless one is quite sleep deprived, which increases its intensity and may in part explain its varying intensity (Lack and Lushington, 1996), the increased sleep propensity during the nap zone generally causes little inconvenience. Despite this regular period of transitory increase in sleep propensity, one has the option not to nap; and indeed the social and other conditions for a period of afternoon sleep are often not available. Moreover, daily experience teaches one that this temporary period of increased sleepiness will soon be replaced by one of sustained arousal, typically the daily period of greatest alertness. The functional significance of the nap zone also remains uncertain. Some have proposed that its evolutionary role has been to get humans out of the mid-day sun (Webb and Dinges, 1989). Others have suggested that it increases flexibility of the timing of sleep giving more opportunities to meet the daily sleep needs (Broughton, 1989b).

It is evident that the data presented here, which are derived from studies mainly of normal young adults most of whom were males, may not be representative of other ages or of females in whom factors relating to the menstrual cycle and menopause also play a role. Moreover, it has not escaped attention that the characteristics of the circadian arousal system may vary producing different sleep/wake styles. For instance, across ontogeny the circadian arousal system appears to strengthen from the neonatal period to adolescence reducing sleep need while producing an initial polyphasic pattern, then in preschoolers a circasemidian one, in adolescence often a purely circadian distribution. This is followed by apparent weakening of the system and a reversion to a circasemidian pattern in adulthood. The relative intensity of this process between individuals may in large part be genetically determined leading to variations in sleep need. It would appear to be phase advanced in morning persons ("larks") and phase delayed in evening persons ("owls"), respectively. Its intensity may be further modulated by CNS active substances including stimulants, especially for those affecting GABA-ergic neurons. Similarly, its intensity may be reduced in neurological disorders involving increased amounts of sleep or excessive daytime sleepiness such as occur in idiopathic CNS hypersomnia, narcolepsy and certain organic hypersomnias resulting from brainstem lesions. Recent studies of patients with narcolepsy/cataplexy syndrome (Broughton et al., 1998) indicate that the appearance and timing of daytime sleep episodes are explicable by reduced intensity of the circadian arousal system which is reversed by the direct action of the new stimulant modafinil. Conversely, increased intensity of the system could explain a number of types of insomnia.

It remains uncertain whether, as has been assumed here, the nap zone reflects the effects of a single arousal process which begins before wake-up and reverses the effects of Process-S. This reversal appears to sculpt the shape of the nap zone. Such a single generator appears to be the most parsimonious explanation. The pattern could, however, also be generated by dual (coupled) arousal processes, the first one of which is expressed as the REM circadian distribution pattern shown in extended sleep studies to have its acrophase around the time of morning wake-up and to continue across the morning (Broughton et al., 1990); and the second to the circadian arousal process. In any event, the increased sleep propensity in the afternoon certainly does not reflect a nadir of core body temperature.

Obvious requirements to further improve this approach to modeling are better charactization of Process-S, the creation of a quantitative mathematical model, and the application of the model to the prediction of the effects of acute changes from the normal entrained condition to other conditions such as night shift work, jet lag and ultrashort sleep schedules.

1. Aschoff J. Circadian rhythms: interference with and dependence on work-rest schedules. In: Johnson LC, Tepas DL, Colquhoun WP, Colligan MJ, eds. Biological Rhythms, Sleep and Shift Work. Jamaica NY: SP Medical and Scientific Books, 1981, pp. 12-24.
2. Aschoff J. Naps as integral parts of the wake time within the human sleep-wake cycle. J Biol Rhythms 1994; 9: 145-55.

3. Aschoff J, Gerecke U, Wever R. Desynchronization of human circadian rhythms. Jn J Physiol 1967; 17: 250-7.

4.Aschoff J, Gerkema MP. On the diversity and uniformity of ultradian rhythms. In: Schultz H, Lavie P, eds. Ultradian Rhythms in Physiology and Behavior. Berlin: Springer-Verlag, 1985, pp. 321-34.

5. Bastugi H, Jouvet M. Successful treatment of idiopathic hypersomnia and narcolepsy with modafinil. Prog Neuropsychopharmacol Biol Psychiatry 1988; 12: 695-700.

6. Beersma DGM, Daan S, Dijk D-J. Sleep intensity and timing: a model for their circadian control. Lectures in Mathematics in the Life Sciences 1987; 19: 39-62.

7. Billiard M, Besset A, Montplaisir J. Modafinil: a double-blind multicenter study. Sleep 1994; 17 (Suppl): 107-12.

8. Blake MFJ. Time of day effects on performance in a range of tasks. Psychon Sci 1967; 9: 349-50.

9. Borbély AA. A two-process model of sleep regulation. Hum Neurobiol 1982; 1: 195-204.

10. Borbély AA, Achermann P. Concepts and models of sleep regulation: an overview. J Sleep Res 1992; 1: 63-79.

11.Borbély AA, Achermann P, Trachsel L, Tobler I. Sleep initiation and initial sleep intensity: interactions of homeostatic and circadian mechanisms. J Biol Rhythms 1989; 4: 149-60.

12. Borbély AA, Baumann F, Brandeis D, Strauch I, Lehmann D. Sleep deprivation: effect on sleep stages and EEG power density in man. Electroencephalogr Clin Neurophysiol 1981; 51: 483-93.

13. Broughton RJ. Biological variations in consciousness and psychological functions. Can Psychol Rev 1975; 16: 217-30.

14. Broughton RJ. The siesta: social or biological phenomenon. Sleep Res 1983; 12: 28.

15. Broughton RJ. Three central issues concerning ultradian rhythms. In: Schulz H, Lavie P,eds. Ultradian Rhythms in Physiology and Behavior. Berlin: Springer-Verlag, 1985, pp. 217-33.

16. Broughton RJ. Sleep attacks, naps and sleepiness in medical sleep disorders. In: Dinges D, Broughton R, eds. Sleep and Alertness: Chronobiological, Behavioral and Medical Aspects of Napping. New York: Raven, 1989a, pp. 267-98.

17. Broughton RJ. Chronobiological aspects and models of sleep and napping. In: Dinges, D, Broughton RJ, eds. Sleep and Alertness: Chronobiological, Behavioral and Medical Aspects of Napping. New York: Raven, 1989b, pp. 71-98.

18. Broughton RJ. Some important underemphasized aspects of sleep onset. In: Ogilvie R, Harsh J, eds. Sleep Onset Mechanisms: Normal and Abnormal Processes. Arlington VA: American Psychological Association, 1994, pp. 19-36.

19. Broughton RJ, Campbell S, Dunham W. Curve Fitting of chronobiological data: some issues and techniques. Sleep Res 1993; 22: 395.

20. Broughton RJ, DeKoninck J, Gagnon P, Dunham W, Stampi C. Sleep-wake rhythms and extended sleep in man. In: Montplaisir J, Godbout R, eds. Sleep and Biological Rhythms: Basic Mechanisms and Applications to Psychiatry. Oxford: Oxford Press, 1990, pp. 25-41.

21. Broughton RJ, Fleming JAE, George CFP, Hill JD, Kryger MH, Moldofsky H, Montplaisir JY, Morehouse RL, Moscovitch A, Murphy WF. Randomized, double-blind, placebo-controlled cross-over trial of Modafinil in the treatment of excessive daytime sleepiness in narcolepsy. Neurology 1997; 47: 441-51.

22. Broughton RJ, Krupa S, Boucher B, Rivers M, Mullington J. Impaired circadian waking arousal in narcolepsy-cataplexy. Sleep Res Online 1998; 1(4): 159-65. http://www.sro.org/1998/broughton/159.

23. Broughton RJ, Mullington J. Circasemidian sleep propensity and the phase-amplitude maintenance model of human sleep/wake regulation. J Sleep Res 1992; 1: 93-8.

24. Broughton RJ, Weber R, Malakhov M, Newman J, Rivers M, Dunham W, Mullington J. Weber-Malakhov north pole expedition: sleep, circadian rhythms, performance. Sleep Res 1994; 23: 480.

25. Brunner DP, Dijk D-J, Tobler I, Borbely AA. Effect of partial sleep deprivation on sleep stages and EEG power spectra: evidence for non-REM and REM sleep homeostasis. Electroencephalogr Clin Neurophysiol 1990; 75: 492-9.

26. Cacot P, Tesolin B, Sebban C. Diurnal variations of EEG power in healthy adults. Electroencephalogr Clin Neurophysiol 1995; 94: 305-12.

27. Cajochen C, Bir S, Khalsa S, Czeisler CA, Dijk D-J. Time course of EEG power density during wakefulness and subjective sleepiness during a 24h constant routine. Sleep 1998; 21 (suppl): 243.

28. Campbell S, Broughton R. Rapid decline in the body temperature prior to sleep: fluffing the physiological pillow? Chronobiol Int 1994; 91: 126-31.

29. Campbell S, Zulley J. Evidence for circadian influence on human slow wave sleep during daytime sleep episodes. Psychophysiology 1989; 26: 580-5.

30. Carskadon M, Dement W. Multiple sleep latency tests during the constant routine. Sleep 1992; 15: 396-9.

31. Chouvet G, Mouret J, Coindet J, Siffre M, Jouvet M. Periodicité bicircadienne du cycle veille-sommeil dans des conditions hors du temps: étude polygraphique. Electroencephalogr Clin Neurophysiol 1974; 37: 367-80.

32. Colin J, Timball J, Boutelier C, Houndas Y, Siffre M. Rhythm of the rectal temperature during a 6 month free-running experiment. J Appl Physiol 1968; 25: 170-6.

33. Czeisler CA, Allan JS, Strogatz SH, Ronda JM, Sanchez R, Rios D, Freitag O, Richardson GS, Kronauer RE. Bright light resets the human circadian pacemaker independent of the timing of the sleep-wake cycle. Science 1986; 233(4764): 667-71.

34. Czeisler CA, Weitzman ED, Moore-Ede MC, Zimmerman JC, Kronauer R. Human sleep: its duration and organization depend on its circadian phase. Science 1980; 210: 1264-7.

35. Daan S, Beersma DGM. Circadian gating of human sleep and wakefulness. In: Moore-Ede MC, Czeisler CS, eds. Mathematical Models of the Circadian Sleep-Wake Cycle. New York: Raven, 1983, pp. 129-58.

36. Daan S, Beersma DGM, Borbély AA. The timing of human sleep: recovery process gated by a circadian pacemaker. Am J Physiol 1984; 246: R161-78.

37. Dawson D, Campbell SS. Timed exposure to bright lights improves sleep and alertness during simulated night shifts. Sleep, 1991; 14: 511-6.

38. Dijk D-J, Beersma DGM, Daan S. EEG Power density during nap sleep: reflection of an hourglass measuring the duration of prior wakefulness. J Biol Rhythms 1987; 2: 207-19.

39. Dijk D-J, Cajochen C, Tobler I, Borbély AA. Sleep extension in humans: sleep stages, EEG power spectra and body temperature. Sleep 1991; 14: 294-306.

40. Dijk D-J, Czeisler CA. Contribution of the circadian pacemaker and the sleep homeostat to sleep propensity, sleep structure and electroencephalographic slow waves and sleep spindle activity in humans. J Neurosci 1995; 15: 3526-38.

41. Dijk D-J, Duffy JF, Czeisler CA. Circadian and sleep/wake dependent aspects of subjective alertness and cognition. J Sleep Res 1992; 1: 112-7.

42. Dinges DF, Orne MT, Orne EC, Evans FJ. Voluntary self-control of sleep to facilitate quasi-continuous performance. US Army Medical Research and Development Command Report No. 80 (National Technical Information Service No. AD-A102264), Fort Detruck, Frederick, MD: US Army, 1980.

43. Edgar DM, Dement WC, Fuller CA. Effect of SCN lesions on sleep in squirrel monkeys. Neuroscience 1993; 13: 1065-79.

44. Folkard S, Åkerstedt T. A three-process model of the regulation of alertness-sleepiness. In: Broughton RJ, Ogilvie R., eds. Sleep, Arousal and Performance. Boston: Birkhauser, 1992, pp. 11-26.

45. Gagnon P, De Koninck J. Reappearance of EEG slow waves in extended sleep. Electroencephalogr Clin Neurophysiol 1984; 58: 155-7.

46. Gagnon P, De Koninck J, Broughton R. Reappearance of electroencephalogram slow waves in extended sleep with delayed bedtime. Sleep 1985; 8: 118-28.

47. Honma K, Honma S. Circabidian rhythm: its appearance and disappearance in association with bright light pulse. Experientia 1988; 44: 981-3.

48. Knowles JB, MacLean AW, Salem L, Vetere C, Coulter M. Slow-wave sleep in daytime and nocturnal sleep: an estimate in the time course of "process-S". J Biol Rhythms 1986; 1: 303-8.

49. Krupa S, Boucher B, Broughton R. Does the circadian clock control the afternoon "nap zone". Sleep 1998; 21 (suppl): 204.

50. Lack L, Lushington K. The rhythms of human sleep propensity and core body temperature. J Sleep Res 1996; 5: 1-11.

51. Lavie P. Ultrashort sleep-waking schedule: III. "Gates" and "forbidden zones" for sleep. Electroencephalograph Clin Neurophysiol 1986; 63: 414-25.

52. Lin JS, Hou Y, Jouvet M. Potential brain neuronal targets for amphetamine, methylphenidate and modafinil induced wakefulness, evidenced by c-fos immunochemistry in the cat. Proc Nat Acad Sci USA 1996; 93(24): 14128-33.

53. Mills JN, Minors DS, Waterhouse JM. Exogenous and endogenous influences on rhythms after sudden time shifts. Ergonomics 1978; 21: 755-61.

54. Minors DS, Waterhouse JM. The use of constant routine in unmasking the endogenous component of human circadian rhythms. Chronobiol Int 1984; 1(3): 205-16.

55. Mitler MM, Carskadon MA, Czeisler C, Dement WC, Dinges DF, Graeber RC. Catastrophes, sleep and public policy: consensus report. Sleep 1988; 11: 100-9.

56. Monk TM, Buysse DJ, Reynolds III CF, Jarrett DB, Kupfer DJ. Rhythmic vs homeostatic influences on mood, activation, and performance in young and old men. J Gerontol Psychol Sci 1992; 47(4): 221-7.

57. Quadens O. L´ Architecture du Rêve du Cerveau à la Culture. Louvain: Éditions Peeters, 1990.

58. Richardson GS, Carskadon M, Flagg W, van den Hoed J, Dement WC, Mitler MM. Excessive daytime sleepiness in man: multiple latency measurement in narcoleptic and control subjects. Electroencephalogr Clin Neurophysiol 1978; 45(5): 621-7.

59. Richardson GS, Carskadon MA, Orav EJ, Dement WC. Circadian variation of sleep tendency in elderly and young adult subjects. Sleep 1982; 5: S82-94.

60. Strogatz SH. The Mathematical Structure of the Human Sleep-Wake Cycle. Berlin: Springer-Verlag, 1986.

61. Strogatz SH, Kronauer RE, Czeisler CA. Circadian pacemaker interferes with sleep onset at specific times each day: role in insomnia. Am J Physiol 1987; 253: R172-8.

62. Vokac M, Vokac Z. Serial section analysis of circadian rhythms by best-fitting periods. Adv Chronobiol 1987; part B: 549-58.

63. Webb WB. Enhanced slow sleep in extended sleep. Electroencephalogr Clin Neurophysiol 1986; 64: 27-30.

64. Webb WB. An objective behavioural model of sleep. Sleep 1988; 11: 48-96.

65. Webb WB. Prediction of sleep onset. In: Ogilvie R, Harsh J, eds. Sleep Onset: Normal and Abnormal Processes. Arlington VA: American Psychological Association 1994, pp. 53-72.

66. Webb WB, Dinges D. Cultural perspectives on napping and the siesta. In: Dinges DF, Broughton RJ, eds. Sleep and Alertness: Chronobiological, Behavioural and Medical Aspects of Napping. New York: Raven, 1989, pp. 247-65.

67. Wever R. The Circadian System of Man. Berlin: Springer-Verlag 1979.

68. Yan H, Rivers M, Boucher B, Broughton R. Quantitative EEG and excessive daytime sleepiness. Sleep 1998; 12(suppl): 207.

69. Zulley J, Campbell S. Napping behavior during "spontaneous internal desychronization". Human Neurobiol 1985; 4: 123-6.

The author thanks the Medical Research Council of Canada for a research grant and Career Investigator Award (1968-97) covering the period of this research. Janet Mullington read an early version of the paper and suggested the evening sleep threshold. Francesca Cañellas read the manuscript at several stages and made many helpful suggestions. The assistance of Brigitte Boucher for graphics and curve-fitting was invaluable.

Original address of this text :

http://www.sro.org/bin/article.dll?Paper&1480&0&0

Please copy this address to the address bar of your internet browser and press the "enter" key.
(We prefer not to put actual links because often page locations change and then our log files get cluttered with error messages
- if the address does not work try to find it from the homepage of the website in question).