Professional Documents
Culture Documents
Summary: Previous field and laboratory studies have revealed that there is a
diurnal variation in subjective sleepiness that is different in form to that of
objective sleepiness and many measures of performance efficiency. The worst
subjective sleepiness occurs at the trough in the circadian temperature rhythm,
the least subjective sleepiness about 7 h before the peak in temperature. A
series of forced de synchronization experiments, in which the endogenous cir-
cadian oscillator controlling the temperature rhythm ran at a different period to
the sleep/wake cycle, revealed that these findings can be explained by postu-
lating subjective sleepiness to be under the control of both factors, with min-
imum sleepiness occurring at the peak in temperature in terms of the tempera-
ture cycle and about 6 h after waking in terms of the sleep/wake cycle. Key
Words: Alertness-Circadian-Human-Sleepiness-Subjective ratings.
INTRODUCTION
Background
As there are dramatic changes over the day in an individual's physiological make-up
(1), it is not surprising that psychological variables also show a diurnal variation, even
in individuals who are well-slept and living on a "normal" routine. Thus, there are
reliable changes over the day in mood (2), subjective (3-5) and objective sleepiness
(6,7), and the ability to perform various tasks (8). Although there are some individuals
who are extreme in their preference for either early morning ("larks" or M-types) or
late evening ("owls" or E-types) activity (9), a sizable majority (about 80%) fall into
neither extreme category, and for them generalizations can be made about the way in
which these variables fluctuate over the day (10).
The diversity of the many physiological circadian rhythms renders it inappropriate to
speak of a single composite "physiological rhythm." Rather there are many different
physiological circadian rhythms, showing different peak timings, levels of lability, and
rates of phase adjustment to a change in routine. A similar argument holds for the
"psychological" variables. Thus, while performance speed at simple repetitive (11,12)
343
344 T. H.MONK
and serial search (13) tasks peaks with temperature in the evening, cognitive perfor-
mance speed peaks in the late morning (14) and short-term memory in the early- to
mid-morning (15,16). These differences in phase are also associated with different rates
of phase adjustment to a change in routine (17) and with differences in free-running
period (18,19) and underlying oscillatory control (20).
One must therefore be careful in making generalizations about the diurnal variation
in global variables like "mood," "ability to perform," or "sleepiness," as the form
that the variation will take will depend on the way in which the variable is measured.
from the sleep time of the individual. Alertness ratings are thus usually restricted to the
"normal" hours (7:00 a.m. to midnight) of wakefulness. However, either by keeping
subjects awake continuously, or by studying nightworkers, one can obtain sensible
readings through the night hours.
Froberg (29) used the former technique keeping subjects continuously awake in a
72-h temporal isolation experiment, with constant conditions, uniform food intake, and
no knowledge of clock time. Although subjective alertness showed the expected de-
creasing trend over the vigil, there was also a clear circadian rhythm in both subjective
'0
H 31.'
...
,," t
, I
30
," II
I I
, 37.0
,
I TEMP
U.t
2'
,/\\,
I
I
I
I
r- J ,, I
I I
20 ,
I \
I
I
n·'
I , I
I
\ ,,
I
I
,,
I' 16.7
I
,,,, , J ~
10 ~
3£.
~
54
"OU~S 1010 SrUDY
FIG. 1. Circadian rhythms in body temperature and self-rated alertness from 15 subjects experiencing 72 h of
sleep deprivation in temporal isolation (29).
v r=====~~====-----------------------------------'9
•
3e·g A1.£ RT N£ S5
UoI ~
a: Z
::I ~
~ Je·8 «
«
a: a:
UoI 8 V1
Q. V1
1: UoI
UoI Z
00 04 08 12 115 20 00 04 08 12 1e 20
TIME OF DAY
FIG. 2. Circadian rhythms in oral temperature (e--e) and subjective alertness (0- - - - -0). Two cycles
of the 12 points defining each rhythm have been plotted (30).
nearly coinciding with the temperature peak, the other occurring considerably earlier
(at about mid-day). The coincidence of the two troughs again confirms the importance
of the temperature oscillator as a determinant of alertness, but the mid-day peak in
alertness suggests that in this situation, i.e., subjects who were not time isolated and
were sleeping normally (at night) while on daywork, some other factor was coming into
play.
This difference in the time of peak between the two measures was confirmed in our
two studies on normal diurnal subjects (36 students, 38 members of the general public)
(5). Figure 3 shows the results of the second of those studies.
Clearly, the most parsimonious explanation for the difference is that although the
temperature oscillator is certainly one determinant of subjective sleepiness, the pro-
cess that controls the sleep/wake cycle [be it like an oscillator (21,32) or capacitor (22)
in action] would also seem to be having an effect. This is not implausible, in view of the
obvious link between sleepiness and going to bed! Indeed, one might conceive of the
subjective sleepiness rhythm as the means by which the physiological circadian system
makes its wishes known to the conscious mind, making the required circadian behavior
become the desired circadian behavior.
If one is to tease apart the influence of the "temperature oscillator" and "sleep/wake
cycle mechanism" on subjective sleepiness, it is necessary to separate them in some
way. We have done this using a forced desynchronization technique in which the tem-
perature oscillator is made to run at a different period to that of the sleep/wake cycle.
METHODS
The analyses reported here comprise one small aspect of a multifaceted series of
studies carried out at Montefiore Medical Center and Cornell University Medical Col-
lege by J. E. Fookson, M. L. Moline, C. P. Pollak, and the late E. D. Weitzman, in
collaboration with the present author. Other aspects of these studies have been and will
be reported elsewhere (33,34).
Resources and environment
Each subject lived alone in a special time-isolated apartment that allowed no knowl-
edge of clock time or external time cues. The facility is described in detail in a separate
~~---------------------------------,
36.8
~
c:
"
~
~
:l5 36.7 a
"-
~ eo !
..
:l a
..
i! i
..
<i
>
.'"
!
~
z; 36.8 ~
.," 50
38.5
40
FIG. 3. Time of day effects in subjective alertness and oral temperature (± 1 SE) for 38 subjects studied over
8 days (5).
publication (35). Social interaction with the technicians (who brought the subjects food
and supervised data collection) was allowed, but the technicians worked a totally
random shift schedule and were specially trained not to give any time cues.
Throughout the experiment, subjects were told when to go to bed, get up, and take
their meals. Lights were extinguished at bedtime and did not come on again until the
prescribed wake-time, at which time the subject was aroused from sleep if necessary
and required to get out of bed. This time will be referred to henceforth as "wake-
time," although subjects sometimes awoke spontaneously before it (but were not al-
lowed to get out of bed). Sleep at all other times was forbidden, and compliance was
monitored using closed-circuit TV.
Measures taken
A broad range of psychological and physiological measures were taken. This report
will concentrate on measures of subjective sleepiness that were assessed using a com-
puterized VAS technique, comprising a question "How sleepy do you feel?" with a
IO-cm line below it with the label "very little" at the left end and "very much" at the
right. Subjects moved a cursor to a position on the line that corresponded to their level
of sleepiness at that moment, considering the whole line to represent their own normal
range. The measure taken was the distance from the left end in (0-36) arbitrary units.
The VAS technique has been used widely in our own laboratory, involving more than
2,000 subject-sessions, as well as in many previous studies in which other psycholog-
ical measures were evaluated (27,36,37). Assessments were made at waketime and at
bedtime, and every time the subject used the bathroom or requested something to eat.
This corresponded to about 6 times per day.
Experimental procedure
The four experiments to be described each lasted several months and were con-
cerned with several different manipulations of day length and their effect on sleep and
the circadian system. These manipulations were accomplished by either stretching or
shrinking an individual's whole daily routine for several weeks, so that the cycle re-
tained its integrity, but at a period length that was different to 24 h. This analysis will be
restricted to day lengths which were engineered to be beyond the range of entrainment
of the temperature oscillator. Under those conditions, a "desynchronization" was in-
duced, with the sleep/wake cycle being driven at one period length (e.g., 26 h), while
the temperature oscillator ran independently at a different period, corresponding to its
"natural" or "free-running" period (typically close to 25 h). That "natural" period
was assessed using a periodogram technique applied to the minute-by-minute around-
the-clock temperature readings (38). Table 1 presents details of the lengths of these
sections, the period lengths of sleep/wake cycle and temperature oscillator, and the age
and sex of the subjects.
Statistical techniques
As subjective sleepiness could only be sampled during wakefulness, sampling was
irregularly spaced, and standard periodogram, autocorrelation, or Fourier techniques
could not be applied. Instead, a sinusoid fitting technique (39) was used. For each
subject, this technique was applied separately at the period of the (imposed) sleep/
wake cyCle and at the "natural" period of the temperature oscillator. At each period, a
least-squares cosine curve-fitting procedure was used, involving the solution of three
simultaneous equations in order to derive estimates of acrophase (time of cosine peak),
Temperature oscillator cycle length was determined by periodogram analysis of minute-by-minute temper-
ature readings.
" More sessions were given, data were lost due to equipment malfunction.
mesor (mean level of curve), and rhythm amplitude (peak level minus mesor). These
estimates comprised the parameters of the cosine curve that best fit the observed data
at that particular period. Measures of goodness of fit were also obtained, including an
F ratio to assess statistical reliability. All of the fits reported here were significant at
the 5% level or better.
This particular technique has been used extensively in this (19,34) and other (40)
laboratories and has been shown to give period (frequency) information that can be as
reliable as that obtained by applying standard periodogram techniques to around-the-
RESULTS
Primary analysis
Unlike the strictly 24-h environment of the world outside, "day lengths" in temporal
isolation studies can be arbitrarily defined according to the dictates (imposed day
lengths) of the experimenter and/or the circadian behavior of the subject. In analyzing
this study, we defined two "day lengths." "Temperature oscillator days" were defined
by the temperature oscillator period (tau) obtained by periodogram analysis and listed
in Table 1. As these "days" were out of synchrony with both clock time and the im-
posed sleep/wake cycle, a special time marker was required for them. For this marker
we used the acrophase (time of sinusoid peak) of the sinusoid fitted to the rectal tem-
perature rhythm at the temperature oscillator tau. Thus, the time measure used the
distance in time from peak temperature, with positive numbers indicating later times,
negative numbers earlier times, and zero a coincidence with peak temperature. The
second "day length" we defined was simply that of the "sleep/wake cycle days" en-
forced during the experiment. Here, the measure taken was the hours passed since the
subject was told to get up ("wake-time").
Sinusoids were then fitted to the time series defining the subjective sleepiness
rhythm separately for "temperature oscillator days" and for "sleep/wake cycle days."
The time of least subjective sleepiness (180 0 after the acrophase) was then obtained for
each and the results summarized in Table 2.
The fact that a coherent pattern of results was obtained both in terms of temperature
oscillator days (where least sleepiness coincided with peak temperature) and in terms
of sleep/wake cycle days (where least sleepiness occurred about 6 h after getting out of
bed), bespeaks the importance of both mechanisms as determinants of subjective
sleepiness.
Secondary analyses
Since we have shown earlier (Fig. 2) that alertness rhythms are not perfectly sinu-
soidal, it is clearly desirable to verify the results summarized in Table 2 without re-
course to any assumptions about the shape of the data. (For technical reasons, these
analyses could not be performed on subject PB01.)
The coincidence of least sleepiness with peak temperature (Table 2, upper panel) was
tested by simply pairing each sleepiness rating with its associated rectal temperature
and calculating a correlation coefficient for each subject. The expected significant neg-
ative correlations (PB02: r = -0.395, df = 332, p < 0.001; PB03: r = -0.511, df =
228, P < 0.001; PB04: r = -0.320, df = 121, P < 0.01) confirmed that rises in tempera-
ture were indeed associated with a lessening of subjective sleepiness.
The occurrence of least sleepiness about 6 h after wake time (Table 2, lower panel)
was tested by indexing each subjective sleepiness rating by the number of hours since
wake-time that it occurred and averaging to give a "time of day function" for each
subject. The three time of day functions are plotted in Fig. 4. A good agreement was
found with the times of least sleepiness obtained by the sinusoid-fitting analysis.
DISCUSSION
These laboratory results very nicely integrate the findings from previous studies and
resolve the paradox regarding the apparent parallelism between alertness and tempera-
ture, but lack of coincidence in their peaks. Thus, a minimum in sleepiness 6 h after
wake-time corresponds, for most people's routines, to about 1:00 p.m., approximating
the time of peak subjective alertness found by Folkard (4) in his review and by the
present author in his field and laboratory experiments (2,5,30). The coincidence be-
tween temperature peak and minimum sleepiness at the temperature oscillator period
is in agreement with the parallelism between temperature and arousal presumed to
underlie the circadian variation in human performance efficiency (28), and also ex-
plains why Froberg (29) (see 'Fig. 1) obtained a perfect parallelism between subjective
sleepiness and body temperature when he suspended the sleep/wake cycle for 72 h.
The paradox that remains is the relation between subjective and objective measures
of sleepiness. Thus, it seems peculiar that subjects rate themselves as at their least
sleepy a few short hours before their peak objective sleepiness as measured by the
PB02
P803
PS04
MSLT. One possible explanation may be that objective sleepiness is related not to the
absolute level of subjective sleepiness, but to its rate of change. Thus, the rate of rise in
subjective sleepiness may be the prime determinant of an individual's ability to fall
asleep.
However, the results do make sense from an intuitive point of view. As the duration
of a sleep episode is dependent on body temperature at sleep onset (42), it is clearly an
advantage to have the subjective sleepiness rhythm, which in a perfect world would
naturally precipitate the bedtime decision, also be under the influence of the tempera-
ture oscillator. However, because of the marked stability of that oscillator, subjective
sleepiness rhythms that were totally under its control would be inflexible in coping with
phase shifts and missed nights of sleep. The mixture of control from both temperature
oscillator and sleep/wake cycle mechanism ensures that the subjective sleepiness
REFERENCES
1. Aschoff J, ed. Handbook of Behavioral Neurobiology, vol 4. New York: Plenum, 1981.
2. Monk TH, Fookson JE, Moline ML, Pollak CPo Diurnal variation in mood and performance in a time-
isolated environment. Chronobiolint 1985;2:185-293.
3. Thayer RE. Toward a psychological theory of multidimensional activation (arousal). Motivation & Emo-
tion 1978;2: 1-34.
4. Folkard S. Diurnal variational. In: Hockey GRJ, ed. Stress and fatigue in human performance. New
York: Wiley, 1983:245-72.
5. Monk TH, Leng VC, Folkard S, Weitzman ED. Circadian rhythms in subjective alertness and core body
temperature. Chronobiologia 1983; I 0:49-55.
6. Carskadon M, Dement W. Sleep tendency: an objective measure of sleep loss. Sleep Res 1977;6:200.
7. Richardson GS, Carskadon MA, Orav EJ, Dement WC. Circadian variation of sleep tendency in elderly
and young adult subjects. Sleep 1982;5:582-94.
8. Colquhoun WP. Rhythms in performance. In: Aschoff J, ed. Handbook of behavioral neurobiology, vol
4. New York: Plenum, 1981:333-48.
9. Horne JA, Ostberg O. A self-assessment questionnaire to determine morningness-eveningness in human
circadian rhythms. Int] Chronobiol 1976;4:97-110.
10. Folkard S, Monk TH. Circadian performance rhythms. In: Folkard S, Monk TH, eds. Hours of work-
temporal factors in work scheduling. New York: John Wiley & Sons, 1985:37-52.
11. Blake MJF. Time of day effects on performance in a range of tasks. Psychon Sci 1967;9:349-50.
12. Colquhoun WP. Circadian variations in mental efficiency. In: Colquhoun WP, ed. Biological rhythms and
human performance. London: Academic Press, 1971:39-107.
13. Monk TH. Temporal effects in visual search. In: Clare IN, Sinclair MA, eds. Search and the human
observer. London: Taylor & Francis, 1979:30-9.
14. Folkard S. Diurnal variation in logical reasoning. Br] Psychol 1975;66: 1-8.
15. Laird DA. Relative performance of college students as conditioned by time of day and day of week. ]
Exp Psychol 1925;8:50-63.
16. Folkard S, Monk TH. Circadian rhythms in human memory. Br] PsychoI1980;71:295-307.
17. Folkard S, Monk TH. Shiftwork and performance. Hum Factors 1979;21:483-92.
18. Folkard S, Wever RA, Wildgruber CM. Multioscillatory control of circadian rhythms in human perfor-
mance. Nature 1983;305:223-6.
19. Monk TH, Weitzman ED, Fookson JE, Moline ML. Circadian rhythms in human performance efficiency
under free-running conditions. Chronobiologia 1984;11:343-54.
20. Monk TH, Weitzman ED, Fookson JE, Moline ML, Kronauer RE, Gander PH. Task variables determine
which biological clock controls circadian rhythms in human performance. Nature 1983;304:543-5.
21. Wever R. The circadian multi-oscillator system of man. Int] ChronobioI1975;3: 19-55.
22. Borbely AA. A two-process model of sleep regulation. Hum Neurobiol 1982;1: 195-204.
23. Hoddes E, Zarcone V, Smythe H, Phillips R, Dement W. Quantification and sleepiness: a new approach.
Psychophysiology 1973; 10:431-6.
24. Wever R. The circadian system of man: results of experiments under temporal isolation. New York:
Springer-Verlag, 1979.
25. Folkard S, Monk TH, Lobban MC. Short and long-term adjustment of circadian rhythms in "perma-
nent" night nurses. Ergonomics 1978;21:785-99.
26. Weitzman ED, Czeisler CA, Zimmerman JC, Ronda JM, Knauer RS. Chronobiological disorders: ana-
lytic and therapeutic techniques. In: Guilleminault C, ed. Disorders of sleeping and waking: indications
and techniques. California: Addison-Wesley, 1982:297-329.
27. Folstein MF, Luria R. Reliability, validity and clinical application of the visual analogue mood scale.
Psychol Med 1973;3:479-86.
. -
Sleep, Vol. 10, No.4, 1987
SUBJECTIVE SLEEPINESS 353
28. Hockey GRI, Colquhoun WP. Diurnal variation in human performance: a review. In: Colquhoun WP, ed.
Aspects of human efficiency-diurnal rhythm and loss of sleep. London: English Universities Press,
1972:39-107.
29. Froberg JE. Twenty-four hour patterns in human performance, subjective and physiological variables
and differences between morning and evening active subjects. Bioi PsychoI1977;5: 119-34.
30. Monk TH, Embrey DE. A field study of circadian rhythms in actual and interpolated task performance.
In: Reinberg A, Vieux N, Andlauer P, eds. Night and shift work: biological and social aspects. Oxford:
Pergamon Press, 1981:473-80.
31. Monk TH. Advantages and disadvantages of rapidly rotating shift schedules-a circadian viewpoint.
Hum Factors 1986;28:553-7.