Sleep, 20(6):399-40 I
© 1997 American Sleep Disorders Association and Sleep Research Society
Speed of Mental Processing In the Middle of the Night
Timothy H. Monk and Julie Carrier
Sleep and Chronobiology Center, Western Psychiatric Institute and Clinic, University of Pittsburgh
School of Medicine, Pittsburgh, Pennsylvania, US.A,
Summary: This study aimed to determine whether human mental processing actually slows down during the night
hours, separately from the previously documented microsleeps, lapses in attention, and general slowing of motor
responses. Eighteen healthy young adults were studied during 36 hours of constant wakeful bedrest. Every 2 hours,
they performed a logical reasoning task. Items phrased in the negative voice took reliably longer to respond to than
items phrased in the positive voice, indicating the need for more mental processing in those items. By subtracting
"negative" from "positive" reaction times at each time of day, we were able to plot a circadian rhythm in the time
taken for this extra mental processing to be done separately from microsleeps, psychomotor slowing, and inattention.
The extra mental processing took longer at night and on the day following sleep loss than it did during the day
before the sleep loss, suggesting that human mental processing slows down during the night under sleep deprivation.
Key Words: Human-Sleep-Circadian rhythms-Performance .
..
1'.
There are circadian rhythms in human performance
much like those in physiology, with peaks at some
times of day, troughs at others, and a gradual transition
between the two (1). While different tasks can show
different time-of-day effects, most tasks show a decrement during the night, which is borne out when the
actual "on job" performance of shift workers is studied
over 24 hours (2,3). At least some of this decrement
can be attributed directly to the pressure to fall asleep,
since this has been well documented as a means by
which night-worker and night-driver performance and
safety are impaired (4,5). Laboratory studies have indicated that poor performance at night can be explained
by lapses in attention and/or "response blocks" (unduly
long reaction times) that may, in some cases, be associated with microsleeps and also be explained by psychomotor slowing (6). The question remains as to
whether all nocturnal performance speed decrements
can be so explained. Our aim was to conduct a relatively pure study of the circadian rhythm in the speed
of mental processing in order to determine whether people actually think more slowly at night.
METHODS
As part of a larger study of circadian rhythms in human performance (7), 18 healthy young adults (19-28
Accepted for publication March 1997.
Address correspondence and reprint requests to Dr. Timothy H.
Monk, WPIC Room E1123, 3811 O'Hara Street, Pittsburgh, PA
15213, U.S.A.
years; 8 females, 10 males) each experienced a protocol
involving 36 hours of constant wakeful bedrest (starting
at 0900 hours), with no knowledge of clock time and
meals replaced by hourly nutritional supplements. Outside of the experiment, the subjects lived a normal diurnal routine (mean habitual bedtime 2354 hours, waketime 0737 hours). On each odd-numbered hour, after
ingesting the food supplement and being offered the opportunity to void, a computer was used to administer a
battery of mood and performance tests that included a
brief (approximately 3-minute) modified version of the
Baddeley reasoning test (8). The test had 32 items presented in a randomly shuffled order; the computer recorded the speed and accuracy of the response to each
item. About 28 practice sessions (896 trials) were given
before the sessions reported here. Subjects were required
to respond with handheld buttons marked "true" and
"false" to indicate whether a sentence was a true or a
false description of the letter pair immediately following
it. Sixteen of the sentences (eight true, eight false) were
framed in the positive voice (e.g. C IS BEFORE MCM; M FOLLOWS C-MC), 16 (eight true, eight false)
in the negative voice (e.g. C DOES NOT FOLLOW
M-CM; M IS NOT BEFORE C-MC). Trials were
then divided into "negative-voice" and "positive-voice"
sentences and the mean latency for each calculated per
subject-session.
RESULTS
Negative-voice sentences took longer to respond to
than positive-voice ones (3.93 vs. 2.55 seconds; see
399
T. H. MONK AND J. CARRIER
400
6
~
-0--
5
N",,,iv.
Positive
~
UQ)
~~~
HwWt
Q)
E 4
i=
c
.2
t5 3
ro
on simple-difference scores (negative voice minus positive voice) revealed that the three sections (day 1,
night, day 2) differed reliably [F(2,34) = 4.46, P <
0.02] with significant pairwise comparisons (paired t
tests) between day one and night (p < 0.02) and day
one and day two (p < 0.02).
jl~
DISCUSSION
Q)
a:
2
«
0«
2.2
Q)
UQ)
Ern
l=~
c>
0'';:::
'.t='
U 0
2.0
1.8
'w
roo. 1.6
~~
c c 1.4
'w'E
uQ)
~£; 1.2
Q)eO
~g'
o~
1.0
0.0
f
'1
'
9:00 13:0017:0021 :00 1 :00 5:00 9:00 13:0017:0021:00
Time of Day
FIG. 1. Top panel: Mean reaction time (:!: I SEM) for the negativevoice items and for the positive-voice items of a modified Baddeley
reasoning test, plotted as a function of time of day during 36 hours
of constant wakeful bedrest. Slower performance is represented by
a rise on the ordinate. Lower panel: Mean difference in reaction
time (:!: I SEM) between negative- and positive-voice items, plotted
as a function of time of day. Each point represents the mean of 18
difference scores (mean of 16 negative-voice items minus mean of
16 positive-voice items), one from each subject.
Fig. 1, upper panel). This implied a greater processing
requirement, part of which can be explained by the
negative-voice sentences having an average of 23%
more syllables. By subtracting each subject's mean response latency for positive-voice sentences from that
of negative-voice sentences, we were able to plot a
circadian rhythm in the time taken for this extra processing requirement (Fig. 1, lower panel). There was
little change over the first day (0900-1900 hours) but
with an increase (slowing) during the night (01000700 hours) [day, 1:1.14 seconds::±:: 0.12 (SEM); night,
1.58 seconds::±:: 0.17]. After the lost night of sleep, the
second day's daytime (0900-1900 hours) readings had
longer latencies than those of the corresponding test
sessions before the sleep loss (day 1, 1.14 seconds : ±:
0.12; day 2, 1.52; SEM = 0.15). Analysis of variance
(ANOVA) (corrected by Greenhouse-Geisser epsilon)
Sleep. Vo!' 20. No.6, 1997
To date, the only suggestions in the literature for a
"cognitive slowing" during the night hours have been
from studies showing an overall increase of reaction time
on self-paced tasks like mental arithmetic, logical reasoning, and tracking (6,9). However, in these studies it
is possible that the increase in reaction time was due
simply to a change in motivation, drowsiness, inattention, and/or psychomotor slowing. In the present study,
it is implausible that these factors were modified by the
subject according to whether the sentence was in positive
or negative voice. Thus, the data suggest that the extra
processing requirement of having to parse a negativevoice sentence rather than a positive-voice one took longer to accomplish at night than during the day, even
when motivation, drowsiness, and psychomotor response
are controlled. This may either have been due to the
brain taking longer to process each step (but doing the
same number of steps) or to the adoption of a more timeconsuming information processing strategy (i.e. one with
more steps) at night. The latter explanation is not implausible because strategy changes (e.g. to one making
more use of subvocal rehearsal) have been used to explain other circadian performance rhythms, and it is quite
possible that people approach a given task quite differently at night than during the day. The change in approach may result from the need to fight off sleep or be
an adaptive response to the altered "operating characteristics" of the brain at night. Other investigators have
described how extremely sleepy individuals can change
information-processing strategy so as to maintain accuracy, even if this results in excessively slow response
times (10). Interestingly, the present accuracies were uniformly high (mean, 96.4%; SD, 6.7%) with no reliable
differences between day one, night, and day two [F(2,34)
= 2.15, ns].
Our finding of a slowing in speed of mental processing during the day after the lost night of sleep is in
agreement with an earlier study (10) that used a similar
technique to demonstrate daytime decrements in information-processing speed following more extreme (two
nights) sleep loss. It suggests that in addition to rhythmic
determinants of information-processing speed, there are
also homeostatic ones relating to the time spent awake.
This is in line with our earlier study of circadian performance rhythms in logical reasoning (11). It should also
be recognized, of course, that homeostatic factors may
MENTAL PROCESSING AT NIGHT
have accounted for some of the day versus night differences, since subjects had been awake for longer when
the night performance tests were given. From a practical
point of view, the present results suggest that in addition
to safety concerns resulting from night workers' and
night drivers' propensity to fall asleep, we should also
recognize that even if they are wide awake, they may be
thinking more slowly.
,.
Acknowledgements: This work was supported by the
National Institute on Aging (AG 06836, AG 13396), the National Institute of Mental Health [MH 01235, MH 30915 (D.
J. Kupfer, P.I.)], NASA (NAS 9-18404, NAS 9-19407), and
the Canadian MRC. We thank our professional colleagues
for their advice on the manuscript and also Mr. Bart Billy,
Mr. Jeff Dettling, and our technicians for running the study.
REFERENCES
1. Colquhoun WP. Circadian variations in mental efficiency. In:
Colquhoun WP, ed. Biological rhythms and human performance. London: Academic Press, 1971:39-107 .
401
2. Monk TH, Folkard S, Wedderburn AI. Maintaining safety and
high performance on shift work. Appl Ergonom 1996;27:17-23.
3. Smith L, Folkard S, Poole CJM. Increased injuries on night
shift. Lancet 1994;1137-9.
4. Akerstedt T. Sleepiness as a consequence of shift work. Sleep
1988; 11: 17-34.
5. Pack AI, Pack AM, Rodgman E, Cucchiara A, Dinges DF,
Schwab CWo Characteristics of crashes attributed to the driver
having fallen asleep, Accid Anal Prey 1995;27:769-75.
6. Dinges OF, Kribbs NB. Performing while sleepy: effects of experimentally-induced sleepiness. In: Monk TH, ed. Sleep, sleepiness and performance. New York: J. Wiley, 1991:97-128.
7. Monk TH, Buysse OJ, Reynolds CF, et al. Circadian rhythms in
performance and mood under constant conditions. J Sleep Res
1997;6:9-18.
8. Baddeley AD. A threc-minutc reasoning tcst based on grammatical transformation. Psychonom Sci 1968;10:341.
9. Angus RG, Heslegrave RJ. Effects of sleep loss on sustained
cognitive performance during a command and control stimulation. Beh Res Meth Instru Comp 1985;17:55-67.
10. Williams HL, Lubin A. Speeded addition and sleep loss. J Exp
PsychoI1967;33l3-7.
II. Monk TH, Weitzman ED, Fookson JE, et al. Task variables
determine which biological clock controls circadian rhythms in
human performance. Nature 1983;304:543-5.
.
"
.
Sleep, Vo!' 20, No.6, 1997