Sleep, 6(4):362-368
© 1983 Raven Press, New York
The incentive Effect and Sleep Deprivation
D. R. Haslam
Army Personnel Research Establishment, Farnborough, Hampshire, England
Summary: In order to examine the effect of a small amount of sleep following
J3j4 days (90 h) of wakefulness, 10 infantry soldiers took part in a laboratorybased experiment. At the end of the vigil, a 2-h sleep was preceded and followed by a cognitive test session consisting of encoding and decoding. In order
to simulate a realistic situation, subjects were not told the scheduled length of
their vigil until a few hours before their 2-h sleep. Following the test-sleeptest period, 27 h were allowed for sleep and rest. Results indicated that after
3 nights without sleep, performance was, on the average, 55% of the control
values. During the test session before the 2-h sleep, performance improved by
30%, to 85% of control values, indicating the considerable effect that incentive
can have on even severely sleep-deprived subjects. The reserve mental capacity demonstrable during sleep deprivation indicates the caution that is
needed if the effects of "undiluted" sleep loss are sought; it also emphasises
once again the lack of knowledge concerning the function of sleep. Key Words:
Sleep deprivation - Incentive- N aps- Performance.
It is important for military commanders to know the least amount of sleep likely to
be beneficial to their men after 2 or more days have passed without sleep. Although
in recent years there have been a number of studies concerned with nap sleep, most
of these have examined the effect offragmented sleep without initial sleep loss (1-5).
Naitoh (6), however, in a laboratory experiment, examined the effect after 45 h of
wakefulness of 2 h sleep in the early morning (from 0400 to 0600 h) and, after 53 h of
wakefulness, of 2 h sleep at midday. Naitoh reported that "the early morning nap
resulted in severe and prolonged sleep inertia," and did not have any recuperative
power, whereas "the midday nap produced a relatively short sleep inertia and had
clear recuperative power." Naitoh concluded that the recuperative power of a nap
depended not only on its duration but, more importantly, on the hours of prior wakefulness and the time of day when it was taken.
Opstad et al. (7), on the other hand, in a field study found an improvement in the
performance of cadets after either a 3-h sleep (from 0300 to 0600 h) or a 6-h sleep (from
0001 to 0600 h) taken half-way through a combat training course lasting either 90 or
114 h. In addition, 4 h sleep in the early morning hours at the end of the course had a
beneficial effect on test performance. Also in a field study, Haslam (8) found that after
90 h without scheduled sleep, 4 h sleep in the early morning hours (from 0145 to 0545)
Accepted for publication August 1983.
Address correspondence and reprint requests to Dr. D. R. Haslam, Army Personnel Research Establishment, Famborough, Hampshire, England.
362
363
SLEEP LOSS, INCENTIVE, AND NAPS
had a marked beneficial effect upon the military performance and mood of 10 infantry
soldiers.
Since both Opstad et a1. and Haslam found 3 and 4 h sleep, respectively, to have a
beneficial effect after prolonged wakefulness, and in view of Naitoh's negative finding,
it was decided to examine further the value, positive or otherwise, of nap sleep in the
early morning hours following 3 to 4 days of wakefulness. Accordingly, as part of a
larger study, it was decided to examine the effect of 2 h sleep following 90 h without
scheduled sleep.
To simulate a military situation in which there is a foreseen lull in operations, a few
hours before the end of the vigil subjects were told they were going to be allowed some
sleep. Up to this time, they were ignorant of the scheduled length of sleep deprivation.
METHODS
Subjects
Ten trained infantrymen, who had not previously taken part in a sleep deprivation
experiment, were used as subjects. All were fit, and none was taking any medication.
They were divided into two groups of five, with one noncommissioned officer and four
private soldiers in each group. To increase motivation, the two groups were in competition against each other for a reward of extra leave. The competition was based on
the maintaining of performance levels in certain key tasks in relation to initial control
(baseline) values. The data from the two groups were combined for the statistical
analysis. The average age of the whole group was 23.1 years (range 20-31).
Trial design
The experiment started at 0630 hours. During the first 33/4 days (89.75 h) of this
period (days El to E3) there was no scheduled sleep. As stated above, subjects were
not told the intended length of the sleep deprivation period until a few hours before
their first allocated sleep. At the 89-h-awake point, the following sequence of events,
which took place at this point, was made known to them:
Hours:
23300010
00150215
02150245
02450330
Cognitive
test
session
Sleep
Interval
to permit
awakening
Cognitive
test
session
0330Unlimited
sleep
For both sleep periods, subjects remained fully dressed in their combat kit (less
boots), and slept on camp beds in a light-proof laboratory, which was free from noise.
From 0330 h, subjects were allowed to sleep until they awoke spontaneously.
The experimental period was preceded and followed by a 2-day control period in
which 7 h uninterrupted sleep (from 2330 to 0630 h) in every 24 h was allowed. The
first of the control periods was preceded by 1 day of training, and the second by II/S
days (27 h) of rest and sleep.
Sleep. Vol. 6, No.4, 1983
D. R. HASLAM
364
Day 1
T1
Training
on all
tests and
procedures
Days 2 and 3
CI, C2
Control:
7/24 h
sleep;
testing at
1000 h.
Days 4 to 7
EI to E4
Experimental:
No scheduled
sleep, followed
by period
detailed above;
on days El to
E3, testing at
0230 and 1000 h.
Day 7
RI
Restl
sleep:
27 h.
Days 8 and 9
C3, C4
Control:
7/24 h
sleep;
testing at
1000 h.
In an attempt to eliminate practice effects over the trial period, during the week
preceding the trial the subjects underwent training on the performance tests.
Throughout the trial, the subjects were observed by military, civilian, and medical staff
members for 24 h a day. One of their duties was to ensure that the subjects had no
unscheduled sleep.
Pattern of activities
During the trial period, the subjects ate and slept (when permitted) in the laboratory,
and a programme of tests and activities (including card games for relaxation) was
devised to keep them occupied for most of the day, and also the night, during the sleepdeprivation period. With regard to individual performance assessment, only the tests
that were given before and after the 2-h sleep will be included in this report, since the
results for these tests are the subject under discussion. Although the test session at
2300 h on day E3 was the only one at that time of day, it was considered to be comparable to the 1000 h sessions, because Kleitman (9-p. 157) has shown that cognitive
performance at 2330 h is approximately the same as at 1000 h. The tests used were as
follows.
Encoding/decoding tests using military ciphers
Encoding and decoding tests, which have been used successfully in previous experiments, give an indication of cognitive ability and are susceptible to sleep-loss effects.
A product-moment correlation of 0.6 for 12 subjects (unpublished data) has been found
between performance on encoding/decoding grid references and performance on the
Stroop test (10). The tests are fairly complex, and accurate performance requires
sustained attention. In addition, these tests are relevant and meaningful to the Army
and are, therefore, motivating.
Encoding/decoding grid references
Subjects were given 5 min for encoding 6-figure map grid references, followed by 5
min for decoding. Performance was scored for the number of correct grid references
and the number of errors. On the experimental days, test sessions were held at 1000
hand 0230 h. During the control days there was only one session daily (at 1000 h),
because of the subjects' sleep schedule. OWing to a circadian effect, performance
would be expected to be better at 1000 h than at 0230 h.
Decoding messages
In this 1O-min test, which always followed the above test, a message was decoded.
Performance was scored for the number of code words correctly decoded and for the
number of errors.
Sleep, Vol. 6. No.4, 1983
365
SLEEP LOSS, INCENTIVE, AND NAPS
Statistical analysis
A mixed-model Type 3 Analysis of Variance was used, with a testing strategy as
follows: the main factors were tested against the subject by main factor interaction
except when the mean square for the interaction was less than the mean square for
the residual, in which case the residual was used for the error term. The same procedure was followed for all interactions not involving the subjects.
RESULTS
All 10 subjects completed the trial; no subject had more than a few seconds at a
time of unscheduled sleep, and the total amount taken was small.
Objective tests
Encoding and decoding grid references. The results are shown in Table 1. The
number correct deteriorated over the sleep deprivation days. This was due, from day
El to E3, to a drop in the number of items attempted and not to an increase in errors.
The difference between daytime and nighttime scores was not statistically significant,
nor was the difference between days Cl, C2 compared with days C3, C4.
The Analysis of Variance showed the days to be significantly different at the p <
0.001 level (F = 2.63 for encoding, 6.06 for decoding, df = 11,99). Further breakdown
using selected contrasts (Table 2) indicated that performance on day E3 at 1000 h was
significantly worse than on Cl, C2 and on C3, C4. Performance before the 2-h sleep
(day E3) was significantly better in the decoding test than at 1000 h on day E3 (p <
0.001), and although there was an improvement in encoding-a difference of 33.8%
(see Table 1)-this did not reach statistical significance.
Scores for days E3 and E4 as percentages of the average control value (Cl, C2, C3,
C4) are as follows (it should be remembered that there was no significant difference
between daytime and nighttime scores):
Encoding
Decoding
Day E3
(1000 h)
Day E3
Before 2-h
sleep (2330 h)
Day E4
After 2-h
sleep (0245 h)
68.0
37.0
91.2
85.0
77.9
70.0
Decoding messages. Table 1 gives the mean number of correct items and standard
deviations. As can be seen, the number correct deteriorated over the sleep deprivation
period; there was no statistically significant change in the number of errors. The difference between daytime and nighttime scores was not statistically significant. The
Analysis of Variance showed the days to be significantly different at the p < 0.001
level (F = 4.32, df = 11,99). Further breakdown using selected contrasts (Table 2)
indicated that performance on day E3 at 1000 h was significantly worse than on Cl,
C2 and C3, C4. Compared with 1000 h on day E3, performance improved before the
2-h sleep, the difference being 28.6%; but this was not statistically significant.
Although scores on C3, C4 were not significantly better than on Ct, C2, there was
a tendency in that direction. Unlike the previous test, there was a tendency for performance to be better after the sleep than before it; this might reflect the length of time
Sleep, Vol. 6, No.4, 1983
D. R. HASLAM
366
TABLE 1. Average scores and standard deviations in the encoding and decoding tests
Decoding
grid
references
(number
correct)
Encoding
grid
references
(number
correct)
Decoding
messages
(number of
correct
words)
Day
Time
Mean
SD
Mean
SD
Mean
SD
C1
C2
E1
1000
1000
0230
1000
0230
1000
0230
1000
2330
0245
1000
1000
10.3
11.4
10.9
9.0
9.1
8.1
7.1
7.7
10.3
8.8
11.5
11.9
4.1
4.0
5.1
5.1
3.9
3.5
5.1
3.9
3.9
5.2
4.7
3.7
7.0
8.0
7.6
5.5
4.7
6.0
3.3
2.8
6.8
5.6
8.4
8.7
4.3
3.8
3.4
3.7
2.9
3.4
3.7
1.5
3.4
3.2
3.5
4.9
35.8
45.9
40.3
30.6
35.3
28.2
27.1
27.6
35.5
37.9
45.8
46.5
16.7
8.5
14.2
12.6
16.4
15.1
17.5
18.7
16.5
13.1
8.8
9.4
E2
E3
E4
C3
C4
since awakening. As stated above, the test on messages was always carried out after
the grid references test.
Scores for days E3 and E4 as percentages of the average control value (CI, C2, C3,
C4) are as follows (again there was no significant difference between daytime and
nighttime scores):
Day E3
1000 h
Day E3
Before 2-h
sleep
Day E4
After 2-h
sleep
62.0
80.3
85.7
General behavior
On the third day of sleep loss, the subjects continually dozed, especially during the
cognitive test session at 1000 h, but they were wakened immediately by observers.
Furthermore, during the afternoon on the third day of sleep loss, subjects appeared to
be in a daze when they went from one laboratory to another for the various test
sessions, and, during the relaxation period, they could not maintain their concentration
on a card game for more than a few seconds at a time.
Contrasted with this was their behaviour, demeanour, and mood during the last few
hours of the vigil, when they knew the 2-h sleep period was imminent. They appeared
cheerful and alert, and did not have the appearance expected of people who had been
awake for almost 90 h.
Recovery period
The average amount of sleep taken when the subjects were allowed unlimited sleep
at the end of the deprivation period was 9.7 h (range 7.5-11.3). This, together with
the 2 h they were allowed before the last test session, amounts to an average of 11.7 h.
Sleep, Vol. 6, No.4, 1983
367
SLEEP LOSS, INCENTIVE, AND NAPS
TABLE 2. Significance levels for selected contrasts in the encoding and decoding tests
Days
Cl, C2 versus C3, C4
0230 versus 1000 h (EI-E3)
E3 (1000) versus E3 (2300)
Cl, C2 versus E3 (1000)
C3, C4 versus E3 (1000)
Cl, C2 versus E3 (2300)
C3, C4 versus E3 (2300)
E4 versus E3 (2300)
C3, C4 versus E4
Encoding
grid
references
Decoding
grid
references
NS
NS
NS
NS
NS
p < 0.05
p < 0.01
p < 0.001
p < 0.001
p < 0.001
NS
NS
NS
NS
NS
NS
p < 0.05
p < 0.01
Decoding
messages
NS
NS
NS
P < 0.01
P < 0.001
NS
p < 0.01
NS
NS
DISCUSSION
It was expected that the anticipation of 2 h sleep would, to some extent, reduce
feelings of tiredness and improve mood. However, it was not known to what extent it
would improve cognitive performance; in the absence of sleep for 90 h it was felt that
it was unlikely to have a great effect, although it is well known that with moderate
amounts of sleep loss, a stimulating event can improve performance considerably. In
the event, average performance in the cognitive tests improved by 30%, and behaviour
changed from a state of extreme lethilrgy to buoyant cheerfulness.
As stated earlier, performance at i330 h has been reported as being approximately
the same as performance at 1000 h (9). However, on day E3, by 2330 h there had been
an additional 13.5 h without sleep compared with the amount at 1000 h that day.
Therefore, in the absence of incentive, it might be expected that performance would
be worse than at 1000 h. However, as indicated above, there was an improvement in
all three tests. Although statistical significance was reached in only one of these (decoding grid references), it is clear that the incentive of impending sleep had an effect.
If it had not been shown that performance had improved by 30% before the 2-h sleep
period, it would have seemed that this sleep had improved performance by 22% (it is
unlikely that any incentive effect would have outlasted the sleep). However, because
of the apparent effect on performance of the anticipation of sleep, no conclusion can
be drawn as to the value or otherwise of the 2-h sleep per se. This is because subjects
were going to be allowed unlimited sleep following the test session, and anticipation
of this may well have affected their performance.
Although in the present experiment the effect of incentive improved performance by
30%, the tests were only of 5 and 10 min duration. The effect on a task of greater
length, and the duration of time over which such mental simulation would be effective,
is unknown at present; but these are questions which it is proposed to address in due
course. It should be remembered, however, that these were the tests that were adversely affected over days EI to E3.
The large effect of incentive on severely sleep-deprived, and apparently very fatigued, subjects indicates that the "pure" effect of sleep loss can be jeopardised by a
change for the better in the'subjects' mental attitude. It is important, therefore, that
care should be exercised in the management of subjects in order not to compromise
the results of sleep deprivation experiments.
Sleep, Vol. 6, No.4, 1983
368
D. R. HASLAM
The finding that average performance for the three tests improved before the 2-h
sleep from 55% of control values to 85% of control values brings to mind the statement
of Whiting and English (11) that fatigue does not directly cause work decrement but
raises the threshold at which work motives are effective. They go on to say that "if
such positive motives are adequate at all, then fatigue has no effect upon work efficiency." If this statement is true, it raises the question of the function of sleep. As
Horne (12) says, "If much of human sleep were for cerebral restitution, then the extent
of cerebral capacity to perform at near optimal levels during limited total sleep deprivation would be less than is found."
Even if it is not apparent what the function of sleep is, and it seems unlikely that it
is for body restitution (13) [Horne, however, is of the opinion that sleep does have a
restorative function (personal communication, 1983)], the need to sleep after sleep loss
seems to have the essential qualities of a physiological drive, which increases in relation
to sleep deprivation (14,15), and has the effect of decreasing the interest in, and motivation to carry out, other tasks. If, however, sufficient incentive is injected into the
situation, for example, by the promise of sleep, then the downward trend of performance can, on short tasks anyway, be reversed, demonstrating a reserve and manipulative mental capacity even in the presence of severe sleep deprivation.
REFERENCES
1. Carskadon MA, Dement WC. Sleep studies on a 90-minute day. Electroencephalogr Clin Neurophysiol
1975;39:145-55.
2. Moses JM, Hord DJ, Lubin A, Johnson LC, Naitoh P. Dynamics of nap sleep during a 40-hour period.
Electroencephalogr Clin NeurophysioI1975;39:627-33.
3. Webb WB, Agnew HW, Jr. Sleep: effects of a restricted regime. Science 1965;150:1745-7.
4. Weitzman ED, Nogeire C, Perlow M et aI. Effects of a prolonged 3-hour sleep-wake cycle on sleep
stages, plasma cortisol, growth hormone, and body temperature in man. J Clin Endocrinol 1974;38:
1018-30.
5. Wilkinson RT, Edwards RS. Performance following a night of reduced sleep. Psychon Sci 1966;5:
471-2.
6. Naitoh P. Circadian cycles and restorative power of naps. In: Johnson LC, Tepas DI, Colquhoun WP,
Colligan MJ, eds, Biological rhythms, sleep and shift work, Advances in Sleep Research, Vol 7. New
York: Spectrum, 1981:553-80.
7. Opstad PK, Ekanger M, Nummestad M, Raabe N. Performance, mood, and clinical symptoms in men
exposed to prolonged, severe physical work and sleep deprivation. Aviat Space Environ Med
1978;49: 1065-73.
8. Haslam DR. Sleep loss, recovery sleep, and military performance. Ergonomics 1982;25: 163-78.
9. Kleitman N. Sleep and wakefulness, Chicago: University of Chicago Press, 1972.
10. Jensen AR, Rohwer WD. The Stroop Colour-Word Test: a review. Acta Psychologica 1966;25:36-93.
11. Whiting HF, English HB. Fatigue tests and incentives. J Exp PsychoI1925;8:33-49.
12. Home JA. Sleep deprivation, stress and sleep function. In: Koella WP, ed, Sleep 1980: Proceedings of
the Fifth European Congress on Sleep Research. Basel: Karger, 1981:95-7.
13. Home JA. A review of the biological effects of total sleep deprivation in man. Bioi Psychol 1978;7:
55-102.
14. Akerstedt T, Gillberg M. Sleep, stress and recuperation. In: Koella WP,- ed, Sleep 1980: Proceedings of
the Fifth European Congress on Sleep Research. Basel: Karger, 1981:98-101.
15. Dement WC, Seidel W, Carskadon M. Daytime alertness, insomnia, and benzodiazepines. Sleep
1982;5:528-45.
Sleep, Vol. 6, No.4, 1983