Physiology & Behavior, Vol. 16, pp. 177-184. Pergamon
Press and Brain Research Publ., 1976. Printed in the U.S.A.
Altered Sleep Duration and Sleep Period Time Displacements: Effects on Performance in Habitual Long S l e e p e r s 1': J O H N M. T A U B
University o f California, Los Angeles AND R A L P H J. B E R G E R
University of California, Santa Cruz (Received 14 April 1975) TAUB, J. M. AND R. J. BERGER. Altered sleep duration and sleep period time displacements: effects on performance in habitual long sleepers. PHYSIOL. BEHAV. 16(2) 177-184, 1976. - Performance was studied in 10 healthy young adult males who characteristically sleep 9 112 - 10 1/2 hr following an electroencephalographically (EEG) recorded habitual sleep night and 4 nights on which their customary sleep was altered by 3 hr as follows: extended (E), deprived (D), delayed shift (DS), and advanced shift (AS). In the E condition sleep was extended by advancing sleep onset 3 hr corresponding to the AS condition which had the same retiring time, but differing from it with awakening occurring 3 hr earlier. In the D and DS conditions time of sleep onset was delayed 3 hr and the subjects were awakened at their customary time in the D condition, but 3 hr later than usual in the DS condition. Subjects performed an auditory vigilance task 35 min after awakening, at midday and in the early evening. Throughout the day after both shifted sleep and altered sleep duration performance was significantly impaired to an equivalent degree as reflected by longer reaction time, increased misses and a dedine of intrinsic perceptual capacity. Changes in the vigilance measures did not correlate with sleep duration or any other specific alterations in the EEG patterns of sleep. The behavioral deficits which resulted from altered sleep schedules are discussed viewing sleep as a biological adaptive process with respect to the feature of its occurrence under natural conditions in a temporally rhythmic sequence. Behavior Circadian rhythms EEG Performance Reaction time Sleep deprivation Sleep extension Sleep stages Vigilance
AS a c o n s e q u e n c e of terrestrial e v o l u t i o n there are behavioral variations in m a n y biological systems of plant and animal life associated with geophysical and diurnal cycles. U n d e r natural conditions the h u m a n sleep-waking r h y t h m is entrained to about 24 hr by periodic environmental agents (Zeitgebers) as living routine and social cues which are the most p o t e n t o f such synchronizing agents for man [3]. When Zeitgebers are volitionally or involuntarily altered such that it is not possible for persons to follow a c u s t o m a r y circadian r h y t h m of sleeping and waking, behavioral deficits would seem to be a predictable consequence. Relatively neglected, b u t a significant q u e s t i o n is the e x t e n t to which o p t i m a l waking behavior is contingent u p o n stability in the individual normal sleep-activity cycle. Probably the popular c o n c e p t i o n of sleep as being primarily a restorative process has led to an emphasis o f research on its experimental elimination m o r e c o m m o n l y
Signal detection
k n o w n as sleep deprivation. A view [6] for which there is some supportive evidence is that sleep loss studies have quite successfully d e m o n s t r a t e d alterations in psychophysiology and behavior due to the i n t e r r u p t i o n of an adapted process, m a n ' s basic 24 hr sleep-wakefulness cycle. Performance deficits on tasks shown sensitive to the direct effects of total sleep loss were observed in the morning, but not in the a f t e r n o o n after a night of unrestricted sleep following 24 hr awake, a finding which indicated to Wilkinson [37] the disruption of diurnal physiological r h y t h m s by sleep deprivation. The regular recurrent t e m p o r a l occurrence of sleep may perhaps be equally i m p o r t a n t to any fixed a m o u n t of sleeping t i m e per se for peak behavioral efficiency. In a previous study we observed that p e r f o r m a n c e declined during the day following acute 2 - 4 hr advances and delays in the times w h e n regular 1 2 : 0 0 - 8 : 0 0 a.m. sleepers were allotted
1 Support for this research was provided by National Institute of Mental Health Research Grant MH18928 and by National Institute of Mental Health Interdisciplinary Training Grant MH06415. 2The authors wish to thank Douglas Clarkson, Donald Guthrie and Robert T. Wilkinson for their advice and comments. Results of this study were presented in part at the annual meeting of the Association for the Psychophysiological Study of Sleep, San Diego, California, May, 1973. 177
178
TAUB AND BERGER
their usual 7 or 8 hr of sleep [27]. In another study of a similar group of sleepers [25] lengthening or shortening sleep by 3 hr caused similarly adverse behavioral deficits as advancing or delaying their regular sleep period by 3 hr. In both the above studies time spent asleep was similar for the shifted-sleep conditions and the 1 2 : 0 0 - 8 : 0 0 a.m. habitual condition, and averaged at least 7 hr. Furthermore, changes in the performance measures were unrelated to sleep duration or any specific changes in the electrophysiologically recorded sleep stages [20]. The findings from these studies and others [12,16] lend support to the hypothesis that alterations in a customary sleep-waking pattern may be more closely related to the efficiency of behavioral functions than total sleep duration or time spent in specific sleep stages [ 11,37]. Often underemphasized is the fact that individual members of any species, including a general population of humans of the same age, exhibit a wide range of individual differences in average sleep length [30,31]. If acutely altering the sleep-wakefulness rhythm produced similar defects of waking behavior in groups of subjects who differ in their habitual sleep durations, this might indicate the more significant behavioral adaptation conferred by maintenance of regular sleeping schedules than by the accumulation of any invariant amount of sleep. The purpose of the present investigation was to compare the behavioral effects of partial sleep deprivation and sleep extension with those following temporal shifts of the sleep period in a longer than normal group of subjects who characteristically sleep 9 1 / 2 - 1 0 1/2 hr per night. It was intended to determine whether similar findings of impaired performance would occur in this group of long sleepers as was observed to occur in habitual 7 - 8 hr sleepers [25] following 3 hr alterations in the length of timing of their accustomed sleep period. METHOD
Subjects Ten subjects were selected from 1,000 male respondents to an inventory distributed among students on 2 college campuses. The screening device was a modified version of the Cornell Medical Index [4[ which consisted of questions about medical and psychosomatic conditions, and sleep characteristics. Subjects were considered for further study only if their responses to the inventory were not indicative of sleep disturbance, medical problems, psychiatric disorders, and frequent alcohol or other drug usage; and if they reported having consistently slept 1 1/2 hr or longer over at least an immediately preceding period of 2 years. Charts similar in form to those used previously [28] were mailed to the 107 respondents who satisfied the above criteria with instructions to record for 2 weeks each 30 min period during which they were asleep. Subjects were eliminated if a discrepancy of more than 1 hr existed between their questionnaire estimate of sleep duration and average sleep during the 2 weeks. Sleep charts were returned by 54 subjects and 28 were selected for further study who showed that they almost always retired at an habitual time and had 9 1 / 2 - 1 0 1/2 hr of uninterrupted sleep nightly with no evidence of unusual fluctuations or daytime naps. In the final stage of screening, subjects were administered the MMPI and rejected if they scored 2 standard deviations above normal on any MMPI scale except Mf (since elevated Mf scores are quite common among male college students).
Ten subjects were randomly selected from 17 who were eligible and paid $75.00 for their participation in the experiment. The subjects ranged from 1 8 - 2 5 yr old with a mean age of 20 yr.
Measurement of Performance A 45 min. Wilkinson [ 39 ] auditory vigilance task was used to measure performance. Auditory vigilance has proven to be an aspect of behavior especially sensitive to moderate manipulations in sleep-waking patterns [9, 25, 27, 38 ]. During testing subjects sat in a sound attenuated cubicle and were presented the task binaurally through headphones. The auditory stimuli were 1/2 sec tones occurring at 2 sec intervals over 85 dB ambient white noise. Thirty tones were slightly shorter than the others (3/8 sec) and it was the subject's task to press a telegraph key immediately whenever he detected a short (critical) signal. The subject's reaction times to the critical signals were recorded in 1 msec units by a timer [29]. The critical signals occurred at irregular intervals such that they seemed random to the subjects. Ten different audiotapes were used so that the order of signals remained unpredictable to the subjects throughout the experiment. The number of signals missed (misses) and incorrectly detected (false reports) were scored, and further transformed to yield measures of intrinsic sensory capacity (6) and decision criterion for reporting signals O). These latter analyses were based on signal detection theory [24] and values of 6 and 0 were obtained from Freeman's [8] tables. Signal detection analysis separates factors of perceptual discrimination presumably from those of expectation and motivation which also affect the responses [23]. At a minimum of 2 days before the experiment, subjects practiced the task with at least 24 hr intervening between the 2 sessions. During the first practice a recorded 20 min preliminary instruction tape was presented which contained explanatory information about the vigilance task, delineated the signal from nonsignal stimuli and provided initial practice in detecting them. A 3 min familiarization period during which signals occurred at a relatively frequent rate to remind the subjects of their perceptual characteristics then preceded a 45 rain test tape. During the second practice session subjects were presented another vigilance test tape.
Design The experiment comprised 5 nights of sleep and an adaptation night of the subject's habitual 9 1 / 2 - 1 0 1/2 hr sleeping period preceded it by a week. The subjects were studied individually with sleep treatments spaced 1 week apart in a 10 x 5 balanced incomplete block design. The order of the sleep conditions was randomly assigned to each subject. The 5 experimental treatments were as follows: an habitual sleep condition; 2 conditions, 1 in which the period allowed for sleep was lengthened and 1 in which it was shortened by 3 hr; and 2 conditions, 1 in which the usual period for sleep was advanced and 1 in which it was delayed by 3 hr. In the habitual (H) condition subjects slept 9 1 / 2 - 1 0 1/2 hr at their accustomed times. Under the extended (E) sleep condition subjects accumulated extra sleep by having them retire 3 hr earlier than usual. In the advanced shift (AS) condition subjects were required to retire at the same time as in the E condition, but were awakened 3 hr sooner
ALTERED SLEEP PATTERNS AND PERFORMANCE IN LONG SLEEPERS than usual thereby obtaining their regular 9 1 / 2 - 1 0 1/2 hr of sleep. In the sleep deprivation (D) condition, subjects were required to remain awake 3 hr later than usual. In the delayed-shift (DS) condition subjects were required to remain awake also until the same time as in the D condition, but were awakened 3 hr later than usual thereby accumulating their regular 9 1 / 2 - 1 0 1/2 hr of sleep. For the D and E conditions of altered sleep duration, time of awakening was the same as in the H condition. The independent variables of sleep duration, retiring time, and awakening time for the 5 sleep conditions are depicted in Table 1. A common awakening time was chosen for the D and E conditions to control for the possible influence of circadian rhythms on behavioral measures taken in the morning. Three-hour manipulations of sleepwaking patterns were chosen to render comparable the magnitude of differences produced by the effects of shifting sleeping time and altering sleep duration between the long sleepers and regular 7 - 8 hr sleepers studied previously under an identical experimental design [25]. Table 2 shows the sleep habits of the subjects and times of testing. The sleeping times varied from subjects who retire at 10:00 or 10:30 p.m., and awaken at 7:30 or 8:00 a.m. to late sleepers who retire after midnight and wake up almost at noon. Testing times were scheduled relative to habitual sleep patterns as exemplified by subjects 9 and 10 who were awakened at 11:00 a.m. and were tested before lunch, at midday which was for them 4:30 p.m., and at 8:30 p.m. in the early evening.
Electrodes were placed for recording electroencephalographic, electromyographic, and electrooculographic activity during all conditions and the records were scored for sleep stages by 30 sec epochs according to standard procedures [20]. The 9 1 / 2 - 1 0 1/2 hr recordings from the H, AS and DS conditions were coded and scored blind with respect to treatment. Sleep records from the D and E conditions differed from these in duration so that it was not possible to employ a blind scoring procedure with them. Upon awakening subjects were served 120 ml of fruit juice as a partial nutritional control for blood sugar level. Thirty-five min later they were given the vigilance task which was administered twice again on the same day before lunch and in the evening at fixed clock times (Table 2) to control for possible effects of diurnal rhythms; and to determine the extent to which any effects of altered sleep patterns persist throughout the following day. Measures taken soon after awakening, however, were independent of the varying amounts of wakefulness preceding the tests taken at fixed times. Tests were not given at midday after the DS condition because postsleep testing for some subjects had finished almost exactly then or was in very close proximity to midday testing for others. The same procedure was, therefore, adopted as in earlier studies [25,27] in considering data obtained from the first DS test session twice in separate statistical analyses both as immediate postsleep and midday measures. There were no cues present in the experimental setting nor from the experimenter to inform subjects about their performance.
Procedure Before and during experimental sessions subjects were instructed not to nap or to drink caffeinated beverages, but to maintain their usual physical activity and food and fluid intake. They were told that the purpose of the experiment was to determine the relationship between sleep, biological time cycles, and personality functioning. All subjects reported to the laboratory at 6:00 p.m. (except as noted below) and sat in bed reading magazines or books. Time cues including daylight, chronometers and radios were absent until termination of postawakening testing. Since subjects 1 - 4 retired relatively early (see Table 1), in the AS and E conditions they were instructed to arrive at the laboratory by 4:00 p.m. allowing at least 3 hr of isolation from time cues and other periodic environmental factors before they were required to retire. The experimenter explained to these 4 subjects that the changes in their times for reporting to the laboratory were due to technical difficulties.
179
RESULTS Nonparametric statistics [22] were used in data analyses so as few assumptions as possible would have to be made concerning population distributions. Spearman rank correlation coefficients were computed to explore the possible relationship between the various parameters of sleep physiology and measures of vigilance. Two tailed values of the Wilcoxon matched pairs signed ranks test were used to evaluate effects of the sleep conditions on parameters of the vigilance task at each time of day. For each performance variable 4 main sets of comparisons were performed : H with D, DS, AS, and E; D with E; D with DS; and E with AS.
Main Effects of the Sleep Conditions There were statistically significant decrements in measures of both speed and accuracy on the vigilance task following each of the experimental treatments compared to
TABLE 1 TREATMENTVARIABLESFOR THE DIFFERENTCONDITIONS
Independent variables Sleep duration Retiring time Awakening time
Sleep deprivation (D)
Delayed shift ( D S )
6½-7½ hr 3 hr later than H Same as H
Same as H 3 hr later than H 3 hr later than H
Sleep Condition Advanced Habitual (H) shift (AS) 9½-10½ hr
Same as H 3 hr earlier than H 3 hr earlier than H
Extended sleep (E) 12½-13½ hr 3 hr earlier than H Same as H
180
TAUB AND BERGER TABLE 2 SUBJECTS' HABITUALTIMESOF SLEEPAND SCHEDULEOF TESTING
Subjects 1 2,3 4 5-7 8 9 10
Habitual Sleep 10:00 p.m.-7:30 a.m. 10:30 p.m.-8:00 a.m. 11:00 p.m.-9:00 a.m. 12:00-9:30a.m. 12:30-10:00 a.m. 12:30--1h00 a.m. 1:00-11:00 a.m.
Postsleep
Time of Test Sessions Midday
Evening
12:00 p.m. 12:00 p.m. 12:30 p.m. 12:30 p.m. 4:30 p.m. 4:30 p.m. 4:30 p.m.
5:00 p.m. 5:00 p.m. 5:30 p.m. 5:30 p.m. 8:30 p.m. 8:30 p.m. 8:30 p.m.
8:00 a.m. 8:30 a.m. 9:30 a.m. 10:00 a.m. 10:30 a.m. lh30a.m. lh30a.m.
the condition of habitual sleep. Mean reaction time to critical signals on the vigilance task during postsleep, midday, and evening testing sessions for each experimental condition is shown in Fig. 1. Reaction times were shorter in the H condition compared to the D, DS, AS, and E (experimental) conditions (Ts< 3, ps
Press and Brain Research Publ., 1976. Printed in the U.S.A.
Altered Sleep Duration and Sleep Period Time Displacements: Effects on Performance in Habitual Long S l e e p e r s 1': J O H N M. T A U B
University o f California, Los Angeles AND R A L P H J. B E R G E R
University of California, Santa Cruz (Received 14 April 1975) TAUB, J. M. AND R. J. BERGER. Altered sleep duration and sleep period time displacements: effects on performance in habitual long sleepers. PHYSIOL. BEHAV. 16(2) 177-184, 1976. - Performance was studied in 10 healthy young adult males who characteristically sleep 9 112 - 10 1/2 hr following an electroencephalographically (EEG) recorded habitual sleep night and 4 nights on which their customary sleep was altered by 3 hr as follows: extended (E), deprived (D), delayed shift (DS), and advanced shift (AS). In the E condition sleep was extended by advancing sleep onset 3 hr corresponding to the AS condition which had the same retiring time, but differing from it with awakening occurring 3 hr earlier. In the D and DS conditions time of sleep onset was delayed 3 hr and the subjects were awakened at their customary time in the D condition, but 3 hr later than usual in the DS condition. Subjects performed an auditory vigilance task 35 min after awakening, at midday and in the early evening. Throughout the day after both shifted sleep and altered sleep duration performance was significantly impaired to an equivalent degree as reflected by longer reaction time, increased misses and a dedine of intrinsic perceptual capacity. Changes in the vigilance measures did not correlate with sleep duration or any other specific alterations in the EEG patterns of sleep. The behavioral deficits which resulted from altered sleep schedules are discussed viewing sleep as a biological adaptive process with respect to the feature of its occurrence under natural conditions in a temporally rhythmic sequence. Behavior Circadian rhythms EEG Performance Reaction time Sleep deprivation Sleep extension Sleep stages Vigilance
AS a c o n s e q u e n c e of terrestrial e v o l u t i o n there are behavioral variations in m a n y biological systems of plant and animal life associated with geophysical and diurnal cycles. U n d e r natural conditions the h u m a n sleep-waking r h y t h m is entrained to about 24 hr by periodic environmental agents (Zeitgebers) as living routine and social cues which are the most p o t e n t o f such synchronizing agents for man [3]. When Zeitgebers are volitionally or involuntarily altered such that it is not possible for persons to follow a c u s t o m a r y circadian r h y t h m of sleeping and waking, behavioral deficits would seem to be a predictable consequence. Relatively neglected, b u t a significant q u e s t i o n is the e x t e n t to which o p t i m a l waking behavior is contingent u p o n stability in the individual normal sleep-activity cycle. Probably the popular c o n c e p t i o n of sleep as being primarily a restorative process has led to an emphasis o f research on its experimental elimination m o r e c o m m o n l y
Signal detection
k n o w n as sleep deprivation. A view [6] for which there is some supportive evidence is that sleep loss studies have quite successfully d e m o n s t r a t e d alterations in psychophysiology and behavior due to the i n t e r r u p t i o n of an adapted process, m a n ' s basic 24 hr sleep-wakefulness cycle. Performance deficits on tasks shown sensitive to the direct effects of total sleep loss were observed in the morning, but not in the a f t e r n o o n after a night of unrestricted sleep following 24 hr awake, a finding which indicated to Wilkinson [37] the disruption of diurnal physiological r h y t h m s by sleep deprivation. The regular recurrent t e m p o r a l occurrence of sleep may perhaps be equally i m p o r t a n t to any fixed a m o u n t of sleeping t i m e per se for peak behavioral efficiency. In a previous study we observed that p e r f o r m a n c e declined during the day following acute 2 - 4 hr advances and delays in the times w h e n regular 1 2 : 0 0 - 8 : 0 0 a.m. sleepers were allotted
1 Support for this research was provided by National Institute of Mental Health Research Grant MH18928 and by National Institute of Mental Health Interdisciplinary Training Grant MH06415. 2The authors wish to thank Douglas Clarkson, Donald Guthrie and Robert T. Wilkinson for their advice and comments. Results of this study were presented in part at the annual meeting of the Association for the Psychophysiological Study of Sleep, San Diego, California, May, 1973. 177
178
TAUB AND BERGER
their usual 7 or 8 hr of sleep [27]. In another study of a similar group of sleepers [25] lengthening or shortening sleep by 3 hr caused similarly adverse behavioral deficits as advancing or delaying their regular sleep period by 3 hr. In both the above studies time spent asleep was similar for the shifted-sleep conditions and the 1 2 : 0 0 - 8 : 0 0 a.m. habitual condition, and averaged at least 7 hr. Furthermore, changes in the performance measures were unrelated to sleep duration or any specific changes in the electrophysiologically recorded sleep stages [20]. The findings from these studies and others [12,16] lend support to the hypothesis that alterations in a customary sleep-waking pattern may be more closely related to the efficiency of behavioral functions than total sleep duration or time spent in specific sleep stages [ 11,37]. Often underemphasized is the fact that individual members of any species, including a general population of humans of the same age, exhibit a wide range of individual differences in average sleep length [30,31]. If acutely altering the sleep-wakefulness rhythm produced similar defects of waking behavior in groups of subjects who differ in their habitual sleep durations, this might indicate the more significant behavioral adaptation conferred by maintenance of regular sleeping schedules than by the accumulation of any invariant amount of sleep. The purpose of the present investigation was to compare the behavioral effects of partial sleep deprivation and sleep extension with those following temporal shifts of the sleep period in a longer than normal group of subjects who characteristically sleep 9 1 / 2 - 1 0 1/2 hr per night. It was intended to determine whether similar findings of impaired performance would occur in this group of long sleepers as was observed to occur in habitual 7 - 8 hr sleepers [25] following 3 hr alterations in the length of timing of their accustomed sleep period. METHOD
Subjects Ten subjects were selected from 1,000 male respondents to an inventory distributed among students on 2 college campuses. The screening device was a modified version of the Cornell Medical Index [4[ which consisted of questions about medical and psychosomatic conditions, and sleep characteristics. Subjects were considered for further study only if their responses to the inventory were not indicative of sleep disturbance, medical problems, psychiatric disorders, and frequent alcohol or other drug usage; and if they reported having consistently slept 1 1/2 hr or longer over at least an immediately preceding period of 2 years. Charts similar in form to those used previously [28] were mailed to the 107 respondents who satisfied the above criteria with instructions to record for 2 weeks each 30 min period during which they were asleep. Subjects were eliminated if a discrepancy of more than 1 hr existed between their questionnaire estimate of sleep duration and average sleep during the 2 weeks. Sleep charts were returned by 54 subjects and 28 were selected for further study who showed that they almost always retired at an habitual time and had 9 1 / 2 - 1 0 1/2 hr of uninterrupted sleep nightly with no evidence of unusual fluctuations or daytime naps. In the final stage of screening, subjects were administered the MMPI and rejected if they scored 2 standard deviations above normal on any MMPI scale except Mf (since elevated Mf scores are quite common among male college students).
Ten subjects were randomly selected from 17 who were eligible and paid $75.00 for their participation in the experiment. The subjects ranged from 1 8 - 2 5 yr old with a mean age of 20 yr.
Measurement of Performance A 45 min. Wilkinson [ 39 ] auditory vigilance task was used to measure performance. Auditory vigilance has proven to be an aspect of behavior especially sensitive to moderate manipulations in sleep-waking patterns [9, 25, 27, 38 ]. During testing subjects sat in a sound attenuated cubicle and were presented the task binaurally through headphones. The auditory stimuli were 1/2 sec tones occurring at 2 sec intervals over 85 dB ambient white noise. Thirty tones were slightly shorter than the others (3/8 sec) and it was the subject's task to press a telegraph key immediately whenever he detected a short (critical) signal. The subject's reaction times to the critical signals were recorded in 1 msec units by a timer [29]. The critical signals occurred at irregular intervals such that they seemed random to the subjects. Ten different audiotapes were used so that the order of signals remained unpredictable to the subjects throughout the experiment. The number of signals missed (misses) and incorrectly detected (false reports) were scored, and further transformed to yield measures of intrinsic sensory capacity (6) and decision criterion for reporting signals O). These latter analyses were based on signal detection theory [24] and values of 6 and 0 were obtained from Freeman's [8] tables. Signal detection analysis separates factors of perceptual discrimination presumably from those of expectation and motivation which also affect the responses [23]. At a minimum of 2 days before the experiment, subjects practiced the task with at least 24 hr intervening between the 2 sessions. During the first practice a recorded 20 min preliminary instruction tape was presented which contained explanatory information about the vigilance task, delineated the signal from nonsignal stimuli and provided initial practice in detecting them. A 3 min familiarization period during which signals occurred at a relatively frequent rate to remind the subjects of their perceptual characteristics then preceded a 45 rain test tape. During the second practice session subjects were presented another vigilance test tape.
Design The experiment comprised 5 nights of sleep and an adaptation night of the subject's habitual 9 1 / 2 - 1 0 1/2 hr sleeping period preceded it by a week. The subjects were studied individually with sleep treatments spaced 1 week apart in a 10 x 5 balanced incomplete block design. The order of the sleep conditions was randomly assigned to each subject. The 5 experimental treatments were as follows: an habitual sleep condition; 2 conditions, 1 in which the period allowed for sleep was lengthened and 1 in which it was shortened by 3 hr; and 2 conditions, 1 in which the usual period for sleep was advanced and 1 in which it was delayed by 3 hr. In the habitual (H) condition subjects slept 9 1 / 2 - 1 0 1/2 hr at their accustomed times. Under the extended (E) sleep condition subjects accumulated extra sleep by having them retire 3 hr earlier than usual. In the advanced shift (AS) condition subjects were required to retire at the same time as in the E condition, but were awakened 3 hr sooner
ALTERED SLEEP PATTERNS AND PERFORMANCE IN LONG SLEEPERS than usual thereby obtaining their regular 9 1 / 2 - 1 0 1/2 hr of sleep. In the sleep deprivation (D) condition, subjects were required to remain awake 3 hr later than usual. In the delayed-shift (DS) condition subjects were required to remain awake also until the same time as in the D condition, but were awakened 3 hr later than usual thereby accumulating their regular 9 1 / 2 - 1 0 1/2 hr of sleep. For the D and E conditions of altered sleep duration, time of awakening was the same as in the H condition. The independent variables of sleep duration, retiring time, and awakening time for the 5 sleep conditions are depicted in Table 1. A common awakening time was chosen for the D and E conditions to control for the possible influence of circadian rhythms on behavioral measures taken in the morning. Three-hour manipulations of sleepwaking patterns were chosen to render comparable the magnitude of differences produced by the effects of shifting sleeping time and altering sleep duration between the long sleepers and regular 7 - 8 hr sleepers studied previously under an identical experimental design [25]. Table 2 shows the sleep habits of the subjects and times of testing. The sleeping times varied from subjects who retire at 10:00 or 10:30 p.m., and awaken at 7:30 or 8:00 a.m. to late sleepers who retire after midnight and wake up almost at noon. Testing times were scheduled relative to habitual sleep patterns as exemplified by subjects 9 and 10 who were awakened at 11:00 a.m. and were tested before lunch, at midday which was for them 4:30 p.m., and at 8:30 p.m. in the early evening.
Electrodes were placed for recording electroencephalographic, electromyographic, and electrooculographic activity during all conditions and the records were scored for sleep stages by 30 sec epochs according to standard procedures [20]. The 9 1 / 2 - 1 0 1/2 hr recordings from the H, AS and DS conditions were coded and scored blind with respect to treatment. Sleep records from the D and E conditions differed from these in duration so that it was not possible to employ a blind scoring procedure with them. Upon awakening subjects were served 120 ml of fruit juice as a partial nutritional control for blood sugar level. Thirty-five min later they were given the vigilance task which was administered twice again on the same day before lunch and in the evening at fixed clock times (Table 2) to control for possible effects of diurnal rhythms; and to determine the extent to which any effects of altered sleep patterns persist throughout the following day. Measures taken soon after awakening, however, were independent of the varying amounts of wakefulness preceding the tests taken at fixed times. Tests were not given at midday after the DS condition because postsleep testing for some subjects had finished almost exactly then or was in very close proximity to midday testing for others. The same procedure was, therefore, adopted as in earlier studies [25,27] in considering data obtained from the first DS test session twice in separate statistical analyses both as immediate postsleep and midday measures. There were no cues present in the experimental setting nor from the experimenter to inform subjects about their performance.
Procedure Before and during experimental sessions subjects were instructed not to nap or to drink caffeinated beverages, but to maintain their usual physical activity and food and fluid intake. They were told that the purpose of the experiment was to determine the relationship between sleep, biological time cycles, and personality functioning. All subjects reported to the laboratory at 6:00 p.m. (except as noted below) and sat in bed reading magazines or books. Time cues including daylight, chronometers and radios were absent until termination of postawakening testing. Since subjects 1 - 4 retired relatively early (see Table 1), in the AS and E conditions they were instructed to arrive at the laboratory by 4:00 p.m. allowing at least 3 hr of isolation from time cues and other periodic environmental factors before they were required to retire. The experimenter explained to these 4 subjects that the changes in their times for reporting to the laboratory were due to technical difficulties.
179
RESULTS Nonparametric statistics [22] were used in data analyses so as few assumptions as possible would have to be made concerning population distributions. Spearman rank correlation coefficients were computed to explore the possible relationship between the various parameters of sleep physiology and measures of vigilance. Two tailed values of the Wilcoxon matched pairs signed ranks test were used to evaluate effects of the sleep conditions on parameters of the vigilance task at each time of day. For each performance variable 4 main sets of comparisons were performed : H with D, DS, AS, and E; D with E; D with DS; and E with AS.
Main Effects of the Sleep Conditions There were statistically significant decrements in measures of both speed and accuracy on the vigilance task following each of the experimental treatments compared to
TABLE 1 TREATMENTVARIABLESFOR THE DIFFERENTCONDITIONS
Independent variables Sleep duration Retiring time Awakening time
Sleep deprivation (D)
Delayed shift ( D S )
6½-7½ hr 3 hr later than H Same as H
Same as H 3 hr later than H 3 hr later than H
Sleep Condition Advanced Habitual (H) shift (AS) 9½-10½ hr
Same as H 3 hr earlier than H 3 hr earlier than H
Extended sleep (E) 12½-13½ hr 3 hr earlier than H Same as H
180
TAUB AND BERGER TABLE 2 SUBJECTS' HABITUALTIMESOF SLEEPAND SCHEDULEOF TESTING
Subjects 1 2,3 4 5-7 8 9 10
Habitual Sleep 10:00 p.m.-7:30 a.m. 10:30 p.m.-8:00 a.m. 11:00 p.m.-9:00 a.m. 12:00-9:30a.m. 12:30-10:00 a.m. 12:30--1h00 a.m. 1:00-11:00 a.m.
Postsleep
Time of Test Sessions Midday
Evening
12:00 p.m. 12:00 p.m. 12:30 p.m. 12:30 p.m. 4:30 p.m. 4:30 p.m. 4:30 p.m.
5:00 p.m. 5:00 p.m. 5:30 p.m. 5:30 p.m. 8:30 p.m. 8:30 p.m. 8:30 p.m.
8:00 a.m. 8:30 a.m. 9:30 a.m. 10:00 a.m. 10:30 a.m. lh30a.m. lh30a.m.
the condition of habitual sleep. Mean reaction time to critical signals on the vigilance task during postsleep, midday, and evening testing sessions for each experimental condition is shown in Fig. 1. Reaction times were shorter in the H condition compared to the D, DS, AS, and E (experimental) conditions (Ts< 3, ps
