Sleep 1(6):6-65, Raven Press, Ltd., New York 1987 Association of Professional Sleep Societies Homeostatic Changes in Slow Wave Sleep during Sleep Following Nocturnal Sleep and Partial Slow Wave Sleep during an Afternoon Nap Andrew Tilley, Frank Donohoe, and Sharon Hensby Department of Psychology, University of Queensland, Queensland, Australia Summary: There has been speculation and some evidence to suggest that certain fractions of sleep, notably, slow wave sleep (SWS), are under homeostatic control. In order to test this hypothesis, the sleep of eight subjects. was terminated on four different occasions after 5% of their normal baseline SWS levels had been obtained. An afternoon nap then followed during which 1%, 5%, or 25% of the SWS debt was reclaimed. A fourth condition contained no afternoon nap. The change in SWS from baseline during subsequent recovery sleep was directly related to the outstanding SWS debt, thus demonstrating that SWS is under homeostatic control. Key Words: Homeostasis-Slow wave sleep. Loss of slow wave sleep (SWS), as a result of short-term selective or total sleep deprivation, is usually substantially reclaimed during subsequent recovery sleep (1-4). For example, reanalysis of the data reported by Agnew et al. (1) shows that 9% of the SWS (stage 3 and 4) lost during 2 consecutive nights of selective stage 4 sleep deprivation was recovered on a single recovery night. In a complementary fashion, BorMly et al. (4) have reported that 1% of the SWS lost following 1 night of sleep deprivation is reclaimed over the next 2 recovery nights. These results suggest that SWS may be under homeostatic control and that there may be a "need" for a set amount of SWS per day. Further incidental evidence supporting the notion that SWS is under homeostatic control is provided by studies showing that normal, baseline levels of SWS are usually preserved during restricted sleep regimes (5-12) or if nocturnal sleep is replaced or augmented by naps (13,14,15). For example, the amount of SWS obtained during 7 nights of sleep restricted to 3 h/night was virtually the same as that which would have been obtained during 7 full night's sleep (5). The amount of SWS obtained in ten I-hour naps evenly spread over a 4-h period (6 to 22 h the next day) was exactly the same as would have been obtained in the single night's sleep which the naps replaced Address correspondence and reprint requests to Dr. A. Tilley at Department of Psychology, University of Queensland, St Lucia, Queensland 467, Australia. 6
HOMEOSTATIC CHANGES IN SWS 61 (14). Finally, an afternoon nap has been shown to produce reductions in the amount of SWS during subsequent nocturnal sleep commensurate with the amount of SWS obtained during the nap (13, IS). The focus of the last study (1S) was on stage 4 sleep. It was reported that the reduction in nocturnal stage 4 sleep following an afternoon nap was greater than the amount of stage 4 sleep obtained in the nap. However, reanalysis of the data in terms of SWS shows that the reduction in the amount of SWS at night is equivalent to the amount obtained during an afternoon nap. The present experiment is a direct test of the hypothesized homeostasis of SWS. If SWS is under homeostatic control, then reducing the amount of SWS at night should result in homeostatic increases in SWS during subsequent sleep. Furthermore, it should be possible to supplement the recovery of the SWS debt by means of a daytime nap. For example, if SWS is restricted to SO% of baseline levels at night and 1% of the debt is repaid during a nap, the level of SWS should remain at near baseline levels during' 'recovery" sleep. If only SO% of the debt is repaid during a nap, then the level of SWS should rise 2S% above baseline levels during recovery sleep, and so on. In short, the amount of additional SWS obtained during recovery sleep should be a function of the amount of SWS debt. The present study was designed to test this hypothesis by systematically varying the amount of SWS debt repaid during an afternoon nap prior to recovery sleep. METHODS Eight healthy young women (mean age 2), all paid volunteers and students at the University of Queensland, took part in the study. Following three, nonrecording adaptation nights in the sleep laboratory, their sleep electroencephalographic (C4/Al electrode placements), electro-oculographic, and electro myographic activity was recorded for 2 consecutive nights to establish individual baseline sleep measures, which were averaged across the 2 baseline nights. Each subject returned to the sleep laboratory on four separate occasions, spaced at least one week apart. On each occasion, following 2 further adaptation nights, the subject's nocturnal sleep was continually monitored from lights out (the subject's habitual bedtime) and terminated after ~SO% of baseline SWS had been obtained. After being awakened, the subject arose, showered and dressed, and was supervised throughout the remainder of the experimental period. Most of the subjects spent their time in the psychology student's common room playing video games, studying, or chatting with other students. The restricted nocturnal sleep was then either supplemented by an afternoon nap, commencing at 13 h, or followed by no nap at all. (The order of the four conditions was counterbalanced across subjects.) The afternoon nap was allowed to continue until ~1, ~SO, or ~2S% of the SWS debt had been repaid, at which point the subject was awakened, arose, dressed, and was supervised as before. The subject returned to bed in the early evening. Bedtime varied slightly from subject to subject and from condition to condition depending upon the total amount (pre-nap + post-nap) of wakefulness, which was fixed at around 16.S h, a normal daily amount. For example, if the subject had been woken at 1 h after restricted sleep and had a l.s-h afternoon nap from 13 to 143 h~ the subject would have gone to bed at 19 h. The subject was allowed to sleep ad libitum until awakening spontaneously the following morning.
62 A. TILLEY ET AL. RESULTS All sleep records were scored blind according to standard procedures (16). Entry into stage 2 sleep was taken as the point of sleep onset. Summary statistics for the various sleep measures are shown in Table 1. Table 2 shows the relationship between the amount of stage 2, SWS, REM, and total sleep time (TST) debt before recovery sleep and the change in these sleep measures from baseline during recovery sleep. TABLE 1. Sleep measures a for baseline sleep and four experimental conditions Latency Awake Stage 1 Stage 2 Stage 3 Stage 4 SWS REM TST Baseline Mean 16 (7) 3 (I) 8 (3) 225 (2) 22 (6) 84 (14) 16 (14) 118 (14) 457 (25) 1% Nap condition Mean 14 (5) () 1 (2) 22 (12) 8 (3) 46 (8) 54 (8) 2 (3) 79 (I5) Debt 23 14 38 52 116 378 Nap Mean 5 (2) () () 12 (6) 8 (5) 48 (8) 56 (1) 9 (7) 77 (15) Debt 191 6-1 -4 17 31 Mean 16 (4) 4 (2) 7 (2) 286 (55) 4 (II) 68 (16) 18 (22) 23 (26) 64 (53) Debt 13-12 6-6 22 154 5% Nap condition Mean 16 (8) () 3 (7) 16 (14) 6 (2) 45 (7) 51 (7) 2 (4) 72 (21) Debt 29 16 39 55 116 385 Nap Mean 5 (I) () () 1 (5) 4 (3) 22 (4) 26 (5) () 36 (6) Debt 199 12 17 29 116 349 Mean 14 (1) 2 (3) 7 (4) 38 (5) 44 (9) 88 (19) 132 (24) 19 (29) 637 (46) Debt 116-1 \3 3 44 169 25% Nap condition Mean 12 (5) () () 17 (19) 7 (4) 47 (6) 54 (8) 4 (5) 75 (26) Debt 15 37 52 114 382 Nap Mean 6 (2) () () 7 (3) 2 (I) 13 (2) 15 (2) () 23 (4) Debt 21 \3 24 37 114 359 Mean 11 (6) 2 (4) 9 (6) 34 (53) 41 (9) 93 (2) 134 (23) 196 (19) 643 (58) Debt 122-6 15 9 36 173 No nap condition Mean 13 (6) () 4 (8) 35 (47) 7 (6) 43 (13) 5 (9) 2 (5) 91 (48) Debt 19 15 41 56 116 366 Mean 5 (2) 5 (6) 7 (7) 334 (48) 4 (9) 18 (18) 148 (23) 184 (22) 673 (48) Debt 81-3 17 14 5 15 SDs are shown in parentheses. Latency, time from lights out to sleep onset; awake, time awake following sleep onset; and TST, total sleep time. Negative numbers signify amount of debt overpaid. a Measured in minutes, rounded to nearest whole number.
HOMEOSTATIC CHANGES IN SWS 63 As shown in Table 1, the amount of SWS obtained during recovery sleep appears to be determined by the amount of SWS debt. In general, the greater the debt, the higher the amount of SWS during recovery sleep. Moreover, the change in SWS from baseline is more or less the same as the SWS debt, although there appears to be a slight overshoot in debt recovery in the 1% nap condition and a slight undershoot in the 5%, 25%, and no nap conditions. As shown in Table 2, overall, the SWS debt accounts for over 8% of the variance in the change in SWS from baseline during recovery sleep. This clearly indicates that SWS is under homeostatic control. The relationship between SWS debt and SWS change during recovery sleep is shown in Figure 1. It is possible that some of the change in SWS from baseline during recovery sleep may be due to the amount of prior TST (restricted sleep + nap), which significantly decreases across conditions from 156 min in the 1% nap condition to 91 min in the no nap condition (F 328 = 5.56, P <.1). However, two stepwise, multiple regression analyses, with sirs debt and prior TST as predictors of SWS change, showed that prior TST added <1% to the variance predicted by SWS debt (82% to 82.3%) whereas SWS debt added almost 6% to the variance predicted by prior TST (25% to 82.3%). Thus, SWS debt is clearly the major predictor of SWS change. The change in SWS from baseline during recovery sleep cannot be related to the amount of prior wakefulness as this was held constant throughout. The means in Table 1 indicate a 6 to 8% reclamation of the REM debt during recovery sleep. At first sight this would appear to suggest a possible homeostatic response by the sleep system to the REM deprivation inflicted by the restricted sleep and nap regimen. However, the amount of extra REM sleep obtained above baseline levels during recovery is not related to the amount of REM debt. Across conditions, the REM debt accounts for < 1 % of the variance of the increase in REM above baseline during recovery sleep. Thus, there is no evidence to suggest that REM sleep is under homeostatic control. Instead, the rise in REM sleep during recovery sleep appears to be related to the SWS debt (r = -.36, p <.5) and the extra TST (r =.32, p <.7). A multiple regression analysis showed that the SWS debt and the extra TST together account for just under 3% of the variance in the extra amount of REM sleep obtained during recovery sleep (F 2,29 = 6.16, P <.1). There is no evidence to suggest that stage 2 sleep or TST are under homeostatic control. Overall, the stage 2 and TST debt only account for 1 and 1.2%, respectively, of the variance in the change in these sleep parameters from baseline during recovery TABLE 2. Relationship (Pearson's r) between amount of sleep stage debt or total sleep time (TST) debt before recovery sleep and the change in sleep stage or TST from baseline during recovery sleepa Nap Stage 2 SWS REM TST 1% -.58 (34).78 b (61).17 (3) -.33 (II) 5% -.24 (5.8).65 (42).19 (4) -.24 (5.8) 25% -.24 (5.8).79 b (62) -.4 (.16) -.39 (IS) No nap -.32 (1).75 b (56).11 (1.2) -.45 (2) Overall -.32 (1).9i" (82).3 (.9) -.11 (1.2) a The percentage of the change variance tbat is predictable from the debt is shown in parentheses. b p <.5 (two-tailed). c p <.1 (two-tailed). Sleep, Vol. 1, No, 6, 1987
64 A. TILLEY ETAL. 6 F t:p.1% Nap en c 5% Nap ~ 4.25% Nap. I o No Nap W <.9 z 2 11 ~ «I FIG. 1. Relationship between!'l () SWS debt and the change in SWS (/) from baseline during recovery s:. ' sleep. (/) -2-4 -2 2 4 6 8 SWS DEBT (mins) sleep (see Table 2). The increase in stage 2 sleep above baseline during recovery sleep can be almost entirely explained by the extra TST, which accounts for 9% of the variance in the rise in stage 2 sleep (r =.95, p <.1). DISCUSSION The present results clearly show that SWS is under homeostatic control. Inflicting a constant SWS debt and then varying the amount of the debt reclaimed during an afternoon nap produces finely tuned homeostatic changes in postnap sleep that recover any outstanding SWS debt. It appears that the sleep system is programmed to obtain a more or less fixed amount of SWS per day, with any shortfall being carried over into the next night's sleep. The SWS debt can be reduced by an afternoon nap. The present results provide no evidence to suggest that stage 2, REM, or TST are under homeostatic control. Interestingly, despite the fact that the restricted sleep and nap regimen inflicted a heavy loss of REM sleep, the rise in REM sleep above baseline during recovery sleep is not related to the amount of REM debt but to the SWS debt and the extra amount of TST. In general, the greater the SWS debt the smaller the rise in REM sleep, and the greater the extra amount of TST the greater the rise in REM sleep. The negative relationship between the SWS debt and REM is probably due to the priority given by the sleep system to procuring SWS and recovering the SWS debt. The positive relationship between extra TST and the rise in REM is probably due to the extended sleep time. REM sleep typically achieves its highest levels in the second part of the night. Allowing sleep to continue beyond a normal amount extends the period of sleep in which REM normally predominates, and hence the rise in REM sleep. The increase in stage 2 sleep above baseline during recovery sleep is also largely due to the extra sleep time. Although the function of sleep is still unclear, Borbely and colleagues (4,17-19) have recently proposed a model of sleep regulation in which SWS is seen as the electrophysiological correlate of a sleep-dependent neurobiochemical process, labeled process S, the level of which is assumed to increase during wakefulness and to exponentially decline during sleep. We also believe that SWS is the electrophysiological accompaniment of the neurobiochemical process controlling sleep. However, this is not to say that the function of sleep is necessarily one of restoration or detoxification, as is com-
HOMEOSTATIC CHANGES IN SWS 65 monly supposed and implied by the above model. Sleep may have essentially a nonrestorative role, for example, behavioral adaptation to the environment (2), that requires a neurobiochemical process to promote and subserve the state of sleep rather than the process being the raison d'etre for sleep. Acknowledgment: We thank Dr. Roderick Ashton, Dr. John Bain, Professor Borris Crassini, and Dr. Ken White for their helpful comments and suggestions. This research was supported by grants from the University of Queensland and the Australian Research Grants Scheme. REFERENCES 1. Agnew HW, Webb, WB, Williams RL. The effects of stage 4 sleep deprivation. Electroenceph Clin Neurophysiol 1964;17:68-7. 2. Williams HL, Hammack JT, Daly RL, Dement WC, Lubin A. Responses to auditory stimulation, sleep loss and the EEG stages of sleep. Electroenceph Clin NeurophysioI1964;16:269-79. 3. Berger RJ, Walker JM, Scott TD, Manguson LF, Pollack SL. Diurnal and nocturnal sleep stage patterns following sleep deprivation. Psychonomic Science 1971 ;23:273-5. 4. Borb ly AA, Baumann F, Brandeis D, Strauch I, Lehmann D. Sleep deprivation: Effect on sleep stages and EEG density in man. Electroenceph Clin NeurophysioI1981;51:483-93. 5. Webb WB, Agnew HW. Sleep: Effects of a restricted regime. Science 1965;15: 1745-7. 6. Dement WC, Greenberg S. Changes in total amount of stage four sleep as a function of partial sleep deprivation. Electroenceph CUn NeurophysioI1966;2:523-6. 7. Webb WB, Agnew HW. The effects of a chronic limitation of sleep length. Psychophysiology 1974;11:265-74. 8. Taub 1M, Berger Rl. The effects of changing the phase and duration of sleep. J Exper Psychology Human Percept Perf 1976;2:3-41. 9. Mullaney DJ, Johnson LC, Naitoh P, Friedman JK, Globus GG. Sleep during and after gradual sleep reduction. Psychophysiology 1977;14:237-44. 1. Carskadon MA, Harvey K, Dement WC. Acute restriction of nocturnal sleep in children. Percept Motor Skills 1981 ;53: 13-12. 11. Tilley AJ, Wilkinson RT. The effects of a restricted sleep regime on the composition of sleep and on performance. Psychophysiology 1984;21 :46-12. 12. Tilley AJ. sleep at different times of the night following loss of the last four hours of sleep. Sleep 1985;8: 129-36. 13. Karacan I, Williams RL, Finley WW, Hursch CJ. The effects of naps on nocturnal sleep: influence on the need for stage-1 REM and stage 4 sleep. Biological Psychiatry 197;2:391-9. 14. Moses J, Hord DJ, Lubin A, Johnson LC. Naitoh P. Dynamics of nap sleep during a 4 hour period. Electroenceph Clin NeurophysioI1975;39:627-33. 15. Feinberg I, March JD, Floyd TC, Jimison R, Bossom-Demitrack L, Katz PH. Homeostatic changes during post-nap sleep maintain baseline levels of delta EEG. Electroenceph Clin Neurophysiol 1985;61:134-7. 16. Rechtschaffen A, Kales A, eds. A manual of standardized terminology, techniques, and scoring systems for sleep stages of human subjects. Washington, D.C.: U.S. Government Printing Office, 1968. 17. Borb ly AA. The sleep process: Circadian and homeostatic aspects. In: Obal F, Benedek G, eds. Environmental Physiology Vol. 18. Oxford: Pergamon Press, 1981:85-91. 18. Borb ly AA. Sleep regulation: Circadian rhythm and homeostasis. In: Ganten D, Pfaff D, eds. Sleep: Clinical and experimental aspects. Current topics in neuroendocrinology, Vol. 1. Berlin: Springer Verlag, 1982:83-13. 19. Borb ly AA. A two-process model of sleep regulation. Human Neurobiology 1982;1:195-24. 2. Meddis R. The sleep instinct. London: Routledge and Kegan Paul, 1977.