171 EFFECT OF SLEEP DEPRIVATION ON SLEEP ... - ScienceDirect

BehaviouralBrainResearch, 14 (1984) 171-182

171

Elsevier BBR00409

EFFECT OF SLEEP DEPRIVATION ON SLEEP AND EEG POWER SPECTRA IN THE RAT

ALEXANDER A. BORBI~LY, IRENE TOBLER and MEHMET HANAGASIOGLU

Institute of Pharmacology, University of Ziirich, 8006 Ziirich (Switzerland) (Received July 18th, 1984) (Revised version received October 3rd, 1984) (Accepted October 9th, 1984)

Key words: sleep - sleep deprivation - EEG spectral analysis - sleep EEG - rat

EEG power spectra of the rat were computed for consecutive 4-s epochs of the daily light period and matched with the scores of the vigilance states. Sleep was characterized by a progressive decline of low frequency spectral values (i.e. slow wave activity) in non-rapid eye movement (non-REM) sleep, and a progressive increase in the amount of REM sleep. During recovery from 24-h total sleep deprivation (TSD) the following changes were observed: an increase of slow wave activity in non REM sleep with a persisting declining trend; an enhancement of theta activity (7.25-10.0 Hz) both in REM sleep and waking; a decrease of non-REM sleep and an increase of REM sleep. In addition, a slow wave EEG pattern prevailed in the awake and behaving animal during the initial recovery period. In selective sleep deprivation paradigms, either REM sleep or slow wave activity in non-REM sleep was prevented during a 2-h period following upon 24-h TSD. During both procedures, non-REM sleep spectra in the lowest frequency band showed no increase. There was no evidence for a further enhancement of slow wave activity after its selective deprivation. The results indicate that: (1) slow wave activity in non-REM sleep and theta activity in REM sleep may reflect sleep intensity; and (2) REM sleep and active waking, the two states with dominant theta activity, may be functionally related.

INTRODUCTION

The sleep-wake cycle is a self-regulating process. Extended waking enhances sleep propensity, while extended sleep raises the likelihood of waking. The physiological processes underlying sleep regulation may be reflected by EEG parameters. We have previously argued that EEG slow wave activity in the rat may indicate the level of 'sleep pressure', since it is enhanced as a function of prior wake time 5. This conclusion was based on data obtained by EEG zero-crossing analysis, a technique that lends itself readily to the quantification of large amounts of data, but does not allow a precise measurement of the EEG frequency distribution. This limitation can be overcome by the use of spectral analysis which yields a power density measure for each frequency band. Moreover, computing spectra for a large number

of consecutive epochs allows the analyzation of the EEG changes as a function of time. All-night spectral analysis of the human sleep EEG has been successfully applied in physiological6, pharmacological7 and psychiatric studies s where it revealed effects that could not be recognized by conventional sleep scoring procedures. EEG data derived from spectral analysis were also used to define a sleep process within a model of sleep regulation 3,9,1°. While spectra of the rat EEG have been computed in several studies 5,~6,21, they have not been analyzed in detail in terms of their frequency distribution and time course. The principal aim of the present study was to obtain such data for baseline conditions and for recovery from sleep deprivation, and to examine whether spectral EEG parameters could serve as indicators of sleep regulatory processes.

Correspondence: A. Borbtly, Pharmakologisches Institut der Universit~t, Gloriastrasse 32, CH-8006 Z~irich, Switzerland. 0166-4328/84/$03.00 © 1984 Elsevier Science Publishers B.V.

172 METHODS AND EXPERIMENTALPROCEDURE

Animals and electrode implantation Adult male Sprague-Dawley-Ivanovas rats (SIV-50) with an initial mean body weight of 293 g (range 268-317 g) were used. The animals were "housed individually in transparent Plexiglas cages with unlimited access to food and water, and were maintainedi~'on a 12-h light/12-h dark schedule (light from 09.00 to 21.00 h) at an ambient temperature of ca. 24 ° C. Implantation of electrodes was performed under pentobarbital anesthesia. The cortical electroencephalogram (EEG) was recorded epiduraUy between the parietal (3.5 P, 1.0 L relative to the bregma) and cerebellar cortex (10.0 P, 0.0 L) with gold-plated, round-tipped miniature screw electrodes in the skull; the electromyogram (EMG) between two gold wire electrodes (0.2 mm diameter) placed upon the occipital skull beneath the neck muscles.

Experimental protocol To allow sufficient time for recovery and adaptation to the recording cables, the experiments were started not earlier than 7 days after surgery. All recordings were made during the first 8-11 h of the 12-h-light period. First, a baseline record was obtained. The animals were then subjected to the following sleep deprivation (SD) procedures: (1) total sleep deprivation (TSD) for 24 h (14 animals); (2) 24-h TSD followed by 2-h rapid eye movement (REM) sleep deprivation (TSD + RSD) (7 animals); (3) 24-h TSD followed by 2-h slow wave sleep deprivation (TSD + SWSD) (8 animals). Sleep deprivation was always started at light onset, and at least 3 days elapsed between the onset of consecutive deprivation experiments. Seven rats underwent only TSD, 3 rats only TSD + SWSD, while in 7 further animals two (TSD and TSD + RSD; n = 1) or all 3 (n = 6) deprivation procedures were applied.

Sleep deprivationprocedure TSD was carried out by placing the animal for 24 h into a slowly rotating cylinder (one revolution per 45 s)5. The consecutive 2-h RSD or SWSD was performed in the recording cage on the basis

of the polygraph record. For RSD the animal was awakened whenever unambiguous REM sleep periods appeared (i.e. a reduced amplitude and theta activity in the EEG; low EMG). For SWSD the animal was mechanically stimulated whenever a continuous non-REM sleep period lasted for 20 s. The stimulus usually caused a cortical desynchronization without a behavioral awakening.

Spectral analysis of the EEG and the scoring of vigilance states The EEG and EMG were recorded by the polygraph (paper speed 3 mm/s) as well as on analog magnetic tape (Hewlett Packard, Model 3968; tape speed 1.19 cm/s). The data were played back from tape, passed through the AD converter (25 Hz low pass filtering at 6 dB/octave; sampling rate 64/s) ofa PDP 11/20 computer and subjected to spectral analysis 6 (FFT routine, DEC Laboratory Subroutine Package). Power density values (~VZ/0.25 Hz) were computed for consecutive 4-s epochs in the frequency range of 0.25-25.0 Hz (0.5-Hz bins between 0.25 and 5.0 Hz, 1-Hz bins between 5.25 and 25.0 Hz). A software routine was applied to offset the attenuation of the EEG in the lowest frequency range due to the frequency characteristics of the recording amplifier. Spectra from epochs with EEG artifacts were eliminated. The EMG was also AD converted, full-wave rectified and integrated to yield a single value for each 4-s epoch. The vigilance states waking, non-REM sleep and REM sleep were scored by a semi-automatic procedure for 4-s epochs to match the sleep scores with the spectral values. The state identification was based on the level of EEG slow wave activity (i.e. the power density in the delta range of 0.25-4.0 Hz) and on the level of the integrated EMG. For both parameters and for each animal, a threshold value was defined to serve as a decision criterion. Thus non-REM sleep was scored whenever suprathreshold slow wave activity occurred together with a subthreshold EMG level; REM sleep when both parameters were below threshold; and waking when subthreshold slow wave activity occurred with a suprathreshold EMG level. The output of the state identification program was carefully compared to the polygraph

173 record as well as to notes describing behavior, and, when necessary, corrected. While the hourly mean values in Fig. 3 are based on the state identification of all 4-s epochs, in the 2-h mean values presented in the other figures (Figs. 2, 5 and 8) the epochs with artifactual EEG spectra were omitted. Slight discrepancies of the results are due to these differences in the computation procedures. RESULTS

Baseline

Fig. 1 shows the frequency distribution of the EEG power density for the 3 vigilance states during the In'st 8 h of the baseline light period. The

3 ' 0 /q /~ / I 2.5~ I

A >. I--

2.0

Z U4 Lg

~ ~\

t "'A.

1.5

"""

.......

WAKING

.... ~

N.EMS REMS

N, \

o= o=

°!?I !;11)

0.5

0.0

.

0

.

.

5

.

.

10

15

20

25 (Hz)

N-R

W-R W-N

Fig. 1. Frequency distribution of the EEG power density during waking (W), non-REM sleep (N) and REM sleep (R). The curves connect the mean values (14 animals) obtained for the first 8 h of the baseline light period, and computed for 0.5-Hz bins between 0.25 and 5.0 Hz, and for 1-Hz bins between 5.25 and 25.0 Hz (the values are plotted at the upper limit of each bin). The power densities were standardized for each animal by defining the mean value during sleep (i.e. non-REM sleep and REM sleep) as 100%. Thus the curves represent relative power densities which are plotted on a logarithmic scale. Lines below the abscissa indicate the frequency bands in which the spectral values of two vigilance states differ significantly (P < 0.001; two-sided, paired Wilcoxon test). The vertical lines at 2-Hz intervals serve as visual aids.

mean power density during sleep computed for the entire frequency range was def'med as 100% and all values were expressed relative to this reference value. This procedure ensured that each animal contributed equally to the mean. It should be noted that the power density values cannot be affected by variations in the amount of vigilance states, since they represent the spectral power per unit of time. Lines below the abscissa indicate the frequencies in which the spectral values of two vigilance states differed significantly (P < 0.001). In non-REM sleep the power density values in the low frequency range (0.25-6.0 Hz) exceeded those of REM sleep and waking by one order of magnitude. Slow wave activity in waking was larger than in REM sleep, a difference that may be due to the fact that drowsy episodes with a high muscle tone were scored as waking. The prominent theta activity during REM sleep and waking is reflected by the peak in the 6.25-9.0 Hz range. The lower values for waking are largely due to the fact that theta waves are not consistently present in the awake animal. It should be noted that the theta level was considerably high also in nonREM sleep where a significant difference from REM sleep was present only in the 7.25-8.0 Hz range. At frequencies beyond 9 Hz the non-REM sleep curve was significantly higher than the two other curves. Fig. 2 shows for consecutive 2-h baseline periods the spectral frequency distribution for each vigilance state as well as the amount of the vigilance state (bars on the right). The values for hours 2-4, 4-6 and 6-8 have been expressed relative to the value of the initial 2-h period (-- 100~o). This type of data presentation was chosen to visualize trends across the entire recording period. For waking and REM sleep the power density values deviated only rarely by more than 10~o from the reference level and did not exhibit systematic trends over the recording period. By contrast, a marked and consistent decreasing trend was seen for the slow wave activity of non-REM sleep. These changes were not limited to the delta range, but extended up to a frequency of 7 Hz. At higher frequencies no comparable trends were observed. The amount ofnon-REM sleep showed

174

VIGILANCE

POWER DENSITY (% O F

HOURS

(% O F

0-2)

HOURS

0-2)

_

100

HOURS -4

WAKING

120-

STATES

2

4-6

/--A\

"~ "~

/ ,/ \ \ A v ~ . ~ . ' /'/-/ ./_',. _ ~ . ^ / / ~ ' Z ~

100

"'" . ~'\

.

.

.

.

.

.

.

.

.

/ " " " /- ~ . . . . . . . ,,_,. _ . , , / _ ~ - , . _ ; " ~,'~--------- ":'--" 7 . -. " .-

6-8

.

--

/ ..... . . . . .

---.

~

~

80 I

I

[

1

I

I

II

|

|

I

~I

I

I

I

I

1

,

W

l

I

I

I

'

I

I

REMS

:t

..-o .. ...... ---°--.....o

........... ',

11o-0 I/

100

.", " ' , '/ . .

,'', .,'/.-%, . .

l," / l ,';

/"

-

\ \\ \ - - -- -, """". . . . . . ,, .-'"/ / ~

-',~

-- -- -\C---/ .

.

~

.

T

. ~

.

-..//

]

140

\\ ~ ~ ~ ~ ~ . . .

.

|

100

9O

I,

NREMS

1oo -\

_

I',\ 80

//

',~.

60

/

. . . .

,"

0-2

4-6

6-8

[dOORS

"." ,

0 u~ I uJ n, Z

2-4

5

i

I

10

'

'

'

'

I

15

'

'

i

,

I

20

i

,

i

i

I

25 (Hz)

~o 2 - 4 C~ 4 - 6 w 6-8 o.

Fig. 2. Frequency distribution of relative EEG power density computed for consecutive 2-h periods, and plotted separately for waking, REM sleep and non-REM sleep. The curves connect the mean values (n = 14) obtained for the first 8 h of the baseline light period, and are plotted for 0.5- or 1,0-Hz bins as in Fig. 1. For each animal a~d each vigilance state the mean value of the first 2-h period was defined as 100% (interrupted horizontal line). Thus the values for the subsequent 2-11periods (i.e. hours 2-4, 4-6, and 6-8) are plotted relative to the value for hours 0-2. On the right, the amount of the ~ states is indicated for consecutive 2-h periods and expressed as a percentage of hours 0-2 (= 100%). Lines below the bottom abscissa indicate for consecutive 2-h periods those frequency bands in which the spectral values of non,REM sleep differ s ~ a a t l y from those of hours 0-2 (P < 0.001; see legend of Fig. 1). Asterisks indicate sLm~i~ant differences in the amotmt of vigilance states from the 0-2 13 value (P < 0.002).

175 no significant changes over the 8-h recording period (bars on the right). A progressive increase in the amount of REM sleep was observed (see also top panel of Fig. 3).

BASELINE 100" WAKING

80" F(3

z

60.

0 0

40.

Total sleep deprivation (TSD)

NREMS

TSD had a marked effect on sleep (Fig. 3, lower part; Fig. 5, right part). The mean amount of REM sleep was increased by ca. 100% inthe first 2 h of the recovery period and remained elevated throughout the recording period. This REM sleep rebound was paralleled by a reduction of nonREM sleep which was most prominent in the first 3 h. Sleep latency after TSD was often longer than under baseline condition. This was the main cause for the increased amount of waking in the first recovery hour. However, a non-significant increase of the mean values was present also during later parts of the recording period when non-REM sleep showed a significant reduction. Figs. 4 and 5 illustrate the effect of TSD on the EEG spectra. Fig. 4 shows the vigilance states

I,l,I

20. 0 O'

SLEEP

DEPRIVATION

( ~ OF BASELINE)

22o t 180

WAKING

'4°tl ,ooi L_j 60

1

L

' '

;

'401 NREMS

BASELINE

1°°1 I-4 6o

*

260 q

.

AFTER

24-h

TSD

W NREMS REMS

* Hz

0-4

REMS

14°111111 100

I

'

'

'

~

'

'

'

4-6

1 '

'

'

J

6-7

HOURS

Fig. 3. Distribution of vigilance states during the baseline day (top) and deviation from baseline after 24-h sleep deprivation (lower part). The curves connect mean hourly values (n = 14 for hours 1-8, n = 7 for hours 9-10, n = 6 for hour 11) computed for REM sleep (shaded area) and total sleep (i.e. REM sleep and non-REM sleep), and expressed as percentage of recording time. Thus non-REM sleep is represented by the interval between the curves, and waking by the interval between the top curve and the 100% line. The bars in the lower part represent mean deviations from baseline ( = 100%). Significant differences from baseline are indicated by asterisks (P < 0.001) and filled circles (P < 0.05) above or below the S.E.M. line.

7-8

3

8-9

6

9-15

15-20

2 0 - 25

48

0

2

4

HOURS

6

8

0

2

4

6

8

HOURS

Fig. 4. Vigilance states and EEG power density plots of a rat for a baseline day and for the day after 24-h sleep deprivation. Values are plotted for 1-min epochs for the first 8 h of the light period. The lower limit of each frequency band is 0.25 Hz higher than indicated on the left (e.g. 0-4 Hz corresponds to 0.25-4.00 Hz). The scale factors indicated on the right were chosen to equalize the area of the plots. Power densities are plotted in arbitrary units.

176 POWER (%

340"

A

VIGILANCE

DENSITY

OF

(% OF

BASELINE)

STATES

BASELINE)

WAKING

HOURS 0-2 2-4 4-6

260-

180

..........

6"i_

I

_

180

100

100 ,

,

i

0 3 o O vl o.

I

I

,

i

,

5

I 10

I

i

,

,

I 15

i

,

I

,

I

i

,

i

l

20

I

25 (Hz)

0-2 2-4 4-6 6-8

REMS -- 180

180

u

140

, -

100

100 '

~0 w

0-2 2-4 4-6 6-8

140

'

'

'

0

I 5

'

i

i

,

I 10

~

i

,

I

I 15

'

[

I

i

i

i

I 20

'

I

a

i

I

!

I

25 (Hz)

NREMS

180-

140

100

_ T ~ 100 I

o

w cL

0-2 2-4 4- 6 6-8

I

!

I

i

5

I

I

~

I

i

10

I

I

I

I

15

!

!

I

20

,

,

,

I

25 (Hz)

0-2 2-4 4-6 6-8

HOURS

177 (top) and the power density of the various frequency bands for a baseline day and a TSD recovery day in the same animal. Note that peaks of slow wave activity (0.25-6.0 Hz) occur during non-REM sleep and show a decreasing trend throughout the recording period. The sharp peaks in the theta band (6.25-9.0 Hz) show a close correspondence with REM sleep episodes. Both slow wave and theta activity were enhanced by TSD. Peaks in two higher frequency bands (9.25-15.0 Hz and 15.25-20.0 Hz) which in part reflect the EEG spindle activity, were correlated with non-REM sleep episodes. The power density in these bands was moderately enhanced by TSD. In the highest frequency band (20.25-25.0 Hz) practically no state-dependent changes were seen. Fig. 5 shows the mean values of relative EEG power density on the recovery day after TSD. As in Fig. 2 the frequency distribution is shown for consecutive 2-h periods with the values being expressed relative to the corresponding 2-h baseline period (= 100~). Both the curves and the bars indicate, therefore, the percentagewise deviation from baseline. It is evident that the effect of TSD was different in REM sleep and n o n - R E M sleep. In REM sleep a significant increase in theta activity was seen. The maximum increase occurred in the first 2-h period with the peak value situated in the 8.25-9.0 Hz band. However, the power density value remained elevated throughout the entire 8-h period. The values showing a significant enhancement (P < 0.001; horizontal lines below the abscissa) shifted towards somewhat higher frequencies in the second part of the recording period (hours 4-8). In nonREM sleep the largest increase in power density occurred in the low frequency band with the maximum at 1.75-2.5 Hz. However, in the first 2-h period, values in the entire frequency range were significantly (P < 0.001) enhanced. The in-

•

SLEEP

DEPRIVED

RAT

WAKING

I

't

]

EEG

EMG

NON-REM

EMG

'~ ........

.....

SLEEP

--

:

"

"

"

:: ......

] 1so v v

i 2 Sec

Fig. 6. EEG and EMG recordings of a rat illustrating a waking and a non-REM sleep period within the first hour alter 24-h sleep deprivation. Note the EEG slow wave activity in the awake and behaving animal.

crease of fast activity persisted in subsequent recording periods. Note that the power density curves are expressed relative to the corresponding baseline curves which exhibited a marked decreasing trend (Fig. 2). Consequently, the persistent slow wave peaks in Fig. 5 correspond to a decreasing trend of the absolute power density values. In waking, the spectral curves showed both a slow wave peak and a theta peak that were separated by a trough at 6.25-7.0 Hz. While a significant increase over nearly the entire frequency range was present in the first 2-h period, persistent significant effects were limited to the theta band (7.25-9.0 Hz). The massive enhancement of low frequency activity during the initial 2-h period was due to the occurrence of a typical slow wave EEG in the awake and behaving (e.g. grooming, drinking) animal (Fig. 6). Such episodes were consistently seen in the hour immediately following TSD and then became progressively less frequent.

Fig. 5. Effect of 24-h sleep deprivation on EEG spectra and vigilance states. The frequency distribution of relative EEG power density was computed for consecutive 2-h periods, and plotted separately for waking, REM sleep and non-REM sleep. The curves connect the mean values (n = 14) for the first 8-h light period of the recovery day. For each animal and each vigilance state, the value of the corresponding 2-h baseline period was defined as 100% (interrupted horizontal line). Thus the curves and the bars indicate the deviation from baseline for consecutive 2-h periods. Lines below the abscissae indicate those frequency bands for which the values differ significantly from the corresponding baseline values (P < 0.001; see legends of Figs. 1 and 2). Asterisks indicate significant differences in the amount of vigilance states relative to baseline (P < 0.001).

178

Selective deprivation of REM sleep and slow wave sleep (S WS) Selective deprivation of REM sleep and SWS followed immediately upon a 24-h TSD period. By this procedure 'sleep pressure' during the selective sleep state deprivation period could be kept sufficiently high to avoid the induction of prolonged awake periods by the arousing stimuli. On the other side, due to the high REM sleep pressure, only a partial REM sleep deprivation could be obtained. For SWS deprivation, the arousing stimulus was administered 20 s after sleep onset. This procedure was chosen to interfere with the sleep period in which slow waves are typically most prominent, while avoiding a total deprivation of non-REM sleep. Fig. 7 shows for consecutive 30-min periods the number of arousing stimuli that were required to deprive REM sleep or SWS. During REM sleep deprivation, the number of stimuli increased progressively over the 2-h period, while during SWS deprivation no clear trend was observed. During the REM sleep deprivation period, the amount of REM sleep was reduced to 90.8~o (+ 15.3) of baseline which corresponds to a ca. 5 0 ~ reduction relative to the first 2-h period after TSD (see Fig. 5). During SWS deprivation, there was a non-significant (135.0 ~o) increase of REM sleep relative to baseline. As indicated by the bars on REMS

-

SWS -

DEPRIVATION

DEPRIVATION

~k 40 --

r--,

-- 40

,_L

20

o

L_ 1

2

3

0 1

4

CONSECUTIVE

3O--MIN

2

3

4

PERIODS

Fig. 7. Number of arousing stimuli required for REM sleep deprivation (n = 7) and slow wave sleep deprivation (n = 8), each of which followed the 24-h total sleep deprivation period. Each bar represents mean values and S.EM. for consecutive 30-rain periods. Asterisks indicate significant differences between adjacent values (P < 0.05).

Fig. 8, the amount of non-REM sleep was depressed during both selective deprivation schedules. Both selective sleep deprivation schedules caused a very prominent REM sleep rebound during the 6-h recovery period (values for consecutive 2-h periods: REM sleep deprivation: 266.2~*; 210.2~*; 157.5~o*; SWS deprivation: 243.3~o**; 188.5~**; 146.7~o**; percentage of corresponding baseline value; * P < 0.02; • * P < 0.01) as well as a reduction of non-REM sleep in the first 2-h period (bars in Fig. 8). Fig. 8 shows the effect of TSD and the selective sleep deprivation procedures on the frequency distribution of the relative EEG power density in non-REM sleep. This sleep state was chosen for an illustration by spectral plots, because the changes of slow wave activity were considered to be particularly interesting. The spectral curves obtained for the REM sleep deprivation period itself (dotted curve) showed no significant increase in the low frequency range, but a prominent, significant rise at frequencies above 5.25 Hz. A minor peak was present in the theta band. During the subsequent recovery period, the values were most enhanced in the tow frequency bands. The pattern was essentially the same as after TSD (Fig. 8, top). During SWS deprivation itself, the values below 2.5 Hz were not significantly increased, while those at higher frequencies were enhanced to a similar extent as after TSD (hours 0-2). By contrast, in the first 2-h recovery period only the values in the lowest frequency band (0.25-1.5 Hz) were increased. The slow wave peak broadened somewhat in the subsequent period (hours 2-4). It is evident from Fig. 8 that in comparison to the other deprivation schedules, the slow wave rebound was least prominent after SWS deprivation, whereas the increase at higher frequencies was similar for all 3 conditions. The effect of selective sleep deprivation on EEG spectra in REM sleep is not illustrated. It consisted during the REM sleep deprivation of a significant (P < 0.02) enhancement of the values over almost the entire frequency range (1.5-6.0 Hz and 8.25-25.0 Hz)with a peak value of 257 ~o relative to baseline at 9.25-10.0 Hz. The recovery period after REM sleep deprivation was

179 POWER

DENSITY

IN

NREMS

NREMS

(% OF B A S E L I N E )

(% O F B A S E L I N E ) HOURS -

AFTER

24h

TOTAL

SLEEP

-

0

-

2

DEPRIVATION(TSD)

2-4 4-6

. . . .

180-

140

% 100

i

i

i

I 5

0 5

o. o w

cL

i

i

,

i

I i 10

i

l

,

I i 15

i

~ ,

I 20

'

i

i

,

t

I 25 (Hz)

0-2 2-4

100

(-

4- 6 6-8

0-2

TSD

180-

140

"4" 2h R E M S -

2-4

4-6

°" "'"" """ """"__ . . . . . . . . . • , . " . . . . . . . . . . . . . . . . . . . . . . . . . . .

.."

100

~ i

I

i

0

l 5

i

i

,

i

HOURS

DEPRIVATION

...o"-/"''"°"""""

I

6-8

i ' 10

i

,

,

l ' 15

i

i

~ l 20

'

'

i

I

100

l

2 5 (Hz)

80

DEP. 0-2 w o.

.

60

2-4 4-6

DEP.

1801

TSD

1,o t

4"

2h

J

0

I

I

•.•..-

•%

[

,°.o . . . . . .

I

I

5

I

I

2-4

4-6

HOURS

SWS-DEPRIVATION

• •% /.;•

0-2

I

I

[

10

I

I

I

I

'..........................

I

I

I

I

15

I

I

I

..... . ...........

I

I

20

|

J

2 5 (Hz)

DEP. o