The Sleep need:

Material Information

The Sleep need: sleep deprivation in the rat
Levitt, Robert Alan, 1938-
Publication Date:
Copyright Date:
Physical Description:
vii, 114 leaves. : illus. ; 28 cm.


Subjects / Keywords:
Activity units ( jstor )
Body weight ( jstor )
Circadian rhythm ( jstor )
Electrodes ( jstor )
Rapid eye movement sleep ( jstor )
Rats ( jstor )
Sleep ( jstor )
Sleep deprivation ( jstor )
Treadmills ( jstor )
Ultrasonics ( jstor )
Dissertations, Academic -- Psychology -- UF ( lcsh )
Psychology thesis Ph. D ( lcsh )
Rats ( lcsh )
Sleep ( lcsh )
bibliography ( marcgt )
non-fiction ( marcgt )


Thesis -- University of Florida.
Bibliography: leaves 110-113.
Additional Physical Form:
Also available on World Wide Web
General Note:
Manuscript copy.
General Note:

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
000554633 ( AlephBibNum )
13418880 ( OCLC )
ACX9478 ( NOTIS )


This item has the following downloads:

Full Text







August, 1965


The author wishes to express his gratitude to

Dr. Wilse B. Webb for stimulating the research idea and

to Dr. Bradford N. Bunnell for supervising the conduct

of the research. The author also wishes to acknowledge

the advice and council of Drs. Frederick A. King, Henry

S. Pennypacker, Robert L. King, and James A. Horel.

He also wishes to express his appreciation to his

wife, Phyllis, for her patience and assistance in the

typing of this manuscript. The help of Frank Johnson,

William Walker, Mario Perez, Lester Cohan, Barbara Weise,

and Alan Goldstein in data gathering and scoring is also

gratefully acknowledged.



ACKNOWLEDGMENTS . . . . . . . . . ii

LIST OF TABLES ... .. . . . . . . iv

LIST OF FIGURES . . . . . . . . .. vi

INTRODUCTION . . . . . . . . . 1

RAT . . . . . . . . . . . .13

(ACTIVITY MEASURE) . . . . . . . .. 28


MEASURE) . . . . . . . . . . 62

CYCLE . . . . . . . . . . . 68


APPENDICES . . . . . . . . . . 95

REFERENCES . . . . . . . . . . 110

BIOGRAPHICAL SKETCH ... . . . . . .. 114



Table Page

1. An Activity Scoring System ... .. . 22

2. Agreement Between Individual Samples of
Correlated EEG-Activity Records . . .. 24

3. Averaged Data on Sleep Cycle and EEG-
Activity Correlation from Experiment I . 25

4. Recording Schedule for the Deprivation
Groups in Experiment II . . . . 34

5. Estimated Post-Dextro-Amphetamine Deprivation
Compensation ...... .. . . . . 40

6. Mean Percentage of Daily Sleep Obtained
During Lighted Hours . . . . . .. 43

7. Mean Percentage of Daily Food Intake
Obtained During Dark Hours . . . .. 50

8. Mean Percentage of Daily Water Intake
Obtained During Dark Hours . . . .. 51

9. Recording Schedule for the Deprivation
Groups in Experiment IV . . . . .. 63

10. Estimated Post-Treadmill Deprivation
Compensation . . . . . . . . 66

11. Sleep Cycle Totals Before and After
Deprivation '.. . . . . . . .72

12. Estimated Post-Deprivation Compensation 73


13. Dextro-Amphetamine Induced Deprivation
and Sleep Cycle . . . . . ... .96

14. Dextro-Amphetamine Deprivation and Food
Intake . . . . . . . . ... .97

Table Page

15. Dextro-Amphetamine Deprivation and
Water Intake . . . . . . ... .98

16. Dextro-Amphetamine Deprivation and
Body Weight . . . . . . ... .99

17. Treadmill Induced Deprivation and
Sleep Cycle . . . . . . ... 100

18. Dextro-Amphetamine or Treadmill Induced
Deprivation and Sleep Cycle . . . .. 101

19. Mean Length of Sleep Epoch as a Function
of Sleep Deprivation . . . . ... 102

20. Length, Frequency, and Percentage of
Paradoxical Sleep as a Function of Sleep
Deprivation . . . . . . ... .103


Figure Paeq

1. The Pendulum-type Calibrator . ... . 16

2. The Environmental Enclosure and Ultrasonic
Activity Unit . . . .... .. . . 18

3. The Environmental Enclosure with Calibrator
in Position . ... . . .. . . . 20

4. EEG Recordings from Two Rats . . . .. 29

5. Dextro-Amphetamine Deprivation Experiment:
Daily Sleep Time . . . . . .. 36

6. Dextro-Amphetamine Deprivation Experiment:
Circadian Rhythm . . . .... .. . 37

7. Mean Minutes Sleep Obtained per Subject
During Dextro-Amphetamine Deprivation . 45

8. Minutes Sleep Deprivation per Dextro-
Amphetamine Injection . . . . .. 45

9. Food Intake During Dextro-Amphetamine
Experiment . . . .... .. . . . 47

10. Water Intake During Dextro-Amphetamine
Experiment . . . . .... .. . . 48

11. Body Weight During Dextro-Amphetamine
Experiment . . . .... .. . . . 49

12. Micro-Sleep Prevalence During Treadmill
Deprivation . . . . .. . . 55

13. Micro-Sleep and Behavioral Motionlessness
on the Sleep Deprivation Treadmill ... . 58

14. Treadmill Deprivation Experiment: Daily
Sleep Time . . . . . . . ... .64

15. Treadmill Deprivation Experiment:
Circadian Rhythm . . . . . ... 65


Figure Paqe

16. Dextro-Amphetamine or Treadmill
Deprivation and Sleep Cycle: Waking,
Paradoxical Sleep, and Normal Sleep . . 74

17. Post-Deprivation Sleep-Waking Cycle as a
Percentage Change from Control Values . 77

18. Length of Sleep Epochs . . . . .. 79

19. Length of Paradoxical Sleep Bursts . .79

20. Minutes of Slow Wave Sleep Separating
Paradoxical Sleep Bursts . . . ... .81

21. Percentage of Total Sleep Time Spent in
Paradoxical Sleep . . . . . ... .81

22. Minutes into Sleep Epoch When First Minute
of Paradoxical Sleep Occurs . . . . 82

23. Difference Between Observed and Expected
Awakenings from Paradoxical Sleep . . 82



Using as our criterion the time spent in consummatory

activity, sleep is by far the most important of man's

activities (Cofer & Appley, 1964). And yet, although sleep

is presently receiving a great deal of experimental atten-

tion, little is known of its biological function. Sir John

Eccles, in opening an international symposium on sleep, put

it this way. "When we consider the immense human signifi-

cance of sleep, the absolute necessity for us to spend a

considerable part of our lives in abject mental annihilation,

it is remarkable how little we know about it, how little

we can say to account for the necessity of sleep" (Eccles,

1960, p. 1).

The experiments to be reported in this paper are a

beginning on an analysis of the necessity for sleep. The

method utilized is to produce a heightened need through

deprivation of sleep and to study the changes in behavior

following this deprivation. The extirpation of an organ

has long been a recognized method of determining function.

Depriving the organism of a particular bodily activity

(sleep) is the functional counterpart (Kleitman, 1963).

The dependent variable in these experiments is the

characteristics of the sleep cycle as measured electro-

physiologically or by means of an activity measuring unit.

The basic assumption underlying this research is that the

compensatory changes following sleep deprivation are part

of the homeostatic bodily mechanisms which function during

sleep. A number of investigators have suggested the

existence of stages or levels of sleep. It may be that

these various stages of sleep subserve different bodily



The first major human electroencephalographic (EEG)

study of sleep was made by Loomis, Harvey, and Hobart

(1937). They classified the EEG into five stages, A to

E, of increasing sleep depth. Other investigators have

preferred somewhat different classifications (Kleitman,

1963; Oswald, 1962). The waking EEG is of a low ampli-

tude and high frequency. As the depth of sleep increases

. ,(from stages A through E or 1 through 4), the amplitude

increases and the frequency is reduced. Although most

-investigators have assumed that the function of-sleep is

accomplished mostly at increasing depths, at the present time

no separate functions can be attributed to the different

stages. The most that can be said for the function of sleep

is that investigators tend to assume a restorative one,

allowing the recovery from deficits produced by waking

activity (Kleitman, 1963; Oswald, 1962). However, new

evidence has suggested a specific function for at least

one stage of sleep.

Dreaming.--In recent years attention has been directed

to a stage of sleep designated the rapid eye movement phase

(REMP). This stage has also been called paradoxical, acti-

vated, rhinencephalic, or low voltage fast sleep (LVFS).

These terms will be used interchangeably in this paper.

REM sleep was first described in human subjects by Aserinsky

and Kleitman in 1953. Since then, an analogous stage of

sleep has been described in a number of other species,

including the cat (Dement, 1958) and the rat (Hall, 1963;

Swisher, 1962)

When Aserinsky and Kleitman (1953) awoke subjects from

the different stages of sleep including the REMP, they made

the exciting discovery that dreams were reported about 80

per cent of the time when the awakening interrupted REM

sleep but only about 20 per cent when the awakening inter-

rupted other sleep stages. This discovery led to a number

of studies exploring this relationship (Dement, 1955;

Dement & Kleitman, -1957a). Dement (1965), upon reviewing

the evidence on dreaming, has concluded that although the

probability of dream recall and by inference the probability

of dreaming is much higher in REM than non-REM (NREM) .sleep,

dreams do occur during NREM sleep. However, Dement has

suggested that REM and NREM dreams differ in their essential

- nature with only REM dreams consisting of perceptual experi-

ences (Class I experiencing). He has theorized that, "during

the REM phase of sleep the brain is somehow generating the

neurophysiological background for Class I experiencing"

(Dement, 1965, p. 203). NREM dreams consist only of Class

II experiencing (abstract thought, imagery) not dependent

on sensory input. Thus, during REM dreams the brain is

generating its own sensory input. Dement (1965) and Jouvet

(1960) have suggested the existence of two qualitatively

different phases of sleep.

Sleep phases.--Jouvet (1960; Jouvet & Jouvet 1963)

has demonstrated that the paradoxical phase of sleep is

triggered by an area in the caudal brain stem. He has

called this phase rhombencephalic sleep to differentiate it

from slow wave sleep which"'he believes is telencephalic in

origin. The paradoxical nature of rhombencephalic sleep is

that the EEG is one of waking or light sleep, but the depth

of sleep as determined by arousal threshold is deeper than

during any other stage (Dement & Kleitman, 1957b; Dillon

& Webb, in press).

Both Dement (1965) and Jouvet (1960) have concluded

that there are two entirely distinct phases of sleep. One

of these phases is the REM stage and the other is all the

slow wave sleep that precedes it. Dement (1965) has

suggested that physiological variations within NREM sleep

are essentially quantitative and do not warrant further

subdivision. These two phases may be expected to subserve

differing functions and.may respond independently to

experimental manipulations.

The measurement of sleep

Until now there have been two means of measuring the

sleep cycle, behavioral observation and the electroencephalo-

graph. Both of these methods have severe disadvantages.

Behavioral observation requires the continuing presence of

an observer to make a subjective judgment of whether the

subject (S) is asleep or awake. The experimenter's very

presence is likely to disturb the sleep cycle and also means

a great expense in man hours. For this expense only a

subjective judgment of sleep or waking is produced.

The EEG surmounts two of these difficulties; levels

of sleep are discriminable and the judgment is objective.

However, these advantages are obtained at considerable

expense. An observer must constantly be present to monitor

the equipment, and electrodes need to be attached to the

subject. In the case of the rat, the animal used in these

studies, the recording electrodes usually are surgically

implanted under the skull. These electrodes are likely

to alter somewhat the nature of the response being measured.

This large expense in observer and equipment severely limits

the length of EEG recording that it is practical to

obtain. As an alternative, the first experiment reported

here presents an activity measuring device for recording

sleep and waking. This method has the advantage that it

is inexpensive, relative to the EEG, and does not require

the continuing presence of an experimenter. This activity

system may be operated continually for long periods of

time. Since the activity system has the disadvantage of

being unable to discriminate phases of sleep, both activity

and EEG recording were used in these experiments.

The production of sleep deprivation

Another difficulty which had to be surmounted was that

of producing sleep deprivation. Licklider and Bunch (1946)

were unable to keep rats awake on beds of nails, but thought

they could successfully maintain wakefulness by forcing the

Ss to walk a treadmill partially submerged in water. The

rats died after 3 to 14 days on the treadmill probably, at

least partially, from fighting with each other (the rats

were not in separate compartments). Evidence to be reported

in this paper raises serious questions as to the effective-

ness of the treadmill as a method of producing sleep depri-


Svorad and Novikova (1960) have reported successfully

producing sleep deprivation lasting seven days in rats by

periodically administering dextro-amphetamine. The authors

do not state their criterion of sleep deprivation, nor do

they report data on the sleep cycle in their Ss. This

procedure was part of a study on the effect of sleep

deprivation on an experimentally induced neurosis in the

rat. In a preliminary study the present author was

successful in replicating the finding of Svorad and

Novikova that sleep deprivation could be produced in rats

with dextro-amphetamine.

According to Beckman (1961) dextro-amphetamine is

used clinically, principally as a stimulant of the central

nervous system and for its anorexigenic effect in the

treatment of obesity. Bradley and Elkes (1957) have found

a correlated EEG desynchronization and behavioral alertness

to follow systemic dextro-amphetamine administration. This

arousal was like that following electrical reticular forma-

tion stimulation and was dependent on an intact mesencephalon

for its occurrence, implicating receptors in the ascending

reticular activating system.

Williams, Lubin, and Goodnow (1959) have studied the

effects of sleep loss in humans on performance at skilled

tasks. They felt the impaired performance was due to the

occurrence of brief sleep episodes in sleep deprived

subjects during performance. This interpretation is sup-

ported by the experiments of Kornetsky, Mirsky, Kessler,

and Dorff (1959), who found that dextro-amphetamine,

which stimulates the reticular formation, greatly reduces

performance impairment in sleep deprived Ss.


The results of sleep deprivation

The only report of a change in sleep behavior follow-

ing sleep deprivation in the rat is that of Webb (1957),

who found that rats sleep deprived on a water-immersed

treadmill had shorter "sleep latencies" (time to fall

asleep) than during control tests.

Kleitman (1963) and Oswald (1962) have reported an

increase in sleep time following prolonged sleep deprivation

in humans. This increase is not equal to the amount of sleep

lost, but amounts to about 11 to 14 hours sleep the first

night following up to 65 hours sleep deprivation. Berger

and Oswald (1962) and Williams, Hammack, Daly, Dement, and

Lubin (1963) have reported data on changes in the EEG stages

of sleep following deprivation. Both studies reported an

increase in stage 4 on the first recovery night, followed

by an increase in REM sleep on the second recovery night.

The authors of these articles infer that these rebound

effects reflect an underlying "need" state of the organism.

Webb and Agnew (1965), in an experiment on continued partial

sleep deprivation in the human, found an increase in stage

4 sleep. This increase was at the expense, primarily, of

stage 3 with essentially no change in REM or stage 2 sleep

as a percentage of total sleep. This result is probably

due to the Ss being allowed only three hours sleep per

night. Since the stage 4 "need" apparently takes precedence

over the REM "need," the Ss did not have time to compensate

for the REM loss.

Oswald (1962) has suggested that there may not be

a fundamental requirement for adult humans to spend 8 out

of every 24 hours asleep, since deprived subjects make up

only a fraction of the deficit on the first night. It

would seem this question cannot be answered at the present,

since compensatory changes in sleep behavior may continue

to occur for a number of days following deprivation. However,

it may be that there is a difference in the "value" of the

various stages or phases of sleep in alleviating a large

sleep debt, and that following sleep deprivation the "most

valuable" kind of sleep will occur in greater quantity than


If, in fact, as Dement and Jouvet have suggested there

are two distinct sleep phases differing in function, then

this assumption of a change in sleep characteristics follow-

ing total sleep deprivation is reasonable. There are a

number of reports of deprivation of one or another stage

of sleep. Dement (1960) has reported that REM (dream)

deprivation by awakening human subjects every time they

enter REM sleep for five consecutive nights results in the

need for an increased number of awakenings on successive

nights in order to prevent occurrences of REM sleep. He

also found a compensatory increase in REM time following

deprivation. Dement also-reported REM deprivation to

produce "personality disturbances" in the subjects, to

a degree that some subjects were forced to terminate this

part of the experiment. These personality disturbances

occurred in spite of a normal, or near normal, level of

total sleep and did not occur when, in a control study,

the same Ss were awakened only from NREM sleep.

These findings have recently been replicated by

Siegel and Gordon (1965) in the cat and Khazan and Sawyer

(1963) in the rabbit. Siegel and Gordon's Ss required an

increased number of awakenings on successive days of depri-

vation in order to prevent paradoxical sleep, and compensated

for the deprivation with an increased amount of paradoxical

sleep after deprivation was discontinued. Siegel and Gordon

recorded EEGs for 10 to 12 hours a day during a so-called

"sleep period." Paradoxical sleep deprivation during this

sleep period was produced by electrical stimulation of the

reticular formation each time the S entered paradoxical

sleep. During the other 12 to 14 hours per day, the authors

report keeping the cats awake by placing them on a brick

in the middle of a pan of water, which was the floor of the

cage. Khazan and Sawyer (1963) discovered that a constant

loud noise (80 decibels) would almost obliterate paradoxical

sleep without influencing the amount of slow wave sleep.

Following 20 hours of paradoxical sleep deprivation, a

reboun- increase occurred. After 20 hours, the paradoxical

L ir]hibi-ion produce by t. vhiteoie tended to

a ,- t, C'

Agnew, Webb, and Williams (1964) have reported a

compensatory increase in stage 4 sleep following depriva-

tion of this sleep stage in humans. Current sleep research

is centered on the importance and function of these two

stages of sleep, REM and stage 4.

In order to answer the questions of (a) the length

of time that compensatory changes continue following total

sleep deprivation, (b) the amount of total compensation

that occurs, and (c) the possibility of changes in the

phase characteristics of sleep following total sleep depri-

vation, the following experiments were undertaken.

The first experiment details the design and develop-

ment of an activity system to record the sleep cycle in

small mammals. Experiment II studies the effect of varying

lengths of dextro-amphetamine induced sleep deprivation on

the sleep cycle measured by the ultrasonic activity units.

Experimrent III is concerned with the character of the ELG

cf rats walking on The w.ater-immersed treadmill. Previous

behavioral observation had suggested that rats may get

-. 3 sleep on the treadmill, and this observation was con-

fircmdc in Experimc. III. In spite of this finding, in

Experiment IV sleep cycle was measured before and after

t., _!,:ill deprivation using the activity units. It was

thought that although the Ss did obtain some sleep, the

amount was less than normal, and this might be evident in

post-deprivation compensation. In Experiment V sleep

cycle was measured before and after dextro-amphetamine or

treadmill deprivation, but the EEG was used instead of

the activity units. Although it is not now considered

reasonable to record EEG for the lengths of time we use

the activity units, it was possible to get 24-hours' data

before and after deprivation. Although these data are not

as stable or reliable as the activity data, the EEG does

.ermit the differentiation of paradoxical from slow wave


These experiments have required lengthy normative

sleep cycle recording and enable the detailing of the

normal parameters of sleep in our subject population.


Dillon (1963) used an ultrasonic recording device

developed by Peacock and Williams (1962) to record sleep-

ing and waking in the rat. Higgins (1964) has built a

transistorized version of the Peacock and Williams unit.

This experiment presents the methodology for an ultrasonic

activity measure of sleep. The system is a modification

of Dillon's. A simple scoring system has been developed,

a pendulum-type calibrator for adjusting sensitivity has

been designed and tested, and permanent enclosures for

the system have been built. Using these modifications

a series of correlated EEG-activity recordings were




Six male Long-Evans hooded rats were used. Three

were 90-100 and three 180-200 days old. They were maintained

on ad-lib. food and water throughout the experiment.


Sleep and waking were determined by a Grass-III-D

EEG unit. Behavioral observations were periodically

recorded on the EEG record. Each minute was scored for

waking, slow wave sleep, or paradoxical sleep. Whichever

of these was most prevalent was the condition assigned to

that minute.

Activity was measured with an ultrasonic device

developed by Peacock and Williams (1962) or a transistorized

version built by Alton Electronics (Higgins, 1964). These

activity devices are intended for use in detecting the

movement of small mammals. The devices are usable within

a one-cubic yard area. The Peacock and Williams unit con-

sists of a transmitting transducer, receiving transducer,

power supply, and readout unit. The readout drives an

ink writing recorder. In the Alton units the power supply

drives the recorder directly.

The principle of operation is as follows. A 40-

kilocycle sine wave signal is generated by the transmitter

and radiated as ultrasonic sound (nonaudible) into the test

volume by the transmitting transducer. These sound waves

permeate the experimental volume, being reflected from its

walls and from objects within the experimental volume.

At the receiver a transducer picks up the sound waves,

converting'them back into electrical signals which are then

amplified. Any motion within the experimental volume into

which the sound wave is directed produces disturbances in

-the received portion of the wave, causing the receiver to

produce electrical pulses. These electrical pulses may

be recorded directly on a strip chart recorder (Alton

unit) or (Peacock and Williams unit) used to operate a

relay which in turn may operate a recording device (Higgins,


The ink writing recorder is an Esterline-Angus Opera-

tion Recorder, Model AW.

The magnitude of movement which causes a pulse can

be adjusted through a broad range of behavior by manipula-

ting a sensitivity control.


The activity units have a sensitivity adjustment for

controlling the level of activity that will activate the

system. A pendulum-type calibrator was used to determine

the sensitivity. Figure 1 is a dimensional diagram of this

calibrator. The pendulum was manually held at a given

distance from the vertical and then released. The distance

between the swing adjustment and pendulum screw in these

experiments was 0.20 inch. The time from release until

the activity unit stopped responding to the decreasing

motion was the unit of calibration. A reading was made

.immediately before data collection began, and this reading

was the calibration setting. A goal of this experiment was

to determine the range of sensitivity settings at which an

1 '

Figure 1. The Pendulum-type Calibrator.





Y P t

acceptable correlation between EEG and activity is


Recording procedure

Care must be taken to ensure that activity and EEG

records correspond when both are being recorded. This

means that sections of data from these two methods should

be within 1 to 2 seconds of each other. Dillon (1963) has

described a simple method to ensure this agreement; the

system mainly involves using a stop watch to turn one

system on an exactly known time after the other.

Environmental conditions

For EEG-activity recording the animal cage was placed

inside an electrically shielded room. The cages were 9 in.

high, 11 in. wide, and 16 in. long. The cage bottom was

1 in. above the floor allowing droppings to fall from the

cage. All six sides were of wire mesh, permitting the

ultrasonic waves to pass through. The cages were individually

housed within a wall-board enclosure measuring 32-1/2 in.

long, 14 in. wide, and 16 in. high. This enclosure was

designed to bounce the waves throughout the cage and elimi-

nate "dead" spots in the corners and also to prevent

leakage of ultrasonic sound into neighboring units, causing

interference and interaction. Figure 2 is a dimensional

diagram showing the wall-board enclosure and a Peacock and

Williams activity unit ready for operation. The Alton unit

Figure 2. The Environmental Enclosure and Ultrasonic
Activity Unit.

does not have the activity unit sitting on top of the

enclosure but on the floor behind the transducers. Figure

3 shows the calibrator position in the enclosure during

calibration. The "movement baffle" faces toward the

activity transducers. The cage was removed from the

enclosure for calibration.

The two activity transducers were 3-1/4 in. apart

parallel to and 10 in. from the cage. The activity unit

placement is shown in Figure 2. The activity transducers

were at a level 3 in. above the cage floor. The recorder

was housed inside an "ice chest" to reduce noise.

A light cycle was maintained in the room with lights

on 9:00 A.M. to 9:00 P.M. and off 9:00 P.M. to 9:00 A.M.

The room was air-conditioned with the temperature main-

tained between 66 and 69 degrees. The activity units are

sensitive to temperature and function best at this level.


Gold bipolar dural electrodes were fabricated and

implanted under nembutal anesthesia at least four days

prior to recording. The method is described in detail by

Dillon (1963). Both electrodes were placed on the same

side of the skull, one in the posterior and one in the

frontal area. This electrode placement is important, since

Swisher (1962) has found it difficult to differentiate

paradoxical sleep from waking if symmetrical electrodes are

Figure 3. The Environmental Enclosure with Calibrator
in Position.

used. For EEG recording,electrode leads were lowered

through the wire mesh top of the cage and attached to S.

The leads were suspended from a rubber band to maintain

a constant tension and prevent the rat getting tangled in

the wires.

Scoring procedure

The scoring rules were developed empirically by taking

a correlated sample of 200 minutes of EEG-activity data

and working out a set of rules that produced a high correla-

tion. The scoring system assigns a number to each minute

based on the amount of activity within the four 15-second

periods making up the minute. Table 1 shows the scoring


The point scores for the four 15-second periods making

up a minute are added together, and the final number defines

the activity level for that minute. Using this system a

minute can have a score from 0 to 20. Any minute with a

total point score from 0 to 3 is considered sleep, and any

with a point score of 4 to 20 is waking except that one

minute of 3 between two minutes of 4 or more is scored as

waking, and one minute of 4 between two minutes of 3 or

less is considered sleep. It should be remembered that

paradoxical and slow wave sleep can be separated by the

EEG but not with activity recording.


An Activity Scoring System*

Points Needle excursions (blips)

0 0

1 1-3

2 4-8

3 9-15

4 16-24

5 25 or more

* Per 15-second observation period

Reliability.--The experimenter and an independent

observer both scored the same 150-minute sample of correlated

EEG and activity data. The two scorers worked independently

and scored the EEG and activity records separately. A 95

per cent agreement on EEG and 100 per cent agreement on

activity was obtained when the records were scored on a

minute by minute analysis for sleeping or waking only.

LVFS was scored sleep, since it cannot be differentiated

from sleep in the activity records. The scoring of EEG and

ultrasonic activity records in this experiment was done

separately without referring to the other record.

Results and Discussion

Table 2 presents data on the individual EEG-activity

records. Each sample is between 100 and 110 minutes long.

SFrom the data in Table 2 it is apparent that the EEG-

activity correlation is satisfactory at all settings between

120 and 200 seconds (individual minute agreement ranges

from 83 to 100 per cent and total correspondence from 93

to 100 per cent). The system would seem to function

effectively for male hooded rats at least between 90 and

200 days old. The total correspondence measure may require

explanation. This measure is defined as the percentage agree-

ment between the EEG and activity units for a lengthy sample

of data. For example, upon scoring a 100-minute sample,

part of the errors are of the sleep-active type and part of

the waking-nonactive type. On scoring for the entire sample,

rather than minute by minute, some of these errors will

offset each other, thereby increasing the percentage agree-

ment. This total correspondence measure is important, since

in succeeding experiments minutes of sleep and waking per

6-to 10-hour period is the dependent variable.

Table 3 summarizes the data from this experiment.

Only data from the calibration settings between 120 and 200

seconds are included. The figures on minutes of waking,

sleep, and LVFS are based on the EEG records.


Agreement Between Individual Samples of
Correlated EEG-Activity Records

Mean calibration Minute by minute Total Animal
reading agreement correspondence age

Peacock and Williams Units



90-100 days



Alton Units













Averaged Data on Sleep Cycle and EEG-Activity
Correlation from Experiment I

Peacock and Williams Alton
Units Units

Total minutes recorded 610 708

Minutes waking 394 402
Per cent of time waking 65% 57%
Minutes waking errors 31 26
Per cent of waking minutes
in error 7.9% 6.5%

Minutes sleeping 177 268
Per cent of time sleeping 29% 38%
Minutes sleeping errors 18 18
Per cent of sleeping minutes
in error 10.2% 6.7%

Minutes LVFS 39 38
Per cent of time LVFS 6% 5%
Minutes LVFS errors 4 2
Per cent of LVFS minutes
in error 10.3% 5.3%

Minute by minute agreement* 91% 94%
Total correspondence* 97% 97%

*Based on individual 100-minute samples.

There is an average 97 per cent correspondence for

100-minute samples with both the Peacock and Williams and

Alton activity units. This correlation is sufficiently

high to warrant using ultrasonically measured activity as

a sleep cycle measure. Both units would seem to be equally

accurate. The figures on the percentage of time spent in

sleep, waking, or LVFS are not acceptable as normative data,

since the animals were disturbed by having electrodes

attached immediately before the 100-minute recording sessions,

and, also, occasionally electrodes required manipulation

during the session. These factors would be expected to

increase the amount of waking. However, it is interesting

to note that the percentage of sleep errors and LVFS errors

is very similar.

A number of investigators have found muscle tone to

,be at a minimum during REM sleep in humans (Dement &

Kleitman, 1957b), in cats (Jouvet, 1960), and in rats (Hall,

1963). The incidence of body movement seems to peak just

prior to REM sleep and be at a minimum during the REMP.

In contrast to gross body movements both human and animal

subjects show a maximum of small twitches (fingers, tail,

vibrissae) during the REM. Also, it should be remembered

that depth of sleep as measured by threshold for arousal is

higher during the REMP than slow wave sleep (Dement &

Kleitman, 1957b; Dillon & Webb, in press; Jouvet, 1960).


These factors might suggest an imbalance in activity errors

between slow wave and paradoxical sleep. The lack of this

imbalance can do no more than suggest that a number of

factors are present and perhaps neutralize each other.


In a pilot study the present author confirmed Svorad

and Novikova's (1960) report of obtaining sleep deprivation

with periodic dextro-amphetamine injections. Rats were

kept awake up to six days, obtaining only small amounts of

sleep between the time an injection wore off and the time

this change in behavior was noticed by an observer. Samples

of EEG were taken and these showed desynchronization during

drug action and highly synchronized activity when the drug

was allowed to wear off. Figure 4 is a comparison of EEG

recordings for two rats during these various states of the

sleep cycle. During the drugged state, the Ss appeared

quite active, shaking their heads from side to side inces-

santly, but not moving around the cage much. When startled,

the S would characteristically react violently with a jump,

followed by rapidly running around the cage for a few

seconds then resuming the head shake. Following drug with-

drawal the animals went into what seemed to be a deep sleep,

from which they were difficult to awaken.



Slow Wave


Rat 1



Slow Wave

^roxical vwWYlWWVvvvWviVVVvV

Rat 2 1 SEC. 50p

Figure 4. EEG Recordings from Two Rats.*

Right frontal to right visual bipolar electrodes



rl__ i.

A, 71


It was expected that sleep deprivation would increase

the rats' sleep need and that, as a result, sleep time

during recovery would increase above normative values. Also,

it was thought that this experiment would answer the question

of whether Ss completely make up for missed sleep or com-

pensate by making up only a part of the loss. This experi-

ment was also expected to provide data on the relationship

between length of sleep deprivation and amount of compensatory

sleep obtained.

Measures of food and water intake and body weight were

also made, since the anorexigenic action of dextro-amphetamine

was expected to influence them. These measures were made

both in the morning and evening, since the circadian rhythm

has been found to strongly influence these behaviors. A

circadian rhythm in behavior is an association between the

day-night cycle and the level of the behavior under observa-

tion. Since rats are known to be relatively active at

night and relatively inactive during the day, highest levels

of waking, eating, and drinking were expected during the

night,and body weight should be higher following an evening's

activity than a day's sleep (Munn, 1950).



Twelve male Long-Evans hooded rats 110-120 days old

at the beginning of the experiment were used.


The apparatus was the same as in Experiment I. Both

Alton and Peacock and Williams activity units were used in

this and the succeeding studies. Occasionally a malfunction

developed in an activity unit. This usually consisted of

a unit becoming insensitive to movement, or less frequently,

picking up "random noise" and responding although there was

no movement within the enclosure. Spare units were available

so that a rat could be transferred to an operable unit. The

maximum amount of lost data for an individual animal was

ten hours. Usually at least the first part of the recording

period (2 to 8 hours' data) was usable. If fewer than ten

hours' data were usable, the data were prorated to ten hours

to produce a number which would fit into the data analysis.


The same pendulum-type calibrator as that shown in

Figure 2 was used in all succeeding activity recording

experiments. A sensitivity setting between 120 and 200

seconds was used.

Design and procedures

There were four subjects at each of the deprivation

levels, 24, 72, or 120-hours. The environmental conditions

were the same as in Experiment I except that an electrically

shielded room was not required. The Ss were fed powdered

Purina rat chow and received water from metered glass

bottles. The chow consistency and water bottles were

different from what the rats had experienced in the main

colony. A ten-day adaptation period was utilized during

which Ss were housed in experimental cages in the experi-

mental room and adapted to the new food and water procedures.

The lights were on approximately 9:00 A.M. to 9:00 P.M.

and off 9:00 P.M. to 9:00 A.M. The room was completely

blacked out during the lights off period. Activity record-

ings were made from 10:00 A.M. to 8:00 P.M. and 10:00 P.M.

to 8:00 A.M. During the morning and night two-hour non-

recording periods (8:00 to 10:00 A.M. and 8:00 to 10:00 P.M.),

the activity units were calibrated, the food and rats weighed,

and water intake measured. The food, water, and body weight

measures were made in only 2 of the 4 animals at each depri-

vation level. The calibration procedures took about 1 to 2

hours to complete. This sometimes required that the lights

be turned on a few minutes before 9:00 A.M. and turned off a

few minutes after 9:00 P.M. Except for the injection phase

of the experiment, the experimenters (Es) were not in the

experimental room except to calibrate.

Data were not scored until at least 15 minutes had

passed since calibration and measurement procedures had

been completed and E had left the room. Following the

deprivation period, data were collected for an additional

eight days.


Preliminary studies had utilized both the subcutaneous

(SQ) and intraperitoneal (IP) routes. The SQ route produced

a longer response and was used in the present study. The

dose was 10 mg./kgm. of dextro-amphetamine sulfate SQ, under

the skin on the dorsal surface (upper back between shoulders).

A stock solution containing 10 mg./cc. was prepared.

In the preliminary studies higher doses had produced

circling and backing-up behavior, followed by coma and

death at still higher doses.

The inter-injection time was determined by Ss response.

When a S first showed signs of sleep, it received another

injection. The observers were directed to constantly watch

the recorder. When a minute of inactivity was seen, they

would observe the rat, and, if asleep, give an injection

(occasionally the S would be awake but in such a position

that the units did not record the wakefulness). If S

was awake, the observer would note this on the recording

paper. Figure 8 shows the average injection intervals.


The data were scored for sleeping and waking according

to the rules illustrated in Table 1. The daily 20 hours

of recording were divided into ten daylight hours and ten

dark hours. The total minutes of sleep per ten-hour period

(600 minutes) was the measure used for statistical analysis.

Occasionally the full 600 minutes of data was not obtained.

In this case the data wereinterpolated to 600 minutes.

For example, if calibration procedures had taken an inordi-

nately long time, perhaps 100 minutes' data would be lost.

If the animal was found to have been asleep 300 out of 500

minutes, this number (300) would be interpolated to a 600-

minute total and 360 would be assigned for this S. Table

4 illustrates the schedule in this experiment by deprivation

groups. A finding of this study was a reduction in circadian

rhythm following five days'deprivation. In other words, the

difference between the amount of sleep obtained during the

day and the amount of sleep obtained during the night was

reduced. For this reason three five-day deprived animals

were recorded on post-drug days 19 and 20 to determine if

the circadian rhythm reduction was still present.


Recording Schedule for the Deprivation
Groups in Experiment II*

Deprivation group Pre-drug Drug Post-drug

24-hours 4 days 1 day 8 days
72-hours 4 days 3 days 8 days
120-hours 4 days 5 days 8 days
* Four Ss per group


The Ss in this study were run in two separate groups;

the replications were approximately six months apart. Each

experiment consisted of six rats, two at each level of

deprivation. Food, water, and body weight data were

recorded only in the original experiment. All other aspects

of the studies were identical. Since the results of the

two replications were essentially identical, supporting all

of the conclusions reached from the grouped data, the data

will be presented in grouped form only.

Results and Discussion

Normative sleep cycle data

Figure 5 is a graph of the daily average sleep time

during this experiment for the three deprivation groups.

Figure 6 shows the circadian rhythm with the deprivation

groups combined. Table 13 in Appendix A contains the

significance levels for the statistically significant

effects in five separate analyses run on thesedata.

If we look at the four pre-deprivation days in Figures

5 and 6, it can be seen that there is a strong circadian

rhythm in sleep cycle (Figure 6) and also a significant

variation in sleep time over days. The analysis of

variance of sleep time for days 1-4 pre-deprivation also

shows significant subject differences in sleep time.

24 HR.

72 HR.
120 HR.

p I




Figure 5. Dextro-Amphetamine
Daily Sleep Time.*

Deprivation Experiment:

* Mean minutes sleep out of 1200 minutes daily record





500 -

w 400


/ "% LIGHT
I es


Ii I
^*-^ / f




Figure 6. Dextro-Amphetamine Deprivation Experiment:
Circadian Rhythm.*

* Mean minutes sleep out of each 600 minute recording


These animals slept an average of 68 per cent of the

time (811 minutes out of 1,200 minutes recorded). They

slept 80 per cent of the day (483 minutes out of 600

minutes recorded) and 55 per cent of the night (328

minutes out of 600 minutes recorded). The Ss obtained a

mean of 60 per cent of their total sleep during the day

and 40 per cent during the night.

Sleep deprivation effects

The significant circadian rhythm continued throughout

the experiment. However, this rhythm was considerably

reduced following deprivation. Figure 6 shows these effects.

Sleep deprivation did increase sleep time above the pre-

deprivation level; however, this effect is statistically

significant only during the first four post-deprivation


There is no difference in total compensatory sleep

due to deprivation level (24, 72, or 120 hours). This

can be seen in Figure 5 where there is no divergence in

sleep time between the deprivation level groups. The days'

effect is significant on days 5-8 post-deprivation, but

not on days 1-4 post-deprivation. However, it can be

seen in Figure 5 that sleep time is considerably elevated

on day one post-deprivation, and gradually returns to the

pre-deprivation level. The 24- and 120-hour deprivation

groups return to a level of daily total sleep within their

pre-deprivation range on day 7 post-deprivation, and the

72-hour group does this on day 6 post-deprivation. Table

5 is an attempt to estimate the average amount of com-

pensatory sleep obtained by each deprivation group. Com-

pensation is arbitrarily considered to have ended on the

first day that the mean sleep time for a deprivation group

overlaps its pre-deprivation range. The most important

point to be made is that total compensation is not sig-

nificantly different between the deprivation groups. Depri-

vation level failed to have an effect in spite of the sleep

deprivation periods being 3 and 5 times as long, respectively,

in the 72- and 120-hour groups as in the 24-hour deprivation

group. Again, there is no increase in amount of compensatory

sleep obtained with increasing sleep deprivation.

There are two possible confounding influences which

may account for this finding. The first is shown in Figure

7. Although the amount of sleep obtained during depriva-

tion was small throughout the drug-injection period, the

amount of sleep obtained does increase after the first day.

This may be due to the increasing sleep need of the Ss

as the deprivation period progressed. It seems possible

that during this high need state one minute of sleep may

be "more valuable" in reducing the post-deprivation com-

pensation than would normally be expected. However, this


Estimated Post-Dextro-Amphetamine
Deprivation Compensation


Mean pre-deprivation sleep

Mean pre-deprivation sleep
time (minutes)

Days of deprivation

Minutes sleep loss


Day 1






Total compensation

Per cent compensation



































* Minutes sleep above the pre-deprivation mean.

sleep during deprivation amounts to only about 20 minutes

per day, which is a very small amount when compared to

the 800 minutes sleep these animals normally get each day.

It se ; unlikely that this small amount of sleep would

be sufficient to eliminate increasing compensation with

increasing amounts of deprivation.

Another consideration is a possible interaction

between the hunger and sleep drives following dextro-

amphetamine. In Figures 9 and 11 it can be seen that the

drug reduced food intake-and body weight considerably,

suggesting that after deprivation the Ss may awaken to eat

because of a strong hunger drive. This could have changed

the relationship between hours of drug administration and

compensatory sleep. However, it can be seen in Figure 9

that food intake did not increase above normal after depri-

vation. Actually, food intake appears to be slightly

depressed. This may be due to stomach shrinkage during

the drug treatment. If stomach shrinkage did occur, it

is possible that the Ss ate more frequently, but consumed

smaller amounts at each "meal."

A curious effect is illustrated in Figure 6. There

is no increase in sleep time during the post-deprivation

light period. As a matter of fact, there is a slight

decrease in sleeping time during the day. The entire

compensation is accounted for by an increased sleep time

during the dark or night phase of the light cycle. An

explanation of this phenomenon may be that the rat, being

a nocturnal creature, sleeps maximally during the day,

awakening only to satisfy other need states which require

attention. The rat, according to this hypothesis, cannot

constantly remain asleep for 8-12 hours because of other

need requirements. Thus, only at night is there time

available to compensate for the heightened sleep need.

Only at night does the rat indulge in activities which can

be curtailed to satisfy an increased need.

Table 6 illustrates the circadian rhythm reduction

following dextro-amphetamine administration. This effect

is particularly noticeable at the longer periods of depri-

vation. The Ss in the 120-hour group obtained 61 per cent

of their daily sleep during the daylight period on the control

days and only 54 per cent on days 5-8 post-deprivation. Two

of these Ss had their sleep cycles recorded on days 19 and

20 following 120-hours deprivation. These are not the

same Ss who received a drug injection on day 11 post-

deprivation (see below). A 57 per cent figure for these

two days suggests that this is only a temporary effect.

This circadian rhythm reduction was not observed in the

treadmill deprivation study (Experiment IV), suggesting

that it is specifically related to the drug treatment.

Since the reduction continues after post-deprivation

compensation has ended, the reduction cannot simply be

accounted for by an increased amount of night sleep with

no change in day sleep. On days 7 and 8 post-deprivation

(Figure 6) after compensation has ended, the Ss are obtain-

ing normal amounts of total sleep, but this amount is

achieved by sleeping less than normal during the day and

more than normal during the night.


Mean Percentage of Daily Sleep
Obtained During Lighted Hours

Deprivation level
Days 24-hours 72-hours 120-hours

Pre-deprivation (1-4) 59% 59% 61%

Post-deprivation (1-4) 53 53 52

(5-8) 57 53 54

(19-20) 57

Sleep during drug administration

Figure 7 shows the mean minutes sleep over days during

deprivation. The diurnal, days and subjects variations were

not statistically significant, although there does appear

to be an increase in sleep time over days, especially

between day 1 and days 2 to 5. On the average the Ss

received about 3 per cent of their normal daily sleep

time during deprivation.

Injection parameters

Figure 8 is a graph of the mean length of sleep

deprivation per injection over the five deprivation days.

An analysis of the original waking times showed the decrease

in length of drug action over days to be significant

(p < .01). This effect could be due either to true drug

tolerance or to an increased sleep need produced by the

increasing deprivation. As a possible test of these

alternatives, a tolerance test was conducted on six of the

Ss (two at each deprivation level) 11 days following drug

withdrawal. A single injection of dextro-amphetamine was

administered and the length of its action determined. The

mean length of wakefulness during the tolerance test was 357

minutes compared to a mean of 354 minutes for the drug on

deprivation day one. It appears that the Ss drug

responsiveness had returned to its initial level; thus,

it is not possible to choose between the alternatives.

This effect may be due to the dissipation of either toler-

ance or sleep need. If the length of drug effect had

still been shortened, tolerance would be suggested since

indications are that the sleep need is normal on day

11 post-deprivation (the 120-hour Ss used in this test

20 -


Figure 7.



Mean Minutes Sleep Obtained per
Subject During Dextro-Amphetamine

2 4


Figure 8. Minutes Sleep Deprivation per
Dextro-Amphetamine Injection.

I I 1 7

were not the same rats as those whose sleep cycle was

recorded on days 19 and 20 post-deprivation).

Drug toxicity

Some of the Ss in the 72- and 120-hour deprivation

groups lost some toes during deprivation. The paws

became white, and the Ss constantly worried and nibbled

at them during deprivation. The most likely explanation

for this seems to be that the dextro-amphetamine produced

peripheral vasoconstriction, reducing circulation and causing

numbness. Tetracycline (10 mg./kgm. twice a day) was

administered on days 1-3 post-deprivation to the injured

Ss. There was no evidence that this side effect had any

confounding influence on the data. All Ss recuperated and

appeared normal following the experiment.

Food and water intake and body weight data

Figures 9, 10, and 11 illustrate the mean food, water,

and body weight measures for each deprivation group over

the entire experiment. These data are for only 6 of the

12 Ss, two at each level of deprivation. Tables 14, 15,

and 16 in Appendix A are summaries of the variance analyses

of these data. The effects of the drug treatment on these

measures are determinable from an inspection of the

figures and tables. The most important conclusions to be

reached are: (a) the drug treatment almost completely

* 120 HR. GROUP

2 4 6


I I I I I I I I I I I I 1 1
4 2 4 6 8 10 12


Figure 9. Food Intake During Dextro-Amphetamine Experiment.


. & . .


2 4 6 2 4 2 4 6 8 10 12


Figure 10. Water Intake During


0 40



*....120 HR. GROUP

oS S^, ^, ...- . .. ......-- --

**-* --*****7*<****C9
~Bq *-*-*

2 4 6 2 4

2 4 6 8 10 12


Figure 11.

Body Weight During Dextro-Amphetamine




eliminates eating, and (b) water intake and body weight

are also decreased secondary to the drug effect on hunger.

These measures all return to normal by the end of the

12 post-deprivation days (readings of these measures were

started two days before and continued four days after the

sleep cycle recording). A reduction in circadian variation,

similar to that seen in the sleep cycle following drug

treatment, is present in the food and water intake data

(Tables 7 and 8).

Tolerance does not seem to have developed to the

anorexigenic effect of dextro-amphetamine during the five

days of treatment.


Mean Percentage of Daily Food Intake
Obtained During Dark Hours

Deprivation level
Days 24-hours 72-hours 120-hours

Pre-deprivation (1-6) 63% 69% 70%

Post-deprivation (1-6) 66 56 56

(7-12) 65 61 58


Mean Percentage of Daily Water Intake
Obtained During Dark Hours

Deprivation level
Days 24-hours 72-hours 120-hours

Pre-deprivation (1-6) 65% 68% 68%

Post-deprivation (1-6) 65 57 58

(7-12) 73 64 56

Body weight was lost during the drug treatment. The

Ss were at approximately 92 per cent of pre-deprivation

body weight after 24 hours, 87 per cent after 72 hours, and

82 per cent after 120 hours. A diurnal rhythm in body

weight is present throughout the experiment, and body weight

has returned to the pre-deprivation level by the end of

the experiment. It is interesting that for the 72- and

120-hour groups body weight continues to fall for a day or

two following drug withdrawal. Food intake is also

relatively low during this period. It is possible that

although there is not enough drug remaining in the body

to maintain wakefulness there is enough to reduce food

intake. The continued low food intake on post-deprivation

days 3-10 possibly results from shrinkage of the stomach

during deprivation.

The inversion of the diurnal effect on body weight

during deprivation is an artifact resulting from the

procedure having been started in the morning. Each con-

secutive weighing was lower than the preceding, and the

daytime weighing followed the night weighing on each day.

The most to be said about the analyses in Tables 14,

15, and 16 in Appendix A is that these measures seem to

be quite variable, and most sources of variances did

produce significant effects. An interesting finding is

that there are significant individual differences in body

weight in spite of no subject difference in food or water

intake prior to deprivation. This would suggest that

differences in body weight among same aged animals are

due to other than intake factors.

Another curious finding is that, although sleep time

increased above pre-deprivation levels following dextro-

amphetamine deprivation, food intake did not. The drug

has deprived the Ss of both of these goal objects, but

apparently the Ss compensate, in this particular situation,

only for the lost sleep, not for the lost food. This may

be due to the aforementioned stomach shrinkage.


In some preliminary observations periodic sleep

waves were found in the EEG of animals on the treadmill.

The rats were seen moving to the front of the wheel to

remain stationary as long as 3 or 4 seconds while riding

to the rear, followed by walking to the front (taking

2 to 4 seconds) to repeat the process. It appeared that

there was good agreement between the EEG synchronization

and behavioral motionlessness. This experiment was

designed to further study this phenomenon.



Three 110-120-day-old male Long-Evans hooded rats

with bipolar EEG electrodes implanted were used (right

frontal to right visual). Food and water were available

throughout the experiment.


This apparatus is the same as the water-immersed

treadmill used by Levitt and Webb (1964). The rats were

placed in individual 5.5 by 9.5 inch cubicles, on wheels

two-thirds submerged in water, which rotated at a constant

speed of 2 r.p.m. Food trays were available in each

cubicle. The animals remained on these wheels continuously.

The total distance covered by an animal was approximately

0.7 mile per 24-hour period. The upper sides of the

apparatus were of plexiglass to facilitate behavioral

observations. The animals remained on the wheels for 32


Results and Discussion

Seven hours of EEG data were collected on each of the

three Ss during their 32 hours on the wheel. Figure 12

shows the increase in "sleep" prevalence during the course

of the experiment. The increase in sleep during the time

on the wheel was statistically significant (p < .05) as was

the difference in sleep prevalence between subjects (p < .01).

The sleep on the treadmill differed from normal slow wave

sleep mainly in the length'of each burst. This "micro-

sleep" occurred in bursts of only 1 to 4 seconds separated

by 2 to 5 seconds of waking activity, while normally, bursts

of sleep last for a number of minutes. There was no micro-

sleep exhibited on the stopped wheel before the experiment

or during the first hour on the moving wheel. It was

possible to inspect the EEG records and estimate the

amount of micro-sleep obtained by the Ss. By hour 32 Ss


StRot 3
m 12 I \ --- Rot 2
O I /

o .
-' 4 0.00
O Rot I

I 8 15 20 21 22 32


Figure 12. Micro-Sleep Prevalence During
Treadmill Deprivation.

2 and 3 were micro-sleeping approximately 20 per cent

of the time. This is compared to a home cage value of

about 65 to 70 per cent of total time asleep for the rats

used in Experiment II. As suggested before, this amount

of sleep may have more value in reducing compensation than

would be expected by the objective amount.

Also, it must be recognized that the relationship

between normal sleep and micro-sleep has not been established.

However, there is evidence that micro-sleep is not equivalent

to paradoxical sleep. First, rats have not been observed

to enter paradoxical sleep directly, but only via slow wave

sleep (Hall, 1963; Swisher, 1962). Since micro-sleep occurs

in very short bursts separated by a waking record, it is

unlikely that it is similar to paradoxical sleep; the Ss

simply do not have time to pass through slow wave sleep and

into paradoxical sleep. Second, Hall (1963) for the rat

and Dement (1958) for the cat have found muscle tension to

be at its lowest during paradoxical sleep. Both investi-

gators reported that whenever the S entered paradoxical

sleep it collapsed due to the reduced muscle tension.

This phenomenon would be inconsistent with the erect posi-

tion maintained by the rats on the treadmill during micro-


The data discussed in the preceding paragraph would

tend to support a skeptical attitude toward the normative

data on paradoxical sleep reported by Siegel and Gordon

(1965). These authors reported control levels of para-

doxical sleep, for the cat, ranging from 27 to 42 per

cent of total sleep. These figures are compared to about

20 per cent in the human (Dement, 1960) and approximately

10 per cent reported later in this paper for the rat

(Experiment V). The peculiarity of the Siegel and Gordon

study is that during 12 to 14 hours out of each day (at

night) the cats were kept awake by being placed on a brick

in the middle of a pan of water. The authors report that

the cats could not lie down; however, they do not mention

having recorded the EEG of these cats on the brick. The

evidence on the occurrence of micro-sleep reported in this

dissertation suggests that Siegel and Gordon's cats did

obtain short bursts of sleep on the brick. However, it

is unlikely that any of this sleep was paradoxical sleep,

since the reduced muscle tension would cause the cat to

fall in the water. Therefore, the normative data on

paradoxical sleep as a percentage of total sleep reported

by Siegel and Gordon are probably artifactually high as a

result of the cats being differentially deprived of para-

doxical sleep for 12 to 14 hours a day while on the brick.

Figure 13 shows examples of micro-sleep and a

behavioral correlation with motionlessness. This behavioral

reading was obtained by an observer watching the rat and

activating a channel on the EEG record whenever the S



1 SEC.

Figure 13. Micro-Sleep and Behavioral Motionlessness
on the Sleep Deprivation Treadmill.*

* Right frontal to right visual bipolar electrodes



Figure 13. Continued

appeared motionless. The E pressed a switch to activate

the behavior channel without observing the EEG record.

This correlation appeared to be essentially 100 per cent.

There were certain sources of error such as latency of

observer's response and the necessity of the S being

motionless for a short period before E could make a decision

and activate'the switch. These error sources seemed to

account for any deviation between the EEG and behavioral


It is most interesting to find the rats on the tread-

mill able to sleep 1 to 4 seconds, wake up for 2 to 5

seconds to perform a goal directed act, and instantaneously

return to sleep. This finding would seem to be of some

theoretical interest. First, these results raise a serious

question as to the sleep deprivation produced on the water-

immersed treadmill (Levitt & Webb, 1964; Licklider & Bunch,

1946; Webb & Agnew, 1962) and on the brick surrounded by

water (Siegel & Gordon, 1965). Second, broadly conceived,

this micro-sleep phenomenon may be interpreted as an instru-

mental response (walking to the front of the wheel) at least

partially motivated by the drive for sleep (micro-sleep).

The author recognizes that escape from water is also a

motive for treadmill walking and initially the only motive.

However, he prefers the interpretation that during the


course of the experiment a second motive for micro-sleep

also comes into force. The possibility of instrumentally

conditioning the sleep response is certainly deserving

of further study. Clemente, Sterman, and Wyrwicka (1963),

and also the present author (Levitt, 1964) have previously

been successful in classically conditioning sleep.


Although the recording of EEGs on the treadmill

showed a sleep-like pattern, it still seemed advisable

to study the effect of this deprivation technique on the

sleep cycle using the ultrasonic activity method, especially

since the Ss may only experience a sleep state analogous to

slow wave sleep on the treadmill. If this were the case,

rats on the treadmill would be deprived of paradoxical or

"dream" sleep at the same time they were receiving some

slow wave sleep. This situation would be similar to the

previously discussed "dream" or paradoxical sleep depri-

vation studies (Dement, 1960; Khazan & Sawyer, 1963;

Siegel & Gordon, 1965).



Ten male Long-Evans hooded rats 110-120 days old at

the beginning of the experiment were used.


Sleep cycle recording was by the same ultrasonic

activity units used in Experiments I and II. The sleep

deprivation treadmill was described in Experiment III.

The calibration procedure was the same as that used in

Experiment II.


There were five Ss at each of two deprivation levels

(24 or 72 hours). Environmental conditions and scoring

methods were described in Experiment II. Table 9 illustrates

the design of this experiment.


Recording Schedule for the Deprivation
Groups in Experiment IV

Deprivation Pre-deprivation Deprivation Post-deprivation

24-hours 3 days 1 day 5 days

72-hours 3 days 3 days 5 days

Results and Discussion

Figures 14 and 15 illustrate the sleep cycle data

from this experiment. Table 17 in Appendix A summarizes

the analyses of variance, and Table 10 in the text is an

estimate of sleep compensation following treadmill depriva-


1000 C

w I .*
I %

$ oo

Scoo. I \ "
I /

I 2 3 1 2 3 4 5

Figure 14. Treadmill Deprivation Experiment:
Daily Sleep Time.*

* Mean minutes sleep out of 1200 minutes daily record




" `-o DARK

2 3

I I I I 5

S4 5


Figure 15.

Treadmill Deprivation Experiment:
Circadian Rhythm.*

* Mean minutes sleep out of each 600 minute
recording period

550 2





Estimated Post-Treadmill
Deprivation Compensation

24-hour 72-hour
group group

Mean pre-deprivation sleep time 810 810

Days of deprivation 1 1

Minutes sleep loss 810 2,430

Post-deprivation compensation*

Day 1 105 187

2 63 126

Total compensation 168 313

Per cent compensation 21% 13%

* Minutes sleep above the pre-deprivation mean.

There was a strong circadian rhythm throughout the

experiment, but unlike Experiment II there was not a

circadian rhythm reduction following treadmill deprivation

(Figure 15). See Figure 5 for a comparison. As a matter

of fact, the circadian rhythm on post-deprivation days 3

to 5 was greater than it was pre-deprivation (p < .05).

It was not possible to maintain treadmill deprivation for

as long as dextro-amphetamine deprivation, since many Ss

at this age would be expected to reach exhaustion between

72 and 120 hours on the wheel (Levitt & Webb, 1964; Webb

& Agnew, 1962).

The major conclusions suggested by the results of

this experiment are: (a) that sleep time temporarily

increased following treadmill deprivation and had returned

to normal by the end of the study, and (b) that there was

no significant effect of deprivation level on sleep depri-

vation compensation. Both these findings confirm the

dextro-amphetamine experiment.

Table 10 shows the estimated compensation. It appears

to be somewhat less than in Experiment II. The Ss returned

to their normal sleep time range on day 3 post-deprivation,

while the Ss in Experiment II did not do this until day 6

or 7 post-deprivation. Also, the total minutes compensation

and percentage compensation are smaller in Table 10 than

in Table 5. This is especially noticeable for the 24-hour

deprivation groups.


The experiments detailed thus far have produced some

interesting and suggestive data on the response of the sleep

cycle to sleep deprivation. A question remaining to be

answered is whether there is a difference in the response of

the two sleep phases (slow wave and paradoxic 1) to the two

deprivation techniques utilized in this paper. In particular,

dextro-amphetamine induced sleep deprivation deprives the

S equally of both slow wave and paradoxical sleep. The

present experiment will enable us to answer the question of

whether the post-deprivation compensation consists of the

two sleep phases in their normative proportions or consists

predominantly of one or the other sleep phase. Unlike dextro-

amphetamine induced sleep deprivation, treadmill induced

deprivation seems to differentially deprive Ss of para-

doxical sleep. Although the Ss also receive less slow wave

sleep than normal, the treadmill seems to deprive them of

relatively more paradoxical than slow wave sleep. This

experiment will enable the description of the response of the

two sleep phases to treadmill deprivation. The information

reported in this experiment should provide a beginning

at answering the question of which phase of sleep is most

"needed," and hopefully also provide part of a foundation

for a functional analysis of the sleep phases.

This experiment will also serve as a partial replica-

tion using the EEG of Experiments II and IV which measured

the sleep cycle with the ultrasonic activity units. The

data on compensation during the first 24 hours in this

experiment can be considered a replication of the compensa-

tion findings on day one post-deprivation in Experiments

II and IV.



Six male Long-Evans hooded rats 110-120 days old at

the beginning of the experiment were used. All Ss had

bipolar cortical electrodes implanted. This procedure is

the same as that in Experiments I and III.


The EEG, treadmill, and dextro-amphetamine procedures

have been described earlier in this paper.


All six Ss were placed in the experimental cages

48 hours prior to EEG recording. EEG recording was begun

at 8:00 A.M. on day one. Data were not scored until

2:00 P.M. This six-hour period was used as an adaptation

procedure. The EEG attachment would occasionally come off

the rat and require reapplication. An observer was always

in the experimental room monitoring the Ss and EEG equip-

ment. The behavioral appearance of S (sleeping or waking)

was periodically noted and marked on the EEG record.

These procedures would be expected to disturb the

Ss somewhat and cause them to be awake more than normal.

Although this did occur, the change would not seem large

enough to effect generalization from these findings.

During the control days in the first dextro-amphetamine

study, the Ss averaged 68 per cent total sleep. During

the 24 control hours in this study, the Ss averaged 52 per

cent total sleep.

Control EEG recordings were made from 2:00 P.M. on

day 1 to 2:00 P.M. on day 2 at which time 3 Ss received

their initial dextro-amphetamine injection and the other

3 Ss were put on the treadmill. Deprivation continued for

24 hours until 2:00 P.M. on day 3. Post-deprivation EEG

recording began at this time and continued for 24 hours.

Each minute of the EEG record was scored waking, para-

doxical sleep, or slow wave (normal) sleep.

Results and Discussion

Figures 16 through 23, Tables 11 and 12 in the text,

and Tables 18, 19, and 20 in Appendix A contain summaries

and analyses of thesedata. For purposes of analysis the

day was divided into four six-hour periods beginning at

6:00 A.M. Table 18 is a summary of nine analyses of

variance performed on thesedata. Figure 16 illustrates

these effects. Dextro-amphetamine deprivation significantly

increased the amount of both paradoxical and slow wave

sleep and decreased waking as compared with their control

values. Treadmill deprivation increased paradoxical sleep

without significantly altering the amount of either waking

or slow wave sleep. Boththese findings (dextro-amphetamine

and treadmill) are consistent with those of Experiments II

and IV. The bottom half of Table 18 is a summary of three

analyses comparing dextro-amphetamine to treadmill depri-

vation. The lack of a significant method effect indicates

that the two groups did not differ significantly before

treatment. The treatment effect on waking and normal sleep

is known from the analyses at the top of Table 18 to be

completely accounted for by the dextro-amphetamine group.

The significant M x T interaction on waking and normal

sleep is an expression of the drug, but not the treadmill

altering these measures. The lack of a M x T interaction

on paradoxical sleep confirms that both deprivation methods

increased paradoxical sleep equally.


Sleep Cycle Totals Before and
After Deprivation

Pre Post Pre




Paradoxical sleep

Normal sleep

728 min.



387 min.



653 min.



587 min.



Per cent of total


Paradoxical sleep

Normal sleep

Paradoxical sleep as
a per cent of total













10 16


11 21



Estimated Post-Deprivation Compensation

Dextro-amphetamine Treadmill
group group

Pre-deprivation 24-hour
sleep time

Normal sleep 641 min. 701 min.

Paradoxical sleep 71 86

Total minutes sleep loss 712 787

Post-deprivation compensation*

Normal sleep 240 -24

Paradoxical sleep 101 90

Total minutes compensation 341 66

* Minutes of normal sleep and paradoxical sleep above the
pre-deprivation level.

Figure 17 shows the deprivation effects as a percentage

of control readings for the two deprivation methods, and

Table 11 presents pre- and post-deprivation waking, para-

doxical, and normal sleep averages for each deprivation

method. Table 12 is an attempt to estimate the amount of

post-deprivation compensation. Again we see that dextro-

amphetamine deprivation increased both normal and para-

doxical sleep and decreased waking; however, the major






200 -


6A.M. Noon- 6P.M. Mid.- 6A.M. Noon-
-Noon 6P.M. -Mid. 6A.M. -Noon 6P.M


Figure 16. Dextro-Amphetamine or Treadmill
Deprivation and Sleep Cycle:
Waking, Paradoxical Sleep, and
Normal Sleep.

- 6P.M. Mid.-
S-Mid. GA.M.







*.. --

6A.M. Noon. 6P.M. Mid.-
-Noon 6P.M. -Mid. 6A.M.


/ \

/ 1
/ 1
/ 1

6A.M. Noon- 6P.M. Mid.-
-Noon 6P.M. -Mid. 6A.M.


Figure 16. Continued.





6A.M. Noon- 6P.M. Mid.-
-Noon 6P.M. -Mid. 6A.M.


6A.M. Noon- 6P.M. Mid.-
-Noon 6P.M. -Mid. 6A.M.


Figure 16. Continued.




*--- 0






1 23 45 6


123 456


123 456

Figure 17.

Post-Deprivation Sleep-Waking
Cycle as a Percentage Change
from Control Values.






effect is on paradoxical sleep which increases from 10

to 16 per cent of total sleep. Treadmill deprivation

strongly increased paradoxical sleep, and this increase

was allowed for by small but nonsignificant decreases in

both waking and normal sleep. Paradoxical sleep increased

from 11 to 21 per cent of total sleep after treadmill

deprivation. The increases in paradoxical sleep produced

by dextro-amphetamine or treadmill deprivation were not

statistically different.

An attempt was made to further analyze the change

in paradoxical sleep following deprivation. Figure 18

shows the mean sleep epoch length before and after depri-

vation. Table 19 is an analysis of variance which con-

firmed that the sleep epochs were longer after deprivation

than on the control day. A sleep epoch is defined as four

or more consecutive minutes of sleep (either slow wave or

paradoxical). In order for a sleep epoch to be terminated,

four consecutive minutes of waking must interrupt the

epoch. It is known from numerous studies of sleep in humans

that REM sleep tends to occur in the second half of a

night's sleep (Aserinsky & Kleitman, 1953; Dement & Kleitman,

1957a). This relationship has not heretofore been demon-

strated in rats, but, if it existed, an increase in length

of sleep epoch might be expected to produce increased para-

doxical sleep as a nonspecific effect.

--- Pre-Deprivotion
---Post-Depriva ion





t \


/ d

6 A. M.
- Noon

Figure 18. Length






Noon- 6P.M. Mid.-
6P.M. -Mid. 6A.M.


of Sleep Epochs.

- ,o-- -*

0-10 11-30 31-60 61-100 101-180


Figure 19. Length of Paradoxical Sleep Bursts.

An attempt was made to further analyze the para-

doxical sleep data so as to test this hypothesis. The

pre- and post-deprivation data were matched according to

length of sleep epoch. Figures 19 through 22 summarize

these data. Table 20 in Appendix A summarizes the variance

analyses for these parameters. Following deprivation para-

doxical sleep bursts occurred significantly more frequently

during a sleep epoch and occupied a larger percentage of

total sleep. Although the length of each paradoxical sleep

burst was not significantly longer, it was in that direction

(Figure 19). Also, the first minute of paradoxical sleep

did not occur significantly earlier in the sleep epoch

following deprivation, although again the data are in that

direction (Figure 22). These results suggest that sleep

deprivation specifically acts to increase paradoxical sleep.

In Figures 19 to 21 it can be seen that the amount of para-

doxical sleep does not increase with increasing length of

sleep epoch. This suggests that the increase in paradoxical

sleep produced by deprivation is not simply the result of

an increase in sleep epoch length. Also, it can be seen

in these figures that at the same point in a sleep epoch

paradoxical sleep is more likely to occur following depri-

vation. This also suggests a specific activation of the

paradoxical sleep state.

Many investigators have reported increased movement

and twitching during paradoxical sleep in the rat (Hall,

- Pro-Deprlvalon





.-C-- o---

0-10 11-30 31-60 61-100 10t-180


Figure 20. Minutes of Slow Wave Sleep Separa-
ting Paradoxical Sleep Bursts.*

; Frequency






/ -D


0-10 11-30 31-60 61-100 101-180


Figure 21. Percentage of Total Sleep Time
Spent in Paradoxical Sleep.

- Pre-Doprivaion
- --Post-Deprivation

6A.M. Noon- 6P.M.
-Noon 6P.M. -Mid.

6A. M.


Figure 22.

Minutes into Sleep Epoch When
First Minute of Paradoxical
Sleep Occurs.


I I I i
0-10 11-30 31-60 61-100 101-180


Figure 23. Difference Between Observed and
Expected Awakenings from Para-
doxical Sleep.

1963; Swisher, 1962). This was not confirmed in Experiment

I, since activity during paradoxical and slow wave sleep

as measured by ultrasonic activity errors was about equal.

However, the hypothesis that Ss awoke more frequently

from paradoxical than slow wave sleep was tested and con-

firmed (x = 60.2, df = 5, p < .001). If the rats awoke

from sleep epochs ten times during 100 minutes of sleep

and paradoxical sleep amounted to 10 per cent of sleeping

time, then we would predict one awakening from paradoxical

sleep. Figure 23 presents these data. The Ss awoke

significantly more frequently from paradoxical sleep than

would be expected. This finding perhaps suggests a

disturbing influence (dreams?) during this stage of sleep.


Experiment I
The Development of an Ultrasonic Activity Device
to Measure Sleep and Waking in the Rat

This experiment established the ultrasonic activity

system as a useful and reliable means of measuring the

sleep cycle. These units are especially valuable for long

term recordings for which the EEG is impractical. It is

realized that it is still necessary to refer to the EEG

periodically to recheck the EEG-activity correlation and

also in order to differentiate sleep stages.

The ultrasonic activity unit can introduce new

experimental possibilities into sleep research. With

these units it should be possible to continuously monitor

the sleep-waking cycle of small mammals for months or

longer. The disadvantages of the EEG, which are eliminated

by ultrasonic activity recording, include: (a) large

expense in time and money, (b) the necessity of surgically

implanting electrodes, (c) the expense and difficulty of

reading and interpreting EEG records, and (d) the necessity

for the animal to trail electrode wires, which probably

disrupt the naturalness of the response under study. The

major disadvantage of ultrasonic activity recording--not

being able to differentiate sleep phases--has already

been mentioned.

The interesting finding of no difference between

the percentage of high voltage slow sleep minutes and

LVFS minutes, which produce errors, would lead to the

conclusion that the Ss were equally active during these

two sleep stages.

Experiment II
The Effect of Dextro-Amphetamine Induced
Sleep Deprivation on the Sleep Cycle
(Activity Measure)

The stability of the sleep cycle in this experiment

and also in Experiments IV and V is quite rewarding. It

must be realized that these are the first explorations

utilizing the techniques of ultrasonic activity and pro-

longed EEG recording. It is expected that future experi-

ments will use longer recording periods and also that as

the Es become more familiar with the equipment, an even

more stable baseline will be found. This consideration

is important, since the magnitude of experimental effect

required to produce statistical dependability is partially

determined by the stability of the baseline measure. The

fact that results of both statistical and psychological

import were found in all the studies reported here suggests

the systems (ultrasonic activity and EEG) are sufficiently

reliable for variables of the magnitude used.

The experiments reported here also have allowed the

description of the parameters of sleep behavior in the

rat. These are the first sleep studies in the rat of

sufficient length to allow a description of the normal

sleep cycle (see especially the normative sleep cycle data

from Experiment II, and also the pre-deprivation measures

from Experiment V shown in Table 11 and Figures 16 to 23).

These data apply only to a limited subject population

but are, at least, a beginning at a description of rat sleep.

The dextro-amphetamine technique was successful in

almost completely eliminating sleep during the deprivation

period. The two major findings of this procedure are:

(a) that the drug induced sleep deprivation resulted in a

temporary compensatory increase in sleep time following drug

withdrawal, and (b) that increasing deprivation from 24 to 72

to 120 hours did not produce any significant change in the

amount of compensatory sleep. The most exciting conclusion

suggested by these data is that increased sleep deprivation

over 24 hours up to 120 hours does not result in an

increased sleep need over that present in the 24-hour sleep

deprivation group. Two possible alternatives to this

conclusion have been discussed earlier. First, although

the amount of sleep during deprivation was small compared

to the normal level, sleep was present. It may be that

sleep during a heightened need is ',more valuable" in

relieving the deficit than under conditions of normal

drive. The data from Experiments III and V would seem

to confirm this hypothesis, since in Experiment III the

Ss obtained sleep in smaller than normal amounts on the

treadmill, and in Experiment V this amount of sleep was

sufficient to eliminate the need for post-deprivation slow

wave sleep compensation. Further experiments which (a)

are more successful in completely eliminating sleep, and

(b) study the effect of small amounts of sleep inserted

at various times during deprivation, will help to answer

this question. Also, the dextro-amphetamine anorexigenia

may interact with the sleep need and change the shape of

the recovery function. Stomach loading during deprivation

may control for this effect. However, the author prefers

the interpretation that the lack of increased compensatory

sleep with increased sleep deprivation from 1 to 5 days

reflects the relationship between deprivation and need.

This finding may be very significant if confirmed. It is

conceivable that similar findings would be found in other

species; of particular interest are the possibilities and

implications for humans in acute behavioral requirement


There was no cultural restraint on the sleeping time

of the rats in this study. We can assume that they received

all of the additional sleep they required. The amount of

compensatory sleep should reflect the biological deficit

and sleep requirement produced by prolonged dextro-

amphetamine induced sleep deprivation.

Dextro-amphetamine is a sleep depriver that requires

no effort to stay awake by the subject; thus, perhaps

leaving S free to perform tasks that an individual intent

on remaining awake could not perform. These possible

effects may have important applications in behavioral

situations such as may be found in military and aerospace


One conceivable fault of the dextro-amphetamine

technique is that sleep loss effects are compounded by

muscle fatigue, but it seems that all methods of producing

sleep deprivation require movement and work on the part

of the subject. At present it is not possible to keep an

inactive S awake. Kleitman (1963) found that the only

way he could keep human Ss awake was to have them engage

in some sort of muscular activity. Whether the beneficial

effects of dextro-amphetamine outweigh the toxic ones in

particular behavioral situations remains an open question

for further study. Further study of behavioral capacity

during and following dextro-amphetamine administration in

doses sufficient to maintain prolonged wakefulness in

animals, and particularly in humans, would be of con-

siderable importance. Of at least peripheral interest is

a review article by Weiss and Laties (1962), who have

examined the effects of dextro-amphetamine on performance,

concluding that physical endurance, capacity, and motor

coordination are enhanced. Dextro-amphetamine seems to

hasten conditioning, to improve discrimination learning

in sleepy Ss, and increase the rate of motor learning.

Dextro-amphetamine apparently does not lead to improved

intellectual performance except when normal functioning

is degraded by fatigue or boredom. Weiss and Laties

conclude that there is no convincing evidence that a

psychological or physiological price is paid for the

enhanced performance.

The curious finding that sleep time does not increase

or increases very little during the daylight but that

large amounts of compensation occur at night following

sleep deprivation is of some interest (see Figures 6 and

14). The suggestion that the rat sleeps maximally during

the day, in the same sense that a human sleeping in bed

from midnight to 7:00 A.M. sleeps maximally, should be


Experiment III
EEG and Behavioral Observation on Rats
Walking the Water-Immersed Treadmill

The finding of micro-sleep in rats walking the sleep-

deprivation treadmill is of considerable interest. This

finding effects the interpretation of a number of experi-

ments which have studied the effects of sleep deprivation

produced by a water-immersed treadmill or similar apparatus

on later behavior. These Ss were not completely deprived

of sleep, to say the least. This experiment is also of

interest, since the micro-sleep phenomenon may be con-

sidered to have the properties of an instrumentally condi-
tioned response motivated by the sleep need.

Experiment IV
The Effect of Treadmill Induced Sleep
Deprivation on the Sleep Cycle
(Activity Measure)

The results of this experiment suggest that, although

rats do experience a "sleep-like" state on the treadmill,

some type of deprivation condition is produced. The major

findings confirm the dextro-amphetamine experiment: (a)

deprivation did result in increased sleep, and (b) increas-

ing the deprivation level from 24 to 72 hours did not

significantly increase the amount of compensatory sleep.

This finding would suggest that the similar observation in

Experiment II was not due to food deprivation factors.

However, these Ss also obtained larger amounts of sleep

(micro-sleep) during increasing deprivation (see Figure 12),

and this problem has not been resolved.

Experiment V
Dextro-AmDhetamine or Treadmill Induced Sleep
Deprivation and EEG Measured Sleep Cycle

The normative data produced by this experiment have

been discussed above.

The most interesting finding is that dextro-amphetamine

deprivation results in a compensatory increase in both slow

wave and paradoxical sleep (confirming Experiment II), while

treadmill deprivation produces a compensatory increase only

in paradoxical sleep (suggesting that the compensation in

Experiment IV was only for paradoxical sleep). The tread-

mill results confirm the finding of micro-sleep in Experi-

ment III and suggest that this phenomenon is analogous to

slow'wave sleep. However, dextro-amphetamine deprivation

also increased paradoxical sleep more than slow wave sleep.

This finding that paradoxical sleep occupies a higher

percentage of total sleep suggests a function in speci-

fically remedying a high sleep need state. Certain specific

psychological and/or physiological activities may occur

during paradoxical, but not slow wave sleep. These acti-

vities apparently have a high need priority above those

occurring during slow wave sleep. The nature of these

activities is not known, although the psychological functions

may be related to the dream state.

This finding of a high priority for paradoxical sleep

compensation following deprivation has been confirmed by

Svorad for the rat (Webb, personal communication) and

Ferguson and Dement (1965) for the cat. Berger and

Oswald (1962) and Williams, et al. (1963) have also con-

firmed these findings in the human. However, the human

studies have found a stage 4 compensation to occur before

REM sleep can increase. Since stages within slow wave sleep

have not been differentiated in the rat, these studies are

not directly comparable. These studies would seem to

indicate a high "need" for both paradoxical and stage 4

sleep. These results, showing specific compensatory

changes in stage 4 and REM sleep to follow sleep deprivation,

are in conflict with Dement's (1965) hypothesis that

physiological variations within NREM sleep do not warrant

further subdivision. The results of Berger and Oswald

(1962) and Williams, et al. (1963) suggest a functional

differentiation of stages within NREM sleep, since stage 4

responds differently than other NREM sleep to deprivation.

A further analysis of the paradoxical sleep effect

indicated a specific activation of this state, which was

similar for both dextro-amphetamine and treadmill depri-

vation procedures. Following deprivation paradoxical

sleep occupied a significantly higher percentage of total

sleep, and bursts of paradoxical sleep occurred more fre-

quently during a sleep epoch. Also, a nonsignificant

increase in the length of each burst and an earlier