~MAY '9-.,. - DTIC · breaks. Both brief breaks and a longer rest period were very helpful in sustaining performance. METHODS Subjects Subjects in the study were 32 male volunLeers,

LEVEV

TECHNICAL REPORT #1 LEVEIi_____-- - ~MAY 816)SLEEP ýOSS •-EFFECTS ON SONTINUOUS SUSTAINED PERFORMANCE,

• (. .0 '9-.,. -. ..00 ANIEL J. LANE I -A LZ/IKE 3A LC B.A.

EPARTMENT OF PSYCHIATR

UNIVERSITY OF CALIFORNIA, SAN DIEGO

00

PREPARED FOR OFFICE OF NAVAL RESEARCH

PERFORMED*UNDER CONTRACT i i00 :A1A4-79-C-0317 ,' I./7'- .

LAVERNE JOHNSON, PH.D. CONTRiBUTED TO THE DESIGN OF THIS RESEARCH.

WILLIAM MASON, M.A., SAM MESSIN, NOBUYUKI OKUDAIRA, M.D., PH.D., AND

JOHN WEBSTER, PH.D. PROVIDED TECHNICAL ASSISTANCE. THIS RESEARCH WAS

ALSO SUPPORTED BY THE MEDICAL RESEARCH SERV'I.CE OF THE VETSRANS

ADMINISTRATION, BY DAM]) 17-78-C-8040, AND tý,;NIMH RESEARCH SCIENTIST

DEVELOPMENT AWARD 1 K02 MHOO117.

REPRODUCTION IN WHOLE OR IN PART IS PERMITTED FOR ANY PURPOSE OF THE

UNITED STATES GOVERNMENT.

U", - - ---_ _ _ _ o

All

SECURITY CLAS .tFICATION OF THIS PAGE ("oen Data E•atered) RD" ~ ~ ~ ~ ~ ~ ~ ~ ~ RA REOTDCMNAINPG , •INSTUCTIONS.REPORT DOCUMENTATION PAGE BEFORE COMPLETING WORM

. REPORT NUM"ER . .2. GOVT ACCISSION NO: 2.- RECIPIENTwS CATALOG NUMBER#114. TITLE (and Subtitle) S. TYPE OF REPORT & PERIOD COVERED

SLEEP LOSS EFFECTS ON CONTINUOUS SUSTAINED Interim Report

PERFORMANCE. C. PERFORMINT OR'' REPORT NUMBER

7. AUTHOR,( 1- CONTRACTOR fi),,' NUMEe)

Daniel J. Mullaney, Daniel F. Kripke, and ONR N00014-79-C-0317 4*,

Paul Fleck

9. PERFORMING ORGANIZATION NAME AND ADDRESS 0I. PROGRAM ELEMENT. PROJECT. TASK II AREA I WORK UNIT NUMEERS

Department of Psychiatry (MOO3)' . EO0014-79-C-0317/KrIpke,RS

University of California, San Diego 2)1381La Jolla, CA 92093

it. CONTROLLING OFFICE NAME AND ADDRESS iig. EPORT DATE

O~ffze of Naval Research (Code 441) 24 April 1981Department of the Navy II. NPMSER OF PAGES

Arlington, VA 22217 -8814. MONITORING AGENCY NAME & ADORESS(If dilferent from Controlling Oflice) 15. SECURITY CLASS. (of this ropot.

Unclassified

IS,,. DECLASSIFICATION/DOWNGRADING

-SCHEDULE

16. DISTRIBUTION STATEMENT (of Ohli Report)

Distribution unlimited.

17. DISTRIBUTION STATEMCNT (of the abstract entered in 41lock 20, if different ktom Report)

4!IS. SUPPLEMENTARY NOTES

19. KEY WORDS (Continue on revere side If nceieuaav and Ide'ntify by block number)

Sleep loss, continuous performance, sustained performance,

sleep deprivation, sleepiness

20. SSTRACT (Continue an,.evero side if necesary and Identify by block number)

• he ability to sustain continuous performance for up to 42 hours wasstudied. During each 10 minutes, subjects performed a tracking task, apattern memory task, an addition task, a simultaneous auditory vigilance task, Iand provided subjective ratings on sleepiness and attention-fantasy scales.Of 10 subjects required to work continuously with no breaks, 3 could not com-plete the 42 hours, and 8 out of 10 suffered symptoms such as hallucinations,visual illusions, and disorientation. Subjects provided either 6 one-hour naps

FORM

I..I

D 0 1 A"7, 473 EDITION OP I NOV 65 IS OBSOLETE

S,'N0 10- LF 014 660 SECRITYCLASSIFICATION OF THIS PAGE (Ifton~ Date* En1tmed,

suffered fewer psychiatric symptomm. '..The group working continuously suffered

those receiving one-hour or sl~x-'iour rest periods sustained their level ofperformance much more effectively. Viese results show that when absolutelycontinuous sustained work is required, performance deteriorates seriously.

even within the first 24 hours.

Aoesshton For-V~j GRA&I

DTIC TAB

Just ification

By-Diut~ributiol/-

SN 0 102- LF 0 14- 6601

SECURITY CLASSIFICATION OF THIS PAG"(When Data Entered)

INTRODUCTION

Previous studies of sleep deprivation have emphasized the duration of sleep1oss required to produce measurable or severe impairments of performance. Inmost of these studies, performance in a variety of psychological tests wasmeasured only intermittently, so that although the subjects were kept awakecontinuously, they had long rest periods when nothing specific was required ofthem.*

In this study, we examined the impairments of performance which occur whensubjects are required to work contlnuously with only negligible opportunitiesfor rest and recuperation. A requirement for absolutely continuous performanceis increasingly found in the tasks of modern life such as long distance driving,flying airplanes, monitoring radar screens, and similar assignments requiringcontinuous monitoring of electronic consoles or machinery.

Previous studies of sleep deprivation have suggested that the tasks mostvulnerable to fatigue and sleep deprivation are those which are work-pacedrather than self-paced., high workload, complex, requiring conLinuots attentionand performance, and which are relatively monotonous (Naitoh, 1969; Johnson andNaitoh, 1974; Woodward, 1974). Immediate feedback tends to sustain performance(Wilkinson, 1961). Vigilance seems to suffer sooner than memory or cognitiveskills (Naitoh, 1969).

Two mechanisms may impair performance early after minimal sleep deprivation(Kjellberg, 1977a, 1977b, 1977c). The first mechanism is the occurrence of

lapses of vigilance, often associated with loss of EEG alpha. Williams, Lubinr and Goodnow (1959) showed that a loss of alpha rhythm was closely associated

with performance failure. A second mechanism may be a failure of attentionalfocus (Kjellberg, 1977b). Tests which require continuously focused attentionmay be differentially impaired, particularly beyond the first 10 minutes of taskperformance (Hockey, 197ia). High arousal increases the selectivity ofattentional focus and may thus actually restrict attention to low priority tasksin a complesx performance model (Hockey, 1977a, 1977b).

We believe that many of the sleep deprivation studies thus far reported maymodel continuous sustained performance poorly, because the performance taskstested were relatively brief, and long breaks were allowed between testingsessions. Studies of physiologic effects of sleep deprivation have emphasizeddepr vations of 60-100 hours or more (Johnson, 1969). The continuousperformances tested were continuous on a scale of days but not on a scale ofhours.

The crucial factor in measuring the performance impairments which occurearly seems to be the duration of the testing session (Wilkinson, 1969).Wilkinson, Edwards, and Hainer (1966) stated:

"...if the durati..n of performance testing is extended so that it•1 approximates to a normal day's work, the reduction of sleep by about

half on a single night can produce a significant fall in working

3

I

efficiency. The reason why previous experiments...have failed toshow this effect of partial sleep deprivation is almost certainlythat the performance tests were too short. The complete loss of onenight's sleep has no effect upon the first five min of work on

, serial reaction teats, vigilance and adding, but a clear impairment ofperformance emerges when these teýts are prolonged for 15 to 40mi[nutese •.

This would suggest that breaks are crucial in sustaining performance oncontinuou'0 high workload tasks. Perhaps the importance of breaks- may help toexplain the relatively favorable results obtained by Alluisi and colleagues with4-on-4-off and 4-on-2-off schedules as opposed to 12-on-12-off, since one wouldcertainly anticipate that subjects would obtain mcre sleep with 12-on-12-affschedules (Alluisi, 1972). The low-demand intervals in Alluisi's 2-hourperformance modules may also have served as breaks. Similarly, breaks mayprovide a benefit favoring persistance of 4-on-4-off schedules in certain Navalsettings, particularly aboard submarines.

With a view to exploring these issues further, we have designed and testeda model where subjects are required to perform virtually without any brt-ak orrest for up to 42 hours. The pressure of the performance tasks is continuous,and the motivation is high. Impairments of performance in this model were.compared with impairments produced when a similar 42-hour work schedule wasinterrupted by several 1-hour breaks or by one prolonged 6-hour sleep period.The results tend to confirm our hypotheses that performance impairments appearvery quickly when tasks must be sustained absolutely continuously withoutbreaks. Both brief breaks and a longer rest period were very helpful insustaining performance.

METHODS

Subjects

Subjects in the study were 32 male volunLeers, mostly university students,recruited from advertisements in the campus newspaper. Ages ranged from 1 8 to31 with a mean age of 21.4 years.

Potential subjects were screened by interview for suitability and formotivation to do well in the study. All subjects r.ompleted written informedconsent forms on a day prior to the experiment.

On a day prior to the study, subjects wEre given 1 .struction and were

invited to practice the performance tasks for 30-40 minutes, so that they wouldbe familiar with the procedures. They were told that even though they were freeto drop out of the study at any time, we preferred that they complete theprotocol and attempt to maintain a sustained high level of performance at alltimes. Subjects were paid $2 per hour for each hour of participation. Inaddition, in order to enhance motivation and compliance, subjects were paid upto an additional $2 per hour depending on their performance and contingent upontheir completion of the study.

4

*[

Approximately onu hour ?rior to their normal bedtimes, subjects came to thelaboratory to be equipped for polygraphic recordings. Right and lefthemisphere perieto-occipital EEG, submental electromyogram (EMG), bipolarhorioontal electro-oculogram (EOG), and activity from the dominant wrist wererecorded during both wakefullness and sleep throughout the 48 hours of thestudy. Signals from an artifact detector attached to a head-mounted EEGpreamplifier were recorded to aid in scoring EEG activity. (Polygraphic reoultswill be described elsewhere.) Each study began with a 6-hour sleep period fromthe time of "lights out," which was chosen as the subject's usual bedtiue.Subjects were then awakened (usually between 0500 and 0600, mean time-0510) andimmediately required to commence performance testing.

Ten subjects were randomly assigned to each of three groups. A structuredrandom assigrnent method wat used so that each subject had an equal chance ofassignment to each group, and the three protocols were run simultaneously fromFebruary through December of 1980, counterbalancing any order effects. Onesubject, who quit the study after only 3 hours of testing, was replaced. Asecond experiment was terminated by the experimenters when it was realized thatperformance data were not being stored due to an equipment failure, and thissubject was also replaced.

In the first group (42-straight), subjects were required to maintaincontinuous performance for the final 42 hours of the experiment, fromapproximately 0500 of the first day to 2300 of the second day. In the secondgroup (6-and-i), subjects were required to work coatinuously except that every 7hours they were given a one-hour rest period, duriLg which they were asked tosleep in bed with the lights out. In the third group (18-and-6), subjects wererequired to work continuously for 18 hours, then they were allowed a 6-hoursleep period, and finally a second 18-hour continuous performance period wasrequired. Thus, within the 42 hours following the initial 6-hour sleep period,the total durations of the rest periods were equivalent in the 6-and-l and the18-and-6 groups. In the 6-and-i group, however, the final 1-hour nap occurredafter all performance testing was completed. Sequences of sleep periods andtesting for all 3 groups are diagrammed in Fig 1.

Subjects worked alone in a well-lighted attraciive 3x3.5m room, but theywere intermittently visited by research staff to attend to subjects' needs andto repair occasional equipment problems. Subjects did not wear wrist watches.but they were not otherwise isolated from time cues such as hallway noises and ahallway clock. Subjects were permitted to walk across the corridor to use therefstroom. Snack food and non-caffeinated beverages stored in a smallrefrigerator were available ad lib, but the performance tests were notdiscontinued foi eating, drinking, or restroom trips.

Tasks

An Apple II microcomuter system was employed to present the performancetasks, to regulate the timing of the performance trials and sleep periods, andto store the data.

1. A 3-minute tracking task displayed a small randomly moving target on a

5

A,_ ý;A

,J '.' -•!llm~ll~lRI~im l • ,, . ,,. -..... ..... . ... . . . . ..

video monitor screen which also displayed a stationary grid. The subject's jobwas to position the target using a hand-held joystick (control stick) to theexact center of the grid, at which time pressing a "trigger" button could score"a "hit" on the target. This task simulated electronically-controlled gunnery ormissile firing. The subject was given immediate feedback of his performance,since the target would simulate explosion when properly tracked and hit. Thenumber of "bits" was tabulated by the computer. This task was largely

self-paced.

2. A visual pattern memory task was presented for 3 minutes. Arandomly-generated 5X5 matrix of 25 squares (each either white or black) wasdisplayed on the video monitor screen for 3 seconds. After this initial patterndisappeared from the screen, a second and a third pattern were sequentiallydisplayed, either of which might be identical to the original pattern. Thesubject then indicated by pressing the appropriate key if the second, the third,

or neither copy was identical to the first pattern. This task was totallyexperiment-paced; 3.2 seconds were allowed for response. Subjects receivedimmediate feedback whether their identifications were correct or wrong.

3. A third 3-minute task tested numerical addition skills. The subjectwas required to add two randomly-generated 3-digit numbers (presented on themon'tor), then add 7, and enter the 3-digit sum on a numeric keyboard. The timeallowed for responses during the task was limited, and the limit was variedadaptively to create time pressure however performance varied. Thus, this taskwas experimente-. paced .%Ithough adaptive to aa individual's ability, and ittended to demand maximal perfcrmance speed. Subjects ware also given immediatefeedback regarding "correct" and "wrong" additions.

4. In experiments with 23 of our subjects, a fourth task, an auditoryvigilance (AV) task, was required simultaneously with the first 3 tasks for 9minutes.. For the other 7 subjects, the technology was not functional. Thistask was an adaptation of the Wilkinson Auditory Vigilancia Task (Wilkinson,1969). In our AV task, long (520 msec) and short (160 msec) tones werepresented every 3 seconds. Ten percent of all tones were short. Short toneswre randomly di,:1ersed among the long tones. Subjects were required toidentify the short tones by pressing a foot pedal which registered bcth correctand incorrect identifications. This task was experimenter-paced, and there wasno immediate feedback on performance.

During performance testing periods, subjects performed the tracking task,the pattern memory task, and the addition task each for 3 minutes. At the endof each task, the video monitor displayed summary feedback on performance in theform of a statement of the amount of bonus money earned during the 3-minute taskinterval. Thus, the performance tests were framed as a repetitive series oftasks where favorable performance was rewarded by money. The scoring systemseverely penalized failures to respoe:1 LiO the task presentations, so thatinattention or sleep were heavily penalized, whereas errors representingcontinuing effort were less heavily penalized. In this way, the reward systemwas able to elicit very determined and motivated effort from almost allsubjects.

At the end of each 9 minutes, a 100 mm vertical line was displayed on the

I 6

IN.. . . . . . . . . . . . . . . . . . .. . . . . . . .on Min

left-hand side of the monitor screen. Seven areas along this continuum wereassociated with items of the Stanford Sleepiness Scale, which was presentedadjacent to the line (Table I). Subjects were instructed to use the joystick toquantify their subjective sleepiness or alertness on this continuous self-reportscale. Next, a similar linear scale was displayed on the right of the screen to

register a subjective self-report on a continuum scale ranging from completeattention to intense fantasy or daydreaming (Table 11). Subjects then indicatedhow much they had eaten or had had to drink and whether they had used therestroom during the prior 10 minutes. Food items (sandwiches, fruit, etc) wereassigned semi-quantitative numeric values by the experimenters (eg, one-fourth

sandwich - 2), and drinks were quantified as the number of 3-ounce cupfuls.Subjects used a numeric keyboard to indicate the units of food and drinkconsumed over the last ten minutes. After this information was provided, thecomputer program then presented AV task performance feedback along with summaryperformance feedback on the total amount of money earned over the past tenminutes. The program also displayed a projected hourly rate reflectingperformance levels during the past 10-minute module. Since in practice, mostsubjects were able to respond to all self-report items quickly within 15-20seconds, they were afforde4 about 40-45 seconds to rest, obtain food from therefrigerator, or visit the restroom before the tasks were again resumed. Ifsubjects were not ready to resume their duties at the end of this 40-45 seconds,they were penalized by the system for failures to respond. This sequencerepeated itself every 10 minutes throughout the periods of performance testing,resulting in 6 scores per hour for each performance task and each subjectivemeasurement.

Follow-up

At the conclusion of the stidy, while electrode!. were being removed,subjects were interviewed and osked to complete a brieZ questionnaire. Subjectswere then driven home unless they preferred to walk to nearby on-campus housing.Telephone follow-up contacts were made the next day with subjects who hadexperienced any physical or psychological difficulties in the course of thestudy, in order to assure the subjects' health. Later follow-ups were alsoobtained from time to timei.

RESULTS

Compliance and Psychiatric Symptoms

A major finding was that almost all of the subjt.2ts in the three 42-hourprotocols completed the studies. Two subjects who were recruited did not finishand vare ,not included in, the data analyses. One of these subjects, who wasassigned to the 6-and-I group, quit after only 3 hours of continuousperformance. This subject expressed a surprising lack of motivation, whichbecame apparent after one and one-half hours of testing, when his performancedeteriorated. A second 42-straight study was terminated by the expe4imentersdue to an equipment failure. Experimental group assignment was not considered asignificant factor in the termination of these two studies, so these twosubjects were replaced to complete the experimental design.

7

aw -

Three additional subjects, all in the 42-straight group, also dropped outof the study after 17, 26, and 40.5 hours of continuous performance. Sincethese 3 had all been randomly assigned to the 42-straight group and were allwell-motivated subjects, these failures were interpreted as experimentaleffects. These dropouts reduce the numbers of subjects included in the42-straight group statistical analyses for the latter part of the 42 hours.

Although the other 27 subjects were all able to complete the protocol, manyreported minor physical discomforts and psychological impairments ordistorbances. Examples of verbatim subjective reports are given in Table III.As is shown in Table IV, about half of all subjects reported some mildhallucinations, visual illusions, or experiences of derealization anddepersona]lization. Eight of 10 subjects in the 42-straight group experiencedone or more of these psychiatric symptoms. Subjects in this group experiencedsignificantly more of these symptoms than subjects in the other 2 groups (TableIV). Only 2 subjects in the 18-and-6 group reported such disturbances, andthese were only mild visual illusions. Four subjects in the 6-and-i groupexperienced mental symptoms including hallucinations. It is apparent that restbreaks were effective in minimizing psychiatric symptoms, and the 6-hour b -eakwas more effective than the 1-hour breaks. Several subjects also mentionedsomatic complaints such as nausea, headache, backache, etc, some of which werevery likely psychosomatic in origin. The majority of these complaints alsooccurred in the 42-straight group.

Performance Data and Subjective Measures

Examples; of performance scores for typical individual subjects appear inFig 2. In order to provide comprehensive analyses of these data, hourly groupmeans were plotted with confidence limits for each group separately see (eg, Fig3). In addition, the hourly means were plotted without confidence limits forall three groups on the same axes. To emphasize intra-subject variations inperformance over time while minimizing the contributions of inter-subjectvariability, the separate 10-minute scores for each subject were normalized(converted to z-scores) by subtracting the mean and dividing by theintra-subject standard deviation. Then normalized group means and confidencelimits also were derived. These normalized data were then replotted.Improvement or deterioration in thes&. z-normalized plots can be interpreted byappreciating the changes in z-normaltzed performance after the start of thestudy. A final plot was prepared to facilitate comparison of the 3 groups.Normalized scores from each group were plotted on the same axes, adjusting thescores so that for each group, the mean z-score for the initial 6 hours wasplaced at the O-level of the ordinate. Assuming that random assignment producedgroups initially equated during the first 6 hours of performance (before thethree protocols diverged), these latter plots present the most cogentdemonstrations of the divergences between experimental groups. Plots ofnormalized mean data were generally similar to non-normalized mean plots, but itwill be seen that the normalized plots give a more precise description of changewith narrower covfidence regions. Results for each measure will be describedseparately.

Tracking. In the tracking task, there was a clear and consistent

8

.t ;.~

divergence in the performance of the three groups during the 42 hours of thestudy (Fig 3). All three groups improved their performance steadily during thefirst 18 hours of performance, presumably because of a learning effect, althougha time-of-day effect could not be excluded. Early improvement was greatest inthe 6-and-i group which seemed to benefit from the one-hour naps. From 18 toabout 28 hours after the beginning of performance testing, that is, fromapproximately 2300 at night until 0900 the next morning, there was a progressivedeterioration in the performance of all three groups, due to a time-of-dayeffect or 24-hour rhythm. Both th% 18-and-6 and the 6-and-i groups then showedcontinuing improvements in performance later in the second day, but the42-straight group showed continuing deterioration due to fatigue. The 18-and-6

group showed superior performance throughout the second day, indicating that theone 6-hour sleep period preserved performance better than the five 1-hour naps.It is interesting that the tracking performance of all groups was always worsein the first hour after awakening than during immediately subsequent hours. Onfour or five occasions, tracking performance was worse in the 6-and-i group inthe first hour after a nap than in the last hour before the nap. Performancewas also poorer in the 18-and-6 group in the first hour after the 6-hour sleepperiod than during any hour of the 6 hours preceeding the sleep period. Thiswaking-up effect was in pact due to the fact that the performance tasks beganimmediately at the end of the sleep period, and subjects required some time tomove from the bed to the computer console to begin work. Nevertheless, since

decrements in performance were sometimes seen for several 10-minute testingepochs, this waking-up effect was not due to mechanical problems in gettingstarted alone.

Pattern Memory., A slight trend for pattern memory to improve was seen inall three groups in the first 18 hours (Fig 4). There was particularimprovement in the 6-and-i group, an apparent benefit of their naps. All threegroups showed some decrement from about 16 to 26 hours after the start ofperformance, due to a 24-hour rhythm or to time-of-day. The dip wasparticularly pronounced in the 42-straight group. Performance was markedlyimproved in the 18-and-6 group after the rest period, and even the 42-straight

group showed some improvement toward the middle of the second day, beforefurther deterioration appeared. On all occasions, performance was worse in thefirst hour after awakening than in subsequent hours, suggesting a waking-upeffect. Poor performance in pattern memory after waking up was seen almostentirely in the first 10 minutes after a rest period, so it could have beenproduced largely by difficulty in getting seated at the computer console,"especially since "no answer" responses were most frequent at this time.

-•ailures in the pattern memory task were produced both by incorrectresponses and by failures to respond within the 2-second time limit. Fig 5, agraph of "no answer" responses to the patt3rn memory task, shows that the42-straight and 6-and-i groups failed to respond much more commonly during thest-cond day of performance, ard the 6-and-i group had many failures to respond inthe first hours after awakening from rest periods. More failures to respondalso occurred in the 18-and-6 group following the 6-hour sleep period thaai atany other time for that group. Failures to respond were always most commonin the first hour after awakenings from rest perirxds.

Addition. After a rapid improvement in the first four or five hours,

9

-''>.

attributed to learning, addition performance deteriorated in both the42-straight and the 18-and-6 groups (Fig 6). It was interesting that thesubjects' addition skills substantially improved in the first 18 hours in the6-and-1 group, indicating that the benefit of learning was greater or thefatigue was less in these subjects because they received brief rest periods.Nevertheless, both the 42-straight and the 6-and-i groups deteriorated markedlyduring the first night from about 20 to 28 hours after beginning work, that is,from about 0100 to 0900. Mild recovery of performance occurred towards noon ofthe second day in these groups followed by further deterioration. The 18-and-6group improved markedly after the 6-hour rest period before deteriorating laterduring the second day. Although the relative improvement of performance duringthe second day for the 18-and-6 group was extremely impressive, the absoluteperformance of this group during the second day was not consistently superior tothe 6-and-i pgroup (Fig 6). This was most probably because the absoluteperformanca in the addi'.ion tF3k was consistently and significantly poorer forthe 18-and-6 group during the first day compired to that of the other two

groups. We have no explanation why the ai'solute performance of the three groupswas not equivalent at the start of V'e Jtudy. The deterioration in addition ofthe 42-straight group was particularly impressive, as these subjects were ableto perform less than half as many correct additions at the end of 42 hours ofcontinuous work as they were able to do at the beginning of the study.

In all three groups, there was a slight trend for the numbers of incorrectaddition responses to decrease as the study progressed (Fig 7). "Incorrectadditions" included both erroneous answers plus instances where subjects failedto answer within the allotted time. This result therefore suggests thatdeteriorations of correct addition completions were due to loss of speed ratherthan to errors of accuracy. Trends in the "no answer" responses to the additiontask, measured separately, were not large or consistent (Fig 8). Thus, thetrends in the "no answer" responses were also consistent with the interpretationthat the reduction in addition performance was largely due to a loss of speed.

Addition performance was always worse in the first hour after rest periods.Because this effect usually persisted 30 minutes or more after awakenings and

was more noticeable in correct and incorrect responses than in "no-answer"scores, it appeared to be due to impairments of speed and not due to mechanicalproblems in becoming seated at the console.

Auditory Vigilance Task. Correct responses in the auditory vigilance (AV)task were approximately stable throughout most of the first 18 hours of thestudy for the 42-straight and the 18-and-6 groups alt ough responses were morevariable in the 18-and-6 group (Fig 4). In the 42-straight group, performancebegan to deteriorate during the early evening. Absolute level of performance inthe 18-and-6 group were almost identical to those of the 42-straight groupduring this time. In contrast, the 6-and-i group demonstrated a possiblelearning effect over the first 6 hours and then maintained relatively improvedperformance during the rethen mainder of the first 18-20 hours. This againsuggests a possible benefit of the i-hour rest breaks in the first half of theperformance testing period. As the study progressed, a definite deteriorationin AV correct responses occurred in the 42-straight and the 6-and-I groupsbetween 18 and 24 hours after the begi-ming of the study. After 24 hours, AVperformance stabilized in the 42-straig'it group -- the rapid deterioration

10

- ----. *-t--

plotted in the final 5 hours vtay not be significant. In contrast., performancein the 6-and-i group was improved after the trough at about 24 hours. AVperformance in the 18-and-6 group was fairly stable during the final 18 hours oftesting and was at an almost identical level as during the first 18 hours.

False responses in the AV task which were previously occurring at a stablulevel increased steadily after 20-22 hours in the 42-straight group (Fig 10).False AV responses were fairly stable in the 18-and-6 group during both 18-hourtest periods except for a brief 3 to 4 hour decrease after the 6-hour rest.False responses in the 6-and-1 grouip increased after 21 hours. This slightincrease in AV false responses in 'he 6-and-i group was trivial compared to themuch larger increase which occurred in the 42-straight group during the lasthalf of performance tesring. The significance of false responses in the42-straight group was increased by the fact that this group made fewer falseresponses at the beginning of the study compared to the other two groups.

Sleepiness. Subjective sleepiness scores increased substantially in allthree groups during the first 18 hours of the study (Fig 11). The range of theincrease was from approximately "Functioning at a high level, but not at peaE;able to concentrate" to "Sleepy; woozy; prefer to be lying down; fighting sleep"on the Stanford Sleepiness Scale. It is interesting that during the first 18hours, even though the 6-and-i group indicated somewhat greater subjectivesleepiness during the first 6 hours compared to the other groups, 6-and-isubjects reported less sleepiness over the remainder of the first 18 hours.This was apparently due to the benefit of the brief naps. Compared to the42-straight group, increases in sleepiness as the study continued were much lessin the 6-and-1 group, also suggesting that the brief naps attenuated sleepiness.A very distinct peak in sleepiness occurred in the 42-straight and 6-and-igroups 24-25 hours after the beginning of the study, that is, at 0500-0600 AM ofthe second day, and sleepiness actually decreased during the course of thesecond day of performance. The increase in sleepiness in the early morninghours indicates that time-of-day or 24-hour rhythm effects were stronger thanthe effects of progressive fatigue in modulating sleepiness. The reportedsleepiness of the 18-and-6 group was less than that of the 6-and-i groupthroughout most of the second day, but this benefit was lost toward tte end ofthe second day when the total duration of the rest periods for the two groupsbecame roughly equivalent. In every instance, the sleepiness reported in thefirst hour after awakening was greater than that in the following hour.Increased sleepiness was prominsnt up to 20-40 minutes after awakening. Asmight be expected, the sleepiness of the 42-straight group was markedly greaterin the second day than that of the two groups which received rest periods.

Attention-Fantasy Scale. In general, the trends in theattention-fantasy scale were the reciprocal of trends in sleepiness, that is,at times when subjects reported increased sleepiness, less attention to theexternal enviromnnt and greater fantasy or daydreaming were reported (Fig 12).The only exception to this observation was that subjects always reported moresleepiness in the first hour after arousal, and they usually reported lessattention (more fantasy) during this hour. There were some occasions whensubjects reported more attention in the first hour after arousal than insubsequent hours. The range of the mean reports was from "awareness of presentsurroundings, thinking of what is happening" to "abstract and fanciful thinking,

X,

''*• ... - . .. - • . .. .. . C .-.. . . . - ,-_-yZ .••. •- . .....- - - . 7.;''-- . ': -L ,[•" ": ".'.•:' _•'," ' '# m N •

especially past and future".

Food Consumption. There was a slight tendency for subjects in all threegroups to report less frequeut food consumption as the study progressed,nevertheless, the trends were not strong compared to the large variability inthe timing of eating (Fig 13). There was a definite trend for subjects to eatin the first hour or two after a rest period than in subsequent hours. Thistrend was not as prcminent in che f'Lrst 10-20 minutes after awakening as laterin the hour.

Drinking. As with consuming food, there was a slight trend for subjects todrink less as the study progressed, but this trend was by no means prominentcompared to the-cverall variability of drinking occurrences (Fig 14). There wasno consistent trend for su,'jects to drink more in the first hour afterawakening.

Restroom Trips. There was no definite trend over time in the numbers ofrestroom trips for any group (Fig 15). Nevertheless, restroom trips were morecommon in the first hour after rest periods, but not especially in the first10-20 minutes.

DISCUSSION

The design of this study was rather unique in that subjects were requiredto display almost -:ontinuous sustained performance for extended periods of time.It was gratifying t*'at four separate performance variables plus the sleepinE.3sand attentiorr-fantasy scales all sbc-4d consistent trends.

In general, subjects had marked difficulty in sustaining truly continuousperformance for 42 hours. Perforvance on all taskE deteriorated, and three

W subjects had to discontinue participation despite high motivation. Mostsubjects who worked without rest periods experienced psychiatric symptoms suchas hallucinations, visual illusions, or sensaLions of derealizAtion. Suchsymptoms appeared much earlier than had been reported in sleep-loss studieswhere only intermittent performance was required. Although our subjects were

r not explicitly isolated from external sensations, interperoonal communicationwith subjects was minimi7ed to prevent distraction. Other sensory iup:ts werereatricted by the attentional focus demanded by the performance tasks. Wespeculate that a narrowing of atLentionial focus onto monotonous, repetitiveinputs led to these symptoms. Thus, without formally restricting the quantityor quality of sensory inputs and sicial contactc, continuous sustainedperformance may have effects m,-.h like thoseo of sensory isolation (Zubek, 1969).In future research, we plan to further explore which factors are mostinstrumental in so quickly producing psychiatric disturbances.

Subjects permitted 6 hours of rest in 42 hours suffered much lessC;'rrioration. Even one-hour ut,'s were beneficial even 6-13 hours fromiegZnning work, however, single 6-hour rest period reduced impairment moreoverall, particularly during the second day of testing.

A mixture of factors was found to influence continuous sustained

12

77

rL.-- - - - - -.-

performance.

At the beginning of the performance testing portion of the experiment, allthree groups showed some progres6ive improvements in several tasks. TheseImprovements were very likely due both to learning and to time-of-day effects.Although the subjects had practiced the tasks for up to an hour prior to the dayof the experiment, it would appear they had not yet reached asymptotalperformance. There was a tendency for improvement of performance to occur fromaround 0500 to 0900 or 1000 in the second day of the experiment as well as inthe first day, so time-of-day effects were also suggested.

Superimposed on learning effects, fatigue effects of varying strengthsupervened, so that improvements in performance related to learning werebalanced by increasing fatigue. Effects of fatigue early in the experiment weredemonstrated by the superior performances of the 6-and-I group after the firstand second rest periods. Since the 6-and-i group showed more improvement thanthe other two groups in the first 18 hours of performance, it is evident thatthe brief reot periods were useful in counteracting fatigue after as little as 6hours of continuous performance. There had been only minimal sleep deprivationbefore 18 hours of performance, since all subjects started from a 6-hour sleepperiod at their usual bedtime.

Approximately 18 hours after performance testing began (nn average, about2300), a new factor became manifest by an abrupt deterioration in performance inthe 42-straight and 6-and-1 groupo. It is evident that performance fell rapidlyduring the subjects' usual hours of sleep, after about 2300. This dip inperformance lasted until about 0900 the next morning, when a partial recovery ofperformance appeared. Since 18-and-6 subjects were allowed to sleep during thenight, their performance during night hours was not tested, but there was astriking improvement in performances of the 18-and-6 group after the 6-hour restperiod. In the second day, tne 18-and-6 group performed consistently betterthan both the 6-and-i group as well as the 42-straight group. This may beeither because the total amount of rest received by the 18-and-6 group beforethc second day (6 hours) exceeded that of the 6-and-1 group, or it may be tbhtthe more extended rest period was of more value than several shorter restperiods for sustaining performance. Another possibility is that rest during thenormal hours of sleep was more valuable to the 18-and-b group than rest periodsscattered around the 24 hours for 6-and-i subjects. There is some evidence thatthe total qu.ntity of rest received was critical, since in several tasks, theperformances of the 6-ard-1 and 18-and-6 groups tended to approach each othertoward the end of the second day. Nevertheless, even up until the final hoursof the expýriment, the 6-and-i group had still received I hour less total sleepthan the it'•-aud-6 group. This led to an unfortunate asymetry of design.

Rest periods had bi-phasic effects. In the first hour (and especially thefirst 10-30 minutes) after a rest period, performance tended to be actuallysomewhat worse and eC.eepiness somewhat increased compared to that prior to therest period. On the other hand, in subsequent hours, the rest periods seemed tobe of benefit for improving sustained continuous performance, and overall, therest periods were of most definite benefit. Poor performances immediately afterrest periods may to some extent have been due to distractions or mechanicalfactors. During the 10-minute modules immediately after rest periods, subjects

13

may have been at a disadvantage because of the time required to move from Zhebed to the computer console and become settled for work. Subjects tended toconsume more food, to drink more, and to visit the restroom more often in thefirst hour after rest periods. These activities may have kept the subjects away -from the computer console during testing and may have provided a psychologicaldistraction. Nevertheless, since impairments continued for several 10-minutemodules after rest periods, and since increasei subjective slcepiness wasreported after rest periods, distractions and mechanical factors alone do notrxplain the results. Wa con-lude that the rest periods did indeed produce abrief hangover effect w-ich impaired performance.

From the reasonably reliable hour-to-hour consistency of these results andthe relatively narrow confidence limits for mean performances for the threegroups of subjects, we infer these results have considerable reliability for theparticular experimental model which we have tested. The generalizability of theresults to other schedules of performance demands, to other physical and social

settings, or to other practical provisions for feeding and restroom visits isuncertain. Generalizability could only be definitely determined by varying theexperimental designs. '4

As we had predicted, performance on the various tasks was affected somewhatdifferentially by the continuous sustained performance model. The tracking taskshowed some actual improvements over *time in the 18-and-C and 6-and-i groups,whereas the deterioration of the -42-straight group was modest. This was to beexpected since the tracking task required only brief periods of sustainedattention, and performance could be largely self-paced. Performance in thepattern memory task also deto.riorated relatively little in the 42-straightgroup, and there was some improvement in the 18-and-6 group over time. Therelative success of subjects with this task may be related to thi fact that thepacing, although experimentally controlled, involved little time pressure. Incontrast, the addition task displayed a profound impairment in the 42-straightgroup, and the 6-and-i group showed considerable impairment beyond 18 hours fromthe beginning of the study. The difficulties with sustained performance on thistask may be related to its very high cognitive load, to its requirement forrelatively extended concentration and memory, to the extreme time pressure, andLlIto the fact that the timing (although adaptive) was entirely experiment-paced. ILess specific and discrete changes were observed in the AV task than in theother tasks, although this task had proven particularly sensitive in formerstudies of sleep deprivation. The relatl.ve lack of sensitivity of this task inour experiment may have been related to the technical difficulties weexperienced, lowering the numbers of subjects who were tested on this task ineach group. The fact that this task was required simultaneously with the otherperformance tasks may also have been a factor. The most likely explanation isthat the auditory vigilance task is not particularly sensitive to a sustainedcontinuous performance model.

In summary, continuous sustained peiformance produced rapid fatigue effectswhich were measurable after only 6 hours of sustained performance, beforesustained sleep deprivation could occur. Within a relatively brief period, lessthan 42 hours, rather serious decrements in performance occurred, and disturbingpsychological symptoms such as visual illusions, hallucinations, derealizatior.iand disorientation were reported. These aberrations in themselves may have

14

7! V

interfered with task performance. Some well-motivared young men were quiteunable to work for 42 hours continuously. Both 1-hour and 6-hour rest periodswere effective in attenuating performance decrements. One hour rest periodswere most valuable within the first 18 hours of sustained performance, but a6-hour rest period was more useful than six 1-hour rest pariods over a totalperformance interval of 42 hours.

These results indicate that impaired performance must be anticipated inpersonnel required to sustain continuous performance at electronic consoles muchearlier than would be predicted from classic studies of sleep deprivationutilizing intermittent performance testing. Substantial perfr.ian-edeterioration should be anticipated after 18 hours of continuous performance.In addition, cirnadian rhythm or 24-hour time-of-day effects produce prominentearly impairments of continuous performance. Subjects requ ied to performduring their normal hours of sleep will show marked impairments even within thefirst 24 hours of sleep deprivation. Manning strategies shoild be adapted toanticipate these dilficulties in any future conflict which would requirecontinuous sustained performance of key personnel.

15

REFERENCES

Allulsi, E.A. Influence of work-rest scheduling and sleep loss on sustainedperformance. In, Colquhoun, W.P. (Ed), Aspects of Human Efficiency.English Universities Press, London, 1972, pp. 199-215.

Hockey, G.J.R. Chances in attention allocation in a multicomponent taskunder loss of sleep. Brit. J. Psychol., 61: 473-480, 1970a.

Hockey, G.J.R. The effect of loud noise on attenticnal selectivity. Quart.J. Exp. Psychol., 22: 32-36, 1970b.

Johnoon, L.C. and Naitoh, P. The Operational Consequences of Sleep Depriva-.tion and Sleep Deficit. NATO/AGARD AGARDograyh AGARD-AG-193, 1974.

Johnson, L.C. Psychological and'physiological changes following total sleepdeprivation. In, Kales, A. (Ed), Sleep: Physiology and Pathology.J.B. Lippincott, Philadelphia, 1969, pp. 206-220.

Kjellberg, A. Sleep deprivation and some aspects of performaxce: I. Problemsof arousal changes. Waking and Sleeping, 1: 139-143, 1977a.iI

Kjellberg, A. Sleep deprivation and some aspects of performance: II. Lapsesand other attentional effects. Waking and Sleeping, 1: 145-148, 1977b.

Kjellberg, A. Sleep deprivation vid some aspects of performance: III.Motivation, comment and conclusi. Waking and Sleeping, 1: 149-153, 1977c.

Naitoh, P. Sleep Lose and Its Effects on Performance. Navy Medical Neuro-psychiatric Research Unit Technical Report No. 68-3, 1969.

Wilkinson, R.T. Interaction of lack of sleep with knowledge of results, re-peated testing, and individual differences. J. Exp. Psychol., 62:263-271, 1961.

Wilkinson, R.T., Edwards, R.S., and Hainer, E. Performance following a nightof reduced sleep. Psycho. Sci., 5: 471-472, 1966.

Wilkinson, R.T. Sleep deprivation: Performance tests for partial and selec-tive sleep deprivation. Prog. Clin. Psychol., 8: 28-43, 1969.

Woodward, D.P. A User Oriented Review of the Literature on the Effects ofi Sleep Loss, Work-Rest Schedules, and Recovery on Performance. Office

"of Naval Research, ACR-206, 1974.

Zubek, J.P. Sensory Deprivation: Fifteen Years of Research. Appleton-Century-Crofts (Meredith Corp.). New York, 1969.

16

TODLE 1. Stanford Sleepiness Scale Items

Almost in reverie; sleep onset soon.; losing struggle to remain awake.

Sleepy; woozy; prefer to be lying down; fighting sleep.

Foggy; slowed down; beginning to~ lose interest in remaining awake.

A little foggy; let down; not a~t peak.

Relaxed; awake, responsive, but not at full alertness.

Functioning at a high level, but not at peak; able to concentrate.

171mlj

TABLE 2. Attention-Fantasy Scale Items

Present-time perceptual scanning and perceptual-motor activities.

Awa.-eness of present surroundings, thinking of what is happening, etc.

Concrete, realistic, and present-time thinking or planning.

Abstract ind fanciful thinking, especially past or future.

Vivid dreamlike fantasies with bizarre or symboli. content and strongei0tions such as wishes or fears.

18

r .. . .. . .l l . . ... .l l. . . . ." .. . . . . .• Z• '

row__-.. - - ..... -'.

Table 3: Examples of Psychiatric Disturbances

Type of Example Subjective Description

Visual Illusion "....it looked like there was water starting underneaththe seat flowing back towards the wall."

"I would look at my jeans and scR lots of red thread inthem.And a kinda sheet of red in thb paint...."

Disorientation "....for a period of 2 hours, the whole room....I wasnever sure where the door was."

Derealization "All of the objects were in the same place as the objectsI was used to, but they looked different."

"I thought I w&s at home doing this and my roomate wouldcome in and I Just wanted to say P-mething to him and Irealized he wasn't here."

Hypnagogic "I was just getting ready to go to sleep and all of aHallucination sudden I heard real low voices and someone talking about

Earth....It was just nonsense."

Hallucination "....it would look like someone was looking out of tbescreen, like it was a window .... like there was someorebehind there looking out."

"I found myself trying to receive something from whal:I thought was a woman standing in front of me, whileactually I was just grabbing at the card next to theTV set."

19

i _ __ _

TABLE 4: Psychiatric Disturbances During Performance Testing

Type of Disturbance

Illusion3 Derealization Hyptiagogic Hallucinations Number ofGroup and and Hallucinations Subjects

Distortions Disorientation with any Symptoms

42 7 5 1 3 8

18&6 2 2

6&W 1 2 4

Total -14

For the number of subjects with aay sy.ptc•min the three groups, X2 -7.49, p<O.0 2 5

20

S ... • '• '-> •' • "'•" . .. ... . , . ' r - " - -

FIG 1. DIAuwmTIC ILLUSTRATION OF E)PERIMENTAL GROUP PROTOCOLS

42 Straight

T T T T. T TI TII

II

III

18 and 65

T T T T-

6 and I

• . T M TNm T w.,T I Tom T ,

0 6 12 18 24 30 36 42 48 KHOURS

; I Ii SLEEP T x TESTING

21

a ,a-' - .... .. _,, ,, -, ,i ,, " 4, .; . "i"• • " • a • •

25

o

15

0 23

-+ --- -----.6 A..-"-'•---------, •,,,r y ̂ •

2 4 6 6 10 12 14 16 I 20 2I 2I 26 30 22 34 6 4 40 2TIME ( Hours)

FIG. 2A IndiVidual data points collected every 10 minutes for tracking,pattern memory, and addition performance plus subjectiveestimates of sleepiness. These data were obtained from a repre-sentative subject in the 42-straight group.

22

--- ------------ -------

10

+ F j•_•SLEEPI NESS •i

2 4 6 6 10 12 14 16 16 20 22 24 26 28 30 32 34 36 36 40 42FTIME (Hours)

FIG, 2R Individual data points collected every 10 minutes for tracking,pattern memory, and addition performance plus subjective estimatesof sleepiness. These data were obtained from a representativesubject in the 18-and-6 group.

Wi .Iý

0 23

AA

29 TRACKING (HITS)

ow 0 L PATTEN WEMORYC qO. CORRCT )

0n24

0 0ADD ION00(NO. L.AMU•CTM

F SLEEPINESS A

2 4 6 8 10 12 14 16 20 2 24 26 '. 30 32 3 6 3 40

TIME (Hours)

FIG, 2c Individual data points collected every 10 minutes for tracking,

pattern memory, and addition performance plus subjective estimates

of sleepiness. These data were obtained from a representative

subject in the 6-and-1 group.

24

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

TRACKING

CD

U)

CJ o

Ln I'

'N-C1{ * 2- TRRI H

I ]I I ' l I

S-I !II

ZCý

CV)

CD

SiB-RNU-60..00 6.00 12.00 18.00 214.00 30.00 36.00 142.00

ELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1100 1700 2300APPROXIMATE TIME OF DAY

FIG. 3A Hourly means of normalized tracking performance for the three experimental

groups. Group data are plotted with the mean z-score of the first 6 hoursset to the 0 level on the ordinate. Means were derived from z-scores forthe entire data of ea'cn subject.

25

TRACKING

an

0

C;

, Z

o

CC

0.0 60.0O0 12.0O0 18.0 2O 4.00 3b. O0 A6. 00 42.O00ELAPSED TIME (HOURS)


FIG. 3B Hourly means of normalized tracking performance for the 42-straightgroup. Means were derived from z-scores for the entire data of eachsubject. Also plotted are 95% confidence limits.

* 26

LU

i cj

0Jwmi 4 •..

|-

TRACKING

18-RND-6

0

o 1

LUC

C

0 '

0.o00 6.00 12.00 18.00 24.t 00 36.00 L12.00ELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1100 1700 2300

APPROXIMATE TIME OF DAY

FIG. 3C Hourly means of normalized tracking performance for the 18-and-6group. Means were derived from z-scores for the entire data of eachsubject. Also plotted are 95Z confidence limits.

27 .

'•'' m l i i - •' .. . '-• - - " . ... . • . . . . '---' i-. .i i- _ _ _ _ _ _ _ _ _ __I_ _I_ _ a

TRACKING

" 6-RND)- 1

r,j

tinv, M

!"J I I

I' I'0.00 6'.00 12. 00 18.00 214.0 o 3b. oo 316.o o iký. ooI [ ELAPSED TIME (HOURS)


FIG, 3D Hourly means of normalized tracking performance for the 6-and-igroup. Means were derived from z-scores for the entire data of eachsubject. Also plotted are 95% confidence limits.

28

• ., .......

..........

TRACKINGC)0

'C L42-STRAIGHT

CC,

U.

Cr

00

0

0-

'1 0

"00oo 6.00 1o2.00 18.00 214.00 3b.00 36.00 42.00


0500 1i00 1700 2300 0500 ii00 1760 2300APPROXIMATE TIME OF DAY

FIG. 3E Houily means of tracking performance for the 42-straight group. Alsoplotted are 95% confidence limits.

29

TRACKING

NL- 18-RND-6

Z- 4CC

oo 0. 00 1*. 00 1 0" 0 4 oo 3b . oo 3 11. 00 ý. oo

ELAPSED TIME (HOURS)SI I . .. I I I p


FIG. 3F Hourly means of tracking performance for the 18-a-d-6 group. Alsoplotted are 95% confidence limits.

30

TRACKINGC

6- IN' '- 1

LiX0

0" .OO 6.0O0 12. 00 1b.0O0 24.0O0 00 0•6 0 2.O00

ELAPSED TIME (HOURS)S 0500 1100 1700 2300 0500 1100 1700 2300


FiG. 3G Hourfy peans of tracking performance for the 6-and-i group. Alsoplotted are 95% confidence limits.

41~

31

IN.ic'aJ /. , . •

4, -

PATTERN MEMORY CO6ECTC

0 #

U,

C)'V)'

o A . " "I

LLJAAt,, LI

• }_.~Lr) !II

i42-STRRIGHT

18-RND-6

Lf, 6-AND-1

0.o00 6.00 12.00 18.00 241.00 30.00 36.00 o2. 00.-ELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1100 1700 2300

APPROXIMATE TIME OF DAYI

FIG. 4A Hourly means of normalized pattern memory performance for thl threeexperimental groups. Group data are plotted with the mean z-score

1o the first 6 hours set to the 0 level on the ordinate. Means werederived from z-scores for the entire data of each subject.

32

w ..... - --- ... ...... . . :, . . .

-. ;

PATTERN MEMORY CORRECTCo

"42-STRAIGHT

0 I0-

f 0

0.00 6.00 1,2.00 1,.00 24..0 3bqoo 31.o0 0 '2.00



FIG. 4B Hourly means of ncrmalized pattern memory performance for '.he 42-straightgroup. Means wern derived from z-scores for the entire data of eachsubject. Also pl4otted are 95% confidence limits.

33

Munn;.,_._,,._,

PATTERN MEMORY CORRECT

•9 -18-AND-6

0 ICC

C'

wo

0

0o0oo 6'.o 1'2.00 i.oo 214.00 3b.00 36.0oo 12.00ELAPSED TIME (HOURS)

0500 i01700 2300 0500 1100 1700 2300APPROXIMATE TIME OF DAY

FIG. 4C .oSurly means of normalized pattern memory performance for the 18-and-6group. Means were derived from z-scores for the entire data of eachsubject. Also plotted are 95% confidence limits.

34I!134

it .~.. .


Co{D• -6-RND- 1

0/

Z ,o

•0 1.0 -- 1.00 I•.00 1,8.00 24.00 30.00 16.00 42.00ELAPSED TIME (HOURS) :

0500 1100 1,700 2300 0500 1100 1700 2300APPROXIMATE TIME OF DAY

FIG. 4D Hourly means of normalized pattern memory performance for the 6-and-I

group. Means were derived from z-scores for the entire data'of eachsubject. Also plotted are 95% confidence limits.

•,, 35

i: .. . .... .


"L42-STRRIGHT

DD-d

z"-4LU

00 6'.00 12.00 16.00 2. oo 3b. co 36.00 o2.00ELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 II00 1700 2300APPROXIMATE TIME OF DAY

FIG. 4E Hourly means of pattern memory performance for the 42-straight group.Also plotted are 95Z confidence limits.

36


18-RND-6

LoUI-

CD

0.00 .0120 15.00 2~4.00 -3'0.00 3'6.00 412.00 IELAPSED TIM E (HOURS)

| I I I !

G500 1100 1700 2300 0500 1100 1700 2300APPROXIMAiL TIME OF DAY

FIG. 4F Hoqrly means of pattern memory performance for the 18-and-6 group. Alsoplotted are 95% confidence limits.

37

37


" B6-RND-1

0

c-"

C[:

= 0' 6'.00 12.00 18.00 214.00 3b.00 36.00 '42.0cELAPSED TIME (HOURS)

0500 ii60 1700 2300 0500 1100 1700 2300APPROXIMATE TIME OF DAY

;I

FIG. 4G HRourly means of pattern memory performance for the 6-and-i gi:oup. Alsoplotted are 95% confidence limits.

38

............-.----!

;if 3B

PATTERN MEMORY--NO ANSWER

0

Ur)II

0!

ZOr

LLJ~

€JII

C) I

I

4,2-STRRI GHT

-- - IS-RND-6C)

-4

'0.00 6'.0o 12.0o 18.00 2'.0oo b.oo 3.oo 3 42.otELAPSED TIME (HOURS)

0500 1100 1700 2.00 0500 1100 1700 230(0


FIG. 5A Hourly means of normalized numbers of failures to respond (no answer)during the pattern memory task for the three experimental groups.Group data are plotted with the mean z-score fo the first 6 hoursset to the 0 level on the ordinate. Means were derived from z-scoresfor the entire data of each subject.

39

~'&

-• -•- i -l I- II


1•- '42-STARIAGHT

c:)uJZl-r

o2

rrd

Cr0

i0 I I I i I0.00 6.00 12.00 Ib. O0 -24.O0 3b0.0o 36.00 1'2. 00



FIG, 5B Hourly means of the number of failures to respond (no answer) during thepattern memory task for the 42-straight group. Also plotted are 95Z

confidence limits.

40


Cr0

0

a;"1 8-RNIJ-6

Co

0.00o 6.0 oo4i.00 oo.0oo 24.0oo 30.00o 36.0 oo. ocELAPSED TIME (HOURS) -

0!


FIG. 5c Hourly means of the number of failures to respond (no answer) during thepattern memory task for the 18-and-6 group. Also plotted are 95%

, confidence limits.

c41

0:, S

- !

00,,


c.c

6-RND-1

0-)DL z=;-I

ao

10.00 6'.00 12.00 18.00 214.00 3'0-00 A'.00 4'2.00ELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1100 1700 2300

FIG. 5D 1gourly means of the number of failures to respond (no answer) during the

pattern memory task for the 6-and-i group. Also plotted are 95%confidence limits.

42

, &

ADD IT ION

LI) %%

Z Ný

CCI

Li V V

L2- STBR IGHT

-18-AND-6

LOl - -AND-1

'0.00 6 .00 12.00 18.00 24.00 30.00 36.00 4~2.OC~ELAPSED TIME (HOURS)


4 FIG. 6A Hourly means of normalized addition performance for the three experimentalgroups. Group data are plotted with the mean z-score of the first 6 hoursset to the 0 level on the ordinate. Means were derived from z-acoresfor the entire data of each subjeqt.

43

Mac ~~

ADDITION

-42-STRRIGHT )

C,-

0

°)Z

Cc

LLJO

C,.

0.00 6. OL 12.00 1'8.00 24.0oo s0.oo 336.o0ELAPSED TIME (HOURS)

0500 1100 1700 2300 0560 1100 1700 2300


FIG. 6B Hourly means of normalized addition performance for the 42-straightgroup. Means were derived from z-scores for the entire data of eachsubject. Also plotted are 95% confidence limits.

44 6j

ADDITION

* Co

-1-RND-6co

oL-Jo

0-

CD,

'0.00 6 ..00 12.0 OB.O 0 2 -. ,0 3b. oo 36.00 Qý-O(ELAPSED TIME (HOURS)

0500 1100 17XO 2300 0500 Ii00 1700 2300APPROXIMATE TIME OF DAY

FIG, 6c Hqurly means of normalized addition performance for the 18-and-6group. Means were derived from z-scores for the entire data of eachsubject. Also plotted are 95% confidence limits.

45

4

:1 ~ ADDITION 0i

6-RND- 1

q:.

1 ' Lo

W

I.

'0.00 6.00 12.0o 1,8.o0 24.00 30.00 3,.0 ýo2.oELAPSED TIME (HOURS)

0500 1i00 1700 2300 0500 1100 1700 2300' 1 APPROXIMATE TIME OF DAY

Nil1FIG, 6D Hourly means of normalized addition performance for the 6-and-i group.

Means were derived from z-scores for the entire data of each subject.Also plotted are 95Z confidence limits.

4646

ADD IT ION

Li2-ST9RRI GHT

Cu

zZ-4

Cr,

'b o61.00 1'2.00 1'6.00 2~4-00 310.00 316. 0 0 45. 00ELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1100 1700 2300


FIG. 6E Hourly means of addition performance for -he 42-straigbt gruup. Alsoplotdae95% cofdne4iis

47

ADDITION

U,

18-RND-6

Cr.

0

0

C -

o1

0o 0 6.0 1'2 0 •.0o 18.0oo 21. 0oo 3b. oo 3'6.0 oo2.0ooELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1100 1700 2300


FIG. 6F Hqurly means of addition performance for the 18-and-6 group. Also

plotted are 95% confidence limits.

JI

48

•' " •' : •Pr' • , ,• L ,",.-- •.,•., : 0; •+,._o.

- - WWWW

ADDITION

CD

0o

'• 6-AND-1

° iJo

oo 0 6'.0oo 1,. 00 1b3.o0 24,.0oo 30.0 3o16. 00 .0o0,l ELAPSED TIME (HOURS)


FIG. 6G Hourly means of addition performance for the 6-and-I group. Also plottedate 95% confidence limits.

I0

49 l

L.

.

ADDITION INCORRECT

cn

0

L0

U,j

0 %

CU)

I4

C ) 1-

U') I BRN -

'I '

S0500 1100 1700 2300 0500 j100 1700 2300APPROXIMATE TIME OF DAY

FIG 7A Houxly means of normalized numbers of incorrect additions for the threeexperimental groups. Group data are plotted with the mean z-score, of thefirst 6 hours set to the 0 l.evel on the ordinate. Meaivi were derivedfrom z-scores for the entire data of each subject.

J50

iI

ADDITION INCORRECT

Cý- 42-STRRIGHT

Wr)

zw

Mcr)

0

Ln

Th.oo 6'.00 12.00 18.00 24.00 3b0.00 A'. 00 -42ELAPSED TIME (HOURS)


FIG. 7B Hiourly means of incorrect additions for the 42-straight group, Alsoplotted are 95% confidence limits.

51

ADDITION INCORRECT

0ý - N -

0lc ZI0C

W A

0KCCC )0

Lfl

Th 00 6.00 12.00 18.00 24.00 30.00 36.00 UP2.00

ELAPSED TIME (HOURS) fI0500 1100 1700230 0500 1100 1700 2300


FiG. 7C Hourly means of incorrect additions for the 18-and-6 group, Alsoplotedare95% confidence limits.

52

iI

ADDITION INCORRECT

a;" 6-RND-1

Ln--0

o

-,"1ooo 6'.0 112I. oo 0 1.00o 24. oo 'b. oo 316. 0o 0 @. ocELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1100 1700 2300APPROXIMATE TIME OF DAYA

FIG. 7D liourly means of incorrect additions for the 6-and-I group. Also

plotted are 95% confidence limits.

a53

S...... .- ••. ••.• ••'..... mwpW6-A N D -i•-' . . .

0 ¢ , •...

ADDITION--NO ANSWER

C3

C:)

-fA

L0'I ' fA

C:-V A\/i

C)

UL-STRAIGHT

CD- 18-PNO-8LI.) - 6-RND-1

'0.00 6.00 12.00 18.00 214.00 3b.00 316. 00 '42.OC 'ELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1.100 1700 2300APPROXIMATE TIME OF DAY

[IG 8AHul en'fnraieIubr o alrst epn n nwrduring the addition task for the three experimental groups. Group

data are plotted with the mean z-score of the first 6 hours set tothe 0 level on the ordinate. Means were derived from z-scores forthe entire data of each subject.

54I

ADDITION--NO ANSWER42-STRRIGHT

Lii

SC

'1

,:0

CbD.00 6.00 12.00 1'6.00 24.00 30.00 36.00 0.



FIG. 8B Hourly means of the number of failures to respond (no answer) during theaddition task for the 42-straigilL group. Also plotted are 95%confidence limits.

S•1 1[i'.1

ADDITION--NO ANSWER

00

•- 18-RND-6

(o

cc0

H uJ

|00!

0'.oo 00o6.00 1.00 1'8.00 2t4.00 3b.oo 36.00 142..I00ELAPSED TIME (HOURS) I

I0500 1100 1700 2300 0500 1100 1760 2300


FIG, 8c Hourly means of the number of failures to respond (no answer) during theaddition task for the 18-and-6 group. Also plotted are 95Zconfidence limits.

5I56

.•,' • ! • r,, • _s"'ll m l,,ir • ...,.,...,,L .•,.•, .:;, ,,I ,•'_,.• • .' <, I

ADDITION--NO ANSWER

6-RND-1

0

CDI

CDLo zI

I00 6.00 1700 2300 2500 ii00 1700 2.00ELAPSED TIME (HOURS) '

0500 1100 1700 2300 0500 1100 170'0 2300APPROXIMATE TIME OF DAY

FIG 8D Hourly means of the number of failures to respond (no answer) during the

addition task for the 6-and-I group. Also plotted are 95%confidence limits.

57

"AUDITORY VIGILANCE CORRECT RESPONSES

I

0*

71 APRXMT IEOAA

oIm

--- SRAGHc'J.

05,,0 '-1mn0 17u0 23 ou 050 11e 1700 23h

z-score of the first 6 hours set to the 0 level on the ordinate.Means were derived from z-scores for the entire data of each subject.

L.'....58.

ELAPSD TIE (HORS)

AUDITORY VIGILANCE CORRECT RESPONSES

'42-STRR IGHT

00:

0.060 `2 0 80 4o b o .0 12 '

ELAPSE TIME HOURS

05016170200501017020APRXMAETMEOrA

FI.9Crymaso nraie uioyvglac efrac o h

02srih ru.Maswr eivdfo -crsfrteetr

daao0ahsbet lopotdae9%cniec iis

05

Co. -gn


18~-ANO-6

0A

ccJ

hii

6.100 6 ou 1'2.00 1,8.00 214.00 3b'0 .00 3'2.00



FIG. 9C Hotirly means of normalized auditory vigilance performance for the18-and-6 group. Means were derived from z-scores for the entire

~J ~ data of each subject. Also plotted are 95%*confidence limits. '~6

LI L I C..'

( I AUDITORY VIGILANCE CORRECT RESPONSES

Ln

'I U7 6-R-AND-1

HI'.

m/C,,

6.0 1ý0 18aoo.o .00 -4. oo 3b.oo Ag.00 2.00ELAPSED TIME (HOURS)


FIG. 9D Hourly means of normalized auditory vigilance performance for the6-and-l group. Means were derived from z-scores for the entiredata of each subject. Also plot:ted are 95%, confidence limits.

61

• iLI . I .• •


"Lq2-STRAIGHT

0 ý7t'

a0o

0J

C

oo B'. o0 12.00 ole.Do 24i.0o 3b. oo 36.00 o,. o


0500 1100 1700 2300 0500 1100 1700 2300APPROXIri',TE TIME OF DAY

FIG. 9E Hourly means of auditory vigilance performance for the 42-straightgroup. Also plotted are 95% confidence limits.

62

1P.9 I


18-ANO--6

0v

a:I

0jZ:

c0cý

91

T00.0 .0 b0 .0 bo3.0 2 0

.050 6100 1700o 23800 05400 11.00 17600 12300


FIG- 9F Hourly means of auditory vigilance performance for the 18-and-6 group.Also plotted are 95% confidence limits.

63

AUDITORY VIGILANCE CORRECT RESPON~SES

0 6-RNO- I

Morv0r

T.ooe0 ý0 ,.0 40 bo- 360

ELPE TIE(HUS

05.00 6100 17200 236D0 0500 110.0 170 2300


FIG. '9G Hourly means of auditory vigilance performance for the 6-and-i group.Also plotted are 95% confidence limits.

"d~

64

AUDITORY VIGILANCE FALSE RESPONSES

U,

NL

L, i v" / '"I

N. ix.!~SI

V V

42-STRAIGHT

m'. 6-AND- I

0.06.00 1'2.00 18.00 24.00 3'0.00 316.00 Ui2.00ELAPSED TIME (HOURS)


FIG. I0A Hourly means of normalized numbers of false responses for the threeeiperimental groups. Group data are plotted with the mean z-scoreof the first 6 hours set to the 0 level on the ordinate. Means werederived from z-scores for the entire data of each subject.

65


.42-STRR I GHT

z

° /Zcc•w

Cr,

ýji

0".00 6.00 1o2.00 1o.00 2.0 oo b. oo 3,6.00 o2. DoELAPSED TIME (HOURS)

0500 1100 1700 2300 0500 1i00 1700 2300APPROXIMATE TIME OF DAY

FIG. 10B Hourly means of numbers of false responses for the 42-straight group.Ki Alsd plotted are 95% confidence limits.

66


, m --

18-RNO-6

zc"

o

CC

KW

0

0 6.00 1,2.00 18.00 214.00 3b.0o 3'6.00 .0cELAPSED TIME (HOURS)


FIG. lOc Hourly means of numbers of auditory vigilance false responses for the18-and-6 group. Also plotted are 95% confidence limits.

67


cr0

CIN

0. 6-1ND-4

ia:

'0500 11.00 12.00 16.o02 00 205 60o0o 310 1700 I230oi ELAPSED TIME (HOURS)

Ij , , I I ,0500 1100 1700 2300 0500 1100 1700 2300

APPROXIMATE TTME OF DAY

FIG. 1OD Hourly means of numbers of auditory vigilance false responses for the6-and-i group. Also plotted are 95% confidence limits.

68

' - -i-- -- ~

SLEEPINESS

CD-

Un

II

C00

Cc,

/1C ID

"11 '

0--

Cr)

L '42-STRAIGHT

-6-AND-1

zI , , I \I

101.00 6 .00 1 12.00 1 18.00 2 14.00 310. 00 316. 0 0 '4.

- ELAPSED TIME (HOURS)


FIG. 11A Hourly me..ns of normalized subject--ve sleepiness estIimates for the. three.

experimental groups. Group data are plotted witY. the mean z-score of

the first 6 hours set to the 0 level on the ordinate. M4eans were

derived from z-scores for the entire data of each subject.

II

I. -R 69

'o~~~~~~~oo~~~ '. oo .' ..o•'.o '. b.o . oo . . o.

SLEEPINESS

Cý

42-STRRIGHT

CD

0 •LE

.0 o.01ý o 18.0oo 24.0oo 30.0 Ao3. 0o ý2. DoELAPSED TIME (HOURS)


FIG. 11B -Hourly means of subjective sleepiness for the 42-straight group.Also plotted are 95% confidence limits.

70

SLEEPINESS

18-RNO-6

L /ý

Cfl

CDO 1.0 6 .00 1.0 0 1,8.00 24. OO 30.00 3b.00 '42.O00ELAPSED TIME (HOURS)


FIG. 11C 'Hourly means of subjective sleepiness for the 18-and-6 group.' ~Also -?lotted are 95% confid]ence limits.

71

SLEEPINESS

C;.

00

"- 6-RND- 1

Zw"Ccna(:

CD

O1.00 6'.O0 12.O0 lb. O0 214.Oo 30..oo 36.0O0 '42.O0



tIFIG. I1D Hourly means of subjective sleepiness for the 6-and-i group.

Also plotted are 95Z confidence limits.

72

ATTENTION-FANTASY

Ln

C)

LO

CE:

LA)

42-STfA IGHT1 B-RND- 6

LA') - 6-RND-1

0'.00 6.00 12.00 18.00 24.oo 310.00 316.00 '2.OC



FIG. 12A Hourly means of normalized attention-fantasy scale estimates for thethree experimental groups. Group data are plotted with the mean z-scoreof the first 6 hours set to the 0 level on the ordinate. Means werederived from z-scores for the entire data of each subject.

73

SA

ATTENT TION-FANTASY

" LI42-STRR IGHT0

8-

0'

oO0

120

0.0ii000 0.00 0 0500 3i0.00 1020

+•, APPROXIMATE TIME OF DAY

i FIG. 12B :Hourly means of attention-fantasy scale for the 42-straight group.Also plotted are 95% confidence limits.

74

.........

_______,__.________-___-- '' ¶ W " ¶, .i • , ,.••• i,- •.¶'w•. * "•, • 1."-

ATTENTION-FANTASY

"18-RND-6

0 • //

LUJ

0

Ob.0 6.0O0 ý. O0 18.00 2l.0O0 30. O0 36.00 42.00O



FIG. 12c H•ourly means of attention-fantasy scale for the 18-and-6 grou~p.Also plotted are 95% confidence limits.

75

07

0

ATTENTION-FANTASY

o

- 6-RNJ- I

C

0

CU

II

h. 0 0 6.00 12.00 1e.00 24.0 30.00 36.00 LI'2i0i0ELAPSED TIME (HOURS)j

0501100 1700 2300 0500 1100 1700 230'0APPROXIMATE TIME OF DAY

FIG, 12D Hourly means of attention-fantasy scale for the 6-and-i group.Also plotted are 95% confidence limits.

76

""Now--

FOOD

CDLn

CD

L$)(.'J

.-4

II

L€:J pi ' XI "I" V,'"A'

'42-STRRIGOHT

5-AND-i

0.0 6.00 12.00 18.00 24.00 30.00 6.00 412.00ELAPSED TIME (HOURS)


FIG. 13A Hýourly means of normalized estimates of food consumed for the threeexperimental groups. Group data are plotted with the mean z-score ofthe first~ 6 hours set to the 0 level on the ordinate'. Means werederived from z-scores for the entire data of each subject.

77

FOOD

CD

-42-STRRIGHT

=0

Cu:v

10

Cr;

.0O0 6. 00 I•..00 18.0O0 24.0O0 .3b. O0 Ag. 00 2L O0000 6ELAPSED TIME (HOURS)


iIFIG. 13B .Iourly means of food consumed for the 42-straight group. Also plotted

are 95% confidence limits.

78 .

WINN = ,

FOOD

18-RND-6

qc-4

ICD0r

0.00 6.00 12.00 1,8.00 214.00 30.00o 36.00 142.00ELAPSED TIME (HOURS)


FIG. 13c h1ourly means of food consumed for the 18-and-6 group. Also plottedK are 95% confidence limits.

V 1 79

FOOD

CE

o•- 6-ANIJ-1

U

,..

I;;

S'.00 6'.00 1 .00 18.00 2o.0,oo 30.0 36.00 41.00ELAPSED TIME (HOURS)

S0500 1100 1700 2300 0500 1100 1700 230APPROXIMATE TIME OF DAYV

FIG. 13D Hourly means of food consumed for the 6-and-i group. Also plottedare 95% confidence limits.

80

DRINK

-!-(U

V to

InIo 1

rjLn .AN -

3. . 00* U,

-- 1S2-STMRIGOHTl

CD fl-RND-6I• • 6-RND-i r

0.00D 6'.OO 1'2.00 1'8.00 2LI.OO 3'0.00 36.00 I42.0OI



FIG. 14A Hourly means of normalized numbers of drinks consumed for the threeexperimental groups. Group data are plotted with the mean z-score ofthe fi::st 6 hours set to the 0 level on the ordinate. Means werederived from z-scores for the entire data of each subject.

81

________________,___,.___+ '

DRINK

,42-STRAIGHT

IL

roq:m

U,

'o. Ioeo I. -o Io • o 3• o ao ,o

0.0 6.01.~ t.0 24. 00 30.00 36.00 'q2.00



FIG. 14B Hotirly means of numb'ers of drinks consumed for the 42-straight group.,Also ploted are 95% confidence limits.

82

DR INi K

0

*" 18-RND-6

N

IIILn"

C

A A0

0.0 60 ( 0 1.0 2.0 3.0 96.004.00


:I II -I

0560 1100 1700 2300 0500 1100 1760 2'300:ii' APPROXIMATE TIME OF DAY

FIG. 14C Hourly means of numbers of drinks consumed for the 18-and-6 group.Also plotted at, )5% confidence limits.

""3

• .0

U , .. . .. F ... . . = ~ - . _ . i: 7 ' . . . .• -.. .. • " ' : ... .± - '• • ? l. . . .. , r : • : ' . . .. . - -". . . . . . . .. ] . . . . .)

DRINK

0:: . 6-ANO- I4

g'dj

* "-

0

(no

w

H6 1l0 1 00 6.00 1200 18.00 24.o00 30.00 A "00 41 2.00


II I


FIG. 14D Hourly means of numbers of drinks consumed for the 6-and-I group.Also plotted are 95Z confidence limits.

84

w". -

RESTROOM TRIPS

Ln

U,

010

-4I2-STRAIGHT

18-RND-S

0

Ln i 6-AND-1

Cu.

1'0 .00 6 1 oOI 12Do 1b8.00 24~.00 3b0.0 oo ?1 00 412. 0* -~ ELAPSED TLME (HOURS)

0500 1100 1700 2300 0500 11600 1700 2300v APPROXIMATE TIME OF DAY

Firs. 15A Hourly meauis of normalized numbers of restroom trips taken for the threeexperimental groups. Group data are plotted with thce mean zv~scoreof the first 6 hours set to the 0 level on the ordinate. Means were

derived from z-scores for the entire data of each subjcct.

85

"p 7 o J'=41A

RESTROOM TRIPS

CC- i-IM Li

Cc

CD

CD

0-

0.gru.As0ltc r 9%cniec iis

08

RESTROOM TRIPS

0- 1B-RND-6

0

(wZ6

Zo

o

0.00o 6.0 oo 2.0 oo 8.0oo 21 . oo 3b. oo 3'6.0 oo 2.0o" ~~ELAPSED TIME (HOURS) ..

0560 ii00 1700 2300 0500 1100 1700 2300 iAPPROXIMATE TIME OF DAY

FIG, 15c Hourly means of numbers of restroom trips taken for the 18.-and-6 group.

Also plotted are 952 confidence limits. .

I

HC87.............................................

RESTROOM TRIPS

01

ýdu

* - 6-RND-1

0.0 o'.o 12-00 18oo.oo 24.o 00 b.oo 41oo •2. oc

,a,


LU II •

,0500 1100 1700 2300 0500 1100 1700 2300APPROXIMATE TIME OF DAY

FIG. 15D Itourly means of numbers of restroom. trips taken for the 6-and-I group.

t .

m pt Vr 9

88I

Documents

~MAY '9-.,. - DTIC · breaks. Both brief breaks and a longer rest period were very helpful in sustaining performance. METHODS Subjects Subjects in the study were 32 male volunLeers,