The Effects of Sleep Deprivation on Performance During Continuous Combat Operations
Gregory Belenky1, David M.Penetar, David Thorne, Kathryn Popp, John Leu, Maria Thomas, Helen Sing, Thomas Balkin, Nancy Wesensten, and Daniel Redmond
Good cognitive performance is central to successful combat operations. Command, control, communication, and intelligence are central to successful operations at all levels, from the crew, squad, and platoon levels through the division and corps levels. Battles are won or lost at the small-unit level (company, platoon, squad, and crew) (English, 1984). A single, small group
delivering modest amounts of fire at the right place and at the right time often determines the outcome of a major engagement (Marshall, 1978).
Sleep deprivation impairs alertness, cognitive performance, and mood. The ability to do useful mental work declines by 25 percent for every successive 24 h that an individual is awake. Thorne et al. (1983) have studied cognitive performance using a variety of computer-based cognitive performance tests during 72 h of total sleep deprivation in normal volunteer subjects. Those data and their analysis are summarized in Figure 7–1. The performance data in Figure 7–1 are expressed as throughput—the product of speed and accuracy. During sleep deprivation, performance declines, but it usually declines in such a way as to preserve the accuracy of response at the expense of speed. The throughput measure captures the combination of speed and accuracy and measures the amount of useful (i.e., accurate) work done per unit of time. Sleep deprivation degrades the most complex mental functions, including the ability to understand, adapt, and plan under rapidly changing circumstances. In contrast, simple psychomotor performance and physical strength and endu-
endurance are unaffected. For example, a soldier can shoot as tight a cluster of rounds at a fixed target after 90 h without sleep as he or she can when well rested, but if he or she has to shoot at targets that pop up at random at random locations on a firing range, then his or her performance drops to below 10 percent of baseline (Haslam and Abraham, 1987).
Pilot data from Thomas et al. (1988) suggest that sleep deprivation-induced decrements in performance are accompanied by decreases in brain glucose metabolism, particularly in the frontal areas. This provides a neurobiological correlate for the performance decrements. Whether the brain is less able to use glucose and hence is less able to do work, or is doing less work and hence uses less glucose, remains to be determined by future research.
Brief fragmented sleep has little recuperative value and is similar to total sleep deprivation in its effects on performance. Bonnet (1987) fragmented the sleep of normal volunteers by sounding an increasingly loud tone every 2–3 min until the subjects met the arousal criterion. For one group of subjects, the arousal criterion was a full awakening, as indicated by a movement and a verbal response. For a second group, the arousal criterion was a simple postural adjustment, with no verbal response required. For a third group, the arousal criterion was simply a change in the electroencephalogram (EEG), with no movement or verbal response required. All three arousal criteria destroyed the recuperative value of sleep as measured by the subjects’ alertness and performance on the next day. Bonnet’s results were not the result of sleep restriction, as subjects in all groups had near normal total sleep times, but rather they were the result of sleep fragmentation. Bonnet’s results are of great relevance to performance during continuous combat operations. If fragmentation of sleep, even fragmentation of sleep with no obvious behavioral manifestation (i.e., the change in the EEG only group), destroys the recuperative value of sleep, then not only duration but also continuity of sleep is important. In consultations to U.S. Army combat units, investigators stressed the need for sleep. Often, commanders have taken and applied the advice only to come back with something like: “I took a four-hour nap and awoke feeling no better than when I went to sleep.” When asked where they slept, a typical answer was “in the corner of my TOC.” (TOC is an acronym for Tactical Operations Center.) During continuous operations, the TOC is a busy, noisy place (people moving around and talking, radios giving off bursts of static) 24 h a day. Behaviorally (as judged by not moving and not talking), these commanders remained asleep for the period of the nap. It is hypothesized that they suffered frequent EEG-only arousals that fragmented their sleep and destroyed its recuperative value.
Continuous combat is characterized by brief fragmented sleep. In anecdotal accounts of actual combat operations and objective studies of simulated combat operations, brief, fragmented sleep is the rule rather than the exception. Pleban
et al. (1990) have studied sleep during the 58 days of U.S. Army Ranger School. These 58 days involve simulated light infantry operations against a superior force. In one study of one class, Ranger candidates averaged 3.2 h of sleep each night over the 58 days of the school (Pleban et al., 1990). In a second study of two classes, Ranger candidates averaged 3.6 h of sleep each night (Popp and Redmond, in preparation). This sleep was not accrued in a single sleep period but in a several naps over each 24-h period. Anecdotally, cognitive performance in Ranger candidates was marginal, with frequent episodes of what the Rangers call “droning,” in which candidates can put one foot in front of another and respond if challenged but have difficulty grasping their situation or acting on their own initiative.
Investigators in our research group have studied sleep during simulated armored and mechanized infantry operations at the National Training Center (NTC) in the high desert of Southern California. The operations involve battalion-sized task forces, consist of force on force and live fire exercises, and last for 14 days. As in the Ranger School study, sleep is brief and fragmented at NTC. Notable in the NTC study was the fact that there was a clear correlation between sleep and rank and between sleep and echelon of command and control. Whereas the personnel at the squad and crew levels averaged between 7 and 8 h of sleep each night, those at the battalion and brigade levels averaged little more than 4 h of sleep each night. Thus, from the perspective of sleep and its effects on performance, one would expect personnel at lower echelons to be more effective than personnel at higher echelons. This is what was observed, with the more junior people improving their performance over the course of the exercise and the more senior people “droning,” to use the Ranger School term, toward the end of the exercise.
McNally and colleagues (1989) have used data from Thorne et al. (1983) on the effects of total sleep deprivation on individual cognitive performance as input to the Army Unit Resiliency Analysis (AURA) model. The AURA model is detailed, modeling the performance of individual soldiers in the unit and, from this, the performance of the unit as a whole. It is this modeling of individual performance that allowed McNally et al. (1989) to use the laboratory data of Thorne et al. (1983) as input data. Specifically, they modeled the effects of sleep deprivation on artillery company performance with 4, 5, 6, or 7 h of sleep each night. They measured artillery company performance in rounds per tube (artillery piece) per day accurately delivered to the target. This is a measure of productivity similar to the throughput measure Thorne et al. (1983) used in their laboratory studies of sleep-deprived normal volunteers. The results of McNally et al. (1983) are depicted in Figure 7–2. What is obvious from Figure 7–2 is that deliberately restricting unit sleep in the hopes of greater output is unproductive. For 2–3 days, the unit that slept less was able, by virtue of having more time in which to work, to put more
rounds accurately on target in any 24-h period, but after the third day, their efficiency degraded to the point that even with this extra time to work their output was less. Even though the unit that slept for 4 h each night had 3 h more in each 24-h period in which to work, their overall output for the 24-h period fell below that for the unit that slept for 7 h each night by the third day of operations. In addition, according to the model, the unit’s aggregate output continued to fall as the days passed. These modeling results provide a qualitative estimate of the effects of partial sleep deprivation on unit performance over days and weeks of continuous operations.
Data from individual subjects in prolonged sleep deprivation show a gradual, systematic decline in performance (Thorne et al., 1983) (Figure 7–1). Modeling of unit performance during continuous operations when subjects are given 4, 5, 6, or 7 h sleep each night shows the same gradual, systematic decline (McNally et al., 1989). In realistic operational simulations and in actual
operations, these systematic declines in performance may be of no great consequence for a few days if the task at hand is simple and familiar and if an accurate although slower response is sufficient to complete the task. However, the task may be complex, unfamiliar, and/or intrinsically time-limited, and then sudden, severe, and even catastrophic failure can occur. Examples are illustrative.
Banderet and colleagues (1981) conducted a detailed, realistic simulation of artillery fire direction center (FDC) team operations. FDCs are manned by five-person teams. Their task is to plot the location of a target as given by observers farther forward and to derive range, bearing, angle of gun elevation, and charge. Targets may be called in on the fly with requests for immediate fire (fire missions) or may be called in advance for firing upon at a later time (preplanned targets). In either case, the FDC should derive range, bearing, etc., upon receipt of the target location, and in the case of a fire mission, the FDC should send this information directly to the guns or, in the case of a preplanned target, hold the information until a call for fire on that target is received. In the process of plotting target location, the FDC updates its situation map and clears the target to be sure that the location plotted is not that of a hospital, school, church, etc. At the time that Banderet et al. did their study, FDC teams were able to process two fire missions concurrently. In that study, FDC teams from the 82nd Airborne Division were run through simulated operations for 36 h. Throughout the 36 h, their ability to accurately derive range, bearing, elevation, and charge was unimpaired. However, after about 24 h they stopped keeping up their situation map and stopped computing their preplanned targets immediately upon receipt. They lost their grasp of their place in the operation. They no longer knew where they were located relative to friendly and enemy units. They no longer knew what they were firing at. Early in the simulation, when the investigators called for fire on, for example, a hospital, they would check their situation map, appreciate the nature of the target, and question the request. Later, without a current situation map, they would fire without hesitation, regardless of the nature of the target. Early in the simulation, when the investigators called in two concurrent fire missions and called for fire on a preplanned target, they would, having already plotted and derived information for the preplanned target, fire on all three quickly and accurately. Later, when the investigators called in two concurrent fire missions and called for fire on a preplanned target, the team, having neglected to plot and derive information for the preplanned target, would try to plot and derive information for three targets concurrently, and the targets, if fired on at all, were fired on only after long delays.
One of the authors of this chapter (G.Belenky) conducted after-action debriefings with personnel involved in friendly-fire incidents in the 100-h ground war during Operation Desert Storm. At dusk, after 48-plus h of
continuous operations (i.e., operations with brief, fragmented sleep), a platoon of six Bradley fighting vehicles was ordered to cease forward advance and to set up a screen line. No further movement was planned until the next morning. The platoon was flanked to the right and left by other Bradley fighting vehicle platoons. Supporting each platoon of Bradley fighting vehicles was a platoon of four M-1 tanks. These tanks took up positions some distance to the rear of the Bradleys. This was the position at last light. At about 0100 h, hot spots were observed in the thermal imaging sights moving toward the screen line. For reasons that remain obscure, even though the debriefing occurred soon after the event, these hot spots were simply observed until they ran into the screen line, at which point they resolved themselves into six Iraqi armored personnel carriers. This was not an attack; the Iraqis were still in column and were presumably just as surprised as the Americans. A firefight ensued in which all the Iraqi vehicles were destroyed. In the process, two Bradley fighting vehicles were destroyed. Fortunately, there were no American casualties. Conversely, there were no Iraqi survivors. Later, it was established that the Bradley fighting vehicles were destroyed by friendly fire. From the debriefings, here is apparently what happened. The two Bradleys that were destroyed were on the left of the platoon’s screen line. The one or two Bradleys that destroyed them were on the right of the platoon’s screen line. The advancing Iraqi column coming from the front and right encountered the screen line at an angle in proximity to the two Bradleys on the far left. The advancing Iraqi column was destroyed in a formation that was still in a column and trailing off at an angle to the right. The two Bradleys on the left were maneuvering in and around the first two Iraqi vehicles that were, by then, destroyed and burning. The Bradleys on the right, believing the Bradleys on the left to be Iraqis, engaged and destroyed them. Thanks to the crew protection built into the Bradleys, the five-men crews in both vehicles escaped unhurt. On debriefing, it was apparent that the two Bradleys on the right believed that they were firing forward and were unaware that they were enfilading their own line. Although no objective measures of sleep duration and continuity were made in this platoon, by self-report sleep for the prior 48-plus h had been brief and fragmented. This friendly fire incident is consistent with the known effects of sleep deprivation on performance. The ability to lay cross hairs on a target and accurately squeeze off rounds remained intact. What was lost was orientation and grasp of the tactical situation. The crews of the Bradley fighting vehicles that fired on their own comrades held to the sound tactical idea that “if it’s in front of us it dies.” However, they were no longer clear as to where front was.
CONCLUSIONS AND RECOMMENDATIONS
In conclusion, soldiers can fight for extended periods of time with only brief, fragmented sleep, but they become progressively less productive as the days pass and are increasingly prone to abrupt and serious failures in command and control. Commanders at all echelons should encourage sleep. Deliberately restricting sleep at any echelon in the hope of getting more out of soldiers and units is unproductive. In contrast, an adequate duration and continuity of sleep will sustain individual and unit performance indefinitely. To paraphrase General George Patton, the idea is not to give up sleep for your country but to make the other poor bastard give up sleep for his.
Banderet, L.E., J.W.Stokes, R.Francesconi, D.M.Kowal, and P.Naitoh 1981 Artillery teams in simulated sustained combat: Performance and other measures. Pp. 581–604 in DHHS (Department of Health and Human Services) NIOSH (National Institute for Occupational Safety and Health) Report 81–127: The Twenty-Four Hour Workday: Proceedings of a Symposium on Variations in Work-Sleep Schedules, L.C. Johnson, D.I.Tepas, W.P.Colquhon, and M.J.Colligan, eds. Washington, D.C.: U.S. Government Printing Office .
Bonnet, M.H. 1987 Sleep restoration as a function of periodic awakening, movement, or electroencephalographic change. Sleep 10:364–373.
English, J.A. 1984 On Infantry. Pp. 145–147 New York: Praeger.
Haslam, D.R., and P.Abraham 1987 Sleep loss and military performance. Pp. 167–184 in Contemporary Studies in Combat Psychiatry, G.Belenky, ed. Westport, Conn.: Greenwood Press.
Marshall, S.L.A. 1978 Men Against Fire. Pp. 138–156. Gloucester, Mass.: Peter Smith. McNally, R.E., A.M. Machovec, D.T.Ellzy, and S.R.Hursh
1989 Evaluation of Sleep Discipline in Sustaining Unit Performance. SAIC Contract No. MDA903–88-D-1000: McClean, Va: Science Applications International Corporation.
Pleban, R.J., P.J.Valentine, D.M.Penetar, D.P.Redmond, and G.Belenky 1990 Characterization of sleep and body composition changes during ranger training. Milit. Psychol. 2:145–156.
Thomas, M., H.Sing, G.Belenky, D.Thorne, T.Balkin, D.Penetar, D.Redmond, J.Norcross, P. Newhouse, H.Mayberg, H.Holcomb, B.Sadzot, R.Dannals, J.Links, and H. Wagner 1988 Positron emission tomography measurement of human regional cerebral glucose utilization following 48 hours of sleep deprivation. Soc. Neurosci. Abstr. 14:48.