Clinical Neurophysiology 111 (2000) 1771±1778

www.elsevier.com/locate/clinph

Oculomotor impairment after 1 night of total sleep deprivation: a dissociation between measures of speed and accuracy Luigi De Gennaro a,*, Michele Ferrara a, Luca Urbani b, Mario Bertini a a

Department of Psychology, University of Rome ªLa Sapienzaº, Via dei Marsi, 78; 00185 Roma, Italy b Department of Otolaryngology, Rieti General Hospital ªS. Camillo de Lellisº, Italy Accepted 28 June 2000

Abstract Objectives: The present study examined the effects of 40 h of sleep deprivation and of time-of-day on saccadic and smooth pursuit oculomotor performance. Methods: Nine normal subjects slept for 3 consecutive nights in the laboratory (one adaptation, one baseline, one recovery). Baseline and recovery were separated by a period of 40 h of continuous wakefulness, during which subjects were tested every 2 h. Oculomotor performance assessed at the following hours: 10:00, 12:00, 14:00, 16:00, 18:00, 20:00, 22:00, of both the days preceding and following the sleep deprivation night, as well as at 24:00, 02:00, 04:00, 06:00 and 08:00 h during the deprivation period. Results: Saccade latency increased and peak velocity decreased signi®cantly during the post-deprivation day; saccadic accuracy was unaffected. As regards smooth pursuit performance, phase (a measure of accuracy) was not affected by sleep loss, while velocity gain signi®cantly decreased during the day that followed the sleep deprivation night. Signi®cant time-of-day effects on the considered oculomotor variables except saccadic accuracy were also found, indicating an overall performance impairment during the night. Conclusions: It is concluded that 40 h of sleep deprivation signi®cantly impaired diurnal performance in pursuit and saccadic tasks. This performance worsening is limited to the measures of speed, while accuracy is not affected by sleep loss. A signi®cant operational relevance of these results is suggested, since saccadic velocity has recently been found to be negatively correlated with simulator vehicle crash rates. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Sleep deprivation; Time-of-day; Smooth pursuit; Saccadic eye movements; Oculomotor performance

1. Introduction Changes in ocular and visual functions during sleep deprivation have been reported since the late 1930s (Clarke and Warren, 1939; Kleitman and Schneider, 1940), including exophoria (the tendency of the eyes to diverge) and diplopia (double vision). Currently, the evaluation of the real effectiveness of voluntary oculomotor performance after sleep loss does have signi®cant implications in many practical and operational contexts, where the consequences of the human error may be catastrophic (e.g. car/truck drivers, air traf®c controllers, pilots). In the human factors research, some oculomotor measures (e.g. eye blinking rate, mean saccade velocity, saccade rate, peak saccade velocity) have extensively been evaluated during simulation of different ¯ight tasks as possible indices of workload and fatigue (e.g. Itoh et al., 1990; * Corresponding author. Tel.: 139-06-4991-7647; fax: 139-06-4451667. E-mail address: [email protected] (L. De Gennaro).

Morris and Miller, 1996; Veltman and Gaillard, 1996). Moreover, in the last decade the effectiveness of visual performance after sleep deprivation has been assessed in different simulation settings, such as a military mission with high visual workload (Quant, 1992), a laboratory simulation of sudden sleep-wake inversion (PorcuÁ et al., 1998) and a high-®delity driving simulation (Russo et al., 1999). Quant (1992) reported that the visual system appears quite resilient to 65 h of sleep deprivation, its effects being limited to a slight decrease in contrast sensitivity and in fusional convergence. Successively, PorcuÁ et al. (1998) found that velocity gain in a smooth pursuit test and accuracy of saccadic eye movements are negatively affected by very high levels of sleepiness, being signi®cantly impaired at 06:00 h, after 1 whole night without sleep. Moreover, Russo and co-workers (1999) showed a signi®cant decrease in saccadic velocity in a cumulative partial sleep deprivation paradigm. Since saccadic velocity was negatively correlated with driving simulator crash rates, a potential usefulness of this

1388-2457/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1388-245 7(00)00393-X

CLINPH 2000565

1772

L. De Gennaro et al. / Clinical Neurophysiology 111 (2000) 1771±1778

oculomotor variable for monitoring alertness in sleep restricted drivers has been suggested (Russo et al., 1999). To further assess the sensitivity of some voluntary oculomotor measures to increasing levels of sleepiness, as part of a larger research program (De Gennaro and Ferrara, 2000), in this study we tested saccadic and smooth pursuit eye movements during the day that followed an undisturbed nocturnal sleep, as well as after 1 night of continuous wakefulness. An impairment of oculomotor performance after sleep deprivation should be expected, as compared to the pre-deprivation wakefulness levels. Since oculomotor performance was assessed 12 times at 2-h intervals (from 10:00 to 08:00 h the following morning), time-of-day effects on the considered oculomotor variables were also evaluated. 2. Methods and materials 2.1. Subjects Nine normal male subjects (mean age ˆ 23.2 years) were selected as paid volunteers for the study. Subjects were selected if they usually went to sleep between 11:00 and 00:00 h and if they usually slept 7±8 h per night. Other requisites for inclusion were: no daytime nap habits, no excessive daytime sleepiness, no other sleep, medical or psychiatric disorder, as assessed by a 1-week sleep log and by a clinical interview. Participants were unaware of the purpose of the experiment and signed an informed consent; their rights were protected throughout the entire course of the experiment. 2.2. Procedure The protocol of the study was reviewed and approved by the local Institutional Review Board. Participants slept for 3 consecutive nights in a sound-proof, temperature controlled room: (1) adaptation (AD); (2) baseline (BSL); (3) recovery (REC). To avoid a progressive improvement of performance during the experiment due to a practice effect, all the subjects familiarized with the oculomotor tests in the evening that preceded the adaptation night by performing each test at least 3 times and, in any case, when subjects were able to adequately carry out the tests. ANOVAs comparing, for each variable, performance on the last practice session vs. the mean diurnal values did not show any signi®cant difference. Every night, sleep recording started at about 23:30 h and ended after 7.5 h of accumulated sleep. In the adaptation and baseline nights, subjects arrived at the sleep laboratory and, after electrode and ear mould montage, their undisturbed sleep was recorded. A pressure-sensitive transducer was encased in the ear mould, to record middle-ear muscle activity during sleep (for further details, see De Gennaro and Ferrara, 2000). A 40 h schedule of sleep deprivation began on morning

awakening following the baseline night. Subjects remained in the sleep laboratory together with at least one experimenter throughout the whole course of the 40-h-long continuous wakefulness. During the sleep deprivation period, oculomotor performance (saccadic eye movements, SAC, and smooth pursuit eye movements, SP) was recorded, as a possible indicator of sleepiness, every 2 h in a soundproof room with subjects sitting in bed and leaning their back against the headboard. Each subject was requested to position his head to a predetermined height, to insure that the light bar (see below) was exactly located in front of his eyes. The oculomotor performance was assessed in the dark. In the present paper we report on oculomotor performance assessed during the 40 h of continuous wakefulness (from 10:00 h of the pre-deprivation day to 22:00 h of the day following the sleep deprivation night). When not involved in testing sessions, subjects were allowed to carry out their own preferred activities, such as reading, writing, listening to music, watching TV or playing games, always under the direct supervision of at least one experimenter. Lying down, sleeping and vigorous physical activity were not permitted. Meals were provided to subjects at 08:30, 14:30, and 19:30 h. Non-scheduled light snacks were permitted, while caffeinated beverages, chocolate, alcohol and medications that can induce sleepiness were not allowed during the deprivation protocol. Time information was available to subjects, and light exposure was not strictly controlled for (although the laboratory was constantly illuminated by 4 neon lamps, blinds only in part attenuated the light coming from the outside). The 40 h schedule of sleep deprivation ended at 22:00 h; the sleep recording of the recovery night began at about 23:30 h. 2.3. Saccadic eye movement recording The test stimulus was a visual target (a red dot of light) moving on a horizontal bar located at 1.2 m from the subject's eyes. The subject was asked to ®x his gaze on the target at the center of the bar without moving his head, and then to visually follow the target which moved horizontally, once every 1.25 s, through a series of stepwise jumps of pseudo-random amplitude (5±30 degrees), within a range of ^208 of the visual ®eld. For each session, about 100 saccadic eye movements were elicited in a time interval of 2 min. Eye movements were recorded with an electronystagmographic (ENG) technique by two electrodes placed about 1 cm from the medial and lateral canthi of the dominant eye; a bipolar recording (AC, time constant: 15.9 s) was carried out with an Automated ENG MASTR Package (ICS Medical Corporation, Schaumburg, IL, USA), installed on a Bull Micral 200 computer. This system controlled the stimulus administration and calibration, and also performed automatic saccade analysis after detection and rejection of eye movements considered as artifacts. Latency, accuracy and peak velocity of saccadic eye movements (limited to the saccades between 11 and 148

L. De Gennaro et al. / Clinical Neurophysiology 111 (2000) 1771±1778

of amplitude) were considered as dependent variables for data analysis. Latency was computed as the time interval between stimulus movement and the onset of the next eye movement of more than 908/s. Accuracy is the amplitude of the ®rst saccade (not considering other possible correcting saccades) divided by the amplitude of target movement, and expressed as a percentage. Peak velocity is the maximum velocity reached in each saccadic movement as measured over an 18.75 ms period from the onset of the saccade. Finally, the algorithm of the analysis software rejected as artifacts eye movements that occurred too early (250 ms before through 75 ms after target movement), too late (more than 600 ms after target movement), or in the wrong direction. 2.4. Smooth pursuit eye movement recording The test stimulus was a red dot of light moving on a horizontal bar located in front of the subject at eye height and at a distance of 1.2 m. The subject was asked to look at the center of the bar without moving his head and to ®x his gaze on the dot of light which moved horizontally in a sinusoidal pattern for a total amplitude of 33.48 of the visual ®eld. Six different stimulus frequencies were consecutively used in each sequence: 0.2, 0.3, 0.4, 0.5, 0.6, 0.7 Hz, which corresponded respectively to a peak velocity of 21, 31.5, 42.5, 52.5, 63 and 73.58/s. The whole sequence was repeated twice and a total of 6 cycles were recorded for each stimulus frequency. Eye movements were recorded with the same ENG technique and system used for the SAC. This system controlled the stimulus administration and calibration, the recording and automatic analysis of SP eye movements. Phase (accuracy of eye movement measured in angular degrees of leading or lagging with respect to target movement) and velocity gain for right and left eye movements (peak velocity of eye movement/peak velocity of target movement) were considered as dependent variables after elimination of saccadic eye movements (de®ned as movements faster than the stimulus by more than 158/s). As regards the former variable, the phase of the fundamental frequency of the eye movement is computed by the computerized system from a discrete Fourier transform, and compared to the phase of the stimulus. If phase shift is greater than 38 leading or more than 208 lagging, the cycle is rejected as an artifact. For computing gain, the velocity of the stimulus over its fastest 250 ms is compared to the eyemovement over the same period. 2.5. Data analysis 2.5.1. Effects of sleep deprivation As regards SAC, latency and accuracy of saccadic eye movements were submitted to a two-way repeated measure ANOVA Condition (Pre-deprivation, Post-deprivation) by Time-of-day (10:00, 12:00, 14:00, 16:00, 18:00, 20:00, 22:00 h), while velocity was submitted to a 3-way repeated measure ANOVA Amplitude (11, 12, 13, 148) by Condition

1773

(Pre-deprivation, Post-deprivation) by Time-of-day (10:00, 12:00, 14:00, 16:00, 18:00, 20:00, 22:00 h). As regards SP, phase and velocity gain for right and left eye movements were submitted to 3-way repeated measure ANOVAs Frequency (0.2, 0.3, 0.4, 0.5, 0.6, 0.7 Hz) by Condition (Pre-deprivation, Post-deprivation) by Time-ofday (10:00, 12:00, 14:00, 16:00, 18:00, 20:00, 22:00 h). 2.5.2. Time-of-day effects Each of the above-mentioned variables was submitted to a one-way ANOVA with Time (10:00, 12:00, 14:00, 16:00, 18:00, 20:00, 22:00, 24:00, 02:00, 04:00, 06:00, 08:00 h). To simplify data handling, saccadic velocity was collapsed across the 4 amplitudes, and SP gain and phase across the 6 frequencies. 3. Results 3.1. Polysomnography These results have been published in detail elsewhere (De Gennaro and Ferrara, 2000). Brie¯y, as a consequence of sleep deprivation, recovery nights were characterized by a decrease of stage 1, stage 2, and intra-sleep wake. SWS percentage was doubled (from about 11 to more than 22%), the sleep ef®ciency index increased, and percentage of REM sleep was unaffected. The latencies of all NREM sleep stages were shortened. 3.2. Effects of sleep deprivation on oculomotor performance The behavior of the speed and accuracy measures for both SAC and SP across the pre- and post-deprivation days is reported in Table 1. 3.2.1. Saccadic eye movements 3.2.1.1. Latency of saccadic eye movements. Two-way ANOVA showed a signi®cant main effect for Condition (F1;8 ˆ 19:03; P ˆ 0:002), indicating that mean latency of saccades increased signi®cantly during the day that followed the night of sleep deprivation (Fig. 1a). The main effect for Time-of-day (F6;48 ˆ 2:12; P ˆ 0:07) and the Condition by Time-of-day interaction (F6;48 ˆ 1:81; P ˆ 0:12) were not signi®cant. 3.2.1.2. Velocity of saccadic eye movements. Three-way ANOVA showed a signi®cant main effect for Amplitude (F3;24 ˆ 48:38; P ˆ 0:0000), indicating the physiological increase of eye movement velocity as a function of the amplitude of saccades (Fig. 2). Trend analysis con®rmed the linearity of such an increase (F1;8 ˆ 88:99; P ˆ 0:00001). The main effect for Condition was also signi®cant (F1;8 ˆ 13:95; P ˆ 0:006), indicating a decrease of mean

1774

L. De Gennaro et al. / Clinical Neurophysiology 111 (2000) 1771±1778

Table 1 Mean raw data of each considered variable (and SEMs) for the 7 pre-deprivation and the corresponding post-deprivation testing sessions a Time of day

10:00 Pre 12:00 Pre 14:00 Pre 16:00 Pre 18:00 Pre 20:00 Pre 22:00 Pre 10:00 Post 12:00 Post 14:00 Post 16:00 Post 18:00 Post 20:00 Post 22:00 Post

Saccadic performance

Smooth pursuit performance

Latency

Velocity

Accuracy

RVG

LVG

Phase

197.2 (5.3) 191.8 (5.4) 195.7 (4.6) 197.6 (4.4) 196.8 (5.1) 200.2 (4.8) 198.7 (6.3) 208.2 (7.1) 202.5 (8.2) 220.1 (6.5) 213.3 (6.4) 217.2 (7.2) 205.6 (5.2) 208.7 (6.5)

336.9 (8.6) 334.5 (5.9) 343.3 (6.9) 339.8 (6.1) 332.9 (12.3) 335.1 (8.8) 327.4 (4.8) 314.2 (7.6) 324.2 (21.6) 331.6 (29.1) 316.7 (9.7) 323.0 (10.1) 320.9 (9.3) 325.8 (10.5)

94.2 (1.8) 101.3 (4.7) 102.2 (7.1) 100.2 (5.5) 96.7 (3.4) 97.3 (3.7) 95.5 (1.5) 94.4 (1.3) 98.0 (6.2) 98.0 (5.7) 97.0 (2.3) 101.7 (6.9) 96.5 (2.7) 95.9 (1.2)

0.92 (0.02) 0.90 (0.02) 0.89 (0.02) 0.91 (0.01) 0.90 (0.01) 0.90 (0.02) 0.90 (0.02) 0.84 (0.02) 0.83 (0.03) 0.86 (0.03) 0.84 (0.03) 0.85 (0.02) 0.87 (0.03) 0.83 (0.04)

0.91 (0.02) 0.90 (0.03) 0.89 (0.01) 0.88 (0.01) 0.89 (0.02) 0.89 (0.02) 0.90 (0.01) 0.82 (0.02) 0.83 (0.02) 0.84 (0.03) 0.84 (0.03) 0.84 (0.02) 0.88 (0.02) 0.83 (0.04)

2.11 (0.17) 2.03 (0.17) 2.34 (0.25) 2.30 (0.24) 2.21 (0.40) 2.40 (0.31) 2.09 (0.20) 2.97 (0.44) 2.63 (0.55) 2.69 (0.35) 2.27 (0.32) 2.45 (0.17) 2.47 (0.27) 2.65 (0.24)

a Saccadic latency is reported in ms; velocity in angular 8/s; accuracy in percentage. Saccadic velocity values have been collapsed across amplitudes. Accuracy values closer to 100 indicate a better performance; values over 100 indicate that saccadic amplitude was bigger than the target amplitude (overshooting), while values under 100 show the opposite (undershooting). Smooth pursuit right velocity gain (RVG) and left velocity gain (LVG) have been obtained as: peak velocity of eye movement/peak velocity of target movement. Consequently, values closer to 1 indicate a better performance. Phase is expressed in angular degrees of delay with respect to the target. Each smooth pursuit variable has been collapsed across the 6 target frequencies.

saccadic velocity during the day following the sleep deprivation night (Fig. 1a). No other main effect or interaction was signi®cant. 3.2.1.3. Accuracy of saccadic eye movements. Two-way ANOVA did not show any signi®cant main effect or interaction (F1;8 ˆ 0:20; F6;48 ˆ 0:55; F6;48 ˆ 0:34, respectively). 3.2.2. Smooth pursuit eye movements 3.2.2.1. Velocity gain of SP rightward eye movements (RVG). Three-way ANOVA showed a signi®cant main effect for Frequency (F5;40 ˆ 19:12; P ˆ 0:0000), indicat-

Fig. 1. Upper part (a): mean latency and mean peak velocity of saccadic eye movements (and SEMs) during the 7 pre- and post-deprivation sessions. Lower part (b): mean velocity gain (and SEMs) of rightward and leftward smooth pursuit eye movements during the 7 pre- and post-deprivation sessions. Values closer to 1 indicate a better performance.

Fig. 2. Mean peak velocity (and SEMs) of saccadic eye movements for each considered amplitude during the 7 pre- and post-deprivation sessions.

L. De Gennaro et al. / Clinical Neurophysiology 111 (2000) 1771±1778

1775

ing a worse performance at the highest frequencies (Fig. 3). Trend analysis showed that the decrease of mean velocity gain as a function of increasing target frequency has a signi®cant linear component (F1;8 ˆ 21:68; P ˆ 0:002), as well as a quadratic one (F1;8 ˆ 21:79; P ˆ 0:002). The main effect for Condition was also signi®cant (F1;8 ˆ 12:73; P ˆ 0:007), indicating a decrease of RVG during the day that followed 1 night of sleep deprivation (Fig. 1b). No other main effect or interaction was signi®cant. 3.2.2.2. Velocity gain of SP leftward eye movements (LVG). Three-way ANOVA showed a signi®cant main effect for Frequency (F5;40 ˆ 21:18; P ˆ 0:0000), indicating a worse performance at the highest frequencies (Fig. 3). Trend analysis showed that the decrease of mean velocity gain as a function of increasing target frequency has a signi®cant linear component (F1;8 ˆ 25:19; P ˆ 0:001), as well as a quadratic one (F1;8 ˆ 22:95; P ˆ 0:001). The main effect for Condition was also signi®cant (F1;8 ˆ 11:73; P ˆ 0:009), indicating a decrease of LVG during the day that followed 1 night of sleep deprivation (Fig. 1b). No other main effect or interaction was signi®cant. 3.2.2.3. Phase of SP eye movements. Three-way ANOVA showed a signi®cant main effect for Frequency (F5;40 ˆ 23:39; P ˆ 0:0000), indicating the negative effect of increasing target frequencies on SP responses (Fig. 4). Trend analysis showed that the increase of mean phase delay as a function of increasing target frequency has a signi®cant linear component (F1;8 ˆ 29:76; P ˆ 0:0006), as well as a quadratic one (F1;8 ˆ 28:14; P ˆ 0:0007). No other main effect or interaction was signi®cant.

Fig. 3. Mean velocity gain (and SEMs) of rightward (RVG) and leftward (LVG) smooth pursuit eye movements at each target frequency during the 7 pre- and post-deprivation sessions. Values closer to 1 indicate a better performance.

Fig. 4. Mean phase delay in angular degrees (and SEMs) of smooth pursuit eye movements at each target frequency during the 7 pre- and post-deprivation sessions.

3.3. Time-of-day effects on oculomotor performance Twenty-four hours oscillation of each considered variable are plotted in Fig. 5. 3.3.1. Saccadic eye movements The ANOVA was signi®cant for both latency (F8;11 ˆ 3:68; P ˆ 0:0003; Fig. 5A) and velocity (F8;11 ˆ 5:90; P ˆ 0:0001; Fig. 5B), but not for accuracy (F8;11 ˆ 0:80; P ˆ 0:64; Fig. 5C). Saccadic latency seems to be fairly stable during the ®rst 7 testing sessions, i.e. until 22:00 h, then it slightly increases during the night; the increase becomes clearer during the last two sessions. A possible circadian effect is evident for mean saccadic velocity: this variable suddenly drops at 24:00 h and then remains stable during the following nocturnal and early morning sessions. On the other hand, saccadic accuracy does not show any signi®cant variation over the 24 h, although from a visual inspection of its trend (Fig. 5C) there seems to emerge a rhythm with maximal accuracy during the midday hours (from 12:00 to 16:00 h). 3.3.2. Smooth pursuit eye movements The ANOVA was signi®cant for the gain of both rightward (F8;11 ˆ 3:02; P ˆ 0:002; Fig. 5D) and leftward eye movements (F8;11 ˆ 3:98; P ˆ 0:0001; Fig. 5E), as well as for the phase (F8;11 ˆ 2:36; P ˆ 0:01; Fig. 5F). As far as gain is concerned, from a visual inspection of the plots (Fig. 5D,E) it is clear that this variable is maintained at fairly stable levels until 22:00 h, and then shows a linear decrease during the nocturnal and early morning hours (from 24:00 to 06:00 h). On the other hand, phase is quite stable in all testing sessions except the last 3 (from 04:00 to 08:00 h), when accuracy levels show a drop.

1776

L. De Gennaro et al. / Clinical Neurophysiology 111 (2000) 1771±1778

Fig. 5. Means (and SEMs) of each considered variable recorded during 24 h at 2 h intervals (from 10:00 to 08:00).

4. Discussion In the present study we found that some measures of voluntary saccadic and smooth pursuit performance, recorded after 1 night of sleep deprivation, are negatively affected by increasing levels of sleepiness. It has to be emphasized that oculomotor performance worsening during prolonged wakefulness is limited to the measures of speed, while both saccadic and smooth pursuit accuracy are maintained at adequate levels even after 40 h of continuous wakefulness. These results are in line with those recently found in a study on the effects of sleep inertia on the same oculomotor variables assessed upon awakening (Ferrara et al., 2000). In that case, a signi®cant increase of saccade latency and a decrease of saccadic velocity and of smooth pursuit velocity gain was found upon awakening from a recovery sleep following 40 h of sleep deprivation, as compared to a diurnal baseline and to performance assessed upon awakening from the adaptation and baseline nights. More speci®cally, results concerning saccadic performance further con®rm the linear increase of saccade velocity as a function of eye movement amplitude in the 11±148 range, in agreement with the Schmidt et al.'s ®nding (1979) of a linear increase of peak velocity for saccades between 5 and 208 and, more generally, with the well-known positive relationship between saccadic amplitude and velocity (Baloh et al. 1975; Becker, 1991; Henriksson et al., 1980). In addition, we found a signi®cant decrease of saccadic peak velocity during the day following the sleep deprivation night. This variable has already been considered a sensitive indicator of the alertness state (e.g. Ron et al., 1972). Velocity of saccades decreases when the subject's level of alertness lowers (Becker, 1991); moreover, this parameter is

sensitive to the effects of drugs that impair the state of alertness, such as benzodiazepines (e.g. Gentles and Thomas, 1971), as well as to cumulative partial sleep deprivation (Russo et al. 1999). In the latter study, Russo and coworkers reported that saccadic velocity decreased signi®cantly during 7 consecutive nights of sleep restricted to 3 or 5 h. Since this oculomotor parameter was negatively correlated with simulator vehicle crash rates during the same partial sleep deprivation paradigm, the authors suggested the potential usefulness of saccadic velocity to evaluate alertness also in environmental contexts or in ®eld studies (i.e. in sleep restricted drivers). In light of this, our ®ndings on the decrease of saccade velocity and on the signi®cant increase of saccade latency as a consequence of 1 night of sleep loss, further con®rm the potential dangerous consequences of sleep loss on human performance. On the other hand, the accuracy of saccades is not affected by sleep loss, since no signi®cant difference between pre- and post-deprivation performance levels was found. A previous study on oculomotor performance after an abrupt shift of the sleep-wake pattern (PorcuÁ et al. 1998), reported a signi®cant deterioration of saccadic accuracy at the highest levels of fatigue and sleepiness, i.e. at 06:00 h after 1 night without sleep. It may be hypothesized that, in the above-mentioned study (PorcuÁ et al., 1998), the increasing sleepiness interacted with circadian factors in in¯uencing saccadic accuracy: in fact, a signi®cant performance worsening was found only at 06:00 h, i.e. near the circadian nadir of many psychophysiological variables (e.g. Naitoh 1981). However, this result was not con®rmed in the present study. As regards smooth pursuit velocity gain (i.e. the ratio

L. De Gennaro et al. / Clinical Neurophysiology 111 (2000) 1771±1778

between peak eye velocity and peak target velocity), that is close to 1 during ideal pursuit tracking, our results further con®rm that smooth pursuit performance is signi®cantly deteriorated at the higher target frequencies (Leigh and Zee, 1991). Furthermore, velocity gain signi®cantly decreased during the day that followed 1 night of sleep deprivation, pointing out again that some measures of speed of oculomotor performance are negatively affected by sleep loss. As far as the accuracy of smooth pursuit performance is concerned, phase - a measure of the temporal synchrony between the target and the eye ± showed an increase in mean delay at higher frequencies, con®rming the wellknown disrupting effect of increasing target frequencies on smooth pursuit responses (Leigh and Zee, 1991). This variable, however, at variance with velocity gain, did not turn out to be negatively affected by sleep loss, further con®rming that oculomotor performance accuracy is maintained at adequate levels even after 40 h of sleep deprivation. Finally, our data showed signi®cant time-of-day effect on the considered oculomotor variables except saccadic accuracy. However, these results have to be considered cautiously, since it is dif®cult to say whether the oculomotor performance ¯uctuations during the sleep deprivation night are due to actual circadian oscillations of those variables, or to the disrupting effects of sleep debt on performance levels, or to an interaction between sleep deprivation and time-ofday effects. Actually, to our best knowledge only Giedke et al. (1996) reported, in a total sleep deprivation paradigm, what they called `a non-random diurnal (`circadian') variation' of smooth pursuit velocity gain with a distinct minimum at 08:00 h. The same variable did not show any clear diurnal maximum. However, in that case the reported effect on pursuit gain could simply be attributed to the high levels of sleepiness, since the 08:00 h recording followed 1 whole night of sleep deprivation, and a pre-deprivation session at the same hour was not scheduled. In our study, the same variable showed a sharp drop at 06:00 h; even though we did not record body core temperature, it is possible to speculate that the high levels of sleepiness and fatigue due to sleep deprivation interacted with circadian factors (body temperature minimum) in causing that slowing in smooth pursuit performance. More generally, our data seem to support a circadian ¯uctuation of oculomotor performance, at least for the measures of saccadic and smooth pursuit speed. As a matter of fact, saccadic velocity showed a clear and sudden drop at 24:00 h, before the accumulation of any sleep debt, remaining than stable at levels lower than those recorded during the day. Moreover, pursuit gain showed a linear nocturnal decrease from 24:00 to 06:00 h, while phase presented a worsening limited to the early morning hours. In conclusion, 40 h of sleep deprivation signi®cantly impaired diurnal performance in smooth pursuit and saccadic tasks administered every other hour from 10:00 to 20:00 h. This performance worsening is limited to the measures of

1777

speed, while both saccade accuracy and smooth pursuit phase are unaffected by total sleep deprivation. These results may have a signi®cant operational relevance. Although in our study oculomotor performance accuracy is maintained at satisfactory levels even after 40 h of continuous wakefulness, at the same time it should be emphasized that saccadic velocity has recently been found to negatively correlate with simulator vehicle crash rates in a cumulative partial sleep deprivation paradigm (Russo et al., 1999). Also in the present study a signi®cant decrease of saccadic velocity during the day following the sleep deprivation night has been found. From a sleep logistics perspective, our results further con®rm that the adverse and potentially dangerous effects of sleep deprivation on oculomotor performance should be avoided, especially by people who perform complex or critical tasks requiring high oculomotor control (e.g. commercial drivers, truck drivers, pilots, air traf®c controllers). Acknowledgements Thanks to Irene Lorusso, Simona Baldanza and Francesca Anzidei for their invaluable help in data collecting. This research was partially supported by a MURST grant to L.D.G. (Finanziamento ricerche di FacoltaÁ 1998). We also wish to thank the anonymous referees, whose comments helped to greatly improve the manuscript. References Baloh R, Sills A, Kumley W, Honrubia V. Quantitative measurement of saccade amplitude, duration and velocity. Neurol 1975;25:1065±1070. Becker W. Saccades. In: Carpenter RH, editor. Vision and visual disfunction, Vol. 8: Eye movements, London: MacMillan Press, 1991. pp. 95± 137. Clarke B, Warren N. The effect of loss of sleep on visual tests. Am J Optom 1939;16:80±95. De Gennaro L, Ferrara M. Sleep deprivation and phasic activity of REM sleep: a dissociation between rapid eye movements and middle-ear muscle activity. Sleep 2000;23:81±85. Ferrara M, De Gennaro L, Bertini M. Voluntary oculomotor performance upon awakening after total sleep deprivation. Sleep 2000;23(6). Gentles W, Thomas E. Effect of benzodiazepines upon saccadic eye movements in man. Clin Pharmacol Ther 1971;12:563±574. Giedke H, Bork S, Buettner U. Diurnal variation of smooth pursuit eye movements during sleep deprivation. J Sleep Res 1996;5((Suppl 1)):73. Henriksson N, Pyykko I, Scalen L, Wenmo C. Velocity patterns of rapid eye movements. Acta Otolaringol 1980;89:504±512. Itoh Y, Hayashi Y, Tsukui I, Saito S. The ergonomic evaluation of eye movement and mental workload in aircraft pilots. Ergonomics 1990;33:719±733. Kleitman N, Schneider J. Sleepiness and diplopia. Am J Physiol 1940;129:398±404. Leigh RJ, Zee DS. Smooth Pursuit and visual ®xation. In: Leigh RJ, Zee DS, editors. The neurology of eye movements, 2nd edn. Philadelphia, PA: FA, Davis Company, 1991. pp. 139±179. Morris TL, Miller JC. Electrooculographic and performance indices of fatigue during simulated ¯ight. Biol Psychol 1996;42:343±360. Naitoh P. Circadian cycles and restorative power of naps. In: Johnson LC, Tepas DL, Colquhoun WP, Colligan MJ, editors. Biological rhythms,

1778

L. De Gennaro et al. / Clinical Neurophysiology 111 (2000) 1771±1778

sleep and shiftwork, advances in sleep research, Vol. 7. New York: Spectrum, 1981. pp. 553±580. PorcuÁ S, Ferrara M, Urbani L, Bellatreccia A, Casagrande M. Smooth pursuit and saccadic eye movements as possible indicators of nighttime sleepiness. Physiol Behav 1998;65:437±443. Quant LR. The effect of sleep deprivation and sustained military operations on near visual performance. Aviat Space Environ Med 1992;63:172± 176. Ron S, Robinson D, Skavenski A. Saccades and the quick phase of nystagmus. Vision Res 1972;12:2015±2022.

Russo M, Thomas M, Sing H, Thorne D, Balkin T, Wesensten N, Redmond D, Welsh A, Rowland L, Johnson D, Cephus R, Hall S, Krichmar J, Belenky G. Saccadic velocity and pupil constriction latency changes in partial sleep deprivation, and correlations with simulated motor vehicle crashes. Sleep 1999;22((Suppl 1)):S297±S298. Schmidt D, Abel L, Dell'Osso L, Daroff R. Saccade velocity characteristics: intrinsic variability and fatigue. Aviat Space Environ Med 1979;50:393±395. Veltman J, Gaillard A. Physiological indices of workload in a simulated ¯ight task. Biol Psychol 1996;42:323±342.