The trait of Introversion Extraversion predicts vulnerability to sleep deprivation

Transcription

1 J. Sleep Res. (2007) 16, The trait of Introversion Extraversion predicts vulnerability to sleep deprivation WILLIAM D. S. KILLGORE, JESSICA M. RICHARDS, DESIREE B. KILLGORE, GARY H. KAMIMORI and THOMAS J. BALKIN Department of Behavioral Biology, Walter Reed Army Institute of Research, Silver Spring, MD, USA Accepted in revised form 30 July 2007; received 1 February 2007 SUMMARY According to EysenckÕs theory of Introversion Extroversion (I E), introverts demonstrate higher levels of basal activity within the reticular-thalamic-cortical loop, yielding higher tonic cortical arousal than Extraverts, who are described conversely as chronically under-aroused and easily bored. We hypothesized that higher scores on the trait of Extraversion would be associated with greater declines in psychomotor vigilance performance during prolonged wakefulness. We evaluated the relationship between I E and overnight psychomotor vigilance performance during 77 h of continuous sleep deprivation in a sample of 23 healthy adult military personnel (19 men; four women), ranging in age from 20 to 35 years. At baseline, volunteers completed the Revised NEO Personality Inventory (NEO PI-R) and completed psychomotor vigilance testing at approximately 10-min intervals from 00:15 to 08:50 hours over three nights of continuous sleep deprivation. In addition, 12 participants received four repeated administrations of caffeine (200 mg) every 2 h each night. Analysis of covariance and stepwise multiple regression analyses showed that, above and beyond the effects of caffeine, higher Extraversion was significantly related to more extensive declines in speed of responding and more frequent attentional lapses, but only for the first overnight testing session. Sub-factors of Extraversion, including Gregariousness and higher Activity level were most predictive of these changes following sleep loss. These findings are consistent with EysenckÕs cortico-reticular activation theory of I E and suggest that individual differences in the trait of Extraversion confer some vulnerability resistance to the adverse effects of sleep loss on attention and vigilance. keywords Extraversion, individual differences, Introversion, personality, psychomotor vigilance, sleep deprivation To varying degrees, we are all subject to the negative influences that fatigue and inadequate sleep impart on cognitive and physiological functioning. For most individuals, prolonged wakefulness extending beyond the normal 16-h period of daytime activity is usually accompanied by a general decline in alertness, vigilance and mood, and a progressive increase in homeostatic sleep pressure, making it increasingly difficult to resist the urge to obtain sleep (Van Dongen and Dinges, 2005). Correspondence: MAJ William D. Killgore, PhD, Cognitive Neuroimaging Laboratory, Brain Imaging Center, McLean Hospital, 115 Mill Street, Belmont, MA 02478, USA. Tel.: ; fax: ; Killgore@mclean.harvard.edu. Most individuals appear to function optimally with about 7 8 h of sleep per night (Belenky et al., 2003), but anecdotal accounts frequently describe individuals who seem to function effectively on very little sleep. Indeed, a growing body of research appears to support such accounts, with accumulating evidence now suggesting that there are trait-like differences among individuals in their vulnerability to the adverse cognitive effects of sleep loss, with some individuals showing great sensitivity to sleep loss and others appearing only minimally affected (Caldwell et al., 2005; Van Dongen et al., 2004, 2005). Despite the emerging scientific consensus that some individuals are more resistant than others to the adverse effects of 354 Ó 2007 European Sleep Research Society

2 Introversion Extraversion and sleep deprivation 355 sleep deprivation, there has been little progress in defining the underlying psychological, motivational and neurobiological bases for these differences (Van Dongen et al., 2005). One possible explanation was suggested by a recent functional magnetic resonance imaging (fmri) study that examined the cortical activation patterns of fatigue-resistant fighter pilots versus individuals identified to be more vulnerable to the adverse effects of fatigue (Caldwell et al., 2005). In that study, the spatial extent of cortical activation produced by a demanding cognitive task when pilots were well rested correlated with their ability to sustain cognitive performance when sleep was deprived for 37 h. Overall, those individuals with the lowest cortical activity were also the most vulnerable to the adverse effects of fatigue on cognitive performance, whereas those with the highest levels of cortical activity showed very little alteration of performance as a function of sleep loss. Other functional neuroimaging studies have shown that 24 or more hours of sleep deprivation produces significant reductions in regional cerebral metabolism within prefrontal cortex, parietal cortex and thalamus (Thomas et al., 2000, 2003), and the functioning of these regions during cognitive challenge tasks appears to be particularly affected by sleep loss (Drummond et al., 2000, 2004). When considered together, the evidence suggests that the cognitive capacities of individuals with higher resting baseline levels of cortical activation may be more resistant to the effects of sleep loss than those with chronically lower levels of tonic cortical arousal. If there is, in fact, a relationship between cortical arousal and vulnerability to the effects of sleep deprivation, several corollary predictions should also be true. Of particular relevance to this paper, cortical arousal and the ability to resist sleep deprivation, may be related to specific aspects of personality functioning. Cortical arousal has been postulated as the primary neurobiological substrate underlying the personality dimension of Introversion Extraversion (I E) (Eysenck, 1967), an aspect of personality that involves the degree of sociability, gregariousness, preferred activity level, emotional warmth and the desire to seek excitement and stimulation preferred by an individual (Zuckerman, 2005). According to Eysenck (Eysenck, 1981), the differences between Introverts and Extraverts in their social gregariousness and sensation seeking emerges from psychophysiological differences in the tonic arousal of the reticulo-thalamiccortical activation system. EysenckÕs model proposes that this activation system is maintained at a significantly higher level of tonic arousal in Introverts relative to Extraverts. Incorporating concepts from the earlier work of Hebb (1955) and that of Yerkes and Dodson (1908), which described the relationship between arousal and performance as an inverted U-shaped curve, with optimal performance occurring at moderate levels of arousal, Eysenck proposed that Introverts and Extraverts are characterized by different levels of basal arousal function along this curve (Eysenck, 1964). Fig. 1 shows a modified form of EysenckÕs hypothesized model to illustrate the typical level of functioning for Introverts and High Performance Low High Performance Low Rested baseline Sleep deprived Sleep/Wake threshold Extraverts Optimal level of arousal Extraverts Introverts Extraverts along a single arousal performance curve. As evident in the top panel of Fig. 1, the typical basal arousal level of Extraverts is slightly lower than that of Introverts, although with all conditions being equal, both would typically function within the optimal zone of arousal. Extraverts, on average, fall slightly to the left of Introverts along the stimulation arousal axis, indicating that they require more stimulation than Introverts to achieve their ÔoptimalÕ level of Introverts Optimal level of arousal Low arousal Normal alertness High arousal Level of arousal Over aroused Figure 1. The figure is adapted from Eysenck (1964) and Hebb (1955) and shows the presently hypothesized model of the effects of sleep deprivation on Introverts and Extraverts based on the theory of optimal levels of stimulation and arousal by Eysenck (1964, 1967, 1981). According to EysenckÕs theory, cortical arousal is related to performance in the form of an inverted U-shaped function, with optimal levels of performance associated with moderate levels of arousal. Eysenck also suggested that Introverts have a higher basal level of cortical arousal than Extraverts, and therefore, require less stimulation to achieve optimal arousal. The hypothesized relationship is represented in black dashed lines for Introverts and gray solid lines for Extraverts. According to the model, both types of individuals may start off functioning within the optimal range of arousal at rested baseline (top panel), but with prolonged wakefulness, the corresponding reduction in cortical arousal, as indicated by the leftward movement of the arrows in the lower panel, would be associated with greater decrements in performance for Extraverts (solid gray lines), whereas the greater baseline cortical arousal of Introverts (black dashed lines) would provide an initial resistance by virtue of remaining nearer to the peak of the curve during the initial decline in arousal. With increasing duration of wakefulness (i.e. reduction in arousal), the initial resistance shown by Introverts would be predicted to diminish.

3 356 W. D. S. Killgore et al. arousal. By contrast, with a chronically high level of basal arousal of the reticulo-thalamic-cortical activation system, Introverts are suggested to generally prefer a lower level of stimulation and can become easily over-aroused as the level of stimulation is increased. The hypothesized relationship between cortical arousal and I E has received support from a variety of psychophysiological and neuroimaging methods (Zuckerman, 1992), including electroencephalography (EEG) (Matthews and Amelang, 1993), evoked potentials (Ditraglia and Polich, 1991) and functional neuroimaging (Johnson et al., 1999; Kumari et al., 2004; Mathew et al., 1984), although negative or inconsistent findings have been reported as well (for review see Zuckerman, 2005). Notably, an early study by Mathew et al. (1984) found that cerebral blood flow negatively correlated with a measure of Extraversion, suggesting that Introverts show greater cortical arousal than Extraverts. Similarly, Johnson et al. (1999) correlated brain activation as measured by H 15 2 O positron emission tomography (PET) with I E scores and found that, consistent with EysenckÕs theory, Introversion correlated with greater cerebral blood flow within the prefrontal cortex and anterior thalamus, brain regions that also show prominent reductions in activity during sleep loss (Thomas et al., 2000). Similarly, Kumari et al. (2004) found that higher Extraversion correlated with lower resting fmri signal intensity within the cortex, again suggesting lower levels of basal cortical arousal in this group. In sum, considerable theoretical and empirical evidence suggests that individual differences in the trait of I E are reflective of trait levels of cortical arousal, particularly in regions where brain activity is most affected by sleep loss. Based on the predictions of EysenckÕs I E theory, Fig. 1 illustrates a hypothesized model of how these trait differences between Introverts and Extraverts in tonic arousal of the reticulo-thalamic-cortical activation system may contribute to trait-like individual differences in the vulnerability to fatigue and sleep loss. As shown in the figure, when well rested and functioning within the optimal level of stimulation arousal, both Introverts and Extraverts should perform similarly on a task of attention and psychomotor vigilance. However, as the level of cortical arousal is decreased by sleep deprivation, the higher tonic level of arousal of Introverts may provide a protective buffer against early degradation of performance because their arousal levels would still remain within their ÔoptimalÕ zone longer than the Extraverts. Extraverts, by contrast, with their lower level of basal cortical arousal, would be predicted to show a more pronounced decline in performance with roughly the same amount of sleep loss. Furthermore, with progressively longer periods of sleep loss, the distance between the curves diminishes and, thus, the advantage of Introverts would be predicted to diminish as the period of wakefulness is extended. Based on this model, we hypothesized that the trait of Extraversion significantly correlated with declines in psychomotor vigilance speed and the number of attentional lapses in a sample of healthy military volunteers deprived of sleep for 77 h. METHODS Subjects Twenty-three healthy military personnel (19 men, four women), ranging in age from 20 to 35 years participated (mean 25.3; SD 4.1). Volunteers had all attained at least a high school education (mean 14.0, SD 1.6 years of completed education). As part of a larger study, separate and unrelated findings from this same sample have been reported elsewhere (Killgore et al., 2007a, Killgore et al., 2007b). All participants were required to be native English speakers and all demonstrated at least an eighth grade or higher reading level on the Wide Range Achievement Test-3rd Edition (WRAT-3; Psychological Assessment Resources, Inc., Lutz, FL, USA). To be included in the study, participants had to be non-smokers, low to moderate habitual caffeine users (<300 mg day), and without significant medical, neurological or psychiatric history as determined by the study physician. At intake, participants were also screened for problems with depression and anxiety and were excluded if they scored above 9 on the Beck Depression Inventory (BDI) (Beck and Steer, 1993) and or greater than 40 on either scale of the Spielberger State-Trait Anxiety Inventory (STAI) (Spielberger et al., 1970). These military participants were relatively homogeneous with regard to habitual bedtimes and rise times and were all within the required weight standards based on age, sex and height. Prior to participation, all volunteers provided written informed consent, were medically cleared by a physician, and were given a urine drug screen to confirm that they were free of any stimulant medications or illicit substances. Participants were compensated for their time in the study. The study was approved by the Walter Reed Army Institute of Research Human Use Review Committee and the US Army Human Subjects Research Review Board. Materials The scales and subscales of the Revised NEO Personality Inventory (NEO-PI-R; Costa and Mccrae, 1992) served as the primary predictor variables of interest. The NEO-PI-R consists of five domain scales including Neuroticism, Extraversion, Openness to experience, Agreeableness and Conscientiousness. Each domain is further subdivided into six facet scales that define specific aspects of its parent scale. The psychometric properties of the test are strong, with high internal consistency among items and high test retest reliability for most scales (Costa and Mccrae, 1992; Kurtz et al., 1999). The NEO-PI-R is one of the most widely used instruments for the assessment of personality functioning and has been validated across a wide range of populations and disorders (Bagby et al., 2004; De Fruyt et al., 2000; Quirk et al., 2003; Sherry et al., 2003; Young and Schinka, 2001). Moreover, the NEO-PI-R has been shown to be particularly useful in measuring the major dimensions of personality suggested by Eysenck (1967, 1990), including Neuroticism, Extraversion and Psychoticism (Costa and Mccrae, 1995).

4 Introversion Extraversion and sleep deprivation 357 Simple psychomotor vigilance (i.e. speed of reaction time to a simple visual stimulus) was assessed with the Walter Reed Palm- Held Psychomotor Vigilance Test (Palm-PVT; Thorne et al., 2005). The Palm-PVT is a portable test of reaction time that is administered on a hand-held PDA device. The Palm-PVT emulates other more cumbersome commercially available psychomotor vigilance tests but was designed for greater flexibility and portability within operational environments. In the present study, each administration of the Palm-PVT was 5 min in duration. During each administration, participants were required to visually monitor the PDA screen and press a response button each time a Ôbulls-eyeÕ target appeared. The interstimulus interval was varied pseudorandomly between presentations to prevent the participants from anticipating the onset of the stimulus. For each Palm-PVT administration, the average reaction time across all trials was scored. This was transformed into a measure of Speed (i.e. Speed = 1 Reaction Time 100). The number of attentional lapses (i.e. failure to respond to the target) lasting 1000 ms or longer was also tabulated. These lapses were converted to a change score representing the number of lapses relative to each individualõs rested baseline. Procedure Participants arrived at the sleep laboratory in the evening of the acclimation day (day 0) and were run in groups of two to four at a time. Following a period of training on study tasks and procedures, participants were allowed to relax for the evening. To ensure an equivalent baseline of rest prior to the sleep deprivation phase of the study, all participants were provided 8 h of uninterrupted time in bed in an individual, darkened, sound attenuated room, beginning at 23:00 hours. Participants were awakened at 07:00 hours the following morning (day 1), and remained awake continuously over the next 77 h. At 16:15 hours on day 1 (9.25 h awake), volunteers completed the NEO-PI-R on a PC to provide a baseline (i.e. non-sleep deprived) assessment of the Introversion extroversion and other major domains of personality functioning. Every 2 h, beginning at 09:45 hours, volunteers completed a single Palm-PVT administration. In addition, participants completed three 8.5-h duration Palm-PVT testing blocks, one block per night, over the next three nights. Each testing block began at 00:15 hours and lasted until 08:50 hours and required the volunteers to complete one Palm-PVT administration approximately every 10 min. Additionally, in a double-blind procedure, participants were randomly assigned to receive either caffeine (n = 12; 200 mg in a chewing gum preparation; (Syed et al., 2005) or an identical flavored placebo gum (n = 11), administered every 2 h each night at 01:00, 03:00, 05:00 and 07:00 hours, for a total of 800 mg of caffeine distributed over the entire night). Analyses Data from the NEO-PI-R were first scored by the automated software package provided with the test to provide summary scale scores. For each participant, summary T-scores with a mean of 50 and standard deviation of 10 were calculated for each of the five domains and the 30 facet scales. In addition to the continuous variables produced by the T-scores, each participant was also classified into one of two categories of Extraversion based on a median split of the T-scores on the Extraversion domain. For this sample, scores of 56 and below were classified into the Introverted group (n = 11; mean I E T-score = 48.2, SD 7.1) while those scoring 57 and above were classified into the Extraverted group (n = 12; mean I E T-score = 64.5, SD 7.7). Mean I E scores did not differ significantly [t(21) = )0.02, P = 0.98] between those assigned to receive caffeine (mean 56.8, SD 8.8) or placebo (mean 56.6, SD 13.6). Baseline Speed performance on the Palm-PVT was defined as the mean Speed performance for the three afternoon administrations of the Palm-PVT on day 1 (i.e. administered at 13:50, 15:55 and 17:50 hours). The Speed scores for the overnight testing blocks of the Palm-PVT were each converted to a score reflecting the percent change from baseline performance. These percent change scores were then averaged for each third of the night (i.e. each third of the night encompassed 13 administrations: Period ÔaÕ was from 00:15 to 03:10 hours; Period ÔbÕ was from 03:15 to 05:55 hours; Period ÔcÕ was from 06:15 to 08:50 hours) as well as the average percent change in speed performance across the entire night (i.e. averaged across all 39 administrations) to provide an indication of the relative decline in Palm-PVT performance as a result of sleep deprivation. Similarly, the number of lapses (i.e. periods of nonresponsiveness equal to or exceeding 1000 ms) were tabulated for the same three baseline sessions and for each overnight testing session. The number of lapses for each session was subtracted from the mean number of lapses during baseline testing to yield a lapse change score. Separate analyses were conducted for each of the three nights. First, for each night, the percent change in Palm-PVT speed scores relative to baseline was entered into a 3(period) 2(Introversion Extraversion Category) mixed model analysis of covariance (ancova), with drug group entered as a covariate in spss Greenhouse Geisser corrections were applied when the assumption of sphericity was violated. Planned comparisons were conducted between the two Extraversion categories after accounting for drug group at each third of the night. The second stage of analysis involved entering the five domain scores of the NEO-PI-R as predictor variables in a hierarchical multiple linear regression analysis. In the first series of regressions, the mean nightly percent change in Palm- PVT was entered as the dependent variable, whereas in the second series, Palm-PVT lapses were entered as dependent variables. Because the effect of caffeine was not of primary interest in this study, membership in the caffeine or placebo group was forced into the regression analyses in the first stage to statistically control the effects of drug administration. The effects of the personality variables, above and beyond that produced by caffeine, were then entered in a stepwise proce-

5 358 W. D. S. Killgore et al. dure and tested by evaluating the significance of change in R 2. If a domain emerged as a significant predictor, then the partial correlations (pr) controlling for caffeine group, between the associated facet scales and Palm-PVT performance were examined directly using one-tailed directional tests of the hypothesis. RESULTS Palm-PVT Speed For illustration, the percent change in Palm-PVT speed across the three nights of testing is shown in Fig. 2, with separate lines showing the performance of the Introverted and Extraverted groups. Solid lines show the percent change in speed performance for the subjects that received caffeine while dotted lines show performance for the placebo group during each third of the night (i.e. Periods ÔaÕ to ÔcÕ). Not surprisingly, on night 1, speed performance declined significantly across the three sessions of the night, F(1.5, 30.6) = 5.75, P = and caffeine administration had a significant effect on speed performances, F(1, 20) = 33.08, P < 0.001, with the caffeine group showing significantly faster performances than the PVT speed (% of baseline) 120% 100% 80% 60% 40% 20% 0% BL 1a 1b 1c 2a 2b 2c 3a 3b 3c Test day CAFF introverts CAFF extraverts PLAC introverts PLAC extraverts Figure 2. Psychomotor speed declines with continuous wakefulness, but the magnitude of this decline differed according to caffeine administration and the trait of Introversion Extraversion. Speed performance (i.e. Speed = 1 Reaction Time 100) is presented as a percent of baseline (BL) performance from the first day (i.e. administered at 13:50, 15:55 and 17:50 hours). Performances are presented for each of the three overnight sessions from 00:15 to 08:50 hours each morning. Each overnight session is divided into thirds ranging from: (a) 00:15 to 03:10 hours, (b) 03:15 to 05:55 hours and (c) 06:15 to 08:50 hours. Mean (±SE) as a percent of baseline is presented for each of these sessions. Participants receiving caffeine (solid lines) significantly outperformed those receiving placebo (dashed lines) for all three overnight testing sessions. Moreover, participants classified as Introverts significantly outperformed participants classified as Extraverts on the first overnight session, even after the effects of caffeine were statistically controlled. Introversion Extraversion group had no effect on subsequent nights. placebo group. Moreover, regardless of drug group, there was a significant main effect of Introversion Extraversion group, F(1, 20) = 5.34, P = 0.032, with Extraverts showing a significantly greater decline in speed performance at the first (P = 0.05) and third (P = 0.04) periods of the night. By night 2, performance continued to degrade throughout the night F(2, 40) = 4.52, P = 0.017, but those receiving caffeine still outperformed those receiving placebo, F(1, 20) = 18.98, P < In contrast to the previous night, however, the Introverted and Extraverted groups did not differ significantly on the percent change in speed performance, F(1, 20) = 2.41, P = Similarly, by night 3, performance again declined across the three test blocks, F(2, 40) = 3.80, P = , and the caffeine group continued to outperform the placebo group overall, F(1, 20) = 6.44, P = As with the previous night, after accounting for the variance associated with caffeine group, the Introverted and Extraverted groups did not differ significantly with regard to the observed change in speed performance, F(1, 20) = 0.11, P = To determine whether the five domains of personality from the NEO-PI-R provided incremental improvements in predicting Palm-PVT speed performance above and beyond that provided by caffeine group, the domain scores were entered into a series of three hierarchical regression analyses (one for each night), with drug condition forced into the equation at the first step. Not surprisingly, the multiple regression analysis indicated that caffeine administration significantly sustained Palm-PVT speed performance throughout the first night relative to placebo (R 2 = 0.52, P < 0.001; see Table 1). Moreover, as initially hypothesized, higher scores on Extraversion significantly predicted poorer Palm-PVT performance above and beyond the effects of caffeine (b = )0.38, R 2 change = 0.15, P = 0.007). By contrast, neither Extraversion nor any of the other domain scales provided significant prediction for night 2 or night 3. Partial correlation coefficients showing the relationship between the five domains of the NEO-PI-R and the average percent change in Palm-PVT speed performance for each overnight testing session are presented in Table 2. As evident Table 1 Summary of hierarchical regression analysis for variables predicting average percent change in Palm-PVT Speed for each night of sleep deprivation Variable b R 2 R 2 change P Night 1 Caffeine group <0.001 Step 2 Caffeine group 0.73 <0.001 NEO-PI-R Extraversion ) Night 2 Caffeine group Night 3 Caffeine group

6 Introversion Extraversion and sleep deprivation 359 Table 2 Partial correlation coefficients between the five domain scales of the NEO-PI-R and percent change in Palm-PVT Speed for three night time testing blocks NEO-PI-R domain Night 1 Night 2 Night 3 All nights Neuroticism (N) Extraversion (E) )0.56** )0.34 )0.25 )0.47* Openness to Experience (O) Agreeableness (A) )0.09 ) )0.06 Conscientiousness (C) )0.06 )0.23 )0.22 )0.21 Partial correlations controlling for drug condition. One-tailed tests, *P < 0.05, **P < from the table, only the Extraversion domain significantly correlated with psychomotor vigilance speed after controlling for caffeine administration. On the whole, higher Extraversion scores were associated with greater percent declines in psychomotor vigilance speed during the first night of sleep deprivation but not on the subsequent two nights (see Fig. 3, top row). When the speed performances were averaged across all three nights, however, the relationship with Extraversion remained statistically significant. To further analyze the components of Extraversion that most strongly related to the ability to sustain psychomotor vigilance speed, we examined the partial correlations between each facet scale comprising the Extraversion domain and Palm-PVT speed, with drug group statistically controlled. Table 3 shows that four of the six facets, including Warmth (E1), Gregariousness (E2), Assertiveness (E3) and Activity (E4), were all significantly related to the magnitude of decline in psychomotor vigilance speed on the first night of sleep deprivation. Those participants with the highest scores on these facet scales consistently showed the greatest percent declines in their Palm-PVT speed during the first 24-h period of continuous wakefulness. By night 2 of sleep deprivation, only Gregariousness (E2) and Activity (E4) significantly correlated with psychomotor vigilance speed, and only Gregariousness (E2) correlated with such performance by night 3. By contrast, Excitement Seeking (E5) and Positive Emotions (E6) were unrelated to the ability to sustain psychomotor vigilance speed for any of the three overnight sessions. Palm-PVT Lapses The data for lapses equal to or exceeding 1 s in duration were subjected to a series of hierarchical regression analyses in a manner analogous to the Speed data described earlier. Specifically, the number of lapses for each NEO-PI-R domain were entered into a series of three hierarchical regression analyses (one for each night), with drug condition forced into the equation at the first step and change in lapse performance entered as the dependent variable. The multiple regression analysis indicated that caffeine administration significantly reduced lapses throughout the first night relative to placebo (R 2 = 0.41, P = 0.001; see Table 4). Moreover, as initially hypothesized, higher scores on Extraversion significantly Night 1 Night 2 Night 3 Speed % change r 2 =.31* r 2 =.12 r 2 = Lapses change r 2 =.20* r 2 =.00 r 2 = Extraversion Extraversion Extraversion Figure 3. Partial correlation plots show the effect of NEO-PI-R Extraversion score on psychomotor vigilance performance after statistically controlling for the effects of caffeine. (top row) After statistically removing the effects of caffeine group, higher Extraversion scores significantly (*P < 0.05) and inversely correlated with psychomotor vigilance speed for night 1, but not for night 2 or 3 (bottom row). Similarly, after statistically equating groups for caffeine administration, higher Extraversion scores significantly (*P < 0.05) correlated with greater numbers of lapses of 1000 ms or greater on night 1, but not on night 2 or 3.

7 360 W. D. S. Killgore et al. Table 3 Partial correlation coefficients between the six facet scales comprising the NEO-PI-R Extraversion domain and percent change in Palm-PVT Speed for three night time testing blocks NEO-PI-R Extraversion facet Night 1 Night 2 Night 3 All nights Warmth (E1) )0.39* )0.23 )0.16 )0.32 Gregariousness (E2) )0.57** )0.45* )0.37* )0.57** Assertiveness (E3) )0.37* )0.28 )0.10 )0.31 Activity (E4) )0.58** )0.41* )0.33 )0.54** Excitement Seeking (E5) )0.18 )0.16 )0.06 )0.16 Positive Emotions (E6) )0.32 )0.26 )0.02 )0.25 Partial correlations controlling for drug condition. One-tailed tests, *P < 0.05, **P Table 6 Partial correlation coefficients between the six facet scales comprising the NEO-PI-R Extraversion domain and average change in 1-s lapses on the Palm-PVT for three night time testing blocks NEO-PI-R Extraversion facet Night 1 Night 2 Night 3 All nights Warmth (E1) 0.25 ) Gregariousness (E2) 0.58** * Assertiveness (E3) ) Activity (E4) 0.44* Excitement Seeking (E5) Positive Emotions (E6) ) Partial correlations controlling for drug condition. One-tailed tests, *P < 0.05, **P Table 4 Summary of hierarchical regression analysis for variables predicting average change in 1-s lapses on the Palm-PVT for each night of sleep deprivation Variable b R 2 R 2 change P Night 1 Caffeine group ) Step 2 Caffeine group )0.64 <0.001 NEO-PI-R Extraversion Night 2 Caffeine group ) Night 3 Caffeine group ) predicted increased frequency of lapses exclusive of the effects of caffeine (b = 0.34, R 2 change = 0.12, P = 0.038). However, while caffeine was effective at reducing lapses for nights 2 and 3, lapses were not significantly predicted by Extraversion or any of the other domain scales on those nights. For illustration, partial correlation coefficients between the five domains of the NEO-PI-R and the average change in lapses of 1 s or longer for each overnight testing session are presented in Table 5. From the table, it is clear that only the Extraversion domain significantly correlated with attentional lapses after controlling for caffeine administration. Overall, higher Extraversion scores were associated with more frequent Table 5 Partial correlation coefficients between the five domain scales of the NEO-PI-R and average change in 1-s lapses on the Palm-PVT for three night time testing blocks NEO-PI-R domain Night 1 Night 2 Night 3 All nights Neuroticism (N) ) )0.04 Extraversion (E) 0.45* Openness to experience (O) )0.09 )0.01 )0.28 )0.16 Agreeableness (A) )0.02 )0.03 )0.05 )0.04 Conscientiousness (C) ) Partial correlations controlling for drug condition. One-tailed tests, *P < s attentional lapses during the first night of sleep deprivation but not on the subsequent two nights (see Fig. 3, bottom row). Because Extraversion was predictive of attentional lapses, we also examined the partial correlations between each facet scale comprising the Extraversion domain and the change in 1- s lapses, after covariation for drug condition. Table 6 shows that two of the six facets, including Gregariousness (E2) and Activity (E4) both significantly correlated with higher rates of 1-s lapses during the first night of sleep deprivation. By night 2 and night 3, these relationships were no longer significant. DISCUSSION Consistent with the hypothesized relationship between I E and cortical arousal (Eysenck, 1967, 1981; Zuckerman, 2005), higher scores on a measure of Extraversion at rested baseline were associated with greater declines in alertness and psychomotor vigilance during the first 26 h of a three-night period of continuous wakefulness. These effects were above and beyond the alerting effects provided by overnight administration of 200 mg of caffeine every 2 h. Of the five major personality domains of the NEO-PI-R, only Extraversion was significantly predictive of measurable declines in Speed of responding and the number of 1-s lapses on the Palm-PVT during sleep deprivation. Specifically, Introversive traits (i.e. lower Extraversion) was associated with relatively preserved speed of responding and minimal increases in attentional lapses following sleep loss, whereas individuals with higher levels of Extraversion (i.e. low Introversion) demonstrated slower response times and were at greater risk for attentional lapses in excess of 1000 ms. These relationships were only statistically significant for the first night of sleep deprivation, with Extraversion failing to correlate significantly with psychomotor vigilance performance on the second and third overnight testing sessions. In short, the present findings suggest that extraversive personality traits confer some vulnerability to the adverse neurobehavioral effects of sleep deprivation, whereas introversive traits appear to confer some resistance to these effects, particularly within the first 26 h of continuous wakefulness. The present findings are consistent with the predictions that logically follow from EysenckÕs cortical arousal theory of I E

8 Introversion Extraversion and sleep deprivation 361 (Eysenck, 1964, 1967, 1981, 1990) and similar theories that postulate that the psychobiological basis of Introversion is rooted in greater tonic arousal of the ascending reticular activating system and its influences on cortical activity (Gray, 1970; Zuckerman, 2005). With sleep deprivation, cortical and thalamic activity decreases, particularly within the first 24 h of wakefulness (Thomas et al., 2000, 2003). Presumably, the more highly aroused reticulo-thalamic-cortical activation system hypothesized to be characteristic of Introverts provides some initial protection against the decreased arousal, reduced cerebral metabolism and associated attentional declines produced by prolonged wakefulness. As evident in the model in Fig. 1, the higher level of tonic arousal hypothesized to be present in Introverts may afford them some protection against performance declines with initial reductions in arousal due to sleep deprivation. As cortical arousal declines with prolonged wakefulness, extraverted personalities, with their lower basal level of arousal, would be expected to show a rapid drop in performance earlier than individuals with introverted personalities, who would still be expected to be functioning nearer to their optimal level of arousal. By the second night of sleep deprivation, other factors unaccounted for in the present analysis presumably overwhelmed the effects of I E on arousal and psychomotor vigilance. The finding that I E is related to the vulnerability resistance to the adverse cognitive effects of sleep loss is consistent with other research examining the relationship of this personality dimension to sleep and arousal-related phenomena. Previous research has suggested that Extraverts are more adversely affected by a single night sleep loss than Introverts on a test of vigilance for repeated digits and a logical reasoning task (Smith and Maben, 1993). Furthermore, consistent with their higher level of basal arousal, Introverts have been shown to be more likely to experience objective insomnia than Extraverts, and higher scores on the trait of Extraversion (and related traits of sociability and impulsivity) negatively correlated with sleep latency. In other words, highly extraverted individuals appear to fall asleep faster and more easily than their more introverted peers, who tend to be more prone to problems with insomnia (Dorsey and Bootzin, 1997). Thus, the same arousal mechanisms that appear to be associated with insomnia and Introversion in other studies may also account for some of the individual differences in vulnerability to fatigue and sleep loss that have been described in the recent literature (Caldwell et al., 2005; Van Dongen et al., 2004, 2005). Although the construct of Introversion Extraversion appears to be a fundamental dimension of the vast majority of personality theories (Zuckerman, 1992), there are slight variations in how it is conceptualized by various theorists (Costa and Mccrae, 1995) and it is possible that only specific aspects of the construct are related to the ability to resist sleep loss. We therefore also examined the specific sub-components of the Introversion Extraversion factor to determine which were most significantly related to the observed vulnerability to sleep loss. The NEO-PI-R domain of Extraversion comprises six facet scales. Of these facets, Gregariousness and Activity most strongly correlated with declines in psychomotor vigilance Speed and increased lapses following a night without sleep. Gregariousness is defined by the NEO-PI-R Manual as the degree to which an individual prefers the company of other people, whereas Activity is defined as the need to keep oneself busy, active and engaged in vigorous movement (Costa and Mccrae, 1992). Thus, individuals that said they prefer to engage with others in large group settings and enjoy a fastpaced and active schedule tended to show the greatest vulnerability to performance decrements during first night of sleep loss, whereas those individuals that described themselves as ÔlonersÕ and those who prefer to keep a more relaxed tempo tended to sustain their speed of reaction time and were less likely to have attentional lapses when deprived of sleep. Interestingly, the facets of Warmth (i.e. affectionate, friendly and closely attached to others) and Assertiveness (i.e. dominant, forceful and strong leadership tendencies) correlated with declines in the speed of reaction time but not with the number of attentional lapses over the first night of wakefulness. While the present findings are consistent with the predictions of the tonic cortical arousal theory of Eysenck, other explanations should also be considered, particularly in light of the correlational nature of these data. For example, it is conceivable that it is not the personality trait of Introversion per se that protects against the vulnerability to sleep deprivation, but instead may be some aspect of the different social learning and experiential environments of Introverts and Extraverts that is the causal factor leading to the observed results. Such a possibility was suggested by a recent study that demonstrated that the sleep need of Drosophilia was heavily dependent upon the recent social environment to which the fruit flies were exposed (Ganguly-Fitzgerald et al., 2006). In that study, fruit flies experimentally segregated to socially impoverished (i.e. housed in isolation) or socially enriched (i.e. housed in large groups) environments demonstrated significantly different needs for sleep. Isolated flies showed significantly less need for sleep than those housed together in groups. Moreover, the larger the social group to which the flies were exposed, the larger the corresponding need for sleep. These effects were reversible in individual flies, dependent primarily on the most recent social encounter, and could not be accounted for by activity level, size of living space, reproductive status or sexual activity. Interestingly, humans engage in a similar process of social exposure via self-selection, with Introverts tending to avoid large or frequent social encounters, whereas Extraverts seek out such environments. While speculative, it is conceivable that the self-selected social environments of individuals of varying personality type could also contribute to general arousal, fatigue levels and changes in the functioning of the cortex and limbic system that could affect the objective need for sleep and associated vulnerability to sleep loss. Because our study was controlled for group size and all subjects maintained fairly regular bedtimes and rise times, it is unlikely that such factors played a significant role in the present findings. However, future research could examine this more closely by

9 362 W. D. S. Killgore et al. manipulating the degree of social contact of Introverts and Extraverts prior to and or during sleep deprivation to see whether personality or social exposure has the greater effect on the vulnerability to sleep loss. Several limitations should be kept in mind when interpreting these results. First, the present sample was limited in size, which may have contributed to the lack of statistically significant relationships between personality variables and Palm-PVT performance for the second and third nights. Insufficient statistical power may have also contributed to the failure to find statistically significant relationships between psychomotor vigilance and other personality variables, such as Neuroticism, that are also believed to play a modulatory role on cortical arousal (Eysenck, 1967; Zuckerman, 2005). Replication with a larger sample will be necessary to examine whether the present findings are stable and whether other aspects of personality may interact with I E to influence vulnerability to sleep deprivation. Secondly, while we included caffeine group membership as a covariate in all statistical analyses, it is possible that this control was insufficient to account for the group differences in the correlations. Future studies with larger samples could directly compare the effects of caffeine on the strength of the associations or examine the effects of personality and vigilance without the influence of stimulants. Finally, the present study used only a single selfreport measure of the construct of I E, based on responses to the NEO-PI-R. Other measures of the I E construct, such as the Eysenck Personality Inventory (Eysenck and Eysenck, 1964) or the Behavioral Inhibition and Behavioral Activation Scales (BIS BAS; Carver and White, 1994) should also be included in future research attempting to replicate these findings. These findings suggest that previously reported individual differences in the vulnerability to sleep deprivation (Van Dongen et al., 2004, 2005), and conversely, the ability to sustain alertness and vigilance performance when sleep deprived (Caldwell et al., 2005), may be partly accounted for by individual variations in the personality dimension of I E. It is postulated that the measurable relationship between I E and performance during sleep deprivation is due to a combination of trait-like differences in tonic cortical arousal as proposed by Eysenck (1967), and selfselection of social environments and activities (e.g. Ganguly- Fitzgerald et al., 2006) consistent with I E related personality preferences that either promote or limit the subsequent need for recuperative sleep. A suggested focus for future research would be to clarify the relative contribution of genetic trait-like factors in tonic cortical arousal versus the role of social exposure and preferred activity level to the ability to resist sleep loss. ACKNOWLEDGEMENTS The views expressed in this paper are those of the authors and do not reflect the official policy or position of the Walter Reed Army Institute of Research, the Department of the Army, the Department of Defense, the US Government, or any of the institutions with which the authors are affiliated. REFERENCES Bagby, R. M., Schuller, D. R., Marshall, M. B. and Ryder, A. G. Depressive personality disorder: rates of comorbidity with personality disorders and relations to the five-factor model of personality. J Personal Disord., 2004, 18: Beck, A. T. and Steer, R. A. Manual for the Beck Depression Inventory. The Psychological Corporation, San Antonio, TX, Belenky, G., Wesensten, N. J., Thorne, D. R., Thomas, M. L., Sing, H. C., Redmond, D. P., Russo, M. B. and Balkin, T. J. Patterns of performance degradation and restoration during sleep restriction and subsequent recovery: a sleep dose response study. J Sleep Res, 2003, 12: Caldwell, J. A., Mu, Q., Smith, J. K., Mishory, A., Caldwell, J. L., Peters, G., Brown, D. L. and George, M. S. Are individual differences in fatigue vulnerability related to baseline differences in cortical activation? Behav. Neurosci., 2005, 119: Carver, C. S. and White, T. L. Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: the BIS BAS scales. J. Pers. Soc. Psychol., 1994, 67: Costa, P. T. Jr and Mccrae, R. R. Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) Professional Manual. Psychological Assessment Resources, Inc., Odessa, FL, Costa, P. T. Jr and Mccrae, R. R. Primary traits of EysenckÕs P E N system: three- and five-factor solutions. J. Pers. Soc. Psychol., 1995, 69: De Fruyt, F., Mervielde, I., Hoekstra, H. A. and Rolland, J. P. Assessing adolescentsõ personality with the NEO PI-R. Assessment, 2000, 7: Ditraglia, G. M. and Polich, J. P300 and introverted extraverted personality types. Psychophysiology, 1991, 28: Dorsey, C. M. and Bootzin, R. R. Subjective and psychophysiologic insomnia: an examination of sleep tendency and personality. Biol. Psychiatry, 1997, 41: Drummond, S. P., Brown, G. G., Gillin, J. C., Stricker, J. L., Wong, E. C. and Buxton, R. B. Altered brain response to verbal learning following sleep deprivation. Nature, 2000, 403: Drummond, S. P., Brown, G. G., Salamat, J. S. and Gillin, J. C. Increasing task difficulty facilitates the cerebral compensatory response to total sleep deprivation. Sleep, 2004, 27: Eysenck, H. J. Crime and Personality. Houghton Mifflin, Boston, MA, Eysenck, H. J. The Biological Basis of Personality. Charles C. Thomas, Springfield, IL, Eysenck, H. J. General features of the model. A model for personality. Springer, Berlin, Eysenck, H. J. Genetic and environmental contributions to individual differences: three major dimensions of personality. J. Pers., 1990, 58: Eysenck, H. J. and Eysenck, S. B. G. Manual of the Eysenck Personality Inventory. University of London Press, London, Ganguly-Fitzgerald, I., Donlea, J. and Shaw, P. J. Waking experience affects sleep need in Drosophila. Science, 2006, 313: Gray, J. A. The psychophysiological basis of introversion extraversion. Behav. Res. Ther., 1970, 8: Hebb, D. O. Drives and the C.N.S. (conceptual nervous system). Psychol. Rev., 1955, 62: Johnson, D. L., Wiebe, J. S., Gold, S. M., Andreasen, N. C., Hichwa, R. D., Watkins, G. L. and Boles Ponto, L. L. Cerebral blood flow and personality: a positron emission tomography study. Am. J. Psychiatry, 1999, 156:

10 Introversion Extraversion and sleep deprivation 363 Killgore, W. D. S., Killgore, D. B., Day, L. M., Li, C., Kamimori, G. H. and Balkin, T. J. The effects of 53 hours of sleep deprivation on moral judgment. Sleep, 2007a, 30: Killgore, W. D. S., Lipizzi, E. L., Kamimori, G. H. and Balkin, T. J. Caffeine effects on risky decision-making after 75 hours of sleep deprivation. Aviation Space & Environmental Medicine, 2007b, 78: Kumari, V., Ffytche, D. H., Williams, S. C. and Gray, J. A. Personality predicts brain responses to cognitive demands. J. Neurosci., 2004, 24: Kurtz, J. E., Lee, P. A. and Sherker, J. L. Internal and temporal reliability estimates for informant ratings of personality using the NEO PI-R and IAS. NEO Personality Inventory. Interpersonal Adjective Scales. Assessment, 1999, 6: Mathew, R. J., Weinman, M. L. and Barr, D. L. Personality and regional cerebral blood flow. Br. J. Psychiatry, 1984, 144: Matthews, G. and Amelang, M. Extraversion, arousal theory and performance: a study of individual differences in EEG. Pers. Individ. Diff., 1993, 14: Quirk, S. W., Christiansen, N. D., Wagner, S. H. and Mcnulty, J. L. On the usefulness of measures of normal personality for clinical assessment: evidence of the incremental validity of the Revised NEO Personality Inventory. Psychol. Assess., 2003, 15: Sherry, A., Henson, R. K. and Lewis, J. G. Evaluating the appropriateness of college-age norms for use with adolescents on the NEO personality inventory-revised. Assessment, 2003, 10: Smith, A. and Maben, A. Effects of sleep deprivation, lunch, and personality on performance, mood, and cardiovascular function. Physiol. Behav., 1993, 54: Spielberger, C. D., Gorsuch, R. L. and Lushene, R. E. Manual for the State-Trait Anxiety Inventory. Consulting Psychologist Press, Palo Alto, CA, Syed, S. A., Kamimori, G. H., Kelly, W. and Eddington, N. D. Multiple dose pharmacokinetics of caffeine administered in chewing gum to normal healthy volunteers. Biopharm. Drug Dispos., 2005, 26: Thomas, M., Sing, H., Belenky, G., Holcomb, H., Mayberg, H., Dannals, R., Wagner, H., Thorne, D., Popp, K., Rowland, L., Welsh, A., Balwinski, S. and Redmond, D. Neural basis of alertness and cognitive performance impairments during sleepiness. I. Effects of 24 h of sleep deprivation on waking human regional brain activity. J. Sleep Res., 2000, 9: Thomas, M., Sing, H. S., Belenky, G., Holcomb, H. H., Mayberg, H. S., Dannals, R. F., Wagner, H. N., Thorne, D. R., Popp, K. A., Rowland, L. M., Welsh, A. B., Balwinski, S. M. and Redmond, D. P. Neural basis of alertness and cognitive performance impairments during sleepiness II. Effects of 48 and 72 h of sleep deprivation on waking human regional brain activity. Thalamus Relat. Syst., 2003, 2: Thorne, D. R., Johnson, D. E., Redmond, D. P., Sing, H. C., Belenky, G. and Shapiro, J. M. The Walter Reed palm-held psychomotor vigilance test. Behav. Res. Methods Instrum. Comput., 2005, 37: Van Dongen, H. P. and Dinges, D. F. Sleep, circadian rhythms, and psychomotor vigilance. Clin. Sports Med., 2005, 24: Van Dongen, H. P., Baynard, M. D., Maislin, G. and Dinges, D. F. Systematic interindividual differences in neurobehavioral impairment from sleep loss: evidence of trait-like differential vulnerability. Sleep, 2004, 27: Van Dongen, H. P., Vitellaro, K. M. and Dinges, D. F. Individual differences in adult human sleep and wakefulness: Leitmotif for a research agenda. Sleep, 2005, 28: Yerkes, R. M. and Dodson, J. D. The relation of strength of stimulus to rapidity of habit-formation. J. Comp. Neurol. Psychol., 1908, 18: Young, M. S. and Schinka, J. A. Research Validity Scales for the NEO-PI-R: additional evidence for reliability and validity. J. Pers. Assess., 2001, 76: Zuckerman, M. What is a basic factor and which factors are basic? Turtles all the way down Pers. Individ. Diff., 1992, 13: Zuckerman, M. Psychobiology of Personality. Cambridge University Press, New York, 2005.