Brain and Cognition 132 (2019) 89–97
Contents lists available at ScienceDirect
Brain and Cognition journal homepage: www.elsevier.com/locate/b&c
Sleep deprivation moderates neural processes associated with passive auditory capture
T
Alexandra Muller-Gassa,b,c, , Kenneth Campbellc,d ⁎
a
Defence Research and Development Canada, Toronto Research Centre, Canada Canadian Forces Health Services, Directorate of Mental Health, Ottawa, Canada c School of Psychology, University of Ottawa, Ottawa, Canada d Brain and Mind Institute, University of Ottawa, Ottawa, Canada b
ARTICLE INFO
ABSTRACT
Keywords: Total sleep deprivation Partial sleep deprivation Attention capture Event-related potentials P3a MMN
Sleep loss has a major effect on cognitive tasks that are dependent on the maintenance of active sustained attention. This study examines the effects of sleep deprivation on automatic information processing, more specifically, its effect on processes leading to involuntary auditory attention capture by task-irrelevant auditory events. Two experiments were run. In the first, 13 participants were totally sleep-deprived (TSD); in the second, 16 participants were partially sleep-deprived (PSD), sleeping only four hours. Event-related potentials were recorded while participants discriminated the duration of equiprobable short and long auditory tones. At rare times, a small change to the pitch of these stimuli occurred. This deviant was however irrelevant to the duration detection task. As expected, TSD had a significant effect on the attention-dependent duration detection task; performance was worse and the P3b, associated with ease of detection, was attenuated. PSD had a similar, but reduced effect. Critically, the small pitch deviant resulted in less behavioural distraction following TSD compared to normal sleep. Consistent with this, the P3a, associated with the attention capture process, was significantly reduced following both TSD and PSD. Processes related to the passive switching of attention to potentially critical, but unattended, stimulus events are moderated by sleep deprivation.
1. Introduction
1.1. Auditory attention capture
A hallmark of modern society is that individuals often choose to forgo sleep in order to pursue recreational and work opportunities. The consequences of sleep loss on an individual’s performance and safety are however far from benign. Sleep loss increases the risk of humanerror-related accidents, and has been implicated in some of the most devastating human and environmental health disasters. There is a general consensus that sleep deprivation mainly affects tasks that are dependent on the maintenance of active, sustained, selective attention, particularly for monotonous tasks (Killgore, 2010; Krause et al., 2017; Lim & Dinges, 2010; Lo et al., 2012). Relatively few studies have investigated the effect of sleep loss on automatic information processing, that is, processing that does not require the maintenance of sustained, endogenous attention. The present study examines whether sleep deprivation affects auditory attention capture, an automatic cognitive process that is biologically adaptive but can also be disruptive to ongoing task performance.
Auditory attention capture is a mechanism by which potentially relevant auditory input that occurs outside the current focus of attention is involuntarily processed, often causing a disengagement of attention away from the current task. The purpose of a mechanism for detecting unattended sensory information is to alert the individual to such potentially relevant and possibly threatening events. It is thus critical that processing of these auditory stimuli results in an interruption of the central executive, resulting in a switch of attention from the current cognitive task-at-hand and toward the potentially much more relevant auditory event. This process is also called passive or involuntary attention (James, 1890). This biological adaptation provides a means by which the individual is able to detect sensory input that is critical for survival regardless of current processing priorities. The attention switch, nevertheless, comes at a cost. Most potentially relevant input turns out to be incidental and is not critical for survival. Because processing resources have been switched to a task-irrelevant
⁎ Corresponding author at: Canadian Forces Health Services Group Head Quarters, Directorate of Mental Health, Carling Campus, Building 9, 101 Colonel By Drive, Ottawa, ON K1A 0K2, Canada. E-mail address:
[email protected] (A. Muller-Gass).
https://doi.org/10.1016/j.bandc.2019.03.004 Received 14 August 2018; Received in revised form 22 February 2019; Accepted 19 March 2019 0278-2626/ © 2019 Published by Elsevier Inc.
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
input, performance on the current task may deteriorate; this process is called distraction and is an obstacle to task execution.
Przuntek, & Daum, 2001). Comparison among different studies is however difficult because of the use of widely varying methodologies. The MMN is much affected by factors such as the type of deviance, extent of deviance, probability of deviant occurrence and the rate of stimulus presentation. Moreover, Atienza, Cantero, and Quian Quiroga (2005) noted that sleep deprivation may result in latency jitter at the time of the MMN. When the latency jitter was corrected, the amplitude of the MMN did not differ between normal sleep and sleep deprivation conditions.
1.2. Event-related potentials It is difficult to study involuntary or passive processing using traditional measures of performance because the subject does not respond to (or may not be aware of) the presentation of the stimuli that lead to the switching of attention. The study of passive processing is facilitated by the recording of event-related potentials (ERPs) because they provide a means of determining the extent to which to-be-ignored stimuli are processed. ERPs are the minute changes in the electrical activity of the brain that are elicited by a physical stimulus or an internal, psychological event. They consist of a series of positive- and negative-going components, thought to reflect different aspects of processing. The process of auditory attention capture starts with the detection of acoustic change, which is reflected by an ERP component called the Mismatch Negativity (MMN). The MMN is often elicited using the socalled passive auditory oddball paradigm. The subject is presented with a regular, homogenous sequence of auditory standard stimuli and instructed to ignore these stimuli. At rare and unpredictable times, a feature of the standard is changed. This odd or rarely occurring stimulus is called a deviant. Both the standard and deviant stimuli elicit an obligatory negative-going auditory ERP component, N1, peaking at about 100 ms following stimulus onset. The deviant however elicits an additional negativity, the MMN, peaking 100–250 ms following stimulus onset. The MMN can be automatically elicited by changes in any physical feature of the stimulus, including its pitch, location, duration, and/or intensity, but also by a violation in prediction formed on the basis of the rules that are automatically detected in recent auditory stimulation (Näätänen, 1992; Näätänen, Kujala, & Winkler, 2011; Paavilainen, 2013; Winkler, 2007; Winkler, Denham, & Nelken, 2009). MMN amplitude varies directly with the extent of change. If the extent of change is large enough, the deviant may also elicit a later positivity, the P3a, peaking about 250–350 ms. It is thought that the P3a reflects a switch of attention from ongoing cognitive tasks toward the acoustic change (Escera, Alho, Winkler, & Naatanen, 1998); although there is current debate about whether the P3a reflects the actual involuntary switching of attention or is simply part of the process that leads to this switch (Parmentier, 2014; Wetzel, Schröger, & Widmann, 2013). There is nevertheless general agreement that the P3a is associated with the process of passive attention. In contrast, active attention to the oddball sequence and the detection of the deviant is associated with a later positivity, the P3b, peaking at about 300–600 ms, depending on the difficulty of detection. P3b has a more posterior centro-parietal scalp distribution whereas the P3a has a more anterior centro-frontal distribution. The P3b is known to be strongly affected by sleep deprivation because it does require sustained attention in order to be elicited (Morris, So, Lee, Lash, & Becker, 1992).
1.4. The effects of sleep loss and sleep quality on the P3a in auditory-visual paradigms The effect of sleep quality and sleep loss on the amplitude of P3a has received limited study using either auditory-visual or auditory-auditory paradigms. An auditory-visual paradigm is one in which the participant is instructed to ignore the auditory stimulation that includes the taskirrelevant auditory change, while attending to a visual task. In contrast, in an auditory-auditory paradigm, the task-irrelevant auditory change occurs within an auditory stimulus sequence to which the participant must attend in order to perform an assigned task. The studies that have examined the effect of sleep loss and quality on the P3a using auditory-visual paradigms have also reported inconsistent findings (Atienza, Cantero, & Stickgold, 2004; Gumenyuk et al., 2010, 2011; Gumenyuk, Howard, Roth, Korzyukov, & Drake, 2014; Salmi et al., 2005). In most of these studies, participants were instructed to ignore the auditory stimulation and read a book or watch a silent movie. Atienza et al. (2004) observed that a pitch deviant elicited a small amplitude P3a and its amplitude was reduced following 48 and 72 h of sleep deprivation. Gumenyuk et al. (2014) also provided evidence of a deficiency in the neural processes involved in passive auditory capture as a result of acute sleep deprivation. They examined the effects of auditory distraction on permanent night workers with and without shift work sleep disorder (SWSD), before and after acute sleep deprivation. The amplitude of the P3a to novel environmental sounds was reduced with sleep deprivation but it did not however significantly differ between the two groups. Smaller P3a waveforms, particularly over frontal scalp regions, have also been shown to occur following task-irrelevant novel environmental sounds in habitual short sleepers compared to normal sleepers (Gumenyuk et al., 2011). Studies investigating the effect of sleep quality or disorders on P3a using auditory-visual paradigms show seemingly contradictory findings. Gumenyuk et al. found that a task-irrelevant novel sound elicited a more prominent P3a in a group of permanent night workers with SWSD than in those without SWSD. A duration deviant, representing a much smaller extent of change, however elicited a P3a waveform that did not differ between the groups. Similarly, Salmi et al. (2005) found that the P3a elicited by a duration deviant increased in amplitude as sleep quality measures deteriorated. The authors speculated that this finding might be attributed to an inability to sustain attention to the assigned reading task following lower quality sleep, thus permitting eavesdropping on (or sampling of) the irrelevant auditory sequence.
1.3. The effects of sleep loss and sleep quality on the MMN In general, the MMN has been demonstrated to be reliably elicited by a deviant stimulus regardless of the direction or strength of attention (Muller-Gass, Stelmack, & Campbell, 2005; 2006). The effects of chronic sleep loss and sleep quality appear to have inconsistent effects on the MMN: some studies have shown that the MMN is not altered in both healthy subjects and patients manifesting poor sleep quality or sleep loss (Gosselin et al., 2006; Salmi et al., 2005), although habitual short sleepers and individuals with shift work sleep disorder (SWSD) may show an attenuated MMN (Gumenyuk et al., 2010). Similarly, the effects of acute sleep deprivation on the amplitude of the MMN are inconsistent, with some studies indicating that it may decrease the amplitude of the MMN (Raz, Deouell, & Bentin, 2001; Zerouali, Jemel, & Godbout, 2010), while other studies showing no effect (Bortoletto, Tona, Scozzari, Sarasso, & Stegagno, 2011; Naumann, Bierbrauer,
1.5. The effects of sleep loss and sleep quality on the P3a in auditoryauditory paradigms A few studies have examined the effect of sleep loss and quality on P3a in an auditory-auditory paradigm. In this type of paradigm, the deviant stimulus is embedded within the task-relevant auditory stimulation and therefore is consistently afforded at least some basic processing. It is for this reason that the P3a is much more prominent even when the extent of deviance is small (Schröger & Wolff, 1998). Gosselin, De Koninck, and Campbell (2005) asked subjects to attend to an auditory oddball sequence that included infrequent pitch deviants and environmental novel sounds. The subject was asked to detect the pitch deviants but withhold their response to the environmental sounds. 90
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
2. Experiment 1: Total sleep deprivation
The environmental sounds elicited a large P3a, the amplitude of which was reduced following total sleep deprivation. The authors suggested that the processes resulting in attention capture were affected by the loss of sleep. Gosselin et al. (2006) examined processing in patients with obstructive sleep apnea syndrome (OSAS) using a paradigm that is commonly employed to study auditory attention capture, originally developed by Schröger and Wolff. Participants were presented with a sequence of equally probable short and long duration auditory tones and asked to discriminate the duration of the stimuli. At rare times, a large change to the pitch of these stimuli was made; this deviant feature was however irrelevant to the duration detection task. The authors observed a reduced P3a in OSAS patients compared to controls following the presentation of the large pitch deviant. They concluded that OSAS inhibits the ability to shift attention to potentially highly relevant input.
2.1. Methods 2.1.1. Participants Thirteen young adults (7 females) between the ages of 20 and 31 years (Mean = 24.1, SD = 3.4 years) volunteered to participate in this study. All participants had self-reported normal hearing, were in good health, and none had a history of neurological or psychiatric disorder. All were right-handed, and were not taking any medications known to affect cognitive function. Absence of sleep disorders was verified using the Pittsburgh Sleep Index (Buysse, Reynolds, Monk, Berman, & Kupfer, 1989). Participants were required to maintain a regular sleep schedule (no night or shift work). They were asked to abstain from alcohol and caffeine in the 24 h period prior to the start of data collection. A written informed consent was obtained prior to the beginning of the experiment. This study was conducted following the guidelines of the Canadian Tri-Council (Health, Natural, and Social Sciences) on ethical conduct involving humans.
1.6. The present study Only a limited number of studies have examined the effects of sleep loss on the ERP components underlying involuntary attention capture by task-irrelevant auditory change. Most of these studies have featured auditory-visual paradigms which are more prone to attentional leakage toward the irrelevant auditory sequence, in particular when the visual task is easy to perform (Muller-Gass et al., 2006). Also, the effect of sleep quality or sleep loss on attention capture has been most consistently demonstrated using novel environmental sounds, which represent a large extent of deviance from the standard stimulus. In general, these studies have found that a decrease in the amplitude of the P3a seems to occur following sleep loss. This would suggest a decreased ability to switch attention from a current cognitive task to a potentially more relevant auditory stimulus. Unfortunately, none of the ERP findings were accompanied with independent behavioural performance evidence to support this claim. Presumably, if attention is switched away from the cognitive task, performance on it should then deteriorate. The present study examines the effects of sleep loss on involuntary attention capture using an auditory-auditory paradigm similar to that developed by Schröger and Wolff. In this context, the benefit of this paradigm is threefold: 1) it provides a demanding task to tightly focus the participant’s attention, 2) it allows for small deviants (representing only a small extent of change from the standard) to elicit a prominent P3a and 3) it provides an independent behavioural measure of involuntary attention capture. Performance on the duration discrimination task often deteriorates following the presentation of the deviant. Accuracy is lower and/or reaction time (RT) longer compared to when the standard is presented. Because participants perform the same duration discrimination task for deviants as for standards, this decay in performance is assumed to be a result of attention capture and distraction - the switching of attention away from the processing of the relevant feature (the detection of the duration of the deviant) and toward the processing of an irrelevant feature (its pitch). Furthermore, this study will investigate attention capture following both total and partial sleep deprivation. While partial sleep deprivation occurs much more often than total sleep deprivation, relatively few studies have examined its effects on cognitive processing. One study has however suggested that automatic auditory processing can be affected by partial sleep deprivation (Zerouali et al., 2010). If sleep deprivation does affect processes related to attention capture, it would be expected that the presentation of the deviant will be less likely to result in an interruption of the central executive and a subsequent switch of attention. Sleep deprivation might thus be associated with less behavioural distraction, and a reduced P3a, following presentation of the deviant.
2.1.2. Procedure The participants were tested in two different sessions, following normal sleep (NS) and following total sleep deprivation (TSD). The order of the sessions was counterbalanced across participants and at least one week separated the two sleep conditions. Participants kept a sleep log for four consecutive nights, prior to both sleep conditions. The sleep logs did not differ prior to the NS and TSD testing sessions. On the sleep-deprived night, a research assistant was present at all times to ensure participants did not sleep, even for brief periods of time. They watched videos, read or played computer games for the entire duration of the night. They were tested between 10:00 and 12:00 the morning following either normal sleep or sleep deprivation. Upon coming to the lab, participants completed the Stanford Sleepiness Scale (SSS), a 7point rating scale (1 = awake and alert, 7 = falling asleep). They were then fitted with an electrode cap, and seated in a sound-attenuated testing lab where they were instructed to perform the experimental task. 2.1.3. Experimental task Participants engaged in a duration classification task following both sleep conditions. They were presented with a sequence of binaural 70 dB SPL auditory stimuli transduced by Sony MDR-V6 headphones. Half of the stimuli had a duration of 190 ms and the other half, a duration of 310 ms. Participants were asked to press either the left or right button of a computer mouse upon detection of the short- and longduration stimuli; this association was counterbalanced across conditions. On 84% of trials, a 1000 Hz standard stimulus was presented. This was changed on the remaining 16% of trials to a 950 Hz deviant stimulus. This change in stimulus frequency was however irrelevant to the duration detection task. Participants were thus required to signal the duration of the stimuli, whether they were standard or deviant tones. The standard and deviant stimuli were presented in a pseudorandom manner. The auditory sequence started with a minimum of 4 consecutive standards, in order to establish a memory for the features of the standard. Deviants were separated by at least 3 standards. A stimulus occurred every 1400 ms. Prior to the start of the experiment, the participant was trained in the classification task. During the experiment, two blocks of 500 trials were presented, each block lasting about 11.7 min. A brief break was given between blocks. Stimulus presentation and response monitoring were controlled by E-prime software (Psychology Software Tools Inc., Sharpsburg, Pennsylvania) using a PC with a Windows XP operating system. Timing is much more precise with the Windows XP compared to newer versions of the operating system. Timing of stimulus onset was verified to be within 2 ms of the trigger with little jitter (SD = 0.20 ms) using a Brain Products’ StimTrak system. 91
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
2.1.4. EEG recording EEG activity was recorded from 63 sites over frontal, central, parietal, temporal, occipital sites and the two mastoids using active silversilver chloride electrodes attached to an electrode cap (Brain Products, GmbH, Munich, Germany). An electrode was also placed on the infraorbital ridge of the left eye to record vertical eye movements. The tip of the nose was used as a reference for all channels. Inter-electrode impedances varied from 20 to 50 kΩ. A low pass filter was set at 250 Hz. The high pass filter was set at 0.08 Hz (i.e., a time constant of 2 s). The EEG was continuously digitized at a 500 Hz sampling rate and stored on hard disk for later analyses. Offline, the data were reconstructed using Brain Products’ Analyzer2 software. The continuous data were digitally filtered using a high filter set at 20 Hz. The EEG was visually inspected for channels containing high levels of noise. These channels were replaced by interpolating the data of the surrounding electrode sites (Perrin, Pernier, Bertrand, & Echallier, 1989). The data of participants containing more than 4 channels with excess noise were rejected. This was the case for one subject during the TSD condition. Thus, data for this participant were rejected for both the TSD and NS conditions. A vertical EOG was computed by subtracting activity recorded at electrode site FP1 from that at the infra-orbital ridge. A horizontal EOG was computed by subtracting the activity recorded at FT9 from that at FT10. Independent Component Analysis (Delorme, Sejnowski, & Makeig, 2007) was used to identify eye movements and blinks that were statistically independent of the EEG activity. These vertical and horizontal eye movements were then partialled out from the EEG.
deviant waveforms at each electrode site in the same condition. The subtraction routine thus removes processing that is common to both the standard and deviant stimuli, leaving processing that is unique to the deviant. The difference wave consisted of a negative deflection peaking at about 150 ms (the MMN) and a later positive deflection peaking at about 300 ms (P3a). The mean amplitude of MMN and P3a were computed by averaging all data points within ± 25 ms of the peak amplitude in the grand average (the average of all participants’ average). Electrode sites were grouped into regions of interest (ROIs). Mean amplitudes for the MMN were computed separately for frontal (Fz, F3, F4), central (C3, Cz, C4) and mastoid (M1, M2) ROIs. These electrodes were chosen based on previous studies reporting maximum MMN at these fronto-central electrode sites and the well-known MMN polarity inversion at the mastoids (Horváth, Winkler, & Bendixen, 2008; Naatanen, 1995). A 2-way ANOVA with repeated measures on sleep condition (NS, TSD) and electrode site was performed on the MMN data for each ROIs (frontal, central and mastoid). The P3a is largest over centro-frontal sites, and therefore frontal (F3, Fz, F4) and central (C3, Cz, C4) ROIs were retained. Again, a 2-way ANOVA with repeated measures on sleep condition and electrode site was performed on the P3a data for each ROI (frontal, central). An α level of 0.05 was used for all statistical tests. Significant interactions were followed up by simple main effects testing. Greenhouse-Geisser correction procedures were applied when the assumption of sphericity was violated. 2.2. Results 2.2.1. Subjective ratings of sleepiness As expected, subjective ratings of sleepiness on the Stanford Sleepiness Scale significantly increased following the TSD compared to the NS condition. The mean subjective rating was 2.00 (SD = 0.77) following NS and 5.91 (SD = 1.36) following TSD, t(10) = 9.43, p < .01.
2.1.5. Data analyses 2.1.5.1. Behavioural performance data. Performance on the classification task was measured in terms of accuracy (hit rate) and reaction time (RT), and determined separately for standard and deviant stimuli. Correct responses occurring between 150 and 1000 ms from the offset of the stimulus were classified as hits. RTs longer than 1000 ms were considered outliers and removed from the analysis. This was the case in fewer than 2% of trials. The mean RT was computed by averaging only those trials that included hits. Participants for whom accuracy of detection was below 0.60 (i.e., near chance detection) were rejected from further analyses for both the performance and physiological data. This was the case for one participant in the sleep deprivation condition, and therefore this participant’s data were excluded from the study. Initial analyses of the performance data did not reveal either main or interactive effects of stimulus duration. The performance data were therefore collapsed across short- and long-duration stimuli. This also allowed for an increase in the number of stimulus trial epochs available for averaging. A 2-way ANOVA with repeated measures on sleep condition (NS, TSD) and stimulus (standard, deviant) was run on the performance data. Separate ANOVAs were run for the accuracy and RT measures.
2.2.2. Behavioural performance data Table 1 presents performance results for the duration classification task. The interaction between the type of stimulus and sleep condition was significant F(1,10) = 5.39, p < .043, η2p = 0.30. Simple effects testing showed that, as expected, the accuracy of classification significantly decreased following the TSD compared to the NS condition, in both standard and deviant trials. The source of the interaction was due to a deleterious effect of the deviant (compared to the standard) on performance in the NS condition only. More specifically, in the NS condition, participants performed well (accuracy = 0.89) on standard trials but showed a significant performance decrement during deviant trials (11% more errors). In contrast, in the TSD condition, participants’ performance was not significantly affected by the occurrence of the deviant (only 5% more errors on deviant trials). For RT, the interaction between the type of stimulus and sleep condition was also significant, F(1,10) = 5.92, p < .03, η2p = 0.37 Sleep condition did not significantly affect RT on both standard and deviant trials (F < 1). The source of the interaction was again due to a
2.1.5.2. ERP data. The continuous EEG was reconstructed into single trial epochs, beginning 100 ms prior to stimulus onset and continuing for 900 ms following stimulus offset. The average of the activity in the 100 ms pre-stimulus period served as a zero-voltage baseline. Drifts in voltage from the baseline were then corrected for each epoch. Epochs in which EEG activity exceeded ± 100 μV were excluded from further analyses. The single epochs were sorted and averaged on the basis of electrode site, stimulus type (standard, deviant) and sleep condition (NS, TSD). Only EEG epochs in which a hit was achieved were included in the average. There were too few epochs containing errors to permit an analysis of the associated ERPs. The processing of the deviants is best observed in a difference ERP wave computed by subtracting point-by-point, the standard from the
Table 1 Accuracy of detection and reaction time (RT) during the duration classification task following normal sleep and total sleep deprivation. Sleep condition
Normal Sleep Total Sleep Deprivation
92
Stimulus
Standard Deviant Standard Deviant
Accuracy
RT (ms)
Mean
SD
Mean
SD
0.89 0.78 0.75 0.70
0.05 0.08 0.08 0.11
576 615 579 599
53 64 51 52
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
after normal sleep, but the difference was not significant, F (1,10) = 1.83 and 2.16 respectively, p > .20. The small amplitude P2 was not affected by the amount of sleep, F < 1. The P3b, associated with the active detection of the short and long duration tones, was visible at about 570–600 ms following the presentation of both the standard and the deviant stimuli. It was maximal over the parietal region of the scalp, and therefore was quantified at Pz as the average of all data points within ± 25 ms of the peak as measured in the grand average waveform. The amplitude of the standard P3b was significantly reduced following the TSD compared to the NS condition, F (1,10) = 4.89, p < .05, η2p = 0.24. The deviant P3b was also reduced following TSD but the difference was not significant, F (1,10) = 3.62, p > 0.08. 2.2.3.2. Difference waveforms. Fig. 2 illustrates the deviant – standard difference waveforms for both sleep conditions at all ROIs. As may be observed, the processing that was common to the standard and deviant stimuli (including the N1 and P3b) was largely removed in the subtraction process, leaving the processing that was unique to the deviant (the MMN and P3a). A small amplitude MMN (at Fz: −1.57 μV) was elicited at about 120 ms following stimulus onset. This MMN was maximal over frontocentral areas of the scalp and did invert in amplitude at the mastoids. The MMN was not however significantly affected by sleep condition, and the interaction between sleep condition and electrode site was not significant, F < 1 in both cases and for all ROIs. The MMN was followed by a prominent P3a (at Cz: 4.41 μV) that was maximal at centrofrontal sites and peaked at about 300 ms. The main effect of sleep condition was significant, F(1,10) = 7.14 and 5.33, p < .02 and .05, η2p = 0.42 and 0.35 for the frontal and central ROIs, respectively. The P3a was significantly attenuated following the TSD compared to the NS condition. The interaction between sleep condition and electrode site was not significant, F < 1 for both frontal and central ROIs. Fig. 3 presents the spherical scalp distribution maps of the MMN and P3a as a function of sleep condition. As may be observed, the MMN displayed a typical frontal maximum negativity with an inversion in polarity over inferior, lateral sites. This scalp distribution was quite similar between NS and TSD conditions. The P3a displayed a much more centralfrontal scalp distribution. Although it was attenuated following TSD, its scalp distribution was similar for both sleep conditions.
Fig. 1. Experiment 1: ERPs associated with the duration detection task. In this and all other figures, positivity at the scalp relative to the reference is illustrated as an upward deflection. Participants were asked to detect the duration of the frequently occurring standard and the rarely occurring deviant stimulus. The ERPs to these stimuli consisted of a prominent N1 at about 100 ms and a large positivity, the P3b at about 600 ms. The standard N1 was slightly larger following TSD, but the difference was not significant. The P3b, prominent over parietal regions of the scalp, was reduced following TSD for the presentation of the standard and the deviant but the difference was only significant for the standard.
stronger deleterious effect of the deviant (compared to the standard) in the NS condition. More specifically, RT was significantly longer following presentation of the deviant compared to the standard in the NS condition (+39 ms); but this effect was reduced in the TSD condition (+20 ms). 2.2.3. ERP data 2.2.3.1. Standard and deviant ERPs. The ERPs elicited by the standard and deviant stimuli at Fz, Cz and Pz electrode sites are illustrated in Fig. 1. The amplitude of N1 at Fz and Cz following presentation of the standard stimulus was slightly larger after total sleep deprivation than
Fig. 2. Experiment 1: Deviant-standard difference waves. The presentation of the deviant was associated with a small negativity, the MMN, maximum over fronto-central regions of the scalp, inverting in polarity at the mastoids (M1, M2). The deviant also elicited a large early positivity, the P3a, maximum over centro-frontal regions of the scalp. The P3a was significantly attenuated in the TSD condition.
93
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
Table 2 Accuracy of detection and reaction time (RT) during the duration classification task following normal sleep and partial sleep deprivation. Sleep condition
Normal Sleep Partial Sleep Deprivation
Stimulus
Standard Deviant Standard Deviant
Accuracy
RT (ms)
Mean
SD
Mean
SD
0.84 0.71 0.78 0.68
0.09 0.10 0.10 0.12
521 625 510 634
53 50 57 57
chloride electrodes attached to an electrode cap. 3.1.5. Data analyses Performance and ERP data were analyzed in a comparable manner to that used in Experiment 1. 3.2. Results 3.2.1. Subjective ratings of sleepiness The subjective ratings of sleepiness on the Stanford Sleepiness Scale significantly increased following the PSD compared to the NS condition, although the difference was smaller compared to those observed with TSD. The mean subjective rating was 1.93 (SD = 0.62) following NS and 3.07 (SD = 1.07) following PSD, t(13) = 3.66, p < .01.
Fig. 3. Experiment 1: Spherical scalp distribution maps of the MMN and P3a as a function of the amount of sleep. A linear, equidistant perspective of a “flattened” head is illustrated. In this view, the projection was extended down 20° below the Fp1-T7-Oz-T8-Fp2 circumference to show data from the mastoids (or TP9 and TP10) and inferior frontal-temporal (FT9, FT10) sites. The maps of the MMN in the NS and TSD conditions are very similar. While the P3a was significantly reduced following TSD, its centro-frontal prominence was similar in both conditions, although the maximum of P3a may be more widespread in the NS condition.
3.2.2. Behavioural performance data Table 2 presents performance results for the duration classification task. Accuracy of detection declined following the PSD compared to NS condition, but the difference did not reach significance, F(1,13) = 2.56, p > .13. As expected, the overall effect of stimulus type on performance accuracy was significant, F(1,13) = 22.32, p < .01, η2p = 0.63; accuracy deteriorated by about 12% following presentation of the deviant compared to the standard. The interaction between type of stimulus and sleep condition failed to reach significance, F(1,13) = 2.93, p > .11. The trend to a greater deleterious effect of the deviant in the NS compared to the sleep-deprived condition was however again observed (as was the case in Experiment 1). Accuracy decreased by 13%
3. Experiment 2: Partial sleep deprivation 3.1. Methods Methods similar to those used in Experiment 1 (Total Sleep Deprivation) were employed; only differences between experiments will be reported here. 3.1.1. Participants Sixteen right-handed young adults (10 females) between the ages of 20 and 30 years (Mean = 23.1, SD = 2.9 years) volunteered to participate in this study. The data of one participant were rejected because of excessive artifact in the EEG and the data of another participant were rejected because of poor performance on the duration discrimination behavioural task during the session following partial sleep deprivation. 3.1.2. Procedure The participants were tested in two different sessions, following normal sleep (NS) and following partial sleep deprivation (PSD). Again, the order of the sessions was counterbalanced across participants and at least one week separated the two sleep conditions. On the night of the PSD session, participants were instructed to sleep at 03:00 and were awakened at 07:00; in order to ensure compliance, participants were required to wear a wrist actigraph. Subsequent analyses of the actigraph data did indicate that participants followed this sleep schedule. 3.1.3. Experimental task The duration classification task employed was identical to that of Experiment 1.
Fig. 4. Experiment 2: ERPs associated with the duration detection task. In the PSD condition, the N1 following the presentation of the standard was significantly enhanced compared to that from the NS condition. The amplitude of the P3b was reduced in the PSD condition for both the standards and deviants, but the difference was not significant.
3.1.4. EEG recording EEG activity was recorded from 32 sites using active silver-silver
94
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
Fig. 5. Experiment 2: Deviant-standard difference waves. Again, PSD did not affect the amplitude of the MMN. On the other hand, PSD was associated with a significant reduction in the amplitude of the P3a.
for deviant compared to standard trials in the NS condition whereas accuracy only decreased by 10% in the PSD condition. Sleep did not affect overall RT, F < 1. RT significantly increased following presentation of the deviant compared to the standard in both sleep conditions F(1,13) = 48.88, p < .01, η2p = 0.68. The interaction between type of stimulus and sleep condition failed to reach significance, F(1,13) = 2.46, p > .14. 3.2.3. ERP data 3.2.3.1. Standard and deviant ERPs. The ERPs elicited by the standard and deviant stimuli are illustrated in Fig. 4. The amplitude of the standard N1 at frontal and central ROIs was significantly larger in the PSD than in the NS condition, F(1,13) = 4.84 and 13.28, p < .05 and .01, η2p = 0.27 and 0.51, respectively. The amplitude of the standard P2 was again small and did not significantly vary between the NS and PSD conditions at either the frontal or central ROIs, F < 1. The amplitude of the standard and deviant P3b measured at Pz was reduced following partial sleep deprivation, but the difference did not reach significance, F(1,13) = 1.69 and 1.26, p > .22 and .26.
Fig. 6. Experiment 2: Spherical scalp distribution maps of the MMN and P3a as a function of the amount of sleep. The maps of the MMN are again very similar in the NS and PSD conditions. Again, while the P3a was significantly reduced following PSD, its centro-frontal prominence was similar regardless of the amount of sleep.
3.2.3.2. Difference waveforms. Fig. 5 illustrates the deviant-standard difference waveforms for both sleep conditions at all ROIs. The spherical scalp distribution maps for the MMN and P3a are presented in Fig. 6. A small amplitude MMN (at Fz: −1.48 μV) was apparent at about 120 ms; but again, its amplitude did not significantly vary with sleep condition and the interaction between sleep condition and electrode site was not significant, F < 1 in both case and for all ROIs. A prominent centro-frontal P3a (at Cz: 3.58 μV) was elicited following the MMN. The P3a was significantly reduced in amplitude following the PSD compared to NS condition, at both frontal or central ROIs, F(1,13) = 6.49 and 7.30, p < .05, η2p = 0.33 and 0.36 for the frontal and central ROIs, respectively. The interaction between sleep condition and electrode site was not significant, F < 1 for both frontal and central ROIs. Following NS, the spherical scalp distribution maps of the MMN and P3a were quite similar to those observed in Experiment 1. These scalp distribution maps did not change following PSD (Fig. 6).
4. Discussion The present study examined the effect of sleep deprivation on the neural mechanisms underlying attention capture by a small task-irrelevant change in auditory stimulation. The most important result indicated that sleep deprivation had a moderating effect on auditory attention capture, evidenced by a reduction of P3a amplitude and a reduction of behavioral distraction following total sleep deprivation when compared to normal sleep. Sleep loss also led to a reduction of P3a in the partial sleep condition; in this condition a trend toward an abatement of the distracting effect on performance was also found, although this result failed to reach significance. Interestingly, there was
95
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
little evidence that the pre-conscious detection of the auditory change, a process that precedes the attention capture and is reflected by the MMN, was affected by sleep deprivation. Together, these results suggest that sleep deprivation reduces the effectiveness of an auditory change to interrupt the central executive, thus decreasing the likelihood that attention will be switched to this potentially highly relevant signal, and that it will be further evaluated.
deviance that causes the switching of attention. ERPs do however provide a means of determining the extent to which such to-be-ignored stimuli are processed. This study made use of an auditory-auditory distraction paradigm (Schröger & Wolff, 1998) that allowed the recording of both ERP and behavioral evidence of distraction. The auditory-auditory paradigm achieves this by embedding the task-irrelevant deviance within the task-relevant auditory stimulus sequence.
4.1. General effects of sleep loss during the total and partial sleep deprivation conditions
4.3.1. Behavioural evidence of the switching of attention In the auditory-auditory paradigm, performance on the primary task often deteriorates following the presentation of the task-irrelevant deviant. The presentation of the deviant causes processing resources to be switched from the task-relevant stimulus feature (its duration) to the task-irrelevant deviant feature (its pitch) and this switch comes at a cost to primary task performance. This is seen as a reduced accuracy and/or a prolonged RT for the deviant compared to the standard stimulus. The deleterious effect of the deviant (compared to the standard) on performance was quite apparent in the NS conditions of both experiments: participants performed well on standard trials but showed a significant performance decrement during deviant trials in the form of more errors and prolonged RT. The deviant was however much less distracting to the participants during the TSD condition: participants’ accuracy was not significantly affected by the occurrence of the deviant, and RT was much less affected by the deviant stimulus during TSD than during the NS condition. During PSD, there was also a trend toward a smaller deleterious effect of the deviant on performance accuracy compared to the NS condition, but this difference did not reach significance. These results suggest that sleep loss may dampen the attention capture mechanisms, that the sleep-deprived participants are less likely to switch attention to a potentially relevant auditory stimulus and thus the extent of deterioration in performance between standard and deviant trials is lessened.
Subjective, behavioral and ERP results indicate that subjects experienced effects of sleep loss in both sleep deprivation conditions, but more so in the TSD compared to the PSD condition. Although the extent of self-reported sleepiness was reduced after 4 h of sleep (PSD condition) compared to total sleep deprivation, it was still significantly greater than following normal sleep. Similarly, accuracy on the auditory classification task decreased with sleep loss in both sleep conditions; however this difference failed to reach significance for the PSD condition. The P3b is an ERP component that reflects processes related to sustained active attention, evaluation, and selection. As it is wellknown that sleep deprivation affects tasks that are dependent on the maintenance of sustained selective attention (Killgore, 2010; Lim & Dinges, 2010; Lo et al., 2012), it was not surprising that the P3b to the standard stimuli was reduced following both sleep conditions, but again this difference did not reach significance in the PSD condition. Finally, the N1 to the standard stimuli was larger following the partial sleep compared to the normal sleep condition; it was also slightly larger after the TSD, but the difference was not significant. Previous studies have generally shown that N1 is either unaffected or enhanced by the effects of total sleep deprivation (Bortoletto et al., 2011; Lee et al., 2004; Raz et al., 2001). The increases in N1 following sleep deprivation may be related to the often-reported compensatory attentional “effort” (Drummond et al., 2005). Other authors (Bortoletto et al., 2011) suggest that sleep deprivation might induce an increase in cortical excitability resulting in the enhanced N1. Together these results suggest that sleep-deprived subjects in this study felt sleepier, had a decreased level of performance on primary task, were less effective in maintaining active attention, but might have exerted more compensatory effort. As expected, this sleep deprivation effect varied according to the extent of sleep loss.
4.3.2. P3a evidence of the switching of attention P3a reflects the neural processes involved in involuntary attention capture by task-irrelevant auditory change, and its amplitude often varies according to the obtrusiveness of the stimulus. The ERP results are in accordance with the behavioral results: In both sleep deprivation conditions, the P3a elicited by the deviant was reduced in amplitude when compared to that elicited during the NS condition. Again, this suggests that sleep loss reduces the strength of the mechanisms underlying the involuntary shift of attention to an auditory change. Only a limited number of studies have previously examined the effects of sleep loss or sleep quality on the P3a. Most of these studies have employed auditory-visual paradigms in which the participant is instructed to attend to a visual task and ignore the auditory stimulation that includes the task-irrelevant. In general, these studies have found that a decrease in the amplitude of the P3a seems to occur following sleep loss, suggesting a decreased ability to switch attention from a current visual task to a potentially more relevant auditory stimulus (Atienza et al., 2004; Gumenyuk et al., 2011, 2014). A few studies have examined the effect of sleep loss or quality on P3a in an auditory-auditory paradigm. Consistent with our results, Gosselin et al. (2005) showed that the amplitude of P3a elicited by obtrusive environmental sounds was reduced following total sleep deprivation; whereas Gosselin et al. (2006) observed a reduced P3a in OSAS patients compared to controls following the presentation of the large pitch deviant. They concluded that OSAS inhibits the ability to shift attention to potentially highly relevant input.
4.2. The effect of sleep loss on the detection of acoustic change The process of auditory attention capture is initiated with the preconscious detection of acoustic change, which is reflected by the MMN. In the present study, a small but distinct MMN was elicited in all sleep conditions. This result is in accordance with the resilience of the processes underlying MMN elicitation to the direction and strength of attention (Muller-Gass, Stelmack, & Campbell, 2006). As was outlined in the Introduction, the effects of sleep loss have had inconsistent effects on the MMN, most studies indicating no significant effect or an attenuation of the MMN following sleep loss. Widely discrepant methodologies across studies however make generalization difficult. In the present study, the MMN elicited by a small frequency change was not altered by either partial or even total sleep deprivation. As such, it would appear that the early automatic, attention-independent detection of stimulus change is relatively well-preserved even following total sleep deprivation.
4.4. Conclusion
4.3. The effect of sleep loss on auditory attention capture
Sleep deprivation is known to affect tasks that require active sustained attention, particularly those that are monotonous and demand vigilance. The finding that P3a is reduced following sleep deprivation lends support to the idea that prolonged wakefulness can affect
It is difficult to study involuntary processes such as those leading up to auditory attention capture using traditional measures of performance because typically the subject is asked to ignore the task-irrelevant 96
Brain and Cognition 132 (2019) 89–97
A. Muller-Gass and K. Campbell
executive processes involved in the capturing of attention; importantly, these processes occur involuntarily and do not necessitate endogenous attentional resources. The results of this study also indicate that both total sleep deprivation and partial sleep deprivation, a condition that occurs much more often in real life, affect the mechanisms underlying involuntary auditory attention capture; but that this effect is reduced for partial sleep loss during which more attentional resources are purportedly available. Detection of change is critical for survival in almost all mammals. In humans, acoustic change forms the basis of many warning and alarm systems. A failure to detect warning signals may of course have devastating consequences. From our results, it would appear that a major effect of sleep deprivation is that an auditory change may be less effective in interrupting the central executive, thus decreasing the likelihood of a switch of attention to the processing of this potentially highly relevant stimulus event. It is important to note that the capture of attention has both negative and positive consequences. As observed in the present study, the processing of the deviant will often come at a cost to the performance of the current task-at-hand because limited attentional resources are switched from task-relevant to task-irrelevant processing. The dampening of the attentional capture mechanism may therefore serve the protective function of being less distracted by irrelevant change when operating under reduced attentional resources (i.e. sleep deprivation conditions), and assure the optimization of task execution.
related potential study. Clinical Neurophysiology, 117(10), 2228–2235. https://doi. org/10.1016/j.clinph.2006.07.130. Gumenyuk, V., Howard, R., Roth, T., Korzyukov, O., & Drake, C. L. (2014). Sleep loss, circadian mismatch, and abnormalities in reorienting of attention in night workers with shift work disorder. Sleep, 37(3), 545–556. https://doi.org/10.5665/sleep.3494. Gumenyuk, V., Roth, T., Korzyukov, O., Jefferson, C., Bowyer, S., & Drake, C. L. (2011). Habitual short sleep impacts frontal switch mechanism in attention to novelty. Sleep, 34(12), 1659–1670. https://doi.org/10.5665/sleep.1430. Gumenyuk, V., Roth, T., Korzyukov, O., Jefferson, C., Kick, A., Spear, L., ... Drake, C. L. (2010). Shift work sleep disorder is associated with an attenuated brain response of sensory memory and an increased brain response to novelty: An ERP study. Sleep, 33(5), 703–713. https://doi.org/10.1093/sleep/33.5.703. Horváth, J., Winkler, I., & Bendixen, A. (2008). Do N1/MMN, P3a, and RON form a strongly coupled chain reflecting the three stages of auditory distraction? Biological Psychology. https://doi.org/10.1016/j.biopsycho.2008.04.001. James, W. (1890). The principles of psychology. New York Holt, 1, 697. https://doi.org/ 10.1037/10538-000. Killgore, W. D. S. (2010). Effects of sleep deprivation on cognition. Progress in Brain Research, 185(C), 105–129. https://doi.org/10.1016/B978-0-444-53702-7.00007-5. Krause, A. J., Simon, E. B., Mander, B. A., Greer, S. M., Saletin, J. M., Goldstein-Piekarski, A. N., & Walker, M. P. (2017). The sleep-deprived human brain. Nature Reviews Neuroscience, 18(7), 404–418. https://doi.org/10.1038/nrn.2017.55. Lee, H.-J., Kim, L., Kim, Y.-K., Suh, K.-Y., Han, J., Park, M.-K., ... Lee, D.-H. (2004). Auditory event-related potentials and psychological changes during sleep deprivation. Neuropsychobiology, 50(1), 1–5. https://doi.org/10.1159/000077933. Lim, J., & Dinges, D. F. (2010). A meta-analysis of the impact of short-term sleep deprivation on cognitive variables. Psychological Bulletin, 136(3), 375–389. https://doi. org/10.1037/a0018883. Lo, J. C., Groeger, J. A., Santhi, N., Arbon, E. L., Lazar, A. S., Hasan, S., ... Dijk, D.-J. (2012). Effects of partial and acute total sleep deprivation on performance across cognitive domains individuals and circadian phase. PLoS ONE, 7(9), e45987. https:// doi.org/10.1371/journal.pone.0045987. Morris, A. M., So, Y., Lee, K. A., Lash, A. A., & Becker, C. E. (1992). The P300 eventrelated potential. The effects of sleep deprivation. Journal of Occupational Medicine, 34(12), 1143–1152. Muller-Gass, A., Stelmack, R. M., & Campbell, K. B. (2005). “…and were instructed to read a self-selected book while ignoring the auditory stimuli”: The effects of task demands on the mismatch negativity. Clinical Neurophysiology, 116(9), 2142–2152. https://doi.org/10.1016/j.clinph.2005.05.012. Muller-Gass, A., Stelmack, R. M., & Campbell, K. B. (2006). The effect of visual task difficulty and attentional direction on the detection of acoustic change as indexed by the mismatch negativity. Brain Research, 1078(1), https://doi.org/10.1016/j. brainres.2005.12.125. Naatanen, R. (1995). The mismatch negativity: A powerful tool for cognitive neuroscience. Ear and Hearing, 16(1), 6–18. Näätänen, R. (1992). Attention and brain function. Attention and brain function. Hillsdale, NJ, US: Lawrence Erlbaum Associates Inc. Näätänen, R., Kujala, T., & Winkler, I. (2011). Auditory processing that leads to conscious perception: A unique window to central auditory processing opened by the mismatch negativity and related responses. Psychophysiology, 48(1), 4–22. https://doi.org/10. 1111/j.1469-8986.2010.01114.x. Naumann, A., Bierbrauer, J., Przuntek, H., & Daum, I. (2001). Attentive and preattentive processing in narcolepsy as revealed by event-related potentials (ERPs). Neuroreport, 12(13), 2807–2811. Paavilainen, P. (2013). The mismatch-negativity (MMN) component of the auditory event-related potential to violations of abstract regularities: A review. International Journal of Psychophysiology. https://doi.org/10.1016/j.ijpsycho.2013.03.015. Parmentier, F. B. R. (2014). The cognitive determinants of behavioral distraction by deviant auditory stimuli: A review. Psychological Research, 78(3), 321–338. https:// doi.org/10.1007/s00426-013-0534-4. Perrin, F., Pernier, J., Bertrand, O., & Echallier, J. F. (1989). Spherical splines for scalp potential and current density mapping. Electroencephalography and Clinical Neurophysiology, 72(2), 184–187. https://doi.org/10.1016/0013-4694(89)90180-6. Raz, A., Deouell, L. Y., & Bentin, S. (2001). Is pre-attentive processing compromised by prolonged wakefulness? Effects of total sleep deprivation on the mismatch negativity. Psychophysiology, 38(5), 787–795. Salmi, J., Huotilainen, M., Pakarinen, S., Siren, T., Alho, K., & Aronen, E. T. (2005). Does sleep quality affect involuntary attention switching system? Neuroscience Letters, 390(3), 150–155. https://doi.org/10.1016/j.neulet.2005.08.016. Schröger, E., & Wolff, C. (1998). Behavioral and electrophysiological effects of task-irrelevant sound change: A new distraction paradigm. Cognitive Brain Research, 7(1), 71–87. https://doi.org/10.1016/S0926-6410(98)00013-5. Wetzel, N., Schröger, E., & Widmann, A. (2013). The dissociation between the P3a eventrelated potential and behavioral distraction. Psychophysiology, 50(9), 920–930. https://doi.org/10.1111/psyp.12072. Winkler, I. (2007). Interpreting the mismatch negativity. Journal of Psychophysiology, 21(3–4), 147–163. https://doi.org/10.1027/0269-8803.21.34.147. Winkler, I., Denham, S. L., & Nelken, I. (2009). Modeling the auditory scene: Predictive regularity representations and perceptual objects. Trends in Cognitive Sciences. https://doi.org/10.1016/j.tics.2009.09.003. Zerouali, Y., Jemel, B., & Godbout, R. (2010). The effects of early and late night partial sleep deprivation on automatic and selective attention: An ERP study. Brain Research, 1308, 87–99. https://doi.org/10.1016/j.brainres.2009.09.090.
Funding This work was supported by the Applied Research Project 14dm funded by the Department of National Defence, Canada. All intellectual property derived from this work, including the proprietary right under copyright law, shall however be retained by the University of Ottawa. This study was also supported by grants from the Natural Sciences and Engineering Research Council (NSERC) of Canada to KC. Acknowledgements The authors gratefully acknowledge the assistance of Dominique Gosselin, Rocio Lopez and Paniz Tavakoli with the collection of the data. References Atienza, M., Cantero, J. L., & Quian Quiroga, R. (2005). Precise timing accounts for posttraining sleep-dependent enhancements of the auditory mismatch negativity. NeuroImage, 26(2), 628–634. https://doi.org/10.1016/j.neuroimage.2005.02.014. Atienza, M., Cantero, J. L., & Stickgold, R. (2004). Posttraining sleep enhances automaticity in perceptual discrimination. Journal of Cognitive Neuroscience, 16(1), 53–64. https://doi.org/10.1162/089892904322755557. Bortoletto, M., Tona, G. D. M., Scozzari, S., Sarasso, S., & Stegagno, L. (2011). Effects of sleep deprivation on auditory change detection: A N1-mismatch negativity study. International Journal of Psychophysiology, 81(3), 312–316. https://doi.org/10.1016/j. ijpsycho.2011.07.017. Buysse, D. J., Reynolds, C. F., Monk, T. H., Berman, S. R., & Kupfer, D. J. (1989). The Pittsburgh sleep quality index: A new instrument for psychiatric practice and research. Psychiatry Research, 28(2), 193–213. https://doi.org/10.1016/0165-1781(89) 90047-4. Delorme, A., Sejnowski, T., & Makeig, S. (2007). Enhanced detection of artifacts in EEG data using higher-order statistics and independent component analysis. NeuroImage, 34(4), 1443–1449. https://doi.org/10.1016/j.neuroimage.2006.11.004. Drummond, S. P. A., Bischoff-Grethe, A., Dinges, D. F., Ayalon, L., Mednick, S. C., & Meloy, M. J. (2005). The neural basis of the psychomotor vigilance task. Sleep, 28(9), 1059–1068. https://doi.org/10.1093/sleep/28.9.1059. Escera, C., Alho, K., Winkler, I., & Naatanen, R. (1998). Neural mechanisms of involuntary attention. Journal of Cognitive Neuroscience, 10(5), 590–604. https://doi. org/10.1162/089892998562997. Gosselin, A., De Koninck, J., & Campbell, K. B. (2005). Total sleep deprivation and novelty processing: Implications for frontal lobe functioning. Clinical Neurophysiology, 116(1), 211–222. https://doi.org/10.1016/j.clinph.2004.07.033. Gosselin, N., Mathieu, A., Mazza, S., Petit, D., Malo, J., & Montplaisir, J. (2006). Attentional deficits in patients with obstructive sleep apnea syndrome: An event-
97