EEG markers of drowsiness do not predict lapses in attention during sleep deprivation

Abstract

During drowsiness, maintaining consistent attention becomes difficult, leading to behavioral lapses. Oscillation bursts in the electroencephalogram (EEG) might predict such lapses, since alpha bursts increase with inattention and theta bursts increase with time spent awake. However, while lapses increase with time awake, paradoxically alpha bursts decrease, and the behavioral relevance of theta bursts is unknown. Therefore, we investigated whether theta or alpha bursts predicted lapses either when well rested (baseline, BL) or sleep deprived (SD). EEG was measured in 18 young adults performing the lateralized attention task, and the timing of bursts was related to trial outcomes (fast, slow, and lapse trials). Against expectations, neither theta nor alpha bursts were more likely during lapses, either at BL or SD. Both were more likely before fast trials, but only at BL. Fast and slow trials were followed by decreases in both theta and alpha bursts, but the effect was reduced during SD, partially due to local increases in bursts. Considering prior literature, these results indicate that bursts have a non-linear relationship with vigilance. The lack of a direct relationship to lapses may be because bursts originate from task-unrelated areas, and the differences during SD suggest these areas change with time awake.