By Alan Z. Segal, MD

Associate Professor of Clinical Neurology, Weill Cornell Medical College

SYNOPSIS: Electroencephalogram studies of humans during periods of “mind wandering” and “mind blanking” have shown regional changes that suggest parts of the brain may be asleep while other areas are activated.

SOURCE: Andrillon T, Burns A, Mackay T, et al. Predicting lapses of attention with sleep-like slow waves. Nat Commun 2021;12:3657.

With every hour spent awake over the course of a day, our “sleep pressure” mounts. It has been believed that falling asleep is the only way for the brain to restore homeostatic balance. However, more recent research suggests that even while awake, localized areas of the brain can rest. This process has been well documented using depth electrodes in rodents and it has been demonstrated to a lesser extent on surface electroencephalogram (EEG) in humans. When local sleep occurs, a specific signature is observed on EEG, showing regional zones of high amplitude slow waves (delta frequency), closely mimicking deep (stage 3), non-rapid eye movement (REM) sleep. Importantly, this pattern is distinct from drowsiness, in which more global changes occur on EEG. The drowsy state more closely resembles stage 1 sleep showing a global dampening rather than a rise in EEG amplitude and showing brain wave frequencies in the low alpha to upper theta range, rather than 1 Hz to 3 Hz delta waves.

Behaviorally, local sleep may be associated with lapses of attention. When subjects are asked to stay on task, especially a boring one, their mind may turn inward, wandering into unrelated thoughts or even becoming vacant entirely. In the Andrillon et al study, states of mind wandering (MW) and mind blanking (MB) are distinguished from the fully attentive state of being “on-task” (ON). Both MW and MB are found to be different from ON, but more importantly, MW and MB are shown to be distinct from each other.

Participants (n = 32) were placed in a dimly lit room and asked to perform go/no-go tasks. In one paradigm, they were shown a series of faces and asked to press a button for all neutral faces (go) and avoid this response (no-go) for any smiling faces. Similarly, they were shown digits (one through nine) and asked to press the button for any digit (go) that was not the digit three (no-go). The task was interrupted at random intervals during which participants also were asked whether they felt they were “task focused” (ON) or whether they were “off task,” either focusing on some other thoughts (MW) or focusing on nothing (MB). Also, they rated their level of vigilance on a scale between one (extremely sleepy) to four (extremely alert).

The results of this investigation showed there was more inattention in the MB state, since misses were recorded when go responses were required. Reaction time also was slower. By contrast, the MW state suggested hyperarousal, with an increase in false alarm activations when no-go responses were required. Reaction times were shortened correspondingly. Overall, MB was a sluggish mental state, while MW suggested impulsivity.

The investigators also used larger pupil size as a measure of vigilance. While both MW and MB showed smaller pupils than ON, there was no difference when MW and MB were compared directly. This occurred despite the observation that subjects reported they were more vigilant when in MW.

However, more striking were the EEG data. When MW was compared to ON, there was an increase in high amplitude, slow frontal lobe activity. This suggests that in MW, the frontal lobe might be downregulated, shifting focus and allowing the mind to wander off topic. MB also showed this frontal lobe activity, but, in addition, it showed slow wave activity in central-parietal areas, suggesting a more widespread distribution of sleep-like brain waves. When comparing MW and MB, frontal lobe, high-amplitude activity was more pronounced with MW, and parietal waves were more pronounced with MB. This parietal activity, with a sharp upward deflection, followed by a wider downward wave, exhibited the morphology of K-complexes as seen in stage 2 non-REM sleep.

In their discussion, the authors noted their findings are robust and consistent over three complementary parameters: behavioral (false activations and misses), phenomenological (subject reporting of MW and MB), and physiological (EEG). Thus, attentional lapses were dichotomized into two distinct footprints. One showed false activations (impulsivity), MW, and impaired frontal lobe function (with high-amplitude slow wave mimicking sleep). The other showed misses (sluggish mentation), MB, and parietal slow waves, which mimicked stage 2 sleep.


It has been proposed that sleep allows for the clearance of toxic proteins (amyloid beta, among others) from the brain by enhancing the so-called “glymphatic” system. During deep sleep, there can be increased drainage of cerebrospinal fluid through channels in perivascular spaces and through widened gap junctions. Crucially, if focal slow wave EEG patterns can occur even in the waking state, as this paper suggests, it can be concluded that areas of the brain may rest and become restored even without the occurrence of sleep. As the authors noted, local sleep is achieved when attention is “turned inward” rather than focused on the external world. It could be suggested meditation (which has been part of the human experience for thousands of years) might help achieve this. By either letting the mind wander or by wiping our thoughts clean entirely, we actually may produce similar restoration of neural homeostatic balance as if we were actually asleep.

One weakness of this study is the authors did not address REM, a state of cortical activation during sleep (dreaming) that also is thought to be a crucial restorative stage. Following sleep deprivation, not only slow wave sleep but also REM can be observed to rebound. REM is a mixed state, since there is muscle relaxation but also upregulation of the sympathetic nervous system, with increases in heart and respiratory rates. If the concept of “daydreaming” were true to its name, it is possible that while some areas are slowed (such as the frontal lobe in MW), deep brain regions known to be involved in REM, such as the so-called peduncular pontine reticular formation, enter an activated state.