“The repose of the night does not belong to us. It is not the possession of our being. Sleep opens within us an inn for phantoms. In the morning we must sweep out the shadows.”

—Gaston Bachelard, French philosopher, 1960

Flies, birds, mice, dogs, monkeys and people all need to sleep. That is, they show daily periods of relative immobility and lack of response to external stimuli, such as light, sound or touch. This reduced sensitivity to external events distinguishes sleep from quiet resting, whereas the capacity to awaken from slumber distinguishes sleep from coma. Why sleep should be such a prominent feature of daily life across the animal kingdom, despite the fact that it leaves the sleeper unable to confront potential threats, remains mysterious.

Still, much progress in characterizing the physiology and capabilities of the sleeping brain has occurred over the past century, driven by the ability to record electrical activity of the brain (via electroencephalography, or EEG, on the surface of the skull), of the eyes (via electrooculography, or EOG), and of facial or other muscles (via electromyography, or EMG). For scientists, it is this triad of simultaneous measurements that operationally defines the state of sleep, leading to both surprising and counterintuitive insights.

Even without these tools, there are some basic things we do know about sleep. It is essential for our brain to function properly. Most of us have pulled all-nighters or have wanted to sleep but could not, unable to switch off our mind. The next day we are irritable, have trouble keeping our eyes open, and are terrible at tasks that demand sustained attention. Indeed, sleep deprivation causes many traffic accidents—the reason countries have laws that mandate a minimum rest period and maximum working hours for truck drivers.


Sleep is for the brain rather than for the body. Otherwise, eight hours of binge watching our favorite TV series in bed could replace sleep. Thus, we need to look at the sleeping brain to better understand the why and how of a state in which we pass one third of our life.

The Study of Shut-Eye

One milestone in the scientific study of sleep came in 1953, when Eugene Aserinsky and Nathaniel Kleitman of the University of Chicago discovered an until then unnoticed distinction in two distinct forms of sleep: rapid eye movement (REM) sleep and non-REM (NREM) sleep. When subjects are awake, before entering either of these states, their brain waves, recorded via EEG electrodes on the skull, display a typical pattern of electrical activity—low-amplitude, high-frequency signals—while their EMG reveals elevated muscle tone [see illustration above].

As individuals fall asleep and enter lighter and then deeper stages of NREM, also known as deep sleep, their brain waves progressively slow while increasing in amplitude. Eye movements, a hallmark of wakefulness, cease, and muscle tone diminishes. As sleep deepens, assayed by how difficult it is to wake a sleeper, so does the person's EEG. In the most restful form of sleep early on during the night, the EEG is dominated by high-amplitude waves, or oscillations, that slowly wax and wane. Electrical recordings of individual nerve cells in the neocortex directly underneath the skull show regular occurrences of on periods, when cells fire a series of all-or-none electrical pulses, called spikes, as happens when a person is awake. Pulses alternate with off periods, when neurons turn silent. These on and off periods and the associated slow waves in the EEG, termed slow-wave activity (SWA), occur as often as four times every second or as infrequently as once every four seconds (covering a frequency range from 0.25 to four hertz).

NREM sleep is interrupted by shorter episodes of REM sleep during which the EEG has a drastically different character: the slow and large waves are replaced by fast and choppy ones that superficially resemble the awake brain. The same paradoxical activation is seen at the level of individual neocortical neurons that fire spikes with the same intensity as they do during the day. Muscular tone is gone—to all intents and purposes the body is paralyzed—except for the breathing musculature and the jerky, rapid and symmetric movement in each eye that give this phase of sleep its name.


Most of the night is spent in NREM, with the most restorative deep sleep and its associated SWA, taking up 20 to 25 percent of a full night's slumber. Slow-wave activity is homeostatically regulated—that is, the longer somebody stays awake, the deeper and more frequent slow waves occur the following night. Conversely, early in the morning when sleep pressure has lessened, SWA diminishes, and sleep becomes shallower. Likewise, taking a nap reduces nighttime slow waves.

A number of consumer devices now on the market play regular soft tones through headphones at the same frequency as the SWA to entrain deep-sleep waves and thereby induce a more restful power sleep.

One Hemisphere on Watch

Until recently, deep sleep in humans was thought to be a global condition: a person is either asleep or awake but not both simultaneously. Put differently, their brain is either in deep sleep, as characterized by slow-wave activity, or awake, but not both. Yet birds and aquatic mammals such as dolphins and whales display the remarkable phenomenon of unihemispheric slow-wave sleep: one half of their brain is awake, including an open eye, and the other half shows the electrical signatures of sleep. This is most likely a protective mechanism, enabling the animal to fly or swim and monitor its environment for threats with one hemisphere while the other gets some rest.

It now turns out that even for people, there is more to sleep than meets the (shut) eye. Frequent travelers will be familiar with the first-night effect, the observation that the initial night in an unfamiliar place, whether a hotel, a friend's apartment or a tent, is less restful than subsequent nights. We find it more difficult to shut our mind down and wake up groggy. A team of researchers under Yuka Sasaki and Takeo Watanabe of Brown University set out to investigate this phenomenon [see “Why We Toss and Turn in an Unfamiliar Bed.”]

Eleven healthy volunteers slept for two nights inside an advanced neuroimaging scanner that allowed recordings of the brain's weak but ever changing magnetic field. The scientists focused on slow-wave activity, measuring its strength in four networks in the left cortical hemisphere and four in the right. Intriguingly, they found that the left cortical default-mode network, a set of interacting regions associated with mind wandering and daydreaming, had less SWA than the right one during the first night. No such imbalance emerged during a second night sleeping in the scanner. Also, the more asymmetric the pattern of SWA during the first night, the longer it took subjects to fall asleep. Part of the left hemisphere, in essence, is not sleeping as deeply as the right one during the first night.

To test the extent to which the left hemisphere is more vigilant in an unfamiliar environment, the team delivered tones via earphones to 13 subjects (different from the initial 11 volunteers, who were by now used to the sleep setup). Most of the tones were the same, but on rare occasions a different one sounded: beep, beep, beep, boop, beep, beep. The oddball tone drew attention and triggered a signature electrical response. When the deviant tone is played to the left ear, its output is predominantly relayed to the right cortical hemisphere, which shows the characteristic vigilance response.

During the first night, the left hemisphere had a more pronounced vigilance response to these deviant tones as compared with the right hemisphere. The enhanced vigilance response also led to the left brain being more frequently aroused (as defined by EEG criteria) than the right hemisphere.


During the second session in the scanner, both left and right hemispheres responded weakly and in the same manner to oddball sounds, as they did to the stereotyped beeps during both nights. If a brain network in the left cortical hemisphere acts as a night watchman for the sleeper, then an irregular event registered only by the left brain (via the right ear) should elicit a faster response than an oddball sound delivered to the right brain (via the left ear). This idea was tested in a third group of 11 volunteers: they had to lightly tap their fingers whenever they heard the sounds while asleep in the scanner. (I know, it doesn't sound like the most restful way to sleep; they were also not permitted to drink caffeine or alcohol or to take a daytime nap.) Sounds delivered to the ear projecting into the left hemisphere were much more likely to trigger an awakening during the first night than sounds to the opposite ear and hemisphere. This left-right asymmetry disappeared during the second night's sleep. Furthermore, it took the left brain less time to awaken in response to the deviant sound than the right brain.

In short, while sleeping in an unfamiliar place, the left cortical hemisphere is more vigilant and responds stronger and faster than the right one. Evolutionarily, this reaction makes a great deal of sense. It is important that a sentinel—here the left cortical default-mode network—monitors the unknown environment for threatening events while we sleep. The human brain, it turns out, is endowed with a less dramatic form of the unihemispheric sleep found in birds and some mammals. For humans, familiarity with a place breeds a deep night's sleep.

If we consider the individual we routinely share a bed with—whether spouse, partner or child—to be the most important social component of the environment, then I suspect that the left hemisphere might also be more watchful during the first night we sleep alone in our familiar bedroom. It knows something is amiss, and we'll sleep less restfully as a consequence.

In my next column, I will discuss another recent discovery: how deep sleep can intrude into our waking brain.