On the morning of May 24, 1987, sometime after 1:30 A.M., a 23-year-old Canadian named Kenneth Parks drove 14 miles to his in-laws' home, strangled his father-in-law to the point of unconsciousness, and beat and stabbed his mother-in-law to death. A year later he was acquitted of both assault and murder. After a careful investigation, specialists reached the astonishing conclusion that Parks had been sleepwalking—and sleep driving and sleep attacking—during the incident.

This story inspired a 1997 made-for-television movie, The Sleepwalker Killing, starring Hilary Swank as Parks's wife. Although such extreme cases are rare, unintended acts of violence during sleep are quite common among those with sleep disorders. In a 1995 study of 64 sleep clinic patients suffering from sleepwalking or sleep terrors, more than half exhibited harmful behavior during sleep. An analysis at a different clinic that same year concluded that 70 percent of their 41 sleepwalking patients acted in a potentially injurious way.

Evidence from population surveys confirms that sleep violence is not a trivial threat. In a 2010 review of nearly 20,000 telephone interviews across six European countries, about 1.7 percent of the respondents reported behaving violently during sleep. Because this study is based on self-reports, it may be an overestimate. Nevertheless, the findings echoed an earlier survey, in which 2.1 percent reported acting in dangerous ways while slumbering.

Ultimately sleep violence is a symptom of an underlying condition. Scientists who study these behaviors, including the authors of this article, seek to identify its psychological and neurological determinants and to produce effective treatments. What makes these reports so alarming, however, is the total lack of self-control they imply. The ability to unwittingly carry out complex actions while asleep poses a serious challenge to our sense of being in charge. Using imaging techniques, we have learned that while certain important regions of a sleepwalker's brain behave as if the person is deeply asleep—such as the frontal lobe—others are unusually active, as if the person is wide awake. These emerging findings allow us not only to explore the subtle boundaries separating normal and pathological sleep but also to probe the mysteries of consciousness and free will.

Neither Awake nor Asleep

For as long as we have recognized walking and talking in our sleep, we have also been aware of more extreme nighttime behaviors. Homer's epics mention a sleeper's tragic suicide. In 1313 a church-led council concluded that a sleepwalking killer was not culpable for his crimes. One of the first legal cases involving sleep violence occurred in the central European region of Silesia in 1791, in which a woodcutter killed his wife with an ax and later insisted he was asleep at the time. We have no way of knowing the truth of those matters; nonetheless, the medical literature reflects many complex actions executed during sleep, including driving, eating and sex, as well as murder, suicide and rape. In fact, much of the evidence that scientists use to study extreme cases of sleep violence comes from criminal investigations and court cases.

Sleep violence tends to emerge from three main conditions: rapid eye movement (REM) sleep behavior disorder, arousal disorders and epilepsy. We will focus primarily on arousal disorders, which occur during non-REM sleep. In arousal disorders, a sleeper enters a so-called dissociative state, as though beginning but failing to completely awaken. The first brain-imaging study to observe this dissociative state was led by one of us (Bassetti) while at the University Hospital of Bern in 2000. A 16-year-old sleepwalker was monitored for two nights with electrodes placed on his scalp to produce a polysomnogram of his brain activity. On one of those nights, when the polysomnogram showed the teenager to be in deep sleep, he rose from his bed and opened his eyes, a scared expression on his face. Half a minute after he began sleepwalking, Bassetti's team injected him with a weak radioactive tracer. Several hours later the tracer would allow the researchers to produce scans of his brain activity at the time of sleepwalking.

We then compared the boy's brain activity when sleepwalking and when in deep sleep. In the sleepwalking state, scans revealed greater activity in areas of the brain involved in motor control, including the posterior cingulate cortex and parts of the cerebellum, located in the middle and at the base of the brain, respectively. Compared with the brain activity of healthy, awake subjects, the sleepwalker showed less engagement in regions responsible for higher cognitive functions, such as attention, insight, planning and judgment.

A similar pattern was found in 2009 by sleep specialist Michele Terzaghi and her colleagues at Niguarda Hospital in Milan, Italy. The researchers implanted electrodes under the cranium of a patient who suffered from both epilepsy and sleepwalking. During the study the subject sat up and spoke briefly while asleep. As in Bassetti's study, parts of the sleepwalker's posterior cingulate cortex, tucked into the middle of the brain, appeared as active as in an awake person, whereas other regions remained in a sleeplike state.

Rude Awakening

One of the important results from these studies is that during a sleepwalking episode, the brain's frontal lobe functioned as if in deep sleep. Among other things, the frontal lobe enables a person to understand and evaluate an action's consequences. Dysfunction in this area, seated directly behind the forehead, has been linked to violent behavior.

Low frontal lobe activity, however, does not fully explain sleep violence. Sleepwalking without incident is common in children, and for many adults the only injury comes from bumping into furniture. Mark Pressman, a doctor of sleep medicine at Thomas Jefferson University, investigated this question by analyzing 32 cases of nocturnal violence documented in the medical and legal literature. In 2007 he reported that most aggressive behavior may be provoked by encounters with other people while the sleeper is somnambulating.

Disturbing dreams can also accompany abnormal sleep behavior. A team under medical doctor Isabelle Arnulf of Piti-Salptrire Hospital in Paris interviewed 38 patients in the sleep disorder unit with questions about the content, frequency, time and activity of their sleep disorders. Sleepwalkers reported experiencing intense, nightmarish images. In the study, published in 2009, 84 percent of these images inspired fear and more than half were unhappy in content. About a quarter of individuals questioned had dreamed of being physically attacked.

Getting through the Night

Sleep is not an all-or-none phenomenon. At times, the boundaries between sleep and wakefulness are disrupted, and individuals become caught between these states. The sleepwalker who attacks a beloved family member, the narcoleptic who is conscious but suddenly rendered unable to move by a bout of laughter, and the lucid dreamer, perfectly aware of the fact that his or her experiences are not real, are all examples. Such cases of unusual sleep offer a window into consciousness. Not only does consciousness vanish when we doze off and reappear in full on waking, it can assume a variety of forms. It can range from brief images that flash by as sleep sets in to vivid hallucinatory experiences in dreams later in the night.

These observations inevitably raise difficult questions. What determines the level of consciousness during sleep and wakefulness? Which parts of the brain must be awake to carry out actions deliberately, with full knowledge of their consequences? How culpable is a person like Kenneth Parks for his behavior? Only further study of the brain and behavior, awake and asleep, will yield the answers.

So far this work underscores that sleep and wakefulness can coexist in the brain. Sleep can occupy certain populations of brain cells but not others. This observation has consequences for a healthy person's waking life as well. Think of the last time you had a poor night's sleep. There is a good chance that the next day, parts of your brain were off-line while the rest was humming along in a normal waking state. This is what one of us (Tononi) and colleagues showed in a breakthrough study published in 2011. In the brain of sleep-deprived, awake rats, isolated groups of neurons briefly ceased firing, a phenomenon that increased with the amount of sleep deprivation. Working with researchers at the University of California, Los Angeles, they also reported that same year that when humans sleep, some parts of the brain can be observed behaving as if they are already awake, especially toward the end of the night.

Because we can identify the brain regions involved in sleep disorders, these conditions provide an excellent case study for clarifying how the brain creates an integrated conscious experience. The discoveries being made in sleep violence may have moral, ethical and legal implications that society has barely begun to recognize.