There was nothing outwardly unusual about the man who showed up at the Minnesota Regional Sleep Disorders Center on June 27, 2005. Like thousands of other clinic patients, Benjamin Adoyo (not his real name) was a sleepwalker. A 26-year-old college student, originally from Kenya, Adoyo had been wandering at night since childhood. Lately, though, the behavior had been getting worse. Adoyo had gotten married in February, and his wife would wake to him shaking her while looming over their bed and babbling unintelligibly. Scared, she would simply do her best to rouse Adoyo, who, once awakened, never remembered a thing. They lived in a one-bedroom apartment in Plymouth, a suburb of Minneapolis, and the sleepwalking was straining their young marriage. The referral form from Adoyo's primary care doctor noted that the patient's wife was “sometimes startled by his behavior, but no injury, per se.”

After evaluating Adoyo, the sleep center's clinicians directed him to return on August 10 for an overnight electroencephalography (EEG) study of the electrical waves generated by his brain during sleep. In the middle of the night, Adoyo began thrashing about and yanking at the wires connected to the electrodes, pulling out tufts of hair as he ripped them off. But he did not wake up. The next morning Michel Cramer Bornemann, director of the center, told Adoyo that the study supported a diagnosis of a sleep disorder known as a non-REM parasomnia. Recounting when Adoyo ripped off the sensors, Bornemann asked, “Do you recall feeling any pain or pulling?”

“Nope,” Adoyo replied without any hesitation.

Adoyo's next visit to the sleep center was on October 17. He said that the antianxiety medication Bornemann had prescribed to treat the sleepwalking was not helping much, so Bornemann bumped the dosage from one milligram to two. The doctor sincerely hoped that he could help his patient. “He was the nicest guy—friendly, engaging,” Bornemann recalls. “I had no premonition at all that there was a malignant bone in his body.”

Adoyo never came back. The sleep clinicians found out why several months later, when they received a letter from the Minnesota Public Defender's office informing them that on October 19, only two days after the last clinic visit, Adoyo had been arrested for killing his wife and was now charged with the crime. “We are looking for someone to consult with regarding any relationship this sleep disorder may have with his offence,” the letter stated.

Perchance to Dream

The most basic and seemingly indisputable fact about sleep is that you are either asleep or awake. Sure, scientists subdivide the unconscious state into rapid eye movement (REM) and non-REM (NREM) cycles, and the latter is further divided into three substages. Overall, however, for most of the century-plus that scientists have been studying human rest, they have supported the notion that sleep and wakefulness are two distinct states with well-defined limits.

These supposedly firm boundaries are why judges and juries are skeptical when a sleep disorder is presented as an explanation for a crime such as Adoyo's. “I was asleep when I did it” sounds like a classic Twinkie defense, one in which science is twisted to abrogate personal responsibility. How, after all, could a person be anything other than totally awake if he is able to molest, injure or kill someone else? In the past two decades, though, sleep science has been revolutionized by a new theory that helps to explain everything from sleep crimes to the fundamental nature of sleep itself. As Bornemann puts it, “Sleep-or-wake is not an all-or-none, black-and-white phenomenon. It occurs on a spectrum.”

The idea that a person could be physically animate but mentally off-line is well established in popular culture—recall the sleepwalking of Shakespeare's Lady Macbeth—and in courtrooms. The first time in American legal history that sleepwalking was successfully used as a defense against murder was in the 1846 trial of Albert Jackson Tirrell, who killed a prostitute by nearly beheading her with a razor. More recently, in Toronto in 1987, a 23-year-old man named Kenneth Parks drove 14 miles and murdered his mother-in-law, allegedly all while sleepwalking unaware. He was subsequently acquitted.

Sleep murders grab headlines but are thankfully rare; a 2010 review in the neuroscience journal Brain listed 21 sample cases, with the defendant being acquitted about a third of the time. Nonlethally violent, sexual and otherwise illegal behaviors during sleep, however, are more common than the public might suspect. Some 40 million Americans suffer from sleep disorders, and a telephone survey in the U.S. from the late 1990s estimated that two people in 100 have injured themselves or others while sleeping.

Bornemann, along with his colleagues Mark Mahowald and Carlos Schenck, is among the world's preeminent experts on parasomnias—the umbrella term for unwanted sleep behaviors—and they frequently receive lawyers' requests for help. To distinguish between their medical and legal work, the doctors launched a separate entity in 2006 with Bornemann at the helm and Mahowald and Schenck serving as consultants. They call themselves Sleep Forensics Associates.

Sleep Forensics operates as a kind of scientific detective agency. Its more than 250 cases so far have been divided equally between work for the prosecution and the defense. Regardless of who is paying the fee, the agency's approach is not simply to serve up a medical opinion that supports a desired verdict. Instead the doctors try to discover the truth. The title Bornemann gave himself is “lead investigator,” and he says that “in many ways, what I am is a neuroscientific profiler.”

The outcomes of investigations are unpredictable. “If I can refute a parasomnia defense, the prosecuting attorney can say, ‘Now I have the potential for a conviction,’” Bornemann says. But his work also offers the possibility of absolution. “True parasomnia behaviors are done without awareness, intent or motivation,” Bornemann says. “Therefore, from a defense attorney's perspective, you have the grounds for a complete acquittal.” He knows, however, that judges and juries struggle to accept the idea of sleep existing on a spectrum. In the courtroom, then, it is not just the accused who is on trial but the very definition of consciousness itself.

Awake and Unaware?

the essence of what is known as local sleep theory is obvious from the name: parts of the brain can be asleep while others are awake. If true, the theory helps to account for people driving less safely when they are tired and for somnambulists scarfing pints of Chunky Monkey ice cream. It also explains “sexsomniacs” who fondle their partners or their children while unconscious. The concept of local sleep was first articulated neuroscientifically in a 1993 paper co-authored by James Krueger, who is currently at Washington State University, Spokane. At the time, the idea was heretical among senior sleep researchers. “It still is heretical,” Krueger says, although the localists now form a significant and well-regarded subset of sleep scientists throughout the world.

Conventionally, sleep has been understood as a whole-brain phenomenon and, what is more, a state that is controlled top-down by regulatory circuits. But this view has never made much sense to Krueger. He points out that scientists already have real-world evidence for partial brain sleep among other mammals. Dolphins, for instance, snooze with half of their brain at a time, swimming with one eye open. Krueger has also reviewed the scientific literature on brain lesions in humans and found that no matter what part—or how much—of the brain is damaged or missing, people are always able to sleep. This argues against the presence of a centralized sleep command center in the brain.

In a 2011 paper entitled “Local Use-Dependent Sleep,” Krueger summarizes the alternative view—that of a scattered, bottom-up process. “The new paradigm views sleep as an emergent property of the collective output of smaller functional units within the brain,” he wrote. Krueger and other like-minded researchers suspect that individual parts of the brain—neural networks and perhaps even individual neurons—go to sleep at different times around the clock depending on how much they have been taxed recently. (This is why the researchers describe sleep as both local—affecting only distinct parts of the brain at different times—and use-dependent—occurring only after the region has been sufficiently taxed.) Only when most of the brain's neurons are in the sleep condition does sleep's characteristic behavioral state kick in—that is, stillness, closed eyes, slackened muscles. Well before that point, though, tiny chunks of the brain are effectively taking a snooze.

Some of the most direct evidence for the theory has come out of the lab of David Rector, a colleague of Krueger's at W.S.U., Pullman. Rector works with rats, twitching their whiskers in a precise, controlled fashion. Each whisker is associated with a particular cortical column, a group of hundreds of tightly interconnected neurons that are located at the surface, or cortex, of the brain. He inserts probes through the rats' skulls into these cortical columns and can thus measure their electrical responses to the whisker twitching.

First, Rector established what the electrical response to whisker twitching looked like when the whole animal was behaviorally awake and when it was behaviorally asleep. Then he uncovered exciting exceptions to the rule. “The findings that columns can exist in a sleeplike state during whole-animal wake episodes and, conversely, that columns can exist in awakelike states during whole-animal sleep suggest that sleep is a property of individual cortical columns,” he and Krueger reported in a 2008 paper.

Human lab subjects, needless to say, do not like having metal probes inserted into their brain, so researchers have devised less direct experimental gauges. In work by Hans Van Dongen, another scientist at W.S.U., Spokane, subjects look at a computer screen and must press a button as soon as a reaction-time counter pops up. The subjects are directed to perform this action repeatedly for 10 minutes, and their response times slow as the task progresses. Vigilance tests like this one repeatedly tax the same neural pathways, and the excessive use during the experiment essentially forces them into a sleep state, Van Dongen says. He sees this as evidence of local sleep rather than more globalized fatigue or boredom because the performance of his subjects improves immediately when they are allowed to switch to a different task that calls on another area of the brain.

If people can be partially asleep while otherwise outwardly awake, then you also have to consider the opposite proposition—that they can be partially awake while behaviorally asleep. This possibility would help explain something that has long puzzled sleep scientists: insomniacs who report after a night of monitoring in the lab that they “didn't sleep a wink,” even though EEG measurements clearly show the brain-wave patterns that are characteristic of sleep. Looking for an explanation for this conflict, Daniel Buysse of the University of Pittsburgh's Sleep Medicine Institute performed a series of brain-imaging studies on insomniacs at night. He concluded that while the subjects were asleep as gauged by EEG and, for that matter, by behavioral observation, the parietal cortices of their brain—where the perception of alertness is formulated—remained active overnight. In that sense, the insomniacs' reports that they were awake were quite true.

Following the Clues

“what's going on?” the 911 operator asks.

“You just get here,” the man on the other end of the line replies tersely.

“You need to tell me what's going on,” the operator insists.

“Somebody is dead,” the man says.

“Somebody is dead?

“Yes.”

“Where are they at?”

“In their house. Somebody is dead. Get here.”

The call, which was received by the Hennepin County's emergency communication center at 3:41 a.m. on October 19, 2005, had been placed by Benjamin Adoyo. He used the cell phone of his wife, who at that moment was lying on the bathroom floor in a pool of her own blood.

When news of the killing reached Sleep Forensics Associates via Adoyo's defense attorney, Bornemann set out to understand both the alleged criminal and the crime. After being briefed by the attorney, he read the police reports and the transcripts of Adoyo's interrogation in the predawn hours after the homicide. He even visited the apartment and had a computer-animated video made to help him reconstruct the events leading up to the murder.

The peculiar syntax of the 911 call was one of the first things to catch Bornemann's attention. Adoyo did not say, “My wife is dead,” Bornemann noted, but rather, “Somebody is dead.” He did not say, “In our house,” but rather, “In their house.” In other words, Adoyo sounds like someone who does not know who he is, who the dead woman is or what has happened. He sounds like someone who is just waking up.

There are alternative interpretations, of course. Maybe Adoyo was knowingly guilty and wanted to reveal as little information as possible when he called 911. But when Bornemann read through the police reports, he did not see evidence of concealment or evasion. When officers from the Plymouth Police Department arrived on the scene, Adoyo was waiting for them on the front steps. At the police station, after he had been read his Miranda rights, Adoyo readily confessed to attacking his wife, although he seemed hazy about the details. “How is she?” he asked an officer at one point in the interrogation.

These initial findings—the detachment of the 911 call, the lack of concealment, the partial amnesia—all suggested to Bornemann that it was at least possible that Adoyo had been sleepwalking when he killed his wife. But a judge or jury would question the science behind this explanation before ever considering an acquittal. Could somebody really kill unknowingly while asleep and, if so, how?

To answer that question, first consider how sleep works for people without parasomnias. The shifts between wakefulness and REM and NREM sleep states are established by “literally hundreds” of hormonal, neural, sensory, muscular and other physiological variables, Bornemann's colleague Mahowald remarks. “Amazingly, these variables usually cycle together, and you've got billions of people in the world all cycling through wake, REM and NREM states multiple times every 24 hours.” Sure, there will be pockets of “awake” neural networks when the rest of the brain is asleep, and vice versa—that is what local sleep theory tells us—but overall, the transitions are clear.

In people with parasomnias, though, the myriad regulatory variables become out of sync, and the switching between awake and sleep states gets mixed up. The result, Mahowald says, is what amounts to an extreme form of the local sleep phenomenon, a condition known as state dissociation, in which the physical and mental attributes of alertness, deep sleep and dreaming overlap. Afflicted people effectively suffer from having significant parts of their brain off-line even when their body is active.

Many Sleep Forensics cases illustrate how state dissociation can lead to criminal behavior. In late April of this year, for example, Bornemann was investigating a sleeping U.S. soldier who, when his wife attempted to rouse him, savagely pistol-whipped her. After the fact, he claimed that he had no intention of attacking her, nor did he have any memory of doing so. What he does remember is dreaming about using a knife to fend off an attacking Nazi spy. To Bornemann, this sounds like a possible example of REM behavior disorder, in which the afflicted person lacks the slackened muscles that normally accompany dreaming and is able to get up and physically act out the fantastical scenarios running through his head.

Another case Bornemann was investigating in late April concerned a well-off businessman in Utah. The businessman was asleep one night when his nine-year-old daughter slipped into his bed, which she apparently did when she was having trouble sleeping. The father awoke later and discovered, to his horror, that he was pelvic thrusting against his daughter and that his hand was touching her genitals.

The businessman had no past record of sex crimes. After the incident, he was evaluated by a psychologist, took a polygraph test and even had his penile tumescence measured while being shown inappropriate images of children. None of these measures indicated that he was a pedophile. Bornemann suspects that his behavior may instead be caused by an arousal disorder, the subcategory of state dissociations that includes sleepwalking, sleep eating and sexsomnia. What all of them have in common is that they arise when the neurophysiological attributes of NREM sleep overlap with the complex motor abilities of wakefulness.

Knowing just which parts of the brain are working and which are slumbering helps to explain the perversity and violence that parasomniacs sometimes exhibit. Brain-imaging studies reveal that during NREM sleep, the prefrontal cortex—a section of the brain located just behind the forehead, where reason and moral judgments are formulated—is much less active than it is when an individual is awake. The midbrain, meanwhile, is active and capable of generating simple behaviors known as fixed-action patterns. “These tend to be very primal in nature,” Bornemann says. “You can have standing, walking, predatory attack, eating and drinking, grooming, and sexual and maternal behaviors.” The prefrontal cortex normally checks such patterns when they are inappropriate, but during NREM sleep, this part of the brain is no longer on the job. People become more like wild animals, governed by instinctive urges and impulsive reactions.

The Verdict

The crux of a Sleep Forensics investigation is when Bornemann interviews the accused. Face-to-face is best. The two questions he must answer is whether the person legitimately has a sleep disorder and, incorporating all of the other evidence, whether that disorder might have been active at the time of the criminal act.

With Adoyo, Bornemann was in the highly unusual position of having treated the accused as a patient, so he knew that the young man was not a faker. Family members also vouched for the fact that Adoyo had been a sleepwalker since he was a boy. The second question, though, was tougher: Was Adoyo's sleep disorder the reason he committed the crime? That query could not be answered with total certainty because Bornemann could not travel back in time and enter Adoyo's mind to see what he was or was not thinking during the criminal act. That said, it is not easy to fake a sleepwalking defense. “The general public has the impression that anything can happen during sleepwalking,” Bornemann says. “But only certain behaviors can occur and, in general, for a limited amount of time.”

For instance, “proximity is the key in the vast majority of sleepwalking violence,” Bornemann says. The victims are often lying next to the parasomniac or are attacked when they attempt to rouse the sleeper. The latter was the case with the soldier who was dreaming about the Nazi spy as well as with the sleep-driving Parks, who attacked his family only after they tried to wake him. Sleep crimes are also usually inexplicable—motiveless and out of character, such as in the case of the Utah businessman who fondled his daughter.

During the Adoyo investigation, Bornemann learned that his former patient had not in fact been physically proximate to his wife before the attack; he had nodded off on the couch while she was asleep in the bedroom. Also, the violent outburst was not brief and random, as sleep-based ones typically are, he notes, and instead was prolonged and “procedural,” meaning several complex behaviors were involved. Adoyo first entered his wife's bedroom and assaulted her with a hammer; then he chased her into the hall outside the apartment and back inside to the bathroom; and he finally stabbed and strangled her. “It is highly unusual to see three mechanisms of sleepwalking violence” at once, Bornemann says.

Any remaining doubts were erased by Adoyo's admission—and the account that Bornemann read in the dead woman's own diary—that the couple had been fighting on the last day of her life. Adoyo suspected that his wife was having an affair and confronted her with what he believed to be evidence—condoms in her laundry—before she stalked off to bed. The crime, in short, was out of character but not motiveless, and Bornemann reported this and all of his other findings to the public defender. Adoyo ultimately pled guilty to second-degree murder, and he is currently serving a 37-year prison sentence.

Bornemann, for his part, says he is not personally invested in the guilt or innocence of the people he investigates. To him, what his work offers is the opportunity to do behavioral research on extreme sleep disorders that could never be replicated in the lab. The goal is to gather enough evidence to help shift the attitudes of jurors, judges and the general public, for whom on/off notions of consciousness still hold sway. “Neuroscience has moved far ahead of the paradigms of the legal community,” he says, “and the legal community needs to catch up.”