Sleep—the inscrutable state that has dimmed consciousness for millions of years—is finally trending. Social media ads hawk wearables that track circadian rhythms. Mattress start-ups pledge immaculate rest. Supplements put us under with hormones and exotic herbs. Sleep-hacking sites extol blue-light-blocking glasses, blackout curtains and reserving the bedroom as a sanctuary for repose. After decades of being revved into hyperproductivity, we lie anxiously in bed, so cognizant of sleep’s rewards that we’re afraid of missing out.
Sleep’s prophet, were it to have one, might be William Dement, who, in 1970, founded the Stanford Sleep Disorders Clinic (now the Sleep Medicine Center), the world’s first medical facility of its sort. In 1971, he began teaching Sleep and Dreams, which went on to become one of the most popular courses in Stanford’s history. Over nearly half a century, the professor of psychiatry and behavioral sciences warned about the dangers of sleep debt not only for brain health but also for safety on the highways, in the skies and on the high seas. He educated more than 21,000 students, sending ranks of sleep experts into industry and academia while making the university a hub and place of pilgrimage for those intent on the mysteries of Nod.
Five years ago, Dement began priming his Sleep and Dreams successor: Rafael Pelayo, a clinical professor in the psychiatry department’s division of sleep medicine. Pelayo—who, in 1993, as a medical student in the Bronx, found his passion for sleep research upon reading about Dement in National Geographic—took over Sleep and Dreams three years ago. But Dement, at 91, still attends, with a wireless mic clipped to his lapel so he can interject with songs (he is a trained musician) and stories (for instance, about working with the scientists who discovered REM—the sleep phase with rapid eye movement and dreams—in the 1950s).
To get a sense of Dement’s legacy in sleep research, one need only browse the roster of guest lecturers in Sleep and Dreams. Take Cheri Mah, ’06, MS ’07, who, as an undergraduate, showed how longer sleep duration is associated with higher scoring in basketball games. She developed a formula to predict NBA wins on the basis of fatigue, factoring in travel, recovery time, and the locations and frequency of games. Its predictive power was so great that the NBA adjusted its schedule to allow players more rest. Or there’s Mark Rosekind, ’77, the first sleep expert appointed to the National Transportation Safety Board and later the 15th administrator of the National Highway Traffic Safety Administration. Back when he was a teaching assistant in Sleep and Dreams, Rosekind joined a waterbed study conducted by Dement in which Rosekind’s future wife, Debra Babcock, ’76, also participated. “The way Mark has told it publicly,” Pelayo recalls, “that’s how they met, during a waterbed research study. That was the ’70s.”
Having spent those decades railing against people who bragged about skimping on sleep, Dement is now being vindicated by a host of new, rapidly evolving technologies. Millions of people wear sleep trackers whose data is processed by machine learning. Millions of sequenced genomes give insights into how humans are programmed to sleep. Scientists better understand sleep’s complex relationship with physical and mental health. And pop culture has been quick to respond. Clickbait features the sleep habits of famous CEOs: Elon Musk snoozes from
1 a.m. to 7 a.m.; Bill Gates is tucked in by midnight. The rested, productive brain is the new flexed biceps.
Here we look at a number of the shadowy domains on which the current generation of sleep scientists are shining their lights. Some of their studies read like the premises to sci-fi films, while others are sufficiently disturbing to keep you up at night.
The Genes of Nightmares
Hanna Ollila, a visiting instructor in psychiatry and behavioral sciences, became interested in sleep during her high school years in Finland, when she and her friends were discussing why people sleep. Five years later, she started a PhD in sleep science. She partnered with a fellow graduate student—appropriately named Nils Sandman—to research nightmares, clinically defined as negative dreams that cause the dreamer to wake up. In a 2013 study, they found that veterans of war, especially those who saw combat, were comparatively likely to have nightmares, and that the nightmares correlated with insomnia, depression and anxiety.
Post-traumatic nightmares made sense, but Ollila became increasingly curious about idiopathic nightmares—those without a known cause. Although nightmares were rare in the population at large, previous studies had shown that if one twin had them, the other often did as well. Ollila wondered whether idiopathic nightmares had a genetic basis.
Often viewed as profoundly personal, dreams and nightmares have eluded scientific explanation for decades. “When people think about dreaming,” Ollila says, “they think about Freud. It’s not very serious science. We wanted to do a study that would give us scientific proof that nightmares are actually important and dreaming is important. Genetics is a nice way to do that because the genes don’t change during your lifetime.”
Ollila and her team conducted a genome-wide association study in which 28,596 people were given sleep questionnaires and had their genomes analyzed. The results: Two genetic variants were more common in people who reported nightmares. The first variant is located near PTPRJ, a gene correlated with sleep duration, and the second is near MYOF, which codes for a protein highly expressed in the brain and bladder.
Untangling causality in genetics is tricky, and in this case, deciphering the results is particularly challenging, since the variants are in unexpressed regions of the DNA: those that don’t code for traits but could affect the regulation or splicing of many nearby genes. So it’s not yet clear what, if anything, the two closest genes have to do with nightmares per se. Given that people are most likely to recall the dreams in which they wake up, those with the variants might not have more nightmares. They may simply wake up more often, either because PTPRJ affects sleep duration or because MYOF results in nighttime trips to the bathroom. Or the variants could have far different and possibly more complex relationships with nightmares.
Sleep, the New Health Biomarker
While Dement’s efforts to educate people about the dangers of sleep debt have borne fruit—with industry having adjusted safety standards to allow periods of rest—new studies suggest that sleep influences health in more ways than previously imagined. A growing body of research reveals that individuals are programmed to sleep differently. Some are refreshed after a mere six hours, whereas others need nine. And a recent study in which Ollila participated found 42 genetic variants associated with daytime sleepiness. For individuals and employers, knowledge of sleep genes could avert automobile or work accidents while leading to greater happiness and productivity.
Genes associated with sleep are also involved in other biological processes and have a role in health problems. “Sleep is kind of a central anchor that connects a lot of different types of diseases,” says Nasa Sinnott-Armstrong, a PhD student in genetics who works with Ollila. Genes implicated in sleep are linked to cardiac, metabolic and autoimmune diseases as well as obesity, type 2 diabetes, schizophrenia, bipolar disorder and depression. Such health problems, Sinnott-Armstrong says, appear to evolve in tandem with sleep disruption: “I think of sleep as a really good self-reported biomarker for many important diseases.”
The broader research on sleep and health dovetails with that on nightmares, since, as assistant professor of psychiatry and behavioral sciences Rebecca Bernert has shown, frequent nightmares correlate with psychiatric disorders and suicidal behavior, even in the absence of depression. The question then, asks Ollila, is whether managing sleep according to our genetics could have mental-health benefits. “If you treat the sleep component efficiently,” she says, “it may have an impact on the psychiatric disorder.”
Not Letting Sleeping Dogs Lie
In 1974, Dement brought a French poodle named Monique to Stanford. The dog had narcolepsy, a condition that affects 1 out of every 2,000 people, causing them to fall asleep repeatedly over the course of each day. Its symptoms range from collapsing during moments of excitement to hallucinating as if dreaming to plunging into REM sleep. Narcolepsy presents constant dangers, whether a person is driving, cooking, carrying a child or going for a dip in the ocean.
By 1976, Dement had established a colony of narcoleptic dogs, and in the 1980s he founded the Stanford Center for Narcolepsy. Emmanuel Mignot, a French sleep researcher, arrived in 1986 to study the dogs, and in 1999 he discovered narcolepsy’s cause: a lack of hypocretin—a signaling molecule that controls wakefulness and is produced in part of the hypothalamus, a small area in the brain that regulates processes such as circadian rhythms, body temperature and appetite. The area that produces hypocretin contains only 70,000 of the brain’s 86 billion neurons, and in narcoleptics, they have been decimated. The culprit: certain strains of the influenza virus, especially H1N1. Receptors on the virus resemble those on the neurons. White blood cells targeting the flu inadvertently destroy the neurons as well, causing lifelong narcolepsy. “It’s an autoimmune disease that’s triggered by the flu,” says Mignot.
A professor of psychiatry and behavioral sciences and director of the narcolepsy center, Mignot is now using large genetic databases to evaluate whether certain people are more vulnerable to having their hypocretin-producing neurons destroyed. He and his collaborators are also researching and testing possible treatments. “It’s very exciting,” Mignot says, “because new drugs based on this hypocretin pathway are coming now on the market.”
As for Stanford’s narcoleptic dogs, the last one died in 2014. By then, the colony had long since closed and the remaining dog—named Bear—was living with Mignot and his wife. But the next year, a dog breeder contacted Mignot and asked if he wanted a narcoleptic Chihuahua puppy. Today, when Mignot guest lectures for Sleep and Dreams, he brings Watson, the Chihuahua. “Any student anywhere in the country can learn about sleep,” Rafael Pelayo says, “but only here at Stanford can they actually hold a narcoleptic dog in their arms as they are learning about it.”
Dream On
As a teenager, Jonathan Berent, ’95—another guest lecturer in Sleep and Dreams—read about lucid dreaming and, following the instructions in a book, taught himself to remain aware in his dreams and even, to some extent, to control them. At night, he could fly or explore fabulous landscapes, real or imagined. “It really does feel like a superpower,” he says.
At Stanford, Berent read the work of Stephen LaBerge, PhD ’80, who researched lucid dreaming. Berent contacted him and, with his mentorship, wrote a paper exploring lucid dreaming’s potential to shed light on the nature of consciousness. After completing a degree in philosophy and religious studies, Berent went into the tech industry; he now works at Alphabet, Google’s parent company. Independently, he continued to research lucid dreaming, and in 2015 he collaborated with Ken Paller, director of cognitive neuroscience at Northwestern, to develop a smart sleep mask that helps people choreograph their dreams. The prototype uses subtle light pulses to make sleepers aware that they are dreaming. It also gives them sound cues using targeted memory reactivation, a technique in which selected activities are paired with tones during the day. When sleepers hear the tone, they recall the associated activity: visiting a place, meeting a person or working out a practical challenge during sleep.
Better yet: The dreamer may be able to answer. During REM sleep, the brain shuts off the neurons that control virtually all muscles, paralyzing the body. Only the eyes can move. In the 1980s, LaBerge proposed that bidirectional communication during sleep was possible by lucid dreamers who learn to control their eyes; if information were transmitted to them, they could reply with eye movements. Berent envisions using a series of coded eye motions that a mask transmits to a computer for interpretation. He contemplates scenarios in which a scientist connects with dreamers. “Can you ask a specific question,” he says, giving the example of a simple arithmetic problem, “and can the person stay asleep, do the math and respond?”
For Berent, harnessing the power of the unconscious is the ultimate goal, but the mask may have more commercial uses: It can be synced with virtual reality headsets, so that the dreamer can be cued to pick up where he left off in VR, gaming from dusk till dawn. (Berent’s team also created an app, Lucid Reality, that trains people to lucid dream without the mask.)
But if pinging the subconscious for inspiration or engaging in video game firefights during REM sounds less than restful, Berent agrees. Despite the energizing effects of lucid dreaming, he feels slightly less refreshed the next morning. When he was most actively exploring lucid dreams, he says, “I did it as many times as I felt like I wanted to, and that ended up being two times a week. I needed those other nights off.”
The Sleep Revolution
The challenge in studying sleep and dreaming has been in connecting them with the biological processes that underpin them. Until now. “The field of sleep is at the confluence of three revolutions,” says Mignot. “One is genetics and proteomics—the biological revolution. The second is the hardware revolution. And then there’s a data revolution.”
In genetics, the complex science behind polygenetic traits—those with origins in multiple genes—requires finding correlations between millions of variants and traits. Doing so is the focus of the computational genetics lab of biology and genetics professor Jonathan Pritchard (with whom Ollila and Sinnott-Armstrong work on sleep, health and nightmares). Pritchard explains that genetic correlation with traits has historically been found by looking at similarities among family members. Now, researchers deploy computational techniques and machine learning to examine vast numbers of genomes, determining whether individuals with similar characteristics have similar genotypes. “The first genome-wide association studies were around 2006. Those were very small. The largest study then was 20,000 people,” Pritchard says. “The technology has moved really quickly. Now for a lot of phenotypes, we are getting up toward a million.”
With technological improvements and the decrease in the cost of gene sequencing from $100 million in 2001 to $1,000 today, Ollila points out, our understanding of genetic behavioral traits is rapidly growing. “Big data has revolutionized how we understand what modifies sleep at the genetic level,” she says. And some patient registries are combining genetic data with medical records, allowing researchers further insight. “UK Biobank revolutionized the study of common sleep features,” she says, referring to a genetic registry of half a million people, “as it had both a sufficient sample size and it had questionnaires of sleep traits.”
But whereas the genetics revolution has been well publicized, proteomics—the study of proteins—hasn’t received much media attention. Yet every drop of blood contains thousands of proteins, each one coded by a different gene before being modified within cells and secreted into our blood. Through proteomics, scientists can see the expression level of genes and learn how protein modifications reflect our health.
Mignot can now send blood to SomaLogic, a company based in Boulder, Colo., whose technology can measure as many as 5,500 different proteins in each 100-microliter sample—one-tenth of a milliliter. “It’s a little like [nuclear] fusion,” Mignot says of proteomics. “For 20 years, we’ve been saying that it’s going to be the next revolution, and we’re still waiting. However, I think it’s really happening now.” Proteins can be used to identify physiological states and health problems long before other diagnostic tools. For instance, they may reveal precisely what is happening with someone’s circadian rhythms. “If you travel from Tokyo,” Mignot says, “I can take a blood sample and see that you’re still on Tokyo time.”
Until now, much sleep data has been based on self-reported descriptions—whether study subjects snore or move about—but Mignot’s goal is to shift away from this model. He explains that many people who complain of insomnia think that they are asleep for only an hour or two but are actually asleep for five or six hours. Since perception of one’s own sleep is often foggy, he wants the hard data—the knowledge of which brain waves correlate with which proteins in the context of specific genotypes.
Alongside the revolution in hardware—wearable devices that track brain waves and sleep stages—the data revolution has transformed how much information can be processed. Now, in a sleep clinic, computer algorithms can analyze brain waves and draw correlations between sleep patterns and thousands of proteins and genes. Eventually, once data has been compiled from a sufficient number of patients, the information from gene sequencing and a drop of blood might suffice to diagnose a sleep disorder.
“There are hundreds of companies—literally hundreds—that are trying to measure sleep as a vital sign for health,” Mignot says. He himself designed a trial that would survey 30,000 people—a large enough number to derive accurate patterns from the data—using proteomics, genetics, hardware and computer analysis. An initial request for funding did not come through, but he is undeterred. “I am very persistent,” he says. “Sleep is going to be one of the next waves. It is ready for an explosion of understanding because of all of this new technology.”
If sleep research continues apace, the decades ahead promise advances in the understanding of how our biology regulates sleep and in the gadgets helping us to do so. Just as artificial light has thwarted our internal clocks, devices will attempt to reset them by managing sleep patterns and light exposure according to our genetics. As we forge ahead into one scientific revolution after another, our devices may steer us away from health problems by holding us to biological rhythms from the dawn of our evolution—while also allowing us to live our wildest dreams every time we close our eyes.
Deni Ellis Béchard was a senior writer at Stanford. Email him at stanford.magazine@stanford.edu.
