How we long for a good night's sleep. More than 40 million Americans suffer from chronic sleep disorders, and this month, a National Sleep Foundation survey found that 60 percent of American women say they get enough sleep only a few nights a week (men weren't asked). Small wonder, then, that U.S. sales for sleeping pills hit $3.7 billion last year and are on the rise, according to IMS Health, a health-care information company. We are taunted by Lunesta's bioluminescent moth.* We are seduced by ads for the sleeping pill Rozerem that wistfully tell us, "Your dreams miss you." We have made sleep "the new bottled water," as the New York Times recently put it in an article about slumber salons that the tired pay to visit during the day. We may envy CEOs and coaches who fire up their treadmills at 5 a.m., but what we want for ourselves is an afternoon nap.
Why the obsession? And what, exactly, is sleep for? A little more than a half-century ago, most scientists believed that sleep was an inactive state, a kind of parenthesis in living. Then in 1951, Eugene Aserinsky, a clever graduate student at the University of Chicago, hooked his son Armond to a retooled "brain wave machine" and monitored the boy's sleep deep into the night. Aserinsky observed sharp spikes of activity on his readout, suggesting that Armond's eyes were darting back and forth. This turned out to reflect the distinctive state within sleep dubbed rapid eye movement, or REM— a "new continent in the brain," as a colleague later put it. Aserinsky and his adviser, Nathaniel Kleitman, hypothesized that rapid eye movement was related to dreaming. Soon Kleitman and others were mapping the basic cartography of sleep, which alternates between non-REM and REM periods throughout the night, in roughly 90-minute cycles.
Sleep is not an optional enterprise. All mammals do it. So do birds, reptiles, and even fruit flies. Rats deprived of sleep apparently die faster than those deprived of food. Sleep deprivation is a ruthlessly effective means of torture, as the new movie The Lives of Others shows in a stomach-turning scene. Yet the bedrock question—what purpose does sleep serve for us and the rest of the animal kingdom—remains oddly unsettled.
After sleeping on it, I went in search of some theories. I started with Robert Stickgold, a cognitive neuroscientist at Harvard, whose office is located in one of the quieter precincts of Beth Israel Deaconess Hospital, in what used to be a patient room. "There's an old joke that the function of sleep is to cure sleepiness," he says, smiling, white hair mussed and thick eyebrows animated. "We think we can deprive ourselves of sleep and drink triple espressos or take Modafinil. But that's like saying, 'I've figured out how to cure hunger in Africa: I'm going to send over amphetamines so people won't feel hungry anymore.' It doesn't address the underlying need."
Except that with sleep, as opposed to food, no one knows exactly what the underlying need is. Stickgold is one of the foremost sleep researchers in the country and has long argued that sleep's crucial function is to boost memory and learning. His theory is that during sleep, the brain evaluates recently learned information and decides what to do with it. In the process, memory consolidation takes place—memories or skills that were acquired during waking are stabilized or enhanced, or perhaps moved to new locations. The brain may also extract patterns and rules from large amounts of information.
Evidence that sleep plays a role in memory consolidation and mental processing comes largely from behavioral research in which scientists teach people a task and then try to pinpoint a boost in performance that they can attribute to sleep. Frequently, they try to correlate the gain with a particular portion of sleep, as well—REM versus non-REM. In an experiment published in 2000, for instance, Stickgold and his colleagues asked subjects to identify, as rapidly as possible, the orientation of diagonal bars against a background of horizontal ones. Some people returned for retesting later the same day and showed little improvement. Others returned after a night of sleep and did significantly better. The sleepers' improved scores were proportional to the amount of slow-wave sleep (a stage of non-REM) they had gotten early in the night and the amount of REM sleep they had gotten late in the night. The researchers also found that when subjects were deprived of sleep the night after learning the task, their performance did not significantly improve later, even after two subsequent nights of recovery sleep. And so they concluded that sleep "within 30 hours of training is absolutely required for improved performance."
Dozens of studies similarly show evidence of a link between a learning or memory task and some portion of sleep. (Here's an example.)
Sleep may facilitate more complex forms of insight as well. For a study reported in 2004, a German group asked subjects to work on a series of interrelated math problems, which contained a "hidden rule" that allowed them to be solved more quickly. People who returned to the puzzles after a night of sleep found the shortcut roughly twice as often as those who had spent an equal amount of time awake. What enabled the flash of insight, the researchers suggested, may have been the "restructuring of representations in memory" during sleep.
New work from Stickgold's lab also suggests that sleep helps people to extract themes or meaning from information they were presented with during the day. When shown lists of related words, for instance, subjects were better able to recall the unifying principle of the list—what the words had in common—after sleeping.
Stickgold is sure that sleep is crucial to "how our memories are initially stabilized and ultimately shaped," including how new memories are integrated with older ones. And this function may help to explain sleep's evolution. During sleep, animals cannot hunt for food or produce more offspring and may be more vulnerable to predators. So to have endured all these millennia, snoozing is likely to offer some adaptive advantage that outweighs its risks. But what? Stickgold suggests that the different stages of sleep—with their distinctive patterns of brain activity—were selected for because they help the brain to perform different kinds of memory tasks.
Yet questions remain. In some of the older work on sleep and memory, methodological issues make it hard to tell if gains in performance are due specifically to sleep (as opposed to, say, the passage of time). More recently, there has been a huge push to tackle which stages of sleep enhance memory—and which sorts of memory get a critical boost from sleep. When you sort through them, the distinctions that crop up can sometimes seem a little baroque. Why should learning to discriminate among visual patterns depend on slow-wave sleep early in the night and REM sleep late, as Stickgold has shown? Are there more thoroughgoing effects throughout the night that we have not yet picked up on? Also, apparent contradictions emerge with regard to particular sleep stages. For instance, as this review points out, one study found that depriving people of all sleep in the second half of the night (when REM predominates) impaired their improvement on a given task. But depriving them of REM sleep for the entire night did not have the same effect.
Maybe future research will resolve all of this. Or perhaps we'll learn that the mapping of specific memory tasks onto portions of sleep tells only a partial story. The effects of sleep on memory could turn out to be byproducts of something more basic—as the slumber of peculiar creatures like the platypus hints. More on that tomorrow.
In the past several years, researchers like Harvard's Robert Stickgold have made the case that sleep plays a critical role in boosting memory and learning. Many scientists believe that sleep must serve some crucial purpose, since sleeping animals can't do useful things like search for food and may be easier targets for hungry predators. Memory processing has been held up as the critical task that might have made sleep worth the risk.
For a dose of concentrated skepticism, I called Jerry Siegel, a maverick professor of psychiatry at the University of California at Los Angeles. He has been jousting with Stickgold since 2001, when he wrote a bruising takedown in Science of the evidence that memory consolidation depends on REM sleep.
"The literature's all over the place in terms of what stage of sleep and what type of memory" is supposedly affected, says Siegel. "The evidence isn't converging. It's contradictory."
Will sleep turn out to play some role in boosting memory? I ask.
"Maybe, but the evidence isn't there. And I doubt it will be an essential role," he says.
Siegel has turned his attention to early egg-laying mammals, like the platypus and echidna (otherwise known as the spiny anteater). He has probed the relationship between these and other animals' snoozing habits and their ecological niches to theorize about the ancient origins of sleep and its evolutionary progression. Not surprisingly, he ends up telling a different story than the memory camp about why we spend up to a third of our lives in slumber.
Human sleep is divided into stages of non-REM and REM, which alternate in a repeating cycle throughout the night. During non-REM sleep, the brain's activity is highly synchronized, with large groups of neurons firing simultaneously. "Imagine a stadium full of people, all quiet, then all shouting at the top of their lungs together, then quiet, then shouting," one sleep researcher told me. During REM, on the other hand, groups of neurons fire at many different frequencies, and brain activity overall resembles that during waking. Also, blood pressure rises, the body's muscles are paralyzed (except for those that control breathing and darting eyes), and men typically get erections (leading another researcher to quip that these are the antennae for catching dreams).
What original functions might non-REM and REM sleep have served? Researchers have long speculated that non-REM plays a housekeeping role, removing a toxin or replenishing a substance depleted during waking. Siegel and his colleagues observed that sleep-deprived rats show damage to some brain-cell membranes, damage that sleep may serve to prevent or repair. As for REM sleep, researchers once believed that it was a recent evolutionary development—absent, for instance, in early egg-laying mammals like the platypus and echidna. In the late 1990s, however, Siegel and colleagues in Australia demonstrated that these throwback animals do experience a form of REM sleep, although only in their brainstems. Siegel argues that REM's original function may have been to stimulate the brainstem, perhaps to warm the brain physically or to ready it for waking.
Siegel also argues that the sleep habits of different animals evolved to help them adapt to their ecological niches. The way they slept varied according to what dangers they faced, when they needed to hunt for food, what kinds of shelter were available to them. In contrast to the usual evolutionary assumptions, he suggests that sleep is not dangerous or maladaptive at all. Rather, it provides a period of safety for animals—keeping them immobile, tucked away in a cave or a burrow, hidden from predators, and less likely to be injured or get into fights.
Consider that different animals have different time slots during the day for productive foraging. Bats, for instance, come out at dusk, when the flies they eat are abundant. "The rest of the time, bats are better off sleeping," Siegel says. "Their reproductive success is going to be diminished if they are active at any other time of the day." Newborn dolphins, on the other hand, do not appear to sleep for more than a minute at a time and may not sleep at all (in contrast to the heavy slumber of most mammalian babes). Siegel suggests that this pattern emerged because dolphins cannot retreat to a safe spot while they're learning to survive in the open ocean. He also connects how much a species sleeps to how time-intensive its foraging needs are. Carnivores, which eat periodic high-calorie meals, tend to sleep more than herbivores, which must spend more time grazing. (Omnivores like us fall somewhere in between.)
Viewed through this lens, memory consolidation does not look like the key function of sleep. Siegel notes that animals can take care of all manner of physiological business (besides eating and having sex) as they while the hours away in slumber. They can secrete hormones, repair cells, or shore up their immune systems. Humans, for one, may engage in various kinds of emotional processing; sleep may "knit up the raveled sleeve of care," as Shakespeare put it. We and other species may consolidate memories. But for Siegel, the variation—and especially the unusual ability of some marine mammals to do without consolidated sleep for periods of time—calls into question the essential nature of any one of sleep's functions.
Sleep researcher Jerry Siegel's contrarian stance is philosophically appealing. In his telling, sleep becomes simply another portion of living, its purposes as complex and multifaceted as those of our waking lives. Perhaps, as writer Fran Lebowitz quipped, "Life is something to do when you can't get to sleep."
Still, the basic premise that sleep provides a survival advantage, keeping animals out of harm's way, is not entirely persuasive. Wouldn't lying quietly in a burrow or cave, with full awareness of the surroundings, provide the same safety—and conserve nearly as much energy? Why did animals evolve to spend so much time unconscious? And why the massive changes in nervous system function that accompany sleep?
The variation in sleep among animals does not necessarily undermine the case that sleep serves an essential function either. Consider again the baby dolphins. Perhaps they evolved in some way that allows them to buck the usual mammalian pattern and stay awake early on. Perhaps at birth their brains are already more developed than those of other animals that spend large amounts of time in REM. Siegel himself has pointed out a trend that seems to offer indirect evidence for this possibility: The more an animal can move around on its own at birth, the less REM it tends to get. Dolphins are able to paddle about early on, and get little or no REM during this time. Platypuses, on the other hand are born blind and defenseless and are REM champions, spending more time in this state than any other animal. But that suggests that dolphins' relative lack of need for baby sleep does not mean slumber is any less critical in the animals that get a lot of it after birth. (And surely sleep must be important for dolphins later on, since they go to the trouble of sleeping with one hemisphere of the brain at a time.)
Human dreams raise even more tantalizing questions. Evolution covered some serious ground between sleeping platypuses, which experience REM only in their brainstems and almost certainly do not dream, and sleeping people, who have REM throughout the cortex and can leap off cliffs or stand naked in front of classrooms in the course of a night. Harvard sleep researcher J. Allan Hobson has pointed out that if our brains did what they do during dreams while awake, we would be diagnosed as mentally ill. Yet we enter into this cousin of delirious insanity every night. "Do we go mad at night to prevent ourselves from doing so in the day time?" Hobson has asked. "Or do we go mad because the brain temporarily gives up certain of its controls in order to regain them, in better order, when sleep ends?" Dreams are only one aspect of REM, of course. But the expansion of REM into the cortex that makes them possible may have aided the development of more advanced mental processing—including, perhaps, memory consolidation. Maybe we got smarter around the same time we developed full-blown dreams. (And perhaps it's no accident that dreams have provoked some of the world's most celebrated aha moments.)
Still, more evidence—especially of the cellular and genetic sort—is needed to flesh all of this out. So I turn next to a neuroscientist who focuses on how sleep may help to promote changes in brain wiring, strengthening some connections and weakening others. Time to peer down at the neurons themselves, through a small hole cut in the cranium.
