Scientists from around the world gathered
in Chicago on June 5, 2003, to celebrate the
50th anniversary of the discovery of rapid eye
movement, or REM, sleep, an achievement
that became the launching pad of modern
sleep research and the identiﬁcation and
clinical treatment of sleep disorders. The
discovery of REM sleep also ushered in the
scientiﬁc investigation of dreaming, a realm
dominated throughout human history by awe,
terror, claims to prophecy, and free-for-all
interpretation by Freudian analysts. We now
understand sleep paralysis and people who act
out their dreams while asleep, and we know
that dolphins sleep in only half of their brains
at a time. Still, surveying 50 years of investi
gation, sleep researcher Adrian Morrison
suspects that the real mystery of why we sleep
—at all—awaits explanation by the next
generation of scientists.
It was in the 1950s that Eugene Aserinsky, a graduate student of Nathaniel Kleitman at the University of Chicago’s sleep research laboratory, ﬁrst observed and recorded what we now call rapid eye movement (REM) sleep. The investigation began after Kleitman was satisﬁed, as Aserinsky puts it in a humorous and valuable memoir, 1 “that my mind was a clean slate devoid of any self-generated ideas.” Aserinsky came to graduate studies with no post-high school diploma, but with a keen eye and, it turned out, a willingness to spend tedious hours watching babies sleep—his initial project.
Kleitman, then the world’s leading sleep researcher, became interested in the idea that eyelid movements might be a measure of the depth of sleep after reading a report in Nature from a physicist who claimed that he detected sleep onset by noting an abrupt cessation of blinking in two fellow passengers in his rail compartment. Kleitman was disturbed by this report because he did not believe that blinking would end immediately when sleep began. Aserinsky—as he admits being less scientiﬁcally minded—was simply amazed that a prestigious journal would have published such casual observations.
After countless hours trying to glean signiﬁcance from the babies’ eyelid movements, Aserinsky realized that it might be more promising to study the eye movements underneath the lids. His ﬁrst signiﬁcant observation was what he called no eye movement periods—approximately 20 minutes per hour of ocular quiescence. Only when he attached his balky brain wave monitors to his sleeping eight-year-old son, however, did he observe periods of high ocular motility (spontaneous movement). The immediate trigger for this discovery was the decision to use continuous recording with an electroencephalograph (EEG). This, he wrote, consumed “up to a half-mile length of paper per sleep session” to graph the sleep subject’s brain waves and, because it led to the discovery of the periodically recurring REM state, “practically made the denuding of the world’s timberlands inevitable.” Aserinsky almost ended up calling this ocular motility “jerky eye movement periods” or JEMS. Thinking ahead to possible taunts, however, he shied away from using “jerk,” and “rapid eye movement” won the day.
Aserinsky later commented on his difﬁculties in winning Kleitman’s attention to his REM observations, because Kleitman— like most scientists of his time—had theorized that sleep was a completely passive phenomenon. The rigorous Kleitman insisted on repeating Aserinsky’s experiments on his own baby daughter. After further long hours of study, Aserinsky and Kleitman recognized that these episodes of REM were associated with a speciﬁc stage of sleep; the episodes appeared to accompany dreaming. In September 1953, they reported these discoveries to the world.2 When Kleitman died in August 1999, at the age of 104, the University of Chicago’s announcement judiciously described the discovery as follows: “In September, 1953, as a result of work in his laboratory, Kleitman and one of his students, the late Eugene Aserinsky, reported the discovery of rapid eye movements.”
Suddenly, brain science had a brand new world to explore. By the early 1960s, neurophysiologists were exploring the brain structures and mechanisms underlying REM sleep; psychologists and psychiatrists were studying dreams; and neurologists and pulmunologists were ﬁnding that sleep presented its own medical problems. Within 20 years after Aserinsky began his work, a new medical specialty—sleep disorders medicine—was born. I am fortunate to have spent 40 years studying REM sleep and to have witnessed steady progress in unraveling its puzzles.
A NIGHT’S SLEEP
In humans, a night of sleep begins when the low-amplitude, high-frequency brain waves of wakefulness, which can be recorded on the sleep researcher’s classical tool, the EEG, gradually become intermixed with higher amplitude, lower frequency waves as an individual passes from stage 1 through stage 3 of what we now call non-REM (NREM) sleep. The highest amplitude, lowest frequency waves appear in the fourth stage of non-REM sleep, just before the onset of an episode of REM sleep. But during REM sleep, our EEG resembles one recorded when we are awake—in contrast to the lazy, slow brain waves of NREM sleep. This sequence of sleep stages recurs through the night, although later not all stages repeat, and REM sleep becomes more prominent. Slow eye movements do appear in early NREM sleep, but a dramatic shift occurs during REM sleep, when the striking rapid eye movements that Aserinsky ﬁrst detected always make their appearance. The alternation of NREM sleep with REM sleep constitutes a sleep cycle of 90 to 110 minutes or so. REM sleep thus occurs about four to six times per night and comprises approximately 25 percent of total sleep in adults.
Other warm-blooded vertebrates— mammals and birds—exhibit the same sequence, although the NREM stages are not as clearly individuated, and the cycle lengths vary, shorter in smaller animals and longer in animals larger than people. Cats and dogs have a cycle length of about 25 minutes during their sleeping periods, whereas elephants cycle through NREM sleep and REM sleep about every 2 hours and rats do so every 9 minutes. REM sleep has not been observed in insects, ﬁsh, amphibians, or reptiles, and in most birds each episode of REM lasts only a few seconds. Penguins and a few other birds are different; their REM sleep episodes are longer, catlike. The only warm-blooded vertebrate species that have been studied that appear to lack REM sleep are the dolphins and closely related species of whales, although further investigations may yet reveal very small amounts of REM sleep in these animals. Interestingly, dolphins sleep one hemisphere of the brain at a time, presumably to allow them to sleep while surfacing to breathe.3
Paradoxically, although our brain waves are active during REM sleep, we are physically paralyzed (a state of complete lack of skeletal muscle tone called atonia), not to mention unaware of our surroundings. Furthermore, during REM sleep the tight regulation of our internal environment—our homeostasis —is greatly suppressed. We (and other mammals) are to all appearances defenseless, raising puzzling questions about REM sleep’s role from the point of view of evolution.
Scientists still do not agree on the function of sleep itself. Kleitman spent his career studying sleep—a lonely job for a physiologist, because most of his colleagues saw little of interest in the period of our lives when not much seemed to be happening physiologically—and is reported to have said that it was wakefulness that needed explaining, not sleep. Allan Rechtschaffen, the scientist who more than any other has pursued the question of sleep’s purpose, once said that the answer “will probably come at four in the morning in a dingy laboratory in Minneapolis to a graduate student in biology who never read this paper. God bless him.” Later, Rechtschaffen exposed the difﬁculties that the mythical graduate student would face by analyzing other scientists’ attempts to explain the function of sleep. He found all existing hypotheses lacking in what he described as universality and parsimony, but he remained conﬁdent that there is “a primary, essential functional core to sleep that is not captured by a potpourri of lesser functional attributes.”4
Rechtschaffen summarized the reasons for thinking that sleep must be important and ultimately enhance survival, with the implication that those who would dismiss sleep are doomed to failure (I have added a few comments in brackets):
- Sleep is ubiquitous among mammals, birds, and reptiles [and even invertebrates, according to recent experiments].
- Sleep has persisted in evolution even though it is apparently maladaptive with respect to other functions.
- Accommodations are made to permit sleep in different environments and lifestyles [for example, some marine mammals sleep one hemisphere at a time, presumably allowing the animal to surface and breathe].
- Sleep is homeostatically regulated [in other words, rebound occurs after deprivation].
- Serious physiological changes result from prolonged sleep deprivation of animals [including death, if prolonged enough].
The real purpose of sleep, however, remains a mystery to science. My own bet, though, is that various neurons are “resting” during different stages of sleep, the net result being expressed by different EEG patterns.
WHEN DO WE DREAM?
Soon after the initial report on REM sleep, dream research was off and running with a new tool for identifying when to waken subjects in order to record their dreams for analysis. The entire ﬁeld of dream research owes much to a painstaking study published in 1957 by William C. Dement (another Kleitman graduate student who became a sleep research legend) and Kleitman that involved 126 nights of recording from 33 sleeping subjects.5 The idea that some people did not dream at all, or that people who did dream did not always do so, was soon dispelled when subjects awakened during REM sleep always reported dreams. The common assumption that dreams do not take place in real time was dispelled after researchers woke sleeping subjects at various times after the beginning of a REM sleep period and then counted the number of words in their reports on their dreams. Dreams do occur in real time: the longer the episode of REM sleep, the more words in the report.
In this initial human study, REM sleep was still not called by that name. Dement and Kleitman referred to the rapid eye movement periods as “emergent stage 1,” as distinguished from “descending stage 1” at the onset of sleep, when there were no eye movements. They published the now familiar ﬁgure that showed the cyclic pattern of four stages of NREM sleep, followed by a period of REM sleep. As Dement said, “Variations on this picture of all-night sleep have been seen over and over in normal human beings of both genders, in widely varying environments and cultures, and, to all intents and purposes, across the life span.”
We begin our lives (even in utero) with a predominance of REM sleep, or active sleep as it is termed in infants. NREM stages 3 and 4, or slow wave sleep, are at their maximum in young children, and early in the night it is almost impossible to awaken them from this sleep. In the second decade of life, slow wave sleep begins to decline and can even be absent after we are 60. REM sleep declines to 20 to 25 percent by adulthood and remains steady thereafter.
Initially, REM sleep was simply equated with dreaming sleep and NREM with non-dreaming sleep. Those researchers performing the initial studies reported less than 10 percent of dreams from NREM sleep and more than 70 percent from REM sleep. Very soon, though, other sleep researchers reported collecting dream reports from other stages of sleep. These investigators put the incidence of dream recall from NREM sleep at 23 to 74 percent, depending on the study. The difference seems to lie in what one considers a dream. Some NREM sleep dreams approach REM dreams in complexity, but many seem to be simply thoughts going through one’s head without an interesting story line. Tests of discrimination have shown that REM and NREM dreams can be distinguished with a high level of conﬁdence.6
The debate rages on, though, as does the debate about where dreams “arise” in the brain. A few sentences here would not do the problem justice.
DREAMING CATS AND DOGS
As is often the case in investigating the brain mechanisms that underlie a phenomenon found in humans, further study of sleep required experiments in animals. Neurophysiological studies conducted in animals in the 1960s gradually shifted the focus of sleep research from dreams to pathophysiology. Today, as a result, the American Academy of Sleep Medicine counts a mere 5 percent of its members as psychiatrists. Nearly one third are pulmonologists (lung specialists), outnumbering even neurologists.
Why are pulmonologists so prominent in sleep medicine? Just a dozen years after REM sleep’s recognition, simultaneous reports of the discovery of obstructive sleep apnea in France and Germany provided a major insight into daytime sleepiness. Ironically, neurologists made this discovery, not pulmonologists. But thanks to subsequent studies by pulmonologists and respiratory physiologists, we now know that sleep apnea occurs in both NREM and REM sleep when a patient’s airway, compromised because of upper airway malformations or excessive fatty tissue in the throat, occasions multiple protective arousals from sleep. Because the person does not fully awaken, he (most sufferers are men) is not aware of why he is so sleepy during the day.
Years ago, I suggested to my then student, Joan Hendricks, that, if she studied pug-faced English bulldogs brought to our veterinary clinic, we would have a wonderful way to explore the mechanisms of naturally occurring sleep-disordered breathing. Sure enough, we discovered that bulldogs are sleepy fellows—all bulldogs tested fell asleep within 15 minutes in a busy room—whereas test dogs with other shaped heads remained wide awake after an hour and a half. Hendricks and other colleagues are now studying drugs that can enhance muscle tone in the upper airways of bulldogs and other dogs with similar breathing problems.
The far more important experimental subjects, as sleep research moved to animals, were cats, simply because, by the 1950s, so much was known about the structure and function of the cat’s nervous system. For sleep researchers cats were a dream come true: They not only sleep a lot, but they also sleep during the day, when scientists like to do their work. In 1958, Dement published his studies on cats, which were conducted by continuously recording their brain activity through implanted electrodes. He demonstrated that cats have periods of REM sleep with characteristics similar to those of people, but he had a hard time getting the word out. His paper was rejected by several journals, because many scientists could not accept the idea that recordings from a sleeping brain could look like those obtained in wakefulness. Finally, however, to Dement’s “everlasting gratitude, Editor-in-Chief Herbert Jasper accepted the paper without revision for publication in Electroencephalography and Clinical Neurophysiology.”
The idea that animals dream was already well established in folk wisdom. What dog owner did not speak of dogs dreaming of the hunt? Their sleeping dogs paddled their feet, whined, and yelped as if they were on the track of something. But, of course, no layperson was thinking of what the EEG might show. Dogs do act like they could be dreaming and will vocalize like hounds during REM sleep even if they normally do not sound so when awake. I recognized this while listening to my old Labrador, Nera, as she slept on the kitchen ﬂoor. Her strange vocalizations were clearly the result of her irregular respiratory movements coupled with the equally irregular twitches of upper airway muscles. Cats, however, are silent sleepers—and silent hunters. Even big cats, such as lions and tigers, are silent during REM sleep, as I learned when I spent a day at the zoo, monitoring their sleep in between visits of elementary school classes.
THE SLEEPING RUNNING BACK
Discovery that atonia (or paralysis) was also an important characteristic of REM sleep awaited another experiment on cats in 1959 by Michel Jouvet and François Michel in Lyon, France. They were studying the startle reﬂex in cats, portions of whose brains had been surgically removed or altered. During their experiments, the French workers keenly noted that the rigidity characteristic of these cats periodically melted away and that there were eye movements and twitching of the whiskers while the cats were atonic. They soon demonstrated the same atonia during REM sleep in normal cats, and, of course, atonia is something we humans also experience four or ﬁve times every night during REM sleep. We are temporarily paralyzed, but various muscles are repeatedly excited, including those moving our eyeballs.
Yet another surprise awaited the world of sleep research thanks to basic research with animals. Many of us have occasionally experienced an inability to move on awakening. This phenomenon is called sleep paralysis and occurs when the various mechanisms that support wakefulness are brieﬂy out of phase, causing us to perceive brieﬂy the paralysis of REM sleep while we are awake. It would seem an obvious protective mechanism to be unable to move while we are asleep and experiencing rather wild, sometimes socially unacceptable, thoughts like those of many dreams. But in fact, atonia is not necessary for REM sleep. Roughly 35 years ago, Jouvet, and subsequently our research team, demonstrated that in cats damage in the part of the brain called the pons permitted quite elaborate behavior during REM sleep, including mock attacks on a nonexistent mouse.7 Thanks to these experiments with animals, in 1986 sleep disorders specialists in Minneapolis recognized a serious new syndrome: REM sleep behavior disorder (RBD), in which patients—generally older men—are capable of acting out their dreams.8
I address schoolchildren on occasion about the use of animals in biomedical research. Not long into my talk, adolescent boys begin to exhibit a clear wish to be doing something else. That is when I pull out my story of a patient in the ﬁrst series of RBD cases who reported this dream. “I was a halfback playing football, and after the quarterback received the ball from center he lateralled it sideways to me, and I’m supposed to go around the end and cut back over the tackle and—this is very vivid—as I cut back over to tackle there is this big 280-pound tackle waiting, so I, according to football rules, was to give him my shoulder and bounce him out of the way, supposedly, and when I came to I was standing in front of our dresser, and I had knocked lamps, mirrors, and everything off the dresser, hit my head against the wall and my knee against the dresser.”
This story brings the boys back to me with howls of delight and then descriptions of their own dreams. But in this now-enthusiastic setting, I stand alone with other thoughts. Being in my 60s, I can appreciate the sweet sadness that patient must have experienced, as he stood there, not a young football player hearing the cheers of the crowd but a middle-aged man, injured and standing alone in his pajamas.
For people with RBD, however, it is not a laughing matter. They can sustain severe injuries, bone fractures, and severe lacerations and bruises. To protect themselves, sufferers use strategies that range from tying themselves with various types of leashes to the bed, pillow barricades, and padded waterbeds, to sleeping in a barren room on a mattress on the ﬂoor. Fortunately, 85 percent of these patients ﬁnd relief with the anticonvulsant clonazepam.
As bad as the trauma might be, something worse awaits quite a few patients with RBD. When people with RBD were ﬁrst studied, about 50 percent had, remarkably, normal ﬁndings in terms of the results of their neurological examinations; the remainder were afﬂicted with various conditions, in particular neurodegenerative diseases and narcolepsy. Now, however, various neurodegenerative diseases, including Parkinson’s disease, are beginning to make their appearance even in patients apparently normal neurologically when awake. This development suggests the possibility that dopamine abnormalities may underlie RBD. Although RBD can be controlled quite nicely pharmacologically in most patients, it, unfortunately, heralds the onset of even more serious problems.
An interesting footnote exists to the RBD story. In 1981, I and my colleagues reported on a young cat that had been referred to us because of behavior during sleep violent enough to throw it off the couch—or worse, the refrigerator. The veterinary neurologists at the Veterinary Hospital of the University of Pennsylvania brought the cat to Joan Hendricks and me for study because they did not think the animal was truly epileptic. She was not. In a review written for a veterinary journal, we said: “Recently we have identiﬁed a movement disorder conﬁned to REM sleep in a domestic cat…recordings indicated that episodes coincided with REM sleep periods. The cat awoke from REM sleep with no confusion and had no EEG signs of epilepsy or any other abnormality. It remains an alert, healthy animal, with no neurologic disturbance when awake.”9
Hendricks kept the cat, Checkers, as a pet because the owners could not deal with her. As time passed, neurological disturbances began to appear. Checkers developed an unsteady gait and walked with stiffer than normal forelimbs, but the postmortem examination performed after she died at about age 11 revealed no obvious brain damage. Whether she, too, was a case of an individual seemingly normal while awake but transformed later into one with a neurodegenerative disease, we just do not know.
A LINKAGE GONE AWRY
The opposite problem from RBD afﬂicts those who suffer from narcolepsy. This disease is characterized by excessive daytime sleepiness, hallucinations at sleep onset (called hypnogogic hallucinations), sleep paralysis (inability to move on wakening), and cataplexy (partial or total paralysis of skeletal muscles) that may occur at anytime when the patient is awake. The overall problem seems to be maintenance of a steady behavioral state. Cataplectic attacks may be brought on by a variety of exciting, normally non-sleep-promoting stimuli or emotions, such as laughter, surprise, or anger. One can imagine the dangers of becoming suddenly paralyzed while going about one’s daily tasks and so falling uncontrollably in inappropriate locations. But an exciting discovery was recently made in understanding the causes of narcolepsy. Researchers in the laboratories of Emmanuel Mignot and Jerry Siegel examined the brains of narcoleptic patients and learned that neurons in their hypothalamus that normally contain a peptide variously named hypocretin or orexin had degenerated. They looked in this direction because mice that lack a gene for the peptide, and also dogs with damage to a related gene, exhibit signs of narcolepsy.10 Now comes the important task of ﬁnding substances that might substitute for the missing natural peptide in humans with narcolepsy.
These two sleep disorders with opposite effects—too much movement during sleep at night in the case of RBD and paralysis (lack of movement) stimulated during waking activities in people with narcolepsy —can be explained by considering that both an activated brain and suppression of movement during REM sleep are exaggerations of what is called the orienting reﬂex. When we are awake, an unexpected or novel stimulus—for example, a sound or the sudden appearance of something in our visual ﬁeld— produces a very brief hesitation in our ongoing movement and then our attention is directed toward the stimulus. Rather than acting with foresight to prevent dangerous behavior during dreams, nature (being parsimonious) has retained the linkage between a quite activated brain and suppression of motor activity, the result being REM sleep. RBD and narcolepsy are expressions of this linkage gone awry, but in opposite directions.
This hypothesis, which no one has disputed since I proposed it more than 20 years ago, came to me as a result of my penchant for ignoring red lights when crossing a street. Commuters arriving on foot from the train station enter our university campus at an intersection where vehicles approach after a very long block. The alert commuter can safely cross before the cars arrive and the light has turned green. A number of times, though, I lagged a few yards behind other pedestrians, not paying attention to the fact that my fellow commuters were crossing at the last minute. Thus, when I arrived at the corner moments later, a vehicle sometimes suddenly appeared, and I would then note a slight sinking in my knees or a brief prolongation in the stance phase of my stride. One day, having safely reached the other side, I suddenly remembered an event from my college swimming days that provided the insight into the brain excitation-motor suppression linkage of REM sleep. While making the last turn, leading in an exciting 100-yard freestyle race, I suddenly could not move. I felt paralyzed, but I did not “tie up” in muscle spasms as my coach suggested when he tried to console me. No, the fuel line clogged brieﬂy as my brain’s reticular formation, overexcited because of the emotions of the moment, inhibited all my spinal motor neurons. As I stood on the corner of 34th and Spruce, 25 years later, I realized this event could provide the explanation for why REM sleep appears as it does.
My Italian friend, the scientist Pier Luigi Parmeggiani, who ﬁrst demonstrated the suppression of homeostasis during REM sleep, “conﬁrmed” my hypothesis a few years later. We were in a car driven by a Sicilian driver—and therefore at high velocity— suddenly emerging through the gate of an ancient Sicilian town and surprising a man walking just outside the gate. When the man stopped dead in his tracks, Pier Luigi turned to me with a big grin, saying, “You’re right, Adrian, you’re right.”
WHY REM SLEEP?
Of course, my explanation of what happens in REM sleep does not answer the more basic question: Why have REM sleep at all? Why not just let the brain idle, as during NREM sleep? A clue lies in the observation that different groups of neurons in the brain are differentially active during different states. Monoaminergic neurons located in the pons (part of the brain stem), which have widespread projections and many inﬂuences in the brain, are almost completely inactive during REM sleep. Is this stage, then, a period when these neurons have a chance to rest, to replenish themselves? Furthermore, during REM sleep we have seen that the hypothalamus loses its control over homeostatic mechanisms, so those neurons, too, are also not “on watch.”
But if REM sleep is necessary for certain neurons to rest, why can horses do completely without REM sleep for an extended period? Horses can sleep while standing, thanks to passive locking mechanisms in their legs; indeed, they spend 92 percent of a 24-hour period standing.
But they must lie down to enter REM sleep; when they do, their muscles are atonic. More than 30 years ago, French researcher Yves Ruckebusch, whose specialty was the digestive physiology of farm animals, also studied sleep in these animals, including horses. To achieve proper measurements, Ruckebusch had to place electrodes on the horse’s body and hook them up to an EEG machine. After several weeks of habituation to this unusual situation, the horses were able to exhibit all the normal behavior of sleep, including lying ﬂat for REM sleep. (But after being frightened, even a horse well habituated to the recording apparatus will only sleep standing up for the next week.) If awakened once at the beginning of REM sleep, horses will remain awake the entire night. What is it about a horse that allows it to skip an aspect of life so important to fellow mammals, whereas rats, for example, become debilitated and will die if prevented from having REM sleep? One interesting experiment would be to compare the sleep patterns of horses allowed free movement in their stalls with those kept in a tie stall (as cart horses used to be) where they could not lie down but could easily spend time in NREM sleep. When allowed to roam free, would the latter lie down in easily observable REM sleep more than those that had not been tied (the controls)? Would the need for total sleep after REM sleep deprivation be subsumed to their innate need for vigilance if startled at the beginning of their sleep period?
Horses, dolphins, and many other species with special sleep needs and abilities remind us that we are still far from understanding sleep in all its glory, especially REM sleep. At the REM 50th anniversary meeting of the Associated Professional Sleep Societies, Craig Heller, a Stanford biologist who has worked on the problem of the function of sleep, pointed to the many students in the audience, the next generation of scientists who, he said, are likely candidates for providing a deﬁnitive answer. They will be working with a variety of exciting new tools, from genetically engineered mice and fruit ﬂies to increasingly sophisticated brain imaging technology, which may someday help to explain the function of that increasingly less mysterious third of our lives when our brains are on night shift.