Sleep

Sleep is of interest to psychologists for a variety of reasons. There is the persisting puzzle of its function. Does it serve physiological restoration, mental restoration, both, or neither? There is the role of sleep in memory consolidation. For clinical psychologists, there is insomnia and other disorders of sleep, to be diagnosed and treated. I will have little to say about these matters in these lectures, which will focus on sleep as an altered state of consciousness.

Fortunately, UC Berkeley offers other courses that deal with these and other aspects of sleep.

There is, first and foremost, a whole course on "Psychology of Sleep" (Psychology 133), offered by Prof. Allison Harvey, a clinical psychologist who runs the Sleep and Psychological Disorders Laboratory, and Prof. Matt Walker, a cognitive neuroscientist and expert on the role of sleep in learning and memory, who runs the Sleep and Neuroimaging Laboratory.
And in the past, UCB also offered a course on "Psychology of Dreams" (Psychology 106), offered by Prof. Eleanor Rosch, a cognitive psychologist who, early in her career, completely overturned 2000 years of philosophical and psychological thinking about the representation and organization of conceptual knowledge. Rosch is now retired, but maybe she can be coaxed into offering the course again. Or, seek out alumni who took her course, and get their notes!

However, because most college students -- and, for that matter, probably most college faculty -- don't get enough sleep, I provide a couple of questionnaires that will help you evaluate your own sleeping behavior.

Sleepiness and Sleep Experiences

Link to surveys of sleepiness, "sleep debt", and sleep experiences.

Interested in keeping a sleep diary? Here's a form that you can use (thanks to the instructors of Psychology 133).

Having trouble sleeping? Complete the Insomnia Severity Index (Bastien et al., 2001; thanks again to Psych 133). Add the scores up for all 7 items (1a-c), and 2-5). A score of 0-7 means "No Clinically Significant Insomnia"; 8-14 means "Subthreshold Insomnia"; 15-21, "Clinical Insomnia (Moderate Severity)"; scores of 22-28 mean Clinical Insomnia (Severe)".

And now, on to sleep as an altered state of consciousness.

Philosophical discussions of consciousness commonly make reference to sleep and dreams.

Sleep is sometimes seen as a lapse in consciousness. Philosophical discussions often contrast consciousness with "dreamless sleep", and so the nature of sleep and dreams becomes relevant to the scientific study of consciousness.
Dreams raise the question of whether, and in what sense, the sleeper might be conscious, despite the appearance of being unconscious.
Assuming that the sleeper is unconscious of things going on in the surrounding environment, there are questions about whether such events can be processed unconsciously, leading to implicit perception and memory.
The physiological correlates of sleep and dreams appear relevant to the mind-body problem.

Sleep as a Biological Rhythm

In behavioral terms, sleep appears to be a homogeneous state, one of many circadian rhythms (from the Latin circa diem, meaning "about a day") evidenced in cyclical changes in activity levels, hormonal levels, and body temperature that alternate over a period of approximately one day. Although the function of sleep is unknown, one effect of the sleep-wake cycle is to synchronize behavior and physiological states to the changing state of the environment. Thus, humans and some nonhuman animals are diurnal, meaning that they are active during the day, while other nonhuman animals are nocturnal, meaning that they are active during the night.

The sleep-wake cycle is a circadian rhythm regulated (we think) by the suprachiasmatic nucleus (SCN) of the hypothalamus, a neural structure that is located above the optic chiasm (Stephan & Zucker, 1972; Moore & Eichler, 1972). The SCN receives information about the presence of environmental light through the retinothalamic pathway, which permits behavior and physiology to be entrained to the actual pattern of light and dark in the environment. Thus, environmental light serves as a Zeitgeber (from the German, "time-giver"). People who are blind because of damage to the eyes suffer from abnormalities in the sleep-wake cycle, and various sleep disorders, in a way that people who are blind from damage to the visual cortex do not.

In addition to the SCN, there may be other circadian oscillators. For example, in the absence of light cues, our internal biological clock produces its own "free-running" circadian rhythm of approximately 25 hours in length. Apparently, the Zeitgeber is reset from signals from specialized cells in the retina of the eye (cells that are different from the rods and cones involved in visual sensation).

The pineal gland, made famous by Descartes as the hypothetical link between mind and body, appears to play a role in the sleep-wake cycle, releasing melatonin ("the darkness hormone") during periods of darkness.

Circadian rhythms are a subset of a larger class of biological rhythms.

Infradian rhythms, with periods longer than one day (e.g., the menstrual cycle in humans and the estrous cycle in many other animals).
Circannual rhythms, or seasonal cycles, with periods of one year (e.g., hibernation or mating in some animals).
Ultradian rhythms, with periods shorter than 1 day, which may be superimposed on circadian rhythms. Among these ultradian rhythms is the 90-minute cycle or NREM and REM sleep, discussed below.

The diurnal cycle of waking and sleeping seems to be governed by two somewhat different mechanisms:

A "wake gauge"keeps track of the body's increasing need to sleep by accumulating adenosine, a hormone involved in the transfer from chemical energy to mechanical energy, beginning after about 14 hours of wakefulness, and reaching peak intensity after about 18 hours (caffeine is an antagonist of adenosine).
A "body clock" tied to the retinothalamic pathway, which synchronizes the sleep-wake cycle to the pattern of light and dark in the environment, and causes the pineal gland to release melatonin at just the right time to put us to sleep.

For more about the physiology of sleep, various sleep disorders, and proper sleep hygiene, see:

The Mystery of Sleep (2017) by Meir Kryger, and physician who specializes in sleep disorders.
Why We Sleep (2017) by Matt Walker, director of UCB's Center for Human Sleep Science.

The Culture of Sleep

Sleep isn't just for physiologists and psychologists. There's also a sociology and anthropology of sleep. Consider, for example, what life was like before electrification. The sun went down candles were expensive, and so there was little to do at night except sleep. (Well, yes, there was that, too.) Some of the sociocultural aspects of sleep are explored by Benjamin Reiss in Wild Nights: How Taming Sleep Created Our Restless World (2017; reviewed by Jennifer Senor in the New York times, 03/02/2017). Reiss has produced a social history and anthropology of sleep -- not by looking at the scientific and medical literature on sleep, but by looking a literature -- including old medical textbooks. He argues persuasively that "Virtually nothing about our standard model of sleep existed as we know it two centuries ago". There used to be "sleep diversity"; now, with industrialization and standardization, never mind the 24/7 information economy, everybody sleeps the same way. And unequally: Reiss has a discussion of what he refers to as "sleep inequality" between upper and lower classes. And he also suggests that our desire to "tame sleep" is part and parcel of our desire to tame nature in general -- and probably to similar effect.

The Diagnosis of Sleep

How do we recognize that sleep has occurred in others, and in ourselves?

First, sleep is manifested in our overt, gross behavior: There is the appearance of physical relaxation (e.g., the person is often lying prone), and a pattern of slow, even breathing. In fact, before the advent of sophisticated electrophysiological techniques (such as the EEG) for monitoring sleep, sleep was often diagnosed solely in terms of patterns of respiration.

Second, sleep is manifested in subjective experience. We experience, at least retrospectively, an interruption of our normal waking consciousness or awareness. There also may be a momentary disorientation upon awakening. We usually cannot remember events that transpired while we were (apparently) asleep, and we often remember having dreamed.

The Architecture of Sleep

To the sleeper, and to the external observer, sleep appears to be a homogeneous state. But when we go beyond superficial introspection and behavioral observation, psychophysiological studies of sleep reveal that there are many different kinds of physiological changes and mental activities during sleep, and that we shift back and forth from one state to another, throughout the night, on a 90-minute cycle (i.e., an ultradian rhythm) of shifting psychophysiological states.

The stages of sleep are defined principally by three physiological indices:

The electroencephalogram (EEG), or "brain waves", measured from one or more sites on the skull. Traditionally, the EEG was recorded from only a single point, at the vertex of the skull, which another lead attached to the earlobe to serve as a ground or reference point. With the advent of sophisticated computer technology, it is possible to record the EEG from a "bonnet" of 256 or more electrodes, spread all over a sort of "bathing-cap" slipped over the skull.
The electrooculogram (EOG), or eye motility, measured from one or more sites surrounding the eyes;
The electromyogram (EMG), or muscle activity, usually measured from one or more sites around the chin (though, for reasons that will become clearer later, other measurement sites would be very interesting).

Taken together, these measurements constitute polysomnography (from the Greek poly = many; somnos = sleep, and graph = writing).

What's depicted above is the minimal setup for polysomnography, involving only a single EEG channel. Nowadays, when EEG records are displayed on a computer screen rather than paper, and computers have a tremendous amount of processing and data-storage capacity, most sleep laboratories record multiple channels of EEG from electrodes placed at various locations around the scalp, in the hopes of more discretely localizing the source of the signals being recorded.

The aggregate EEG can be broken up into various bands, waves, or rhythms, based on their frequency (cycles per second), amplitude (microvolts), shape, and the scalp location where they are most prominent.

There are lots of these EEG rhythms, but only three are most important in the context of sleep.

Alpha activity: Rhythmic waves, moderate frequency 8-12 cycles per second (Hz), high amplitude 5-100 microvolts)

Beta activity: Desynchronized waves, fast (18-30 Hz), low amplitude (2-20 microvolts)

Delta activity: Rhythmic waves, very low frequency (0.5-4 Hz), high amplitude (20-200 microvolt)

The normal waking state is characterized by a mix of alpha and beta activity in the EEG. The relative amounts of alpha and beta activity depend on the mental activity of the subject. "Active looking" tends to block (diminish) alpha activity. In addition, the amount of alpha activity is related to arousal by an inverted-U-shaped function: alpha is low at both low and high levels of arousal, high at moderate levels.

The EMG may be more or less active, depending on the subject's state of relaxation.
The EOG shows saccadic eye movements when the subject looks at something.

Drowsiness is characterized by an increase (as the eyes close or defocus), and then a decrease in EEG alpha. As the sleeper moves down a continuum of arousal, the EMG shows signs of decreasing activity in the skeletal nervous system (SNS), including relaxation of muscle tension and decreasing bodily motility. There are also signs of relaxation in the autonomic nervous system (ANS), including reduced heart rate, respiration, and body temperature.

The Architecture of Sleep: The 90-Minute Cycle of Waking and Sleeping

The sleeper's eyes eventually close, but the onset of sleep is actually defined by the disappearance of alpha activity from the EEG.

Descending Stage 1 sleep is defined by an alpha-free EEG, dominated by low-voltage, fast, desynchronized beta-like activity resembling the waking EEG, except for a somewhat lower amplitude (voltage) -- partly, perhaps, because of the absence of high-voltage alpha activity in the mix. In this and the remaining stages of sleep, through Stage 4, the EMG signs of increasing relaxation in the SNS. The EOG also shows slowly oscillating (rolling) eye movements (SREMs) through this entire period.

In a behavioral test, the subject's lack of responsiveness confirms that the person is in fact asleep. However, upon awakening from this stage people will frequently deny having been asleep.

Stage 1 is a transitional phase in the sleep cycle, usually accounting for about 5% of a night's sleep.

Stage 2, the next stage "down" in sleep, is characterized by a continued deepening of relaxation and SREMs. The EEG is characterized by spindles (bursts of 14-18 Hz activity) and K-complexes (paroxysmal bursts of high-amplitude activity). The spindles reflect a burst of electrical activity that may help bind recently acquired information into the pre-existing network of knowledge stored in memory. When subjects learn a new task during the day, the amount of spindle activity increases that night; and the amount of spindle activity the night before is correlated with performance on the new task the next day.

Stage 2, with the disappearance of alpha activity in the EEG, and the appearance of spindles and K-complexes, marks the true onset of sleep, and consumes about 45% of the sleeper's time during the night.

Stage 3 is characterized by more profound relaxation in the EMG, and the continuation of SREMs. The EEG shows the first appearance of delta activity. Awakening the subject requires a loud noise, or perhaps repetition of the subject's name.

Stage 4 is dominated by EEG delta activity (more than half the time). The person's muscles are very relaxed, and he or she rarely moves. It is very difficult to awaken the subject (the technical term for this is "high auditory arousal threshold"). If awakened, the subject feels very groggy, and comes into focus only very slowly -- a phenomenon known as sleep inertia.

Stage REM occurs after sleep moves "upward", back through stages 3 and 2, and ascends into a new stage, which resembles Stage 1 in some ways. Desynchronized beta activity dominates the EEG, and alpha activity is absent (otherwise the person would be awake). The EOG shows bursts of rapid conjugate eye movements (REMs), as if the sleeper were looking at something (though the eyes remain closed), which gives this stage its name.

There are also signs of autonomic arousal, including bursts of high heart rate and irregular respiration. In males, there is commonly erection of the penis. In females there is an increase in vaginal blood flow.

And, of course, subjects who are awakened during REM are very likely to report that they were dreaming.

In Stage REM, however, the EMG still shows signs of deep muscle relaxation, virtually a paralysis, known as motor atonia -- at least below the neck. There is a central inhibition of those parts of the skeletal musculature mediated by the spinal cord (i.e., the spinal nerves), though not necessarily those muscles served by the cranial nerves (or else REMs could not occur). This paralysis effectively prevents sleepers from acting out their dreams. But it can sometimes continue after awakening, resulting in the relatively rare phenomenon of sleep paralysis, which terminates either spontaneously or after touching the subject or calling his or her name. (Sleep paralysis is commonly thought to be implicated in many reports of alien abduction.)

Stages 2-4 are collectively known as Stage NREM, owing to the absence (N) of rapid eye movements. It is also known as slow-wave sleep (SWS), because all three stages are characterized by relatively low-frequency activity in the EEG. The intervening stage, of course, is known as Stage REM. Before the discovery of REMs, the combination of EEG and autonomic arousal characteristic of waking with EMG characteristic of sleeping led investigators to call "ascending" Stage 1 paradoxical sleep. It seems likely that SWS serves the hypothesized restorative function of sleep: it's when the body and brain appear to be most inactive. This is especially the case for Stage 4, which in electrophysiological terms resembles a coma -- a coma into which we spontaneously enter, and out of which we spontaneously emerge, several times per night.

Although muscle relaxation is a prominent feature of REM sleep, it occurs in NREM sleep as well. In a study of mice, Liu et al. (Science, 01/24/2020) noted activity in the substantia nigra pars reticulata (SNr) both induced NREM (effectively, Stage 1) sleep and inhibited movemen via connections to the motor thalamus. This pathway is quite different from the one that induces motor atonia in REM sleep, which involves brain-stem circuits that suppress activity in the spinal cord.

The sleep cycle consists of a succession of these stages, each epoch consisting of Stage NREM followed by Stage REM.

One period of Stage NREM, consisting of reduced ANS and SNS activity, slow-wave activity in the EEG, and SREMs.
One period of Stage REM, consisting of increased ANS activity, suppressed SNS activity, fast-wave EEG activity, and REMs.

Each epoch takes approximately 90 minutes. Thus, in eight hours of sleep, a person would go through approximately 5 complete epochs, typically awakening out of Stage REM.

Early in the night, Stage NREM predominates, especially Stages 3-4.
Later in the night, REM predominates, and NREM is largely confined to Stage 2.

There are also interesting developmental trends in the sleep cycle.

In infants, about 50% of sleep time is consumed by Stage REM, and Stage 4 dominates NREM.
In young adults (i.e., college students), there is decreased REM, and more Stage 2 within NREM.
In the elderly, there is relatively little REM; instead, the night is punctuated by brief awakenings.

Some cultures incorporate a siesta period after lunch, in which people nap -- and I really mean nap -- none of this French l'heure bleue business. In fact, napping seems to serve some of the same positive functions as nighttime sleep -- provided that the nap is long enough. Ideally, a nap would last around 90 minutes, or just time for one sleep cycle -- 30 minutes of descending Stage 1, 30 minutes of Stage 2, and then some REM-time. Wake up from a nap out of REM, and you feel, and are, refreshed; wake up out of NREM, and you're likely to experience sleep inertia, so groggy you wish you were dead. Although admittedly the prospect is a little scary at first blush, some researchers have reasonably suggested that workers on long shifts, such as airline pilots and even air-traffic controllers, would benefit from being allowed to nap on their shifts -- provided, of course, that someone else stays awake to fly and land the plane!

Sleep in Coma and PVS

As indicated in the lectures on Anesthesia and Coma, polysomnography should greatly improve the differential diagnosis of coma and the persistent vegetative state. Comatose patients should not show any sleep cycles, whether behaviorally (the usual criterion for clinical diagnosis) or physiologically (which is what polysomnography does). See Cologan et al. (2009) for a review.

In spindle coma, sleep spindles and K-complexes appear against a background of delta and theta activity, resembling NREM sleep. Although their presence does not necessarily predict a good prognosis, their absence is a reliable predictor of a poor prognosis.
By definition, PVS patients show some preservation of the sleep cycle, although phasic REM activity may be reduced. When a patient diagnosed with PVS does not show physiological sleep cycles, that patient may be misdiagnosed.

The Physiology of Sleep

The diurnal sleep-wake cycle is regulated by a number of specialized brain modules.

During daytime activity, a hormone called adenosyne builds up in the body. When it reaches a critical level, this substance triggers the ventrolateral pre-optic nucleus (VLPO) of the hypothalamus, which in turn signals arousal centers in the midbrain -- specifically, in the pons to stop producing histamine and other chemicals which maintain arousal and alertness.
In response to decreasing levels of illumination, the pineal gland (yes, Descartes' pineal gland) produces a hormone, melatonin, that also facilitates sleep.

Some species, such as the leatherback sea turtle, have a dedicated sensory system sensitive just to ambient light. The top of the leatherneck skull is extremely thin, allowing light to fall directly onto the pineal gland -- it's actually called a "skylight".
Some other reptiles and amphibians have what can only be called a "third eye" (complete with a lens and a retina), in addition to their forward-facing eyes, connecting to the pineal gland, dedicated to detecting changes in ambient light.

The thalamus blocks sensory inputs during sleep.
The pons blocks motor signals that might be generated during dreaming.
The pons also generates the autonomic and cortical arousal associated with REM sleep.
In response, the cerebral cortex activates the neural networks that account for dreaming.
Toward the end of the night, cells in the retina of the eye pick up increasing levels of ambient illumination.
In response, to signals from the retina, another hypothalamic structure, the suprachiasmatic nucleus, inhibits the activity of the VLPO, thus disinhibiting the midbrain arousal centers, re-generating arousal and alertness.

We'll have more to say about the physiology of the 90-minute cycle of waking and sleeping in the Lecture Supplement on "Dreaming". To make a long story short: REM is regulated by PGO waves, bursts of neural activity that arise in the pons, and then pass through the lateral geniculate nucleus of the thalamus. From there they end up in the visual areas of the occipital lobe, and produce the vivid visual images that are characteristic of dreams. At the same time, there is activity in the motor cortex, which is prevented by inhibitory centers in the pons from reaching the spinal cord and the skeletal musculature.

The Function of Sleep

So, sleep is something we do, and we do it every day, pretty much like clockwork. Every organism with a nervous system has a sleep-wake cycle (and even many flowers close up for the night!). Maybe sleep is just a holdover from our phylogenetic heritage. But why do we do it? Surprisingly, we don't really have very many good ideas about this.

Every animal sleeps. Even jellyfish, who don't even have a brain, appear to sleep (if you define sleep as a state of diminished mobility and responsiveness from which the organism can be easily aroused). Dolphins sleep, turning off one hemisphere at a time so they can keep swimming; so, perhaps, do frigatebirds, so they can keep flying.

Sleep as a Biological Motive

First, it's clear that we have something like a biological need for sleep. Following a period of sleep-deprivation, people perform poorly on various mental and physical tasks, they're less alert, and they have more negative moods. And when they're allowed to sleep, they fall asleep quickly. If they're deprived of REM or NREM sleep, they'll spend more time the next night in REM -- a phenomenon known as REM rebound. There's a similar rebound for NREM sleep.

A series of experiments by David Dinges and his colleagues at the University of Pennsylvania dramatically reveals the effects of sleep-deprivation on the performance of cognitive tasks. In one study, the longest sleep-restriction study conducted to date, they had subjects spend two weeks in the laboratory, and allowed different groups a full 8 hours' sleep, or only 6 or 4 hours. Every two hours (except when they were sleeping), the subjects completed a 10-minute psychomotor vigilance task (PVT), commonly used in studies of sustained attention, in which subjects must respond whenever a number appears on a computer screen. Even short delays in responding, on the order of 500 milliseconds, indicate microsleep -- that is, that the subject has fallen asleep, however briefly. Subjects who got a "full" eight hours of sleep performed consistently well on this task, responding within a half-second to the stimuli. But the subjects who only got four or six hours performed significantly worse, even on the first day, and their performance deteriorated progressively as the two-week trial went on. By the end of the two-week session, subjects who got only six hours of sleep per might were performing as poorly as those who had been sleep-deprived for 24 hours straight. Similar research by George Belenky found that subjects who got 9 hours of sleep per night did no better on the PVT than those who got 8 hours of sleep; but those who got only 7 hours showed the same cumulative performance decrement observed in Dinges' 6-hour subjects.

A survey by the National Sleep Foundation indicates that the average American gets only about 6.9 hours of sleep per night, indicating that all too many of us are walking around sleep-deprived. Adolescents -- this means many of you -- need more sleep time. The elderly appear to get along with less sleep than the young or middle-aged -- though those who sleep longer tend to stay healthier, both physically and mentally.

Sleep as Somatic and Psychic Restorative

But what exactly do we get from sleep? The obvious answer is that sleep performs some sort of restorative function, allowing us to recover from the day's physical activity

Another, less obvious possibility is that sleep enables the disposal of some sort of waste product accumulated in the brain during the day. The lymphatic system performs this function for the rest of the body, clearing out the toxic byproducts of muscle activity, but the brain lies outside its territorial reach. Research by Maiken Nedergaard and his colleagues (Science, 2013) indicates that sleep performs a similar function in the brain, clearing away beta-amyloid and other waste products that accumulate by virtue of waking mental activity. Nedergaard calls this the "glymphatic" system, because glia cells seem to perform this custodial role. Apparently, this process is accelerated during sleep (or general anesthesia). One implication of this is that poor sleep habits may increase a person's risk for Alzheimer's disease and other dementing illnesses. If the glia cells can't do their work efficiently, and they're most efficient during sleep, then beta-amyloid and other waste products build up in the brain -- exactly what happens in AD.
A related proposal is that sleep promotes energy conservation, by virtue of the reduced rate of metabolic activity that is characteristic of NREM sleep.
Another possibility is that somehow, sleep stimulates the production of growth hormones, compensating for the normal "wear and tear" on the body during the day.
Evolutionary biologists have proposed that the sleep cycle permits different species to share the same ecological niche: some species hunt and forage during the day, others at night.
A related proposal is that sleep somehow promotes predatory avoidance -- one species is safely tucked into its house, or cave, or hole, or tree, while another is prowling around.
It's also a possibility that sleep facilitates memory and problem-solving, as when we say we're going to "sleep on" a problem. In fact, there is now considerable evidence that sleep enhances the consolidation of memory -- a matter we'll discuss later.

All of these proposals are eminently reasonable, and quite likely at least one of them is true. But nailing down the exact biological function of sleep hasn't been easy.

Turning the Tables on Adaptationism

With the possible exception of altruism, nothing bedevils evolutionary psychology more than sleep (and dreams). Why should we do something that renders us vulnerable to predation for roughly 1/3 of every day? Why should we have dreams, when we can't remember them?

Martin E.P. (Marty) Seligman, the "founding father" of positive psychology, once posed this question to William K. (Bill) Estes, the "grandfather" of cognitive psychology: "What do you think the evolutionary function of dreaming is?"

The famously reticent Estes replied with one of his famously long, thoughtful pauses (Seligman clocked it at 30 seconds, which seems like an eternity, but entirely plausible to those of us who knew Bill), and then: "What, Marty, do you think the evolutionary function of waking is?".

We usually think of sleep as an altered state of consciousness, and philosophers often point to "dreamless sleep" as a contrast to normal, waking, consciousness. But maybe not: Maybe sleep is the normal state.

Commenting on a series of articles on "The Secrets of Sleep", appearing in the National Geographic magazine, Walter Weller wrote:

Your article treats waking as the normal state of organisms and sleep as a special state entered for special purposes. Sleep research might be more productive if sleep were considered to be the normal state of an organism and waking to be an alert state entered for special purposes only, such as feeding, reproducing, and defense. Really, there is little point in being awake unless one needs to be. Otherwise, wakefulness is a waste of energy. Cats and babies know this.

---National Geographic, 9/2010.

Sleep and Memory Consolidation

In addition to these biological functions, sleep may also have cognitive functions. In particular, there has also been considerable interest in the effect of sleep on memory for information learned while the subject is awake.

Consider, first, an early study conducted by Jenkins & Dallenbach (1924) as a test of the consolidation theory of forgetting. Subjects learned a list of nonsense syllables, and then were tested after retention intervals of 0-8 hours. Some of the subjects remained awake during this period of time, others went to sleep. Memory was actually better when subjects slept between study and test, compared to trials in which they stayed awake for an equivalent interval. Based on these results, Jenkins and Dallenbach proposed that forgetting is caused by interference, rather than the mere passage of time. Because "nothing" happens during sleep, less interference occurs.

But in addition to preventing interference, sleep may actively enhance memory by promoting the consolidation process.

In fact, neurophysiological studies of rats have examined the role that sleep may play in the consolidation of memory. McNaughton and his colleagues (Wilson & McNaughton, 1994; Skaggs & McNaughton, 1995) made single-unit recordings of neural activity while rats were exposed to a spatial-learning task. The rats were placed in a pool of water, and had to swim around until they located a platform on which they could rest. Over time, the animals learn to find the platform fairly quickly. Using this procedure, the researchers identified an ensemble of neurons in a particular area of the hippocampus (CA1) whose activity changed during learning -- an area representing the spatial location of the platform. They then examined the activity of these neurons before and during learning, and during a post-learning period of sleep. The same neurons were active during "slow-wave" sleep (essentially, Stage NREM) as were active during learning -- suggesting that consolidation is mediated, in part, by the reactivation of memories during sleep. As Wilson & McNaughton (1994) put it:

"[D]uring slow-wave] sleep, the activity of hippocampal cells engaged in place-specific activity during prior waking retains the structure of distributed representations of the visited locations.... [I]nitial storage of event memory occurs through rapid synaptic modification, primarily within the hippocampus.... During… slow-wave sleep, synaptic modification within the hippocampus itself is suppressed and the neuronal states encoded within the hippocampus are "played back" as part of a consolidation process by which hippocampal information is gradually transferred to the neocortex".

Interestingly, however, the kind of sleep that seems to be involved is "slow-wave" sleep, analogous to Stage NREM. Assuming that rats have rat-dreams during REM, it's possible that the relative mental quiescence of NREM sleep is critical for consolidation.

Recently, studies of humans have strengthened the case for memory consolidation during sleep. Stickgold (2005) has reviewed a number of studies involving procedural learning -- discriminating foreground from background, motor sequencing, and motor adaptation -- in which subjects showed further improvements in performance, even after formal training had ceased, after subjects had one night of sleep, compared to a control period in which they stayed awake. (Much of this research was performed in collaboration with UCB's Prof. Matt Walker, who has now extended the essential findings to declarative learning as well.)

One experiment involved a visual-texture discrimination task, in which subjects had to identify the orientation of an array of diagonal bars presented against a horizontal background. (a) When retested 6, 12, 18, or 24 hours after initial training, subjects showed further improvement if the post-training interval contained a period of sleep. (b)Improvement was correlated with the amount of SWS in the first quarter of the night, and the amount of REM in the last quarter of the night. (c)Sleep deprivation during the night after initial training prevented the normal post-training improvement
Another involved motor-sequence learning, in which subjects were instructed to type a sequence of 1-digit numbers on a computer keyboard. (d) Again, there was more post-training improvement in performance following an interval of sleep. (e) Improvement was correlated with the amount of Stage 2 NREM sleep in the latter portion of the night. (f) Sleep-deprivation the night after training reduced the post-training improvement normally seen the day afterwards.
And yet a third task involved motor adaptation, in which subjects traced a line between two points on a computer screen against a.countervailing virtual "force". (g) As in the other two experiments, post-training performance improved further after a night's sleep. (h) Performance improvement was correlated with the amount of SWS. (j) The correlation was particularly strong for EEG slow waves recorded over the right parietal cortex.

Interestingly, SWS (Stage NREM) appears to play a large role in these consolidation effects, compared to REM. This is congruent with McNaughton's rat study.

The function of sleep poses a big problem for biological psychologists, but in this course we are primarily interested in the effects of sleep on mental life. Sleep (at least SWS) is often referred to as a state of unconsciousness, but anecdotal reports suggest that some kind of mental activity persists, even when we are asleep.

The sleeper may make bodily movements to adjust his position in bed, or adjust bedclothes.
People can awaken to sounds in their environment -- and it appears that people can selectively awaken to sounds that are particularly meaningful to them -- like a child's cry from the next room.
Some people claim that they can awaken at a preselected time, without recourse to an alarm clock.
And it often happens that people incorporate environmental stimuli -- from the television or from the street outside, for example -- into their dreams.

Sleep after learning is good for memory consolidation. But a nap before learning also has its benefits, apparently by clearing out working memory, to prepare the learner for something new.

The same process that facilitates memory consolidation may also enhance incubation in problem-solving -- an example of implicit thought discussed in the lectures on Implicit Cognition.

Kristin Sanders and her colleagues (Psychological Science, 11/2019) posed a range of difficult spatial, verbal, and conceptual problems to their subjects, each problem associated with a distinctive sound. While the subjects were sleeping, the researchers played the sounds associated with three of six puzzles that they had not been able to solve -- a technique called targeted memory reactivation. When presented with the six unsolved puzzles during a second session the next day, the subjects were more successful solving the ones associated with the sounds presented during sleep than with the controls. They suggest that the sounds evoked memories of their associated puzzles, so that the process of incubation continued even while the subjects were asleep.

Note: there is a prominent theory that incubation as such -- that is, unconscious problem-solving -- does not occur; rather, on the second trial subjects simply start the problem-solving process from a different "entry point". But that can't explain these results. All of the unsolved problems would have received the benefit of a new entry point on the second day, but still the ones whose memories were reactivated were solved more successful. So incubation does appear to occur after all, and it's facilitated by "sleeping on" unsolved problems -- provided that memories of those problems are reactivated during the night.

Falling Asleep and Waking Up

While the function of sleep remains a mystery, what interests us in this course is sleep as an altered state of consciousness -- a radical departure from the wakeful awareness that we experience in the ordinary course of everyday living. Which brings us to the question: what is consciousness -- particularly cognition like during sleep? Is there any at all?

Webb (1975) has noted that subjects are more likely to report that they were asleep if awakened from Stages 2, 3, and 4 than from descending Stage 1 -- even though they are actually asleep, as indicated by the absence of alpha activity in the EEG.

The cognitive changes occurring while falling asleep are illustrated by a study performed by Wyatt et al. (1994). After lights out, subjects were presented with 16 pairs of word associates (e.g., BREAD-BUTTER) at the rate of 1 per minute while they were falling asleep, and instructed to repeat each word pair aloud after hearing it. After the subjects no longer responded, and alpha was absent from their EEG for at least 15 seconds, presentation was discontinued and the subject was allowed to sleep for another 30 sec or 10 min. At this point, the subject was awakened, and asked to recall and recognize the word pairs s/he had heard. Control subjects were instructed to stay awake during list presentation. This procedure was repeated for 10 cycles -- which must have been extremely annoying to those subjects who were trying to go to sleep!

Tests of recall and recognition indicated a sharp drop-off in memory for the items presented during the final three minutes of sleep onset -- including items that the subjects had repeated while they were still awake. Memory was especially poor for subjects who were allowed to sleep for 10 minutes after presentation of the last item. Recognition was somewhat better than recall, of course, but there was still a big difference between sleepers and controls, and especially after 10 minutes of sleep. Wyatt et al. suggested that these findings indicated a temporally graded retrograde amnesia induced by sleep onset.

And of course, the next thing to determine was whether this sleep-onset amnesia dissociated explicit and implicit memory. For the explicit memory test, Wyatt presented the cue term and asked subjects to respond with the item with which it had been paired. For the implicit memory test, subjects were asked to respond simply with the first word that came to mind. Wyatt found that subjects in the sleep condition performed equally poorly on both explicit and implicit tests of memory (control subjects showed a significant priming effect). So, sleep-onset amnesia impairs both explicit and implicit memory.

Ellenbogen and his colleagues have performed an analogous study of the process of waking up (Current Biology, 2010). Subjects slept for three nights in the laboratory, the first being a baseline night. The next two nights, they played various common sounds in the bedroom -- the sound of a toilet flushing, or a phone ringing, people talking, etc. -- as soon as they fell asleep, beginning at low volume and increasing until they awakened. The more baseline spindle activity the subjects showed during Stage 2, the more difficult it was to awaken them. The investigators suggested that spindle activity has the function of shutting out sounds from the external environment.

REM and Dreaming

Aserinsky and Kleitman (1953) first observed the cyclical shift from SWS to REM, with bursts of REM, lasting 10-20 minutes, occurring approximately every 90 minutes. On Aserinsky's hunch, they awakened sleeping subjects, and asked them a simple question: "Were you dreaming?".

The result was a much higher frequency of dream reports during Stage REM (74% of awakenings) than Stage NREM (only 17% awakenings). So, Aserinsky and Kleitman concluded that dreams do not occur throughout sleep (as had been suggested by Freud, for example), but are largely confined to REM. When William Dement, a sleep researcher at Stanford, reviewed this literature in 1976, a number of studies had confirmed A&K's findings -- though, as we will see, there is some controversy here.

What is the connection between dreams and REMs? According to the scanning hypothesis, REMs reflect visual activity on the part of the dreamer -- the dreamer is actually scanning the visual world represented by the dream. On the other hand, even congenitally blind people show REMs, and when awakened from REM sleep report dreaming. So, to be more precise, we should say that according to the scanning hypothesis the dreamer is scanning the spatial world represented by the dream. Or, maybe, REMs don't have anything to do with looking at all -- they're just an adventitious consequence of nervous system activity during a particular stage of sleep.

Before Aserinsky and Kleitman, nobody had ever made all-night EEG recordings of sleeping individuals. The EEG changes associated with falling asleep -- passing from Descending Stage 1 through Stages 2 and 3 to Stage 4 -- were known. But nobody had carried the observation further, on the assumption that the brain quieted down as the person fell asleep, and remained quiet until the person woke up. Also, at the time computers were not widely available, so that polysomnographic recordings had to be made on huge (and expensive) rolls of graph paper. Recording further was believed by everyone -- including Kleitman himself! -- to be a simple waste of paper. But Aserinsky did it anyway, and after that first night everyone in Kleitman's lab became very interested.

Implications of REM and Dreaming

In the end, Aserinsky's discovery had two implications:

The discovery of two distinct types of sleep, REM and NREM, showed that sleep was not a passive state characterized by the relative absence of neural activity. Instead, the brain remains active during sleep, cycling between two distinct states -- one of relative activity (REM), the other of relative inactivity (NREM).

The discovery of the association between REMs and dreams showed that, despite appearances, the sleeper is not unconscious after all. But of course we've known about dreams for a long time -- they're described in myths and Bible stories, for example. Nobody needed Aserinsky and Kleitman to tell them that they dreamt while they were asleep. The real achievement of Aserinsky and Kleitman was to tie dreaming to a specific aspect of physiology -- REMs. By tying dreams to something other than the subject's unverifiable self-report, Aserinsky and Kleitman made the scientific study of dreams possible.

And by showing that REMs, and thus dreams, occurred periodically, reflecting an intrinsic biological rhythm, Aserinsky and Kleitman began to undermine the Freudian, psychoanalytic theory of dreams as symbolic expressions of repressed instinctual urges. Instead, they appear to be the automatic consequence of a particular cycle of biological activity, occurring largely independently of what the person thinks, feels, and desires.

Kleitman, Azerinsky, and Dement

Our understanding of sleep was revolutionized by Eugene Azerinsky's discovery of rapid eye movements, and their association with dreaming. Despite never having completed college, Aserinsky enrolled as a doctoral student at the University of Chicago (Chicago is fairly notorious for permitting this). Despite having no intrinsic interest in sleep, he began working with Nathaniel Kleitman, a physiologist who is regarded as the "father" of modern sleep research -- his book, Sleep and Wakefulness (1939), was a landmark text in the field. Kleitman assigned Aserinsky to study eye movements in sleeping infants, but his attention soon shifted to sleeping adults. Despite Kleitman's belief that the sleeping brain was of no interest, Aserinsky made the first all-night recordings of the sleeping brain (his pilot subject was actually his 8-year-old son, Armond; another early subject was Elaine May, later to achieve fame as a comedian, writer, and film director). William Dement, then a medical student at Chicago, was assigned to assist Aserinsky with his all-night studies. After completing his doctoral dissertation and publishing his paper with Kleitman, Aserinsky left the field. Dement continued working with Kleitman, and founded the world's first sleep clinic -- as opposed to sleep laboratory -- at Stanford. Aserinsky died in an auto accident in 1998, at the age of 77. Kleitman died in 1999, at the age of 104 -- still not convinced that Stage REM was anything other than a shallow phase of sleep.

Aserinsky provided his own account of his discovery in the Journal of the History of the Neurosciences, 1996; see also "The Stubborn Scientist Who Unraveled a Mystery of the Night", Smithsonian, 10/03.

The early findings of Aserinsky and Kleitman were subsequently replicated by a number of other laboratories, and so the traditional conclusion has been that dreams occur in Stage REM. However, it's not completely clear that dreams are solely a phenomenon of REM. However, as we'll see in the lectures on Dreams, the situation is actually a little bit more complicated than that -- and the complications have serious implications for our understanding of the nature of dreams.

Nightmares

As noted, nightmares are a special category of dreams, defined by their content and their consequences: they evoke anxiety, and they awaken the dreamer. Nightmares tend to occur late in the night's sleep.

The term nightmare is derived from Old English: a mare is a spirit that suffocates a sleeping person or other animal. In medieval mythology, there were actually two of these: the incubus, a female demon who gathered semen from men while they slept; and the succubus, a male demon who deposited the semen in women while they slept. The classic clinical study of nightmares is Ernest Hartmann's The Nightmare (1984). And the classic artistic depiction is Henry Fuseli's "The Nightmare" (1781, now in the Detroit Institute of Art), which painting depicts an incubus sitting on a sleeping woman. What's the horse doing there? Maybe it's a pun -- night mare, get it? Or maybe not. For a discussion of the painting (a copy of which graced Freud's study in Vienna),see "Mystery Stalks a Haunting Dream" by Johm J. Miller, Wall Street Journal, 10/31/2020.

Mind and Behavior in Stage NREM

The findings of Aserinsky and Kleitman showed clearly that the sleeper is not unconscious after all -- at least not in Stage REM. But what about Stage NREM? Possibly the sleeper is unconscious in this stage of sleep, but NREM awakenings do typically yield reports of isolated thoughts and images, and REM-awakening studies do yield some clear dream reports after NREM awakenings as well. Dement and Kleitman (1957) solved the problem by suggesting that NREM "dreams" were really just memories of dreams from a previous REM state (Dement, like Aserinsky, was a student in Kleitman's lab, and went on to found and direct the Sleep Research Laboratory at Stanford University).

If NREM "dreams" are really memories of REM dreams, then it might be the case that the sleeper is actually unconscious in Stage NREM after all. But it is now clear that the sleeper is not unconscious even in NREM. Still, the consensus view is that NREM mentation is typically non-dreamlike thinking, reverie, and static imagery.

Subjects awakened in Descending Stage 1, typically report reverie and daydreaming.
Subjects awakened in Stage 2 typically report thinking, reverie.
Subjects awakened in Stage 3 typically report thinking, and floating images.
Subjects awakened in Stage 4 typically give few coherent reports of anything, probably because of the residual grogginess of sleep inertia.

Note that reports of daydreams and reveries are obtained in Descending Stage 1, before any REM dreams are available to be remembered in later NREM reports. Therefore, it cannot be the case that all NREM mentation reflects memories of previous REM dreams. However, we might revise the Dement and Kleitman hypothesis to suggest that later NREM reports are memories of the hypnagogic state. Alternatively, we might simply conclude that mental activity takes place in Stage NREM as well as in Stage REM, but that NREM and REM mentation are qualitatively different.

The Hypnagogic State

The images, reveries, and daydreams characteristic of Descending Stage 1 have a special name: the hypnagogic state (Schacter, 1976). In this state, the person experiences abundant, vivid visual and auditory imagery -- less bizarre than in dreams, somewhat fleeting and disorganized, and with flat affect; but still, abundant and vivid imagery.

Although hypnagogic imagery appears virtually full-blown, and requires little time to develop, its content changes over time as the individual, falling asleep, diminishes contact with reality.

Visual imagery may begin with geometric patterns, and proceed through static objects to complex scenes.
Auditory imagery may include music; a voice speaking one's own name; or the visual images may speak.

The hypnagogic state sometimes ends in a myoclonic jerk, an involuntary muscle spasm. This phenomenon is purely physiological, and reflects the different rates at which muscles and tendons relax; it may occur more frequently in the sleeper is tense. Interestingly, the myoclonic jerk is often accompanied by mental imagery of falling, crashing, and the like.

Before Sleep and Dreams

The hypnagogic state is portrayed musically in "Before Sleep and Dreams", a 1990 piece by the American composer Aaron Jay Kernis. The piece attempts to capture the state of being half-awake, with hints of an image -- particularly, Debussy's "Engulfed Cathedral".

Pavor Nocturnus

These "night terrors" , also known as incubus (a name inspired by a famous painting, shown earlier, of a sleeping woman with a winged devil sitting on her torso), are usually observed in children (Gastaut & Broughton, 1966). The child will awaken with an inarticulate scream, often followed by a cry for help, and an expression of fear, accompanied by signs of an intense discharge of autonomic nervous system activity (sweating, shivering, weeping). Upon awakening, the child is unable to give an account of his or her distress. Typically, the child will fall back asleep, sleep peacefully through the night, and have no memory of the incident in the morning.

Night terrors are distinct from nightmares, which are really just REM dreams with themes of anxiety or catastrophe (danger, injury, death, social affront, embarrassment). Upon awakening from a nightmare, the sleeper can give an account of the dream, and remember it in the morning. While nightmares occur in Stage REM, usually toward the end of the night's sleep, night terrors usually occur in periods of arousal from Stages 3 or 4 (i.e., moving toward Stage 2), early in the night.

Somnambulism

In sleepwalking, the sleeper engages in activity that resembles waking life (Kales et al., 1966a, 1966b). On closer inspection, these behaviors are typically simple, random, and apparently purposeless -- aimless walking, opening doors, turning lights or the television on and off, fumbling with objects. Occasionally the sleeper engages in more complex episodes. Sleepwalking is occasionally accompanied by talking, or at least incoherent mumbling. The sleepwalker typically does not display any affect, and has a "dazed" appearance.

Sleepwalking typically occurs in Stage NREM, usually in an early episode of Stage 4. Of course, sleepwalking during REM would be generally prohibited by the muscle paralysis that is one of the defining features of that stage. However, motor activity can occur in REM, due to incomplete muscle paralysis (see below).

An episode of sleepwalking does not always progress to actual walking. If the person physically relocates, the episode typically ends in a spontaneous awakening. If there is no physical relocation, the sleeper resumes his or her normal sleep cycle uninterrupted. In either case, the sleepwalker has no memory for the episode.

It has been suggested that sleepwalking comes in two types:

In children (age < 12), sleepwalking may be attributable to central nervous system immaturity (children aged 4-6 tend to have more Stage 4 sleep than other children, or adults).
In adolescents and adults, it may be related to periods of life stress.the incidence of sleepwalking has been attributed to an increase in the use of sedative sleeping medications.
A recent upsurge in

Sleeping with a sleepwalker? Maybe install a baby monitor, or some other form of motion sensor, in the sleepwalker's bedroom, to alert others in the house that an episode has begun. Children who sleep walk generally outgrow it. In adults, it's important to distinguish between true sleepwalking, which occurs in Stage NREM, and REM sleep behavior, described below. Either way, the experts advise against waking a sleepwalker, unless they're about to do something that might put them or others in danger. Instead, speak to them in a low tone of voice, gently guide the sleepwalker away dangerous things like stairs or knives, and back into bed. They won't remember anything about the episode in the morning, but you will.

La Sonnambula

Somnambulism provides the plot for La Sonnambula, an opera by Vincenzo Bellini (1831) -- pictured here is a publicity still from Metropolitan Opera production, with Natalie Dessay in the title role, broadcast 03/21/2009.

Amina and Elvino are engaged to be married. Count Rodolfo, traveling incognito to his newly inherited castle nearby, stops to spend the night at an inn owned by Lisa, who was once Elviono's sweetheart. Rodolfo flirts with Lisa, who has discovered his identity, and who hides in his closet when they hear a noise: a phantom has been seen lurking around town. Amina, who is prone to sleepwalking, enters through an open window: Rodlofo infers that she is the "phantom" that everyone's been talking about. Rodolfo is attracted to Amina, but decides not to take advantage of the situation (or to wake her up). He closes the window, lest she fall and hurt herself; she curls up on his couch, asleep, and he leaves the room. Meanwhile, Lisa sneaks out of the room, and tells Elvino that Amina is with the Count. Elvino confronts Amina, breaks their engagement, and resolves to marry Lisa instead. Rodolfo protests Amina's innocence. Elvino doesn't believe him, however, until Amina reappears on the scene, sleepwalking again -- and sleeptalking, too, about her love for Elvino. The couple reconcile, and go off to be married, while Lisa is left to fend for herself.

For more background, see "Sleep Disorder" by Steven Blier, Opera News, 03/2009.

Somniloquy

Sleeptalking refers to the utterance of speech, and other meaningful sounds, by someone who is asleep (Arkin, 1978). If the sleeptalker is awakened, he or she will have no awareness of talking.

Sleeptalking is almost ubiquitous, though it occurs more often in children than in adults.

Sleeptalking typically occurs during periods of transient arousal from NREM sleep, though it can also occur in REM.

In REM, sleep-speech has better syntax and semantics. The sleep-speech usually takes the form of a monologue, or only a single side of a conversation. There is a relatively high concordance between the content of the sleep-speech and that of dreams reported on awakening. It is sometimes possible for an observer to enter into the conversation.
In NREM, sleep-speech is typically meaningless in terms of both semantics and syntax, and there is a relatively low concordance with any mentation reported on awakening.

Linguists take note: The differences between REM and NREM sleep speech have not been documented all that well, though it's been claimed that NREM sleep speech may resemble certain forms of aphasia.

Roommates take note: sleeptalkers are fun to listen to, but they rarely reveal any secrets.

Parasomnias

Sleepwalking and night terrors are part of a larger category of sleep disorders or parasomnias, which include various forms of insomnia: initial insomnia (difficulty falling asleep), middle insomnia (awakening after falling asleep, and difficulty getting back to sleep), and terminal insomnia (awakening too early).

For a literary account of insomnia, see Wide Awake: A Memoir of Insomnia by Patricia Morrisoe (2010), whose own story provides a framework for a wide-ranging discussion of both normal and pathological aspects of sleep -- and the pharmaceutical "insomnia industry". See also The Family that Couldn't Sleep by D.T. Max, which focuses on a genetic disorder known as fatal familial insomnia -- which is just what it sounds like.

Speaking of which, a large number of pharmaceutical sleep aids drugs are available, either by prescription or over the counter, for the treatment of insomnia. Barbiturate and benzodiazepine drugs are not generally used for this purpose, as they inhibit REM sleep. Instead, physicians will often prescribe nonbenzodiazepine hypnotics such as Lunesta, Sonata, and Ambien. These sleep aids appear to work primarily by enhancing the activity of a neurotransmitter known as GABA.

In addition to sleepwalking and night terrors, parasomnias occurring in NREM include bruxism (grinding teeth) and restless leg syndrome, in which the sleeper's legs jiggle (remember, there is no skeletal muscle paralysis in NREM).

Parasomnias occurring in REM include REM sleep behavior disorder, in which muscle relaxation is incomplete, so that the person may act out some of the imagery in dreams (this is commonly observed in post-traumatic stress disorder). On the other hand, in recurrent isolated Sleep paralysis, the sleeper awakens with the muscle relaxation still in place, so that he cannot move (this dissipates relatively quickly).

There is sleep apnea, pauses in breathing that can last for as little as a few seconds to several minutes, and which may occur as frequently as every two minutes or so.

There's narcolepsy, in which people go from waking straight into REM sleep without passing through NREM -- often collapsing in the process, due to muscle relaxation.

And then there's encephalitis lethargica (EL), also known as "sleepy sickness" (not "sleeping sickness, which is something else entirely), or "von Economo disease"first described by Constnatin von Economo -- the same von Economo who described the "von Economo neuron" which supposedly plays a role in social cognition. As its name implies, EL is characterized by chronic drowsiness and excessive sleep -- a condition which lasts for weeks or even months. An epidemic of EL occurred worldwide, beginning in 1916 and subsiding only in 1923; only a few cases have been documented since then. The disease is caused by an infection of brain tissue in the midbrain where the brainstem immediately below the thalamus. This region is known to be part of the circuit involved in the regulation of the sleep-wake cycle. For more on encephalitis lethargica, see "Sleep Without End" by Christof Koch, Scientific American Mind, 03-04/2016, from which the illustration -- including von Economo's own drawing of the lesion -- is taken.

Sleep Suggestion

Evans' studies of sleep suggestion suggest that information from the external environment can be processed during sleep and be retained into subsequent REM periods the same night and even on later nights. Evans' research was inspired, in part, by hypnosis, in which the subject acts on various suggestions administered by the hypnotist, and he wondered if something similar might occur during sleep. Evans understood clearly that, despite the origins of its name (hypnos is Greek for sleep), hypnosis wasn't anything like sleep, either physiologically or behaviorally. Still, an early study by Lawrence LeShan (1942), a clinical psychologist, had suggested (sorry) that subjects could respond to suggestions presented during sleep. LeShan worked with a group of boys at a summer camp who were prone to nail-biting, and he played them a phonograph record containing the suggestion that "My fingernails taste terribly bitter" (applying a bitter-tasting liquid to the fingernails was a common treatment for nail-biters at the time). After 300 repetitions per night over 54 nights (!), 40% of the subjects who received the suggestion stopped biting their nails, compared to none in a control group. Unfortunately, LeShan was not able make use of the EEG to confirm that his subjects actually slept during the presentations. T.X. Barber (1959), also a distinguished hypnosis researcher, found that subjects showed increased responsiveness to hypnotic suggestions during descending Stage 1 sleep.

Evans' general procedure was to recruit subjects for a standard sleep study, taking place over two nights, but making no mention of suggestions. After subjects fell asleep, he administered various hypnotic-like suggestions, typically over an intercom:

Itch: Your nose will feel itchy until you scratch it.
Pillow: Your pillow will feel uncomfortable until you move it.
Blanket: You will feel cold until you pull your blanket over yourself.
Leg: Your left leg will feel cramped until you move it.

The suggestions were administered only a single time, during Stage REM, and were not repeated. The order of suggestions was randomized across subjects. And, of course, Evans insured that, when the suggestions were administered, there was no alpha activity in the subject's EEG (otherwise, he might be awake). The general format of the two-night experiment was as follows:

Night 1, 1st REM Period: Administer the first suggestion, and test it by recording whether the suggested behavior actually occurred.
Night 1, 2nd REM period: Administer the Suggestion 2, and test both Suggestions 1 and 2.
Night 1, later REM periods: Continue testing Suggestions 1 and 2.
Morning 1: Assess subject's memory for, and response to, suggestions.
Night 2, REM 1: Administer Suggestion 3, test Suggestions 1-3.
Night 2, REM 2: Administer Suggestion 4, test Suggestions 1-4.
Night 2, later REMs: Continue testing Suggestions 1-4.
Morning 2: Assess subject's memory for, and response to, suggestions.

A pilot study by Cobb et al. (1965) seemed to show that subjects responded to the cues when they were presented during REM sleep (i.e., about 20% of the time), but not when the same cues were presented in the morning, after awakening. 20% may not seem high, but it's a lot higher than zero, which is what we'd expect to happen.

A more thorough replication and extension of the Cobb study by Evans et al. (1969, 1970) appeared to confirm this result, with about a 20% response rate. Evans also noted a fairly long latency between cue and response, as if the subjects had to mobilize themselves to respond to the cue. The response rate on the delayed tests, 23%, was about the same as on the immediate tests. However, the latency of response was longer on the delayed tests (c. 35 seconds) than it was on the immediate test (c. 19 seconds).

Upon awakening the next morning, the subjects displayed little memory for events of the previous night. On direct interview, 10 of the 19 subjects reported some awareness of the experimenter's presence. While they could occasionally recall parts of the suggestions or cues, they typically did not recall responding. When given a word-association test designed to elicit the suggested responses to the cues in the waking state, subjects gave the appropriate verbal response (e.g., itch-nose), but the suggested behavioral response was elicited in only 6% of the cases.

Moreover, some subjects continued to respond to the cues the next night, and even when tested 5 months later -- but not while awake. This suggested that response to sleep suggestions was in some sense state-dependent -- that is, it seemed the cues had to be administered in the same state as the suggestion, in order to be effective.

However, a carefully controlled investigation by Perry et al. (including Evans as a co-author), with blind judges evaluating response to "dummy" as well as "real" cues, casts doubt on these earlier findings. What appeared to be specific responses to suggestions administered during REM sleep were probably artifacts of spontaneous activity during sleep, interpreted by the non-blind experimenter as responses to suggestions.

Sleep Learning

If dreams are so poorly remembered, and subjects are not responsive to suggestions administered during sleep, what are the possibilities for sleep learning?

Sleep learning is a long-standing fantasy. In Aldous Huxley's dystopian novel Brave New World (1932), set in the 26th century, Bernard Marx, a psychologist at the Central London Hatchery and Conditioning Centre, employs sleep-learning to indoctrinate new citizens of the World State into the official ideology of "Fordism", and acceptance of their assignment within a hierarchical caste system. It's a good plot device, but does that really work?

Perhaps the earliest study of sleep learning was conducted by Thorndike (1916), on military recruits learning Morse code (this was in the run-up to World War I). Records containing letter-Morse equivalents (e.g., A -- dit-dah; B -- dah-dit-dit-dit; C -- dah-dit-dah-dit) were played to the soldiers while they slept in their barracks. Thorndike reported that soldiers who heard the recordings while they slept learned code more quickly in their daytime classes; but, in a foreshadowing of things to come, he also reported that they were very fatigued the next day. Due to the circumstances of the experiment, Thorndike was not able to rule out the possibility that the soldiers in the experimental group had also engaged in surreptitious learning.

A study by Fox and Robins (1952), conducted in the context of the Cold War (when the Soviet Union was also engaged in sleep research), engaged students in the learning of Chinese-English paired associates. The pairs were presented to the subjects during sleep, and then the subjects were asked to learn the same pairs the next morning. In a control condition, the subjects were asked to learn incorrect pairs. Compared to subjects who were presented with music rather than paired associates while they slept, subjects learned the correct pairs more quickly, and incorrect pairs more slowly, the next morning.

The Simon-Emmons Critique

While these results were promising, Simon & Emmons (1955, 1956) criticized the studies of sleep learning then available on the grounds that sleep was not confirmed using EEG criteria (e.g., disappearance of alpha activity). In formal experiments, they monitored EEG while sleeping subjects were presented with new information in question-and-answer format. Subjects were instructed to report if they heard anything during the night, and then tested for recall and recognition of the information in the morning. Reports of hearing the material when presented at night, and recalling and recognizing it in the morning, all varied directly with the amount of alpha activity in the EEG, with recognition falling to chance levels in (alpha-free) Stage NREM.

They also found that memory was best when alpha activity occurred within 10 seconds of stimulus presentation.

These findings led Simon and Emmons (1955) to their now-famous conclusion:

Sleep learning is possible, to the extent that the subject remains awake.

Simon and Emmons' conclusion virtually killed research on sleep learning, except in the Soviet Union and Eastern Europe, where sleep was generally diagnosed by respiration and other signs of autonomic arousal rather than EEG criteria.

State- and Context-Dependency

On the other hand, it is known from the encoding specificity principle that memory is best when the conditions of retrieval are congruent with the conditions of encoding. In line with observations of "state-dependent" or "context-dependent" memory induced by drug states, emotional states, and environmental settings, it is possible that dreams and sleep-learning episodes are encoded in secondary memory, but that access to them is state-dependent -- that is, we would remember dreams and other information very well if only retrieval were attempted in Stage REM! Unfortunately, state-dependency is not an easy hypothesis to test in the context of sleep, and the fate of Evans' studies of sleep suggestion suggest that neither our inability to remember many dreams nor the failure of sleep learning is due to state-dependent learning and memory.

Still, some kind of context-dependency is suggested by studies by Kenneth Paller, Delphine Oudiette, and their colleagues (summarized in Scientific American, 11/2018) on the role of sleep in the consolidation of memory for material learned prior to sleep (as in the Stickgold and Walker studies). These investigators had subjects learn to insert pictures (e.g., of a cat or a helicopter) into their correct locations on a computer screen. During learning, presentation of each picture was accompanied by a related sound (like a "meow" or the whirring of rotor blades). Then the subjects were allowed to nap (a proper nap, it turns out, enhances consolidation just like a full-night's sleep does). But while the subjects slept, the investigators played the same sounds that had been presented during the learning task (a procedure they call Targeted Memory Reactivation, or TMR), while the subjects were awake. Compared to a control group that did not receive these sounds, the experimental group showed better post-sleep retention for the pictures' locations. Later research indicated that this improvement in memory consolidation was correlated with the extent of spindle activity during Stage 2. As with the studies of Stickgold and Walker, the studies of Paller and Oudiette indicate that a considerable amount of "off-line" information processing goes on during sleep, much of which promotes memory consolidation.

One theory holds that sleep enhances memory consolidation by virtue of continued rehearsal -- as in the McNaughton rat study of place learning and the hippocampus. Playing the same sounds during sleep as had been presented during learning seems to promote this rehearsal, perhaps along the lines of the encoding-specificity principle (see the General Psychology lectures on "Memory"). At the very least, the effect indicates that sleepers can process information from the environment, while remaining asleep.

Explicit and Implicit Memory

Returning to the problem of sleep-learning per se, some post-Simon and Emmons studies, employing standard EEG criteria for sleep, did find some evidence for sleep learning. However, in retrospect, these experiments employed testing procedures that could have drawn on implicit rather than explicit memory.

Evans & Orchard (1969) obtained some evidence for sleep learning in alpha-free REM, with paired associates of the form A IS FOR APPLE. Because the stimuli capitalized on pre-existing letter-word associations, the subjects cued-recall performance could have been enhanced by repetition or semantic priming.

Similarly, Cooper and Hoscovec (1972) obtained some evidence for sleep learning of Russian-English paired associates in alpha-free REM (however, learning was much more efficient when the items were presented during waking). Again, paired-associate learning can always be influenced by priming effects between the cue and the target.

Like many of the successful studies from the Soviet Bloc, then, both the Evans & Orchard study and the Cooper & Hoscovec study employed procedures that, in retrospect, could have capitalized on implicit rather than explicit memory, raising the question of whether it might be possible to demonstrate sleep learning with tests of implicit memory.

Wood et al. (1992) performed just such a study, presenting two types of paired associates (homophones + disambiguating context, e.g., WAR-PEACE; category + instance, e.g., METAL-GOLD) during REM and NREM.

Each of the subjects spent three nights in the laboratory. Night 1 was an adaptation night (sleep is typically disrupted the first night that subjects spend in the laboratory). The other two nights were study nights. On each night, after falling asleep, the subjects were presented with paired associates via earphones -- on one night, homophone pairs; on the other, category pairs, counterbalanced across nights. For most of the subjects, the stimuli were presented during Stage REM; for some subjects, they were presented during Stage 2. Each list was presented for four or five times per night. Presentation of the stimulus lists was discontinued if the subject showed alpha activity in the EEG, or any other signs of arousal, such as body movements or SREMs. After each list presentation, the subjects were allowed to sleep for two minutes, and then awakened. A control group heard the lists while lying awake in a darkened room.

On tests involving homophone spelling and category generation, sleeping subjects showed no evidence of priming; by contrast, priming effects were substantial in waking control subjects. For the present, then it appears that sleep learning does not occur, whether measured by explicit or by implicit memory.

Amazingly, the Wood et al. study remains the only published study to date that assessed sleep learning with tests of both explicit and implicit memory. But it seems definitive, why should we repeat it? Because it actually leaves a couple of interesting questions unanswered.

First, Wood's criteria for sleep, while rigorously responsive to the Simon-Emmons complaint, may have been too strict. Perhaps sleep learning would have occurred under more naturalistic conditions, in which there are transient arousals during the night. The subjects might still have displayed no learning on tests of explicit memory, but they might have showed some implicit memory. Something similar has been observed in studies of implicit memory for anesthesia. Explicit memory is abolished at BIS levels of 60 or less, while implicit memory is abolished at BIS levels below 20. But between those two values, anesthesia causes a profound impairment in explicit memory, while sparing implicit memory.

Speaking of anesthesia brings us to the second question. If general anesthesia spares implicit memory, why doesn't sleep? By any standard, the brain is more active during sleep, especially REM sleep, than it is in adequate anesthesia. Recall, however, that Wood et al. employed tests of semantic priming, while the successful experiments on anesthesia employed repetition priming. The sleeper may not have the cognitive resources available to perform the sorts of meaning analyses required to support semantic priming, but may well have had enough resources to perform the sorts of perceptual analyses required to support repetition priming.

So, we'd really like to see the Wood et al. experiment done again, with measures of repetition as well as semantic priming. If that experiment failed to yield positive results, we'd really be certain that, as Simon and Emmons put it, "Sleep learning is possible, to the extent that the subject stays awake".

Memory for Dreams

One of the greatest puzzles of sleep concerns memory for dreams. We probably have, on average, at least 4-5 dreams every night, but we typically remember, at best, only one of these. The contrast between the large number of dreams that occur in the night, and the small number of dreams remembered in the morning, raises the question of what accounts for the remembering and forgetting of dreams.

Which takes us to perhaps the most puzzling thing about sleep: dreaming. It has been said of dreaming that "Every time we experience REM sleep, we literally go mad. By definition, psychosis is condition characterized by hallucinations and delusions. Dreaming, some sleep scientists say, is a psychotic state -- we fully believe that we see what is not there, and we accept that time, location, and people themselves can morph and disappear without warning. From ancient Greeks to Sigmund Freud to back-alley fortune-tellers, dreams have always been a source of enchantment and mystery...."Want to Fall Asleep? Read this Story" by Michael Finkel (National Geographic, 08/2018, p. 72).

Link to the Lecture Supplement on Dreaming.

This page last modified 11/02/2020.