Sleep

Goals

• To discuss what sleep is and to consider why we sleep.
• To discuss a two process sleep model consisting of a circadian rhythm process and a homeostatic sleep process.
• To discuss the neural basis of sleep
• To discuss sleep and memory
• To discuss the metabolic, endocrine, emotional, and cognitive effects of sleep deprivation.
• To discuss REM sleep and dreaming.

Topic slide

Nathanial Kleitman (1895-1999) was a pioneer in the area of sleep research. He established the University of Chicago as a leading sleep research center. Among his many accomplishments was the discovery (with Aserinsky) of REM sleep.

Eve Van Cauter is the current director of the University of Chicago sleep lab. She is an endocrinologist and has studied the effects of sleep deprivation on endocrine function and obesity.

Reading

• Reading: PN6 Chapter 28

What is sleep?

A brief and interesting 2008 essay by Cirelli and Tononi about whether sleep is essential can be found here.

Most, if not all, organisms display periods of rest and activity linked to light/dark cycles. Sleep is a recurring state of inactivity that involves a diminished interaction with the environment. It can be characterized by changes in brain electrical activity in mammals, birds, and reptiles. It can also be demonstrated behaviorally by difficulty in arousing a resting organism.

Depending upon how one defines sleep, it can also be recognized in insects and fishes. One criterion that has been advance is behavior quiescence combined with increased arousal threshold. Other criteria are whether sleep deprivation causes rebound sleep – which suggests that sleep is homeostatically regulated.

Cirelli and Tononi note:

Studies with Drosophilamelanogaster .. , however, demonstrated that flies, also, become less responsive, i.e., sleep, when they remain quiescent for a few minutes. Moreover, sleep pressure increases if flies are kept awake, their sleep patterns change with the life span, and they are sensitive to hypnotics and stimulants … Finally, the fly brain undergoes changes in gene expression between sleep and wakefulness similar to those observed in mammals …, and shows changes in brain electrical activity … Similar criteria have now been provided for zebrafish …, and there is evidence that even the worm C. elegans shows a sleep-like state at a certain stage of development …

Birds and mammals demonstrate two forms of sleep – REM (rapid eye movement) and NREM (non-rapid eye movement). These forms of sleep are distinguished by their EEG patterns. NREM sleep occurs in stages of progressively deeper sleep with the deepest phase called slow wave sleep (SWS). SWS is characterized by slow delta EEG rhythms.

REM sleep (sometimes called paradoxical sleep) has a cortical EEG pattern that is similar to a waking EEG. However, an animal in REM is disconnected from the sensory world and muscularly paralyzed. REM was first discovered by noting that a sleeper's eyes were moving back and forth rapidly under their closed eyelids.

There are controversies about whether reptiles have REM sleep (and not every animal has been tested with EEG). While it was thought that reptiles only had NREM sleep, a recent report suggests that the Dragon Lizard might have a form of REM sleep.

In humans, REM sleep is the time when we have our most vivid ('hallucinatory') dream narratives. We will consider dreaming further below.

The pattern of sleep is different in different animals. Large predators (like humans) sleep for long uninterrupted periods of time. But a sleeping animal is vulnerable, and prey animals may sleep for only very short periods.

Some animals must keep moving while sleeping. Dolphins sleep in one hemisphere at a time, as do some migratory birds.

Why do we sleep?

The short answer to this question is that animals and humans die if totally sleep deprived. Also, if we deprive ourselves of sleep (particularly SWS or REM sleep, we show rebound on the following night). So, there must be some critical biological function that sleep fulfills. However, despite a growing understanding of the deficits caused by NOT sleeping, the precise function that sleep provides is not understood.

There are several hypotheses for why we sleep.

• Metabolic
  • A 155 pound human uses ~46 calories per hour while asleep, ~102 calories per hour while sitting at a desk, and ~930 calories per hour while running at a brisk pace.
  • However, sleeping exposes organisms to predators.
• Restorative
• Learning, memory, consolidation requiring disconnection from new sensory/motor input

I discussed recent data from studies of the glymphatic system that has revived the idea that a core function of sleep may be to remove metabolities and other molecules that build up during the day. I related this research to other data suggesting that the glymphatic system may play a key role in removing toxic proteins from the brain, such as beta-amyloid and tau. Both of these proteins have been implicated in Alzheimer's disease.

I also discussed the idea of normalizing synaptic strenthening.

Again from Cirelli and Tononi:

A net increase of synaptic strength at the end of a waking day would result in higher energy consumption .., larger synapses that take up precious space, and saturation of the capacity to learn. Also, a net strengthening of synapses likely represents a major source of cellular stress …, due to the need to synthesize and deliver cellular constituents ranging from mitochondria to synaptic vesicles to various proteins and lipids. In this view, then, sleep would be necessary to renormalize synapses to a baseline level that is sustainable and ensures cellular homeostasis.

Off-line activity may be necessary to stimulate synapses that remain underused during the waking day …, so they can be ready when their turn comes. It may also be an excellent way of maintaining old memories by keeping them “exercised,” or of weakening nonadaptive memory traces while strengthening the adaptive ones …

I discussed a study where rats were deprived of sleep (either total sleep, or REM sleep). Rats die without either within ~10 days. The animals eat more but lose weight, have difficulty in thermoregulating, and die of massive infections. Thus sleep deprivation is associated with metabolic, thermoregulatory, and immunological processes.

Humans afflicted with fatal familial insomnia also die after months of no sleep.

In the 1960s, researchers deprived young healthy adults of REM sleep for one week. By day 3, they were showing signs of psychosis (hearing things, becoming paranoid, hallucinating).

Two processes involved in sleep

• A circadian process (C) that involves a biological clock located in the suprachiasmatic nucleus of the hypothalamus.
  • This clock is informed of the dark/light cycle from the retino-hypothalamic pathway that conveys information from melanopsin containing retinal ganglion cells.
  • The light cycle stimulates the production of orexin (from the lateral hypothalamus. Orexin stimulates ascending neurochemical systems causing wakefulness and wakeful behaviors (such as eating).
  • The dark cycle stimulates the production of melatonin. This occurs in a somewhat circuitous pathway, whereby the SCN stimulates neurons in the cervical ganglia that, in turn, stimulate the pineal gland to produce melatonin. Melatonin signals darkness, and initiates activity in other clocks that are synchronized to the master clock in the SCN.
    • Administration of melatonin or its analogues can phase-shift the circadian clock, and is used clinically for that purpose.
    • Melatonin can cause drowsiness in humans, but it is not a sedative per se. Indeed, in nocturnal animals, melatonin can signal wakefulness.
    • Melatonin production is diminished by the presence of light – particularly light in the blue end of the visible light spectrum. LED devices have a strong blue peak in their light emissions, and thus watching LED screens before sleep can negatively affect sleep.
• A homeostatic sleep process (S) creates 'sleep pressure'.
  • The homeostatic mechanism may be triggered by the accumulation during the day of adenosine. Adenosine is a byproduct of ATP metabolism that accumulates in the extracellular space over the course of the day. It is also a neuromodulator, and adenosine agonists (barbiturates) cause sleep.

Under normal circumstances, sleep occurs when the two processes are aligned. New data from experiments by Gandi and colleagues in zebra fish have demonstrated that melatonin may be the bridge that connect the S and C processes. Melatonin appears to potentiate the neuromodulatory effects of adenosine.

Molecular clock

Our molecular clock is located in the suprachiasmatic nucleus (SCN). The clock does not run at exactly 24 hours, as demonstrated by a study by Kleitman in Mammoth Cave (KY). There is variation in most individual's circadian timing, but on average, the clock runs a little long (i.e., more like 24 hours and 11 minutes than 24 hours).

Our molecular clock synchronizes other processes in the brain and body that have a circadian rhythm, such as cortisol levels and core body temperature.

Our molecular clock can be synchronized to the light/day cycle by ambient light carried from the retina through the retinohypothalamic tract to the SCN. The retina has a special photopigment – melanopsin – that exists within a special class of ganglion cells that make up this tract.

How long do we sleep, and how long do we need to sleep?

Sleep needs vary across the lifespan. An new born infant might sleep 16 hours per day while a 20 year old sleeps 8 hours per day. The average sleep time declines with aging to about 7 hours. However, many diseases affect sleep, and so it is not certain that less sleep is needed with aging, or whether sleep is affected by illness.

There has been controversy about how much sleep individuals need. There is evidence that people now sleep less previous generations. A 2005 survey indicates that Americans sleep (on average) 6.8 hours per night on a workday and 7.4 hours per night on the weekend. Thus, Americans are chronically sleep-deprived.

However, it is maddenly difficult to get a precise figure on how much sleep is needed. I have been researching this for months, and haven't found consensus. A recent study of three pre-industrial socieites (Yetish et al. Current Biology) showed that they slept 5.7-7.1 hours per night.

Sleep stages

There are five generally recognized stages of sleep. NREM sleep consists of stages 1-4, in which the depth of sleep progresses to stage 4 known as SWS. REM sleep is a distinct stage where the EEG is similar to a waking EEG.

Sleep stages occur on a roughly 90-minute cycle, sometimes punctuated by a brief awakening. In the first half of the night, we spend more time in the SWS phase of the cycle. In the second half of the night, we spend more time in the REM phase.

REM sleep is accompanied by increases in heart rate, respiration, penile erections in males, and eye movements. It is also associated with a marked loss of EMG activity.

Homeostatic sleep process

Over the course of the day, we accumulate adenosine in our brains. This is a byproduct of metabolism, but also acts as a neurotransmitter. Adenosine makes us sleepy, as do adenosine agonists such as benzodiazapines and barbiturates. Adenosine antagonists, such as caffeine, make us wakeful. Caffeine has a half-life of about 6 hours (although this varies among individuals), and so a 100 mg caffeine dose (cup of coffee) at noon is reduced to 50 mg at 6 PM, and 25 mg at midnight. Thus, caffeine can keep us from having good sleep if taken too late in the day.

I showed data from rats showing the adenosine peak over a 24 hour day, and PET data from humans showing how adenosine accumulates in the brain in the sleep deprived.

Neural basis of sleep

Historical studies

Moruzzi and Magoun showed that the 'Ascending Reticular Activating System' was required for wakefulness, and lesions within the ARAS led to permanent sleep.

Walter Hess demonstrated that slow frequency electrical stimulation of the thalamus (in cats) led to sleep.

Suprachiasmatic Nucleus

The SCN is stimulated by the retinohypothalamic tract to synchronize its biological clock to the light/dark cycle. The SCN influences two systems, the Orexin system for waking, and the Melatonin system for sleeping.

Orexin

Orexin is a neuropeptide released by the lateral hypothalamus (LH). It excites the neurotransmitter systems (dopamine, serotonin, norepinephrine, acetylcholine) to activate cortex.

The LH is inhibited by leptin (satiety hormone) and blood glucose. It is stimulated by the amygdala and by ghrelin (the hungry hormone).

Melatonin

The SCN stimulates melatonin release, which in turn synchronizes several body processes and ancillary 'clocks' related to sleep. The release of melatonin is a bit circuitous. The SCN stimulates the paraventricular n of the hypothalamus (PVN) which stimulates the spinal cord and the then neurons in the superior cervical ganglia. These neurons (finally!) stimulate the pineal gland, which releases melatonin.

Up and Down states in the Thalamus

In NREM sleep, all of the neurotransmitter systems are turned down, and this reduces the arousal state of the brain. One of the most interesting aspect of sleep is the development of very slow rhythms in the thalamus. These rhythms consist of UP and DOWN states. In the DOWN states, the neurons in the thalamus become hyperpolarized (below their normal resting potential) and so do not fire. This deprives cortex of sensory input.

In UP states, the thalamic nuclei return to their normal resting potential, and this is accompanied by a synchronized burst of neuronal firing. This burst is believe to play a role in the consolidation of memories.

REM sleep

During REM sleep, intrinsic acetylcholine neurons in the border of the Pons and midbrain begin firing. Their electrical signal – call pontine-geniculate-occipital waves, or PGO waves – are the harbingers of REM sleep (and may be the neural basis for the rapid eye movements).

During REM sleep, the acetylcholine system comes back to activation (which the other neuromodulator systems remain quiescent). This activates cortex (but not all of cortex, see below) and results in an activated EEG cortical pattern.

During REM sleep, there is inhibition of motor neurons in the spinal cord, which leads to paralysis. This may be an adaptation that keeps us from acting out our dreams and wakening ourselves. There is also inhibition of the dorsal column nuclei (DCN) which is where cutaneous information from the spinal cord synapses. This leaves us less responsive to sensory input.

In a sense, during REM, we have an active cortex that is disconnected from sensory input and motor output.

Not all areas of cortex are activated during REM sleep. Dorsolateral prefrontal cortex is not active. We will return to this point later when we consider dreaming.

Sleep and memory

One idea about necessity for sleep is that is it necessary for memory. Hundreds of studies have shown that sleep is necessary to consolidate memories. I showed one such example study.

A recent 2017 paper in Science demonstrated that, during sleep, synapses are remodeled (weakened) and that this participates in the consolidation of memory.

I showed a model by Matthew Walker from UC Berkeley, who posits that, during SWS, the hippocampus trains cortex with memories, and thus restores its ability to encode more information.

Sleep also improves procedural memories. Many musicians have anectdotally noted that a good night's sleep can improve the consolidation of difficult musical sequences. This has been tested experimentally (usually by having subjects press button sequences) in sleep labs and confirmed.

Sleep deprivation

I discussed several metabolic, performance, and cognitive effects of sleep deprivation. In these studies, the sleep deprivation was NOT total, but relatively mild (similar to shift workers, or college students).

Metabolic/endocrine

After 4 hours of sleep on 6 consecutive nights, healthy young adults:

• have increased evening cortisol levels
• increase in sympathetic activation
• increase in insulin resistance (i.e., they show signs of being prediabetic)
• lower cerebral glucose

After 4 hours of sleep for two nights, healthy young adults:

• 18% reduction in leptin (satiety hormone)
• 28% increase in ghrelin (hunger hormone)
  • increased desire for carbs

After 5 hours of sleep for 1 week, healthy young males:

• decreased testosterone by 15% (but NOT cortisol – like the subordinate male baboons)
• showed signs and symptoms of androgen insufficiency syndrome

Memory and performance

• Memory consolidation is poor.
• Motor vehicle accidents and performance errors show a circadian rhythm.
  • 460% increase in diagnostic errors in the ICU in medical residents working a 30-hour shift.

Emotional

• increase in likelihood to false confess after 24 hours of sleep deprivation
• increase in emotional volatility
• increase in endorsing unethical behaviors in the workplace

Wages

Gibson and Shrader (2014) studied wages and sleep, by taking advantage of time zone differences between cites on the extremes of the same time zone (where individuals both need to be at work at the same clock time, but where people in the extreme west of the time zone go to bed later).

A one-hour increase in average sleep was associated with 16% increase in wages.

Modern life

Modern lighting (blue LED light) is strongest at the wavelength of melanopsin. Late night exposure to blue light inhibits the melatonin response from the SCN and leads to poor sleep.

Night caps of alcohol can help us fall asleep faster and increase SWS in first half of sleep, but suppress REM in the second half of sleep, and lead to more awakenings.

Dreaming and REM

The key feature of REM is that the brain is detached from sensory and motor inputs, while demonstrating an otherwise activated pattern. It must be doing something it can't do while simultaneously processing the demands of the day.

As discussed earlier for SWS, REM sleep might play a role in learning. As animals learn new tasks, they spend more time in REM sleep. Francis Crick (who won the Nobel Prize with Watson for DNA) thought that REM sleep was a way to prune memories through 'unlearning'. Others have argued that REM strengthens old memories to keep them from decaying.

The content of dreams depends upon cognitive and neural development. It is tightly linked to the development of visual imagination.

There are two aspects of rapid eye movement (REM) sleep to consider.

• The function of REM sleep with respect to physiology.
• The experiential aspect (conscious experience) of dreaming.

There are several key features of REM sleep.

• The brain is detached from sensory inputs and is unable to produce motor outputs, but otherwise displays an activated pattern. It logically follows that it must be doing something it can't do while simultaneously processing the demands of the day.
• If individuals are deprived of REM sleep for a night, there is a rebound of REM sleep on the following night. This strongly suggests that REM is a biological necessity and is managed by a homeostatic mechanism.
• Birds and mammals have REM sleep, however, reptiles do not (although this is now a contentious point, recent results suggest that Dragon Lizards have REM sleep).
  • This indicates that REM sleep evolved sometime before the emergence of birds and mammals. This suggests that REM must have contributed in some way to biological fitness (as there are clear opportunities for a nocturnal predator encountering a sleeping, paralyzed animal).
• In humans, not all areas of cortex are equally active. The data suggests that visual areas and limbic areas are more active during REM than in either SWS or wakefulness. However, the frontal-parietal attention network seems to be less active in REM. The vmPFC also seems less active during REM. These differences may be reflected in dream content (considered below).
• Our most vivid hallucinatory dreams occur during REM sleep (we dream less frequently in other sleep stages, and these dreams appear more procedural – e.g., we dream about the last thing we were doing before sleep in a repetitive way). However, many animals (perhaps even Dragon Lizards!) have REM, so we cannot know whether they have any experiential component of dreaming.

Potential physiological function of REM sleep.

• REM sleep might play a role in learning. As animals learn new tasks, they spend more time in REM sleep.
• Francis Crick (who won the Nobel Prize with Watson for discovery of DNA) thought that REM sleep pruned memories through 'unlearning'.
• Others have argued that REM strengthens old memories to keep them from decaying.
• Similar arguments have been put forth for SWS, so it is not clear what distinguishes the memory and learning effects of REM and SWS. Some authors have argued that SWS is special for strengthening procedural memories while REM is special for strengthening declarative memories, but there is little empirical data to support these distinctions.

Hypotheses concerning the function of dreams.

There have been many models advanced for the experiential aspects of dreaming. None of these models have achieved scientific consensus.

• Freud had the idea that dreams represented repressed wishes appearing in disguise, and that dreams thus provided important information regarding the state of the psyche.
• Francis Crick, Owen Flanegan and many others have argued that dreams are epiphenomenal and the conscious experience of dreaming has no biological significance.
  • If 'neural noise' is played through the memory system to weaken or strengthen memories, then one might expect random hallucinatory images with sharp transitions.
  • By detaching from sensory experience, Crick argued that subjecting semantic networks to random neural 'noise' would weaken synaptic strengths and, thus, weaken unnecessary ‘parasitic’ memories and associations.
• Similar in some ways to Crick, Alan Hobson has argued that the interpretive parts of cortex attempt to create a narrative out of essentially random input. In Hobson's view, it is this narrative that we experience as dreams. This is reminiscent of Gazzaniga’s ‘interpreter’ that we considered in our discussion of hemispheric differences.
  • It is interesting that the prefrontal regions (dlPFC and OFC) are relatively suppressed during REM and that cingulate, hippocampus, amygdala and visual regions of cortex are activated. This may be the basis for the visual and emotional component of our dreams, and the lack of inhibition.
• David Foulkes (1985) also proposes that dreaming originates in random activation, but with a focus on systems for semantic and episodic memories. In Foulkes model, "… the events depicted in the dream are related to the recent or distant past of the dreamer, not as a simple replay of a past experience but rather as a variation of the past as something that really could have happened to the dreamer.' (from Revonsuo, BBS, 2000).
• Others, such as Revonsuo (2000) have pointed out that dream content is NOT as random as one would expect if dreams were the result of random neuronal firing. Using data from investigators who analyze dream content from thousands of dreams, Revonsuo argues that dreams are emotional (80% negative emotions, 20% positive), overwhelmingly include themes of misfortune (411/1000 misfortune to the dream-self, 58/1000 good fortune), and unpleasant and threatening environments (wild animals, monsters, burglars, storms, fires, floods). Most of the latter dreams required the dream-self to hide or run away. These scenarios are particularly dominant in those who have suffered trauma. Based on the consistency of dream themes – between individuals and within individuals
  • Revonusuo has argued that dreams provide a simulation of threat environments in which different coping strategies can be simulated by the dream-self.

Lucid dreaming

• Lucid dreaming is that in which you have awareness that you are dreaming, and you have a sense of control within your dream (others believe it may be a confusion between a brief awakening during a dream).
• One of the interesting things about Lucid dreaming is that EEG recordings suggest that the dlPFC might be active during Lucid dreams.

Videos

Prerecorded videos for 2020

Live recordings from 2019

The two embedded videos that follow were both recorded in Fall 2019. The sleep topic was not completed in the first lecture, and so part of the following lecture was used to complete the discussion. Note that the second lecture begins with a summary of the major points of the first lecture before moving on to the topic of REM sleep and dreaming.

Here is the first lecture. It is ~ 64 minutes long.

Here is the second lecture that completed the sleep topic. It is ~35 minutes long.