Loughborough University
Leicestershire, UK
LE11 3TU
+44 (0)1509 263171

School of Sport, Exercise and Health Sciences

The phenomena of human sleep

Professor Jim Horne

Human sleep seems to have much in common with that of other mammals. It occurs regularly each day, largely under the influence of the 24h circadian rhythm. There is a typical body posture, a specific place to sleep, other behaviours and physical activity cease, the eyes close, and there is a generalised reduction in sensory awareness. The organ showing the clearest changes during sleep compared with relaxed wakefulness, is the brain. This is particularly obvious in the electroencephalogram (EEG), the brains electrical activity. Focussing on the brain in this way is appropriate in other respects, as not only does the brain contain the control mechanisms of sleep, but of all the body's organs, it the brain and behaviour (especially the cerebral cortex) for which sleep seems to be the most vital. This is especially the case in those mammals with a more advanced development of the cerebral cortex, and culminating in humans.

For many small mammals such as rodents, sleep has other particular benefits, as it provides the only real opportunity for physical rest, and confines the animal to the thermal insulation of a nest. In these respects sleep conserves much energy in such mammals, particularly as sleep can also develop into a torpor, whereby metabolic rate drops significantly for a few hours during the sleep period. On the other hand, humans can usually rest and relax quite adequately during wakefulness, and there is only a modest further energy saving to be gained by sleeping. We do not enter torpor, and the fall in metabolic rate for a human adult sleeping rather lying resting but awake, is only about 5-10%.

It has been thought that sleep is a lesser form of hibernation, and although those small mammals which hibernate enter hibernation through sleep (via torpor), more recent and intriguing evidence indicates that the two states are distinct. Hibernating mammals have periodically to arouse and enter sleep, as it seems that a need for sleep also accumulates during hibernation. Such sleep periods involve a lengthy arousal from hibernation and are costly in energy terms. This strongly suggests that sleep must serve some vital function other than just energy conservation, even for these less cerebrally advanced mammals. Sleep probably serves a variety of functions which may well alter subtly as the evolutionary scale is ascended, depending on various interrelated circumstances of the mammal, particularly body size and need to conserve energy, level of cerebral development, amount of relaxed wakefulness, type of diet, and whether it is predator or prey. Also, for most mammals including ourselves, the roles of sleep may, for different reasons, alter as a night's sleep progresses, initially serving more important purposes, then changing to those of less benefit, and eventually to a sleep that is superfluous, luxurious and just pleasant to take.

Perhaps sleep can be compared with eating and drinking? A certain amount is vital, but we can easily consume more than we really need. Just because we can, on days off, all sleep an hour or so, longer than our average daily amount, it does not mean that we really need this extra sleep. Currently this is a very controversial topic, and the counter- argument is that because most people have the capacity to take an hour or so more sleep than their usual daily 7-8 hours, then they must be chronically sleep deprived. This is the same as claiming that all drinking and eating is physiologically necessary, including the eating of more luxury foods such as cakes, chocolate and ice cream. When people are encouraged to take as much sleep as they can, their alertness in the daytime is only improved marginally, mainly in the early afternoon. If this regimen is continued for several days and nights, then night-time sleep quality deteriorates, and people find that it takes longer to get to sleep at night and they wake up more frequently. It is as if their sleep is becoming stretched out, inefficiently, to fill the extended time available for sleep. Lying in bed in the morning to obtain more sleep is not cost effective time-wise, especially as in healthy people the simplest remedy to daytime sleepiness is a short nap. There is no substantive evidence to show that our daily amount of sleep has declined much over the course of history, and evidence to the contrary is very debatable.

The Measurement of Human Sleep

The human sleep EEG is obtained from innocuous electrodes glued to the scalp, and able to detect the relatively small electrical activity generated largely by the cortex itself. For animals, these electrodes are usually placed on or within the brain, and the EEG can be a little different from that obtained by scalp electrodes. Much of human sleep can be assessed from the EEG alone. But for the determination of rapid eye movement (REM - "dreaming") sleep, there is a problem. In primates the EEG of REM sleep resembles that of stage 1 sleep, the lightest form of sleep (see below), and for this reason REM sleep used to be called "stage 1 REM" sleep. The situation is different for most other mammals, where the EEG of REM sleep resembles that of an alert waking state, and for this reason is often also called "paradoxical sleep" (i.e. the animal is asleep but the EEG indicates wakefulness). So, for one reason or another, additional measures have to be used to detect REM sleep. Electrodes are placed around the eyes to detect the rapid eye movements, and over muscles in the chin or neck, in order to to measure the tonus of resting muscle. For reasons that are not understood, in REM sleep these particular muscles profoundly relax, as their resting tonus is lost - hence, this phenomenon can be used as a further guide to REM sleep. Although other voluntary muscles throughout the rest of the body do not lose this tonus, they are unable to move, as there is a generalised "atonia" (i.e. paralysis of voluntary movement) apparent during REM. The locus coeruleus (situated in the brainstem) is responsible for the paralysis, and the destruction of this area in cats produces "REM sleep without atonia", whereby the sleeping cat moves about in bursts of activity (see below), seemingly enacting its dreams. Rapid eye movements themselves occur in phases or bursts, occupying about one third of REM sleep. This "phasic" activity is particularly intriguing and can be seen within the brain as bursts of "ponto-geniculo-occipital (PGO)" waves, and externally in the limbs as twitches, or during REM sleep without atonia, as more intense movement, with the cat often seeming to be searching around quickly in what appears to be rapid re-orienting to a succession of new and imagined stimuli. The EEG consists of waves that can be measured in terms of amplitude, which rises as sleep becomes deeper, and in terms of frequency (number of waves per second and expressed as "Hertz" - Hz - cycles per second), which falls with deepening sleep. In sleep, the effective frequency range for the human EEG is from about 0.5 Hz to around 25 Hz, although there is other interesting EEG activity up to and around 40 Hz. Generally speaking, EEG frequencies below 3.5 Hz are referred to as "delta" or "slow wave activity" (SWA), which has a high amplitude and is prolific in the deeper sleep stages 3 and especially 4. Other EEG frequency bands can readily be identified in sleep, especially "theta" activity, which is typical of drowsiness and the lightest sleep (stage 1).In contrast, "beta" activity, typically 15-25 Hz, consists of fast waves of low amplitude that are found mostly in the waking, alert cerebrum. "Alpha" EEG activity is in the range 8-11 Hz, and appears during relaxed wakefulness, when there is little or no visual input (especially when the eyes are shut or staring at a blank wall). There are some other, more transient EEG activities found only during sleep, such as "K complexes" and "spindles", which are both most evident in stage 2 sleep. The former seems to be associated with brief arousals often in response to external stimuli, and the latter, probably more with the the brain's active blocking of arousals.

The EEG consists of waves that can be measured in terms of amplitude, which rises as sleep becomes deeper, and in terms of frequency (number of waves per second and expressed as "Hertz" - Hz - but nevertheless, is still generally accepted. Stage 1 is really a transition from wakefulness or drowsiness to what many view to be "real sleep"(stages 2-4 and REM sleep), and in healthy adult good sleepers usually occupies only about 5% of the night. In this stage there is also much slow "eye rolling" of the eyes upwards and downwards, with the eyelids slowly closing and opening. This can easily be seen when watching someone fall asleep. Much of human sleep, around 45 %, is made up of stage 2 sleep. Stage 3 is more of a transition phase from stage 2 to stage 4 sleep, comprising about 7% of sleep, and contains up to 50 % of SWA. When SWA exceeds this level, then the "deepest" sleep, stage 4, is reached, which makes up about 15% of sleep in the young adult.

As can be seen from the "hypnogram" in Figure 1, sleep stages cycle throughout the night, shown especially by REM sleep, which appears about every 90 minutes in human sleep, with each episode lasting around 20 minutes, and in total constituting 20-25% of sleep. Two other characteristics of sleep seen in this hypnogram are: a rapid descent to stage 4 sleep soon after sleep onset, and a prevalence of stages 3 and 4 sleep in the first two sleep cycle. The first period of REM sleep is usually a little shorter than the average.

Why Do We Sleep?

Theories about the functions of sleep date back thousands of years. Most see the purpose of sleep in terms of rest and recovery from the "wear and tear" of wakefulness. One cannot really argue with this idea as it makes so much sense, and besides, we all know that we feel the "worse for wear" without sleep, and so much better after sleep. However, there is little or no evidence that outside the brain any other organ undergoes any heightened degree of repair during sleep. All the evidence so far shows that these other organs undergo their restoration equally effectively (and probably more so) during relaxed wakefulness. The stimulus to the repair of tissue wear and tear is an increase in amino acid levels in the blood, following food absorption by the gut. These amino acids are selectively taken up by cells to be synthesised into new protein(i.e. anabolism) to replace old and degraded protein (apart from water, tissue is largely comprised of protein). This repair is facilitated by physical rest, but it does not have to be sleep - relaxed wakefulness is sufficient. During human nighttime sleep, blood amino acid levels usually fall owing to the absence of eating, and consequently, tissue repair is reduced. Anabolism requires much energy, and it has been estimated that over half of our resting metabolic rate is due to anabolic activities. If these increased significantly throughout the body during sleep, then metabolic rate and oxygen consumption would rise substantially. This doesn't happen, but falls somewhat instead (see above).

Humans show a marked surge in growth hormone output during the early part of sleep, which is largely associated with stages 3 and 4 sleep. If one remains awake at this time then this surge is absent. This sleep-related growth hormone release is rare in other mammals, and has little to do with tissue growth and repair, at least in the human adults. It is probably linked to the rather unusual fasting state which develops in human sleep, and it should be remembered that human night-time sleep tends to be quite lengthy in comparison with that of many other mammals. Few other mammals actually enter a fasting state during sleep. Herbivores continue to digest food throughout their sleep, carnivores gorge themselves on meat which can take up to a day to digest, and rodents wake up periodically to nibble more food. The human sleep-related growth hormone surge may be a mechanism to protect tissue protein against potentially detrimental effects of this fast, and also to promote the body's mobilisation of its fat reserves.

Cell division in many tissues also shows a daily surge late at night and in the small hours of the morning, often coinciding with stages 3 and 4 sleep, the growth hormone rise, and again suggesting that sleep promotes general growth and repair. However, the increase in mitosis is not due to sleep, or to the growth hormone, as it is still evident if one remains awake at this time. This daily cycle in cell division is largely associated with feeding activity and a sleep-independent circadian rhythm. An increase in mitosis typically occurs a few hours after a meal (i.e. allowing for the preliminary digestion, food absorption, cell repair and growth), particularly when we are resting.

Dreaming

Many of the theories about the function of sleep concentrate on REM sleep, and many people believe that we only go to sleep for the purpose of dreaming or having REM sleep. Dreams are usually pleasant and entertaining. We tend to dream in the same style as we think during wakefulness. As most dreams are made up of an amalgam of the previous days thoughts and events, the "interpretation" of dreams can only be accomplished by knowing about these idiosyncratic phenomena. Much of what is claimed about dream interpretation is nonsense, and the fantasy exists not only within the dream but also within the mind of those people who "interpret" dreams. Dreams total around 100 minutes a night, but we can seldom recollect more than a few minutes worth. Dreams cannot be remembered unless one wakes up out of one and immediately thinks about it, when it can then be stored into one's memory. Dreams are probably meant to be forgotten and have little more than entertainment value, simply being the nightly "cinema of the mind".

The underlying REM sleep may well stimulate and tone up the sleeping brain, thus preparing it for wakefulness. People can be deprived of REM sleep for several nights either by being woken up whilst in the laboratory, or for months at a time through the use of drugs such as the tricyclic antidepressants which are typically used to treat depression. Most (not all) of these tricyclics very effectively suppress REM sleep. If REM sleep had some vital role with regard to memory, as is often believed, then this loss of REM sleep for such a long period of time would be expected to have a major impact on memory. But, for example, depressed patients treated with tricyclics show no memory impairment.

Whatever roles REM sleep may have, these may centre on the developing brain, as for most mammals REM sleep is at its most prolific in the developing foetus. One of the most plausible theories as to why this should be; is that REM sleep helps with brain development by providing some sort of substitute stimulation for the brain, owing to the relative lack of external stimulation from within the uterus.

Side views of the brains of a shrew and human. Apart from having a very much smaller brain in proportion to its body size, the shrew has a poorly developed cortex, but well developed olfactory bulb and lobe. Humans have a relatively huge cortex, massive frontal lobe, but only rudimentary olfactory region. The frontal lobe is almost non-existent in the shrew. If sleep offers a special form of recovery for the cerebral cortex, particularly for the frontal lobe, then in these respects it is likely that sleep function will differ radically between these two mammals.

Whilst REM sleep has traditionally been seen to offer benefit to the brain, non-REM sleep, particularly SWA, has been viewed to facilitate repair for the rest of the body. "Dualism", the body versus mind controversy in biology has strongly, but erroneously influenced perspectives on sleep. However, apart from anything else, one must not to fall into the trap of thinking about each type of sleep in isolation, each having its own distinct function, separate from whatever the other types of sleep are doing. Sleep is a complex process and it is likely that different types of sleep interact with one another to promote a variety of functions.

Other sleep theories

Apart from viewing sleep as some sort of recovery process, most of the early theories about sleep function only looked at the "how" of sleep and mechanisms that produce sleep, rather than whatsleep does. For example, believing sleep to be the result of a build up of some substance in the brain during wakefulness, that is dissipated during sleep. Even Aristotle thought along these lines two thousand years ago, and considered that sleep resulted from warm vapours rising from within the stomach. In the last century, with more advancements in the understanding of the brain, heart and vascular system, one school of thought considered sleep to be caused by the "congestion of the brain" by blood. This contrasted with another popular theory at the time, of "cerebral anaemia", due to blood being drawn away from the brain and diverted elsewhere in the body, especially to the gut. Such ideas even led to opposing beliefs about how to induce "better" sleep. Some propounded sleeping without pillows to encourage blood flow to the head, and others encouraged the opposite - use plenty of pillows to drain the blood away. "Behavioural" theories were also common in the 19th century, particularly that sleep was due to an absence of external stimulation, with wakefulness only being possible if the organism was constantly stimulated. Take the stimulation away and the animal will fall asleep. To some extent this notion is true, as we can all testify, but it is not the answer.

At the turn of this century a popular view was that sleep was not so much a passive response, but an active process like an instinct, to avoid fatigue occurring. At this time there were also many "humoral" theories, whereby various sleep inducing substances were proposed to accumulated in the brain. These ranged from known chemicals like lactic acid, carbon dioxide and cholesterol, to the vaguely described "leucomaines" and "urotoxins". Nevertheless, by 1907 some headway began to be made when two French researchers, Drs Rene Legendre and Henri Pieron, claimed to have obtained a substance they called "hypnotoxin" from sleep deprived animals. This gave a large boost to the humoral theories for the next twenty years or so, with much activity by several groups of researchers. However, success was hard to come by and interest dwindled. That is, until the 1960s, when great headway has since been made into "sleep substances".

In those interim years there were advances in neurophysiology that could be related to sleep, and a spate of different neural "inhibition" theories for sleep appeared. Many had had their early impetus from Pavlov's views on "cortical inhibition" - that sleep originated from a form of blocking within the cerebral hemispheres. Although Pavlov dismissed the alternative, of sleep inducing "centres" in more basic parts of the brain below the cortex, these have since been found to exist, and have become the centre of one of the prominent fields of sleep research, especially, after the discovery in the late 1940s, of arousal centres in the reticular formation. Nevertheless, sleep centres and humoral theories still do not tell us much about the purpose of sleep, in the same way that knowing about centres in the brain that regulate eating behaviour explain little about the purpose of eating.

Sleep deprivation

One obvious route to the understanding of what sleep does, is through the total deprivation of sleep. In humans, and with the exception of the brain, sleep deprivation is surprisingly uneventful for the rest of the body. But there are still gaps in our knowledge, especially in relation to the immune system. There is no strong evidence to show that the human immune system is rejuvenated during sleep. In humans, sleep deprivation causes changes to certain aspects of host defence. But these are not impairments, but best described as adjustments due to an increased surveillance by the immune system, which return to "normal" after recovery sleep. In this difficult area of research it is all to easy to jump to conclusions, and one has to look at any specific changes in host defence within the context of what the rest of the highly complex immune system is doing. Also, especially in this field, one can not assume that small changes of statistical significance are of any physiological significance.

Over the last fifteen years, substantial sleep deprivation work has been undertaken with rats, primarily at Chicago University, and under the direction of Dr Allan Rechtschaffen. Animals totally sleep deprived die after about 14 days, for reasons that are still not clear. In the course of this deprivation rats show reliable and consistent changes. There is an initial body temperature rise, which then for a while returns to normal. Animals eat voraciously, but lose weight and develop malnutrition- like symptoms. The output of thyroid hormones falls, whereas catecholamines (adrenaline and noradrenaline) rise - changes which are obviously linked to those with eating and body weight. Animals develop peculiar necrotic skin lesions, and in the final stage of sleep deprivation body temperature drops and weight loss continues. It should be noted that the weight loss at death is far less than that causing rats to die through starvation alone.

Why exactly these animals die still remains an open question, as there is no obvious cause. Interestingly, rats allowed recovery sleep towards the end of this terminal phase generally fully recover, and very quickly (although, some still die). No organ apart from the skin (see above), seems to show signs of failure. Whether or not the immune system fails is a matter for much debate. There are two camps, both armed with much experimental evidence, and both using very similar experimental procedures with their respective rats. One argues that the sleep deprived animals develop widespread systemic infections due to a breakdown of the immune (host) defence system, with bacterial toxins building up in the blood, and probably the eventual cause of death (i.e. septicaemia). In the early stages of this infection, at least in part, this may cause the earlier weight and temperature changes (infections in rats do not generally cause fever - often the opposite). Others don't necessarily agree with the septicaemia hypothesis, as these rats have little by way of obvious serious infection. Treating animals with antibiotics to remove any signs of infection does not affect the temperature and weight changes, nor prolongs progression towards death.

The method for keeping these animals awake involves the principle that rats detest water and getting wet. They sit on a confined circular platform located over a water bath. Signs of sleep in the EEG automatically causes the platform to rotate, and unless the animal moves accordingly, it is pushed into the water. To attempt to control for the stress of the procedure, there is a control animal isolated on the other side of the same platform. This animal has to move whenever its totally sleep deprived companion causes the platform to rotate. But providing that it sleeps mostly when the sleep deprived animal is not attempting to sleep, then the control animal obtains a reasonable amount of sleep, and shows fewer of the symptoms of sleep deprivation.

Many studies have deprived humans of sleep, usually for around one to three days, and a few for five to eight days without sleep. Of course, the circumstances are much more pleasant than those for the rat. There are no reports of physical illness, and no hormonal signs of any alarm response - there is simply a general physiological slowing up. In humans, minor changes or re-settings occur in the regulation of certain physiological processes, particularly with thermoregulation. However, this is very minor - for example, body temperature drops by about 0.5 oC and then stays at this level. This nil outcome with respect to body functioning, from over fifty sleep deprivation experiments on humans, is not what might be expected from a viewpoint that sleep is a condition of heightened tissue growth and repair, overcoming the wear and tear of wakefulness. It seems that for us, these restorative processes are carried out effectively during the wakefulness of sleep deprivation.

Sleep and the Human Cerebral Cortex

Unlike many organs, the cerebral cortex is unable to switch off to any degree, outside sleep. Even when we lie down, turn off the light, relax but remain awake, block out sound and try and clear our minds of all thoughts, the cortex is always in a state of quiet readiness, prepared to respond immediately to any stimulus. In this respect the waking cerebrum is like a computer, where the power consumption is near to its peak, whether it is idling on standby, awaiting instructions, or is involved with programs and taking in information. This may be why the obvious effects of human sleep deprivation are on behaviour, because the cerebrum needs to sleep, and cannot obtain any rest during wakefulness. If such rest is required for the recovery and restitution of neural and related tissues, then sleep is the only provider of this facility. The more advanced the cerebrum, as in humans, the greater the role sleep may have in this recovery.

Most standard psychological tests are generally unhelpful from the viewpoint of sleep function. They are far too simple, unstimulating and tediously boring. This is why they are very sensitive to sleep loss, but if sleep deprived subjects are motivated to do well at the task then the performance deficit can be reversed. Hence, these tests simply measure tolerance to boredom, which shortens with sleep deprivation, and tells us little about the sleep function with respect to the cerebral cortex. So let me take another perspective. The most metabolically and neuronally active part of the waking cortex is the frontal cortex, which comprises over 30% of the cerebrum. It is responsible for directing and sustaining attention, inhibiting distraction by other stimuli, planning many aspects of behaviour (including speech), working memory, and innovative and flexible thinking. If sleep provides cerebral recovery, then this busiest part of the cortex may require the most recovery during sleep, and if so, then it may be the first to start faltering during sleep loss. The types of (neuro) psychological tasks needed to test this hypothesis have to be much more subtle that the more usual tests used to measure sleep loss. This is the approach that our own research has taken, where we have found that despite the subjects' best efforts to perform well, one night of sleep loss impairs the ability to comprehend a rapidly changing situation, increases distraction by irrelevant information, makes people think and plan in a more rigid and less flexible way, perseverate more and be less able to produce innovative solutions to problems. The number of words in one's vocabulary is reduced both verbally and in writing, and articulation becomes more laboured and intonation becomes flatter. All these effects certainly suggest a deterioration within the frontal cortex, and are pertinent to many real-world situations and for those people working without sleep.

Concerning the type of sleep that might particularly facilitate cerebral recovery, SWA more than any other type of sleep, seems the best suited. It is the form of sleep most highly correlated with the length of prior wakefulness. When we go to sleep there seems to be a great pressure to obtain SWA, which takes precedence over all other sleep states. To these ends, SWA best fits a restorative role. More importantly, SWA is particularly evident in the frontal cortex, and brain scanning measures of cerebral metabolism show what appears to be the greatest degree of cerebral shutdown occurring during SWA, especially in the frontal region.

Inasmuch that human sleep seems to provide a vital role for the frontal cortex, and whereas this brain region is only poorly developed in the rodent, this again suggests that the functions of sleep probably change somewhat with mammalian evolution. Perhaps beginning largely as an immobiliser and energy conserver, and culminating with us humans as a facilitator for the recovery of high level cerebral function.