Introduction
Although it may seem trivial to the reader to discriminate whether a subject is sleeping or not, it turns out that an objective external verification is not trivial to achieve without significantly affecting the possible sleep state. In order to determine, for example, whether a subject had fallen asleep, researchers used to ask the subject to hold an object in his hand and wait until the object was dropped.
A major milestone of sleep research was achieved in the 1920s-1930s with the use of the electroencephalogram (EEG) and subsequently the electro-oculogram (EOG).
Behavioral characteristics of sleep
Some typical behaviours associated with sleep include:
Postural recumbancy
Quiescence
Closed eyes
Dream imagery
Less typical behaviours, some of which may be associated with anomalies of sleep, include:Sleepwalking
Sleeptalking
Toothgrinding
Intrusion of sleep into wakefulness
Intrusion of muscle weakness into wakefulness
Intrusion of dream imagery into wakefulness
EEG of sleep
Scalp recordings during sleep show specific changes during the night. H. Berger was the first to show that electrical recordings could be obtained from the human brain during the 1920s. His ideas and techniques were met with skepticism as is usually the case with a new technique. In this particular case, scientists were concerned with the numerous possible electrical noise artifacts that, as every electrophysiologist knows, are so common when attempting to register the electrical activity of the brain. Using scalp recordings, Berger was able to show that the electrical activity changed when subjects closed their eyes and relaxed. The resulting brain had an important component with frequencies arouond 8 Hz. This is typically called the alpha rhythm. The low voltage high-frequency signals characteristic of the awake state were recovered when subjects opened their eyes.
It was Nathaniel Kleitman's group in Chicago who started using the EEG to characterize the brain activity during sleep. His results constitute the best phenomenological description of the electrical activity up to now and are used in sleep clinics throughout the world to distinguish sleep disorders as well as in sleep research laboratories.
As sleep progresses, after the alpha rhythm, the EEG signal is characterized by continuously increasing voltages and lower frequencies. Four different stages are usually distinguished based on the EEG signals from the different electrodes attached to the scalp:
Subjects typically progress from the awake state, through alpha rhythm to an intermediate stage 1 where awakening thresholds are very low, to stage 2 and then stage 3 and then stage 4. This is what was discovered in the initial EEG description of sleep. It was not until many years later that Eugene Aseinsky, in Nathaniel Kleitman's lab, was studying eye movements and attention in infants. He deviced what is now known as the electro-oculogram (EOG) to capture eye movements in real time and he noticed that his subjects often fell asleep and that there were peculiar rapid eye movements while subjects were asleep.
Given that it is possible to observe these rapid eye movements even without the aid of any specific device by simply looking at someone else's eyelids, it is quite intriguing that no one seems to have reported these periods of rapid eye movements (REM) during sleep until 1953. The EEG signal during these periods of REM was characterized by low amplitudes and high frequencies. Actually, it is indistinguishable from that occuring during the awake state. This led to researchers calling it "paradoxial sleep".
Research money to investigate sleep was scarce in those times. When there was enough paper to record a full night of sleep in the EEG, Kleitman's group made the observation that there is a cycic nature to the EEG patterns. After REM sleep, subjects go back to stage 2, then 3 and 4, then REM sleep and so on. The period is approximately 90 minutes. The exact proportion of each stage varies during the night. For example, the proportion of REM sleep is much larger towards the end of the night than at the beginnig.
REM sleep and dreaming
A period of "brain activation" in the middle of sleep immediately lead to the suggestion that it could be associated to the well known phenomenon of dreaming.
William Dement woke up subjects during REM sleep and non-REM sleep. When they were awaken from REM sleep, in 80 % of the cases, subjects reported vivid dream recall. During non-REM sleep, subjects ocasionallly reported having been elaborating plans or thinking about activities or episodes but did not report vivid dreams. Paradoxical sleep became the brain correlate of dreams.
While there are many evolutionary theories about why eyes are moved in this fashion during this particular period of sleep, none has clearly been proved:
Scanning the environment for protection?
Following the dream content?
Randomness?
A "normal" night of sleepSleep starts with non-REM sleep.
Cyclic alternations with a period of approximately 90 minutes.
Slow wave sleep predominates in the first third of the night
REM sleep predominates in the last third of the night and is linked to the circadian rhythm of body temperature.
There are approximately 4-6 discrete episodes of REM sleep.
Factors modifying sleep stage distribution
Age: Newborns have a radically different sleep pattern: Sleep starts with REM in newborns, the period of NREM-REM is 50-60 minutes and the percentage of REM sleep is much larger in newborns.
Prior sleep history: Sleep deprivation, for example, can alter the sequence, proportion and amount of the different sleep stages (to be discussed later in the class).
Time of sleep and circadian rhythms: REM sleep seems to be linked to the circadian variations so that starting very early in the morning can cause more REM sleep to occur at the beginnig of the sleep period.
Temperature
Drug ingestion
What do these sleep stages mean?
The EEG represents some overall aggregate activity in the brain (We don’t yet fully comprehend how it arises precisely: is it synchronous firing of many neurons? is it synchronous dendritic activity? what is the spatial resolution/sumation that yields the signal?) The brain is, I am afraid, much more complicated. Different brain regions, different groups of neurons within each region, different layers within each group, different neuronal types within each layer.
The transitions between the different stages are typically not sharp, with a gradual change which may take from seconds to a few minutes. While the EEG changes are our best indicators nowadays of these pressumed changes in the brain, it is conceivable that there are other dimensions that define more clearly what changes in the brain along the night. Very little research has been carried out to try to understand more carefully what goes on in the brain specifically during each of these tradditional sleep stages. It is conceivable that different neurons are activated and inhibited during, say, stage 1 and stage 4. What neuronal circuits and in which areas yield the phenomenological EEG description of sleep? One way to investigate these issues is through single unit electrophysiology in animals at the same time as EEG. This has the advantages of specificity both in space and time but it has the disadvantage of looking at one or at best a few units at a time. A promising new approach are the new imaging techniques like PET and fMRI. Although at this point the temporal and spatial resolution is worse than that in electrophysiology, it offers the possibility of working with humans in a non-invasive manner and look at the whole brain at once. Part of the techonological complications will involve recording EEG, EMG and EOG at the same time given the high magnetic field in the fMRI machine.
Sleep researchers would also like to understand the functional aspects of each of these stages. In particular, if there is a particular area of the brain, say X, which is particularly activated during stage 3, what is the funtion of X during that period? Part of the way towards assigning functional roles to these stages would be to deprive subjects from each stage and study their behavior. First of all, it is not clear exactly which behaviors should be studied. Memory function is comparatively easier to measure and researchers have started investigating memory recollection during REM and NREM sleep deprivation. But it is conveivable that other important behaviors are affected by the amount of the different sleep stages. Part of the difficulties in this approach arise also because of the difficulties in specifically deprive subjects of some stages without affecting the other sleep stages.
Another approach would be to attempt to create animal models that lack specific sleep stages. Knock-out technology has succeded in isolating mouse/rat strains that are defficient or enhanced performance in specific behaviors. In most of those mutated animals, little care has been given to the sleep period. It may be possible to find mutations that selectively disrupt particular sleep stages. Or it might be possible to perform a specific disruption pharmacologically. Or perhaps there are some lesions in animals or in humans that disrupt particular sleep stages. Again, the literature on the subject is very scarce. But where to start? Recent fMRI data suggests specific regions that may be activated during NREM sleep.
The case of REM sleep and dreaming is particularly interesting to analyze. A correlation between vivid dreaming and REM sleep has been extensively demonstrated by many researchers. But the causal link is less clear. In many cases, REM sleep has been taken as a synonimous of dreaming. It seems possible that dreaming does not occur through the entire REM stage. Furthermore, it may that the specific activation of certain brain areas (such as the higher visual areas in the temporal lobe) is what causes our rich visual sensations during dreaming. This may be normally triggered by a mechanism that is linked to the eye movements but the two processes could be dissociated. Indications of this can be found in the neurology literature where some patients do dream in spite of the fact that they do not show signs of rapid-eye-movement sleep and viceversa. Brain activity during REM sleep has indeed recently been studied in humans with fMRI and higher visual areas do seem to be particularly activated during REM sleep.
Unresolved issues: Hot topics for a Ph.D
1. What are the functional differences between the tradditional called stage 1, stage
2, stage 3, stage 4?2. Which brain regions and circuits are involved in generating these stages?
Are there any molecules/groups of molecules whose level oscillate in a perioc
fashion through the night so as to yield the 90 minute cycles through sleep?3. Which areas in the brain are active only during dreaming (exclusively dreaming, not REM)?
4. Can we do without certain sleep stages?
5. Animal disruption of particular sleep stages. Pharmacology, Knock-outs, Lessions. fMRI data as a guidance?
References
This is a partial list; for a complete list of references, please go to the Bibliography page
Rechtschaffen, A. and A. Kales (1968). A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. Wasington, D.C., NIH Publication 204.
Kryger, M., T. Roth, et al. (1994). Principles and Practice of Sleep Medicine. Philadelphia, W.B. Saunders Company.
Hobson, J. (1995). Sleep. New York, Scientific American Library.