This essay is in two parts. The first part offers a brief overview of the main functional theories of sleep. In the second part, attempts to falsify the theories, based on both extant evidence and possible future evidence, will be made.
Why do animals sleep? Functional theories of sleep offer answers to this question. The two most influential theories over the past twenty have been restorative theories and adaptive theories. The following paragraphs briefly describe these two theories. In addition, Horne's core-optional theory and some theories of dreaming and learning will be outlined.
Restorative theories (aka recuperative, restitutive theories) are based on the idea that animals sleep so that physiological and biochemical repairs can take place. Thus, without sleep, an animal's health (physical and/or mental) deteriorates. Three major mechanisms have been proposed: (i) the need to breakdown an unwanted substance which accumulates in the body during activity; (ii) the need to carry out some essential process of chemical synthesis, which is inefficient or impossible during wakefulness; and (iii) to allow the recovery of neural components or pathways which fatigue during arousal. Oswald (1974) has proposed that body restoration occurs during S-sleep, and that brain restoration processes occur during D-sleep, partially through stimulating neural protein synthesis.
Adaptive theories (aka behavioural theories) are based around the idea that sleep is a useful behaviour which keeps an organism out of harms way, both from predators and from inefficient energy expenditure. Hence, sleep is an innate period of immobilisation which confers evolutionary fitness on a species, with no special restorative properties (Webb, 1975). The theory rests on three assumptions: (i) sleeping involves reduce energy expenditure; (ii) immobility affords security against predators; and (iii) sleep occurs during periods of the circadian cycle when an animal is least competent to be active.
A major obstacle for adherents of the purely adaptive position is to explain
the profoundly different stages which comprise sleep. One proposal (Berger &
Walker, 1980) is that D-sleep evolved first in cold-blooded species, S-sleep
evolving later in warm-blooded species to overcome the lack of heat control
in D-sleep, hence avoid non-homeothermic physiological crises during long periods
of sleep.
Following an idea first proposed by Parmeggiani, Horne (1988) has developed a theory that combines element from both restorative and adaptive theories. According to this theory, sleep is composed of two parts: core sleep and optional sleep. Core sleep, which typically constitutes the first three sleep cycles, is the indispensable portion of sleep that an organism requires for essential brain processes, particularly in the cerebral cortex. These processes are presumed to be for the recovery and restitution of neural and related tissue (though Horne is still doubtful of this). Optional sleep - comprising the second half of a night's D-sleep, plus much of stage 2 sleep - is not essential; indeed, it is dispensable. Horne believes that optional sleep satisfies some form of behavioural drive, having the function of occupying unproductive hours and, in the case of small mammals, of conserving energy.
Very many ideas for the purposes of dreams and dreaming have been proposed: e.g. in resolving emotional preoccupations, in altering mood, in development, in adaptation, in creativity and for amusement (Hobson, 1988).One of the most long-standing ideas has been the link between dreaming and learning. This has recently been updated following the work of Crick and Mitchison (1983). Their theory is based on an analogy between the cerebral cortex and artificial neural networks. The problem with such networks is that adding new information can cause overload and disruption, leading to undesirable modes of interaction. The proposed function of D-sleep is to tune and debug the system via random neuronal firings which help eliminate spurious connections, hence consolidating 'real' memories. According to this view, D-sleep dreaming is nothing more than an epiphenomenon brought about by these random firings (as proposed by Hobson & McCarley, 1977, in their Activation-Synthesis hypothesis).
Having surveyed the theories, let me now turn to look at the evidence which might help decide between them. The emphasis will be on attempts to falsify the theories.
Sleep theories can be assessed by comparing the sleeping patterns of different animals and how sleeping patterns relate to the animals's form, behavious and environment. It should be stated from the outset, that comparative studies may not shed much light on the functional nature of human sleep. Horne (1988) points out that the questions 'why does a bird sleep?', 'why does a rat sleep?' and 'why does a human sleep?' may have very different answers. Having said that, let me briefly mention some evidence.
Strong comparative evidence against restorative theories (and for adaptive theories) comes from the fact that although animals may have very similar sizes and complexity, their amounts of sleep can vary enormously. An example of this is that bats sleep for 20 hours per day, but shrews hardly at all (Allison & Van Twyer, 1970). In addition, Meddis (1975) claims that shrews, swifts and Dall porpoises all survive without sleep. Seemingly damning evidence against restorative theories.
What of comparative evidence against adaptive theories? This would surely come from finding animals in which sleep seemingly decreases their (evolutionary) fitness, but in which sleep has not been "deselected". A candidate for such a case is the Indus Dolphin. Pilleri (1979) reports that these dolphins sleep for 7 hours per day in numerous naps lasting less than a minute each. However, sleep in this species would seem to be a maladaptive behaviour, since these dolphins must always keeping swimming to avoid being harmed by dangerous currents and large amounts of debris.Another odd case is that of bottlenose dolphins and porpoises (Mukhametov, 1984) in which the two hemispheres of the brain sleep independently. The question arises; why have such a complex mechanism if sleep has no restorative properties?
Two questions arise from Berger and Walker's homeothermic proposal: (i) why do both mammals and birds go into S-sleep first?; and (ii) why has D-sleep not 'evolved out' completely? Unfortunately, there is no fossil evidence to say how sleep patterns have evolved, and studies involving extant species (especially reptiles) have proved controversial and inconclusive, so we may never know the correct answer to such questions.
Attempts to find a natural sleep-inducing chemical have proved inconclusive (e.g. methodological problems with twin studies and CSF transfusions, and temperature effects of muramyl peptide). The proposal that growth hormone mediates restorative processes appears flawed. The finding that protein is built up during the day and not during sleep (Clugston & Garlick, 1982) is further evidence against a possible restorative process. However, the fact that glucose metabolism increases during tasks that require mental activity and alertness (Roland, 1985) does suggest a restorative period may be required.
A number of laboratory-based studies have deprived animals of sleep to observe the biochemiocal and behavioural effects. Unfortunately, the evidence from such studies is virtually worthless due to the methodological flaw associated with the stressors involved in keeping the animal awake (Horne, 1988). If animals could be kept awake by disrupting their sleep triggering mechanisms (by lesioning, electrode stimulation or genetic engineering), then important evidence may result.
Humans, deprived of sleep for a few days, suffer virtually no physiological damage. However, they do suffer in psychological ways: e.g. lack of prolonged concentration and some perceptual distortions (though as Meddis, 1983, has pointed out, the deleterious effects of sleep deprivation may be indicative of nothing other than a disruption in the mechanisms that control sleep). Lack of sleep, then, is not devastating to humans apart from the intense desire to want to sleep. Add to this the fact that some people apparently stay healthy with very little (or possibly zero) sleep and one has reasonable evidence against a purely restorative position. The case would be trengthened, if someone could be found who displayed zero sleep under strict laboratory conditions.
Lavie et al (1984) describe the case of a man who displayed virtually no D-sleep due to a shrapnel injury to the pons, yet showed no obvious health or learning problems. This offers strong evidence against both restorative and learning theories. However, a learning theory which does accommodate such evidence is that of the Ontogenetic hypothesis (Roffwarg et al, 1966) - the idea that very young mammals (even foetuses) require D-sleep for cerebral development, but that in later life this function has mostly been fulfilled (the percentage of D-sleep is much higher in very young humans (Parmelee & Stern, 1972).
A number of studies have shown the link between D-sleep and learning. For example, Bloch et al (1977) showed that D-sleep increased by 10-15% over 6 days while rats learned a complex maze task, and decreased when the task was accomplished. A recent study by Marczynski et al (1992) lends support to Crick and Mitchison's ideas. These investigators used intracellular electrical recordings in freely moving cats to determine neuronal firing patterns during learning tasks and sleep. Basing their model on artificial neural networks, they confirmed that random firing patterns during both S-sleep and D-sleep removed unwanted data and strengthened the cat's learning.
In studies on both healthy subjects given 6 weeks complete rest (Ryback et
al, 1971) and on quadriplegics (Adey et al, 1968), lack of exercise
does not result in significantly lowered S-sleep; strong evidence against Oswald's
position. Horne has provided a new insight into the fact that heavy physical
exertion during the day will lead to large increases in S-sleep. In a series
of experiments (including people exercising while being sprayed with cool water;
and people taking long, hot baths), it was shown that it is not the exercise
that is causing the increase in SWS but core temperature raising by 1 or 2 degrees
(Horne & Staff, 1983; Horne & Moore, 1985). Horne proposes that a rise
in brain temperature leads to increased metabolic rate, resulting in a need
for more restoration during S-sleep. If this causal path could be demonstrated
then it would be strong evidence against purely adaptive theories. (Alternatively,
the temperature rise could simply be disrupting sleep control mechanisms).
In conclusion, what has been learned? Sleep is a complex phenomena which possibly serves many functions which may vary across different species. Serious sleep research has only been going for some fifty years, and is still in its infancy. As new facts become apparent about the complex nature of sleep - often through new and more sophisticated measurement techniques - the data base becomes larger, and theories must adapt or die. Unfortunately, most theories appear to have been dealt fatal blows - though adaptive theories and the ontogenetic hypothesis appear to be least harmed. Given the general move over the past decade away from functional research towards investigation of sleep mechanisms and disorders, the immediate prospects do not look too bright for uncovering the mysteries of why we sleep and why we dream.
Carlson, N.R., (1991). Physiology of Behaviour. Boston: Allyn & Bacon.
Crick, F., & Mitchison, G. (1983). The function of dream sleep. Nature, 304, 111-114.
Hobson, J. A., (1988). The Dreaming Brain. New York: Basic Books.
Hobson, J. A., & McCarley, R. W. (1977). The brain as a dream state generator. American Journal of Psychiatry, 134, 1335-1368.
Horne, J. A., (1988). Why We Sleep: The Functions of Sleep in Humans and Other Animals. Oxford: Oxford University Press.
Lavie, P., Pratt, H., Scharf, B., Peled, R., & Brown, J. (1984). Localized pontine lesion: nearly total absence of REM sleep. Neurology, 34, 1118-1120.
Marczynski, T. J., Burns, L. L., & Monley, C. A. (1992). Empirically derived model of the role of sleep in associative and recuperative processes. Neural Networks, 5, 371-402.
Mayes, A. R., ed, (1983). Sleep Mechanisms and Functions in Humans and Animals: an Evolutionary Perspective. Wokingham: Van Nostrand.
Meddis R. (1983). The evolution of sleep. In Sleep Mechanisms and Functions in Humans and Animal, edited by A. Mayes. Wokingham: Van Nostrand.
Roland, P. E., (1985). Metabolic measurements of the working frontal cortex in man. Trends in neurosciences, 7, 430-435.
Toates, F. (1980). Animal Behaviour. A Systems Approach. Chichester: Wiley.