studied.Much of our slumber is spent in slow-wave sleep, also known as stage 3 or deep sleep (see diagram),during which there are easily detectable waves of electrical activity across the whole brain, caused byneurons firing in synchrony about once a second. This is interspersed with other phases, includingrapid-eye-movement sleep, where brain activity resembles that seen during wakefulness, andtransitional stages between the two states.It is slow-wave sleep that is generally thought to do whatever it is that sleep actually does. As well asappearing to be the most different to the brain's waking activity, the waves are larger at the beginningof sleep, when sleep need is presumably greatest, and then gradually reduce. And if you go withoutsleep for longer than usual, these slow waves are larger when you do eventually nod off.Explanations for sleep fall into two broad groups: those related to brain repair or maintenance, andthose in which the sleeping brain is thought to perform some unique, active function. There has beenspeculation over the maintenance angle for over a century. It was once a fashionable idea that somekind of toxin built up in the brain during our waking hours which, when it reached a certain level, madesleep irresistible. Such a substance has never been found, but a modern version of the maintenancehypothesis says that during the day we deplete supplies of large molecules essential for the operationof the brain, including proteins, RNA and cholesterol, and that these arereplenished during sleep. Ithas been found in animals that production of such macromolecules increases during slow-wave sleep,although critics point out that the figures show a mere correlation, not that levels of these moleculescontrol sleep.The unique function school of thought also has a long pedigree. Sigmund Freud proposed that thepurpose of sleep was wish fulfilment during dreaming, although scientific support for this notion failedto materialise.There is good evidence, however, for sleep mediating a different kind of brain function - memoryconsolidation. Memories are not written in stone the instant an event is experienced. Instead, initiallylabile traces are held as short-term memories, before the most relevant aspects of the experience aretransferred to long-term storage.
Several kinds of experiment, in animals and people, show that stronger memories form when sleeptakes place between learning and recall. Some of the most compellingsupportfor this idea camewhen electrodes placed into rats' brains showed small clusters of neurons "replaying" patterns ofactivity during sleep that had first been generated while the rats had been awake and exploring."Memory representations are reactivated during sleep," says Jan Born at the University of Tübingen inGermany.Many labs remain focused on how memory systems are updated during sleep, but since 2003 a newidea has been gaining traction. It straddles both categories of theory, concerned as it is with neuronalmaintenance and memory processing.The hypothesis concerns synapses, the junctions between neurons through which they communicate.We know that when we form new memories, the synapses of the neurons involved become stronger.The idea is that while awake we are constantly forming new memories and therefore strengtheningsynapses. But this strengthening cannot go on indefinitely: it would be too expensive in terms ofenergy, and eventually there would be no way of forming new memories as our synapses wouldbecome "maxed-out".The proposed solution is slow-wave sleep. In the absence of any appreciable external input, the slowcycles of neuronal firing gradually lower synaptic strength across the board, while maintaining therelative differences in strength between synapses, so that new memories are retained (see diagram).There is now much evidence to support what is known as the "synaptic homeostasis hypothesis". Inhumans, brain scans show that our grey matter uses more energy at the end of the waking day than at