In the 1950s, scientists discovered that sleep is an active process, not, as formerly believed, a passive state not worthy of scientific investigation. By recording brainwave activity (the electroencephalogram, EEG), eye movements, and muscle tension, it was determined that sleep is broken down into two major states, rapid eye movement (REM) and nonrapid eye movement (NREM). NREM sleep is further subdivided into sleep stages 1 through 4, which represent a continuum from light sleep (stage 1) to deep sleep (stages 3 and 4). REM sleep is characterized by rapid eye movements similar to those seen in the waking state, as well as muscle atonia.

Healthy human individuals show stable and distinct patterns of sleep architecture, that is, the cyclic alternations of the different sleep stages. Sleep is entered through the transitional stage 1, and followed by stages 2, 3, and 4, respectively. After about 90 min, through stage 2 sleep, REM sleep begins. The first appearance of REM sleep during the night is termed REM latency. This 90- to 100-min cycle repeats itself typically four to five times during the night. However, the internal temporal structure changes, so that at the beginning of the night stages 3 and 4 are long while REM sleep is short, and as the night progresses, the amount of stages 3 and 4 decreases, while the amount of REM sleep increases.

Both REM sleep and non-REM sleep stages 3 and 4 are homeostatically driven; that is, selective deprivation of each of these states subsequently causes a rebound in their appearance once the person is allowed to sleep. This finding leads to the ubiquitous assumption that both are essential in the sleep process and its many functions. REM sleep may also be driven by a circadian oscillator, as studies have shown that REM is temporally coupled with the circadian rhythm of temperature.