A night's sleep cycles through stages roughly every 90 minutes: light sleep (stages N1-N2, the majority of total sleep time, involved in initial transition and some memory processing), deep slow-wave sleep (N3, concentrated in the first half of the night, most associated with physical restoration and declarative memory consolidation), and REM sleep (increasingly dominant in later cycles, associated with vivid dreaming, emotional processing, and procedural/creative memory consolidation). This staged structure was first mapped in detail using EEG recordings starting in the 1950s, when researchers Aserinsky and Kleitman discovered REM sleep itself by noticing distinctive rapid eye movements and brain-wave patterns in sleeping subjects.
Because slow-wave sleep concentrates early in the night and REM concentrates later, cutting sleep short doesn't just reduce total sleep proportionally — it disproportionately cuts into REM sleep specifically, since that's back-loaded toward the final hours. This is one concrete reason why 'I'll just sleep a bit less' costs more than the raw hour lost suggests: it's not an even trim across stages, it's a targeted cut into the stage that happens last.
Aserinsky, a graduate student, was using a leftover EEG machine to monitor sleeping subjects — including, in the earliest sessions, his own young son — when he noticed periods during the night marked by bursts of rapid eye movement and a brain-wave pattern that looked almost like wakefulness, even though the subject was clearly still asleep. Working with his advisor Kleitman, he confirmed the pattern was consistent and connected it to vivid dream recall when subjects were woken during those specific periods. Their work established REM as a genuinely distinct sleep stage, not just a lighter phase of the same uniform process researchers had assumed sleep to be.