Dreams occur only during REM sleep.

The discovery of rapid eye movement sleep in the 1950s seemed to offer an objective physiological marker for something previously accessible mainly through subjective reports. When researchers awakened sleepers during REM, they frequently obtained vivid accounts of dreams. Awakenings from other stages produced dream reports less often.

That strong association was easy to turn into an identity: REM sleep is dreaming. But the experimental method also provided the means to test the claim. Researchers who systematically awakened people during non-REM sleep found that they too could report imagery, thoughts and sometimes elaborate dream experiences.

Later neurophysiology and brain-imaging work reinforced the separation between a sleep stage and the conscious experience occurring within it. REM remains especially associated with vivid, emotional and story-like dreams, but it is neither necessary for every dream nor a guarantee that a remembered dream report will follow.

Dream research always has a measurement problem: investigators normally know that a dream occurred because a person wakes and reports it. Recall therefore depends on timing and memory as well as on the underlying sleep state.