Sleep researchers have long tracked how much REM sleep a person gets, but a new study published in the journal Sleep argues that the more important number may be what's happening moment to moment inside it. Analyzing overnight brain recordings from 627 healthy adults, researchers found that the internal architecture of REM sleep degrades steadily with age — and does so faster in women than in men.
Two Kinds of REM, One Declining With Age
REM sleep isn't a single uniform state. It alternates between two microstates: "phasic" REM, marked by bursts of rapid eye movements and heightened brain activity, and "tonic" REM, a quieter, more stable interval in between. Phasic REM is thought to reflect the most active, dream-intensive processing the brain does overnight.
The research team — led by Celia Bertin and Jean-Baptiste Eichenlaub, with collaborators across French and Swiss institutions — pooled retrospective polysomnography data from five separate databases to assemble one of the largest lifespan samples of its kind: 290 women and 337 men aged 18 to 80, none with diagnosed sleep or neurological disorders.
As age increased, the proportion of REM sleep spent in the phasic microstate steadily declined, along with the number and density of eye movements themselves. That decline held up even after accounting for total REM sleep duration, meaning a person's overall REM time can look unremarkable on a standard sleep report while the "active" portion of it is quietly eroding underneath.
Why the Decline Hits Women Harder
The steepest finding involved sex differences. Women in the study showed a significantly sharper drop in both phasic REM percentage and eye movement density as they aged, compared with men — a divergence the researchers say indicates women's REM sleep deteriorates at a faster rate across the lifespan, not just that it starts or ends at a different baseline.
Spectral analysis of the brain-wave data added a second layer to the finding. In tonic REM, brain activity shifted toward higher-frequency bands with age. In phasic REM, the opposite happened: EEG activity progressively slowed across the decades. The two REM microstates aren't just declining in volume — their underlying neural signatures are changing shape.
A Possible Early-Warning Signal
The researchers frame these microstructure changes as a potential biomarker, one that could flag early neurodegenerative or psychiatric changes well before conventional sleep-stage measures — total REM minutes, REM latency — show anything unusual. Loss of REM atonia and disrupted REM microarchitecture have both been linked in prior research to early stages of conditions like Parkinson's disease and REM sleep behavior disorder, and the authors suggest that finer-grained tracking of phasic versus tonic REM could eventually sharpen those early-detection tools.
Because the analysis pooled five independently collected cohorts rather than relying on a single lab's data, the pattern held across different recording setups and populations — a step that strengthens confidence the effect is a real biological signal rather than an artifact of any one dataset.
What This Means for Patients
For now, phasic-versus-tonic REM breakdown isn't part of a standard sleep study report, and there's no consumer wearable that reliably distinguishes between the two. But the findings offer a useful reframe for anyone puzzling over "poor quality" sleep despite normal-looking sleep-tracker numbers: REM quantity and REM quality are not the same thing, and the gap between them appears to widen with age, particularly for women navigating the sleep disruptions common around perimenopause and menopause.
Patients with unexplained daytime grogginess, especially those with a family or personal history of neurodegenerative disease, may want to discuss overnight polysomnography — the gold-standard test that can, in principle, capture this microstructure — with a sleep specialist rather than relying solely on wearable-derived sleep-stage estimates. As this kind of finer-grained REM analysis moves from research labs toward clinical practice, it may eventually help identify at-risk patients years before symptoms of cognitive or neurological decline appear.
