"The Sleep Ramp"
REM sleep doesn't arrive on a fixed schedule. Its probability depends on how long NREM sleep has been accumulating. Akhavan and colleagues build a metric โ REM propensity โ that rises with NREM duration, peaks, then falls. The peak isn't just a statistical artifact: it predicts how long the subsequent REM episode will last and whether the animal will continue cycling through sleep stages.
The ramp holds across species with radically different sleep architectures. Humans consolidate sleep into one long block; mice and rats sleep in fragmented bursts throughout the day. The propensity curve looks the same. NREM duration drives REM timing regardless of whether the NREM accumulates in a single stretch or in scattered episodes.
This cross-species invariance constrains the mechanism. Whatever biological process tracks NREM duration and triggers REM must operate the same way in consolidated and fragmented sleepers. The most parsimonious explanation: a process that integrates time-in-NREM and resets after each REM episode. The integration is the clock. The reset is the discharge.
The decay past the peak is the least intuitive finding. After enough NREM, the probability of entering REM actually decreases. This could mean that very long NREM bouts shift the brain into a different state โ one where the accumulating REM pressure is overridden by sustained deep sleep. The ramp has a maximum, and past the maximum, the system stabilizes rather than continuing to build pressure. The propensity isn't a reservoir that fills without limit. It fills, peaks, and then the system accommodates.
