I have used the Watch for sleep monitoring for 10 days now and I've noticed something peculiar about how sleep score is measured. I've had good scores, between 67 and 82 so far. The variety seems mostly caused by differences in "mental recovery" which is based on REM sleep - it ranged from 0% to 80% for me, much more than the other aspects.
Last night I had my shortest sleep, less than 6 hours, and only 3 sleep cycles (normally I have 4 or 5); yet I had my highest overall sleep score so far (82). I saw that this is due to the score for "mental recovery" which measures REM sleep. Last night I had the highest score for mental recovery so far. Why?
I had noticed before that sometimes, when I have had plenty of REM sleep, I still get a very low score for mental recovery. (e.g. 1 hour 52 min of REM sleep but mental recovery score 0%). I have now discovered why: if the watch registers that you wake up shortly after REM sleep, this is "punished" with a low score. Last night I slept late and short so I had no awake moments after REM, resulting in a high score. In nights where I had "awake" episodes after REM sleep, my mental recovery dropped steeply below the required 60%. The watch assumes that I have been disturbed in my sleep.
That's a bug/mistake. It's part of the normal sleep cycle to awake very shortly after an REM episode.At the end of REM sleep, sleep is always the lightest. But the watch assumes that you have been actively disturbed in your REM sleep. This causes the "awake" moments to have too much influence on the mental recovery score and total sleep score.
Good hypothesis. I have been getting low mental recovery scores despite long REM times (2hrs.) and a high % of that night's sleep (24%). I do wake up in the middle of dreams in the morning often, and as you suggest, this may be lowering the mental recovery score despite the volume of REM sleep.