Now that we’ve pretty much established that sleep is crucial for consolidating memory, the next question is how much sleep we need.

A new study compared motor sequence learning in 16 people with mild obstructive sleep apnea to a matched control group (also attending the sleep clinic). There were no significant differences between the groups in total sleep time, sleep efficiency and sleep architecture (time spent in the various sleep stages), subjective measures of sleepiness, or performance on a psychomotor vigilance task (a task highly sensitive to sleep deprivation).

Nor were there any differences in learning performance during the training phase on the motor task.

But the interesting thing about consolidation is that skills usually improve overnight — your performance the next day will usually be better than it was at the end of your training. And here there was a significant difference between the groups, with the controls showing much greater overnight improvement on the motor sequence task. For sequences learned in the morning and tested 12 hours later on the same day, however, there were no differences between the groups.

So given all the factors relating to sleep that were the same between the two groups, what was the factor behind the group consolidation difference? It turns out it was (principally) the arousal index (arousals were scored on the basis of abrupt shifts in EEG frequency that last at least 3 seconds with 10 seconds of stable sleep preceding), and to a lesser extent the apnea-hypopnea index.

It seems likely, then, that arousals from sleep may (depending, presumably, on timing) interrupt the transfer of labile memories from the hippocampus to the neocortex for long-term storage. Thus, the more arousals you have, the more likely it is that this process will be interrupted.