Sleep is good for muscle memory. Specifically, REM sleep has been associated with consolidation of non-declarative (muscle) memories.
I first heard of this when I was learning to do a Mohawk turn when ice skating, and it came back to me last night as I was deciding whether to finish watching Craig Ferguson or pack it in for the night. I packed it in.
This morning I found a few discussions of this phenomenon on the web, including one regarding speed skating, and another in Wikipedia.
I learned from these entries that, surprisingly, a good sleep can allow you to perform better than your performance during practice the previous day. According to the model advanced by the speed skating reference above, after you repeatedly activate a nerve pathway by practicing a movement over and over, the nerve cells grow additional synapses and receptors, and also synthesize more neurotransmitter to move the signal between them. This growth takes time (hours), but after this delay there is a performance gain. There is apparently some debate as to whether time alone is enough for this effect, or whether REM is essential.
Sorry Craig, but if I've been practicing new skills on the unicycle earlier in the day, I'll be sleeping through your monolog.
