Practice and sleep form different aspects of skill.

Because I am a pianist I find this work by Song and Cohen totally fascinating. It conforms to my own experience in learning new note sequences in a piano composition (currently I'm working on Scriabin's Etude Op. 8 no. 12 D sharp minor). A fingering sequence that I find difficult I can discover to be playing in head during momentary waking during the night, and on the next day the notes come much more easily. Song and Cohen's distinction of transition and ordinal forms also matches with my experience of being able to verbalize (declarative memory) versus 'just knowing' (procedural memory) a passage.

Performance for skills such as a sequence of finger movements improves during sleep. This has widely been interpreted as evidence for a role of sleep in strengthening skill learning. Here we propose a different interpretation. We propose that practice and sleep form different aspects of skill. To show this, we train 80 subjects on a sequence of key-presses and test at different time points to determine the amount of skill stored in transition (that is, pressing ‘2’ after ‘3’ in ‘4-3-2-1’) and ordinal (that is, pressing ‘2’ in the third ordinal position in ‘4-3-2-1’) forms. We find transition representations improve with practice and ordinal representations improve during sleep. Further, whether subjects can verbalize the trained sequence affects the formation of ordinal but not transition representations. Verbal knowledge itself does not increase over sleep. Thus, sleep encodes different representations of memory than practice, and may mediate conversion of memories between declarative and procedural forms.