More interesting research from Northwestern's Insight group...
Insight's (Aha!) way of solving problems is different from deduction because insight solutions are not fully conscious. When we solve problems by insight, we there's a period of incubation (when we seem to turn away from the problem), then suddenly answers appear, perhaps fully formed. Unlike deduction, with insight there may be no causal chain or line of individually thought out steps.
In previous work (top row of figure below), Kounios and Jung-Beeman found that the right temporal lobe seem to have a special role in verbal problem solving by insight. Now this group seems to have found that there's a favorable preparatory state in the brain just before the insight solution appears, and this state seems to allow scientists to predict whether an answer will be arrived at by insight or by conscious problem solving (bottom row of figure below). For non-insight solutions, the brain looks like it's visually attentive, or looking at the question expectantly. Interestingly, for insight solutions, this external looking seems to turn off, and internal looking (looking within, using frontal lobe reflection, stored temporal lobe memory files) turns on.
Maybe Einstein was rather smart when he reflected, "I do my best thinking after playing the violin for a half hour."
Prepared Mind and fMRI
PLoS Biology: Neural Activity When People Solve Verbal Problems with Insight
Eide Neurolearning Blog: Easy and Hard Problem Solving