Dave Munger has written two posts about music in his Cognitive Daily blog. Music and IQ relates recent research on whether musical training improves intelligence (the so-called Mozart Effect). Glenn Schellenberg presented much of this research at the Leipzig conference, making the very convincing case that music is a special case in intellectual development. By comparing music lessons with drama lessons, Schellenberg handles the objection that any extra-curricular activity would help. There is a big "But" in this study, though. The best predictor of future academic performance for grade-school kids is not IQ or current grades, but their social behavior. And acting lessons were the only activity that improved adaptive social behavior.
Some insight into how we develop preferences shows a study that found that the perception of familiarity is more important than actual familiarity on liking a piece of music. Listeners heard several pieces of classical music, followed by an hour-long pause. Then a randomized mix of new excerpts and the previous excerpts were played. The listeners indicated whether they liked each excerpt, and whether they had heard the excerpt before. They were much more likely to like the piece if they thought they heard it before, even when they hadn't. This supports the hypothesis that we like things that we are certain about, not things that create uncertainty.