Abstract
Music analysis raises interesting problems for the theory of mental representation and meaning, and poses new challenges for epistemology. When an analysis purports to show the structure an analyst or reader hears a piece as having, what relation must thereby hold between hearing and analysis, and how does the analyst or reader know that it does? A paradox of analysis arises: if an analysis correctly captures the information content of a hearing, then it is bound to be uninformative. The solution is to distinguish different levels of content, where analysis and hearing share content on one level and diverge on another. The question then arises of how an analyst or reader knows that she hears a piece in the way described by an analysis. The nature of this (apparently a priori) knowledge is an important, and heretofore unappreciated, problem for epistemology. Music analysis is, finally, a fertile ground for investigating the age-old problem of the relation between perception and concepts.