Abstract
A number of important issues in the construction of cognitive models of music listening are explored. These include the essential starting-point of the model, the kinds of abstractions a model will make, and the features of the musical experience which a model preserves and those which it obscures. A number of approaches to modeling music listening are surveyed; these include linguistically-oriented models, those based on schemata, and simulation models, including those embodied in computer programs. Three main issues concerning the development and use of a model of music listening are considered: choice of a model's formalism, choosing which elements of cognition to include in a model, and ways in which a model may be evaluated. Special attention is given to the questions raised in modeling musical processes as computer programs. A computer model of melodic learning is described and evaluated as an example of how these considerations may be addressed.