In this thesis, I exploit the use of intentionality in music, particularly in performance systems. I have created a model for an ensemble of performers. This model stresses intention and negotiation to cooperatively add to the musical performance of multiple agents. My thesis project is a limited version of this ENSEMBLE model, focusing on three basic intentionalities: changing tempo, changing dynamics, and negotiating for a solo performance. The problem being addressed is the lack of a cognitive basis for current representations of music. My system can produce musical output, or modify its planned events on the basis of interpretations of other performers intentions. As a result of this research, I have formalized criteria for interpreting clues that result in belief. These clues represent gestures and musical events. The system is implemented in ORBS, an object-oriented, rule-based programming language developed for artificial intelligence applications. A performance is more than just the production of notes. This methodology defines a performance as the visual and musical effort, and the score as only a blueprint.