This article examines the phenomenon in which musical lines establish what Edward T. Cone calls virtual agents, making the argument that listeners are more likely to ascribe such agency to lines that have high information content (in the formal, information-theoretic sense). I use a computational model, which I have proposed elsewhere, to estimate the information content a listener would perceive when listening to a piece of music. I then compare this model’s output to traditional musical analyses of several string-quartet expositions, demonstrating a qualitative correlation between the information content of these expositions and the virtual agents therein. Additionally, I assemble a larger corpus of expositions, many of which deploy virtual agents in a similar manner. Using formal statistical analysis, I show that these similarities can be traced back to the expositions’ information content. Thus, both qualitative and quantitative evidence suggest a connection between information and agency in music.

This content is only available as a PDF.
You do not currently have access to this content.