Aarhus University Seal / Aarhus Universitets segl

CFIN & MIB guest talk: Marcus Pearce

Marcus Pearce from Queen Mary University of London gives a talk on Statistical Learning and Probabilistic Prediction in Musical Style Acquisition.

28.11.2016 | Hella Kastbjerg

Dato man 12 jun
Tid 13:30 14:30
Sted Meeting room 5th floor, DNC Building 10G, Aarhus University Hospital, Nørrebrogade

Title:
Statistical Learning and Probabilistic Prediction in Musical Style Acquisition

Abstract:
Drawing inspiration from both the 19th Century musicologist Eduard Hanslick and post-war development of the mathematical foundations of information theory by Shannon and Weaver, Leonard Meyer proposed in the 1950s that musical styles can be viewed as complex systems of probabilistic constraints which are acquired through experience by composers, performers and listeners thus enabling successful communication of musical meaning. However, until recently, there was surprisingly little detailed scientific work attempting to establish this proposal. I will introduce a probabilistic model of auditory expectation which learns through unsupervised exposure to music and generates musical expectations that can be characterised in information-theoretic terms. Empirical studies have successfully tested the efficacy of the model in accounting for the cognitive process of generating melodic pitch expectations during musical listening. I will discuss ongoing work that extends this research programme in three directions. First, extending the representations used by the model to account for musical parameters other than pitch including rhythm and harmony. Second, extending the computational approach to account for other cognitive processes in music perception such as recognition memory, similarity perception, grouping structure and emotional response. Third, extending the range of the populations covered to investigate and simulate perceptual differences between individuals with varying degrees of exposure and training in different musical styles.

Forskning, Alle grupper, CFIN, Musicinthebrain, Seminar, Musicinthebrain