Skip to main content

Hierarchical processing of temporal information during naturalistic music production and perception

Hierarchical processing of temporal information during naturalistic music production and perception

Session 1

Presenter Name:Riesa Cassano

School/Affiliation:University of Rochester

Co-Authors:Jamal Williams, Marius Cătălin Iordan, Uri Hasson, Elise Piazza

Abstract:

Recent work using naturalistic stimuli—such as movies, verbal narratives, and music—has used scrambled versions of these stimuli to reveal hierarchies of temporal processing in the brain. A new fMRI dataset of expert pianists sight-reading naturalistic music scrambled at different timescales offers the opportunity to investigate hierarchical processing during music production. Intersubject correlation is used to measure the reliability of responses in a region. We expected early auditory areas to have reliable responses across all scrambled conditions, whereas higher-order areas should have more reliable responses for relatively more intact conditions. Surprisingly, both early auditory cortex and motor cortex showed sensitivity to the effects of scrambling: intersubject temporal correlation increased systematically for more intact conditions. However, when comparing average intersubject pattern correlation across varying segment lengths within the context of the “intact” condition, we found a more expected hierarchical effect: early auditory and motor regions showed reliable response patterns over shorter timescales (phrases and half-phrases), whereas higher-order regions (posterior medial cortex) showed more reliable response patterns over longer timescales (half-sections). This analysis may be more sensitive to hierarchical effects due to asymmetries in attentional load across the scramble conditions. During listening, response patterns are more reliable over longer timescales (half-sections and sections) in both auditory areas and higher-order areas. This suggests that subjects may devote more attentional resources to longer-term prediction and chunking during listening compared to playing. These analyses are a first attempt to characterize the hierarchical processing of highly natural musical structure during live performance.

Poster PDFPoster PDF Meeting LinkMeeting Link