How Naturalistic Music Listening Encodes Acoustic and Emotional Features in the Brain and Is Shaped by Extra-Musical Context

Emotional responses to music arise from an interplay between low-level acoustic cues, predictive and reward-related mech- anisms, and top-down contextual interpretation. However, most neuroimaging approaches capture only fragments of this process, relying on constrained stimuli and models that do not generalize well to naturalistic listening. This dissertation pro- vides three complementary studies on the link between music and emotion in the neuroscience domain. These studies move from naturalistic, idiographic frameworks, through feature- specific functional connectivity, to top-down contextual fram- ing of felt emotions, outlining a progression from bottom-up to top-down mechanisms by which music and emotions are en- coded in the brain and shaped by extra-musical information. The three studies, illustrated in the thesis, are organized in a multi-layered, bottom-up fashion to understand how natural- istic music listening can elicit emotions and how they manifest in the brain. At a first level, I address the challenge of group- level inference in fMRI when voxel-wise encoding models are spatially idiographic, and directionality is not shared across participants. To this end, I validate a Non-Parametric Combi- nation framework for one-sample, unsigned statistics that en- ables population-level voxel- and cluster-wise inference while preserving subject-specific encoding patterns, and apply it to fMRI data acquired during music listening. At a second level, I introduce BOSS (Brain Orchestra Synchronization Study), a modular toolbox for functional connectivity that quantifies the relative contributions of acoustic and emotion-related fea- tures to connectivity with auditory and affective-related brain regions. In the final study, I investigate how extra-musical information shapes emotional experience by manipulating the valence of brief textual framings presented before symphonic listening in an online naturalistic paradigm, thereby quanti- fying how top-down interpretive cues bias the trajectory of felt affect. Across all three levels, the work prioritizes nat- vii uralistic stimuli so that statistical evidence accumulates over extended listening and generalizes beyond short, constrained excerpts. Together, these studies offer an integrated account of how music and emotion are encoded, integrated, and con- textually shaped in the brain, and they motivate future work on how controlled emotional listening may reveal individual differences in both voxel-wise encoding and functional connec- tivity.