<EM>FROM PREDICTION TO PLEASURE:</EM> INVESTIGATING THE NEURAL ENCODING OF OPTIMAL SURPRISE IN MUSIC

When listening to music, the human brain engages in a predictive process that is inherently rewarding. Driven by the intrinsic motivation to learn, this process relies on achieving an optimal balance between uncertainty and predictability in musical features. Yet, the specific brain circuits that encode “optimal” or “learnable” surprises remain unclear. To address this gap, we measured the neural activity of 100 participants with fMRI while they performed the novel MUSIc COmplexity Sensitvity (MUSICOS) task. MUSICOS is a decision-making task that orthogonalizes musical surprise (information content) and uncertainty (entropy) while simultaneously capturing preference-based behavioral measures. Behaviorally, participants exhibited the expected Wundt effect, an inverted U-shaped relationship between information content (IC) and preference, with optimal preferences shifting toward larger or smaller surprises depending on entropy. Importantly, fMRI analyses revealed that both the Wundt effect of IC alone and the interaction between IC and entropy are encoded in the right anterior superior temporal gyrus (aSTG), mirroring the pattern of “learnable” surprises. This suggests that the aSTG serves as a core auditory hub for detecting moderate, context-dependent surprises, thereby linking auditory processing with reward circuitry in music listening.