Time-warped representational similarity analysis reveals acoustic contributions to musical pleasure are weakly shaped by autonomic neural inputs.

Most people enjoy music and often use it to regulate their emotions. Although pleasure derived from music-listening has been shown to be mediated by dopaminergic signals in the mesolimbic reward network, its relationship with physiology has been inconsistent. Here, we introduced time-warped representational similarity analysis (twRSA) to directly map dynamic representations of multiple modalities across variable-duration stimuli. Our method revealed that while time-varying spectral and tonal acoustic features predicted changes in autonomic neural responses (measured via cardiac, pupil, and respiratory activity) during music-listening, only a small subset was relevant to listeners' on-line pleasure ratings. Furthermore, structural equation modeling showed that the effect of acoustic changes on musical pleasure was only weakly mediated by physiological changes. Our results thus suggest that while musical pleasure may be embodied in bodily responses, the mapping between subjective experience and autonomic neural activity is likely highly context-dependent and weak at best.