Non-linear relationships between the auditory mismatch responses and the level of inharmonicity of complex sounds

Predictive processing accounts of perception suggest that the brain is using an internal generative model to form predictions about the incoming stimuli. Precision weighting is an important facet of predictive processing theories and has been extensively studied with mismatch responses such as mismatch negativity (MMN). Harmonicity is a feature of sound that is important for auditory perception and previous research has shown that it modulates the brain mismatch responses. Since inharmonic sound spectra contain more information (have higher information entropy), inharmonicity has been suggested to be involved in precision weighting. In this study we explored this issue by parametrically modulating the level of inharmonicity applied to synthetic sounds and recording MMN and P3a responses from healthy volunteers. Our results show that a sigmoid function models the relationship between inharmonicity and MMN amplitude better than any linear or polynomial function. Furthermore, P3a amplitude has an inverted-U relationship with inharmonicity and peaks at inharmonicity levels just below the threshold for pitch discrimination. These results are consistent with the hypothesis that inharmonicity impairs F0 extraction above a certain threshold and does not serve as an index of precision in the auditory system.