Modeling of Perceived Musical Rhythms Using Electrocorticography

Numerous studies have explored the neural correlates of musical rhythms using various neuroimaging modalities. Non-invasive neuroimaging modalities lack either the spatial or temporal resolution to reveal the nuances of neural processes involved in perception of musical rhythms. Intracranial recordings of electrophysiological activity such as electrocorticography (ECoG) can jointly provide spatial and temporal resolution for improved characterization and modeling of the underlying processes. The present study examines anticipatory and perceptual models that use ECoG recordings to estimate simple perceived and imagined musical rhythms in human participants. The resulting models are characterized and compared across participants. The results show that the anticipatory and perceptual models can reconstruct the auditory stimulus envelope with statistically-significant correlations when trained and tested on independent listening data. However, these models are unable to reliably reconstruct the expected rhythm pattern when trained on listening data and applied to imagining data. This suggests, similar to recent findings in overt and imagined speech decoding using intracranial signals, that there are likely distinct neural substrates activated during listening and imagining of musical rhythms.