The polyphonic brain:Extracting the neural representation of rhythmic structure for separate voices of polyphonic music using ERPs

Rapid changes in the stimulus envelope (indicating tone onsets) elicit an N1-P2 ERP response, as has been shown for clicks and sine waves, musical tones and for speech. Canonical Correlation Analysis with temporal embedding (tkCCA), a multivariate correlation-based method, allows to extract brain responses to these changes in continuous auditory stimuli. Here, we(1) probe, whether tkCCA can be applied to track changes in the stimulus envelope in the EEG of subjects who were presented with semi-artificial monophonic music clips of three instruments. (2) On polyphonic trials, composed of the same parts as in 1), we explore, whether the tkCCA-filters 10 20derived in 1) can recover a representation of each instrument’s part from the EEG where subjects listened to the polyphonic stream. (3) We explore, whether, eventually, such a representation is influenced by focused attention