Toward an Objective Measure of Listener Engagement with Natural Music Using Inter-subject Eeg Correlation

This study extends existing research on inter-subject correlations (ISCs) of brain responses as a measure of engagement, with a focus on the relationship between the structural coherence of a musical stream and the listener’s degree of engagement with it. EEG was recorded while subjects listened to naturalistic music (popular Hindi songs) presented in original versions and in temporally disrupted, phase-scrambled conditions. ISCs were computed from the EEG data using Reliable Components Analysis. Overall, original versions of songs yielded significantly higher ISCs than phase-scrambled versions, and were also rated as more pleasant, well ordered, and interesting by subjects. The most reliable spatial component extracted from responses to the original songs concurs with past EEG findings involving naturalistic music. The time course of the ISCs is resolved at a musically relevant time scale. The sum of our findings suggests that ISCs show promise toward finding time-critical measures of engagement and attention in typically noisy EEG signals, and, specifically, as a means to find correlations between structural features of music and brain responses in listeners. We discuss possible links between heightened ISCs and regions of musical interest, and implications for future research.