Name that tune: decoding music from the listening brain.

Area of Science:

• Neuroscience
• Cognitive Science
• Music Perception

Background:

• Understanding music perception involves analyzing brain responses to auditory stimuli.
• Electroencephalography (EEG) offers a method to capture the temporal dynamics of neural activity.
• The time structure of music may provide unique neural signatures for decoding perception.

Purpose of the Study:

• To investigate the feasibility of decoding music perception from EEG signals.
• To determine if time-domain information, specifically event-related potentials (ERPs), can differentiate between musical fragments.
• To assess the generalizability of neural representations of music across participants.

Main Methods:

• Utilized electroencephalography (EEG) to record brain activity while participants listened to music.
• Classified perception of seven distinct, vocal-free musical fragments (approx. 3 seconds each) using only time-domain EEG data (ERPs).
• Performed both individual and cross-participant classification analyses on single and multiple trials.

Main Results:

• Achieved up to 70% accuracy in classifying individual musical fragments in a seven-class problem using single EEG trials.
• Reached 100% classification accuracy after six stimulus presentations when using multiple trials.
• Attained a maximum accuracy of 53% in cross-participant classification, suggesting shared neural representations of musical fragments.

Conclusions:

• EEG, particularly ERPs, can effectively decode the perception of distinct musical fragments.
• The findings support the existence of generalizable neural representations for music across individuals.
• Further research is needed to explore stimulus-specific factors influencing EEG responses to music.