Music can be reconstructed from human auditory cortex activity using nonlinear decoding models.

Area of Science:

• Neuroscience
• Auditory Perception
• Brain-Computer Interfaces

Background:

• Music perception is fundamental to human experience, but its neural underpinnings are not fully understood.
• Previous research has primarily focused on broader auditory processing, leaving specific musical dynamics unclear.

Purpose of the Study:

• To investigate the neural dynamics of music perception using intracranial electroencephalography (iEEG).
• To apply stimulus reconstruction techniques, adapted from speech processing, to analyze music decoding from neural data.
• To identify brain regions and organizational principles involved in processing musical elements like rhythm.

Main Methods:

• Analysis of a unique iEEG dataset from 29 patients listening to a Pink Floyd song.
• Application of a stimulus reconstruction approach to decode musical information from neural recordings.
• Combined encoding and decoding analyses to assess neural responses and predictive accuracy.

Main Results:

• Successful reconstruction of a recognizable song from direct neural recordings.
• Quantification of factors influencing music decoding accuracy.
• Evidence of right-hemisphere dominance in music perception, particularly involving the superior temporal gyrus (STG).
• Identification of a novel STG subregion specialized for musical rhythm processing.
• Definition of an anterior-posterior organization within the STG, showing distinct sustained and onset responses to musical elements.

Conclusions:

• Predictive modeling is feasible even with short datasets from single patients.
• The superior temporal gyrus plays a crucial role in music perception, with distinct subregions and organizational patterns.
• These findings open avenues for integrating musical elements into brain-computer interface (BCI) applications.