Decoding power-spectral profiles from FMRI brain activities during naturalistic auditory experience

Area of Science:

• Neuroscience
• Auditory Neuroscience
• Computational Neuroscience

Background:

• Computational acoustic features correlate with neural brain activities.
• Understanding auditory information processing is advancing.
• Naturalistic paradigms offer insights into complex auditory encoding.

Purpose of the Study:

• To investigate if power spectral density (PSD) profiles can be decoded from brain activity during naturalistic auditory experiences.
• To examine the relationship between neural decodability and PSD profile characteristics like power intensity and frequency deviants.
• To validate the use of naturalistic paradigms for studying neural encoding of complex auditory information.

Main Methods:

• Utilized a high-resolution functional magnetic resonance imaging (fMRI) dataset.
• Participants freely listened to the audio-description of the movie "Forrest Gump".
• Employed clustering to identify representative PSD profiles and Support Vector Machine (SVM) classifiers to decode PSD profiles from fMRI data.

Main Results:

• Demonstrated that PSD profiles can be reliably decoded from brain activities.
• Identified a sigmoidal relationship between neural decodability and power intensity deviants of PSD profiles.
• Confirmed the feasibility and advantages of naturalistic paradigms for auditory neuroscience research.

Conclusions:

• Power spectral density profiles are decodable from neural activity during naturalistic auditory perception.
• Neural decodability exhibits a sigmoidal relationship with power intensity variations.
• Naturalistic auditory paradigms are effective for studying complex auditory information processing in the brain.