Characterizing dynamic functional connectivity subnetwork contributions in narrative classification with Shapley...

Area of Science:

• Neuroscience
• Cognitive Science
• Machine Learning Applications in Brain Imaging

Background:

• Functional magnetic resonance imaging (fMRI) is crucial for studying brain activity.
• Dynamic functional connectivity analysis models temporal brain networks.
• Understanding narrative comprehension requires exploring brain network dynamics.

Purpose of the Study:

• To model dynamic functional connectivity during narrative tasks as temporal brain networks.
• To classify narrative modality and content using a supervised machine learning model.
• To investigate subnetwork contributions to narrative comprehension using Shapley values.

Main Methods:

• Modeled dynamic functional connectivity from fMRI data during narrative tasks.
• Employed a supervised machine learning model for classification of narrative features.
• Utilized Shapley values to analyze subnetwork contributions within Yeo parcellations.

Main Results:

• Successfully classified narrative modality and content using the machine learning model.
• Identified specific subnetwork contributions to understanding narrative modality and content.
• Demonstrated the involvement of the temporal parietal subnetwork in narrative comprehension.

Conclusions:

• The study provides novel insights into the functional aspects of the brain during narrative tasks.
• Narrative schematic representations may emerge from bottom-up processing driven by the temporal parietal subnetwork.
• Findings challenge the notion that narrative comprehension relies solely on top-down processes and pre-existing knowledge.