A simple clustering approach to map the human brain's cortical semantic network organization during task

Area of Science:

• Neuroscience
• Cognitive Science
• Network Science

Background:

• Understanding brain function relies on mapping large-scale brain networks during cognitive tasks.
• Partitioning brain networks aims to group functionally similar regions, but this is complex as regions often serve multiple functions.

Purpose of the Study:

• To propose and validate a novel clustering method for partitioning large-scale brain networks based on specific cognitive functions.
• To use semantic representation as a target cognitive function to evaluate the proposed method.

Main Methods:

• Analyzed functional magnetic resonance imaging (fMRI) data from 11 subjects exposed to 672 concepts.
• Correlated fMRI data with semantic rating data and utilized multidimensional semantic activation clustering.
• Validated network partitioning robustness using multiple methods and compared results with resting-state and traditional task-based networks.

Main Results:

• Identified distinct and reliable semantic brain networks with high cross-model consistency (semantic ratings, GPT-2 word embeddings).
• Demonstrated that these semantic networks differ significantly from resting-state and traditional task-based networks.
• Revealed functional differences among semantic networks regarding semantic representation capabilities, information modalities, and general cognitive domain associations.

Conclusions:

• Introduced a novel, function-specific approach for brain network analysis, establishing a standard semantic parcellation with seven networks.
• The findings provide a robust framework for future research into complex cognitive processes and their neural underpinnings.
• This method enhances understanding of brain organization by tailoring network analysis to specific cognitive functions.