Researchers use Graph Neural Networks to decode how the brain organizes visual categories through functional brain connectivity patterns

Scientists analyzed high-resolution 7T fMRI data from the Natural Scenes Dataset to map how brain networks represent visual categories like sports, food, and vehicles

A signed Graph Neural Network was trained to model both positive and negative interactions between brain regions, using sparse edge masking and class-specific saliency

The model successfully identified reproducible, biologically meaningful subnetworks along the ventral and dorsal visual pathways that process different visual information

This framework bridges machine learning and neuroscience by moving beyond individual neuron selectivity to understand how connectivity patterns represent visual processing