fMRI Transfer Learning Maps Brain Task Relations Across 23 States

Researchers trained 1,127 fMRI models to map how cognitive tasks transfer knowledge across the brain. Using Boolean Integer Programming on 23 Human Connectome Project task states, they found that single-source transfer is directional and paradigm structured: motor states transfer well within the motor paradigm but provide limited support to most non-motor targets. Multi-source transfer depends on the composition of the source set. Across supervision budgets, Boolean Integer Programming repeatedly allocated direct supervision to several 0-back and 2-back working-memory states, even though these states are not consistently the strongest individual sources. These findings reveal a cross-paradigm-limited motor cluster and working-memory states with high priority under the specified global allocation objective, extending reconstruction-based fMRI taskonomy from one-to-one to many-to-one task relations and budget-constrained dependencies.