Recovering Whole-Brain Causal Connectivity under Indirect Observation with Applications to Human EEG and fMRI

ABSTRACT

Recovering the true causal connectivity of the human brain from indirect observations like EEG and fMRI is a fundamental but ill-posed problem due to the presence of unobserved confounders and the low temporal resolution of imaging. We propose a new causal inference framework that utilizes state-space modeling and structural constraints to recover directed connectivity patterns from multi-modal neuroimaging data. By explicitly modeling the latent neural dynamics and the measurement processes, our method can distinguish between direct causal influences and spurious correlations. We validate our approach on synthetic data and demonstrate its utility in identifying disease-specific connectivity changes in human EEG and fMRI datasets, providing a more reliable map of the brain's functional architecture.