Causality in variance in electrophysiological data using the ARCH model

Measurements of electrophysiological activity can be used to infer interactions between different regions of the human brain. In this work, we consider the use of an autoregressive conditional heteroscedasticity (ARCH) model to estimate causality in variance between different brain regions in simulation and continuously measured EEG data. We propose an efficient new algorithm for ARCH model estimation and demonstrate that the proposed approach provides promising results that are distinct from the causality estimates obtained from simpler and more conventional signal causality models.