Time-varying effective connectivity for investigating the neurophysiological basis of cognitive processes

This chapter describes the methodological advancements developed during the last 20 years in the field of effective connectivity based on Granger causality and linear autoregressive modeling. At first we introduce the concept of Granger causality and its application to the connectivity field. Then, a detailed description of both stationary and time-varying versions of Partial Directed Coherence (PDC) estimator for effective connectivity will be given. The General Linear Kalman Filter (GLKF) approach is described an algorithm, recently introduced for estimating the temporal evolution of the parameters of adaptive multivariate model, able to overcome the limits of existing time-varying approaches. Then a detailed description of the graph theory approach and of possible indexes which could be defined is given. At the end, the potentiality of the described methodologies is demonstrated in an application aiming at investigating the neurophysiological basis of motor imagery processes.