On the extraction, classification and causal analysis of EEG signal components

Drawing on results from time series analysis, this thesis provides more theoretical guarantees for time-reversed Granger causality, a recently proposed approach which is more robust with respect to noise.