Short-window spectral analysis of cortical event-related potentials by adaptive multivariate autoregressive modeling: data preprocessing, model validation, and variability assessment

In this article we consider the application of parametric spectral analysis to multichannel event-related potentials (ERPs) during cognitive experimen ts. We show that with proper data preprocessing, Adaptive MultiVariate Auto Regressive (AMVAR) modeling is an effective technique for dealing with nons tationary ERP time series. We propose a bootstrap procedure to assess the v ariability in the estimated spectral quantities. Finally, we apply AMVAR sp ectral analysis to a visuomotor integration task, revealing rapidly changin g cortical dynamics during different stages of task processing.