Caution on causality analysis of ERP data

The event-related potential (ERP) technique is one of the most widely used methods in cognitive neuroscience research to study the physiological correlates of sensory, perceptual and cognitive activity associated with processing information (1,2). An ERP eeg segment data usually involves three subsegments: pre-stimulus,stimulus and post-stimulus. Obviously, the mathematic models for pre-stimulus (or post-stimulus) and stimulus should be different. However, the general approach for ERP analysis in the literature always views the whole segment data as one model. In this paper we use Granger causality (GC) and recently proposed New causality (NC) methods to demonstrate that NC is more sensitive than GC to reveal true causality influence based on ERP data.