Jan R. Wessel

Temporal independent component analysis (ICA) is applied to an electrophysiological signal mixture (such as an EEG recording) to disentangle the independent neural source signals-independent components-underlying said signal mixture. When applied to scalp EEG, ICA is most commonly used either as a pre-processing step (e.g., to isolate physiological processes from non-physiological artifacts), or as a data-reduction step (i.e., to focus on one specific neural process with increased signal-to-noise ratio). However, ICA can be used in an even more powerful way that fundamentally expands the inferential utility of scalp EEG. The core assumption of EEG-ICA-namely, that individual independent components represent separable neural processes-can be leveraged to derive the following inferential logic: If a specific independent component shows activity related to multiple psychological processes within the same dataset (e.g., elicited by different experimental events), it follows that those psychological processes involve a common, non-separable neural mechanism. As such, this logic allows testing a class of hypotheses that is beyond the reach of regular EEG analyses techniques, thereby crucially increasing the inferential utility of the EEG. In the current article, this logic will be referred to as the 'common independent process identification' (CIPI) approach. This article aims to provide a tutorial into the application of this powerful approach, targeted at researchers that have a basic understanding of standard EEG analysis. Furthermore, the article aims to exemplify the usage of CIPI by outlining recent studies that successfully applied this approach to test neural theories of mental functions.