Improving source localization of event-related potentials in the GO/NOGO task by modeling their cross-covariance structure

Nineteen-channel event-related potentials (ERPs) were studied in the GO/NOGO test in 517 healthy subjects. Multi-channel ERPs were decomposed into independent components by the method of modeling the cross-covariance structure of transient process. The research results showed that low-correlated ERP components could be obtained by this method with acceptable reliability. The identified components were associated with psychological processes, such as the attended sensory mismatch operation, decision making on the subsequent action, the action inhibition operation, and the conflict monitoring operation. In conclusion, decomposition of ERP recordings into independent components and the use of sLORETA help to localize ERP sources more accurately than the conventional ERP analysis.