Statistical evaluation of slow brain potentials on the basis of single trials

Abstract The amplitude of slow brain potentials is related to the complexity of the task performed. Therefore, the large potentials related to the processing of complex tasks (tracking, writing, mental acts) may be analysed statistically on a single trial basis. This opens up a range of new facilities for within-session tests of significance. Prior to evaluation, signals are passed through several automatic error compensation routines: smoothing, trend removal, subtraction of eye artifacts, rejection of trials with random artifacts. The shape of the task potential approximates well to a rectangular elevation of cortical negativity, with the mean amplitude over a pre-event baseline being used as the main statistical parameter. Non-parametric statistical tests are preferred for evaluation. Application of the method is demonstrated for a tracking paradigm.