MOMENT ANALYSIS OF EEG AMPLITUDE HISTOGRAMS AND SPECTRAL-ANALYSIS - RELATIVE CLASSIFICATION OF SEVERAL BEHAVIORAL TASKS

Previous studies indicate that EEG amplitude probability density funct ions are Gaussian (normal) during rest and non-Gaussian during perform ance of mental tasks. In the present study we compared measures of nor mality, including higher central moments (e.g., skewness, kurtosis) an d relative spectral power, to classify data sampled from several diffe rent behavioral tasks (resting eyes closed and mental arithmetic). Ana lysis shows significant classification in 22 of 25 subjects, based upo n a total of 46 EEG variables. However, only two of these variables in volved Gaussian properties of the amplitude distribution, Relative spe ctral power, on the other hand, contributed 33 predictor variables in delta, theta, alpha, and beta frequency bands (alpha was the best sing le predictor). These results lend support to studies demonstrating the robustness of EEG relative spectra but cast doubt upon the utility of Gaussian patterns in EEG amplitude distributions as predictors of beh avioral states.