Multivariate Statistical Methods and Their Capability to Demarcate Psychophysiologically and Neurophysiologically Sound Frequency Components of Human Scalp EEG

Abstract

For many fields of applied electroencephalography—especially in psychology, psychiatry, psychophysiology and several neurosciences—defining frequency bands for human spontaneous scalp EEG measurement is a crucial and problematic decision. Modern basic EEG research, as represented by this book, usually directs only limited attention to this issue. Whereas increased interest in time domain, evoked, and topographical EEG phenomena is to be noted, research in the frequency domain has been declining. Today choice of frequency rationales more and more seems to become “a matter of footnotes” (John et al., 1988). In particular, the frequency band issue is usually treated as a marginality. For instance, the Methods volume of the revised Handbook of Electroencephalography and Clinical Neurophysiology (Gevins and Remond, 1987) does not offer more than a glimpse of the problem of frequency band methodology for human scalp EEG. Nevertheless, many researchers and practitioners of the disciplines mentioned above rely more and more on uniformly applied frequency band schedules as apparently self-evident rules of data reduction. Although these contrasting trends can be understood in terms of EEG research history and economy, severe objections from the scientific point of view must be brought forward.