Power-law scaling in human EEG: Relation to Fourier power spectrum

We discuss a method of analyzing spontaneous human EEG time series, which emphasizes scale-independent behavior. We use detrended fluctuation analysis to quantify the temporal fluctuations as a function of window width, and show how power-law scaling behavior is frequently manifest over two distinct temporal ranges. These ranges encompass time scales associated with meaningful aspects of cortical physiology. This paper shows a simple way of quantifying the existence of such scaling behavior, and determining the characteristic time scale which separates the two regions. By making a qualitative connection with the discrete Fourier transform, we show how the violation of scaling between the two regions is associated with the normal human alpha rhythm, but that the existence scale-independent behavior on either side of the alpha rhythm enables a succinct description of the complex dynamics not accessible in the Fourier power spectrum. © 2002 Elsevier Science B.V. All rights reserved.