Abstract
“1/f-like” power spectrum scaling is a ubiquitous feature of complex systems. In such scaling, the power spectrum of a given time series is dominated by an inverse power law, resulting in an inverse linear relation between log power and log frequency. The 1/f-like power spectrum scaling properties of human, resting eyes-closed and eyes-open EEG were examined. For both eyes-closed and eyes-open EEG, log power had a significant inverse linear relation with log frequency. The slope of this relation was not correlated with previously-calculated Grassberger-Procaccia dimension estimates for the same data, indicating that the EEG is not a 1/f-like stochastic process. Further, the degree of deviation from perfect log power versus log frequency linearity was related to power in the 8–12 Hz alpha frequency band. It is speculated that this deviation may be related to the low dimension of a chaotic alpha rhythm relative to other EEG rhythms.