EEG bands during wakefulness, slow-wave, and paradoxical sleep as a result of principal component analysis in the rat.

Rat EEG has been empirically divided in bands that frequently do not correspond with EEG generators nor with the functional meaning of EEG rhythms. Power spectra from wakefulness (W), slow-wave sleep (SWS), and paradoxical sleep (PS) of Wistar rats were submitted to Principal Component Analyses (PCA) to investigate which frequencies are covariant. Three independent eigenvectors were identified for SWS: a band between 1-6, an intermediate band between 7-15, and a fast band between 16-32 Hz (90.74% of the variance); two independent eigenvectors were extracted for PS: slow frequencies between 1-6 covarying together with frequencies between 11-16 Hz, and activity between 6-10 covarying together with fast frequencies between 17-32 Hz (80.38% of the variance); four eigen-vectors were obtained for W: 3-7, 8-9, 10-21 and 21-32 Hz (81.47% of the variance). Vigilance states showed significant differences in AP from 1 to 22 Hz. PCA extracted broad bands different for each vigilance state, which included the most representative EEG activities characteristic of them. These results indicate that during SWS, slow oscillations include frequencies up to 6 Hz, and spindle oscillations frequencies down to 7 Hz. No alpha frequencies were identified as an independent band. Frequencies within theta and beta were gathered in the same eigenvector during PS and in different eigenvectors during W suggesting coordinated activation of hippocampal and cortical systems during PS. These bands are consistent with the underlying neurophysiological mechanisms of sleep and wakefulness and with firing frequencies of generators of rhythmic activity obtained in cellular studies in animals.