Exponentially weighton running statistics for the analysis of non-stationary bioelectric signals

Exponentially weighted running statistics of electroencephalograms (EEG) from the motor cortex and dorsal hippocampus of rabbits were obtained using a special purpose analog computer designed for this research. The EEG of spontaneous arousal and drug-induced behavioral states under amphetamine and pentobarbital was studied. It appears that variability in the central nervous system (CNS) as expressed by the running variance contains valuable information about the state of the CNS. The results show that amphetamine arousal is different from spontaneous arousal in that under amphetamine a state of "hyperarousal" appears to exist. The transition from arousal to pentobarbitalinduced sedation appears to be a stage-wise process and exhibits "undershoots" and "overshoots" in the cortical-hippocampal interstructural variability. It appears that the running mean and running variance are particularly useful statistics for the analysis of nonstationary bioelectric signals.