Abstract: Based on calculation of the complexity of mutual information transmission time series obtained from multi-channel EEG signals, this paper gives the main results of our studies on the characteristics in EEGs from subjects with different cognition levels and under different conditions. These conditions include doing mental arithmetic and resting with eyes closed. Our results show that all experiment conditions, cognition levels and areas of the brain have influences on the value of mutual information complexity of EEG time series. Those with depression problems have a significant lower (P<0.001) complexity under arithmetic task than that under resting conditions. However, the difference in the normal group is not significant. We conclude therefore that the mutual information complexity of EEG time series demonstrates the cognition level of the subjects to some degree.

Key words: cognitive function, electroencephalogram (EEG)
, mutual information complexity