Non-linear dynamic complexity of the human EEG during evoked emotions.

The present study examined EEG dimensional complexity (estimated correlation dimension) in 76 healthy volunteers in response to emotionally valenced (i.e. neutral, positive and negative) film clip stimulation. EEG was recorded from 18 sites (10-20 system). We estimated the dimensional complexity by the Grassberger and Procaccia and Skinner's point-wise dimension (PD2i) methods. The results were compared to spectral measures of the EEG. Only the PD2i algorithm (i.e. the one that did not require data stationarity) discriminated among all the three film categories. The main results showed that both negative and positive emotions occurred with higher values (at some posterior locations) of EEG DCx estimates compared to the neutral viewing condition. The topographical differences (frontal vs. posterior temporal) between positive and negative evoked emotions were obtained. There were also some significant direct relationships between dynamic complexity estimates and intensities of subjective emotional feelings. It is concluded that dimensional complexity estimates turned out to be sensitive to subtle aspects of emotional processing not accessible by linear EEG analyses.