AIM: The universal emotions defined independently of cultures, are induced by stimuli depending on subjective experiences, and cause neuro-physiological changes Therefore, analysis of emotional EEG has been used for emotion recognition in order to detect cortical dysfunctions originated from psychiatric disorders. The aim of this study is to propose a new emotional EEG marker.
METHODS: 14-channel EEG series, downloaded from a dataset called DREAMER, were measured from 9 females and 14 males aged between 22-33 years old with 128 Hz sampling frequency with respect to international 10-20 placement system in response to watching 18 different video films. The largest principal components of full-band EEG phase space trajectories were used as emotional EEG markers. Firstly, statistical differences between males and females in emotional states were calculated through one-way ANOVA test, then these differences were transformed into cortical maps.
RESULTS: No significant differences (p>0.5) were observed in basic emotions, while meaningful differences (p<<0.5) were obtained in a mixed type emotion (amusement) between them. Regarding cortical maps, ignorable group differences (p>=0.8) were mostly observed at right temporal lobes. Regarding deep learning applications, emotional states, fear and excitement, calm and anger, surprise and amusement, sadness and amusement, happiness and surprise, happiness and sadness, were classified with the accuracies of %97.28, %94.47, %95.75, %94.15, %92.38, %91.83 respectively. Regarding histograms (Figure-2), the lowest EEG complexity levels were generated in calm and disgust, while the largest levels were observed in happiness and anger. In females and males, identical neuro-cortical functions were generated in basic emotions, whereas their responses were affected by subjective experiences in mixed type emotions.