Abstract: In view of the differences in Electroencephalogram (EEG) induced by different emotions and the insignificant effects of traditional emotion induction methods, an EEG analysis method for different emotions based on event-related potential technology and virtual reality technology was proposed. First, Virtual Reality (VR) technology was adopted as the experimental emotion induction environment to solve the problem of insignificant emotion induction effects; second, the third-order FastICA algorithm was used to process the original EEG data for artifact signals; finally, Event-Related Potential (ERP) technology was employed to analyze EEG signals under different emotional stimuli.The experimental results show that the EEG signals of positive and negative emotions, and neutral emotions after de-artifact processing have significant differences in the N1 and P2 stages of ERP. In the N1 time window, the low-titer negative stimuli pay more attention to the allocation of resources than the high-price positive stimulus. In the P2 stage, it is found that the brain preferentially processes not only negative stimuli that threaten the individual, but also various stimuli with motivational significance.