Emotion Recognition with Portable EEG in Immersive 360-Degree Environment

This study aimed to explore the feasibility of using portable single-channel dry electrode electroencephalography (EEG) headbands to identify and distinguish human emotions elicited by multimodal stimuli presented in a 360-degree immersive environment. Such an environment was specifically chosen to facilitate naturalistic perception, in contrast to the conventional presentation of stimuli through a flat screen and headphones in the laboratory setting. To this end, this study designed multimodal stimulation and recorded the subjective scores of the subjects in multiple emotional dimensions through a self-rating scale. The differential entropy (DE) feature was used to capture the dynamic changes and complexity of the EEG signal. A variety of classic machine learning (ML) models were used for classification, and the feature performance and model effectiveness were compared and analyzed. The results show that after removing most artifacts and applying DE features, single-channel EEG signals can effectively distinguish different emotional states measured under multimodal stimulation. In summary, this study provides empirical support for emotion recognition using single-channel EEG in a 360-degree immersive environment, which allowed for naturalistic perception while maintaining the advantages of a controlled setting. This marks a step toward multi-user applications by leveraging the portability and convenience of portable devices.