Emotion recognition of serious game players using a simple brain computer interface

Abstract

Understanding player's cognitive state and emotional response is necessary to make serious games more challenging and enhance the quality of human-machine interaction and gaming experience. Previous emotion recognition using brain computer interfaces (BCIs) is either relying a large number of wet electrodes or combining both brain signals with other peripheral physiological signals. In this paper, we investigate whether a simple BCI with only a few electrodes can identify basic or complex emotions in more natural settings like playing a game. We performed an experiment with 42 participants, who played a brain-controlled video game wearing a single electrode BCI headset and provided a self-assessed valence/arousal feedback at the end of each trial. By analyzing the data obtained from the self-evaluated questionnaires and the attention and meditation recordings from the BCI device, we introduce an automatic emotion recognition method that classifies four emotional states with the average recognition accuracy 66.04%.