Empirical Evaluation of the Emotiv EPOC BCI Headset for the Detection of Mental Actions

Abstract

This study evaluated the detection accuracy of one of the first brain-computer interfaces intended for personal use by normal, healthy users: the Emotiv EPOC. This system allows the user to directly interact with computer software through thoughts alone. The system was evaluated on its ability to accurately detect and classify six sets of paired mental actions over three evaluation phases. Results found the system to perform significantly better than chance for all mental actions, and improve over time with additional training data. The system detected all actions with equivalent accuracy. Additional investigation of individual differences revealed that a user’s gender, age, handedness, attentional control, vividness of visual imagery, and mental rotation ability all had no bearing on the detection accuracy of the system. These results indicate the Emotiv EPOC system performs its function as a brain-computer interface with an acceptable level of accuracy, yielding many new possibilities for human-computer interaction.