User Privacy in the Metaverse: On the Potential of Person Identification from EEG Signals

Electroencephalogram (EEG) signals have been shown to convey numerous details about a user, including gender, age, mental state, and health. It has also been shown that EEG signals can convey information about user identity. As virtual and augmented reality (VR/AR) headsets with embedded EEG sensors are starting to emerge, it is crucial to better understand the impact that this may have on user privacy in the metaverse. In this study, we are interested in comparing the identification performance achieved during motor imagery and compare to results achieved with actual movements that are ubiquitous during VR/AR usage. We compare the accuracy achieved under both scenarios, as well as with and without artifact removal algorithms. We show that artifacts present in EEG signals (e.g., head movements, eye blinks) may provide additional user-specific cues that may further exacerbate the issue. Recommendations for future applications and user privacy are discussed, especially in the realm of healthcare applications.