Twenty-three participants completed two calibration sessions and one online BCI session. The team operationalized interaction intent as comprising affordance-related evaluation (whether an object affords interaction) and approach-avoidance evaluation (the directional tendency toward or away from an outcome). Offline analyses showed above-chance decoding of approach-avoidance classification, with a grand-average accuracy of 66.28%; this transferred to online closed-loop gameplay, reaching 69.64%. Accuracy reached 80.84% for the clearest reward-versus-punishment pairing but dropped to near chance (59.03%) for more ambiguous pairings, while affordance-related classification accuracy ranged from 77.76% to 83.50%. Despite usability limitations, participants reported the BCI-based interaction paradigm was more fun than the controller baseline. The researchers state that this is the first demonstration of real-time EEG decoding of interaction intent during dynamic VR gameplay.
Passive BCI Decodes VR Intent With Eye Gaze
Summary
Researchers combined electroencephalography (EEG)-based passive brain-computer interface (BCI) with eye tracking to decode users' interaction intent in real time during dynamic VR gameplay, reaching 69.64% accuracy in the online closed-loop phase.
Why it matters
Most VR interaction still depends on explicit controller input; this is the first real-time demonstration of EEG-based passive BCI combined with eye tracking decoding interaction intent during dynamic VR gameplay, with online closed-loop accuracy well above chance — a step toward more natural immersive human-computer interaction.
Compiled by BCIwiki from public sources
Sources · 1
biorxiv.org 2026-07-10