Developing a multi-user SSVEP-based BCI

Abstract

Steady-State Visual Evoked Potential (SSVEP)-based Brain-Computer Interfaces (BCIs) enable users to control systems using neural responses to periodic visual stimuli. This project extends traditional single-user SSVEP paradigms to a multi-user competitive setting, investigating whether competition modulates signal strength and user engagement. A total of 14 participants engaged in real-time trials under both single-user and multi-user conditions while observing shared SSVEP stimuli on the same screen. The system employed a synchronised dual-BioSemi ActiveTwo EEG setup, with a custom MATLAB interface handling real-time visual cue presentation, TCPbased inter-device communication, and feedback generation based on Filter Bank Canonical Correlation Analysis (FBCCA) classification outputs. While all trials were conducted with fixed 5-second gaze durations, an additional post-hoc offline analysis explored the relationship between gaze duration and signal quality. Objective metrics included FBCCA correlation coefficients, spectral signal strength via power spectral density (PSD), and signal-to-noise ratio (SNR) analyses. Subjective feedback on engagement, focus, and usability was collected through surveys. Classification accuracies did not significantly differ between conditions. However, FBCCA correlation values were generally higher during the multi-user trials, particularly at occipital sites (O1, O2, Oz). Post-experiment analysis revealed enhanced SSVEP responses at the second and third harmonics in the multiuser condition, supporting the possibility that extrinsic social factors—such as competition—may improve attentional engagement and neural entrainment. These findings lend partial support to the Social Facilitation Theory. To ensure consistent visual exposure, participants viewed shared stimuli on a single display, avoiding occlusion while maintaining equivalent fixation conditions. The system also featured precise cue timing, synchronised feedback, and structured data storage. Overall, while classification performance remained stable, the results suggest that multi-user dynamics may enhance the quality of SSVEP responses and engagement. These findings inform the design of socially embedded BCI systems, particularly in contexts such as gaming, rehabilitation, and collaborative interfaces.