Decoding Decision Correctness from EEG Under High Cognitive Workload in Virtual Reality: Implications for Collaborative Brain-Computer Interface Teams

Collaborative Brain-Computer Interfaces (cBCIs) aim to improve team decision-making by aggregating individual inputs. Traditional methods rely on post-hoc metrics like confidence or reaction time, which are only available after a decision is finalized. This study investigates whether a pre-emptive neural signal—extracted from EEG during the decision-making process itself—can predict decision correctness and improve team accuracy. The researchers used a virtual reality target-detection task with 23 participants, manipulating cognitive workload (High vs. Low) to test the robustness of this signal. They employed a Riemannian spatial-covariance classifier on a strictly motor-safe EEG window to ensure the signal reflected perceptual processing rather than motor preparation.

The study found that the neural signal successfully decoded decision correctness under High Workload (61.4% accuracy). In contrast, the classifier was ineffective under Low Workload, primarily because participants performed near the ceiling, leaving insufficient error trials for the model to learn. When applied to simulated team aggregation, weighting votes by this pre-emptive neural signal significantly improved accuracy on contested (evenly-split) trials under High Workload, increasing team accuracy from 57% to 88% as team size grew. However, this same weighting strategy was detrimental under Low Workload, highlighting that the utility of the cBCI is highly dependent on the operational environment.

This research demonstrates that cBCI systems are not universal tools for team augmentation. Instead, they are workload-conditional. While post-hoc behavioral signals like confidence often provide higher absolute accuracy, they cannot inform real-time, time-critical decisions. The ability to identify decision reliability before a response is committed offers a unique, actionable advantage for teams operating in high-stress, high-workload environments where traditional behavioral reports may be delayed or unreliable.