Abstract
The electroencephalogram (EEG) hyperscanning technique has been demonstrated to facilitate the applicability of a collaborative brain-computer interface (cBCI) to augment human performance with respect to the collective intelligence of multiple brains. However, assembling a hyperscanning platform with commercial products inevitably introduces practical and cost burdens regarding labor and hardware setup, hindering group scalability. This work thus explores how effectively a low-cost, custom-made EEG hyperscanning platform can be achieved by demonstrating a cBCI framework. This work quantifies BCI performance in collaborative and single-brain scenarios applied to the EEG dataset collected from three subjects simultaneously participating in a target-distractor differentiation task over 10 days. This work also compares various brain-fusion scenarios with feature-extraction methods for multiple brains. Given the 30 pseudo brains (i.e., 3 subjects × 10 day sessions), the decision-level committee voting outperformed the single-brain BCI performance and considerably improved by leveraging more pseudo brains. The 27-brain setting achieved the best information transfer rate (ITR) of 116.6±5.6 bits/min, which was a nearly 817% enhancement over the single-brain ITR (12.7±9.2 bits/min). In addition, the cBCI decision augmented the actual button-pressing time by 25 ms. Such a low-cost, custom-made hyperscanning infrastructure economically and practically favors multiple-brain applications in a larger group.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have