Robust Brain-Computer Interface (BCI) Control During Frequency-Tuned Transcranial Alternating Current Stimulation (tACS).

Integrating frequency-tuned, adaptive brain stimulation with brain-computer interfaces (BCIs) allows direct tests of the causal role of brain oscillations and advances BCIs toward bi-directional operation. A central challenge is that stimulation-induced artifacts overlap with endogenous brain rhythms, undermining robust real-time signal decoding and contingent BCI feedback. Here, we overcome this limitation by introducing an artifact-suppression approach that enables robust motor-imagery BCI control during frequency-tuned amplitude-modulated transcranial alternating current stimulation (AM-tACS). We developed a real-time spatial filtering pipeline that combines spatio-spectral decomposition (SSD) with beamforming and evaluated its performance against a standard Laplacian filter in 14 healthy participants. BCI control was assessed both in the absence of stimulation and during stimulation. We hypothesized that only the SSD-beamforming approach would preserve robust BCI control under stimulation. In the absence of stimulation, both pipelines supported robust BCI control (SSD-beamforming: $73~\pm ~9$ %; Laplacian: $72~\pm ~8$ %; ${p} =$ .849). During AM-tACS, Laplacian filtering showed a marked performance decline to near chance level ( $58~\pm ~9$ %; ${p} \lt $ .001). However, with SSD-beamforming, robust BCI control was preserved ( $76~\pm ~9$ %). These results demonstrate that robust BCI control during frequency-tuned AM-tACS is achievable. By enabling simultaneous stimulation and decoding, this approach establishes a new paradigm for testing the causal contributions of brain rhythms during ongoing BCI control and for advancing stimulation-informed, bi-directional BCI interventions. Future work will determine how AM-tACS can be leveraged to enhance BCI performance but also promote neuroplasticity during restorative BCI applications.