Training and transfer effect of evoked brain responses by brain-computer interaction

OBJECTIVE: Electroencephalography (EEG)-based neuro feedback training (NFT) guides users to regulate their neural activity via sensory feedback and further modulates their cognitive state or function. It is a crucial approach to improving brain-computer interface (BCI) performance. However, previous studies rarely report cross-task transfer effects of existing NFT approaches beyond the training task, suggesting that they may fail to improve task-related common functions. METHODS: Here, we propose a steady-state visual evoked potential (SSVEP)-based table hockey BCI game as an NFT approach to improve SSVEP-BCI performance. 40 healthy subjects were randomized into four groups: the 10-frequency NFT group, the 5-frequency NFT and transfer group, the placebo group, and the blank control group. All completed a 10-frequency online SSVEP task before and after five days of training, with EEG and subjective experiences recorded throughout. RESULTS: The two NFT groups achieved significant improvement in online SSVEP classification accuracy. This was accompanied by increased SSVEP power, inter-trial phase coherence (ITPC), and expansion of activated cortical areas. Notably, enhancements generalized to untrained transfer tasks, likely due to the common impulse response between the training and the transfer tasks. CONCLUSION: These findings demonstrate that the proposed NFT approach not only improves SSVEP-BCI performance on the trained conditions but also induces transferable neural changes to adjacent frequencies, suggesting training augments the neural population engaged in processing steady-state visual stimuli. SIGNIFICANCE: This work advances understanding of how self-regulation during NFT improves task-related common functions.