Exploring Training Effect in 42 Human Subjects Using a Non-invasive Sensorimotor Rhythm Based Online BCI

EEG-based brain-computer interfaces (BCIs) show promise for providing an alternative communication channel between the brain and an external device. It is well acknowledged that BCI control is a skill and could be improved through practice and training. In this study, we explored the change of BCI behavioral performance as well as the electrophysiological properties across three training sessions in a pool of 42 human subjects. Our results show that the group average of BCI accuracy and the information transfer rate improved significantly in the third session compared to the first session; especially the significance was reached in a smaller subset of low BCI performance group (average accuracy<70%) as well. There was a significant difference of event-related desynchronization (ERD) lateralization for BCI control between the left and right-hand imagination task in the last two sessions, but this significant difference was not shown for the first training sessions. No significant change of R-square value or event-related desynchronization and synchronization (ERD/ERS) for either channel C3 or channel C4, which were used for online control, was found across the training sessions. The change of ERD lateralization was neither significant across the training sessions. The present results indicate that BCI training could induce the change of behavioral performance and electrophysiological properties quickly, in just a few hours’ training distributed into three sessions. Multiple training sessions might be especially beneficial for the low BCI performers.