Spatiotemporal neural correlates of brain-computer interface learning

Abstract

Brain-computer interfaces (BCIs) have been largely developed to allow communication, control, and neurofeedback in human beings. Despite their great potential, BCIs perform inconsistently across individuals and the neural processes that enable humans to achieve good control remain poorly understood. To address this question, we performed simultaneous high-density electroencephalographic (EEG) and magnetoencephalographic (MEG) recordings in a motor imagery-based BCI training involving a group of healthy subjects.After reconstructing the signals at the cortical level, we showed that the reinforcement of motor-related activity during the BCI skill acquisition is paralleled by a progressive disconnection of associative areas which were not directly targeted during the experiments. Notably, these network connectivity changes reflected growing automaticity associated with BCI performance and predicted future learning rate. Altogether, our findings provide new insights into the large-scale cortical organizational mechanisms underlying BCI learning, which have implications for the improvement of this technology in a broad range of real-life applications.

Learning in brain-computer interface control evidenced by joint decomposition of brain and behavior

“…We use MEG data from 20 healthy adult individuals who learned to control a motor-imagery based BCI over four separate sessions spanning a two week period. Consistent with prior reports of this experiment [32], we find a significant improvement in performance across the four sessions (oneway ANOVA F (3, 57) = 13.8, p = 6.8 −7 ) ( Fig. 2).…”

“…In BCI learning specifically, the strength of white matter connectivity between frontal and occipital regions predicts control of motor imagery based BCIs [78]. Additionally, analyses of this same experiment have shown task related changes in functional connectivity were spatially diffuse, and found in frontal, temporal, and occipital regions in the α band [32], and were strongest in frontal, motor, central, and parietal regions in the β band. Our results add to these findings by demonstrating that the most consistent regions that covary in their functional connectivity are interactions between the frontal lobe and other regions.…”

“…Recent studies have demonstrated that patterns of functional connections can provide clearer explanations of the learning process than activation alone [30], and changes in those functional connections can track changes in behavior [31]. During BCI tasks, functional connectivity reportedly increases within supplementary and primary motor areas [15] and decreases between motor and higher-order as-sociation areas as performance becomes more automatic [32]. Data-driven methods to detect putative cognitive systems as modules in functional brain networks [33,34] have been used to demonstrate that a particularly clear neural marker of learning is reconfiguration of the network's functional modules [31].…”