A new generation of non-invasive biomarkers of cognitive-motor states with application to smart brain-computer interfaces

The design of assistive technologies such as non-invasive brain computer interfaces (BCI) requires an improved understanding of the cortical dynamics of the human brain when interacting with new tools and/or adapting to novel environments in ecological situations. Therefore the aim of this study was to investigate potential biomarkers able to reflect dynamic cognitive-motor states of subjects who had to learn a new tool. These biomarkers were derived from the power bands of electroencephalographic (EEG) signals. The EEG and hand kinematic signals were analyzed for subjects of a Learning Group (LG; n=10) who performed self-selected/initiated center-out hand movements during a visuomotor adaptation task. A Control Group (CG; n=5) was also tested with the same task; however, no adaptation was required. For the LG, the findings indicated that the alpha ([9-13] Hz) and high theta ([6-7] Hz) power computed at the frontal and temporal sites showed a consistent linear and bilateral increase during movement planning during tool learning that may represent the update of the internal model of the new tool. The power increases were correlated with enhanced kinematics as the task progressed. No such differences appeared for the CG. These non-invasive biomarkers appear able to track the human learning/adaptation status and may play a role to overcome specific pitfalls in BCI applications such as the need for frequent recalibrations and the management of the co-adaptation/cooperation between the user's brain and the decoding algorithm required to design the next generation of smart neuroprostheses.