De novo brain-computer interfacing deforms manifold of populational neural activity patterns in human cerebral cortex

Abstract

Human brains are capable of modulating innate activities to adapt to novel environmental stimuli; for sensorimotor cortices (SM1) this means acquisition of a rich repertoire of motor behaviors. We investigated the adaptability of human SM1 motor control by analyzing net neural population activity during the learning of brain-computer interface (BCI) operations. We found systematic interactions between the neural manifold of cortical population activities and BCI classifiers; the neural manifold was stretched by rescaling motor-related features of electroencephalograms with model-based fixed classifiers, but not with adaptive classifiers that were constantly recalibrated to user activity. Moreover, operation of a BCI based on a de novo classifier with a fixed decision boundary based on biologically unnatural features, deformed the neural manifold to be orthogonal to the boundary. These principles of neural adaptation at a macroscopic level may underlie the ability of humans to learn wide-ranging behavioral repertoires and adapt to novel environments.