Improving performance in motor imagery BCI-based control applications via virtually embodied feedback

Abstract

Objective: Brain-computer interfaces (BCIs) based on motor imagery (MI) are commonly used for control applications. However, these applications require strong and discriminant neural patterns for which extensive experience in MI may be necessary. Inspired by the field of rehabilitation where embodiment is a key element for improving cortical activity, our study proposes a novel control scheme in which virtually embodiable feedback is provided during control to enhance performance. Methods: Subjects underwent two immersive virtual reality control scenarios in which they controlled the two-dimensional movement of a device using electroencephalography (EEG). The two scenarios only differ on whether embodiable feedback, which mirrors the movement of the classified intention, is provided. After undergoing each scenario, subjects also answered a questionnaire in which they rated how immersive the scenario and embodiable the feedback were. Results: Subjects exhibited higher control performance, greater discriminability in brain activity patterns, and enhanced cortical activation when using our control scheme compared to the standard control scheme in which embodiable feedback is absent. Moreover, the self-rated embodiment and presence scores showed significantly positive linear relationships with performance. Significance: The findings in our study provide evidence that providing embodiable feedback as guidance on how intention is classified may be effective for control applications by inducing enhanced neural activity and patterns with greater discriminability. By applying embodiable feedback to immersive virtual reality, our study also serves as another instance in which virtual reality is shown to be a promising tool for improving MI.