Mattia Rigotti
BME PhD Defense Presentation

Date: 2026-02-09
Time: 12:30 PM ET
Location / Meeting Link: EBB CHOA Seminar Room / https://emory.zoom.us/j/98098280681?pwd=Cgh0t8W92gTUOh9yNb79kj4D1jrlfS.1

Committee Members:
Chethan Pandarinath, PhD (advisor); Nicholas Au Yong, MD, PhD; Jonathan Kao, PhD; Garrett Stanley, PhD; Anqi Wu, PhD


Title: Hierarchical movement encoding in motor cortex for high-performance brain-computer interface control

Abstract:
Intracortical brain-computer interfaces (iBCIs) are a promising avenue for restoring movement capabilities to individuals with tetraplegia, by recording neural activity from implants in the motor cortex and translating these signals into intended actions. Despite significant advances over the years, iBCI control performance still falls short of able-bodied behavior. One potential source of this limitation is that standard iBCI control paradigms largely assume that the motor cortex only encodes low-level, instantaneous movement variables, despite growing evidence for richer and higher-level representations. In this dissertation, we investigated the organization and tuning properties of multi-level movement-related neural computations in the motor cortex, with the goal of informing next-generation iBCI solutions. First, we studied the motor cortical activity underlying cross-modality movement generation (e.g., different effectors and physical settings), through high-channel-count recordings from non-human primates and analyses using state-of-the-art computational tools. We found that neural population activity within the motor cortex was consistent with hierarchically organized computations, spanning neural subspaces from high-level, modality-independent to modality-specific. Next, we investigated the encoding properties of neural activity in the human motor cortex during motor preparation, through intracortical recordings from research participants in the BrainGate2 clinical trial. We found that preparatory activity encoded diverse multi-level features of intended movement prior to movement onset, including direction, curvature, distance, and speed. Finally, we designed and implemented a proof-of-concept iBCI control paradigm that predicts intended movement features in advance from preparatory activity, enabling rapid, self-paced control of a computer cursor by human participants. Together, these findings provide insight into the hierarchy of motor cortical computations underlying movement and demonstrate how high-level movement representations can be leveraged for high-performance iBCI control.