The performance of intracortical brain-computer interfaces (BCIs) has advanced substantially over the last
decade, but these devices are not yet robust enough for the home environment, where they can truly improve
quality of life for individuals with disabilities. To date, BCIs have depended on experienced technicians to operate
large and complicated systems comprised of multiple computers, signal processors, neural recording
headstages, and custom software. Our laboratory has developed a portable, battery-powered intracortical BCI
system that enables independent in-home computer access. However, to achieve the long-term goal of true
clinical viability, BCIs must also offer reliable and robust functional performance in the less well-controlled home
environment. We have achieved robust and generalizable control of a computer cursor using a biomimetic
approach that combines reach-based velocity control of cursor position with grasp-based control of mouse click
onset and offset. This transient-based neural decoder allows for generalized click function, adding the ability to
‘click-and-drag’ when accessing a computer (similar to carrying an object) to the ‘point-and-click’ functionality
that is typically implemented in BCIs. Independent home use of the BCI system will provide an opportunity to
collect neural data over long periods of time during unstructured and varied tasks, enabling quantification of
context-dependent neural variability due to subject-state (e.g., fatigue, pain, or stress) as well as plasticity due
to learning. Understanding how neural signals vary over time will be critical for clinical BCI systems that must be
robust, generalizable, and autonomous (i.e., operate for extended periods without technician intervention).
This project will first quantify the impact of subject-state on movement-related neural activity and performance
during in-home BCI use. This understanding is critical to developing robust BCIs that eliminate the need for
recalibration even in uncontrolled environments. The extent to which subject-state information is represented in
motor cortex and overlaps with BCI control dimensions will inform development efforts to mitigate the impact of
these nuisance variables. Participants will use the BCI for a variety of self-selected computer access tasks over
periods of many months that will challenge decoder performance. This project will investigate motor learning
mechanisms that may be engaged to facilitate improvements in performance that generalize to many different
tasks. Experiments will test the hypothesis that stable population-level neural activity emerges to strengthen
movement-related activity while minimizing non-task-related neural variability. Finally, participants will undergo
targeted neural training to determine if motor learning can be accelerated and whether different mechanisms of
neural reorganization are engaged in response to interventions that challenge the speed and accuracy properties
of the decoder in different ways. This project will improve our understanding of neural plasticity mechanisms
during extended BCI use in a real-world environment. Ultimately this knowledge will enable stable, high-
functioning BCI performance during independent home-use, which is critical for clinical translation.
