Interpreting Neural Population Activity During Feedback Motor Control Matthew Golub 10.1184/R1/7461863.v1 https://kilthub.cmu.edu/articles/thesis/Interpreting_Neural_Population_Activity_During_Feedback_Motor_Control/7461863 The motor system routinely generates a multitude of fast, accurate, and elegant movements.<br>In large part, this capacity is enabled by closed-loop feedback control systems in the brain.<br>Brain-machine interfaces (BMIs), which translate neural activity into control signals for driving<br>prosthetic devices, also engage the brain’s feedback control systems and offer a promising<br>experimental paradigm for studying the neural basis of feedback motor control. Here, we address<br>both the engineering challenges facing current BMI systems and the basic science opportunities<br>afforded by them.<br>Previous studies have demonstrated reliable control of the direction of movement in cursorbased<br>BMI systems. However, control of movement speed has been notably deficient. We provide<br>an explanation for these observed difficulties based on neurophysiological studies of arm<br>reaching. These findings inspired our design of a novel BMI decoding algorithm, the speeddampening<br>Kalman filter (SDKF) that automatically slows the cursor upon detecting changes<br>in decoded movement direction. SDKF improved success rates by a factor of 1.7 relative to a<br>standard Kalman filter in a closed-loop BMI task requiring stable stops at targets.<br>Next, we transition toward leveraging the BMI paradigm for basic scientific studies of feedback<br>motor control. It is widely believed that the brain employs internal models to describe our<br>prior beliefs about how an effector responds to motor commands. We developed a statistical<br>framework for extracting a subject’s internal model from neural population activity. We discovered<br>that a mismatch between the actual BMI and the subjects internal model of the BMI explains<br>roughly 65% of movement errors. We also show that this internal model mismatch limits movement<br>speed dynamic range and may contribute toward the aforementioned known difficulties in<br>control of BMI movement speed. <br> 2015-05-01 00:00:00 Interpreting neural population activity