Single-trial dynamics of motor cortex and their applications to brain-machine interfaces.

Jonathan C Kao,Krishna V Shenoy,Paul Nuyujukian,John P Cunningham,Stephen I Ryu,Mark M Churchland

doi:10.1038/ncomms8759

Jonathan C Kao, Krishna V Shenoy + Show 4 more

Open Access

https://doi.org/10.1038/ncomms8759

Copy DOI

Abstract

Increasing evidence suggests that neural population responses have their own internal drive, or dynamics, that describe how the neural population evolves through time. An important prediction of neural dynamical models is that previously observed neural activity is informative of noisy yet-to-be-observed activity on single-trials, and may thus have a denoising effect. To investigate this prediction, we built and characterized dynamical models of single-trial motor cortical activity. We find these models capture salient dynamical features of the neural population and are informative of future neural activity on single trials. To assess how neural dynamics may beneficially denoise single-trial neural activity, we incorporate neural dynamics into a brain–machine interface (BMI). In online experiments, we find that a neural dynamical BMI achieves substantially higher performance than its non-dynamical counterpart. These results provide evidence that neural dynamics beneficially inform the temporal evolution of neural activity on single trials and may directly impact the performance of BMIs.

Highlights

Increasing evidence suggests that neural population responses have their own internal drive, or dynamics, that describe how the neural population evolves through time
An important prediction of neural dynamical models is that neural population activity observed up to the current time is informative of neural activity that has yet-to-be-observed on noisy single-trials
While previous studies have performed systems identification to characterize the neural dynamics in motor cortex[3,5], we sought to see if neural dynamics were informative of future neural activity and could aid in single-trial neural state estimation

Summary

Introduction

Increasing evidence suggests that neural population responses have their own internal drive, or dynamics, that describe how the neural population evolves through time. An important prediction of neural dynamical models is that neural population activity observed up to the current time is informative of neural activity that has yet-to-be-observed on noisy single-trials Performing dynamical estimation may have beneficial smoothing and denoising properties; in our example, a neural state trajectory rotating counterclockwise should not instantaneously traverse a clockwise path, as might be observed due to single-trial noise, just as a falling cannonball should not defy gravity and float up. This dynamical estimation should result in more accurate neural state trajectories than could be inferred by merely smoothing the observations without knowledge of neural dynamics. If dynamical neural state estimation improves closed-loop BMI performance, this may have significant implications on BMI design[23]

Methods

Results

Conclusion