Duration-modulated neural population dynamics in humans during BMI controls

TL;DR

This study investigates how motor cortex neural dynamics differ during BMI-controlled movements of varying durations, revealing sustained inputs' role in maintaining activity for longer movements.

Contribution

It demonstrates that MC activity during BMI control involves sustained inputs, which modulate neural dynamics across different movement durations, extending understanding of BMI neural mechanisms.

Findings

Long-duration movements sustain MC activity, pausing neural dynamics.

MC initial conditions are shared across different movement durations.

External inputs influence the differences in MC population dynamics.

Abstract

The motor cortex (MC) is often described as an autonomous dynamical system during movement execution. In an autonomous dynamical system, flexible movement generation depends on reconfiguring the initial conditions, which then unwind along known dynamics. An open question is whether these dynamics govern MC activity during brain-machine interface (BMI) control. We investigated MC activity during BMI cursor movements of multiple durations, ranging from hundreds of milliseconds to sustained over seconds. These durations were chosen to cover the range of movement durations necessary to control modern BMIs under varying precision levels. Movements shared their MC initial condition with movements of different durations in the same direction. Long-duration movements sustained MC activity, effectively pausing the neural population dynamics until each movement goal was reached. The difference across durations in MC population dynamics may be attributed to external inputs. Our results highlight the role of sustained inputs to MC during movement.