Sensory Constraints on Volitional Modulation of Neural Activity in the Motor Cortex

A fundamental way in which we interact with the world around us is voluntary movement. The primary motor cortex (M1) is known for its central role in driving this intentional movement, and decades of research in the field of motor control has shown that individuals can learn to modulate even single neurons in M1 at will. Indeed, the fact that neural activity in M1 can be volitionally controlled makes it a powerful target for brain-computer interface (BCI) devices and their clinical applications.

Yet M1 also encodes non-volitional information. This brain area receives pronounced sensory inputs. It also contributes to sensory-evoked—as well as voluntary—motor responses. What does this duality in M1 imply for its volitional control? To what extent do nonvolitional signals restrict our ability to voluntarily modulate M1?

This thesis provides insight into the constraints imposed by sensory context on M1 by explicitly probing voluntary control of individual neurons in this area under a variety of sensory conditions. To do this, three macaque monkeys were trained in a BCI paradigm that decoupled volitional modulation of M1 from specific aspects of sensory feedback. In these experiments, the firing rate of a single neuron—termed the command neuron—directly determined the position of a computer cursor along a one dimensional axis. Altering the orientation and location of this movement axis and the posture in which the animal’s arm rested created distinct sensory contexts for the subject without changing the neural requirements of the task. We leveraged this paradigm to assess monkeys’ ability to modulate individual neurons in M1 under different sensory contexts.

We found that sensory context persistently affected volitional control of single neurons in M1. For all three subjects, the ability to perform the task significantly changed based on movement orientation: rotating the feedback axis could render the same neural task effortless or problematic. Axis location within the workspace also affected the ability to modulate individual neurons, albeit to a lesser extent than orientation. Notably, the disparity in single-neuron control across sensory contexts was not resolved even after extended training in the task. We found that additional practice within a session or across multiple days was not sufficient to erase the interaction between movement orientation and the ability to voluntarily modulate individual neurons.

Overall, these findings suggest that sensory context does limit the degree to which M1 activity is under volitional control. This interplay between sensory and volitional signals, which manifested at the level of individual neurons within M1, may constitute an additional constraint on motor learning. The notion that sensory inputs may impose bounds on the voluntary modulation of M1 could also have direct implications for the development of clinical neuroprosthetic devices and other advances in neurotechnology.