Each movement we make represents the final output of complex processes in the nervous system. Studies of motor control often attempt to minimize further complication by using controlled environments to generate repeated movements. However, in natural situations, the motor system faces the much more complicated task of interacting with an uncertain environment. Often this uncertainty arises from noise. For example, when attempting to swat a fly, we must rely on an erratic visual cue (the fly’s flight path) to estimate its future location and then develop the appropriate motor command. In this case, the goal of the sensorimotor system is to translate noisy sensory information into the single most appropriate action. At other times, uncertainty arises from the presence multiple discrete options. For example, when picking out an apple at the store, we examine the color and size of each one first and then reach to the one we deem best. In this situation, the uncertainty does not pertain to the executed movement itself—after all, the locations of the apples are static—but rather to the expected consequences for each option. The goal of the sensorimotor system in this case is to use sensory cues (and/or previous experiences) to decide between the multiple possible actions. Regardless of the source and type of uncertainty encountered during motor control, the sensorimotor system should also use the result (missing the fly or picking a rotten apple) to inform future situations. In this work, I examine how primary motor cortex (M1) and dorsal premotor cortex (PMd) plan and execute reaching movements when faced with noisy sensory information or multiple potential reach targets. I show that when faced with noisy information about a target location, PMd develops a low-fidelity reach plan, reflecting a subjective sense of uncertainty about the decision. I then show that when faced with two potential reach targets, neither PMd nor M1 contains multiple plans for both targets. Instead, motor cortex appears to decide on one option very quickly and later switches if required. Finally, I show that both PMd and M1 contain differential responses to the consequences (success or failure) of executed movements, potentially revealing a mechanism for driving motor learning within the network. These results together provide insight into how motor cortices cope with uncertainty when planning and executing movements.

Last modified

• 11/18/2019

Creator

• Brian M Dekleva

DOI

• https://doi.org/10.21985/n2-ey2t-6q34

Subject

• Biomedical Engineering

Keyword

• Decision-making
• Bayesian behavior
• Motor cortex
• Uncertainty
• Motor control

Date created

• 2018-01-01

Resource type

• Dissertation

Rights statement

• In Copyright