Dissertation: An investigation of the effects of reward and metabolic cost on movement vigor
Published in CU Boulder, 2021
Recommended citation: Garrick W. Bruening, "An Investigation of the Effects of Reward and Metabolic Cost on Movement Vigor." University of Colorado at Boulder, Colorado, 2021. https://www.proquest.com/openview/bb4d6a41ab32289b7cbf8f14ff6b5a55/1?pq-origsite=gscholar&cbl=18750&diss=y
Nearly every movement we make reflects a decision. This decision likely involves a trade-off between maximizing reward and minimizing effort. How fast we move, vigor, can reveal how we value reward and effort when making movement decisions. The goal of this dissertation is to gain further insight into how reward and effort affect movement vigor, and how we can use computational models of effort to improve predictions of movement effort.
First, I evaluated how subjects responded to reward when given in a probabilistic manner. Subjects made reaching movements to multiple targets, while we altered the probability of receiving a fixed reward amount from completing the movement. Vigor was influenced by the probability of reward. Interestingly, the vigor of the following movement was faster following a reward compared to the absence of reward, and faster still the more surprising the reward was. This suggests that vigor is also modulated by the history of reward and reward prediction error.
Next, I focused on the effect of effort on vigor. I measured both the metabolic cost and vigor of reaching with increasing mass and found that metabolic cost increased and movements slowed with added mass. I found that movement slowing was best explained by the maximization of a utility that considered the rate at which both reward was to be acquired, effort was to be expended, and where effort, critically, was represented as metabolic cost.
Lastly, I focused on determining the ability of oft-used models of metabolic cost to explain the metabolic cost of reaching. I developed a neuromechanical model of the arm to estimate the metabolic rate of experimental reach data. I found that while many metabolic cost models can reasonably predict the experimental data, effort representations used in motor control research fail to adequately capture the metabolic cost of reaching.
Together, these studies provide a better understanding of the effects of reward and effort on movement vigor and how we can use computational models of effort in research, potentially informing future work aimed at understanding the neural, psychological and biomechanical determinants of why we move the way we do.