Title: “Adaptation responses to altered physics law of standing: is it a consequence of exploratory actions and minimizing muscle activation?”
Thesis Supervisor: Dr. Jean-Sébastien Blouin
Committee Members: Dr. Calvin Kuo (School of Biomedical Engineering), Dr. Romeo Chua
Chair: Dr. Tim Inglis
Abstract: Computational models help us understand how the brain controls movement while enabling testable predictions regarding the sensorimotor control of movement. Models for standing balance typically follow optimal feedback control principles, that is, they optimize well-defined rules or cost functions based on feedback provided to the system (Todorov, 2004). A central question to our understanding of brain function relates to the characterization of the cost function for standing balance or in lay terms, what is the overall goal of standing balance. Current opinions of the topic diverge: researchers have argued that minimization of metabolic cost (Selinger et al., 2019), overall muscle torque (Kiemel et al., 2011; Jamali et al., 2017), and movement variability (Hidenori et al., 2006) could all represent fundamental principles of balance control. Most of existing models, including Proportional-Integral-Derivative and linear quadratic controllers, however, rely on minimization of movement variability (Van der Kooij et al., 1999; Hidenori et al., 2006; Welch & Ting, 2008; Li, Levine and Leob, 2012; Lockhart & Ting, 2017). Nevertheless, variability in motor commands is not always detrimental and may help to explore the environment and facilitate motor learning (Wu et al., 2014). An alternative modeling approach is the Markov Decision Process framework that models stochastic decision-making processes during which the system constantly explores possible actions and their related consequences. The Markov Decision Process framework can represent the characteristics of both stochastic and discrete motor control tasks (Michimoto et al., 2016; Jamali et al., 2017; Selinger et al., 2019).
The aims of the proposed thesis work are (1) to model human standing balance control using the Markov Decision Process framework (2) to test physiological predictions from the model using a novel custom-designed robotic balancing platform. I predict that humans will actively explore the consequences of their actions when exposed to novel standing dynamics while minimizing the joint torques they apply during standing balance.
The significance of this thesis is two-fold. First, I will provide physiological evidence supporting or refuting predictions from a Markov Decision Process model that simulates standing balance in both the anterior-posterior and medial-lateral directions. Second, the findings of the proposed model and experiment will provide insights on the overall goal of standing balance and whether humans continuously explore distinct outcomes of their actions while following the overarching rule (cost function) of standing balance.