UC Berkeley

In his 1945 book The Phenomenology of Perception, French philosopher Maurice Merleau-Ponty wrote, “The body is our general medium for having a world.” When learning to move, our bodies implicitly know what to do, letting experiences tweak our neural circuits to etch in precise models of the world. These experiences wire up complex capabilities in bodies that were born ready to move, but did not yet know what a pen, a cartwheel, or a hug was.

What is arguably more difficult is using our minds to gain explicit understanding of the processes governing learning and movement. For this, we have science. Merleau-Ponty later wrote, “Science manipulates things and gives up living in them. It makes its own limited models of things; operating upon these indices or variables to effect whatever transformations are permitted by their definition, it comes face to face with the real world only at rare intervals.” In this dissertation, I present the outcomes of my own manipulations, my own limited models which, at rare intervals, come face to face with the ways in which our bodies construct our world.

In Chapter 2, I investigated a potential problem for the use of functional magnetic resonance imaging (fMRI) to study human motor behavior. Task-related changes in heart rate can cause changes in the blood oxygen level dependent (BOLD) signal, potentially masking effects of interest. Correcting for heart rate fluctuations did not fundamentally change the brain's responses to arm movements. However, because these corrections did improve the explanatory power of fMRI analyses, my work stands as an illustration of the efficacy of this correction.

In Chapter 3, I had people learn to produce movements of certain speeds in a virtual shuffleboard game. When learning movement directions, delayed feedback impairs performance due to processing limitations in a brain structure called the cerebellum. Because feedback is naturally delayed in shuffleboard, I asked whether delays affect learning of speeds, as well. They did not, suggesting that the cerebellum is not involved in learning movement speeds from errors.

In Chapter 4, I sought to confirm this speculation using fMRI. I scanned the brains of people as they played shuffleboard. I found no evidence that the cerebellum was involved in the processing of errors of movement speed. However, successful task performance robustly activated the dorsal striatum, a structure involved in learning from rewards and forming habits. These results suggest that, when people learn to move at certain speeds, they depend more on memories of times they got it right than on feedback telling them to speed up or slow down.