PhD Defence Tonio Weidler

Keywords: computational neuroscience, deep learning, sensorimotor system, dexterity
 

"The mechanism at hand: A goal-driven approach to modeling the human sensorimotor system"

The human brain controls the body to interact with the world. And in turn, both the body and the individual’s intentions shape the brain’s activity. “The Mechanism at Hand” introduces a novel framework for computational neuroscience that allows researchers to study the brain under this perspective. By combining machine learning, neuroanatomy, and robotics, it presents an approach that links motor behaviour to the neural activity that produces it. Specifically, the thesis developed a biologically plausible model of the large network of brain regions that enables the human to produce fine motor movements in the hand. Machine learning techniques then trained this model to manipulate objects by controlling a simulated robotic hand. The thesis then analysed the neural activity the model exhibited while manipulating an object and thereby revealed a novel mechanism that may explain how the brain plans movements. In sum, this research highlights the potential of analysing brain functions in the context of bodily action and supports this approach by offering practical tools and conceptual guidelines. 

Click here for the live stream.