Finalist EPFL doctorate Award 2011 - Biljana Petreska

Modeling the neural correlates of imitation from a neuropsychological perspective.

The neural control of movement is a key problem in neuroscience. The only way we can we can interact with the outside world is through the contraction of muscles, whether it is to act upon objects or communicate with others. It is thus essential to understand how the brain generates and controls movement. We have investigated this still largely unresolved problem with three complementary approaches: human psychophysics, cerebral disorders and computational modeling.

We first recorded the kinematics of how humans perform natural movements. The collected data allowed us to develop a nonlinear dynamical model of reaching. Given a point to reach, this model generates the appropriate three-dimensional trajectories of the hand. Then, we looked at how the imitation of these reaching movements is affected by a brain lesion. Brain lesions make it possible to “reverse-engineer the brain”, by identifying fundamental building blocks whose functions remain preserved despite the impairment. By studying patients with a motor disorder called apraxia, we found that the brain decomposes an imitation gesture into several simpler components. Surprisingly, some of these patients were able to imitate the simple components separately, but could not coordinate these components when asked to imitate the whole gesture. Simulating different brain lesions with neural networks allowed us to identify which pathway in the brain may create such a disorder.

These results suggest ways to improve current rehabilitation methods and provide new insights into the underpinnings of the acting brain.