Humans, brains and robots: motor compositionality and action-perception coupling

Tamar Flash
Dept. of Computer Science and Applied Mathematics
Weizmann Institutes of Science
Rehovot, Israel.

Our research combines behavioral and computational modeling studies aimed at uncovering the principles underlying movement generation and perception. I will first describe the kinematic and temporal characteristics of different arm and locomotion movements.  Reviewing observations indicating the strong coupling between path geometry and timing as revealed in both motion generation and perception studies, I will describe optimization and geometric modelling approaches and their integration aimed at investigating the origins of the observed kinematic and timing regularities. I will also   describe recent results regarding motor compositionality from basic primitives, focusing on the possible roles of non-Euclidean geometries in brain representations of motion. Finally, I will describe brain imaging and neural decoding studies examining the possible neural underpinnings of the observed kinematic and temporal invariants and possible implications of our work for robotics.