The Neural Coding of Force Fields: Implications for Control, Learning and Brain-Machine Interfaces Sando Mussa-Ivaldi (Northwestern University)
The output of the biological motor apparatus is mediated by complex musculo-skeletal structures, with nonlinear geometrical and mechanical properties. I will discuss theoretical and empirical evidence that through the interaction with the peripheral mechanics, the nervous system establishes a set of control primitives in the form of force fields. These fields constitute a system of control policies that the brain may combine to construct – by summation – a broad repertoire of behaviors. This approach provides a framework for understanding how the brain controls movements and contact forces by forming adaptive representations of the environment. Furthermore, I will discuss how the concept of control through the generation of force fields defines a new approach to the development of bi-directional brain-machine interfaces. The communication between brain and external devices takes place via two elements: (a) a motor interface decoding activities recorded from a motor cortical area, and (b) a sensory interface encoding the state of the controlled device into e stimuli delivered to a somatosensory area. In this framework, the enforcement of desired system’s dynamics is reduced to the problem of force-field approximation.