Field computation in motor control

Abstract

Abstract Field computation deals with information processing in terms of fields , continuous distributions of data. Many neural phenomena are conveniently described as fields, including neuron activity from large (brain area) to small (dendritic) scales. Further, it is often useful to describe motor control and sensorimotor coordination in terms of external fields such as force fields and sensory images. We survey the basic concepts of field computation, including both feed-forward field operations and field dynamics resulting from recurrent connections. Adaptive and learning mechanisms axe discussed briefly. The application of field computation to motor control is illustrated by several examples: external force fields associated with spinal neurons (Bizzi & Mussa-Ivaldi 1995), population coding of direction in motor cortex (Georgopoulos 1995), continuous transformation of direction fields (Droulez & Berthoz 1991a), and linear gain fields and coordinate transformations in posterior parietal cortex (Andersen 1995). Next we survey some field-based representations of motion, including direct, Fourier, Gabor and wavelet or multiresolution representations. Finally we consider briefly the application of these representations to constraint satisfaction, which has many applications in motor control.