NESTED REENTRANT AND RECURRENT COMPUTATION IN EARLY VISION - A BAYESIAN NEUROMORPHIC MODEL APPLIED TO HYPERACUITY

Hyperacuity is demonstrated in a neuromorphic model of the early visua l system. The model incorporates Bayesian principles which are embodie d in the dynamics of reentrant and recurrent feedback processes. Each retinotopically mapped area in the model represents a transformation o f data from the visual field. Sensory information propagates in a bott om-up direction from one area to the next, while information based on Bayesian priors propagates in a top-down direction through reentrant c onnections. The 'bottom-up' and 'top-down' information maintain a sepa rate existence in distinct layers of the model, but they interact thro ugh local connections within each area. Transformations between one ar ea and the next are defined by the reentrant synaptic connections betw een areas, while local prior probability maps are defined by local rec urrent connections within layers. The representation of hyperacuity is accomplished using a model of functional multiplicity: the large rati o of neurons in striate cortex compared with the number of afferent fi bers projecting from the lateral geniculate nucleus. High functional m ultiplicity, in conjunction with hierarchical reentrant processing, al lows the model to represent a fine-grained restoration of the line str ucture of visual input.