MODELS OF PERCEPTUAL-LEARNING IN VERNIER HYPERACUITY

Performance of human subjects in a wide variety of early visual proces sing tasks improves with practice. HyperBF networks (Poggio and Girosi 1990) constitute a mathematically well-founded framework for understa nding such improvement in performance, or perceptual learning, in the class of tasks known as visual hyperacuity. The present article concen trates on two issues raised by the recent psychophysical and computati onal findings reported in Poggio et al. (1992b) and Fahle and Edelman (1992). First, we develop a biologically plausible extension of the Hy perBF model that takes into account basic features of the functional a rchitecture of early vision. Second, we explore various learning modes that can coexist within the HyperBF framework and focus on two unsupe rvised learning rules that may be involved in hyperacuity learning. Fi nally, we report results of psychophysical experiments that are consis tent with the hypothesis that activity-dependent presynaptic amplifica tion may be involved in perceptual learning in hyperacuity.