Discrimination of aperiodic two-dimensional patterns: a comparison of feature detection and frequency analysis models/
Clinton F. Jenne · ScholarWorks@UMassAmherst (University of Massachusetts Amherst) · 2021
Several theoretical models have been proposed of human and lnfrahuman vision and employed, with some success, to explain the characteristics of a biological system which detects stimulus -equivalences (Deutsch, 1955; Sutherland, 1957; Thomas, 1970; Kerr and Thomas, 1972). The neural correlate of pattern vision, according to Deutsch and Sutherland, was a two-dimensional network of cells which transform the stimulus pattern into coded form that is qualitatively dissimilar from the original stimulus figure. A model of this sort was suggested in order to account for shape perception independent of retinal location. Later, Thomas (1970) outlined a detector system in which neural units responded optimally only If the stimulus has a certain width, orientation and position. The response properties that Thomas endowed to neural units were based upon electrophysiological evidence supplied by Hubel and wiesel (1962, 1965); the model proposed by Thomas was an extension of the neural network concept proposed by Deutsch, whereby neural units that comprise the two-dimensional network were now driven by very special stimulus charcteristics.