Abstract

The results of a series of spatial frequency and orientation masking studies in my laboratory have recently been incorporated into a nonlinear mathematical model for suprathreshold pattern discrimination tasks.^1,2 The model contains six distinct spatial frequency tuned mechanisms, each of which consists of an orientation selective linear filter followed by a nonlinear contrast transfer function. Pattern discrimination thresholds are computed by pooling the difference in mechanism responses to each of the two patterns in a two alternative forced choice paradigm, with the pooling being over spatial frequency mechanisms, orientations, and spatial nearest neighbors. Thus, the model is formally analogous to line element models for color discrimination. Following a summary of the model, it will be applied to a number of suprathreshold pattern discrimination tasks including spatial frequency discrimination, curvature discrimination, a variety of hyperacuity tasks, and phase or position discrimination between pairs of spatial patterns with bandwidths of 1.0 octave. The model will also be applied to the effect of contrast on spatial frequency discrimination, Vernier acuity, and phase discrimination. Some discrepancies between model computations and data may be attributed to individual differences among subjects, differing temporal presentations, and differing mean luminances in various studies.

© 1985 Optical Society of America