Bivariate Response Distributions in Visual Detection and Discrimination

Abstract

Two patterns can sometimes be discriminated from one another better than they are detected. A Signal Detection analysis of the information used to perform these psychophysical judgments suggests that properties exist in the visual system which favor the ability to discriminate over the ability to detect. Two candidate properties are inhibition and/or correlated noise between tuned spatial frequency analyzers. The extent to which such processes occur is reflected by the parameters defining sensory response distributions in a Signal Detection model of pattern perception. A method by which these parameters can be estimated is described, and several alternative response spaces are discussed.