Information processing of stimulus modules and dimensions in visual perception: Independence and interactions

Helena Kadlec, Purdue University

Abstract

Early vision is believed to be moderated by a number of anatomically separate modules in the cortex. Specifically, two anatomically separate cortical pathways subserve the processing of different types of information: The occipito-temporal pathway processes information related to form, texture and color, while the occipito-parietal pathway is involved in the perception of motion, disparity and location. Evidence for the "independence" of processing of these two pathways is reviewed from three sources; neuroanatomy and neurophysiology of visual system, neuropsychology, and visual perception in normal adults. The term "independence", however, needs to be clarified. Three definitions have been advanced in the framework of the General Recognition Theory (GRT; Ashby & Townsend, 1986), a multidimensional extension of Signal Detection Theory (SDT; Green & Swets, 1966); (1) perceptual separability, (2) decisional separability, (3) and perceptual independence. Within this framework, a new methodology to test for these three types of "independence" was recently developed (Kadlec & Townsend, 1992a, 1992b). The present study employed this methodology to test the separabilities and independence of three stimulus dimensions from the "Form" and "Location" modules/pathways. Two Form dimensions were chosen, Curvature of an arc and Orientation of a line segment in modified circle-radial line stimuli (Shepard, 1964). One dimension of Location was employed, the spatial position in the upper half of the retina. Stimuli were presented monocularly to the left or to the right visual hemifield. Three subjects were tested in three conditions, at low, medium and high discriminability on the three dimensions. Results supported the hypothesis that Form was perceptually separable from Location; however, Location was not separable from Form. Further analyses indicated that, as predicted, Curvature and Orientation interacted while Curvature and Location were often perceptually separable. Contrary to hypothesis, however, Location was not separable from Orientation. These results were corroborated by fitting and comparing two classes of geometric models, GRT models and Multidimensional Scaling (MDS)-Choice model (e.g., Getty, Swets, Swets, & Green, 1979; Nosofsky, 1989). The conclusions reached by the model fitting procedures were globally similar to those reached by the SDT-based methodology, but not as powerful in revealing the asymmetric failures of separability. Implications for future research are indicated.

Degree

Ph.D.

Advisors

Townsend, Purdue University.

Subject Area

Psychology|Experiments

COinS