Biological visual systems make use of many different cues for visual judgments. For depth and shape estimation, these include occlusion, texture, perspective, motion parallax, disparity, individual cues, but cannot occur until cues are promoted to a commensurate scale by filling in one or more needed parameters (e.g. the fixation distance for depth estimates, the illuminant color and intensity for estimates of surface color, etc.). These parameters are also estimated using multiple cues (e.g., both retinal and oculomotor cues for the fixation distance). We propose experiments intended to clarify how human observers promote and combine cues. The experimental methods used are bases on perturbation analysis which permits examination of a system that can potentially react to distortions and inconsistencies in stimuli. The proposed research consists of three major tasks. (1) We will examine depth judgments and motor responses in simulated 3-D scenes to determine whether behavior can be well-understood by modeling observers as Bayesian decision makers. If so, any difference between visual judgments and motor responses may be due to different weights given to visual cues due to differences in the corresponding risk factors. (2) The notion of cue promotion suggests that there are parameters that observers must estimate along the way to determining scene geometry and surface properties such as color. Again, a Bayesian model will be used to shed light on the process of estimating these internal parameters. (3) We will continue our studies of the form of representation used by human observers for curves and surfaces as well as 3-dimensional motion paths. These studies will inform our understanding of how object shape and motions are determined from sparse and often conflicting visual data.
Showing the most recent 10 out of 113 publications
