College of Science Distinguished Speaker: Bill Geisler
College of Science Distinguished Speaker
Finding Targets in Noise and Natural Backgrounds
Dr. Bill Geisler
Director, Center for Perceptual Systems
David Wechsler Regents Chair in Psychology
University of Texas at Austin
Abstract:
This talk will describe evidence for a two-step theory of visual search based directly on the statistical properties of natural images and on the variation in neural processing with retinal location. The first step is a model of detectability at known locations in natural images. This model incorporates measurements of the statistical properties of the natural images and known properties of the early visual system. It has almost no free parameters. The second step is a theory of covert search that combines optimal processing (given the model in step one) with a principled attentional mechanism that efficiently allocates sensitivity gain across the visual field. This attentional mechanism correctly predicts a substantial loss of sensitivity in and around the fovea (“foveal neglect”) when humans are searching over large regions.
Speaker Bio:
Bill Geisler received his Ph.D. from the Indiana University before coming to the University of Texas. He is director of the Center for Perceptual Systems, holder of the David Wechsler Regents Chair in Psychology, a member of the Institute for Neuroscience and on the faculty of the Biomedical Engineering Program. Geisler has broad interests within the general areas of vision, visual perception and the evolution of perceptual systems. In his lab, scientific questions are often attacked with multiple techniques: psychophysics (behavior), neurophysiology (in collaboration with Professors Albrecht and Seidemann), image and scene analysis, and mathematical and computational modeling. Current projects are concerned with perceptual grouping, visual search, natural scene statistics, and the neurophysiology of primary visual cortex.
Intended Audience:
Those with interest in the topic.
Event Snapshot
When and Where
Who
Open to the Public
Interpreter Requested?
No