The broad goal for the proposed research is to understand spatial vision and the degradations found in amblyopia and peripheral vision. The proposed experiments will be coordinated with the development of a quantitative, physiologically plausible, model of spatial vision. A """"""""test-pedestal"""""""" approach will be used. This approach compares the discriminability of patterns A and B to the detection threshold of the difference pattern, A - B. This approach makes better predictions for several hyperacuity, resolution and phase discrimination tasks than assumption-ridden filter models. The proposed research will use deviations from the """"""""test-pedestal"""""""" predictions to develop an improved filter model of spatial vision. The proposed research seeks to gain a better understanding of the nature of the visual losses in the periphery and in amblyopia. To a first approximation the visual deficits of strabismic amblyopes resemble those of normal periphery and the vision of anisometropic amblyopes resemble the deficits caused by optical blur and by degraded contrast. This simple story is not satisfactory, as will be pointed out in the following list of proposed experiments to be done as part of developing a comprehensive spatial vision model. The proposed experiments will clarify several ambiguities of present models. a) Discrete and """"""""highest"""""""" spatial frequency mechanisms. The proposed experiments will settle once and for all the question of whether there are discrete channels in foveal and peripheral vision. Of special interest is the question of whether there is a """"""""highest channel"""""""". There are theoretical arguments for the existence of a highest channel in peripheral and amblyopic vision. b) Spatial sampling. There is presently no satisfactory method of modeling the cortical sampling grain. Sparse sampling is a likely factor in peripheral and strabismic amblyopic loss. Several of the proposed experiments explore this possibility. c) Gain control and local nonlinearities. Probably the main ingredient missing from present models of spatial vision is a robust understanding of how the sensitivity of mechanisms depends upon local luminance fluctuations. Several experiments will separate out the effects of luminance gain control from the effects of local masking. The possibility of a degraded contrast gain control in anisometropic amblyopes will be tested. d) Phase effects, Phase effects have been found in amblyopes and in peripheral vision that defy the predictions of present models. The proposed experiments should clarify these issues and also a controversy about whether the losses are really phase losses or position losses. e) spatial frequency vision, Stimuli consisting of repetitive features will be used to investigate visual processing within 2 octaves of the cutoff spatial frequency where amblyopes and periphery are most disadvantaged. These stimuli will be simple both in Fourier space and in feature space making them easier to model. f) Use of motion to distinguish between competing theories. Many position tasks can be accomplished either by attending to the output of a single mechanism, or by comparing the outputs of multiple mechanisms. With moving stimuli the comparison method is degraded, allowing one to distinguish between methods.
Showing the most recent 10 out of 82 publications
