The long term goal of this project is to elucidate visual performance by blind individuals whose vision is partially restored through a prosthesis. Phosphenes (localized dots or blobs of light) elicited in blind individuals, through implanted electrodes in visual cortex or through intra-operative retinal stimulation, have been reported. In this project, in narrow collaboration with retinal and cortical prosthesis research teams, phosphenes will be simulated as blob-like light stimuli to normally sighted subjects. Each simulated phosphene represents a stationary electrode, and must thus be stabilized in the visual field: It moves with the observer's direction of gaze. In this project, this will be accomplished through eye movement compensation, either of a defocused mask in a dual Purkinje eyetracker through which the subject observes a monitor screen- with small openings in the mask acting as phosphenes-or of a virtual mask in a head-mounted display system (Low Vision Enhancement System and two successors). Simulated phosphenes will be presented in arrays, initially with small (16-100), then with larger (144-400) and finally with large (up to 1,024) phosphene numbers, and with various layouts: """"""""retinal"""""""", i.e. minimally distorted, and """"""""cortical"""""""", i.e., transformed according to the striate cortical projection map. Strategies for reconstruction of irregular cortical phosphene maps will be evaluated in these sighted subjects. The project will address issues crucial to prosthetic vision: How many phosphenes, in what spacing and configuration, are required for specific vision tasks? What are the effects of irregularities and distortions in the mask, and of missing phosphenes? Does fading due to image stabilization occur; can it be prevented with eye movements; if not, what other methods work? Can tasks be carried out with few (4-8) gray levels? Representative visual tasks-letter recognition, reading, object recognition, eye-hand coordination, and mobility through virtual and real mazes-will be performed. Tasks will be computer-generated, but in years 4 and 5 will include live input from a head-mounted wide-field camera.
