Retinal neuroprosthetics can potentially be used to address some of the major degenerative disorders that cause blindness, including Retinitis Pigmentosa and Macular Degeneration, by bypassing the degenerated photoreceptor layer, and interfacing directly the more viable Retinal Ganglion Cells (RGCs). I will describe the development of new optical and computational tools aimed at allowing controlled experimental emulation of activity patterns in a large population of retinal ganglion cells and their correlation structure. First, we introduce new optical systems allowing control of increasingly complex spatiotemporal activity patterns in neural populations, focusing on holographic photo-stimulation which has several fundamental advantages in this application. Next, we introduce a general new computational strategy based on correlation distortions, for controlling and analyzing the pair-wise correlation structure (defined in terms of auto- and cross-correlation functions) in multiple synthetic spike trains. This approach can be used to generate stationary or non-stationary network activity patterns with predictable spatio-temporal correlations.
In a final part of the talk I will describe a new approach for exact, flexible control of neurite outgrowth in three-dimensional neural structures, and its possible applications.
