Information transmission in normal vision and optogenetically resensitised dystrophic retinas

Abstract

The retina is a sophisticated image processing machine, transforming the visual scene as detected by the photoreceptors into a pattern of action potentials that is sent to the brain by the retinal ganglion cells (RGCs), where it is further processed to help us understand and navigate the world. Understanding this encoding process is important on a number of levels. First, it informs the study of upstream visual processing by elucidating the signals higher visual areas receive as input and how they relate to the outside world. Second, it is important for the development of treatments for retinal blindness, such as retinal prosthetics. In this thesis, I present work using multielectrode array (MEA) recordings of RGC populations from ex-vivo retinal wholemounts to study various aspects of retinal information processing. My results fall into two main themes. In the rst part, in collaboration with Dr Geo rey Portelli and Dr Pierre Kornprobst of INRIA, I use ashed gratings of varying spatial frequency and phase to compare di erent coding strategies that the retina might use. These results show that information is encoded synergistically by pairs of neurons and that, of the codes tested, a Rank Order Code based on the relative order of ring of the rst spikes of a population of neurons following a stimulus provides information about the stimulus faster and more e ciently than other codes. In the later parts, I use optogenetic stimulation of RGCs in congenitally blind retinas to study how visual information is corrupted by the spontaneous hyperactivity that arises as a result of photoreceptor degeneration. I show that by dampening this activity with the gap junction blocker meclofenamic acid, I can improve the signal-to-noise ratio, spatial acuity and contrast sensitivity of prosthetically evoked responses. Taken together, this work provides important insights for the future development of retinal prostheses.