The vertebrate retina supports two distinct functions: image-forming vision, which is concerned with the detection of spatial patterns and objects in visual scenes, and non-image-forming vision, which detects the overall intensity of ambient illumination. Non-image-forming vision influences mood, sleep and body temperature, entrains circadian rhythms to the solar day, and drives light-evoked behaviors. The primary sensory neurons that enable non-image-forming vision are a population of ganglion cells in the retina that express the photopigment melanopsin and are therefore called intrinsically photosensitive retinal ganglion cells (ipRGCs). IpRGCs come in multiple subtypes, distinguished by differences in their morphology, circuit connectivity, cellular physiology, and projection targets in the brain. Analogous to other RGCs, adult ipRGCs receive synaptic inputs from the retinal circuits that convey inputs from rod and cone photoreceptors, but are also uniquely capable of signaling light intensity to the brain in the absence of rod and cone inputs, such as in disease states that cause rod and cone degeneration or disruption of retinal circuits. Interestingly, ipRGCs signal light intensity during early development, prior to the developmental maturation of the synapses that relay information from rods and cones to RGCs .

The focus of this dissertation is to understand how the cellular specializations and circuit connectivity of ipRGC subtypes contribute to the encoding of light intensity in the developing retina and drive a specific non-image-forming function, an innate light avoidance behavior exhibited by newborn mammals. In Chapter 1, I summarize the current understanding of the neural implementation of non-image-forming-vision, with an emphasis on developing mammals. In Chapter 2, I use calcium imaging, an unsupervised clustering analysis, and genetic and pharmacological manipulations of gap junction coupling in the retina to investigate how the neonatal retina encodes light stimuli and whether cell-intrinsic properties or circuit connectivity determine the functional output of ipRGCs. I find that populations of ipRGCs and other retinal neurons in the developing mouse retina encode light stimuli as six functional groups that are mixtures of ipRGC subtypes and other retinal neurons. I also demonstrate that ipRGCs are anatomically and functionally gap junction coupled to one another and to other retinal neurons in a circuit arrangement that likely leads to functional mixing of subtypes and allows modulation of gap junction coupling to dramatically regulate the population light response. In Chapter 3, I investigate the neural basis of light avoidance behavior in neonatal mice using an automated assay of light avoidance behavior coupled with genetic and pharmacological manipulations of ipRGC phototransduction and gap junction circuitry. I find that a specific ipRGC subtype is necessary and sufficient for the behavior. Together, these results reveal the relative contributions of cell-intrinsic properties and gap junction circuitry of ipRGC subtypes in the developing retina to the encoding of light intensity and to light-evoked behavior.