Exploring the neural machinery of color vision: A bold fMRI-based investigation

The chromatic content of the environment acts as a powerful sensory cue to our visual system. Perceptually, we are intimately aware of associations that can be made based on chromatic variation within the visual scene. For instance, spotting a piece of ripe fruit amongst dappled foliage (or amongst other fruit at the market) is relatively trivial, yet intuitively quite powerful. The roles of chromatic cues in other aspects of vision are less intuitive but presumably still very important in guiding behavior. Chromatic signals are linked to circadian circuitry and it has long been appreciated that chromatic information can be used in the processing of motionbased visual stimuli. These are two examples of how chromatic information can be used for visual functions that are distinct from hue perception. The focus of this dissertation is chromatic signal processing in the retina and brain. Our hope is to show that understanding the many roles of chromatic signals is important, and by exploring these roles we can obtain a better understanding of color vision.;The short-wavelength sensitive (S-) cone photoreceptor generates a signal that is critically important for encoding chromatic information and there is evidence that neural circuit features associated with the S-cone are conserved across mammalian species. At the heart of this proposal is a straightforward circuit model, wherein, the following has classically been proposed: at the level of the retina, Scone signals are conveyed exclusively by metabotropic glutamate signaling (mGluR6) from S-cones to S-cone specific bipolar cells. Thus it follows that all centrally projected S-cone signals must traverse the mGluR6 mechanism. We tested the foregoing circuit model in humans (genetics) and animals (pharmacology) using blood oxygenation dependent functional magnetic resonance imaging (BOLD fMRI). BOLD fMRI is a powerful neuroimaging technique that allows whole brain function to be assessed at different temporal and spatial scales. This allowed us to not only test the current circuit model but also infer the role(s) of S-cone signals by evaluating the response properties of different brain areas across species.;We used the well-known mGluR6 agonist 2-amino-4-phosphonobutanoic acid to selectively block the putative pathway in animals. In humans, we studied patients with genetic mutations that render the mGluR6 signaling pathway nonfunctional. We presented diffuse stimuli that were chromatically modulated to stimulate only S-cones. In both 'lesion' models S-cone signals should have been absent from all downstream pathways. In each model studied S-cone responses were selectively attenuated, but not for all visual areas and/or experimental conditions. In animals, selective attenuation was observed in subcortical structures implicated in circadic functions: the olivary pretectal nucleus, pretectal area, pregeniculate, and ventral lateral geniculate nucleus. In human patients, dorsal stream visual areas were less responsive to S-cone stimuli modulated at high temporal frequencies. The attenuation of S-cone signals in both models is consistent with the mGluR6 theory and is further supported by central projections of 'S-cone carrying' ganglion cells. However, in all models tested we observed persistent Scone signals in the primary retinogeniculostriate pathways. These findings allow us to reject the current circuit model. That is, in addition to the mGluR6 mechanism, there must be an alternative mechanism-pathway that makes a major, if not sole contribution to the retinogeniculatestriate pathway. One potential candidate is the direct feed-forward GABAergic signaling from H2 horizontal cells. This pathway may be the primary source of S-cone signals supporting hue percepts.