The vertebrate retina is the gateway to the visual world. Retinal ganglion cells (RGCs), carry parallel streams of information to the visual areas of the brain, where it is interpreted to guide behaviour. A full catalogue of the types of RGC and precisely what information they encode is incomplete. Moreover, how populations of neurons interact in the brain to give rise to a coherent percept of the visual world is unknown. Technological advances in functional imaging promises progress, allowing the recording of activity across populations of neurons during visual presentations. However, methods to interpret such rich datasets are lacking. Using confocal and lightsheet imaging of larval zebrafish expressing genetically encoded reporters of neuronal activity, this thesis first investigates the spatial selectivity of RGCs and tectal neurons, and asks whether those selective for specific directions also encode information about spatial scale. This revealed three discrete populations of RGC and tectal neurons able to distinguish between small (8°), medium (17°), and large (27°) gratings, and an apparent specialisation of direction selective units to fine spatial detail corresponding to prey sized objects (8°). An unbiased, non-parametric clustering approach is then developed to group neuronal responses by their similarity without previous knowledge of the number or character of groups required, addressing a problem in the field’s ability to interpret large scale datasets. This approach is used to dissect responses to a wide variety of visual stimuli, identifying several novel types of RGC and tectal neuron. It also provides further evidence for a retinal origin for the identification of prey, and an intratectal computation generating responses to an aversive stimulus. This study provides some new insights into a major unanswered question in visual neurobiology; whether behavioural responses to salient stimuli is driven by specialised neurons in dedicated visuomotor pathways, or by coincident activity of multiple feature channels.
|Date of Award
|1 Dec 2017
|Martin Meyer (Supervisor), QueeLim Ch'ng (Supervisor) & Eric Blanc (Supervisor)