Retinal computations
Animal eyes do not transmit a raw picture of the world to the brain. Instead, the retina transforms visual input into a set of structured output signals carried by parallel channels of retinal ganglion cells. Each channel emphasizes particular features of the visual environment, effectively implementing a set of algorithms that compress and reformat visual information before it ever leaves the eye. These retinal computations determine what aspects of the world are highlighted for downstream brain circuits, and therefore strongly shape what an animal can perceive and respond to. Because different species inhabit very different visual environments and behavioural niches, the algorithms implemented by their retinas are often tuned to the specific problems those animals need to solve. Studying retinal output across species therefore provides a powerful window into how evolution tailors neural computation to ecological demands. But then, are these algorithms general-purpose strategies for efficient vision, or specialized solutions to species-specific challenges? Our work explores how retinal computations reflect the particular worlds that different animals need to see
