Population coding in the retina
Broadly speaking, I am interested in how network of neurons together can code visual information and be the basis of our perception. Although I have worked on the visual cortex in the past, now I tackle this issue at the level of the retina. This thin piece of tissue at the back of our eye, with several layers of cells, is much more than a camera. The last decade has shown that it performs many complex computations, and we still have to understand how the light impinging the photoreceptors is transformed by the retinal network into a set of spike trains that is sent to the brain. Fortunately, recent experimental techniques have made the quantitative study of this retinal code possible. So the retina is a good compromise between complexity of the computations and simplicity of the experimental access to study population coding in the visual system.
On one side of the retina, the photoreceptors receive the visual scene and transduce the light into electrical currents. This electrical activity is then processed by the retinal network through different layers of cells down to the ganglion cells. The last layer is composed of cells that will send their spike trains to the brain. My purpose is to understand the retinal "code", i.e. characterizing the transformation of the visual stimulus into the spike trains sent to the brain. For that purpose, I use an interdiscplinary approach combining large-scale recordings of the retinal activity, and theoretical tools to make sense of these data. This work can be divided in three complementary questions. Decoding: given the spike trains received by the brain, how can we reconstruct what happened in the visual stimulus ? Encoding: For a given stimulus, can we predict the response of the ganglion cells ? Statistics: can we understand the statistical organization of the retinal "words" ? To address these issues I have also developed large-scale recordings and data analysis to record the activity of hundreds of ganglion cells simultaneously. Ultimately, I would like to test new concepts and develop new tools that will not only help understanding population coding at the retina level, but also in other sensory structures like the sensory cortex.
While this work is primarily carried out in the normal retina, I am also collaborating with other research groups in the Vision Institute to understand the neural computations still at work in degenerated retinas whose light sensitivity has been restored by optogenetics, with the hope to find treatments for blindness in the long-term.