Are dendritic integration rules in retinal ganglion cells adapted to the statistics of the natural environment?

01/12/2019 to 31/05/2020

Animals live in environments in which different visual information is present in distinct parts of the visual field and, hence, falls onto different retinal regions. For example, the world of many terrestrial animals is bisected by the horizon: For mice, predators are more likely to appear in the upper visual field, whereas food is rather found in the lower visual field. To encode survival-relevant information more efficiently and robustly, the visual system is thought to be adapted to the statistics of the natural environment. This adaptation starts already in the retina, which often can be divided into different zones: in these, for example, the spectral composition of the photoreceptor population seems to be specialized for distinct tasks, such as raptor detection in mice or hunting paramecia in zebrafish larvae. Currently, we are only at the beginning of understanding these region-specific adaptations. However, without an in-depth knowledge of the regional circuits and the underlying adaptational mechanisms, a comprehensive understanding of how the retina processes an animals’ natural visual environment is difficult.In this project, we will focus on the output neurons of the retinal network, the retinal ganglion cells. In these cells, changes in dendritic morphology are often a tell-tale sign for regional specializations. Therefore, we will study the rules according to which ganglion cell dendritic arbours integrate input from the presynaptic circuitry and evaluate if and how these rules change across the retina. We then ask how such adaptations could subserve efficient processing of the natural visual environment.Specifically, using functional dendritic imaging, morphological reconstruction, and computational modelling, we will study how different types of mouse RGC process the synaptic input of the upstream circuitry and how regional variations in an RGC type’s dendritic integration profile may contribute to encoding the natural scene encountered by this retinal region. We will capture the essence of these adaptions in "minimal" biophysical RGC models which allow us to match measured data from different retinal regions and, thereby, extracting important parameters that determine RGC function. In addition to insights into the mechanisms of dendritic integration and processing in RGCs, we expect to further our understanding of how RGCs contribute to the generation of the more than 32 output channels of the retina (Objective 1). We expect to learn how and to what extent the dendritic integration function of selected types of RGC changes with retinal location and visual field position-dependent natural input statistics (Objective 2) and extract the underlying general principles of these adaptations using minimal biophysical models (Objective 3).