With EM, 2p-imaging and computational modelling we show that on the synaptic level of single neuron types in the retina there exist highly specialized mechanisms which are advantageous for the encoding of different visual features. 

Already in EM we see different geometries of the ribbon depending on the location of the UV-cones.

We confirm these differences by “dual-color” 2-photon imaging of pre-synaptic calcium and glutamate release of the synapse.

We use these recordings to build a computational model of the ribbon synapse, which reproduces region-specific response properties. An interactive model tool can be found on #github and you can run it on #googlecolab here: http://www.tinyurl.com/h3avl1ga

By using simulation-based inference we get full posterior estimations of the parameter distributions and can compare these between different retinal regions.

Finally, the computational model allows us to extrapolate to new stimuli and predict response behaviours of different synapse configurations. Hereby we identify principles which are advantageous for the encoding of different visual features, which may be more relevant in different locations of the visual field.

This work builds on previous findings, where we showed fovea-like photoreceptor specializations in UV Cones in zebrafish, which drives prey-capture behavior: https://www.cell.com/neuron/pdf/S0896-6273(20)30313-5.pdf