, 2003, Segev et al., 2006, Zeck and Masland, 2007, Farrow and Masland, 2011 and Marre et al., 2012), in particular for extracellular recordings where the morphologies of the recorded neurons are not available. Similarly relevant as the question how ganglion cells integrate visual signals over their receptive field centers is the question how they pool signals in their receptive field surrounds and how center signals and surround signals are combined. high throughput screening Evidence for nonlinear interactions between center and surround comes from the finding that

the surround appears to act in a divisive fashion rather than in a linear, subtractive way (Merwine et al., 1995). Furthermore, it was observed that the effect of surround inhibition strongly differs for On-type and Off-type responses

of On–Off ganglion cells in the frog retina, pointing towards further intricate receptive field structure (Barlow, 1953). As discussed above, stimulus integration in the surround is an check details important component for specific ganglion cell types, in particular object-motion-sensitive cells and W3 cells. More generally, it may be interesting to see whether stimulus integration in the surround allows similar classifications as for the linear or nonlinear integration over the receptive field center. The models that have been used to describe nonlinear spatial integration in center and surround have been inspired by retinal anatomy, typically using bipolar cells as subunits, assumed to cover the receptive field of the ganglion cell in some regular fashion. Two recent

methodological advances ought to provide opportunities to bring this substrate for nonlinear integration in closer alignment with the actual circuitry. First, large-scale reconstructions at the electron-microscope-level can provide circuit diagrams for individual cells after they have been physiologically characterized (Helmstaedter et al., 2008, Briggman et al., 2011 and Denk et al., 2012). This may help relate the spatial only sub-structure of receptive fields to actual circuit elements on a single-cell basis. Second, physiological mappings of receptive fields at very high spatial resolution have shown that it is possible to identify the locations and identities of individual cone photoreceptors that provided signals for a measured ganglion cell (Field et al., 2010). It is conceivable that this can lay the foundation for detailed assessments of nonlinear transformations in the transmission from cones to ganglion cells, for example, by measuring iso-response stimuli when activating pairs of individual cones. The focus of this review has been on spatial integration. Yet, different nonlinear effects also occur in temporal integration by retinal ganglion cells.