Spatial integration of visual signals in ganglion cells of the mouse retina

• Applicant: Professor Dr. Tim Gollisch
• Subject Area: Cognitive, Systems and Behavioural Neurobiology
• Term: from 2013 to 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 247302823

Ganglion cells of the vertebrate retina collect visual signals over their spatial receptive fields. This spatial integration is often described as a simple linear filtering operation. Yet, several recent examples have shown that certain types of ganglion cells obtain their specific functional characteristics by integrating stimuli across their receptive fields in nonlinear fashion. However, little is known regarding the functional characteristics of these spatial nonlinearities. A key problem for experimentally assessing spatial nonlinearities is that these have to be separated from subsequent, cell-intrinsic nonlinear operations, such as the transformation of the integrated signals into spike trains by the ganglion cell itself. To overcome this challenge, we have recently introduced the method of measuring iso-reponse stimuli to elucidate spatial integration by retinal ganglion cells. By identifying different spatial patterns that yield the same responses of a measured ganglion cell, we can infer the underlying nonlinear transformations that govern spatial integration. After having applied this method to analyze spatial integration by ganglion cells in the salamander retina, we have now developed new spike detection and sorting methods which will allow us to extend the scope of these analyses to a mammalian system, the retina of the mouse. We will aim at characterizing spatial integration not only in the receptive field center of the ganglion cells, but also in the surround as well as in the interaction of center signals and surround signals to obtain a comprehensive picture of spatial integration in the retina. These investigations will be based on extracellular recordings of spiking activity from ganglion cells in isolated mouse retina. The visual stimulation of the retina will be coupled to the incoming recorded data in a closed-loop fashion in order to actively search for stimulus patterns that yield the same predefined response level of a cell. We will further connect the identified spatial integration characteristics to standard ganglion cell subtype classifications. Finally, we will study the functional roles of nonlinear spatial integration for vision by developing computational models for different ganglion cell types and comparing their performance to those of standard linear-filter models and to recordings obtained from ganglion cells under stimulation with natural images. Together, these investigations aim at helping provide a new perspective on signal processing through receptive fields and underscore the importance of nonlinear spatial integration in the retina.

• DFG Programme: Research Grants