The goal of the proposal is to elucidate the neural computations carried out by the visual system to identify an impending threat and their use for the generation of collision avoidance behaviors. The proposal will develop advanced genetic, behavioral and imaging techniques to address these questions. The experimental data will be summarized in a biophysical model of the neural circuits generating collision avoidance behaviors that will be validated using the experimental data acquired during the project. The project will advance our understanding of how the visual system goes about reliably identifying a threat in the natural visual environment without reacting to irrelevant visual stimuli. The knowledge gained from the project is expected to allow the future design of efficient, neurally-inspired collision avoidance systems.
The proposed experiments will be carried out in the fruit fly Drosophila, a model system in which sophisticated genetics tools are available, including genetically encoded Ca2+ indicators and modifiers of neural activity that can be expressed in specific neural subpopulations. These tools, paired with the recent anatomical description of visual pathways at the electron microscopic level, offer the possibility of investigating how networks of neurons process information leading to visually guided escape behaviors at an unprecedented level of detail. In particular, these tools will allow (1) to carry out behavioral experiments where specific populations of neurons belonging to the visually-guided escape pathway are silenced; (2) to perform imaging experiments allowing to study the activation of the neurons belonging to the visually-guided escape pathway at all successive stages of the visual system and determine how/when the specificity for looming stimuli arises; (3) to apply localized stimuli and advanced microstimulation techniques allowing to isolate the contribution of individual photoreceptors to the processing of visual information related to looming stimuli in single neurons; (4) to develop genetic tools allowing to silence populations of neurons by using novel anion channel rhodopsins and allowing to sparsely label neurons of the pathway at two successive stages, either with an indicator of neuronal activity or with an optogenetic activator; (5) to test the functional connectivity at successive stages of the pathway using these tools in conjunction with three dimensional random access imaging; and (6) to model the neural computations carried out along the visually guided escape pathway.
