PROJECT SUMMARY (See instructions): Multimodal integration is a fundamental feature of animal nervous systems that allows them to extract useful information from complex environmental signals and guide behavior. A defect in this process may lead to communication disorders. However, there is currently limited understanding of this pervasive phenomenon. Due to its simplicity, the honeybee antennal lobe (AL) provides an excellent system for studying sensory integration. Chemical (i.e., odor) and mechanical (i.e., wind speed) information converge within the AL, likely in service of two intermingled tasks facing honeybees - tracking highly turbulent odor plumes and discriminating odors. Both are critical for foraging success. This proposal seeks to: (1) determine the impact of mechanical input (wind speed) on AL odor responses and odor classification; (2) determine the functional roles and mechanisms of multisensory integration within the AL. We postulate that the AL can switch between two distinct dynamic regimes – an odor tracking regime (triggered by high mechanical input) and an odor discrimination regime (triggered by low mechanical input). In other words, input from one modality affects the coding scheme of the other. To test this hypothesis, our experimental work will entail a suite of electrophysiological experiments that disentangle the contributions of each modality to AL dynamics, determine the impact of mechanical input on correlations across the AL, and assess the dependence of AL odor classification on mechanical input. Computationally, we will construct a realistic, experimentally benchmarked spiking network model of the AL integrating mechanical and olfactory inputs, and use it to study the network mechanisms that underlie AL dynamics within the two postulated regimes. The model will be used to explore conceptual ideas and generate specific hypotheses that will be tested in subsequent experiments. Finally, we will incorporate the fundamental principles uncovered in our work into novel machine learning algorithms for solving multimodal problems. The PIs are excellently suited for the proposed work – Dr. Lei is an expert in olfaction and the electrophysiological studies of the AL, Dr. Patel has extensive experience in biologically realistic modeling of AL dynamics, and Dr. Bazhenov is an expert in computational neuroscience, data analysis and machine learning.