Most animals rely on olfaction to find resources such as food, mates or hosts, which poses difficult challenges. Target odors occur in turbulent plumes and quickly break into thin filaments that intermingle with background odors. Our objective is to investigate how the brain solves the fundamental underlying problem of odor-background segregation in natural olfactory environments. We will use honey bees to test the following hypotheses: 1) The insect brain exploits fast stimulus dynamics for odor-background segregation, as target and background odors arrive at different times; 2) the inhibitory network in the first olfactory brain area (antennal lobe) is strongly involved in odor-background segregation; 3) learning-induced changes of neuronal odor representations support odor-background segregation. We will measure the temporal structure of natural plumes of multiple odorants originating from separate sources with a flying vehicle equipped with moth antennae. We will use this information to mimic the temporal structure of natural multi-odorant plumes in behavioral and physiological experiments with honey bees. We will use behavioral experiments to determine the timescales relevant for odor-background segregation, investigate the role of inhibitory networks in the antennal lobes with pharmacological treatments and RNA interference (RNAi), and investigate the effect of odor learning on odor-background segregation. In parallel we will determine the temporal resolution of olfactory receptor neurons and measure neuronal responses to naturalistic odor-background stimuli in the brain. We will combine neuronal recordings in the brain with pharmacological treatments and RNAi to investigate the role of inhibitory networks in the antennal lobe in encoding of target and background odors, and we will investigate the effect of odor learning on the encoding of target odors. We will use computational models of the olfactory pathway and the premotor center to explore mechanisms of odor segregation based on our experimental observations and validate the models by testing predictions in verification experiments. Finally, we will use the resulting computational model to enable an autonomous flying robot to perform odor-background segregation and source localization in natural environments.
