Circuit motifs for sensory integration, learning, and the initiation of adaptive behavior in larval Drosophila
1 University of Cologne
Animals navigate life in complex and dynamic environments by seeking out advantageous circumstances and avoiding harm by employing adaptive approach or avoidance behaviors based on the sensory stimuli surrounding them. Such goal-directed behavior must be situationally appropriate and flexibly take into account the animal's internal state and physical abilities. We study the entire process, from the encoding of sensory stimuli to adaptive behavior, using computational models of the Drosophila larva olfactory pathway and the mushroom body as the primary insect learning center in combination with computer simulations of virtual larvae in their environment. The availability of the complete connectome, a restricted and traceable behavioral repertoire, established learning protocols, and the numerical simplicity of the entire neural system make the larva an ideal model organism. The expected outcomes of an encounter often drive stimulus approach or avoidance behavior. Such expectations can be based on inherent, innate meaning or learned from experience. We investigate the transformation from sensory input to behavior, starting with mechanisms of sparse, specific, and separable representation of sensory stimuli to adaptive expectation-driven behavior through biased execution of specific motor programs. At the core are computational motifs that continously compute expectations from experiences. If momentary expectations are not fulfilled, expectations and behavior are dynamically updated. These computational motifs can account for learning phenomena such as prediction error, delay and trace conditioning, higher-order learning, and conditioned inhibition. Funded by MGK-NRW (Networks 2021), BMBF (DrosoExpect), and DFG (SPP 2205)