From noisy sensory signals to reliable decisions: neural computations for navigation in the Drosophila larva

While behavioral strategies for chemotaxis have been characterized across phyla, the neural computations underlying navigational decisions remain poorly understood. Through a combination of electrophysiology, behavioral analysis and computational modeling, we investigate how the fruit fly Drosophila larva processes olfactory signals during navigation with its compact nervous system of fewer than 10,000 neurons. Using optogenetic manipulations in virtual sensory environments, we demonstrate that larvae achieve robust chemotaxis by temporally integrating olfactory information, even in the presence of significant sensory noise. Our results reveal an evidence accumulation mechanism that parallels recent findings in the adult fly and mammals, suggesting the existence of conserved neural computations for perceptual decision making across developmental stages and species. This work helps explain how a numerically-simple brain can make reliable decisions in natural environments where sensory information is noisy and uncertain.