Highlights d The metabotropic glutamate receptor MGL-1 is necessary for local search behavior d Parallel chemosensory and mechanosensory circuits redundantly generate local search d Sensory neurons encode information about the time since the last food encounter d The motor circuit for local search becomes resistant to sensory input over time

SUMMARYForaging strategies that enable animals to locate food efficiently are composed of highly conserved behavioral states with characteristic features. Here, we identify parallel multimodal circuit modules that control an innate foraging state --local search behavior --after food removal in the nematode Caenorhabditis elegans. Two parallel groups of chemosensory and mechanosensory glutamatergic neurons that detect food-related cues trigger local search by inhibiting separate integrating neurons through a metabotropic glutamate receptor, MGL-1. The chemosensory and mechanosensory modules are separate and redundant, as glutamate release from either can drive the full behavior. Spontaneous activity in the chemosensory module encodes information about the time since the last food encounter and correlates with the foraging behavior. In addition, the ability of the sensory modules to control local search is gated by the internal nutritional state of the animal. This multimodal circuit configuration provides robust control of an innate adaptive behavior.