Deconstructing a behavioral state: parallel neural integrators control distinct features of an aversive behavioral state in C. elegans

Transient sensory stimuli can trigger long-lasting changes in behavior and cognition. For example, repeated aversive experiences can lead to persistent changes in arousal and mood. Specialized integrator circuits can store information about recent experiences, but how they are embedded within the architecture of the brain to control internal brain states remains poorly understood. We find that C. elegans accumulates evidence of past aversive experiences over minutes to generate a scalable behavioral state consisting of multiple behavioral changes. Using a brain-wide imaging screen, we identified a set of neurons that integrate aversive sensory stimuli over different timescales, from seconds to minutes. Neuronal perturbations show that these integrator neurons are critical for the aversive behavioral state. They control distinct behavioral features, such as changes in locomotion speed or increased responsiveness to subsequent aversive stimuli. Moreover, the integrator neurons function in parallel to one another, utilize different mechanisms to generate persistent neural activity, and signal through different transmitters. Our result show how a behavioral state can be decomposed into distinct behavioral features that map onto different neural integrators that act in parallel. Combined action of the integrators then generates the full set of behaviors that comprise the global state.