A zebrafish circuit for behavioral credit assignment

An important question in systems neuroscience is how the brain links choices and their outcomes to modify future behavior. Despite being a fundamental computation, we lack explicit circuit-level models of how this is achieved. Here, we provide evidence for a cellular substrate underlying this computation in a vertebrate brain. We develop an operant thermoregulatory assay in larval zebrafish, and show that adaptive responses depend on the precise temporal coincidence between choices and contingent thermal feedback. Chemogenetic ablations reveal that the dorsal habenula (dHb), the homolog of the mammalian medial habenula, plays an essential role in this adaptive behavior. Functional and anatomical mapping of the downstream circuitry reveals that the intermediate interpeduncular nucleus (iIPN) integrates choice- and temperature-related signals, conveyed by the dHb and a prepontine hindbrain nucleus, respectively. While dHb axons exhibit a diffuse connectivity pattern, hindbrain neurons project onto specific, lateralized iIPN regions. We show that these iIPN glomeruli function as sensorimotor detectors, integrating choices and outcomes. Specifically, when a swim in a given direction is closely followed by a temperature change, iIPN activity shows increased responses. We further show that this coincidence detection, witha precise temporal profile, links an action to its subsequent sensory feedback, relies on a presynaptic mechanism within the terminal segments of dHb axons. This mechanism is mediated by GABA _B -dependent modulation of dHb sensory axons by choice-related prepontine GABAergic neurons. In this study we are therefore able to link an adaptive behavior to a specific circuit mechanism.