Abstract

Animals are able to accumulate sensory evidence over considerable timescales in order to select behaviors fundamental for their survival. Despite the importance and ubiquity of this phenomenon, how activity in different brain regions contributes to this process is not understood. In this study, I develop a novel perceptual decision making assay in the larval zebrafish, based on whole-field visual motion of varying strength. Upon presentation of motion, fish integrate this noisy sensory evidence in time before swimming in the direction of perceived motion, a behavior known as the optomotor response. Behavioral parameters such as the latency to initiate swimming and the fraction of correct turns are modulated by motion strength. Whole-brain functional imaging experiments with single-cell resolution enable identification of almost all neural activity relevant to the different stages of the decision making process, including evaluation of momentary sensory input, accumulation of this sensory evidence, and behavioral output. Fitting a generalized integrator model to every neuron reveals a wide range of time constants, which are distributed in functional clusters across different brain regions. Based on the behavior and the imaging data, a model is proposed where integrating units set the left and right turning rates. An unbiased whole-brain analysis revealed that the interpeduncular nucleus, a circular structure located ventrally on the midline of the brain, reliably encodes these rates.