Brain function relies on the joint dynamics of neural networks distributed across multiple brain regions. Functional light-sheet imaging allows for activity measurements in most of the 100.000 neurons in the brain of the larval zebrafish at single-cell resolution. We combined light-sheet techniques with a virtual-reality setup for larval zebrafish to allow for whole-brain cellular imaging during behavior. To analyze the resulting datasets, we developed analytical methods based on large-scale distributed computation, available as open-source software, which can be applied to the data to elucidate brain function in the context of sensorimotor transformations, motor learning and patterning of spontaneous behavior.

We applied both the setup and the analysis framework to try to locate the locus of modulation of spontaneous behavior. Spontaneous swimming follows well-defined temporal statistics, where the fish string together multiple turns in one direction before switching to the other. Correlating fictive turns to whole-brain activity isolated four clusters of cells in the hindbrain exhibiting activity tightly coupled to turning behavior. Lesion and activation studies support a causal role for this nucleus in the slow modulation of spontaneous behavior. The identification of this small ~100-cell nucleus controlling temporal patterning in spontaneous behavior forms an example of the utility of whole-brain neuron-level analysis, which we hope will be useful for shedding light on the neural basis of further behaviors in the future.