A Model of Conserved Global Neuronal Dynamics Predicts Future Behaviors in Caenorhabditis Elegans

It is now possible to simultaneously record activity from hundreds of neurons. In some simple organisms such as nematode C. elegans, whole brain imaging with single neuron resolution has been successfully performed. Here we show how such complex datasets can be used to generate quantitative models of the nervous system at the behaviorally-relevant scale. Our model predicts the timing of changes in locomotor direction at least 15 seconds prior to the event. These predictions are valid for individual instances of locomotion and can be applied across individuals including those not used in model construction. In addition, the model predicts behavioral dwell time statistics, sequences of behaviors, and neuronal activation. To develop this model we extracted loops in the trajectories spanned by neuronal activity using novel methodology. The model uses only two variables: the identity of the loop and the phase along it. Current values of these macroscopic variables predict subsequent behaviors. Remarkably, our model based on macroscopic variables succeeds despite consistent differences in activation of individual neurons and neuronal populations between different C. elegans. Thus, our theoretical framework reconciles the variability of neuronal activation amongst individuals with global dynamics that operate universally.