The Structure of Mouse Behavior

Abstract

Complex animal behaviors are likely built from simpler modules, but their systematic identification in mammals remains a significant challenge. Here we use depth imaging to show that three-dimensional (3D) mouse pose dynamics are structured at the sub-second timescale by using a newly developed 3D imaging and machine learning-based automated phenotyping system, which we term Motion Sequencing (MoSeq). Computational modeling of these fast postural dynamics effectively describes mouse behavior as a series of reused and stereotyped modules with defined transition probabilities, which collectively encapsulate the underlying structure of mouse behavior within a given experiment. By deploying MoSeq in a variety of experimental contexts, we show that it unmasks strategies employed by the brain to generate specific adaptations to changes in the environment, and captures both predicted and previously-hidden phenotypes induced by genetic, neural, and pharmacological manipulations. We directly compare the predictive power of behavioral representations built by MoSeq against traditional measurements of behavior, including speed, length, and allocentric position, and demonstrate MoSeq is able to discriminate between subtle pharmacological manipulations of behavior, while traditional methods are not. This work demonstrates that mouse body language is built from identifiable components and is organized in a predictable fashion; deciphering this language establishes a framework for characterizing the influence of environmental cues, genes, neural activity and pharmacology on behavior.