Social behaviors involve interaction and communication between multiple individuals and are frequently crucial to the success and survival of an animal in its environment. These interactions range across sensory modalities, length scales, and time scales, and can be extremely subtle or complicated. Contextual effects based on social relationships and physical features of the environment can drive how an animal will behave. For example, actions of courting flies will appear wildly different from those in a same-sex pairing. The behaviors animals produce throughout interaction are difficult to quantify due to this complexity, as well as the visual occlusion that occurs when tracking interacting individuals.

I introduce a method for quantifying behavior of interacting fruit flies which combines high-throughput video acquisition and tracking of individuals with unsupervised learning methods for capturing an animal's entire behavioral repertoire. I characterize the behavior of flies placed in various social contexts and find behavioral differences in paired and solitary flies of both sexes, and identify specific behaviors that are affected by social and spatial context.

I introduce techniques to probe the structure of social behavior at various scales. Behavioral density maps provide a summary of behavior over long time scales and across many individuals, models of behavioral context dependence inform us how animals respond to specific configurations of their environment, and temporal feature analysis highlights how animals react to stimuli rapidly and often predictably. The special case of courtship interaction illustrates the subtlety and specificity of communication. I discuss how these results can refine our definition of behavior and lay the groundwork for modeling complex social interactions.