The design of control systems used in many robotic foraging systems bear a remarkable similarity to the models of behavior biologists suggest for social insects. This is no accident of course; roboticists have been "stealing" from the biological literature for years. In particular, social insect systems provide a number of useful mechanisms for coordinating effective multi-agent behavior using only local interaction rules. In this paper, however, we will turn the tables to show how computer science and robotics technologies can be applied to the task of understanding social insect behavior. We will also examine the feasibility of automatically "extracting" models of behavior through observation, then running those models on teams of mobile robots.