Submarine navigation and flight control tasks are difficult for humans to learn because they require the coordination of a task strategy (evasive maneuver) and the skills to implement that strategy (a visualmotor servo-loop). Under the Hybrid Learning Program of the Office of Naval Research, Devika Subramanian helped to create the NRL Navigation Task: design algorithms that learn probabilistic models of control policies used by humans from non-invasive recordings of their visualmotor activity. These models can be used by human or machine trainers to shape learning and also shed light on how humans acquire tasks with strategic and visualmotor components. From the Series:CSE Colloquia - 2002