In order to achieve an increasingly dynamic and nuanced commander's intent, warfighters must understand when and where to apply their skills most effectively. Current training methods, though very effective at producing skilled warfighters, focus primarily on lower level skills and outcome-based performance. However, there is a need to assess the warfighter at the level of intent and how the warfighter factors that into their process of skill selection and skill execution. Cognitive models appear as a promising solution to understanding warfighter processes and intent. Yet, traditional cognitive models designed to replicate human cognitive processes are cumbersome to develop and maintain, requiring large amounts of data.

An innovative capability was designed to address these challenges by leveraging advances in training technology that increase data availability to capture warfighter actions and behaviors during training while applying recent research findings focused on understanding intent from actions (Baker, Saxe & Tenenbaum 2007). This capability integrates a modeling method to infer intent from actions, by employing Markov Decision Processes and Bayesian inverse planning.

This paper will describe initial testing and evaluation of this technology with novice remotely-piloted aircraft operators and show the model's ability to infer intent and predict operator actions with a satisfying level of reliability. Initially implemented in a basic research setting, this modeling method is currently being transitioned to simulation and training environments with gradually increasing level of fidelity, beginning with an operationallyrelevant, game-based training environment. This paper will describe the transition plan and discuss how this modeling approach constitutes an example of a new generation of practical, lightweight, and extremely useful cognitive models.