Acquiring the cognitive skills necessary to perform effectively as a member of a tactical decisionmaking team is neither a smooth nor a consistent endeavor. In order to extend training technology into a more dynamic domain we have created a system that utilizes expert defined problem solving skills and strategies, and compares them to those used by the trainee. Trainee models are inferred on the bases of monitored trainee behaviors and the use of probe techniques (such as verbal reports or questioning). Concurrence and divergence between the trainee and expert models, assessed as a function of outcome (was the answer correct and was it gained using a process similar to that of an expert), serves as the basis for feedback and skill building. Such systems could be embedded within the operational context to meet "train like you fight, fight like you train" requirements. This new generation of training systems is referred to as Intelligent Embedded Trainers (IET).
One ongoing program directed by the Naval Air Warfare Center Training Systems Division is to develop a standard, modular architecture for the development of IET systems. Critical aspects of the architecture include the use of a proven process model of human decision making and flexible knowledge engineering/artificial intelligence techniques in combination with structured training objectives, cognitive feedback techniques, performance assessment and tracking methods. The objectives of this paper are to describe the architecture used, outline the functional modes for development and operation of IET systems, and to demonstrate how the architecture addresses shipboard electronic warfare training.