EasyCog: Affordable and Easy to Develop Cognitive Models Navy SBIR FY2009.1 Sol No.: Navy SBIR FY2009.1 Topic No.: N091-086 Topic Title: EasyCog: Affordable and Easy to Develop Cognitive Models Proposal No.: N091-086-0321 Firm: Charles River Analytics Inc. 625 Mount Auburn Street Cambridge, Massachusetts 02138-4555 Contact: Peter Weyhrauch Phone: (617) 491-3474 Web Site: www.cra.com Abstract: Applications that incorporate Human Behavior Models (HBMs) have the potential to improve the training, decision-making, and even the operational effectiveness of US forces. This software can mimic human decision-making, reasoning, learning, social and cultural biases, as well as perceptual, cognitive, and motor limitations in support of many applications, including analysis, simulation-based acquisition, training, and mission rehearsal. Unfortunately, the development of such complex software systems is currently costly, time consuming, and error-prone due to a lack of appropriate high-level languages and corresponding development tools. To improve the affordability, efficiency and robustness of the HBM-development process, we propose to design and demonstrate the feasibility of EasyCog, a high-level cognitive language, compiler, and development environment that is designed to improve the HBM-development process across multiple cognitive architectures. EasyCog combines cognitive, behavioral, social, affect, and cultural models with high-level programming language features to improve the breadth and scalability of HBMs. Visual Editors and Debuggers work at the level of the high-level language to improve the efficiency of model development. Finally, EasyCog can generate configurable source code for a variety of existing cognitive architectures, meaning runtime users can reuse models in a variety of domains based on a single high-level model. Benefits: EasyCog lowers the cost of developing HBMs through a high-level language that combines cognitive, behavioral, social, affect, and cultural models with high-level programming language features, and generates configurable runtime models that lets end users reuse models in a variety of domains based on a single high-level model. One concrete government/military application is synthetic ASW training for the Navy, though there are potential applications in all branches of the U.S. military, as well as other government organizations, such as FEMA, homeland security, FBI, or even local police. Incorporating novel HBM-development techniques and tools into our commercial AgentWorksT product line will make it more marketable to medical and financial customers who can also benefit from affordable HBM-based training and decision-support applications. Return