COGNITECTURET: Cognition-based, Integrative Architecture for Next Generation Intelligent Agents

Computational architectures for intelligent agents and human behavior models have advanced to the point where one can benefit from the other in creating more realistic Intelligent Synthetic Force behaviors. This Phase I effort is directed to analyzing therespective strengths and weaknesses of agent architectures and cognitive/human behavior models with a view to incorporating high payoff cognitive/behavioral constructs into agent architectures. To this end, a comparative framework will be constructed toexplore the strengths and limitations of candidate agent architectures and human behavioral models with a view to identifying and exploiting common features, combining unique strengths, and circumventing the limitations of each within a next generationagent architecture. The resultant architecture will improve both the fidelity and validity of distributed training and wargaming simulations while facilitating the design, development, and deployment of agents. Psychologically-principled, socially-awareagent architectures will enable rapid development of increasingly usable systems in both the military and commercial sectors. In addition, distributed training, distributed mission rehearsal, wargaming, virtual reality systems development and constructivesimulations, all stand to benefit from these next generation agent architectures.