In this talk, I will lay out a functional approach to diagrammatic reasoning, and propose diagrammatic representations and processes inspired in part by aspects of human cognition. In doing so, I will present certain differences between diagrams in the external world and diagrams that are "inside" the agent's head, that are important both for AI and cognitive modeling. I will present a system that accommodates these difference within the constraints of the cognitive architecture Soar. I will provide examples of how this augmented cognitive architecture, called biSoar, performs in AI as well as cognitive modeling tasks.

link: http://www.ukurup.com/media/files/ICCM07.pdf