Carnegie Mellon University
Ann Arbor, Michigan
Mature science and engineering disciplines have long converged on levels of description that best and most efficiently capture the set of phenomena on which they are focused. This convergence has led to the evolution of frameworks that allowed each discipline to build on the results of the others while focusing at different levels of abstraction. Obvious examples in the basic sciences are the different levels of focus in biology, chemistry, and physics. Such consensus has escaped up to now the study of human behavior, with various levels of description fighting for supremacy rather than building upon each other. However, recent developments in the cognitive sciences have started to lead toward such a convergence. Detailed study of neural anatomy and physiology reveal details of the computational properties of individual neurons and neural circuits. Functional brain imaging techniques are generating a wealth of data about brain organization. Integrated computational cognitive architectures have enabled precise modeling of quantitative human performance data of increasingly complex tasks in dynamic environments, including social interactions with other entities. Artificial intelligence techniques inspired by human cognition are being used to control robots and provide realistic partners and opponents in virtual reality training. Thus, these disciplines are starting to converge to create an empirical and computational framework in which to understand, model, and simulate human behavior at multiple levels of abstraction, from detailed neural simulations to high-fidelity cognitive models of behavior, to complex intelligent agents acting and interacting in dynamic environments.