Problem-Solving Variability in Cognitive Architectures

Cognitive architectures provide a promising means to model human behavior in complex systems and problem domains. The fields of social simulation and cognitive architectures can be linked more effectively if cognitive variability can be modeled in a realistic way. In particular, if individual differences in the ways humans solve problems can be captured in computational models, the dynamic patterns of change and diversity in human systems can be explored in greater depth. Kirton's Adaption-Innovation theory provides a robust foundation for the study of creativity, problem solving, and decision making based on individual differences in cognitive level (capacity) and cognitive style (preferred approach) of problem solving. This paper examines four well-known cognitive architectures (SOAR, ACT-R, CLARION, and DUAL) in light of Adaption-Innovation theory to explore if and how cognitive style and level variables are manifested within them. This analysis leads to a proposed cognitive style continuum for cognitive architectures, as well as other possible architectural mechanisms to incorporate problem-solving variability.