Cognitive Architectures: Definition, Examples, and Challenges

A cognitive architecture integrates the components that support the epistemic functions of an agent. This entry presents an overview of cognitive architectures, starting from a conceptual and historical analysis. After the fall of behaviorism, the notion of cognitive architectures came to full fruition in the cognitive revolution of the second half of the twentieth century and the associated growth of artificial intelligence and its computer metaphor. In this entry, the focal point is the proposal of Allen Newell to formulate Unified Theories of Cognition as architectures. We look at pertinent proposals on cognitive architectures, followed by those with links to robotics. Subsequently, we analyze contemporary contenders from cognitive robotics. This entry shows that the investigation and formulation of cognitive architectures are still in an early phase. In response, the entry proposes a taxonomy for candidate cognitive architectures. This taxonomy is illustrated by showing how it can accommodate various and putatively contradicting candidate-cognitive architectures. The proposed taxonomy argues that arbitration and conflict resolution processes must be considered in addition to the components of the mechanisms underlying cognition. In this respect, architectures for advanced intelligent systems should not be considered monolithic and static but rather pluralistic and dynamic. Cognitive architectures are at the heart of integrating robotics, artificial intelligence, and the study of mind and brain. In turn, robotics is the platform to advance an integrated real-world, embodied, and situated perspective toward understanding and synthesizing intelligence.