This symposium will explore the flexibility of our architectural assumptions in the light of increasing robotic sophistication. Robots provide perhaps the best opportunity for testing cognitive architectures in the real world, but can current architectures address the complexity that robotics brings to the table? In particular, this symposium will discuss the following questions – Can Cognitive Architectures provide the “brains” for a robot? What aspects of the architectural approach will (or won’t) work for robotic systems, and what requires significant modification? Finally, what new capabilities should architectures posses before they can be suitable for robotic high-level cognition?