The flexibility of generic architectures: Lessons from the human nervous system

Abstract

Many engineered systems are biologically inspired. In this paper, we examine the structure of the human nervous system with an eye toward understanding how its internal architecture may inform the design of large-scale engineered systems. Specifically, we examine four types of "generic" architectures - treestructured hierarchies, layered hierarchies, diffuse networks, and teams. We observe all four types of these hierarchies in the human nervous system. Consistent with prior theory, tree-structured hierarchies are relatively inflexible, but simple and easy to control. Layered hierarchies are moderately flexible, more complex, yet still largely controllable. Diffuse networks are easy to describe and therefore relatively simple yet flexible; however, they can lead to unexpected emergent behaviors undermining controllability. Finally, team structures are extremely flexible, but can lead to instabilities. Implications for system design are discussed.