Flexible Brains + Structured Bodies: A Developmental Framework for Biological and Artificial Intelligence

Abstract

What are the origins of intelligence? We argue that intelligence develops when flexible, domain-general brains adapt to experiences produced by structured, domain-optimized bodies. Brains are highly plastic and can assume many possible functional configurations, whereas bodies (e.g., retina, cochlea, skin, muscles) form relatively stable structures that shape the information available for learning. By filtering raw input into structured patterns, bodies generate the experiences that allow domain-general brains to develop domain-specific knowledge. Drawing on controlled-rearing studies of newborn animals and embodied AI models, we show how object perception—a core mental skill often viewed as innate—develops automatically when flexible neural networks adapt to first-person sensory experience. The body contributes at multiple stages: generating prenatal activity that establishes initial cortical organization, generating postnatal experiences that drive object learning, and structuring visual input to enhance object perception. Intelligence across biological and artificial systems emerges from the interaction between flexible brains and structured bodies.