Diversity-regulated cumulative cultural evolution

Kenji Itao, Assistant Professor, Tohoku University, Tokyo

As living systems evolve through natural selection and mutation, cultures evolve through teacher selection and learning. Cumulative cultural evolution enables knowledge and skills to increase in complexity across generations and is considered a key factor in human success. In biological evolution, strong selection can accelerate short-term adaptation but may trap populations at local optima on rugged fitness landscapes. Here, I show that weaker selection can also benefit cultural evolution, even on a smooth landscape. I model the social learning of partially observable input–output relationships, in which individuals learn by mimicking teachers’ outputs and through individual trial and error. Strong teacher selection initially accelerates learning but rapidly erodes cultural diversity, reducing the long-term effectiveness of social learning. Moderate randomness in teacher selection instead preserves useful variation and sustains cumulative improvement. I further examine how optimal teacher diversity depends on task complexity, population size, and the rates of social and individual learning.