Revealing inductive biases by simulating cultural evolution

People are remarkably good at acquiring complex knowledge from limited data, as is required in learning causal relationships, categories, or aspects of language. Successfully solving inductive problems of this kind requires having good "inductive biases" - constraints that guide inductive inference. Viewed abstractly, understanding human learning requires identifying these inductive biases and exploring their origins. I will argue that an effective way to solve this problem is to take a step outside the traditional methodology of cognitive psychology, which focuses on individuals learning from the world, and instead look at what happens when people learn from one another. A mathematical analysis using a simple Bayesian model predicts that inductive biases will determine the outcome of processes of cultural transmission in which learners learn from other learners. This provides the foundation for an experimental method in which cultural evolution is simulated in the laboratory as a means of magnifying the effects of human inductive biases. I will present the results of using this method to explore questions about both individual cognition and cultural evolution, and highlight some surprising connections to algorithms more commonly used in computer science and statistics.