Alison Gopnik: UC Berkeley Department of Psychology

How do we accurately infer the causal structure of the world around us? Thirty years of developmental research has shown that human children form and revise intuitive theories of everyday physics, biology and psychology. These theories are similar to the formal theories of science. Recent work in philosophy of science, computer science and statistics has developed formal computational models of this kind of theory formation, particularly in the framework of causal graphical models or Bayes nets. I’ll argue that we can think of intuitive theories as causal maps, analogous to the spatial maps animals use in navigation. These maps allow us to consider counterfactual alternatives, design complex plans and act to change our environments, and they may be uniquely human. I will present evidence showing that preschool children construct such maps in ways that accord with the Bayes net formalism.
http://www.bec.ucla.edu/papers/Gopnik_4.6.07.pdf