Alison Gopnik, Professor of Psychology and Affiliate Professor of Philosophy at the University of California, Berkeley, will deliver a plenary lecture, and will be introduced by H. Clark Barrett, Professor of Anthropology at the University of California, Los Angeles. Talk title and abstract below.
Childhood is evolution’s way of performing simulated annealing: Life history, variability and learning.
I will argue for a theoretical link between the development of an extended period of immaturity in human evolution and the emergence of powerful and wide-ranging causal learning mechanisms, particularly the use of causal models and Bayesian learning. In the past 15 years we’ve discovered that even young children are adept at inferring causal structure from statistical patterns. But can they also learn more abstract theoretical principles? And are there differences in the ways that younger children, older children and adults learn that might be relevant to our extended immaturity? I will present two case-studies showing that preschoolers can learn abstract higher-order principles from data. The examples involve abstract relations (same vs. different), and abstract logical forms (AND vs. OR). In each case, younger learners were actually better at inferring unusual or unlikely principles than older learners. I relate this to computational ideas about search and sampling and to evolutionary ideas about human life history. Our hypothesis is that childhood is evolution’s way of performing simulated annealing – a computational strategy in which an early period of broad exploration is followed by later more focused exploitation. Our distinctively long human childhood allows a period of broad “high-temperature” hypothesis search. The variability and noisiness that is so characteristic of young minds and brains has distinct cognitive advantages particularly in variable and unpredictable environments. It comes, however, with the cost of greater caregiving investment, and humans also seem to be particularly adapted for such investment with the distinctive “triple-threat” of pair-bonding fathers, grandmothers and alloparents.