Mark Ho, Princeton University

People can teach one another and can learn from teaching. These capacities are important for acquiring skills and learning about how the world works. But how does intentional teaching work? What are the cognitive and computational mechanisms that support teaching others how to do things or how to act? In this talk, I will discuss two types of non-verbal teaching interactions: teaching by demonstration and teaching with reward and punishment. Understanding these forms of interactive teaching requires characterizing dependencies between teachers, learners, and their shared environment. It also requires specifying how teachers and learners reason about those dependencies over time. To elucidate the dynamics of such processes, I draw on mathematical tools from optimal control and probabilistic epistemic logic. This approach leads to a number of empirical insights. In particular, a teacher’s intention to convey information and learner’s recognition of that intent play a crucial role, even in non-verbal interactions involving demonstration and rewards. This has broad implications for our understanding of social learning in people as well as the design of artificial social learners.