Modeling Cognitive Strategies in Teaching

Sevan K Harootonian, Yael Niv, Thomas L. Griffiths, Mark K Ho

Track: Full track

Keywords: Cognitive Science, Inverse reinforcement learning, teaching, bayesian modeling

TL;DR: This paper investigates how people use different cognitive strategies, like effortful planning or simple heuristics, to teach effectively.

Abstract:

Teaching is a complex social behavior that sometimes results from goal-directed processing. However, goal-directed teaching is cognitively demanding since it requires actively assessing and correcting gaps in a learner's knowledge. When do people teach using such mentally effortful strategies versus falling back on more cognitively frugal ones? Here, we investigated this question using a combination of novel behavioral experiments and computational theory. We found robust individual differences in people's teaching strategies: some participants spontaneously teach using high-effort processing (e.g., Bayesian theory of mind and model-based planning) while others engage in low-effort processing (e.g., model-free heuristics). Our results and analyses provide a novel demonstration of how people engage in planning versus heuristics when teaching, as well as how people adapt processing to avoid mental effort in social interactions.

Submission Number: 29