- This event has passed.
Thursday Morning Lecture with Falk Lieder: “Understanding and Improving Human Learning and Decision-Making”
One of its most remarkable features of human intelligence is the mind’s ability to discover and continuously refine its own algorithms. This enables people to discover clever heuristics for mastering most everyday decisions very efficiently. But some less familiar situations require different decision strategies that many people haven’t had a chance to discover yet. In this talk I will illustrate how investigating how people learn how to decide can enable advances in artificial intelligence and open up new avenues to improving human decision-making. This line of work started with reverse-engineering how people discover efficient planning strategies. We found that many aspects of how people’s decision strategies change with experience can be understood in terms of metacognitive reinforcement learning. The resulting cognitively-inspired learning algorithms make it possible to discover planning strategies that reach a super-human level of computational efficiency and outperform existing planning algorithms. Teaching these automatically discovered strategies to people significantly improved their performance in simple planning problems. Encouraged by these findings, we have scaled up our approach to larger and more complex sequential decision problems, made it robust to uncertainty about the environment, and extended it to generating human-interpretable descriptions of optimal planning strategies in the form of flowcharts and procedural descriptions. These advances make it possible to improve human decision-making in a wider range of decision problems. I will close with an outlook on improving goal-setting, goal pursuit, and helping people learn how to make better decisions in the real world.
***Want to attend this lecture? Click here