Metagames and Imprecise Probability - Game Theory and Agent Behavior

Watch a Topos Institute Berkeley seminar exploring the intersection of agent behavior, recursive type equations, and game theory through the lens of imprecise probability and domain theory. Delve into how agents form beliefs about other agents, leading to a recursive relationship expressed as Agents = [Beliefs about agents → Actions]. Learn how combining imprecise probability with Dana Scott's domain theory provides solutions to these equations, creating a framework reminiscent of Nigel Howard's metagame theory. Discover applications in open-source game theory, examining both conventional and innovative agent types that emerge from this mathematical construction. Presented by Nisan Stiennon, this hour-long talk bridges theoretical foundations with practical implications for understanding complex agent interactions.