Reasoning about Bounded Reasoning
Interactive decision-making relies on strategic reasoning. Two prominent frameworks are (1) models of bounded reasoning, exemplified by level-$k$ models, which keep reasoning implicit, and (2) epistemic game theory, which makes reasoning explicit. We connect these approaches by "lifting" static complete-information games into incomplete-information settings where payoff types reflect players' reasoning depths as in level-$k$ models. We introduce downward rationalizability, defined via minimal belief restrictions capturing the basic idea common to level-$k$ models, to provide robust yet well-founded predictions in games where bounded reasoning matters. We then refine these belief restrictions to analyze the foundations of two seminal models of bounded reasoning: the classic level-$k$ model and the cognitive hierarchy model. Our findings shed light on the distinction between hard cognitive bounds on reasoning and beliefs about co-players' types. Furthermore, they offer insights into robustness issues relevant for market design. Thus, our approach unifies key level-$k$ models building on clear foundations of strategic reasoning stemming from epistemic game theory.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Proceedings Eighteenth Conference on Theoretical Aspects of Rationality and Knowledge
The TARK conference (Theoretical Aspects of Rationality and Knowledge) is a biannual conference that aims to bring together researchers from a wide variety of fields, including computer science, artificial intelligence, …
Epistemic ReasoningPhilosophyBounded Conditioning: Flexible Inference for Decisions under Scarce Resources
We introduce a graceful approach to probabilistic inference called bounded conditioning. Bounded conditioning monotonically refines the bounds on posterior probabilities in a belief network with computation, and converge…
Proceedings Nineteenth conference on Theoretical Aspects of Rationality and Knowledge
The TARK conference (Theoretical Aspects of Rationality and Knowledge) is a conference that aims to bring together researchers from a wide variety of fields, including computer science, artificial intelligence, game theo…
Epistemic ReasoningPhilosophyDepth-Bounded Epistemic Planning
In this paper, we propose a novel algorithm for epistemic planning based on dynamic epistemic logic (DEL). The novelty is that we limit the depth of reasoning of the planning agent to an upper bound b, meaning that the p…
AI Mental Models: Learned Intuition and Deliberation in a Bounded Neural Architecture
This paper asks whether a bounded neural architecture can exhibit a meaningful division of labor between intuition and deliberation on a classic 64-item syllogistic reasoning benchmark. More broadly, the benchmark is rel…