Algorithmic Rationality: Adding Cost of Computation to Game Theory JOSEPH Y. HALPERN and RAFAEL PASS Cornell University We summarize our work on a general game-theoretic framework for reasoning about strategic agents performing possibly costly computation. In this framework, many traditional gametheoretic results (such as the existence of a Nash equilibrium) no longer hold. Nevertheless, we can use the framework to provide psychologically appealing explanations to observed behavior in well-studied games (such as nitely repeated prisoner s dilemma and rock-paper-scissors). Categories and Subject Descriptors: F.m [Theory of Computation]: Miscellaneous; I.2.11 [Arti cial Intelligence]: Distributed Arti cial Intelligence Multiagent systems; J.4 [Social and Behavioral Sciences]: Economics General Terms: Economics, Theory Additional Key Words and Phrases: Nash equilibrium, costly computation 1. INTRODUCTION Consider the following game. You are given a random odd n-bit number x and you are supposed decide whether x is prime or composite. If you guess correctly you receive $2, if you guess incorrectly you instead have to pay a penalty of $1000. Additionally you have the choice of playing safe by giving up, in which case you receive $1. In traditional game theory, computation is considered costless ; in other words, players are allowed to perform
/lp/association-for-computing-machinery/algorithmic-rationality-adding-cost-of-computation-to-game-theory-CpyOz9kWgj