S. M. Amadae, University of Helsinki, FI

In this paper I argue that the rationality characterizing strategic action in game theory is computable. In making this argument I discuss parametric ordinal decision-theory developed by Kenneth J. Arrow, and the parametric expected utility rankings of John von Neumann and Oskar Morgenstern. I next discuss von Neumann and Morgenstern’s two-person zero-sum game theory. I argue that even though von Neumann and Morgenstern introduce rational decision-making predicated on a randomizing device, that this procedure is subject to computation. Moreover, whereas the Arrovian actors and von Neumann Morgenstern expected utility maximizing agents can be subject to indecisiveness due to indifference among elements in an optimal choice set, if we assume repeating choice contexts, then von Neumann and Morgenstern’s introduction of randomization in mixed-strategies can solve the problem of computing decisions in these cases. This argument is a fundamental part of a larger project that argues that the strategic rationality formalized by von Neumann and Morgenstern is computable in the sense of the Church thesis. If this is true, then insofar as strategic rationality (also called rational choice) is paradigmatic of instrumental rationality, then these agents are in principle no different from artificial intelligences with the same instructions for action (rules linking choices and outcomes) and identical preferences and beliefs.