Genetic drift resolves Selten's Chain Store Paradox

Abstract

Recent behavioral evidence suggests that mutation-susceptible, best-practice emulation is a common strategy updating mechanism among real world human actors. Unlike purely analytical models of non-cooperative strategic behavior, computational models employing mutation-susceptible emulation-based strategy updating mechanisms (e.g. elitist Genetic Algorithms) are susceptible to a process similar to genetic drift. This drift is known to disrupt the stability of an equilibria. This paper uses a computational, Genetic Algorithm based model to demonstrate that such equilibrium-disrupting drift resolves Selten's Chain Store Paradox. More broadly, this paper hopes to modestly demonstrate how results from behavioral game theory can fruitfully be used to select the mechanisms used in computational game theoretic models.