The normativity of Lewis Conventions

David Lewis famously proposed to model conventions as solutions to coordination games, where equilibrium selection is driven by precedence, or the history of play. A characteristic feature of Lewis Conventions is that they are intrinsically nonnormative. Some philosophers have argued that for this reason they miss a crucial aspect of our folk notion of convention. It is doubtful however that Lewis was merely analysing a folk concept. I illustrate how his theory can (and must) be assessed using empirical data, and argue that it does indeed miss some important aspects of real-world conventions. Acknowledgments: Research for this paper was made possible by the ESRC grant RES000-22-1591 and the Computable and Experimental Economics Laboratory of the University of Trento. Previous versions were presented at the universities of Amsterdam, Cambridge, Exeter, and Trento; the Institut d’Histoire et de Philosophie des Sciences in Paris, a seminar of the British Society for Philosophy of Science, the annual meetings of the Italian Society for Analytical Philosophy (SIFA) and the Italian Society for the History of Economic Thought (STOREP). I’m grateful to members of these audiences and in particular to Ivan Moscati, Mario Gilli, Philippe Mongin, Giacomo Sillari and an anonymous referee for their detailed feedback. The usual disclaimers apply.