Economics in the lab: Completeness vs. testability

Two important arguments in the methodological literature on experimental economics rely on the specification of a domain for economic theory. The first one is used by some experimenters in their skirmishes with economic theorists, and moves from the assumption that theories have ( or ought to have) their domain of application written in their assumptions. The other one is used to play down the relevance of certain unwelcome experimental results, and moves from the symmetric assumption that the domain of economic theory is more limited than a literal reading of its assumptions would suggest. Of course, only one of them can be right. In this paper I criticise the former, and outline some well-known arguments that strongly point in the direction of the incompleteness of economic theory. Some remarks on the role of methodological arguments conclude the paper.