As has been noted before (e.g., Hoffman et al. 1996, Ortmann et al. forthcoming), experimental participants' typical everyday experiences are those of indefinitely repeated games (e.g. Binmore 1994, 1997; Smith 1759.) For many of these games, the game-theoretic predictions for one-off or finitely repeated games on the one hand and indefinitely repeated games on the other, sharply differ (e.g., social dilemma games, ultimatum and dictator games, trust games, principal-agent games, etc.). To the extent that real-world and laboratory environments are congruent, experimental participants thus have to manage to suspend (i.e. unlearn instantaneously) their everyday experiences, and then to understand the very peculiar circumstances of the laboratory environment. We first present evidence from psychology on the so-called base-rate fallacy (Bar-Hillel 1980): experimental participants, when given bits of information about a population or a set of events or facts and "individuating" information about a member of that same population or set, tend to under-weigh the population information and to over-weigh the individuating information. However, there is also strong evidence that the opposite happens in many other scenarios. In fact, a consensus seems be emerging that everyday experiences typically induce potent and resilient base rates that are not easily unsettled by individuating information and that the base rate fallacy may well-be a myth for most of those situations that people typically encounter in real life (e.g. Koehler 1996; Cosmides and Tooby 1996).

We next note that many of the so-called anomalies in economics are connected to experimental designs that try to implement one-off or finitely repeated game or decision scenarios (e.g., results from social dilemma games, ultimatum and dictator games, trust games, principal-agent games, etc.), i.e. situations where experimental participants have to overcome the very potent and resilient base rate experiences that serve them well in everyday life. We then connect these findings to those from psychology. We conclude that there is a good chance that many of the experimental results for the classes of games discussed here are artifacts of experimental design very similar to the base-rate fallacy. Unfortunately, while these anomalies are easy to construct, they are very difficult to deconstruct because such deconstruction involves, at least for the games considered here, the construction of a counterfactual reality.

Co-author RALPH HERTWIG