Strategic reasoning in persuasion games: An experiment

We study experimentally persuasion games in which a sender with private information provides verifiable but potentially vague information (i.e., about the quality of a product) to a receiver. Sequential equilibrium and iterated admissibility predict unraveling of information. Iterated admissibility also provides predictions for every finite level of reasoning about rationality. We observe behavior consistent with relatively high levels of reasoning. Iterated admissibility implies that the level of reasoning required for unraveling is increasing in the number of quality levels. Yet, there is only insignificantly more unraveling in a game with two quality levels than in a game with four quality levels. There is weak evidence for learning. Participants display difficulties in transferring learning from a game with two quality levels to a game with four quality levels. There is a significant but small positive correlation between cognitive abilities in Raven's progressive matrices test and levels of reasoning.