Abstract: We present a game-theoretic model of debate and a laboratory experiment that explore how strategic incentives to make potentially persuasive arguments vary across different informational and institutional contexts. In our model, a key feature of the informational environment is the extent to which members of a debate audience are able to extract informational content from exposure to an argument that they nd unconvincing. Our theoretical results show that when the informational content of unconvincing arguments is relatively high, speakers are discouraged from arguing irrespective of the distinct institutional rules of debate that we consider. By contrast, when the informational content of unconvincing arguments is relatively low, debate rules matter: speakers may be lead towards maximally or minimally informative debate, depending on the debate rule. In a laboratory experiment, we vary the informational and institutional settings for debate across four distinct treatments, and observe patterns of behavior which are broadly consistent with the predictions of our model.

Download the paper