We find Judge Posner's comment, like Judge Easterbrook's, agreeable enough. We think the study of interpretation by non-judges still has value, and don't understand him to disagree. Obviously judges know a lot more than our respondents did, as we acknowledged early. The fact remains that judges often are confronted with a basic cognitive task similar to the one that faced the takers of our surveys: they have to figure out what a piece of English text is best understood to mean. Sometimes their impressions from the wording and punctuation are heavily influenced by things they know that our respondents didn't know. Sometimes they aren't. In the gun use case, a lot of the arguing was all about what a normal usage of the statute's words would mean; there wasn't very much reliance on knowledge that our respondents lacked. In other cases the reliance on additional knowledge is greater, but the judges still argue about the simple English meaning of the sentence as one component of their analysis. That is the component we are studying. We suspect that the cognitive hazards we describe are common to judges just like anyone else.
As for the general differences between the positions of our respondents and the positions of judges, in some respects doing experiments is like creating economic models. Either way you have to simplify; either way you abstract certain features of the situation for rigorous study, and end up with a stylized result. The experiment (or model) isn't the same as reality, but it may capture something important about the reality that is harder to see when all the complexity is left in the picture.
In this case, the experiments suggest a cognitive issue in interpretation and a way to help overcome it. In real judging, the issue and the corrective may both be less important than they were in our experiments, because (as we have said, and as others have said) judges operate in different and richer circumstances than our respondents. But we think it is a mistake to too quickly assume judges are able to avoid the hazards that our experiments illustrate, whether because of their training or because they have contextual details to consider that our respondent lacked. After all, those other materials that judges consider contain ambiguities of their own. As I mentioned in reply to Professor Elhauge, perhaps—we might even say probably—those other things (ambiguous legislative history, ambiguous statutory context, etc.) may themselves be subject to interpretation, and perhaps those interpretations are subject to the same tendencies we have identified here. We hypothesize that policy preferences color the handling of ambiguous legal materials of all kinds, and that our studies just show a simple example of it. The argument to the contrary would be a lot stronger if judges, as a group, were famous for deciding hard statutory cases in ways that are hard to predict on the basis of their policy preferences. They aren't.