Export Reviews, Discussions, Author Feedback and Meta-Reviews

Too many technical details for simulations are missing-e.g., sure rewards, cost of sampling, discount factor. As noted on line 256, coherence is observable to the experimenter, but not to the subject. The virtues of normativity for modeling human behaviour are not adequately defended here. Why would normativity be good in and of itself? Does the model make any clear, testable predictions?