Believing is not the same as testing: A reply to Beyth-Marom and Arkes

Several of Beyth-Marom and Arkes' criticisms of our original article arise from their misunderstanding our conclusions. This reply clarifies those misunderstandings. We reiterate our conclusion that even though people may not use normative rules in laboratory tasks, they still can make inferences that are nearly as optimal as those they would have made if they had been using those rules. This reply also elaborates upon our concern about the use of quantitative word problems to test human inference. We submit that additional evidence is needed to justify continued reliance on these problems.