The Strategy Behind Belief Revision: A Matter of Judging Probability or the Use of Mental Models?

Research in the field of human reasoning has shown repeatedly that people find it reasonably easy to detect inconsistencies. The question that still remains is how people revise their beliefs to undo these inconsistencies. We report two experiments in which subjects had to make belief revision choices on modus ponens (MP) and modus tollens (MT) sets of problems that contained conditionals with different levels of probability. After the final statement of each set, which was stated to be true, they had to decide which of the first two statements they believed more. The results showed that with high and low probability problems, subjects revised their beliefs as a function of the probability of the conditional. However, when the conditional had a near 50% chance of occurring, the pattern of belief revision more resembled the mismatch principle as suggested by the mental model theory. Thus, people use different strategies when revising their beliefs depending on whether they are guided by semantics or mental