No abstract

We tackle the problem that arises when an agent receives unbelievable information. Information is unbelievable if it conflicts with the agent's convictions, that is, what the agent considers knowledge. We propose two solutions based on modifying the information so that it is no longer unbelievable. In one solution, the source and the receiver of the information cooperatively resolve the conflict. For this purpose we introduce a dialogue protocol in which the receiver explains what is wrong with the information by using logical interpolation, and the source produces a new assertion accordingly. If such cooperation is not possible, we propose an alternative solution in which the receiver revises the new piece of information by its own convictions to make it acceptable.