How Is “Conceptual Engineering” Rational? Solving Some Puzzles by Connecting Rationality and Attention

Some concepts generate paradoxes by licensing inconsistent beliefs. We can try to revise some of those beliefs by doing some “conceptual engineering,” but this leads to a puzzle. However exactly the details of conceptual engineering get filled in, it seems that what happens is we revise our beliefs without gaining any evidence against our old views or in support of our new views. But epistemically rational belief change always requires new evidence—or so it seems. I respond to the puzzle by proposing a view of epistemic rationality on which we are required to hold a belief only if two conditions hold: Our evidence supports the belief, and our attention is directed at the evidence and the supported belief's content. Thus, we can rationally withdraw a belief by withdrawing our attention to its content, even without having any new evidence. Its ability to solve this puzzle for conceptual engineering, in addition to some other important puzzles that I briefly show it also solves, gives us a strong case for this proposed connection between rationality and attention.