On the consistency of choice