Abstract: This paper introduces the first method for uniquely extending any given iterated belief revision operator so as to handle so-called conditional beliefs as inputs. This allows us to model agents capable of acquiring new beliefs not only regarding matters of fact, but also regarding which rules of inference they ought to use in reasoning about these facts. The suggestion is discussed against the backdrop of past research on the topic, including the work of Kern-Isberner, Boutillier & Goldzsmidt, Hansson and Nayak et al.

NB: Tea starts at 15:00!