Sorry I wasn't clearer first time around. Yes, rationality is a distinct topic; but it has some overlap with AI, inasmuch as learning how to think better is served by understanding more of how we can think at all. The discussions around decision theory clearly belong to that overlapping area; Eliezer makes no bones about needing a decision theory for FAI research. Analogy in the Hofstadterian sense seems underappreciated here by comparison. To my way of thinking it belongs in the overlap too, as Kaj's post seems to me to hint strongly.