Normative reasons and full rationality: reply to Swanton

Abstract

In The Moral Problem I suggest that an agent, A, has a normative reason to 4 in certain circumstances C, just in case, in nearby possible worlds in which A is fully rational, A desires that, in those possible worlds in which she finds herself in circumstances C, she Os. If we call the possible world in which A is fully rational the 'evaluating' world, and the possible world in which A is in circumstances C the 'evaluated' world, then my proposal is that facts about A's normative reasons in the evaluated world are constituted by facts about the desires she has in the evaluating world about what she is to do in the evaluated world (Smith 1994: 156-61; 1995a). The idea of full rationality employed in the analysis clearly needs to be spelled out. My suggestion is that to be fully rational an agent must not be suffering from the effects of any physical or emotional disturbance, she must have no false beliefs, she must have all relevant true beliefs, and she must have a systematically justifiable set of desires: that is, a set of desires that is maximally coherent and unified (1994: 158-61). In figuring out what an agent is like in the evaluating world we must therefore abstract away from various aspects of her actual psychology: the effects of any physical and emotional disturbances, her false beliefs, incoherence in her psychology, and so on and so forth. Furthermore, I argue that it is part of what we mean when we say that a set of desires is systematically justifiable that the desires that are elements