Effective Altruism and Decision-Making for the Clueless

Various philosophers have recently claimed that normative uncertainty in general, or decision-theoretic, moral, or prudential uncertainty in particular, should be taken into consideration in decision-making, a position sometimes labelled Metanormativism. In this thesis, I argue against a restricted form of Metanormativism, pertaining only to uncertainty about imprecise decision principles, i.e., accounts of rational choice for decisions modelled using imprecise credences (‘decisions under cluelessness’). To factor uncertainty about imprecise decision principles into decision-making, a Metanormativist account of rational choice is required, which tells decision-makers how to act on the basis of (among other things) uncertainty about imprecise decision principles. Such a Metanormativist account of rational choice can take one of two general forms. First, it might take the form of a single second-order decision principle, i.e., an account that specifies which alternatives are permissible on the basis of our confidence in imprecise decision principles and the permissibility of alternatives according to these principles. Second, the Metanormativist account could be comprised of an infinite hierarchy of second- and higher order decision principles: In case decision-makers are uncertain about second-order decision principles, decision-makers should resort to a third-order decision principle; uncertainty about third-order decision principles should be handled by means of a fourth-order decision principle; and so on. I argue that there isn’t any single second-order decision principle that decision-makers should conform to, nor any plausible account of choice comprised of an infinite hierarchy of second- and higher order decision principles. Hence there is no plausible Metanormativist account of rational choice, from which it follows that Metanormativism about imprecise decision principles is false. To arrive at this conclusion, I’ll have to lay considerable groundwork, since the possibility of imprecise credences has so far only been acknowledged in passing in the literature on Metanormativism. Notably, I don’t hold the discussion about Metanormativism in the abstract, but embed it in the context of effective altruism. Considering that effective altruists often face decisions best modelled using imprecise credences, the conclusion that Metanormativism about imprecise decision principles is false has important implications for this social movement.