It is standard to talk about different sources of normativity. Epistemic rationality is aimed at truth, or knowledge, or understanding. The norms of epistemic rationality concern gaining this goal in the best way. Individual practical rationality is aimed at achieving the individuals goals, whatever they may, in the best way. Social practical rationality is aimed at achieving the groups goals, whatever they may be, in the best way. Moral rationality is aimed at doing what is right. All of these norms come in at least two flavors: structural/wide-scope norms tell us the abstract ‘shape’ our reasoning should take, like deductive consistency or expected-utility maximization, and substantive/narrow-scope norms, like expert-deference principles or codes of conduct, that tell us what the true, good, or valuable things or actions in fact are. 

Sometimes, these norms appear to conflict. I defend the view that these conflicts are not subject to rational reconciliation. One example. My in-progress paper Countable Sure-Thing vs. the Axiological Principal Principle explores a conflict between a structural norm of individual practical rationality and a substantive norm of individual practical rationality. Isaacs and Russell’s countable sure-thing principle is designed to shield agents from certain kinds of dutch books that arise in infinitely situations like the St. Petersburg Game, and is an extension of the finite sure-thing principle standard to expected utility theory. The axiological principal principle, inspired by David Lewis’s chance-credence-connecting principal principle, instructs agents to desire or value things in accordance with their objective goodness. These principles come into conflict in cases where there exist an infinitely-ascending series of goods, each of which improves on its predecessors without limit. For example, suppose an agent comes to possess a bottle of EverBetter Wine, a bottle of wine whose flavor and quality improves every day. The axiological principal principle tells her to prefer drinking the wine tomorrow to drinking it today, all else equal. The countable sure-thing principle tells her that this preference is unreasonable. 

But how can these kinds of disputes be resolved? We may choose one principle or the other, but in order to do so in a way that is not ad-hoc, we must introduce a meta-principle for choosing our principles. I argue that any meta-principle we try to introduce to settle the dispute will not succeed. It, too, will have some constitutive goal, one with its own constitutive goal. Inevitably, this meta-principle will face competitors, resulting in a higher order normative disharmony. In order to resolve this disharmony in a principled way, some higher higher order principle will be required, which will face competitors, and so on indefinitely upward. There is no neutral ground. We must simply choose our ends without normative guidance; reason only steps in once the ends are settled.