The Poverty of Nonconsequentialism

Now, some people take the intuitive insanity of refusing to torture even to prevent the total annihilation of the universe as a devastating counterexample to non-consequentialist moral theories. I think this is a mistake based on a misunderstanding of the nature of morality and moral theory. Even the very best moral theory ever—one that organises and codifies our considered moral judgments better than any other—will sometimes generate the wrong advice. The usually unarticulated requirement that the very best moral theory ever have no notable counterexamples is arbitrarily over-demanding. A moral theory isn't a machine that takes in the specification of scenarios and spits out inerrant prescriptions. It is an intellectual refinement of our lived, evolving, socially-embodied morality, which is a body of largely tacit, often conflicting conventional rules and norms. The application of a moral theory requires the exercise of judgment at every step. Recognising the morally-relevant features of a scenario requires judgment. Identifying the rules and norms relevant to the circumstances requires judgment. Applying the relevant rules and norms appropriately requires judgment.

You may ask, "How do you know when a moral rule, such as 'don't torture', renders the wrong advice if the best moral theory always tells you that it is the right advice?" The answer is that you don't know. Sometimes exercising judgment amounts to little more than guessing and sometimes you'll guess wrong. Torture is categorically wrong, but it's not inconceivable that there are circumstances in which you should do it. However, there can be no general account of when you should do it, because generally you categorically shouldn't.

Will Wilkinson, "Torture Is Wrong"