What should an ordinary household thermometer show at temperatures close to absolute zero? There is no answer to this question. The question asks about what should happen in circumstances too far beyond the normal operating conditions of the instrument.
I wonder if something analogous doesn't happen in ethics. We have our normal operating conditions. These are very broad, because we are very adaptable beings, but nonetheless they are limited. Are there always going to be answer to questions about what we should do beyond these conditions?
One way of going beyond the conditions is to consider metaphysically impossible situations. If you promised to bring three oranges to the party and you are in the impossible world where four is less than three, do you fulfill your promise by bring four? Would you be obligated to honor yourself as your parent if you were your own cause? These questions seem to make little sense, and even we philosophers rarely think about them. But analytic ethicists do, on the other hand, sometimes ask questions about nomically impossible situations, and certainly we ask questions about situations far beyond ordinary life.
I think we should, however, take seriously the possibility that as we depart far enough from the normal operating conditions of human beings, some of the questions (a) have no answer or at least (b) have no answer available to us. This possibility undercuts some arguments.
For instance, one can argue that utilitarianism gives deeply implausible answers (e.g., that every action is equally permissible) in cases where there are infinitely many people. But suppose that there aren't in fact infinitely many people, and the situation of there being infinitely many people is far beyond humans' normal operating conditions. Then the fact that utilitarianism predicts something that seems implausible to us beyond those conditions is not a problem for the utilitarian--as long as she is willing to modestly limit the scope of ethics to humans in or near their normal operating conditions (if she's not, the argument is fair game).
Or consider this argument against deontology: It seems permissible to kill one innocent person to save a billion. But circumstances where we choose between one life and a billion lives might well be so far beyond our normal operating conditions that they fall beyond the scope of ethics.
The last case is interesting. For it raises this question: Might we not actually find ourselves in circumstances so far beyond our normal operating conditions that ethics doesn't apply, much as someone could actually throw a household thermometer into liquid helium? After all, it is sadly all too easy to imagine how someone might end up choosing whether to kill one innocent to save a billion: it seems physically possible for someone to end up in that position. It seems deeply troubling to suppose that some people end up in circumstances that go beyond the presuppositions in the moral law.
I think Christians have reason based on revelation to think this doesn't actually ever happen. The moral law is also embodied in revelation, and revelation presents itself as a guide to us in all the vicissitudes of life. But note that even if nobody ends up in circumstances that go beyond the presuppositions of the moral law, going beyond these presuppositions could be physically possible but for God's providential protection. A case of choosing whether to kill one innocent to save a billion may be like that: God makes sure we're not tried beyond the edge of ethics.
But what could one say without revelation?
Of course, the above line of thought fits best either with (a) divine command metaethics or (b) natural law metaethics on which what grounds ethical truths is our nature and an Aristotelian metaphysics of human beings on which it makes sense to ask what our normal operating conditions are. All this won't be an issue given a utilitarian or perhaps even Kantian metaethics. So that limits the applicability of the line of thought. But if we do find plausible the Aristotelian metaphysics and a natural law metaethics, then I think we should take seriously the worry that sometimes an analytic philosopher's ethics examples will go too far beyond our normal operating conditions. An argument for the above line of thought, and hence indirectly for either (a) or (b), is given by the apparently insuperable difficulties in ethics when one supposes that one's actions affect an infinite number of people.