Lots of people have the intuition that you shouldn’t cause harm even to prevent greater harm. For example, most people think that you shouldn’t kill one person to prevent one extra killing. Similarly, most people think that you shouldn’t kill one person to harvest the person’s organs and save multiple lives. (I’ve argued against this before).
But we don’t have these intuitions about doing good. For example, most people think that if you’re choosing between you saving a life or two others saving two lives, you should choose for the others to save the lives. If you saving a life will cause two other deaths, it would be wrong to do so.
What’s the reason for this asymmetry? Kagan suggests in the beginning of normative ethics—not specifically in response to this problem—that when thinking about the good you do, we think about all the good that you do, which includes the good foregone by your action. But then why wouldn’t we do that on the negative end—why would, when thinking about a killing to save two, we not conclude that it’s good because, while you cause some harm, you prevent more harm?
If the explanation is grounded in the fact that we count extra benefits that you prevent against the good that you do when you do good, then why wouldn’t we count extra harm that you prevent when you cause some harm? Thus, this supposed explanation just pushes the puzzle up a level.
I can’t think of a very satisfying account of this. Thus, I think the non-consequentialist will just have to take this puzzling asymmetry as a brute fact. They’ll be unable to provide a deeper explanation for it. The fact that they have to posit lots of brute facts—lots of unexplainable distinctions—should count against it. The more puzzling brute facts one must posit, the worse for the theory.