It’s often been alleged by various critics of utilitarianism that utilitarianism implies implausible things about self-defense. Utilitarianism will, of course, in most cases imply nothing out of the ordinary. Having robust norms around self-defense has good outcomes—if people know that carrying out acts of violence will be met with reciprocal violence they’re less likely to engage in violence. So utilitarianism makes sense of these norms. However, like with most norms that utilitarianism makes sense of, there are rare edge cases where utilitarianism diverges from common sense. Consider the following example:
Maimer: someone comes at me trying to harm me significantly. The only way I can stop him is by harming him more significantly. He’s gone temporarily insane, so if I don’t harm him, he won’t harm anyone else in the future. He’s also a Yankees fan.
Intuitively, most people think that harming him would be permissible. Even if harming people in self-defense makes the world worse, it’s still permissible. Utilitarianism makes sense of there being social norms that permit this, for they maintain effective deterrence. But it would hold that if one is in the Maimer scenario, and not acting in self-defense would have no other spillover effects, what they really ought to do is not act in self-defense. Many find this to be quite counterintuitive.
Now, as I’ve argued before, there is a sound methodological criticism of such objections. We have ample and indeed quite overwhelming reason to think that our intuitions about cases are quite unreliable, and thoroughly methodologically inferior to intuitions about broad principles. I won’t rehash the claims in that article, but if one buys any of the many arguments made there, they should give this objection virtually no weight.
Still though, it’s worth considering whether, upon thorough reflection, the claims made by utilitarianism about self-defense are, in fact, unintuitive or whether, like pretty much every alleged counterintuitive implication of utilitarianism, upon reflection they become hard to deny. I haven’t thought about the self-defense cases for very long, only around 20 minutes, and I’m sure more thorough reflection would turn up more arguments, but even based on this minimal reflection, it’s clear that rejecting the utilitarian view of self-defense has steep costs, ones much more severe than those of accepting it.
Before I provide reasons to accept the utilitarian result, let me provide three reasons why the intuition about cases like Maimer is not especially good evidence against utilitarianism.
First, this implication is not that counterintuitive. When we think about cases like Maimer, it’s quite plausible that there’s something noble about failing to act in self-defense. Indeed, it’s not hard to believe that failing to act in Maimer would be ideal. Thus, our intuition about the case is that failing to act would be ideal but a reasonable person might act in self-defense. But a utilitarian can agree with both of those! Not acting in Maimer requires an extraordinary level of benevolence that is not expected of most people. In fact, if one adopts the type of scalar utilitarianism advocated by Richard here, as well as various other utilitarians, one can even accommodate the intuition that acting in self-defense in Maimer is permissible but not ideal.
The primary disagreement between utilitarianism and the common-sense verdict about self-defense is metaphysical in nature. Common sense seems to affirm that there are some actions that are supererogatory—above the call of duty. Utilitarians will generally deny this, instead thinking that there are just different strengths of reasons people have to take various acts, rather than some invisible line that demarcates the permissible from the impermissible. But if this is true then the self-defense objection is really the supererogation objection repackaged. If so, then the objection fails, because we have ample reason to reject the idea that some actions are supererogatory1.
Second, our intuitions about cases like this are very likely a byproduct of our culture. From a very young age people are told—for good reason—that self-defense is permitted. Thus, it’s no more surprising that people have the intuition that self-defense is permitted even in cases where it has bad outcomes than it is that people think other things that have been drilled in them from the time they could walk.
So I think that our intuitions about supererogation don’t actually conflict that much with common sense and even if they did it wouldn’t be a big problem. But we have a smorgasbord of good reasons to give up on the non-utilitarian judgment about cases like Maimer.
First, this verdict is especially implausible when paired with general deontological verdicts. The deontologist will hold:
It’s wrong to kill someone to harvest their organs and save five, if all else is equal and there are no other negative side effects.
However, given that the organ harvesting case is a counterexample to consequentialism, they should also hold:
Killing someone and harvesting the organs to save five improves the world, if all else is equal.
They should next accept:
A perfect being would not kill someone in a way that enabled them to harvest their organs and save five people in ordinary cases, if all else is equal.
Then, if they accept the verdict around self-defense, they should accept:
A perfect being would kill someone in a way that enabled them to harvest their organ and save five people if this was in self-defense against being killed.
But these imply the absurd:
If a person has organs that could be harvested to save five people, it would be fortunate for them to aggress against a perfect being so that they’re killed in self-defense and have their organs harvested, all else equal.
Generally the people who have the non-utilitarian intuition about self-defense have the non-utilitarian intuition about organ harvesting. But together these intuitions produce absurd results. So one has to choose…
A second worry for the non-utilitarian intuition about self-defense cases is that it violates a modest principle according to which people acting permissibly rather than impermissibly won’t be infinitely bad for everyone. To see how it violates this, suppose that the world consists of infinite long-lived people who are mostly peaceful but occasionally aggress. Suppose that when they aggress, another can stop their aggression by imposing a greater harm on the aggressor than they would have endured. If they all act in self-defense, rather than taking some other impermissible act that merely harms them for no benefit, that would be infinitely bad for everyone.
You could avoid this by saying that self-defense is impermissible if abstaining from it is part of a sequence of actions that could be taken and would result be better for some and worse for no one. But this is an arbitrary ad hoc stipulation, raises troubles about how to specify a sequence of acts, and conflicts with our intuitions about self-defense cases—intuitively it seems permissible to use self-defense against those who you previously aggressed against and who didn’t exert their right to self-defense.
Third, and perhaps most decisively, the common-sense approach to human shield cases involving self-defense is implausible. Most people intuitively hold:
Whether one can be harmed in the pursuit of stopping an atrocity depends on whether they’re being used as a human shield.
For example, it would be wrong to kill a random innocent person to stop a few deaths. However, if there is a killer plotting to kill multiple people who is using someone as a human shield, it’s permissible to kill the human shield to stop the killing. But this has the implausible result that whether someone is being used as a human shield is of intrinsic moral significance and affects what reasons one has. This, in fact, violates the Pareto principle.
For example, suppose you aren’t sure if someone is being used as a human shield—there’s a 50% chance that they’re being used as a human shield. Some killer is rampaging and will cause enough damage that it would be worth killing the human shield to stop them, but not worth killing an unrelated innocent to stop them. You have the following options:
Kill the person if and only if they’re being used as a human shield in a way that would be pretty painful.
Kill the person if and only if they’re not being used as a human shield in a way that would be painless.
The common-sense view of human shields would hold that 1 would be permissible but 2 wouldn’t be, if one was choosing between them and inaction. But if 2 is worse than inaction but 1 is better than inaction, then 1 must be better than 2. But this is implausible! 1 is worse for the victim in expectation and equally good for everyone else. Morality shouldn’t instruct us to make people worse off rather than better off for the benefit of no one.
You might object that in a binary choice 2 is worse than inaction and 1 is better than inaction but in a three-way choice between 2, 1, and inaction, 2 is better than 1. But this violates the irrelevance of unchosen alternatives—if choosing between any two actions, the existence of other actions that you’re not going to take shouldn’t affect which one you choose.
Fourth, self-defense norms are hard to explain. There doesn’t seem to be a deeper account of them—they just seem to be brute. This makes positing them as part of the moral fabric less plausible.
Thus, we have ample reason to abandon the common sense verdict about self-defense to the extent that it conflicts with utilitarianism. I don’t think it conflicts very sharply, but if it does, then utilitarianism is correct.
I have a paper in the works that provides more arguments against supererogation.
I think this supercedes the collaborative chat I was attempting on you-know-what. Although I'd still like to do something on moral reasoning in general, at some point.
There is an un-noted shift from "It’s wrong to kill someone to harvest their organs" to "A perfect being would not kill someone in a way that enabled them to harvest their organs" in that the latter phrasing drops any suggestion of intentionality. Organ harvesting could be an impermissible reason for killing without being an impermissible outcome.
Imagine that in a city with budgetary issues, a very wealthy person is caught breaking the law and incurs a massive fine. It seems perfectly plausible to think that: the city shouldn't just seize one person's money to fund its budget, but the city should impose fines for wrongdoing, and the city can regard it as a fortunate byproduct of the latter that their budget issues are solved. This is like the organ self defense case.
If you're not already a consequentialist, I don't think there will be much tension between thinking that a certain outcome is fortunate (acquiring money/organs) and that whether you can bring it about depends on other factors.