By In Applied Ethics Comments (9)

A worry regarding the necessity of defensive force

I've got, well, a worry regarding the "necessity" requirement on the legitimacy of self- or other-defensive force. I don't really work on this stuff, so it's entirely possible that there's an easy, pat answer to this worry. Anyway, I'd be interested to hear what you all think.


It's traditionally thought that defensive force is justified only if that force is necessary to prevent harm to oneself or another. Or rather, what's traditionally thought is a specification of this. For we need to draw a distinction between two types of cases in which the use of force is unnecessary.

Type #1 (the type you were probably thinking of): Biff attacks Marty. Marty uses force to fend off Biff's attack. But Marty could have escaped by skateboard instead, and so Marty's use of force is unnecessary. 

Type #2 (the type you may not have been thinking of): Biff attacks George at the soda shop. George uses force to fend off Biff's attack. But George could have avoided Biff's attack by not going to the soda shop in the first place (and let's say George knew that Biff would be at the soda shop at that time, and would fly into a violent rage upon seeing George there). So George's use of force is unnecessary. Let us suppose, for the sake of contrast with Type 1 cases, that once Biff attacked, George could not have avoided harm except through self-defensive force.

According to what we might call "classical" Anglo-American criminal law, defensive force is impermissible in Type 1 cases. "Stand your ground" laws are, then, a deviation from classical A-A criminal law. But the law traditionally regards defensive force in Type 2 cases as permissible, assuming other requirements — proportionality, imminence, etc. — are met.

In the aftermath of the George Zimmerman trial, I wondered whether the law might should be amended to count Type 2 defensive force as unnecessary in the relevant sense. After all, Zimmerman didn't *have* to approach Trayvon Martin; he could have just driven on and maybe called the police if he was so inclined. I posted something to this effect on Facebook, and received plenty of "likes", which I take as evidence that others concurred. (I'm not on Facebook any more.)

On reflection, though, I rather doubt that my and others' intuitions about the Zimmerman case were driven by our adopting this more restrictive conception of necessity. Instead, I suspect that most people who suspected Zimmerman was in the wrong really thought so because they believed one or more of: (a) Zimmerman's killing of Martin was disproportionate, (b) The killing was not in self-defence, (c) Zimmerman did not reasonably believe that it was in self-defence, (d) Even if the killing was in self-defence, it was in response to *Martin's* use of force in *permissible* self-defence, and so was itself impermissible.  

At any rate, I think it is wrong to classify defensive force in Type 2 cases as impermissible-because-unnecessary. It seems to me that George's response to Biff above is appropriate. Or suppose that a female undergraduate has just successfully fought to force a campus club to admit women. Let us suppose that, if she attends an end-of-year party at which members of this club will be present, she will be assaulted. And let us suppose that her credence in this is justifiably high. She attends the party — perhaps partly out of determination not to give in to misogynist hostility, perhaps out of confidence that she can defend herself — and repels an assault with proportionate force. It does not seem to me that her use of force is impermissible-because-unnecessary. 

Anyway, those are my intuitions about Type 2 cases. So suppose we wanted to uphold those intuitions while hanging on to the traditional necessity requirement (i.e. the one that pertains to Type 1 cases). What differences between the two case-types might we cite? I'm at least initially attracted to a roughly-drawn view that we might call "the world is your oyster" (or "TWIYO" — trademark and apparel pending). On this view, you have the right to be anywhere you want, subject (maybe) to the property rights of others. Doesn't matter if it's a rave at an abandoned warehouse, a neo-Nazi rally, or the "cheap" seats at a Philadelphia Eagles game. The criminal law should never force you to choose between avoiding certain places and placing yourself at the mercy of attackers by going to those places. The law would be forcing this choice by counting defensive force in Type 2 cases as unnecessary, and so it oughtn't to do so. 

But recognize that if proscribing defensive force in Type 2 cases offends against your right to be where you want, so does proscribing such force in Type 1 cases. In our imagined case, for example, Marty would be forced to either flee or just sit there and take it as Biff attacked him. The law and Biff's aggression would conspire to make it impossible for Marty to safely stay where he was. 

Now, maybe you think there's an important distinction making it impossible for a person to *go* somewhere safely, and making it impossible for her to *stay* there safely. But why? What could justify placing a higher value on people's ability to venture out into new places than on their ability to stay in one place? (I find it hard not to think of the post-Bennett literature on acts and omissions here. Imagine a world of conveyor belts, etc., etc.)

I would think the answer would have to take the form of a claim to the effect that the goods to be attained by going to another place are higher or more valuable than the goods to be attained by staying where one is in the face of a threat. We liberal elites romanticize exploration, expanding one's horizons, experiencing new pleasures, broadening one's perspective, and so on. It's less clear to us what's so great about standing up to aggressor when flight is an option. We would look askance at Marty were he to claim that his social status would suffer by standing down from an asshole like Biff, or that his sense of himself, of the boundaries that separate this self from others, of his personal integrity would suffer by allowing Biff's aggression to push him around the world-space.

But I'm leery of this answer. First of all, there are some Type 2 cases in which goods one would forgo by avoiding the would-be aggressor are not so great. I mean, suppose Biff only hangs out at the soda shop from 4:45 to 4:50 PM; what would George lose by avoiding being there exactly then? Secondly, there are Type 1 cases in which the goods one would forgo by fleeing are great on any measure. Suppose a guy assaults me just as Springsteen brings my wife up on stage to sing "Thunder Road", dedicated to me, on our wedding anniversary. Finally, and a bit more seriously, I'm reluctant to dismiss claims of the sort I imagined Marty making in the previous paragraph. Some people's sense of self-respect, of integrity, etc. may indeed depend on standing their ground. And the reputational effects of running from an aggressor may be very real, even if not in my social circles. Of course, you may say that these goods are not worth, say, an aggressor's life/limb. Perhaps not. But then neither are the goods in many Type 2 cases (on any reasonable reading of this "worth" claim).

Now, obviously one response to all of this is to say that the criminal law must, for reasons of "followability", draw sharper lines than the moral law draws, and so we should just continue treating defensive force as "unnecessary" in Type 1 cases and as "necessary" in Type 2 cases notwithstanding the putative difficulties I've raised. But of course it is partly an empirical question just how much moral complexity it makes sense for the law to reflect.

9 Responses to A worry regarding the necessity of defensive force

  1. Victor Tadros says:

    Hi Andrew
    the problem that you raise is quite similar to the escalation problem that just war theorists have discussed.
    Suppose that it’s disproportionate to kill a person to stop him stealing my wallet.
    Now consider:
    1) X says to Y – hand over your wallet or I’ll kill you.
    2) Y refuses knowing that X will then attempt to kill him, and knowing that he will be able to prevent him doing this only by killing X.
    3) X attempts to kill Y
    4) Y kills X to prevent X killing him.
    If it’s disproportionate for Y to kill X to stop X taking Y’s wallet, I think that the conjunction of Y’s acts in 2) and 4) are wrong. Many just war theorists, though, disagree.
    Now suppose that in 4), Y had an option to run away rather than killing X, but if he did so he would drop his wallet, which X would then pick up and keep. He will suffer no other harm. I think that if it is wrong for Y to kill X to stop X taking his wallet, Y is also required to run away.
    So I think that when confounding factors are excluded, there is no moral difference between Type 1 cases and Type 2 cases that you mention.
    Now, we might also evaluate whether Y is permitted to defend himself given his refusal in 2). Suppose that Y refuses, knowing that X will threaten him. X threatens him. Is Y permitted to defend himself despite the fact that the conjunction of his acts is wrong?
    Similarly, consider this more extreme variation on your undergraduate case. Suppose that Y knows that if she goes to the party X will attempt a serious assault on her, serious enough that killing X would normally be a proportionate response. Y knows that she will be able to do this, and when threatened that is what she will do. One question is whether it is wrong for Y to go to the party and defend herself against X, killing him. Another question is whether if she goes to the party she is permitted to defend herself against X.
    I am not sure whether it is coherent to say this, but I am tempted to say that it is wrong for Y to go to the party and kill X, but that if she goes to the party and X attempts this serious assault, she is permitted to kill X. This is so, I think, even though if she kills X the conjunction of her acts of going to the party and killing are wrong, whereas if she does not kill X she will not have acted wrongly. I certainly think that if Y goes to the party, a third person, Z, ought not to prevent Y from killing X, say by taking away her gun, if otherwise Y will be seriously assaulted.

  2. Jussi Suikkanen says:

    Hi Andrew
    great post and I very much agree with your intuitions! I also think that it is brilliant question what the moral difference between Type 1 and Type 2 cases. Can I just run one attempt to draw the distinction which is different from going and staying?
    Could it be that there are two different moral principles for two different type of situations? One principle governs where you may go or stay when you are not under attack. This would be a liberal one, which entitles you to go anywhere really unless you violate other people’s property rights etc. (Victor’s last cases might show that in some situations even this principle is overriden by other considerations).
    Another principle might apply only when you are already under attack. This principle requires you to flee if this is an alternative to avoid an attack. So, when you are under attack some of your entitlements to go or stay are being limited. There is thus a way in which the attack transforms the normative landscape you are in: different principle applies to you as you are in different situation now.
    Of course this proposal still leaves it open why such different principles apply in the different circumstances: when you are under attack and when you aren’t (but might be in the future). Why does being under attack make such a difference to what you may do? Perhaps there are general justifications for this: it could be that having distinct rules like this have good consequences or they create less serious burdens for individuals in long term or something like this.

  3. Matt King says:

    Hi Andrew,
    This is an interesting question. My initial reaction is that the criminal law enforces a general principle restricting the use of violence, and operates on a pretty robust understanding of individual’s ability to refrain from using violence.
    So, everyone is supposed to refrain from violence. The effect of this principle varies depending on the circumstances. If you’re considering where to go and spend your time, then the principle is silent. Even if you can reasonably be sure that your presence will cause someone else to become enraged and assault you, the general principle against violence applies to them as well. So they’re obligated not to assault you, even though you are not obligated not to show up wherever they are.
    Once attacked, however, the general principle against use of violence means that you only use it as a last resort. So the principle is still in play even for the victim’s of assault, who should modulate their response appropriately.
    Thus, the difference between Type-1 and Type-2 is that you are not obligated by the principle against violence to ensure that others comport with that same principle. I’m not sure that’s a fully satisfactory answer; but it’s my initial thought.

  4. Larry Becker says:

    It’s a fine topic. Many thanks.
    I agree with previous comments that call attention to the whole landscape of duties, rights, and obligations involved in Type-1 and Type-2 cases, and the fact that this normative landscape involves both self-defenders and attackers. I also agree that circumstances are important. (Whether one is already under attack or not; whether one is at home or not; etc.) May I just add an additional worry about these cases? It isn’t exactly necessity that law and presumably morality are worried about. That’s too strict a standard. It is rather the presence and reasonability of the self-defender’s belief that his or her life is in danger and that there is no effective defensive alternative to the use of violence. And sometimes even that is also too strict a standard. Honest rather than reasonable belief sometimes suffices for people who have a legal duty to stand their ground – e.g., police officers. And then proportionality considerations apply.

  5. Hi Andrew,
    A quick suggestion:
    One difference between the two types of cases that I think might be driving the different intuitions is this: In the Type II case, the agent’s action of going to the place of aggression leaves it in the hands of the aggressor whether he will be harmed or not: even if the agent goes there, the aggressor has a perfectly legitimate way of avoiding the harm, namely by not attacking the agent, and the aggressor would be responsible for not choosing this way. In Type I cases by contrast, the aggressor would not be able to avoid the unnecessary harm inflicted on the aggressor by the agent. The aggressor is of course responsible for the aggression, but as the aggression has begun, it is now in the hands of the agent.
    One way to bring out the role of responsibility is to think of a Type II* case, which is like the Type II case, except the aggressor would not be responsible for his aggression should the agent show up. Perhaps, through no fault of his own been given a drug. It is the drug that causes the aggressive disposition, and it does so in a way that removes the aggressors responsibility for his reaction. If the agent knows this, the agent doesn’t seem to me to have the right to go to that place, knowing that he will have to harm the aggressor to avoid being harmed himself.

  6. Oops. I was a bit too quick in writing down that suggestion. The sentence
    “The aggressor is of course responsible for the aggression, but as the aggression has begun, it is now in the hands of the agent.”
    should have said
    “The aggressor is of course responsible for the aggression, but if the agent resorts to violent defence, the aggressor cannot then avoid being harmed by it.”

  7. Simon Rippon says:

    Great post. I am worried about the fitting attitude analysis of the dangerous as that which merits fear, because some things seem to merit fear which are not at all dangerous. For example, viewing the shower scene in ‘Psycho’, a magician doing a sawing the person in half type of trick, or going on a roller coaster ride. Whether things merit fear, to put it bluntly, seems to have more to do with how dangerous they seem to a fully informed and otherwise generally normal observer than with how dangerous they actually are. It would be no surprise, of course, to learn that ‘seemingly dangerous’ is a response dependent property. So you need to prove that your response-dependent analysis is of the underlying property of actual dangerousness, if you want to show there’s a surprising and significant kind of response dependence here.

  8. Simon Rippon says:

    Whoops I don’t know how that ended up here, sorry!

  9. Hi everyone — Sorry for the very late response; my mom was in town, etc., etc., etc….
    Victor — Yes, this “wallet” case is very helpful — especially in, as you say, removing “confounding factors”. For in your case, the envisaged loss to Y in failing to give up his wallet in response to the threat is the same as the envisaged loss to Y in running away — namely, his wallet. That wasn’t true in my original cases.
    Now, you also raise this matter of how to evaluate combinations of actions as opposed to individual actions. I appreciate this, as it helps to highlight some unclarity in the original post. I am uncomfortable ascribing deontic status to combinations of actions (well, really, to combinations of actions each of which is a distinct object of will; all actions are in some sense composed of other, “smaller” actions). Given the chance to express my thoughts more clearly, I would put my question as follows: “Should we count acts of self-defence in Type 2 cases as impermissible-because-unnecessary”, where by “count[ing] them as I-B-U” I mean “treating them, for the purposes of the criminal law, as we would treat paradigmatic cases of unnecessary defensive force — i.e. Type 1 cases”. It’s an interesting question how reasons/deontic statuses would have to be distributed over, say, *refusing to give up your wallet* and *using self-defensive force* in order for such treatment to be appropriate. If I had an answer to this question, I’d have something more satisfactory to say regarding Matt’s suggestion (see above).
    Jussi — I’d have to hear a bit more about why different principles apply in the different cases. I mean, one explanation — which is really more of an explaining-away — is that we have trouble accepting that someone who hasn’t yet been attacked could be as justified in believing she’ll be harmed by an attack as at least some people who have been attacked. (See my reply to Larry as well.) Another explanation — and again, I think this is more of an explaining-away — is that we think of those who stand their ground against an already-initiated attack as motivated by a desire to, well, stand their ground — to be tough guys or what-have-you; by contrast, we think of those who want to go to locations at which they’ll be attacked as motivated by other, better desires. But as I tried to point out at the end of my post, it’s not at all clear that a desire to stand one’s ground is the only reason for standing one’s ground. Maybe you just want to catch your wife singing “Thunder Road” with The Boss. And furthermore, I’m not so persuaded that a motivation to stand one’s ground is somehow worse than these other motivations.
    Other than that, I’m not so sure how the “moral landscape” can be changed by an attack in the way you suggest. If anything, we might think that your conduct in Type 2 cases is WORSE than your conduct in Type 1 cases. When you go to the scene in Type 2 cases, you’re unnecessarily setting in motion a causal chain that will culminate in violence against someone who has not yet attacked anyone, whose moral record is clean, as it were. But when you attack in Type 1 cases, we can at least say that the person you’re unnecessarily harming has crossed the moral Rubicon, or whatever.
    Matt — Let me see if I’ve got your idea right: Is it that what’s prohibited is unnecessary violence, which is violence at a time when one is capable of avoiding harm non-violently at the same time? So on this view, there’s no unnecessary violence in Type 2 cases, since unnecessarily going to the site of an attack is not violence, and violence in response to the attack is not unnecessary? This is an interesting idea. Now, one thing worth noting is that certain cases that both the law and ordinary morality will count as impermissible-because-unnecessary are not going to involve “unnecessary violence” by this standard. For suppose Biff attacks Doc at t1. Doc is able to escape from t1 through t5 but doesn’t, but nor does he fight back in that interval. And then, by the time t6 rolls around, is only way to escape harm from Biff is through self-defensive force. There’s no time at which Doc chose violence when he had an option of non-violent escape.
    Larry — Yes, you’re absolutely right that “mistaken” self-defence is treated as a justification in A-A criminal law. I generally tried to steer clear of this because I think it adds some complication that I’d preferred to avoid. Also, for reasons of theoretical neatness, I prefer to think of true defence as a justification and mistaken defence as an excuse. For one thing, counting mistaken defence as a justification seems inconsistent with the principle that at most one side in a conflict of force may be justified. When it comes to morality, though, yes, I think there ‘oughts’ relative to all the facts, to the evidence, to one’s actual beliefs, etc. etc.
    Gunnar — This is an interesting suggestion as well. I’m unclear about one thing, though: You’re right that *going to the place of aggression* in Type 2 cases gives the aggressor a choice re: being harmed in a way that *harming the aggressor* in Type 1 cases does not. But then again, *going to the place of aggression and harming the aggressor* or simply *harming the aggressor* in Type 2 cases do not seem to give the aggressor a choice re: being harmed. So I guess I’d have to hear a bit more about your suggestion: is it that THE AGENT gives the aggressor a choice in one case but not the other, or that the ACTION UPON WHICH CRIMINAL LIABILITY IS BASED gives the aggressor a choice in one case but not the other, or something else?
    I agree with your intuition about the innocent aggressor case, although I’m less confident that it supports your view about the difference between my two case-types. I mean, one thing that’s true of the aggressor in your case that’s not true of the aggressor in either of mine is that she’s non-culpable. That’s what seems to be driving my intuition that the agent in your case should avoid “setting the aggressor off”, as it were, and then harming her in self-defence.