A Journal of Philosophy, Applied to the Real World

Necessity and Liability: On an Honour-Based Justification for Defensive Harming

University of St Andrews and University of Stirling



This paper considers whether victims can justify what appears to be unnecessary defensive harming by reference to an honour-based justification. I argue that such an account faces serious problems: the honour-based justification cannot permit, first, defensive harming, and second, substantial unnecessary harming. Finally, I suggest that, if the purpose of the honour based justification is expressive, an argument must be given to demonstrate why harming threateners, as opposed to opting for a non-harmful alternative, is the most effective means of affirming one’s honour. Along the way, I also suggest why I think that internalism about the constraints on defensive harming (the view that the satisfaction of the necessity constraint is a necessity condition of a threatener’s liability) is correct. Most importantly, externalism implies that threateners can be liable to suffer gratuitous harm. I take this to be an unattractive consequence of the view.


1. Introduction

Let us begin with a case:

Standing One’s Ground.1 Threatener will culpably kill Victim unless Victim takes some preventive action. She has two options, both of which she knows will be effective. She can (1) retreat from the confrontation without risk or cost and alert the police, who will then be able to subdue Threatener without harming him, or (2) stand her ground and kill Threatener in self-defence.

Some believe that Victim is justified in standing her ground and killing Threatener. They believe that, because of the threat he poses, Threatener is liable to suffer defensive harm. To say that a individual is liable to be harmed is to say that harming him would not wrong him nor violate his rights, and so he would not be justified in defending himself.2 However, this verdict is at odds with a condition commonly applied to liability to defensive harm—necessity. Standardly, necessity requires that, for a defensive act to be justified it must be the least harmful of the agent’s available effective means of averting the threat. Such is not the case in Standing One’s Ground.

This paper considers whether victims can justify what appears to be unnecessary defensive harm by reference to an honour-based justification. The honour-based justification suggests that when a victim is threatened, there are two threats she faces: (1) the direct physical threat; (2) a secondary threat to her honour. It is defending herself against this secondary threat to her honour that renders certain defensive acts necessary (such as Victim’s standing her ground). The primary purpose of this paper is to argue that the honour-based justification faces serious problems. (Along the way, I also suggest why I think that internalism concerning necessity—the view that satisfying the necessity constraint is a necessary condition of a threatener’s liability—is correct, despite its initial intuitive implausibility.)

The paper proceeds as follows. Section 2 considers the relationship between necessity and liability. Section 3 advances what I take to be the most plausible version of the honour-based justification. Section 4 raises two worries with the honour-based justification: 4.1 questions whether the honour-based justification can justify defensive harming; 4.2 questions whether the honour-based justification can permit substantial unnecessary harming. Section 5 question what is meant by an honour-based justification in the first place. I argue that, if the purpose of the honour-based justification is expressive, an argument must be given to demonstrate why harming threateners, as opposed to choosing a non-harmful alternative, is the most effective means of affirming one’s honour.

2. Necessity and Liability

Necessity requires that, for a defensive act to be justified it must be the least harmful of the agent’s available effective means of averting the threat.3 There is debate as to the precise role that necessity plays within a theory of self-defence. Internalism holds that necessity is an internal condition of liability to defensive harm: threateners are liable to defensive harm only if that harm is the least harmful means of averting the threat they pose.4 This means that Threatener is not liable to be killed in Standing One’s Ground. (For brevity, “Victim/Threatener” refer directly to Standing One’s Ground, whereas “victim(s)/threatener(s)” characterise the generic.) If internalism is true, and Victim nonetheless stands her ground, this seems to imply that Threatener is, amongst other things, (1) wronged, (2) owed compensation, and (3) permitted to engage in counter-defence against Victim (by hypothesis, Victim is now posing a threat to which Threatener is not liable). It further permits, or even obligates, (4) third parties to defend or assist Threatener in counter-defence, as well as obligating them to refrain from interfering with Threatener’s counter-defensive action. (These implications hold, absent other arguments.5 I set these other arguments aside because our purpose is to discover the work that the honour-based justification can do in justifying what appears to be unnecessary defensive harming.)

Given what some take to be the implausibility of these implications, externalism denies that necessity is internal to liability, though it may bear on the all-things-considered permissibility of defensive action.6 On externalism, because satisfying the necessity constraint is not a necessary condition of a threatener’s liability, victims do not wrong threateners if they violate the putatively external necessity condition, but merely act impermissibly. Given that this still means that victims act wrongly in these cases, externalism remains somewhat intuitively implausible.

Despite its apparent intuitive implausibility, one reason for endorsing internalism about necessity is that liability justifications seem to be instrumental: for a threatener to be liable to defensive harm, that harm must serve some purpose (in Standing One’s Ground, preventing harm to Victim). This is what distinguishes liability-based justifications from desert-based justifications, which hold that imposing harm on a culpable individual is valuable as an end in itself. It is for this reason that Jeff McMahan is an internalist: ‘Threatener is clearly wronged by being killed in Standing One’s Ground because the unnecessary harm is entirely gratuitous and no one can be liable to suffer gratuitous harm’ (2016b, 196).

Let us consider what the externalist says about liability. I take the most plausible version of externalism to be Helen Frowe’s “proportionate means externalism”. Frowe thinks that, ‘to say that a person is liable to harm means only that a usual reason not to harm her—that she has a right not to be harmed—is absent’; accordingly, she thinks that ‘a person who is liable to defensive harm may use non-harmful means to prevent herself from suffering the harm, even though she may not harmfully defend herself’ (2014, 91, 106). So on proportionate means externalism, threateners are permitted to non-harmfully prevent themselves from suffering harm; yet, because their victims have not forfeited their rights not to be defensively harmed, threateners may not harmfully prevent themselves from suffering harm. However, it must follow that, were a threatener unable to non-harmfully defend herself in a case of unnecessary defensive harm and yet were the victim to nonetheless proceed with such unnecessary action, the threatener must suffer some gratuitous harm. (Namely, the amount of harm which was not necessary.) Even when the threatener can and may non-harmfully prevent himself from suffering unnecessary harm, he remains liable to suffer such gratuitous harm. This is a disturbing consequence of the view. Persons have the right to not be gratuitously harmed. There is no instrumental purpose to be served in the forfeiture of that right.

We might question how, on proportionate means externalism, liability is distinct from desert. After all, it seems that threateners are liable to suffer gratuitous harm. The proportionate means externalist might reply that, whilst necessity is not internal to liability, an effectiveness condition is internal to liability. For example, Frowe writes that proportionate means externalism ‘recognizes that liability to defensive harm means liability to harm that can avert a threat’ (2014, 105). This means that an instrumental purpose must still, potentially, be served in cases of unnecessary defensive harm.7

The purpose of this section has been twofold: first, to situate the debate that the necessity condition figures in; second, to demonstrate that, if we understand liability as an instrumental notion, there is good reason to be internalists about the necessity condition—to think otherwise would mean that agents are liable to some gratuitous harm.

3. An Honour-based Justification of Defensive Harming

Despite its theoretical plausibility, the internal necessity condition seems to have some counterintuitive implications: for example, that if a victim engages in unnecessary defensive harm, this wrongs the threatener and the threatener is thereby granted counter-defensive permissions. Even the external necessity condition implies that a victim acts wrongly if she engages in unnecessary defensive harming. In Standing One’s Ground, one might be willing to accept either of these implications. However, consider the following case:

Rape.8 Eric is in the midst of culpably raping Fran. There is nothing she can do to stop him from continuing to rape her. Fran is able to break Eric’s wrist, though this will have no effect but to cause him to suffer. The only way Eric can stop Fran breaking his wrist is to quickly break her wrist first. All of these facts are known by both Fran and Eric.

Both internalism and externalism imply that Fran acts wrongly if she breaks Eric’s wrist. This seems implausible. If internalism is correct, it seems to imply further that Fran wrongs Eric and that Eric is permitted to break Fran’s wrist to prevent her from unnecessarily breaking his. As it stands, the necessity condition seems implausible.

It is important to clarify four ways in which harm might be justified even if it will not avert the unjustified threat faced by a victim. A harm may be justified on the grounds that: (i) threatener deserves to be harmed; (ii) it has some deterrent effect by preventing that threatener or others from perpetrating future harm; (iii) it will lessen further psychological harm to the victim; or, (iv) it defends or affirms a victim’s honour. Because we are discussing the last of these justifications, we must be careful to bracket out the proceeding considerations.

(As point iii above indicates, we must not count any benefits that Fran may receive from breaking Eric’s wrist under the umbrella of honour. Suppose that if Fran does not break Eric’s wrist, she will suffer from Post Traumatic Stress Disorder (PTSD) for two years; if she does break his wrist, this will significantly lessen the length of time that she suffers from it. In these circumstances, breaking Eric’s wrist is not necessary with respect to averting the continuation of the rape, but might be necessary with respect to preventing or lessening psychological harm to Fran. Accordingly, the honour-based account is not required to justify her breaking Eric’s wrist (provided that breaking it would be proportionate in relation to mitigating Fran’s PTSD). We must suppose that Fran’s breaking Eric’s wrist will not prevent any psychological harm to Fran (or, that it will not prevent enough harm to render breaking Eric’s wrist proportionate).)

Whilst some defensive options might be unnecessary as means of averting the physical threat, one might argue that one goal of defensive harm is for the victim to affirm her equal moral status (in Daniel Statman’s and Frowe’s terms, her honour), which is challenged or violated by the threatener’s action. As Statman writes, ‘in the eyes of the aggressor, [the victims] are just items to be used, mere objects’ (2008, 689). It is therefore reasonable to suppose that the victim is permitted to assert or vindicate her equal moral status (her honour) by violent resistance, even if doing so will not avert the physical threat she cases. (For brevity, hereafter I stick to Statman’s and Frowe’s terminology of “honour”. See Frowe (2014, 109) for an indication that talk of “moral status” might be more accurate.) To put it another way, perhaps when a victim is threatened there are two threats she faces: (1) the direct physical threat; (2) the threat to her honour. It is defending herself against this secondary threat to her honour that renders certain defensive acts necessary (such as standing one’s ground or Fran’s breaking Eric’s wrist).

The honour-based justification is appealing because it allows the internalist and externalist to explain why victims may sometimes employ what would otherwise be unnecessary (or ineffective) harm. Frowe has recently made two refinements to the view: (i) by limiting the proportionality of the honour-based justification; (ii) by denying deferred honour-based harming. We will consider these in turn.

(i) Proportionality of Honour-Based Harming. One might think that the amount of harm which is justifiable in defence of one’s honour is that which would be proportionate in defence of the primary threat (Statman 2008, §V). In Standing One’s Ground, because killing Threatener is proportionate to the threat Victim faces (were Victim not able to avert the threat by retreating, it would also be necessary), Threatener is liable to the harm of death in defence of Victim’s honour.

This justification goes too far. Whilst the gravity of the threat to one’s honour will usually rise with the seriousness of the physical threat, this does not necessarily imply that the threat to one’s honour is as serious as the physical threat. As Frowe writes, ‘[t]he wrongness of being treated as the sort of thing that may be subjected to a harm is not the same (and is not as bad) as being subjected to that harm’ (2014, 112).

That said, I do not agree with Frowe that the threat to one’s honour is always ‘not as bad’ as being subjected to the physical harm. Consider the following case.

Bigot. Victim is of a persecuted minority race. Threatener is planning to flick Victim in the ear because he thinks that she is less worthy of equal moral consideration. The harm to Victim will be minor (including, let us suppose, the psychological harm).

This case shows that the primary physical threat may be less serious than the secondary harm to a victim’s honour. Further, it seems there will be cases in which there are threats to one’s honour that do not involve harm at all. We return to the connection between the threatened physical harm and the threat to one’s honour in 4.1.

(ii) Deferred Honour-Based Harming. One might also think that, ‘[u]nlike the defense of life, in which Victim must act before the threat materializes, acts to restore Victim’s honor can take place afterwards’ (Statman 2008, 673). Let us call this “deferred harming”. This implies, for example, that Fran’s actions would be permissible in the following case.

Deferred Harm.9 On Monday, Eric rapes Fran. Fran inflicts no harm upon Eric during the rape. On Tuesday, Fran sees Eric in a bar having a drink. Fran breaks Eric’s wrist.

Frowe thinks that the honour-based justification is most plausible when ‘conceived of as a defence against a threat to honour rather than an attempt to restore honour ex post’ (2014, 113). Whilst Fran’s breaking Eric’s wrist appears to be permissible in Rape, Frowe is less sure in Deferred Harm. In order to drive a wedge between the ex ante defence of one’s honour and the ex post restoration of one’s honour, Frowe suggests that we should conceive of the secondary threat to a victim’s honour as supervening upon the physical threat that the victim faces. I take the idea to be that the higher-order properties of the threat to a victim’s honour are determined by, and dependent upon, the lower-order physical properties. Accordingly, once the physical threat has ceased to exist, ‘nothing that the victim can do can avert that threat to her honour’ (Frowe 2014, 114).

In summary, victims face two threats: a primary physical threat and a secondary threat to their honour. (i) The secondary threat to a victim’s honour is determined by the gravity of, and yet is less serious than, the physical threat. (ii) The secondary threat to a victim’s honour supervenes upon the primary physical threat. Accordingly, victims can justify some defensive harm vis-à-vis defending their honour only as long as they face the primary physical threat.

4. Against the Honour-Based Justification

I think both of the alterations made above, whilst necessary for the honour-based justification, are unsuccessful. I argue against them in reverse order.

4.1 Supervenience

The problem with Frowe’s supervenience argument arises when considering the point at which the threat to a victim’s honour emerges. She faces a dilemma. In order to deny that deferred harming is permissible, it must be that Fran can break Eric’s wrist only to avert the ex ante threat to her honour and not to restore her honour ex post. However, on the one hand, if the threat to the victim’s honour emerges only when the threatener brings the physical threat to fruition, then the victim’s avoidance of the primary threat will itself dissolve the supervening threat to her honour.10 In Standing One’s Ground, Victim’s retreating itself nullifies the threat to her honour which would have occurred had Threatener been able to engage in harming Victim.

If, on the other hand, the threat to a victim’s honour emerges earlier than the onset of the primary physical threat, her honour is not being threatened but has already been violated, namely when the physical threat emerged. If this is correct, any “defence” of one’s honour is actually a deferred restoration of that honour. Consider Rape. There is not merely a threat to Fran’s honour unless she breaks Eric’s wrist. The physical threat is already being realised, which means that, if the claim about supervenience is correct, the violation of Fran’s honour has already occurred. By breaking Eric’s wrist, Fran is affirming her honour. (Due to space constraints, I cannot discuss when a victim’s honour is violated. It seems reasonable to suppose that it occurs when threateners begin to act on their intention to do what will wrong their victim. It could be argued that it begins with the formation of the intention itself.)

Frowe might suggest that, because the rape is still occurring, there is still a threat to Fran’s honour (or, that Fran is preventing the threatened continued violation of her honour). However, this entails that the further into the rape Fran can break Eric’s wrist, the less harm can be justified by reference to the honour-based justification. This is because, if the threat to a victim’s honour supervenes upon the physical threat, the remaining violation of Fran’s honour must decrease in severity as the rape gets nearer to ending. This seems absurd.

Perhaps it could be suggested that the harm of rape cannot be disaggregated as I suggest (i.e., one cannot say, at t2, there is less harm threatened than at t1). We might think that the threat to Fran’s honour supervenes upon the total harm of the rape. However, this does not seem quite right either. Other things equal, it appears that a longer rape must be more harmful than a shorter rape. What might not change is the wronging of the victim. But this is exactly the point: the threat to one’s honour depends more upon the wrongdoing than the threatened harm; and, in Rape, there is not a threat that Fran will be wronged—she already has been wronged.

The threat to one’s honour does not seem to depend purely upon the magnitude of the physical harm faced (as also suggested at 3.2.i in the case, Bigot)—harm and honour come apart in a way that the supervenience argument cannot account for. It is this half of the dilemma (that victim is affirming, not averting a threat to her honour) that supports the suggestion that a victim’s honour is being defended—we are just equivocating in our understanding of defence. Fran’s breaking of Eric’s wrist may be necessary (and effective) in respect to affirming her honour.

At this stage, whilst the honour-based justification might not permit defensive harming, it may still permit what appears to be unnecessary harming ex post.

4.2 Justifying Standing One’s Ground

My second concern is whether the honour-based justification can justify the infliction of what appears to be substantial unnecessary harm on threateners. As suggested in 3.2.i, with Frowe I think that the honour-based justification is most plausible when justifying less severe harms. Let me illustrate this with Standing One’s Ground. Victim can (1) retreat, or (2) engage in defensive action, killing Threatener. It seems implausible that Victim may kill Threatener merely in defence of her honour. It would seem, though, that Victim may (3) break Threatener’s leg in defence of her honour. That Victim would be justified in breaking Threatener’s leg has implications on her actions if she nonetheless engages in lethal defensive action. When victims engage in what appears to be unnecessary defensive action which is also disproportionate in relation to the threat to that victim’s honour (as (2) Victim’s standing her ground appears), we can subtract from the disproportionate defensive harm that which is justified by appeal to the honour-based justification.

If internalism concerning necessity is correct, and Victim nonetheless stands her ground by killing Threatener, this means that she wrongs Threatener to the extent of a harm that would be equivalent to the harm of death, minus the harm of a broken leg. This is because we have stipulated that Threatener is liable only to a broken leg as a means of defending Victim’s honour, meaning that he is not wronged by a harm of that degree. (For brevity, let us refer to the harm that would be equivalent to the harm of death minus the harm of a broken leg as “harm [death—broken leg]”.) Threatener may harm Victim in counter-defence to the extent that is proportionate to harm [death—broken leg]. This is because agents are permitted counter-defence against unjustified threats which, by hypothesis, Victim is now subjecting Threatener to.

Harm [death—broken leg] is an awfully large harm. One way of understanding harm [death—broken leg] is on the following preference-based view. Begin by considering how much life one would give up in order to avoid the harm of a broken leg. Plausibly, one might give up one month of their life in order to not suffer such a harm. If the badness of death is that future good which one would be denied through dying, harm [death—broken leg] might be thought of as the harm of being killed in one months time from now—this is equivalent to the harm of death, minus the harm of a broken leg.11

Victim wrongs Threatener to an extent that is equivalent to harm [death—broken leg] and Threatener is permitted counter-defence to the extent that is proportionate to harm [death—broken leg]. Even with the honour-based justification in place, internalism concerning necessity has massive intuitive implausibility when justifying what appears to be substantial unnecessary harm. (Again, I would like to stress that these implications hold, absent additional arguments that do not concern the honour-based justification.)

Before turning to how proportionality relates to externalism, let me illustrate the problem for internalism further. What can be drawn from what is said above is that there is a difference between that harm which Threatener unjustifiably threatens to inflict upon Victim (the harm of death), and that harm which Victim unjustifiably threatens to inflict upon Threatener (the unnecessary harm, [death—broken leg]). Whilst there is a difference between Threatener’s and Victim’s wrongdoing (Threatener commits a more serious wrong), the harm of a broken leg that differentiates their wrongdoing is not substantial enough to mitigate Victim’s wrongdoing if she engages in unnecessary harming. She still commits a grave wrong.

To put it another way, let us amend Standing One’s Ground.

Standing One’s Ground Bystander. The case unfolds as above, except that Bystander is watching. Bystander cannot stop Threatener from attacking Victim. Victim chooses option (2), and stands her ground and is about to kill Threatener. Bystander is now faced with the following three options: she can (A) stop Victim from engaging in unnecessary defensive action by killing Victim; (B) let Victim kill Threatener; (C) stop Victim from killing Threatener by imposing harm [death—broken leg] onto Victim, thereby also stopping Threatener from killing Victim.

The only difference with respect to unjustified harm between choosing options (A) and (B) is the justifiable harm of a broken leg—all other harm is, by hypothesis, unjustified. With the honour-based justification in place (and the stipulation that all Victim is justified in doing in defence of her honour is breaking Threatener’s leg), option (C) is what Bystander ought to do. This is because option (C) is the option whereby everyone receives only the harms to which they are liable. Even with the addition of an honour-based justification, the internalist still faces implausible implications.

Things do look a little better for externalism, for it does not imply that Victim wrongs Threatener or that Threatener is permitted counter-defence. However, if externalism concerning necessity is correct and Victim nonetheless stands her ground, this does imply that Victim acts, all-things-considered, impermissibly to an extent that would be equivalent to unjustifiably threatening to cause someone to suffer harm [death—broken leg]. If necessity is external to liability to defensive harm (about which I raised reservations above), then even with the addition of the honour-based justification, externalism still faces implausible implications.

The plausibility of the honour-based justification increases when the defensive action Victim takes is smaller (compare (3) Victim breaking Threatener’s leg with (4) Victim paralysing Threatener merely in defence of her honour). Accordingly, even with the honour-based justification, the difference in wrongdoing between Threatener and Victim is smaller than necessary to rule out, on internalism, substantial counter-defence on the part of Threatener and, on externalism, substantial impersonal wrongdoing.

5. What is Honour Anyway?

This final section asks what is meant by honour and how can it be appropriately and effectively “defended”. Statman writes, ‘[c]oncrete acts of resistance are needed in order to communicate to the aggressor, to ourselves, and to an actual or potential audience that we are not just passive objects to be trodden upon. By carrying out such acts, we reaffirm, or protect, our honor’; he continues, ‘[t]he idea that self-defensive acts aim at blocking the degrading message sent by Aggressor is reminiscent of expressive theories of punishment’ (2008, 669, fn 22). I think this reading of honour is the most plausible: there is expressive value realised when victims defend themselves, even when they know that this defence is not going to avert the physical threat they face; and, because this value is realised by certain defensive acts, when proportionate, threateners are liable to them.

If we take forward Statman’s remark that the honour-based justification is reminiscent of expressive theories of punishment, we run into a problem. Expressive theories of punishment say that the function of punishment is, amongst other things, to show society’s moral condemnation of the criminal’s behaviour. Joel Feinberg writes that punishment is a ‘conventional device for the expression of […] resentment and indignation, and of judgements of disproval and reprobation’ (1965, 400). Yet, as H. L. A. Hart notes, expressive theories of punishment treat ‘the infliction of suffering as a uniquely appropriate or ‘emphatic’ mode of expression’; however, the normal way in which we express moral condemnation is through words: ‘if denunciation is really what is required, why a solemn public statement of disapproval would not be the most ‘appropriate’ or ‘emphatic’ means of expressing this?’ (1963, 63, 66)

If the instrumental value to be achieved by the honour-based justification is merely the expression of the “defence” of one’s honour, why must threateners be harmed to achieve this goal? Consider the peaceful protest of Gandhi’s followers in the face of British soldiers’ brutality during the Salt March. In response, it might be suggested that harming threateners is the conventional method by which victims protect their honour. However, if this is the case then we ought to reject those conventions. To paraphrase Victor Tadros, if harming threateners is simply conventional, then making wrongdoers suffer in order to protect victim’s honour is barbaric: ‘We ought to find some other way of effectively communicating about wrongdoing’ (2011, 102).

Statman takes the line of suggesting that acts of violence are the conventional devices by which we protect our honour. Considering the example of a Gandhi-figure, he suggests that ‘[i]t is not the nonviolence itself that would protect a Gandhi from degradation but his prior overcoming of his concern about honor’ (2008, 680). This reply misses the point. We ought to not be concerned with harmful affirmations of honour. (At least, as concerns ideal theory. The most this argument might show is that, given the non-ideal circumstances we find ourselves in, honour-based harming is currently permitted. Even that, I am unsure of. Perhaps Statman could recast this argument and suggest that, with some additional account of demandingness, such Gandhi-like responses are supererogatory.)

The aim of this section has been modest: all I mean to do is shift the burden of proof onto those who endorse the honour-based justification of defensive harming. If defensive harming is a means of affirming a victim’s honour, an argument must be presented as to why harming threateners is the only, or the most effective way to achieve this. To put it another way, if there are less harmful ways of defending (or restoring) one’s honour, harming threateners is not going to satisfy necessity in any case.

6. Conclusion

This paper has considered whether one can pursue an honour-based justification to account for cases in which it seems that victims should be permitted to unnecessarily harm threateners. The primary purpose of this paper has been to argue that such an account fails. In 4.1 I suggested that if the threat to a victim’s honour emerges when the physical threat occurs, victim’s avoidance of the primary threat will itself prevent the supervening threat to her honour; if the threat to victim’s honour emerges earlier, it has already been violated—in which case, any defence of one’s honour is actually a deferred restoration of it. Accordingly, the honour-based justification can only permit harming ex post. In 4.2 I suggested that even with an honour-based justification in place, the difference between the victim’s and the threatener’s wrongdoing is not going to be substantial enough to mitigate victim’s wrongdoing if she engages in unnecessary defensive harm. In section 5 I suggested that if the purpose of the honour based justification is expressive, then an argument must be given to demonstrate why harming threateners is the only means by which to achieve this end.

The problem of reconciling the necessity condition and accounting for intuitively plausible cases in which it seems that victims should be permitted to unnecessarily harm threateners remains. This task is even more pressing if, as I have suggested at the close of section 2, internalism concerning the necessity condition (the view that meeting the necessity condition is partly constitutive of a threatener’s liability) is correct.

Acknowledgements: My thanks go to audiences in Monterey, Warsaw, Oslo, and Oxford, and for written comments from Jonathan Parry, Henry Phipps, Michael Prinzing, and Michael Robillard. Most of all, my thanks go to Jeff McMahan and Stephanie Miller for reading countless versions of this paper.


Fabre Cécile (2012a). ‘Internecine War Killings’, Utilitas, 24/02, pp. 214-236.

——— (2012b). Cosmopolitan War (Oxford: Oxford University Press).

Feinberg, Joel (1965). ‘The Expressive Function of Punishment’, The Monist, 49/3, pp. 397-423.

Frowe, Helen (2014). Defensive Killing (Oxford: Oxford University Press).

Firth, Joanna Mary and Quong, Jonathan (2012). ‘Necessity, Moral Liability, and Defensive Harm’, Law and Philosophy, 31, pp. 673-701.

Hart, H. L. A. (1963). Law, Liberty and Morality, (Oxford: Oxford University Press), pp. 60-69.

McMahan, Jeff (2005). ‘The Basis of Moral Liability to Defensive Killing’, Philosophical Issues 15/1, pp. 386-405.

——— (2009). Killing in War (Oxford: Oxford University Press).

——— (2016a). ‘Review of Helen Frowe, Defensive Killing’, Ethics 126.

——— (2016b). ‘The Limits of Self-Defense’, in Christian Coons and Michael Weber (eds.), The Ethics of Self-Defense (New York: Oxford University Press), pp. 185-211.

Tadros, Victor (2008). The Ends of Harm (Oxford: Oxford University Press).

Statman, Daniel (2008). ‘On the Success Condition for Legitimate Self-Defense’, Ethics, 118, pp. 659-86.

1. This case is taken, with slight alterations, from McMahan (2016b, 195). See also Frowe’s Lucky Escape (2014, 88).

2. Some might add that, for an individual to be liable to some harm, he must have forfeited rights that he previously held against that harm. See, e.g., McMahan (2005, 386); Frowe (2014, 3).

3. I am tempted to think that satisfying the necessity condition does not require satisfying the (evidence-relative) effectiveness condition, but follow McMahan in the main text for ease of exposition (2016b, 192). It appears to me that one can satisfy the necessity condition vis-à-vis some likelihood of effectiveness, even when that likelihood is below that which is deemed high enough to satisfy the effectiveness condition.

4. E.g., McMahan (2009), (2016a), (2016b); Fabre (2012a), (2012b).

5. E.g., McMahan (2016a), (2016b, 199-203).

6. E.g., Firth and Quong (2012); Frowe (2014).

7. For an account of externalism which does not have an internal effectiveness condition, see Firth and Quong (2012).

8. Similar cases appear in: Firth and Quong (2012, 689); Frowe (2014, 99).

9. Frowe (2014, 108) calls this case ‘Late Rape’, but the point of the case is that the harming is deferred, not the rape.

10. I owe a lot of this horn of the dilemma to discussions with Henry Phipps.

11. Thanks to Jeff McMahan for this illustration.