*NB: Thread Extended until Monday, 20 March, 17:00 PDT*
Welcome to our latest P&PA discussion forum, on David Clark’s “Refusing Protection”. You can find the paper open access here.
We kick off the discussion with a critical précis by Kim Ferzan:
For years, I’ve been at conferences where someone will occasionally mention the ability a victim has to refuse another’s defense. It is the kind of question that gets a passing glance, but David Clark’s fabulous new piece, Refusing Protection, reveals precisely how much we’ve missed in these cursory discussions. Refusal is a layered and deep philosophical puzzle that can teach us quite a bit about consent, self-defense, lesser-evils, and normative powers generally. Clark does a masterful job of explicating all the moving pieces—his piece is both admirably clear and thoroughly rigorous—and I hope this brief précis can do his work justice.
Here’s the kind of case at issue:
Siblings: Adam, Teresa, and Dan are adult siblings. Jealous about being left out of their mother’s will, Adam attempts to break Teresa’s leg. The attack can only be stopped … if either Teresa or Dan break Adam’s leg first. Dan prepares to harm his brother, but Teresa—who is a committed pacifist and who would rather be harmed than see Dan harm their brother in her name—demands that he stand down. (1)
Working off the intuition that Dan’s intervention is now impermissible, Clark’s objective is to understand precisely why this is so. The real-world payoff is the permissibility of humanitarian intervention when there is disagreement within a particular populace as to whether they want to be rescued; specifically, the threshold at which refusal renders intervention impermissible.
Clark’s central claim is that “the power to forbid protection is an upshot of a more basic normative power—the power to mitigate specific normative effects of our rights.” (2) He argues, “we have the power to selectively restore some of the rights that one’s attacker forfeits by threatening our rights. One way in which this power is exercised is by the refusal of protection.” (2) His foil is Jonathan Parry, who has introduced a different explanation.
To understand how Clark contrasts his position with Parry’s, let’s begin with liability to defensive force as well as a framework for consent. One common understanding of liability in self-defense (and the view that Clark adopts) is that the aggressor has engaged in an action such that she has forfeited her rights against defensive force. If Culpable Aggressor is attacking Innocent Victim, Innocent Victim does not wrong Culpable Aggressor by defending herself b/c Culpable Aggressor has forfeited her rights against proportionate force by attacking. But, liability alone is not sufficient to make defensive force permissible. There must also be a reason to impose that force. So, if Innocent Victim can simply pinch Culpable Aggressor, then even if Culpable Aggressor is threatening deadly force, there is no reason for Innocent Victim to do any more than pinch Culpable Aggressor to defend herself.
Now here’s my take on consent, one that I hope is generally uncontroversial. First, consenting is not requesting. Requesting not only drops a claim right but gives a person a positive reason to engage in the act. Second, some consenting is just dropping a claim right. You ask if you can borrow my lawnmower. I am not excited about that, but I nevertheless agree. That’s consent. Third, sometimes one consents to an action that is harmful. A and B agree to a boxing match. Not only does A’s consent mean that B does not wrong her but also A’s consent excludes the harmful impact on A as a reason when B considers her own reasons for acting.
Clark begins by dismissing that refusal is as simple as consenting. The problem is that Teresa may still wish to maintain her right to defend herself or (even if as a pacifist she doesn’t want to), she may still want it to be the case that Adam is wronging her and that compensation is owed.
Clark maintains that refusal is either going to effect the “reasons dimension” or the “liability dimension.” (2) For liability, protection is impermissible because the attacker is no longer liable to the third party’s force. For reasons, the refusal changes the reasons the third party has for intervention, either removing reasons that did exist or contributing to reasons not to defend. Parry pursues the reasons path; Clark opts for liability.
Parry’s view (EXCLUSION) is “[a] person X has the normative power to make it so that the fact that Y’s Φing would benefit (or harm) X is not a reason for Y (not) to Φ.” (5) EXCLUSION allows Teresa to defend even though she refuses Dan’s protection because the exclusion of reasons is person-relative.
Clark presents three objections (to which I am going to pepper in some concerns along the way). First, he thinks this view has implausible implications, as he imagines a boulder rolling at A, B, and C. A and B will die, but C will only have his toes broken. B tells you not to take his interests into account. This now means that you should roll the boulder at B, something Clark finds “very implausible.” (7)
Though the set up does make this result counterintuitive, it ignores the way in which we consent. Consider a case in which a boulder is rolling along a trolley track at B (who will die) who is in front of C (who will have broken toes). You can press a button that switches the track so the boulder will hit C instead. If B says, “I consent to be hit,” this seems to me to present the same compelling kind of case where we are just flummoxed that B wants to sacrifice so much to prevent so little harm to C. But if we reject B’s consent, we are going to need a reason to doubt the validity of the consent on that occasion.
Clark has more objections up his sleeve. He claims EXCLUSION is under motivated, rejecting Parry’s assimilation of the need for a power to exclude with “paternalism’s distinctive wrongfulness.” (8). Clark finds this unavailing because one can be wrongfully paternalistic without someone exercising a normative power, such as when A hides his brother’s cigarettes to get him to stop smoking. This, notes Clark, has nothing to do with reasons exclusion.
I am not so sure. Let’s assume that Teresa is being attacked but she is sleeping. Dan knows that she is a committed pacifist—maybe she’s a philosopher who has devoted her life to trying to convince others of her views. Although this may be difficult to imagine because we might wonder whether Dan can truly know that Teresa would refuse protection, it seems that if he were certain, he would be as bound by her theory of the good as if she had voiced her refusal on that occasion. (Perhaps just as there is “hypothetical consent” there is “hypothetical refusal.”)
Additionally, Clark’s argument, that ignoring the exercise of a normative power is why paternalism is objectionable, seems to undermine the standard assumption of how we wrong the boxer in the case of actual consent. Does Clark deny that in some cases it is specifically because the person exercises their will that paternalism is wrong? Maybe there is not just one way in which paternalism is wrong. Perhaps it is wrong when we ignore someone’s will, when we undermine our knowledge of their theory of the good, and when we manipulate them and fail to take them seriously when they are selecting their good (as in the case of nudging that Clark mentions (9)).
Finally, Clark also rejects that if Dan could stop Adam with only a light punch that Dan should not intervene if Teresa refuses. But Clark simply rejects this as counterintuitive. (10)
As I have noted, my worry is how Clark’s objections generalize to consent itself. Now, perhaps Clark can point to the fact that consent impacts the “liability”/rights dimension, not just the reasons dimension. But I think this ignores how important the reasons dimension is in determining the permissibility of third-party behavior. Returning to our boxing match, if A will be seriously injured by fighting but B will lose almost nothing if the match does not proceed, shouldn’t a third party still respect A’s decision? Indeed, even though it is true that B would not wrong A by punching her, it seems that A’s interests, even if not protected by a right, would generate reasons for third parties to intervene unless A also has the power to exclude the harm to her as reasons for others to act. On Clark’s view, all that Teresa does is to reveal “certain of her preferences” which give Dan “some reason to refrain from protecting her,” but “it is rare that such reasons are weighty enough to tip the scale.” (11) Yet, in some cases, we are going to need far more than this to get consent off the ground.
Still, one might think that even if the reasons approach is plausible; there might be something (more) to the liability approach. Indeed, it is notable that although I rely on the reasons aspect of consent in raising my objections to Clark, consent does have a rights dimension.
Here’s how Clark argues for a liability approach. He claims that when someone attacks one person, he renders himself liable not just to his victim but to everyone. He notes that waiver/consent changes all sorts of rights and duties, but he wishes to slice and dice more finely. He argues that “it often matters to us that we have this more fine-grained form of control,” (13) noting it allows Teresa to both deny defense on her behalf as well as to receive compensation. So, too, someone who wants to practice her mixed-martial art skills may wish to defend herself and not have anyone intervene.
His view is CONTRACTION: “[W]e have the power to restrict the domain of persons with respect to whom our attacker is liable.” (14) Clark immediately addresses the most central implication of CONTRACTION—that it seems that Adam, but not Teresa, is wronged if Dan ignores Teresa’s refusal. Clark denies this, noting that we are wronged when others fail to respect the exercise of our normative powers so he believes he can claim that Teresa is wronged.
The more interesting question, though, is whether Dan wrongs Adam if he intervenes. Clark notes the three markers of a rights infringement: 1) Adam would have standing to complain, 2) Adam could use counter defense, and 3) Dan would owe Adam compensation. One might expect Clark to bite the bullet on these implications. Instead, he both grants that these are the typical implications of a right’s infringement and simultaneously denies that they apply in this case, while still maintaining that Adam would be wronged by Dan’s defense.
First, Clark denies that Adam would have standing to complain if Dan broke his leg because Adam would be complaining about the very same act that he engages in himself. Second, Clark rejects that Adam may fight back. Although Adam is not liable to the defensive force, if Dan chooses to engage in it anyway, then, Clark argues, as between Adam and Dan, Adam is more responsible and thus cannot defend. After all, Adam is the one who created the forced choice for Dan such that he must choose not to intervene at the cost of seeing his sister receive a broken leg, and Adam is the more culpable and responsible of the two. Third, for these same reasons, Dan does not owe Adam compensation if he goes ahead and breaks Adam’s leg to stop him. Notably, the second and third implications will depend upon how responsible the attacker is for the threat; minimally responsible attackers may be entitled to counter defense and compensation.
CONTRACTION now appears to be a pretty anemic normative power! We are left with Teresa having standing to complain that her normative power was not respected, but to the extent that any benefit of the exercise of this normative power was supposed to inure to Adam’s benefit, it is hard to see how. Though we may objectively appreciate that Dan violates Adam’s rights if Dan breaks Adam’s leg, Adam can’t defend, can’t be compensated, and can’t even say a peep about it! So, if culpably attacking renders one liable, on Clark’s view, Teresa has the power to erase the effect of the attack but not to render the culpable attack itself irrelevant to Adam’s standing vis-à-vis Dan. She can’t restore Adam’s status so that he is the equivalent of an innocent bystander. He’s still the culpable cause, and when push comes to shove, culpable causes are going to lose in the lesser-evils balance, even when Adam goes against his sister’s expressed wishes.
Let me close with how Clark articulates the different implications of EXCLUSION and CONTRACTION for humanitarian intervention. Clark imagines a case where Attacker is going to throw a grenade into an elevator that will result in the loss of a hand to the elevator’s occupants. He then stipulates that it is proportionate to kill the Attacker if it will save three hands, but not two. Accordingly, under CONTRACTION, if there are three people in the elevator and one refuses, then it would be disproportionate to kill Attacker to protect two hands. Clark does not articulate Parry’s view here, but it does not generate a different conclusion. In a nutshell, Attacker is liable to defensive killing to protect three hands, but a defender cannot access all three hands as a reason to harm the liable Attacker because the refuser has excluded her good from consideration.
According to Clark, the distinction between the views, and the payoff of CONTRACTION over EXCLUSION, occurs within the realm of lesser-evils balancing. In instances in which a defender can thwart a serious harm by an attacker by imposing a minor harm on the attacker, EXCLUSION yields that the act is impermissible (because the victim’s well-being is not a reason defender can avail herself of) whereas CONTRACTION yields that the rights infringement of the attacker is permissible because the victim’s well-being is still a reason.
There are further implications when there are bystanders. Clark imagines 100 people are under attack, but to rescue them, the defender will have to throw a grenade that kills 1 bystander. What if 99 refuse protection? Under EXCLUSION, the 99 lives are excluded, and so defender may not kill one bystander to essentially rescue one person. In contrast, Clark sees it as a benefit of his view that it yields that defender may throw the grenade because the other 99 lives are not excluded as reasons.
Ultimately, my concern here is the same as above. Why can’t the 99 then consent to the harm, which would therefore exclude their well-being? Does Clark find that equally implausible? And if so, are Clark’s objections more to the way in which consent operates generally? And if not, why can’t we have a lesser normative power to exclude ourselves from others’ reasons? At the end of the day, I worry that Clark bestows individuals with a normative power that is not particularly powerful at all as it grants individuals neither the power to truly restore the status of their attackers nor the power to limit the impact that their well-being has on their defenders.
That said, I can’t say I have my finger on the answer to this puzzle, and I know I have I learned a tremendous amount for the piece. I am so delighted that Clark has pushed us to investigate the normative machinery involved in these cases. I look forward to learning more from him and others on this thread!
43 Replies to “David J. Clark: “Refusing Protection”. Précis by Kim Ferzan”
Thank you so much for these careful, insightful, and challenging comments, Kim. There are so many great thoughts worth engaging with here. I can’t get to them all at the moment, so let me respond to a couple of the more central ones.
Response #1. A number of concerns come back to worries about the implications my account has for consent. (This comes up in a discussion of the boulder case, as well as in discussion of the case where one can save 100 people by throwing a grenade that kills their attacker and one innocent bystander.) And towards the end of the paper you ask whether there might be disagreement between us regarding the way in which consent operates generally.
I think there likely is.
Some claims you make regarding consent:
(1) Consenting is not requesting.
(2) Some consenting is just dropping a claim right.
(3) Sometimes one consents to an action that is harmful.
I am fully on board with all three of these claims. But I think our central point of disagreement concerns your fourth claim about consent — the claim that when A consents to be harmed by B that they are also “excluding the harmful impact on A as a reason when B considers her own reasons for acting.”
Take the boxing case. A and B have consenting to the boxing match, and thus consented to all the harm that is allowed within the rules of the bout. But now suppose B has A “against the ropes”, and is in a position to win the fight with one more punch. B can drop A with a mildly painful, but not injurious, punch to the abdomen. Or B can drop A with a brutal punch to the head that will leave A with a serious concussion and long-term brain injury. Both options are equally good for finishing the fight, let’s suppose, and both options are permitted by the rules of the bout.
Even so, I find it intuitively compelling that B should opt for the punch to the abdomen rather than the head, even though I agree that B would not wrong A by punching them in the head. This suggests that A’s consent does not go so far as to entirely exclude the harmful impact on A as a reason for B to consider when acting.
I want to say the same thing about the boulder case (and likewise for the case with the grenade that can save the lives of one-hundred people). The boulder rolls towards A. It will kill her if you do nothing. You can redirect the boulder either towards B or C. If towards B, he will be killed. If towards C, they will suffer a bruised toe. B, however, consents to be killed. It seems to me consistent with acknowledging his consent as valid that we still have more reason to redirect the boulder towards C than B. This is because I think that valid consent here is merely the dropping of B’s claim right not to be killed. But his consent does not also exclude all of the reasons we have not to harm him; it excludes only the reasons supplied by his rights. There is still reason not to seriously harm B — not to kill him in this case — and that reason is, I think, sufficient to override the reasons we have not to infringe C’s right against a bruised toe.
In sum: I think consent does alter the reasons dimension, but I think it does so only insofar as it alters the liability dimension. When A consents to be harmed by B, A removes the reasons that are supplied by her claim rights. But I think that A’s well-being still supplies reasons for B not to harm A.
Response #2. Another worry raised is that, on my account, the power of liability contraction turns out to be a pretty anemic power. “At the end of the day, I worry that Clark bestows individuals with a normative power that is not particularly powerful at all as it grants individuals neither the power to truly restore the status of their attackers nor the power to limit the impact that their well-being has on their defenders.”
This is another nice challenge. Let me try to clarify what my account says about the extent of this purported power of liability contraction.
When Teresa refuses Dan’s protection from Adam, she makes it the case that:
(i) Adam’s “pre-attack” right that Dan not break his leg is restored (which supplies weighty reasons for Dan not to break Adam’s leg, making it impermissible for Dan to break Adam’s leg), and
(ii) Teresa would be wronged by Dan’s protecting her from Adam, since he would be failing to respect her exercise of her power to effect (i).
What Teresa’s refusal of Dan’s protection does not do is to make it the case that:
(iii) Adam is permitted to fight back if Dan does what he should not do and attempts to break Adam’s leg in defense of Teresa,
(iv) Adam has the standing to complain if Dan attempts to break his leg, or
(v) Adam is owed compensation if Dan breaks his leg.
That’s the extent of the power. (Though worth noting one wrinkle: there are some cases — usually involving minimally responsible attackers — where the refusal of protection does grant the attacker a permission to defend himself, the standing to complain about intervention, and a claim to compensation.) It’s, admittedly, a less grand power than the power that Parry attributes to refusal. But I think (i) and (ii) are still quite significant alterations to the normative landscape! In any case, I don’t see it as a cost of the account that it does not grant Teresa with the power to effect (iii), (iv), and (v) by her act of refusing protection. In fact, I see it as a virtue of the account that it does not grant Teresa with the power to effect (iii), (iv), and (v), since I find it pretty counterintuitive that Adam, a culpable attacker, has the standing to complain if Dan intervenes, a permission to fight back, and a claim to compensation.
Now there remains the worry (expressed in the quote above) that my account does not grant Teresa the power to limit, by her act of refusing protection, the impact that her own well-being has on the defender. But my account does predict that refusal will have some impact in this respect. For one thing, there is (ii) above. By refusing protection, Teresa makes it so that Dan’s protection would fail to respect her exercise of her normative powers; it would treat her as though she lacked such powers. It’s in Teresa’s interests that Dan not treat her thus. Likewise, I think it is in Teresa’s interests that her revealed preferences are satisfied and that others exhibit respect for her values and commitments (above and beyond respect for her normative powers). When Teresa refuses protection, she may well reveal something about her preferences, values, and commitments — and in so doing alter the way in which intervention by Dan would impact her well-being.
Again, this impact is much less grand than the impact predicted by Parry’s theory. But again, I’m pretty content with that result.
Thanks for the paper David, which I really enjoyed (and for the great reply from Kim). One thing I’ve been wondering about is how your view will deal with risk. (Sorry for the long comment.
In *Risky Other-Defence*, suppose that Threatener unjustifiably threatens to paralyse Victim from the waist down. Victim is unable to defend herself. There is a 99 percent chance Threatener will succeed unless Rescuer shoots Threatener in defence of Victim. This carries a 5 percent chance of causing Victim to lose the use of one of her arms. Victim competently refuses her consent to Rescuer’s defending her.
(Sorry, I’ll hold my hands up that this is a case from my paper, and I’m being lazy, copy and pasting it!)
Stipulate that being paralysed from the waist down is worse than being exposed to a 5 percent chance of losing the use of one of one’s arms. This means, I think, were Victim unconscious, it would be permissible for Rescuer to kill Threatener in defence of Victim, because she may presume consent when refraining from acting is worse for Victim. Nonetheless, given that Victim’s refused consent, I think it’s impermissible to intervene.
And in this case, I feel like the Exclusion Account gets the explanation right. It’s one thing to subject me to a risk that’s ex ante in my benefit. It’s another thing to justify imposing that risk on me by citing the fact you’ll ex ante benefit me by so doing. I think this might just be the way consent works (perhaps echoing some of Kim’s thoughts). I think if someone’s fallen over and is struggling to get back up, I might have most reason to help them up. But if they don’t consent to me helping them up, that reason is excluded from my deliberation about whether to help them. Analogously, when Victim refuses consent in Risky Other-Defence, that I can expect to benefit Victim shouldn’t figure in my deliberation.
So my worry is, without the exclusion account, how d’you explain Risky Other-Defence in a satisfying way? It’s not merely that Rescuer would subject Victim to risk of harm that they don’t consent to. It’s that, additionally, they use the good to Victim to justify imposing that risk on them. And if that complaint has force in Risky Other-Defence, won’t that complaint also have force in non-risky cases in which rescuers nonetheless use the good of victims to justify their interventions?
(I think this is also going to lead to problems in risky other-defence cases with groups of people…but this comment is quite long already!)
Hi all. Thanks for this! Great paper and great précis.
So, I think Parry’s account might suffer from a problem which (I think) Clark’s account avoids (which is a point for Clark’s account). At least since JJ Thomson’s groundbreaking work on defensive rights, we’ve pretty much come to recognize that the right to self-defense has a fundamental and ineliminable agent-neutral element to it. There might be agent-relative reasons that kick in under certain circumstances – this is how some people have analyzed “falling man” cases (in my view, wrongly). But either way, in cases where we have a responsible or culpable individual aggressing against a morally innocent victim, there will typically be agent-neutral reasons to “take the side” of the victim.
If Parry’s account (EXCLUSION) is correct, Victim has the normative power to make it so that no one has reasons to attack Aggressor except for Victim. But in such a case, are Victim’s reasons to resist Aggressor agent-relative, or agent-neutral? If they’re agent-neutral, then they seem agent-neutral “in name only” since they are by hypothesis reason-providing only for Victim and no one else. If they’re agent-relative, then this problematically constrains what wide-proportionality would permit in Victim’s defense against Aggressor.
At least at first glance, Clark’s view (CONTRACTION), doesn’t seem to have this problem since it doesn’t traffic in reasons-talk. Rather, Victim can exercise a normative power to restrict the domain of persons with respect to whom her attacker is liable. And it is well recognized that liability is a multi-part relation; i.e., to be liable is to be liable to specific persons and to specific deprivations. So restricting the liable parties to just one – viz., Victim – seems in principle unproblematic.
Or maybe I’m wrong – Maybe Clark inherits the relativity/neutrality challenge that I‘ve attributed to Parry. Or, alternatively, maybe the relativity/neutrality isn’t an issue for all of them after all. Either way, I have to think more about this. I thank Clark (and Ferzan) for helping me do so!
Hi David, thanks for the excellent paper. It is very clear, precise, and well-argued. Your objections to Parry’s Exclusion Account seem to me to be decisive. Your reply to the objection that your Liability Contraction Account implies that Adam could justifiably harm Dan in self-defense if Dan does intervene also seems right on the mark.
I found your objections to the Refusal-As-Waiver Account unconvincing, however, because I don’t share your assumption that “[w]hen T waives a right that A not 𝝫, she eliminates all of the protections afforded by that right”. It seems to me that Theresa can waive her right not to be attacked by Adam without giving up her right to defend herself against the attack and without giving up her right to be compensated for any injuries the attack might inflict upon her. In principle, Theresa could truly say to Adam, “I waive my right that you not attack me, but not my right to defend myself against the attack nor my right to compensation from you for any injuries that you inflict on me.” That would result in Adam’s attack not posing a threat to Theresa’s rights, which would mean that Adam would not be liable to defensive violence on the part of Dan. And that would explain the impermissibility of Dan’s intervention.
Indeed, I think that, absent consent to the attack, Theresa does not have the authority to demand that Dan stand down. Absent that consent, Dan’s right to liberty includes the right to defend Theresa against Adam. Theresa can request that Dan not intervene. Maybe out of respect for Theresa’s values and desires, Dan shouldn’t intervene. But he is within his rights to intervene because Adam is solely responsible for an unjust threat to Theresa and so is liable to defense on the part of third parties like Dan.
Now obviously in your example, Theresa does not consent to being attacked. She’s a pacifist and so why would she consent to Adam engaging in violence against her? And that means I am not convinced that in your example Dan is not acting within his rights if he does intervene.
terrific paper, and I’m really enjoying the discussion so far. I had two initial questions.
1) The first draws out something from Kim’s discussion of consent. What is wrong with a qualified version of Parry’s view? Consider a generic case where Attacker attacks Victim, and a third party, Defender, can prevent the attack. On the version you advertise Victim can withdraw their welfare as a reason for Defender to prevent the attack for any reason. This implies that Victim’s life is not a reason for Defender to act even where they can save Victim’s life by inflicting some low cost on another person (either attacker or someone else), as you point out. But can’t we just qualify the view: Victim is able to withdraw their welfare for Defender to prevent the attack when doing so is [here are some alternatives]…a) rational; b) reasonable; c) consistent with Victim’s self-regarding duties. As it would be irrational, unreasonable and a violation of one’s self-regarding duties to sacrifice one’s life for the sake of Attacker’s toe, this view does not imply that Defender acts wrongly where they save Victim’s life by inflicting the cost of a toe on Attacker, or on third parties. And that view chimes with limits on the power of consent.
2) One thing that it seems impossible to do is for Victim to restore Attacker’s rights against Victim. Victim cannot make it the case that Attacker has a right not to be harmed against them. For example, suppose that Victim is a pacifist, and so thinks it would be wrong for them to defend themselves. But they are wrong about this. They attempt to restore Attacker’s rights against Victim, so that Victim would wrong Attacker were they to defend themselves. Victim then gives in to temptation, and defends themselves and so harms Attacker. It is not plausible that Victim wrongs Attacker. You claim that Victim has the power to restore Attacker’s rights against Defender. What explains the asymmetry? (There is a parallel question for Parry’s view – Victim also can’t withdraw their welfare as a reason for their own action – and it would be interesting to see which view can offer a better explanation of the asymmetry).
I find this issue fascinating. I thought Parry’s paper was an excellent exploration of one potential response, and David’s new paper continues this rich seam.
I want to focus on the boulder case. David says it is implausible that we should listen to Bob when he says ‘Don’t save me! Save Charlie’s toe!’ and that Parry’s view implies that we should. As VT points out above, Parry’s view can address this by introducing limits to our power to remove our wellbeing from consideration (or by acknowledging that we have non-wellbeing-related reasons to defend and protect each other).
There are two aspects of the boulder case I want to draw attention to — it’s a LARGE DIFFERENCE case (toe vs life), and it’s a NATURAL THREAT case. Now, once we play with these two variables, I think two problems arise for David’s view.
1. Large difference vs. Small difference (natural threat)
Imagine in the boulder case the boulder will break either Bob’s big toe or Charlie’s little toe. (I am imagining it is slightly worse to have your big toe broken). Bob says ‘save Charlie! Don’t save me!’ Jonathan’s view suggests we should save Charlie. David’s view is inert here — you can’t restore a boulder’s rights. So if David is right BOTH about his positive view, AND that JP’s view doesn’t work, we should save Bob’s big toe despite his protestations. I find this implausible. If that’s right, it’s important — Jonathan’s view has something to say about this case and David’s doesn’t AND it suggests that what goes wrong in the original boulder case is along the lines VT outlined above. It’s the big difference, not the ability to control whether our wellbeing counts, that does the intuitive work. But if it’s the big difference that is the problem (rather than any difference) David’s diagnosis and preferred solution is off the mark.
2. Natural Threat vs. Attacker
Now let’s take the other variable. As we have seen, David’s view appears to say that what people want in natural threat cases is irrelevant to whom we should save. But this means that it makes a massive difference whether the threat is a natural threat vs. an Attacker. Compare the small difference boulder case with this one:
Bob and Charlie are both being attacked by Aggressors. You can fend either Aggressor off with narrowly proportionate harm. Bob says ‘Please! don’t save my big toe! Please save Charlie!’ (Charlie’s little toe is under threat). According to David, we should listen to Bob in this case, since Bob has the power to restore his attacker’s rights. Imagine in both the Boulder case and the Attacker case Bob’s motivations are the same — save Charlie, for Charlie’s sake. Yet apparently we should not listen to Bob if the threat is natural (or maybe innocent), and listen to him if the threat is culpable. I find this contrast between natural threats and aggressors implausible.
Now consider cases where we restore the attacker’s rights for the sake of the attacker (like David’s sibling cases). It means that we can direct others to save our loved ones if they attack us, if they wrong us, but not (for their sake) if they are innocent, and fellow victims. We can only save our innocent siblings via the mechanism of saving those who wish to do us wrong.
Super paper! Thanks to David for writing it and to Kim for her excellent comments (and for protecting me [with my permission!] from David’s attack!])
A few quick thoughts (I’ll hopefully have time to contribute some more later)
1. Picking up Pat’s natural threat vs attacker point
So one key difference between my view and David’s view is the scope of our powers to refuse protection: David’s view applies to cases of unjust aggression and operates on liability justifications (when I refuse protection, I restore the potentially-liable attacker’s right not to be defensively harmed), whereas my view applies to both unjust aggression cases and natural threat cases (when I refuse protection, my good/wellbeing does not stand in a justifying relation to harming the liable attacker or to over-ridding innocent people’s rights as the lesser-evil).
To report my own intuitions: I find it more intuitive that we have a power to control whether it is permissible to harm innocents as the lesser-evil, than that we have a power to control whether it is permissible to harm liable aggressors (of course, I think we have both powers). As Kaila points out above: one might think that we can proceed with using defensive force against a liable attacker, even in the face of victim’s refusal, simply because we have strong reasons/a right to uphold justice. So I think it is a virtue of my view that it applies to lesser-evil cases too.
Why do I think that lesser-evil cases are the more central/intutive cases? I think we can get a handle on this by thinking about the underlying values that might justify our moral powers. Why is is valuable for us to have the specific power to control the use of our good? I think one reason is that when people justify their actions by appealing to our good, that implicates us in what they do in a distinctive way. And we care about what we are implicated in. Now, it seems to me that it is more obvious that I have reason to care about whether my good is used *to justify overriding the rights of innocent people* (lesser-evil justification cases) than that I have reason to care about whether my good is used to *to justify harming people who have already made themselves liable* (liability-justification cases).
2. On the limits of the power to control one’s good
I agree with the comments above that the best version of the reasons-exclusion view will need to explain why the power (like all our other rights and powers) has various limits and qualifications to it. I’m reasonably optimistic that a more fleshed out version of the view might be able to accommodate the various counter-examples that David has skilfully identified (I mean, I would say that wouldn’t I?).
But one general thought on this: Though the power to control the use of our good/wellbeing has lots of clear similarities and connections to other kinds of powers (like our powers to control our bodies and property, consent to sex, etc, about which there is much written), it is still sui generis: It involves having normative control *over a purely normative object*. In light of this, I think we need to be careful about inferring that this power is limited/qualified from the ways in which our powers over non-normative objects (like our bodies or property) are limited/qualified. Given the difference in the object of the power, the interests/values that justify the power will likely be different to those that justify our powers to control non-normative objects. I think that figuring out principled limits and qualification to the power to control one’s wellbeing/good will require figuring out which interests and values justify the power in the first place.
3. On Saba’s agent-neutral/agent-relative objection
I’m not quite seeing the problem. As I understand my view, the idea is that: (i) My good/wellbeing grounds agent-neutral reasons, (ii) but individuals have control over whether these reasons grounded in their good stand in a justifying-relation to other people’s actions, (iii) so when I refuse protection from you, this defeats (at least some of) the agent-neutral reasons you have to protect me. So, I’m not seeing why the fact that I have a power to defeat a (putatively agent-neutral) reason creates any problems for classifying that reason as agent-neutral. Sure, the view implies that different people will be differently subject to reasons grounded in my wellbeing. But, to my mind, that doesn’t seem like kind of agent-relativity that people have in mind when talking about the neutral/relative distinction.
I really enjoyed this paper! It has given me lots to think about. My question may be a paraphrase of Victor Tadros’s first question.
You say that ‘If there is such a thing as a power of reasons-exclusion, Bob clearly exercises it’ in RUNAWAY BOULDER (39). But I don’t think this is clear. True, Bob sincerely tells you he doesn’t want his interests taken into account. But successfully exercising a normative power plausibly requires the satisfaction of validity conditions beyond sincerity for its exercise. We know this from other normative powers such as consent. We can divide these validity conditions into procedural and substantive conditions. Procedurally, a valid exercvise of a normative power requires that the sincere expression is not induced by coercion, deception, etc. But even if all the procedural conditions are satisfied, it may not be possible to validly consent to certain kinds of conduct on purely substantive grounds (e.g. being tortured, being enslaved, etc). If RUNAWAY BOULDER is such a case, then it is unclear that Bob exercises the power of reasons-exclusion. And I think something similar might be true in EASY DEFENSE (42). If all that is correct, then it might undercut one of your key objections to Parry-style accounts. So, along the lines of Victor Tadros’s first question, I ask: couldn’t we explain away the putatively counterintuitive consequences of a Parry-style view by appealing to the notion of substantive constrains on the valid exercise of the normative power of reasons exclusion?
Really cool paper, David, and great discussion from Kim and everyone here. I have to admit, with the Siblings case as the beginning, I surprised myself to find that I had some pretty paternalistic intuitions. And then, as I read through the paper, I wondered if I was happy with some of the later results of your CONTRACTION account because they were the results I liked… possibly for reasons of paternalism. Let me think out loud here, explaining my paternalistic intuitions and seeing if there’s any challenge to CONTRACTION that comes out of it.
You set up Siblings in a perfect way to make it clear that Dan’s protecting Teresa would be objectionably paternalistic: Teresa is a committed pacifist; Dan knows her well because they are siblings; and Dan can credibly trust Teresa when she says that she doesn’t want Adam to be harmed by Dan in her name. Being a committed pacifist is a morally strong reason for refusing protection. So is the other reason for refusing protection that I noticed in the paper (apologies if I missed something): Not wanting someone else to protect you because it’s important to you to have the opportunity to act in self-defense on your own behalf. Another strong reason relevant in the humanitarian intervention context: Victim conceptualizes self- and other-defense as being subject to a success condition, and Victim reasonably thinks that, based on what they know about Attacker and Defender, Defender’s actions will tick Attacker off and make things worse. And/or Victim reasonably thinks that Defender’s intervention would make the conflict more protracted and result in more overall harm. (In the real world, other-defense usually isn’t a matter of pressing a Looney Tunes-style Acme Corporation button and Attacker is gone forever.)
However, there are some bad reasons I can think of to refuse protection:
(1) You have extremely low self-esteem and don’t believe your interests matter.
(2) You are brainwashed into thinking the attacker is entitled to harm you. (Perhaps brainwashing is done by the attacker.)
(3) You have been raised in a sexist culture and you’re a woman and also (1) and/or (2).
(4) You think that your mixed martial artist skills are way better than they are, thus you ignore all the evidence suggesting that you could use some outside help based on your fiery yet delusional belief that you got this.
(5) You have been raised in a machismo culture and you’re a dude and also (4).
(6) Your thing is psychological warfare and you think that by refusing protection and being harmed, this will make the attacker feel really bad later, which is a punishment you want to inflict on them.
(7) Etc. etc. etc.
I’m thinking that in a lot of cases, especially in an interpersonal other-defense situation that one might find themselves in involving strangers, it’d be really hard to determine the quality of Victim’s reasons for refusing protection. This is my paternalistic intuition: In a lot of cases Victim and Defender aren’t siblings who know each other, so Defender is permitted to intervene because they might reasonably feel that it’s morally risky to take Victim at their word when there could be bad reasons for Victim’s refusal. Very paternalistic, I know! But I also have to suspect that David might have similar inclinations/worry that this would make everything messier, which is why Siblings was so carefully constructed to avoid considerations along these lines. (Is this the case? And is it actually bad to hide cigarettes from a nicotine addict?)
If we accept that there may be significant epistemic uncertainty surrounding an agent’s reasons for refusing protection, I wonder how this affects some of the cases at the end of the paper, e.g., “Grenade Defense: Attacker launches an attack. If successful, each of one hundred people will be killed. Only Defender can stop Attacker, and he can do so only by throwing a grenade at him. The blast will kill Attacker, but it will also kill an innocent bystander. Ninety-nine members of the target group refuse protection” (p. 55).
If I understand you correctly, Parry’s EXCLUSION account gets the wrong result: The ninety-nine refusers exclude their interests from Defender’s calculation; Defender must do harm or allow harm; allowing harm is better; so all hundred are killed by Attacker. On your CONTRACTION account, what the refusal of the ninety-nine does is impact Attacker’s liability: Attacker is only liable to Defender’s harm for the life of the non-refuser; but the interests of the ninety-nine refusers are not excluded from the overall lesser-evil tallying-up that Defender is required to do; and it’s OK to kill the bystander for lesser-evil reasons. The right result!
However, I wonder if paternalism-based permissibility reasoning can explain the right result more simply. In the moment, Defender is faced with a hard choice, and it must be pretty shocking to suddenly hear that the ninety-nine members of the target group say they don’t want protection at the cost of their lives. It would be understandable for Defender to suspect that there may be bad reasons for refusing protection, like mass brainwash. Such a refusal of protection is hard to understand, and if there are bad reasons underlying refusal, it’s natural to not want these reasons to be why a hundred people die. So Defender is permitted to intervene for paternalistic reasons.
This may in fact be compatible with CONTRACTION, since CONTRACTION is an account of liability, not permissibility. But suppose that CONTRACTION is correct and that it’s also correct to say that refusing protection has to be for good reasons. If in interpersonal self-defense situations, it can be hard to know how good Victim’s refusal reasons are in the moment, doesn’t this mean that Attacker’s liability is quite opaque to Defender? Don’t we need an account of permissibility to supplement it?
The thing is, I have to admit that my paternalism intuitions are weaker in thinking about more realistic real-world humanitarian intervention cases of refusing protection. My sense is that success condition reasons, as outlined above, are pretty good reasons and extremely pertinent to the humanitarian intervention context—they are likely to be *the* reasons for refusing protection, whereas (1)-(6) above are more pertinent to interpersonal self-defense. But I haven’t figured out how that affects my thinking about CONTRACTION vs. EXCLUSION.
I’m absolutely thrilled by the quantity and quality of the comments thus far. Thank you all so much. I’ll try to get back to as many of these as possible today.
I’ll start with Kaila’s nice pushback. Kaila, you stick up for the Refusal-as-Waiver account. On that account, refusal is an exercise of the power of rights-waiver: when T’s refusal succeeds in undermining D’s otherwise-permissible protective intervention, it does so by making it so that T gives up her right that A not attack her. T thereby makes it so that, although A poses a threat of harm to her, he no longer poses a *rights-infringing* threat of harm.
You say, “It seems to me that Theresa can waive her right not to be attacked by Adam without giving up her right to defend herself against the attack and without giving up her right to be compensated for any injuries the attack might inflict upon her. In principle, Theresa could truly say to Adam, ‘I waive my right that you not attack me, but not my right to defend myself against the attack nor my right to compensation from you for any injuries that you inflict on me.’” On the one hand, I agree with the basic idea that our powers over our rights can be exercised in more or less general ways. That’s the idea at the heart of my positive account. So I’m with you there.
But on the other hand, I am not sure the power can be exercised in the particular way you mention here. Here’s my thinking. On my view, the very reason T has a permission right to use defensive force against A in the first place has to do with the fact that such force is necessary to prevent A from infringing her rights. So if T were to give up her right not to be harmed by A, she would thereby make it the case that defensive force is *not* necessary to prevent A from infringing her rights — she would thereby give up the explanatory grounds of her permission right. (Likewise, I think the thing that would explain why T is owed compensation if A succeeded in harming her is the fact that A would succeed in infringing her rights. Only rights-infringing harms call for compensation. So if T were to give up her right not to be harmed by A, she would thereby give up the explanatory grounds of her compensatory claim.)
Another thought about the Refusal-as-Waiver account: even if it is possible, in principle, for T to consent to A’s attack without undermining her own permission to fight back and her claim to compensation, I’m doubtful that this is what one is doing when they exercise whatever power they exercise when they refuse protection. It just seems like whatever one is doing — or at least trying to do — when they refuse protection, it is not to grant their attacker permission to attack them.
But while I’m very down on the Refusal-as-Waiver account, I think some of your other comments might suggest a rival theory of refusal that has more going for it (or that I find harder to make the case against). On this view, refusal does not involve the exercise of any normative power. It might alter the balance of reasons to some extent, in virtue of the fact that the act of refusal expresses the preferences, values, and commitments of the refuser. But by refusing protection one is not consenting to be attacked (as on the Refusal-as-Waiver view), exercising a power of reasons-exclusion (as on Jonathan’s view), nor exercising a power of liability-contraction (as on my view). You’re just making a request, stating your preferences, expressing your commitments, or making a demand. But you’re not exercising any normative power. Call this the Powerless Account. (It’s of course compatible with this account that T could render D’s protection impermissible by consenting to be attacked by A. It’s just that the act of refusal isn’t itself an act of consent.)
I have a good number of friends and colleagues who share your judgment about the Siblings case, or who at least don’t find it obvious that Teresa renders Dan’s protection impermissible by refusing his protection. Most of them seem to hold some version of the Powerless Account. Kaila, is your view more along the lines of the Refusal-as-Waiver Account or the Powerless Account?
Victor and Karamvir ask: “Couldn’t we explain away the putatively counterintuitive consequences of a Parry-style view by appealing to the notion of substantive constraints on the valid exercise of the normative power of reasons exclusion?”
That’s a really nice challenge. Not all consent is valid; not all consent succeeds in doing what consent is supposed to do (e.g., suspending certain rights). Consent can fail to be valid on procedural grounds. And presumably also on substantive grounds: a competent adult cannot validly consent to be tortured, for instance, even if they give their consent free of coercion, deception, etc. Why not think the same of the power of reasons-exclusion?
To properly answer this question, it would be awesome to have an account of the substantive constraints of *consent* to hand (where this includes an explanation why there are these constraints in the first place). We could then think about what substantive constraints this account would predict for this other power of reasons-exclusion. Then we could see how well our judgments about cases involving refusal fit with the predictions we get when we amend Jonathan’s Exclusion account with this account of the substantive constraints on the power of reasons-exclusion.
I don’t have an account of the substantive constraints of consent to hand, but let’s think through this in a less thorough way. Take the Boulder case. The boulder rolls towards A. It will kill her if you do nothing. You can redirect the boulder either towards B or C. If towards B, he will be killed. If towards C, they will suffer a bruised toe. B, however, consents to be killed. He goes so far as to demand that you ignore his interests and redirect the boulder his way. Let’s suppose that B’s consent is invalid on substantive grounds. Let’s suppose the problem is one Victor suggests: B’s giving up his right not to be killed is inconsistent with his self-regarding duties. He has a duty to not sacrifice his life so that someone else’s toe is not bruised. If this duty blocks his consent from waiving his right not to be killed, then it seems plausible it would also block his refusal of protection from excluding his well-being from your reasons for action.
This diagnosis seems plausible in cases where there is this kind of mismatch between the severity of harm threatening one person and the severity of harm threatening another person (e.g., death and a bruised toe). At least, it’s not obvious that we don’t have self-regarding duties to not make *massive* sacrifices for very small goods. But consider cases where there isn’t any *pairwise* mismatch, but only a mismatch that results from interpersonal aggregation. Here’s a variant of the boulder case. The boulder rolls towards one-thousand people (the As). It will kill them all if you do nothing. You can redirect the boulder right or left. If you redirect it to the right, it will break the legs of one-thousands people (the Bs). If you redirect it left, it will break the legs of one person, C. Each of the Bs tells you to ignore their interests and consents to have their legs broken.
Here I find it hard to resist the idea that each of the Bs *validly* consents to have their legs broken. If you redirect the boulder their way, it seems they are not wronged, they have no standing to complain, they are not owed compensation, etc. Perhaps it seems so clear to me that they validly consent because I don’t think each of the Bs has a self-regarding duty to not sacrifice broken legs to save someone else from broken legs.
If the Bs satisfy the conditions for valid consent, however, then, given Jonathan’s Exclusion account, we might also expect them to satisfy the conditions for the valid exercise of the power of reasons-exclusion. And yet here I still find it very counterintuitive that you are required to direct the boulder towards the Bs rather than towards C. (But maybe my intuitions are idiosyncratic here. Maybe I’ve internalized my own theory to the point where it’s corrupted my intuitions about these kinds of cases. I’d be curious to know what other folks think here.)
In short: I agree that the appeal to validity conditions on the exercise of the power of reasons-exclusion might help the Exclusion account to handle some of the cases I’m worried about. But I’m not sure the appeal enables the account to handle every such case. This is because it seems like we can validly consent under conditions where the refusal of protection does not succeed in having the moral effects predicted by the Exclusion account.
Because I think that, in your Siblings case, Dan retains the right to defend Theresa in spite of Theresa’s refusal, technically I have what you are calling a Powerless Account. My qualification would be that Theresa does in fact have the normative power to render Dan’s intervention unjust by waiving her right not to be attacked by Adam. I would add that she could waive her right not be attacked without waiving her right to defend herself or her right to compensation. That is, she could set as the conditions for waiving her right not to be attacked that she retain a right to defend and that she be compensated for any injury that results from the attack.
oops, anonymous is me.
Thanks for the great paper, David. I have also really enjoyed the discussion so far. I share many of the concerns raised above (esp. Victor’s point about qualifying Jonathan’s view, Patrick’s point about David’s view implying a radical distinction between natural and agential threats, and Jen’s point about the potential relevance of epistemic considerations). I also have many other thoughts – here is one that hasn’t been raised yet.
It concerns the distinction between the “reasons dimension” and the “liability dimension”. David says that these are “two standard conditions on permissible defensive harm”. I wonder whether and how this understanding is compatible with being an internalist about narrow proportionality and liability.
In Siblings, if one thinks that narrow proportionality is internal to liability (it is difficult to see that how one could be an externalist), one thinks that Adam is liable to a defensive harm only if inflicting the harm is narrowly proportionate – Adam is liable to a defensive harm only if the harm averted (the saving of Teresa’s leg) compares favourably to the harm inflicted (the breaking of Adam’s leg). The liability dimension and the reasons dimension are thus connected: we cannot determine whether Adam is liable to a defensive harm without considering the reason for inflicting this harm (the reason to save Teresa’s leg).
The disagreement between Jonathan and David thus seems not to be a disagreement over whether victims’ refusal affects reasons or whether it affects liability – it is best thought of as a disagreement over whether the refusal affects reasons or whether it affects other conditions of liability. (On Jonathan’s view, Adam is also not liable to be harmed by Dan, as there isn’t sufficient reason for harming him.)
The key question is what, on David’s view, other liability conditions victims’ refusal can affect. I find this question difficult to answer. Adam is responsibly and culpably threatening Teresa. Teresa’s refusal to be saved by Dan cannot change this. There is a reason to harm Dan, as doing so will save Teresa’s leg. But because David rejects Jonathan’s view, this reason remains in place. It may be said that Adam is responsibly and culpably threatening *to violate Teresa’s right*, and perhaps Teresa’s refusal affects this. But David thinks that Teresa retains her right not to be harmed by Adam (which is why Adam will still wrong Teresa, and will owe her compensation), and so it cannot be that Teresa’s refusal makes it the case that Adam will no longer threaten to violate Teresa’s right.
What David says is that Teresa has the normative power to “mitigate specific normative effects of [her right not to be harmed by Adam]” – in particular, Teresa has the normative power to make it the case that Adam is not liable to harmed by Dan. But why? By what mechanism is this achieved?
It may be helpful to step back and reflect on liability more generally. Suppose we conceptualise liability as localised distributive justice (along the lines proposed by McMahan and others). On this view, it seems that despite Teresa’s refusal, Adam is still liable to be harmed by Dan. This is because Teresa’s refusal does not make it any less just that Adam is harmed (as compared to when she herself is harmed). After all, the harms at stake are equivalent, and Adam is a culpable attacker and Teresa is wholly innocent. (Perhaps it may be said that Teresa’s refusal does make it somewhat less just that Adam is harmed *by Dan*. But still, it is not plausible that her refusal can make this outcome *less just than* when she herself is harmed by Adam.)
Suppose we instead adopt the “moral status account” developed by Jon Quong. The key question is then (roughly) whether Adam treats Teresa as if she lacked a right that she in fact possesses. The answer is yes. On David’s view, Teresa retains her right not to be harmed by Adam, and so, harming Adam will still prevent him from violating a right that Teresa in fact possesses (despite Teresa’ refusal). Adam is thus again liable to be harmed by Dan.
And so, contrary to Clark’s view (and, in fact, Jonathan’s view), Adam should be liable to harmed by Dan despite Teresa’s refusal. I agree that Dan acts impermissibly if he harms Adam, and so Teresa’s refusal does make a moral difference. But I don’t think her refusal makes a moral difference by way of affecting liability. It may be that no one has developed quite the right view yet.
Some other thoughts:
 David, your “Powerless Account” is interesting, but it cannot be the whole picture, I think, because it is incapable of distinguishing between the refusal of a victim and the refusal of an unconnected bystander. Take the paradigm case in which a victim’s refusal makes a significant moral difference. Son culpably threatens to kill Mother. Defender can prevent this by killing Son. Mother says no. Most people would agree that Defender cannot permissibly intervene. But compare this with a case in which Attacker culpably threatens to kill Victim, and Defender can prevent this by killing Attacker. Bystander, an unconnected person who just happens to be on the scene, urges Defender not to intervene because she is a committed pacifist. (She may also urge Attacker not to attack.) Mother’s refusal makes a significant moral difference; Bystander’s refusal makes little to no moral difference. If we conceive of someone’s refusal purely as an expression of a preference, we cannot plausibly explain the difference between these cases. (To dramatise the contrast between these cases, we can even imagine the pacifist Bystander having a stronger preference than Mother – still, it is not plausible that her refusal makes a greater moral difference.)
 I find David’s case of 1,000 legs vs. 1 leg really interesting (and very difficult). It seems to make a difference whether each of the 1,000 people knows (or reasonably believes) what the others will do (or have done).
Imagine first a version of this case in which each of the 1,000 people has to decide whether they want to be saved sequentially. The first person is a pacifist, and says no. The second person is also a pacifist, and say no. And so on. In that case, it seems that the people who have to decide later cannot validly refuse. Just as one’s refusal can be invalid because of self-regarding duties, one’s refusal can also be invalid because of other-regarding duties. (This may or may not be in tension with what Jonathan says in section V of his paper.)
Imagine next a version of the case in which each of the 1,000 people has to decide when they do not yet know how the others have decided (suppose that they have to decide simultaneously by pressing a button). Suppose further that no one has reason to think that the others are different from most ordinary people – they all have reason to think that the others will not refuse (because they all have reason to think that the others care more about their own self-interests). It just so happens, however, that all 1,000 people are pacifists, and they all refuse to be saved. In that case, the victims’ refusal seems valid.
I do still think that David’s general point here is right. Even in the second version of the case, it is not plausible, as Jonathan’s view would imply, that Defender is required to break the 1,000 legs rather than the 1 leg.
 Connected to , here is another view, which we may call DISCOUNT. The idea is that a victim’s refusal does not exclude her wellbeing from providing a reason entirely; it undermines the strength of that reason. This view allows us to avoid the implication in  – even the discounted reasons, once aggregated, still make it the case that Defender is required to break the 1 leg. And this view may have a plausible rationale. My wellbeing grounds a reason for action typically because it is valuable from an agent-neutral perspective (it is valuable impartially) *and* it is valuable from an agent-relative perspective (it is especially valuable to me). If I don’t want my wellbeing to ground a reason, I make it the case that the action which promotes my wellbeing is no longer especially valuable (as it is no longer especially valuable to me). But it is still valuable from an agent-neutral perspective; my refusal, or preference, cannot affect that.
Jonathan does have some objections to this point (which Seth Lazar, among others, has raised), but perhaps those objections are not decisive. I am still not sure whether DISCOUNT is ultimately right, however, as it may or may not face the same problem concerning liability that I raised.
Thank you David for a great paper, Kim for a thought-provoking precis, and everyone for their stimulating engagement with the paper!
I am intrigued by many of the points made, but one that I find very worrying for David’s view and that David (puzzlingly!!) is not too worried about is Kim’s point that CONTRACTION comes out as a “pretty anemic normative power.” I agree with Kim here. The worry, it seems to me, is not (merely or primarily) the intuition that “surely the power must be stronger than that!”, but is, instead, a worry that there is not, after all, a contraction of liability. If Adam is not entitled to fight back against the harm Dan wants to inflict on him, is not entitled to compensation from Dan, and has no standing to complain, it sounds very much as if Adam is liable to the harm that Dan inflicts on him and the wrongness of what Dan does if he harms Adam does, pace David, reside entirely in the wrong that Dan does to Theresa by ignoring her refusal to be protected by Dan. That Dan also wrongs Adam seems like a mere stipulation, it does not come to anything substantive, really.
Really fun paper, David! I want to pile on with Kim and Susanne on the anemia of the normative power a bit. If I’ve understood Contraction correctly, you would endorse these two claims:
(1) Adam has a right that Dan not impose the unavoidable costs of Adam’s wrongdoing on Adam.
(2) Adam owes it to Dan to bear the unavoidable costs of Adam’s wrongdoing.
Dan has a duty to respect Adam’s right in (1). How stringent is this duty? It’s seems much less stringent than if (2) weren’t true. Dan’s duty to not impose C on an innocent bystander (if that would somehow prevent C from befalling any of the siblings), for example, seems much more stringent. But if the stringency is sufficiently weak, then it won’t take much to outweigh the duty. Dan, for example, may bear some cost himself if he stands by and watches Theresa be harmed – he would be emotionally harmed to see this happen. If the stringency of Dan’s duty to Adam is sufficiently weak, the cost of this emotional harm would be enough to outweigh Dan’s duty to Adam. If that’s so, then Adam would still be wronged, of course. It just wouldn’t take much to justify wronging him (sans the wronging of Theresa). Perhaps there’s something to be said about why Dan’s duty is in fact quite stringent. I’m having difficulty coming up with much of story about why that would be.
Shifting gears, given that (2) is used in support of the claim that Adam cannot defend himself against Dan, etc. I have a hard time seeing why this doesn’t follow:
(3) Adam owes it to Dan to allow Dan to impose the unavoidable costs of Adam’s wrongdoing on Adam.
Is the idea that if Adam could dodge Dan’s cost-imposition, Adam would be morally permitted to? That strikes me as rather odd given that (2) is sufficient to disable the connection between Adam’s being wronged and Adam possessing a counter-defensive permissions. While there’s not exactly a contradiction between (1) and (3) looming, it strikes me as an awfully bizarre pair of claims to hold at the same time. So much so, I’m inclined to give up (1). But at minimum, it seems to further reduce the significance of Adam’s right.
And this brings us back to Susanne’s helpful formulation: It doesn’t look like much of a contraction has taken place. If there’s not much of a contraction, then Theresa hasn’t exercised much of a normative power. If Theresa hasn’t exercised much of a normative power, then her right that her normative powers be respected looks rather weak. If her right that her normative powers be respected is weak, then the explanation for why she would be wronged in Siblings looks weak and even implausible. If she hasn’t been wronged (or perhaps only wronged to a very small degree), then it’s unclear why Dan isn’t permitted to harm Adam.
Hi Jonathan! Thanks for the reply.
So, I grant that Victim’s “good/wellbeing grounds agent-neutral reasons” and that Victim enjoys a sort of control in which she can “dial down” these reasons. She can even choose for whom to dial down those reasons. So far, so good.
Suppose we adopt the following pair of claims. A) Victim dials down neutral reasons everyone has to stop Aggressor’s attack. B) Victim does *not* dial down neutral reasons for Aggressor to stop his own attack. But this pair doesn’t seem coherent. If Victim deprives *everyone* of neutral reasons to stop Aggressor, then doesn’t that mean she deprives Aggressor of neutral reasons to stop herself?
But then I guess we could change ‘A’ so that it says: Victim dials down neutral reasons everyone has to stop Aggressor’s attack, except Aggressor herself. That makes the pair coherent. But then my worry was that this sort of move is illicit, since it gives others neutral reasons to allow Aggressor to do what Aggressor has neutral reasons not to do. I thought that is was verboten, at least on my perhaps naïve reading of agent-relativity. But as you suggested, maybe this isn’t problem. Even if there is some agent-relativity sneaking in here somewhere, I agree that it isn’t of the type, or otherwise doesn’t play the role of those that characterized earlier theories of self-defense.
Apologies to everyone for my slow replies. We actually had to evacuate our home yesterday due to mudslides with this unusual quantity of rain in southern California (my vegetable garden was taken, may it R.I.P.), and I’ve been busy working with the fire department today to determine if it’s safe to return home. It’s kept me from being as active in the discussion as I would have liked.
Anyways, speaking of natural threats … let me reply to Patrick.
First off, let me say that these challenges are tough ones! This is great stuff. While I have something to say to them, these challenges do make me wonder if my account needs some additional pieces and if the nature of refusal is more complicated than I’ve thought. I’ll explain.
Let me start with the “small difference” case you give. The boulder rolls towards Alice. It will kill her unless you redirect it towards Bob, breaking his big toe, or Charlie, breaking his little toe. A broken big toe is only a bit worse than a broken little toe. Absent anyone’s refusal of rescue, you should save Bob (we assume all else is equal). However, Bob says, “Save Charlie, not me!” The worry: my view predicts that Bob’s speech act has no moral effect with respect to whether you should direct the boulder towards Bob or Charlie. (By contrast, Parry’s view clearly gets the result that Bob’s speech act makes it so that you should save Charlie rather than Bob.)
My reply: while the mechanism I focus on in the paper — the power of liability contraction — is clearly not exercised by Bob’s refusal of protection in this case (since the boulder isn’t a rights-holder!), it doesn’t follow that Bob’s speech act has no moral effect, even if we reject Jonathan’s Exclusion account. I think there are at least two ways Bob’s refusal can change the moral landscape without exercising the power of liability contraction. First, it can alter the way your options affect Bob’s well-being, since his refusal expresses his preferences, values, and commitments, and Bob has some interest in your acting in ways that satisfy or respect those things. But second, when Bob tells you not to protect him, he sure seems to be waiving a right that he holds against you — namely, his right that you not harm him. By removing this right he thereby removes one weighty reason for you not to harm him. And since Charlie does not likewise refuse protection, nor waive his right that you not harm him in any other way, Charlie’s right remains in effect. This difference in rights can explain why you should direct the boulder towards Bob, even though this will result in more harm. The difference in harm just isn’t big enough to generate a lesser-evil justification that would permit you to infringe Charlie’s right that you not harm him.
This seems to me a plausible way to analyze the case. But now a new worry emerges! This move would commit me to a more complicated view about refusal than the one I present in the paper. It would commit me to saying that some instances of refusal — such as the refusal of protection from natural threats or non-liable threats — do not involve the exercise of the power of liability contraction, but only the expression of preferences, commitments, and/or values together with (if applicable) the waiver of certain rights that the refuser holds against the refusee. It’s only when the threat is posed by someone who is liable to defensive harm that the power of liability contraction becomes part of the package.
I like simplicity as much as the next person, so I’m not thrilled about that commitment. But let me say two things to (maybe) make the pill a little less bitter to swallow. First, there may also be a problem here for Jonathan. For although his account can explain why we should break Bob’s big toe without any appeal to rights-waiver, I think there is nonetheless pressure for Jonathan to say that Bob engages in an act of rights-waiver. I think that — whatever else Bob does when he says, “Save Charlie, not me!” — he is *also* waiving his right that you not harm him. After all, you don’t wrong him when you harm him. (Maybe Jonathan will want to say that this is because Bob’s well-being doesn’t contribute to your reasons for action. But even if the link between Bob’s well-being and your reasons for action have been severed, it’s still true that you *harm* Bob. And since Bob normally has a right to not be harmed, we need to explain why Bob lacks this right under the circumstances.)
Second, it doesn’t actually seem that weird to me to think there is something different about refusing protection from a wrongful attacker as opposed to a natural threat. Compare two scenarios. In the first, Teresa refuses Dan’s protection from a boulder. In the second, Teresa refuses Dan’s protection from their brother Adam (my original Siblings case). In the first case, Teresa is only expressing to Dan how she would like to be treated. In the latter case, she is plausibly expressing to Dan, not just how she would like to be treated, but also how she would like *Adam* to be treated. She is telling Dan not to harm Adam. I think what Teresa is trying to do in the latter case is to raise a moral shield between Adam and Dan. But she’s plainly not trying to raise a moral shield to protect the boulder in the first case. The two acts of refusal *feel* different from the perspective of the refuser. So maybe it’s not that weird that the refusal *actually* effects a moral shield in the one case and not the other.
This second point relates to your worry about the asymmetry my account predicts between the refusal of protection from natural or non-liable threats and liable threats. The worry: my view predicts that Bob’s refusal of protection from an attacker that threatens to break his big toe has greater moral effects than his refusal of protection from a boulder that threatens to break his big toe. My reply: yes, my view definitely has this implication (and yes, Jonathan’s view does not). It’s just not an implication I find counterintuitive or otherwise troubling.
@Jen, you say that you suspect the Siblings case was carefully constructed to avoid considerations along the lines you mention. Your suspicions are correct. (It’s hard enough to theorize about refusal when it is free of complicating factors like irrationality, coercion, malevolence, etc.!) But you’re right to push me to think about the relevance of these complicating factors.
I have a pretty strong intuition, in the original Siblings case, that Teresa’s refusal of protection renders it wrong for Dan to harm Adam in defense of Teresa. But like you (I take it), this intuition gets weaker as the grounds for refusal get “worse”, and the intuition downright flips when the grounds get bad enough. For example, if Teresa refuses Dan’s protection because she’s been brainwashed by Adam to think he is entitled to attack her (reason #2 on your helpful list), it definitely seems that Dan is permitted to defend her. In fact, this seems much more obvious to me than that Dan is *not* permitted to harm Adam in the original Siblings case!
An interesting question (for anyone who takes refusal to have the power to render protection impermissible) is what explains this relevance of the quality of the grounds of refusal. You mention the possibility that uncertainty is doing some work here. That may well be. But it’s worth noting that Dan would surely be permitted to defend Teresa in the brainwashing case even if he had perfect knowledge about all the relevant facts. So it seems that the quality of the grounds of refusal make a difference even when they’re not generating conditions of uncertainty. (I’d say something similar about the case with the grenade where 99 people refuse your protection. I still have the intuition that you are permitted to throw the grenade even if you were in a position to know that the 99 people had the right kinds of grounds for their refusal. Perhaps we know that they’re all members of a pacifist political group who are deeply committed to the ideals of the group, that they’ve come to those beliefs in autonomy-respecting, rational ways, etc. So uncertainty can’t be the whole story.)
Taking a page from Victor and Karamvir’s earlier comments, one possible explanation is just this. Just as only some acts of consent succeed as exercises of the power of rights-waiver, perhaps only some acts of refusal succeed as exercises of the power of liability contraction. Refusal is subject to “validity” conditions just as consent is subject to validity conditions. This seems plausible, especially when we consider that the sorts of conditions (e.g., brainwashing) that make refusal invalid are the same sorts of conditions make consent invalid.
Now there’s still the question of the relevance of uncertainty when it comes to the permissibility of protecting someone who has refused our protection. Suppose D’s evidence has him 50/50 whether T’s refusal of protection is valid. Is D permitted to harm A in defense of T? I think the answer is straightforward if we’re talking about fact-relative permissibility: D is fact-relatively permitted to harm A just in case T’s refusal is actually invalid.
The answer is much less straightforward if we’re talking about belief- or evidence-relative permissibility. I have to confess this is the first I’ve thought about how refusal alters belief- or evidence-relative permissibility in cases where the validity of the refusal is uncertain, or where it is uncertain whether an act of refusal has taken place. (So I’m very grateful you’ve put me on to this!) I’m eager to hear from others who will have better things to say on this point. As a first-stab, I wonder if the belief-relative thing for D to do is to choose the option that he has the greatest *expected* reason to choose. Suppose D’s total reason to harm A conditional on T’s refusal being *invalid* is greater than his total reason to not harm A conditional on T’s refusal being *valid*. Then, if D is 50/50 whether T’s refusal is valid, the belief-relative thing to do is to harm A.
Further question. Why can’t we accept both views, suitably qualified?
The exercise of normative powers normally involves communication. Suppose that Victim (a morally sophisticated fast talker) communicates to defender specifically that they do not wish to restore Attacker’s rights. However, they communicate that they want defender not to take their well-being into consideration when deciding what to do, because although they think that Attacker has no right against being killed, and have no wish to restore that right, they care about Attacker’s welfare and they don’t want Attacker to be killed for their sake (a la Jonathan). So they ask Defender not to act for this reason. If you reject Jonathan’s account, they have not exercised their normative power to restore Attacker’s right. Does it then follow that Defender would be permitted (or perhaps required if the costs to Defender are minimal) to kill Attacker? That doesn’t seem very plausible.
Now suppose that Victim does want their welfare to be a reason for action for Attacker. They don’t want it off the table. However, they want to restore Attacker’s rights. They just don’t want their life to be the reason why Attacker has lost their rights, and they communicate this to Defender. So they think it wrong for Defender to harm Attacker for this reason. If Jonathan rejects your account, he should conclude that Attacker has nevertheless lost their rights, Victim’s welfare is still a reason for action for Defender, and so Defender should act against the wishes of Victim. We might find this doubtful too (though I’m a little bit unsure).
An alternative view, then, is that we should accept a conjunctive view where Victim has the power either to remove their welfare from consideration (subject to constraints suggested earlier to make it more plausible); or to restore Attacker’s rights. One way of understanding this view is that as Defender is acting for the sake of Victim, they should respect Victim’s judgements, decisions, and reasoning in deciding not only how to act but how to reason how to act (subject to validity conditions, substantive constraints, and so on). And for that reason we should accept the conjunctive view.
@Joseph, thanks so much for another great and challenging comment.
I hope I’m appreciating the worry properly, but I’m thinking my account is compatible with there being a complaint here. On my account, when you protect Victim in the face of her refusal of your protection, you are wronging her: you are treating her as though she lacks certain controls over her own rights. This would be true even if you protect her, not for her own sake, but for the sake of harming her attacker (maybe you’ve been looking for justification to harm him for a long time). But Victim, I think, would have a further complaint if you tried to justify your treating her in this way by citing her own well-being. If you say to her, “Yes, but I’m treating you in this way because it’s for your own good,” then this only adds to her complaint, because now she’s being treated in an objectionably paternalistic way. It’s true that my account, itself, doesn’t explain the wrong of this extra paternalism bit, but I’m thinking our account of paternalism can handle that. (Granted, one of the coolest things about Jonathan’s theory is that it purports to deliver a unified treatment of both the power of refusal and paternalism! My theory needs to be supplemented with an independent theory of paternalism.)
Another quick thought. In your example where someone has fallen and is struggling to get up, I agree that it’s normally wrong for me to help them up if they refuse my help. But I’m skeptical that their refusal has the result of completely excluding the benefit of helping them up from my reasons for action. Imagine the following. A, B, and C have all fallen and are struggling to get up. A and B refuse my help. C does not refuse my help. It’s really cold out, however. If I don’t help A, she’ll suffer minor frostbite and lose the use of one finger. If I don’t help B, he’ll suffer extreme frostbite and lose the use of an entire leg. If I don’t help C, they’ll die. The catch: I can only help C up by helping one of A or B up as well (maybe I need to get A or B out of the way to get to C). I have two strong intuitions about this case. First, I am permitted to infringe A or B’s rights not to be touched without their consent in this case in order to save C’s life. Second, I have *decisive* reason to help C by helping B rather than by helping A. The degree of benefit to B is clearly still making a difference in this case.
Now there’s Victor and Karamvir’s point about substantive constraints on valid refusal that is relevant to this case. One might think that B’s refusal simply failed to be valid, and that’s why B’s interests aren’t entirely excluded. But I think we can tell this story in a way where it is plausible *both* that (i) there is more reason to help C by helping B than to help C by helping A, and (ii) you should *not* help either A or B if you could help C without involving A or B. Maybe we can imagine that A and B are in a mountaineering competition that requires them to summit a mountain without any assistance. Under these conditions, and assuming they’re fully aware that they will lose the use of a finger or leg without help, of sound mind, etc., then their refusal seems to me to be valid. And yet I still think you’re permitted to help C by helping either A or B (and that if you help C, then you should help them by helping B rather than A).
^ Although maybe it is less clear that we can tell the story in such a way where B’s refusal is not invalidated by their *other-regarding* duties (a point Kida raises in regards to a different case). But it’s also not clear to me that B’s refusal *is* invalidated by their other-regarding duties in this case. I’ll have to think more about this…
What a fascinating paper and great discussion! Thank you, David, Kim, and all! Let me add a few words to it. I will try to be brief.
1. I doubt that we have the power to deny others the right to protect us, particularly while maintaining (not waiving) our claim-right not to be harmed. I doubt the “reasons” account supports such a power (mostly because a request by the victim (“V”) that the defender (“D”) stay away is a reason—perhaps a very strong, but certainly not the only reason—for D’s actions; I think David would agree with that). But I also doubt the “liability” account works the magic. Here is why: to eliminate or reduce the liability of the attacker (“A”), V has to do something that would render either A less culpable or A’s attack on V’s rights less wrongful. David chooses the latter route: he maintains that, by saying “no” to D’s protection, V eliminates the wrongfulness of A’s attack, at least as far as D is concerned. But there is a problem: unlike the “reasons,” the “liability” account is objective (agent-neutral), i.e. A’s liability indeed has to be reduced by V’s actions (not merely from V’s or D’s or even A’s perspective). And here I wonder whether it is possible to say that objectively both things happen: the attack on V is wrongful (V retains a claim-right against A’s attack) and the attack on V is not wrongful (D may not infringe on A’s rights to protect V).
2. Perhaps my doubt stems from my more general skepticism about “defense of another” being congruent to self-defense. In other words, it may be the case that the privilege to use force in defense of another is not entirely derivative of the privilege to use force in self-defense; and if I am right about that (I am actually working on this puzzle now), then V has no control in terms of liability (although still has certain control in terms of “reasons”) over D’s choice to protect or not protect V.
3. It follows that I do not believe that, by not respecting V’s wishes, D wrongs A (I agree that D wrongs V to a certain extent). In addition, I am not convinced by David’s claim that A’s privilege to defend himself against the V-defensive attack by D depends on the degree of A’s culpability (“wrongfulness” or “responsibility” for the attack). According to David, a wrongful attacker A may defend himself against the use of force by D (who is acting in defense of V in spite of V’s refusal of protection) if A has attacked V under duress or due to a mistake. Let’s play it out:
– V is an innocent bystander;
– A attacks V with deadly force in order to avoid being beaten up by some villainous person X (the threatened harm to A would normally be less serious than the harm A threatens to V–otherwise A would not need to rely on the excuse of duress; A would be able to claim the privilege of lesser-evils justification);
– V tells D not to protect her against A, but D nevertheless attacks A (in defense of V) threatening A with deadly force.
According to David, but for the duress, A would not be permitted to use deadly force in self-defense; however, duress essentially restores A in the rights of an innocent bystander. I find this implausible: duress certainly constrains A’s choices, but nevertheless it is A’s choice to shift harm from himself to V and to prefer killing V rather than suffering beating from X. A still implements a malevolent plan of killing an innocent person V, while D implements a benevolent plan of defending the innocent V. Should they be deemed morally equivalent? I think not. Disrespecting V’s wishes is wrong, but attempting to murder V (or V’s defender) is much more wrong. I am all in favor of comparative responsibility, and I would agree to hold A responsible of a lesser offense than murder should A kill D, but I don’t think it is morally right to treat A’s and D’s wrongs as morally equivalent by treating A’s use of deadly force as permissible. The same would be true with respect to a mistaken A. The fact that A is “justified” in the epistemic sense still does not make A morally equivalent to a deontically-justified person. Assuming D has the privilege of defending V, D should be permitted to use deadly force against any aggressor, including “innocent aggressors,” whereas they should not be permitted to use deadly force in response. Intentionally causing or threatening harm has independent moral significance, and V or D should be allowed to try to avoid it by the exclusive use of deadly force.
Finally, I have trouble with determining how much responsibility should A bear for the attack to lose the right to defend himself against D’s unauthorized intervention. What if the villainous person X threatens A with less than severe harm? From whose perspective shall we measure the degree of harm with which A is threatened? What if A is partially responsible for his own mistake etc.?
I would love to go on and on, because the paper is so rich and well written, but I realize I need to stop, at least for now. Thanks again for the opportunity to think about such fascinating issues!
Thanks to all commentators for outstanding levels of engagement with David and Kim! In view of this, and more importantly David’s exceptional circumstances at home*, our editors have decided that this thread will remain open for an extra weekday, i.e. until MONDAY 5PM PDT.
* Permission thoroughly sought to consider David’s well-being in this decision
There are so many nice issues raised here. Thanks for this. I’ll probably have to break my thoughts into two different posts.
First, let me address the thought that, “On Jonathan’s view, Adam is also not liable to be harmed by Dan, as there isn’t sufficient reason for harming him.” I actually pressed this thought to Jonathan in conversation previously, with the intention of showing that his account is in the same boat as mine in having the implication that unwanted protective intervention *wrongs the attacker*. But I’m actually not so sure anymore his account does have this implication. (Here I go defending my sworn enemy…)
Take the Siblings case. On his account, Teresa’s refusal doesn’t change the fact that her well-being is threatened. Teresa will still be harmed by the attack; her interests will still be set back. It’s just that she severs the reason-giving force of this harm *for Dan*. Thus, whether Teresa renders protective intervention narrowly disproportionate will depend on the relata of the proportionality relation (What is X in the schema: “Defensive harm H is disproportionate just in case the severity of H is “too great” as compared to X”?) If the proportionality limit is determined by the strength of the defender’s reasons for action (as I take your comments to be suggesting), then Teresa’s refusal would render Dan’s protection narrowly disproportionate. But why think that proportionality is determined by the strength of the defender’s reasons, as opposed to by the severity of the attacker’s threat (or by something else like the stringency of the rights threatened by the attacker or the demandingness of the attacker’s duty to avert his threat)?
I worry that if the proportionality bar is set by the strength of the defender’s reasons, we’ll get some really counterintuitive results. Take an example that doesn’t involve the refusal of protection. Suppose A attempts to break T’s leg. D1 and D2 are both in a position to protect T by harming A. But suppose that if D1 is the one who harms A, it will have the result that A is pushed in front of a boulder, shielding a random bystander from suffering a broken leg.
This benefit to the bystander surely makes a difference to wide proportionality, but on the view I take you to be suggesting it would also make a difference to narrow proportionality. But that seems wrong. It seems to me that the degree of harm that A is liable to suffer is the same whether it is D1 imposing the harm or D2. (Notice that we get the rights results in this case on a view according to which proportionality is determined by the degree of harm threatened by the attacker, or by the stringency of the rights he threatens, or by the demandingness of his own duties to avert the threat.)
All that to say: I think Jonathan’s account is one on which refusal makes a difference only to wide proportionality and not also to narrow proportionality. (An upshot is that this lets him avoid having to tackle the worry — that I’m stuck dealing with — about attackers having a permission to fight back against defense that is unwanted by the victim.)
@Kida. Now to your worry about how it is that the refuser is able to mitigate an attacker’s liability (a worry that I think Vera shares).
The worry is that the conditions of liability are still satisfied for the attacker even after his victim refuses a defender’s protection. And indeed this is true on either a localized-responsibility theory of liability (like McMahan’s) or on Quong’s Moral Status account.
Now I should be clear that what I’m trying to do in this paper is to argue that we actually need to rethink the conditions of liability. We’re missing a small piece of the story. I’m claiming that among the conditions of liability are conditions about what powers have been exercised by the victim of the attack. I think existing theories of liability, since they all leave a condition like this out, are incomplete.
That said, this “addendum” I’m proposing definitely plays better with certain existing theories of liability than others. I think it plays poorly, for instance, with a view on which liability just is a matter of localized distributive justice. But consider a theory of liability like Victor’s, on which Adam is liable in virtue of the duty he owes to Teresa to avert his threat. Specifically: Adam is liable to harm that serves this duty (and he is not liable to harm that is greater than the stringency of that duty). The ground of Adam’s liability is Teresa’s own correlative right that Adam avert his threat. I think it’s plausible to think that, just as Teresa has the power to waive that right entirely (the exercise of that power would be one way of undermining Adam’s conditions of liability), so too she has the power to mitigate only some of the effects of that right — in this case, the scope of its enforceability. That’s just a different way of undermining Adam’s conditions of liability.
An account like Victor’s already needs to be supplemented with a theory for distinguishing between duties that are enforceable (i.e., those that make the duty-holder liable to be harmed in ways that serve the duty) and those that aren’t. My claims about the power of liability-contraction might be thought of as claims about some of the conditions of enforceability. One of the things that can render a duty non-enforceable is that the person who holds the correlative right has chosen for it to be so.
So in short: Yes, my claims in this paper are not compatible with existing theories of liability. But that’s exactly the thought. I think we need to slightly amend our theory of liability to take notice of the way in which the victim can alter the conditions of liability.
(1/4) Super interesting paper, David. Thanks for writing it!
I’m sorry to be late to the discussion. I’ll start by confessing that I don’t find the liability/forfeiture literature especially helpful in thinking through cases of defensive force, since it obscures the salience of the defender’s explanatory reasons. It also has the odd implication (odd to me, anyway) that one does not violate an aggressor’s rights (breach directed duties to them) despite doing a host of horrible things to them – killing, torturing, breaking legs, etc. So, suffice it to say, I approach these questions from a different starting point, which I’ll call a reasons-based account.
With that in mind, here are a few thoughts (broken into separate posts) that might explain how our views intersect. I suspect they put me more in harmony with Parry’s Exclusion Account – but, as you’ll see, I think there’s more going on in cases of refusing protection than either the Exclusion Account or the Liability Contraction Account accommodates.
(2/4) Four Senses of Refusal (and another sense that I reject)
It seems to me there are at least four things that might be going on in any given case of refusal, and it might be helpful to tease them apart. Before doing so, I would start with the observation that (at least in cases like “Sibling” and “Easy Defense”) Defender faces a conflict of duties. On the one hand, Defender has a duty to intervene to protect Target. On the other hand, Defender has a duty not to harm Aggressor. One thing Target’s refusal does is to release Defender from the first duty. Target’s refusal can also do at least three other things, which bear on the justifiability of Defender breaching the second duty. Let’s take each in turn:
1. Refusal-as-Declining-Intervention-by-Defender: Suppose Defender asks Target, “Is it okay with you if I don’t intervene?” and Target replies, “Sure, that’s fine with me.” In so doing, Target waives her claim right against Defender that Defender protect Target from Aggressor. We can think of this as roughly equivalent to Target consenting to Defender’s non-intervention. In this sense, then, Target’s refusal is a waiver of a right – it’s just not a waiver of a right against Aggressor (rather, it’s a waiver of right against Defender).
2. Refusal-as-Directing-Defender-Not-to-Act-for-Target’s Well-Being: Here, we can imagine Defender saying to Target, “Should I intervene to protect you?” and Target replying, “Whatever you do, don’t intervene on my account.” This sense of refusal, I take it, is Perry’s Exclusion Account. Note that, even where this sense of refusal is in play, it might still be justified, all-things-considered, for Defender to intervene – but, in so doing, Defender cannot act for reasons grounded in Target’s well-being. (That is, those reasons cannot serve as Defender’s explanatory reasons.)
3. Refusal-as-Requesting-Defender-Not-to-Intervene: Here, Target’s response goes something like, “No, don’t intervene.” This third sense of refusal will ordinarily go hand-in-hand with the second, but note that it has a different normative effect: rather than excluding reasons from Defender’s rational horizons, refusal-as-request adds a new (first order) reason for Defender not to intervene.
4. Refusal-as-Triggering-Duty-Not-to-Intervene: In the simple, interpersonal cases, I don’t think targets have this normative power. Defender already has a duty not to harm Aggressor (on my account), and Target’s refusal does not trigger a new duty not to do so. However, when it comes to humanitarian intervention, considerations of state sovereignty arguably ground the Target state’s authority to trigger a duty of non-intervention.
If I were to translate your Liability Contraction Account into a Reasons-Based Account (which, again, strikes me as more helpful insofar as it takes into account the salience of explanatory reasons), I think it would go something like this:
5. Refusal-as-Failing-To-Extend-Exclusionary-Permission-in-Harming-Aggressor: First, let me just translate liability/forfeiture accounts into a reasons-based account at the level of self-defense: to say that Aggressor is liable to defensive force by Target is to say that Aggressor’s wrongful aggression against Target grounds an exclusionary permission for Target – such that, in defending herself against Aggressor’s wrongful attack, Target can opt to exclude (at least some) reasons grounded in Aggressor’s well-being. When it comes to defense-of-others, does this exclusionary permission extend to third-party Defenders? On your view, I think the answer is, “Yes, unless Target refuses to extend the exclusionary permission.” That is, Target’s refusal, communicated to a particular Defender, has the normative force of depriving that Defender of the exclusionary permission that would otherwise permit that Defender, in defending Target against Aggressor’s wrongful attack, to exclude (at least some) reasons grounded in Aggressor’s well-being. In this sense, Target’s refusal contracts the normative force of Aggressor’s wrongful attack, such that Aggressor is no longer liable to defensive force used by a Defender as to whom Target has refused protection.
If I’m understanding your claim correctly, I disagree. I think the normative force of Aggressor’s wrongful attack is that it grounds an exclusionary permission for any Defender, such that, in defensively harming Aggressor, any Defender has an exclusionary permission to disregard (at least some) reasons grounded in Aggressor’s well-being. That is, I don’t think Target’s refusal of protection impacts the normative force of Aggressor’s wrongful attack. So, while I think refusals of protection typically involve #1-3 above, and may involve #4 (especially in humanitarian intervention cases), I doubt that refusals of protection contract the normative force of Aggressor’s liability to defensive harm.
(3/4) On a different topic (unrelated to the normative force of refusals), I wonder how the law re: “citizen’s arrest” authority impacts your analysis of the case, “Siblings 2”? Where such laws apply, all citizens, not just law enforcement, have the legal right to enforce laws, so the fact that Pam is a police officer might not bear the weight that you put on it. Of course, as a matter of dealing with posited law, you can simply argue that citizen’s arrest laws are unjustifiable and should be eliminated. But, I think Malcolm Thornburn’s work on justification defenses presents a more difficult challenge to the kind of sharp separation you wish to draw between private actor Defenders and law enforcement officials. (See, his “Justification, Power, and Authority” (2008), where argues that “the best way to make normative sense of the power wielded by citizens… [in cases involving justification such as] self-defense, lesser-evils [and, no doubt he would include defense-of-others] is as public officials pro tempore of necessity.” It’s available here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1139030)
(4/4) Finally, a quick response to Kim‘s suggestion that there might be “hypothetical refusals.” That notion strikes me as entirely plausible, and I think it provides the most satisfying analysis to the last episode of “The Last of Us” (which was a topic of hot philosophical debate this past weekend at the Dempsey house).
Spoiler Alert: If you’ve been watching the show, or are a player of the game, you’ll know that Ellie (who is immune to Zombie bites) travels with her protector, Joel, to find the doctor who can create a vaccine based on Ellie’s immunity. (Until the last episode, it’s left unclear how Ellie would contribute to the vaccine – perhaps drawing her blood? In the last episode, while Ellie is unconscious, we learn that the doctors will have to harvest her brain, killing her.) Ellie has made it clear to Joel that she wants to “see this through, no matter what” – and Joel knew (even if he couldn’t admit to himself), that if Ellie had the chance to decide whether to sacrifice herself, she would’ve done so. Not only would she have consented to the brain harvesting surgery (thus waiving her rights-claim against the surgeon), she would’ve refused any intervention by Joel. Thus, Ellie’s case illustrates what Ferzan calls, “hypothetical refusal.” Still, Joel kills everyone (including the only doctor who can make a vaccine) and rescues Ellie (who is unconscious throughout). As we’ll see in season two, when Ellie finds out what Joel has done, she is furious – and her rage is partly grounded in precisely what Jonathan writes above: “when people justify their actions by appealing to our good, that implicates us in what they do in a distinctive way. And we care about what we are implicated in.” Joel’s saving Ellie wasn’t justified in terms either lesser-evils or liability-to-harm. (That is, everyone is now going to die the Zombie apocalypse and the people Joel killed were not wrongfully aggressing against Ellie.) In light of her “hypothetical refusal,” Ellie’s complaint against Joel is not only that he ensured the end of humanity, but that – by acting on reasons grounded in her well-being – he implicated her in this tragedy.
@Vera Hi! You mentioned that you “doubt that we have the power to deny others the right to protect us, particularly while maintaining (not waiving) our claim-right not to be harmed”. I tried to make a similar point in a couple of my posts, above, by pointing out a possible inconsistency in the putatively agent-neutral reasons that the the victim affects by exercising her normative power. (I thought, though, that the liability account might fare better, but I think you’ve convinced me otherwise!)
@Saba Hi! Excellent points — sorry I missed them earlier (just came back home from a long flight and was in a hurry to post my comments). I suppose one reason for my critique is David’s switching between two different arguments, i.e. reasons and liability. Reasons can be different for different actors, but liability is supposed to be “real.” In other words, just because “for me” you are less liable does not make you less liable indeed. So, to reduce Attacker’s liability, Victim must change her claim not w/re to Defender, but w/re to Attacker, since the liability is based on Attacker’s actions. I don’t see how requesting something from Defender can do that.
Thanks David – this is very interesting. (Apologies for the long post – there is absolutely no pressure to respond.) On the conception of narrow proportionality:
I don’t think X can be “the severity of the attacker’s threat”. Consider:
1,000 for 1. Torturer responsibly and culpably threatens to inflict 1,000 electric shocks on Victim for 8 hours. Each electric shock will cause a barely perceptible pain; 1,000 such shocks constitute torture. Defender can press a button, which will avert 1 such shock for Victim (and so Victim will still suffer 999 shocks), and it will also kill Torturer.
Surely, we want to say that the harm inflicted on Torturer is narrowly disproportionate, and that Torturer is not liable to the harm.
If X is “the severity of the attacker’s threat”, however, the harm inflicted on Torturer would be narrowly proportionate. In relation to the threat of inflicting 1,000 shocks on Victim, the harm of death is not a disproportionate harm to inflict on Torturer.
The issue here is that by killing Torturer, Defender will only partially, rather than fully, avert Torturer’s threat. This is why killing Torturer is narrowly disproportionate (even though it would be narrowly proportionate if it were to avert all 1,000 shocks for Victim).
And so, at the very least, X has to be “the severity of the part of the attacker’s threat that the defender will in fact avert” (“the severity of the harm averted”, for short).
It should be noted that Quong’s Stringency Principle seems to face the same problem. Torturer threatens to violate a very stringent right – the right not to suffer 1,000 shocks. And so, if X is “the stringency of the right threatened”, inflicting the harm in 1,000 for 1 would also be narrowly proportionate.
I haven’t had the chance to read your article in Ethics yet (I’m looking forward to doing so). I am thus unsure as to the details of your view. But it seems that if X is “the demandingness of the attacker’s duty to avert his threat”, killing Torturer is likely also narrowly proportionate. Torturer, after all, has a very demanding duty to prevent Victim from suffering *1,000 shocks*.
Now, I also don’t think X really is “the severity of the harm averted”. I can think of at least two reasons for this.
The first is that when we think more clearly about what X is compared to – the harm inflicted on the attacker – we see that the severity of the harm is not the only fact that matters. It also matters, for example, how the harm is inflicted. Suppose that Attacker responsibly and culpably sets in motion a trolley which now threatens to break Victim’s arm. In one version of the case, Defender can save Victim by pushing Attacker from a footbridge, causing Attacker to suffer a harm H1. In another version, Defender can save Victim by diverting the trolley, which will then cause Attacker to suffer a harm H2. It seems to me that the narrow proportionality threshold for H1 is lower than that for H2. For example, while it may be narrowly proportionate to kill Attacker in the second version of the case, it may not be in the first version. (I will confess that my intuition is not entirely firm here.) If this is right – if the severity of the harm is not all that matters for the harm inflicted – it gives us reason to think that the severity of the harm averted is also not all that matters. (In fact, if the mode of agency matters for the infliction of a harm on the attacker, it should also matter for the aversion of a harm for the victim.)
Secondly, the culpability of the victim also affects narrow proportionality:
Enemies. Alice and Benjamin hate each other, and they both want to make the other suffer. They both thus attempt to press a button. If Alice presses the button first, this will threaten to cause Benjamin to suffer 1,000 shocks. Vice versa for Benjamin. Alice successfully presses the button first, and Defender now can prevent Benjamin from suffering all 1,000 shocks by pressing a button which inflict will X number of shocks on Alice.
It seems that, for Defender’s intervention to be narrowly proportionate, the maximum harm that she can inflict on Torturer is greater than that which she can inflict on Alice.
Button line: X must account for not only the severity of the harm averted, but also facts about, among other things, the mode of the defender’s agency for averting the harm, and the victim’s culpability (if any).
X may thus be “the reason for harming the attacker”, which takes into account these and any other morally relevant facts.
Your case in which harming the attacker produces a side-effect benefit to a bystander challenges this (I really like the case). I agree that, while this fact is a reason to harm the attacker, it should not affect narrow proportionality. My first though is this. Suppose that an attacker threatens to inflict a harm H on a victim. And Defender can avert a harm H* for Victim, and H* is either part of the harm threatened by the attacker, or the whole of it. In that case, X should be “the reason to avert H* for the victim” (which takes into account the severity of H*, the mode of agency, the victim’s culpability, and so on). This is true even if, by acting, Defender will also achieve some other good (such as preventing Bystander’s leg from being broken by a falling boulder).
But now the issue regarding the causal condition of liability looms large. Suppose that Attacker1 responsibly and culpably threatens to inflict 1,000 electric shocks on the innocent Victim. And, independently, Attacker2 responsibly and culpably threatens to inflict 1,000 shocks on the same Victim. Defender can press a button which will avert 1,000 shocks – in particular, this will avert 500 shocks threatened by Attacker1, and 500 shocks threatened by Attacker2 (and it will thus leave Victim to still suffer 1,000 shocks). And suppose that pressing the button will also inflict a harm H on Attacker1. To determine the maximum narrowly proportionate harm that Defender can inflict on Attacker1, should we count only the reason to avert the 500 shocks threatened by Attacker1, or the reason to avert the 1,000 shocks (500 of which are threatened by Attacker1, and the other 500 are threatened by Attacker2)? What if, by acting, Defender will (i) avert 500 shocks threatened by Attacker1 against Victim1, (ii) avert 500 shocks threatened by Attacker2 *against Victim2*, and (iii) inflict a harm H on Attacker1? To determine the maximum narrowly proportionate harm that Defender can inflict on Attacker1, should we count the reason to (ii)?
(I wonder what Victor would say about this; the difficulty we have in answering questions like this seems to be why Jeff rejects Victor’s view that we should abandon the causal condition of liability.)
My suspicion: among the harms averted, if a harm is directly threatened by the individual whom the defensive act will now harm, it counts for more than a harm threatened by a different individual against the same victim, which then counts for more than a harm threatened by a different individual against a different victim. The aversion of natural threats against unconnected bystanders does not count at all (for narrow proportionality). (I’m sure that the truth is much more complicated than this.)
(I should add that the complication I noted towards the end does not seem to affect my point that, on Jonathan’s view, Adam is not liable to be harmed by Dan. In that case, the reason to harm Adam is the reason to avert the threat caused by him against Teresa – there is no complication there.)
@Susanne. There’s at least three folks who have the “anemia” worry about my Contraction account, so I definitely should try to say more to that worry!
I thought Susanne’s way of expressing the anemia worry was super helpful: “The worry, it seems to me, is not (merely or primarily) the intuition that ‘surely the power must be stronger than that!’, but is, instead, a worry that there is not, after all, a contraction of liability.” There’s not really a contraction of liability because none of the upshots or significance of liability is left behind. The account says that Adam (in the Siblings case) is not liable to be harmed by Dan, and yet Adam can’t fight back if he is attacked, he isn’t owed compensation if Dan harms him, and he has no standing to complain. So in what sense should we say Adam has a right that Dan not harm him? That’s the worry.
Now there’s definitely a problem for my account here if *part of what it is* to have a right against harm H is to have a permission to fight back against the imposition of H, a claim to compensation for H, and the standing to complain about H. But I don’t think these three things are part of what it is to have a right. They may seem to be, since they’re the typical *explanatory byproducts* or upshots of a right (hence why we can use them as evidential marks of the presence of a right). But they are, I think, *defeasible* upshots. I think it is possible to have a right against H and yet lack the permission to fight back, a claim to compensation, and the standing to complain.
Let me offer a case to illustrate why I think these are defeasible “rights byproducts,” rather than part of what it is to have a right, before saying something about what is left to a right if we take that view. Here’s the case. I’m a farmer. There’s a man — call him Josh — who occasionally flies into a violent rage. His family loves him and wants him cared for, but knows he is a danger and that they need him to be out of their house. They decide that what is best is to “commit” him to a state-run institution. But he’s a great worker. No one can harvest potatoes and shuck corn like Josh. And since it’s just me on the farm and since I fancy myself a strong guy, I figure I can protect myself if Josh slips into one of his fits of rage. I offer to take him in. But his family wants reassurances that Josh will be cared for. They tell me they’ll let Josh live with me only if I promise that I won’t harm him even if he poses a threat of harm to me, and only if I waive my claim that: if Josh harms me, then he owes compensation. I make that promise and waive that right. A few nights after Josh comes to live with me, he slips into a rage and attempts to punch me in the face.
I think I clearly have a right that Josh not punch me in the face. I never consented to that. I never waived *that* right. But, nonetheless, I’ve done something to undercut some of the normal byproducts of that right. In this case, my permission to use defensive force and my claim to compensation. (The standing to complain byproduct is still left standing in this case. But I think there are other sorts of cases that illustrate the defeasibility of that byproduct as well.)
Now come back to a case like Siblings. Of course, Adam hasn’t made certain promises that undercut his permission to fight back, nor has he chosen to waive his compensatory claim. Nonetheless, there are things he has done that have had that same effect. Given facts about Adam’s responsibility and culpability for creating a situation where Dan must choose between harming Adam and letting his sister be harmed, Adam has a duty to assume the costs of Dan’s wrongful choice to attack Adam.
I think it’s very plausible to think that a person can have a duty to assume the costs of *someone else’s* attack. Here’s a case (borrowed from the paper I’m currently working on for your Zurich workshop, Susanne) that I think illustrates this. Call it Human Shield. Testifier wants to see Innocent harmed by Defender, and so he lies to Defender: he tells her that Innocent is about to kill her. Defender has excellent reason to trust Testifier’s testimony, and so she attacks Innocent, attempting to break his leg. Testifier can intervene to defend Innocent, but in only one of two ways: by breaking Defender’s leg or by shielding Innocent such that Testifier’s own leg is broken.
Defender attacks Innocent. She is the most immediate cause of the fact that there are costs that must fall on someone or other. She poses a wrongful (though non-culpable) threat. And yet I think Testifier has a duty to assume the costs of Defender’s attack. He should shield Innocent rather than redirect the costs onto Defender or let the costs fall on Innocent.
And I want to say something similar of the Siblings case. Like Defender, Dan poses a wrongful threat. And yet there is someone else who has a duty to assume the costs of that threat. It just so happens, in the Siblings case, that the target of Dan’s attack and the person who bears the most responsibility and culpability for those costs are the same person.
So that’s my case for two ideas. 1) The permission to fight back and the claim to compensation are defeasible byproducts of a right against harm. 2) Those byproducts can be defeated by a duty to assume the costs of someone else’s threat (just as they can be defeated by a promissory duty).
Now that still leaves the “What’s left?” worry. If someone has a right but has lost these byproducts, do they have a right in anything other than name only? Is there any significance to a right devoid of these upshots? I think so! As I noted in my initial reply to Kim, the right still has its reason-giving force. It still explains why a third-party could criticize Dan for harming Adam. It still explains why it is morally impermissible for Dan to harm Adam. I think that’s a big deal!
As a final thought (to an already way-too-long comment): I wonder if I should have introduced a notion of *partial liability* to better capture my idea here. There are a number of ways in which theorists have previously noted that liability can be partial. It can be partial in the sense that you are liable only to a certain amount of harm. Or it can be partial in the sense that you are liable only relative to some people and not others. My idea in this paper is that liability can be partial in yet another sense: one can enjoy some, but only some, of the usual “perks” of having a right.
Sorry for type-o’s in my post above (calling Defender “Defendant”). That’s what I get for prepping class while drafting post! 🙄
Hi Michelle, thanks for these fantastic comments. A few quick thoughts/questions.
1) You note that one thing the refusal of protection does is to waive a right, not against the attacker, but against the would-be protector: the refuser waives a right of protection against the refusee. I think that’s got to be correct, at least in cases where there is such a right to be waived. That’s not something that I discuss in my paper, but I agree it needs to be a feature of a more complete account of the moral effects of refusal. (Michelle, are you thinking that this is a generic upshot of the refusal of protection, or only of refusal that is expressed in a certain manner? There’s a more general question here, concerning the three sorts of refusal you endorse. Are you thinking these are three distinct acts of refusal, or rather that these are three generic effects of a common act of refusal?)
2) In mentioning your third sense of refusal, you note that refusal can have the effect (in addition to possibly excluding reasons *for* intervention) of adding reasons *against* intervention. This is also something agree with, and that does come up in my paper. I argue that these reasons are ones that have to do with the value of respecting the preferences, values, and commitments of others. The refusal of protection often makes a difference because it expresses and makes known these preferences, values, and commitment. Are you thinking of the added reasons as of this sort, or do you think there are different reasons that refusal contributes?
3) This non-liability approach to the ethics of defense is super interesting. I want to make sure I have my head around it properly. And one place I’m having a bit of trouble getting clear on is the relationship between the duty not to harm that is owed the attacker and the power others have to opt to exclude (at least some) reasons grounded in Aggressor’s well-being. Suppose Target has this power. She exercises this power, thereby (at least partially) severing the link between Aggressor’s well-being and what she has reason to do. I take this to mean that the fact that defense would harm aggressor is no longer a reason for Target to use force against him. But on your view, there is no rights forfeiture: Target still has a duty not to harm; Aggressor still has a right not to be harmed. That right thus retains its normal reason-giving force for Target not to harm Aggressor. So is it just that these reasons are outweighed by the reasons Target has to defend herself? Does every instance of justified defense admit of a lesser-evil justification? (This isn’t an objection. I just want to make sure I understand the ideology.)
4) What do you think are the grounds of Target’s power (and the power that third-parties have) to opt to exclude Aggressor’s well-being from contributing to her reasons not to harm him? Jonathan thinks we can exclude our *own* well-being from contributing to the reasons of others, and for reasons having largely to do with the value of autonomy and our interest in control over our own good. On your view, we (also?) have this power over *others* provided they satisfy certain conditions. Presumably it isn’t the value of autonomy and our interest in control over the effects our own well-being that explains why we have such a power, since this power would, if anything, “cut the other way”, as it goes some way towards giving others control over the effects of one’s well-being.
Hi Vera, thanks for these great challenges!
Your first worry, I think, relates to one of Kida’s worries. You say: “But I also doubt the “liability” account works the magic. Here is why: to eliminate or reduce the liability of the attacker (“A”), V has to do something that would render either A less culpable or A’s attack on V’s rights less wrongful. David chooses the latter route: he maintains that, by saying “no” to D’s protection, V eliminates the wrongfulness of A’s attack, at least as far as D is concerned. But there is a problem: unlike the “reasons,” the “liability” account is objective (agent-neutral), i.e. A’s liability indeed has to be reduced by V’s actions (not merely from V’s or D’s or even A’s perspective). And here I wonder whether it is possible to say that objectively both things happen: the attack on V is wrongful (V retains a claim-right against A’s attack) and the attack on V is not wrongful (D may not infringe on A’s rights to protect V).”
That’s a potent worry. Let me say two things in reply. First, I’m actually not opting for the latter route. In fact, I want to deny (against what I call the Refusal-as-Waiver account) that V’s refusal does anything to eliminate, or even weaken, the wrongfulness of A’s attack. On my view, V’s right not to be harmed by A is untouched by her act of refusal — and thus the wrongfulness of A’s attack and the stringency of A’s duty not to harm V is also left untouched.
My move is to actually resist the premise that “to eliminate or reduce the liability of the attacker (“A”), V has to do something that would render either A less culpable or A’s attack on V’s rights less wrongful.” I think we should reject this disjunction; I think there is a third way to eliminate liability. A’s liability to defensive harm is grounded in the threat he poses to V’s right against harm; his liability is an effect of *her* right. Now most everyone will agree that V has the power to remove that effect by giving up her right entirely. My claim is that whatever explains why V has such a general power (e.g., our interest in having control over our rights and the limits they impose on others) will also ground V’s having a more limited power to give up only *some* of the effects of her rights. A’s liability w.r.t D is one such effect. And so I think V has the power to give up this protective benefit of her right if she chooses. (If A’s liability was grounded in something other than V’s own rights, then I agree that V wouldn’t have the power to contract it unless she could somehow render A less responsible/culpable or his threat less wrongful.)
I understand that my claims here cut against common views about the conditions of liability, but I take that to be a feature, rather than a bug, of the account. As I said in my reply to Kida, I think we need to revise our theories of liability to take notice of the powers we have over the effects of our own rights.
Now to your other thoughtful and thought-provoking comments about attacker counter-defense. First off, I didn’t mean to claim that, where V has refused D’s protection, A can fight back against D *so long as* A’s initial aggression is the result of duress or mistake. I didn’t mean to suggest that durress/mistake was sufficient for the permission for A to fight back. I was only trying to make the weaker claim that it is not *always* the case that A would be forbidden from counter-defense, and I was trying to use cases involving minimally-responsible attackers to make the point. Let me try to clarify the view.
On my view, V’s refusal of D’s protections makes it so that A is not liable to be harmed by D. That is, A has a right not to be harmed by D. And yet I think that if D does attack A, A may nonetheless lack a permission to fight back. Not because he lacks a right against D’s attack, but rather because the defensive permission his rights would otherwise grant him is undercut by his duty to not redirect the costs of D’s wrongful attack. (The Human Shield case I gave in my reply to Susanne illustrates the possibility of X’s having a duty to not redirect onto Y the costs of Y’s wrongful attack.)
So the central question, then, is this. When does someone have a duty to not redirect the costs of someone else’s wrongful attack back onto that person? I think considerations of comparative responsibility and culpability are relevant here. Start with a case where we scale the attacker’s responsibility and culpability way up and the defender’s responsibility way down. Suppose A really wants to harm both V and D. He knows that D cares about V and will defend her regardless of whether she refuses protection. And so he attacks V in order to provoke D, giving him the opportunity to use counter-defensive force against D. Here it seems clear to me that A has a duty to assume the costs of D’s engagement *even if* that engagement is wrongful in virtue of V’s refusal of protection. And this for reasons of A and D’s comparative responsibility and culpability for the costs of D’s wrongful attack.
Now compare with a case where we scale the aggressor’s responsibility and culpability way down and the defender’s way up. In the paper I used a case involving duress. For reasons you mention, that sort of case might not be the best example of a minimally-responsible attacker. Let’s instead take a case of someone who doesn’t intend harm at all. A, an otherwise-upstanding person, is hungry. In his haste to get home for dinner, he drives just a wee bit faster than his duties of care require, and loses control of his car, threatening V. (I’m assuming this is enough to make V liable to some defensive harm.) V refuses D’s protection. D knows that this makes it wrong for him to harm A, but he’s been long looking for an excuse to harm A (who was selected ahead of him for a promotion at work). And so D attacks A. Here it seems clear to me that A does not have a duty to assume the costs of D’s wrongful attack.
So the only point I wanted to make in the relevant section of the paper was this. It isn’t generally true that A is permitted to fight back against D, but neither is it generally true that A is forbidden from fighting back. Sometimes A can and sometimes he can’t. Whether he does or not depends if he has a duty to assume the costs of D’s wrongdoing. And whether A has such a duty depends, in part, on considerations of *comparative* responsibility and culpability for the costs of D’s wrongdoing.
Now of course there’s a great deal more to be said here. Some unfinished business (just to mention a few): (i) What other factors — apart from comparative responsibility and culpability — are relevant to determining whether A has a duty to assume the costs of someone else’s wrongdoing? (ii) How great must be the responsibility/culpability gap between A and D for A to have (lack) a duty of cost assumption? (iii) What about when responsibility and culpability come apart?
@David Dear David, thanks for your reply. I would like to read it more carefully and respond to it, but I am now running out of the house for my 3:55 seminar (Eastern time). Will try to respond after that if the discussion is still open,
@David The anonymous above is Vera. )
@David Hi David, thanks for your thought provoking response. Here are my quick thoughts to the points you make:
1. I understand of course (and completely agree) that, by refusing D’s protection, V does not waive any rights against A. I simply disagree that V’s refusal of protection is effective (i) to make D’s protection wrongful (it is certainly unauthorized, but even unauthorized act may be justified on other grounds); and more importantly, (ii) to eliminate A’s liability to D.
You say: “My claim is that whatever explains why V has such a general power (e.g., our interest in having control over our rights and the limits they impose on others) will also ground V’s having a more limited power to give up only *some* of the effects of her rights. A’s liability w.r.t D is one such effect. And so I think V has the power to give up this protective benefit of her right if she chooses.”
I understand the argument, but you still need to persuade me that A’s liability w.r.t D is an effect of V’s right not to be harmed by A. Or put differently, are you sure that we have the right that other people protect us? Because if we don’t, we cannot take it away. And I don’t think we have this right; otherwise, V would have a claim against any bystander who chose not to defend her against A.
2. Comparative liability and other questions: I think (and to the extent you argue the same, agree) that in any conflict, it matters what interests are at stake (e.g. life v. property); how much damage is threatened (a pinch v. death); what is the probability of harm; and how culpable each party is. (sorry for self-promotion, but I am only citing this piece of mine as a short-cut in case you find it helpful to your argument, “Self-Defense and Risks,” in: THE ETHICS OF SELF-DEFENSE (Oxford University Press, 2016).
Since we disagree on whether A is liable to be harmed by D despite V’s refusal of protection (I believe he is), we answer the question about A’s right to use force against D differently. Under my account, D’s use of force against A would be limited only by proportionality and not A’s culpability (i.e. no comparative culpability here). And therefore, I would argue that, when harm to V is imminent, A’s culpability or lack thereof makes no difference for permissibility of A’s fighting back (it certainly makes a difference though for the degree of A’s responsibility for any ensuing harm or attempted harm): except for the constraints of proportionality (magnitude of the affected right; extent of threatened harm; and sometimes probability of harm)–i.e. D using disproportionate force–A never has that right. If an innocent toddler A is about to shoot V to death, D has the right to shoot the toddler to prevent him from killing V, and if the “precocious toddler” (sorry, couldn’t resist 🙂 ) realized the danger he was in and shot D first, the toddler would be in the wrong. The same applies to any aggressor, culpable, not-very-culpable, or innocent. But this is the point where we disagree because we disagree on A’s liability to D.
Thanks for your excellent paper and the fascinating discussion, they gave me a lot to think about! Hope to continue our conversation some other time.