Jeremy Williams: Paper as provided for discussion

The paper by Jeremy Williams on anti-abortion violence was provided, July 26, 2021, for blog discussion by fellows.



Responses to Romanell Center readers of ‘On Anti-Abortion Violence’

Jeremy Williams, University of Birmingham


I’m exceptionally grateful to the participants on the Romanell Center blog for their many fascinating and challenging comments about my paper, ‘On Anti-Abortion Violence’. I have attempted, while keeping my own remarks to a reasonable length, to respond below to as many of these comments as possible (barring some of the participants’ asides to each other – entertaining, if somewhat baffling to an outsider, as they are). But there’s much, inevitably, that I have not managed to address, and plenty that I will be continuing to ponder for quite some time.[1] Hopefully, however, there will be opportunities for other conversations in future on these and related issues.



Let me begin with a couple of preliminary matters: my use of the term ‘Restrictivism’, in preference to the more common moniker ‘pro-life’, and the question of who counts as a Restrictivist, on my understanding. Phil Reed writes that the renaming of the pro-life position as “restrictivism” is distracting and not well-justified. The move assumes his view is correct (they can’t be “pro-life” if they’re committed to AAV!). If we regularly renamed our opponent’s position on the basis of what we argued they were logically committed to (even as they explicitly disavow being committed to that thing), this would needlessly muddy the philosophical discourse.

David Hershenov (Aug. 2nd, 4.04 pm) endorses these complaints. But I confess I found their reactions surprising. For my own observation is that philosophers quite often name and rename the positions occupied by others to suit their own purposes – in the name of clarity or consistency when constructing a schema of available positions, for instance, or (crucially in this context) to avoid distracting or misleading implicit associations of the original label. These practices generally seem to be regarded as unexceptional, and I see no particular reason why the label at issue here, uniquely, should be treated with additional reverence.

Perhaps what Phil and David object to primarily is not my failure to use the term ‘pro-life’ per se, but only to the reasons I gave for not doing so. I said, in the paper, that ‘if my argument succeeds’, the pro-life label ‘is highly inapposite’. Phil complains, in the quote above, that my move here assumes my conclusions to be correct. And David adds, similarly:

I agree with the (small) point on naming. It is also very tendentious as it assumes his view is correct. He should, AT MOST, have said if his view is persuasive, then the pro life view is really best renamed Restrictivist.

Well, I did say ‘if’, as you can see. But I also think that, whether the argument of the paper succeeds or not, the mere fact that its aim is to investigate whether the ‘pro-life’ view is committed to violence is sufficient to justify refusing to make use of the disputed term in the article itself. For while David and Phil both insist that it is a misreading to suppose that the ‘pro-life’ label refers to any general pro-attitude towards life or peace, as opposed to a narrow commitment to protecting (specifically) fetal lives against (specifically) abortion, I think it is clear that this is not how the term sounds, in many people’s ears, nor how it is uniformly used, including by those who associate themselves with it. If I understand rightly, when the label ‘pro-life’ came to prominence in the 1960s and ‘70s, it was indeed often employed by activists who were, in addition to being opposed to abortion, also anti-war, and so on. The term continues to be used by some in that way today. And given those broader associations, and the fact that they are often intentionally courted, making use of the label ‘pro-life’, in the context of this particular investigation, would have been an unnecessary risk to clarity of expression and thinking. I agree that I could have said that explicitly. But I still think the convention I proposed was right.

Incidentally, while we are on this issue, I cannot agree with Phil (Jul. 30th, 1.46pm) that those who make complaints of the form ‘If you were really pro-life you would accept policy X’ are necessarily engaged in merely fallacious reasoning. Rather than attributing inconsistency to their opponents, what these people are often doing is arguing that self-identified ‘pro-lifers’ cleave to an impoverished or ersatz interpretation of the value in relation to which they have chosen to define themselves. The complaint, in other words, is that their opponents are not ‘pro-lifers’ worthy of the name. Opponents of abortion themselves, note, make this charge against each other. (And so I’m unconvinced that Phil is correct, either, when he claims that this ‘error’, as he sees it, is one that opponents of abortion themselves tend not to make.) To give but one example, I recall one supporter of Texas-style heartbeat laws, who also wants to see greater efforts to combat maternal mortality in Texas, writing in the NYT that ‘[i]t is not pro-life to save the child, only to lose her mother in the process.’ It would, I think, be at best a feeble response to this writer’s exhortation to say that, strictly speaking, the ‘pro-life’ label only signals opposition to abortion, and not a commitment to saving lives more generally. For what she is deploring is what she sees as others’ misplaced or arbitrary priorities. And I say the response is ‘at best’ feeble because the author’s words also show the response to be false: she clearly understands the term ‘pro-life’ to convey a set of commitments broader than just opposition to abortion, and wants us to do likewise. In short, then, ‘pro-life’ is a contested term. We have ample reason to avoid it, and its baggage, for present philosophical purposes.


Who counts as a Restrictivist?

Let me next address David’s questions about who exactly falls within the scope of Restrictivism, and the paper’s critique of that position.

David asks why I refer to Christopher Kaczor, for instance, as a Restrictivist, in spite of certain of his claims, which my paper cites, to the effect that the death of a fetus is, in certain respects, less harmful than the death of an older child or adult. And David asks for clarification on what ‘evils’ refers to, in the early account which the paper offers of what Restrictivism is. For how we understand that term ‘evils’ makes a difference, as David notes, to who ends up being a Restrictivist, including David himself. I say in the paper that Restrictivists take fetuses to be persons in three ways, of which the third is that fetal deaths ‘are ordinarily equal (or at least comparable) evils to ours in what Derek Parfit (2011, 38) calls the ‘reason-implying sense’.’ In response, David writes:

Perhaps Williams means by “evils” in condition #3 something more akin to “wrongs” than “harms.” Kaczor certainly believes the harms aren’t as great. If evils mean something like harms, then I, too, am not a Restrictivist.

I agree that it would have been wise to explicitly clarify this, but I intended ‘evil’ in the paper to read as ‘disvaluable’. It did not mean ‘wrong’, though we need to know how great an evil death is in order to know when bringing it about is wrong, of course. The evil of death could, but need not, be cashed out in terms of its harmfulness. (And, attempting now to retrace my mental steps, I assume I opted against referring to the ‘badness of death’ because that phrase tends to be read as referring to death’s harmfulness, even if, strictly, it need not be.) We need to weigh and compare the evil or disvalue of death, for various victims, for a variety of moral purposes. And while it is frequently thought that we should cash it out in terms of harm, or the frustration of interests, there are alternatives – one might instead weigh lives and deaths in some other currency, like the victims’ ‘worth’, for the relevant purposes. One could, thus, be a Restrictivist, on the understanding of the paper, while thinking that a fetus is not significantly harmed by its death, as long as one does not weigh its life in the currency of harm when determining how it is to be treated, and the degree of priority attaching to its life. Hence the reference, in the wording I quoted a paragraph ago, to death’s being an evil in the reason-implying sense.

Is Kaczor, then, a Restrictivist? I shan’t try to reconstruct his full position on fetal harm and moral status, not all of which I find easy to interpret or reconcile. In the passage, however, which the paper cites, he does seem to accept that fetuses lose less from death than older children or adults, due to their not being invested in their lives and relationships in the same way that the rest of us usually are. Whether that is consistent with his being a Restrictivist depends, then, on how great the disparity is, and whether, if it is great, that fact matters when we are comparing outcomes for practical purposes. Although he invokes the equality of worth of the fetus as driving the conclusion that abortion is wrong, Kaczor does also seem to think that it is the harm of death to the fetus that determines what we may do with it in at least some cases. He believes, for instance, that whether a woman is permitted to have a hysterectomy that ends the life of a fetus as a side-effect is to be determined by the doctrine of double effect, and has indicated that the proportionality condition of the doctrine weighs the good to the woman against the harm to the fetus (see his The Ethics of Abortion, 1st Edition, p. 162). His view, moreover, if I’m interpreting him rightly, is that, for the proportionality condition to be satisfied, the life of the woman must be at stake (as when, e.g., she needs the hysterectomy to remove a potentially fatal cancerous tumour). And that suggests in turn that he thinks the harm of death to a fetus is indeed comparable to that of a pregnant woman, notwithstanding the fetus’s lack of investment in its life, relationships, etc. On the basis of these claims, at least, Kaczor would indeed remain a Restrictivist in my sense.

Is David a Restrictivist? He argues not, at some length. But even if that is right, I take it that this question is orthogonal to the disagreements between our respective papers on AAV, since David and Phil’s work claims to take up cudgels on behalf of the ‘mainstream pro-life view’ (at p. 415 of the published version), and does not — unless I have missed it — depend upon the particular account of prenatal moral status which David describes in his blog comments. In any case, however, despite his explanation, I don’t think I understand in what sense David thinks his position is non-Restrictivist.

David says that he is prepared to grant that a revised time-relative interest account applies to embryos and fetuses, though he thinks the harm incurred through death by an embryo, even absent psychological connectedness, is still great. He also thinks abortion is generally wrong, notwithstanding the burdens of pregnancy, birth, and unwanted parenthood (the only exception to this acknowledged, in the co-authored paper with Phil, is for cases of life-threatening pregnancy). Indeed, if I understand the lesson of his rooftop case rightly, David thinks that even in the absence of any special relationship, a person must accept serious bodily trauma, of the kind one might incur in, e.g., falling off a building, as a cost of rescuing an embryo or fetus (and not only, then, as a cost of abstaining from killing them). But this suggests either that the harm of death to the embryo must be comparable to the harm of death to an older person after all, with the absence of psychological connectedness making little difference, or that it is not harm but something like equal worth that does the work when we are specifying the parameters of the duty to save. For otherwise, David’s view seems to require an implausible amount of self-abnegation from rescuers. (How much weaker can the interests and value of a being be, consistently with our being under a duty to incur the costs of falling off a roof for their sake?) Either of those commitments, however, would make David a Restrictivist in my sense.

That David is a Restrictivist after all seems to be confirmed by the paper, co-authored with Rose Hershenov, to which he directs me for further elaboration of his position – ‘Health, Moral Status, and a Minimal Speciesism’. For there, the authors say that ‘the fetus and newborn can suffer great harms… and these harms are nearly as great [my emphasis] as those you and I can suffer’. Again, then, if David does not fit the mould of a Restrictivist as I defined that position, I’m unclear as to why. Could it be that there has been a misreading of the account of Restrictivism that I provided? In his comment on Aug. 2nd, 3.47 pm, David writes that ‘Williams’s condition #3 states that their [i.e. fetuses’] deaths are equal evils to ours’. That is to omit the parenthetical qualification ‘or at least comparable’.


My paper’s thesis

With those preliminaries out of the way, let me next say something about the interpretation of the thesis of my paper in a number of the blog comments, and contrast it with the positions of others in the blog.

Steve Kershnar opens the discussion with a summary of his own arguments, developed in his book, for the conclusion that non-pacifist Restrictivists are committed to anti-abortion violence (AAV). His position, in both his book and his comments here, differs from mine (he does not say otherwise, of course). Steve seems to want to cast the problem for Restrictivism entirely in terms of the conditions of forfeiture by doctors of the right not to be killed, or, put in what I shall treat here as equivalent terms, their liability to be killed. This is a mistake, I think, both because doctors are not the only potential targets, or victims, of defensive force about whom we need to worry, and because forfeiture is not the only potentially relevant justification for harming them.

In his book, Steve argues at some length that ‘forfeiture theory best explains just killing’ (p. 67). I find this puzzling. If ‘just killing’ means killing without transgressing any existing or operative rights, then ‘forfeiture theory’ does not explain or justify just killing in any general sense (for it does not, e.g., explain the justness of voluntary euthanasia). And where defence specifically is concerned, just killing is not all that we want to explain, since we will also want to explain our sense that rights not to be killed or harmed for defensive purposes can sometimes be justifiably overridden even where they exist. Steve is not unaware of these other candidate justifications for harming. But he nonetheless treats ‘just killing’ as interchangeable with permissible or justified killing, and claims that ‘forfeiture theory’ is the ‘best’ theory not only of just defensive killing but of ‘permissible killing’ (p. 67), and of ‘when individual and state violence is justified’ (ibid.). There is, however, no clear sense in which an account of the conditions of forfeiture would represent the ‘best’ theory of permissible killing, or harming, tout court, as opposed to an (admittedly important) element of a full theory. And we seem certain to end up with a defective account of forfeiture if we formulate it with the aim of making sure it justifies all the cases of killing which we take to be permissible.

Steve’s argument is, I think, caught between two stools. On the one hand, he wants to develop an account of the conditions for forfeiting rights against defensive harm, and apply it to the case of abortion doctors. But the details of the account are contentious, and in my view sometimes dubious. (I could not, for instance, quite understand Steve’s reasons for thinking that innocent aggressors forfeit rights against defence, at p. 71, nor his reasons for rejecting the idea of agent-relative prerogatives to employ defensive force against the innocent, at pp. 137-8.)

And when Steve attempts in his book to accommodate the diversity of positions on the justifiability of defensive force, he runs into trouble. He says, for instance, that even if one thinks that rights against defensive harm are forfeited only when the aggressor is culpable or blameworthy for an unjust threat (as opposed to morally innocent, or responsible without culpability), it still follows that an abortionist may be harmed in defence of a fetus, notwithstanding any excuses of which they may be able to avail themselves, because ‘abortion doctors are more blameworthy than fetuses or their defenders on any plausible pro-life position’ (p. 91). The claim appears to be that, because fetuses are wholly innocent, abortion doctors must be culpable by comparison. But this is not right: blame does not have to go somewhere, and need not attach to anyone.

In criticising Restrictivism in my paper, on the other hand, my aim was to be more ecumenical about the conditions of justifiable defensive force. My strategy was to try to show that the justifiability of AAV, in one form or another, is the outcome on which Restrictivists will converge if they do not endorse, as I put it, ‘an especially strict and unattractive form of pacifism’. To be sure, the precise nature and extent of the violence justified will turn somewhat on the account of defensive force to which a Restrictivist subscribes. But the only Restrictivist who entirely avoids becoming committed to violence, I argued, is one who is unacceptably resistant to action in cases where it is intuitively required - in particular in the case I called Death Camps.

I cannot, then, agree with Phil’s suggestion (30th Jul., 1.46 pm), which Steve appears to endorse, that it is a problem that the paper did not proceed on the basis of one or other ‘theory of defensive violence’. That the paper does not depend on some particular theory is part of the point. Nor is it the case that, as Phil claims, I acknowledge that my argument throughout the paper depends on such a theory, which I do not go on to provide. Phil’s quotation from the paper to the effect that ‘much may depend on complex interactions between our account of moral status and theory of defensive harm’ is out of context, since there I am not referring to my case against Restrictivism, but to future inquiries into the compatibility of various non-Restrictivist conceptions of prenatal moral status with what I call the Non-Violence Constraint (NVC). I do think it is likely that, as these inquiries progress, they will require us to engage in increasingly complex parallel investigations in defensive ethics. Indeed, I think that, more generally, defensive ethics and the study of moral status ought to be brought into closer contact, for the benefit of both fields, and I hope to contribute to that work. But that’s not the same as conceding, in this paper, that the case against Restrictivism can only be prosecuted with some full theory in hand.

Phil writes that ‘[t]here is no formal inconsistency in saying that one can intervene to save children in death camps and one ought not to commit AAV’, and I agree. The question is what one can do to drive a wedge between the two. I can’t see any viable way for the Restrictivist to do so, and I’m not sure why one needs a theory to show that this is correct (as opposed to envisaging and rejecting candidate proposals, as I try to do). And let me add that I’m unsure what Phil thinks constitutes a theory of the relevant matters, since it is hard to see in what sense his own co-authored paper presented such a theory if mine did not.

Finally, let me address a misinterpretation about the conclusion of my paper. I say there that it is much harder than has been previously anticipated to provide a conception of prenatal moral status that avoids violence, without resorting to (a particularly hard to swallow version of) pacifism, and while meeting certain other common-sense conditions. I hope that this can be done. And, as Phil anticipates, I hope to have a stab at it myself elsewhere. But there is no abandonment at the end of the paper, of the kind Phil observes (Jul. 30th, 1.47 pm), of the reflective equilibrium method. The further work I outline should be conducted, I think, by means of that method. It is conceivable, as the paper notes, that, in doing so, we will have to revisit conclusions reached in this paper, unsettling as I find that prospect. I also say that there are certain outcomes of this further work that would be so unwelcome that, if they turned out to be unavoidable, would justifiably shake one’s confidence in moral realism. But that is not to endorse, as Phil puts it, ‘moral nihilism (or anti-realism or whatever)’, precisely because that crunch point has not yet been reached.


The assimilation argument

Much of the criticism directed at my paper over the course of the discussion concerns my treatment of what I called ‘the assimilation argument’. That argument holds that AAV (at least of a serious kind, that violates the NVC) is unjustified because abortion, even when it involves killing the fetus, is to be treated on a par with letting the fetus die, where a threat of wrongful letting die (so the assimilation argument invites us to accept) does not merit a serious violent response. I originally envisaged this argument being deployed by a Thomsonian defender of abortion, though, as I then found out, David and Phil have mounted a related defence of Restrictivism. So it’s no surprise, then, that this would be where they in particular would be most interested in pressing. I think, for what it’s worth, that the assimilation argument is even less suited to being applied in service of Restrictivism than it is in the context I originally described. More on that later.

My paper’s response to the assimilation argument first appeals to intuition, in the form of what I call the Dismemberment case – a variant on Thomson’s violinist example, in which ‘it is sufficient to save the violinist that one be hooked up to him for a short time, during which attachment causes no significant ill-effects, while detaching early would require dismembering.’ David (Aug. 2nd, 3.54pm) objects to that case, on grounds that I don’t quite understand. Inspired by Kamm’s methodology, he says that, to hold other things truly equal, in comparing the significance of killing vs letting die, we should stipulate that, if the violinist is allowed to die rather than being killed, his fate would be comparably gruesome to dismemberment. But insofar as that makes a difference, it is to elicit the intuition that a threat to impermissibly disconnect the violinist, as well as to kill him, may be met with serious force, which is not the result David is looking for.

As well as being unintuitive as applied to cases like Dismemberment, the assimilation thesis seems to me, let me add, unintuitive when held up to examination as a general principle. It tells us that dependants are at the mercy of their carers in an unappealing way: that dependants, and those who act for them, lack meaningful recourse when the former are subjected to unjust threats of killing, by wrongdoers who seek to avoid bearing costs no greater than those which they are required to bear for the dependants’ sake. It seems objectionable for dependant people to be so comprehensively unprotected. Curtailing dependent victims’ defensive options to that extent is not necessary as a moral acknowledgement of the burdens of helping, or the moral status of the aid provider as more than an object of use. For providers of aid may still discontinue aid when the cost becomes too high, and, when it does, discontinuation is a rights-protected option. Moreover, my argument grants what Thomsonians centrally contend – that, when the costs of aiding are sufficiently high, it can be permissible to kill as well as to withdraw aid in a way that does not involve attacking the dependent. But the cases of concern are not those cases. And it is unclear why aid-givers ought, as a general matter, to receive yet a further indulgence, by having not only the option to withdraw aid, including by killing, when the costs are too high but also thorough insulation against the normal repercussions when they propose to kill to avoid costs which are not too high – costs which they owe it to their victims to bear rather than resort to a killing, yet which they seek to avoid nonetheless.

That point about costs brings me to an objection to my argument with which David, Phil and Steven all seem to be in sympathy. I say in the paper that the assimilation argument ‘greatly exaggerates… the degree of moral equivalency between killing and letting die that it is plausible to assert, in relation to the conduct of both the woman obtaining an abortion and her doctor.’ Where harm to the woman is concerned, David and Phil both argue (and Steve agrees) that my argument relies on too speculative a set of judgements about the extent to which several attributes of the woman’s act, distinguished in section 5.1, together affect the degree of force that can be brought to bear on her. Phil writes:

Williams says that the assimilation view only offers an act of abortion as intermediate between a paradigmatic killing and a failure to save. But then, without offering a theory of defensive violence or principles of defensive violence, he says this intermediate position will allow non-lethal harms incompatible with the NVC. This all seems very messy.

Steve says he finds Phil’s point ‘excellent’. And David, similarly, says that, when I argue that a woman who wrongfully aborts a fetus with the moral status of a person would be justifiably subject to serious defensive harm, I am misleadingly presenting a merely impressionistic judgement as having ‘pseudo mathematical rigor’. How do I know, he asks, that, when the various moral considerations which the paper highlights as relevant to the cases at hand are properly accounted for, the amount of defensive harm justified will exceed that allowed by the NVC?

The characteristics of the woman’s act which I identified as relevant are (to quote the paper, while altering the formatting and punctuation) these:

(1) the fact that, in being killed, the fetus loses only what it could have had with the woman’s efforts;

(2) the fact that, to avoid defensive harm, the woman would have to endure the considerable burden of completing an unwanted pregnancy;

(3) the fact that the woman crosses the boundary between separate persons, in directly killing the fetus, thereby becoming the original cause of its death;

(4) the fact that she has an opportunity to avoid defensive harm by incurring only costs that she owes it to the fetus to bear;

(5) the fact that the costs with which she unjustly threatens the fetus are much greater than those she seeks to avoid.


I conclude the discussion of these considerations by saying the following:

If we tally up these considerations without giving arbitrary preeminence to any, what again seems to emerge is, as suggested above, the view that the harm that may be imposed, in the ex ante enforcement of the woman’s duty not to abort, is intermediate between the levels of defensive harm that are justifiable in paradigmatic cases of killing and omitting to save respectively. That implies that the woman may not be killed, but may be subject to non-lethal harms incompatible with the NVC.


Looking back at the paper, I am also dissatisfied with what I say at around this point. For a start, the conclusion here suggests a firmer endorsement of what we might call the

intermediate harm thesis than I am happy with on balance. I am somewhat ambivalent about it. I can still see its facial appeal when one focuses on the Kammian observation that the relevant killings bear a key hallmark of letting die, and in that respect seem to fall midway between the two. But against this is the point that I made above: that the fact that the victim relies for life on the benefactor is already reflected in the cost threshold on saving, and that to deprive the victim of meaningful defence even in cases where it is agreed that the costs of saving are not excessive seems to be to consign dependant people to live by the grace or whim of their aid givers, rather than by right. So I am at least open to the view that the fact that the dependent loses only what the killer provides them with is not as morally significant, on the side of blunting an intervenor’s defensive permissions, as the intermediate harm thesis indicates, and perhaps not significant at all. (In fact, I am also at least somewhat tempted by the view that wrongful, intentional letting die justifiably incurs comparable harm to wrongful killing, which is why – to answer David’s query at August 2nd, 3.55pm – I say, at one point, that I am unsure that Jones, in the revised bathtub case, ought to be harmed less severely than Smith).

A little earlier in the same section as the conclusion quoted above, I am more circumspect about the intermediate harm thesis, saying, ‘The most we are justified in concluding, it seems, is that killing to discontinue aid has a moral status intermediate between those of paradigmatic killings of independent victims on the one hand, and omissions to save by withholding one’s effort and resources on the other.’ That is, I think, how my position is better put. The intermediate harm thesis is more a strategic concession, for the sake of further argument, than a conclusion to which I would want to firmly pin my colours. It is concession because, as the paper argues, and as I have suggested above, there is no clear basis for thinking that the usual significance of killing is substantially counteracted by consideration (1), above, for defensive purposes, and reason, indeed, for skepticism about that claim. Since the intermediate harm thesis is concessive, it is not, I think, exposed to the objection that it pretends to ‘pseudo mathematical rigour’. The point is not: ‘This is precisely how much harm I have calculated is justified in the relevant cases.’ Instead, the point is: ‘Even if we meet proponents of the assimilation argument half way, by accepting that the harm justified in response to killing a dependent is intermediate between the harms justified in response to paradigmatic killings and omissions to save respectively, that’s not enough to save their position’.

The ‘concessive’ reading of my argument is stated at a point where only (1) and (3) are as yet on the table. I then introduce (2), (4) and (5). And I can see, again, why having done so, the eventual conclusion of the section, which reaffirms the intermediate harm thesis, might look ‘messy’. But of those additional considerations, it seems to me that (2) is cancelled out by (4). Or, perhaps more precisely: (4) shows that (2) is not really what matters. The fact, that is, that the costs of not aborting are substantial does not seem relevant, given that the costs are ex hypothesi the mandatory alternative to killing. The benefactor, in the cases of interest, has sufficient opportunity to avoid killing – not, to be sure, according to her own preferences, but in the relevant, moral sense. It is hard to see how a wrongful attacker can make a claim for leniency, in the use of defensive force, on grounds that their only alternative to killing was unacceptable to them, when the costs have already been taken into account, morally speaking, and found insufficient to release the attacker from the duty not to resort to killing in the first place.

In sum, then, I agree that my original presentation of my response to the assimilation thesis, in 5.1, was inexact. But I think it can be clarified, and that it does not involve an implausible pretence at pseudo mathematical rigour. Unjust, intentional lethal threats can generally be met with serious defensive force – killing, in many cases. Is a person who threatens to kill a dependant insulated against such force, on grounds of (1), above? Case-based intuition, and reflection on the principle itself, make it hard to see why, I have argued. It’s true, as per (2), that, in wrongful abortion cases, abstaining from killing the dependant will also be burdensome to the benefactor. But as I’ve suggested here (and as the paper comes frustratingly close to saying, at pp. 290-1), what matters is that the cost of avoiding killing is a mandatory cost under the circumstances, as per (4). Since we can ignore (2), we’re thus not in a position of trying to calculate, in what David calls a pseudo mathematical fashion, and as my own language of tallying suggested, the value of multiple considerations on both sides of the moral equation. The real question, I think, is what (1) counts for, relative to (3), (4) and (5). Because I cannot find strong evidence that (1) matters, I think acceptance of the intermediate harm thesis is charitable, yet still not enough to fend off the threat of AAV, for those who accept the personhood of the fetus.


Harming to prevent failures to save

There’s another respect in which the argument of 5.1 is charitable. For, as I go on to argue in the paper, I think we should reject the foundational premise of the assimilation argument that wrongful letting die can, as a sufficiently general matter, be met with only an insubstantial level of force. I make an admittedly brief case to that effect in the paper’s section 5.3. In ‘How not to Defend the Unborn’, meanwhile, David and Phil, as far as I can tell, bypass the question of whether wrongful letting die can attract substantial defensive harm, and therefore how much harm, short of death, they would countenance in response to unjust abortion. David says, in his comment of Aug. 2nd, 3.52pm, that the aim of his and Philip’s paper was only to rebut the specific charge that Restrictivists are committed to defensive killing. But as their paper rightly begins by acknowledging, it is the spectre of AAV, not only of lethal violence, that threatens to undermine Restrictivism. Having acknowledged that concern, however, they pull back from squarely engaging with it. I’m still unsure exactly what David and Phil’s position on non-lethal violence is. But since a number of their comments are aimed at overcoming the claim that Restrictivism licenses serious but non-lethal violence, I will assume that that is what they hope to show.

Why would someone think that wrongful failures to save, generally speaking, do not attract significant defensive force? It’s worth emphasising that the cases of interest involve intentionally letting die. So we shouldn’t appeal to our judgements about, for instance, what may be done to those who fail to give enough to Oxfam. As David and Phil themselves point out in their own paper (at n12, pp. 427-8),


abortion is standardly worse than typical cases of letting individuals die in a far-off land from, say, malnutrition, because in such cases the non-saver … hopes someone else will save them. … [T]hose aborting do not want the child saved. As Patrick Lee observes “In most abortions, if by some miracle the baby did survive and the attendants at the clinic brought the baby into the mother, all of those involved in choosing the abortion would no doubt protest that the abortion clinic had not done their [sic] job”


It is for that reason that, in my paper, I turn to a revised version of James Rachels’s oft-cited bathtub case. In my revised version of the case, Jones watches a child drown in the bath to avoid the burden of having to look after him (in the original, Jones acts to obtain the child’s inheritance, which exacerbates his wrongdoing, by rendering it opportunistic). Intuitively, I continue to find, Jones may be subject to substantial defensive force – force, that is, in excess of what the NVC allows - if this will facilitate the rescue of the child.

David objects (Aug. 2nd, 3.55 pm) to my use of the bathtub case. Part of his problem with it is based on necessity — ‘adoption is an option’ — though that is a feature of the case that David has added, not a feature that is essential, and not one that it seems to make any intuitive difference explicitly to remove. David also objects that the burdens, for Jones, of caring for the child, come ‘nowhere close’ to those involved in pregnancy. But that depends on what we imagine Jones’s child to require – the child might have a high level of need. And I’ve already cast doubt on the thought that would-be wrongdoers ought to be met with lesser force just on grounds that doing their duty would be costly to them.

When David invokes the costs of saving here, however, what may lie behind his comment is a concern I have not yet discussed: a wrongful non-rescuer’s excuses, and particularly the availability of the excuse of duress. The availability of this excuse is connected to the cost of rescue, but only contingently: it is a matter of one’s firmness in the face of the costs. That duress is what David has in mind here is suggested by his description of what a more sympathetic Jones would, in his view, look like: exhausted through work on the child’s behalf. (David also mentions a water-phobic Jones, who would have an excuse on grounds of diminished responsibility). It is at least plausible that an exhausted Jones would indeed be liable to only a lesser harm than one who is not similarly excused (though not everyone accepts that liability is sensitive, in this way, to a wrongdoer’s degree of culpability). But I doubt that duress, in this form, makes enough of a difference to serve’s David’s purposes. If I imagine myself in Jones’s position, as finessed by David, I still think that a third party would be permitted to harm me pretty substantially to save the child’s life. And I don’t need to rely on a disputable view about the conditions of liability here, since, importantly, even if one were of the view that I would not be in any way liable to harm in this case, owing to my excuse, it does not seem an at all excessive cost to impose on me, say, a broken limb as a matter of the lesser evil.

In the place of the case of Jones, David proposes the case of ‘someone on a deserted isle breastfeeding, changing, and co-sleeping with a child that one never wanted nor had a chance to give it away.’ David writes, ‘It isn’t clear how much harm we can impose on this woman. Perhaps far less than the Non-Violence Constraint (NVC) of killing, burning and bone breaking.’ It is also unclear, however, at least to me, that the woman is under the relevant duties, or at what cost, intimate and burdensome as the needed activities are. In addition, David does not specify how, in this case, the baby comes to die. But if the woman merely walks away, and makes camp on the other side of the island, she need not intend the death, even if she foresees it with certainty. Yet if, on the other hand, she does what clearly constitutes an intentional letting die – watching, say, while the baby drowns in the bath, for surety that its crying will cease – we’re back to Jones. David’s case is not, thus, an improvement on mine.

Similar remarks apply to David’s rooftop case, a version of which he and Phil employ in their co-authored paper. The version in their paper, as I noted in mine, was unsatisfactory, as it was worded in a way that was suggestive of killing, when its point was to demonstrate the demandingness of the duty of rescue. Indeed, I could have added that it was easily read as a case of opportunistic killing, since the rooftop man appears, in the paper’s rendition of the case, to save himself by cushioning his fall with the baby. (He is said, true, to land on his feet. But he is also said to land on top of, and squash, the baby). So if, as he reports, David uses the squashing version of the case elsewhere as an analogy to abortion by killing, I think he is in danger of putting people in mind of killing by a different, and worse, mode of agency. In his blog comments, meanwhile, David describes a version of the case without these particular problems, in which the rooftop man is plummeting to the ground with the baby on top of him, and offloads the baby via a half-turn, allowing it to die, in order to avoid impacting the ground in a manner that would (a) save the baby, but (b) be very costly to him. No one, David claims, may impose significant harm on the falling man to force him to carry out the costly rescue, though he is obligated to do so nonetheless. As with the island case, however, I’m not convinced – I have no clear intuition – that the rooftop man is under the relevant duty. And if he does rotate, it seems he doesn’t intend the baby’s death. For his plan to save himself does not involve or require the baby, and he passes David and Phil’s test, quoted above, in that he would presumably not be regretful if he were to learn, on reaching the ground, that the baby had somehow miraculously survived unscathed.

David’s cases do not, then, seem to provide clear or relevant evidence against the view that wrongful non-rescuers can be subject to substantial force. Before moving on, let me also add the following, semi-digressive point about the rooftop case (semi-digressive because it goes to David’s case for Restrictivism itself, rather than his case against AAV). Suppose that everyone had, as I do not, a clear intuition that the man must use himself as a shield or cushion for the baby, at high cost to himself. It is still unclear that we are justified in inferring anything from this about duties in pregnancy. For the rooftop case is an exceptional emergency. And it may be, as some think, that in unpredictable emergencies, when one happens to be on the scene, the costs we can be obligated to incur for the sake of effecting a rescue are elevated, relative to those which we must undergo for the sake of helping the needy on a more routine basis (as when people are in poverty, or their kidneys are failing). I’m not sure I agree that we can be required to do more on an emergency basis. And I am especially sceptical that, if such a duty exists, it is a duty which we owe to the imperilled (for it is hard to imagine them having a complaint against us, if we do not act, to the effect that we should have been a hero for their sake). But some people do think that there are duties to make heroic sacrifices in exceptional cases. And that makes less obvious the relevance of the rooftop case to abortion. For the question is then whether the duties to support a fetus are a matter of routine Minimally Decent Samaritanism, or of heroism. And arguably, they are the former, precisely because pregnancy is such a routine part of life.



As I say at the end of section 5.3, even if it is true that an individual cannot be subjected to significant defensive harm to prevent a single letting die, the disvalue of deaths by letting die aggregate. Even if, thus, you think that Jones is not properly subject to significant harm when he watches the child drown, it is hard to imagine that nothing changes if there are, e.g., drowning triplets in the bath. And a fortiori, an abortion doctor who carries out many dozens or hundreds of abortions seems justifiably subject to serious force, even if each individual abortion has the status of a letting die (and attracts, taken singly, only force falling below the level that violates the NVC). David disputes these points, claiming that both woman and doctor are ‘immune’ to harm, irrespective of numbers. But despite his long exchange on this topic with Steve, I still find myself unclear exactly why he thinks this. He explains in detail how what (on Aug. 3rd, 12.32 pm) he calls ‘immunity transfers’, from woman to doctor, would work, if they work. And he says (Aug. 2nd, 3.46 pm) that, since the doctor ‘inherits’ an immunity to harm from each woman, for the abortion he carries out on her behalf, it doesn’t matter how many abortions a doctor performs. I agree with Steve that this seems like magical thinking.

But even if David is right that both women and doctors are completely immune to harm, in the sense of retaining their rights against it, that is compatible with their rights being overridden, as a matter of the lesser evil. It is also compatible with the rights of bystanders being overridden on those same grounds. If many hundreds of wrongful deaths by letting die will take place within some abortion facility, and this can be prevented by planting a bomb that will destroy the facility, but seriously injure a single bystander as a side-effect, what is the objection to doing this? To cite the bystander’s immunity would be beside the point, since a lesser evil justification grants that they have it. One of David’s comments suggests that he may think that lesser evil justifications for harming the innocent can only be engaged when the negative consequences of abiding by the constraint against harming are so extreme as to reach a level comparable to ‘genocide’. So perhaps this is what he would now say in response: that sufficient good is not at stake. But that response proves too much, since, by that token, we cannot close the camps in Death Camps at the cost of a single collateral injury either.


Restrictivism and the assimilation argument

I said earlier that I think that the assimilation argument serves Restrictivists even less well than Thomsonians. David and Phil’s words in their paper illustrate some of the reasons why. In explaining why, on the Restrictivist view, abortion is a graver injustice than typical omissions to save, notwithstanding the parallel they have drawn between them, the authors write:

abortion is standardly worse than typical cases of letting individuals die in a far-off land from, say, malnutrition, because in such cases the non-saver is not responsible for their predicament nor has a special obligation as a relative to aid them, and hopes someone else will save them. These features are all absent in a typical abortion. Abortion standardly involves a woman deciding to kill a child who is in a state of precarious dependence on her because of her earlier decision to engage in consensual sex. It also involves killing one’s own child to whom one has special obligations.

There are three considerations here being highlighted as exacerbating the wrongness of abortion: that death is deliberately sought, that the woman is responsible for the fetus’s need for aid, and that she has the special obligations of a parent. As I pointed out earlier, the first of these considerations undermines David’s attempt to draw conclusions about defensive harm by appeal to cases like his roof case. And something similar applies to the second and third considerations which David and Phil mention here. If I had been able to discuss ‘How Not to Defend the Unborn’ in greater detail in my own paper, I would have noted that, as with the relevance of being an original cause of death, these considerations are employed by David and Phil in an ad hoc way.

Thomsonians do not tend to accept that pregnant women are parents, or responsible for their fetuses’ plight, in a morally-relevant sense. But Restrictivists like David and Phil frequently do. And having pressed that point, it seems they ought to follow through when defensive violence is at issue. Consider parental obligations. Parents are under a special duty to accept additional burdens, relative to a stranger, in the course of caring for their child. And the duty to care for the child, at the relevant cost, is enforceable, we usually think. Thus, an absent father can have his duty to pay child support enforced, even when the burden of paying outstrips what he would owe to a stranger. Insofar as parents must incur higher costs for the sake of their children, and the duty is enforceable, can they not also, a fortiori, be forced to bear higher costs than others when they themselves come to constitute a threat to their dependent child? It seems highly plausible that they can. And even if, on further examination, that is not right, David and Phil’s separate appeal to the woman’s responsibility for dependence through her prior act of procreative sex seems to remove a crucial, load-bearing Jenga piece from their own argument. It is not the case after all, David and Phil parenthetically note, that the woman is a mere bystander to the harm that will accrue to the fetus if she refuses to aid it. She is responsible for its needy condition. Why, then, can she claim the status of a bystander? The charge of ad hocness which I briefly make against David and Phil in the paper (and which, at 3.50pm on August 2nd, David says he has difficulty understanding) is a charge of failing to apply consistently moral considerations that they claim bear on the wrongness of abortion, and the aptness of its legal prohibition, but which one would also expect to bear on defence. I was referring at the time to their treatment of the significance of killing. The point applies here too, I believe.


Liberal and illiberal justifications of violence

In my paper, I put aside, for reasons of space and scope, the suggestion that AAV might be ruled out by a public reason constraint. As I suggest in the paper, I think that public reason liberalism is a deeply defective doctrine – primarily on grounds that it handles various practical policy questions in an intuitively unacceptable way. I have written on that in the past, and hope to return to and develop that critique further in future (for I think that, if public reason in practice is defective in the particular way I believe it is, rather than in the ways some other critics have argued it to be, this has implications for the sort of position on political justification that one ought to accept instead, and tells against various forms of accommodation or compromise with public reason views). As I mention in a note, meanwhile, an appeal to the Rawlsian variant of public reason was part of David and Phil’s case in ‘How Not To…’. But David, at least, would now not want to associate himself with, specifically, Rawlsian public reason. He suggests, however, that there is a general distinction between liberal and illiberal forms of violence that can do work in resisting AAV. He writes:

Our point is that there is a liberal sort of violence (one defended in liberal terms such as autonomy, self-ownership, bodily integrity, self-defense and the like) which is different from an illiberal violence of say the Nazi (defended on the grounds of illiberal racism, anti-Semitism, eugenics, Aryan myths, national purity etc.). Our view is that liberal societies should tolerate liberal violence and use the ballot box rather than the gun to remove it. But illiberal violence has no place in a liberal society and so is permissibly a target of righteous violence of private individuals.

He then asks:

Isn’t there clearly a liberal form of defending abortion on the basis of liberal sounding notions such as individual rights, autonomy, bodily control, bodily integrity self-ownership etc. that is lacking in the cases of violence that he and Kershnar mentions involving Nazis? So doesn’t our distinction stand up independently of Rawls’s framework even if it would also seem to follow [from] Rawls’s pubic reason [?]

For what it’s worth, I don’t mention, specifically, Nazis in my paper: I left it to the imagination what the underlying ideology or justification was in my Death Camps case. We might imagine, thus, that the justification given is clothed in language consistent with or drawn from liberalism. Indeed, at one point, I propose that the personnel in the camp be imagined to sincerely believe that the killings constitute justifiable non-voluntary euthanasia of those unable to give or withhold consent. Inserting that particular detail makes no discernible difference to the case. Nor, more generally, as far as I can tell, does the liberal/illiberal distinction. Consider a different kind of case: X is about to tread on Y’s toe. Y’s killing X in response to that minor threat would be grossly disproportionate. It would, in turn, attract justifiable defensive violence from someone, Z, acting on X’s behalf. It doesn’t seem to matter whether Y cites, in self-justification, his liberal right to bodily integrity, or his racial animus towards X. It seems bizarre to conclude otherwise – that there can be no substantial defensive force employed between avowed liberals, even when they threaten to violate each other’s fundamental rights. So in my view, David and Phil’s back-up argument, from liberalism, like their main argument, does not succeed.


[1] I have in particular reluctantly omitted various comments on the killing/letting die distinction, the inclusion of which was starting to make this document get a bit out of hand. Interesting as that issue is, David, who presses the relevant objections, seems to agree with me that doctors who perform wrongful abortions kill, even when the abortion is extractive, though he disagrees with what I say in the paper about the reasons why.