Central to non-consequentialist morality and any moral theory which attempts to hew even moderately closely to common sense is a morally relevant distinction between doing harm and allowing it to happen. Doing harm to others is intuitively much harder to justify morally than is allowing harm to befall them. Whereas I may allow you to die in order to save myself (as when I must rush off to the hospital, leaving you to drown, to have my poisonous snake bite treated), I may not kill you even if doing so is the only way I can save myself (as when, in order to get to the hospital, I must push a boulder out of the way and, unfortunately for me, onto you). Whereas I may allow you to die to save some other greater number of individuals (as when I must rush off to a different part of the beach, leaving you to drown, in order to save five others from drowning), I may not kill you in order to save a greater number of people (as when, in order to get to the other part of the beach where the five are drowning, I must, because you are in the way, run you over with my truck). Fiona Woollard's admirable Doing and Allowing Harm is an in-depth investigation and vigorous defence of the moral relevance of this distinction.

Woollard's focus is on what we might call the intrapersonal doing/allowing distinction:

(This is to be contrasted with what we might call the interpersonal doing/allowing distinction – ‘it is not (generally) permissible to do harm in order to avoid allowing greater harm, because doing harm is harder to justify than merely allowing harm’ (p. 14).) The book comprises four parts. Part I consists in a search for an analysis of the doing/allowing distinction. Part II is a defence of the moral significance of the distinction so analysed. Part III considers a certain class of limitations on the moral relevance of the distinction. And Part IV explores the suitability of Woollard's version and defence of the distinction for incorporation into either one of two prominent contemporary ethical theories.

Wise counsel, before trying to argue for the moral relevance of a distinction, would be to explicate it first. In Part I, Woollard considers a variety of different cases, marshalling a bevy of intuitions, and borrows a few theoretical distinctions, from Philippa Foot, Jonathan Bennett and Jeff McMahan, en route to an account of the morally relevant doing/allowing distinction. Whether an agent counts as doing or allowing a harm, Woollard maintains, depends on the relation between the agent's behaviour and the sequence that leads to the harm. An agent does a harm if and only if a fact about the agent's behaviour is part of the sequence leading to it; she allows it if and only if a fact about her behaviour is relevant to but not a part of the sequence leading to it. (Because none of Barack Obama's behaviour was relevant to the bombing of Hiroshima – he neither caused it, nor was there anything he could have done to prevent it – he neither did it nor allowed it to happen.) And a fact about the agent's behaviour is a part of the sequence leading to a harm if and only if ‘the agent is relevant to that sequence through a complete series of substantial or relatively substantial facts’ (p. 35). A fact is substantial if and only if it is a ‘positive fact’ – a fact which says that something is the case (e.g. Bob pushed the boulder), as opposed to a ‘negative fact’, one which merely says that something is not the case (e.g. Bob did not interpose his car between Victor and the oncoming boulder) – or it is ‘contrary to our normal presuppositions’ (p. 35). And a fact is relatively substantial if and only if it is a fact about the absence of a barrier either belonging to the victim or one both upon which the victim has a non-need-based claim and which isn't sustained by resources belonging to the agent.

This account is quite complicated, with many moving parts. The complications arise from Woollard's sincere desire to sculpt the distinction so as to capture as many of our pre-theoretic intuitive judgements as possible. Not only must the distinction classify Bob's pushing the boulder towards Victor as his doing him harm, so too must it likewise classify Bob's driving Victor's car out from in between the oncoming boulder and Victor as his doing him harm. Bob's driving his own car out of the way of the boulder, on the other hand, would be merely his allowing Victor to die. Thus, the need for the clause about relatively substantial facts – Bob's driving Victor's car out of the path of the boulder is the removal of a barrier belonging to the victim, whereas Bob's driving his own car out of the way isn't. Not only must the distinction classify Bob's injecting cyanide into Victor's bloodstream as a doing of harm, so too must it likewise classify Bob's pumping all of the oxygen out of the room in which Victor is trapped as a doing of harm. Thus, the need for the ‘normal presuppositions’ clause – it is a normal presupposition that we inhabit an oxygen-rich environment and so the absence of oxygen in the room, though a negative fact, counts as a substantial fact that is a part of the sequence leading to Victor's death.

There are a couple of deeply unsatisfying features of this account. First, it is incredibly baroque. It feels more like a hodgepodge of unrelated conditions pasted together than it does an analysis of a morally relevant real distinction in the world. True, some of our intuitive judgements about doing and allowing seem to motivate some of the various different conditions Woollard sets down, but sometimes some intuitive judgements must be sacrificed in favour of others. For surely it is also one of our intuitive judgements about doings and allowings that the difference between them isn't so higgledy-piggledy an affair. The choice on Woollard's account isn't simply that standard trade-off between theoretical simplicity and an honest adherence to the facts, as Woollard suggests it is (p. 60), but a clash of intuitions about doings and allowings themselves. Second, adverting to ‘our normal presuppositions’ makes the analysis conventional (a fact of which Woollard is aware and one which she openly embraces (pp. 90–2)), and objectionably so. Surely, another of our intuitions about a real distinction between doing and allowing in the world is that it isn't, even in part, a matter of convention. Whether something counts as a doing of harm, as opposed to an allowing, we don't think depends on what we happen to be generally presupposing at the time. (Would pressing the button on the oxygen-removing device really have been a letting die, as opposed to a killing, had we not advanced scientifically to the point where we knew that respiration requires oxygen and that the atmosphere contains it?) In so far as what we face here is a clash of intuitions, it might have been better to secure the non-conventionality of the distinction and give up some of the intuitions that conflict with it (e.g. the intuition that Bob's driving Victor's car out of the path of the boulder is an instance of Bob's doing Victor a harm), especially as many of those intuitions, arguably, are driven not by the thought that a harm was done, but, rather, by the thought that the agent acts particularly wrongfully and the feeling that classifying that action merely as an allowing would somehow not do sufficient justice to its wrongness. (Note, also, that proceeding in this way could be quite in keeping with much of the rest of what Woollard argues for in the book, for she herself notes not only that some allowings of harm are very morally wrong, but, more importantly, that there are different kinds of allowings with different moral valences.)

In Part II of the book Woollard aims to offer a defence of the moral relevance of the distinction as analysed in Part I. This is the heart of the book and, I believe, the most interesting part of it. The moral relevance of the distinction is supposed to be that it is harder to justify doing harm than it is to justify merely allowing harm. But how so and why? The moral relevance of the distinction is made manifest in a dual structure of constraints against doing harm and permissions to allow harm. Both the constraints and the permissions, Woollard argues, are bulwarks against harmful imposition. Here, borrowing an insight from Frances Kamm, Woollard notes that having harm done to one and being required to help others both involve a kind of imposition. When X harms Y, X pretty clearly enough causally imposes on Y. A little less obviously, when X is required to assist Y, Y normatively imposes on X; in being required to help Y, X is (normatively) forced to use what belongs to her, either her body or her resources, in the service of Y’s interests. But why are protections against harmful imposition morally necessary? Taking a page out of Warren Quinn's book, Woollard suggests that protection from harmful imposition is necessary if anything is to truly belong to anyone. And some things do truly belong to us. Some things, in the very first instance our own bodies, are such that we do have a fundamental and unique normative authority over them. Woollard goes further, digging a little bit more deeply than Quinn, and identifies the ways in which belonging requires protection from harmful imposition. In brief: a right to possess and a right to use a thing are necessary for that thing to belong to one; constraints against doing harm undergird the former and permissions to allow harm ground the latter.

This is deep stuff and it has an air of plausibility to it. There is a tension, however, between this rather straightforward explanation of the moral relevance of the doing/allowing distinction and the analysis of the distinction Woollard gives in Part I – some doings of harm, e.g. the removal of certain barriers, don't involve any causal imposition on the victim. This counts less, I'm inclined to think, against Woollard's illuminating grounding of the moral relevance of the distinction as it does against her overly ecumenical account of doing harm. An analysis of the doing/allowing distinction which left all barrier-removals on the allowing side of the divide, for instance, would fit much more nicely with Woollard's arguments in Part II than does that which she defends in Part I.

After having defended the moral significance of the doing/allowing distinction in Part II, in Part III Woollard sets about outlining some of the limits to the moral relevance of the distinction. Doing harm may well be harder to justify than merely allowing harm, but, because the well-being of others matters morally, some harm doing is morally permissible and some harm allowing is not. Chapters 7 and 8 explore the latter limits by considering the well-known arguments of Peter Singer and Peter Unger to the effect that we are morally required to sacrifice far more than is intuitively thought necessary to assist those in dire need. Woollard contends that there are two distinct obligations to aid those in need: a requirement ‘to make one-off substantial sacrifices when personally involved in a crisis’, and ‘to make non-trivial contributions in response to continuing need’ (p. 158). Both requirements arise out of the interplay between morality's protection against normative imposition upon that which belongs to us and the moral importance of the welfare of others. The two requirements thus emerge as the embodiment of a ‘maximal requirement to aid compatible with recognizing that our bodies and other resources genuinely belong to us’ (p. 148). A requirement to make large one-off substantial sacrifices when personally involved in a crisis places one under a severe obligation to aid when such crises arise, but because such crises have a low relative frequency, the obligation, though severe, does not count as an intolerable imposition upon that which belongs to one. And a requirement to make continued non-trivial contributions to respond to recurring need also, though a constant obligation, isn't so intrusive as to constitute a threat to one's body and resources belonging to one, for it isn't as onerous or severe as that which holds in one-off crisis cases. (It should be emphasized, though, that this latter requirement might indeed be a much more severe requirement than is generally thought to hold of us. It seems perfectly consistent, for example, with one of the principles Singer puts forward in his landmark ‘Famine, Affluence, and Morality’ (Philosophy and Public Affairs, 1972), the one which states that when we can prevent something very bad from happening without sacrificing anything of moral significance at all (i.e. anything that doesn't count as a luxury), we morally ought to do so.)

Finally, in Part IV, Woollard tries to demonstrate how usable her account of the morally relevant doing/allowing distinction is by showing how it can be incorporated into the very different contemporary moral theories of Tim Scanlon's contractualism and Brad Hooker's rule consequentialism. In each case, Woollard shows how her belonging-based grounding of the moral relevance of the distinction might fit naturally within the theoretical frameworks of each philosopher's moral theory.

A morally relevant doing/allowing distinction is indispensable to common-sense morality. So far, however, the distinction has proven elusive; no fully satisfactory analysis of the distinction has yet been offered. Woollard's Doing and Allowing Harm does a laudable job of trying to limn the contours of the distinction and offers a plausible account of the grounding of its moral significance and limitations. All who champion some version of non-consequentialist moral theory would do well to engage with the arguments and ideas in it.