Iwao Hirose's book Moral Aggregation is a clear and insightful discussion of the nature and role of aggregation in moral reasoning. The book has two parts. In the first, Hirose discusses the nature of aggregation. His chapter on the clarification of what aggregation is, is admirably clear and accessible. He then draws a distinction between substantive and formal moral aggregation (more on this below), and makes a very valuable contribution to the debate by carefully discussing various interpretations of the Separateness of Persons criticism against moral aggregation. He also includes a chapter on intra-personal aggregation, an interesting issue that has not received enough attention. The second part of the book concerns the debate about moral aggregation that followed in the wake of John Taurek's famous Rescue case (see below). Hirose's discussion of the various responses is an important contribution to the debate.

Hirose's main claim in the book is a defence of what he calls formal aggregation, in contrast to substantive aggregation. Overall, this claim plays a quite small role in the book. But since it is the main claim, it is definitely worth some discussion. The standard form of aggregation is substantive aggregation. Its purpose is to help us decide what to do; it requires that we already have decided what the morally relevant factors are, and that the task left is that of combining their values into an overall value for each option. For example, classical utilitarianism takes the morally relevant factors to be people's happiness or pleasure, and combines the value of each individual's happiness or pleasure into an overall value of happiness. Substantive aggregation is, however, often criticized, and Hirose seems more or less to agree with that criticism. Formal aggregation, on the other hand, is not about telling us what to do; rather, it is a tool for illuminating the structure of the morally relevant factors. Therefore, Hirose argues, it is not vulnerable to the criticisms raised against substantial aggregation. To see how formal aggregation works, consider the following case: suppose we can save one person, A, or we can save another person, B, and also cure person C's sore throat. According to classical utilitarianism, curing C tips the balance between saving A and saving B. However, many people have the intuition that we wrong A if we let the presence of C and his/her sore throat tip the balance in such a way that A has no chance of being saved. One way of interpreting that wrong is as unfairness: A and B have equally strong claims to be helped, and fairness requires that we treat both of them the same. Therefore, saving B outright means that we are unfair to A, because we are not treating them the same. With the help of formal aggregation, Hirose presents the decision problem in the following way. The expected good of saving B and curing C's sore throat is <one life saved and one sore throat cured – u>, where u stands for the unfairness to A. This value should then be compared with the expected good of determining what to do on the basis of tossing a (fair) coin, which is <saving one life plus ½ (sore throat cured)>. If not curing C's sore throat is more than twice as bad as the unfairness to A, then C's sore throat should tip the balance, otherwise it should not. According to this formalization, fairness can thus sometimes be outweighed by other values. However, we could have decided that the unfairness to A should be taken into account in some other way, and if we had, we would have rewritten the possible outcomes accordingly. Formal aggregation doesn't specify the way in which the unfairness to A should be taken into account; it is only a tool for clarifying the way we do decide to take it into account.

However, formal aggregation's complete flexibility with respect to accommodating different views also constitutes its major weakness. Since there is no restriction on how morally relevant factors are to be combined, it is completely up to us how we want to do it. This makes formal aggregation ad hoc. Hirose acknowledges that, but thinks that, given the way formal aggregation can help us clarify the structure of morally relevant factors, this is a price we should be willing to pay.

I agree that, sometimes, writing down our decision of how morally relevant factors should be structurally related to each other can help us clarify our thoughts (and that is not a minor feat!). But the benefit stems from the extra clarity that formalization brings, not from aggregation per se. Sometimes formalization will show that our reasoning does indeed rely on aggregation; sometimes it will show the opposite. And sometimes, I think, formalization will not add much to our reasoning at all. For example, we don't need formal aggregation to understand the claim that sometimes a little unfairness can be outweighed by other values. Overall, then, I accept Hirose's argument that formal aggregation escapes the common criticisms against substantial aggregation, but I have not yet been convinced that formal aggregation is all that useful or important. More examples are needed which illustrate the benefits of formal aggregation, in particular cases in which it is aggregation rather than merely the formalization that does the work.

As mentioned, the second part of the book concerns Taurek's famous Rescue case and the various responses to that case. The original scenario is this: in the Rescue Case, six people are in need of a life-saving drug; without it, they will die. Five of them only need one-fifth each of the amount of the drug that you have, so you could save all of them. The sixth person, however, without any fault of his/her own, needs all of the drug in order to survive. The question then is: who should you give the drug to? One answer is to save the greatest number; that is, give the drug to the five who can share it and all survive, and let the sixth person die. Another answer is to give each person an equal chance of surviving: toss a (fair) coin. If it lands heads, give the drug to the five people; if it lands tails, give it to the single person. That way, everyone has a 50 per cent chance of surviving, and Taurek argues that since we owe it to everyone to show them equal concern and respect, this is what we ought to do. The Rescue case has generated a lot of responses. Hirose provides an admirably clear overview and discussion of them, and his comments are insightful and valuable. To mention just one example, I found particularly beneficial his comments on the views put forward by Frances Kamm and Thomas Scanlon.

I also found Hirose's discussion of interpretations of Taurek's demand that we treat everyone with equal concern and respect very valuable. This requirement can be interpreted in very different ways, Hirose points out (even utilitarians would agree with it). Even if we take it to mean ‘give everyone an equal chance of being saved’, there are some deeply counterintuitive possible interpretations, counterintuitive in ways that presumably would trouble Taurek too. For example, Hirose points out, suppose there are five people stranded on one rock and a single person on another. You can only go to one rock. The rock the single person is on is surrounded by a dangerous current that makes an attempt to save him only 50 per cent likely to succeed, whereas there is no current around the rock that the five people are on. If we ought to equalize their chances of being saved, we need to arrange a weighted lottery such that there is a 2/3 chance that we head to the rock with the single person, and 1/3 to the other rock. But clearly it is absurd that we should be more and more likely to try to save a particular person the less likely it is that a rescue attempt will be successful. The same awareness of the vagueness of common claims about showing respect also characterizes Hirose's last chapter, on weighted lotteries.

However, in a few places, I found myself disagreeing with Hirose's discussion of responses to the Rescue case, for example with regard to Thomas Schelling's non-aggregative argument for saving the greatest number (pp. 174–5). Schelling argues that we ought to focus on what rule for rescues we ought to adopt, rather than whom to rescue in the particular case. Behind a veil of ignorance we would choose the rule that says the rescuer should rescue the greatest number. That is the rule that maximizes your chance to be rescued. But it is also the rule that maximizes the chance that we save individuals especially worthy of being rescued, if there are any such individuals, since they too are more likely to be members of the large group than of the small one.

Hirose criticizes this argument in two main ways. First, by pointing out that just as there can be people more deserving of being rescued, there can be people not deserving of being rescued, and both kinds of people are more likely to be found in the larger group. It seems that Hirose assumes that the rescue of a morally bad person counts negatively, so that saving one such person outweighs saving one especially deserving person. But I don't think there is, at least typically, something negative about saving even a morally bad person, it just isn't morally pressing to do so. If that is correct, we still ought to save the greatest number, because that maximizes our chances of saving especially deserving people. Hirose's second criticism is directed against Schelling's claim that self-interest will make us support the rule to save the greater number. Hirose claims that the rescuer would know she was not at risk (since she was the rescuer), and therefore she would have no self-interest-based reason to save the greater number. However, note that the rescuer's self-interest is supposed to guide her choice of rule; the decision about whom to save in the actual case is just about the implementation. Of course, we might doubt that she is interested in saving anyone at all, since she is lucky enough not to find herself among those in need of rescue. But that is a much more general criticism than Hirose notes: it is a problem of commitment to rules that were chosen for self-interested reasons behind a veil of ignorance, but have to be complied with even when we do know that in actual life they don't work to our advantage.

But small disagreements like this one aside, I found the book a great contribution to the debate on moral aggregation, and highly recommend it to anyone interested in the nature and role of such aggregation in moral reasoning.