This is an inventive and ambitious book, and it offers its readers much to explore. Each chapter brings a wealth of surprises, distinctive ideas and provocations. Readers will resist many of Peterson's proposals, to be sure, but his arguments repay consideration, and provide an imaginative route into some difficult problems.
Peterson's core claim is that the deontic status of an act will often depend upon multiple irreducible aspects of the consequences. An act that is right with respect to one aspect may be wrong with respect to another. For instance, an act might have beneficial effects on equality, while resulting in lower welfare. Both equality and welfare influence the deontic status of an act, and they are not straightforwardly exchangeable. On Peterson's view, such acts are ‘a little bit right and a little bit wrong’ (p. 2). When an act is right with respect to one aspect and wrong with respect to another, we should resist the temptation to arrive at a unitary deontic judgement, for if we do so then we will commit a ‘deontic leap’ – our conclusion will not adequately reflect the richness of our evaluative judgements. Instead, Peterson suggests, each aspect should ‘remain visible in the final analysis’ (p. 29). Deontic status, in other words, will be a many-place function rather than a one-place function (p. 3), and a natural way to represent it will be as a vector of real numbers, with each representing the standing of the act with respect to a particular aspect (p. 175).
Peterson hopes that his proposal will allow consequentialists to address a number of important objections to consequentialism, such as the objection from separateness of persons. On traditional utilitarianism, consequences are assessed by aggregating welfare across persons. If an act improves welfare by ten units for one person at a cost of five units for another, then other things being equal the act is right. Critics have argued that this fails to respect the separateness of persons; we may have moral reasons related to overall outcomes, but we also have moral reasons related to concern for individuals. Peterson hopes that multi-dimensional consequentialism can offer a response to such objections. Each person, he says, should be understood as a separate ‘aspect’ – as considerable in his or her own right, with interests that cannot simply be aggregated with those of others. On Peterson's view, the act described above would at best be right to some degree and wrong to some degree. Peterson offers many such proposals, and they are often imaginative, controversial and thought-provoking.
The book as a whole is dedicated to making Peterson's initial proposal more precise; using it to analyse traditional problems for consequentialism; and examining upshots for moral decision-making. Chapter 2 defends the idea that rightness comes in degrees. This claim is one of the trickier features of the book, and it seems to me that the discussion is sometimes too elliptical to give the reader a full picture of how the concept is to be used. At times it sounds as though degrees apply to an act as a whole – that a given act is right to some degree and wrong to some degree. At other times, it sounds as though degrees apply to individual aspects in such a way that an act could be entirely right with respect to one aspect while only being partly right with respect to another (p. 118). It seems to me that this possibility is not fully explained by the ‘Conflict of Aspects’ argument given in chapter 2. At any rate, precisely how degrees are to be assigned to each aspect remains obscure, and it would be nice to see this made more transparent.
Chapters 3 (‘Persons’), 4 (‘Equality’) and 5 (‘Risk’) each address irreducible dimensions of consequences, and show how a multi-dimensional approach might allow for distinctive analysis of diverse aspects. These chapters contain inventive and provocative variations on familiar themes such as separateness of persons, loss-aversion, the Taurek lifeboat problem and the analysis of risk.
Chapter 6, on multi-dimensional decision-making, offers a proposal for how we ought to make decisions in cases in which there are conflicting aspects. Specifically, Peterson argues that when our alternative acts are all partly right and partly wrong, we should randomize across those acts. That is, if A, B and C are right to some degree and wrong to some degree, then we should sometimes take A, sometimes B and sometimes C. This proposal initially looks highly controversial. In an objection raised by Daniel Hausman and discussed by Peterson, what if there is an act that appears to have some aspect in its favour, but which we should never take because other aspects dominate against it? In such cases, randomizing appears to give the wrong answer. For instance, imagine that act A involves breaking a minor promise in order to save a life, while B involves failing to save a life but keeping a minor promise (p. 120). In that case it seems clear that we always ought to take A, even though it involves breaking a promise. Peterson's response is that what appeared to be an aspect, in that case, was not an aspect after all: it is entirely right to save the life, and doing otherwise would simply be wrong.
This response reveals some perhaps surprising features of the theory. We might have wondered up to this point whether aspects could simply correspond to incommensurable values – for instance, whether equality and welfare might be understood as representing distinct and incommensurable values, which accordingly warrant being treated as separate aspects. The example above shows that we cannot identify aspects simply by looking towards incommensurable values. Different features only count as separate aspects if they factor into deontic status, and values might be incommensurable and important but dominated by other concerns, and thus not contribute directly to deontic status. This supports Peterson's claim that the theory he has proposed is genuinely distinct from value pluralism. However, it also has the result that the theory is less expressive of the complexity of our normative judgement than we might initially have hoped – a limitation to which I return below.
Chapter 7 analyses what Peterson believes to be the best one-dimensional consequentialist theory, and explains why two problematic axioms make this theory inferior to multi-dimensional consequentialism. Chapter 8 examines the idea that we might be able to ‘consequentialize’ any moral theory. Peterson suggests that consequentializing moral theories is desirable because it allows us to make their claims more precise, and to employ tools of analysis that might be difficult to employ otherwise.
As the above summary should make clear, Peterson has imaginatively set out to show a route through which consequentialists might respond to a range of classic objections from non-consequentialists. Does he succeed?
I suspect that most critics of consequentialism will say that the answer is ‘no’ – or at least, ‘not yet’. My own view is that the account is limited by the way in which Peterson has chosen to demarcate the task of moral theory. Peterson thinks of moral theory as being concerned solely with deontic status. He says, ‘a moral theory can be conceived of as a device that attaches deontic predicates to acts’, where deontic predicates include ‘right’, ‘wrong’, ‘permissible’, forbidden’, ‘obligatory’ and ‘supererogatory’ (p. 170). But this is a narrow view of moral theory, and one that many critics would reject.
Bear in mind that Peterson's core criticism of one-dimensional theories is that they ‘do not provide adequate conceptual resources to account for this type of complex moral judgement’, i.e. judgement in cases where multiple aspects conflict (p. 11). I agree with Peterson that this is a serious flaw – indeed that it is a fatal one. But I strongly suspect that no moral theory that reduces moral status to deontic status will be able to capture the complexity of our moral judgements.
How might multi-dimensional consequentialism be developed to reflect the full range of our moral evaluative attitudes? One option would be to start by replacing ‘deontic’ with ‘normative’ and then see how the account would need to change. For instance, we could begin by saying that the normative (rather than deontic) status of an act depends on several irreducible moral aspects (C1) and agree that the binary relation ‘at least as good consequences as’ is not a complete ordering (C2) (p. 8).
Even if we focus on deontic status, however, we could give a less reductive account by replacing Peterson's (C3), which states: ‘Moral rightness and wrongness are non-binary entities, meaning that moral rightness and wrongness vary in degrees’ (p. 10). Presumably rightness could be non-binary and not vary in degrees. We might reject degrees in favour of a much richer picture of what it means to say that an act is right or wrong – a picture that captures more of the complexity of our moral judgement.
