1 What I have just defined is standard, or optimizing (maximizing), consequentialism. There is also non-optimizing or satisficing consequentialism, which dispenses with optimality, holding instead that an action is right just in case its consequences are good enough. Cf. Slote, M., Common-Sense Morality and Consequentialism (London, 1985). In the article I will be referring to optimizing consequentialism, except when I say otherwise. As far as I can tell, though, the arguments I offer against optimizing consequentialism apply equally to satisficing consequentialism. There are also forms of rule-consequentialism, of course, but I will ignore these entirely.
2 Cf. Ewing, A. C., Ethics (London, 1953), pp. 113–14: ‘[Metaphysical and naturalistic definitions of ‘ought’] would destroy what Kant calls the autonomy of ethics by refusing to recognize the uniqueness of its fundamental concepts and trying to reduce it to a mere branch of another study.’
3 See Moore, G. E., Principia Ethica (Cambridge, 1903), p. 25, for the original claim, and Ethics (Oxford, 1912; 1971 reprint), p. 73, for the recantation.
4 This last claim is perhaps a bit anachronistic as far as Moore is concerned, as he rarely talks of ‘properties’ at all.
5 See Bales, R. E., ‘Act-Utilitarianism: Account of Right-Making Characteristics or Decision-Making Procedure?’, American Philosophical Quarterly 8 (1971), pp. 257–65. In opening up for anti-reductionist consequentialism I at least ostensibly go against one influential tradition in moral philosophy that simply defines consequentialist (or ‘teleological’) theories as reductionist. For the best-known example, see Rawls, J., A Theory of Justice (Cambridge, Mass., 1971), p. 24. However, it is not clear whether representatives of that tradition, Rawls included, were explicitly concerned to exclude the anti-reductionist alternative. The distinction between the two versions of consequentialism is but rarely at the forefront of discussion.
6 For a clear endorsement of this claim, see Molnar, G., Powers (Oxford, 2003), pp. 209–10, who also uses the triangle example.
7 See Singer, P., ‘Famine, Affluence, and Morality’, Philosophy & Public Affairs 1 (1972), pp. 229–43, and Kagan, S., The Limits of Morality (Oxford, 1989).
8 This is not to say that this second type of reduction is a necessary feature of any meta-ethical position meriting the label ‘naturalist’ (though it plainly is sufficient). Some naturalists claim to defend a version of that doctrine that does not involve any reduction of this second type. See e.g. Sturgeon, N., ‘Moral Explanations’, reprinted in Essays on Moral Realism, ed. Sayre-McCord, G. (Ithaca, 1988).
9 Note that though G will contain species, these are not essential to it. A particular species could disappear (due to its not being instantiated anymore), yet the genus may persist. The function PG is essential, however.
10 It is possible to classify the set of animals differently from the suggestion in the text – say into aquatic, aerial and terrestrial animals. That decision would yield a different genus. In this case M is trivial: within this genus, species is determined simply by the type of environment in which a given animal spends the majority of its time. Similarly, the set of ‘deontic objects’ can also be classified in a variety of ways, some interesting, others not. It is important to note, then, that a ‘genus’, as I use that term, incorporates a method of classification.
11 These new insights in hand, we may restate in our new terminology the conclusion of the preceding section: if TD has the function of determining species only when applied to objects of a certain sort, though it is equally applicable in many other contexts, that fact is in need of explanation, an explanation that anti-reductionist consequentialists cannot provide. It is crucial here that TD is general, applicable also to certain other genera. As I said, consequentialism implies that PD consists of the idea that a member of the deontic genus, characterized by CD, however understood, is right just in case it is at least as good as any relevant alternative. This idea does not apply exclusively to actions. Take the hurricane value genus, for instance, whose two species are hurricane optimality and hurricane suboptimality. The type of its species-determining mechanism is exactly the same as that of the deontic genus (but its criterion, C, is of course different).
12 The example illustrates another fact I mentioned earlier, that P is relativized to G (and so to C): when G changes, so does P. In our example, we find that to each property p serving as an argument for PP there corresponds the property p & CP in PM. For concreteness, the property o, which distinguishes orangutans from all other primates, corresponds to the property o & CP, which distinguishes orangutans from all other mammal species. In this case it might not be practically necessary to mention the additional property, CP, but it is metaphysically possible for some non-primate mammal to have o, and to distinguish the orangutan species from that hypothetical mammal species we also need to bring in the fact that orangutans are primates.
13 By contrast, in n. 10 I suggested the possibility of dividing animals into terrestrial, aquatic, and aerial. This genus is not a realm. The reason is that it is subordinated under a more general genus, one that divides all material objects into these three categories. The type (i.e. T) is the same: it is where the objects ‘spend’ most of their time. A rock on the ocean floor, for instance, would be aquatic.
14 Inspiration for this diagram comes from Wlodek Rabinowicz.
15 This premise is not true of non-realms. We may again use the case of dividing animals into aquatic, terrestrial and aerial. Algae, for instance, are aquatic but are not animals.
16 Condition (i) is weak, as it does not require that the species of R also be species of R* necessarily. Perhaps that is too weak for some people's taste. However, I can see no reason to believe that there is any possible world in which any species of R is not also a species of R*. And we know also that there is at least one world in which there is a species in R* that does not also belong to R. Some might think that, too, is required for inclusion.
17 Here I am simplifying Kant's position. What I am referring to is what he calls the ‘morality’ of an action, as opposed to its ‘legality’, which is independent of the agent's maxim. For the distinction, see I. Kant, The Metaphysics of Morals, trans. M. Gregor, Practical Philosophy (Cambridge, 1996), p. 380 (Ak., p. 225).
18 I owe a big debt to Wlodek Rabinowicz. When I presented an ancestor of this article at a departmental seminar, Wlodek blindsided me with a simple yet seemingly devastating objection, and I was astonished that I had not seen it myself. Only later did I realize that, being (apparently) such an inveterate anti-reductionist myself, I had simply overlooked the possibility of a reductionist form of consequentialism, and hence that my argument was not quite as sweeping as I had thought – but still workable. I should also thank Robin Stenwall for piloting me through the unknown waters of species and genera, though my presentation will no doubt still strike him as a bit dilettantish.