1In his excellent Sidgwickian Ethics, David Phillips argues that Sidgwick’s argument for utilitarianism from the axioms is less successful than Sidgwick believes. He also argues that Sidgwick’s argument for egoism is more successful than this argument for utilitarianism. I disagree. I close by noting, briefly, a possible solution to an epistemological puzzle in Sidgwick that Phillips raises.

I. Utilitarianism

2Phillips takes the argument for utilitarianism to have two premises:

4Phillips thinks (R) is analytic and that Sidgwick “should be happy to admit” this (124). (R) is analytic because for Sidgwick, “to say something is good just is to say there is reason to aim at it” (124).

6But Phillips does not think Sidgwick’s argument fails because (R) is analytic. He takes (U) to be the problem. The first problem is that an egoist can deny “that there is such a thing as universal goodness” (125).

7I do not think that (U) unpacks the concept of “universal goodness.” Phillips argues that it does mainly by citing the first edition version of (U) (121-2). But even if he is right about the first edition, in the later editions Sidgwick seems to unpack not “universal goodness” but rather what it is to take up the point of view of the universe (or at least has this as the antecedent of (U)). And putting the egoist’s disagreement in terms of denying that there is universal goodness is puzzling. Phillips’s idea is that the “egoist can admit…that some people have the concept of universal goodness; but the egoist will deny that that concept is ever instantiated” (152n3). The egoist, then, does not find the concept confused (like “round square”), but rather uninstantiated (like “unicorn”). But this seems an odd thing to say: surely there is something I ought to desire, assuming myself to have an equal concern for all existence (or, alternatively, surely there is something that is the set of the goods of each individual). It seems preferable to say instead that the egoist is uninterested in this thing—or, as Sidgwick says, refuses to take up the point of view of the universe.

8But this disagreement is not so important, since Phillips notes that Sidgwick, given the dualism, might admit this first problem (125). Whether the problem is put in terms of taking up a point of view or instantiating a concept does not affect this admission. The more serious, second problem Phillips raises for (U) is that, even if true, and even if one does take up the point of view of the universe (or think “universal good” is instantiated), (U) cannot justify utilitarianism. Consider a view on which “(a) any person’s happiness is good; i.e., there is a requirement of reason for any agent, ceteris parabis, to promote it; and (b) there is a special requirement of reason for any agent to promote his own happiness three times as much as he promotes the happiness of anyone else” (125). On this view, (U) is true, but what one ought to do is not act as a utilitarian, but rather “promote the weighted sum of her own good and everyone else’s” (125). Phillips puts his objection this way: “Sidgwick slips from the (proper, but nonsubstantive) idea that if there is such a thing as universal goodness there is some reason to aim at it, to the (substantive, but not properly supported) idea that it is the only thing that ought to be aimed at or promoted” (125) The real problem, he thinks, is that (U), like (R), is analytic, and so cannot rule out aiming at other things. (Phillips does not say explicitly that (U) is analytic, but he does write that “the argument…contains only tautologies” (125).)

9I have three comments on this objection.

10(1) Insofar as the objection is that Sidgwick cannot go directly from the axioms to utilitarianism, Phillips is surely correct. Sidgwick thinks common sense moralists agree with his axioms (ME 421, GSM 331-2). He also thinks Clarke and Kant agree with them (ME 384-6). So he cannot understand the axioms as ruling out reasons to act in non-utilitarian ways. But the obvious fix is to think that Sidgwick supposes that he has already, in Methods III.XI, ruled out any additional axioms (such as Phillips’s (b)). Whether that argument works is another matter (considered by Phillips in his pretty convincing chapter 4). But the issue is the success of that further argument; Sidgwick is not defeated simply by noting the possibility of a view like (a)-(b).

11(2) The objection in (1) can be made without thinking that (U) is analytic. One might think that (U) says that “from a point of view from which I give no special weight to myself, I ought to be indifferent between distributions that result in the same amounts of good.” This is not analytic, in part because giving no special weight to myself is consistent with favouring one distribution over another on grounds other than the amount of goodness. But if so, (U) by itself (or with (R)) does not secure utilitarianism.

12(3) I am not sure how Phillips understands (U). He seems to gloss (U) as saying that “there is such a thing as ‘goodness’ period” (124), or as saying that “his happiness is good, not just good for him” and “his happiness cannot be a more important part of good, taken universally, than the equal happiness of any other person” (125). He also thinks of “the premise” of the argument for utilitarianism as saying that “there is such a thing as…good-from-the-point-of-view-of-the-universe” (125). But (U) seems to say only that from the point of view of the universe, I ought to be indifferent between equal amounts of good. (U) does not, then, entail that one does take up the point of view of the universe. Nor is it analytic, even on what I take to be Phillips’s reading: if (U) claims that “universal good” is instantiated, Phillips says the egoist disagrees, but presumably not with an analytic truth.

II. Egoism

13Phillips takes the argument for egoism to have two premises:
1.The distinction between any one individual and any other is real and fundamental.
2. If the distinction between any one individual and any other is real and fundamental, then I ought to be concerned with the quality of my existence as an individual in a sense, fundamentally important, in which I ought not to be concerned with the quality of the existence of other individuals.
Therefore,
3. I ought to be concerned with the quality of my existence as an individual in a sense, fundamentally important, in which I ought not to be concerned with the quality of the existence of other individuals (127-8).

14Phillips notes that 3. is weaker than egoism, since special concern for myself is not exclusive concern for myself. But he thinks the argument does better than the argument for utilitarianism, for two reasons: 2. is “both substantive and self-evident,” and 3., though not egoism, is the significant claim that there are agent-relative reasons (129).

15Phillips glosses 1. as follows: “I have a special connection to certain goods and bads (e.g., I directly experience certain pains and pleasures)….The key idea is just that of a kind of special connection: that the fact that a certain pleasure or pain is mine means I experience it in a way others do not” (129, 131). 2. is compelling because if 1. is true, “that fact gives me special reason to want and pursue those goods, and to not want and avoid those bads….My reasons are reasons for me. How could the fact that something has a special effect on me not affect my reasons?” (129)

16I have four comments on this argument.

17(1) Phillips describes 2. as a normative premiss (130). This lets him avoid the objection that the argument deduces a normative conclusion (3.) from wholly non-normative premises (1.). But this still violates Sidgwick’s rule that normative conclusions cannot be deduced from wholly non-normative premises. For 2. says that a wholly non-normative claim entails a normative claim.

18(2) As it stands, the argument seems to depend on a particular view of the good. The special connection is that I directly experience the things that are good. This may be plausible if the things that are goods are mental states. But I would not have thought that the argument for agent-relative reasons depended on holding any particular view about what is good. For example, some think that what is good is the state of affairs (whether that is a mental state or not) that satisfies my preference. I have no special connection to the obtaining of that state of affairs—I might not know that it obtains, and no state of me need be a part of it. Phillips might reply that for such goods, there is no case for agent-relative reasons. But that has odd results: for example, I then have no agent-relative reason to care about how my children fare after my death.

19(3) I take it that Phillips’s idea in 1. is this: we are creatures such that if I experience a pleasure or pain, it does not follow that anyone else experiences it. In this sense, we are distinct individuals, as would not be true if, for example, we were creatures with direct empathetic connections, or there was a sea of pleasures and pains without distinctions between individuals. The question is why this gives me a special reason to care about my own pleasure or pain, over and above any reason I have to care about the occurrence of pleasure or pain somewhere. Phillips’s thought seems to be that since it is a state of me (and not you), it must give me a special reason. But this seems awfully close to simply asserting what was supposed to be the conclusion of the argument, viz. that I have a special reason to care about states of myself.

20I can see one way in which 1. is part of an argument for 3. Phillips goes on to endorse Roger Crisp’s “Two Doors 3”:

21You are confronted with two doors. If you do not pass through one or other of them, you will suffer an extremely painful electric shock. If you pass through door A, you will experience a less painful but significant shock. If you pass through door B, you will not experience this shock, but some other person, a stranger and out of sight, will suffer a shock of the same intensity….Surely you have a reason—a strong reason—to choose door B over door A grounded in the fact that it is that door which will significantly promote your well-being (148-9).

221. makes this kind of case possible, and our intuition about this case then constitutes an argument for 3.

23I do not share this intuition. Since some others do, I see how this might be an effective argument for agent-relative reasons. It is not really, however, the argument of 1.-3.. That argument promised a justification for agent-relative reasons that goes beyond just giving cases in which some think we have them.

24(4) Phillips reconstructs Sidgwick as giving a deductive argument for egoism. I think Sidgwick intends something weaker. In the Methods, after noting 2., he writes that “I do not see how it can be proved that this distinction is not to be taken as fundamental in determining the ultimate end of rational action for an individual” (ME 498). The point seems to be that if someone thought that the distinction between one individual and another did matter to choosing between egoism and utilitarianism, it is hard to see how to show that such a person is wrong. It does not follow that one must think it matters. Similarly, in “Some Fundamental Ethical Controversies,” where the argument first appears, Sidgwick writes that “the proposition that this distinction is to be taken as fundamental in determining the ultimate end of rational action for an individual cannot be disproved” (FEC 484). Of the utilitarian position, Sidgwick writes “I do not deny this position to be tenable; since, even if the reality and essentiality of the distinction between one individual and another be granted, I do not see how to show its fundamental practical importance to anyone who refuses to admit it” (FEC 485). Sidgwick’s view seems to be that the distinction between individuals is something to which more than one response is permitted. He, and egoists, take it to be relevant to setting ultimate ends. Others, such as utilitarians, may disagree.

III. Epistemology

25Phillips notes that in Methods IV.II, the “proof” of utilitarianism proceeds by ad hominem arguments directed at the egoist and the common sense moralist. The application of the tests for highest certainty in III.XIII, which seemed at least to have established consequentialism, and which elsewhere Sidgwick endorses as the best way of justifying beliefs, seems to have been forgotten (65-76). Earlier I suggested that there is a gap between the axioms and utilitarianism. If so, Sidgwick has reason to think the tests are insufficient: they secure the axioms but not utilitarianism. This is slightly different than Phillips’s solution to the puzzle, according to which Sidgwick simply saw that the appeal to the tests failed. On my view, the appeal to the tests succeeds in establishing the axioms—which is why Sidgwick does not note that the argument of III.XIII fails.