sidgwick's axioms and consequentialism

Sidgwick’s Axioms and Consequentialism

Robert ShaverUniversity of Manitoba

Sidgwick gives four tests for highest certainty.

(1) The terms of the proposition must be clear and precise.

(2) The proposition must be (as far as I can tell by reflection on it) self-

evident.

(3) The proposition must be consistent with other propositions I take

to be self-evident.

(4) The proposition must not be denied by someone of whom I have no

more reason to suspect of error than myself (1981 [1907], 338–42).

When he applies these tests to commonsense morality, he findslittle of highest certainty. In contrast, when he applies these tests to hisown axioms—axioms that he claims lead to utilitarianism—he findsthese axioms to have highest certainty. The key axioms supporting utili-tarianism are:

Personal Irrelevance: “The good of any one individual is of no more impor-

tance, from the point of view (if I may say so) of the

Universe, than the good of any other; unless, that is,

there are special grounds for believing that more

good is likely to be realised in the one case than in

the other” (Sidgwick 1981 [1907], 382).

Thanks to Roger Crisp, Tom Hurka, Joyce Jenkins, David Phillips, Anthony Skelton, andreferees of this journal for detailed comments on earlier drafts, to Jeff Verman and SandyVettese for conversations, and to audiences at Manitoba and at ISUS New York 2012.

Philosophical Review, Vol. 123, No. 2, 2014

DOI 10.1215/00318108-2315306

q 2014 by Cornell University

173

General Good: “As a rational being I am bound to aim at good gener-

ally,—so far as it is attainable by my efforts,—not merely

at a particular part of it” (1981 [1907], 382).

From these axioms Sidgwick claims to deduce:

Benevolence : “Each one is morally bound to regard the good of any other

individual as much as his own, except in so far as he judges it

to be less, when impartially viewed, or less certainly know-

able or attainable by him” (1981 [1907], 382).1

Some think that this is Sidgwick’s main argument for utilitarian-ism, and indeed a convincing argument (P. Singer 1974, 516, 517; 2005,351). Others find the main argument instead in a coherentist argumentthat presents utilitarianism as the product of a cleaned-up commonsensemorality.2 Oddly, it is fairly rare for anyone to give careful attention to theargument from the axioms.3 This is especially odd given the accolades theMethods has received—Broad (1930, 143) finds it “the best treatise onmoral theory that has ever been written,” Parfit (2011, 1:xxxiii, xl) “thebest book on ethics ever written,” “the book it would be best for everyoneinterested in ethics to read, remember, and be able to assume that othershave read.” The lack of attention may be, in part, because Sidgwick him-self admits that the argument does not work against an egoist who refusesto take up the point of view of the universe. But Sidgwick does think thatthe argument works against the commonsense moralist. He thinks a com-monsense moralist must agree with the axioms and so with benevolence asone principle and that there are no other axioms that pass his tests andcould conflict with these (1981 [1907], 421). Hence it is worth askingwhether Sidgwick has a good argument from the axioms againstnonegoists.

The axioms face challenges from two sides.First, Sidgwick thinks that he has no more reason to suspect the

egoist of error than himself. By test (4), the egoist must not deny the

1. I put aside the issue of whether benevolence has the same status as the other axioms.Sidgwick first presents it as deduced from some of them. If this is the only way of arriving atit, it fails test (2). But Sidgwick treats it as having the same status. Perhaps he thinks that itcan be arrived at just by understanding it, as well as by being inferred. For one discussion,see Schneewind 1977, 294–95.

2. For this debate, see P. Singer 1974; Skelton 2010; Sverdlik 1985; Brink 1994;Shaver 1999, 62–71; Schultz 2004, chap. 4; Phillips 2011, chap. 3.

3. I note some of the careful treatments below, mainly in notes.

R O B E R T S H A V E R

174

axioms. But it would seem that an egoist would reject general good andbenevolence.

Second, Sidgwick thinks that he must show that the commonsensemoralist agrees to the axioms. Personal irrelevance and benevolence seem tosay that the only reason for departing from being indifferent to who getssome good, or from pursuing the good of another just as one pursuesone’s own good, is that more good would be produced. But the common-sense moralist will not agree that this is the only reason. The common-sense moralist thinks there are many reasons—say, I have made a promiseor owe gratitude to one party, or one party deserves the good.

In reply to the threat of an egoist’s disagreement, I argue, in sec-tion 1, that not only personal irrelevance but all the axioms above should beread as having as their antecedent “from the point of view of the uni-verse.” I reply to the worry that, if so, the axioms become tautologous. Ialso reply to the worry that, since the egoist now agrees with the axioms,there is no dualism of practical reason.

In reply to the threat of a commonsense moralist’s disagreement,I argue, in section 2, that each axiom states one duty. None states that it isthe only duty. The argument against the commonsense moralist concernsnot benevolence but whether there are further duties that pass the tests.I reply to the worry that, if so, Sidgwick is wrong to conclude that theaxioms by themselves yield consequentialism. I also reply to the worry thatbenevolence is not even one duty.

In section 3, I note what Sidgwick needs to move from benevolence

to consequentialism. It is especially important for him to be able to rejectduties other than benevolence, where these duties are understood in effectto be prima facie duties. I raise the worry that here Sidgwick is unfair tocommonsense morality and argue that he does not have wholly successfulreplies. In an appendix, I review some of the textual changes Sidgwickmade.

Some preliminaries: I ignore the hedonist side of utilitarian-ism, as Sidgwick does in Methods 3.13. The issue is the argument forconsequentialism.

I take the axioms presented in 3.13.3 as canonical. I ignore Sidg-wick’s sloppy labeling elsewhere of (for example) egoism and utilitar-ianism as axioms.4

4. For discussion and references to the many commentators who make the slop-piness point, see Schneewind 1977, 304–8; Shaver 1999, 74–77, 79–80. I put aside theissue, noted to me by David Phillips, of whether there is, or ought to be, an axiom of


175

In interpreting Sidgwick, my aim is to have the axioms pass histests. Many interpreters seem not to have this aim. Many (understand-ably) aim to have the axioms yield consequentialism. But that doomsthem to fail the agreement test. The axioms would fare no better thancommonsense morality. Sidgwick’s best hope is to find axioms that passthe tests and then argue that additional considerations support conse-quentialism. This mimics his strategy of adding to consequentialism con-siderations (namely, hedonism) that support utilitarianism. This strategymotivates the main distinctive features of my interpretation—stressingthat the axioms include reference to the point of view of the universe,that they do not claim to state our only duties, and hence that themain argumentative action consists in moving from the axioms toconsequentialism.

I say little about egoism other than arguing that egoists can agreewith the axioms. If there is a Sidgwickian argument against egoism, it liesnot in the axioms, but in the worries that egoism is arbitrary or that“Humean” views of personal identity undercut it—worries raised by Sidg-wick (1879, 106–7; 1981 [1907], 418–19) and, perhaps prematurely,rejected.

1

It seems that egoists would deny general good and benevolence. Personal irrel-

evance, however, is worded to avoid this. It states what follows from takingup the point of view of the universe. Egoists could agree about whatfollows since they do not take up this point of view.5

This reply faces two objections: it threatens to make personal irrel-

evance tautologous, and it does not save general good or benevolence.Consider the tautology worry first. Sidgwick (1981 [1907], 374–

75) starts the axioms chapter by cautioning against “sham-axioms” that“appear certain and self-evident because they are substantially tautologi-

egoism (or something close to egoism) outside 3.13. For an argument in favor, see Phillips2011, 126–33, 138–39; for an argument against, see Shaver 1999, 83–98.

5. Broad (1930, 244–45) notes that the egoist can agree with personal irrelevance by“remark[ing] that, after all, he is not the Universe, and therefore it is not obvious that heought to take the Universe’s point of view.” One might object that egoists would say thatonce they leave their own point of view, there is nothing they ought or ought not to do.(I owe this objection to Jack Bailey.) But egoists need not say this, and saying it seems odd.Egoists, for example, would have to deny that if I take up the point of view of being yourfriend, I ought to promote your good.


176

cal.” He catalogs these in Plato, Aristotle, the Stoics, and Butler, pokingfun at “it is right to act rationally,” “it is right that the Lower parts of ournature should be governed by the Higher,” “we ought to give every man hisown,” and “live according to Nature” (375–78). When formulating oneaxiom, he discards an attempt because “it does not clearly avoid tautolo-gy” (381). He prefaces his axioms by wanting “self-evident moral prin-ciples of real significance” rather than “doctrines that lead us round in acircle” (379). And after giving his axioms, he compares them to one ofEuclid’s axioms, “if equals are added to equals the wholes are equal,”which he seems to find synthetic (383; also 507).6 But one might worrythat personal irrelevance is really a sham. If the point of view of the universejust is the point of view from which the good of any one is no more impor-tant than the good of any other, personal irrelevance seems tautologous.

Sidgwick could make two replies.7

First, personal irrelevance is best read as having a consequent thatmakes an “ought” claim: one ought to be indifferent between distri-butions that result in the same amount of good (though not betweendistributions that result in different amounts of good). The antecedent ofpersonal irrelevance says that I have adopted the point of view of the uni-verse. According to Sidgwick (1981 [1907], 98; 1879, 107–8; 1877c, 412),no “ought” claim is entailed by strictly nonnormative claims. If so, forSidgwick, the consequent does not follow from the antecedent. Personal

irrelevance is not, then, tautologous in the way that, for example, “it is rightto act rationally” is: “I have adopted the point of view of the universe” and“I ought to be indifferent between distributions that result in the sameamount of good (though not between distributions that result in differ-

6. Sidgwick (1905, 45) writes that “geometry as commonly accepted does giveus . . . synthetical universals: I know that all triangles inscribed in a semicircle must beright-angled triangles, and I could not obtain this knowledge by mere analysis of thenotions of ‘semicircle’ and ‘right-angled triangle’.” This does not, however, directlyaddress the axioms. Elsewhere, Sidgwick (1883, 80) finds Kant guilty of a “hasty extensionto Arithmetic of a view originally formed by a consideration of Geometry,” which can beseen in his “famous distinction between ‘analytical’ and ‘synthetical’ propositions.” Sidg-wick goes on to argue that arithmetic is analytic, perhaps implying that geometry is not.

7. At one point, Sidgwick (1877c, 411) suggests that either personal irrelevance orbenevolence is analytic: “That one ought not to prefer one’s own good to the greater good ofanother . . . is evolved immediately out of the notion of ‘good’ . . . if this notion is usedabsolutely.” This cannot be his considered view, unless he takes up the suggestion made inthe next note.


177

ent amounts of good)” are not mere verbal equivalents, unlike (for Sidg-wick) “right” and “rational” (Sidgwick 1981 [1907], 375).8

This is not the usual reading of personal irrelevance. Many readpersonal irrelevance as making a point only about goodness: the goodnessof some good does not vary with mere changes in who possesses the good(Broad 1971 [1953], 263; Schneewind 1977, 299, 301; Skelton 2008,190).9 But to say that the good of A is no more important than thegood of B seems to say that one ought to be indifferent between them.And to add “unless more good is produced by giving to A” is to add a pointabout how goods ought to be distributed. The “unless” clause could notbe a condition for the truth of the claim that the goodness of some gooddoes not vary merely with who possesses the good. That claim is true evenif more good would be produced by giving the good to A. Personal irrele-

vance, then, claims more than that the goodness of some good does notvary with mere changes in who possesses it. It claims that I ought to be

8. This leaves open the status of personal irrelevance. (I owe the raising of this issue toan anonymous referee.) Sidgwick (1981 [1907], 353, also 229) allows unobvious analyses.He suggests, for example, that “the uninstructed majority of mankind could not define acircle as a figure bounded by a line of which every point is equidistant from the centre: butnevertheless, when the definition is explained to them, they will accept it as expressing theperfect type of that notion of roundness which they have long had in their minds.” Personal

irrelevance might, then, be analytic, though not like the useless “it is right to act rationally,”or it might be synthetic a priori. Given Sidgwick’s view on “is” and “ought,” I think heclearly favors the latter, but in the context of the tautology objection, this does not matter.One thing that the debate over “every cube has twelve edges” and the like made clear isthat it is hard to tell whether some claims are unobvious analyses or synthetic a priori. Theimportant distinction is between useful and useless claims.

9. Broad thinks that an egoist might agree with personal irrelevance even if “point ofview of the universe” is cut. Egoists can agree that the good of any one is of no moreimportance than the good of any other, in the sense that, say, five units of pleasure foranother really is more good than four units of pleasure for themselves. On this suggestion,Sidgwick’s careful inclusion of “from the point of view of the universe” does no work. Theaxiom is consistent with egoism because the egoist can hold that there is no duty toproduce more good rather than less. This simply moves the problem of passing theagreement test from personal irrelevance to general good. See Broad 1930, 243–45; 1968[1942], 45; 1971 [1953], 263–65. Schneewind objects that Broad’s interpretation doesnot fit Sidgwick’s view of how the debate with the egoist runs. This debate turns on takingup the point of view of the universe, rather than on arguing for a duty to produce moregood rather than less (Schneewind 1977, 365n2). But since I shall argue that general good

should be understood as having “from the point of the view of the universe” as its ante-cedent, Broad would have an explanation.


178

indifferent between equal distributions of good, though not betweenunequal ones.10

This interpretation, on which personal irrelevance makes an “ought”claim, is supported by Sidgwick’s treatment of the parallel axiom

Temporal irrelevance : “the mere difference of priority and posteriority in

time is not a reasonable ground for having more

regard to the consciousness of one moment than

to that of another.”

In alternate formulations and applications, Sidgwick (1981 [1907], 381;also 124n1) says that “hereafter as such is to be regarded neither more norless than Now” or “a smaller present good is not to be preferred to agreater future good” or “present pleasure or happiness is reasonably tobe foregone with the view of obtaining greater pleasure or happinesshereafter.” Temporal irrelevance does not say only that the goodness ofsome good does not vary with mere temporal changes. It also says thatone ought to ignore mere changes in temporal position.11

The second reply to the tautology charge turns on glossing “pointof view of the universe.” When Sidgwick explains how it is reasonable foran egoist to avoid the argument for utilitarianism—and so what it is not totake up the point of view of the universe—he writes that “the distinctionbetween any one individual and any other is real and fundamental, andthat consequently ‘I’ am concerned with the quality of my existence as anindividual in a sense, fundamentally important, in which I am not con-

10. For other interpretations (though not explicitly of personal irrelevance) thatascribe ought-claims, see Rashdall 1913, 62; 1907, 1:185; Hayward 1901, 145. Rashdallgives an interpretation like Broad’s in 1907, 1:90–91, 147, and in 1914, 41–42. At onepoint, he argues (not as an objection) that the best version of the axiom—“one man’sgood is of as much intrinsic worth as the like good of another”—is analytic since “thatwhich I recognize as having value in one man I must recognize as having the same value inanother, provided it is really the same thing that is implied in the assertion that it hasvalue” (Rashdall 1907, 1:147). But he does not ascribe this version to Sidgwick (Rashdall1907, 1:147n; see note 22 below).

11. This interpretation may also be supported by what may be an alternative state-ment of personal irrelevance : “one ought not to prefer one’s own good to the greater goodof another . . . is [a principle] evolved immediately out of the notion of ‘good’ or ‘desir-able’, if this notion is used absolutely; as it then must mean ‘desirable from a universalpoint of view’, or ‘what all rational beings, as such, ought to aim at realising’” (Sidgwick1877c, 411). One might read this as claiming that from the point of view of the universe, Iought not to produce the smaller good, which is close to personal irrelevance. But if, as thefinal clause suggests, this involves thinking that I ought to bring about the good, theprinciple may be better identified with benevolence.


179

cerned with the quality of the existence of other individuals” (1981[1907], 498). I take the point of view of the universe to be a point ofview from which I have no special concern for myself. This seems to fitSidgwick’s (1877c, 411; 1981 [1907], 421) examples of taking up thepoint of view of the universe—one does so when one thinks that “thedesign of the creator of the world is to realise Good” or that “‘naturedesigned him to seek his own happiness’.” Neither the creator nor naturehas special concern for me.12

On this reading, personal irrelevance is not tautologous. Personal

irrelevance says not only that I should be indifferent to who has somegood, but also that I should depart from indifference when more goodwould be produced by distributing to one person over another. Presum-ably this addition is also contingent on adopting the point of view of theuniverse. (It would be odd to think that the addition lies outside the scopeof taking up the point of view of the universe, so that “if I adopt the pointof view of the universe, then the good of one is of no more importancethan the good of another” is itself conditional on it’s not being true thatmore good would be produced by distributing to one person ratherthan another.) But “I have no special concern for myself” and “I oughtto depart from indifference when more good would be produced” arenot mere verbal equivalents. However “point of view of the universe” isdefined, it seems intended to exclude only egoistic considerations.

Even if one disregards Sidgwick’s “unless . . . there are specialgrounds for believing that more good is likely to be realised,” the samereply holds. “I have no special concern for myself” is not a mere verbalvariation on “I ought to be indifferent between equal amounts of good.”As the objection considered in the next section indicates, there are possi-ble grounds of preference between equal amounts of good that do notrest on a special concern for myself.

(The tautology worry can also be raised for general good, uncon-nected to adopting the point of view of the universe. One might worrythat general good is a tautology on the ground that Sidgwick [1981 (1907),112, 381; 1877c, 411; 1902, 331] defines “good” as what one has reason toaim at.13 Since Sidgwick [1981 (1907), 381] has, one page before, worriedthat “‘one ought to aim at one’s own good’ . . . does not clearly avoid

12. Similarly, Schneewind (1977, 299) and Parfit (2011, 2:447) take the point of viewof the universe to be an impartial one.

13. For the point (not presented as an objection), see Phillips 2011, 124. Paul Bernaysmakes the same point as an objection (quoted in Schneewind 1977, 307–8).


180

tautology, since we may define ‘good’ as ‘what one ought to aim at’,” it isodd that he does not raise the worry here. But the worry seems misguidedin both cases. Sidgwick [1981 (1907), 112] defines “ultimate good on thewhole for me” as “what I should practically desire if my desires were inharmony with reason, assuming my own existence alone to be con-sidered.” “I ought to aim at my own good” becomes “I ought to aim atwhat I ought to desire assuming my own existence alone to be con-sidered.” This is not a tautology. Similarly, Sidgwick [112] defines “‘ulti-mate good on the whole’, unqualified by reference to a particular subject”as “what as a rational being I should desire and seek to realize, assumingmyself to have an equal concern for all existence.” Suppose, as is plau-sible, that “ultimate good on the whole, unqualified by reference to aparticular subject” is the same as “general good.” “I ought to aim at thegeneral good” becomes “I ought to aim at what I ought to desire assumingmyself to have an equal concern for all existence,” which again is nottautologous.)

Even if the tautology worry is avoided, however, there remains theworry that Sidgwick does not affix “from the point of view of the universe”to general good or benevolence. But there is evidence indicating exactly thatthis is his intention.14

In the second and third editions, where Sidgwick first introducesgeneral good, he writes that “as rational beings we are manifestly bound toaim at good generally, not merely at this or that part of it; we can onlyevade the conviction of this objectivity by denying that there is any suchuniversal good” (Sidgwick 1877b, 355; 1884, 381). The “we can onlyevade . . . ,” dropped in later editions, makes being bound to aim at thegeneral good conditional on thinking that “there is any such universalgood.” This conceptual claim is dubious—an egoist seems capable of

14. There is also evidence against it since Sidgwick sometimes seems to treat general

good as inconsistent with egoism—although given the agreement test, this must be slop-piness. In the second and third editions, he claims that one can deny aiming at goodgenerally only by “denying that there is any such universal good,” which sounds just likewhat he supposes the egoist can say to avoid the argument for utilitarianism. Sidgwick1877b, 355; 1884, 381. He also contrasts “aiming at happiness generally” with “aiming atone’s own” (Sidgwick 1981 [1907], xii). Later, he writes of the “‘sense of the ignobility ofEgoism’ . . . which . . . is the normal emotional concomitant or expression of the moralintuition that the Good of the whole is reasonably to be preferred to the Good of a part”(500). “Intuition” suggests that Sidgwick is referring to one of his axioms. He might bereferring to personal irrelevance since that is introduced by noting parts and wholes (382).Or he might (more naturally) be referring to general good since that tells one to prefergood generally to a part of it.


181

talking about the universal or general good, or of recognizing that somestates of other people are goods, while not being committed to thinkingthat he or she must aim at it or them. Perhaps this explains why Sidgwickcut the addition to general good after the third edition, as part of his revis-ing away from talk of what concepts entail.15 But general good could insteadnote a consequence of adopting the point of view of the universe. Anegoist could then agree to it, but refuse to take up this point of view.16

This may be implicit in later editions. Sidgwick begins the axiomwith “as a rational being I am bound. . . . ” “As a rational being” seemssuperfluous. But at one point Sidgwick (1877c, 411) treats “what allrational beings, as such, ought to aim at realising” as equivalent to “desir-able from a universal point of view.”

There are two arguments for reading benevolence in the same way.First, personal irrelevance contains as its antecedent “from the point

of view of the universe.” I have argued that general good should be read asalso having this antecedent. Sidgwick deduces benevolence from personal

irrelevance and general good. One would think that benevolence should havethe same antecedent.

Second, although Sidgwick almost always omits this antecedentin stating benevolence, on occasion something very close to benevolence hasit: “No doubt it was, from the point of view of the universe, reasonable toprefer the greater good to the lesser, even though the lesser good was theprivate happiness of the agent”; “I certainly could will it to be a universallaw that men should act in such a way as to promote universal happiness;in fact it was the only law that it was perfectly clear to me that I coulddecisively will, from a universal point of view” (Sidgwick 1981 [1907], xx,xxii).17

15. For evidence of this trend, see the appendix.16. This is similar to Schneewind’s (1977, 363) suggestion that general good can be

read so that an egoist can accept it if it says only that I ought to aim at more good ratherthan less, where my point of view sets the relevant goods. Elsewhere, however, Schneewind(1977, 299–300, also 301, 307) gives a different weakened reading of general good: it saysthat “if it is the goodness of some state of affairs which is an agent’s reason for an action,then the goodness of a precisely similar state of affairs is also a reason for action, regardlessof the ownership of what is good.” General good says more than this; it says that there is areason to promote good states of affairs (wherever they occur). For a similar point aboutSchneewind’s treatment of the axioms, see Irwin 2009, 493–94.

17. Parfit (2011, 2:554) gives, in passing, the same interpretation. If the principlequoted earlier from “Mr. Barratt” is read as benevolence, this is another place in which “fromthe point of view of the universe” is attached to it (see note 11).


182

I conclude that Sidgwick can get the egoist’s assent by making theaxioms conditional on taking up the point of view of the universe.

One might object that, since Sidgwick believes that there is a dual-ism of practical reason, an interpretation on which egoists agree to theaxioms cannot be right.18 (I take the dualism to be that egoism andutilitarianism are equally well justified and sometimes conflict in thesense of giving guidance toward two actions such that if one is performed,the other cannot be.) But, first, Sidgwick does not explicitly say that ego-ists reject any of the axioms;19 there is no hint in the chapter on theaxioms that anyone disagrees with them; and if Sidgwick did allow dis-agreement, his crucial claims for the superiority of his axioms over com-monsense morality (and their status as similar to mathematical axioms)would be false (Sidgwick 1981 [1907], xxii, xxiii, 383, 421, 496, 507; 1876,564). What is explicit is that the egoist avoids utilitarianism by refusing tosay that “his happiness . . . is Good, not only for him but from the point ofview of the Universe” (Sidgwick 1981 [1907], 420). The egoist “may avoidthe proof of Utilitarianism by declining to affirm” that “his own greatesthappiness is . . . a part of Universal Good” (497–98).

Second, there is a gap between the positions that constitute thedualism—egoism and utilitarianism—and the axioms.20 The relevant

18. Versions of this objection were raised by Tom Hurka, Anthony Skelton, and ananonymous referee.

19. He does come close for general good; see note 14. He also comes close in his sketchof the history of the Methods. There he claims that because of the conflict between egoismand utilitarianism, “I was forced to recognise the need of a fundamental ethical intuition”(Sidgwick 1981 [1907], xviii, also xxi). He seems to identify this intuition as “That arational agent is bound to aim at Universal Happiness” (xxi). (This is inserted in squarebrackets into Sidgwick’s text, presumably by E. E. Constance Jones, but it seems a correctinsertion, given Sidgwick’s reference there to his discussion of More and Clarke. Thatintuition, unlike my version of benevolence, would conflict with egoism. But since Sidgwickalso gives, in the same sketch, versions of benevolence that do seem to include the conditionthat one take up the point of view of the universe (xx, xxii, quoted above), perhaps he isbest read as saying in the sketch that an intuition is needed to choose utilitarianism overegoism, although that intuition is not the unimpeached axiom (my benevolence) found inMethods 3.13. One needs the intuition that one ought to take up the point of view of theuniverse—an intuition Sidgwick does not claim passes his tests.

20. See the references to Schneewind and Shaver in note 4. Seeing a gap helps tosolve a puzzle raised by Phillips. He notes that in Methods 4.2, the “proof” of utilitarianismproceeds by ad hominem arguments directed at the egoist and the commonsense mor-alist. The application of the tests for highest certainty in 3.13 seems to have been for-gotten (Phillips 2011, 65–76). If there is a gap between the axioms and utilitarianism, thisis understandable: the tests secure the axioms but not utilitarianism.


183

axioms include the condition that one takes up the point of view of theuniverse; they state only one set of duties; and they do not specify thegood. (For further gaps, see section 3.) This allows for the possibility thatall agree to the axioms and there is still a dualism between egoism andutilitarianism. Admittedly, Sidgwick (508) is misleading when, in statingthe dualism, he writes that we must “admit an ultimate and fundamentalcontradiction in our apparent intuitions of what is Reasonable in con-duct” and that “the apparently intuitive operation of the Practical Reason,manifested in these contradictory judgments, is after all illusory.” If benev-

olence includes the condition that one takes up the point of view of theuniverse, there is no contradiction between it and any egoist axiom(if there is one). But even if one thought that there were unconditionalconsequentialist and egoist axioms, Sidgwick is misleading here sinceeven then there is no contradiction—again in the sense of guidancetoward two actions such that if one is performed, the other cannotbe—without adding to the axioms the claim that what maximizes mygood may not maximize the general good (a claim Sidgwick justifies bymeans of his particular view of the good, as pleasure, and an empiricalclaim about what maximizes pleasure).21 On my view, similarly, conflictoccurs only when more is added, and in particular when it is added thatsomeone does, and another does not, take up the point of view of theuniverse, and that either choice is reasonable. This could be described asa contradiction in our judgments about “what is reasonable in conduct”and as a failure of practical reason—so perhaps Sidgwick is not so mis-leading. In any case, his pessimistic conclusion remains.

2

Personal irrelevance says that when I have no special concern for myself,I ought to be indifferent between distributing x to A and x to B, unlessmore good would be produced by distributing to one of us. The common-sense moralist might object that there are reasons for departing fromindifference that do not depend on special concern for myself or on

21. On some views of the dualism, nothing need be added to secure one sort ofcontradiction: suppose egoism says that the only right-making feature is the happinessof the agent, and utilitarianism says that the only right-making feature is the generalhappiness. This reading, however, cannot explain how Sidgwick (1981 [1907], xxii, also508) thinks the “moral government of the world” would avoid the dualism. For variousreadings of the dualism, see Schneewind 1977, 372–74; Shaver 1999, 78–80; and Phillips2011, 134–40, all of which give references to further literature.


184

producing more good. Perhaps I ought to distribute to A, who is virtuous,rather than to B, who is vicious, even if no more good would be produced.Perhaps I should distribute to A, to whom I have promised the item, or towhom I owe gratitude, rather than to B, even if no more good is produced.Personal irrelevance says that I ought to be indifferent between A and B. Thesame points can be made for benevolence.22

One could reply to the commonsense moralist that more goodcomes of distributing things to A. Perhaps (for example) distributing to Aencourages virtue and so produces more goods such as virtue and hap-piness in the long run. Perhaps the deserving getting what they deserve isitself a good. But some will still disagree: I might say that I ought todistribute to A, without thinking that rewarding A produces more goodthan not doing so; I might think (with H. A. Prichard and E. F. Carritt)that there is a duty of justice that is independent of any duty to producemore good. The standard reply on behalf of the consequentialist is thatthis position has no clear advantage over one that admits goods such asdistribution according to virtue. But this is rejected by those who thinkthat, say, I ought to give the deserving what they deserve even when theresult is that five other people fail to give the deserving what they deserve.Considerations of goodness favor my not distributing according to des-ert here.

There is a different reply to the objection that there are reasons fordeparting from indifference that do not depend on special concern formyself or on producing more good. This reply relies more on Sidgwick’stext. Personal irrelevance claims that one of my duties is to be indifferentbetween distributing to myself and distributing to another, unless moregood would be produced by distributing to one of us. As far as personal

irrelevance goes, I may have other duties, such as to distribute according todesert or gratitude. (Again, the same reply can be made for benevolence.Henceforth I omit this rider.)

This interpretation can be supported in three ways.First, Sidgwick (1981 [1907], 421) thinks that the argument from

the axioms “only shows the Utilitarian first principle to be one moral

22. Rashdall (1907, 1:234–42, 262–63) raises a related objection, concentrating oncases in which A can attain a “higher” good than B and cases of differences in merit. Hethinks the axiom should read “Every man’s good to count as equal to the like good of everyother man,” to allow for distributing more to one with the capacity for higher goods(1:240). He seems to omit the “unless,” at one point interpreting Sidgwick as holdingthat “I ought always, so far as possible, to produce an equal distribution of good among allthe people who can be affected by my action” (Rashdall 1914, 42).


185

axiom: it does not prove that it is sole or supreme. The premises fromwhich the Intuitionist starts commonly include other formulae held asindependent and self-evident.”23 This suggests that personal irrelevance isnot intended to rule out all reasons for departing from indifference otherthan the amount of good produced. If it did, it by itself would imply thatmany other principles are false. One could not, for example, hold thatalthough giving the resource to A produces less good than giving it to B, itshould be given to A because A deserves or has been promised it. ThatSidgwick does not consider this sort of objection to personal irrelevance isevidence that he did not intend it to rule out all reasons for departingfrom indifference other than bringing about more good.

Similarly, Sidgwick notes elsewhere that commonsense moralistsagree at least with benevolence.

No Intuitionist ever maintained that all our conduct can be ordered

rightly without any calculation of its effects on human happiness. On

the contrary, this calculation . . . is expressly inculcated by the maxims

of Prudence and Benevolence, as commonly understood. (Sidgwick

1876, 564)

The judgment of good and ill is . . . ethical. And this is admitted by Com-

mon Sense. In the recognition of the virtues of Prudence and Benevo-

lence it is allowed to be (1) right . . . that a man should prefer greater to

lesser good in his own case, and (2) virtuous to promote good generally

for others. . . . And the only difference between Common Sense and

Utilitarianism as regards the outward act is, that in the view of Common

Sense these rules are but rules among others, while in the view of Utilita-

rianism, the rule of promoting the greatest good is supreme. (Sidgwick

1902, 331–32)

Second, Sidgwick (1981 [1907], 384–86) claims that Clarke andKant agree with his axioms. This may be true if personal irrelevance states

23. One might object that Sidgwick writes that “as addressed to the Intuitionist, thisreasoning only shows . . . ” Perhaps Sidgwick does suppose he has shown that there are noreasons for departing from indifference other than the amount of good produced. Hegrants here that there might be other duties only in the course of an ad hominem argu-ment addressed to the intuitionist. (I owe this objection to Tom Hurka.) But—puttingaside the vexed issue of whether Sidgwick’s argument is just ad hominem—in the textsgiven below, where there is no suggestion of a concession made just for the sake of argu-ment, Sidgwick again claims that the commonsense moralist would agree with benevolence.


186

one duty. It is false (at least for Kant) if personal irrelevance rules outall reasons for departing from indifference other than the amountof good.24

Third, in the first edition’s discussion of (what becomes) personal

irrelevance, Sidgwick does not say that the only reason for departing fromindifference is that more good would be produced. The emphasis is ondenying egoism. A “rational being . . . cannot regard the satisfaction of hisown personal desires as intrinsically more desirable (as being his own)than the satisfaction of the desires of any other person” (Sidgwick 1874,363, also 360, 364). The chapter concludes by noting that “we have onlyevolved the suppression of Egoism” (366). In later editions (and else-where), when Sidgwick glosses what is self-evident, the gloss similarlydoes not make the claim that the only reason for departing from indif-ference is that more good would be produced:

I ought not to prefer my own lesser good to the greater good of another.

(Sidgwick 1981 [1907], 383; also 1876, 565; 1877c, 411; 1889, 483)

The good . . . of any one individual must as such be an object of rational

aim to any other reasonable individual no less than his own similar good.

(Sidgwick 1981 [1907], 385)

[One is to seek] others’ good no less than one’s own. (392)

Universal Benevolence . . . sets before each man the happiness of all

others as an object of pursuit no less worthy than his own. (496; also

1877a, 31)

I ought to sacrifice my own happiness, if by doing so I can increase the

happiness of others to a greater extent than I diminish my own. (Sidgwick

1889, 483)

My own good is no more to be regarded than the good of another. (484)

Sidgwick (1981 [1907], 420) glosses the axioms chapter as giving an argu-ment “addressed to Egoism.”

Why, then, does Sidgwick include the “unless” clause, given that hedoes not intend personal irrelevance to rule out departing from indiffer-ence on grounds such as desert or fidelity?25 I think he is working toward

24. Irwin (2009, 499, 501, 504–5, 510, 512–13, 516, 518) makes this point, though asa criticism of Sidgwick.

25. Some commentators, such as Schneewind (1977, 294, 296), simply leave offthe “unless” clause. When he first quotes personal irrelevance, he adds—not as part of


187

stating the main duty concerned with promoting the good. He is denying,in one way, a conclusion one might be tempted to draw from the claim ofequal importance, just as he denies a conclusion one might be tempted todraw from temporal irrelevance. There he writes that “it is not, of course,meant that the good of the present may not reasonably be preferred tothat of the future on account of its greater certainty: or again, that a weekten years hence may not be more important to us than a week now,through an increase in our means or capacities of happiness” (Sidgwick1981 [1907], 381).26

Sidgwick is not denying that there might be other reasons for notbeing indifferent. These would be found in other duties. Perhaps com-mon sense thinks I ought to favor (say) my parents, when that does notbring about the greatest good, but it does so only in virtue of duties otherthan benevolence, such as gratitude (248, 259).27 In the case of duties of

the axiom—that “[a]s in the case of [temporal irrelevance], quantitative differen-ces . . . are discounted here” (294). Irwin (see below), Lacey (1959, 218), and Skelton(2008, 187) include the “unless.” Phillips (2011, 121), Seth (1901, 182), Broad (1971[1953], 262), Parfit (2011, 2:447), Rashdall (see note 10), and Wallace (1988, 42) donot. A similar split occurs on the “except” clause in benevolence.

26. Irwin (2009, 498–99) notes the importance of the “unless” addition. (He seemsto be the only one to do so.) He treats it as a momentous shift between the first edition andlater ones, a shift to an argument that takes maximizing the good to be the fundamentalaxiom. (Skelton gives some textual evidence for thinking Sidgwick finds maximizing self-evident; see 2008, 188, 193–94.) Irwin (2009, 504, 510, 512–13) objects that Sidgwickcould not expect agreement to taking maximizing the good to be the fundamental axiomand gives various textual objections to doing so (504, 507–8). This, along with the textualevidence I give (from editions other than the first), favors a reading on which personal

irrelevance is just one duty. Irwin may want his reading because he supposes that Sidgwickmust prove consequentialism directly from the axioms, rather than from a combination ofthe axioms and an attack on any alternative axioms. He objects to a similar combinationsuggestion by Schneewind (1977, 307, 308) that “failure to find any other universal right-making property does not show that maximizing utility is a viable candidate; perhaps weought to conclude that we cannot find any single right-making property” (Irwin 2009,498n30). But this objection is avoided if a case is first made for maximizing utility, andthen a case is made against any other property (see section 3 below).

27. Ross (1939, 76) writes that it is on the prima facie duty of gratitude that “ourspecial duty to parents and friends in the main depends.” He links the duty of parents tochildren to promising (273). Broad may think that special relationships are sufficient toground duties, without reference to further duties. He writes that common sense “holdsthat . . . special relationships are the ultimate and sufficient ground for . . . specially urgentclaims on one’s beneficence” (Broad 1971 [1953], 280). “Prima facie the special urgency ofthe claims of certain others upon one’s beneficence seem to be founded directly on certainspecial relationships of those others to oneself ” (269, also 265). Sidgwick may disagree


188

parents to children, Sidgwick (347) suggests (and rejects) a “special duty. . . aris[ing] from the fact that I have brought them into being.” Benevo-

lence does not rule out these duties.One might object that there is no such duty to promote the good.

Consider cases familiar from the antiutilitarian literature, such as Scan-lon’s World Cup case: A is trapped in a TV transmission unit, sufferingpainful shocks; A can be removed, but only by stopping transmission forfifteen minutes; since a World Cup match is on, with millions of viewers,more good would be produced by leaving A in until the game ends in anhour; doing so is clearly wrong (Scanlon 1998, 235).

Sidgwick can reply, however, that personal irrelevance does not saythat favoring the viewers is my duty all things considered. Personal irrele-

vance says that it is one duty, one thing I have reason to do. This is hard todeny. Even if the loss of viewing has little significance compared to thepain of the shocks, the loss of viewing still is a loss, something there issome reason to avoid. It is much more plausible to think that the reason toallow continued viewing is outweighed than it is to think that, althoughthe viewing is a good, it is a good there is no reason to bring about(especially, but not only, on a view of the good as what there is reasonto bring about).

Sidgwick could note that his later critics, speaking on behalf ofcommonsense morality, agree. Ross writes that “it seems self-evident thatif there are things that are intrinsically good, it is prima facie a duty to bringthem into existence rather than not to do so, and to bring as much ofthem into existence as possible.” To say that “there is no prima facie duty toproduce as much that is good as we can” is “plainly false” (Ross 1930, 24,25, also 27, 39, 58; 1939, 271, 273). Broad (1971 [1953], 264, also 1930,221) writes that “on any view, one important prima facie duty is to produceand conserve good and to avert and diminish evil.”

One might instead object that “the duty of Benevolence as recog-nised by common sense seems to fall somewhat short” of what benevolence

requires. Common sense favors aiming at “the good of a limited numberof human beings . . . generally in proportion to the closeness of theirconnexion with him” (Sidgwick 1981 [1907], 382, also 431). Sidgwick(252–53, 348–49, 431–32) also notes that one might worry that the

with Broad; this would explain why he so brusquely finds aiming at the happiness ofone’s family, as an ultimate end, arbitrary (Sidgwick 1981 [1907], 10). But even ifBroad is correct, this does not affect agreement on benevolence.


189

ordinary duty of benevolence does not rule out my sacrificing morethan another gains, and that it requires only a small sacrifice for a largegain. One might add that perhaps benevolence is better captured bygiving priority to the worse-off.

These worries are best answered by claiming that the common-sense moralist accepts benevolence as one duty—as Ross and Broad do—whether or not benevolence should be described as the duty of benevo-lence. But I admit that this is not what Sidgwick says. In reply to theimpartiality worry, Sidgwick (1981 [1907], 382, also 327, 431–39;1877a, 31–32) writes that “practically each man, even with a view to uni-versal Good, ought chiefly to concern himself with promoting the good ofa limited number of human beings, and that generally in proportion tothe closeness of their connexion with him.” He then gives

Plain man: “a ‘plain man’ . . . if his conscience were fairly brought to

consider the hypothetical question, whether it would be mor-

ally right for him to seek his own happiness on any occasion if

it involved a certain sacrifice of the greater happiness of

some other human being,—without any counterbalancing

gain to any one else,—would answer unhesitatingly in the

negative” (382)

to confirm that when these practical considerations are absent, commonsense agrees with benevolence.

If Sidgwick’s point is that the prevalence of partial duties does notshow the absence of an impartial duty, he is correct, and it backs up what Itake to be the best answer. But if his point is that the plain man acceptsbenevolence as an all-things-considered duty, Sidgwick is saying somethingboth unnecessary to his argument and probably false.28 Even if plain man

is true, it does not follow that common sense forbids me to favor thesmaller good of, say, my children or my parents, over the greater goodof some strangers. Sidgwick makes his work easier by choosing oneself asthe part.29

28. Parfit (2011, 1:453) thinks plain man is “simply false.”29. Thanks to Nick Novelli for this point. Sidgwick may cheat in a further way. The

plain man is asked whether it is “morally right” to refuse to sacrifice his good for the sake ofthe greater good of another. (Similarly, benevolence concerns what one is “morally bound”to do.) One might agree with plain man when thinking of morality as by definition rulingout special concern for oneself. But this is not Sidgwick’s view: although he expressesdoubt about whether egoism counts as a moral theory, he does not make anything hangon this taxonomic point (Sidgwick 1981 [1907], 119). His concern is with what it is mostreasonable to do.


190

I have argued that benevolence states just one duty. One might try togive Sidgwick a more ambitious view. There are arguments one could givefor moving from adopting the point of view of the universe, understood ina particular way, to being concerned only with amounts of good, all thingsconsidered.

First, in “Mr. Barratt on ‘The Suppression of Egoism’,” Sidgwick(1877c, 411) writes that “one ought not to prefer one’s own good to thegreater good of another . . . is [a principle] evolved immediately out of thenotion of ‘good’ or ‘desirable’, if this notion is used absolutely, as it thenmust mean ‘desirable from a universal point of view’, or ‘what all rationalbeings, as such, ought to aim at realising’.” One might gloss “desirablefrom a universal point of view” as requiring that everyone desire some onestate of affairs. This admits only agent-neutral reasons. One might thenargue that all reasons other than a concern with the amount of good areagent relative. For example, deontological constraints add reasons inaddition to the amount of good, but many have argued that they areagent relative. I see no evidence for this in Sidgwick, however, and think-ing of deontological constraints as agent relative is controversial.30

Second, the same passage from “Mr. Barratt” identifies a “universalpoint of view” with what “all rational beings as such, ought to aim atrealising.” Personal irrelevance thus claims that qua rational being Iought to be indifferent between different distributions that producethe same amount of good, and more generally, that only amounts ofgood are relevant to judging between distributions. All I know aboutmyself when I consider myself merely as a rational being is that I act onwhat I have most reason to do. The good is what it is reasonable to aim at(Sidgwick 1981 [1907], 112, 381; 1902, 331). So I should be concernedonly with amounts of good.

If this is Sidgwick’s thought, there is a familiar objection. He wouldbe “consequentializing”—the good expands to cover anything one oughtto aim at, so there are no considerations other than the amount of good.But this simply relocates the debate between (traditional) consequen-tialists and their opponents. And Sidgwick himself seems to deny thistaxonomy. When he repeats his opening view that moral theories takeeither the right or the good as fundamental, he notes that taking the rightas fundamental has been defeated because he has shown that most duties“contain an implicit subordination” to axioms concerning the good and

30. For a sample discussion, see Kamm 1996, chap. 9.


191

that only the axioms I have discussed, all of which but consistency concernthe good, pass his tests (Sidgwick 1981 [1907], 391). The consequen-tializing tactic would give a much quicker victory.

Putting aside these more ambitious readings, Sidgwick’s argumentfor benevolence goes as follows:

1. From a point of view from which I have no special concern for myself,

one thing I ought to do is be indifferent between distributing x to A

and x to B, unless I could produce more good by one distribution or

the other ( personal irrelevance).

2. From a point of view from which I have no special concern for myself,

one thing I ought to do is aim at the good of everyone (general good).

3. Therefore from a point of view from which I have no special concern

for myself, one thing I ought to do is aim at any other’s good as much as

at my own, unless I could produce more good by one distribution or

the other (benevolence).

This reconstruction gives personal irrelevance and general good separate andnecessary roles. Personal irrelevance claims that I ought to be indifferentbetween equal distributions of good to myself and A, as well as betweenequal distributions to A and B. It also says that, when more good can beproduced by not being indifferent, I need not be indifferent. But it doesnot tell me that I ought to aim at the good of others. General good tells meto aim at the good of others, excluding no one, but does not forbid givingsomeone or some group special weight. (One might think that this isruled out by adopting a point of view from which I have no special con-cern for myself. But I might well favor those close to me without specialconcern for myself.)

One might object that my version of benevolence is far weaker thanSidgwick intends.31 In the next section, I note the many things Sidgwickneeds to argue for to move from benevolence to the consequentialism hewants. Most obviously, he needs to argue in favor of taking up a point ofview from which I have no special concern for myself, in favor of maxi-mizing, and in favor of thinking that benevolence is the only duty. But—sothe objection goes—Sidgwick claims that he has established his conse-quentialism very soon after stating the axioms. He writes that “I find that Iarrive, in my search for really clear and certain ethical intuitions, at thefundamental principle of utilitarianism” (Sidgwick 1981 [1907], 387).“Utilitarianism is thus presented as the final form into which Intuitionism

31. I owe this objection to Tom Hurka and an anonymous referee.


192

tends to pass, when the demand for really self-evident first princi-ples is rigorously pressed” (388). The “Intuitional method rigorouslyapplied yields as its final result . . .Utilitarianism” (406–7). Elsewherehe writes that “considerations of intuitionism in its most philosophicalform . . . would have led me at once to utilitarianism” (Sidgwick 1876, 564).

These claims cannot, however, claim so much. Again, Sidgwickcannot say that consequentialism passes the agreement test. He has justnoted that Clarke and Kant agree with his axioms (Sidgwick 1981 [1907],384–86); if he thinks he has established consequentialism, he cannoteven claim the agreement of those he has just cited as agreeing. Andagain, Sidgwick (421) later notes that the argument from the axioms“only shows the Utilitarian first principle to be one moral axiom: it doesnot prove that it is sole or supreme.” Fortunately, there is some indicationthat Sidgwick is aware that he has established something more modest.Directly before saying that he has arrived at “the fundamental principleof utilitarianism,” he writes that “the axiom of Rational Benevolenceis . . . required as a rational basis for the Utilitarian system.” This is thenglossed as “accordingly, I find that I arrive . . . ” (387). Benevolence, on myreading, is plausibly required as a basis for utilitarianism. Later Sidgwickglosses his argument as showing that “such abstract moral principles as wecan admit to be really self-evident . . . seem required to furnish a rationalbasis for such a [utilitarian] system” (496, also xxii–xxiii). Utilitarianism“rest[s]” on the intuitions (Sidgwick 1877a, 31; also 1981 [1907], xxi). Ithas benevolence as its “first principle” (Sidgwick 1981 [1907], 505). As forthe claim that intuitionism “tends to pass” into utilitarianism, the “tends”suggests that there is some slack between accepting the axioms and adopt-ing utilitarianism, just as my interpretation predicts. (Perhaps, with morestrain, “yields as its final result” also permits some slack.) And the claimthat consideration of intuitionism “at once” leads to utilitarianism is fol-lowed by this explanation: “because I hold that the only moral intuitionswhich sound philosophy can accept as ultimately valid are thosewhich . . . provide the only possible philosophical basis of the Utilitariancreed” (Sidgwick 1876, 564).

3

To reach the maximizing, agent-neutral consequentialism Sidgwick wantsfrom benevolence, he has at least these tasks.

(i) Sidgwick’s strategy is to show that the commonsense moralistmust agree to benevolence, as the utilitarian does, and then argue that


193

commonsense morality contains nothing on a par with benevolence (Sidg-wick 1981 [1907], 421).32 If so, he needs to drop “from the point of viewfrom which I have no special concern for myself” from benevolence. Forutilitarians believe not just that if I have no special concern for myself,I ought to aim at any other’s good as much as at my own (etc.). Theybelieve I have this duty, period.

I think Sidgwick assumes that the commonsense moralist adoptsthe point of view from which one has no special concern for oneself. Hemay claim more than this—that this is on reflection the only point of viewcommon sense adopts. But provided commonsense moralists always takethis point of view to be relevant, they agree that one thing I ought to aim atis any other’s good as much as my own. Even in cases in which they thinkthat this “ought” is outweighed by, say, a claim that I ought to favor myfamily over the greater good of a stranger, they think that there is some-thing to be outweighed. Critics often argue that Sidgwick slides from theclaim that benevolence is one duty to the claim that benevolence is theonly duty. But if Sidgwick (here) only wants benevolence as one duty, andthe critics agree with that claim, there is no objection.33

(ii) Benevolence says that if more good would be produced by dis-tributing to A rather than to myself, it is not the case that I ought to beindifferent between distributing to A and distributing to myself. Strictly, itdoes not make the further claim that when more good would be pro-duced by distributing to A, I should not only favor either A or myself,but in particular ought to favor A. Presumably Sidgwick takes that to beunderstood.

(iii) Benevolence concerns myself and another, but personal irrele-

vance, from which benevolence is deduced, concerns anyone. PresumablySidgwick intends the wider application. Although benevolence notes only

32. This fits Sidgwick’s account of his thought: “I found the axiom I required for myUtilitarianism. . . . [But] why not further? The orthodox moralists such as Whewell . . . saidthere was a whole intelligible system of intuitions. . . . But the result of the examination[of commonsense morality] was to bring out with fresh force and vividness the differencebetween the maxims of Common Sense Morality . . . and the intuitions which I had alreadyattained” (Sidgwick 1981 [1907], xxi–xxii).

33. For example, Broad (1971 [1953], 264) writes that “on any view, one importantprima facie duty is to produce and conserve good and to avert and diminish evil” but readsgeneral good as saying that the “only legitimate ground for devoting more of one’s benefi-cent activities to one person . . . rather than to another . . . is that by doing so one willpromote more good or avert more evil.” For complaints like Broad’s, see Skelton 2011,608; and Phillips 2011, 125–26.


194

the egoist case, the objection he considers to benevolence is that commonsense thinks one ought to favor friends and family (Sidgwick 1981 [1907],382). This understands benevolence as more than antiegoist.

(iv) Sidgwick must argue for maximizing. Benevolence says that myattitude toward another’s good must be the same as my attitude towardmy own good. If I think I ought to maximize my own good, maximizing isthe correct attitude toward the good of each person. Sidgwick assumes,I think, that maximization is the correct attitude to take toward one’s owngood. He does not consider satisficing, or, say, preferring slightly less totalgood with the best parts toward the end to slightly more total good withthe best parts toward the start.34

Sidgwick does not, however, understand “I ought to aim at anyother’s good as much as at my own” as telling me to maximize the goodof each person in the sense that I ought to put each person at a level ofgood such that there is no possible world in which he or she has moregood. In most cases, doing so is impossible. (Say I can save A or save B andC.) This impossibility is what often makes it false that I ought to maximizethe good of each person in this sense—not, as benevolence claims, that Ican produce more good by not maximizing the good of each. Indeed, ifI can maximize the good of each in this sense, producing more good bynot doing so is impossible (for a fixed population). I think, then, thatSidgwick understands “I ought to aim at any other’s good as much as at myown” as telling me to maximize the equal good of everyone.35

It is worth stressing the role of equality. Later Sidgwick (1981[1907], 417) claims that when equal amounts of good could be produced,one should favor an equal distribution over an unequal one. Critics objectthat there is no ground in maximization or utilitarianism for favoring anequal distribution, and that there is no reason to think that an indepen-dent principle of equal distribution need always be subordinate to maxi-mizing the good.36 But the equality claim follows from benevolence : I candepart from maximizing the equal good of each only if a greater totalamount of good can be produced by doing so. Sidgwick’s utilitarianismhas a concern for equal distribution built in. Structurally, it resembles

34. For the latter, see, for example, Broad 1930, 225–26; Irwin 2009, 510–12.35. Irwin (2009, 495) suggests this as what the first edition supports. Gerald Postema

has interpreted Bentham as endorsing a version of this principle. See, for example, Post-ema 1998.

36. See Ross 1939, 71–72 (though he is willing to accept a consequentialist reply,making certain distributions themselves goods, on 73–74); Wallace 1988, 45–48; Irwin2009, 513–16.


195

Rawls’s difference principle, although the condition for departing fromequality is of course different.37

(v) Sidgwick must argue that one thing I ought to do is not tomaximize the equal good of each when I could produce more good bynot doing so. The argument for this parallels the treatment of the WorldCup case.

(vi) Most important, Sidgwick must argue that there are no otheraxioms that pass the tests. This is the job of Methods 3.11.38

Many have charged that Sidgwick is unfair: he holds other pro-posed axioms to a standard higher than that he applies to his own.39 Myreconstruction makes this charge clearly justified—in some cases. Sidg-wick cannot complain that, say, the axiom “One ought to keep one’spromises” is not clear and precise because it fails to tell me what to dowhen keeping a promise conflicts with some other duty. Benevolence fails inthe same way. Nor can Sidgwick argue that “One ought to keep one’spromises” fails the agreement test because there are cases in which, givenother duties, we think we ought to break the promise. The commonsensemoralist can say that there is still a reason to keep the promise, just as

37. Thanks to Emily Muller for pointing out the structural similarity.38. After stating some of the axioms, Rashdall (1885, 202) gives this reconstruction of

the plan of the Methods: “(1) That there are the two above-mentioned formal intuitiveprinciples of Morality; (2) That there are no others: hence the necessity for the elaborateexamination of ‘Common Sense’ Morality.” Rashdall does not specify whether eliminat-ing other principles is needed because the two principles are, in effect, prima facie duties,or because they state duties all things considered that must be shown to meet the agree-ment test. Since Rashdall does not mention agreement, the former is more likely.

39. For discussion of the unfairness charge, see Crisp 2002, 74–75; Skelton 2008,205–7; Skelton 2011, 608; Lillehammer 2010, 371; Shaver 2011. Earlier commentatorswere more impressed by the negative examination of commonsense morality. Hayward(1901, viii, xviii) writes that “Sidgwick’s discussion of this ‘common sense’ doctrine isadmitted by all critics to be extremely able, to be, in fact, the most irrefutable part ofthe book. The student . . . will thus come to see the weakness of popular intuitionism, andthe necessity for a sounder ethical theory. . . . Most of the third book requires no com-mentary whatever; by common consent its chapters are so lucid, and the conclusions theyembody are, for the most part, so indisputable, that even to point out their merits wouldbe to gild refined gold.” Rashdall (1907, 1:83) writes that the “strongest part of Sidgwick’sgreat work consists in its analysis of common-sense Morality. The loose statements ofIntuitionists as to the clearness, certainty, adequacy, and self-evidence of the ordinarilyreceived rules of conduct have never been subjected to so searching, so exhaustive, and soilluminating an examination. That task has been done once for all.” Broad (1952 [1938],64) writes that Sidgwick’s “discussion of the Morality of Common Sense seems to me to bethe best work of the kind since Aristotle’s.”


196

Sidgwick can reply to standard objections to consequentialism that thereis still a reason to bring about the most good.

Reviewing Sidgwick’s examination of commonsense morality isa task for a paper devoted to the unfairness charge. Here I give someexamples—some good for Sidgwick, some not so much.

Sidgwick (1981 [1907], 316, 355) notes that there is controversyabout lying to invalids. The commonsense moralist could reply that thereis always a duty to tell invalids the truth, though it can be outweighed by aduty of benevolence. But Sidgwick also worries that it is not clear whether“it is our actual affirmation as understood by the recipient which weare bound to make correspondent with fact . . . or whatever inferenceswe foresee that he is likely to draw from this, or both” (355, see316–17). This seems to be a problem internal to the duty. But it mightbe met by copying the strategy used to defend benevolence. One mightclaim that according to common sense, there is always a reason to saywhat is true and always a reason to make the inferences the recipientdraws from what we say result in truths.40 There is not one duty with aninternal problem, but rather two unproblematic duties.

Sidgwick’s treatment of “one ought to distribute according todesert” does better (where this is understood as a rival to producingthe good rather than an instance of doing so). Sidgwick (1981 [1907],283–85) objects that

the question . . . recurs: whether . . . we are to apportion the reward to the

effort made, or to the results attained. For it may be said that the actual

utility of any service must depend much upon favourable circumstances

and fortunate accidents, not due to any desert of the agent: or again, may

be due to powers and skills which were connate, or have been developed

by favourable conditions of life, or by good education, and why should we

reward him for these? . . . [Going to moral excellences does not help,

since] it may still be said that good actions are due entirely, or to a great

extent, to good dispositions and habits, and that these are partly inherited

and partly due to the care of parents and teachers; so that in rewarding

these we are rewarding the results of natural and accidental advantages,

and it is unreasonable to distinguish these from others, such as skill and

knowledge, and to say that it is even ideally just to reward the one and not

the other. Shall we say, then, that the reward should be proportionate to

40. Broad (1930, 219) suggests that it is fitting to tell the truth but also fitting to avoidfalse beliefs, so one’s duty is determined by whichever is most fitting. Ross (1939, 81–82)instead argues that veracity requires saying what is true, though utility may favor avoidinginferences to false beliefs.


197

the amount of voluntary effort for a good end? But Determinists will say

that even this is ultimately the effect of causes extraneous to the man’s

self. . . . [T]here seems to be no justice in making A happier than B, merely

because circumstances beyond his control have first made him better. . . .

[I]n any case it does not seem possible to separate in practice that part

of a man’s achievement which is due strictly to his free choice from

that part which is due to the original gift of nature and to favouring

circumstances.41

These objections do not rely on noting conflict with other duties. Nor canone plausibly retreat to thinking that there remains a reason for distrib-uting according to desert. The objections note a problem internal tothe duty.

In the case of promising, some of Sidgwick’s worries can be seen asworries for thinking that there is a reason to keep a promise. He notes thatit is controversial whether one is bound by a promise one makes given thata false statement, or suggestion of one, or concealment, or a falsehoodshared by the promiser and promisee, was one consideration in favor ofmaking it. It is also controversial whether changed circumstances, orchanged beliefs about them, release the promiser. As a special case ofthis, Sidgwick considers that, after making a promise, I come to see thatmy keeping it will injure the promisee; the promisee, falsely believing thatkeeping the promise is harmless, refuses to release me from the promise.Suppose I come to see that the food I have promised is poisoned (Sidg-wick 1981 [1907], 306–8, 353–54). In these cases, many will not thinkthat there is a reason for keeping it, with the various complications weak-ening this reason or suggesting an outweighing duty. Many will thinkthere is no reason.

Ross agrees. When Ross (1939, 94–97, 110–11) considers variouscases of changed circumstances and falsehood, he claims that there is noprima facie duty, since the prima facie duty is conditional on variousconditions that are now not met. He does not, then, hold the simpleview that there is always a prima facie duty to keep promises. In onecase of changed circumstances—a promise to pay a pauper £100, withthe pauper getting rich before the payment is due—he thinks that onemight say either that one is bound, or that one is not bound because animplicit condition was not met (97–98). When he notes a case in whichkeeping the promise turns out to be unexpectedly costly to either the

41. This appears in the first edition of the Methods, before Mill’s similar argument waspublished in Chapters on Socialism but after Chapters was written.


198

promiser or the promisee (as in the poison case), he does think that one isbound, though the duty may be outweighed (109, 111–12). But else-where he argues that for me to be bound, I must think that my keepingof the promise produces some good to the promisee (162; 1928–29,267–68). Of course, it is tricky to distinguish between a weak reasonand no reason.42 Ross (1930, 28) suggests that feeling “compunction”and that “it is our duty to make up somehow to the promisee for thebreaking of the promise” is a sign of an outweighed, rather than absent,reason. But I think these tests leave it controversial whether there is areason to keep the poison promise, just as it is controversial whether thereis a reason to keep the promise to the pauper.

Ross might insist that there is an unproblematic prima facie dutyto keep some promises. Sidgwick (1981 [1907], 310–11, also 353) himselfwrites:

It appears that a clear consensus can only be claimed for the principle that

a promise . . . is binding, if a number of conditions are fulfilled: viz. if the

promiser has a clear belief as to the sense in which it was understood by the

promisee, and if the latter is still in a position to grant release from it, but

unwilling to do so, if it was not obtained by force or fraud, if it does not

conflict with definite prior obligations, if we do not believe that its fulfil-

ment will be harmful to the promisee, or will inflict a disproportionate

sacrifice on the promiser, and if circumstances have not materially chang-

ed since it was made.

Perhaps Sidgwick can reply by asking what separates these promises fromothers. Sidgwick’s (354, 443–44) answer is that there are clear conse-quentialist rationales for keeping these promises but not for others. Itseems harder to give an explanation of what makes these promises specialthat is internal to the duty of fidelity.43

So far, I have considered the unfairness charge by looking at com-monsense morality. One might also press the charge by looking forinternal problems with benevolence. The obvious worry concerns “moregood.” Sidgwick (415–16) intends “more total good.” But he has no argu-ment for choosing total over average (or some other alternative), orat least no more argument than a commonsense moralist might havefor, say, coming to a particular verdict about whether desert should be

42. There are brief discussions of this issue in the literature on particularism. See, forexample, Shafer-Landau 1997, 590, 592–94; Hooker 2000, 7–11.

43. Of course, there may be such an explanation. For one nonconsequentialistaccount of some of these conditions, see Thomson 1990, 310–20.


199

proportioned to effort or result.44 Nor does it seem plausible to arguethat there is a duty to maximize the total and a duty to maximize theaverage.45 In lieu of a decisive argument, Sidgwick might reply that what-ever internal problems benevolence has are problems for commonsensemorality as well as for consequentialism since commonsense moralityaccepts benevolence. This form of the unfairness charge, then, might bemet by noting that commonsense morality faces extra internal pro-blems—though this is hardly the sort of contrast Sidgwick has in mindwhen he contrasts, for example, “‘I ought to keep my promises’” with hisaxioms that “present themselves as self-evident; as much (e.g.) as themathematical axiom that ‘if equals be added to equals the wholes areequal’” (383).

***

Stepping back from the details, Sidgwick’s argument is simple. ConsiderRoss’s list of prima facie duties. Four are presented as instances of the dutyto promote the good. Three—gratitude, recompense, fidelity to prom-ises—are not. Sidgwick’s axioms support the duty to promote the good.Since they do not say that this is the only duty, Ross, and commonsensemoralists, would agree. Perhaps this is why Sidgwick was so careless inpresenting the axioms—the only people he thought might reject themwere egoists, and here he was careful (sometimes) to word the axioms sothat egoists would agree. The issue between Sidgwick and commonsensemorality is not the axioms but the much longer portion of the Methods thatargues, sometimes unfairly, that the tests rule out any additional axioms.Those who favor the argument for consequentialism from the axiomsmust direct their energies to showing that Sidgwick is fair after all, byshowing that he raises decisive internal problems for all prima facie dutiesother than the duty to promote the good.

44. For this point, see Phillips 2011, 103; Wallace 1988, 41–44; M. Singer 1961, 201.Broad (1930, 250) thinks a utilitarian must favor maximizing the total, and that that“would quite certainly be wrong.” Ross (1939, 70–71, 73–74), citing Broad, agrees,though accepts that taking certain distributions as good avoids the problem.

45. See, for example, some of the problems for maximizing the average in Parfit1984, 420–22.


200

Appendix

In the first edition of the Methods, in a discussion of what becomes personal

irrelevance, Sidgwick (1874, 360) writes that we

are supposed to judge that there is something intrinsically desirable—

some result which it would be reasonable for each individual to seek for

himself, if he considered himself alone . . . . [But] the Good of any one

individual cannot be more intrinsically desirable, because it is his, than the

equal Good of any other individual. So that our notion of Ultimate Good,

at the realization of which it is evidently reasonable to aim, must include

the Good of every one on the same ground that it includes that of any one.

Sidgwick (366) sums up what has been shown: “We have . . . evolved thesuppression of Egoism, the necessary universality of view, which is impliedin the mere form of the objective judgment ‘that an end is good’ just as itis in the judgment ‘that an action is right’.” Later he notes that the egoistcan escape by denying that “his happiness . . . is objectively desirable orGood” (391).

In the second edition, personal irrelevance is stated as it is in the finaledition, except that “as a part of universal good” occupies the place latertaken by “from the point of view of the universe” (Sidgwick 1877b, 355).Sidgwick adds general good: “as rational beings we are manifestly bound toaim at good generally, not merely at this or that part of it; we can onlyevade the conviction of this objectivity by denying that there is any suchuniversal good” (355). He also adds benevolence and claims that a “plainman” would find it wrong to choose his own lesser happiness over thegreater happiness of another. The egoist escapes by denying that his goodis good “absolutely” (389).

In the third edition, “from the point of view of the universe” appearsin personal irrelevance (Sidgwick 1884, 381). The egoist escapes by refusingto speak of good “taken universally” (417).

In the fourth edition, Sidgwick (1890, 382) drops the claim, attachedto general good, about evading the conviction of objectivity by denying thatthere is a universal good. The egoist escapes by avoiding “the point of viewof the Universe,” as in the final edition (417).

It is noteworthy that in the earlier editions, Sidgwick argues that usinga certain concept or concepts—“intrinsically desirable,” “ultimate good,”“universal good,” “objective good,” “good” or “desirable” “used abso-lutely”—commits one to thinking that the good of one is no more impor-tant than the good of another.


201

For some of the concepts listed, Sidgwick’s argument seems too ambi-tious. I can speak of my happiness as “intrinsically desirable” withoutbeing committed to thinking that my happiness is no more importantthan the happiness of another. I seem committed only to thinking that Iought to desire my happiness as an end rather than merely as a means.

Sidgwick also sometimes puts his point in an ambiguous way. When hewrites, for example, of “denying that there is . . . universal good,” he mightbe denying the existence of a concept, or of instances of the concept, or ofuse of the concept.46 The last is best.

Perhaps for these reasons, in later editions, Sidgwick stresses not whatis entailed by certain concepts but instead what is entailed by taking up acertain point of view. On this view, the egoist uses the same concepts aseveryone else but refuses to take up the “point of view of the universe.”

References

Brink, David. 1994. “Common Sense and First Principles in Sidgwick’s Methods.”Social Philosophy and Policy 11: 179–201.

Broad, C. D. 1930. Five Types of Ethical Theory. London: Kegan Paul.———. 1952 [1938]. “Henry Sidgwick.” In Ethics and the History of Philosophy,

49–69. London: Routledge and Kegan Paul.———. 1968 [1942]. “Certain Features in Moore’s Ethical Doctrines.” In The

Philosophy of G. E. Moore, ed. P. A. Schilpp, 43–67. 3rd ed. Lasalle: OpenCourt.

———. 1971 [1953]. “Self and Others.” In Broad’s Critical Essays on Moral Philos-

ophy, ed. David Cheney, 262–82. London: George Allen and Unwin.Crisp, Roger. 2002. “Sidgwick and the Boundaries of Intuitionism.” In Ethical

Intuitionism, ed. Philip Stratton-Lake, 56–75. Oxford: Clarendon.Hayward, F. H. 1901. The Ethical Philosophy of Sidgwick. London: Swan Sonnen-

schein.Hooker, Brad. 2000. “Moral Particularism: Wrong and Bad.” In Moral Particular-

ism, ed. Brad Hooker and Margaret Olivia Little, 1–22. Oxford: Clarendon.Irwin, Terence. 2009. The Development of Ethics. Vol. 3. Oxford: Oxford University

Press.Kamm, F. M. 1996. Morality, Mortality. Vol. 2. Oxford: Oxford University Press.Lacey, A. R. 1959. “Sidgwick’s Ethical Maxims.” Philosophy 34: 217–28.

46. I think this ambiguity slightly mars Phillips’s treatment of the axioms. He thinksSidgwick’s argument turns on the claim that “there is a general reason to suppose thatthere is such a thing as universal goodness. And Sidgwick, so far as I can see, supplies noreason” (Phillips 2011, 125). It is not the existence of the concept or instances of it thatcan be denied but rather that one must employ it.


202

Lillehammer, Hallvard. 2010. “Methods of Ethics and the Descent of Man:Darwin and Sidgwick on Ethics and Evolution.” Biology and Philosophy 25:361–78.

Parfit, Derek. 1984. Reasons and Persons. Oxford: Oxford University Press.———. 2011. On What Matters. 2 vols. Oxford: Oxford University Press.Phillips, David. 2011. Sidgwickian Ethics. Oxford: Oxford University Press.Postema, Gerald. 1998. “Bentham’s Equality-Sensitive Utilitarianism.” Utilitas

10: 144–58.Rashdall, Hastings. 1885. “Professor Sidgwick’s Utilitarianism.” Mind, o.s., 10:

200–26.———. 1907. The Theory of Good and Evil. 2 vols. Oxford: Oxford University Press.———. 1913. Ethics. London: T. C. and E. C. Jack.———. 1914. Is Conscience an Emotion? London: T. Fisher Unwin.Ross, W. D. 1928–29. “The Nature of Morally Good Action.” Proceedings of the

Aristotelian Society 29: 251–74.———. 1930. The Right and the Good. Oxford: Clarendon.———. 1939. Foundations of Ethics. Oxford: Clarendon.Scanlon, Thomas. 1998. What We Owe to Each Other. Cambridge, MA: Harvard

University Press.Schneewind, J. B. 1977. Sidgwick’s Ethics and Victorian Moral Philosophy. Oxford:

Clarendon.Seth, James. 1901. “The Ethical System of Henry Sidgwick.” Mind 10: 172–87.Shafer-Landau, Russ. 1997. “Moral Rules.” Ethics 107: 584–611.Shaver, Robert. 1999. Rational Egoism. Cambridge: Cambridge University Press.———. 2011. “Sidgwick’s Axioms and Common Sense Morality.” In Proceedings of

the Second World Congress on Henry Sidgwick, ed. Placido Bucolo, Roger Crisp,and Bart Schultz, 562–91. Catania: Cooperativa Universitaria Editrice Cata-nese di Magistero.

Schultz, Bart. 2004. Henry Sidgwick: Eye of the Universe. Cambridge: CambridgeUniversity Press.

Sidgwick, Henry. 1874. The Methods of Ethics. 1st ed. London: Macmillan.———. 1876. “Professor Calderwood on Intuitionism in Morals.” Mind, o.s., 1:

563–66.———. 1877a. “Hedonism and Ultimate Good.” Mind, o.s. 2: 27–38.———. 1877b. The Methods of Ethics. 2nd ed. London: Macmillan.———. 1877c. “Mr. Barratt on ‘The Suppression of Egoism’.” Mind, o.s., 2:

411–12.———. 1879. “The Establishment of Ethical First Principles.” Mind, o.s., 4:

106–11.———. 1883. “A Criticism of the Critical Philosophy I.” Mind, o.s., 8: 69–89.———. 1884. The Methods of Ethics. 3rd ed. London: Macmillan.———. 1889. “Some Fundamental Ethical Controversies.” Mind, o.s., 14:

473–87.


203

———. 1890. The Methods of Ethics. 4th ed. London: Macmillan.———. 1902. Lectures on the Ethics of T. H. Green, Mr. H. Spencer, and J. Martineau.

London: Macmillan.———. 1905. Lectures on the Philosophy of Kant and Other Philosophical Lectures and

Essays. London: Macmillan.———. 1981 [1907]. The Methods of Ethics. 7th ed. Indianapolis: Hackett.Singer, Marcus G. 1961. Generalization in Ethics. New York: Knopf.Singer, Peter. 1974. “Sidgwick and Reflective Equilibrium.” Monist 58: 490–517.———. 2005. “Ethics and Intuitions.” Journal of Ethics 9: 331–52.Skelton, Anthony. 2008. “Sidgwick’s Philosophical Intuitions.” Ethics and Politics

10: 185–209.———. 2010. “Henry Sidgwick’s Moral Epistemology.” Journal of the History of

Philosophy 48: 491–519.———. 2011. “Utilitarian Practical Ethics.” In Proceedings of the Second World

Congress on Henry Sidgwick, ed. Placido Bucolo, Roger Crisp, and Bart Schultz,592–633. Catania: Cooperativa Universitaria Editrice Catanese di Magis-tero.

Sverdlik, Steven. 1985. “Sidgwick’s Methodology.” Journal of the History of Philos-

ophy 23: 537–53.Thomson, Judith Jarvis. 1990. The Realm of Rights. Cambridge, MA: Harvard

University Press.Wallace, James D. 1988. Moral Relevance and Moral Conflict. Ithaca, NY: Cornell

University Press.


204

sidgwick's axioms and consequentialism

Documents