I

It is customary to divide normative concepts into two broad categories, the evaluative (‘the good’) and the deontic (‘the right’). Among the evaluative concepts are the concepts of what is good, what is bad and what is neutral. It is not clear (to me) just how this list might be extended, except that it should include the comparative concepts of what is better and what is worse. The deontic concepts are more numerous (or at least we have more names for them). They include the concepts of what is right, fitting, suitable, appropriate, correct, justified, warranted; of what is wrong, unfitting, unsuitable, inappropriate, incorrect, unjustified, unwarranted; of what is obligatory, required, imperative, mandatory; and also floating around somewhere in this mix is the concept of a reason.

There are normative concepts other than the evaluative and the deontic – those of praise- and blameworthiness, for example, and those of super- and suberogation. Just how these concepts relate to those of the evaluative and the deontic is an interesting question: the former appear to overlap with the latter but also to extend beyond them. I will not address this question here. Instead, I will concentrate on the interconnections between the good and the right. More particularly, I will focus on a thesis that has recently been endorsed by T. M. Scanlon but whose origin can be traced back through A. C. Ewing to Franz Brentano and perhaps beyond.Footnote ¹ That thesis may initially be put simply as follows:

• G: x is good = df. x is valuable.

The plan of this article is the following. In Section II, I will explain how thesis G is to be understood and develop some refinements of it with a view to focusing on intrinsic value in particular. In Section III, I will discuss the relation between G and consequentialism. In Section IV, I will critically examine Scanlon's reasons for thinking G true. In Section V, I will consider other reasons for thinking G true and reasons for thinking it false and discuss what its being false implies. There will then come two appendices. In the first, I will discuss whether it is good to do right. In the second, I will discuss whether an act's being wrong provides a reason not to do it.

II

To say that something, x, is valuable is not simply to say that it can be valued. It is to say that it ought to be valued; valuing x is the right way to respond to it. Thesis G thus asserts a strong connection between the good and the right. As just indicated, other terms can be and have been used to express the deontic side of this connection. We could say that it is fitting to value x, or suitable to do so, or appropriate, and so on; or that x is worthy of being valued, or deserves to be valued, and so on. Perhaps there are subtle differences here that merit investigation, but undertaking this would take us too far afield. Instead let me use yet another term to express the idea at issue, one that is currently all the rage: there is a reason to value x. Thus:

• V: x is valuable = df. there is a reason to value x.

In light of V, G becomes

• G1: x is good = df. there is a reason to value x.

To understand what G1 comes to, it is important to take note of a vexing ambiguity in the verb ‘to value’. Surely one common way of understanding an expression of the form ‘S values x’ is as follows:

• v1: S values x = df. S ascribes value to x, i.e.

  S judges x to be of value, i.e.

  S judges x to be good.

But this poses an immediate problem: G1 and v1 are jointly circular. Another common wayFootnote ² to understand ‘S values x’ is as follows:

• v2: S values x = df. S judges x to be valuable.

But this too poses an immediate problem: V and v2 are jointly circular. To avoid such problems with circularity, we must accept that the verb ‘to value’ is doing double duty. Apart from the common sense(s) identified in v1 and v2 there is also this sense: to value something is to have what is often called a ‘pro-attitude’ towards it, that is, to be disposed to respond positively towards it in some way. I will use the verb ‘to favour’ to refer to this disposition. Thus:

• v3: S values x = df. S favours x.

In light of v3, G may now be recast once again as follows:

• G2: x is good = df. there is a reason to favour x.Footnote ³

(Since Scanlon will be the focus of discussion later, it is worth noting now some difficulties with his own use of the terms ‘to value’ and ‘valuable’. He says: ‘To value something is to take oneself to have reasons for holding certain positive attitudes towards it and for acting in certain ways in regard to it.’Footnote ⁴ This corresponds pretty well with the sense of ‘to value’ given in v2. He continues: ‘To claim that something is valuable . . . is to claim that others also have reason to value it, as you do.’Footnote ⁵ This is odd. First, in this sentence ‘to value’ is most naturally understood in terms of v3 rather than v2, although Scanlon gives no hint that he takes himself to be switching senses. Second, it is implausible to maintain that what is valuable must be impersonally so. Surely some things – old family photos, for instance – can be, and one can find them to be, valuable for or to oneself but not for or to others; that is, they are, and one can find them to be, such that one oneself has a reason to favour them even though others do not.)

Thesis G2 concerns the concept of goodness in general. It can and should be refined. One refinement has to do with the distinction between non-derivative and derivative goodness; the former has to do with what is good ‘for its own sake’, whereas the latter has to do with what is good ‘for the sake of something else’. Thus:

• G2a: x is non-derivatively good = df. there is a reason to favour x for its own sake;

• G2b: x is derivatively good = df. there is a reason to favour x for the sake of something else to which x is related in some way.

Notice three things about these theses, in light of which still further refinement can be made. First, the sort of reason at issue has not been specified; it could be prudential, moral, legal, aesthetic, and so on. Second, just who (if anyone) has a reason to favour x has not been specified. Third, the sort of relation at issue in G2b has not been specified; it could be a means–end relation, a part–whole relation, a sign–signified relation, and so on.

As I noted in Section I, my aim in this article is to focus on intrinsic value in particular. I have in mind the sort of value that G. E. Moore ascribes to pleasure, beauty and knowledge when he says that they (among other things) are intrinsically good.Footnote ⁶ Evidently, Moore has a kind of non-derivative value in mind.Footnote ⁷ But what kind? Well, one thing that seems clear is that it is an impersonal kind; something cannot be intrinsically good, in the relevant sense, for one person but not for another. I would also say that it is an ethical or moral kind. This may not be so clear – indeed, it is quite controversial – but consider the following. Suppose that you are attracted to the idea that there is a reason to favour that which is good. Imagine that John becomes aware of something that is intrinsically good (in the Moorean sense) – for example, he sees a news report of some mother in a foreign land crying tears of joy when her child is rescued from a burning building – but he does not respond positively towards it. He is unmoved. You accuse him of being hard-hearted. He replies: ‘What has that child's being saved got to do with me? I have no reason to be glad about it.’ Surely this reply is defective. But how? It is true that John has no prudential reason to be glad at the news; he is no better off for it. Likewise, he has no legal or aesthetic reason to be glad at it. I submit that he has a moral reason to be glad at it; his failure so to respond is indicative of a moral deficiency, an ethical insensitivity, in him.

To forestall one possible misunderstanding: in saying that John has a moral reason to be glad at the news, I am not saying that he is morally obligated (even prima facie) so to respond. One can be morally obligated to do something only if one is in control of doing it, and John might not be in control of such a response.Footnote ⁸ But even if he cannot control his response, it makes sense to say that his being glad at the news would be a morally fitting way to respond to it – that the news calls for or requires such a response from him. To say, then, that John has a moral reason to be glad at the news is to say that there is a moral requirement that he so respond, even if this requirement is not tantamount to an obligation.Footnote ⁹

This observation brings to light another vexing ambiguity, this time in the term ‘reason’ itself. What exactly constitutes the reason that I have alleged that John has to be glad at the news? Does the reason consist in his being required to respond in this way, or does it consist in that which requires him to respond in this way? As far as I can tell, both the former sense (the ‘requirement’ sense) and the latter sense (the ‘requirer’ sense) are legitimate senses of the term. That is, we can equally well say that the news is a reason (‘requirer’ sense) for John to be glad or that the news gives him a reason (‘requirement’ sense) to be glad. The distinction between these two senses of ‘reason’ is easy to overlook, since for every requirement there is a ground that plays the role of requirer, and nothing can play the role of requirer unless it grounds a requirement. Nonetheless, our linguistic conventions seem sometimes to mark the distinction. To say ‘Jane has more reason (singular) than John to respond in that way’ is to invoke the ‘requirement’ sense of ‘reason’; it is to say that (a) Jane's requirement to respond in the way in question is stronger than John's. To say ‘Jane has more reasons (plural) than John to respond in that way’ is to invoke either the ‘requirement’ or the ‘requirer’ sense of ‘reason’; one may mean either that (b) Jane has more requirements than John, or that (c) there are more grounds for Jane's requirement(s) than for John's. It is important to note the strict independence of (a), on the one hand, from (b) and (c), on the other. Jane may have more reasons than John to respond in the way in question, yet John may have more reason than Jane so to respond; it is not necessarily the case (although it may frequently be the case) that the strength of one's requirements increases as the number of their grounds increases. I will return to this issue in Section IV below.

To forestall another possible misunderstanding: in saying that John's failure to be glad at the news is a moral failing, I am not saying either that the mother's joy is morally good or that the child's presumed suffering is morally bad. All that I am saying is that the joy is intrinsically good and the suffering intrinsically bad and that, in virtue of this fact, John has a moral reason to be glad at the news. You might agree but think that it is true only of some intrinsic goods that there is a moral reason to favour them and true only of some intrinsic evils that there is a moral reason to disfavour them. Not so, I believe; it is true of all. Suppose, for the sake of argument, that Michelangelo's artistic works are intrinsically good. If so, then I am committed to saying that we have a moral reason to favour them. This may seem odd: you may accept that we have an aesthetic reason to favour them but balk at saying that we have a moral reason to do so. Now, I acknowledge that there is an aesthetic reason to admire Michelangelo's works for their own sakes and that they are therefore non-derivatively aesthetically good. If we do admire his works for this reason, we thereby evince an appropriate aesthetic sensitivity. But Michelangelo's works’ being non-derivatively aesthetically good does not imply their being intrinsically good in (what I take to be) the Moorean sense. If they are intrinsically good, then we should favour them for their own sakes, regardless of our aesthetic capacities. If they are intrinsically good, then a blind man has, once he has been made aware of them, just as much reason to favour them for their own sakes as someone who is looking directly at them. What kind of ‘should’, what kind of reason is this? I submit that it is a moral one.

I will not belabour this point further here,Footnote ¹⁰ however, since, as will become apparent, I will not need to rely on it. All that needs to be noted is that the type of reason at issue is a normative one of some sort, and this is something that everyone can acknowledge. Nonetheless, let me now state, in the terms that I prefer, the particular version of thesis G on which I wish to concentrate:

• IG: x is intrinsically good = df. anyone who is aware of x has a moral reason to favour it for its own sake.

III

In Principia Ethica, Moore gives the following account of moral rightness:

• R: some act, a, is the morally right thing for someone, S, to do = df. a is the best thing for S to do (i.e. a ‘together with its consequences presents a greater sum of intrinsic value than any possible alternative’).Footnote ¹¹

It is clear that Moore is hereby proposing that we understand moral rightness in terms of intrinsic goodness. The proponent of IG, on the other hand, is claiming that we should understand intrinsic goodness in terms of moral rightness. It may therefore seem that such a person is ‘reversing’ Moore's position.Footnote ¹² But, actually, this is misleading, for IG and R are compatible. The proponent of R, such as Moore, is concerned with what it is morally right to do overall or all things considered, that is, with what there is conclusive moral reason to do. The proponent of IG, on the other hand, is concerned with what there is some moral reason to do, that is, with what is prima facie morally right.Footnote ¹³ No one wishes to say that, whenever one is aware of something intrinsically good, one is overall morally required to favour it for its own sake, for there could well be some overriding countervailing moral reason not to do so. The upshot is that it is perfectly consistent to analyse overall moral rightness in terms of intrinsic goodness, as R does, and then to analyse intrinsic goodness in turn in terms of prima facie moral rightness, as IG does. So we should not think of IG as being opposed in any way to the Moorean brand of consequentialism captured by R.Footnote ¹⁴

It has been claimed that IG in fact implies a form of consequentialism. (Clearly, the term ‘consequentialism’ is now being used more broadly, since IG is concerned with favourable responses generally rather than, as R is, with the performance of acts in particular, and also since, as just noted, IG is concerned with prima facie rightness generally rather than, as R is, with overall rightness in particular.) Jonathan Dancy characterizes this form of consequentialism as the thesis that all moral reasons are ‘value-involving’.Footnote ¹⁵ The idea is that, since IG asserts an identity between something's being intrinsically good and there being a reason to favour it, it implies not only that wherever there is value there is a reason to respond favourably, but also that wherever there is a reason to respond favourably there is value. Deontologists have typically resisted the latter claim. Dancy thus argues that IG ‘threatens’ to resolve the debate between consequentialists and deontologists in favour of consequentialism.

This argument can be resisted. Consider the moral theory advocated by W. D. Ross in The Right and the Good. According to Ross, we have a number of prima facie moral duties or obligations, i.e. a number of different moral reasons to act in various ways. He says that it is prima facie morally obligatory to (a) fulfil one's commitments, (b) make reparation for past wrongdoing, (c) display gratitude for benefits received, (d) distribute goods justly, (e) act beneficently towards others, (f) act beneficently towards oneself and (g) refrain from maleficence towards others.Footnote ¹⁶ At one point he explicitly groups duties (d)–(f) together and says that they all come under the general principle that ‘we should produce as much good as possible’,Footnote ¹⁷ but he explicitly denies that duties (a)–(c) fall within the ambit of this general obligation. Ross, then, would reject the claim that all prima facie duties are value-involving and thus deny that his theory reduces to consequentialism. Is he thereby committed to rejecting IG?

Perhaps not. It seems to me that Ross could, consistently with his theory, accept IG if he made one of the following two moves.Footnote ¹⁸ He could interpret IG as follows:

• IG1: x is intrinsically good = df. there is some limited set of kinds of favourable responses such that anyone who is aware of x has a moral reason to make some such kind of response towards x for its own sake.

The underlying idea here is that intrinsic value does not simply call for just any kind of positive response; it calls for a positive response of some particular kind or kinds.Footnote ¹⁹ For example, it might be said that what is intrinsically good warrants ‘promoting’ for its own sake, whereas what we have reason to do when, for example, we have made a promise is not to ‘promote’ fidelity but rather to ‘exhibit’ it. Alternatively, Ross could interpret IG as follows:

• IG2: x is intrinsically good = df. there is some limited set of kinds of moral reasons such that anyone who is aware of x has a moral reason of some such kind to favour x for its own sake.

The underlying idea here is that there are not just different moral reasons (e.g. reasons to fulfil one's commitments, reasons to act beneficently, and so on) but different kinds of moral reasons to behave in various ways, and that only some of these kinds of reasons are value-involving.Footnote ²⁰

I am not endorsing (or rejecting) either of these moves. To evaluate them, it would be helpful to have some story about the different kinds of responses or reasons that are supposed to be at issue, and of why these kinds have a tie to intrinsic value that other kinds do not, and I am not sure just how such a story might go. I mean only to point out that there does seem to be some conceptual room to manoeuvre here, and that we should therefore not rush to the conclusion that IG implies that some form of consequentialism is true.

IV

Whether IG implies that some form of consequentialism is true seems to me not so important as whether it itself is true. Should we accept it?

Scanlon believes so. What he says is this:

This is a long and complex passage. I think it is pretty clear what Scanlon's view of goodness is. What is not so clear (to me) is just why he holds this view, and it is for this reason that I have quoted him at length: my doing so should help you judge whether what I say below misrepresents his position in any way.

Before I lay out what I take to be Scanlon's view of goodness, let me note a complication concerning his use of the terms ‘good’ and ‘valuable’ in the quoted passage. He starts by saying, ‘We judge things to be good or valuable . . .’, thereby indicating that he already accepts

• G: x is good = df. x is valuable.

This impression is reinforced when he continues, ‘But being good or valuable cannot be identified with any such “natural” property . . . .’ Remember also, from Section II above, that Scanlon gives the following account of what it is for something to be valuable: ‘To claim that something is valuable . . . is to claim that others have reason to value it, as you do.’Footnote ²² This suggests that he begins by accepting, more particularly,

• G2: x is good = df. there is a reason to favour x.

But, of course, the question at issue is whether G2 is true; presuming its truth at the outset is not a good way to argue for its truth. This problem is heightened when, in the second paragraph, Scanlon drops his reference to goodness and talks only in terms of valuableness. He distinguishes two views regarding ‘the relations between [(a)] these natural properties, [(b)] the property of being valuable [note: here he does not say, “the property of being good”], and [(c)] the reasons that we have for behaving in certain ways in regard to things that are valuable [note again: he does not say, “things that are good”].’ He then characterizes the first view, which he rejects, as follows: ‘when something has [(a)] the right natural properties it has [(b)] the further property of being valuable [note once again: he does not say “good”], and that property [i.e. valuableness] gives us [(c)] reason to behave or react in certain ways with regard to it . . .’. But of course this is a crazy view, if we accept, as I suggest and Scanlon indicates, that for something to be valuable is for it to be such that we have a reason to value (i.e. to favour) it. On this understanding, the view that he is attacking is the view that something's being such that we have a reason to favour it gives us – that is, presumably, either grounds our having or explains our having – a reason to favour it. But who would want to say this? Who (apart, perhaps, from some theologians) would ever want to say that it is possible that x's being F grounds or explains x's being F?

In order to avoid prejudging the question at issue, I therefore propose that we couch it in the following terms: what is the relation between (a) something's having some natural property, (b) its being good and (c) its being valuable (i.e. its being such that we have a reason to favour it)? The two views that Scanlon discusses are these:

• View 1: (a) grounds (b), which grounds (c).

• View 2: (a) grounds (b), which is identical with (c).

View 2 is the buck-passing account, which of course constitutes an endorsement of

• G: x is good = df. x is valuable.

Before examining Scanlon's reasons for accepting View 2, let me briefly attend to some further complications. First, Scanlon moves in the quoted passage from talk (in the second paragraph) of a buck-passing account (singular) of goodness and value, to talk (in the third paragraph) of buck-passing accounts (plural), back to talk (in the fourth paragraph) of a buck-passing account, back to talk (in the final paragraph) of buck-passing accounts. The singular talk suggests that he identifies goodness with value, whereas the plural talk suggests that he distinguishes between goodness and value. I do not think we should take the plural talk seriously; I will assume that Scanlon is proposing just one buck-passing account of goodness, i.e. of value.Footnote ²³

A second complication is that the buck-passing account that Scanlon embraces is slightly more detailed than my statement of View 2 reveals. What he actually says is that (b) something's being good is identical with (c′) its being such that its having some natural property (i.e. its being such that (a) is true of it) provides a reason to favour it.

A third complication is that Scanlon has said that he did not intend to commit himself to the claim that such reason-providers must in every case be restricted to natural properties. He concedes that they may sometimes involve certain evaluative properties that are more specific than the general property of goodness.Footnote ²⁴ This raises difficulties that I will not address here.

A final complication is that the natural properties that Scanlon cites as illustrations of what might ground goodness or valuableness are ill suited to a discussion of intrinsic goodness in particular, which, as I have said, is the kind of goodness on which I wish to concentrate here. He talks of something's being good because it is pleasant, and of a discovery's being valuable because it provides new understanding of how cancer cells develop. In each case, it would seem implausible to ascribe intrinsic value to that which is said to be good or valuable. There is some plausibility to saying that pleasure and understanding are intrinsically good, but this would only warrant ascribing extrinsic value to that which is pleasant or provides understanding.

In order to focus on intrinsic value, let us suppose for the sake of illustration that pleasure is intrinsically good, that is, that states consisting of someone's being pleased are intrinsically good. Consider some such state, x. Let ‘P’ stand for x's being a state of pleasure, ‘G’ for x's being intrinsically good and ‘V’ for x's being intrinsically valuable (i.e. for x's being such that anyone who is aware of it has a moral reason to favour it for its own sake). (Although the distinction between the ‘requirement’ and ‘requirer’ senses of ‘reason’ will come into play later, we need not worry about it at this stage, given the fact, noted above, that there is no requirement without a requirer and no requirer without a requirement. For present purposes, we can understand ‘V’ as standing for x's being such that anyone who is aware of it is morally required by some fact about it to favour it for its own sake.) Let ‘→’ signify the grounding relation and ‘=’ signify the relation of identity. (The grounding relation is the inverse of supervenience; if x's being F grounds x's being H, then x's being H supervenes on x's being F.) View 1 gives us the following picture of the relation between P, G and V:

View 2, which incorporates IG, gives us the following picture of this relation:

Scanlon opts for View 2/Picture 2 over View 1/Picture 1. Why? He gives two arguments, which Roger Crisp has dubbed the Redundancy Argument and the Argument from Pluralism.Footnote ²⁵ Let us consider them in turn.

The Redundancy Argument is contained in the third paragraph of the passage quoted above. It is not clear to me just what the argument is supposed to be. Here is one possible reconstruction (when applied to a supposed instance of intrinsic value):

• RA1: Suppose 1. x is a state of pleasure.

  2. Whatever is a state of pleasure is intrinsically good.

  Therefore 3. x is intrinsically good.

  4. Whatever is intrinsically good is intrinsically valuable.

  Therefore 5. x is intrinsically valuable.

  6. x's being a state of pleasure fully explains x's being intrinsically valuable.

  7. If x's being a state of pleasure fully explains x's being intrinsically valuable, then x's being intrinsically good does not also explain x's being intrinsically valuable.

  8. If x's being intrinsically good does not explain x's being intrinsically valuable, then x's being intrinsically good does not ground x's being intrinsically valuable.

  9. If x's being intrinsically good does not ground x's being intrinsically valuable, then View 1/Picture 1 is false.

  10. If View 1/Picture 1 is false, then View 2/Picture 2 is true.

  Therefore 11. View 2/Picture 2 is true.

I find this argument unpersuasive. I will not quarrel with any of steps 1–5. The hedonistic judgement expressed in premiss 2 may be too sweeping, but it is made solely for the sake of illustration and is not essential to the argument. Someone might well balk at premiss 4, especially since I have said that ‘valuable’ is to be construed in terms of what there is a moral reason to favour. But in fact, as noted in Section II, the reason's being a moral one is not essential to the argument, so long as it is recognized as a normative reason of some sort. Moreover, premiss 4, so understood, is endorsed by both Picture 1 and Picture 2 and so does not itself favour either picture over the other.

The trouble begins with premiss 6. Scanlon claims that x's natural properties ‘provide a complete explanation’ of the reason we have to favour x. In the preceding paragraph he claims that these properties ‘can be grounds for concluding’ that x is valuable. I think these expressions may be regarded as equivalent. What, though, of the claim (suggested by Scanlon's talk of lower-order properties ‘providing reasons’ to behave in certain ways) that x's natural properties ‘ground’ x's being valuable? There is a distinction between saying, on the one hand, that x's being F is grounds for concluding that x is H and saying, on the other, that x's being F grounds x's being H. The latter concerns a purely ontological relation, whereas the former concerns an ontological-cum-epistemological relation.

Consider Picture 2. According to that picture, x's being a state of pleasure (P) grounds its being intrinsically good (G). Of course, P also grounds x's being intrinsically valuable (V), since on this picture G is identical with V. But does P provide an adequate explanation of either G or V? This is not so obvious. Suppose that Jane is aware that, in general, something's being a state of pleasure grounds its being intrinsically good but not that its being intrinsically good is identical with its being intrinsically valuable. Then pointing out to her that P is the case will serve to explain to her how it is that G is the case but not how it is that V is the case. You might reply that P is nonetheless grounds for Jane's concluding that V is the case, whether or not she is aware of this. But that may not be so. Whether something is grounds for someone's reaching some conclusion is a function of that person's evidence, and it may be that Jane has no evidence that something's being intrinsically good is identical with its being intrinsically valuable.

Now consider Picture 1. According to that picture, P grounds G, which in turn grounds V. The grounding relation is transitive. It thus follows that P grounds V. Suppose that Jane has evidence that P grounds G but not that G grounds V. Then pointing out to her that P is the case will explain how it is that G is the case but not how it is that V is the case.

Return now to premiss 6 of RA1. I find it tendentious. Even if it is agreed that P grounds V, there is no reason to think that everyone is so situated epistemically that the former explains the latter. (For this reason, premiss 8 is also suspect.) But let us put this aside. Suppose we restrict our attention to people who are ideally situated epistemically, and so grant premiss 6, and that we now turn our attention to premiss 7. I see no reason to accept this premiss. Consider Picture 1. Suppose that Jane is aware that in general something's being a state of pleasure grounds its being intrinsically good, which in turn grounds its being intrinsically valuable, and that therefore something's being a state of pleasure grounds its being intrinsically valuable. Then pointing out to her that P is the case will fully explain how it is that V is the case, but this does not mean that G does not also explain V. On the contrary, G will also fully explain V, in that citing the former will under the circumstances suffice for Jane's concluding the latter.

You may object that what I have just said misuses the expression ‘fully explain’, in that there cannot be more than one ‘full’ explanation of some fact. But if that is so, we must reject premiss 6 after all, even on the supposition that we are dealing with people who are ideally situated epistemically. For there is no reason to think that, on either Picture 1 or Picture 2, P represents the most fundamental ontological ground that is implicitly ‘at work’ in the picture. On the contrary, there is good reason to think otherwise. Presumably, P will itself supervene on some neurophysiological fact, N, that would serve to explain, to anyone in the appropriate epistemic position, how it is that P is the case. Thus if we are required to ‘pass the buck’ on from G to P, in light of G's failure to give a ‘full’ explanation of V, we must pass it on yet again from P to N. But this is not in keeping with Scanlon's own examples. These examples strongly suggest that he takes the buck to stop with the higher-level natural properties that he cites; he gives no indication that he would pass it on to whatever lower-level natural properties ground these higher-level properties. I can see no rationale for stopping at such a mid-way point. More importantly, and to repeat my main point, tracing the explanation of V to some natural ground such as P (or N) does not preclude G's being a ground of V and thus playing a role in the explanation of V.Footnote ²⁶

I conclude that at least one of premisses 6, 7 and 8 is false and that RA1 is therefore unsound. Perhaps it is also worth noting that premiss 10 is false. (Premiss 9 is clearly true.) There are, after all, other pictures concerning the relation between P, G and V that might be considered. Here are some:

Still, I will not pursue this point, for two reasons. First, I find no reason to endorse any of these other pictures. Second, in the quoted passage Scanlon merely claims to provide reasons in ‘support’ of the buck-passing account, and so premiss 10 may oversimplify his position.

RA1 constitutes just one way to construe Scanlon's Redundancy Argument. When discussing this argument, Dancy indicates that he finds in it a key idea that does not appear in RA1. When commenting on two views, the second of which is the view that I have called View 1 and depicted in Picture 1, Dancy says:

In saying this, Dancy evidently takes himself to be endorsing something in Scanlon's Redundancy Argument, since he shortly goes on to say: ‘Remember that we agreed with Scanlon that the badness of the pain cannot add to the reasons generated by the features that make the pain bad . . .’.Footnote ²⁸ Dancy seems to be supposing that, when Scanlon says that the pertinent natural properties provide a ‘complete explanation’ of the reasons we have for favouring something that is good, and that the goodness of the thing therefore does no ‘further work’ in providing such reasons, he is thereby invoking the idea that goodness adds no reasons to those generated by the natural properties in question. Now, I am not sure that this is Scanlon's intention but, if it is, then his argument is not properly represented by RA1 and requires a different reconstruction. But what reconstruction?

At this point, the ambiguity between the ‘requirement’ and ‘requirer’ senses of ‘reason’ becomes relevant. Much of what Scanlon says suggests that he is invoking the former sense. He talks repeatedly of natural properties ‘providing’ reasons to respond in certain ways and also of their ‘providing an explanation’ of these reasons. (Dancy talks similarly of such properties ‘giving’ or ‘generating’ reasons to act.) Such talk strongly suggests that the properties are one thing and the reasons are another. This seems to be good evidence that Scanlon takes reasons to be requirements to behave in certain ways, requirements that are grounded in the natural properties in question. Yet Scanlon also talks at times of natural properties as ‘constituting’ reasons, and this might seem to suggest that such properties are themselves reasons to respond in certain ways. So construed, reasons are requirers rather than requirements. Moreover, it is clearly the ‘requirer’ sense that Scanlon has in mind when he says that the fact that a resort is pleasant is a reason to visit it, and the fact that a discovery casts light on the causes of cancer is a reason to applaud it.

In light of this ambiguity, two distinct reconstructions of Scanlon's Redundancy Argument (construed in light of Dancy's observation about value giving us no further reason(s) to act beyond the reason(s) generated by the ground for that value) emerge, one couched in the ‘requirement’ sense of ‘reason’, the other in the ‘requirer’ sense. In the ‘requirement’ sense, Scanlon's argument may be put as follows (when applied, once again, to a supposed instance of intrinsic value):

• RA2: 1–6. [same as steps 1–6 of RA1]

  7. If x's being intrinsically good also explained x's being intrinsically valuable, then x's being intrinsically good would provide an additional explanation of x's being intrinsically valuable.

  8. If x's being intrinsically good provided an additional explanation of x's being intrinsically valuable, then x's being intrinsically good would add to the reason, i.e. would strengthen the requirement, for anyone who is aware of x to favour x for its own sake.

  But 9. x's being intrinsically good does not add to the reason (already provided by x's natural properties – in particular, by x's being a state of pleasure) to favour x for its own sake.

  Therefore 10. x's being intrinsically good does not explain x's being intrinsically valuable.

  11–14. [same as steps 8–11 of RA1]

When understood in terms of the ‘requirer’ sense of ‘reason’, Scanlon's argument may be put as follows:

• RA3: 1–6. [same as steps 1–6 of RA1]

  7. If x's being intrinsically good also explained x's being intrinsically valuable, then x's being intrinsically good would provide an additional explanation of x's being intrinsically valuable.

  8. If x's being intrinsically good provided an additional explanation of x's being intrinsically valuable, then x's being intrinsically good would add to the reasons for, i.e. would be an additional ground that requires, anyone who is aware of x to favour x for its own sake.

  But 9. x's being intrinsically good does not add to the reasons (already constituted by x's natural properties) to favour x for its own sake.

  Therefore 10. x's being intrinsically good does not explain x's being intrinsically valuable.

  11–14. [same as steps 8–11 of RA1]

In both RA2 and RA3, it is of course premiss 9 that is intended to give expression to the key idea, which Dancy ascribes to Scanlon, that values are redundant in the sense that they add no reason(s) not already furnished by their grounds.

Each of RA2 and RA3 inherits some of the problems of RA1, in that some of the troublesome premisses in RA1 (namely, 6, 8 and 10) reappear in RA2 and RA3 (as premisses 6, 11 and 13 – the problems arise, regardless of whether ‘reason’ is interpreted in the ‘requirement’ or ‘requirer’ sense, which is why I did not attend to this distinction when discussing RA1). But each also faces an additional problem.

Consider RA2 first. Its premiss 8 is simply false. It is of course the case that on Picture 1 there are more grounds than on Picture 2 for the requirement to favour x for its own sake. But this gives us no reason whatsoever to think that Picture 1 implies that the requirement in question is somehow amplified. As noted in Section II above, the strength of a requirement is strictly independent of the number of its grounds. It would be absurd to think that the badness of a toothache somehow amplifies the (prudential) requirement, already provided by its painfulness, to go to the dentist. But, again, Picture 1 entails no such absurdity.

Now consider RA3. Its premiss 8 is unexceptionable (as long as one recognizes that adding grounds for a requirement does not entail strengthening the requirement). But now premiss 9 – Dancy's key claim – is problematic. Why agree that G constitutes no ground for the requirement contained in V beyond that constituted by P? Perhaps the idea is that, if G were to constitute such a ground, it would have to do so independently of P. But that is not the case. Picture 1 is not Picture 6. According to the latter picture, G is a ground for V that is separate from P. I think we can agree that there is no good reason to accept this picture. But Picture 1 is quite different. According to it, P grounds V by way of G (just as the neurophysiological property N grounds V by way of P). Premiss 9 of RA3 simply begs the question against the proponent of View 1/Picture 1.

I conclude that Scanlon's Redundancy Argument is unsuccessful. What about his Argument from Pluralism? This can be dealt with much more briefly. As I see it, the following is what he has in mind (when applied to intrinsic value):

• AP: 1. There are many different kinds of things (e.g. states of pleasure, beauty, and knowledge) that are intrinsically good.

  2. Whatever is intrinsically good is intrinsically valuable.

  Therefore 3. There are many different kinds of things that are intrinsically valuable.

  4. If there are many different kinds of things that are intrinsically valuable, then there are many different kinds of grounds for things’ being intrinsically valuable – that is, the grounds for some things’ being intrinsically valuable are distinct in kind from the grounds for other things’ being intrinsically valuable.

  5. If the grounds for some things’ being intrinsically valuable are distinct in kind from the grounds for other things’ being intrinsically valuable, then being intrinsically good is not the only kind of ground for things’ being intrinsically valuable.

  Therefore 6. Being intrinsically good is not the only kind of ground for things’ being intrinsically valuable.

  7. If being intrinsically good is not the only kind of ground for things’ being intrinsically valuable, then View 1/Picture 1 is false.

  8. If View 1/Picture 1 is false, then View 2/Picture 2 is true.

  Therefore 9. View 2/Picture 2 is true.

I find this argument just as unpersuasive as the other(s), even if we grant (as I suspect we should) the pluralism contained in steps 1–3, and even if we overlook the oversimplification contained in premiss 8. First, premiss 4 is tendentious. We have been given no reason to think that, just because there are many different kinds of things that have intrinsic value, a thing's being intrinsically valuable cannot in every case be attributed to its being intrinsically good. Picture 1 is, after all, perfectly compatible with the following pictures (where ‘B’ stands for x's being a state of beauty, and ‘K’ for x's being a state of knowledge):

Second, even if premiss 4 is granted and hence its proximate conclusion, 6, is too (for there is no doubting premiss 5), this does nothing to impugn View 1, since premiss 7 is false. The reason for this is the same as that given earlier when discussing the Redundancy Argument: to claim that a thing's being intrinsically good is not the sole ground for its being intrinsically valuable is not to claim that the former is not a ground for the latter.

V

The verdict on View 2/Picture 2 thus remains open. We have not yet been given sufficient reason to prefer it to View 1/Picture 1. Are there any other considerations that might help tip the balance one way or the other?

One consideration is simply that of simplicity. Someone might say: Picture 2 is simpler than Picture 1; hence Picture 2 is to be preferred.

The problem with this proposal is that simplicity is surely at best a tie-breaker: all else being equal, a simpler view is perhaps to be preferred to a more complicated one, but not necessarily otherwise. Of course, the question is whether all else is equal in this context.

Another consideration concerns the relation between goodness and valuableness. The idea that something's being intrinsically good is equivalent to its being intrinsically valuable is very attractive. Someone might say: Picture 2 captures this idea, whereas Picture 1 does not; hence Picture 2 is to be preferred.

The problem with this proposal is that it misrepresents Picture 1, in that this picture can accommodate the idea in question, which I agree is very attractive.Footnote ²⁹ It is true that Picture 1 is incompatible with

IG: x is intrinsically good = df. anyone who is aware of x has a moral reason to favour it for its own sake

since Picture 1 depicts V as supervenient on, rather than as identical with, G, whereas Picture 2 of course endorses IG. But Picture 1 is perfectly compatible with a thesis that is slightly weaker than IG, namely,

IG*: necessarily, x is intrinsically good iff anyone who is aware of x has a moral reason to favour it for its own sake,

and this thesis gives full expression to the idea at issue.

A third consideration concerns the following thesis:

1. (1) if x is intrinsically good, then anyone who is aware of x has a moral reason to favour it for its own sake because it is intrinsically good.

Someone might say: (1) is true; Picture 1 accommodates this thesis, whereas Picture 2 does not; hence Picture 1 is to be preferred.Footnote ³⁰

I find this a very difficult matter. I think there is no denying that thesis (1) strikes many, myself included, as intuitively plausible, whereas its near-converse

1. (2) if x is intrinsically good, then x is intrinsically good because anyone who is aware of x has a moral reason to favour it for its own sakeFootnote ³¹

strikes no one as plausible. But what are we to make of this?

Proponents of IG/View 2/Picture 2 have tended simply to deny that (1) is true, pointing to the fact that whatever natural properties x has fully account for the reason to favour it.Footnote ³² But we have been through all of that: in whatever sense x's natural properties fully account for the reason to favour it, their doing so does not preclude its goodness from also providing such an account. Of course, this last point does not show that (1) is true, but if, as IG asserts, goodness and valuableness are the same thing, why should (1) strike so many as plausible whereas (2) does not?

One answer, suggested by Ewing, is this: there can sometimes be a reason to favour x in one way because there is already a reason to favour it in some other way. (For example, perhaps something is the sort of thing that we ought to pursue because it is the sort of thing that we ought to welcome if attained.) The latter reason constitutes x's goodness, and so we can indeed say in such a case that there is a reason (i.e. the former reason) to favour x because x is good.Footnote ³³ Now, presumably such a chain of reasons can arise, but noting this fact does nothing to allay the sense that (1) is true in all cases in which something has intrinsic value, and thus that there is a reason to favour x in the first place (i.e. that the latter reason exists) because x is good.

Another answer is that, although analyses such as IG assert an identity between analysandum and analysans, they nonetheless involve an asymmetry between the former and the latter;Footnote ³⁴ and it is, or at least can on occasion be, correct to say that the analysans is true because the analysandum is. Suppose, for example, that two people are discussing something that they both recognize to be a circle. One of them asserts that the circle is equidistant at all points from some fixed point; the other expresses doubts about this. The first then says, ‘Look, it's equidistant at all points from some fixed point because it's a circle. That's just what it is to be a circle.’Footnote ³⁵ If this sort of statement is coherent, and if it is this sort of ‘because’ that is at issue in (1), then we have an explanation both of how (1) does not threaten IG/View 2/Picture 2 after all and of why (2) does not have the same plausibility as (1).

I doubt, however, that it is this sort of ‘because’ that is at issue in the claim under discussion. The explanation just given about what it is to be a circle presupposes that the person to whom it is offered fails to recognize that something's being a circle requires that it be equidistant at all points from some fixed point. But this sort of ignorance is precisely not what is at issue in the present case. Someone (like me) might acknowledge

IG*: necessarily, x is intrinsically good iff anyone who is aware of x has a moral reason to favour it for its own sake,

and still claim that something's being good explains why there is a reason to favour it. Indeed, there seems to be some pressure to say that a thing's evaluative properties provide a better or deeper explanation than its non-evaluative properties do of why there is a reason to respond to it in some way. Consider the following remark by Crisp:

Here, I take ‘the reason for admiration’ to refer to that which explains, rather than simply grounds, the fact that the painting is fit to be admired. After all, presumably the non-evaluative properties in question do ground the painting's beauty and, thereby, its being fit to be admired. Something similar might be said about the case depicted in Picture 1: although P grounds G and thereby also grounds V, it is G that provides the better or deeper explanation of V.

How might we account for this? Here is one attempt. G is a better explanation of V than P is because (a) V would still be the case, as long as G was, even if P were not the case, whereas (b) V would not still be the case, if G were not, even if P were still the case.

But I think this is hopeless. First, (a) provides no support for Picture 1 over Picture 2; the value-pluralism to which it gives implicit expression is as compatible with the latter as with the former. Second, given that P does indeed ground G, (b) constitutes, as I see it, a counterfactual with an impossible antecedent, and I have no idea whether such counterfactuals are meaningful and, if so, how they should be handled. Third, even if (b) is meaningful, it too seems not to favour Picture 1 over Picture 2. In short, even if (a) and (b) are true, they do not appear to entail that G explains V, let alone that it does so better than P does.

Here is another attempt. Consider again Pictures 1, 7 and 8:

Not only are these pictures, as noted earlier, compatible with one another, but they clearly have something in common. In each case, G grounds V, even though what grounds G differs from case to case. G is thus a better explanation of V than either P, B or K is, because it has greater generality.

I am not sure what to make of this argument. Compare the following pictures:

These pictures also clearly have something in common: in each case, G is identical with V, even though what grounds G differs from case to case. Why should we prefer the first set of pictures (Pictures 1, 7 and 8) to the second (Pictures 2, 9 and 10)? Embracing either set commits one to value-pluralism. In the second set, no explanation is given of how it is that V (= G) can supervene on a plurality of natural grounds (P, B and K). In the first set, although V supervenes in each case on a common ground (G), no explanation is given of how G itself can supervene on a plurality of natural grounds (P, B and K). How then does the generality of G as an (alleged) explanation of V render it a better explanation, when its grounds are as disparate in the first set of pictures as in the second? Perhaps there is a good answer to this question, but I do not know what it might be.

Here is a final attempt. Return to Crisp's illustration. He notes that someone might recognize all the pertinent natural properties of the Madonna but, lacking any aesthetic sense, not recognize its beauty and so not recognize how it is fit to be admired. Similarly, someone might recognize that P is the case but, lacking any moral sense, not recognize that G is the case and so not recognize how it is that V is the case. However, someone who recognizes the beauty of the Madonna, but is unable to identify precisely the underlying natural properties in virtue of which it is beautiful, can nonetheless recognize how it is that it is fit to be admired. Similarly (although the simplicity of the example makes this harder to imagine), someone who recognizes that G is the case, but is unable to identify P as the underlying natural ground of G, can nonetheless recognize how it is that V is the case. Now, this observation does not by itself favour Picture 1 over Picture 2, since someone who subscribes to IG can acknowledge the possibility that someone recognize that ‘both’ G and V are the case without recognizing that P is the case. Thus the observation does not itself show that G explains V, let alone that G explains V better than P does. Nonetheless, it is suggestive.

Compare the following example of causal explanation. Imagine that we have a glass container, x, and some boiling water, and that pouring the boiling water into x (W) causes x's cracking (C). Presumably, W supervenes on certain microphysical facts and events (M) that are causally sufficient for C. But that does not render W causally irrelevant to C. On the contrary, W explains C. Indeed, there is reason to think that W explains C better than M does, since someone might recognize that M is the case but, lacking any macrophysical sense, not recognize that W is the case and so not recognize how it is that C is the case; whereas someone who recognizes that W is the case, but is unable to identify M as the ground of W, can nonetheless recognize how it is that C is the case.Footnote ³⁷

It is of course not necessary to agree that W explains C better than M does in order to accept that W explains C. Likewise, it is not necessary to agree that G explains V better than P does in order to accept that G explains V. There is little doubt that W does explain C, nor that W is distinct from C; for W is a cause of C. I am inclined also to accept that G does explain V, and that G is distinct from V. Of course, G is not a cause of V; I am inclined to take it to be a ground of V. I do not have any argument for this view, but nor do I know of any good argument against it. If I am right, Picture 1 is to be accepted and Picture 2 rejected; if I am wrong, Picture 2 may be correct after all. As far as I can tell, the verdict remains open.

One final point. Let me reiterate that Picture 1, though incompatible with IG, is perfectly compatible with IG*. If you are wondering why anyone might accept both Picture 1 and IG*, consider this analogy. Many people assert that there is a strict equivalence between moral rights and duties that may be put as follows:

1. (3) Necessarily, S has a right against S* that S* do a iff S* has a duty to S to do a.

It is far less common to find people asserting the following:

1. (4) S has a right against S* that S* do a = df. S* has a duty to S to do a.

I suspect that this is in part because it seems natural, at least in many cases, to say that, if S* has a duty to S, this is because S has the relevant right against S*. This would seem to indicate that S*'s duty is grounded in S's right. The sort of right in question is a ‘claim right’, and we could borrow this terminology to express what is at issue in Picture 1: something's being good ‘lays claim’ to our favour of that thing.

APPENDIX 1

I have tentatively embraced a view according to which certain evaluative properties ground certain deontic ones. Might it also be that certain deontic properties ground certain evaluative ones? I think so.

Suppose that some act a is morally right. Is there anything good about that? Nothing that I can see. But what if a is performed: is there anything good about that? It seems reasonable to think so. More generally, it seems reasonable to say that it is good when something morally right or fitting occurs and bad when something morally wrong or unfitting occurs.

It may seem that this view threatens a regress, but I do not think it does. I have suggested that something's being intrinsically good grounds its being intrinsically valuable. If I were now to suggest that something's being intrinsically valuable grounds its being intrinsically good, there would be a regress. But I am not suggesting that. I am suggesting that something's being intrinsically good grounds its being intrinsically valuable, and that its being both intrinsically valuable and valued is itself a good thing. There is no regress in the offing here. Suppose, for example, that x is a state of pleasure and therefore intrinsically good. Its being good calls for a certain response. If that res-ponse is forthcoming, then some extra good has occurred; but if not, not.

APPENDIX 2

Although Scanlon rejects the thesis that x's being good (or bad) provides a reason to respond favourably (or disfavourably) towards x, he appears to accept the thesis that x's being morally right (or wrong) provides, or may provide, a reason to respond favourably (or disfavourably) towards x. He says:

I have exactly the opposite view. Whereas (I am inclined to say) goodness does give us a reason to respond in a certain way towards something, wrongness does not; an act's being morally wrong just is its being such that there is a moral reason not to do it.

Dancy reports a suggestion of Derek Parfit's in support of Scanlon's view about the reason-giving nature of wrongness. Suppose that its being wrong to do some act consists in there being reasons of a certain sort not to do it. If our performing this act would give us a reason to feel guilty and others a reason to feel resentful, then that would appear to give us a further reason not to perform the act in the first place.Footnote ³⁹ Dancy resists this suggestion, pointing out that we can simply say instead that whatever grounds the wrongness of an act also grounds the reason to feel guilty or resentful if it is performed.Footnote ⁴⁰ This seems right. I would add that, if ‘further reason’ is supposed to signify a strengthening of the requirement at issue, Parfit's suggestion seems to pose the risk of an infinite regress. Suppose that the wrongness of some act consists in there being a reason of strength n not to do it and grounds a further reason of strength m to feel guilty or resentful if it is performed, and that this latter reason amplifies the requirement not to perform the act in the first place. Then there is a combined reason of strength n + m not to perform the act. (Addition is not essential to the argument. What is crucial is that the original requirement is amplified to some extent.) If there is a reason of strength m to feel guilty or resentful when an act is performed which is such that there is a reason of strength n not to do it, then (it would seem) there is a reason of greater strength, m + l, to feel guilty or resentful when an act is performed which is such that there is a reason of greater strength, n + m, not to do it. So (it would seem) there is a reason of strength n + m + l not to perform the original act. This line of reasoning can be reiterated ad infinitum. The upshot would seem to be that we might have a reason of infinite strength not to perform an act that is wrong – an absurd thesis.

Thus I side with Dancy against Scanlon and Parfit here. However, I do not think that the regress argument establishes my thesis that an act's being wrong just is its being such that there is a reason not to do it. What the argument attacks is the idea that an act's being wrong strengthens the requirement not to do it. Someone might accept this but still claim that an act's being wrong grounds a requirement not to do it. For example, let ‘P’ stand for x's being a state of pain, ‘W’ for x's being such that it is (or would be) wrong to bring x about, and ‘N’ for x's being such that there is a reason not to bring x about. We could then picture the claim in a way that mirrors Picture 1:

This is not a picture according to which W strengthens the requirement, already provided by P, not to bring x about. Hence the regress argument just given does not count against Picture 11. Nonetheless, I reject Picture 11 in favour of the following analogue to Picture 2:

I have no argument to give here. I see no reason in principle why a deontic property should not supervene on some other deontic property (on the contrary, see the discussion of (3) above); it is just that in this particular case there seems to me to be just one deontic property at issue, not two. Picture 12 is not the ‘whole’ picture, of course. Where ‘B’ stands for x's being intrinsically bad, a more complete picture would, on the view to which I am inclined, be this:Footnote ⁴¹