‘Normativity’ is a relatively new term in metaethics and what is now called ‘metanormative theory’, but the concept is an old one. It is the idea that drives Moore’s ‘open question’ argument, which spawned so much of the metaethics that followed throughout the twentieth century to the present day (Moore, Reference Moore1993).Footnote 1 Moore denied that, as we shall see presently. First, though, we should fix the idea of normativity itself, at least as it has operated in contemporary debates.
1. Rule-Involving v. Reason-Involving Normativity
A good place to start is Parfit’s distinction between ‘rule-involving’ and ‘reason-involving’ conceptions of normativity (Parfit, Reference Parfit2011, I, pp. 144-148). On the rule-involving conception, ‘normativity involves requirements, or rules, that distinguish between correct and incorrect, or what is allowed or disallowed’ (I, p. 144). The rules might be those of the law, a code of honor, etiquette, a game, or a language, just to take some examples. The crucial point is that whether something is normative in the rule-involving sense is settled by actual practice or, as it is said, ‘social construction’. From the fact that something is normative in this sense, nothing whatsoever follows about what there is any reason for anyone to do, to feel, or to have an attitude of any sort toward; nothing follows about normativity in the reason-involving sense.
For ‘reasons fundamentalists’, like Parfit, Scanlon, and Raz, the most basic normative notion is that of a normative reason. As Raz puts it, ‘the normativity of all that is normative consists in the way it is, or provides, or is otherwise related to reasons’ (Raz, Reference Raz1999, p. 67; see also Scanlon, Reference Scanlon2014, p. 2).
But how can we distinguish normative from explanatory or motivating reasons without some independent grasp of normativity? A natural answer is that normative and explanatory reasons have different functional roles. We seek normative reasons when we deliberate: about what to do, what to desire, what to feel, or what to believe. And we seek explanatory or motivating reasons when we want to explain or understand something that has already been done or happened, or that is set to happen.
Normative reasons are sought from the standpoint of someone making up their mind about – or as Pamela Hieronymi calls it, ‘settling the question’ of – whether to X, where ‘X’ may be filled in by some attitude or other: intention, belief, desire, and so on (Hieronymi, Reference Hieronymi, O’Brien and Soteriou2009). It is only when deliberation is complete, our minds made up, that the question of motivating and explanatory reasons arises: what moved, or is moving, Jones to X? Roughly speaking, attitudes just are the states of mind we form for reasons (Hieronymi, Reference Hieronymi, O’Brien and Soteriou2009), where the relevant explanatory reasons constitute reasons for or on which we formed the attitude.Footnote 2 They are one’s reasons. Agents’ reasons are considerations that, from the deliberative perspective, they themselves took to be normative reason to X and on the basis of which they X-ed as they did.Footnote 3
2. Reasons, Ought, Fit, ROF, and Value(s)
Another way of delimiting the normative is through oughts. Ought fundamentalists, like John Broome, hold that oughts are fundamental and understand normative reasons as what supports and explains the relevant oughts (Broome, Reference Broome and Star2018). The idea goes back to Sidgwick, at least. In The Methods of Ethics, Sidgwick says that all ethical judgments contain ‘the fundamental notion represented by the word ought’ (Sidgwick, Reference Sidgwick1962, p. 25).
For our purposes, we need not decide between reasons and ought fundamentalism. Allan Gibbard, whose Wise Choices, Apt Feelings did much to extend philosophers’ attention to normativity beyond ethics, is fond of Sellars’s way of delimiting the normative as that which is ‘fraught with ought’ (Sellars, Reference Sellars1962, p. 44; Gibbard, Reference Gibbard2003, p. 21).
Yet another way of marking out the normative is in terms of ‘fittingness’ or ‘fit’. This idea goes back to Broad (1930) and Ewing (1948) in the Anglophone analytic tradition, and earlier to Brentano (Reference Brentano2009, originally published in 1889). A number of philosophers have argued recently, however, that fit is the fundamental normative idea (Chappell, Reference Chappell2012; McHugh and Way, Reference McHugh and Way2016; Howard, Reference Howard2019). Fit is a relation between an attitude (including action under choice or intention) and some situation, which is in a broad sense the attitude’s object. For example, a desire is fitting just when it fits its object in the sense that its object is desirable. An action or intention is fitting just when it fits what confronts the agent (the object) in the deliberative situation they face in the sense of being choiceworthy in that context. And so on.
Now it might seem that saying that an attitude is fitting is just another way of saying that it is supported by reasons or is an attitude one ought to have. But while these may come to the same thing, they will not if there can be reasons for attitudes that are, as it is said, of ‘the wrong kind’ to support the relevant fittingness judgments (Rabinowicz and Ronnøw-Rasmussen, Reference Rabinowicz and Ronnøw-Rasmussen2004; Hieronymi, Reference Hieronymi2005). Consider, for example, pragmatic reasons for holding a belief that do not bear on whether the belief is credible, that is, whether it is fitting to its epistemic situation. If someone offers to benefit my favorite charity if I believe some obvious falsehood, then there seems to be a sense in which that creates a reason for me to believe it. But such pragmatic reasons for belief have nothing to do with the credibility of what I believe. In this sense, they are said to be reasons of the wrong kind for belief. We can construct similar cases for attitudes of any kind.
The same considerations appear to hold with respect to oughts. There seems to be a sense in which, if my favorite charity can be benefited by my believing something manifestly false, I ought to hold that belief, other things being equal, at least. But this seems to be an ‘ought of the wrong kind’ in the same sense, and if there is no way of explaining why without appealing to credibility, then this might seem to make fittingness a more fundamental notion than ought.
One way that reasons and ought fundamentalists can resist arguments of this sort is, like Parfit, to insist that reasons of the wrong kind cannot exist, since all putative such reasons are not strictly speaking normative reasons for their respective attitudes, in this case, for belief, but rather reasons to desire that one have the attitude (here, belief) (Parfit, Reference Parfit2011). Or they can grant that reasons of the wrong kind exist but attempt to explain what makes them so without appeal to fit. There are some obvious possibilities here even if the details have not yet been adequately worked out. Parfit argues, for example, that the right kind of reasons (which he believes are the only genuine normative reasons) are all ‘object-given’ rather than ‘state-given’ (Parfit, Reference Parfit2011, I, pp. 44–52). Object-given reasons for an attitude are reasons concerning the attitude’s object, by contrast with ‘state-given’ facts about the attitude, say, that it would be valuable to hold it – see Rabinowicz and Ronnøw-Rasmussen (Reference Rabinowicz and Ronnøw-Rasmussen2004) for a critique. Another possibility is to say that reasons of the right kind are those on the basis of which an agent can form the attitude directly, without trying to bring it about in some indirect way as one would have to to get oneself to believe something manifestly false. Reasons might then be determinable as of the wrong kind by whether they could function as the agent’s reason in this way.
For our purposes, we do not need to decide between reasons, ought, or fit fundamentalism as to which is the most fundamental normative notion. Perhaps all are equally fundamental and interdefinable. Rather, we should focus on what is common ground between these views. This is that normativity always concerns a relation between an attitude (again, construed broadly to include action under intention or choice) and a presenting object or situation.Footnote 4 Normative questions, we might say, always concern what attitude or action to take in or toward some situation or object that presents itself for consideration to a deliberating agent (using ‘agent’ in Hieronymi’s broad sense that involves forming any attitude, in a reason-, ought-, or fittingness-guided way). Normativity, at least in what we might now call the reason-, ought-, or fittingness-involving sense (ROF-involving sense, for short), always concerns some attitude or other.
On all three of these approaches, there is a fundamental normative notion, ROF, which is an ineliminable part of any normative concept. Less fundamental, more specific normative notions all conjoin the fundamental normative notion with different attitudes in different ways. The desirable conjoins ROF with desire; the credible conjoins ROF with belief; the estimable conjoins ROF with esteem; and so on. Some normative notions are more complex. I argue, for example, that if the concept of well-being or welfare is a normative notion, it is the concept of what would be ROF desired for someone for their sake (alternatively what, insofar as one cares for someone, one would ROF desire for them; also alternatively, what if one ROF cares for someone, one would ROF desire for them) (Darwall, Reference Darwall2002; Gibbard, unpublished).
An apparent alternative to the ROF account of normativity is what is sometimes called a value-based account.Footnote 5 A significant problem with any value-based approach to normativity, however, is, as Elizabeth Anderson has argued, that there are values of irreducibly different kinds that are conceptually related to different attitudes (Anderson, Reference Anderson1993). The desirable, the estimable, welfare, and dignity are different kinds or dimensions of value. There is no such thing dimensionless value, or value having generic dimension. It makes no sense to ask which has more value, what is most desirable or what is most estimable?
The point is not that, say, the desirable and the estimable are incommensurable values, that they have value in the same sense, but that these cannot be ranked or compared to one another. Moore, as we shall see, thought there was only one form of intrinsic value, goodness or, as he preferred to say ‘good.’ He could have held, consistently with that, that things having (this kind of) intrinsic value, say beauty and friendship, are incommensurable and incapable of being traded off against one another.
Anderson holds what seems to me the obviously correct view that ‘value’ is ambiguous as between different normative notions, for example, desirability, estimability, and the dignity of persons, to name just a few. What makes these distinct normative notions is their conceptual connection to distinct attitudes, to desire, esteem, and (recognition) respect, respectively. To account for the distinctive normativity of these different forms of value, it seems, we must appeal to attitudes of these distinctive kinds that are warranted or fitting, that is, to ROF attitudes. I take this to be a decisive reason to reject a value-based account of normativity as opposed to an ROF account.
2.1 Moore, The Open Question and Normativity
This brings us to Moore and Moore’s ‘open question’ argument, which has loomed so large in contemporary metaethics. Although Tom Hurka may be right that Principia Ethica did not seem terribly original to Moore’s contemporaries, there is no doubt that later generations were deeply influenced by Moore’s claims and arguments about the irreducibility of fundamental ethical notions to non-ethical, naturalistic or super-naturalistic, ‘metaphysical’ notions (Hurka, Reference Hurka2003; Reference Hurka2015). Ayer, Stevenson, and Hare all mention Moore as a major source of their respective forms of irrealist expressivism (Ayer, Reference Ayer1952; Stevenson, Reference Stevenson1937; Hare, Reference Hare1952). For example, Stevenson says that ‘no matter what set of scientifically knowable properties a thing may have (says Moore, in effect), you will find, on careful introspection, that it is an open question to ask whether anything having these properties is good’ (Stevenson, Reference Stevenson1937).Footnote 6
Moore made clear that he thought that the open question showed the irreducibility of the ethical not just to the natural, but also to supernatural, metaphysical claims, for example, those concerning what is loved by God (Moore, Reference Moore1993, p. 91). This is what suggested to irrealist expressivists like Stevenson and Hare that Moore had effectively shown that ethical judgments had a different function than non-ethical ones (whether they were naturalistic or metaphysical).
But what made the open question argument so compelling, at least, at the conceptual level? I don’t think there is any doubt that the plausibility of Moore’s open question relies, as Frankena pointed out over eighty years ago, on the implicit normativity of ethical claims and concepts, including those concerning good.
[T]o my mind, what makes ethical judgments seem irreducible to natural or metaphysical judgments is their apparently normative character; that is, the fact that they seem to be saying of some agent that he ought to do something (Frankena, Reference Frankena and Schilpp1942, p. 102, emphasis added).Footnote 7
Moore himself had cited Sidgwick, indeed, as the ‘one ethical writer’ who had ‘clearly recognized and stated’ the unanalyzability of good (Moore, Reference Moore1993, p. 69). But Sidgwick held, as we saw, that the fundamental ethical notion is ought rather than good. His view was that all ‘ethical judgments’ include ‘the fundamental notion represented by the word ought’ (Sidgwick, Reference Sidgwick1962, p. 25). Moreover, Sidgwick held, as Frankena points out against Moore, ‘ought’ expresses not a simple unanalyzable concept or property, but a relation. The ‘ethical sense of the term ought’, Sidgwick writes, is that of a ‘“dictate” or “precept” of reason to the persons to whom it relates’ (Reference Sidgwick1962, p. 34). The reason why questions of good, and ethics more generally, are conceptually open even after all non-ethical facts, natural or metaphysical, have been stipulated is that ethical questions always concern what attitudes or actions an agent (in Hieronymi’s broad sense) ought to take toward some presenting situation or object. Oughts, and ROF more generally, always concern a relation between an attitude, understood broadly, and a presenting situation or object to which it responds.Footnote 8
Moore may have thought his view did not differ from Sidgwick’s because he held that the claim that something is good or has intrinsic value is identical to the claim that it ‘ought to exist for its own sake’ (Moore, Reference Moore1993, p. 34). But it is unclear how to understand such a nonrelational ought claim. If we take the bearers of intrinsic goodness or value to be possible states of affairs (as Moore-inspired philosophers generally do), then it might mean that the state of affairs’ obtaining is intrinsically desirable. That would identify good or intrinsic value with ROF desire, but what would be doing the normative work would seem to be the ROF relation.
It is useful to note here what Moore himself says about desirability. Remarkably, Moore holds that ‘desirable’ means ‘what it is good to desire’ (Moore, Reference Moore1993, p. 119). In other words, if a possible state of affairs X is desirable, that means that a desire for X has the property of being intrinsically valuable or good. But if we are to understand the normativity of good itself in terms of what ought to exist for its own sake, and to understand the latter in terms of the desirability of that that possible state of affairs obtaining, then it will turn out that a possible state of affairs is desirable just when a desire for that state of affairs is itself desirable. This is a weird view. It makes the desirability of things entirely state-given rather than object-given, to use Parfit’s vocabulary. Surely, the desirability of something turns on whether desire is fitting to the nature of the (potentially desired) thing itself. It turns on whether a desire is supported by reasons or oughts of the right kind, and this concerns the relation of the desire to its object, not some property of the desire for the object considered in itself.
It was normativity in the ROF sense that made Moore’s open question argument so influential for philosophers who followed him, like Stevenson and Hare, and made ‘noncognitivism … the real beneficiary of the open question argument’ (Darwall, Gibbard, and Railton, Reference Darwall, Darwall, Gibbard and Railton1997, p. 119).Footnote 9 The reason is that once we focus on normativity as fundamentally concerned with the deliberative question of what attitude to take toward something, then there will seem to be a conceptual or metaphysical connection between the agent’s judgment that one should have a given attitude and actually having that attitude, barring certain defeaters, at least.
For example, if the normative judgment concerns action, then there will be an obvious attraction to holding that if S were to settle the question of what to do with the normative conclusion that they ought to do A, then, necessarily, they would be moved to do A, again absent defeaters. This is what is known as ‘judgment’ or ‘motivational internalism’ (Darwall, Reference Darwall1983; Shafer-Landau, Reference Shafer-Landau2000). It is part of what gives rise to Michael Smith’s formulation of the ‘moral problem’ (Smith, Reference Smith1994). How can judgments about what someone ought to do be capable of truth and falsity (like beliefs), if they necessarily motivate, and if belief alone does not (the ‘Humean Theory of Motivation’). Smith’s form of naturalism is an example of how one need not be a noncognitivist to feel the force of the ‘moral problem’, or as we might say more generally, the ‘normative problem’, once one sees the connection between moral or normative judgment and the attitudes – in the case of normative practical judgment, intention or choice – they distinctively involve. Mackie’s argument from ‘queerness’ for his ‘error theory’ is another example (Mackie, Reference Mackie1977).
3. Normativity and ‘Modern Moral Philosophy’
To this point, I have been emphasizing that normativity is a relation between an attitude and a presenting situation, which can be considered the attitude’s object, and that different normative notions conceptually implicate different attitudes. The relation might be formulated in terms of normative reasons, or oughts, or fittingness. For our purposes, it does not matter.
In what follows, I want to consider the relation between the concepts that loom large in Sidgwick and Moore – ought and good, respectively – on the one hand, and the notions of deontic morality – right, wrong, and moral obligation – on the other. I believe that an important issue still facing contemporary ethics is to place the latter properly on the normative map. To do that we need to appreciate the nature of the attitudes that deontic morality conceptually implicates.
Sidgwick famously held that a focus on deontic morality was distinctive of ‘modern moral philosophy’, as Anscombe would later call it, echoing Sidgwick’s theme, in contrast with the ethical philosophy of the ancient Greeks and its derivatives, including the classical natural law tradition of Aquinas. Sidgwick put his contrast in terms of fundamental normative notions and faculties (forms of practical reason).Footnote 10
[I]n Platonism and Stoicism, and in Greek moral philosophy generally, but one regulative and governing faculty is recognised under the name of Reason – however the regulation of Reason may be understood; in the modern ethical view, when it has worked itself clear, there are found to be two, – Universal Reason and Egoistic Reason, or Conscience and Self-love (Sidgwick, Reference Sidgwick1962, p. 198).
Sidgwick understood the Greeks as eudaimonists, therefore, as identifying practical reason with ‘egoistic reason’ or ‘self-love.’ By contrast, ‘the modern view, when it had worked itself clear’, holds that there are two fundamentally distinct forms of practical reason, ‘egoistic reason’ or ‘self-love’, on the one hand, and ‘universal reason’ or ‘conscience’, on the other.
Since Sidgwick interprets ‘universal reason’ through the principle of rational benevolence, this may seem to yield a distinction wholly within the good, between prudential and impartial good (Sidgwick, Reference Sidgwick1962, p. 381). Sidgwick stresses, however, that ‘conscience’ in the modern view is conceived with the ‘quasi-jural notions’ of obligation, ‘duty’, and right (Sidgwick, Reference Sidgwick1962, p. 106). We can scarcely understand ancient Greek ethics, he says, unless ‘we throw the quasi-jural notions of modern ethics aside, and ask (as they did) not “What is Duty and what is its ground?” but “Which of the objects that men think good is truly Good or the Highest Good?”’ (Sidgwick, Reference Sidgwick1962, p. 106).
To be sure, the notion of impartial good is also distinctive of an important strain of modern moral philosophy. But even utilitarian or consequentialist theories in which this notion most prominently figures are generally put forward as theories of right and not just of good, whether prudential or impartial. As generally understood, act utilitarianism is not simply the doctrine that actions that produce the greatest overall utility or happiness are impartially best, either in what Parfit calls the ‘impartial reason-involving sense’ or in Moore’s (Parfit, Reference Parfit2011, I, p. 41). It is the deontic thesis that actions are morally right, and so not morally wrong, if, and only if, they produce no less overall utility than would any other available action.
We might put Sidgwick’s contrast, then, by saying that although ancient ethical philosophy focused entirely on the good in its various forms (eudaimonia – well-being – and kalon – the estimable or fine), modern moral philosophy, when it ‘worked itself clear’, also theorized the right conceived in ‘quasi-jural’ or deontic terms.
4. ROF and Deontic Moral Concepts
We might note in passing that philosophizing about normativity is a signal feature of modern philosophy, and relatively recent philosophy at that. I speculate that a major reason for this is the phenomenon Sidgwick identified. So far as ethics is concerned at least, it was only when philosophers had identified at least two mutually irreducible normative ethical ideas, the good and the right, that the question, ‘What feature do these ideas have in common?’ could arise. Only then could the question of the nature and sources of normativity come into view.
What, then, is the relation between deontic moral notions, like moral obligation, right, and wrong, on the one hand, and ought in the general sense involved in ROF, on the other? The first thing to appreciate is that Sidgwick cannot consistently hold that ‘ought’ (in the ROF sense) expresses the same concept as do deontic moral terms like ‘obligation’. The reason is that Sidgwick holds, plausibly, that all ethical judgments contain ‘the fundamental notion represented by the word ought’, (Sidgwick, Reference Sidgwick1962, p. 25). ‘Ought’ in what Gibbard calls the ‘flavorless’ sense that concerns warrant or justification in general (ROF, as we are calling it) cannot express moral obligation in particular. According to Sidgwick, good in the sense of desirability must involve the idea of ought no less than do ‘quasi-jural’ deontic moral notions on pain of not being an ethical notion.
Ewing made the same point about the notion of fittingness, distinguishing between fittingness in the ROF sense and moral obligation. It is a central claim of Hurka’s British Ethical Theorists from Sidgwick to Ewing, that Prichard, Ross, Broad, and Ewing all tended to identify a ‘flavorless’ ought with fittingness (Hurka, Reference Hurka2015). Prichard and Ross are, of course, frequently seen as deontological theorists of the right who oppose Moore’s analytical consequentialism in Principia. Moore held in Principia that the right is conceptually reducible to the good as the property of bringing about greater intrinsic value than any other available act (Moore, Reference Moore1993, §89).Footnote 11 Ross showed that Moore’s own irreducibility arguments about good apply equally to right. If Moore’s analytical consequentialism were correct, Ross observes, act consequentialism would not be normative, much less a controversial, claim (Reference Ross2002, p. 8). It would be what Parfit calls a ‘concealed tautology.’
But although Ross clearly thought ‘right’ expressed a different normative notion than ‘good’, he did not think that it expressed a concept that differed from that expressed by ‘ought’ in Sidgwick’s broad sense or from fittingness as it was understood by Ewing and Broad. This means that Ross’s famous claims about prima facie duties and duties sans phrase were actually claims about what one ought to do in the broadest sense. They were not assertions of moral obligation, right, and wrong, as these might be contrasted with practical ought or normative reasons claims more generally. They were normative practical claims in the ROF-involving sense. Only Ewing, among the philosophers Hurka canvasses saw that deontic moral ideas like obligation, right, and wrong must be understood in some other way. In The Definition of Good, Ewing says, ‘“ought” really covers two different concepts, the concept of fittingness and of moral obligation’ (2012, p. 145).
That the concept of obligation and its correlates must be treated differently from ROF is a consequence of a distinction made by the early modern natural lawyer, Francisco Suárez, between ‘law’ and ‘counsel’, a distinction that was carried forward in modern moral philosophy, especially in Grotius, Hobbes, and Kant, and that is now enshrined, I believe, in moral common sense (Schneewind, Reference Schneewind1990, I, pp. 74–75; discussed in Darwall, Reference Darwall2012). Suárez noticed that deontic concepts of (moral) law and obligation have distinctive entailments of accountability that are not shared by the concept of a normative reason. No matter how weighty the reasons, the claim that reasons recommend or counsel an action is a claim of a different kind than the deontic claim that the action is morally obligatory, required, or demanded. The latter entails accountability for performance in a way that the former does not. Similarly, the claim that an action is morally permissible, not obligatory not to do, cannot be translated seamlessly into the language of reasons.
Although Suárez was in many ways a classical natural lawyer, he thought that Aquinas’s version left out natural law’s (morality’s) power to obligate those subject to it by authoritatively addressing demands to them that they are accountable for complying with.
[C]ounsel is excluded from law …. The word promulgation implies an order for the purpose of creating an obligation and it is in this respect most of all that counsel differs from law [i.e., in not being ‘promulgated’ and obligation creating] (quoted in Schneewind, Reference Schneewind1990, I, pp. 74 –75).Footnote 12
Suárez thought that any such obligation could only result from God’s authoritative commands. And he tied being subject to God’s superior authority to being accountable to him for compliance. If we do not ‘voluntarily observe the law’, he writes, we are ‘legal culprits [i.e., culpable] in the sight of God’ (Suárez, Reference Francisco1944, p. 132). However, the conceptual connection Suárez notes between law or right and accountability does not require any form of theological voluntarism. We might alternatively assume a (transcendental, if you like) community of moral agents with moral obligation entailing accountability to one another as representative members (Darwall, Reference Darwall2006, Reference Darwall2013a, Reference Darwall2013b).
To be sure, there are also rule-involving concepts of obligation, right, wrong, and the like, which can be understood entirely within rule-governed practices, like the rules of a club or game or the laws of nation (or, indeed international law), in which liability to rule-defined sanctions may be involved. But these socially constructed notions are not normative in the sense with which moral philosophers have been concerned. Of course, ethics takes an interest in rule-involving obligations when it asks questions like is there an obligation to obey the law. But the concept in terms of which this ethical question is raised is not itself a rule-involving concept like that of legal obligation where it simply follows from something’s being against the law that there is a legal obligation not to do it. Ethics is concerned with further normative questions like whether there is a moral obligation to comply with (rule-given) law.
Now it might be thought that all this is much ado, since even if there is a difference between general practical ROF claims and claims of moral obligation, the latter might nonetheless be understood as a domain-restricted ROF claim concerning what one morally ought to do, or what there is most moral reason to do. The normativity of moral obligation and its correlates would then not be fundamentally different from ROF and the flavourless practical ought. It would just be that the considerations relevant to the moral ought would come from a restricted domain.Footnote 13
We can use a Moore-style open question argument, however, to show that that cannot be so. We only have to imagine the possibility of a coherent disagreement between two people who agree that a given action is best supported by moral reasons, hence is most morally choiceworthy and in that sense what the agent morally ought to do, but who disagree about whether the action is morally obligatory. They could agree that an act is morally ROF, that is, but coherently disagree about whether it is morally required or something one morally must do (Darwall, Reference Darwall, Lord and Maguire2016, pp. 265–266).
Ewing, again, distinguishes between fittingness and moral obligation. The former is the fundamental normative notion that is involved in any normative judgment, whether about the good in any of its forms, the right, or some other normative idea like credibility. All normative judgments concern the relation between an attitude of some kind and a presenting deliberative context that is the attitude’s object. It follows, for Ewing, that if judgments of moral obligation are normative, they must concern some attitude that is distinctive of the concept. However, whereas the attitude that is implicated in practical ROF judgments is intention or choice, Ewing holds in The Definition of Good that moral obligation judgments conceptually implicate an attitude he calls ‘moral disapproval’. To say that something is morally obligatory or wrong not to do, he there writes, is to say that ‘we should be fitting objects of moral disapproval if we did not do [it]’ (Ewing, Reference Ewing2012a, p. 173). And he adds: ‘“To be a fitting object of disapproval”, is equivalent to “deserving blame”’ (Ewing, 2012, p. 169).
To put the claim in a way that provides for the possibility of excused wrongdoing, I prefer to formulate Ewing’s idea this way: ‘Necessarily, an act is wrong (violates a moral obligation) if, and only, if it is an act of a kind that it would be blameworthy to perform, were the agent to do so without adequate excuse’ (Darwall, Reference Darwall, Darwall, Gibbard and Railton2019, p. 5). This is the account of deontic moral concepts that I have defended in my work (Darwall, Reference Darwall2006, Reference Darwall2013a, Reference Darwall2013b, Reference Darwall, Darwall, Gibbard and Railton2019). And, following Strawson, I have emphasized that the attitudes that are implicated in moral obligation are ‘reactive attitudes’ through which we hold one another and ourselves accountable (Strawson, Reference Strawson1968). When we have reactive attitudes like blame, we take up a distinctive ‘participant’ standpoint, one of participation in a relationship with the blamed, even if only the relationship of fellow person, and implicitly relate to them (second personally).
The claim that an action is blameworthy is itself an ROF claim. It says that blame is a fitting attitude toward the action. Now, on the face of it, the claim that an act is morally obligatory, or wrong not to do, seems to say something normative about the act and not about attitudes one might have in response to its being performed or not. But we have seen that deontic moral claims cannot be understood as practical ROF claims on pain of running over Suarez’s distinction between law and counsel. Ewing’s solution, which I endorse, is to understand the distinctive normativity of deontic moral claims in terms of blameworthiness claims. Wrongness and blameworthiness are not identical concepts, since an action can be wrong without being blameworthy if the agent has a valid excuse. Nevertheless, it is a conceptual truth that if an act is wrong (violates moral obligation), then it is an act of a kind that would be blameworthy were it done without excuse.
The puzzle persists, however. Deontic moral claims seem to be about acts themselves, not, or at least, not just, about attitudes toward the performance of those acts. I argue that the right way to dissolve this puzzle is to see that although deontic moral claims are not themselves ROF practical claims, they nonetheless entail ROF practical claims. What makes this so, I claim, is a presupposition of blame as a reactive attitude (Darwall, Reference Darwall, Lord and Maguire2016).
Bernard Williams, among others, noticed that it is incoherent to blame someone for doing something one thinks they had sufficient normative reason to do (Williams, Reference Bernard1985, p. 193; Gibbard, Reference Allan1990, pp. 299–300; Skorupski, Reference Skorupski1999, pp. 42–43; Shafer-Landau, Reference Shafer-Landau2003, pp. 181–183; Darwall, Reference Darwall2006; Portmore, Reference Portmore2011, p. 44). Some kind of inconsistency seems to be involved in having the attitude of blame toward someone while simultaneously fully accepting that they had good and sufficient reason for doing what they did. After all, if someone can establish that they had sufficient reason, then they will have adequately accounted for themselves. They will have justified what they did and shown, therefore, that it was not actually wrong. Blame must therefore assume that they cannot, that there was insufficient reason for what they did.
Now, if an act is morally obligatory, all things considered, it is conceptually necessary that it is an act of a kind that, in the circumstances, it would be blameworthy to fail to perform without excuse. But if that is so, it follows that the act must be one we cannot have sufficient reason to omit, because if we did have such reason that would show that blame would be unjustified (Darwall, Reference Darwall, Lord and Maguire2016).
On the Ewing-inspired approach I favour, the distinctive normativity of deontic morality consists in the fact that deontic moral judgments are normative, in the first instance, for reactive attitudes rather than for action. The deliberative question they distinctively concern is whether to blame someone for an act of the relevant kind in the relevant circumstances, assuming the agent lacks an excuse. This gives us the conceptual space to hold with Suarez that an act can be morally ROF without being morally required. If the act would be wrong, that is, culpable lacking excuse, then it is conceptually guaranteed that it is not one the agent could have sufficient reason to choose. The choiceworthiness of obligatory conduct (their being ROF acts) is not the same thing as their being obligatory but is rather a conceptual consequence of presuppositions of blame, which is the attitude for which deontic moral judgments are normative in the first instance.
Ewing later gave up this fitting-attitude analysis of deontic moral concepts like wrongness. In Second Thoughts on Moral Philosophy (1959), he says that if we hold that being wrong is being something ‘it would be reasonable to blame [someone] for not doing’, we will not be able to say that someone ‘ought to be blamed for not doing [what is morally obligatory] only because it was morally wrong not to do it’ (Ewing, Reference Ewing2012b, p. 96, emphasis added). However, a Ewing-style fitting-attitude analysis of ‘wrong’ should not say that being wrong is the same thing as being something it would be fitting, or ROF, to blame, since it can recognize that there is excusable wrongdoing. To say that an act is wrong is to say that it is the kind of act there is reason to blame someone for doing should their action be unexcused. So we can still meaningfully say that an action can be justifiably blamed (at least partly) because it is wrong.
Somewhat similarly, in his discussion of the concept of wrongness in On What Matters, Parfit distinguishes between what he calls the ‘reactive attitude’ sense and the ‘ordinary’ sense of ‘wrong’ (Parfit, Reference Parfit2011, I, p. 165). To say that an act is wrong in the ‘reactive attitude’ sense is to say it is ‘an act of a kind that gives its agent reasons to feel remorse or guilt and gives others reasons for indignation and resentment’ (Parfit, Reference Parfit2011, I, p.165). This is a variant of the account I propose. The ordinary sense of ‘wrong’, by contrast, Parfit says, is ‘indefinable.’ As with the later Ewing, what Parfit calls the ‘ordinary’ sense of wrong may seem to provide a more robust way of saying that someone is to blame because they have done something wrong than seems to be available on the ‘reactive attitude sense’.
If it does so, however, it comes at a significant cost, since it undercuts the possibility of Parfit and the later Ewing explaining what makes wrongness and other deontic moral concepts normative concepts. For Parfit, normativity in the reason-involving sense must bottom out in normative reasons. And for Ewing, the fundamental normative concept is fittingness. But unless wrongness can itself be understood in ROF terms, then it will be possible to stipulate that an act is wrong and it will remain a conceptually open question whether there is any reason, or whether it is fitting, to take any attitude or action toward it. For normative concepts, this is precisely the kind of question that should not be open. Or to put the point the other way around, it is the fact that the questions Moore and his followers consider are open that shows that the candidate natural or metaphysical concept is not a normative concept, so not an ethical concept. On ROF accounts of normativity, like Parfit’s and Ewing’s, wrong can be a normative concept only if the fact that an act is wrong has conceptual entailments about ROF attitudes. This point about wrong is the analogue to Frankena’s objection to Moore about good.
We began with Parfit’s distinction between rule-involving and reason-involving conceptions of normativity and saw that reasons, ought, and fit fundamentalists can all agree that normativity always concerns a relation between an attitude and a presenting situation that can be considered the attitude’s object. Different normative concepts conceptually implicate different attitudes. We turned to the question of how to locate the deontic moral concepts of right, wrong, and obligation on an ROF account. Suarez’s distinction between law and counsel helps us see both that the deontic ‘must’ of obligation must be distinguished from the ‘flavorless’ ‘ought’ that, on ROF accounts, is implicated in all normative judgments. I follow Suarez, the early Ewing, Strawson and others, in holding that the attitudes for which deontic morality is normative in the first instance are ‘reactive attitudes’ through which we hold one another and ourselves accountable. As Strawson has taught us, we hold such attitudes from a second-person point of view. It follows that the deliberative perspective that is implicated in deontic morality is the second-person standpoint. The implications of that, however, is a topic for another day.
Open access
