Introduction: characterizing ethical realism
It is useful to begin a survey of recent work on ethical realism with a look at current disputes over what makes a theory of ethics count as ‘realist’ in the first place. Nearly all characterizations of ethical realism include some version of the following two core claims:
Ethical discourse is assertoric and descriptive: ethical claims purport to state ethical facts by attributing ethical properties to people, actions, institutions, etc., and are thus true or false depending on whether their descriptions of things are accurate or not (and similarly, the ethical beliefs expressed by such claims are true or false depending on whether their representations of things are accurate or not).1
At least some ethical claims, when literally construed, are true in the above sense.
Sayre-McCord (2008) maintains that these two conditions are necessary and sufficient to characterize ethical realism; others argue that further conditions must be added to yield ethical realism, or at least ‘paradigmatic’ or ‘robust’ ethical realism. Before turning to that, however, it is worth noting that some have recently denied that the above two conditions are even necessary for ethical realism. In probing new work on metaethical taxonomy, Miller (forthcoming) argues that the core of ethical realism is simply the metaphysical commitment to objective ethical properties and facts, which is strictly independent of the semantic claims above. If one holds that there are objective facts about the wrongness of slavery, for example, then one ought to count as an ethical realist regardless of whatever views one might hold about the semantics of the contingent forms of discourse that have evolved among language users. Even if our ethical discourse turns out primarily to express conative states (as most expressivists have claimed, in violation of (i)), or all moral claims turn out to be false because they’re misguided or infected by false presuppositions (as error theorists have claimed, in violation of (ii)), what matters as far as ethical realism is concerned is simply the existence of objective ethical properties and facts.
On this approach, then, our focus in demarcating the realist/antirealist opposition should be on moral ontology rather than on semantics. Of course, this will make little practical difference if few realists will want to defend the existence of objective ethical properties and facts while holding that we never correctly attribute or represent them in our ethical judgements. And it's likely that few will take such a position, since this combination of views would be very difficult to motivate: even the compelling attribution of wrongness to slavery would fail to point us to the alleged sort of real property or fact, and so couldn’t help to nudge us toward realism. Realists, then, will tend to accept the claims in (i) and (ii), while anti-realists will reject at least one of them.
Still, the proposed shift from semantics to metaphysics may be important. As Dreier (2004) has lucidly brought out, the very existence of a substantive debate between ethical realists and antirealists is threatened by the problem of ‘creeping minimalism’, wherein sophisticated antirealist (or irrealist) positions have become so flexible in their accommodation of the trappings of realism, by deflating talk of ‘truth’, ‘facts’, ‘properties’, ‘beliefs’, etc., that it has become difficult to characterize what, if anything, sets realists apart from antirealists. For example, if saying that a claim is true is nothing more than a way of affirming that claim – i.e. if the notion of truth is exhausted by the platitude that p is true iff p – then even expressivists can allow for ascriptions of truth to moral claims. If one expresses one's disapproval of slavery by saying ‘slavery is wrong,’ one can equally express that disapproval by saying ‘it is true that slavery is wrong.’ But then similar deflationary moves might likewise be made for the notions of ‘description’ and ‘accuracy’: claim C describes X as being F iff C has the content that X is F; and this description is accurate iff it is true that X is F, i.e. iff X is F. And now it seems that minimalist antirealists may be happy to accept conditions i and ii after all. If so, then what has become of the realist/antirealist debate? Should we declare it a non-issue and take up a quietest position?
There have been various attempts to deal with this problem and preserve the debate (Copp 2007: 156–59, 195–97, Dreier 2004: 32–42, Gibbard 2003). Miller's suggestion is that the metaphysical focus on the realist's claim of mind-independence for moral properties and facts best captures the divide between the two camps – at least once mind-independence is properly understood. This requires, of course, resisting Blackburn's (1998: 311) familiar deflationary attempt to convert the idea of mind-independence from a second-order claim about the metaphysics of ethical properties and facts into a first-order expression of ethical commitment that can again be accommodated by the antirealist. (For example, Blackburn would note that we do not make our disapproval of cruelty to animals contingent on how people happen to feel about it, which means we can express this attitude by saying ‘cruelty to animals is wrong and would be wrong even if people approved of it’, or equivalently: ‘the wrongness of cruelty to animals is mind-independent’.) Miller argues against this form of ‘creeping minimalism,’ freeing up the idea of real mind-independence to do useful work in characterizing realism.
Many who, unlike Miller, retain conditions (i) and (ii) above as central to ethical realism would agree in any case that an objectivity condition needs to be added to the characterization (FitzPatrick 2008a, Huemer 2005, Oddie 2005, Shafer-Landau 2003), though some do not (e.g. Cuneo 2007: 45–49, who rejects the mind-independence condition and allows for constructivist views to count as paradigmatically realist). Miller's condition requires a relevant independence of ethical facts from both our intentional attitudes and our conceptual schemes: the realist fact that slavery is wrong, say, requires that this fact ‘exist unchanged in nearby worlds in which … human beings believe that slavery is permissible, and furthermore the fact's existing in those nearby worlds must have nothing to do with our conceptual schemes or intentional attitudes pertaining to slavery in the actual world’. It is not enough just to be able to say that slavery is wrong even in other possible worlds inasmuch as we will evaluate it as wrong there too because we will (naturally) be applying our actual conceptual scheme; for that would be compatible with the concession that facts about wrongness are merely relative to contingent conceptual schemes. For the realist, the wrongness of slavery, whether in the actual world or in other possible worlds, is not dependent on our contingent attitudes or conceptual schemes: the reality of the wrongness of slavery, we might say, is deeper and more robust than any such dependency would capture.
This is related to Shafer-Landau's (2003: 15) condition of stance-independence: ‘there are moral truths that obtain independently of any preferred perspective, in the sense that the moral standards that fix the moral facts are not made true by virtue of their ratification from within any given actual or hypothetical perspective.’2 The addition of this stance-independence condition has the potentially confusing result that many views typically taken to be paradigms of realism – such as Railton's (1986) and Smith's (1994) – fail even to count as realist after all. For this reason, others opt to leave this condition out of the description of ethical realism per se, but then distinguish stance-dependent forms of realism from a more ‘robust’ realism that incorporates stance-independence and other conditions. FitzPatrick (2008a: 166) understands robust ethical realism to require (among other things) that ‘ethical standards and facts are independent of us in the sense that they are not constituted by the actual or hypothetical results of any ethically-neutrally specifiable set of conditions or procedures applied to our beliefs, desires, attitudes, etc.’ The thought here is that the ethical facts may always in principle outrun the outputs of any ethically-neutral procedure applied to elements of our psychology (such as correcting for empirical ignorance, inconsistency, or lack of deliberation), and cannot properly be held hostage to such outputs – a point tied to a form of ‘Open Question Argument’ and to claims about the autonomy of ethics. (On the idea of ‘robust’ realism, see also Oddie 2005 and Enoch 2007.)
Cuneo (2007) and FitzPatrick (2008a) both also include in ‘paradigmatic moral realism’ or ‘robust ethical realism’, respectively, a feature left out of many theories (especially naturalistic ones) described as more broadly realist: namely, the categorical normative force of moral facts. Slavery's being wrong in a robustly realist sense is not just a matter of its possessing a real property referred to by our term ‘wrong’, but essentially involves its having an irreducibly evaluative property that gives us genuine reasons to reject slavery, regardless of our contingent desires and interests. (Cf. also Huemer 2005.)
Though there is much more that could be said about recent characterizations of ethical realism, we will turn now to some particular argumentative strategies, developments and defences of ethical realism. It should be emphasized that space allows for only a selective survey, and there is much excellent and important work that will regrettably have to be left out.
2. Some recent developments of non-naturalist ethical realism
There is probably no better way in to the topic of ethical realism than Shafer-Landau's (2003) excellent and highly accessible book, which offers a comprehensive defence of a version of realism in connection with the central debates of twentieth century metaethics. It is thus a valuable teaching resource as well as an important contribution to the metaethical debates. (Much the same can be said for Huemer's (2005) rich and wide-ranging book, which defends most of the same realist territory in the guise of rationalist intuitionism. Along similar lines, see also Audi (2004) for an important development of intuitionism combining a refined and expanded version of Rossian intuitionism with elements of Kantian ethics; the result is a ‘value-based Kantian intuitionism’ with potentially attractive implications for normative ethical theory as well. Stratton-Lake (2002) also provides a first-rate collection of recent work on ethical intuitionism from a variety of perspectives. For an intriguing contrasting account of moral epistemology within a robust value realism, focused not on intuition but on desires as experiences of value through which we causally gain knowledge of value, see Oddie 2005.)
Shafer-Landau defends a version of realism that is non-reductive and non-naturalistic, according to which the sphere of ethical inquiry is discontinuous from scientific inquiry and admits of a priori knowledge (with at least some moral principles being self-evident in a qualified sense – a view also developed by Audi 2004 and Huemer 2005). In the process, he develops arguments against several claims that have often been used to support antirealism, such as ‘Motivational Humeanism’ (the view that beliefs in themselves are always motivationally inert), ‘Motivational Judgement Internalism’ (the view that moral judgements necessarily pack some motivational punch for the person making the judgement) and ‘Reasons Internalism’ (the view that the normative reasons that exist for one to do something are contingent upon one's ability to be motivated by them, at least under conditions of full empirical information and procedurally rational deliberation). He also defends ‘Moral Rationalism’ (the view that moral obligations entail genuine normative reasons for so acting, so that morality is ‘intrinsically normative’).
While Shafer-Landau's view is non-naturalistic in terms of methodology and epistemology, its metaphysics of ethical facts and properties is roughly equivalent to Brink's (1989) non-reductive ethical naturalism, where ethical facts and properties are exhaustively constituted by natural ones. This raises questions about how non-naturalistic Shafer-Landau's view really is, and whether it can effectively handle the worries about normativity that typically arise for ethical naturalism (though see Copp 2007 for attempts to deal with such problems). That is, although Shafer-Landau's view is non-reductive (rejecting any type-reductions of ethical facts or properties to natural ones of the sort identified by the sciences, describable in non-ethical terms), one may wonder whether it has the resources to make sense of the normative aspect of ethical facts and properties, such as the categorical reason-giving force attributed to them at least by robust realist views. Specifically, it is unclear how Shafer-Landau's view that all evaluative and normative facts and properties are exhaustively constituted by natural ones coheres with his own commitment to the ‘intrinsic normativity’ of morality. The latter entails that there are normative facts (or ‘metafacts’) such as: stealing's being wrong is a reason (for every agent) not to steal, or merits being taken seriously in deliberation.3 But what set of natural facts, one might wonder, could possibly constitute that normative fact – especially if we have already rejected, as Shafer-Landau has, any stance-dependent account of such normative facts (such as a desire-based account of reasons), and so cannot reduce them to facts about potential motivation under certain idealized conditions? A realism that incorporates ‘intrinsic normativity’ may require not merely a rejection of reductionism, but a rejection of the exhaustive constitution thesis and an acceptance of clearly non-natural properties and facts (FitzPatrick 2008a).
There are different ways of developing such a view. At one end of the spectrum, with heavy (many would say extravagant) metaphysical investment, FitzPatrick (2008a) proposes a dual-aspect view modelled on similar views in the philosophy of mind, according to which many features of the world that are empirically investigable are also inherently value-laden: in addition to their empirical aspects that can be examined scientifically (and that are considered ‘natural’ to that extent), they also have irreducibly evaluative aspects discoverable through ethical experience and reflection; suffering, for example, is subject to empirical investigation, but it also has an evaluative and normative aspect that we know through experience of our own and others’ suffering and from reflection on its evaluative and normative significance – its badness and the pro tanto reasons this badness provides to avoid or mitigate suffering.4
At the other end of the metaphysical spectrum, Parfit (2006, forthcoming) likewise argues for non-naturalist normative realism, but rather than understanding reasons as grounded in metaphysically real values in the world, reasons are taken as fundamental and facts about value are reduced to facts about reasons (i.e. to be good just is to possess features that give us reasons of certain kinds). According to this ‘Non-Metaphysical Intuitionism’ (Parfit forthcoming), there are irreducible and non-natural facts about reasons (and so derivatively about values), but these do not carry metaphysical commitments: they do not involve any Platonic commitment to normative facts about some non-spatio-temporal part of reality, and Parfit doesn’t think they require metaphysical claims about ordinary reality either; indeed, on his view, normative facts aren’t ‘part of the world’ at all, any more than facts about numbers are. (Compare, in this respect if not others, Putnam's (2004) ‘ethics without ontology’, which rejects the typical realist appeal to ethical truth-makers in the world, taking this to be unnecessary for establishing objectivity, again as in mathematics.) While this view thus avoids what many might regard as the metaphysical extravagance of views that posit objective values in the world, however, the challenge will be to make sense of the rich spectrum of posited irreducibly normative facts about reasons without any appeal to any prior values and without any metaphysical commitments beyond those recognized by naturalists.
3. The problem of supervenience
Another metaphysical topic central to recent debates is the explanation of supervenience. It seems that in any given case the ‘descriptive’ properties and facts (i.e. those picked out by non-ethical predicates or claims) fix the ethical properties and facts, such that two cases could not differ in the latter without differing in the former.5 Indeed, this seems to be a conceptual truth: even if it is not a conceptual matter as to which descriptive properties underlie which ethical properties, it is a conceptual truth that given that a set of descriptive properties D underlies ethical property E in one case, it must do so in other cases (at least where D was fully specified to take account of the absence of potential defeaters, as implied by speaking of D as ‘underlying’ E in the first case); if something lacks E, it must differ from the first case in some descriptive respect. The question is how non-reductive realists can explain this conceptual constraint, given that it's not a conceptual matter which descriptive properties fix which ethical ones.
Shafer-Landau (2003: 85f.) seeks to dissolve the mystery by noting that the last point doesn’t imply a lack of any entailment between descriptive properties and ethical ones, as if they must just be free-floating: there may still be a metaphysical entailment, making their covariance unsurprising. (It may even be a conceptual matter that there is such a metaphysical entailment if there are any moral properties at all, even if the details are not a conceptual matter. Compare the relation between molecular compositions and chemical substances; though it is not a conceptual matter that H2O is water, it may be a conceptual matter that if H2O is ever water then it always is.) And on his exhaustive constitution view, which he shares with non-reductive naturalists (Brink 1989), supervenience is exactly what we would expect: if it is granted that D constitutes E in one case, then it would be a conceptual matter that another instance of D will likewise constitute E, so that two cases that are exactly alike in terms of D will not vary with respect to E.
This last move, of course, will not be available to more robustly non-naturalist views that reject the exhaustive constitution thesis. So, for example, FitzPatrick (2008a) takes a different tack, appealing to a combination of metaphysical entailment and a standard-based account of (non-basic) ethical facts: non-natural ethical standards will be fixed (and metaphysically entailed) by irreducibly evaluative aspects of the same world we call ‘natural’ when considered in its descriptive aspects, so that the standards will be the same in any two worlds with relevantly similar descriptive natures; and an action will be right or wrong depending on whether its particular descriptive properties qualify it as meeting or failing to meet those standards – thus explaining why any two things with all the same descriptive properties must possess the same ethical properties, just as two watches with all the same descriptive properties must be evaluatively identical as good or bad watches. (Cf. Copp 2007, who offers a related but naturalistic standard-based account of moral judgment.) Others offer further accounts of supervenience in defence of real, irreducible value, some of them quite technical (Oddie 2005: Ch. 6). In any case, this remains an area of controversy, with critics raising difficulties not only for Shafer-Landau's handling of supervenience (Mabrito 2005), but for any non-reductive accounts, whether non-naturalist or naturalist (Ridge 2007).
4. Realist expressivism
Ethical realism has long faced a powerful challenge stemming from Hume and pressed by expressivists over the past century, emphasizing the connection between ethical judgement and motivation: if ethical judgements necessarily carry some motivational force, and beliefs in themselves cannot motivate, then it looks as if moral judgements are not, or do not express, beliefs (at least not beliefs as usually understood, with representational content). In illuminating recent work on moral semantics and pragmatics, Copp (2007: Ch. 5) has taken a major step in defusing this line of resistance to realism, while also answering certain objections to ethical naturalism.
Copp's strategy is to capture the phenomena that have seemed to support antirealist expressivism within a fully realist model, through what he calls ‘realist expressivism’. On this view, moral terms refer to real moral properties, and moral beliefs have objective truth conditions specified in terms of a standard-based account within a society-centred theory; an act's wrongness in a given society, for example, consists roughly in its being prohibited by the set of standards whose currency in the society would best enable that society to meet its needs. Hence the realism: when we assert moral claims, we express such beliefs, and some such beliefs are true. This is only half the story, however. What we typically express in sincerely asserting moral claims is a complex state of mind comprising both a cognitive moral belief and a conative state – a state of subscribing to the standards in question, involving endorsement and an intention to comply with them. This conative state is not part of the moral belief, and does not affect its truth conditions, but because moral terms have a distinctive ‘colouring’ due to associated linguistic conventions, one does conventionally implicate that one is in such a motivational state by sincerely asserting a moral claim. And this captures the expressive quality of moral discourse that expressivists have highlighted – the fact that sincere moral assertions, unlike assertions in non-moral language, carry with them the implication of a motivational element of endorsement or commitment, much as uses of loaded terms such as ‘cur’ implicate certain attitudes. Yet all of this is compatible with realism.
Copp's account is a model for how to bring sophisticated work in other areas – in this case the philosophy of language (going back to Frege) – to bear on metaethical issues, and if it is successful it shows that one should not be tempted away from realism and toward expressivism by the practicality or motivationally-laden nature of moral discourse: realist expressivism can equally capture this aspect of moral discourse. It also advances the prospects for a naturalist form of realism, as it helps to explain the intuitions behind ‘Open Question Arguments’ against ethical naturalism, or convictions that there is an unbridgeable fact/value gap. On Copp's view, it's true that the meaning of moral terms is not fully captured by non-moral descriptions, just as the meaning of ‘cur’ is not fully captured by ‘mixed-breed dog’, in both cases due to the ‘coloring’ of the terms whereby one implicates certain attitudes in using them (though the truth of one's claims is not contingent on whether or not one actually has the attitude). This gives the impression of ‘open questions’ or gaps whenever non-moral descriptions are proposed to explicate moral properties. Yet this sort of gap, due to the conventional implicature surrounding moral terms, does not tell against the identification of moral properties with natural ones (199).
5. Developments in naturalist ethical realism
There have been other important recent developments in naturalist forms of realism, of which we will note two. Sturgeon (2003, 2006) has emphasized that while ethical naturalists hold that ethical properties are natural properties, they need not claim that an ethical property such as goodness can be identified with some natural property for which there exists some non-ethical term or description, such as pleasure. ‘Goodness’ refers to a natural property, but this need not be a property we can also pick out in purely descriptive terms, and so there may be no way of completing the sentence: ‘Goodness = _____’, as naturalists have often assumed there is. Sturgeon's proposed ‘one-term’ non-reductive naturalism, as Dancy (2006: 127) calls it, would explain why naturalistic identifications repeatedly run into trouble, as various forms of ‘Open Question Argument’ have brought out, but in a way that leaves naturalism intact: the identifications seem always to leave open questions (e.g., ‘granted X is pleasant, but is X really good?’) not because ethical properties are not natural properties (they are, for Sturgeon), and not just because ethical and non-ethical terms are not synonymous (though that may be true), but because ethical properties simply aren’t identical with any independently identifiable natural properties. Another advantage of this one-term naturalism is that it avoids a redundancy objection that plagues other versions. If goodness were pleasure, for example, then the claim ‘this pleasant thing is good’ would be redundant in a way it does not seem to be, with ‘is good’ just restating the fact already stated by describing the thing as pleasant, which does not seem to capture the sense of an evaluation. By refusing to make any such identification, however, the one-term naturalist avoids this form of objection (Dancy 2006: 132).
Brink (2001) has also significantly advanced non-reductive naturalist realism with helpful new work on moral semantics. The ‘Cornell Realism’ or ‘Synthetic Semantic Naturalism’ associated with Brink (1989), Boyd (1988) and others, has come under attack by Horgan and Timmons (1991), who employ a ‘twin earth’ style argument to show that the causal regulation model of reference developed by Boyd (1988) fails: if moral terms rigidly designate the objective, natural properties that causally regulate the use of those terms, then many cases of what seems clearly to be moral disagreement would turn out (implausibly) just to be cases where people are talking about different things and not disagreeing at all; such a semantics also implies far more relativism than a realist should allow. Brink responds by revising the semantics. One possibility involves an interpretation of causal regulation that appeals to counterfactuals incorporating conditions of reflective equilibrium, rather than focusing on actual, historical causal regulation of moral terms. Another possibility is to look instead to shared referential intentions to link speakers’ uses of terms to the relevant causal-historical chains, which Brink argues establishes – when combined with a common concern with interpersonal justification – common meaning and reference in moral discourse sufficient to avoid the twin earth style problems. For a related defense of naturalist semantics against twin earth objections, see Copp 2007: Ch. 6.
6. The ‘companions in guilt’ strategy
One popular way of defending ethical realism is to exploit a ‘companions in guilt’ strategy: when objections are raised to allegedly problematic features of realism, find parallels from outside of ethics that share the same features and yet do not appear problematic (at least not sufficiently so to make the same critics reject them), thus defusing the objections to realism. This strategy – which might more precisely be labelled the ‘companions in innocence’ strategy – is explored in depth by Lillehammer (2007), who critically examines its employment by Davidson, Putnam, Hampton, Korsgaard, McDowell and others to defend various forms of ethical objectivity. Hampton (1998) and Korsgaard (1997), for example, employ such a strategy in defending categorical moral norms, which is a form of objectivity important to robust realists (though Korsgaard, for example, rejects other elements of realism, so her use of this strategy is not intended to be a defence of ethical realism as a whole). They argue that Humeans, who reject categorical imperatives in morality on the grounds that there are no categorical norms at all, and who thus accept only hypothetical imperatives, are in an untenable position. For the normativity of particular hypothetical imperatives depends on the normativity of the formal hypothetical imperative (i.e. the imperative to be instrumentally consistent), and that formal imperative is itself categorically prescriptive (because rooted in the very conditions of agency); thus, the allegedly objectionable feature of categorical moral norms is something to which Humeans are already committed just in accepting hypothetical imperatives.
Lillehammer (2007: Ch. 3) argues that while this argument works against crude neo-Humeanism, it is possible to avoid it with a minimal concession. Humeans can grant that there is categoricity in the case of the formal hypothetical imperative, understood as a formal normative requirement, since there is a compelling explanation of the grounds of such categoricity; but they can still reject categorical prescriptivity in ethics, since there is no similarly compelling account of that sort of categoricity, which involves categorically required ends (unlike the mere means/end consistency requirement of the formal hypothetical imperative). Humeans often go too far in rejecting categoricity across the board as ‘queer’, but if they make the above sorts of distinctions they can allow the kind of categoricity they need and still reject anything like categorical moral imperatives. So, Lillehammer argues, this employment of the companions in guilt strategy against Humeans fails to support categorical moral norms.
An excellent recent example of a book-length development of a ‘companions in guilt’ strategy is Cuneo (2007), which is the most extensive examination to date of the parallels between moral normativity and epistemic normativity with a view to defending a realist account of both. Many moral antirealists reject moral facts because of certain allegedly problematic features of such facts. Cuneo, however, argues that these same features are shared by epistemic normative facts, such as facts about epistemic justification or rationality, which these same critics mostly accept: there is a parity between moral and epistemic normative facts. He also argues that we should accept the latter, which means that we should therefore not reject moral facts simply because they have the features in question. Antirealist arguments against moral facts based on the alleged queerness of these features would prove too much, leading equally to an implausible antirealism about epistemic facts.
To take an example, we have already seen how the supervenience of the ethical on the natural has been taken by antirealists to pose a problem for (non-reductive) realists, and how realists have attempted to resolve this problem. Cuneo adds to this defence by noting that a closely parallel supervenience equally obtains for epistemic normative facts and properties which, again, most philosophers admit into their ontology. For example, if a belief has the normative property of being warranted or rational, this will be so by virtue of certain underlying descriptive properties, and (as a conceptual matter), properties such as being warranted or rational cannot vary if all the descriptive properties are held fixed. This does not seem to be a deep mystery that leads people to reject epistemic facts, and so by the same token we should not take moral supervenience to be a reason to reject moral facts (2007: 90–2). Another very interesting application of this strategy is to the familiar problem of moral disagreement (107–12): for just as people have persistent disagreements over first-order moral judgements, so too they have persistent disagreements over first-order epistemic normative judgements, e.g. over the rationality of religious belief, epistemic norms governing the role of authority, faith, and so on. Yet we don’t tend to deny the existence of facts about rationality or warrant just because there are such persistent disagreements. Why, then, should disagreement about moral norms lead to moral antirealism? (For another, extended defence of realism against the argument from disagreement, see Huemer 2005: Ch. 6.)
As in the case of categoricity, ethical antirealists might respond with a combination of concession and an attempt to show that the required concession is not damaging to their position. They might concede that we should be realists about epistemic facts (and so admit that there are some real normative facts), but deny that this forces us to realism about moral facts, since there are good grounds for epistemic realism that don’t carry over to the case of moral realism (cf. Lillehammer 2007: 170). Suppose, for example, that the existence of epistemic normative facts is tied to the fact that belief has a distinct constitutive aim: namely, truth. That is, mental states will not even count as beliefs unless they are regulated by norms of accountability to evidence of truth (Railton 1997) or are ‘constitutively regulated by mechanisms designed to ensure that [they are] true’ (Velleman 2000: 16–17). If so, then the normative fact that R is a reason to believe B, for example, will be an objective matter having to do with evidential relations between R and B: if R shows B to be true, then it is an objective reason for believing B, given what belief is. But an ethical antirealist might grant all this while remaining sceptical that there is any similar analogue of a constitutive aim for action, desire or intention – at least one that would support realism about moral facts (Darwall 2003). Thus, they might seek to drive a wedge between epistemic realism (which is well-founded in terms of the constitutive aim of belief) and moral realism (which may require a different basis that the antirealist is not prepared to grant).
7. Other topics
There is much other work deserving of discussion, but space only to mention two more topics in closing. Korsgaard (1996, 2003, 2009) has raised powerful challenges to all forms of realism, naturalist or non-naturalist, focusing on the ability of realist views to account for the binding normative force of moral requirements on the will: for normative facts are not themselves derived from the operations of the will on realist views, thus setting up a gap between such facts and the will. Realists have responded to these challenges, arguing both that realism has the resources to account for binding moral normativity and that the neo-Kantian alternative, which attempts to derive normativity ultimately from necessary operations of the will (even identifying normativity with practical necessity associated with the functioning of agency as such) is itself deeply problematic (FitzPatrick 2005, Parfit 2006, Enoch 2006).
Finally, ethical realism has recently come under renewed fire from those who find antirealist implications in evolutionary biology. Street (2006) has argued that because the contents of our moral beliefs have been pervasively influenced by our evolutionary history, ethical realism would saddle us with radical ethical scepticism; we would have no reason to expect that our moral beliefs even approximately track the kinds of independent ethical truths the realist posits, since the forces of natural selection that shaped our ethical beliefs would not have been sensitive to such independent ethical facts. Since such radical ethical scepticism is implausible, she argues, we should thus reject ethical realism. Joyce (2006) offers a related but different debunking argument focused on moral concepts, the origins of which he argues are to be explained in terms of natural selection, which again would have rewarded reproductive success rather than the tracking of moral truths. Given this strictly non-moral genealogy of moral concepts, and hence (with the addition of sociological and psychological explanations) of our moral beliefs employing them, the notion of moral truths becomes ‘explanatorily superfluous’ – just as truths about witches are explanatorily superfluous given a complete genealogy of beliefs about witches that dispenses with any appeal to truths about witches. We should thus, he argues, at best remain agnostic about moral truth and give up any claims to moral knowledge, embracing moral skepticism. Realists, unsurprisingly, will have much to say in response to such debunking arguments. For some discussion and responses, see Copp 2008 and FitzPatrick 2008b.