Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. Presumptive Arguments for Moral Realism By Neil Sinclair (University of Nottingham) Please do not quote without permission Disclaimer: This is a copy of my text for a talk I gave at the Oxford moral philosophy seminar on Monday 4th February 2008. The text was intended to be read aloud, and not as a draft of a written paper. I thought it best to make a copy available as soon as possible, so that those who attended the seminar but were unable to ask questions could do so online. For this reason, please excuse the unchecked spelling and punctuation, the lack of references and sometimes patchy explanations. Introduction It is often supposed that realist theories of ethics start with an inherent advantage over their rival theories. The apparent advantage is supposed to derive from the fact that moral judgements exhibit certain features, and those who make moral judgements make certain assumptions, which can only be accommodated on, or at any rate are most concordant with, the realist theory. Thus we are told that only realism straightforwardly preserves ordinary moral talk; that only realism takes moral discourse at its ‘face value’ or that only realism can ‘save the appearances’ of moral discourse. Realists are modest, however, in not viewing this advantage of their theory as tantamount to decisive proof. Instead they feel that it generates a presumption in favour of their theory. It is this presumption that gives realists the courage to take on the various metaphysical and epistemological problems that have plagued their view, and can even be used to nullify some of the implausibility that their theory exhibits in these areas. But though there can be little doubt that this presumptive case for realism has been hugely influential in structuring meta-ethical debate, there is cause to doubt that the presumptive case itself has been submitted to sufficient scrutiny. In this paper I am to address this deficit. My argument will proceed as follows. After outlining the core claims the meta-ethical theories I will be discussing I set out two desiderata for such theories: first, the ability to ‘save the appearances’ of moral discourse and second consistency with our wider philosophical theories. These desiderata will help us distinguish three broad species of presumptive argument: what I label weak, moderate and strong presumptive arguments (the last of which subdivides into two further sub-species). I will then outline two common errors made by those deploying presumptive arguments. The first error is to suppose that some of the distinctive claims of moral realism have infected the forms and assumptions of ordinary moral discourse that stand in need of saving. The second error is to underestimate the resources available to the expressivist in saving the (disinfected) appearances. Since a full exposé of the second error would involve an exhaustive analysis of the expressivists’ ability to save the appearances, my conclusions in this area will be circumspect. Nevertheless I hope to provide cause for optimism regarding the expressivists’ ability to save the appearances. If this optimism is well-founded, it follows that there is no presumptive case for moral realism. 1 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. Realism and opponents First then: what is moral realism? What is a presumptive argument? Moral realists hold that in forming moral opinions and constructing moral theories, our primary goal is to map the contours of a genuinely existing moral reality. Realists are cognitivists, that is, they hold that moral judgements such as “Torture is wrong” express beliefs concerning moral states of affairs. Here a belief is held to be the paradigm example of a cognitive or ‘descriptively representational’ state, that is, a state of mind that represents the world, or some aspect of it, as being thus and so. According to cognitivists, moral judgements express moral beliefs. A belief is a moral belief just in case the representational content of that belief can be captured using a moral sentence, that is, when it represents the world as realising distinctively moral state of affairs. So according to cognitivists moral judgements express moral beliefs. Realists differ from other cognitivists in holding, in addition, that there exists a moral reality the nature of which is sometimes correctly represented by our moral beliefs. Realists in turn differ amongst themselves concerning the nature of that moral reality. [More traditional characterisations of realism often include the claims that moral judgements are truth-apt and that some of them are true. But these truth-based characterisations of realism are secondary to my characterisation, given that we can understand truth-aptness and truth via the notions of belief and (accurate) representation. So, for example we can understand a truth-apt sentence as one that offers a putative description of the way the world might by (say, be expressing a cognitive state), and is therefore true when that description is accurate, false otherwise. Thus realism as I have characterised it leads not only to the claim that moral judgements are truth apt and some of them are true, but to a particular understanding of that claim.] In this paper I will contrast realism with moral expressivism. Expressivists deny that moral judgements express moral beliefs. Instead, they claim, the primary role of moral judgements is to express affective non-cognitive attitudes whose contents and expression play a distinctive role in the mutual co-ordination of attitudes and actions. According to expressivists, the distinctive import of moral judgements arises from such a co-ordinating role, not from expressing states that offer representations of the way the world might (morally) be. According to expresivists, therefore, when I judge that “Torture is wrong” I am expressing a non-cognitive attitude of disapproval towards vivisection, and hoping to persuade others to share that attitude. 2 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. Desiderata for meta-ethics In order to assess whether or not a presumption exists in favour of realism, and against expressivism, it is first necessary to outline the desiderata for such theories. Following common convention, I take there to be two independent desiderata for meta-ethics, the first of which is as follows: D1. A meta-ethical theory should be able to vindicate all of the forms of ordinary moral practice and the deeply embedded assumptions of those who engage in it. To elaborate. Actual moralising takes various forms and agents who engage in moralising make certain assumptions about the nature of the practice they are engaging in. For example, moral claims are standardly expressed in indicative sentences with both a subject and a predicate and most agents who moralise assume that there are better and worse ways of doing so. A meta-ethical theory must be able to vindicate these forms and assumptions. To vindicate a form or assumption is to justify the continued engagement in a practice with that form or which makes that assumption. Any theory that fails to vindicate a form or assumption of moral practice is revisionary. According to D1 there is reason to reject any revisionary theory of moral practice. But as stated D1 is too strict. Notice a curiosity: when considering non-moral types of discourse, it is not generally true that being revisionary provides a reason to reject a meta-theory. For example, the fact that our best meta-theory of witch discourse recommends abandoning that discourse is not a reason against accepting that account. The discrepancy between this and the case of moralising must lie in what may be called the pragmatic assumption: the assumption that moral discourse, unlike, say witch-discourse, is, at least in the most part, a useful practice that we have good reason to go on engaging in. With this assumption in hand, being revisionary is a reason to reject a meta-theory of moral practice. But we mustn’t overstate the content of the pragmatic assumption. The pragmatic assumption is supported by, among other things, the observation that practically all known societies have developed something akin to moral practice. But the support for the pragmatic assumption certainly does not justify the claim that all forms and assumptions of moralising as it is practiced today are pragmatically justifiable. Indeed there are reasons to think that this is not the case: first because it is unlikely that our actual practice manifests a single teleologically unified practice, second because false meta-theory may have itself infested the practice, so that some more reflective moralisers engage in the practice with a (false) understanding of their activities built in. For this reason, it is likely that any meta-ethical theory will be revisionary of some of the forms and assumptions of actual moral practice, making D1 unreasonably strict. With what can it be replaced? I suggest the following: 3 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. D1’: A meta-ethical theory should be able to vindicate all of the pragmatically important forms of moral practice and the pragmatically important assumptions of those who engage in it. So any actual form or assumption of moral practice that is revised carries an explanatory cost, but this cost can be met so long as the resulting meta-theory can show how the important things that we want to say and do using moral discourse, and all of those substantial debates moral discourse involves, are preserved in the recommended revised practice. For example, it will be no good recommending that we replace moralising with a system of communal chanting, if by doing so we lose an important aspect of the way we interact with each other and the world. What the pragmatically important aspects of moral practice are will, of course, be a matter of meta-ethical dispute. Thus in order to be plausible a meta-ethical theory must first defend an account of the way moralising helps us relate in worthwhile ways to the world and to each other and then show how, given this account, some (if not all) of the features of actual moral practice are to be expected. In any case, most meta-ethicists agree on a core set of pragmatically important forms and assumptions of moral practice. They include: (1) Actions can be right, wrong, permissible, supererogatory. States of affairs can be good, bad, morally neutral. Characters can be virtuous or vicious. (2) Indicative moral sentences are capable of intelligible embedding in various logical constructions such as negations, conditions, etc. (3) The rightness and wrongness of actions does not, in general, depend on our thoughts about those actions. (4) Some moral judgements are true. (5) Moral claims can feature as explanations of (non-moral) events. (6) Genuine moral agreement and disagreement is possible (7) Moral discussion is sometimes a fruitful way of resolving such disagreement (8) Moral claims can be supported by reasons According to D1’ then meta-ethical theories are to be judged in their ability to vindicate such forms and assumptions. A note on labelling Now it has, of course, long been recognised that realism can quite easily vindicate these forms and assumptions of moral practice. A more recent trend in metaethics holds that expressivists too might be able to vindicate these forms and assumptions. Nor is there any immediate reason to rule this possibility out. Expressivism simply holds that moral judgements do not express moral beliefs but do express affective non-cognitive attitudes. This is at least prima facie compatible with the possibility that moral sentences can intelligibly embedded, and moral judgements can true, known, explanatory and the rest. As we all know, quasirealism is the view that not only is expressivism the correct view of moral practice, but that moral practice so understood legitimately possesses these forms and features. 4 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. Now it is worth heading off some possible misunderstandings of quasirealism. The above forms and assumptions of moral discourse are sometimes called the ‘appearances’ of moral discourse quasi-realism as the task of ‘saving the appearances’ for expressivism. These characterisations, though not strictly false, are seriously misleading and best avoided. There are two important reasons for this. First the description of the forms and assumptions as appearances, implying as it does that a practice might appear to possess them, but not really do so, is inappropriate in several cases. For example, having sentences of subject-predicate form is a purely syntactic matter – nothing can appear to have subject-predicate form and yet not really do so. Second, the phrase ‘saving the appearances’ is crucially ambiguous. There are at least two ways in which a theory might ‘save the appearances’. First, it may preserve the appearances as mere appearances. To save an appearance in this sense is to justify our right to continue engaging in the practice as if is possessed the feature in questions, when in fact it doesn’t. For example, to save the appearance of truth-aptness for moral claims in this sense would be to justify our behaving as if moral sentences are the sorts of things that can be true or false, when in fact they aren’t. There is, however, a second sense in which one can ‘save the appearances’ and that is by vindicating them. To vindicate the appearances is to explain them as a perfectly natural and expected result of the actual nature of the practice. For example, to vindicate the appearance of truth-aptness for moral discourse would be to show that moral discourse is truth-apt (and that is why it appears to be). To save the appearances in this sense is to justify the forms and assumptions of the discourse as legitimate. Quasi-realism intends to save the appearances of moral discourse in the second, vindicating sense. That is, it hopes not to justify our right to behave as if moral discourse is truth-apt, knowledge-apt and the rest, but moral discourse’s right to be truth-apt, knowledge-apt and the rest. This is why a necessary part of the programme is to ‘domesticate’ the notions of truth, knowledge and so on, that is, show that they can be legitimate upshots of more than just cognitive practices. This crucial ambiguity in the phrase ‘saving the appearances’ is all the reason we need to abandon this terminology. Instead I recommend the following: call the forms and assumptions of moral practice listed above the ‘propositional clothing’ of moral practice. The project to which both realism and expressivism are addressed is that of weaving the propositional clothing for moral discourse, that is, showing how moral discourse can legitimately come possess all of these forms and assumptions. Quasi-realism is best described as the project of weaving, from an expressivist thread, the propositional clothing of moral practice. And, if the project succeeds, the expressivist will not have weaved the emperors new (propositional) clothing, rather it will be the same type of clothing worn by other discourses, only weaved from non-cognitive thread. Of course, the project may not succeed, but it is important to properly understand its aims. The second desiderata for meta-ethical theories is as follows D2. A meta-ethical theory should conform to one’s wider philosophical views about the topics it makes claims about. Since a meta-ethical theory makes claims about the semantics, psychology, metaphysics and epistemology of moral practice, it should not conflict with our best 5 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. general theories of semantics, psychology, metaphysics and epistemology. A metaethical theory that postulates a unique and mysterious type of meaning possessed by all and only moral terms, for example, would fail to meet this desideratum. Call this the desideratum of ‘placement’. Presumptive arguments By a presumptive argument I mean one that establishes a presumption in favour of a theories’ truth. The conclusion of a presumptive argument is that, for all that has so far been said, the theory in question is more likely to be true than any of its rivals. A presumption in favour a theory can be overturned (as when a stronger reason is raised against the theory, leaving the original presumptive reason intact but outweighed) or undermined (as when what was initially thought to be a reason in favour of the theory turns out to be not reason at all). Presumptive arguments for moral realism can be divided into three species: weak, moderate and strong. Weak presumptive arguments accept that other theories can vindicate the requisite forms and assumptions of moral practice but argues that the realist vindication is more the moral natural, simpler or otherwise more appealing vindication. Propositional clothing weaved from a realist thread is smoother to the touch. It follows that all realists have to do to present a conclusive case for their position is show that they can also satisfy the desideratum of placement According to moderate presumptive arguments we cannot rule out the possibility that the realists’ opponents may be able to do the requisite weaving, but it remains the case that such weaving hasn’t been done yet. These arguments provide a case for the ongoing acceptance of realism as our best working hypothesis given that it has been, to date, the only theory that can be shown to meet one of the two desiderata of meta-ethics. According to strong presumptive arguments, only moral realism can weave the propositional clothing for moral discourse. That is, only realism can satisfy the first desideratum for meta-ethical theories. This argument is presumptive because there is a second desideratum by reference to which realists’ opponents may draw an advantage that adequately compensates for their poor weaving skills. If a strong presumptive argument can be sustained, it follows that all that is required to present a conclusive argument for realism is to show that it can also satisfy the desideratum of placement. For having done so realists would have shown that they, and only they, can satisfy both of the desiderata for meta-ethical theories. We can then close the meta-ethical casebook, without having to deal with the worrisome details of the other theories. Strong presumptive arguments in turn can be divided into two subspecies. Type-1 strong presumptive arguments (or type-1 arguments) hold that the forms and assumptions of moral practice that require vindication include some of the metaethical claims distinctive of realism. For example, realists claim that moral judgements express descriptive states of mind whose descriptive content is captured by moral sentences. If this is one of the pragmatically important assumptions of moral 6 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. practice, then it follows that only realism will be able to weave all of the propositional clothing, thus generating a strong presumptive argument. According to type-2 strong presumptive arguments (or type-2 arguments) the forms and assumptions that require vindication do not involve any of the metaethical claims distinctive of realism, nevertheless, these arguments run, such forms and assumptions as there are can only be vindicated by moral realism. According to type-1 arguments, moral realism itself is on the surface of moral practice; according to type-2 presumptive arguments the surface is populated by features that admit only of realist vindication. Note that all of the presumptive arguments noted here are types. Individual tokens can be differentiated by the particular form or assumption of moral practice that they focus on. Putting forward one type of presumptive argument with regard to one of the items of propositional clothing does not commit the realist to presenting the same type of argument for the whole wardrobe. Real presumptive arguments A study of the literature on presumptive arguments provides credence to the view that most actual presentations of the presumptive case for realism are ambiguous. (See Appendix I). This is unfortunate, not only because clarity is a virtue that should be prized, but because conflating types of presumptive argument can snatch unearned argumentative capital. Here is a common example. It is a case of argumentative capital being earned by conflating type-1 and type-2 presumptive arguments. Putative presumptive argument According to expressivists there are no moral facts. But surely one of the most pragmatically important assumptions of moral practice is that there are moral facts. For example, it is a fact that Hitler acted wrongly in ordering the exterminations of the Jews. Therefore, expressivists cannot vindicate this assumption of moral practice. This argument trades on an ambiguity. There are two senses that might be ascribed to the claim that ‘there are moral facts’. In the first, meta-ethically loaded sense, the claim is that there are moral states of affairs such as the realist postulates. In the second, meta-ethically neutral sense, the claim is that it is sometimes appropriate to sincerely utter moral sentences (such as the sentence that ‘Hitler acted wrongly in ordering the extermination of the Jews’). Reading the disputed phrase in the first way, we get a type-1 argument. Reading it in the second way generates a type-2 argument. But we get a bad argument – as we have here – if we read ‘moral facts’ in the meta-ethically neutral way when describing the commitments of moral practice and then in the meta-ethically loaded way when assessing the capability of expressivism to vindicate those commitments. Such an argument illicitly trades on the ambiguity in the phrase ‘moral facts’. 7 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. Assessing presumptive arguments But aside from ambiguity in presentation, are presumptive arguments good arguments? - Weak arguments, IN BRIEF: what is natural/simple is not always what is correct; simplicity is hard to discern. Moderate arguments, IN BRIEF: at best establish the right to treat realism as our best working hypothesis; do not justify refusing to consider the details of the opponents’ position. There are two principle mistakes that defenders of presumptive arguments make. First Error The first error is to assume that the forms and assumptions of moral practice that require vindication involve some of the meta-ethical claims distinctive of realism. This is to assume that the pragmatically important forms and assumptions of moral practice and meta-ethically loaded, and in the realists favour. It is to assume, more particularly, the pragmatically indispensible parts of ordinary moralising commit one to the view that moral judgements express descriptively representational states of mind that represent the world as realising distinctively moral states of affairs. I think it is unlikely that the pragmatically important forms and assumptions of moral practice include the meta-ethical claims of moral realism. This is for two reasons. First, because most everyday moralisers seldom show appreciation of metaethical issues and categories, let alone appreciation of a particular meta-ethical theory. The categories that define meta-ethical theories (descriptive, representational and so on) are to a large extent philosopher’s constructs rather than tacit categories of everyday moralisers. Of course, everyday moralises hold that some things are right, others wrong, that moral judgements can be supported by reasons, that they are sometimes true than that their truth depends on the way the world is and so on. But these are not categories that define realism. What defines realism is that moral judgements have a characteristic linguistic function, express characteristically descriptive (as opposed to affective) states of mind and therefore that their truth consists in correspondence between the descriptively representational content of such states and the moral way of the world. These are sophisticated philosophical claims, involving, for example, an appreciation of the distinctions between descriptive and non-descriptive linguistic functions, between descriptive and non-descriptive psychological roles and between truth understood as correspondence and truth understood in other terms. It is one thing to claim that everyday moralisers assume moral judgements to be true, quite another to claim that they assume moral truth to be understood as the realist sees it. Ask the fabled man on the Clapham omnibus whether torture is wrong and he may well reply that it is. But ask whether his judgement that torture is wrong is descriptive of moral reality, or made true by correspondence between the descriptive 8 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. content of the mental state his judgement expresses and the distinctively moral state of the world and one is more likely to be faced with an uncomprehending silence. Of course, he may take it that to say that “The judgement that torture is wrong describes moral reality” is to say no more than the original “torture is wrong”. But in that case this is not to offer any meta-ethical claim; it is rather just to repeat the original assertion is a different (perhaps more emphatic) idiom. The second reason to doubt the claim that the pragmatically important forms and assumptions of moral practice include the meta-ethical claims of moral realism is more fundamental. For regardless of whether most common moralisers engage in moral practice with a tacit assumption of moral realism, there are certainly some people that do. Perhaps the largest part of this constituency are those who believe that moral truths are constituted by God’s will, although philosopher’s who espouse moral realism are another important part. If meta-ethics is to provide an account of these moralisers, then realism looks to be the presumptively preferred option. But that would be too swift. Remember, meta-ethical theories are only required to vindicate those forms and assumptions of moral practice that are pragmatically important for that practice. And the assumption of the truth of moral realism, in some moralisers, is not pragmatically important, as is demonstrated by the fact that such moraliser can engage in moral dispute with those who make no assumption of moral realism. Thus even if moral realism were assumed by all who engaged in moral practice, it would not make it a pragmatically important part of that practice. To further illustrate the mistake I will briefly discuss two examples. The first (more obvious) case comes from Shafer-Landau: “We use indicative mood when issuing moral judgements. We assert that practices, character traits, or states are vicious, morally attractive or deserving…Moral talk is shot through with description, attribution and predication. This makes perfect sense if cognitivism is true.” (p.24) Here Shafer-Landau groups together three notions: description, attribution and predication. The latter notion is purely syntactic and meta-theoretically neutral. But if the other two notions are to add to the argument then we must assume that to say that moral talk is descriptive or attributive is to say more than that it involves applying predicates to subjects. What more? Well to say that moral talk is descriptive could be to say that moral talk involves the expression of mental states that descriptively represent the world in moral ways. And to say that moral talk is attributive could be to say that moral talk involves attributing moral properties (such as the realist believes in) to actions, states of affairs and characters. Both these claims are part of the realist meta-theory. 9 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. However, when we read these terms in this way then it becomes controversial whether description and attribution really are some of the pragmatically important forms of moral practice. It is surely undeniable that moral talk involves moral predication, and that such a practice is pragmatically important. But to assume that it also involves commitment to a certain meta-theoretical understanding of that predication is much more controversial. Better to say that the uncontroversial phenomena here is the meta-theoretically neutral one of moral predication, with it then being a further issue how that predication is to be understood. Realism and expressivism give two competing explanations, but neither is assumed by the practice. [A second, more subtle manifestation of the first error can be found in the following passage in McNaughton. Criticising Blackburn’s quasi-realist construction of moral truth, McNaughton’s expounds Blackburn’s position as explaining how, from within the practice of moralising, it makes sense to suppose that there is a singularly best set of attitudes, and thus from within that practice we can make sense of the notion of a correct moral opinion as an opinion expressed an attitude that belongs within that set. But to this account McNaughton responds: “It is difficult to resist the conclusion that, although we talk and think within our moral practice as if there were correct answers to moral questions, Blackburn’s theory can make no sense, from the external standpoint, of the notion of a correct moral answer.” (p.188) It may be true that Blackburn’s theory cannot provide an account of a correct moral answer from an external standpoint - that is from a point of view disengaged from “moral action, argument and thought” (p.187) – but why think that moral practice is committed not just to the existence of correct moral answers, but to the claim that their correctness can be understood from an external standpoint? Presumably correctness from an external standpoint would be realist correctness – that is correspondence between the descriptively representational content of the moral belief and the moral state of the world (one can see this correctness even if one is not involved in the moral cut and thrust). Moral practice is surely committed to the idea that some moral judgements are correct, others incorrect. And, if Blackburn’s account success (which McNaughton doesn’t dispute) we can explain why those engaged in moral practice should remain committed to this idea. But it is a further claim entirely (and an unsubstantiated one) that one of the pragmatically important features of moral practice is that this correctness is understandable from a perspective outside of engagement in moral practice, and, therefore, in realistic terms. Thus McNaughton is here unjustifiably reading realist meta-theory into the forms and assumptions of moral practice.] If it is indeed an error to read into the forms and assumptions of moral discourse some of the distinctive claims of realism it may help to diagnose the possible sources of that error. One has already been mentioned: namely the ambiguity in terms such as ‘moral facts’ between a meta-ethically neutral interpretation and a meta-ethically loaded one. A second motive may be the thought, easily acquired if one is introduced to expressivism through the work of Ayer, that expressivism is necessarily a revisionary theory. If so, then whatever the efforts of recent expressivists to weave the 10 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. propositional clothing, there must remain some element of moral practice that they cannot weave. Second error Of course, it may be that though the realism is not a tacit assumption of ordinary moralisers it remains true that those moralisers make certain moves within moral practice that can only be made sense of if moral realism were true. This is to adopt a type-2 presumptive argument. But though such arguments necessarily avoid the first error, they are unfortunately prone to a second. This is the error of underestimating the resources available to expressivists with which they can weave the propositional clothing. The core of this error derives, I think, from systematic downplaying of certain parts of the expressivist theory. It is common, for example, to understand expressivism as committed to simply two claims. The negative claim that moral judgements do not express moral beliefs and the positive claim that moral judgements express affective attitudes. This is a bit like describing 9/11 as the delayed arrival of American Airlines flight 11. What such an account misses out is the purpose or function of expressing those attitudes. In fact, all expressivists (with the possible exception of the 1936 A.J. Ayer) hold that we can only properly understand the meaning of moral judgements when we have understood not only the state of mind they express, but also the function of that expression. For expressivists, moral judgements are not mere ‘sounding off’ - they do not simply display one’s attitudes as one’s choice of football shirt displays one’s sporting allegiance. Rather, when we express our moral attitudes we do so with the intention of getting others to share them. To quote Stevenson, moral judgements are ‘instruments used in the complicated interplay and re-adjustment of human interests’ (Stevenson 1944 p.13). It is only once we have understood this persuasive role, the expressivist claims, that we have understood the meaning of moral terms. The placing of moral judgements as instruments in the practice of coordination attitude provides many of the materials with which expressivists can weave propositional clothing. For example, two of the items of propositional clothing were that moral disagreements are possible, and that moral discussion is sometimes a fruitful way of resolving that disagreement. Moral realists weave these articles thus: moral disagreement occurs when one party offers a description of the state of a certain part of the world; another party offers a different description; both descriptions cannot be accurate; so neither can accept the other’s commitment. In so far as both have an interest in providing correct descriptions of the world, both have a motive for changing the opinion of the other. For expressivists, moral disagreement can seem more ephemeral. What can moralisers be disagreeing about, if it isn’t the (moral) state of the world? The answer, of course, is that moral disagreement is practical disagreement. A moral disagreement, according to expressivists, occurs when agents ‘have opposed attitudes to the same object – one approving of it, for instance, and the other disapproving of it – and when 11 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. at least one of them has a motive for altering or calling into question the attitude of the other’.1 So whereas disagreement in scientific matters is born from a concern to change each other’s beliefs, disagreement in morals is borne from a concern to change each other’s attitudes. So long as there is a motive to insist on change there is a disagreement. [How about the role of moral discussion, the practice of giving reasons in favour of moral judgements, in resolving disagreement? Given the above account of moral disagreement as resting on a motive for altering attitudes, reasons in support of a moral claim will be reasons in support of taking the particular attitude towards the particular object of evaluation that that claim recommends. Likewise, reasons against a particular moral claim will be reasons against taking the particular attitude towards the particular object of evaluation that that claim recommends. A reason in support of or against an attitude typically involves citing a putative feature of the object to which that attitude is a response. Therefore, in so far as our attitudes can be altered by changes in our beliefs about the nature of the object being evaluated, discussions of that nature can have a role in moral argument. See Appendix II]] There is nothing new about these expressivist weavings. But they are instructive in so far as they show what the epxressivist can achieve once he emphasises the co-ordinating, practical role of moral discourse. A more recent example of realists underestimating the expressivists’ resources concerns the apparent mind-independence of morals. One of the items of propositional clothing is the assumption that the rightness and wrongness of actions does not depend on our thoughts about those actions, in particular it does not depend on our thoughts about their rightness or wrongness. Can expressivists weave this item? I can hear provide an outline of how they can. First, we must avoid the first error of reading some of the claims of moral realism into this feature of moral practice. It is tempting to understand the commitment to mind-independence embedded in ordinary moral though in an ontological way that assumes realism. For example, it is tempting to understand the claim that “the wrongness of torture in no way depends on us thinking it wrong” as claiming that the possessing by instances of torture of the property of wrongness in no way depends on our judgements concerning such a property. But expressivists can reject this ontological reading. Instead, claims of moral mind-independence can be understood in a way that assumes neither expressivism nor realism. For example, the above claim about the wrongness of torture can be understood as maintaining simply that the correct application of the predicate ‘wrong’ to the action of torture in no way depends on our propensity to so apply it. It is then a further question how this predicate and the conditions governing its application are to be understood, with realism and expressivism providing distinct alternatives. Generalising, expressivists can urge that we understand the embedded claims of moral mind-independence as asserting not that the distribution of moral properties is independent of us but that the correct application of moral predicates is independent 1 Stevenson 1944 p.3 12 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. of us. Whether those predicates are property-ascribing (as realists hold) or attitude expressing (as expressivists hold) is then a further issue. Having taken the realist sting out of the embedded claims of mindindependence the expressivist is in a position to weave them from expressivist threads. For expressivists, moral predicates are used to express attitude, so to hold that the correct application of moral predicates is independent of us to accept that one’s moral attitudes should not depend on facts about oneself. This is to endorse a certain moral sensibilitiy, that is, a particular way of forming and regulating one’s moral attitudes in response to inputs in belief and perception. According to expressivists, an important part of the co-ordinating role of moral practice is served by evaluating moral sensibilities, in so far as people’s moral sensibilities affect their behaviours. It is such an attitudinal evaluation that is expressed in the commitment of mind-independence. It is therefore false to claim that, according to expressivists, moral facts such as the wrongness of torture are ultimately dependent on our attitudes. But why should moral agents endorse just these way of forming moral attitudes and not others? The answer takes us back to the co-ordinating role of moral practice. Suppose, for example, there was an agent, call him Brian, who accepted minddependence for morals. On the expressivist reading, such an agent endorses moral sensibilites on which our moral attitudes depend on facts about the person whose attitudes they are. Suppose this agent forms moral attitudes and, in the spirit of coordination, intends others to share those attitudes. For example, Brian morally disapproves of torture and expresses this by claiming that torture is wrong. Someone else, who understands that Brian’s use of the term ‘wrong’ expresses disapproval asks Brian: ‘Why is torture wrong?’. What might Brian say? Since he thinks it appropriate to disapprove of something solely on the basis of facts about himself, Brian can only cite such facts in hoping to persuade others to share his disapproval. But such facts are unlikely to be persuasive. For example, Brian might say that torture is wrong because it disgusts him, or even that torture is wrong because he thinks it is. Neither is likely to persuade others to share Brian’s attitude, since others are looking for a reason to change their evaluation of torture, and there is no reason why they should take Brian’s reactions to be indicative of anything significant about torture itself. More generally, for agents like Brian, reasons that can be cited in favour of one’s moral evaluations are not reasons that are likely to persuade others to share those evaluations 13 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. Compare Ben. Ben accepts mind-independence for morals. On the expressivist reading, this is to endorse moral sensibilities on which our moral attitudes towards, say, torture, depend on facts about torture itself, not on facts about us. Suppose, for example Ben endorses sensibilities that form moral attitudes on the basis of beliefs about the pain-causing properties of types of action. When someone asks Ben: ‘Why is torture wrong?’ he can reply: because of the pain caused. This is the reason that Ben thinks that torture is wrong, but also a reason that can potentially persuade others to share the disapproval that that judgement expresses. Thus, for agents like Ben, reasons that can be citied in favour of one’s moral evaluations are also reasons that can be used to persuade others to share those evaluations. I think this account of the commitment to the mind-independence of morals is structurally identical with a plausible account of our commitment to the mindindependence of other facts, such as facts regarding circularity. Whether or not an object is circular doesn’t depend on facts about us. In saying this, I am endorsing those sensibilities according to which our beliefs (about circularity do not depend on facts about us. Why endorse such sensibilities? Because of the function of the mental states involved. Beliefs aim at correct representation of the world, and a belief about circularity formed on the basis of facts about us is unlikely to be correctly representational. Thus the function of beliefs justifies endorse certain sensibilities guiding their formation. In just the same way, I have argued, the co-ordinating function of moral attitudes justifies the endorsement of certain sensibilities guiding their formation. Though beliefs and moral attitudes have different functions, both these functions are aided when we endorse sensibilities which reject the idea that the commitment is correctly held on the basis of facts about us. In so far as the claims of mind-independence have us expressing these endorsements, the claims of minddependence in the two cases are dealt with in fundamentally the same way. What differs is the reason why we make these endorsements and that in turn depends on the differing functions of the commitments in question. The key insight of the foregoing account is that it is not only the descriptive function possessed by beliefs that necessitates endorsing the sort of sensibilities that make the correct application of a predicate dependent on something other than facts about us. 14 Neil Sinclair: “Presumptive Arguments for Moral Realism” Draft of talk for Oxford Moral Philosophy Seminar, 04/-2/08 Please do not quote without permission. We can see, therefore, that emphasising the co-ordinating function of moral practice provides the expressivist with powerful resources to vindicate the forms and assumptions of moral practice. This is not of course to argue that all of the forms and assumptions of moral practice can be vindicated. But it does provide some scope for optimism. In particular, the above highlights a powerful expressivist pattern that can be used for weaving the propositional clothing. The first part of this pattern corresponds to the first error of presumptive arguments. At this point, the expressivist emphasises that there is a meta-ethically neutral account of the forms and assumptions of moral practice. For example I argued that the claims of moral mind-independence can be given a meta-ethically neutral reading if they are understood as claims concerning the application conditions governing moral predicates. This ‘domestication’ of claims of mind-independence is structurally isomorphic with popular minimalist treatments of truth, which discern a meta-theoretically neutral role for a truth-predicate. [Note this is not to be committed to the claim that the only notion of truth available is the minimalist one, only to the claim that the notion of truth employed in moral practice is compatible with a minimalist interpretation.] The second part of the expressivist pattern is corresponds to the second error made by presumptive arguments. This part emphasises that the expressive practice which is at the centre of ethics is not merely a communal show-and-tell; the purpose of the expression of attitudes is a distinctive practical one. This practical role is what necessitates the adoption of certain standards governing the application of moral terms understood expressively and also what allowed Stevenson to claim that the practice of giving reasons is as admissible in ethics as it is in descriptive disciplines. In so far as the first part of this pattern is available, what needs accommodating becomes less daunting. In so far as the second part is available, the resources available to the expressivist in doing so are increased. Conclusion When it comes to moral practice, meta-ethics should leave everything as it is. At least, any theory that forces us to jettison some of the pragmatically important elements of moral practices is by that token implausible. But to assess whether realism is less revisionary than, say, expressivism, we must be careful to do two things. First, we must delineate the proper content of the pragmatically important elements of the practice. Second, we must appreciate the full extent of the resources available to the alternative views. Few actual presentations of presumptive arguments do both. This is not to say, of course, that there is no successful presumptive argument for moral realism. But it should at least make us more cautious when using such arguments to structure the meta-ethical dialectic. 15