Ethic Theory Moral Prac (2013) 16:923–937 DOI 10.1007/s10677-013-9423-6
Moral Error Theory, Entailment and Presupposition Wouter Floris Kalf
Accepted: 14 March 2013 / Published online: 2 April 2013 # Springer Science+Business Media Dordrecht 2013
Abstract According to moral error theory, moral discourse is error-ridden. Establishing error theory requires establishing two claims. These are that moral discourse carries a nonnegotiable commitment to there being a moral reality and that there is no such reality. This paper concerns the first and so-called non-negotiable commitment claim. It starts by identifying the two existing argumentative strategies for settling that claim. The standard strategy is to argue for a relation of conceptual entailment between the moral statements that comprise moral discourse and the statement that there is a moral reality. The non-standard strategy is to argue for a presupposition relation instead. Error theorists have so far failed to consider a third strategy, which uses a general entailment relation that doesn’t require intricate relations between concepts. The paper argues that both entailment claims struggle to meet a new explanatory challenge and that since the presupposition option doesn’t we have prima facie reason to prefer it over the entailment options. The paper then argues that suitably amending the entailment claims enables them to meet this challenge. With all three options back on the table the paper closes by arguing that error theorists should consider developing the currently unrecognised, non-conceptual entailment claim. Keywords Moral error theory . Non-negotiable commitment claim . Entailment . Presupposition
1 Introduction The core commitment of moral error theory is that moral discourse is in error (Mackie 1977; Joyce 2001). Moral discourse consists of various moral statements, including ‘stealing is morally wrong’ and ‘keeping promises is morally obligatory’.1 Such statements attempt to In this paper, ‘moral statement’ refers to a moral judgment, understood as a belief with representational (propositional) content, which is communicated (uttered) by being spoken or inscribed (Mackie 1977: 9). This usage of the terminology entails the falsity of non-cognitivism, according to which moral judgments are not beliefs. As I only inquire into standard, cognitivist versions of error theory, this is unproblematic. Also, henceforth ‘error theory’ denotes ‘moral error theory’ as opposed to other error theories, for instance about mathematical discourse. 1
W. F. Kalf (*) School of Philosophy, Religion and History of Science, Woodhouse Lane, University of Leeds, Leeds LS2 9JT, UK e-mail:
[email protected]
924
W.F. Kalf
describe moral reality and aim to gain a positive truth-value by at least sometimes doing that correctly. Unfortunately, there is no moral reality and so no moral statement can be true. This is only a rough characterization of error theory. One of many issues is that error theorists have to specify what they mean by ‘moral reality’. They need to understand it in a fairly robust sense because a non-robust understanding, such as one according to which moral reality consists of purely natural properties, probably won’t yield an error theory. For skepticism about purely natural properties is probably unwarranted. The following generic and maximally non-committal formulation of error theory allows error theorists to make their claims more precise. It does so by indicating the four choice points at which they have to undertake theoretical commitments: Moral discourse is non-negotiably committed to the Moral Reality Thesis (MRT) Substantive Claim MRT is flawed Conclusion Moral discourse is flawed2
Non-Negotiable Commitment Claim
The first choice point is ‘moral discourse’, by which error theorists may mean ordinary folk moral discourse, philosophers’ discourse about morality (Mackie 1977: 35), or any mixture of these (Blackburn 2005: 327).3 The second choice point is ‘is non-negotiably committed’, which concerns the nature of moral discourse’s mode of commitment to MRT. This commitment has to be non-negotiable in the following sense: denying the commitment amounts to changing the subject, such as when we deny that discourse about witches can remain discourse about witches if we cease to associate witches with supernatural powers (Joyce 2001: 5). Since ordinary moral discourse is comprised of moral statements of various kinds we can use the familiar options from the philosophy of language for understanding how such linguistic utterances can carry semantic and pragmatic commitments, which are, in the relevant sense and to varying degrees, non-negotiable. Hence the options include, but are not limited to, various entailment and presupposition relations (Joyce 2001). The third choice point concerns the content of MRT, which captures the (or a) core commitment of moral discourse.4 Most error theorists are impressed by morality’s inescapable authority—its seeming ability to require us to perform certain actions even if we don’t want to perform them (Joyce 2001: 104). Mackie argued that the best way to make sense of this feature of morality is to postulate mind-independent objective moral properties with ‘prescriptive force’ that our moral judgments are responsive to (1977: 34). The properties had to be mindindependent and objective to make sense of morality’s inescapable demands—had moral properties been dependent on our attitudes and thus subjective in that sense then we would have been able to escape our moral obligations by altering our attitudes. And moral properties
2
Actually a third premise is required to make the argument formally valid; namely, that when moral discourse is committed to a flawed thesis then this renders moral discourse flawed (Shah and Evans 2012: 86n.13). Usually however this premise is left suppressed. I follow suit. 3 A completely different option is to have an error theory not of moral discourse but of morality itself, understood for instance as a collection of ought facts and moral properties (Clark 2009: 204). Such an error theory gets little airtime. I set it aside in this paper as well. 4 I say ‘the or a’ commitment to leave room for error theories that work on the basis of many small problems that together, albeit not individually, pose “A Big Problem” (Joyce 2008: 52). In the remainder of this paper however I assume for simplicity that the sole commitment that error theorists identify is a ‘Big Problem’ all by itself. In particular, I assume that an error theory of deontic concepts like RIGHT and WRONG can sink moral discourse even though moral discourse also deals in evaluative concepts like GOOD and BAD and thick moral concepts like RUDE (I will use SMALL CAPS to refer to concepts).
Moral Error Theory, Entailment and Presupposition
925
had to be ‘prescriptive’ to make sense of the inescapable demands. Mackie probably did not always have the distinction between normative and motivational force clearly in sight, and the various formulations of morality’s objective prescriptivity that he offers oscillate between a purely normative, a purely motivational and a combined normative and motivational interpretation of prescriptivity. Joyce (2001) argued that the strongest version of error theory interprets ‘prescriptivity’ along purely normative lines and formulated morality’s flawed thesis in terms of categorical moral reasons. These are reasons that exert normative force over agents irrespective of whether acting on them furthers their desires (this is what makes them categorical moral reasons) and that have recognisable moral content, such as that stealing is wrong (this is what makes them categorical moral reasons).5 In addition to zooming in on morality’s prescriptivity, there is a whole raft of alternative options, including, but not limited to, taking MRT to be about a metaphysically robust sort of free will (Joyce 2008: 52). The final choice point is ‘flawed’, which occurs twice in the template. Ways for MRT and moral statements to be flawed include being false, untrue, neither true nor false, or unjustified, and again these don’t exhaust the options (Kirchin 2010). The standard formulation of error theory undertakes the following commitments: the object of error theory is ordinary folk moral discourse, the mode of commitment of moral discourse to MRT is one of conceptual entailment, the content of MRT is that there exist categorical moral reasons, and moral statements are false (Olson 2011: 62).6 This paper concerns the Non-Negotiable Commitment Claim, which advocates of the standard formulation of error theory aim to settle with the Conceptual Entailment Claim
Moral discourse’s mode of commitment to MRT is one of conceptual entailment
According to the Conceptual Entailment Claim (CEC), moral statements entail statements about categorical moral reasons because it is necessary to the applicability of moral concepts, as expressed by moral statements, that there exist categorical moral reasons (Joyce 2008: 65). To see how this works, first consider a non-moral example. The sentence ‘this is a vixen’ entails ‘this is a fox’ in virtue of the concept VIXEN expressed by the term ‘vixen’ in the sentence ‘this is a vixen’. For the applicability conditions of VIXEN are twofold: in order to use VIXEN correctly, you have to apply it to something that is both female and a fox. Take either femaleness or foxhood out of VIXEN and the schvixens you end up referring to simply don’t deserve the label ‘vixen’. Likewise, ‘stealing is morally wrong’ entails ‘there is a categorical moral reason not to steal’ in virtue of the concept MORALLY 7 WRONG expressed by the term ‘morally wrong’ in ‘stealing is morally wrong’. For the applicability conditions of MORALLY WRONG are two-fold: in order to use MORALLY WRONG correctly you have to apply it to something there is both categorical reason not to do and that is of recognisable moral import. Take categorical reason-giving force out of moral concepts, and the schmoral concepts you end up with are too “wimpified” (Joyce MS: 14) to be mistaken for the real thing:
5
This does not imply that inescapable authority would have to supply overriding reason to act morally. Indeed it is so standard that, to take just one example, Smith (1994: 65) and others following him (e.g., Robertson 2008: 108) call what I have dubbed the Non-Negotiable Commitment Claim ‘the Conceptual Claim’. 7 I take ‘stealing is morally wrong’ to be equivalent to ‘stealing morally ought not to be done’, which in my experience to some ears more clearly licences the claim that there is a categorical reason not to steal. 6
926
W.F. Kalf
“[A]ny value system with the flawed element(s) extirpated simply would not deserve the name “morality”” (Joyce 2008: 52)8 Joyce argues that if the standard formulation of error theory fails when its conceptual entailment claim is false then error theorists must revert to their back-up plan, which is to settle the non-negotiable commitment claim with the aid of presuppositions (Joyce 2001: 6–7): Presupposition Claim
Moral discourse’s mode of commitment to MRT is one of presupposition
Philosophers have not always clearly distinguished the Presupposition Claim (PC) from MEC. For instance, Finlay sometimes calls the Non-Negotiable Commitment Claim “Presupposition” (2008: 374), yet from reading his work it becomes clear that what he has in mind is CEC instead. This mix-up of terminology is unfortunate, for the commitments of CEC and PC are very different. One statement presupposes another if denying the latter renders the kind of talk one is engaged in infelicitous or worse (provided that the statement is understood as an utterance) and for this to happen intricate connections between the concepts expressed by these statements are not required. A standard example is that denying that France presently has a king renders a conversation about whether or not he’s bald infelicitous or worse, and for this to happen we don’t need to suppose essential connections between KING, FRANCE, BALDNESS, or what have you. Likewise, on this view, moral statements presuppose that there are categorical reasons as denying this renders moral talk infelicitous or worse even when it is not built into moral concepts that there must be categorical reasons. In addition to PC and CEC there is a third option for settling the Non-Negotiable Commitment Claim. This option is currently not recognised in the literature. CEC takes up unnecessarily strong commitments about what it is in virtue of which the entailment relation between moral statements and MRT holds, if indeed it does. These strong commitments are about connections between concepts. But it is possible to weaken these commitments without compromising on the stringency of the entailment relation, and thereby without compromising on its non-negotiability. Consider, for instance, entailment relations that hold in virtue of a posteriori identity or constitution claims between moral considerations and categorical reasons, much like the way in which the statement ‘this is water’ entails ‘this is H2O’. There is no conceptual connection between WATER and H2O, yet if there is water then there is guaranteed to be H2O. Call this a relation of metaphysical entailment. Thus conceptual and metaphysical entailment relations differ only about whether or not the holding of the relation is somehow built into the concepts themselves. Hence in trying to settle the Non-Negotiable Commitment Claim error theorists can also use the Metaphysical Entailment Claim
Moral discourse’s mode of commitment to MRT is one of metaphysical entailment
This paper discusses linguistic data with which we can decide which of CEC, PC or MEC, if any, we should adopt as our model for thinking about the Non-Negotiable Commitment Claim (§2). I argue that the data initially favour PC over an entailment view (§3–4). I then show that in fact CEC and MEC can accommodate the data (§5). I conclude by
8
Are moral concepts likewise defective if we take reference to such acts as stealing and killing out of the applicability conditions of moral concepts, thus allowing them to be used to refer to such acts as looking at hedgehogs in the light of the moon (Foot 1958: 512)? I won’t explore this here.
Moral Error Theory, Entailment and Presupposition
927
suggesting that because of various other problems with CEC and PC, error theorists should consider developing MEC (§6).
2 Moral Statements and Their Commitment to Moral Reality In this section I use the fact that moral discourse consists of different kinds of moral statements to develop a test with which we can determine whether we should accept an entailment or a presupposition view. Let us therefore be more precise about the various moral statements that comprise moral discourse. Consider: Wrong Stealing is morally wrong Wrong is an atomic moral statement; ‘atomic’ because it has the logical form ‘x is F’, and ‘moral’ because the predicate F is a moral predicate.9 It is possible to embed Wrong under various linguistic operators, such as negation, modal and question operators. Doing that gives us three further, non-atomic moral statements: Negation Stealing is not morally wrong10 Modal Stealing might be morally wrong Question Is stealing morally wrong?11 Additionally, there is a class of moral statements that, although they share their atomic logical form with Wrong, do not trigger the thought that there is objective prescriptivity afoot, and hence do not trigger the thought that there must be categorical moral reasons (recall that error theorists postulate categorical reasons in trying to make sense of objective prescriptivity). This class of moral statements includes: Permissible
Stealing is morally permissible
After all, if stealing is morally permissible then, at least intuitively, there doesn’t have to be a categorical reason to steal and there also doesn’t have to be a categorical reason not to steal. Let’s stick with these five moral statements for the moment. One way to test whether some kind of entailment view is correct, regardless of whether or not we ultimately want to say that the entailment holds in virtue of connections between concepts (for that is a moot point in this context), or whether instead a presupposition view is correct is to see how well these views fare with respect to these five moral statements. Here is what I mean by that. Each of the five moral statements commits its utterer to MRT to varying degrees (this will be argued in the remainder of this section) and so if it turns out that the entailment view or PC accounts for this better than the other then that speaks in favour of adopting that view as our account of the Non-Negotiable Commitment Claim. I identify predicates with predicate words such as ‘wrong’, not with expansions of these to include connective verbs to get ‘is wrong’ or even statement-frames like ‘___ is wrong’. Nothing of substance hangs on this. 10 To avoid unnecessarily longwinded formulations I take Negation to be equivalent to ‘it is not the case that stealing is morally wrong’. I also use the negation in Negation in its ordinary truth-functional rather than metalinguistic way. 11 Earlier, and in the context of a discussion solely involving atomic moral statements, I defined moral statements as linguistic utterances that express moral judgements understood as beliefs. From this point onwards, and for ease of exposition, I broaden my understanding of ‘moral statement’ to include non-atomic contributions to moral discourse, such as questions, that do not straightforwardly express beliefs. 9
928
W.F. Kalf
Wrong, given its atomic logical form, straightforwardly commits its utterer to MRT. Negation straightforwardly does not commit its utterer to MRT. How about Modal, Question and Permissible? Consider moral nihilists, ordinary participants in moral discourse deeply sceptical about moral truth (Dreier 2006: 242).12 Nihilists are not willing to utter Modal and Question. Modal suggests that some things can be morally wrong even though nihilists believe that nothing is morally anything. Question leaves it open that some things are morally something whereas nihilists think that this question is closed (and that the answer is ‘no’). This behaviour of nihilists—their reluctance to utter Modal and Question—suggests that Modal and Question carry some kind of commitment to MRT. The eventual account of the mode of commitment of moral discourse to MRT should explain this. Now consider Permissible. It is possible to have the following sense of permissibility in mind when saying that something is permissible: ‘stealing is positively morally permissible; i.e., stealing has the property of being permitted by morality’. But sometimes when people say that something is permissible they mean: ‘stealing, like everything else, is permissible because there is no morality’. In that case we have: Permissible*
Stealing is permissible, but not morally permissible or permissible according to morality, because there is no morality
Nihilists will be willing to utter Permissible* for obvious reasons. But they won’t be willing to utter Permissible. This is because: “facts about moral permissibility … [are] not … objectively prescriptive but rather, as we might say, objectively permissive” (Olson 2011: 79n.4) To understand why nihilists are unwilling to utter Permissible, recall that error theorists are impressed by morality’s apparent inescapable authority: its ability to bind us to perform certain actions (authority) such that there is nothing we can do, not even altering our desires, to escape this authority (inescapability). It is built into this notion of inescapability that moral facts would have to be objective in the sense of being independent of human desires or attitudes more broadly. Otherwise we could change what is morally wrong by altering our attitudes and thus ‘escape’ morality’s authority. But this means that the prescriptive force of moral considerations would have to be “simply there, in the nature of things” (Mackie 1977: 59). And this is intolerably queer—nothing can be objective and prescriptive at the same time.13 But if it is mysterious how things can out of themselves demand the performance of actions then it is equally mysterious how things can out of themselves permit the performance of actions. The nihilists’ reluctance to utter Permissible tells us that Permissible carries a commitment to:
MRTpermissible:
There exists objective permissibility
I use the locution ‘nihilist’ to talk about ordinary moralists and reserve ‘error theorist’ to talk about metaethicists subscribing to error theory. 13 Mackie thought that things can be subjective and prescriptive; he believed in hypothetical reasons (1977: 29). He probably thought this because he didn’t distinguish normativity and motivation, thus believing that since desires are subjective and motivationally efficacious, they are subjective and ‘prescriptive’. But if we do distinguish normativity and motivation then it is not clear that error theorists are entitled to hypothetical normativity—how is it that desires are normative (and not just motivationally efficacious) in a naturalistically respectable way? 12
Moral Error Theory, Entailment and Presupposition
929
Facts about objective permissibility, as remarked earlier, do not entail facts about categorical reasons, so the error theorist can’t specify MRTpermissible further in terms of such reasons. But this is not a problem for me, as it is my purpose to investigate whether PC or an entailment view best accounts for the fact that Permissible carries a commitment to there being a moral reality, and for that purpose I don’t need to understand MRTpermissible in more detail. Nihilists’ unwillingness to utter Permissible has a further implication: they will also be unwilling to utter Negation. Consider a suitable kind of deontic logic, one that treats various deontic operators as inter-definable, including ‘not impermissible’ and ‘permissible’. If we further assume that ‘not wrong’ means, or is (semantically) equivalent to, ‘not impermissible’ and hence ‘permissible’, we can conclude that Negation logically entails Permissible.14 This means that although it remains true that Negation carries no commitment to the moral reality thesis understood as MRT (which mentions categorical reasons), we do find that Negation carries a commitment to the moral reality thesis understood as MRTpermissible. For Negation entails Permissible and Permissible carries a commitment to MRTpermissible. The fact that people sometimes have Permissible* in mind when they say that something is permissible does not threaten my claim that the error theorist’s eventual account of the Non-Negotiable Commitment Claim needs to explain in virtue of what Permissible is committed to MRTpermissible. What would threaten this claim is when people always have Permissible* in mind when they say that something is permissible, but that is currently an empirically underdetermined claim (and, I venture to suggest, probably false). The same, by the way, holds for Modal. It too can be given an alternative interpretation according to which nihilists will be willing to utter it. This alternative interpretation is one according to which the ‘might’ in Modal gets an epistemic rather than metaphysical reading. The nihilist can say: ‘I acknowledge that I am a fallible epistemic agent and thus, for all I know, stealing might be wrong’. She will be willing to utter this because worries about epistemic access to (moral) reality don’t imply a commitment to there being a (moral) reality. But again this would threaten my conclusion that Modal is committed to MRT only if Modal is never used in its non-epistemic sense. But this too is a currently unsettled empirical claim (and again one that is probably false). We need an explanation of the fact that each of the five moral statements carries a commitment to MRT or MRTpermissible, at least to the extent that it does (that is, the commitments of the non-atomic moral statements and Permissible may or may not be non-negotiable; I haven’t argued this point either way). It seems that this favours PC over CEC and MEC. This will now be argued.
3 The Entailment Claim It is a perfectly general feature of entailments that embedding atomic sentences carrying them under negation operators causes the embedded sentence to lose that entailment. Stalnaker writes: “A entails B if and only if B is necessitated by A but not by its denial” (1999: 54) 14
I won’t defend this assumption here—it looks like the natural thing to say and hence it seems that it is incumbent on those who disagree to come up with an argument that ‘not wrong’ is not semantically equivalent to ‘permissible’.
930
W.F. Kalf
Thus ‘this is a vixen’ entails ‘this is a fox’ but ‘this is not a vixen’ does not entail ‘this is a fox’. Likewise, Negation doesn’t entail MRT. This is problematic because Negation is committed to the existence of a moral reality: it entails Permissible, which carries a commitment to MRTpermissible. But it has not yet been argued that the latter commitment is an instance of a (conceptual) entailment. So here is an explanatory task for the entailment view: explain how Permissible entails MRTpermissible and hence how Negation entails MRTpermissible. In addition to Negation and Permissible, consider Modal and Question. Modal and question operators are also entailment-cancelling operators; e.g., ‘stealing might be morally wrong’ does not entail that it is wrong, and hence does not entail there is a categorical reason not to steal. This is also problematic, for as we saw Modal and Question do carry a commitment to MRT. The entailment view is extensionally inadequate, and indeed very much so. My reasoning in this section is open to two objections. First, advocates of the entailment view can “treat [moral discourse] as a term of art … meaning a widespread linguistic practice of uttering atomic [statements]” (Joyce 2007: §4; also see Sinnott-Armstrong 2006: 34–6, Robertson 2008: 109). If error theorists treat moral discourse in this way then my explanatory challenge no longer applies to them, for then moral discourse, by hypothesis, no longer contains non-atomic moral statements. But making this move is to abandon the aim of providing an error theory about ordinary moral discourse, which consists of lots of other kinds of moral statements besides atomic ones. Yet that was the nature of the game. This objection doesn’t stick. The second objection can be understood as Joyce’s attempt to explain away the difficulties that Modal, Question et cetera present. Joyce can “agree that people are not always the best judges of what they are talking about” (2011: 527) and insist that the telling evidence is how nihilists respond to cases where the commitment to MRT is explicitly denied, not how they respond to complicated statements like Modal. In reply, I agree that the folk may be mistaken about their own discourse, thinking that a certain non-atomic statement like Modal carries a commitment to MRT even though, by hypothesis, it doesn’t. But I insist that in that case we need an explanation of why the folk are mistaken in the systematic way that they are. We have a large number of statements that seem to carry a commitment to MRT—Modal, Question, Permissible, Negation via Permissible, and probably many more—and if none of them in fact carries that commitment then the fact that they appear to requires explanation. Either way, Joyce and advocates of the entailment view more generally have explanatory work to do. This objection doesn’t stick either. The entailment view is unable to account for the fact that various moral statements embedded under entailment-cancelling operators seem to carry commitments to MRT. It also has yet to explain how, exactly, Permissible (and hence Negation as it logically entails Permissible) is committed to MRTpermissible. I will now argue that the error theorist’s back-up plan, PC, does a much better job here.
4 The Presupposition Claim It is instructive to think about presupposition relations, just as we did with entailment relations, as holding between statements. Consider: Bald The present king of France is bald Not-bald The present king of France is not bald King France presently has a king
Moral Error Theory, Entailment and Presupposition
931
Intuitively, although Bald and Not-bald contradict each other in a sense because they ascribe mutually incompatible properties about his hairstyle to the present king of France, they both agree that France presently has a king. That is, again intuitively, Bald and Not-bald both presuppose King. We can make the notion of presupposition more precise with the projection test (Simons et al. 2010: 309). This test is used in linguistics and the philosophy of language to determine whether or not a piece of information as carried by an utterance belongs to its meaning. It is widely used as a diagnostic for presupposition. The term ‘projection’ refers to a phenomenon in natural languages where an implication survives as an utterance implication when the target statement (Bald) is embedded under an entailment-cancelling operator (Not-Bald). The implication King as carried by Bald ‘projects’ because it survives as an implication of Not-bald. The currently most inclusive and empirically adequate explanation of why the information as captured by King survives as an implication of Not-Bald is that this information is not at issue (Simons et al. 2010: 315–8). This hypothesis fits our case: in debating the hairdo of the present king of France, his existence is not at issue and hence projects. In the case of Bald and Not-bald it is fairly easy to determine what information projects. Indeed we may go so far as to say that we can simply ‘read off’ the utterance implication from Bald and Not-Bald. Often however figuring out whether a piece of information projects requires priming an appropriate context. Consider: Hammock
If she didn’t sleep in the hammock, then I don’t know where she slept (Simons et al. 2010: 5)
Hammock is felicitous as a contribution to our conversation only if a unique woman is salient in the common ground. We achieve the felicity of Hammock by priming as our conversational context that this woman is in the common ground, thus satisfying the presupposition of the pronoun ‘she’ as it occurs under the entailment-cancelling embedding of this ‘if-then’ construction. Hence presuppositions can be trigged by various things. They can be triggered by particular words without help of a previously primed context (as we had in the case of Bald, Not-Bald and King) or by the interplay of words and context (as we had in the case of Hammock). Although it is tempting to straightforwardly associate the class of semantic presuppositions with the former and more pragmatic ones with the latter, recent studies suggest that the relation between whether a presupposition is semantic or pragmatic on the one hand and what triggers it on the other is more complex (Beaver and Geurts 2011: §1). Thus a presupposition can depend on context and still count as a semantic presupposition. This is good news for error theorists: pragmatic presuppositions are cancellable, therefore negotiable, and therefore a poor model for understanding Non-Negotiable Commitment Claim. If Wrong, Negation, Modal, Question and Permissible presuppose MRT or MRTpermissible we can’t simply ‘read off’ this information from these statements. Contrary to Bald and Not-Bald, where King is pretty much contained within these statements, the information as captured by MRT and MRTpermissible isn’t likewise ‘contained within’ our five moral statements. For MRT mentions categorical reasons and MRTpermissible mentions objective permissibility and that specific information is not contained within our five moral statements. The information that is contained is that there exist moral properties, but that isn’t specific enough. Are moral properties constituted by purely non-normative properties, by normative properties and if so
932
W.F. Kalf
what kind of normative properties (categorical or non-categorical ones, intrinsic or non-intrinsic ones), or different still? So let’s prime an appropriate conversational context by stipulating that our conversation takes place between two moralists; ordinary people debating the wrongness or otherwise of stealing. One of them utters Wrong and his interlocutor utters Modal. On this assumption, given that ordinary morality seems to so many to be invested in objective prescriptivity, what is not at issue is the claim that there is such a thing as moral wrongness understood in terms of categorical reasons. What is at issue instead is whether moral wrongness understood in terms of categorical reasons applies to stealing. This means that the information that stealing is wrong, as entailed by Wrong (every statement entails itself) doesn’t project, but that the claim that there is such a thing as wrongness understood in terms of categorical reasons does project. This is exactly what we intuitively find, and so we may perhaps say that the claim that there is wrongness understood in terms of categorical reasons is presupposed by Wrong and Modal. Applying this to Question, we find that, at least when the intonation is on stealing (as in ‘is stealing wrong?’) the information that there is wrongness understood in terms of categorical reasons projects. For that information is not at issue if you wonder about the wrongness of stealing. And something similar holds for Permissible. If I utter Wrong and you utter Permissible then that there are things that have moral properties is not at issue and hence this information projects; for the only thing that is at issue is whether stealing has the property of being permissible or of being wrong. But whether the information that projects must be understood in connection with categorical reasons, objective permissibility, or a mixture of these is not clear; I assume that an eventual, more fully developed account of PC can work this out. There is a second assumption we have to make to be able to say that PC can account for Modal, Question and Permissible presupposing that there is a moral reality (however precisely understood). It is that we have aptly and yet sufficiently minimally primed our conversational context. This is important because if we put too much into the conversational context (i.e., if we simply ‘enrich’ the context with objective prescriptivity, objective permissibility, or both) then we might find a presupposition that wouldn’t otherwise be there. A more fully worked-out version of PC would have to defend this second assumption as well. As noted above, semantic presuppositions are in one respect similar to entailments: they are non-cancellable. Or at least this holds for atomic statements and their presuppositions. Consider entailments: adding ‘mind you, this is not a fox’ to my original statement ‘this is a vixen’ is not a felicitous contribution to our conversation. Likewise, denying King after having uttered Bald is not felicitous either. But presuppositions are cancellable when embedded (Beaver and Geurts 2011: §3). If I say ‘The present king of France is not bald—there is no present King of France!’ then I have felicitously contributed to our conversation about the hairdo of the present King of France even though I’ve cancelled the presupposition of Not-Bald. Likewise, ‘stealing is not wrong—nothing is wrong!’ is a felicitous contribution to a conversation about the wrongness or otherwise of stealing even though we have a cancellation of the presupposition of Negation. Presuppositions’ limited cancellability doesn’t affect PC negatively. After all, if the presupposition of Negation is cancellable then this may just show that Negation’s commitment to MRT was weaker than the commitment of Wrong to MRT—i.e., that this commitment of Negation to MRT was negotiable, as indeed it seems to be. For, as was just implied,
Moral Error Theory, Entailment and Presupposition
933
nihilists have no problem uttering Negation when its (negotiable) commitment to MRTpermissible has been overtly cancelled. Are the presuppositions of Modal and Question cancellable? Can we say ‘stealing might be wrong—but mind you, nothing is morally anything’? This is not entirely clear, so I will make a third assumption: a more fully worked-out account of PC can get these details right. We had to make three substantial assumptions to make PC work, but with these assumptions in place we do seem to have prima facie reason for using it in settling the Non-Negotiable Commitment Claim. The reason is that it provides at least the beginnings of an explanation of the behaviour of nihilists vis-à-vis our five moral statements, whereas the entailment view seems unable to provide any explanation at all. I will now argue that in fact slightly altering the entailment view allows it to have an explanation as well.
5 Entailment and Pragmatics Our discussion of the entailment view in section 3 shows that it consists of the following theses: Positive Thesis Atomic moral statements (except those about permissibility) entail MRT Negative Thesis Non-atomic statements do not entail MRT Negative Thesis bears its name because it says that the entailment view is only committed to the, precisely, negative thesis that these statements do not entail MRT (recall that non-atomic statements are formed with entailment-cancelling operators). We found this problematic because our non-atomic moral statements do seem to carry a commitment to MRT. The entailment view understood as comprised of Positive Thesis and Negative Thesis is also silent about Permissible and its relation to the moral reality thesis, which is also problematic as Permissible too seems to carry a commitment to their being a moral reality. Here is a proposal for solving these problems. Modal, Question and Permissible keep the existence of a moral reality open as a live option because they conversationally implicate: Reasons There exists objective permissibility as well as objective prescriptivity and hence, in virtue of the latter fact, categorical moral reasons Conversational implicatures are standardly understood along Gricean lines. Grice explains implicatures, or how statements convey information without that information being part of their meaning, with the aid of two phenomena: the cooperative principle and conversational maxims. The cooperative principle says: “[m]ake your conversational contribution such as is required at the stage at which it occurs, by the accepted purposes or direction of the talk exchange in which you are engaged” (Grice 1989: 6) The maxims tell you what to do to meet the principle; e.g., the maxim of quantity tells you to make your contribution neither less nor more informative than is required, and the maxim of relation tells you to make your contribution relevant. According to Grice, we are justified in assuming that people conform to the cooperative principle and hence to the maxims (absent special circumstances such as when people are acting). So for example, if you ask me ‘Will Pete be at the meeting tomorrow?’ and I answer ‘his car broke down’ then my utterance implicates that he won’t be at the meeting. For the implicature is calculable:
934
W.F. Kalf
you make the justified inference that my original statement has this implicature because that is the best way you have of making sense of my obligation to abide by the maxim of quantity as a means of abiding by the cooperative principle. After all, if my statement didn’t carry this implicature then I would be saying something less informative than was required to answer your question whether Pete will be at the meeting. Note that the implicature is also cancellable: it is a felicitous addition to my statement ‘his car broke down’, and ultimately to our conversation, if I continue to say ‘but mind you, Pete will be at the meeting tomorrow—he’ll go by bike’. I myself might go through your calculation of the implicature, realise that you justifiably make the inference that my statement implicates that Pete won’t be at the meeting, and hence make this implicaturecancelling addition to save the conversation. In our case, the implicature between Modal, Question and Permissible on the one hand and Reasons on the other holds because, looking at Permissible for example, we have in our society a moral standard according to which morally relevant actions are either permissible or forbidden (but not both) and because to state that information explicitly would be to violate the Gricean maxim of quantity, which forbids us to be overly informative when we speak. Here is the calculation of the implicature. You utter Permissible. I am justified in thinking that you abide by the maxim of quantity as your means of abiding by the cooperative principle, and since I know that it is part of our conversational context that we have a moral standard in place according to which various morally relevant actions are either morally prescriptive or morally permissive, I know that your utterance of Permissible carries the implicature that there are other actions that carry categorical moral reason-giving force. This explains why nihilists are not willing to utter Permissible. By uttering Permissible they commit themselves, via its implicature to Reasons, to the existence of categorical reasons. But nihilists can also reveal their hands and clarify that they mean to say that stealing isn’t wrong because nothing is morally anything. Although Wrong still entails MRT in that case, the implicature of Permissible to Reasons gets cancelled, and so nihilists will be happy to say that something is permissible—for then it has become clear that what nihilists intend to convey is Permissible*. Note that Negation, since it (logically) entails Permissible, gets a similar treatment. Nihilists are initially unwilling to utter it because it expresses allegiance to the moral game: Negation entails Permissible which conversationally implicates Reasons. But when that conversational implicature gets cancelled, nihilists are willing to utter Negation. They might say: ‘yes, stealing is not wrong, and hence (non-morally) permissible, because nothing is morally anything’. Something similar again holds for Modal and Question; these are also statements that nihilists are initially unwilling to utter as they are suggestive of things having moral status, and thereby trigger recognition of Reasons. And nihilists will be happy to utter Modal and Question once they have cancelled these implicatures. For then Modal may take up its epistemic reading and Question could be used as a means of testing the allegiance of others to either moral realism or moral skepticism. ‘Is stealing wrong?’ will elicit a positive response from a moralist and the following, implicature cancelling response from the nihilist: ‘No, stealing is not wrong—nothing is wrong!’ Hence I propose to keep Positive Thesis and to reformulate Negative Thesis as follows: Negative Thesis* Non-atomic moral statements as well as statements about moral permissibility implicate Reasons, and they lose this implicature if nihilists explicitly deny it
Moral Error Theory, Entailment and Presupposition
935
With Positive Thesis and Negative Thesis* advocates of either entailment view can explain the behaviour of nihilists vis-à-vis the various moral statements we have been considering. This means that PC has lost its dialectical advantage over the entailment views.
6 Metaphysical Entailment At this point our choice between CEC, MEC and PC is wide open again. Is there anything else we can say? There is, for there are well-known objections to CEC. Most of them derive from the fact that CEC states that the entailment holds in virtue of connections between concepts. One such objection is most forcefully formulated by Finlay (2008). It is that there is an alternative way of interpreting the appearance of moral discourse dealing in categorical reasons that doesn’t build that information into (the applicability conditions) of moral concepts. For we can instead build that information into a false folk belief that the referents of moral terms standing for moral concepts are categorical moral reasons even though in fact, and unbeknownst to the folk, moral terms successfully refer to properties that do not have categorical reason-giving force. An example would be the purely natural property of being an act of stealing. We can liken this situation to statements about water as they were made before Lavoisier, in which the concept WATER was used with the intention to refer to an element instead of the actual compound that it is. Had we not had the option of locating that error in the attitudes of otherwise competent participants in pre-Lavoisier water discourse to the concept WATER, thereby avoiding having to locate it in the concept WATER itself, then we would have had to infer that pre-Lavoisier water discourse licences an error theory (Finlay 2008: 351–2). For the concept WATER-AS-AN-ELEMENT is uninstantiated. But we can shift the locus of the error from the level of an allegedly infected concept to the level of false beliefs about an otherwise error-free concept, thereby saving water discourse from error. Likewise for moral discourse: relocate the error from moral concepts to folk attitudes towards these concepts and an error theory no longer seems appropriate.15 This objection, if successful, would completely undermine CEC. If the objection’s alternative interpretation of the datum that moral discourse evinces allegiance to categorical reasons is correct then MRT is not built into moral concepts. The appearance of moral discourse’s commitment to MRT is solely a function of false folk beliefs; moral concepts are error-free. There have been various replies to this objection on behalf of CEC (Olson 2011; Joyce 2011, MS), though Finlay has responded (2011), and it is not clear who is right here. No matter, for MEC can completely avoid Finlay’s objection. After all, if one only wants to show that two statements stand in a relation of metaphysical entailment then it is not necessary to invoke considerations about connections between concepts—various a posteriori considerations, including (but not limited to) speaker-referential intentions, suffice. So even if Finlay is right that a commitment to MRT isn’t part of moral concepts that doesn’t justify moving away from an entailment view entirely. We still have MEC. This gives MEC an important dialectical advantage over CEC, although it remains incumbent on advocates of MEC to show on exactly what basis we can conclude that the kind of entailment relation it puts forward obtains.
An ‘attitude error theory’, according to which the error with morality solely resides in our attitudes towards moral concepts doesn’t seem to deserve to be called an error theory at all. It presents an error that is too easy to rectify: simply change your attitudes towards moral concepts and you’ll have licensed a ‘success theory’.
15
936
W.F. Kalf
Regarding PC, we saw that it makes at least three assumptions that stand in need of defence. One was that we can specify a conversational context in which the presupposition that there is a moral reality (specified in a certain way) arises even when that context is described sufficiently minimally in order to avoid the charge that we are building into the context what we want to get out of it. This is not evidently possible, and since the advocate of MEC, dealing with entailments, doesn’t have to prime conversational contexts in the same way she has a dialectical advantage over PC. The second assumption was that in a conversation in which Wrong and Permissible get uttered, it is not clear which of MRT or MRTpermissible (or both) gets presupposed, and this too would need to be settled. Again, this task is one that a friend of MEC doesn’t have to perform. The final assumption was that an advocate of PC will be able to explain to what extent the presuppositions of Modal and Question are cancellable with felicity. Here however the advocate of MEC faces a similar problem—she has to explain how the implicatures of Modal and Question are cancellable with felicity.16 Yet even though there may be a tie when it comes to this point, the fact that MEC doesn’t have to defend the other assumptions of PC means that it enjoys a dialectical advantage over PC as well. I don’t think that any of these considerations provide decisive reason to prefer MEC over CEC and PC. But since MEC seems able to avoid many of the difficulties that beset CEC and PC, I do think that error theorists should at least take MEC seriously. Acknowledgments Many thanks to Pekka Väyrynen, Andrew McGonigal, Brian McElwee, Simon Kirchin, Gerald Lang, Ulrike Heuer, John Divers, Ben Saunders, Michael Bench-Capon, Paul Ramshaw, Thomas Brouwer, Carl Baker, Tim Taylor and an anonymous referee for BSET for instructive feedback on earlier drafts, as well as audiences at BSET Annual Conference 2012, University of Stirling; 5th Annual ArchéCSMN Philosophy Graduate Conference, University of St. Andrews; Third Annual Dutch Conference on Practical Philosophy, University of Amsterdam; Centre for Ethics and Metaethics Seminar, University of Leeds; Postgraduate Seminar, University of Leeds. Research on this paper has been financially supported by an FP7 (2007/13) Marie Curie International Reintegration Grant and a Jacobsen Fellowship from the Royal Institute of Philosophy (2012/13)
References Beaver DI, Geurts B (2011) Presupposition. In: Zalta EN (ed) Stanford Encyclopedia of Philosophy. http:// plato.stanford.edu/entries/presupposition Blackburn S (2005) Quasi-realism no fictionalism. In: Kalderon ME (ed) Fictionalism in metaphysics. Oxford University Press, Oxford, pp 322–338 Clark P (2009) Mackie’s motivational argument. In: Sobel D, Wall S (eds) Reasons for action. Cambridge University Press, Cambridge, pp 200–19 Dreier J (2006) Moral relativism and moral nihilism. In: Copp D (ed) The oxford handbook of ethical theory. Oxford University Press, Oxford, pp 40–64 Finlay S (2008) The Error in the Error Theory. Australas J Philos 86:347–369 Finlay S (2011) Errors upon errors: A reply to joyce. Australas J Philos 89:535–47 Foot P (1958) Moral arguments. Mind 67:502–13 Grice P (1989) Studies in the ways of words. Harvard University Press, Cambridge MA Joyce R (2001) The myth of morality. Cambridge University Press, Cambridge Joyce R (2007) Moral anti-realism. In: Zalta EN (ed) Stanford encyclopedia of philosophy. http:// plato.stanford.edu/entries/moral-anti-realism Joyce R (2008) Morality, schmorality. In: Bloomfield P (ed) Morality and self-interest. Oxford University Press, Oxford, pp 51–75 16
This ties into the possibility that MEC plus implicature mechanisms may turn out to be a notational variant of PC plus cancellable presuppositions of non-atomic moral statements. I leave researching this option to another day.
Moral Error Theory, Entailment and Presupposition
937
Joyce R (2011) The error in “the error in the error theory”. Australas J Philos 89:519–34 Joyce R (MS) Enough with the errors! A final reply to Finlay. http://www.victoria.ac.nz/staff/richard_joyce/ acrobat/joyce_enough.with.the.errors.pdf Kirchin S (2010) A tension in the error theory. In: Joyce R, Kirchin S (eds) A world without values. Springer, Dordrecht, pp 167–82 Mackie JL (1977) Ethics: inventing right and wrong. Penguin, Harmondsworth Olson J (2011) In defense of moral error theory. In: Brady M (ed) New waves in metaethics. Palgrave Macmillan, Basingstoke, pp 62–84 Robertson S (2008) How to be an error theorist about morality. Polish Journal of Philosophy 2:107–25 Shah N, Evans M (2012) Mental agency and metaethics. In: Shafer-Landau R (ed) Oxford studies in metaethics 7. Oxford University Press, Oxford, pp 81–109 Simons M et al (2010) What projects and why. In: Li N, Lutz D (eds) Proceedings of SALT (Semantics and Linguistic Theory) 20. CLC Publications, Ithaca NY, pp 309–27 Sinnott-Armstrong W (2006) Moral skepticisms. Oxford University Press, Oxford Smith M (1994) The moral problem. Basil Blackwell, London Stalnaker RC (1999) Context and content. Oxford University Press, Oxford