In the last chapter, I explained what the moral error theory is, and the kinds of con- siderations that might lead us to believe that it is true. For the remainder of this the- sis, however, I want us to assume that the moral error theory is true and, moreover, that we believe that it is true. Our concern will be the question that follows this as- sumption: what ought we to do next?
WNQs are certainly not unique to meta-ethics; they currently occupy the minds of philosophers of mathematics, philosophers of mind, and even those who have em- braced an error theory about colour. But I do not take the sole justification for ad- dressing our WNQ to be an argument from precedent. There is more to be said to motivate the project, and saying more is the task of this chapter.
I will begin by explaining what kind of question our WNQ is (§2.1). As I have characterised it, the question is a normative one—a question regarding what we ought to do if the moral error theory is true. However, it is not obvious that our WNQ can be a normative question. The arguments marshalled in support of moral error theory may support an error theory about all normative reasons—or so some philosophers have argued. If they are right, then we are to have no recourse to normative language. Others have claimed that the moral error theorist is committed to denying the exist- ence of epistemic reasons, which (it is alleged) are also categorical in nature. If they are correct, then we cannot speak of reasons for belief. I will explore and defend what I take to be the best strategies for responding to these over-generalisation prob- lems. Doing so can, I hope, earn me the right to speak of various kinds of (non- moral) reasons for the remainder of the work.
The tasks for §2.2 and §2.3 respectively will be to specify to whom our WNQ is addressed, and to motivate the project of providing an answer to it. I will then out- line the different answers to the WNQ for moral discourse that have been proposed
in the current literature (§2.4). (All but one will be explored in the remainder of the thesis.) The chapter concludes by pointing towards the philosophical pay-offs of the project (§2.5). These dividends are not confined to moral error theorists. The project should also be of interest to those contemplating error theories in their own sectors of philosophy, and indeed, to meta-ethicists more generally.
§2.1 CAN THE WNQ FOR MORAL DISCOURSE BE A NORMATIVE
QUESTION?Our first order of business will be to explore whether the WNQ for moral discourse can be a normative question. It is not uncommon for philosophers to characterise the question in this way; as a question regarding what we should do, or ought to do, or have most reason to do with moral discourse, following the moral error theory (e.g., Joyce, p.177; Nolan, Restall & West 2005, p.310; Lutz 2014, p.352; Svoboda 2017, p.3). In doing so, they would appear to presume that the moral error theorist’s rejec- tion of categorical reasons leaves many other, non-moral reasons intact.
Yet it is not obvious that the moral error theorist’s arguments can be insulated in this way. Some think that she is committed to an error theory about epistemic rea- sons as well (Cuneo 2007, Rowland 2013). Others argue that she is committed to an error theory about all normative reasons (Hampton 1995, Korsgaard 1997, Shafer- Landau 2003, Raz 2005, Bedke 2010) These are, of course, over-generalisation wor- ries; for an error theory about all normative reasons (or even just epistemic reasons) seems undesirable. And insofar as the moral error theory entails this more global the- sis, so too, it seems, is moral error theory. These worries would also seem to be bad news for the project of answering our WNQ. If there are no reasons to do or to be- lieve anything, then, a fortiori, there can be no reason to do anything with or believe anything about moral discourse going forward.
My goal in this section will be to earn the right to normative language for the re- mainder of the work. I shall argue that the moral error theorist is not committed to an error theory about epistemic reasons (§2.1.1), nor to a more thoroughgoing error theory about all normative reasons (§2.1.2).
§2.1.1 The over-generalisation problem for epistemic reasons
I shall tackle the over-generalisation problem for epistemic reasons and the more global over-generalisation problem separately, beginning with the former.56 Some
philosophers argue that the error theorist’s attack upon moral reasons is likewise an attack upon reasons for belief. Epistemic reasons, they hold, are similarly categorical; whether or not an agent has a reason to believe some proposition p does not depend upon whatever ends she happens to have (Cuneo 2007). Richard Rowland, for ex- ample, writes that
…our understanding of epistemic reasons and justification also entails that
there are categorical reasons…it seems that there is reason for everyone to
believe that dinosaurs once roamed the earth regardless of what they want to believe. (2013, p.4)
I think that Christopher Cowie’s is the strongest response to this line of argument, and I will draw heavily upon his work in what follows, before adding some develop- ments of my own.
As Cowie (2014a, pp.117-8) notes, intuition-eliciting cases like Rowland’s do not make for a very dialectically effective response to the moral error theorist. She will simply deny that anyone has a categorical reason to believe that dinosaurs once roamed the earth. There is, of course, a claim in the vicinity that she will endorse; she will concede that the fossil record provides evidential support for the claim that dino- saurs once roamed the earth. But this is just to say that the evidence supplied by the fossil record raises the probability that dinosaurs once roamed the earth. It is not to say that the evidential support relation supplies anyone with a categorical reason to be- lieve that dinosaurs once roamed the earth.
Jonas Olson enlists a similar strategy. As he observes, ‘epistemic reason’ is ambig- uous between ‘evidence’ and ‘reason for belief’ (2014, p.158). The former notion is not itself normative, and, he suggests, it is enough to get us by; for evidence can still supply hypothetical reasons for belief—reasons that depend upon our having adopted
56 There are good grounds for addressing the specific problem for epistemic reasons directly. The strategy of enlisting epistemic reasons as partners in guilt (or innocence) is well-known, and so, worthy of discussion in its own right. It will be also helpful to have in hand an account of epistemic reasons that is consistent with moral error theory.
the end of being evidence-responsive, for example.57 (I return to the latter idea be-
low.)
At this stage, it is open to the moral error theorist’s opponent to maintain that ev- idential support relations just are categorical reasons for belief. But it’s not obvious that this claim is plausible. To be sure, it might seem intuitive to say that the fact that some evidence e raises the probability of some hypothesis h is a reason to believe h. The crucial question, however, is whether this ‘reason’ is properly thought of as a normative reason.
Cowie is sceptical that it is, and I share his scepticism. On the common view of things, a normative reason for believing some proposition p is evidence that one ought to believe it (Kearns & Star 2009), or an explanation as to why one ought to respond to this evidence in a particular way (Broome 2013). And it just does not seem true that evidence for some proposition p is always evidence that one ought to believe that p. Cowie makes the point by appealing to banal truths:
… Suppose that I possess, and am aware of possessing, evidence e that
bears on some proposition p. But suppose that I have no interest in arriving at a true or evidentially supported belief about that proposition. And sup- pose that it would not serve any practical end for me to do so. If one never- theless maintains that e is evidence that I ought to believe that p (and not merely that e is evidence for the truth of p), the burden is surely very much on them to explain why. (2014a, p.121)
So the error theorist will concede to her opponent that the fossil record provides ev- idential support for the proposition that dinosaurs roamed the earth. Likewise, she will concede that this evidential support relation is a reason to believe that dinosaurs roamed the earth in one sense of ‘reason’. But she will deny that we are speaking here of a normative reason.
Just what are we speaking of, then? Cowie (2014a, pp.120-1) alludes to “institu- tional reasons”, such as those supplied by the norms of etiquette. These norms may provide us with reasons to say, wear pants to dinner. But again, it does not necessari-
ly follow that anyone has a normative reason wear pants to dinner.58 Presumably,
whether some individual has a normative reason to wear pants to dinner depends upon whether she has a normative reason to participatein the institution of etiquette in the first place. Cowie suggests that the same is true of epistemic reasons: if one is to have a normative reason to believe some proposition p, then one must “also require a reason to engage in the business of believing (the truth) with respect to that proposi- tion” (2014a, p.121; see also Heathwood 2009, p.96; Olson 2014, pp.165-8).
I think this idea is important, and worth developing. The basic thought, I take it, is that there is a kind of epistemic activity or institution that is in some sense norma- tively optional for us. We may or may not have normative reasons to be in the business of responsible believing—to shape our beliefs in accordance with our evidence, and the like.
Thankfully, it seems that many of us do in fact have such reasons; for many of us do have ends that are served by responsible believing—the end of believing truths, and the end of not believing falsehoods, say. Assuming that many of us do have such epistemic goals, many of us will have epistemic reasons to respond appropriately to our evidence. (Insofar as responding appropriately to one’s evidence is likely to be an effective means for arriving at truth, that is. I assume here that it will be—at least in typical circumstances.59) Moreover, even those who don’t value truth per se may have
good instrumental reasons to be responsible believers. After all, most of us care about furthering our ends, and we’re generally better positioned to further those ends—whatever they may be—if we track truth effectively (Kornblith 1993, p.371; 2001, pp.158-9). This way of seeing things dovetails nicely with the error theorist’s claim that all normative reasons are hypothetical—that is, contingent upon our (in- trinsic) ends or desires.
58 This assessment is not uncommon. It is, for example, plausible that undesirable social policies may supply legal reasons to harm others, without supplying any normative reasons to do so (Lillehammer 2002, pp.54-5; see also Joyce, pp.34-42).
59I do not assume that believing in accordance with one’s evidence will always result in the acquisi- tion of true beliefs, nor that believing truly will always further one’s various goals—that would be too strong (see Stephens 2001). There are bound to be exceptions to such unqualified claims. But follow- ing Cowie, these complications should not “…obscure the basic point that forming and revising be- liefs on the basis of one’s evidence is of great practical utility” (2014b, p.4007; see also Wedgwood 2002).
The approach is, of course, a variety of epistemic instrumentalism; the view that epistemic rationality is a species of instrumental rationality, or rationality in the pur- suit of one’s goals. (I remain neutral here on the matter of whether these goals must be distinctively epistemic ones—the goals of believing truth and avoiding falsehoods, say.60) Loosely following Adam Leite (2007, p.456), we can characterise epistemic in-
strumentalism as follows:
Epistemic instrumentalism
A belief is epistemically rational when (and because) holding it is instrumentally ra-
tional given one’s goals, and one has an epistemic reason to believe some proposi- tion p when (and because) doing so would be instrumentally rational given those goals.
This view has an impressive fan base. (See, for example, Foley 1987; Kornblith 1993, 2002; Laudan 1990, 1991; Maffie 1990; Papineau 1999). But I do not pretend that it is uncontroversial. (For criticism, see Kelly 2003, Lockard 2013.) There will be some who are not particularly taken with this account of epistemic reasons. But my ambi- tions here are rather modest. I am only attempting to show that given epistemic in- strumentalism, the error theorist has a reply to the over-generalisation worry for epistemic reasons, and there is a respectable notion of an epistemic reason to which she can appeal.61 Adopting this understanding can, I hope, earn me the rightto speak
of epistemic reasons for the remainder of this thesis. (The reader is therefore free to take my conclusions in what follows to be conditional on the truth of epistemic in- strumentalism—a matter that I haven’t fully adjudicated here owing to considerations of space and priorities.)
60On this issue, see Lockard (2013), who distinguishes “intellectualist” varieties of epistemic in- strumentalism, according to which epistemically rational beliefs are those which serve an agent’s cog- nitive or epistemic goals, from “pragmatist” varieties, which do not require that the relevant goals be epistemic ones.
61 It is no surprise that many moral error theorists (e.g., Joyce, pp.178-9; Olson 2014, pp.158-9) are (or very much seem to be) attracted to epistemic instrumentalism. The position is often hailed as a promising and respectable strategy for naturalising epistemic normativity, and shirking talk of categor- ical reasons. But it is not the only such strategy. One could instead interpret claims about epistemic rationality as expressions of preferences or tastes (see Field (2000)). Indeed, Stephen Ingram (2017) has recently recommended that moral error theorists be epistemic expressivists. I put this possibility to the side here.
There is, however, still a problem in need of address. We have assumed that indi- viduals typically have good hypothetical reasons to be responsible believers (e.g. to re- spond appropriately to their evidence) because responsible believing is a means of realising certain (intrinsic) ends that they have. But perhaps these hypothetical rea- sons simply cannot exist without categorical ones. It is to this latter complaint that I now turn.
§2.1.2 The over-generalisation problem for all normative reasons
The moral error theory, recall, is premised upon an instrumentalist conception of normative reasons and rationality: “practical instrumentalism”. On this view, an agent is practically rational to the extent that she takes what she justifiably believes to be the necessary means to her (intrinsic) ends. The practical instrumentalist also holds that there are only hypothetical reasons—specifically, reasons to adopt particular means when one has particular ends (Joyce, p.51).
Yet quite a few have voiced the suspicion that there must be categorical reasons lurking in the background of these hypothetical ones. Here is the problem. Practical instrumentalism tells us something about the transmission of reasons: they are trans- mitted from ends to means. But if reasons are to be transmitted, then there must be reasons to transmit. So—and here’s the rub—it seems that the instrumentalist re- quires a categorical, non-instrumental principle as a foundation for hypothetical reasons; she must posit a categorical reason to have certain ends (e.g., the end of taking the necessary means to one’s ends) if there is to be anything to transmit in the first place (Hampton 1995, pp.70-71; Korsgaard 1997, p.223; Raz 2005, p.23).62
Call this the grounding challenge; it is the challenge of saying just what (if anything) grounds or explains the normativity of hypothetical reasons, if not categorical rea- sons. If the challenge cannot be met—if, that is, hypothetical reasons must presup-
62 There is a slightly different articulation of the charge that the moral error theorist is committed to an error theory about all normative reasons, according to which hypothetical reasons-relations are just as metaphysically queer as categorical reasons-relations (Shafer-Landau 2003, Bedke 2010). However, this challenge seems more pertinent to Olson’s variety of moral error theory (which appeals to queer- ness concerns) than to that of Joyce (which does not). Since I am assuming the latter, I will not ad- dress this variant of the objection.
pose the existence of categorical ones—then (insofar as she continues to deny that categorical reasons exist) the moral error theorist seems committed to an error theory about practical reasons as well.
One tempting response to the grounding challenge is to dismiss any request for an explanation here as incoherent.63 Joyce seems to flirt with this possibility when he
suggests that asking for a reason to be rational is utter nonsense (p.49).64 One is ef-
fectively demanding a practical reason to be guided by practical reasons (‘what reason do I have to do what I have reason to do?’) and so, one is presuming that there are practical reasons.
Though tempting, this line of response is unlikely to be dialectically effective. Many philosophers agree that asking for a reason to take the means to one’s ends seems incoherent (e.g., Dreier 1997, p.95; Railton 2003, p.317). But they draw a ra- ther different conclusion. The nonsense involved in challenging the claim that we have a reason to take the means to our ends is not that of asking for a justification for it. It is the nonsense of trying to use the claim that we have a reason to take the means to our ends to do the justificatory work! The nonsense evaporates as soon as we pos- it a categorical reason to pursue our ends. (See Dreier 1997, p.96; Railton 2003, pp.318-9.)
Here is what I take to be a more promising strategy. The instrumentalist’s adver- sary assumes that she must offer a particular kind of explanation of an agent’s reasons: she assumes that the instrumentalist must explain why someone has a reason to per- form some action by showing that performing that action is a means of doing some- thing else that she has a reason to do (Schroeder 2007b, ch.3). Suppose, for example, that Luke desires coffee and desires to go to a nearby cafe because there is coffee there.
63 Some have also objected to the grounding challenge on account of its failure to distinguish be-