1. Reply to Ludwig

Let's start with Ludwig, and let's begin by getting one misunderstanding out of the way.

Ludwig takes me to task for directing my arguments against an overly simplistic version of the Humean account of pro-attitudes, according to which a pro-attitude toward winning at chess is, at bottom, a disposition to do whatever one believes will increase one's chances of winning.Footnote ³ He complains that, with the exception of intentions, individual pro-attitudes are not strictly speaking dispositions to act at all; their functional role is more accurately characterized in terms of the influence they exert first on one's judgements of expected utility and ultimately on the formation of one's intentions. There is consequently no need for Humeans to deny that a person could have a pro-attitude toward winning at chess, together with a belief that playing against weak opponents would maximize her chances of winning, yet no disposition whatever to act in that way. For, given the whole set of her pro-attitudes and beliefs, the expected utility of some other course of action could be much greater, and her intentions and thus her dispositions to act could reflect that fact.

However, I acknowledged that Humeans could interpret this case after their own fashion; my worry was that they would still be attributing to this person some disposition to play against weak opponents, when, in fact, it seems quite possible that she is in no way disposed to perform such actions.Footnote ⁴ Admittedly, if we adopt Ludwig's stricter manner of speaking, we should be putting the worry differently; instead of saying that Humeans would still be attributing to this person some disposition to play against weak opponents, we should say that they would still be attributing to her some disposition to judge that playing against weak opponents would have positive expected utility. To my mind, however, this remains a grave worry; for it seems to me quite possible that this person is in no way disposed to make such a judgement, for example, because she believes she has reason to try to win at chess only inasmuch as that would help her develop certain skills, and she does not believe playing weak opponents would help her develop those skills.

Of course, as Ludwig notes, Humeans will maintain that her belief that she has reason to try to win at chess only inasmuch as that would help her develop certain skills must itself simply be an expression of further pro-attitudes that she has. And if, for example, she has a pro-attitude toward winning at chess only because she has a pro-attitude toward developing certain skills and a belief that trying to win at chess would help her develop them, it is indeed quite possible that she will not be disposed to judge that playing against weak opponents would have positive expected utility, even though she believes it would maximize her chances of winning, since she might not believe it would help her develop those skills.

But would a pro-attitude toward developing those skills dispose this person to find positive expected utility in any course of action she believed would help her develop them? It seems to me quite possible that it would not; so, our original worry would re-emerge. And, while Humeans could again respond by suggesting that she has this pro-attitude only because she has some other, I don't see how they could outrun the worry forever. Sooner or later, they will have to confront the fact that their account is at odds with the appearances. Whereas it appears to be the case that people's pro-attitudes are standardly sensitive to their judgements about their reasons, even the most sophisticated of Humeans must deny this.

So, I disagree with Ludwig that questions about the proper formulation of the Humean account are what mainly divide us.Footnote ⁵ I believe the main point of contention is whether we should be looking for an account of pro-attitudes that takes their apparent judgement-sensitivity at face value, or whether we should be looking to explain away those appearances compatibly with the Humean approach. I agree with Ludwig that, if we are to take the revisionary line, we will do well to adopt the most sophisticated version of Humeanism we can come up with. But that is not the line I am inclined to take, so a simpler version of Humeanism seemed sufficient for my purposes in the book.

If pressed to explain why I'm disinclined to take the revisionary line, there is evidently the danger of arguing in a circle. After all, if we are trying to establish that normative beliefs and pro-attitudes have contents of a sort that might ultimately need explaining via triangulation, our argument in favour of adopting a non-Humean account cannot simply be that it secures that result. But by the same token, of course, the argument against adopting a non-Humean line cannot simply be that it avoids that result. So, how are we to determine whether we should be taking a revisionary or a non-revisionary line on questions about the nature of normative beliefs and pro-attitudes?

My thought in the book was at least to make the circle here as broad and as welcoming as possible by arguing, not just that adopting a non-Humean account of pro-attitudes and normative beliefs enables Davidson's triangulation argument to be applied to them, but also that applying Davidson's triangulation argument to pro-attitudes and normative beliefs enables us to defend attractive versions of meta-normative realism and meta-ethical contractualism.Footnote ⁶ By contrast, if we insist on sticking with a Humean account of pro-attitudes and normative beliefs, we end up having to take revisionary lines on meta-normative and meta-ethical issues as well, no matter how sophisticated our Humeanism may be. This may not trouble the more zealous of the Humeans; they may even challenge the insinuation that the meta-normative and meta-ethical appearances are against them. But my hope was that many other people, and perhaps even some of the less zealous of the Humeans, would be encouraged by these results to take non-Humean accounts of pro-attitudes and normative beliefs seriously.

Ludwig doesn't take up these broader questions, turning instead to the exegetical question of whether Davidson himself actually adopted a non-Humean account of pro-attitudes and normative beliefs. And it must be acknowledged that Davidson occasionally made claims that seem to tell against this, especially in the early years. Perhaps the most problematic of these is one Ludwig quotes at length, from “Intending,” where Davidson first conceded that “there is no short proof that evaluative sentences express desires and other pro attitudes in the way that the sentence ‘Snow is white’ expresses the belief that snow is white” but nonetheless maintained that “all pro attitudes may be expressed by value judgements that are at least implicit.”Footnote ⁷ As Ludwig notes, claims like these would appear to pose a challenge to me; either I must explain how they can be understood compatibly with a non-Humean account, or I must demonstrate that Davidson changed his mind about the nature of pro-attitudes and normative beliefs at some point after 1978.Footnote ⁸

Even in earlier papers, however, Davidson was also making claims suggesting that he did advocate a non-Humean account. For example, in “How Is Weakness of the Will Possible?,” he claimed that “a judgement of value must be reflected in wants (or desires or motives).”Footnote ⁹ Ludwig says that he finds it very natural to read this paper in light of the Humean approach, but it seems to me that Davidson's choice of “reflected in” rather than ‘expressions of’ points in the opposite direction. Indeed, the whole paper seems to me to presuppose a sort of objectivism about normative judgements that is extremely difficult to square with Humeanism.

Perhaps even more telling is Davidson's claim, in “Mental Events,” that, when interpreting another person, we should “try for a theory that finds him consistent, a believer of truths, and a lover of the good (all by our own lights, it goes without saying).”Footnote ¹⁰ The parenthetical qualification here mustn't be understood as somehow compromising Davidson's commitment to objectivity, either about descriptive matters or about normative ones. The fact that we must each interpret others “by our own lights” doesn't alter the fact that, in forming their beliefs and pro-attitudes, people are always aiming, in some important sense, to get things right.

So, I don't feel any need either to fight over the early Davidson's every claim about the nature of pro-attitudes and normative beliefs or to demonstrate that the later Davidson underwent a radical change of mind. I agree that there was an important change over time, but I think it was from a less secure grip on the inadequacies of Humean accounts to a more secure one.Footnote ¹¹ Be that as it may, however, I don't see any way of denying that, by one route or the other, he did end up clearly rejecting Humean accounts. How else could he claim, as he did, for example, in “The Objectivity of Values,” that the contents of people's normative beliefs are in the first instance fixed through their triangulations with one another and “attitude-causing properties” of their circumstances?Footnote ¹²

As Ludwig suggests, this could conceivably be the result of a confusion. Perhaps Davidson did hold steadfastly to the Humean view that people's normative beliefs are little more than expressions of their pro-attitudes but somehow still concluded that the contents of people's pro-attitudes are in the first instance fixed through their triangulations with one another and attitude-causing properties of their circumstances. Surely, however, we can't attribute a mistake of this order to Davidson without offering some explanation of why he made it. What might a Humean say about the manner in which the contents of pro-attitudes are fixed that could have led Davidson to suppose triangulation was involved?

Ludwig does not tell us what he thinks Humeans should say about the manner in which the contents of pro-attitudes are fixed, so it's hard to know how he would explain this alleged mistake. Other Humeans, however, have made suggestions that might plausibly be thought to have a bearing on this question. For example, Sharon Street has argued that basic pro-attitudes are often adaptive links of common beliefs and useful behaviours forged through familiar evolutionary processes, while Gilbert Harman has argued that the contents of non-basic pro-attitudes are often fixed through a kind of implicit bargaining.Footnote ¹³ The idea of implicit bargaining is especially intriguing in the present context. For it introduces the possibility that Davidson mistook the complex interactions involved in implicit bargaining with those involved in triangulation, and that this is what explains how he could hold that the triangulation argument applies to pro-attitudes and normative beliefs while also holding that the Humean account of those attitudes is correct.

But surely this hypothesis is absurd. After all, the appeal, for Humeans, of the idea of implicit bargaining is precisely that there is no triangle, because there is no third point, so to speak, no independent truth about their circumstances that the bargainers are trying to get right; whereas Davidson's idea of triangulation, by contrast, is precisely that of interlocutors jointly responding to aspects of their circumstances and trying to get them right.Footnote ¹⁴ It is simply not plausible, therefore, to suppose that Davidson might have mistaken the one idea for the other, and I certainly don't mean to insinuate that Ludwig is suggesting this. But, since I can't think of any other way to explain why Davidson might have held that the triangulation argument applies to pro-attitudes and normative beliefs while also holding that the Humean account of those attitudes is correct, I submit that, pace Ludwig, he cannot, in fact, have held that the Humean account is correct. At some point, whether early or late, he must have gone over to the non-Humean side.

But, of course, this is not enough to establish that Davidson adopted the specific version of the non-Humean account I am advocating. So, let us now turn to that.

2. Reply to Hurley

Hurley raises several interrelated objections at this point. He sums them up nicely by saying that Davidson should be read as following Aristotle and G.E.M. Anscombe, not as anticipating Thomas Scanlon and Michael Smith.

As I argue in Chapter 6 of the book, I think Davidson anticipated Scanlon's idea that motivating attitudes are typically judgement-sensitive. However, as Hurley reads Scanlon, the idea is that motivating attitudes are typically judgement-sensitive because they are typically normative beliefs, and not independent pro-attitudes. That is to say, Hurley reads Scanlon as rejecting the Humean theory of motivation, not the Humean account of pro-attitudes. And since Hurley agrees with me that Davidson accepted the Humean theory of motivation, and rejected the Humean account of pro-attitudes, he worries that my appeals to Scanlon are bound to lead my interpretation of Davidson astray.Footnote ¹⁵ The way they do this, he argues, is by forcing me to interpret Davidson on the model of Smith. For if we go on to ask how motivating attitudes could be sensitive to normative judgements if they were never normative beliefs, but were always independent pro-attitudes, the only available answer seems to be a broadly dispositional one. As I put it earlier, a pro-attitude toward winning at chess might typically dispose one to act only in ways one believes will speak to the reasons one believes one has for trying to win at chess in the first place because pro-attitudes aim, as a system, to get normative matters right.Footnote ¹⁶

Hurley's worry about this is that it still follows Scanlon too closely in holding that pro-attitudes are constitutively independent of normative beliefs, even as it departs from him to some degree by allowing that they might be systemically sensitive to them. Hurley insists that a properly non-Humean account of pro-attitudes should instead allow that pro-attitudes are not merely systemically sensitive to normative beliefs but actually partially constituted by them. And he suggests that this is what Davidson must have meant to convey when, in “Intending,” he said that “all pro attitudes may be expressed by value judgements that are at least implicit.”Footnote ¹⁷ As Hurley reads Davidson, his claim in this passage was that value judgements are at least implicit in all pro-attitudes because all pro-attitudes commit their bearers to the desirability of their contents. Unlike normative beliefs, pro-attitudes also dispose their bearers to act in ways they believe will speak to the reasons they believe make those contents desirable. So, Davidson's claim, according to Hurley, was not that pro-attitudes are fully identical to normative beliefs but simply that they are partially constituted by them. Thus, he was, indeed, still accepting the Humean theory of motivation, even as he rejected the Humean account of pro-attitudes.

As Hurley notes, this interpretation of Davidson makes straightforward sense of his oft-repeated claim that people's pro-attitudes supply the evaluative judgements that serve as major premises in the practical syllogisms articulating their practical deliberations. As he further notes, it also makes straightforward sense of Davidson's claim, in “The Problem of Objectivity,” that the contents of people's pro-attitudes are dependent on one another.Footnote ¹⁸ Given that the contents of people's normative beliefs are dependent on one another, the fact that people's pro-attitudes are partially constituted by their normative beliefs ensures that their contents are dependent on one another as well. By contrast, on my interpretation of Davidson, the contents of people's pro-attitudes are not strictly speaking dependent on one another; they only appear to be because the contents of the normative beliefs to which they are sensitive are dependent on one another.Footnote ¹⁹

However, even on Hurley's interpretation of Davidson, the contents of people's pro-attitudes are dependent on one another only because the contents of their constituent beliefs are dependent on one another. Thus, the “intra-attitudinal” holism that Davidson attributed to pro-attitudes still has an “inter-attitudinal” component; Hurley's Davidson doesn't differ so much from mine in this respect. Likewise, even on my interpretation of Davidson, there is a sense in which pro-attitudes formed entirely against one's better judgement still commit one to the desirability of their contents, and thus still supply evaluative judgements that can serve as major premises in the practical syllogisms articulating one's practical deliberations, since their systemic aim, however badly realized on these occasions, is to get normative matters right. Davidson did hold that intentional behaviour can be essentially surd; my Davidson just takes this a little further than Hurley's does.Footnote ²⁰

So, neither of these points is dispositive. And what of Davidson's claim, in “Actions, Reasons, and Causes,” that “a man may all his life have a yen, say, to drink a can of paint, without ever, even at the moment he yields, believing it would be worth doing?”Footnote ²¹ Wasn't this to say that someone could have such a yen and yet not think acting on it would be worth doing at all?Footnote ²² Hurley might insist that the early Davidson was unclear about the full extent of his opposition to Humean accounts of pro-attitudes, and that it was only later that he stopped counting states of this sort as examples of such things. But I can't think of any place where the later Davidson actually took this step. On the contrary, I think he was always most naturally read as holding that pro-attitudes are modally separable from normative beliefs, not just in the sense that one can have a normative belief without any corresponding pro-attitude, but also in the sense that one can have a pro-attitude without any constituent normative belief.

Still, I am prepared to concede that there remains room for debate here on both the substance and the exegesis. While I remain convinced both that pro-attitudes are best understood as being systemically sensitive to normative beliefs, rather than as being constitutively dependent on them, and that Davidson understood them in this way, I acknowledge that more work must be done in order to make these claims stick. Rather than pursue these issues any further here, however, I want now to take up a different, though related, point of contention. It again has both a substantive and exegetical dimension. The exegetical question is whether I am right in concluding that Davidson embraced a kind of realism — indeed, even a kind of naturalism — about normative properties; the substantive question is whether the kind of meta-normative naturalism I end up attributing to Davidson is as compelling — or, in fact, even as coherent — as I argue it is.

Hurley thinks I clearly get the exegesis here wrong, citing Davidson's remark, in “Objectivity and Practical Reason,” that objectivism about values “should not be confused with realism.”Footnote ²³ As the ensuing paragraphs made clear, however, what Davidson was worried about in that paper was a tendency among philosophers to construe values as non-abstract objects, which then must be located somewhere, rather than as properties, which, as he put it, are nowhere; and in calling Davidson a realist I certainly am not suggesting that he was actually guilty of making this mistake himself. My suggestion in the book is simply that we can and must take seriously his claim, in “The Objectivity of Values,” that the evaluative properties of things are “attitude-causing.”Footnote ²⁴ We can take this claim seriously because Davidson's anomalous monism shows how a non-reductive and non-revisionary form of meta-normative naturalism (we might call it ‘anomalous naturalism’) is conceivable; we must take this claim seriously because, without it, we cannot make sense of Davidson's claim that the triangulation argument applies, not just to the contents of descriptive beliefs, but also to the contents of normative beliefs and pro-attitudes.Footnote ²⁵ Let me say a little more about each of these two points, starting with the second.

Hurley insists that the applicability of the triangulation argument to pro-attitudes is secured by the fact that they are partially constituted by normative beliefs. My desire to improve the taste of the stew is partially constituted by my belief that improving the taste of the stew would be desirable, and such judgements, Hurley says, “essentially involve triangulation among the world, my desires, and the desires of other agents.” As we saw earlier, however, the contents of people's pro-attitudes could be determined by various sorts of back-and-forths among them without being determined through triangulation in the sense Davidson intended. Triangulation in Davidson's sense requires a third point, as I put it, something for triangulators to triangulate on. And now we must add that the third point cannot simply be “the world” but must rather be some causal property of objects or events corresponding to the content being determined. As Claudine Verheggen explains in Part One of the book, Davidson's externalism about content, although interpersonal, is still causal; the necessity for people to interact with one another in order to fix the contents of their most basic attitudes, and the possibility that different people's most basic attitudes could end up with somewhat different contents, doesn't mean that causes have dropped out of the account.

For better or worse, then, there is no way to apply Davidson's triangulation argument to the contents of pro-attitudes and normative beliefs, as he wanted to, without allowing that normative properties are causal properties, as he actually did. Given how he understood causality, this in turn required him to hold that the normative properties of objects and events supervene on and are determined by the physical properties of those objects and events. So, it ultimately was a kind of meta-normative naturalism that Davidson was advocating. And since meta-normative naturalism is widely assumed either to be an unacceptably revisionary position or to be a hopelessly incoherent one, it is perhaps understandable that Hurley is so reluctant to interpret Davidson as embracing it. As I argue in the book, however, I think Davidson had the resources with which to construct credible defences against such concerns, although he unfortunately left much of the work here to us.

I won't reiterate those arguments here.Footnote ²⁶ And I don't deny that the concerns are daunting. It is difficult to convince oneself (let alone others!) that a property could be determined by physical properties in the way all causal properties must be, and that it could exercise an influence on people in the way attitude-causing properties would do, while being truly and fully normative. The non-reductive character of anomalous naturalism is not so troubling; the difficulty is to prevent it from becoming too revisionary. But the same is true of anomalous monism in the philosophy of mind, and of the causal theory of action, yet no one denies that those views were really Davidson's, or that we should take their substantive merits seriously. If the question before us here, therefore, is whether anomalous naturalism, with its commitment to attitude-causing normative properties, is too implausible to attribute to Davidson, or to take seriously ourselves, then I think the answer is clearly that it's not.

And it is important not to focus on the concerns about anomalous naturalism to the exclusion of its promise. For what exactly are the alternatives? Non-naturalists like Scanlon lack any theory of normative content and thus struggle to provide a credible epistemology for normative beliefs. Scanlon models his epistemology on John Rawls's method of reflective equilibrium, but because he lacks a theory of normative content he has no credible way to limit the extent to which people's considered judgements about normative matters might differ and so no credible way to explain why the method of reflective equilibrium should be thought to promote convergence on the normative truths.Footnote ²⁷ Davidson, by contrast, is potentially in a much stronger position here, for if the contents of people's normative beliefs are in the first instance fixed through their triangulations on attitude-causing normative properties, then their considered judgements about normative matters will be presumptively veridical.

Meta-normative constructivists sometimes do possess theories of normative content (recall Street's appeals to the notion of an adaptive link, and Harman's appeals to the notion of implicit bargaining), but these theories invariably threaten us with an outlandish kind of relativism, according to which people might lack not just agent-neutral reasons but even agent-relative reasons to respect and accommodate their agent-relative differences. As the worry here is sometimes put, meta-normative constructivism threatens to disavow the public character of normative reasons.Footnote ²⁸ Kantian constructivists try to avoid doing this by imposing more stringent requirements on the constructive procedure, but they notoriously struggle to come up with plausible requirements capable of securing that result. If our goal is to acknowledge the public character of normative reasons, as I think it surely should be, Davidson's anomalous naturalism again strikes me as a better bet.

Hurley does not tell us what he thinks a properly Aristotelian or Anscombean meta-normative view would look like, or why he thinks it would be preferable to the anomalous naturalism I attribute to Davidson. I think it's safe to assume, however, that we are fighting over the same turf. For I take it that Aristotelians and Anscombeans are also hoping to defend a non-reductive and non-revisionary form of naturalism that avoids the problems bedevilling non-naturalism and constructivism. The question is how they can do this without treating normative properties as attitude-causing properties. Is there one life form common to all possible agents?Footnote ²⁹ Can we know what it is? Does it generate normative reasons that are suitably public in character? If Hurley can answer all these questions in the affirmative, he'll have produced a serious rival to anomalous naturalism and we'll need to talk. I'm sceptical, but at the same time happy to grant that I could be wrong; we'll have to see. What I cannot grant is that the resulting view might be Davidson's. Just as we cannot deny that Davidson held a causal theory of action, even though we might disagree about its precise form and ultimate merits, so too we cannot deny that Davidson held a causal account of normative properties, even though we might disagree about its precise form and ultimate merits. Causalism is at the very heart of Davidson's philosophy.Footnote ³⁰

What, finally, of Hurley's complaint that Davidson's meta-ethics was more likely to have been virtue ethical than contractualist? Here, I am willing to be much more open-minded about the exegesis. As I remark at several points in the book, while Davidson had a great deal to say about normative reasons generally, he had next to nothing to say about specifically moral reasons. So, any suggestions about where he might have been headed on this issue can only be highly speculative. And, as I note in Chapter 8, Section 2, the kind of contractualism I think he was headed toward in effect identifies moral principles as those no virtuous person could reasonably reject, where a virtuous person is both someone who is concerned to interact with others on terms of mutual respect — or, as I sometimes also put it, in a manner befitting people intent on cooperating with one another in the promotion of the overall good — and someone who understands how such concerns put limits on what she can reasonably reject. It could, therefore, be that the differences between my speculations about Davidsons’ meta-ethics and Hurley's don't run very deep. That's a question we won't be able to take up further until Hurley tells us more. But it's certainly a good question, and one to which I hope we'll be able to return very soon.