Engel on doxastic correctness

In this paper I discuss Pascal Engel’s recent work on doxastic correctness. I raise worries about two elements of his view the role played in it by the distinction between icorrectness and e-correctness, and the construal of doxastic correctness as an ideal of reason. I propose an alternative approach. It is a great privilege to contribute to this festschrift for Pascal Engel, and thereby pay a small tribute to his important, wide-ranging contributions to Philosophy in recent decades. I also offer this piece as a token of gratitude. As well as lively, stimulating, and, on Pascal’s side, erudite philosophical discussion, Pascal has also offered me much generous personal and professional support over the years. In addition, I think that all philosophers in the so-called analytic tradition, and all who endorse the values of clarity, humility and open-mindedness that, at its best, this tradition represents, owe Pascal thanks for his tireless efforts in its defence. My topic here is one on which much of Pascal’s recent work has focused: belief and its norms. In particular, I want to discuss how we should best characterise doxastic correctness. 1. Doxastic Correctness It is a platitude that beliefs can be correct or incorrect. And it is very widely held that the standard of correctness for belief is truth: that is, that beliefs are correct just when true. 1 But what do these claims amount to? Pascal Engel, like many others, holds that doxastic correctness is normative (Engel 2013a, 2013b). On this view, to say that true beliefs are correct is to say that true beliefs satisfy some norm; 1 But see Engel (2004), Smithies (2012). 2 to say that false beliefs are incorrect is to say that they violate this norm. It’s not hard to see why this is a natural view. When we say that someone has an incorrect belief, we do not seem to be merely describing some feature of her, or of her belief--a feature whose presence she could sensibly remain unconcerned by. We are saying that she believes wrongly. This looks normative. It is also natural to think that this normative standard of doxastic correctness not only applies necessarily, but is essential to the attitude of belief. An attitude not assessable as correct or incorrect according to the truth or falsity of its content would thereby fail to count as a belief. I am sympathetic both to the claim that doxastic correctness is normative, and to the claim that being subject to this standard is essential to belief. Here I will focus primarily on the first claim. I do not want to question the truth of the claim, but rather to ask about the normative property that is involved here. What kind of property is it? A deontic property a property in the same family as obligation and permission? Or an evaluative one, like goodness and badness? Or something else? These are questions with respect to which Pascal Engel has done much to advance our understanding. Here I will focus on the view that he expounds in his recent article, ‘Doxastic Correctness’ (Engel 2013b). I will raise worries about two elements of that view the role played in it by the distinction between i-correctness and e-correctness, and the construal of doxastic correctness as an ideal of reason. I will propose an alternative construal of doxastic correctness, that requires rejecting Engel’s claims about i-correctness and e-correctness. 2. I-Correctness and E-Correctness Following Thomson (2008, Ch. VI), Engel distinguishes between ‘internal’ correctness and ‘external’ correctness, or i-correctness and e-correctness. Consider the act of asserting some proposition. This act can be performed correctly in the sense that one utters a grammatical sentence that expresses that proposition, one pronounces all the words in the right way, one speaks sufficiently loudly, and so on. In that case, one’s act of assertion is i-correct. We can also say that one has made a correct assertion in the sense that what one has asserted is true. In that case, one’s assertion is e-correct. Clearly, these two ways of being correct can come apart. One can perform in an exemplary way the act of asserting a proposition that is in fact false, and one can do a very bad job of asserting a proposition that is true. 3 The truth-standard of doxastic correctness seems to correspond to e-correctness. But this, for Engel, raises a worry about whether doxastic correctness is really normative. He writes: is it clear that [e-]correctness is a normative property? The standard for a tune is fixed by a set of notes, the standard for a map is fixed by the similarity between the map and the territory represented, the correct spelling is fixed by a certain pronunciation of the word. These are descriptive properties, not normative ones. ... The normative concept of correctness is distinct from this descriptive one. It concerns the way, or the operation which, an agent has to perform in order to meet the descriptive condition (Engel 2013b, 200). The worry here seems to be that e-correctness for various kinds requires merely the possession of a certain descriptive (i.e. non-normative) property, such as containing certain notes in a certain order. Note that this seems to be true in particular for the kind belief, since truth appears to be a purely descriptive property. 2 On the other hand, i-correctness seems to be a matter of doing what one does more or less well--something that does look normative. On the basis of this and other considerations, Engel concludes that for a kind K to be correct it has to meet both the eand the i-correctness conditions. It would be wrong to reduce correctness to either one of these two dimensions (ibid., 201). The suggestion, then, seems to be that correctness, and doxastic correctness in particular, comprises both eand i-correctness, and that it is only because it includes i-correctness that it is normative. It’s worth noting that this isn’t exactly Thomson’s view of things. She does not talk of some overarching correctness property consisting of the conjunction of e-correctness and i-correctness. Rather, she simply holds that e-correctness and i-correctness are distinct normative properties. Indeed, 2 More precisely, the relevant property for belief is that of having a true propositional content. It’s not uncontroversial that this is descriptive. Both content and truth are famously held by some philosophers to be normative. 4 we might want to say that e-correctness and i-correctness merely correspond to distinct kinds of standard of correctness to which certain things, such as assertions, can be held--an external standard and an internal standard--rather than being different kinds of property. In that case it would be hard to see how one of them could be a normative property and the other not. Of course, one could use ‘correctness’ in a stipulative way to refer to the property that consists in being both e-correct and i-correct. But the question would remain whether this property has any theoretical interest. One worry about Engel’s proposal is that it might entail that no beliefs are correct, since, on the face of it, the notion of i-correctness does not obviously apply to belief at all. Pauline can assert that p better than Pierre asserts it because of her superior grammar or diction, say. But, as Thomson points out, it would be odd to talk of one person believing that p better than another person. Pauline might believe that p more strongly than Pierre does, or with more certainty than him. She might understand better what is involved in its being the case that p. But none of this seems to amount to believing better; believing as such doesn’t seem to be the sort of thing one can do a better or worse job of. After all, as Engel says, believing is “not a performance” (ibid.). This suggests that, in so far as the distinction between e-correctness and i-correctness is in good standing, doxastic correctness is just a species of e-correctness (or, as we might better say, correctness with an e-standard). In response to this sort of worry, Engel suggests that we understand i-correctness for belief as follows: “Believing for bad reasons, or on the basis of insufficient evidence, is poor believing and thus i-incorrect” (ibid.). Thus, doxastic i-correctness is a matter of basing one’s beliefs on appropriate grounds. This ingenious idea takes advantage of the point that, while believing per se is not something one can do more or less well, basing one’s beliefs arguably is. Nonetheless, I am sceptical that this is really a kind of doxastic correctness. Or, at any rate, I am sceptical that this property, together with doxastic e-correctness, forms part of an interesting, overarching correctness property possessed by beliefs. This is for several reasons. 5 First, the appropriateness of the way in which a belief is based is, as Engel acknowledges, in the first instance a property not of the belief itself but of some broader state or activity that includes it. It seems that what we have here is, at best, two things that can be assessed as correct or incorrect, rather than one thing that can be assessed as correct or incorrect along two different dimensions, in such a way that it might be thought of as ‘fully correct’ when correct along both dimensions. This is not to say that the appropriateness of the way in which a belief is based is not a property of the belief at all. It is a relational property of the belief (e-correctness is also relational, of course, but it is not in the same way a property of something else in the first instance). Nor is it to say that we can’t consider whether a belief possesses the conjunctive property of being e-correct and being appropriately based. It is to question whether this conjunctive property is itself a further correctness property possessed by the belief, instead of being a conjunction of two rather different properties. Consider a belief that is appropriately based but turns out to be false. We would not say that this belief is partly, but not fully, correct. We would simply say that it is incorrect, though appropriately based. Second, it seems odd to think of appropriate basing as a form of correctness at all. Perhaps we can talk about correct and incorrect reasoning though I think it’s more accurate to talk about good and bad reasoning but it doesn’t seem natural to talk about the basing relation that comes to be instantiated through such reasoning as itself correct or incorrect. It seems to me closer to properties like rationality, reasonability, justification, and the like. It might be argued that these terms are synonymous with a certain sense of ‘correctness’. I doubt this, but even if so, the previous point suggests that this is not a sense of ‘correctness’ that applies to the belief itself. It is very plausible that a belief can be correct but, say, unreasonable, or vice versa. Third, it is natural to think that what counts as appropriate doxastic basing depends in part on the standard of correctness for belief: it is appropriate to base one’s beliefs on evidence precisely because evidence is connected to truth, and true beliefs are (e-)correct. If that’s right, then the truth-standard of doxastic correctness is more fundamental than the standard for appropriate basing of beliefs. It would be odd, then, to think that the truth-standard, and the standard for appropriate basing, somehow 3 That epistemic norms of appropriate basing and the like are derived from the fundamental standard of true belief is a very widely held, though often implicit, assumption. For examples of its being made explicit see e.g. Bonjour (1985), Wedgwood (2002), Alston (2005), Sosa (2007). 6 come together to form a further, conjunctive property of correctness. Perhaps they do form an interesting conjunctive property, but this would seem to be a different kind of property that incorporates correctness, rather than a further correctness property. I am suggesting that the notion of i-correctness does not apply to belief. That leaves e-correctness. But, as we saw, being e-correct often seems to be simply a matter of possessing some descriptive property whatever property is fixed as the standard of e-correctness for the relevant kind. In the doxastic case it is truth. Are we therefore forced to concede that doxastic correctness is not normative after all? I don’t think we should concede so easily that e-correctness is not normative. We must distinguish between the standard of correctness for a given kind the property in virtue of which something counts as correct of that kind and the property of correctness itself (having argued that the notion of i-correctness has no role in this context, henceforth I will drop the ‘e’ prefix). It would be a mistake to think that we give an account of doxastic correctness simply by citing the standard of correctness for belief. Truth is the property a given belief must have in order to have the further property of being correct. Citing it doesn’t tell you the nature of the further property of correctness that the belief thereby has. To see this, note that this is an instance of a property that other attitudes can also have. For example, you can correctly intend a certain course of action. It’s correct to intend to phone your sister on her birthday, and incorrect to intend to murder everyone you know. The standard for correct intention is not truth. It is tied instead to the choiceworthiness of available courses of action (as well as, perhaps, the likelihood of successful execution). It doesn’t matter for our purposes what exactly it takes for intended action to be choiceworthy. What matters is that correctness is a property in common between correct beliefs and correct intentions, and is therefore not identical to truth. Rather, truth is what makes beliefs correct. Some other property makes intentions correct. This point can be obscured by formulations to the effect that correctness for belief is, or consists in, truth. Once we distinguish between the standard of correctness for a given kind and the property of correctness itself, there is room to suppose that the standard of correctness is non-normative, but that correctness itself is normative. That is, it may be that the standard of correctness for a certain kind can be specified in wholly non-normative terms--for example, the standard might be having a true 7 propositional content--but satisfying that standard gives a thing of that kind the normative property of being correct. In general, it is not obviously problematic to say that something’s having a non-normative property gives it a distinct normative property. In ethics, utilitarians may say that what makes an action right is that it maximises pleasure. This is a non-normative property. Utilitarians need not claim that rightness itself is therefore non-normative. They can say that pleasure-maximisation makes actions right, but deny that the property of rightness is identical to the non-normative property of pleasuremaximisation. Defenders of other ethical theories, like Kantianism and contractualism, can make similar moves. Perhaps none of these views is sustainable, but they are not obviously confused from the start. 4 But, it might be asked, why should being true give beliefs a normative property? Granted, true beliefs tend to have certain benefits--for example, they help us to get what we want--but it is largely a contingent matter whether any particular true belief will lead to such benefits. So one might wonder what the big deal is about true belief. 5 It is at this point that it becomes attractive to think that the nature of belief does some work in explaining the norms governing belief. For it seems that the attitude of belief is, in a certain way, made for truth. When what you believe is true you thereby believe rightly, whether or not this brings any benefits. However, even leaving aside the question of how exactly these norms are explained, it might be objected that my suggestion here ignores the lesson of Engel’s other examples of (e-) correctness: the correctness of a performance of a tune, a map, or a spelling of a word. Suppose you misspell a word. Have you done something wrong? Well, you have spelled the word wrongly. But it’s not clear that this alone is a normative matter in anything other than a deflated sense. After all, you may not have been aiming to spell the word correctly, and may have had no reasons to do so. In that case, it doesn’t seem as though you should necessarily care that you spelled it incorrectly. Doesn’t this show that correctness per se is not normative? 4 Compare Parfit (2011, pp. 368ff.) on what makes actions wrong. 5 See, e.g., Papineau (2013), for these sorts of worries. 6 Compare Rosen (2001), Hattiangadi (2007). 8 It may show that the term ‘correctness’ need not always be used to pick out a normative property. But it does not show that uses of ‘correct’ never pick out a property that is normative. To this point I have been talking about correctness as though this is simply a property in common to all things that can be described as correct, be they spellings, maps, beliefs or intentions. But in fact, as Engel notes (ibid., 200), when something is correct it is always correct qua something or other. Being a correct spelling of a word correct qua spelling of that word is not the same property as being a correct belief correct qua belief. While both of these may be instances of the more general property of correctness qua K (and hence, again, must not be confused with the corresponding standards of correctness, namely conforming to a certain orthographic convention, and having a true propositional content, respectively), it may be that one of them is normative and the other isn’t. The nature of the kind K may make a difference to whether correctness qua K is normative. And, as noted in sec. 1 above, the sense in which false beliefs are wrong or incorrect does seem to be a sense that the believer must care about. So it remains plausible that some kind of normative property is involved here. An objector may persist: this shows at most that doxastic correctness has normative significance, in the same way that bringing about pleasure, according to the utilitarian, has normative significance. That is, whenever it is instantiated, so too is some normative property. It does not show that doxastic correctness is itself a normative property. While I think this objection has some force, we should not overestimate it. For one thing, the impression remains that labelling a belief incorrect has the flavour of a normative claim, in a way that noting that an action fails to maximise pleasure does not. Secondly, the view suggested by the objection raises the question why we bother talking about correctness at all. Why not just say that, when beliefs are true, this gives them whatever further normative property it is they are claimed to have? Correctness seems like an idle middleman, on this view. I take it that those, like Engel, who think of doxastic correctness as normative, do not think of it as merely playing this middleman role between truth and whatever property is doing the real normative work. They think of it as something normative in its own right.