We are pleased to present the latest installment of PEA Soup's collaboration with Ethics, in which we host a discussion of one article from an issue of the journal. The article selected from Volume 121, Issue 1 is Gunnar Björnsson and Stephen Finlay, "Metaethical Contextualism Defended." We are very pleased that Ralph Wedgwood is providing a précis of the article to introduce the discussion. His commentary begins below the fold.
In their paper, “Metaethical Contextualism Defended”, Gunnar Björnsson and Stephen Finlay defend a contextualist account of normative judgments – that is, of the kind of judgments that we express by making statements what people ought to do, or the like. The account defended by Björnsson and Finlay (henceforth, “B & F”) is contextualist in two ways: first, it interprets normative judgments as implicitly indexed or relativized to bodies of information; secondly, it interprets normative judgments as implicitly relativized to ends or standards.
The first sort of contextualism seems eminently plausible to me. E.g. imagine a speaker on the top of a tower, keeping track of someone on the ground below who is making his way through a maze. It seems that the speaker might quite naturally say both:
(1) He has no way of knowing it, but he ought to turn left at this point
(2) Since his evidence all suggests that turning left would be the wrong thing to do, he ought not to turn left at this point.
We seem to be able to hear both statements as true in this situation, and the easiest way to explain this is to suppose that these different occurrences of ‘ought’ are implicitly indexed to different bodies of information.
The second sort of contextualism, on the other hand, seems much more controversial. Admittedly, it is plausible that there is an “end-relative” use of ‘ought’. In an unpublished lecture Bernard Williams used the following example:
(3) He ought to be using a Phillips screwdriver to open that safe.
A speaker might quite compatibly make this statement, while at the same time claiming that – in all other salient senses of the term – the agent in question ought not to be trying to open the safe at all. So, this occurrence of ‘ought’ seems to be end-relative – relativized to the end of opening the safe.
Most philosophers, however, would deny that all occurrences of ‘ought’ are end-relative in this way: as Kant would put it, some occurrences of ‘ought’ are categorical rather than hypothetical. The sort of contextualism defended by B & F, on the other hand, implies that all occurrences of ‘ought’ are end-relative or hypothetical in this way.
As B & F rightly see, both forms of contextualism have prima facie difficulties with explaining how different speakers' statements are logically related to each other. The first sort of contextualism (which postulates informative-relativity) faces a prima facie problem that is due to some unpublished work by Niko Kolodny and John MacFarlane.
Suppose that Agent and Adviser are deliberating together about which of four available options – A, B, C, and D – Agent should choose. Both Agent and Adviser know for certain that the objectively best option is either A or B (although unfortunately they do not know which of the two it is), while C and D are certainly no better than second-best. Even though neither Agent nor Adviser has perfect information, Adviser’s information is better than Agent’s. In this case, Agent might ask Adviser what he ought to do, tentatively suggesting that he ought to do C. Adviser might respond “No, you’re wrong! It’s not true that you ought to do C – you ought to do D instead!” Intuitively, it seems, Adviser’s statement could be perfectly true – even if C is what Agent ought to do relative to the body of information that Agent had at the time when he asked his original question.
As Kolodny and MacFarlane point out, the contextualist has a prima facie problem explaining this. Neither Agent’s statement nor Adviser’s statement can involve an objective ‘ought’, relativized to the body of information that contains all relevant truths whatsoever, because they both know that what Agent ought objectively to do is either A or B (and not either C or D). So, perhaps, for example, Agent’s statement is implicitly indexed to the information that he possesses, while Adviser’s statement is indexed to the information that she possesses. However, if the information to which Adviser’s statement is indexed is different from the information to which Agent’s statement is indexed, then Adviser’s statement does not really answer the question that Agent originally asked. Moreover, Adviser’s use of the language of disagreement (“No, you’re wrong!” and the like) is out of place, since strictly speaking the two are speaking past each other.
In response to this objection, B & F first sketch a form of contextualism, “news-sensitive contextualism” (p. 14), which allows both the agent and the adviser to be evaluating the very same proposition. According to news-sensitive contextualism, the information that a statement is indexed to need not be the information that is actually possessed by any of the speakers at the time of utterance; instead, it may be the total body of information that the deliberating agent “will or can acquire by the time by the time he must decide what to do.” This body of information includes not only the information that the agent already possesses, but also all the information that any adviser will give him before he has to decide what to do. So, according to news-sensitive contextualism, the statements of both the Agent and the Adviser can be indexed to the very same body of information.
However, as B & F correctly point out, this sort of “news-sensitive contextualism” cannot be the whole story. We also need to explain why the adviser chooses to introduce new information into the relevant body of information at all.
The key point, B & F claim, is that the aim of advice is not just to enable the advisee to accept true ‘ought’-propositions. The point of advice is to help the advisee with whatever the advisee is trying to do. This is why it is typically helpful for advisers to give their advisees better information. As B & F put it, deliberating agents are aiming to promote certain values; an agent typically prefers fuller information “because it puts him in a better position to promote his values” (p. 16).
In fact, I would quibble with B & F’s claim that the goal of deliberating agents is always to “promote certain values”; but in the interests of brevity I shall not raise these quibbles here. What matters most for present purposes is the point that almost all deliberating agents have an interest in fuller and better information, and this point seems clearly correct.
As B & F rightly point out, however, it is still necessary to explain why it is so natural to use the language of disagreement, even when we are assessing other speakers’ statements that are strictly speaking indexed to bodies of information that are different from the information that we ourselves possess. On this point, they make some very perceptive observations (pp. 19–20) about how the proposition picked out by the demonstrative ‘That’ in statements like ‘That’s not true’ is often not exactly the same as the proposition that the previous speaker has asserted, but the most conversationally salient proposition instead.
Suppose that a speaker S makes a statement of the form ‘x ought to do A’, about the situation of some agent x at some time t. Suppose, moreover, that you possess better or fuller information than the speaker S about the situation that x is in at t. Given that deliberating agents typically have an interest in better or fuller information, it seems easy to explain why the proposition that is most conversationally salient to you is not the proposition that S has asserted, but the proposition that you could make using those words, indexed to the better, fuller body of information that you possess. This is why it would be natural for you to use the language of disagreement, by saying things like “No, S is wrong”, “What S is saying is not true”, and the like.
Overall, the defence of information-relativity that B & F give in the first half of their paper seems to me highly plausible. In the second half of their paper, they argue that essentially the same manoeuvres can be redeployed to defend standard-relativity against the famous objection that standard-relativity implies that speakers who adhere to different standards are not genuinely disagreeing with each other. Unfortunately, it seems to me that B & F’s arguments in the second half of the paper are less successful than the arguments that they gave in the first half.
As we have seen, B & F are defending the strong claim that all instances of ‘ought’ are implicitly relativized to ends or standards. Most other philosophers, as I have already noted in connection with the example (3) above, would be happy to concede that some occurrences of ‘ought’ are relativized to some end or standard. However, the intuition that there are genuine disagreements between the adherents of different standards does not concern the overtly end-relative or hypothetical ‘ought’-statements. They concern instances of the apparently categorical ‘ought’. After all, no one ever thought that there was any sort of disagreement between:
(4) To get to London, you should catch the train from Platform 5
(5) To get to Bristol, it is not the case that you should the train from Platform 5.
So to explain away the appearance of disagreement in a satisfactory way, it seems that B & F would have to focus on some special features of the apparently categorical ‘ought’.
In the second half of their paper, when they are discussing standard-relativity, B & F do not develop any analogue of “news-sensitive contextualism”. That is, they do not try to find any subtle set of standards such that both of the apparently disagreeing statements are in fact implicitly relativized to the same set of standards.
Instead, they rely on the claim that in the conversations that seem intuitively to involve a disagreement about what someone (in some apparently “categorical” sense) ought to do, the proposition that is most salient to each speaker in those conversations is the proposition that is implicitly indexed to the standards that he himself endorses. B & F call this feature of normative discourse “quasi-expressivist” (p. 22). But it is not clear how their sort of standard-relativity can explain why normative discourse has this feature. (To be fair, Finlay has defended this claim in more detail in some of the work that he has published elsewhere; but in this paper the point seems to be asserted rather than justified.)
At all events, the arguments of the first half of the paper crucially relied on the fact that deliberating agents have a general preference for better and fuller information. After all, we typically would not voice a disagreement with another speaker about what an agent x ought to do at a time t if we took ourselves to have an inferior body of information about x’s situation at t than that other speaker. It is not clear that what can play this role in the second half of the paper. So it is not clear that the defence that they offer for information-relativity carries over quite so readily to standard-relativity.