Chapter 7: Truth

In Chapter 7, Ridge takes on the challenge of accommodating the notion of normative truth within the overall programme of Ecumenical Expressivism (EE). His main aim in doing so is to provide a workable alternative to what could currently be seen as the standard expressivist approach to truth. Popularised primarily by Blackburn, this standard approach is deflationary, taking truth-aptness as a mere matter of declarative form and the truth-predicate as a mere device for disquotation. Ridge, however, is concerned that a commitment to deflationism would leave too many hostages to fortune, considering the numerous objections to which this theory of truth has been subjected more broadly within the philosophy of language. For this reason, he devises a conceptual machinery meant to enable the ecumenical expressivist to take on board any other plausible theory of truth (his focus in the chapter is on correspondence) and to modify it in a principled way so that it fits snuggly within EE. There are lots of interesting ideas and insights in this chapter, all worthy of much more attention than I can give them here. In the interest of brevity I’ll focus on what I take to be the main thread of Ridge’s argument, leaving aside some of his subsidiary moves. I’ll raise a number of critical issues and questions along the way, though I confess that on the whole I find myself more often agreeing than disagreeing with Ridge.

Expressivism, Truth-Aptness, Deflationism

For a few decades now, expressivists have taken a deflationary approach to truth. Deflationism works on two levels. First, there’s deflationism about truth-aptness, according to which truth-aptness is simply a matter of grammar: if statements in a certain domain of discourse have the declarative form, then they’re truth-apt. Normative statements are obviously declarative, so they qualify. Then, there’s deflationism about the truth-predicate. On this view, ‘is true’ is merely a device for disquotation, whose meaning is given by instances of the famous schema:

     (T) ‘p’ is true if and only if p.

In other words, the truth-predicate adds nothing to the semantic content of the proposition to which it is ascribed. When I say that it is true that charity is good, I say nothing more than that charity is good. So whenever I’m entitled to make the latter claim I am also entitled to make the former. Deflationism thus gives us truth “on the cheap”.

Ridge rejects both these kinds of deflationism. With respect to truth-aptness, he subscribes instead to minimalism, which adds the idea of discipline to the deflationist’s insistence on declarative form. And since the work completed in previous chapters of the book amounts to a cumulative argument to the effect that normative discourse is subject to the appropriate kind of discipline, Ridge takes himself, rightly in my view, to have already established that normative discourse is truth-apt. 

With respect to deflationism about the truth-predicate, Ridge is equally critical. Although he doesn’t formulate a decisive objection to it, in §2 he lists a number of problems that this type of deflationism faces more widely within the philosophy of language: its incompatibility with the idea of truth-value gaps, the formal problems arising out of the incompleteness results, and, more importantly, the difficulty of accounting for the normativity of meaning and truth. According to Ridge, these problems are enough to show that deflationism is on philosophically shaky ground - shaky enough, that is, for the expressivist not to want to pin all her hopes on it, but instead to be looking elsewhere for a sort of “insurance policy”.

Ridge’s Account of Truth in EE     

Ridge couches his alternative proposal in terms of the correspondence theory, which he chooses because of its friendliness to representationalism. Here’s my reconstruction of the main building blocks of Ridge’s account:

1. As the objects of propositional attitudes like belief and desire, propositions are the primary truth-bearers. (p. 204) 
2. A normative proposition does not fix a single representational content - rather, it expresses a hybrid state (a relation between a normative perspective and a representational belief) and is thus multiply realisable: the representational belief that partly constitutes it can take any of an indefinite range of contents. (p. 205) 
3. From (A) and (B) it follows that the business of representing the world is done not by propositions, but by their representational contents. Token normative propositions represent only indirectly, via the contents that are assigned to them by the normative perspectives that partly constitute them (ibid.).
4. From (C) it follows that the same kind of indirectness must be replicated in our theory of truth for propositions. Specifically, the truth of a proposition must be interpreted in terms of the truth of its associated representational contents.
5. Hence, we get the following indirect account: using ‘p’ to denote an arbitrary proposition, a claim of the form ‘p is true’ means that on any acceptable construal, the representational content of p corresponds to reality. (p. 208)
6. Since the clause on the right-hand side of the meaning equivalence in (E) deploys a normative concept (‘acceptable construal’), it follows that ascriptions of truth are essentially normative judgments.
7. Given EE’s account of normative judgments, it follows from (F) that an ascription of truth of the form ‘p is true’ expresses a hybrid state made up of: (i) a normative state; (ii) the belief that on any of the admissible specifications of the representational content of p, that content corresponds to reality (p. 209).   

Ridge packs a lot in here - much of which, I think, is on the right track. In what follows, I want to express (excuse the pun!) two cheers and two associated fears about his proposal.

Here’s my first cheer. The view spelled out in claims (F) and (G) above, namely that truth-ascriptions are normative, has a lot going for it. I am independently persuaded that truth is an essentially normative concept, mainly due to an argument put forward by Huw Price in his 2003 paper, “Truth as Convenient Fiction”. Moreover, it is a virtue of Ridge’s account that he doesn’t restrict his approach only to ascriptions of truth taking normative judgments as their objects - rather, he states that the approach “generalises to arbitrarily complex claims and thoughts in which ‘true’ figures” (p. 211). This, for instance, enables him to say that both ‘It is true that charity is good’ and ‘It is true that grass is green’ are normative judgments, whereas the deflationist could only say that about the former. To my mind, this gives Ridge an advantage over deflationism, though it’s not an advantage that he explicitly claims for himself. On the other hand, one advantage that Ridge does claim for his view is that it can fend off a counterexample to Hume’s Law (the injunction against deriving ‘ought’ from ‘is’). The counterexample is due to Mark Nelson:

     (1) The Pope believes that charity is good.
     (2) All the Pope’s beliefs about charity are true.

     Therefore, (3) Charity is good.

On a standard understanding of truth-ascriptions, (2) is descriptive. Since (1) is also descriptive but (3) is normative, the inference is supposed to have derived an ‘ought’ purely from a set of ‘is’ premises. However, Ridge rightly points out that on his view, (2) counts instead as a normative judgment, and therefore the inference from (1) & (2) to (3) already contains an ‘ought’ hidden in the premises.

But here comes my related worry. Ridge explicitly commits himself on p. 210 and elsewhere to the “transparency” of truth, namely to the claim that the following two entailments are valid: ‘p, therefore it is true that p’, and ‘it is true that p, therefore p’. Now take the first of these entailments. If that’s valid, then the following instance of it is valid, too: 

     (4) Grass is green.
     Therefore, (5) It is true that grass is green. 

Now, (4) is a descriptive claim, whereas (5) is a normative one, at least if Ridge is right about truth-ascriptions. Thus, we have an ‘is’ that directly entails an ‘ought’ - a fine counterexample to Hume’s Law! However, Ridge states that he finds Hume’s Law independently plausible (p. 218). So something has to give here. For my part, I would be inclined to give up on Hume’s Law, given that I have a stronger commitment to the normativity of truth. But I would also be curious to hear what Ridge thinks about this, and whether or not his own inclination is the same as mine.

Be that as it may, here is my second cheer. As Ridge himself notes (p. 207), his proposal that propositional truth be interpreted as the truth of associated representational contents on all acceptable construals (as per claim (E) above) is structurally similar to the supervaluationist account, on which a sentence is true just in case it is super-true, i.e. true on all admissible precisifications. Supervaluationism is a theory of truth designed to account for propositions containing vague predicates, and in this connection Ridge suggests at one point (particularly in footnote 15 on p. 207) that normative predicates are maximally vague - having determinate non-instances but no determinate instances. As someone who has been thinking long and hard about the vagueness of moral predicates, all of this is music to my ears! 

But I also have a related worry. When glossing the notion of maximal vagueness, Ridge says that normative predicates are maximally vague in the sense that there are no cases to which they apply as a matter of sheer semantic and conceptual competence. But this, he adds, doesn’t entail that there are therefore no determinate normative truths. Instead, according to him “there are many determinately true and non-trivial normative propositions. The point is that which ones are determinately true is something which competent speakers as such can disagree about even if they agree on all the relevant non-normative facts” (p. 207, n. 15). It seems to me that this is one of the strongest claims Ridge makes about truth in this chapter. Effectively, he’s saying that knowledge of normative truths is entailed neither by factual knowledge of non-normative truths, nor by knowledge of semantics. And of course, as a quasi-realist committed to mimicking realist talk about the normative domain, this is exactly what he should be saying. For it is indeed a staple of realism that normative truths are epistemically unconstrained in the ways specified above, as well as in other ways. However, my question is this: how is Ridge going to interpret such claims of epistemic unconstrainedness with respect to normative truth in a quasi-realist fashion, without taking on the full-blown commitments of robust realists? The old deflationist has a ready-made solution to this - she can simply adopt a deflationary view of such claims of epistemic unconstrainedness as:

     (7) If you thought kicking dogs was permissible, you’d be wrong.

     (8) If you knew all the facts about dog pain but somehow thought that kicking dogs was

            permissible, you’d be wrong. 

     (9) Kicking dogs is wrong even if everybody else approves of it. 

   (10) Even if we could never know what to make of it, kicking dogs would still be wrong.

    (11) Even if we all approved of it, kicking dogs would still be wrong.

According to the deflationary expressivist (e.g. Blackburn), statements like (7)-(11) all have essentially the same function: they express our first-order con-attitude towards kicking dogs, as well as various second-order attitudes about that first-order attitude. In (7), for instance, we’re expressing our first order attitude together with a second-order con-attitude to having a pro-attitude towards kicking dogs. In (11), we’re expressing our first-order attitude again, this time together with a second-order con-attitude towards changing our mind about that first-order attitude. And so on. The point is that by providing these deflationary readings, the quasi-realist appears to make good on the promise of mimicking realist talk without taking on the metaphysical baggage of realism. He does so by refusing to hear (7)-(11) as anything other than so many more normative claims, made from within our normative perspectives.

But how can Ridge hope to perform the same trick once he’s rid himself of deflationism? Read in the manner encouraged by a correspondence theory of truth, each of (7)-(11) suggests a commitment to a normative reality making some of our judgements true and others false. At this stage, it may seem as though Ridge has dropped the ‘quasi’ qualification and has opted instead for full-blown realism.

One way for Ridge to reply to this would be to point out that his eschewal of deflationism about truth does not necessarily mean that he cannot take a deflationary view of other notions. Perhaps he could cherry-pick, combining, for instance, a robust view of truth with a deflationary view of facts. This leads me on to my final worry, which I’ll outline in a separate section.

Correspondence and Deflationist Facts

Here is one way of thinking about this. Given (G) above, a truth-ascription like ‘It is true that charity is good’ expresses a hybrid state constituted by (i) a normative perspective, and (ii) the belief that on any of the admissible specifications of the representational content of ‘Charity is good’, that content corresponds to reality. But according to EE, ‘Charity is good’ expresses a hybrid state constituted by (a) a normative perspective, and (b) the belief that charity would be highly ranked as an end by any acceptable ultimate standard of practical reasoning. Featuring in the representational content of ‘Charity is good’ is a reference to admissible standards of practical reasoning. And since this representational content is supposed to correspond to reality, it must follow that reality contains facts about admissible standards of practical reasoning. These are normative facts. Therefore, normative facts exist.

Unless something’s gone wrong in my reasoning above, this seems to indicate that talk of correspondence in Ridge’s theory will ultimately force him to commit to the existence of at least some normative facts (viz., facts about admissible standards). But isn't this just the starting point of a slippery slope leading all the way to full-blown realism?

Well, one could say this depends on how we interpret talk of facts. If facts are understood realistically, then yes, this is realism. But why could Ridge not take a deflationary understanding of facts, according to which ‘It’s a fact that p’ simply means ‘p’? 

The problem here is that deflationism about facts seems to simply nullify the appeal to the correspondence theory of truth. If Ridge were to say that p is true iff p corresponds to the facts, but then add that it’s a fact that p iff p, then it would seem to me he’d be effectively backtracking. In that case, any talk of “corresponding to the facts” would become idle, and Ridge’s theory of truth would just boil down to deflationism - ‘p’ is true iff p.

Summary of critical points

Let me end this lengthy exposition by summarising my main critical points (such as they are) in the form of three brief questions, to facilitate discussion:

(Q1) Is there a tension between Ridge’s commitment to the normativity of truth, his commitment to the transparency of truth, and his commitment to Hume’s Law? (Recall that combining the first two appears to provide a counterexample to the latter.)

(Q2) How can Ridge interpret claims about the epistemic unconstrainedness of normative truth without collapsing into full-blown realism? 

(Q3) Is the correspondence theory of truth compatible with a deflationist view of normative facts?