What is ‘global expressivism’ (GE)? In this piece I'll explain what I mean by the term by contrasting GE (as I understand it) with a range of other views – more familiar views, to many readers, in most cases. In other words, I want to explain what GE is by saying what it is not. This indirect approach has some notable neo-pragmatist champions. Recall Dummett's suggestion that ‘we know the meaning of a sentence when we know how to recognize that it has been falsified’,Footnote 1 and the line from Lear that Wittgenstein is said to have had in mind as an epigraph for the Investigations: ‘I'll teach you differences’.
More precisely, I propose to ‘pentangulate’ on GE by saying how it differs from five other positions in the contemporary philosophical landscape. These five views are: (i) the so-called ‘Canberra Plan’; (ii) Moorean non-naturalism and platonism; (iii) ‘relaxed realism’ and quietism; (iv) local expressivism; and (v) response-dependent realism. Imagine that GE sits in the interior of a pentagon, and that I am describing five possible exit routes from this central and (in my view) privileged location. (Some other familiar positions, including fictionalism, error theories, and idealism, will also be mentioned, but won't merit an exit all of their own, in my map of the territory.)
Before we begin, an important note on terminology. The term ‘expressivism’ is often introduced via the proposal that the function of certain claims (or apparent claims) is (i) not to describe some aspect of the world but rather (ii) to express a psychological state (other than a belief) – an affective state, say. (Think of old-fashioned emotivism, for example.) Things then get confusing when we encounter a case in which we want to say (i) but not (ii). Some views of truth hold that claims of the form ‘P is true’ are not ascriptions of some distinctive kind of property, but rather have some other function – perhaps saying what P itself says, but in a usefully different way (one that permits generalisation, say, as in ‘Everything Mary says is true’).
Should we use ‘expressivism’ for these latter views, too, because they share (i); or choose some other term (perhaps ‘pragmatism’), restricting expressivism to the case in which we also have (ii)? This is simply a terminological choice, and the important thing is to recognise that it needs to be made. Claims (i) and (ii) don't necessarily go together, and there isn't an unambiguous ready-made term that allows for that possibility. As just noted, ‘pragmatism’ is an option (one I have used myself in some contextsFootnote 2), but it has its own ambiguities.
At any rate, I stress that what I here call global expressivism is expressivism in the broad sense (so that (ii) is inessential). This seems to me a very natural usage, especially when one has been interested, as I have, in the convergence between rather different uses of the term ‘expressivism’ in the modern Humean tradition associated with writers such as Simon Blackburn, and in what one might call the Hegelian tradition associated with Robert Brandom.Footnote 3 To avoid confusion, however, it is important to realise both that some contemporary writers use ‘expressivism’ in the narrow sense that requires (ii);Footnote 4 and that some writers are coming to use ‘neo-pragmatism’ for what I call expressivism.Footnote 5
1. The Canberra Plan
Now to the first of our differences. The Canberra PlanFootnote 6 (CP) begins with location or placement problems: Where do normativity, meaning, mentality, and other puzzling domains ‘fit’ in the kind of world described by science? Canberra Planners propose to answer questions of this kind with a generalisations of the Ramsey-Carnap-Lewis approach to the meaning of theoretical terms (or the nature of theoretical entities, to put it in material mode).
The proposed solution comes in two steps. At Step 1 we collect the core truths or platitudes about the target entity or property – the entity or property Target, let us say – and conjoin them to form the Ramsey sentence, R(Target). At Step 2 we ask in the world satisfies or makes true the sentence R(Target) – or to what the term ‘Target’ refers. As Haukioja puts it, Step 1 is a matter of ‘a priori analysis of our philosophically interesting everyday concepts and folk theories’; Step 2 of ‘consult[ing] the best scientific (typically, physical) theories to see whether … referents [for the terms so analysed] are to be found in reality’.Footnote 7 Typically, as here, this is understood to mean natural reality, the world described by natural science, but this isn't essential to the method. A non-naturalist could also frame her investigations in these terms.
Accepting Step 1. How does GE differ? So far as I can see, it need have no distinctive objection to Step 1. Other objections may be raised to Step 1 – for example, that it pays insufficient attention either to the analytic–synthetic distinction, or to the grey zone that results from taking seriously Quinean objections to such a distinction. But if anything such objections are likely to trouble GE less than they do Canberra Planners, I think, because they threaten Step 2, which is where GE and CP really differ.
Trivialising Step 2. GE simply denies that Step 2 leads to non-trivial results, in general. Clearly, Target satisfies R(Target), if anything does. But expecting a non-trivial alternative, in general, relies on a non-deflationary reading of the semantic terms such as satisfies, makes true, or refers – a reading that GE rejects.
Elsewhere I have explained this point in terms of Quine's notion of ‘semantic ascent’. Quine insists that the move from ‘Snow is white’ to ‘“Snow is white” is true’ doesn't change the subject – either way, we're just talking about snow. In the latter case it looks as though we are talking about language, but really we are just talking about the material world, just as before. Similarly, as I put it,
[a]sking “What makes it true that snow is white?”, or “What makes ‘Snow is white’ true?”, is just another way of asking what makes snow white – a reasonable question, in this case, but a question to be answered in terms of the physics of ice and light, not in terms of the metaphysics of facts and states of affairs. There is no additional semantic explanandum, and no distinctively metaphysical question.Footnote 8
Similarly, if you are a competent English speaker, familiar with the use of the term, then the question ‘To what does “snow” refer?’ can be construed as an awkward way of asking ‘What is this stuff, snow?’ That's a reasonable question, in this case, but one for natural science. There's no reason to suppose either that it remains a reasonable question in other domains, or that the talk of reference and the like played any substantial role in framing it.
As I note in the same context, Blackburn makes a similar point:
Blackburn notes that on Ramsey's view, the move from ‘P’ to ‘It is true that P’ – “Ramsey's ladder”, as he calls it – doesn't take us to a new theoretical level. He remarks that there are “philosophies that take advantage of the horizontal nature of Ramsey's ladder to climb it, and then announce a better view from the top.”Footnote 9
I take it that CP is one of the philosophies that Blackburn has in mind here. GE agrees with Blackburn, arguing that talk of truthmakers, denotations, and the like adds nothing to the repertoire of metaphysics, unless the semantic notions in question are more robust than those of Ramsey, Wittgenstein and Quine – and rejecting such a view of semantics. If a proponent of CP tries to embrace this conclusion, saying that their own use of semantic notions is similarly ‘thin’ – that in effect, Step 2 simply asks ‘What is the X such that R(X)?’ – then GE says again that we already have a trivial answer to that question, but no reason in general to expect a non-trivial one.
Once again, the qualification about generality is important. GE may have no need to challenge particular cases, including those of theoretical identification in science. But it will argue that in these cases the semantic characterisation is inessential – the questions can be phrased without it. Some proponents of CP might agree, and argue that the science model is all we need – CP should be simply be seen as generalised functionalism. On this view, R(Target) encodes the causal and functional role of Target, and Step 2 simply enjoins us to look for whatever it is that plays this causal role – a question for natural science, in principle. However, as Peter Menzies and I have pointed out, this version of CP doesn't have the generality to which CP aspires – it cannot handle the investigation of the causal relation itself, for example.Footnote 10
GE is not just Step 1. Thus GE rejects (any non-trivial reading of) Step 2, in general. But it would be a mistake to characterise GE by saying that it simply amounts to Step 1 of CP without Step 2. This would be to ignore what GE takes to be the most interesting question, or group of questions: viz., questions about the function of the term ‘Target’. What are the use-rules for the term? And what is it ‘for’ – what difference does its possession make, to creatures like us? As Michael Williams points out, both these questions can be seen as asking about the function of a term, but in different senses of ‘function’. The first asks a descriptive question about ‘how it works’, or functions in that sense; the second a potentially explanatory question about the role the term or concept plays in our lives – its function in a sense closer to the claim that the function of thirst is to make us drink when our body needs fluids.Footnote 11
Of course, views other than GE may ask these questions, too. What is distinctive about GE is that it eschews semantic notions in answering them. Thus GE asks a question about the function of the term ‘Target’, and expects an answer which doesn't rest substantially on semantic notions (though see Section 4.3 below).
Deflationism about truth conditions. It is important to note that eschewing semantic notions does not commit GE to denying that the language in question ‘has truth conditions’, and the like. On the contrary, GE claims, it is a more or less trivial matter that moral language (say) does have truth conditions, in the only sense GE takes to be available – viz., the deflationary sense. ‘Cruelty is wrong’ is true if and only if cruelty is wrong. Here is Blackburn making this sort of point about his own version of expressivism:
Q. 18. Aren't you really trying to defend our right to talk ‘as if’ there were moral truths, although in your view, there aren't any really?
Ans. No, no no. I do not say that we can talk as if kicking dogs were wrong, when ‘really’ it isn't wrong. I say that it is wrong (so that it is true that it is wrong, so it is really true that it is wrong, so this is an example of a moral truth, so there are moral truths).
This misinterpretation is curiously common. Anyone advancing it must believe themselves to have some more robust, metaphysically heavyweight conception of what it would be for there to be moral truths REALLY, and compared with this genuine article, I only have us talking as if there are moral truths REALLY. I deny that there is any such coherent conception.Footnote 12
In this respect, like Blackburn's view, GE differs from some of its early non-cognitivist ancestors. They were inclined to regard the claim that moral statements have truth conditions as false, rather than trivially true. It is deflationism that recommends this shift, of course. But far from making things problematic for the ur-insight of non-cognitivism – namely, that moral language is in a different line of work than standardly assumed – it actually supports it, in a dramatic fashion. If truth is sufficiently ‘thin’, then for no kinds of claims at all do we do any interesting work by saying that they have the function of ‘stating truths’, or anything of that sort. (More on this in Section 4.1.)
GE, CP and naturalism. As noted, CP typically assumes naturalism. The placement problem is find a place for morality, or meaning, in the natural world. GE rejects naturalism (of this sort), though it puts the point in ‘meta-linguistic’ rather than metaphysical vocabulary – i.e., by saying not that moral properties are not natural properties, but that moral vocabulary is in a different ‘line of work’ from the language of science. (It might be better to say that GE rejects the question about naturalism, as CP conceives it, rather than accepting the question and offering a different answer. Both sides think of the question ‘Are moral properties natural properties?’ as a kind of taxonomic question – Should this go in that box? – but they have very different views of what needs classifying. For GE it is uses of language.)
Subject naturalism. Finally, it is important to note that GE may retain naturalism in a different sense – what I have called subject naturalism, as opposed to the object naturalism of CP.Footnote 13 Subject naturalism is naturalism in the sense of Hume. It takes for granted that we humans are natural creatures, and that language is at base a natural behaviour. It seeks an understanding of the origins and functions of particular discourses on that basis.
2. Non-Naturalism and Platonism
Turning in a different direction, it is clear that in declining to embrace (object) naturalism, GE has something in common with various forms of non-naturalist realism and platonism, such as a Moorean view about morality (Moore, 1903), or platonism about meaning, mathematical objects, or abstract entities. How does GE differ from such a view? In two main ways.
Meta-linguistics, not metaphysics. First, the non-naturalism of GE is expressed, as noted above, in meta-linguistic rather than metaphysical mode. GE does not find it philosophically illuminating to say that the world contains moral properties, as well as natural properties. It may agree that moral properties are not natural properties, but this comes with a crucial clarification: this is to be understood as a loose way of expressing something that we put more clearly by shifting explicitly to the meta-linguistic frame, and saying that moral terms and concepts are in a different ‘line of work’ to the terms and concepts of natural science.
A non-representationalist account of meaning. What does it mean to be in a ‘line of work’? In explaining this we point to the second difference between GE and typical non-naturalist realist and platonist views. As we saw above, GE embraces a non-representationalist functional account of what we do with such terms and concepts. Orthodox non-naturalists and platonists are typically orthodox, among other things, in their representationalism. They take for granted that the function of moral terms is usefully characterised as that of ‘referring to moral properties’ (or something similar, expressed in related semantic terms). Once again, GE does not deny such claims, but regards them in the Quinean spirit as empty of substantial theoretical content. The substantial work takes place elsewhere, according to GE, in a subject naturalist functional genealogy of moral properties.
2.1 Non-naturalism below the bar – fictionalism and error theories
With this contrast between GE and non-naturalist realism in mind, we can treat as a subcase the contrast with non-naturalist irrealism. Non-naturalist irrealists agree with their realist cousins that moral terms are usefully characterised representationally: they are the kind of terms that ‘claim to’ refer to properties in the world. Where they differ from realists is in maintaining that such terms systematically fail to achieve such reference, for there are no such properties. Either our moral claims are flatly in error, or, at best, have the status of useful fictions.
GE differs from such views in two closely related ways. As before, it rejects the representationalist characterisation of the vocabularies in question, except in the trivial deflationary sense. And, in a move that has the effect of extending the same deflationary spirit to the metaphysical side of the ledger, it denies any sense to the irrealist's negative claims. This point was well made by Blackburn in early work, defending his own ‘quasi-realist’ position. (As we shall see, GE differs only in wanting to eliminate the qualification ‘quasi’.) As Blackburn puts it:Footnote 14
What then is the mistake of describing such a philosophy [quasi-realism] as holding that ‘we talk as if there are necessities when really there are none’? It is the failure to notice that the quasi-realist need allow no sense to what follows the ‘as if’ except one in which it is true. And conversely he need allow no sense to the contrasting proposition in which it in turn is true.
Blackburn continues:
Quasi-realism no more need allow such sense than (say) one holding Locke's theory of colour need accept the view that we talk as if there are colours, when there are actually none. This is doubly incorrect, because nothing in the Lockean view forces us to allow any sense to ‘there are colours’ except one in which it is true; conversely neither need it permit a sense to ‘there are actually none’ in which that is true.
My late Sydney colleague David Armstrong used to complain that the problem with Wittgensteinians is that they won't allow you to say what you want to say. Armstrong was right about Wittgensteinians, I think (if in my view wrong about it being a problem, at least in general), and Blackburn's point here is an example of it. This may be one of those Wittgensteinian things for which Ramsey deserves some of the credit. The point has much in common with Ramsey's own famous dig at the (early) Wittgenstein's view: ‘What we can't say we can't say, and we can't whistle it either.’Footnote 15
3. Relaxed Realism and Quietism
Our next contrast is with a different cluster of non-naturalist views, for which I'll borrow Sarah McGrath's excellent term ‘relaxed realism’. McGrath uses this label to characterise some recent positions in normative ethics. As she puts it ‘relaxed realist themes are central to Dworkin's Justice for Hedgehogs (2011), Parfit's On What Matters (2011), and Scanlon's Being Realistic about Reasons (2014)’.Footnote 16 She says:
I … call this picture relaxed realism … to capture the way in which its proponents combine a commitment to realism with a certain lack of anxiety about the status and standing of morality, despite understanding morality in ways that might naturally encourage such anxiety.Footnote 17
We can find similar views under other names, both in the normative case and in others. I'm thinking of McDowell's ‘re-enchanted naturalism’; of ‘minimal realism’, as used by many writers; of John Campbell's ‘simple realism’ about colour; of ‘liberal naturalism’, as used by writers such as Macarthur and de Caro; and of the kind of ‘neo-Fregean platonism’ associated with Bob Hale and Crispin Wright.Footnote 18
This is a large basket, and by no means homogeneous, but I hope it is clear that there are common themes – particularly, the rejection of a certain sort of metaphysical stance, the one that encourages anxiety, as McGrath puts it. A catch-all term for this rejection might be ‘metaphysical quietism’? Often attributed to Wittgenstein, this kind of quietism is characterised by McDowell as the rejection of a sideways metaphysical perspective on our practices.
Metaphysical quietism, yes; explanatory quietism, no. GE agrees with these views about the attractions of metaphysical quietism – of a deflationary approach to metaphysical issues. Where it disagrees, if at all, is in insisting on the interest and respectability of another project – the functional and genealogical project. Concerning McDowell, for example, my own strategyFootnote 19 has been to present him with a dilemma. Either he has to be more quietist than even he wants to be, in being unable to explain the sense in which (in his words), ‘[v]alues are not brutely there—not there independently of our sensibility—any more than colours are’.Footnote 20 Or he has to endorse what is in effect an expressivist genealogy – a ‘sideways’ explanation of how our value and colour judgements come to depend on aspects of our sensibility (different aspects, in each case).
Stepping back a little, we could say that relaxed realists face a trilemma. Faced with what seem to be legitimate questions about particular discourses – why we have them, how they differ, how they relate to our sensibilities – there are three main options. In the metaphysical corner are views that appeal to the nature of the properties or entities in question (e.g., again, colours and values) to answer such questions. In the extreme quietist corner are views that simply fail to engage with such questions. And in the third corner is expressivism. The first corner seems off limits for anything worth calling relaxed realism – but that leaves a choice between what is arguably an excessive quietism, and expressivism itself.Footnote 21
Of course, it is entirely possible that a relaxed realist might choose different corners in different cases. In particular, they might thereby end up endorsing our next complement to GE itself.
4. Local Expressivism
Local expressivism (LE) agrees with GE locally in response to the previous folk, in stressing the importance of pragmatist genealogy (for normative discourse, say). But it disagrees in wanting to maintain a bifurcation between cases in which this pragmatist stance is appropriate and cases in which it is not. A classic statement of this commitment is that of Robert Kraut:
The bifurcationist often undertakes the task of determining which of our well-formed declarative sentences have truth conditions and which ones, though meaningful, are simply the manifestations of attitudes or the expressions of ‘stances’. He wants to know which of our predicates get at real properties in the world, and which, in contrast, merely manifest aspects of our representational apparatus—‘projections borrowed from our internal sentiments’. On different occasions he articulates his task in different ways; but they all point to some variant of the bifurcation thesis …, the thesis that some declarative sentences …
— describe the world
— ascribe real properties
— are genuinely representational
— are about ‘what's really out there’
— have determinate truth conditions
— express matters of fact
— limn the true structure of reality
whereas other declarative sentences …
— express commitments or attitudes
— manifest a ‘stance’ (praise, condemnation, endorsement, etc.)
— are expressive rather than descriptive
— do not ‘picture’ the world
— lack truth conditions, but possess ‘acceptance conditions’ or ‘assertibility conditions’
— merely enable us to ‘cope’ with reality
— are true (or false) by convention
— do not express ‘facts of the matter’.Footnote 22
As I say, LE wants to maintain some distinction of this kind, and to state its central insight as the idea that some interesting discourses – moral or modal discourse, for example – fall on the latter, ‘expressive’, side of it. For such views the question as to how precisely to formulate the bifurcation thesis becomes crucial. Many early LE views tended to do it terms of truth, saying that moral claims lack truth conditions, or something of that kind. However, as Blackburn again deserves much credit for pointing out, such a view is at best incomplete: it leaves us with the question as to why such claims look so much like the claims which do have truth conditions, according to this view. If moral claims don't have truth conditions, why do we call them true and false, in ordinary conversation? Blackburn's quasi-realism (QR) can be thought of as an attempt to answer this important question, and a generalisation of it: If there is a bifurcation, why is it so well hidden in ordinary usage?
GE agrees with QR in pressing this kind of question against early versions of LE. Against QR, it argues that by QR's own lights, there is no satisfactory basis for a bifurcation, at least in the broadly semantic territory in which it LE has tried to find it. In semantic terms, QR comes under pressure to extend whatever it says about the semantic features (e.g., the use of ‘true’ and ‘false’) of supposedly expressive discourses to all discourses, thus eliminating the bifurcation. This is what David Macarthur and I have called the global challenge to QR.Footnote 23 For present purposes I'll call it the semantic global challenge, so as to distinguish it from a second pragmatic global challenge.
4.1 The semantic global challenge
In more detail, the global challenge works from two directions: pulling from the outside, and pushing from the inside, as Macarthur and I say. Pulling from the outside, the argument appeals to semantic deflationism, or minimalism, observing that such a view threatens to deflate the QR's residual representationalism. Deflationism is often characterised as the view that the notions in question don't do explanatory work.Footnote 24 But grounding the bifurcation thesis certainly would be explanatory work.
It is important to realise that this is not an old argument that minimalism defeats non-cognitivism, by making it ‘easy’ to be truth-conditional.Footnote 25 Macarthur and I meet that argument on LE's behalf by distinguishing what we call the positive and negative theses in traditional expressivism. The negative thesis is that moral claims (say) do not have truth conditions. The positive thesis is that moral claims have some non-semantically-characterised function, e.g., that of expressing affective attitudes.
Deflationism does undermine the negative thesis, but doesn't challenge the positive thesis. On the contrary, it suggests that the positive thesis has to be a model for everything, in the sense that it implies that for no vocabulary at all can it be informative to say that it has a semantically-characterised function – deflationism denies us such a theoretical role for semantic notions. Deflationism is thus a friend not an enemy of expressivism, and militates strongly in favour of the global version. We take Blackburn's version of QR to be vulnerable to this argument because Blackburn is (usually – forgetting occasional lapses into ‘success semantics’) a card-carrying deflationist.
So much for pulling from the outside. By pushing from the inside, Macarthur and I mean the argument that QR threatens to be too successful for its own good, so long as it retains the ambition to be a merely local view. After all, if QR can show why we talk the truth talk without walking the representational walk in hard cases, such as ethics, why not in easy cases, too? For example, if the explanation of truth talk in the case of ethical language is that it encourages us to align our affective attitudes in a useful way, why not say the same about other mental states, such as the ones QR thinks of a genuine beliefs. Why not think of their truth talk as explained in the same way?Footnote 26
4.2 The pragmatic global challenge
There is another ingredient to the case for preferring GE to LE, in my view. It rests on the central insights of expressivism, and on the realisation that, once in view, the kind of pragmatic factors important to the formulation of expressivism in familiar ‘local’ cases can be seen to be universal. No discourse is wholly free of them, and expressivism thus becomes a global view.
Briefly, the case goes like this. Expressivism links particular assertoric ‘vocabularies’ to particular ‘pragmatic grounds’ – i.e., to the practical features of speakers on which the use of a particular vocabulary depends. In the moral case, for example, the pragmatic grounds are (in the simplest version of the view) the affective attitudes that moral claims are taken to express.Footnote 27
I have appealed to rule-following considerations to argue that pragmatic grounds are universal – no part of language is free of them. If nothing else, we always rely on contingent dispositions to generalise in the same way from finite classes of training examples. In Facts and the Function of Truth I put this in term of what I called ‘no-fault disagreement’ (NFD). NFD arises in cases in which two speakers seem to disagree, but the apparent difference of opinion turns out to rest on some non-obvious difference in their situation – e.g., in certain cases of probability judgements, which were one of my main examples, on the fact that they have access to different bodies of evidence.
I take NFD to be a characteristic symptom of variation in pragmatic ground. To use an example I give elsewhere,Footnote 28 two speakers might disagree as to whether Canberra is a bustling place. When it turns out that their ‘bustle receptors’ are simply set at different levels – one comes from a rural village, one from a busy city – we are inclined to say that neither has made any mistake. The rule-following considerations show that in principle, all uses of language are subject to this kind of possibility. What this reveals is a particular sort of pragmatic ground underlying all language whatsoever. In my view, this provides a further powerful reason for preferring GE to LE.Footnote 29
4.3 Isn't a different bifurcation possible?
On behalf of LE, it might be objected that GE's rejection of the bifurcation thesis is too swift. Even from a pragmatist's perspective, isn't there something to said for the idea that some of our claims and mental states are more in the business of keeping track of our external environment than others? Analogies with other animals provide one way to develop this objection. Surely they have internal states that function to keep track of their environments, for various purposes. And don't we do the same thing?
This a very helpful objection, and it is useful to think first about what QR should make of it. Even for QR, there's clearly a dilemma lurking here. If QR tries to put weight on some such notion of environment-tracking, in order to ground a semantic bifurcation thesis, the same internal tensions in the position will be pushed to the foreground: roughly, the more QR says that real truth is to be understood in terms of environment-tracking, the less plausible it will be that QR can offer some satisfactory ‘quasi-truth’ in the cases it wants to treat in expressive terms; while the more plausible QR makes its account of ‘quasi-truth’, the less plausible it will be that any separate account of truth is needed in the (claimed) environment-tracking cases.
The solution I have recommended, in the light of these considerations, is to be clear that we have two different notions (or clusters of notions) in play. There isn't a univocal notion that works both in the environment-tracking cases and as an account of the notion of truth in play in language at large. But once we recognise this, and keep these notions distinct, everything goes smoothly. I have put the distinction in terms of two notions of representation: an environment-tracking notion I call e-representation and a broader, linguistically-grounded notion I call i-representation. As I have noted, this distinction does much the same job as Sellars’ distinction between two notions of truth, notions that ‘belong in different boxes’, as Sellars puts it.Footnote 30
So long as we recognise that the narrower notion (my e-representation) should itself be regarded as a pragmatic notion, a bifurcation cast in these terms doesn't in any way undermine the global character of GE. Environment-tracking is one pragmatic function among many others, in effect. The appeal to rule-following should counter any tendency to think that the parts of language that are in the business of environment-tracking are somehow less pragmatically-grounded than other parts of language. On the contrary, the rule-following point shows that there's at least one pragmatic element that underpins them all.Footnote 31
5. Response-Dependent Realism
For our last contrast, I turn to a view once widely seen as an attractive alternative to what we are here calling expressivism, an approach I shall call response-dependent realism (RDR). Leading early versions of this view included those of Mark Johnston and Crispin Wright.Footnote 32 RDR can be seen as a proposal for defending the ‘factual’, ‘cognitive’ or ‘realist’ character of various discourses, by putting pragmatic factors – e.g., desires, in the moral case – into the content. In effect, it proposed to retain factuality (and the like) by reading a discourse as more subject-involving than initially it seems, on the model of the view that colours are dispositions to affect normally-sighted humans in certain ways. Accordingly, as Johnston's title suggests, it can be seen as a form of pragmatism; but not pragmatism as Humean expressivists know it. A Humean expressivist doesn't take moral claims to be talking about a speaker's affective reactions, of course – that's mistake number one in the expressivist's list of common misinterpretations.
By the standards of contemporary expressivism, however, it is hard not to see RDR as a solution to a non-existent problem. As we have seen, contemporary expressivists in the Humean tradition (especially at the GE end of it) don't deny that moral claims have truth conditions, or reject simple speaking-with-the folk realism about moral properties and the like. On the contrary, they affirm these things, in the minimal sense – that's where they agree with relaxed realists – while continuing to insist on an expressivist functional genealogy. (The point of QR was to show how this is possible, starting where Hume does.)
As we noted above (Section 4.1), semantic minimalism is a friend not an enemy of expressivism of this sort (at least until the dispute between LE and GE comes into play, when it favours GE). And minimal semantics seems to bring minimal content, in the obvious way. The content of the belief that X is good is that X is good.
This means that RDR needs some other ‘non-obvious’ notion of content, or truth conditions. And now the dialectic is much as in the fictionalist case, and indeed the Canberra Plan case. GE simply challenges the entitlement to any further fact of the required kind, expressed in semantic vocabulary. GE is not opposed to further facts in the neighbourhood tout court, of course. It simply insists that the right vocabulary in which to express them is one of use conditions, or something similar, not truth or content conditions. As Blackburn himself puts these points, commenting on RDR:
[RDR] goes bullheaded at the issue [of meaning] in terms of finding truth conditions, whereas from the point of view of [expressivism], … if you want to talk in these terms [i.e., in terms of truth conditions], then the best thing to say about ‘X is Φ’ in the cases considered is that its truth-condition is that X is Φ—but this will not be the way to understand matters [i.e., to say anything interesting about the meaning of the claims in question].Footnote 33
In my own early criticism of RDR,Footnote 34 I also argued that the right place for pragmatic factors was in use conditions, not truth or content conditions. Among other considerations, I argued that use conditions make better sense of actual usage, in cases exhibiting no-fault disagreements (e.g., again, about whether Canberra is a bustling place). I also argued that the content condition view is incoherent, as a global view – something has to go in the background, pragmatically presupposed but not stated, on pain of vicious regress. RDR cannot be a model for a global pragmatism.
5.1 GE is not Idealism
Some proponents of RDR may have felt that it offered an attractive alternative to two ways of denying that claims about colour, value, and the like, are answerable, as they seem on the face to be, to a reality beyond ourselves. On the one side (so such proponents thought) was expressivism, which they took as the view that such utterances are not answerable to anything, not being genuine claims in the first place. On the other side was an unattractive idealism, which regarded such claims as entirely subjective – entirely ‘about ourselves’, in some sense, and so not answerable to external reality for that reason.
I have explained why contemporary expressivists, especially global expressivists, would reject the characterisation of their view on which the first part of this contrast depends. It may be helpful to finish by saying why this does not put them on the other side, saddled with an unattractive idealism. Part of what needs to be said has already been mentioned. GE doesn't take moral claims to be about anything other than what they seem to be about, but for the most banal of reasons: ‘about’ simply isn't one of GE's words, in any interesting sense.
At this point the concern that GE is ‘really’ a form of idealism tends to surface as the concern that to the extent that expressivism allows moral facts, it makes them ‘depend on us’ in some implausible way. The expressivist is thought to be committed to claims such as these: kicking dogs is only wrong because we disapprove of it; if we approved of it, it would be good.
The expressivist responds with some careful distinctions. If the question is what we should say about an imagined world, similar to ours except that people enjoy kicking dogs, then of course we assess by our own standards. What those unpleasant imagined people are doing is wrong, even though they enjoy or approve of it. (Sadly, we can also think of plenty of non-imaginary cases of this kind.)
So in this case we assess ‘from the inside’ – from our own standpoint. If we leave this standpoint behind we can talk about what those other folk would say, but not about whether what they say would be true. Attempting to do that involves a kind of use–mention confusion. Folk who approved of kicking dogs might say that it was ‘good’, and their term might have a very similar expressive function as ours, but this licenses no sense in which it is appropriate to say that the facts would be different – for there's no stance to talk about the facts other than our own.
6. Postscript: Two Allies
This concludes our tour of five ways to disagree with global expressivism. To finish, let me mention two views that I take to be very congenial to GE. One is very well-known, one less so.
6.1 Pittsburgh Pragmatism
To introduce the first of these views, recall the local expressivist's bifurcation thesis, and frame it, as we did at some points above, in terms of content. QR can be thought of as claiming to explain how there can be assertions with (say) moral contents, even though the job of moral claims is not being characterised (in any substantial theoretical sense) as that of keeping track of any corresponding aspects of reality. The content of moral judgements cannot be considered to be ‘upstream’ of moral discourse, in other words, in some realm to which the ability to use moral terms gives us access. Rather it emerges ‘downstream’, a product of the practice, and when cast in these terms, the task of QR is to tell us how the trick is turned.
Generalising this thought, we can characterise the bifurcation thesis as the proposal that we need two kinds of accounts of propositional content. For some of our claims and beliefs, on this view, something proposition-shaped or content-like lies upstream, needed to explain what it is to hold a belief with that content. For others claims and beliefs, propositional content emerges only downstream, being explained as Blackburn wants explain the content of moral or modal language – i.e., in terms of what we do with the language and concepts in question.
Characterised in these terms, GE amounts to the view that the downstream model is appropriate in all cases. But here is Brandom, making what I take to be the same distinction in terms of direction of explanation:
An account of the conceptual might explain the use of concepts in terms of a priori understanding of conceptual content. Or it might pursue a complementary explanatory strategy, beginning with a story about the practice or activity of applying concepts, and elaborating on that basis an understanding of conceptual content. The first can be called a platonist strategy, and the second a pragmatist (in this usage, a species of functionalist) strategy.… The pragmatist direction of explanation, by contrast, seeks to explain how the use of linguistic expressions, or the functional role of intentional states, confers conceptual content on them.Footnote 35
Brandom says that his own view is ‘a kind of conceptual pragmatism’: ‘It offers an account of knowing (or believing, or saying) that such and such is the case in terms of knowing how (being able) to do something … —in general, the content by the act, rather than the other way around.’Footnote 36 Again:
Starting with an account of what one is doing in making a claim, [pragmatism] seeks to elaborate from it an account of what is said, the content or proposition—something that can be thought of in terms of truth conditions—to which one commits oneself by making a speech act.Footnote 37
Unhindered by the piecemeal starting points of Blackburn's Humean expressivism, and committed to a general inferentialism about meaning, Brandom simply takes for granted that this kind of pragmatism should be global in nature. There is no bifurcation. Content is everywhere downstream of usage. So Brandom counts as a global expressivist, in my terminology.
6.2 Cambridge Pragmatism
The most basic difference between GE and all the above rivals is that GE sticks consistently to the view that the appropriate philosophical stance is the meta-linguistic one, not the metaphysical one (and that the former is nowhere inappropriate – that marks the contrast with LE). I have used various terms for this contrast in various places. In Facts and the Function of Truth I contrasted the project of offering an ‘analysis’ of truth with that of offering an ‘explanation’ – the latter, the one that I recommended, being what I am here calling the meta-linguistic approach. In other places I have drawn a contrast between ‘metaphysics’ and ‘anthropology’, marking more or less the same distinction.
In recent workFootnote 38 I have noted that when F. P. Ramsey comes this way, he speaks of psychology, not anthropology. Here he is in ‘General Propositions and Causality’, reflecting on a possible response to an account of causation he has just sketched – an account that we would now call expressivist, or pragmatist.
What we have said is, I think, a sufficient outline of the answers to the relevant problems of analysis, but it is apt to leave us muddled and unsatisfied as to what seems the main question—a question not of psychological analysis but of metaphysics which is ‘Is causation a reality or a fiction; and, if a fiction, is it useful or misleading, arbitrary or indispensable?’Footnote 39
Ramsey doesn't address this concern directly, but I think it is clear that his view is that metaphysics is the wrong mode of enquiry, in this case. The illuminating enquiry is the one he calls ‘psychological analysis’ – an investigation into how we come to think and talk in causal terms, conducted in a manner that we do not presuppose that the helpful answer will lead us back to the objects. (In other words, we do not presuppose that the answer will be ‘We talk this way because we are keeping track of the causal facts’, or anything of that kind.)
I have dubbed this stance ‘Cambridge Pragmatism’, noting that in can be found in places in the work of many later Cambridge figures – and not just the obvious ones such as Blackburn and Wittgenstein. Other examples include Mellor on tensed language, Anscombe on the first-person, Craig on knowledge, von Wright on causation, and Bernard Williams, arguably, on truth itself. Most of these figures count in my terms as local pragmatists, or local expressivists. As for Ramsey himself, Cheryl Misak argues that under the influence of Peirce, Ramsey was already a global Cambridge Pragmatist.Footnote 40 While I have expressed some reservations about this claim,Footnote 41 I think it is clear that Ramsey was moving in that direction. As Richard Holton and I have argued,Footnote 42 he would have been pushed there by factors related to what we now call the rule-following considerations. In the terminology of the present paper, then, Ramsey is at least a proto-Global Expressivist.
