Pink Elephants

Alex Oliver

  • Articulating Reasons: An Introduction to Inferentialism by Robert Brandom
    Harvard, 230 pp, £21.95, June 2000, ISBN 0 674 00158 3

I have to confess that before starting on this review I hadn’t read Robert Brandom’s massive Making It Explicit (1994). Although it’s famous, very few of my colleagues have read it either (I mean read it, not just bought it or dipped their toes in it). Writing such a walrus of a book is a risky business. Life is short and it’s publish or perish; so a lot is written and little is read. Michael Dummett, one of the very few contemporary philosophers who, like Brandom, have dared to write books of more than seven hundred pages, has even declared that the merit of a publication ‘must be great enough to outweigh the disservice done by its being published at all’. So the present book, billed as ‘an approachable introduction to the complex system that Making It Explicit mapped out’, promises great relief.

Jackets are designed to push one from the shelves to the till, and this jacket has the lot: an enigmatic photograph of a Japanese town, eulogistic quotes and an author who looks for all the world like a member of ZZ Top. A hyperbolical blurb crowns all this, with talk of ‘a near-Copernican shift in the philosophy of language’. This is doubly allusive. First, to Kant’s ‘Copernican Revolution’ in his first Critique, which turned metaphysics and epistemology inside out, making objects conform to the structure of our minds, just as Copernicus explained the apparent motion of the planets in terms of the real motion of the spectator. And then to Thomas Kuhn’s use of the Copernican overthrow of Ptolemaic astronomy as an example of the paradigm shift characteristic of scientific revolutions.

These twin associations illustrate Brandom’s schizoid self-image as the revolutionary – ‘opposed to many (if not most) of the large theoretical, explanatory and strategic commitments that have shaped and motivated Anglo-American philosophy in the 20th century’ – but also as the spokesman and latest member of a tradition which includes the deities (Kant, Hegel, Frege and Wittgenstein) as well as Dummett, Rorty (Brandom’s teacher) and ‘the Sage of Pittsburgh’ (his university), Wilfrid Sellars.

Brandom’s overarching aim is old-fashioned: to find a single, all-or-nothing feature which distinguishes us both from inanimate, reliable indicators such as thermostats and sentient creatures such as parrots. His answer is equally traditional. We are rational animals (his word is ‘sapient’). To be rational is to apply concepts, and concept-use is an ‘essentially linguistic affair’. Thermostats don’t apply any concepts when they switch over. Merely uttering sounds which are indistinguishable from English is insufficient, however. A parrot trained to respond ‘that’s red’ to a swatch of material is not applying concepts either (Locke’s loquacious parrot was merely sounding off). So Brandom needs an account of linguistic practice which will ground his claims about concepts and rationality.

The problem with language, like so many targets for philosophical investigation, is knowing where to start. Gilbert Ryle drew a rich picture of the philosopher as traffic policeman trying to unpick a jam ‘when crowds of conceptual vehicles, of different sorts and moving in different directions meet at some conceptual crossroads’. We can put the same image in terms of facts. Sometimes they come from quite different domains: for example, here are all the facts of particle physics and there are all the facts about minds, or morals, or mathematics, or medium-sized dry goods, or, to take our present case, meanings. How do they fit together?

Moreover, different features of a given domain need to be jointly controlled. In the domain of language, the two most obvious facts are that linguistic expressions have meaning and that we use language to do things. These are clearly connected, but how? Brandom begins with speech acts – what is done with language – and hopes to derive an account of meaning from use, an explanatory order encapsulated in his slogan, ‘semantics must answer to pragmatics.’

Think of all the things we do with language. We tell jokes and stories, sing songs, skip to rhymes, solve crosswords. We insult, swear, apologise, save face and impress. We instruct, persuade, vote and umpire (‘Not out’). Then there are all the forms of ritual (‘I baptise you . . .’). Then all sorts of social glue, starting with Malinowski’s ‘phatic communion’ – the idle and rigid chit-chat of ‘How are you?’, ‘Nice morning, isn’t it?’, ‘Ciao’ – right up to the dextrous mock insults of sounding and rapping.

So, again, where should we start? Brandom fastens onto sober acts of assertion in which we claim that things are thus and so, that the tulips have wilted or that the square of the hypotenuse is equal to the sum of the squares of the other two sides. The circle involving rationality and linguistic practice is now especially tight. Assertions are ‘essentially performances that can both serve as and stand in need of reasons’, and, for Brandom, no set of activities counts as a linguistic practice unless it incorporates assertion.

In order to flesh out the sense in which assertions are located within a ‘game of giving and asking for reasons’ (Sellars-speak), Brandom employs the notions of commitment and entitlement. When I assert that a particular elephant is pink, I undertake a commitment and thus have a certain ‘normative status’. It’s normative because the commitment is characterised using the vocabulary of obligation: given my first assertion, I ought also to be prepared to assert that my elephant is coloured. One is a reason for the other. It may also happen that someone calls on me to justify – to give a reason for – my original assertion. If I can show that I am entitled to the commitment I have undertaken, I come to have a further normative status.

Successful communication requires that we all know what each of us means. For Brandom, this is practical knowledge. In his idiom, conversational partners know how moves in the game of assertion change the score of commitments and entitlements. The elaborate interplay between these, and the corresponding attitudes which attribute such statuses, is spelled out at some length in his big book. There, he is not concerned to show how this realm of normative facts fits in with the facts of physics. There is some suggestion, however, that these commitments and entitlements are grounded in the attitudes which attribute them, and this gets replayed in the present book as the claim that they are ‘creatures’ of, or are ‘instituted’ by, these attitudes. I challenge a reader of either book to tell me what’s going on here.

In any case the focus of this ‘introduction’ is the particular theory of meaning, ‘semantic inferentialism’, which he derives from his account of linguistic practice. The primary unit of meaning is the sentence, the vehicle for assertion. Sentences ‘acquire content by being caught up in inferences, as premises and conclusions’. The game of giving and asking for reasons thus confers meaning on its pieces. Unlike our assertions, neither the thermostat’s nor the parrot’s responses are involved in inference and thus neither have meaning. In other words, there is no application of concepts and, since that’s the touchstone, neither the thermostat nor the parrot is rational.

Using contemporary jargon, Brandom’s theory of meaning is a version of ‘inferential role’ or ‘conceptual role’ semantics. In their book Holism, Jerry Fodor and Ernest Lepore state that ‘quite a lot of the philosophy of language and philosophy of mind since the 1940s’ has operated with some version of this semantics. In contrast, Brandom’s blurb states that his semantics is a ‘near-Copernican shift in the philosophy of language’ and ‘the most important single development in the field in recent decades’.

If there’s anything distinctive in Brandom’s semantics, it must come with his particular version of ‘inferential role’ semantics. Before exploring this, we need to grasp the kind of ‘Ptolemaic’ semantics he is seeking to overthrow. It starts by understanding meaning in terms of representation, not inference. Sentences employed in assertion represent the world as being a certain way. They have truth conditions and when the world satisfies the conditions, they are true; if not, not. These truth conditions are themselves explained in terms of relations of reference holding between bits of sentences and bits of the world: for example, the relation between the name ‘Socrates’ and the person named. Inference comes in late and is explained in terms of truth conditions. To take a simple case, an inference is said to be valid when any situation which makes all the premises true, also makes the conclusion true.

Brandom wants to reverse all of this, by first explaining meaning in terms of inference and only then moving on to the representational idioms of truth and reference. The devil is always in the detail. Semantic inferentialism is made concrete only when one spells out the moves which determine meaning. Brandom wants to include only correct inferences. This connects back to the normative statuses of commitment and entitlement, and the notion of a good reason that they involve. So the question is: which inferences are correct?

Following Sellars, Brandom insists that he’s not just interested in ‘formally valid’ inferences, but a more inclusive category. Roughly, an inference is formally valid when its validity depends solely on the meaning of logical words like ‘not’ and ‘or’. Consider ‘Pittsburgh is to the west of Princeton. So Princeton is to the east of Pittsburgh.’ This doesn’t contain any logical words, but we all recognise that it’s a good inference and doesn’t need an extra premise to make it so. In Brandom’s terms, it’s ‘materially correct’. We want to say that its correctness depends on the meanings of ‘west’ and ‘east’. And that is what he says too: it is ‘the contents of the concepts west and east’ that make it ‘a good inference’.

But how can he say this? We were supposed first to get a grip on the idea of correct inference and then explain the content of concepts in terms of it. It now appears that Brandom is running things backwards. He might fall back on a primitive notion of correct inference, not explained in other terms, but clearly recognisable whenever we come across it. The trouble is that what he says about correct inference is highly controversial when set in the context of an account of linguistic meaning. Take another of his materially correct inferences: ‘Lightning is seen now. So thunder will be heard soon.’ There’s something to be said for treating this inference as in order as it stands without our needing to detect a suppressed premise. But is it really an inference which can be used to determine the meaning of its constituent sentences and their parts? Brandom says so, but it looks to be more a matter of how the world is (consider reversing the relative speeds of light and sound) than a matter of what those expressions mean.

To reinforce the point, take an inference which works well in my college: ‘Tomorrow is Friday. So it’ll be fish for lunch.’ This is hardly one which can help to determine the meaning of ‘Friday’, ‘fish’ or ‘lunch’. I suppose Brandom might insist that there is a suppressed premise here which contributes ‘real-world knowledge’, so the inference is not to be counted in the reckoning of meaning. But he’ll have to characterise what ranks as real-world knowledge and what doesn’t, and then notions of meaning will creep back in, preventing his preferred order of explanation from inference to meaning.

At the end of his fifth chapter, however, rather than make such a distinction, Brandom moves towards a full-strength ‘holism’ which allows any inference with any premises to count as determining the content of its constituent expressions. This makes meaning radically speaker-dependent since the premises I am disposed to accept will differ from yours (not to mention the inferential moves I deem correct). But I find him inconsistent. For in the introductory chapter he claims that the content of the concept molybdenum ‘need not change as my views about molybdenum and its inferential surround change’.

There’s little here (or in the big book) that adds up to a concrete semantics. Furthermore, in this introduction, Brandom does not provide any support for his explanatory direction (from inference to meaning to truth and reference) save for appeals to authorities, the most surprising of whom is the ‘young Frege’, the author of the founding document of modern logic, Begriffsschrift (1879). But his exegesis is ropy. Certainly, Frege makes out a notion of ‘conceptual content’ for sentences according to which identity of inferential role entails identity of content. But there’s no warrant for thinking that he aimed to leave inference unexplained and to make it explain truth and reference. He was merely interested in arguing that grammatical distinctions within the vernacular, such as that between subject and predicate, should not be preserved in his formal language when they do not make any inferential difference.

As I have said, Brandom locates meaning first of all at the level of sentences since they act as premises and conclusions in inference and are the vehicles for assertion. But a meaningful sentence has meaningful parts, whose meaning contributes to the meaning of the whole. How else would we be able to explain ‘semantic creativity’, the ability to construct and understand meaningful sentences which we have not encountered before (like many in this review)?

In his fourth chapter, Brandom gives an account of the meaning of one kind of subsentential expression – namely, singular terms. His examples include proper names (‘Benjamin Franklin’) and definite descriptions (‘The first Postmaster General of the United States’). Naively, we might explain their meaning in terms of their referential role: they purport to stand for objects. But Brandom does things in terms of inference, not representation. He proceeds by first stating a principle which sorts expressions into groups: two expressions belong to the same group if and only if substituting one for the other in a grammatical sentence never results in an ungrammatical sentence. The singular terms are then distinguished by appeal to their characteristic contribution to the inferential properties of the sentences in which they occur.

One might think that this is all very well for English and kindred natural languages, but won’t some cunning linguist produce an esoteric language which does not contain singular terms? Brandom tries to show that he is not hostage to empirical fortune. A language which has subsentential structure and which can express certain basic logical notions (the conditional ‘if’ or negation) must contain singular terms. Since he explains objects as the purported referents of singular terms, he calls this reasoning ‘rather grandly, an expressive transcendental deduction of the necessity of objects’.

How unfortunate, then, that some sordid facts wreck the argument at the very start. For it is easy to see that, by his own classification, his singular terms cannot belong to the same grammatical category. Try one example from many. Put ‘the first Postmaster General of the United States’ for ‘Benjamin Franklin’ in ‘Clever old Benjamin Franklin invented bifocals’ and you’ll make the original grammatical sentence ungrammatical.

It is somewhat ironic that, despite his avowed interest in natural languages rather than the formal languages of logic, Brandom uses a principle of grammatical classification which has never been taken seriously by grammarians or linguists. Formal languages are a different matter since the grammatical simplicity and regularity required for them to count as formal is naturally secured by using Brandom’s principle. Like many philosophical logicians before him, he has illegitimately transposed a feature of artificial languages to their natural counterparts.

The odd thing about this introduction is that it was never going to be one. The six chapters were originally written as professional lectures, versions of some have been published in the journals, and in places great chunks of text are simply lifted from Making It Explicit. Articulating Reason’s own introduction is even trickier, requiring a thorough background in the philosophy of language if it’s to be understood. It is thus absurd to sell the book as putting Brandom’s work ‘within reach of non-philosophers who want to understand the state of the foundations of semantics’, unless reach is supposed to exceed grasp.

There is no doubt that the subject-matter is abstract and difficult. Yet Brandom’s style obscures it further. Macaulay found Kant’s Critique ‘utterly unintelligible, just as if it had been written in Sanskrit’ and Russell ranked Hegel ‘the hardest to understand of all the great philosophers’. There’s something of this Germanic style in Brandom (alongside a large dose of Sellars): a distaste for the concrete combined with a distrust of ordinary vocabulary; a delight in neologisms (‘botanisation’, ‘explicitation’, ‘committive’) and technical jargon (don’t say ‘belief’, say ‘doxastic commitment’, not ‘action’ but ‘discursive exit transition’); and a weakness for taxonomic structures and elaborate architechtonics with elegant but forced symmetry.

One feels that, unable to share Hegel’s ambition ‘to teach philosophy to speak German’, Brandon settles for the next best thing. Witness these pile-ups: ‘incompatibility-inferentially’ and ‘constitutive, pragmatist, relationally linguistic, conceptual expressivism’. All of this is a shame since the occasional aphoristic gem is neutered by the surrounding prose, like Château Margaux on the rocks. Aristotle was spot on when, in his Rhetoric, he warned against such ‘frigidities’.

I suspect Brandom thinks that his philosophical problems and solutions don’t yield to ordinary expression but demand a whole new vocabulary. I have never believed this special pleading which, in any case, could only be a partial excuse. We owe clarity to the nature of our subject (it’s at least half the message) but, more important, as a courtesy to our audience. Clarity is essential even if it isn’t enough.

I am sorry not to be able to soften these criticisms with the customary concluding politeness (‘anyone working in the field will want to read this book’). The brutal fact is that the theory of meaning at the core of Brandom’s project is in places ill-defined, at others inconsistent, and at yet others plain wrong.