C’est mon métier

Jerry Fodor

  • Philosophy in an Age of Science by Hilary Putnam
    Harvard, 659 pp, £44.95, April 2012, ISBN 978 0 674 05013 6

It would take at least two workaday philosophers to keep up with Hilary Putnam. Philosophy in an Age of Science is a case in point. It’s a collection of papers, most of them previously published, devoted among lots of other things to: the philosophical interpretation of quantum mechanics, the philosophy of mind, the philosophy of language, the philosophy of mathematics, philosophical ethics (analytic and otherwise), and the debate between solipsists, phenomenologists and realists about the epistemological and metaphysical status of ‘external’ objects. That is a long list and it is a long book. One might well doubt that there are philosophers positioned to say important things on all of these topics; but there are, and Putnam is one of them.

I, however, am not. What I know about quantum mechanics amounts to rather less than nothing; what I know about the philosophy of mathematics is only that it is very hard; and, though I have my doubts about ethics and epistemology as they are currently practised by philosophers (viz, as attempts to ‘refute’ one or other sort of scepticism: what a bore!), they are at best dark suspicions and at worst mere prejudice. Nobody, myself certainly included, could suppose that my views about any of these things are informed. On the other hand, I have spent a lot of time over a number of decades thinking about mind and language, and I do have fragments of stories (both philosophical and empirical) to tell about them. And if the sorts of story I’m inclined to tell are even close to right, then quite a lot of what Putnam says on these topics is wrong; indeed, wrong-headed.

Putnam starts from a good place. He shares (and presupposes) Quine’s anti-reductionism, his epistemological holism, his respect for the physical sciences, and his rejection of the analytic-synthetic distinction; which is to say, putting this another way, that Putnam shares Quine’s theses that the evidence relevant to the rational (dis)confirmation of beliefs is not restricted to ‘data’ (still less to ‘observational’ data); that the predictive and explanatory success of our best scientific theories is a strong prima facie argument for their truth, and that the distinction between ‘conceptually (or semantically) true’ and ‘true as a matter of fact’ can’t be sustained. Fine. But then Putnam’s Quine gets mixed up with Putnam’s Wittgenstein in aid of theses of which I’m not at all sure either would have approved.

Quine’s worries about the distinction between analytic truths and empirical truths are part and parcel of his epistemological holism. You can’t trust the fact/meaning distinction because considerations of simplicity, elegance, the availability (or otherwise) of alternative theories, and compatibility with theories in other sciences can all be pertinent to the acceptance of empirical beliefs and they can trump putative claims of conceptual, linguistic or semantic necessity. Indeed, they often do. Empirical confirmation spreads through networks of beliefs so that whole theories are what constitute the units of empirical acceptability. No doubt, such very holistic views of confirmation can tend to hyperbole. Data from well controlled and publicly replicable experiments can have a certain intractability even in the face of well-entrenched theoretical commitments. But the traditional empiricist account of confirmation as the comparison of punctuate beliefs about the world with correspondingly punctuate facts about the world is at least equally implausible, and it is, for better or worse, currently out of fashion. So be it. But Putnam has it in mind to add a dash of Wittgenstein to the mix, and I think that gets him into trouble. In fact, Quine and Wittgenstein hold quite different views, ones that I doubt are even compatible.

Quine is, by and large, dismissive about meaning (sometimes he’s behaviouristic about meaning, which comes to much the same thing). Wittgenstein’s semantics is more nuanced: he doesn’t think the meaning of a word is a definition; but he does think that much of the work that philosophers have hoped to do by appealing to meanings can instead be done with the notion of the ‘use’ of a word in a ‘public’ language. For Wittgenstein, the use of an expression is something like its ‘role in a language game’ or even its role in a ‘form of life’; and, though appeals to use perhaps don’t support a notion of analyticity, they do support a notion of ‘criterion’, which is, very roughly speaking, a ‘way of telling’ whether a word/concept applies to a thing. So to have the concept chair (to understand the word ‘chair’) is, among other things, to know that things like this count as chairs. Accordingly, anyone who is sceptical that there even are such things as chairs, or that we can know that there are, has something less than a full grasp of the word or concept. Perhaps holding that sort of view isn’t quite tantamount to endorsing a notion of truth-in-virtue-of-meaning; but it certainly doesn’t sound much like Quine, and it serves to distinguish his project from Wittgenstein’s.

Wittgenstein’s semantic holism is directed primarily against scepticism; Quine’s epistemological holism is directed primarily against empiricism. I think Putnam ignores this crucial difference. Quine doesn’t care much about what meaning is; he cares mainly about what confirmation isn’t. By contrast, Wittgenstein wants very much that meaning ‘supervenes on’ use. ‘As supervene on Bs’ is philosophers’ jargon for the proposition that there are ‘no differences among As without some difference among Bs’. So, whereas Quine rejects analyticity out of hand, in Wittgenstein it just goes underground. Truth in virtue of the (implicit) rules of use replaces truth in virtue of meaning (i.e. analytic truth, i.e. truth by definition); but Wittgenstein fully accepts the positivist thesis that it would be some sort of semantic mistake to claim, for example, that elephants are numbers. Positivists say that ‘Elephants are numbers’ is ‘nonsense’; Wittgenstein says that if you think elephants are numbers – or if you even think they could be – then you haven’t grasped the concept of an elephant (and/or the concept of a number): you don’t know how ‘we’ use the expressions ‘elephant’ or ‘number’. (Who exactly, ‘we’ are is never made clear; but it doesn’t include metaphysicians.) Perhaps you think that makes the difference between Wittgenstein’s sort of semantics and a positivist’s sort mostly nominal. So do I; but don’t tell that to Wittgenstein. Or to Putnam.

Putnam adheres to both Quine’s epistemic holism and Wittgenstein’s semantic holism. But, crucially, where Wittgenstein is a semantic holist, Quine is a semantic nihilist. Putnam apparently wants to add to Wittgenstein’s doctrine of meaning-as-use (perhaps I should say: ‘use instead of meaning’) a very expansive understanding of ‘use’. ‘Pointing out the differences in the way we answer the questions “why is water trickling down the windowpane?” and “why are you cutting the bread?” is clarifying the “grammar” of intentional explanation.’ But that glosses over the question whether intentional explanation is, or can be, itself a species of causal explanation; a question that is not about language or concepts but about the metaphysics of mind. When Descartes asked how the mind makes the body move, he was not making a ‘category mistake’, still less a ‘grammatical’ mistake. He was raising a perfectly sensible question (to which, by the way, the answer is still unknown).

In a nutshell: Wittgenstein thought that meaning is somehow a matter of use; Putnam raises the ante by understanding ‘use’ anthropologically, as the totality of a word’s (concept’s) ‘entanglements’ with how we speak, think and live. Thus Putnam approves of Iris Murdoch’s remark that ‘it is always possible to improve one’s understanding of a concept like “bravery” or “justice”.’ Well, maybe it is; but, at first blush, that appears to conflate improving one’s understanding of the concept of justice with improving one’s understanding of justice, and those seem to be two quite different things. Or maybe they aren’t; maybe there isn’t any such distinction. But you can’t show that there isn’t by appealing to the ‘use theory of meaning’, because there isn’t, in fact, any such theory. A theory of meaning-as-use would have to say, at minimum, which meanings supervene on which uses. Wittgenstein doesn’t even try to do that and, to the best of my knowledge, neither do any of his sisters or cousins or aunts. Worse yet, it appears that taking Wittgenstein’s sort of line about conceptual/linguistic content eventuates in theories about language and thought that, in practice, do nobody any good. To my knowledge, nothing in cognitive psychology or in ‘linguistic semantics’ underwrites an anthropological account of conceptual or linguistic content; and philosophers appeal to it only when they’re doing what they sometimes call ‘philosophical therapy’, i.e. when they want to represent the theories other philosophers hold as insensitive to linguistic usage or as conceptually confused.

I think it’s quite likely that insisting that meanings are inextricably entangled with forms of life is to give up on the possibility of either empirical theories of language or empirical theories of mind. Putnam sometimes seems to agree, but he thinks that’s OK: geology doesn’t translate into, or reduce to physics; why would you expect linguistics or psychology to do so? But this misses the point once again: it conflates Quine’s holism about confirmation (it isn’t just consonance with observation but also relation to basic science that determines a theory’s acceptability) with Wittgenstein’s holism about meaning. ‘Geology doesn’t translate into physics’ is a thesis about semantics, not a thesis about confirmation. ‘This is water’ doesn’t translate as ‘this is H2O’; still, the former is well confirmed if and only if the latter is. (This is, ironically, a point that Putnam is rightly praised for having insisted on.)

There are ironies aplenty for those who like that sort of thing. The ‘linguistic turn’ in philosophy started with positivism, which held that semantics – in particular, verificationist semantics – can save us from metaphysics. Now the tables have turned: ‘analytic’ philosophy holds that semantics – in particular, anthropological semantics – can save metaphysics from us. Normative discourse is ‘entangled’, but science is too; these are, as it were, separate but equal kinds of entanglement. This is this and that is that and everything is quite correct. But, to repeat, that puns on ‘entanglement’. Quine’s kind is epistemic; Wittgenstein’s is semantic. In fact, I detect a soupçon of special pleading: the point of dwelling on the entanglement of the normative (on, as one says, the ‘thickness’ of normative concepts) is to protect it from the depredations of the empirical; namely, by arguing that no sharp fact/value distinction can be drawn in respect of either. This is a predominant theme in Putnam’s book, and in the work of several of the philosophers Putnam approves of (in particular Stanley Cavell and John McDowell). But the trouble with that strategy is that nothing ontological follows from the fact of entanglement. Apollo was much entangled with the forms of life in ancient Athens; but if there isn’t any Apollo, then there isn’t. It is not a defence of theism to point to its entanglement with religious forms of life. As Peter De Vries once brilliantly remarked, it is the final proof of God’s omnipotence that He needn’t exist in order to save us.

There really is a difference between normative matters and matters of empirical fact; both are entangled, but with quite different sorts of things. Observations and data are tangled with theories; norms are tangled with forms of life. I think Hume and Edmund Burke had this more or less right: ethics is intrinsically about us in ways that empirical science isn’t. Consensus about norms rests on – presupposes – convergences of sympathies and sensibilities in ways that empirical consensus doesn’t. In the long run, this may well come down to brute biology: if a lion could speak, it would, perhaps, be possible to converse with it about which empirical beliefs are true; but not, I think, about whether eating people is wrong. ‘C’est mon métier,’ says the lion, with a Gallic shrug.