Hilary Putnam’s latest book collects two series of his lectures with two chapters of ‘afterwords’. Subsidiary topics go by faster than my eye was able to follow, but the main concerns are: ‘representational’ theories of perception, and ‘identity’ theories of the mind/body relation.
The treatment of the mind/body issues, though the dialectic is often intricate, is quite summary: neither philosophical claims for mind/body identity, nor the denials of such claims, are ‘intelligible’. That’s because ‘the notion of identity has not been given any sense in this context’ (sic). I’m not going to discuss this part of Putnam’s book because, truly, I haven’t a clue what it is to give a sense to a notion; the notion of giving a sense to a notion hasn’t been given a sense, either in this context or, as far as I know, in any other. (I’ve been told that senses are sometimes given to concepts at Oxford after the gates close to visitors; but that may be a leg-pull.) Nor am I clear what you’re supposed to do with a notion once a sense has been given it. There used to be a story according to which empirical inquiry works by first giving senses to notions, and then scouting around for something for the notions to apply to. It was a sort of sophisticated version of Nanny’s advice always to start by defining one’s terms. But I wouldn’t have thought that anybody believes that story any more; indeed, lots of us learned not to believe it from Putnam. What a strange business philosophy is.
Putnam’s discussion of representation and perception is, however, less abrupt. I’ll concentrate on that.
Common sense observes that typical states of the cognitive mind are representational. The belief that the door is open represents the door as being open, and is thus false if the door is closed; so far, credulity is unlikely to be strained. But there is a venerable project, in both philosophy and psychology, of trying to construct a general theory of the mind grounded in this common sense observation. Such a theory would construe assignments of content to mental states as hard facts in good standing, and mental processes (thinking, perceiving and the like) as causal interactions among such content-bearing states. The goal, in short, is to make predictive and explanatory science out of commonsense mentalism. I don’t know who first thought up this research programme; quite possibly he lived in a cave. But its avatars in current cognitive science are directly indebted to a philosophical tradition that has been continuous at least since the 17th century. Occasional flirtations with behaviourism not-withstanding, representational theories are the mainstream tradition of Western thinking about the mind.
Putnam is not, however, an enthusiast for this mainstream tradition.
There is no reason to think that the idea of such a psychological theory . . . is anything but utopian . . . The possibility of an ideal psychological theory of this sort is nothing more than a ‘we know not what’ . . . One hears a lot of talk about cognitive science nowadays, but one needs to distinguish between the putting forward of a scientific theory, or the flourishing of a scientific discipline with well-defined questions, and the proffering of promissory notes for possible theories that one does not even in principle know how to redeem.
And so forth at some length and with perceptible rancour.
I don’t, myself, think that cognitive science is more in need of philosophical defence than is, say, ornithology. The warrant of the enterprise, in both cases, is not that the questions pursued are ‘well-defined’, but the truths that are discovered in the course of pursuing them. For someone who repeatedly claims to be a pragmatist, Putnam is strangely insensitive to the methodological truism that success is what justifies. If he really wants to mount a respectable attack on cognitive science (or ornithology for that matter) he has to show that the truths it claims to have discovered are spurious, or that they can be explained just as well without appeals to representational mental states and processes. That, however, would be hard work, and Putnam doesn’t even try. All of his arguments are a priori.
A representational theory of perception (hereinafter RTP) claims that the cognitive relation between a creature and whatever it perceives is typically causally mediated by mental representations of that object; ‘the disaster is the idea that there has to be an interface between our cognitive powers and the external world.’ Whereas the undisastrous alternative, according to Putnam, is the ‘direct’ or ‘commonsense’ realism about perception that he shares with such of his ‘philosophical heroes’ as Dewey, James (W.; certainly not H.), Peirce, J.L. Austin, John McDowell, Husserl (with reservations) and, of course, Wittgenstein. Disappointingly, however, neither Putnam nor anybody else in his direct realist pantheon is prepared actually to offer an account of how perception works. Rather, ‘in my opinion, “direct realism” is best thought of not as a theory of perception but as a denial of the necessity for and the explanatory value of positing “internal representations” in thought and perception.’ (Elsewhere, Putnam says that the defining question is whether, in perception, ‘our cognitive powers . . . reach all the way out to the objects themselves.’ But this isn’t much help, what with the notion of ‘reaching all the way out’ not having been given a sense in this context. Nor is it clear why mentally representing the objects of perception isn’t to count as one of our cognitive powers.)
In fact, it’s hard to see how one is even to begin to think about the mind without thinking a lot about how it represents things. This is a point that’s been made repeatedly since Descartes and, as far as I can tell, it still stands. Consider, for example, perceptual error. It’s surely natural to say that misperception is a failure of match between how the world is and how the mind of the perceiver represents it. Putnam prefers a ‘disjunctive’ account, according to which ‘seeing an illusory pink elephant’ is understood as ‘seeming to see a pink elephant’. But he doesn’t tell us how ‘seeming to see’ is itself to be construed unless in terms of representing. (There’s a direct realist story, owing to Austin, which construes seeming to see a such-and-such as being disposed to believe that one is seeing a such-and-such. But it’s a notorious non-starter. When I find myself seeming to see pink elephants, what I am disposed to believe is that I’m stewed.)
Or consider thinking. Is thinking about Granny also not a representational state but a direct connection that reaches all the way out to the Old Dear? But how could it be, since I can think of her when I’m here in New York and she is in Ohio? Or what about remembering? How can I be in an unmediated relation to Ebbets Field (alas, long since demolished); or to my erstwhile dentist, who passed away a year ago in August? A last-ditch position might hold that direct realism is a story that applies only to the explanation of veridical perception. But if veridical perceiving is really that different from everything else the mind does, how does one remember, or think about, what one has veridically perceived? Thomas Reid, an 18th-century exponent of direct realism, says that ‘the knowledge which I have of things past, by my memory, seems to me as unaccountable as an immediate knowledge would be of things to come; and I can give no reason why I should have the one and not the other, but that such is the will of my Maker.’ There’s nothing I admire more in philosophy than a well-bitten bullet. Reid is exactly right; but for the notion of mental representation, much of what the mind does would be miraculous. The miracle theory of mind is the natural alternative to the representational theory of mind. Putnam gets no closer to facing this than not bothering to mention it.
In fact, there is no direct realist theory of perception (or of anything else that’s mental). So, willy-nilly, Putnam’s polemic turns on whether the indirect, representational theory of perception is intelligible; and, if it is, whether it is defensible. At this point, there’s need for a little background.
‘The’ representational theory of perception, like ‘the’ representational theory of mind itself, has had a long history and undergone many revisions and extensions along the way. Putnam is, by and large, quite insensitive to the differences between the various ways of running representational theories, but they matter to the merits of his arguments. In particular, the sort of RTPs that cognitive scientists endorse have shed a feature that traditional versions of the doctrine, rationalist and empiricist, invariably took for granted: that RTP must provide not just a psychology of perception but an epistemology, too. That’s to say, the same theory about representational mental states and their causal interactions that answers questions like ‘how come, when a piano is in front of me, and my eyes are open, and the lights are on, and the room isn’t filled with smoke . . . blah, blah, blah . . . I come to believe that I am seeing a piano?’ is also supposed to answer questions like ‘how come, when a piano is in front of me, and my eyes are open, and the room isn’t filled with smoke . . . blah, blah, blah . . . I am justified in believing that I am seeing a piano?’ As it turns out, a lot of what was wrong with traditional RTPs was the result of their attempt to bear this double burden.
Suppose that my (veridical) perception of a piano is mediated by an unconscious causal sequence of mental representations that specify (e.g.) its apparent shape, colour, size and so forth. (The early members of such sequences might be an automatic consequence of the way that light bouncing off the piano stimulates my sensory receptors; later members might be inferred from these sensory representations together with my background beliefs.) Putnam apparently thinks this sort of theorising is ‘science fiction’, but there’s no indication that that is an informed opinion. Has he read (for example) David Marr’s Vision (1982)? Has he heard that there are honest-to-God theorems about the inference from one-dimensional retinal representations to representations of two-dimensional form; and from representations of two-dimensional form (and motion) to representations of three-dimensional form? And that these theorems predict a wide variety of previously unexplained perceptual phenomena? If Putnam does know about Marr’s kind of work and thinks it mistaken or confused, why on earth doesn’t he produce the arguments which show that it is? And if he doesn’t, what claim has his view of RTPs to the serious attention of busy people?
But I digress. My point is that transformations of mental representations might make an essential contribution to the causal process of perceptual belief fixation even if the perceiver is unaware of them. But (arguably) they can’t contribute to the justification of perceptual belief unless the perceiver is able to cite them. This being so, and given the assumption that RTPs must do duty as epistemological theories, traditional versions required that we have conscious access to the mental representations that mediate our perceptual accomplishments. The (alleged) consciously accessible sensory representations that serve as the premises of perceptual inferences are what epistemologists used to call ‘sense data’. Since sense data must be available to perception prior to the inferences that they ground, and since they are mental by assumption, it often seemed to traditional RTPs that only mental things could be perceived directly. Scepticism followed (as in Hume). There are, to be sure, plenty of caveats one might enter to this line of thought. But it’s quite true that there are ways of running RTPs that can lead, and have led, either to the view that ‘strictly speaking’ nobody ever saw a piano, or to the view that ‘strictly speaking’ pianos are mental.
Having abandoned its claim to be epistemology, RTP doesn’t need to say anything like that now. We have perceptual knowledge of things in the world: inter alia, pianos, tables, chairs, trees and their various surface colours. Causal processes involving mental representations mediate these perceptual relations, but you don’t (typically) perceive the representations themselves either directly or otherwise. Putnam is quite aware that this is the way that RTPs are usually run these days, but he describes the concession to Realism as merely terminological; it still doesn’t make perception direct.
Well, that depends on what ‘direct’ means, and I fear it hasn’t been given a sense in the present context. Putnam may be supposing that, in their (oh, Lord!) ‘ordinary uses’, ‘direct’ and ‘mediated’ name mutually exclusive ways that a connection can be. But there really isn’t any reason to suppose that. I can step on your toe, even though both of our shoes mediate the transaction. Likewise, I often talk to my wife on the phone. The connection we make is certainly ‘mediated’ in all sorts of ways; still, it is my wife that I talk to, and it is her voice that I hear. (For what it’s worth, the phone company calls the service it provides ‘direct dialling’.) Any number can play the ordinary language game, but nobody ever wins.
Putnam is one of our best philosophers, and he is terrifyingly smart. He simply can’t believe that the sorts of arguments he gives in The Threefold Cord are serious reasons to abandon the best – not to say the only – way we know to theorise about the mind. So what is going on in his book? In fact, it’s a skirmish in a much larger battle. The fundamental issue isn’t direct perception or even mental representation. It’s whether meaning is a ‘natural kind’.
There are, as usual, two sorts of philosophers. One sort thinks that meaning is a not-intrinsically-problematic property that some things have and other things don’t. Among the things that generally have it are, for example: words, sentences, beliefs and thoughts. Among the things that generally don’t have it are, for example: rocks, clouds, cats and grass. What exactly meaning is, is a hard problem. But we’re working on it, and some bits do seem clear: for example, that things that have meanings do so in virtue of their relational properties. (Nothing is intrinsically meaningful, just as no one is intrinsically a mother.) Maybe more about meaning will get clear as the inquiry proceeds. Perhaps it will turn out that, among the things that have meanings, are states of the nervous systems of perceivers, thinkers and other creatures with minds. If so, that’s what mental representations are.
Philosophers friendly to RTPs, and to cognitive science in general, hope that that’s the way it will turn out. Call that kind of philosopher a ‘meaning realist’. His defining property is that he thinks there are facts about meaning, and that they are suitable objects of scientific inquiry.
All the philosophers in Putnam’s pantheon (and lots of others, including Davidson, Quine, Dennett and more or less the entire populations of France and Germany) don’t believe any of that. Rather, they think that meaning arises from our practices of interpretation, so that there no more needs to be a single right answer to ‘what does “pinochle” mean?’ than there is to ‘what does Hamlet mean?’ Our ‘interpretative practices’, in turn, are paradigmatically those involved in our use of language, so it is only for that context that semantic notions have been given a sense. The supposition that brain states might have meanings therefore lacks a sense and amounts to ‘nothing more than a “we know not what.”’ For this sort of reason meaning is not a suitable object of scientific inquiry (and for several other reasons, too, including its alleged context-dependence and holism, and the heterogeneous character of the uses of language to which meanings are assigned; Putnam takes all of that for granted, though quite probably he shouldn’t.) Trying for a science of meaning would be silly; like trying for a science of games. Or of Tuesdays. Call this kind of philosopher a ‘Wittgensteinian’.
It’s as close as Putnam’s book gets to having a saving grace that it almost sees the clash between meaning realists and Wittgensteinians as its real topic. (Only ‘almost’ because, very unfortunately, Putnam consistently conflates the argument over realism about meaning with a quite different argument over reductionism about meaning. In fact, these issues are largely independent. Most meaning realists are anti-reductionists, though you’d never guess that from Putnam’s exposition.) Well, who’s right about meaning realism is a wide open question; among the deepest, I think, that philosophers have thus far learned how to ask. Putnam isn’t the first to try to settle it a priori, but the a priori arguments don’t convince. Here again, one could wish that Putnam were the pragmatist he claims to be. Pragmatists don’t much hold with a priori evaluations of what purport to be programmes of empirical research. It might after all turn out that there are interesting things to be said about games; or even about Tuesdays; or even about perception. Wittgenstein did, once, get something right. He said: ‘Don’t think, look.’