Meaningful words without sense, & other revolutions
Jerry Fodor interviewed by Richard Marshall.
Jerry Fodor is the sharp-shooting killer-app in the philosopher of mind, the Dirty Harry of non-universal modularity, LOT, and representational theories of the mind. In all his books he tells people that the mind works like this and not like that, that being various forms of associationism, behaviourism and semantic holism. Of the theories of McDowell, Churchland, Brandom, Davidson, Dennett and Block et al, he’ll say that they are not to be put aside lightly, but thrown out with great force – on the grounds that you can’t teach an old dogma new tricks. He’s always thinking about whether there can be a science of Tuesdays, what Darwin got wrong, paradoxes lurking in the bushes, whether the materialist theory of mind begs the question, whether he has cooler intellectual ancestors than Chomsky and all the time he’s doing this sensational stuff whilst writing with the wisecracking humour of someone who knows outside of a cat, philosophy is excellent company.Inside of a cat, it gets too crowded. Like, he’s just a totally audacious jive!
3:AM: What made you become a philosopher? You’ve been one for a long time. Has it been what you expected and has the profession changed a lot since you started?
Jerry Fodor: It was because my parents wanted me to be a lawyer. I actually did take a Constitutional Law course in order to please them; but when I’d read some judicial decisions, it seemed to me I could make equally bad arguments without bothering to get a law degree. Hence philosophy since, being very young, I thought of philosophers as particularly rational sorts of people; a view of which faculty meetings soon disabused me.
3:AM: You’re famous for your work on the Language of Thought (LOT). So, this is the thesis that thinking is explained using a theory of mind that describes a modular representational computerised system. You wrote about it in 1975 but then in 2008 wrote LOT2 to update us on how your thinking had moved on in the intervening years. For readers not up on this, could you quickly give a rundown of the basics of how your theory works. Also, a key issue is that you have to come up with a theory that isn’t just running a random string of thought following thought, but rather the theory has to explain how we can have meaning coherence, that is, the thoughts have to be able to be semantically and epistemically causally connected or else we wouldn’t have a theory of ‘making sense’, which is what thinking is. How have you done this?
JF: LOT is a throwing together of ideas, some borrowed from Empiricists, some from Rationalists and some from theories about computers. The book was an attempt to connect these bits and pieces. I arrived at a view that was already much in the air, albeit less than explicitly: the mind gets at the world by representing it; cognitive processes are operations that the mind performs on mental representations. In fact, I think what I recommended taking out was distinctly more original than what I recommended putting in: associationism (which made an utter mystery of the coherence of mental life) and behaviorism (which favored a cavalier rejection of the commonsense view that people do the things they do because want wht hay do and believe what they do. At the time, behaviorism and associationism permeated both psychology and the philosophy of mind, though in somewhat different forms. They still do here and there.
The main change in my views over the (many, many) intervening years is that I now think we should also discard a thesis that most philosophers hold explicitly and that cognitive science has never considered denying: that words, concepts and the like have ‘senses’ (meanings, contents, etc.) as well as referents. Zenon Pylyshyn and I are just finishing a book about why other philosophers and cognitive scientists should abandon it too.
3:AM: So one of the things you think this commits us to is that learning concepts is impossible and that everything is innate, isn’t it? This is what philosophers label an innatist position. There are some LOT theorists who don’t think this, aren’t there? So why do you think the innatist position is plausible? And how could we have an innate concept for Wednesday, say, or doorknob given that our brains have not changed much for about 200,000 years?
JF: There is certainly a paradox lurking somewhere in these bushes. The question, however, is how(/whether) it can be avoided. For a long while, I thought it couldn’t be; indeed, that it followed directly from what practically everybody agrees: that concept learning, if there were such a thing, would have to be some sort of hypothesis formation and confirmation. So, the argument went: learning the concept BACHELOR requires learning that the hypothesis `bachelors are unmarried men’ is true. But this hypothesis already contains the concept BACHELOR, since (by assumption) the concept UNMARRIED MAN is the concept BACHELOR. Conclusion: You can’t learn any concepts that you don’t already have; which is to say that you can’t learn any concepts. Paradox.
I now think there is, after all, a way out, though it requires rejecting the (more or less Fregian) doctrines that concepts and the like are individuated by their ‘senses': What makes BACHELOR the concept that it is, is its having the same sense as the concept UNMARRIED MAN. Frege likewise held that ‘sense determines reference’, so what makes John a bachelor is his being a man and unmarried.
If, however, Frege was wrong about that, the innateness paradox disappears. Suppose, for example, that what determines the reference (the `extension’) of a concept is some sort of causal relation between the (mental) representation that expresses it and the things belonging to its extension. Then, even assuming the hypothesis-formation-and-confirmation story about concept acquisition, the precondition for learning BACHELOR is only that the learner has some concept that is coextensive with BACHELOR. Since this requirement is very weak, there is no residual paradox.
3:AM: How much is this an empirical thesis, and so a branch of the natural sciences, that kind of stimulates its own research programme, and how much of it is rather like a basic map that can then be utilised by different theories of mind, like the Atomic thesis stands in relation to a bunch of theories not all of which are compatible? I mean, apart from people thinking they have a folk psychology, is there any other evidence supporting the existence of mentalese and RTM? Or is it that what you’re really doing is a kind of metaphysics, looking at what must be for any kind of mind, and so is in some sense a priori?
JF: I think the LOT thesis is a priori only in the sense that it’s hard to think of an alternative that seems remotely plausible. I’ll gladly give it up if somebody were to find one. It’s the usual story: most empirical argument is `inference to the best explanation’.
3:AM: The commitments of RTM , particularly compositionality and productivity, leads to disaster for many big philosophical attempts to understand meaning and the mind. So Wittgenstein’s ‘meaning is use’ gets the order of explanation wrong, so is hopeless, and connectionist views of Paul Churchland and other ‘holist’ positions held by esteemed philosophers such as Brandom, McDowell and Block are also in big trouble. Even your old supervisor Hilary Putnam gets it wrong. Why is holism such a bad idea?
JF: It depends on what, exactly, it’s holism about: whether it’s an epistemological or a semantic thesis. I take epistemological holism (essentially, the `Quine-Duhem ‘thesis’) to be correct in spirit and historically accurate to boot. Scientific method assumes that evidence for the (dis)confirmation of an empirical theory can come from anywhere including not just ‘observations’ but also considerations of simplicity, economy, and other such `global’ properties of belief systems. Anything can be given up (though, of course, you can’t give up everything at once, skeptics to the contrary notwithstanding). This sort of epistemological holism is an essential corrective to the hyper-empiricism that was, for a while, the main obstacle to progress in the behavioural sciences (which turn out, on reconsideration, to be not about behaviour but rather about its mental causes).
But holism makes hopeless semantics; it simply can’t be true that the content of each of one’s beliefs depends on the content of each of the others. Since the reasons are familiar, I won’t review them here; suffice it that, if there is to be a belief-desire psychology at all, it must leave room for the piecemeal alterations of beliefs and desires. This is to say that epistemological holism (which I take to be more or less true) is incompatible with semantic holism (which I take to be false root and branch). It is therefore unsurprising that semantic holists (Quine, Davidson, Dennett), the Churchlands, many others (including, alas, even Putnam in some of his moods) refuse to take `propositional attitude’ psychology to be ontologically serious. (It’s sometimes suggested, as a way out, that semantics should endorse some sort of partial holism. If that’s not an oxymoron, what would be?) The muddling of epistemological with semantic and ontological theses was, of course, pursued as a matter of policy by empiricists; they thought they could refute skepticism by showing that empirical claims translate to claims about one’s experience, which they supposed to be incorrigible. In the event, this project did not succeed. I wonder why anyone still cares whether it does?
3:AM: What’s neat about your thesis is that it takes a strong naturalistic view of the mind and also endorses folk psychological theories such as beliefs and desires and so on. This is kind of essential for you, isn’t it, because if we don’t start with the big problem – ie how could anything material be conscious?’ – then we’re ducking the actual problem. Is this right? Along the way, Dan Dennett’s arguments about the realism of these mental representations gets clobbered don’t they?
JF: I’m increasingly unconvinced that the fuss about consciousness being the `hard problem’ has the stick by the right end. Consciousness is itself an intensional state; you can’t be just conscious tout court; if you are conscious at all, there must be something you are conscious of; and this `of’ is intensional and exhibits the usual substitution ambiguities. This is to say that the familiar claim (see eg. Searle) that materialist theories of intentionality beg the problem of consciousness gets things back to front. Still, even if, as I rather suppose, consciousness turns out to be more or less the same thing as attention, the questions about sensory content (`qualia’ ) have to be faced; but it’s at best unclear that they bear much on theories of cognition.
3:AM: One of the targets of your work initially was Skinner and Watson’s behaviourism. It was the mechanism of sssociationism that they used that had to be replaced because it was too weak assed to work. Yet there are still behaviourists about, aren’t there? So how come?
JF: I don’t know why there are still behaviourists. The explanation may lie in the maxim that, in science, progress is made one funeral at a time. Behaviourism is verificationism applied to psychological explanation; and verificationism is the philosophy of science that’s generally taken for granted in Psych 101. A little philosophy is a dangerous thing and inclines a freshman to behaviourism.
3:AM: Your very cool book on Hume was great on so many levels, but one was that you claimed not to have read any Hume before you set off. (And didn’t read so much by the time you’d finished either). This is often leveled against you as a criticism, that you’re oblivious to the history of philosophy and this means that you’re not able to join in with a philosophical conversation that might have been going for centuries. So is this a deliberate strategy of yours, to work it out for yourself from first principles, and what do you say to those who say you’d gain through reading historical predecessors?
JF: This does surprise me since I think of my views as embedded in a tradition of `representational’ theories of mind that includes, among others: empiricists (notably Hume), rationalists (notably Frege and Kant), certain of the scholastics (notably Ockham) and Aristotle; in fact, practically everybody of major philosophical significance prior to the 20th Century. (When I was still at MIT, I had occasional arguments with Chomsky about which of our ideas had been anticipated first. I won. Though Chomsky is, understandably, proud of his indebtedness to Descartes and Port Royale, I trace my ancestry to the 5th Century BC.) My major departures from this tradition are: urging that it should keep its semantics and ontology clear from its epistemology, and that it cut its traditional entanglement with associationism in favour of a computational account of mental processes. On the other hand, it’s true that I don’t care much about the history of philosophy when it is practiced as the explication of text; and I regard the `If only he’d tried a bit harder, Aristotle might have been Quine’ school of historiography as a giggle.
3:AM: And linked to this I guess is the observation that you’re a great fan of Wagner but have never engaged with philosophers like Hegel and Heidegger or anyone from the so-called continental tradition. There’s been some discussion recently about whether the divide between analytic and continental philosophers identifies anything significant. What’s your take on this? Is it that you just think there’s nothing of value in the stuff from what you’ve heard, or is it just that you’ve already got enough to be thinking about?
JF: As a matter of principle, I refuse to read philosophers who write that badly; anyhow, for what it’s worth, I think that being reconciled with `analytic philosophy’ is a blessing that philosophers are well advised to forego.
3:AM: Your recent incursion into the Darwin wars caught some people by surprise but really it shouldn’t have. You basically were arguing the same case against a version of natural selection that you used against behaviourism weren’t you? The pesky intension/intention distinction was something you realised lay at the heart of a hidden piece of fallacious thinking in the Darwinian camp and although disguised it was a potentially devastating element. Can you say why the mechanism for natural selection used by Darwinists is flawed?
JF: That’s easy. Darwinism doesn’t have a mechanism for natural selection; in particular, the `theory of natural selection’ doesn’t provide one. That was the main theme of Piatelli and my book What Darwin Got Wrong. Darwin, like Skinner, is a `black box’ theorist; both insist that the non-random variables in explanations of biological traits are environmental .In consequence, Darwin is left with the hopeless problem, of (which the `linkage’ of traits is a parade example) how environmental variables could account for the effects of a creatures internal organisation in determining its phenotype. Hence the familiar loose talk about `Mother Nature’s role in guiding trait selection. Dawkins (among others) keeps assuring us that such talk is metaphorical. But he never does explain how to cash the metaphor. Skinner’s problem is exactly the same, except he’s interested not in innate phenotypes but in acquired `behavioral repertoires’.
3:AM: The backwash was unusually intemperate and at times personal. I got the sense that the scientists were pretty upset that you’d called them out. Perhaps your style was something they got all prim about – you’re always very funny in the way you ridicule philosophical ideas you think are bogus and never pull punches, but perhaps the science lot aren’t used to that kind of knock-about mode. Clearly you weren’t part of the usual anti-Darwinian creationist crowd but I wondered if they thought that they couldn’t afford to lose any ground at all because of the extra-curricular politics surrounding Darwin in the USA. Were you shocked by the nature of the criticism, or is it what you get a lot of the time?
JF: I wasn’t surprised that the biologists missed the point; for better or worse, evolutionary biology is now a largely statistical science, so it’s practitioners generally don’t think much about the explanatory adequacy of other kinds of empirical theories. But that the philosophers missed the point too struck me as really shocking; contemplating issues about explanatory adequacy is a lot of what they’re supposed to do for a living. It should be kept in mind, however, both in biology and in the philosophy of biology, a lot of careers, a lot of careers are built on firm adherence to (Neo-) Darwinism. So it’s hardly surprising that serious criticism of The Theory of Natural Selection (as distinct from the Creationist kind) provokes spasms. God save us from true believers on either side.
3:AM: It was Alex Rosenberg, a philosopher, who seemed to mount the best counter-attack on your theory rather than any of the biologists. So what do you say to his criticism of your position? It seems that were his idea to work then there would be implications for the arguments against associationalism. Is that right?
JF: Alex says (rightly) that selection for/against are contraries, not contradictories. (There are neutral traits as well as adaptive and anti-adaptive ones). I think Alec thinks we miss this logical point in What Darwin Got Wrong, but we didn’t. Since it is tautological that there can’t be selection for/against a neutral trait, it follows that, if there is selection at all, then it is selection for a trait iff it isn’t selection against it. Still, let’s assume, for the sake of argument that whiteness wasn’t selected for in polar bears. What, in that case, was selected against? Being pink? Being green with orange stripes? Do Darwinists believe that there used to be green polar bears with orange stripes, but they all got eaten up by predators? If not, what does Alex think is gained by rejecting selection for in favour of selection against?
3:AM: You don’t agree with people like Steven Pinker that your LOT thesis is the final word on thought. He thinks modularity potentially explains everything but you are less optimistic. What’s the reason for your disagreement?
JF: I take it seriously that the cognitive psychology of belief formation should be continuous with the philosophy of theory confirmation in empirical science. Science works so well because it’s just the application of rational practice to questions about how the world works (as opposed to, say, questions about why the family car has stopped running). Among the most important insights in the theory of confirmation in the last hundred years or so is the `Quine-Duhem’ thesis): the (dis)confirming evidence for an empirical theory can come (not only from experiments and observations but) from anywhere in the network of received scientific beliefs; including, in particular, global considerations like the overall simplicity, conservatism, plausibility etc. of the theory under evaluation (including, notably, whether there is a plausible alternative). This is to say that there are criteria of evaluation at work in scientific confirmation that are global rather than modular; hence there are no principled limitations on what may turn out to be relevant to what.
The trouble with endorsing an entirely modular theory in psychology is that you would then be forced to not take global constraints on thought into account. It’s worth noting that`expert systems’ (i.e. modular attempts) to simulate human problem solving at large, quite strikingly don’t work. (if you don’t believe that, try explaining to the computer on the other end of the line just why it is that you want to return the article that you ordered last week.) It’s an open secret that the reason that computers can’t pass the Turing test is that they lack `common sense’; and `common sense’ is nonmodular by definition. Cognitive psychology ignores such patent truths at its peril.
3:AM: Philosophers rarely change their minds. What would it take for you to abandon your approach to the mind? Could Chalmers or McGuinn or Dennett or Churchland ever find something to convince you that you were in the wrong ball park?
JF: All that’s required is that someone suggest a serious (hence not verificationist and not reductionist) theory of cognition that is reasonably sensitive to empirical data. I’m betting that is not going to happen. Any takers?
3:AM: You’re known as an opera man. What’s the appeal for you and what’s your favourite? If you were to recommend a recording, what would you offer?
JF: Usually my favourite opera is the last one I’ve listened to; but if I must choose, I think I’d opt for Debussy’s Pelleas. (Last week, it would have been Gluck’s Orpheus; or maybe Verdi’s Falstaff). There is a fine recording of Pelleas conducted by the always-reliable Haitink, with von Otter singing Mellisande. (The competition has von
Karajan conducting and von Stade (for whom I have, for decades, been consumed by an asymmetric passion). Still, the Haitink by a close call.
3:AM: And to the philosophically curious readers here at 3:AM Magazine, are there five books could recommend that would help us further understand this area of philosophy?
JF: Five is a lot; but here’s four: any of Chomsky’s nontechnical/nonpolitical books (Language and Mind is a good place to start).The anthology Concepts: the Core Readings (Margolis and Laurence editors). Things and Places (Zenon Pylyshyn). Contemporary Philosophy of Mind (Georges Rey).
ABOUT THE INTERVIEWER
Richard Marshall is still biding his time.
First published in 3:AM Magazine: Friday, October 19th, 2012.