Jerry Fodor tells Julian Baggini why he thinks Darwin got it wrongJerry has lost it. You can’t take him seriously. He’s gone a bit mad. These aren’t the kinds of comments you’d expect to hear philosophers make about one of their most senior and esteemed colleagues. This is Jerry Fodor they’re talking about: the philosopher of mind and cognitive scientist who has been a driving force behind two of the most influential ideas in his fields of the twentieth century.
First there was the Language of Thought hypothesis, which claimed that the brain has its own language, “mentalese”, complete with both grammar (syntax) and meaning (semantics). Without this, LOT claims, natural languages would not be possible. Then Fodor argued for the modularity of mind, claiming that the different functions of the mind are largely discrete functions. Although there are clearly some senses in which the brain brings things together, its work is done by distinct modules, operating independently of each other.
Both theses were and remain controversial, but few dismissed them as the ravings of a lunatic. So what has Fodor done now to provoke such extraordinarily dismissive reaction form his peers?
The answer is his latest book, What Darwin Got Wrong, co-written with the molecular biologist Massimo Piattelli-Palmarini. His first major public airing of some the ideas of the book came in a piece in October 2007 for the London Review of Books, headlined “Why Pigs Don’t Have Wings”. Reaction was swift and fierce. Over subsequent weeks, critical letters appeared in the LRB by the likes of Simon Blackburn, Jerry Coyne, Daniel Dennett, Philip Kitcher, Steven Rose and Colin Tudge. Fodor replied, the critics responded, and Fodor replied back. Either they were talking past each other or one side is very, very wrong.
So what exactly does Fodor think Darwin got wrong? Several things, it turned out when I met him in London in the autumn, after reading the final manuscript of his already notorious book.
“I’m worried about what seems to me a sort of conceptual incoherence in the foundations of the Darwinian picture.” This is precisely the kind of worry he thinks a philosopher is best equipped to unearth.
“What philosophers are taught to do in graduate school is in some very loose sense, worry about conceptual relations or quite simply about the soundness of arguments. It’s a kind of training which can make a characteristic kind of contribution to a thoroughly empirical programme, which is what you have in the theory of evolution.”
So what does he see as the conceptual problems with Darwinism? First, there are essentially two sets of factors that can contribute to what traits are “selected” by evolution and so are passed on by organisms: those which are internal to the organism (endogenous) and those which are external to it (exogenous). And, to cut to the chase, Fodor believes Darwin made a huge assumption about which mattered most.
“A very approximate but, I think, not entirely misleading account of what Darwin is up to, is that he’s wanting to claim that if you have speciation, if you have evolutionary structures and traits that go from generation to generation, you can divide the factors that are causally at work into two kinds, namely, what’s the structure of the organism? What’s the structure of the environment? Darwin thinks very much that the constraints on evolutionary change are basically contributed by the environment, in that sense, exogenous structures. The organism is thought of as just contributing random re-shuffling of traits to the process.”
This, Fodor argues, is strikingly like the theory of learning developed by the behaviourist psychologist B.F. Skinner in the 1950s, “operant conditioning”.
“Skinner’s worry is, behaviour of organisms is structured, organisms don’t behave at random. How does structure get in? Well, organisms produce behaviour literally at random, but the environment filters it by techniques of reinforcement or something of that sort. This is very much Darwin’s picture as well. So organisms, or more precisely, genotypes, genetic structures, something like that, vary at random from generation to generation and then the environment selects the fit variants from the rest.
“So one kind of question you might want to ask is – and it turned out to be the right question, or a right question, in the Skinnerian case – is it reasonable to think that what the organism contributes to this process is simply random structures? It’s pretty clear by now the answer is no in the case of learning. One of the reasons that organisms learn the things they do and don’t learn other things is that they’re in some sense tuned to do so. There’s a great deal of innate or endogenous structure, so that the behavioural options that the organism as it were proposes to its environments aren’t just a random.
“A question one might want to raise abut Darwin is whether his assumption that organisms essentially produce candidates for evolutionary change at random can be sustained.”
In his book, Fodor is almost mocking of the simplicity of what he calls Darwin and Skinner’s “generate and test” theories. “Each consists of a random generator of traits and a filter over the traits that are so generated. And That’s All.” But isn’t the simplicity of Darwin’s theory part of its beauty and power?
“You make simplicity decisions between theories that are equivalent for explanatory and predictive power. But saying, ‘I’ve got a really simple theory, so you should prefer mine to yours,’ – the reductio of that is, ‘I’ve got a really simple theory: there isn’t anything. You couldn’t get simpler than that.’ The trouble is there’s a lot of data that doesn’t explain.”
So what is the data that Darwin’s “generate and test” theory can’t account for?
“There’s an interesting question that Darwin doesn’t seem to address, and doesn’t seem to be much perplexed by, which is: suppose that’s why you get the kinds of creatures you do. Why don’t you get the kind of creatures that you don’t get? So take my favourite example, why don’t pigs have wings? You might think at first blush that the reason they don’t is that there used to be pigs with wings, roughly speaking, very small ones maybe, and the environment selected against them for some reason or other and now they’re all dead. Of course that’s not what’s happened. What’s happened is that certain sorts of structures aren’t possible for the organism. You just can’t put a wing on a pig and have it work. And that’s got nothing to do with environmental selection. You don’t have to select against the organisms: the structures of the organisms involved simply prohibit them.
“That’s the kind of case where it looks as if the right way to think about things is not that the environment is imposing constraints on organismic structures generated randomly, but simply there are internal properties of the organism which say the environment only has to screen a certain very limited number of possibilities. So that’s where the exogenous/endogenous business comes in.
“In the Darwinian story, all the structures come from the environment essentially. The organism is just a random generator of traits. Well, everybody knows now that that story isn’t true. One kind of reason is that traits may be linked, say genetically, so that if you get one you have to get the other. In that kind of case, there’s no way that the environment can choose between the traits – that’s endogenously fixed. So my guess is that much the same kind of worry that you have with Skinner, that is, that the structure of the creature is being underestimated, is also possibly true in the case of Darwin. And I think one can give examples where that’s very plausible.”
Although there is clearly some truth in this, I wonder if it really does strike a blow to the heart of Darwinism. Take the pig example. Of course, the organism doesn’t just generate any old trait, but that’s just a contingent matter about where it starts from. So in that way, the possibilities of random mutation are limited by endogenous factors. But isn’t that just saying that the organism places limits on which traits are, as it were, put forward to be selected by the environment? Natural selection is still exogenous, even though the traits it selects from are endogenously generated?
Fodor gives a detailed response, but his conclusion sums up its essence. “A Darwinian kind of theory or the Skinnerian kind of theory thinks that to a first approximation, the organism’s constraints are random, and if you want to know why organisms have the kind of property they do, it’s because they’ve encountered environments that select for those kinds of properties. Well it’s simply not an a priori issue which of these two sources of variance is actually the way we work. It’s not that Darwin can comfortably pre-empt one of the possibilities and then say, well the others are just roughly caveats.”
To illustrate his point, and to emphasise the similarities with Skinner, Fodor returns to his real specialism: language learning.
“Here’s one way of looking at how you learn a language, in fact, it’s the standard way: Children babble at random, they’re just random generators of noises, and the environment comes in and pats them on the back when they produce noises of the right kind in the right circumstances. Well that’s one kind of picture. Another kind of picture is, the set of languages you can learn is innately constrained, and in fact all sorts of possibilities you never even try, so the environment doesn’t have to filter against them. That’s not an issue that can be settled a priori.”
Fodor is skeptical that giving endogenous factors their due weight is simply a matter of adding caveats and qualifications to the basic Darwinian picture. “The question is which do you treat as the exceptions and which do you treat as the rules, and that’s not an a priori issue.” He does not, however, dismiss the possibility that Darwinism may be able to absorb his objections. “Well let’s see. It’s one of those put up or shut up things. If you think that what would appear to be counterexamples to the Darwinian story can be assimilated to that story just as add-ons, fine, let’s see it done. The question is whether you can do it, and it turns out to be at least quite possible that you can’t.”
As we push this further, Fodor’s central point increasingly distils itself into a fairly short and clear statement of the problem, which he finds himself forced to repeat, with minor random variations.
“There are two interacting factors, at least two: there’s the structure of the organism and there’s the structure of the environment. Sure, if you hold either constant, the other’s doing all the work, but you can’t hold either constant. You can’t assume that most of the structure is in the environment, most of the structure is in the organism. That’s an empirical issue and Darwin was taking a very strong line on how this empirical issue should be resolved, I suspect wrongly.”
One consequence of taking the importance of the endogenous more seriously is that it challenges the idea that natural selection can operate on individual traits. Worse, Fodor argues that the theory can’t actually tell you which traits have been selected. The root of this problem is, he argues, in the breeding analogies that were so central to Darwin’s development of his idea of natural selection as a kind of breeding without a breeder.
“Darwin was very impressed by the fact, and rightly so, that you can actually introduce variations, selective variations, in creatures by breeding for them. So a picture that Darwin has is, if you want to understand how traits are generated in nature, accept that picture, but don’t allow appeal to the breeder.” But there’s a problem. “The problem is that if you explain why all these sheep have long wool in terms of their being bred for it, then you’re importing into the explanation reference to the breeder’s intentions, his desires, his goals, his beliefs and so on. The question is whether you can take that kind of explanation and just take the breeder out and replace it by a mechanism. And there’s good reason to think you can’t. … There are limits on evolutionary explanation which are masked when you understand it as just the naturalisation of breeding.”
To explain this, Fodor adopts an architectural example from Stephen Jay Gould and Richard Lewontin. “Gould and Lewontin said, look, if you have certain kinds of architectural properties you’ll have other kinds of architectural properties. We don’t have to worry about the details, but they’re linked because of how the geometry and physics and so on work out.” For instance, renaissance architects often used spandrels: curved pieces of masonry used to hold up domes. Although often ornate and decorated, essentially they were not selected for their aesthetic features but were just a necessary by-product of erecting domes.
“But now imagine churches grew in the wild,” says Fodor. “You might wonder, is the point of having the arches to hold the dome up, or is it the point of the dome to keep the arches off the floor? We have an architect to answer that question. Where you don’t, there isn’t an answer, they’re just interrelated features. You have to keep that sort of explanation out of your evolutionary theory and that’s not a trivial problem because in the case of breeding, that kind of explanation is what’s doing the work. So when a lot of Darwinists say, ‘Well it’s straightforward, just start with breeding and take the breeder out,’ what you’re doing is taking out a lot of kinds of causal relations and causal explanations, which it’s not that obvious how you’re going to replace.”
The standard response is that the residue of intentional thinking in evolutionary theory is all just metaphorical.
“That’s fine, but if it’s metaphorical then there’s got to be a literal story. If you say, ‘It’s just a metaphor and as a matter of principle I can’t cash this metaphor,’ then you’re in trouble. What you want, what we all want – there’s no disagreement between the sides on this – is a mechanical theory. And the question is you can’t do it by subtraction, You can’t start out with a non-mechanical theory, subtract out the non-mechanical parts and end up with a mechanical theory. That’s what Darwin thought he was doing, I think. And that doesn’t work. There are fallacies of subtraction. There are some cases where you’ve got A and B, you take out B, and you don’t have A any more.”
What the metaphor contains, but the literal version needs to eliminate, is teleology: a form of explanation that makes goals or objectives central. But isn’t the literal version simply that within these constraints Fodor points out, variation occurs, and if those variations work, then they’re likely to be passed on and, if not, they’re not passed on? There’s no teleology in there.
“There’s no teleology so far, but then you also haven’t said anything about how the thing actually functions. If you get change in an organism, there’s going to be a story about change in the organism, or in the environment or about the two of them which accounts for the change. That’s not a theory of evolution, that’s the principle of sufficient reason.”
For Fodor, this means Darwin’s account is empty, because it doesn’t explain a mechanism by which evolution works. And without specifying the mechanism, the general claims are vacuous.
“It’s not good enough to say there’s some mechanism such that you start out with amoebas and you end up with us. Everybody agrees with that. The question is in this case in the mechanical details. … What you need is an account, as it were step by step, about what the constraints are, what the environmental variables are, and Darwin doesn’t give you that.”
I’m wondering if the charge of emptiness rests on the mistake of thinking that because Darwin’s core insight seems so obvious, it must be truistic, while forgetting that once it didn’t seem obvious at all. The theory could be what Kant called a synthetic a priori: it’s true by definition, but when we appreciate it, we genuinely learn something new. It may seem to be tautological, but in fact, it was a huge step forward in our thinking when we realised that the way to think about organisms and the way that they evolve is that organisms persist if they are fit for the environment.
“Right, but the theory of evolution is not that evolution is a mechanical process. There are people who think that it’s not, but they’re not even in this game. There is no tooth fairy or whatever. It’s a mechanical process. Now let’s see how it works. You can’t decide that by saying [claps hands] ‘I don’t believe in God’, because nobody in this debate is talking about God.”
But not so long ago, not everyone would have accepted that of course it’s a mechanical process. Isn’t Fodor forgetting how far we’ve come on this?
“I don’t think we have come far, or rather, in so far as we’ve come, I don’t think it’s because the theory of evolution is substantive. I think what’s really carrying the explanatory effort is very detailed natural history. If you want to know why the spider behaves the way it does and produces the webs it does, there’s a causal story to tell, a causal history, a series of events such that that series of events terminated in this kind of capacity. But there isn’t any generalisation. The story of the spiders and webs may be totally different from the story about crickets and their chirps. There isn’t any general account of this, except the empty account which says surely it’s going to be a function of interactions between environmental structure and the organismic structure. Of course it is.”
This idea that there may in fact be no general account of how evolution works is one of the more intriguing of Fodor’s suggestions. The idea is that natural history may be just like ordinary history, and that Darwin’s attempt to provide an overarching theory of natural history is as misguided as Marx’s attempt to have a one-size-fits-all theory of human history.
“It’s a very typical nineteenth century phenomenon, to try to get theories of history,” says Fodor. But real history is fine-grained, with much scattering of detail. “One possibility is that all these fine-grained historical sequences that we know about, sequences of cause and effect, have something in common; the other is that no general theory can deal with this, that the stories you can tell are indeed these very fine-grained stories. That’s a possibility. Darwin in his way, and for that matter Skinner, started out by saying, look, there must be some common principles at work here. My impression is that if you try to find the common principles, you have to extract so much from the detailed grain of the causal history that you end up with truisms, like there must be something in the environment, there must be something in the organism or both.”
The assumption Fodor is challenging is that natural history or biology must be more like physics than history. “But suppose it’s more like history than it is like physics? It seems to me that, given the total bankruptcy of the attempt to provide such an actual theory of what these historical changes have in common, it’s a perfectly reasonable possibility that there’s nothing to be said about it.”
That may seem heretical because, at root, biology is about physical systems. But of course, at root, so is human society, but that doesn’t mean we can develop a unifying theory of how it develops. “Exactly. That cuts no ice. Of course they are physical systems and there might even be generalisations at the physical level. But that’s no comfort to Darwin. Darwin wants a biological theory of evolution. He wants a theory where you talk about interactions between, roughly speaking, whole organisms and whole ecologies. The fact that there’s going to be a physicalistic story, explanatory or otherwise, is of no interest here. We knew that when we started.”
Fodor is not advocating intelligent design, let alone creationism. But in questioning some of the central orthodoxies of Darwinian thought, he has set out to slaughter what perhaps should be called secular cows. Of course, critics say that the reason they’re so dismissive is not because of what Fodor is attacking, but because of how badly he’s doing it. Nevertheless, it would not be surprising if defenders of Darwin, tired from their campaign against intelligent design, were a little bit touchy about what they see as a misguided challenge, one that is bound to play into the hands of religious literalists. Surf the web for five minutes and you’ll see that many have already taken Fodor’s criticisms to be evidence that the evolutionary consensus is crumbling and that science will be forced to reinstate God as the cosmic designer one day soon.
“I assume that a lot of people take it for granted that the options are, to put it crudely, random variation or God,” he says. To put it less crudely, there is a worry that if randomness isn’t at the bottom of evolution, then the teleological bogey man pops back up. But, says Fodor, that isn’t the only alternative.
“It doesn’t follow that the possibilities are, on the one hand, choice among randomly generated variables, and on the other hand, mental structures. Physics is imposing constraints on what kind of organisms you have, not because somebody’s designed those constraints, but because organisms are physical systems. So among other things you’ve got to make sure the phenotype you select is one that’s physically possible, and genetically possible, and physiologically possible, and possible in the climate that you have, and so forth and so on. The culmination of all these constraints can do a great deal of work without teleology being involved at all.”
Fodor claims not to care about the hostile reaction he’s provoked. “Biologists just hate this,” he says. “It’s not surprising. My suspicion is that they’re heavily invested in a false theory, just as fifty years ago, many psychologists hated the anti-Skinnerians, because they were heavily invested in learning theory. One biologist said to me, look, in our discipline – in his discipline – progress is made one funeral after another. You just have to wait for the generation that has an investment to die off and the next generation is less biased. I don’t know. A lot of people dislike this theory and there’s no reason why they shouldn’t. They’re being told that a theory that was supposed to be a sort of paradigm of scientific success is in fact bankrupt.”
He believes that when people critique his argument, “Almost always, the reconstruction is just wrong.” He is also dismissive of some of the dismissive comments philosophers have made, such as “It’s hard to take Fodor seriously”.
“Yeah, I find it hard myself. There’s a deep sense in which I don’t care. I want to know what’s wrong with the argument. As soon as you tell me what’s wrong with the arguments – those have to be gone into in detail – then fine, I take it back and apologise humbly. If, on the other hand, all you’ve got to offer is ‘he’s nuts, he denies Darwin,’ I’m not moved.
“Darwin is like NASA. There are lots of people, generations of people now, who are invested in a certain way of proceeding: ‘What do you mean, we don’t want to launch rockets? Of course we want to, they pay me to launch rockets.’ I’m not moved. It may still be silly to launch rockets. Similarly, sure, there’s a consensus view. I think the consensus is false. I could be wrong. I’ve been wrong before. I’m wrong very very frequently. Sometimes I can’t even find the cap to the toothpaste. It doesn’t matter about me. What matters is whether the arguments are any good.
“In this case, I think I’m onto a good thing. I think I have an argument that actually works And under the circumstances, where I think I have an argument that actually works, I’m inclined to insist on attention to the argument.”
There’s an argument going on all right. Whether it’s truly about Fodor’s, however, I really can’t say. A lot of people who know much more about this than I do think Fodor is embarrassingly wrong. Either there’s a complete breakdown of communication or one side is totally mistaken. If you think about, whichever of those options turns out to be correct, there’s something to worry about.