‘How do you get to Carnegie Hall?’ ‘Practice, practice.’ Here’s a different way: start anywhere you like and take a step at random. If it’s a step in the right direction, I’ll say ‘warmer’; in which case repeat the process from your new position. If I say ‘colder’, go back a step and repeat from there. This is a kind of procedure that they call ‘hill climbing’ in the computer-learning trade (hence, I suppose, the title of Richard Dawkins’s new book). It’s guaranteed to get you where you’re going so long as the distance between is finite. (And so long as there are no insurmountable obstacles or ‘local maxima’ in the way: nothing is perfect.)
Hill climbing is often the theory of choice when a scientist’s problem is to explain how something got to somewhere you wouldn’t otherwise have expected it to be. That’s in part because it’s such an abstract and general sort of theory. All it requires is a source of random variation, a filter to select among the variants, and some ‘memory’ mechanism to ensure that the selected variations accumulate. In all other respects, you’re free to adapt it to whatever is worrying you.
For example, the ‘here’ and ‘there’ needn’t be spatially defined. They might be, respectively, the undifferentiated primal, protoplasmic slime and the vast, intricate proliferation of species of organisms that now obtains. Darwinism (or, anyhow, the adaptationist part of Darwinism) is a hill-climbing account of the phenomenon of speciation: genetic mutation takes the place of your random milling about, the inherited genetic endowments of successive generations of organisms correspond to the succession of positions that you occupy between here and Carnegie; and, instead of my shaping your path with gentle verbal cues, natural selection determines the direction of evolution by killing off mutations that happen to reduce organic fitness.
It all sounds pretty plausible. It might even be true. But the fact that a hill-climbing model could, in mathematical principle, find a way from where you started off to where you ended up, doesn’t at all imply that you (or your species) actually got there that way. I could just have told you where Carnegie Hall is, and you could have got there by following my instructions (‘Directed Evolution’). Or I could have picked you up and carried you there (‘Interventionism’). Or you could have started out at Carnegie Hall, in which case the question wouldn’t have arisen (‘Preformation’). No doubt, other possibilities will occur to you. In the present case, one might reasonably wonder whether we did, after all, get to be us in the way that Darwinian adaptationism says that we did; it’s reasonable to want to see the evidence.
Especially so because the scientific success of the hill-climbing style of explanation has often been underwhelming in other areas where it has been tried. Classical economics (by which Darwin was apparently much influenced) wanted to use it to account for the organisation of markets. In a system of exchange where gizmos are produced with randomly differing efficiencies, canny consumers will filter for the gizmos that are best and cheapest. Gizmos that are too expensive to buy, or too cheap to sell at a profit, will be screened out automatically. Eventually an equilibrium will be achieved that comports, as well as can be, with all the interests involved.
That’s a nice story, too. But in the event, what often happens is that the big gizmo-makers buy out the little gizmo-makers and suppress their patents. If there’s still more than one gizmo-maker left in the field, they compete marginally by painting their gizmos bright colours, or paying some airhead to praise them on television. The evolution of gizmos therefore grinds to a halt. Whichever producer a consumer decides to buy his gizmos from, he finds that they don’t work, or don’t last, or cost too much.
For another example, consider a version of hill-climbing theory that used to be popular in psychology. How does behaviour get organised? How, for example, do you get from being a babbling baby to being a fluent speaker of English? Here’s how, according to B.F. Skinner and the tradition of ‘reinforcement theory’: babbling is vocal behaviour that’s produced at random. When you happen to make a noise that sounds sort of like the local dialect, ‘society’ reinforces you; and your propensity to make that sort of sound (or better, your propensity to make that sort of sound in those sorts of circumstances) increases correspondingly. Keep it up and soon you’ll be able to say ‘Carnegie Hall’ or ‘Jascha Heifetz’ or any other of the innumerable things that being able to speak English allows you to. Skinner used to complain, to people who didn’t like his story about learning, that he was just doing for the formation of behaviour what Darwin did for the formation of species. There was, I think, some justice in that complaint, but it’s an argument that cuts both ways.
In any event, language learning doesn’t work by Skinnerian hill climbing: language learners don’t make their errors at random in the course of the acquisition process. Rather, as Noam Chomsky famously pointed out, the grammatical and phonological hypotheses about language structure that children think to try out are sharply endogenously constrained. ‘Who Mummy love?’ is recognisably baby talk, but ‘love Mummy who?’ is not; it just isn’t the kind of thing children say in the course of acquiring English. Ergo, it’s not a kind of thing that society is required to filter out in the course of ‘shaping’ the child’s verbal behaviour. But why isn’t it if children are hill climbing towards the mastery of English grammar, and making mistakes at random as they go?
So there are at least two cases where, pretty clearly, applications of hill-climbing models tell less than all there is to be told about how a system gets organised. These examples have something strikingly in common. Hill climbing wants a random source of candidates to filter; but, in the market case and the language acquisition case, it appears that there are ‘hidden constraints’ on what candidates for filtering ever get proposed. The market doesn’t produce its gizmos at random, and the child doesn’t produce its verbalisations at random either. The market is inhibited by restraint of trade, the child by (quite possibly innate) conditions on the kinds of language that human beings are able to learn and use. No doubt, in both cases, there is some residual random variability, and correspondingly, some filtering which serves to smooth rough edges; so hill climbing gets a sort of vindication. But it’s pyrrhic if, as practitioners in economics and psycholinguistics tend to suppose these days, the hidden constraints are doing most of the work.
Clearly the track record of hill-climbing explanations outside biology isn’t what you’d call impeccable. What, then, about speciation? Nobody with any sense doubts that adaptation is part of the truth about evolution; but are there, maybe, ‘hidden constraints’ at work here, too? Or is the environmental filtering of random mutation most of what there is to how creatures evolve? Nobody loses absolutely all of the time. Maybe speciation is where hill climbing wins.
There is, in fact, currently something of a storm over just this issue, the vehemence of which Dawkins’s book is much too inclined to understate. Palaeontologists, since Darwin’s own time, have often complained about what looks, from an anti-adaptationist perspective, like an embarrassing lack of smooth gradations from species to species in the geological record. Maybe evolution gets from place to place by relatively big jumps (‘saltations’), the intermediate options being ruled out by hidden constraints on what biological forms are possible. Something like this idea is at the heart of the current enthusiasm for evolution by ‘punctate equilibria’. If you want to get to Carnegie, don’t bother with exploring the intermediate loci: take a jet.
Dawkins doesn’t make much of this sort of option; he’s too busy assuring his lay audience that everything is perfectly fine chez classical adaptationism. Issues about evolution have become so politicised that a popularising biologist must be tempted to make a policy of pas devant les enfants. Dawkins has succumbed a bit to this temptation. It’s a disservice to the reader, who thereby misses much of the fun. For a corrective, try Niles Eldredge’s 1995 book Reinventing Darwin.
If classical adaptationism is true, then, at a minimum, the route from species A to its successor species B must be composed of viable intermediate forms which are of generally increasing fitness; there must be, in Dawkins’s metaphor, smooth gradients leading up the hill that adaptation climbs. Much of his book is devoted to an (admirable) attempt to make the case that there could have been such viable intermediaries in the evolution of vision and of winged flight. Dawkins doesn’t (and shouldn’t) claim that any of these intermediate creatures are known to have existed. But he is pretty convincing that they might have, for all that biochemistry, physiology, embryology and computer modelling have to tell us. The naive objection to adaptationism is that random mutation couldn’t have made anything as intricate as an eye. Dawkins’s answer is that, sure it could; there’s a physiologically possible path from bare sensitivity to light to the kind of visual system that we’ve got, and overall fitness would plausibly increase and accumulate as evolution traverses the path. It appears, in fact, that there may be several such paths; eyes have been independently reinvented many times in the course of evolution.
It is, however, one thing to show that evolution might have been mostly adaptation; it is another thing to show that it actually was. Many readers may be disappointed that Dawkins doesn’t discuss the evolution of the piece of biology that they are likely to have most at heart: namely, human cognitive capacities. This is, on anybody’s story, one of the places where the apparent lack of intermediate forms looks most glaring. Cognition is too soft to leave a palaeontological record. And, pace sentimental propaganda on behalf of chimpanzees and dolphins, there aren’t any types of creature currently around whose cognitive capacities look even remotely similar to ours. Moreover, there is a prima facie plausible argument that hidden constraints might play a special role in the evolution of a creature’s psychological traits as opposed, say, to the evolution of its bodily form.
It’s truistic that natural selection acts to filter genetic variation only insofar as the latter is expressed by corresponding alterations of a creature’s relatively large-scale structure (alterations, for example, of the organs that mediate its internal economy or its environmental interactions). The slogan is: genetic variants are selected for their phenotypic fitness. This holds, of course, for the case of nervous systems, too: genetic endowments build neurological structures which natural selection accepts or rejects as it sees fit. Suppose that there is indeed relatively unsystematic variation not only in the genetic determinants of neurological structure, but also in the corresponding neurological phenotypes. Still, brain structures themselves are selected for the fitness of the psychological capacities that they support. They’re selected, one might say, not for their form but for their function. And nothing general – I mean nothing general – is known about the processes by which neurological alterations can occasion changes of psychological capacities.
Gradually lengthening the giraffe’s neck should gradually increase its reach; that seems sufficiently transparent. But it’s wide open what tumultuous saltations gradual increase in (as it might be) brain size or the density of neural connections might cause in the evolving cognitive capacities of a species. The upshot is that even if we knew for sure that both genetic endowments and neurological phenotypes vary in a way that is more or less random and incremental, as adaptationism requires, it wouldn’t begin to follow that the variation of psychological traits or capacities is random and incremental, too. As things now stand, it’s perfectly possible that unsystematic genetic variation results in correspondingly unsystematic alteration of neurophysiological phenotypes; but that the consequent psychological effects are neither incremental nor continuous. For all anybody knows, our minds could have got here largely at a leap even if our brains did not. In fact, insofar as there is any evidence at all, it seems to suggest that reading brain structures onto mental capacities must substantially amplify neurological discontinuities. Our brains are, by any gross measure, physiologically quite similar to those of creatures whose minds are nonetheless, by any gross measure, unimaginably less intelligent.
Dawkins likes to ‘insist ... that wherever in nature there is a sufficiently powerful illusion of good design for some purpose, natural selection is the only known mechanism that can account for it.’ He’s right, I think, but this is another of those two-edged swords. The conclusion might be that adaptation really is most or all of what there is to evolution; or it might be that we don’t actually know a lot about the etiology of what appears to be biological good design. Dawkins is inclined to bet on the first horse, but it’s not hard to find quite reputable scientists who are inclined to bet on the second. Either way, it’s a shame not to tell the reader that what’s going on is, in fact, a horse race and not a triumphal procession.
Dawkins is the kind of scientist who disapproves of philosophy but can’t stop himself trying to do some. That’s quite a familiar syndrome. I should say a few words about what I’m afraid he takes to be the philosophical chapters of his book. They are, in my view, a lot less interesting than the biology. Dawkins says, rightly, that Darwinism teaches us that the biological population of the world wasn’t made for our comfort, amusement or edification. ‘We need, for purposes of scientific understanding, to find a less human-centred view of the natural world.’ Right on. But then he spoils it by asking, in effect, if it’s not all for us, who (or what) is it all for? This is a bad question, to which only bad answers are forthcoming.
The bad answer Dawkins offers in the present book follows the same line that he took in The Selfish Gene: it’s all in aid of the DNA. ‘What are living things really [sic] for ... The answer is DNA. It is a profound and precise answer and the argument is watertight.’ The idea is that, from the gene’s point of view, organisms are just ‘survival machines’ whose purpose is to house and propagate the DNA that shaped them. A creature’s only function in life (or in death, for that matter; see Dawkins’s adaptationist treatment of the evolution of altruism) is to mediate the proliferation, down through the generations, of the genes that it carries. Likewise for the parts of creatures: ‘The peacock’s beak, by picking up food that keeps the peacock alive, is a tool for indirectly spreading instructions for making peacock beaks’ (i.e. for spreading the peacock’s DNA). It is, according to Dawkins, the preservation of the genetic instructions themselves that is the point of the operation.
But that doesn’t work, since you could tell the story just as well from the point of view of any other of the creature’s heritable traits; there’s nothing special, in this respect, about its genetic endowment. For example, here’s the Cycle of Generation as it appears from the point of view of the peacock’s selfish beak:
Maybe genes think what beaks are for is to help make more genes, but what do they know about philosophy? Beaks see life steadily and they see it whole, and they think what genes are for is to help make more beaks. The apparatus – a survival machine, if that amuses you – works like this: beaks help to ensure the proliferation of peacocks, which help to ensure the proliferation of peacock DNA, which helps to ensure the proliferation of instructions to make more peacocks’ beaks, which helps to make more peacock beaks. The beaks are the point; the beaks are what it’s all ‘for’. The rest is just mechanics.
What’s wrong with this nonsense is that peacocks’ beaks don’t have points of view (or wants, or preferences), selfish or otherwise. And genes don’t either, not even ‘unconsciously’, though Dawkins is often confused between denying that evolutionary design is literally conscious and denying that it is literally design. It’s the latter that’s the issue. All that happens is this: microscopic variations cause macroscopic effects, as an indirect consequence of which sometimes the variants proliferate and sometimes they don’t. That’s all there is; there’s a lot of ‘because’ out there, but there isn’t any ‘for’.
In a certain sense, none of the teleological fooling around actually matters (which is, I guess, why Dawkins is prepared to indulge in it so freely). When you actually start to do the science, the metaphors drop out and statistics take over. So I wouldn’t fuss about it except that, like Dawkins, I take science philosophically seriously; good science is as close as we ever get to the literal truth about how things are. I’m displeased with Dawkins’s pop gloss on evolutionary theory because I think it gets in the way of seeing how science shows the world to be; and that, I would have thought, is what the populariser of science-as-philosophy should most seek to convey. Dawkins is rather proud of his hard-headedness (he writes ‘sensitive’ in sneer-quotes to show how tough he is); but in fact his naturalism doesn’t go nearly deep enough. Certainly it doesn’t go as deep as Darwin’s.
It’s very hard to get this right because our penchant for teleology – for explaining things on the model of agents, with their beliefs, goals and desires – is inveterate, and probably itself innate. We are forever wanting to know what things are for, and we don’t like having to take Nothing for an answer. That gives us a wonderful head start on understanding the practical psychology of ourselves and our conspecifics; but it is one of the (no doubt many) respects in which we aren’t kinds of creatures ideally equipped for doing natural science. Still, I think that sometimes, out of the corner of an eye, ‘at the moment which is not of action or inaction’, one can glimpse the true scientific vision: austere, tragic, alienated and supremely beautiful. A world that isn’t for anything; a world that is just there.