Darwin's Dangerous Idea (21 page)

Moreover, mutations accumulate at the rate of about 100 per genome per generation in mammals. "That is, your children will have one hundred differences from you and your spouse in their genes as a result of random copying errors by your enzymes or as a result 8. This is an oversimplification, leaving out the role of messenger RNA and other of mutations in your ovaries or testicles caused by cosmic rays" (Matt Ridley 1993, p.

complications.

45).

114 THE POSSIBLE AND THE ACTUAL

The Complex Relation Between Genome and Organism
115

brought into the right relations with each other. The process begins with a We will cautiously help ourselves to some poetic license, too. Suppose we single fertilized cell, which then divides into two daughter cells, which proceed
as if
the Library of Mendel were equipped with a single or standard divide again, and so forth ( each with its own duplicate copy of all the DNA DNA-reader that can equally well turn out a turnip or a tiger, depending on that is being read, of course ). These newly formed cells, of many different the recipe it finds in one of the genome volumes. This is a brutal varieties (depending on which proteins are jiggled into which places in which oversimplification, but later we can reopen the question of the developmental order), must in turn migrate to the right locations in the embryo, which grows or embryological complications.10 Whatever standard DNA-reader we by dividing and dividing, building, rebuilding, revising, extending, repeating, choose, relative to it the Vast majority of DNA sequences in the Library of and so forth.

Mendel will be utter gibberish. Any attempt to "execute" such a recipe for This is a process that is only partly controlled by the DNA, which in effect creating a viable organism would quickly terminate in absurdity. We wouldn't
presupposes
(and hence does not itself
specify
) the reader and the reading change this picture appreciably if instead we imagined there to be millions of process. Compare genomes to musical scores. Does a written score of different dialects of DNA-readers, analogous to the different actual languages Beethoven's Fifth Symphony
specify
that piece of music? Not to Martians, it represented in the Library of Babel. In that Library, the English books may be wouldn't, because it presupposes the existence of violins, violas, clarinets, gibberish to the Polish readers and vice versa, but Vastly most of the volumes trumpets. Suppose we take the score and attach a sheaf of directions and are gibberish to all readers. Take any one volume at random, and no doubt we blueprints for making (and playing) all the instruments, and send the whole can imagine that it is composed in a language, Babelish, in which it tells a package to Mars. Now we are getting closer to a package that could in wonderful tale. (Imagination is cheap if we don't have to bother with the principle be used to re-create Beethoven's music on Mars. But the Martians details.) But if we remind ourselves that real languages have to be compact would still have to be able to decipher the recipe, make the instruments, and and
practical
things, with short, easily read sentences that depend on then play them as the score directed.

systematic regularity to get their messages across, we can assure ourselves This is what makes the story of Michael Crichton's novel
Jurassic Park
that, compared with the Vast variety of texts in the Library, the possible (1990)—and the Steven Spielberg movie made of it—a fantasy: even com-languages are Vanishingly few. So we might as well pretend, for the time pletely intact dinosaur DNA would be powerless to re-create a dinosaur being, that there was just one language, just one sort of reader.

without the aid of a dinosaur-DNA-reader, and those are just as extinct as The second complexity we may acknowledge and postpone concerns dinosaurs (they are, after all, the ovaries of dinosaurs). If you
have
a (living) viability. A tiger is viable
now,
in certain existing environments on our dinosaur ovary, then it, together with dinosaur DNA, can specify
another
planet, but would not have been viable in most earlier days, and may become dinosaur, another dinosaur ovary, and so forth indefinitely, but dinosaur inviable in the future (as may all life on Earth, in fact). Viability is relative to DNA by itself, even complete dinosaur DNA, is only half (or, depending on the environment in which the organism must make its living. Without how you count, maybe less than half) the equation. We might say that every breathable atmosphere and edible prey—to take the most obvious species that has ever existed on this planet has had its own dialect of DNA-conditions—the organic features that make tigers viable today would be to reading. Still, these dialects have had a lot in common with each other. The no avail. And since environments are to a great extent composed of, and by, principles of DNA-reading are apparently uniform across all species, after all. That is what makes genetic engineering possible; the organismic effect of a particular permutation in DNA can often be predicted in practice. So the idea of bootstrapping our way back to a dinosaur-DNA-reader is a coherent instance, he suggests, of the Great Chain of Being fallacy. "Humans, of course, are more closely related to dinosaurs than either is to frogs. Human DNA would have been better idea, however improbable. With a helping of poetic license, the film-makers than frog DNA. Bird DNA would be better still."

might pretend that acceptable substitute readers could be found (introduce the dinosaur-DNA text to the DNA-reader in a frog, and hope for the best).9

10. A recent theme often heard among evolutionary theorists is that the "gene centrism"

that is more or less standard these days has gone too far. According to this complaint, orthodoxy vastly overestimates the extent to which the DNA can be considered to be a recipe, composed of genes, specifying a phenotype or an organism. Those who make this claim are the deconstructionists of biology, elevating the reader to power by demoting the text. It is a useful theme as an antidote to oversimplified gene centrism, but in 9. The film-makers never really address the problem of the
DNA-reader
at all, and use frog overdose it is about as silly as deconstructionism in literary studies. This will be given DNA just to patch the missing parts of the dinosaur DNA. David Haig has pointed out to more attention in chapter 11.

me that this choice of a frog by the film-makers manifests an interesting error—an 116 THE POSSIBLE AND THE ACTUAL

The Complex Relation Between Genome and Organism
117

the
other
organisms extant, viability is a constantly changing property, a Similarly, a gene which caused total blindness would also prevent reading, moving target, not a fixed condition. This problem is minimized if we join but it would not usefully be regarded as a gene for not reading. This is Darwin in starting in the middle, with currently existing environments, and simply because preventing reading would not be its most obvious or de-extrapolate cautiously to earlier and later possibilities. We can leave till later bilitating phenotypic effect. [Dawkins 1982, p. 23. See also Dawkins 1989a, a consideration of the initial bootstrapping that may (or must) have happened pp. 281-82, and Sterelny and Kitcher 1988.]

to set this coevolution of organisms and their environments in motion.

The third complexity concerns the relationship between the texts of the The indirect way in which groups of codons—triplets of DNA nucle-genomes that do determine viable organisms, and the features those organ-otides—instruct the building process does not prohibit us, then, from speak-isms exhibit. As we have already noted several times in passing, there is no ing of a gene for
x
or for
y,
using the familiar geneticists' shorthand, and
simple
mapping of nucleotide "words" onto Mendelian genes—putative bearing in mind that that is what we are doing. But it does mean that there carriers of the "specs" (as an engineer would say)
for
one feature or another.

may be fundamental differences between the space of genomes and the space It is simply not the case that there is a sequence of nucleotides that spells of "possible" organisms. The fact that
we
can consistently describe a finished

"blue eyes" or "webbed feet" or "homosexual" in any descriptive language.

product—say, a giraffe with green stripes instead of brown blotches —does And you can't spell "firm" or "flavorful" in the language of tomato DNA—

not guarantee that there is a DNA recipe for making it. It may just be that, even though you
can
revise the nucleotide sequence in that language so that because of the peculiar requirements of development, there simply is no the effect is firmer, more flavorful tomatoes.

starting point in DNA that has such a giraffe as its destination.

When this complication is acknowledged, it is usually pointed out that This may seem very implausible. What could be impossible about a giraffe genomes are not like descriptions or blueprints of finished products, but with green stripes? Zebras have stripes, drakes have green feathers on their more like recipes for building them. This does not mean, as some critics have heads—there is nothing biologically impossible about the properties in contended, that it is always—or even ever—a mistake to speak of a gene
for
isolation, and surely they can be put together in one giraffe! So you'd think.

this or that. The presence or absence of an instruction in a recipe can make a But you must not count on it. You'd probably also think a striped animal with typical and important difference, and whatever difference it makes may be a spotted tail was possible, but it may well not be. James Murray (1989) has correctly described as what the instruction—the gene—is "for." This point developed mathematical models that show how the developmental process of has been so frequently and influentially missed by the critics that it is worth distributing color on animals could readily make a spotted animal with a pausing to expose its error vividly. Richard Dawkins has come up with an striped tail, but not vice versa. This is suggestive, but not yet—as some have example that does this so well that it is worth quoting in full (it also rashly said—a strict proof of impossibility. Anyone who had learned how to highlights the importance of the second of our complications, the relativity of build a tiny ship in a bottle—a hard enough trick— might think it was flat viability to environment):

impossible to put a whole fresh pear in a narrow-necked bottle, but it isn't; witness the bottles of Poire William liqueur. How is it done? Could the Reading is a learned skill of prodigious complexity, but this provides no molten glass somehow be blown around a pear without scorching it? No, the reason in itself for scepticism about the possible existence of a gene for bottles are hung on the trees in the spring so that the pears can grow inside reading. All we would need in order to establish the existence of a gene for them. Proving that there is no
straightforward
way for biology to accomplish reading is to discover a gene for not reading, say a gene which induced a some trick is never a proof of impossibility. Remember Orgel's Second Rule!

brain lesion causing specific dyslexia. Such a dyslexic person might be In his account of Biomorph Land, Dawkins stresses that a tiny—indeed normal and intelligent in all respects except that he could not read. No minimal—change in the genotype (the recipe) can produce a strikingly large geneticist would be particularly surprised if this type of dyslexia turned change in the phenotype (the resulting individual organism), but he tends to out to breed true in some Mendelian fashion. Obviously, in this event, the gene would only exhibit its effect in an environment which included slight one of the major implications of this: if a single step in the genotype normal education. In a prehistoric environment it might have had no can produce a giant step in the phenotype, intermediate steps for the detectable effect, or it might have had some different effect and have been phenotype may be simply unavailable, given the mapping rules. To take a known to cave-dwelling geneticists as, say, a gene for inability to read deliberately extreme and fanciful example, you might think that if a beast animal footprints. In our educated environment it would properly be called could have twenty-centimeter tusks and forty-centimeter tusks, it would a gene 'for' dyslexia, since dyslexia would be its most salient consequence.

stand to reason that it could also have thirty-centimeter tusks, but the rules
Possibility Naturalized
119

118 THE POSSIBLE AND THE ACTUAL

"more possible" than others—that is, nearer in the multidimensional search for tusk-making in the recipe system may not allow for such a case. The 1

space, and more accessible, "easier" to get to. Things that would have been species in question might have to "choose" between tusks ten centimeters "too viewed as biological impossibilities just a few years ago—such as plants that short" or ten centimeters "too long." This means that arguments that proceed glowed in the dark in virtue of having firefly genes in them—are now not only from engineering assumptions about which design would be the optimal or best possible but actual. Are twenty-first-century dinosaurs possible? Well, the design must be extremely cautious in assuming that what seems intuitively to vehicles for
getting there
from here have been developed to the point where we be available or possible is actually accessible in the organism's design space, can at least tell a cracking good story—one requiring remarkably little poetic given the way it reads its recipes. (This will be a major topic in chapters 8, 9, license. ("There" is a portion of the Library of Mendel through which the Tree and 10.)

of Life stopped meandering about sixty million years ago.) What rules govern travel through this space? What rules or laws constrain the relations between genomes and their phenotypic products? So far, all we have acknowledged are 4. POSSIBILITY NATURALIZED

logical or mathematical necessities on the one hand and the laws of physics on the other. That is, we have proceeded as if we knew what both logical possibility With the help of the Library of Mendel, we can now resolve—or at least and (mere) physical possibility were. These are difficult and controversial unite under a single perspective—some of the nagging problems about issues, but we may consider them
clamped:
we simply assume some fixed