February 16, 2009
A Linguistic Analysis of Your Genes
When it comes to evolution these days, scientists tend to present a uniform front of agreement for political and rhetorical reasons, so you maybe didn’t know that, in private, some theoretical biologists have grawlix-laced thoughts about certain colleagues, whose work on one issue in particular they regard as not only wrong but stubbornly, perversely so, crumbling on clearly termite-eaten logic, and vice versa for the second group against the first—but there you go.
A divisive example: While most female lions are dutiful about guarding the borders of their camps against attacks, there are definitely some Cadillac Queens among them who don’t help out at all. The lionesses lazy in this regard benefit disproportionately because they don’t put themselves in danger when attacks come and can concentrate on breeding in the meantime and yet still get all the benefits of the others’ work, since they cannot help but be warned by all the scrambling around and yelling during any breach of security. Natural selection therefore favors lazy lionesses who defect—and if you want to be reductive, it seems to favor genes that make lions lazy defect.
The catch is that if there are too many lazy lionesses, the entire group will get wiped out in one attack, which isn’t good for anyone’s genes. So for the long-term survival of the species natural selection must favor the genes for self-sacrifice. Except that’s not quite right, either. Day to day, the lazy lions still have an advantage over the dutiful lions, and day to day, the lazy lions’ genes are still more likely to spread. In which case natural selection is selecting both for and against genes that are less fit, which isn’t natural selection in any real sense. It gets even knottier when you look at competition between groups, because when individuals decide to cooperate and coalesce into groups, complicated properties emerge. It’s no different than collections of limp neurons firing themselves up into a mind with memory, emotion, and volition. Can one neuron think? Can natural selection meaningfully be said to “work” on individuals when it only favors groups of those individuals working together, and not the individuals themselves?
This is what cleaves biologists. No one argues that natural selection is a monolithic force propelling evolution onward without purpose or design (the uniform front), but what does it act on?—genes, individuals, whole groups at once? Until the 1960s, most biologists were too busily focused on squaring Mendel with Darwin, what’s known as the Modern Synthesis, to ponder this problem. Most, as Charles D. did, lazily assumed selection happened on multiple levels. Ever since then, biologists realized they needed to be a lot more explicit about the assumptions undergirding their models.
With a Moore’s Law-explosion in molecular genetic knowledge, most biologists began to deny it even made sense to talk about selection on any level above genes. After all, if only individual fast-twitch gazelles survive in environments with jackals, it’s really the fast-twitch genes that are propagating themselves. Individuals (in this context) reduce to sums of genes. This gene-centric view of natural selection becomes even more convincing when you consider the many documented cases of rogue genetic fragments that reproduce themselves at the expense of the organism. These rogues might chop up important DNA or disrupt vital metabolic processes, but they don’t care—as long as they clone themselves, the higher structure be damned.
Nevertheless, this view—that it’s genes all the way down—seems to founder on the question of altruism. (I.e., sacrificing your time or even your body in ways that might help pass on, inadvertently or not, the genes of another.) Genophilic biologists long explained altruism in terms of selfish sacrifice: Altruistic animals will sacrifice themselves as long as doing so gives the very similar genes of their close relatives a chance to live. Under this hegemony, you’d readily sacrifice yourself to save two siblings with fifty percent of your genes in common, or four cousins with twenty-five percent of genes, or eight second cousins, etc., because their living to breed is just as good—from the gene’s point of view—as your living to breed. There are variations on “kin selection” theory, but most boil down to this.
The problem is, does anyone really believe that? Not believe that kin selection happens sometimes—it does. But that it explains everything about altruism? That animals like humans evolved to work in groups and desire group contact not because groups succeed more often in the wild; but because the animals really—deep down on the 1’s and 0’s level of their unconscious biochemical beings—are calculating what percentage of their own genes will get passed on if they take an overtime shift of guard duty?
How much simpler to argue for “group selection”: that altruistic genes might hurt individuals but help groups, and therefore give groups a better chance to beat other groups. And notice that this doesn’t deny a red-in-tooth-and-claw version of nature. The “altruistic” behavior that gets favored could be organizing war parties to wipe out rivals with too many selfish members. There’s blood aplenty. The genius of this view—which contradicts the 1960s “selfish gene” dogma that genes and only genes drive evolution and is therefore (the view is) radioactively controversial in some departments—the genius is that it uses group selection to explain altruism and social cohesion without having recourse either to fuzzy angelic crap about animal “essences” being inclined toward goodness, or to unconvincing explanations of kin selection. Ah, ha!
But here’s the reversal: The selfish-gene ilk—those like Richard Dawkins, who hew tight to the 1960s dogma—make the very frustrating point that no matter what holistic, greater-than-the-sum properties you want to impute to a group being selected for, at the end of the day, when the sperm of males in a successful group meets the egg of equally successful females in the group, what get passed on are genes and genes alone. Group selection might look tempting, but it’s still got to account for traits getting passed along, altruism included. And given that creatures within a species are 99.9... percent identical molecularly, perhaps genes still push us into altruism just for the sake of our species’s genes, not just the family jewels. This sort of not-kin-but-kith selection still makes a lot of sense for most genes.
I don’t have nearly enough academic abbreviations behind my name to sort through the subtleties of this debate (no matter however shining an exemplar I am of how wrong-footed and turned around you can get while working through it). Yet something about the nature of the problem strikes me as analogous to—not perfectly mappable to, but analogous to—something I’m a little more fluent in, language.
“This sentence is false.” Just like the different organizational levels of biology (cells, organs, individuals, groups, populations), language has different levels, and the objects of one level glom together to form the higher level. The higher levels therefore contain the lower levels, too. The most common level of language is the normal sign-signifier relationship of words referring to things in the outside world: “He ate the hot dog with relish.” But there are other levels: We can talk about language itself functioning as language (“‘Type’ is a verb), and also talk about words as individual sensory phenomena, the sounds of them and their curvy looks. The really confusing part is that all of this language talking about language necessarily has to be spoken with words. And paradoxes, contradictions, and ambiguities sprout when you mix different levels without being careful.
Read the following sentences: “‘Computer’ has three syllables.” “Computer has three syllables.” The first is true, the second false (since a computer itself, the plastic and metal object, cannot have syllables). Another way of saying this is that the first sentence properly keeps track of which level the word “computer” belongs to (the level of language talking about language), while the second doesn’t. Similarly with, “This sentence is false.” For that paradox, if you persist on a basic level of analysis, you’ll end up on a pendulum where its swings both true and false and then true again but still somehow false, and so on. It’ll keep whipping you around.
But the game changes when you realize that language—when it’s talking about itself—doesn’t work the same as language that’s talking about something outside itself. This is more or less the problem Bertrand Russell (among others, though later) set out to solve in his philosophical system of sets and of sets of sets, and of sets of those sets, ultimately producing a careful hierarchy of different levels. And when translated into this hierarchy, the words “This sentence is false” become not paradoxical but meaningless, the equivalent of “Computer [no quotes] has three syllables.” The problem disappears, albeit a little dissatisfactorily.
Working through multiple levels is messy enough with easily separable and easily manipulated units like words, and the problem is a fortiori more snarly in biology. There, you still have multiple levels interacting—genes run cells, cells comprise individuals, individuals comprise groups, groups interact among themselves—but it’s much harder than in language to isolate the effects of any individual units on the whole. And just as you cannot help but talk about words with anything but more words, genes will show up at all levels of the analysis because they’re the currency, the unit that passes meaning and value.
Again, the analogies are imperfect. But there’s an intuitive parallel between determining whether selection pressures on high-level groups can be reduced to pressure on low-level genes and determining when it’s appropriate for language to talk about language. The point isn’t that biologists should somehow perform a linguistic analysis of genes, although genes are in an abstract sense a language like any other. But there may be underlying structural or mathematical similarities between logico-linguistic analysis like Russell’s and biological analysis that hops around among different levels of natural selection, schemes and strategies in one that can illuminate what’s going on in the other, just as the arcane mathematics of knot theory in the 1980s turned out to be darn useful for describing how DNA refolds and recombines itself.
If nothing else, perhaps biologists can at least learn from linguists to live with the maddening ambiguities that language has built into it. Other philosophers disagree with Russell, et al., and say that “This sentence is false” isn’t meaningless so much as true and false. There’s ambiguity there. Deal. Similarly, while saying that someone is “not unattractive” is logically equivalent to saying he’s “attractive,” it’s not really of course. Or, the sentence cited above, “He ate the hot dog with relish” is holographically ambiguous, meaning one thing or another depending entirely on where you stand. You can parse these phrases till you’re blue, but they’re irresolvable, and the ambiguity becomes an irreducible part of the analysis. Perhaps it’s impossible for biologists to sort out whether groups or genes or individuals are the one thing being acted upon by natural selection—not impossible in the sense of too mysterious or too hard a problem, but impossible in the same sense that “with relish” is simply not resolvable or that “not unattractive”/“attractive” isn’t amenable to clean-jointed logical segmentation. And perhaps accepting that duality will open up new alleys of investigation.
When Darwin wrote, “There is grandeur in this view of life,” he was speaking about his science as something like art. We prize art partly for its ability to console us about paradoxes and ambiguities, even if it cannot explain them. Perhaps there’s consolation for biologists in the conundrums of everyday language.
Posted by Sam Kean at 12:25 AM | Permalink