Wednesday, December 5, 2012

Race and Intelligence

(Originally posted by Pat on 1/6/2012)

Can you feel the norepinephrine surging through your sympathetic nervous system? Does the very idea of talking about race, especially when intelligence is involved, make you angry?
That's all right. I used to feel the same way, and what's more, there is a very good reason you feel this way. You see, good people like us who try very hard not to be racists or bigots or even just arrogant bastards have been systematically pushed away from the study of race and the study of intelligence. Over decades of bad research used to justify horrible behavior, the field has become poisoned. It is now controlled almost exclusively by eugenicists and racists, most of whom are also arrogant bastards. (Exhibit A: Satoshi Kanazawa.)

Yet, the sky does not stop being blue no matter what your favorite color is. No matter how much you don't like spiders, Arachnida is still one of the most ecologically important clades. No matter how many innocent people have been murdered by them, firearms still obey the laws of physics. The truth is the truth, and we must be strong enough to face it, whatever it may be.
So, first of all, we must ask: Is there such a thing as “race”? Originally, “race” meant “subspecies”, which has always been a problematic concept—subspecies by definition must be both well-defined (to be worth naming) and yet not too well-defined (so they can interbreed). Species are a real phenomenon, a cluster of probability in thingspace; though it's more complicated than we might have thought at first (witness Ensatina and Larus, and for that matter Canis), interbreeding still is a real phenomenon that is genuinely important in evolution. Other categories above species, on the other hand, are pure fiction; frankly I think they should be deprecated as much as possible. Clades exist—but nature does not distinguish between a subphylum and a superclass, so why should we? As for subspecies, they may be slightly more meaningful than categories like “class” and “order”, but since by definition they share genes back and forth it's hard to see how we might precisely define them.
In any case, this is not really what “race” means anymore. The language has moved on; we no longer speak of “races” of dog (even though Canis lupus is one species where subspecies actually make pretty good sense). “Race” in modern parlance means a human division, especially the division between “White people” and “Black people”.
Yet this is already a problematic distinction, for several reasons.
First of all, “White” is a club—Nordics and Anglo-Saxons have always been members, but it took awhile for Celts and Germans to get in, a bit longer for Greeks, Italians and Spaniards (even today we speak of the PIGS nations); Arabs used to be included, but since 9/11 have been largely kicked out; Latinos have never been fully admitted. So we can say for certain that Swedes are White—they are as White as can be—but for Arabs we could go either way.
“Black” is not a club, but it's a really bizarre choice so far as haplotypes go. Aborigines are closer to Japanese people than they are to Kenyans, yet we call them both “Black”. Anglo-Saxons are closer to Egyptians than Egyptians are to Hutus, yet Egyptians are always either “Arab” or “Black” while Anglo-Saxons have always been “White”. Africa, where most “Black” people live, is by far the most genetically diverse population in Homo sapiens; about 85% of human genetic diversity can be found on that continent alone.
Even worse, there are many intermediates—as one might expect from interbreeding populations. Dawkins points out a very good example: We usually speak of Colin Powell as “Black”, but his skin tone is as light as mine. Something about his facial features does “seem Black” in a certain sense, but it's really difficult for me to pin down exactly what that means. One of our past members, Carver Shields, actually has more African ancestry than Colin Powell and much the same skin tone, but his facial features “seem White”.
Yet, it may be an overreaction to say “there is no such thing as race”. If you ask a thousand Americans—or even a thousand people from all over the world—to identify which “race” a series of photographs are, there will be substantial, indeed almost perfect, agreement. Virtually no one would hesitate to mark Satoshi Kanazawa as “Asian” or Robert Mugabe as “Black”, nor to mark Tarja Halonen as “White”. If it really were the case that race is completely made up, we could not explain this level of agreement. Indeed, even if it were socially constructed (like, say, democracy and money are socially constructed), we'd be hard-pressed to explain the cross-cultural agreement. Yes, different cultures will make different subdistinctions—I couldn't tell a Hutu from a Tutsi, a Finn from a Swede, or a Czech from a Slovakian if my life depended on it—but nonetheless, the big categories—like Black and Asian—are largely agreed upon.
So what is going on here? Well, it turns out that about 15% of human genetic variance is correlated with the categories that we call “races”. This is not a huge chunk, especially since human genetic variance is so low to begin with and humans are so phenotypically plastic. You can't make precise judgments about someone's bowling average or favorite color or IQ score (more on that last one in a minute) based purely on this information. They could be a moron, or a genius; they could be a professional athlete or a complete couch potato.
Yet, you can make certain judgments pretty accurately. A Black person is much more likely to have sickle-cell anemia. A White person by contrast is more likely to have cystic fibrosis. Almost all Asian people have epicanthic folds on their eyelids, which hardly anyone else has.
Indeed, even their IQ score is not completely unpredictable. Perhaps it would be more convenient if this were not so; but the fact is, a correlation exists in observed data between “race” and IQ score. Asian people (and Jews, intriguingly) tend to score the best, followed by White people, followed by Native Americans and Hispanics, with Black people usually clustering at the bottom. It's a fact. The correlation is there.
The problem of course lies in how we interpret that correlation. Racists immediately take it as proof that not only is race a real phenomenon, but furthermore racism is totally justified and egalitarians are just deluding themselves. Egalitarians usually plug their fingers in their ears and hum loudly, which isn't helping our case.
In fact, there is a much better answer. It consists in keeping in mind what IQ scores really mean. They are not in fact a measure of innate general intelligence, whether or not there is such a thing. (Clearly there is such a thing as general intelligence—what differentiates Homo sapiens from Pan troglodytes—and obviously genes are involved. But whether there is any genetic variation in intelligence within human populations actually remains to be seen. It could well be that the intelligence genes which separate us from chimps are so powerfully selected that they were driven to fixation, much as our genes for, say, bipedality were driven to fixation.)
IQ scores are a measure of cognitive performance. Now, you can try to correct that measure for known factors, and measure heritability coefficients and all that; but the fact remains that without looking at DNA itself (or at least doing very carefully controlled twin studies), you simply can't tell how much of the variance is genetic. The IQ score itself tells how you did on a particular kind of test. It's really not much different from the SAT or the GRE (and indeed SAT and GRE scores are strongly correlated with IQ scores).
Indeed, I think heritability is largely a dead end. It's not too difficult to measure, and it does seem like it would have something to do with genetics, but in fact it's very possible to have highly heritable traits with no genetic causes and even (what's more shocking) poorly heritable traits that are completely genetic. One thing I will say in its favor is that a highly heritable trait can definitely be selected for, even if it's completely non-genetic. (Agriculture is a very successful and heritable meme indeed.)
As an example of the former I present “speaking French”. If your parents speak French, odds are you will too (with a heritability, I might add, much higher than that of IQ scores). Yet if you pick up a random human baby and raise them speaking French, you'll get a perfectly good native speaker, no matter where the baby came from. French is highly heritable yet completely environmental. English, on the other hand, is not as heritable as French, because cultural pressures incentivize people to learn English.
As an example of the latter, consider neurofibromatosis. It's a genetic disorder, no doubt about that—we have even traced specific sections of the specific DNA sequence that cause it. But it's actually not very heritable! It's a little heritable, mind you—about as heritable as, say, “being a mathematician”. But de novo mutations are so frequent at this genetic locus—and neurofibromatosis in general is so rare—that almost half the variance in phenotype is completely unrelated to genetics. Yes, that's right; almost as many people have neurofibromatosis without affected or carrier parents as have it with affected or carrier parents. (Don't confuse this with your chances of getting it; you're still much more likely to get it if your parents have it.)
So the heritability of IQ scores—about 80%—really does not tell us very much. It could be that parents with good genes have kids with good genes; or it could be that parents who learn good cognitive skills (like reading, multiplication tables, study habits) pass those skills on to their children. It could be that cultures which encourage academic success and scientific advancement turn out smarter citizens. Most likely, it's some combination of these—but what combination exactly?
It's every bit as scientifically lazy to simply presume that all the variance is environmental as it would be to presume that it's all genetic. You can't just presume either way; you have to actually look at the world. In fact, we have some pretty good evidence that at least a small part of the variance in human intelligence is genetic, no matter how much people don't like that (see Pinker's The Blank Slate). Some people are born smarter than others.
This fact should not scare people nearly as much as it does. If we had found that all the variance in intelligence were genetic, that might be a little worrisome, as it would imply that stupid people will be stupid no matter what we do (short of radical eugenic policies like sterilization). But even then, it would have a certain upside, as it would imply that no matter what you do, smart people will be smart; even if society tries to hold them down, geniuses will ultimately prevail. If genes were our destiny, well, at least we'd have destiny. (Most people seem to have an intuition that destiny is good, at least when you are the chosen one who will have a heroic destiny. It never seems to occur to people that this implies that millions of other people are destined to live pointless, painful lives.)
But in fact, hardly anyone thinks this (it may even be too extreme for Kanazawa, which is saying quite a lot). Clearly some of the variance is environmental. Ramanujan was born in poor India, and did eventually make it as a mathematician in England, but it took a lot more work than it would have if he'd just been born in England in the first place. In fact, he'd probably have been even better in such a circumstance, as like Einstein and Euler he'd have been given the best possible education. (Indeed, we would probably remember him on the same list as Einstein and Euler, whereas right now he's more on a par with Russell or Leibniz. Still a good place to be—but no Euler.)
And if it sounds like I'm confusing intelligence with performance, well, there really isn't much difference actually. If someone is a good physicist, it's really not important whether that is their innate talent or years of hard work and training; the point is, they are a good physicist. (In fact it's probably both.) If we're interested in making the world better, phenotypes are all that matter.
It might, actually, be the case that the genotypes for higher intelligence are concentrated in certain haplotypes. If that were the case, we would expect that it would be the haplotypes that are dominant in wealthy, advanced societies—because otherwise you'd be saying that dumber people are more likely to be successful in science and industry, which makes no sense. So the idea that Anglo-Saxons are innately smarter than Kenyans is not, a priori, ridiculous. It could be true.
It's just that there isn't good evidence for it. There's too little diversity, too much interbreeding, too much within-group variance. One need only encounter an Anglo-Saxon idiot (there are several on any given television channel, especially Fox News!) or one genius of Kenyan descent (like Barack Obama) to see that whatever between-group variance there might be is swamped by within-group differences.
Moreover, there's no particular reason to think that the haplotypes which vary in intelligence are the same as the ones we easily recognize as “races”. It could just as well be blue-eyed people who are smarter, or taller people (actually there's evidence for that latter!). It could be that penis size is directly proportional to intelligence, or—before you get too smug—inversely proportional. The point is, you have to actually study these things, look at the world, in order to know. And until we do that, well, we won't know.
One common response to all this is, “Why study intelligence at all?” With so much nonsense, so much bigotry, so much bad history attached to the study of intelligence, why even bother? Go study botany or quantum physics or something.
It's tempting, certainly; and I've got nothing against botany or quantum physics. But it's important to remember what we are studying here—we are studying intelligence, the very thing that makes us human, the very thing that makes us capable of botany and quantum physics in the first place.
Could we, by some medical or technological intervention, actually make people smarter? As yet, we have no way of doing this. But surely it is possible—the human brain isn't anywhere near the physical limits of computational power. We don't know how to do it, and of course it's obviously difficult for any machine to make itself better—and all the recursion that implies. So maybe we never will, alas.
But if we were to do that, we would have to start by studying intelligence. We would have to understand what causes it, how it works, where it comes from. We would have to look at variations in it, see who has more of it and why. We would have to find ways of measuring it and manipulating it.
And if you don't think we could actually get a whole lot smarter than we are, I submit that this is just a failure of imagination. Consider a young Homo erectus man on a warm night in the African plains, looking up at the full moon in the starry sky. Imagine how he would react if you told him that one day his descendants will stand on that moon; his grandchildren's grandchildren's grandchildren (add a few thousand “grands”) will actually be up there, standing on that heavenly body. He might well think that the moon is a god watching over him; but no, you explain, it is a rocky, dusty landscape, not too unlike the sands of that fledgling Sahara to the north; and one day men will stand upon it. Would he understand? Could he even imagine what a rocket would look like? Could you find the words to describe a spacesuit to someone who had never even worn cotton, much less a synthetic fiber? If you tried to explain what a computer was, or how a laser worked, do you think it would make any sense to him?
Yet we are only about 50% more intelligent than that Homo erectus man. Our brains are only about half again as big. True, we also have undergone many cultural and technological changes—indeed our young erectus could probably fit in pretty well with a hunter-gatherer tribe of sapiens—but that's part of the point. Changes in society and biology can effect absolutely massive changes in cognitive performance, indeed so massive that they are difficult to predict or even imagine.
My own feeling about this is that the variance in intelligence among healthy modern humans is too small to be concerned about. Perhaps it is true that Jews are on average 4% smarter and Blacks are on average 3% dumber or something like that. It still doesn't tell you anything about individual people—I've met some very dumb Jews and some very smart Blacks—and it still wouldn't justify unequal treatment.
Imagine passing a law that you must score 110 or higher on an IQ test in order to vote. Would this not be obviously undemocratic? Would it not be easily abused to protect entrenched special interests? The consequences of such a policy would be predictably disastrous. It might (or might not) be well-intentioned, but it clearly wouldn't be a good idea.
All the worse if the legislation were based on something that merely correlates with IQ scores, like “race”. It would be bad enough to actually use IQ scores; but if you use “race”, you're now another step removed. You'd exclude millions of smart Black people and include millions of dumb White people.
Actually, there is one issue for which we permit this kind of statistical prejudice: Age. It is presumed that older people are wiser and more responsible than younger people, and statistically this is probably true. But we only let people over 18 vote. Only people over 21 can purchase alcohol. Car insurance rates are much better for people over 25. We only let people over 35 run for President. I'm going to call this what it is: Discrimination. It's different from the race case, because young people grow up into old people; but still, it's obviously unfair.
What would I replace it with? For voting, I think I would just say “any citizen can vote”. There aren't enough 4-year-olds who'd even want to be politically active for that to be a serious problem; and the 14-year-olds and 16-year-olds who want to vote (I used to be one) would probably vote every bit as responsibly as 18-year-olds and 20-year-olds. For driving, we can use driving tests, as we already do, and simply remove the age requirement. (Car insurance rates could be based on driving test scores!) For drinking, we can do what is done in Europe, where almost anyone is allowed to drink, but social custom forbids binge drinking and public drunkenness. For elections, we can do the same thing as voting; you don't have to vote for the 14-year-old if you don't think she is qualified.
Similarly, if a job recruiter wants to base their hiring decision on grades and test scores, so be it. But basing it on racial characteristics will only bias the results. Honestly even test scores are overrated; all tests have a margin of error, and typically two candidates for a position will differ by less than this margin. But the point is, test scores are (in some cases, e.g. hiring for academic positions) directly relevant; they aren't merely a statistical correlate of something that's relevant.
If race or sex ever is directly relevant—e.g. in hiring an actor to play Rosa Parks in a film—then it can of course be used. But in most cases it's purely prejudicial, and remains so even if there is some small correlation between race and intelligence.
Hence, we simply don't need to say that there is no variation in human intelligence. We can merely point out that judging entire groups of people based on these small and hard-to-measure differences would be unjust. We should judge people as individuals, using the best systems of measurement we can.
But at the same time, yes, we should study intelligence. We should study its causes and its effects. We should try to find interventions that can make people smarter—either through education, or if necessary through biological or cybernetic enhancement. The cost of such interventions must of course be considered—but so must the benefits, which promise to be enormous.
And we must never fear the truth, for what can be destroyed by truth should be.

No comments:

Post a Comment