Daily Archives: December 26, 2011

The argument that different races have genetically determined differences in intelligence

The presumption being examined here is that humans are divisible into different groups (races would be one term for those groups) that are genetically distinct from one another in a way that causes those groups to have group level differences in average intelligence, as measured by IQ. More exactly, this post is about the sequence of arguments that are usually made when people try to make this assertion.

The argument usually starts out noting that there are dozens of papers that document group differences in IQ. I’ll point out right now that most of those papers are published in journals with editorial boards staffed in part or in total with well known racist scientists such as J. Philippe Rushton. That fact is not too important to what I have to say here, but since the usual argument about race and IQ starts out with “Hey, look at all these papers in these great journals” it is worth noting.

Heritability of IQ measures is then proffered, often in reference to the famous “twin studies” which show a high heritability for IQ. Heritability is a measure derived from covariance between relatedness and some phenotype. Heritability is not genetic inheritance. It is scientifically incorrect and probably academically dishonest to assume or insist that a high heritability value means that something is genetic. It often is, but it need not be. The truth is, that there are many things that could have a high heritability value but that we know are not genetic, so we don’t make a heritability estimate. There are other things for which we have strong a priori biological arguments that hey are genetic, and we thus make heritability estimates as part of the research on those things. Then there are things that we don’t know the cause of, and in those cases, making an estimate of heritability is useful as an exploratory tool. But, and this is important, arriving at a high value for heritability does not indicate genetic inheritance.

If you apply the methodology of the twin studies to language, you would find that having the capacity of language is of a similar heritability of having one head (as opposed to zero or two heads, for instance): Undefined. The number of heads does not vary, and heritability is a measure of covariation (I use the term “covariation” in a non-technical sense here). If you apply these methodologies to what language someone speaks, the heritability for that trait is very high, much much higher than for IQ. If you apply the same method to heritability of geography (the lat/long of where someone lives), it is even higher, especially for babies or people living in traditional societies.

Does everyone understand why that is the case? Familial or cultural causes may be very strong but not genetic. Using this method, if high heritability means that IQ is genetic, then so is which language you speak and so is what part of the world you live in.

The smoke and mirror part of this is equating heritability with inheritance. We speak the language we speak because it is the language of the culture we grow up in, not because of a gene for speaking French vs. a gene for speaking Sumerian.

This makes sense because we know how a person acquires language, so no one even tries to measure heritability of which language someone speaks. (Same with heritability of geographic location. It would be an absurd measure.) But people make the assumption that intelligence is inherited. Why do they make that assumption? Because lots of people for a long time wanted to, and in some cases, needed to believe this so, and thus it has become part of our culture. It is part of our uncriticized received knowledge, along with other racialized ideas and various sexist ideas, and so on. But recent research (meaning over the last 30 years) has shown us that other than in the case if inherited neuro-developmental diseases, it is impossible to imagine how intelligence can be inherited in such a way as to explain the variability we see in the most inter-group differences. Maybe a little, but not that much. That there is some genetic component is not impossible, but it is very hard to maintain the idea that it is genetic and ethnic, or genetic and racial, or genetic and explanatory of more than a few IQ points in most people. There are no genes, there are no developmental mechanisms, that have been identified. So, to many the issue of inheritance (not heritability, but inheritance via genes) of intelligence is not really an issue.

However, there are many who still need to hang on to this belief. Why they need to hang on is itself an interesting question. I can’t say for a given individual but I’ve been engaged in this conversation for 30 years and in my experience it is very often because of a desire to support a racialized model of human behavior.

The evidence for the usual IQ/Group/Race/Ethnicity/Genetic model we see is always given first as group differences. When the language and geography analogs are brought up, we always see the twin studies brought in. But twins are raised together in the same environment. So they have the same language, the same cultural customs, the same geography etc. That they have the same IQ is not surprising.

There is an interesting set of interactions between familial effects and environmental effects with any of these twin studies results, but it has to be understood that heritability is not inheritance. If you have a genetic mechanism that is real (not inferred or made up) that integrates with a developmental process that can manifest a phenotype based on a genotype (that is real, not made up or inferred) then you can translate heritability to genetic inheritance, roughly. We seem to see this in a number of psychological conditions/diseases, for instance, and obviously we see it for a lot of physical traits. If on the other and you have familial effects that would cause offspring to resemble their parents without genes then cultural/social/familial context is more likely to be the explanation.

Variation in IQ across groups in a single society (like in the US) (which is not the same as a single culture) is known to be primarily caused by SES and home environment, and is indicated by such things as parents’ educational level. Educational levels of Americans have been going up for a hundred years. So has IQ. IQ can jump up in a generation if one generation is educated and changes home environment and SES etc., and thereafter those offspring and grand offspring have higher IQ’s. No new alleles were introduced to cause those changes. Cultural differences were introduced, and we have a concept of the mechanism by how that works.

The difference in IQ across time within a given population is sometimes much greater than the difference in IQ across the usual groupings of people (i.e., “race”). When scientist seek societal, cultural, nutritional and educational explanations for differences in IQ they find them easily. When scientists who have this need for group differences to be genetic seek those genetic explanations for differences in IQ they have to invent new and shall we say “interesting” statistical techniques to justify how their usually cooked data underlie their biologically implausible explanations. The latest is “there are thousands of genes and there are so many we can’t see the pattern,and that is the pattern.” Funny that. The number of genes with tiny variants that “must be” the cause of variation in IQ is going up and up and up and the number of genes that are estimated actually exist in the human genome has gone way down. At this point, we are very close to saying that individual variation in IQ is best explained by … which individual you measured the IQ in!

Let me explain that in another way, which is an analogy though it looks like a statistical argument (don’t mistake the two). If I show you two points on a graph, I can describe a line indicating their relationship with the formula Y = mX + b (the formula for a line). I can use the same formula to describe the line representing a scatter of points, but the line might be a poor describer of the scatter. How bad it is may be indicated by a statistic (a correlation value or a “R” value or something). But, if I change the formula to Y = m1X1 + m2X2 + b then I get a curvy line that may match the points better. But it will still be imperfect. But, if I add even more coefficients so there is one coefficient per point, then I go back to a (nearly) perfect describer of the line once again. Because, I’ve drawn a line (more or less) that starts with the first point, then goes to the second point, then to the third point, etc. etc.

And that would be cheating.

And that would be pretty close to what some of the more recently implemented statistical models of genes and IQ do. If I include every allelic variation in humans (hypothetically) and correlated that to individually measured IQ, I’ve drawn a line from one human’s genetic value (along one axis) and IQ value (along another axis) to the next person, the next person, and then the next person so on down the line. At this point, ladies and gentlemen, we show that IQ correlates (almost) perfectly with fingerprint.

The next argument in favor of the genetic inheritance of intelligence is often to link IQ to head size or brain size. However, much of the data related to this research is very made up or cooked, and the causal arrow is problematic. Also, a third or fourth level factor in IQ is diet, which may affect brain size. Separately, a primary factor in skull shape and bone thickness is also diet (though in totally unrelated ways) which in turn is ethnic/regional… Bottom line, the system is complex, but the data do not support the assertion unless you make a big part of the data up, and Rushton has famously done so.

Another argument that is often made to salvage the genetic determination (by racial group) of intelligence is the between national data that has been more recently assembled and foisted on us. This is no different than ethic groups in the US. IQ is a standard measure, and groups vary in this value. Other measures will also result in variation. The variation is there, and the group level distinction is there. But finding more examples of that does not lead towards the conclusion that this is racial or genetic. Across nations we see a lot of measures that we know change (often in predictable directions) over time with industrialization or various other transitions. National IQ, fertility, various health measures, and so on all do this. And, of all these measures, the most suspect in terms of quality of data is IQ (excepting some more obscure health related data). These IQ comparisons don’t tell us much.

The final argument in favor of the inheritance of IQ via genes passed on from parent to offspring is usually to cite the twins separated at birth studies. These studies, however, simply do not show this. These twins are not separated at birth in the way most people think they are. Usually, the twins knew each other as they grew up, and/or knew commonly held family members. They lived in the same culture, usually in the same city, often in the same neighborhood, and sometimes even in the same physical house. They went to the same school and had the same diet. Separated at birth in these studies usually means grandma and grandpa took one of the twins to raise because mom and dad were strapped. Grandma and grandpa may have lived down the street. The kids may have attended the same school, even the same classes, and spent a lot of time together outside of school.

I was separated (though not from birth) from my older brother, because he lived on the second floor of a two family house, and I lived on the first floor. By the exact criteria of the twin studies, we would be counted as separated because it happened early enough in my life. But, that household I grew up in was a single household that happened to be set up in a two family house. The two floors were connected by an internal rear stairway that led to locked doors (had we locks). I was rather shocked to realize at one point as a child that we were the only family in my neighborhood with two kitchens. (Or two bathrooms, for that matter.)

There may be a small component of intelligence that is inherited, but it seems to be swamped by other factors. The insistence that genes determine intelligence and that these genes are divided up in our species by groups that are often defined racially is usually misguided, and is scientifically wrong. The supra-ultimate argument, after the final argument, brought up in this sort of conversation is usually that the anti-racist argument is a Politically Correct argument, yada yada yada. But it is actually a scientific argument, and the racialized intelligence argument is not. Making the latter a politically incorrect argument.

Which is kind of funny.