Sunday 15 December 2013

Genetics, genetic groups, intelligence, expertise


I have been musing over the results of Rindermann, Coyle and Becker’s survey of intelligence experts presented at the ISIR conference. It looks as if many of the invited participants turned down the invitation. It may have been caution on their part that the survey would not remain anonymous (people can lose their jobs just for reporting intelligence results on group differences) or perhaps they were put off by the format and length (62 main questions) or even the idea of trying to investigate underlying opinions in a way that might lead to a conclusion “experts agree that…..”

You will already know that a general theme of this blog is that humans are a pesky lot, and difficult to study. They don’t fill in questionnaires, they forget about how much chocolate they eat, they look with envy at their neighbour’s wife, and very probably his ox, and have strong opinions, secret thoughts, impossible dreams and some highly questionable habits. Experts seem to be just as pesky. Drosophila are much better. Fruit flies are the future.

On another sour note, how does one qualify to be an expert nowadays? Publishing a recent paper in a peer-refereed journal is certainly a start, and should put you ahead of most of the multitude of commentators, journalists and citizens who knew a bright guy in their class who couldn’t tie his shoelaces,  but is it enough? Can we be sure these 228 authors have done the necessary reading? Should we start getting into impact factors and other measures of expert worthiness? Best of all, should we just ferret out the best papers with the best arguments, and tell students that, until they can be torn down, those papers presents the best approximation to the truth?

Anyway, here is the US centric question on racial differences in intelligence (there are other places on the planet like Brazil with different races living together, and there is also Africa, but once again the US leads the news).




Personally, having looked at the literature off and on for 40 years, and more of it in recent years, I cannot see how anyone can really believe that none of the difference is due to genes or that all of the difference is due to genes. I mean, the tiniest contrary finding would prove you wrong. A few genes linked to brain even in a minority of cases, a few environmental toxins linked to brain in a minority of cases and the presumption would be falsified. How could one be “expert” and be so sure of one’s self, given that there is an error term in all investigations?

For example, there is general agreement in this sample (though not universal agreement) that there is strong evidence for the heritability of intelligence within genetic groups. Given a proven cause of intelligence from genes, how could one be so sure that the difference between genetic groups will have no genetic component? It is not proven thereby,  but certainly not disproven, and the possibility remains open, and can be investigated by genetic research.

Equally, given that we can show a likely effect on intelligence of very bad environments, how can one be sure that all the effect is genetic? Jensen, for example, thought the difference was 50% genetic, and then over the years raised that to about 70% genetic (in the US context). All these estimates depend on the circumstances in which the measures are taken. Universally provided reasonable environments boost the genetic estimate by reducing environmental effects. Reduce the quality of the environment, and it jumps back into the equation with a vengeance.

I think that it is the central section in blue which represents main stream opinion among experts, which is those who did not say 0 or 100. Of those, 40% say up to 40% of variance is genetic, 40% say it is as high as 60-100 genetic, and the remaining 20% are in the middle at 50%. This is a split jury in terms of the extent of genetic influence, but the general tendency is towards agreement that genes are involved to some extent. That is a useful finding.

The paper will be presented for publication soon enough, and then we might speed up the tempo of surveys, to say every 5 years or so. Perhaps there needs to be fewer questions and even more consultation with the relevant experts about what should be included. Perhaps there should be even more stratification of experts, setting the bar higher in terms of number of publications on intelligence. However, we are now in a better position to know what experts really think, and that is a very valuable addition.


  1. It's not true that the tiniest genetic effect would require an above 0% genetic contribution to the US B-W gap. If the role of genes were tiny, and environmental effects were very close to the size of the whole gap, then genetic influences could slightly favor or slightly disfavor people of African descent, and we would see the same things.

    Your survey did not allow for negative values (Africans have some degree of genetic advantage, which is overwhelmed by environmental effects), so this view, or this view with uncertainty about sign, could be best represented by answering 0%.

    On the response rate, which was much lower than for the earlier survey, you might do better by teaming up with luminaries from different schools of thought. For example, you could have teamed up with James Flynn and company. This might have put the exercise in a more favorable light, and they could have had input on the questions.

  2. If your results are based on 60 out of 1200 people it's hard to take the results seriously. For example, people are more likely to answer requests from their friends, and Rindermann at least is known for pushing a hereditarian line. Just the authors' close friends and fellow travelers might have made up a very large portion of the tiny number answering.

  3. The method was to invite everyone who had published in the relevant journals. No one was left out of the invitation, so all schools of thought were able to participate. The low participation rate is clearly stated, and is a problem, of that there can be no doubt. Other surveys in later years may do better. At the moment this is the only contemporary result. On the question about the contribution of genetics and environment the method is the analysis of variance, and so the judgment is about variance accounted for, which is always positive.

  4. For every distinct evolutionary population there is a fitness landscape. That landscape, itself, creates the environmental factors that impact specific individuals, but it is part of that fitness landscape, having been caused by it. Therefore, in distinct populations in distinct geographic locations the variance in intelligence is always 100 percent genetic. Certainly, variance between individuals is due to environmental factors but between groups it is entirely 100 percent genetic (sans significant environmental inputs between groups).

  5. Maybe it's important to distinguish between the two following environmental factors:

    A) Exogenous, not part of the local genetic fitness landscape. An example might be invasion concluding with mass slavery, which are almost certainly short term effects since the fitness landscape adjusts to it.

    B) Endogenous. An example might be lead paint, and could be long term, depending on how quickly it is identified.

    It seems to me as if endogenous factors are going to produce much greater variance within the local population.

  6. “I cannot see how anyone can really believe that none of the difference is due to genes or that all of the difference is due to genes”

    If we begin with the assumption that the U.S. environment is detrimental to sub-Saharan African cognitive ability then asking how much of the observed U.S. B-W difference is due to environment makes sense. But if the U.S. environment is actually beneficial to sub-Saharan African cognitive ability, then the question itself is misguided. Not only is 0% of the difference due to environment, but the expected developmental gap has been substantially reduced through advantageous circumstances.

    This does not preclude any number of hypothetical ways that the U.S. environment could also still be damaging to I.Q. vis a vis the African environment (e.g. local toxins), since the question is merely about the observed difference.

  7. Jason, that is exactly what I've long held. Further, it is important to note that the environment in the US is not provided by the black fitness landscape, meaning that it falls into my category of exogenous factors, listed above.

  8. Put another way, both the question and the fundamental way that social scientists think about this topic are deeply tainted by an environmentalist bias that is incommensurate with both the available evidence and common sense:

    Does being raised from birth in an advanced Western economy IMPEDE or MAXIMIZE the potential of people from disadvantaged family backgrounds.

    IMPEDE 95%

    1. I have explained this concept countless times to people, never with that parsimony of language. Nice. Can I crib it?

    2. Jason, many thanks for your comments. The fundamental problem is trying to resolve questions of racial difference by looking only at one nation. Nations have particular histories. For example, Brazil has a history of far greater integration and inter-marriage, yet has similar findings on intelligence. Africa, though less well studied, produces results for African intelligence which are lower than US figures. This is partly due to racial admixture, partly due to far better environments. You will be able to tell me how many Americans don't have passports, but research on intelligence should have a passport, and get around the world a little more.Geography is meta-analysis in this context.

  9. It is worrying that only 60 out of a pool of 1,200 responded. It's still interesting though.