Genes And IQ: An Update, Ctd

A reader quotes me:

“There’s not a huge debate about the heritability of IQ, but a huge amount of debate about how much intelligence can be tied to genes and how much to the environment.” You’ve been careful in the past not to let IQ stand as a proxy for intelligence, so I’m not sure why you absolutely conflate them here. Also, you have a false dichotomy here: genes and environment are not separate influences. Gene expression is mediated by the “environment”.

A well-known example of this is the desert locust, which has a solitary (grasshopper) and gregarious (locust) forms which were, at one time, thought to be different species. Which form develops depends on the social environment of the insect during its development (specifically, the gregarious form develops when its hind legs are often stimulated by other individuals due to crowding.) A particular pair of individuals, one solitary and one gregarious, may have exactly the same genome, but exhibit severe morphological differences upon full adulthood, and these differences are controlled entirely (or nearly entirely) by the genome.

And genomic expression is far more complicated than you make it sound.

There is rarely a straight gene-to-trait pathway of expression. A gene might regulate the expression of another gene, which turns off a separate gene and turns on two more, one of which slows down the expression of the first gene, another of which causes a side effect that … etc, etc. It’s a bewildering tangle of interactions that permeates the cell and well beyond. Do you think it’s possible to document the whole ecology of a rain forest: all of the interactions and feedback loops between plants, insects, animals, the air, the soil, bacteria, worms, sunlight, wind and weather, ocean currents, river sediment, etc, etc? Trying to find which genes affect intelligence is like trying to find particular Amazonian fish species that are responsible for global warming.

Another quotes the article I cited:

“A copy of each variant accounts for only 0.3 points on a standard IQ test (with a mean of 100 and standard deviation of 15).” With 126K points, the margin of error for something with a standard deviation of 15 and a 95% confidence interval (meaning that we’re 95% sure that this isn’t just chance) is about 2*15/355 = 0.085 IQ points, so at least the 0.3% looks to be statistically significant, although just barely, since the margin of error is more than 1/4 of the entire effect. And the connection to IQ (as opposed to years of schooling) was based on a sample of 24,000, where the margin of error would be 2*15/155 = 0.2 IQ points, or nearly the entire effect. So, I’m not sure whether these results are statistically significant – the algorithm for mapping the 126K data points of years of schooling through the 24K data points of IQ isn’t specified.

Worse, however, what they did is identify 6 variants that were associated with additional years of schooling. If I told you that several of those variants were associated with Ashkenazy Jews, for example, I would have just found an expensive way of saying that Jews as a group spend more time in school than average. Correlation really doesn’t equal causation. My guess is that they found some genetic markers that are associated with some ethnic groups, and then “discovered” that those ethnic groups spend a little more time in school on average.

The author in the end says “We haven’t found nothing.” But that’s most likely exactly what they found.

Another also scrutinizes the statistics:

I was rather surprised that you went there again:

What to make of this with respect to our cultural and political debate about genes and intelligence? For me, some relief that the area is so complex, and varied, and hard to decipher that we may have more time ahead before these things become more knowable, and thereby may avoid any of the worst social implications for longer than some of us feared.

As a scientist (albeit one completely unfamiliar with the field in question), a result that shows such a minuscule effect, well within the variance in the response of the system is equivalent to there being no effect at all. You should probably become more familiar with statistics but such a result essentially means that there is absolutely no correlation whatsoever between specific genetic markers and cognitive abilities. Such a small effect (an order of magnitude smaller than normal variance in IQ) is very likely the product of the random nature of the sample. Your statement seems to suggest that things are really complicated. They are not, at least when it comes to the conclusions of this study; there are no meaningful correlation between genetic variants and IQ.

Without knowing the literature at all, I am willing to bet that there are studies that find much larger correlations between socio-economic factors and IQ scores. Based on IQ scores of whites vs other groups, I am willing to say that such effects are perhaps an order of magnitude larger than the effects in this study.

Let it go (to quote from Frozen) ...