Thread by @benwilbrink

Today's theme: does it make sense to talk about differences in genes as a cause of differences in intelligence and highest degree obtained?

Note: it concerns possibly 10 to 15% of the variance in intelligence or achieved educational level, there are also other causes, haha.

If you want to embarrass a psychometrician, ask how it is possible that differences cause differences. And not only that psychometrician, it is also a question for me to which I do not have a clear answer. We talk glibly about differences and causes. Nothing is obvious here.

Focusing attention on differences ignores levels. It matters what the level of intelligence in the population is: Flynn effect! Or how many years of education have been followed on average.
(Was it Cummings who suggested reducing the education budget, having listened to Plomin?)

The genetics part of the discussion (via Harden 'The genetic lottery' ch. 3):
Lee et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
sci-hub.se/10.1038/s41588…

I'll try to explain. The data are the genomes of 1.1 million individuals (yes, that is correct!) in the US (of European descent). The genomes have been correlated with college completion in a GWAS study. For this technique see the exposition in Harden, ch. 3.

The analysis assigns each individual a PGS (polygenic score): the chance of having completed college. The figure shows horizontally the quintiles from the lowest to the highest PGS (chances). The blue bars are cohorts of adolescents, the yellow ones are individuals > 50 years.

The large gains from left to right indicate a substantial predictability of college attainment.
The quintiles / horizontal axis for PGS is based on the genomes (DNA) only, college completion / vertical axis is known college completion.

With a million individuals to develop the 'prediction formula' for college completion, the danger is to capitalize on chance. Therefore the prediction formula (to construct PGS scores) was applied on new cohorts of adolescents and >50 years old.

I have no issue with this research, the facts are what they are. And to be honest, this kind of sophisticated genetic research is way above my head.
At issue is: does it make sense to speak of genes causing differences in educational attainment, given the scientific results?

What bothers me is the belief among educationalists and in the Western world at large (thanks a century of IQ testing) that intelligence is an inborn characteristic, restricting the ed. level one might attain. Emphasizing the causal role of genes does nothing to correct it.

Let me elaborate a bit.
(1) Every extra year of education adds about 3 IQ points. Intelligence is not something one is born with, it is acquired.
You do not believe it? Check it with Ritchie & Tucker-Drob 2018 journals.sagepub.com/doi/full/10.11… open access.

Psychologists have known this all along, of course. To begin with Alfred Binet himself. A classic study published by Nancy Bailey in 1955 sci-hub.se/10.1037/h00438….
I like especially a Dutch study by Oort in my home town 1928 The same tests taken from 1 until 12, raw scores.

Is intelligence a personal characteristic (even if a growing one)? Robert Sternberg comes up with an answer, december 2021: 'Adaptive Intelligence: Intelligence Is Not a Personal Trait but Rather a Person × Task × Situation Interaction' mdpi.com/2079-3200/9/4/…

Intelligence, as we understand it after the intelligence testing revolution, is whatever it is that intelligence tests measure: pure.uva.nl/ws/files/23764… open access. By the way, the authors state "the search for a genetic of brain instantiation of “g” is deemed useless."

What intelligence tests measure is 'school cognition', not at all representative for the kind of cognition that will count in society: expertise (in one's vocation). The world of intelligence testing is somewhat (or strongly?) disconnected from the world 'after schooling'.

Research on expertise has been popularized by Ericsson & Pool in 'Peak. Secrets from the new science of expertise'. Reviewed by Daniel Willingham: danielwillingham.com/daniel-willing… My own review, in Dutch: benwilbrink.nl/Wilbrink_Piek_…

There is an extensive philosophical and methodological literature on causation and explanation of phenomena (Hempel, Salmon, Strevens, Pearl). I'll leave it all aside, and try to develop some sober arguments about genes as causes of (differences) in intelligence.

Many years, if not decades, lie between genes as a cause and (differences) in intelligence as a result. It's a special kind of 'action at a distance', so to speak.
With 'action at a distance', I must think of Newton, who developed a brilliant model for 'action at a distance':

his mechanic laws using gravity 'g'. Gravity 'g' is what keeps the Earth in its orbit around the Sun. Newton's problem was that 'action at a distance' is impossible, so what is that 'g' anyway? He declined to speculate about it ("Hypotheses non fingo" en.wikipedia.org/wiki/Hypothese…).

'G' is still kind of a mystery, physicists are working hard to solve it. en.wikipedia.org/wiki/Gravity
Mystery or not, Newton's Laws are working fine, (enabling the discovery of Neptune).
Can the same be said of a theory of genes 'causing' differences in intelligence? Just asking.

There are so many steps/stages/levels/categories between the DNA-molecule and the intelligence test or highest ed. level attained. Of course, one may be content with just a black box approach: never mind that no mechanism is known, we can prove causality.

Newton came away with it, in actual practice there is no need to know the 'mechanism' of 'g'. Why would that be different for the 'g' of intelligence? Because of the societal impact of beliefs about differences in intelligence we need to develop and test models of mechanisms.

Nothing new here. Cognitive psychology sports models of cognitive architure that specify the nuts and bolts. The late Allen Newell (SOAR): "I have got to know how the gears clank and how the pistons go and all the rest of that detail." Or Anderson's ACT-R goo.gl/rzwjb8

Back on topic: we need to know the mechanics of the the relation between genetics and intelligence or ed. level attained, if only to be sure the supposed causal relation is not the result of a society, or a school (system) that systematically enlarges particular differences.

Look at the black box as an algorithm that has been developed using sophisticated techniques of artificial intelligence. Nobody knows WHY the algorithm makes particular decisions ('computer says no') - in the social domain that is just unacceptable. A black box of biases is evil.

Turn the question of genes as a cause differences of intelligence around: It is of first importance to know exactly the many ways how differences of intelligence/attainment are or can be generated. In that research genes might turn out to play particular roles. Fine. No problem.

Partial solutions there are plenty. A particularly apt theory is that of Dickens: 'Heritability estimates versus large environmental effects: The IQ paradox resolved' Psychological Review www3.nd.edu/~ghaeffel/Dick… Download and store it. Reading it is even better.
End of thread.

Unroll @threadreaderapp