What is intelligence?

You may have heard that a recent GWAS study found three genes for heritable intelligence, though with tiny effects. There was a great quote in a Nature News article on the topic:

“We haven’t found nothing,” he says.

Yeah, you don’t want that to be your money quote.

Kevin Mitchell has been tweeting about the study – I hope he storifies it! – and linked to an old post of his suggesting that the genetics of intelligence are really the genetics of stupidity: it’s not that these genes are making you smarter, but that they’re making you less dumb (as I gather, a lot of evidence suggests that ‘intelligence’ is related to overall health.)

Anyway, the SNPs that the GWAS identified are in KNCMA1, NRXN1, POU2F3, and SCRT which all are involved in glutamate neurotransmission. This is always troubling to my tiny brain, because I never understand quite how ‘intelligence’ works. People like to think that is some kind of learning, so if we can just learn better we’ll be smarter. And that’s what the authors of the article hint at.

But how does that even make sense? Learning faster is, in a way, like being hyperreactive to the world. There’s a reason that overlearning is a problem in machine learning! There is an optimal level of learning that, presumably, evolution has stuck us with. So is the supposition that we’re overreactive to conforming to stimuli in the world something that is good? Or is it that the modern world favors it whereas historically it would not have? Or what?