Commenting on Hotel Concierge Commenting on IQ Tests

Someone in one of the discussion down below brought up Hotel Conierges old post. In it, he argues that IQ tests dont really test intelligence, but instead Desire To Pass Tests. Some quotes to give you a general idea (If you read the original, you only need about the first third, until "III Its hard to become a doctor"):

I’m not convinced that the Stanford Marshmallow Experiment tests for anything even remotely resembling “innate willpower,” because waiting fifteen minutes for a single marshmallow is a stupid thing to do. The opportunity cost of wasting fifteen minutes is way greater than the utility of one marshmallow. My mom has a sweet tooth—it’s not like the marshmallow was a rare treat in my otherwise Dickensian life—and I’m more of a Reese’s peanut butter cup guy anyway.

That said, if the experiment predicts SAT scores then it’s clearly testing for something. It’s hard to tease out what that something is. Perhaps the delayed-gratifiers want to impress authority figures, perhaps they recognize the challenge and have some internal desire for achievement, perhaps they are simply used to doing as they are told. I’m going to sum all these motivations into The Desire To Pass Tests [1]. And it makes intuitive sense that TDTPT would predict SAT scores and number of degrees, because these are cultural tests of intelligence. It makes sense that TDTPT would predict BMI, because this is a cultural test of appearance. It makes sense that "preschool children who delayed gratification longer in the self-imposed delay paradigm were described more than 10 years later by their parents as adolescents who were significantly more competent,“ because parental approval is the oldest and most universal test there is.

The most extreme interpretation of TTA is that IQ-type tests have nothing at all do with “intelligence”: doing well on a test predicts success on future tests, being good at logic puzzles means that you are good at logic puzzles. I am not an extremist, and I do think that IQ has real meaning. A person with an IQ of 140 is probably more “innately intelligent” than someone with an IQ of 100, at least with regards to the “quantitative reasoning, visual-spatial processing, and working memory” skills that are tested, and this type of intelligence probably does have something to do with success as an economist or biochemist or whatever. SAT scores are less innate than IQ, but even so, someone with a perfect math SAT will probably have a brighter future in math-associated fields than someone with a subpar score [4]. I am not saying tests are useless—I am saying that they have far more Test Taking Ability noise to intelligence signal than most people will admit, and that this noise is more than enough to explain racial score disparities [5]. The map is not the territory.

Ill really only talk about that point. Im not really happy with his argument for it, but I wont address that either. Just imagine for a moment that this is completely true: The thing we call IQ really only measures a desire to pass tests or a willingness to please authority or something of that sort. This feels like a big upset to our beliefs about IQ, that maybe the argument weve been building collapses now. My argument will be that it would be great insight porn but not really change anything.

Remember that all those other studies about IQ are still... there. Its still highly genetic, including in adoption studies. We still dont have a way to increase it long-term. It still predicts future income and productivity on the job and health and martial status and most other good things. It still explains the gaps in these things. Or maybe it doesnt, but thats an argument fought with studies: the conceptual shift doesnt really do anything.

I see really only two ways for this to matter. First, if you believed something about IQ just because it has intelligence in the name, you should go check that. But also youre stupid. In any case, theres been a lot of debate about IQ and pretty much anytime you think "hey this seems related to intelligence, propably correlates with IQ", theres already a dataset for that. So from the perspective of science as a whole, this too doesnt matter. Second, it might suggest new things to study. Equiped with this new informal understanding, maybe we do finally get the idea for how to make lasting increases in IQ. Again, even if the theory is true, that doesnt seem super likely. Weve already tried things all over the place, including some based on theories similar to the one advanced (that it is empty, only measures test-taking ability, etc). And again, the empirics of that are as they are.

So overall, even though it seemed to make over everything at the start, we are now left with, uh, maybe some new expectations extending the old, test results soon? I made this a top-level because it so well ilustrates the mechanism of ideology: How everything snaps into place, the world to be transformed only to, when examined step by step, deflate back to nothing or almost nothing. At the same time, its a relatively small example, without much social support yet, so I can still show the problem in a relatively short post.

All ideological frameworks rely on this mechanism of ABSTRACTION OF TAUTOLOGY, a loop of ideas with a wide circumference, ideally with a curvature so subtle that the logic seems flat.