Well, there’s sort of a spectrum of different positions one could take with regards to the realism of g:
One could argue that g is pure artifact of the method, and not relevant at all. For instance, some people argue that IQ tests just measure “how good you are at tests”, argue that things like test-taking anxiety or whatever are major influences on the test scores, etc..
One could argue that g is not a common underlying cause of performance on tests, but instead a convenient summary statistic; e.g. maybe one believes that different abilities are connected in a “network”, such that learned skill at one ability transfers to other “nearby” abilities. In that case, the g loadings would be a measure of how central the tests are in the network.
One could argue that there are indeed common causes that have widespread influence on cognitive ability, and that summing these common causes together gives you g, without necessarily committing to the notion that there is some clean biological bottleneck for those common causes.
One could argue that there is a simple biological parameter which acts as a causal bottleneck representing g.
Of these, the closest position that your post came to was option 2, though unlike e.g. mutualists, you didn’t commit to any one explanation for the positive manifold. That is, in your post, you wrote “It does not mean that number causes test performance to be correlated.”, which I’d take to be distancing oneself from positions 3+. Meanwhile, out of these, my comment defended something inbetween options 3 and 4.
You seem to be asking me about option 4. I agree that strong versions of option 4 seem implausible, for probably similar reasons to you; it seems like there is a functional coordination of distinct factors that produce intelligence, and so you wouldn’t expect strong versions of option 4 to hold.
However, it seems reasonable to me to define g as being the sum of whichever factors have an positive effect on all cognitive abilities. That is, if you have some genetic variant which makes people better at recognizing patterns, discriminating senses, more knowledgeable, etc., then one could just consider this variant to be part of g. This would lead to g being composed of a larger number of heterogeneous factors, some of which might possibly not be directly observable anymore (e.g. if they are environmental factors that can no longer be tracked); but I don’t see anything wrong with that? It would still satisfy the relevant causal properties.
(Of course then there’s the question of whether all of the different causes have identical proportions of effect on the abilities, or if some influence one ability more and others influence another ability more. The study I linked tested for this and only included genetic variants that had effects in proportion to the g factor loadings. But I’m not sure their tests were well-powered enough to test it exactly. If there is too much heterogeneity between the different causes, then it might make sense to split them into more homogeneous clusters of causes. But that’s for future research to figure out.)
Thanks, very clear! I guess the position I want to take is just that the data in the post gives reasonable evidence for g being at least the convenient summary statistic in 2 (and doesn’t preclude 3 or 4).
What I was really trying to get at in the original quote is that some people seem to consider this to be the canonical position on g:
Factor analysis provides rigorous statistical proof that there is some single underlying event that produces all the correlations between mental tests.
There are lots of articles that (while not explicitly stating the above position) refute it at length, and get passed around as proof that g is a myth. It’s certainly true that position 5 is false (in multiple ways), but I just wanted to say that this doesn’t mean anything for the evidence we have for 2.
I agree that a simple factor analysis does not provide anything even close to proof of 3 or 4, but I think it’s worth noting that the evidence on g goes beyond the factor-analytic, e.g. with the studies I linked.
Thanks for pointing out those papers, which I agree can get at issues that simple correlations can’t. Still, to avoid scope-creep, I’ve taken the less courageous approach of (1) mentioning that the “breadth” of the effects of genes is an active research topic and (2) editing the original paragraph you linked to to be more modest, talking about “does the above data imply” rather than “is it true that”. (I’d rather avoid directly addressing 3 and 4 since I think that doing those claims justice would require more work than I can put in here.) Anyway, thanks again for your comments, it’s useful for me to think of this spectrum of different “notions of g”.
Well, there’s sort of a spectrum of different positions one could take with regards to the realism of g:
One could argue that g is pure artifact of the method, and not relevant at all. For instance, some people argue that IQ tests just measure “how good you are at tests”, argue that things like test-taking anxiety or whatever are major influences on the test scores, etc..
One could argue that g is not a common underlying cause of performance on tests, but instead a convenient summary statistic; e.g. maybe one believes that different abilities are connected in a “network”, such that learned skill at one ability transfers to other “nearby” abilities. In that case, the g loadings would be a measure of how central the tests are in the network.
One could argue that there are indeed common causes that have widespread influence on cognitive ability, and that summing these common causes together gives you g, without necessarily committing to the notion that there is some clean biological bottleneck for those common causes.
One could argue that there is a simple biological parameter which acts as a causal bottleneck representing g.
Of these, the closest position that your post came to was option 2, though unlike e.g. mutualists, you didn’t commit to any one explanation for the positive manifold. That is, in your post, you wrote “It does not mean that number causes test performance to be correlated.”, which I’d take to be distancing oneself from positions 3+. Meanwhile, out of these, my comment defended something inbetween options 3 and 4.
You seem to be asking me about option 4. I agree that strong versions of option 4 seem implausible, for probably similar reasons to you; it seems like there is a functional coordination of distinct factors that produce intelligence, and so you wouldn’t expect strong versions of option 4 to hold.
However, it seems reasonable to me to define g as being the sum of whichever factors have an positive effect on all cognitive abilities. That is, if you have some genetic variant which makes people better at recognizing patterns, discriminating senses, more knowledgeable, etc., then one could just consider this variant to be part of g. This would lead to g being composed of a larger number of heterogeneous factors, some of which might possibly not be directly observable anymore (e.g. if they are environmental factors that can no longer be tracked); but I don’t see anything wrong with that? It would still satisfy the relevant causal properties.
(Of course then there’s the question of whether all of the different causes have identical proportions of effect on the abilities, or if some influence one ability more and others influence another ability more. The study I linked tested for this and only included genetic variants that had effects in proportion to the g factor loadings. But I’m not sure their tests were well-powered enough to test it exactly. If there is too much heterogeneity between the different causes, then it might make sense to split them into more homogeneous clusters of causes. But that’s for future research to figure out.)
Thanks, very clear! I guess the position I want to take is just that the data in the post gives reasonable evidence for g being at least the convenient summary statistic in 2 (and doesn’t preclude 3 or 4).
What I was really trying to get at in the original quote is that some people seem to consider this to be the canonical position on g:
Factor analysis provides rigorous statistical proof that there is some single underlying event that produces all the correlations between mental tests.
There are lots of articles that (while not explicitly stating the above position) refute it at length, and get passed around as proof that g is a myth. It’s certainly true that position 5 is false (in multiple ways), but I just wanted to say that this doesn’t mean anything for the evidence we have for 2.
I agree that a simple factor analysis does not provide anything even close to proof of 3 or 4, but I think it’s worth noting that the evidence on g goes beyond the factor-analytic, e.g. with the studies I linked.
Thanks for pointing out those papers, which I agree can get at issues that simple correlations can’t. Still, to avoid scope-creep, I’ve taken the less courageous approach of (1) mentioning that the “breadth” of the effects of genes is an active research topic and (2) editing the original paragraph you linked to to be more modest, talking about “does the above data imply” rather than “is it true that”. (I’d rather avoid directly addressing 3 and 4 since I think that doing those claims justice would require more work than I can put in here.) Anyway, thanks again for your comments, it’s useful for me to think of this spectrum of different “notions of g”.