Sat in a philosophy class today.

[Alawen]

I don't think you read that paper. I just did and it has several problems. He is operating under the assumption that IQ tests from different sources are designed to correlate. That assertion has no basis. More importantly, the entire article is devoted to debunking Spearman's unitary g factor model. If you've been paying attention, you will note that I, following Cattell, never mentioned that earlier model, and that I discussed the two-factor model, Gf and Gc. The more recent study I linked today identifies three distinct factors.

What really gets my nuts twisted, though, is that after spending almost two hours reading his horrible prose and looking up everything I wasn't positive that I understood, he writes off the whole exercise as too distracting from his real work to finish and delivers a weak conclusion: he doubts that there is a general factor of intelligence, but he's been wrong before. Along the way, he pretty much trashes all social science. I bet he's popular with other departments at CMU.

Well, that straw man is fucking dead and burned to the ground. You sure did a number on him.

[Alawen]

I just read through all of his notes while thinking about your idea of motivation as the real distinguishing factor. On a complete tangent, I like his ideas about the purpose of democracy.

I'm trying to wrap my mind around your general model of how the human brain works. Are you saying that turning on one burner of the stove is a different thought process than turning on a different burner? Or just that finding my keys is different from locking the door? In either case, essentially no commonality at all and that learning something is not useful for learning any other thing however related?

Regarding your idea that despite some possible biological differences, everything really comes down to motivation, isn't that a bit rhetorical in practical terms? You seem to be saying that person A isn't really smarter than person B, he just wants to think more. How is that functionally different?

[Splorf22]

Originally Posted by Alawen [You must be logged in to view images. Log in or Register.]

He is operating under the assumption that IQ tests from different sources are designed to correlate. That assertion has no basis.

I actually rechecked and didn't see this. Anyway I'm not sure you understand his point, maybe because its mostly technical and I'm not sure I can summarize it easily. The key idea he is trying to get across is that G is based on a misuse of mathematics. Maybe one way to put it is the difference between correlation and causation: because various tests are correlated you can run a regression and it will automatically spit out G. But G is just a mathematical artifact that doesn't necessarily represent anything physical, and he shows this in two ways: the fact that G fails confirmatory factor analysis, and the fact that a different model can give the same conclusions. Which is why you probably find his conclusion so wishywashy, he's not actually disproving G, simply saying that its not necessary (what makes it unlikely is the kind of neuroscientific argument I have been making).

Originally Posted by Alawen

More importantly, the entire article is devoted to debunking Spearman's unitary g factor model. If you've been paying attention, you will note that I, following Cattell, never mentioned that earlier model, and that I discussed the two-factor model, Gf and Gc. The more recent study I linked today identifies three distinct factors.

Well he mainly focuses on G but he also talks about the five factor model. Really the mere fact that his alternate model works already casts serious doubts on any other one.

What really gets my nuts twisted, though, is that after spending almost two hours reading his horrible prose and looking up everything I wasn't positive that I understood, he writes off the whole exercise as too distracting from his real work to finish and delivers a weak conclusion: he doubts that there is a general factor of intelligence, but he's been wrong before. Along the way, he pretty much trashes all social science. I bet he's popular with other departments at CMU.

Personally I think this makes him fun to read, but maybe that's just me [You must be logged in to view images. Log in or Register.]

[Alawen]

"The correlations among the components in an intelligence test, and between tests themselves, are all positive, because that's how we design tests." You proposed this as proof that "all of those theories of intelligence are basically BS," but it does no such thing. He doesn't even propose such a thing. What he is asserting is that exploratory factor analysis does not (and cannot) prove the existence of g.

I can't figure out if you're trying to bamboozle me or if you don't understand the essay.

[Alawen]

Incidentally, the five factor model (more commonly referred to as the Big Five) is a personality model analogous to Myers-Briggs, not a theory of intelligence.

[Alawen]

I'm reading the Glymour paper now, just out of curiosity. Herrnstein and Murray really get hammered from every possible direction.

[Splorf22]

Originally Posted by Alawen [You must be logged in to view images. Log in or Register.]

I just read through all of his notes while thinking about your idea of motivation as the real distinguishing factor. On a complete tangent, I like his ideas about the purpose of democracy.

I'm trying to wrap my mind around your general model of how the human brain works. Are you saying that turning on one burner of the stove is a different thought process than turning on a different burner? Or just that finding my keys is different from locking the door? In either case, essentially no commonality at all and that learning something is not useful for learning any other thing however related?

Regarding your idea that despite some possible biological differences, everything really comes down to motivation, isn't that a bit rhetorical in practical terms? You seem to be saying that person A isn't really smarter than person B, he just wants to think more. How is that functionally different?

I guess I am not saying this very well. What I am saying is I believe what the brain learns is extremely specific and there is not much transfer. Learning about American history doesn't help you with mathematics. Learning about integration by parts won't help you with whether power series converge. And so on.

But at the same time we clearly see correlations between various intellectual tasks. On average people that are good with math DO know more about American history than people who aren't. And if you believe the brain is specific, the only logical explanation is that a brain that has been exposed to math is more likely to have been exposed to american history. And the mostly likely reason for that is that this person chose to study.

Also my theory makes vastly different predictions than g. For example g predicts that black people are dumb: every IQ test has blacks about 1 standard deviation below whites. On the other hand my theory predicts that blacks would simply rather play basketball. And in fact IIRC there are studies that black children raised by white parents have comparable IQ scores to white children (don't quote me on that).

[Splorf22]

Originally Posted by Alawen [You must be logged in to view images. Log in or Register.]

"The correlations among the components in an intelligence test, and between tests themselves, are all positive, because that's how we design tests." You proposed this as proof that "all of those theories of intelligence are basically BS," but it does no such thing. He doesn't even propose such a thing. What he is asserting is that exploratory factor analysis does not (and cannot) prove the existence of g.

Ah its in the notes. I actually find that a bit puzzling but I seem to recall reading somewhere that test designers do look at other tests. Regardless I don't think it's a critical point; the fact is the tests are correlated and the reason has no meaning on what he is saying.

When I say "all those theories of intelligence are bs" I am referring to all of these factor models that attempt to break up intelligence (and by extension personality) into nice linear combinations. Just because you can do some statistics and find some correlations and make models that may even predict with a moderate degree of accuracy doesn't mean that those models directly reflect reality.

I can't figure out if you're trying to bamboozle me or if you don't understand the essay.

Well then, in your opinion what is the summary of the essay?

[Alawen]

Originally Posted by Splorf22 [You must be logged in to view images. Log in or Register.]

I guess I am not saying this very well. What I am saying is I believe what the brain learns is extremely specific and there is not much transfer. Learning about American history doesn't help you with mathematics. Learning about integration by parts won't help you with whether power series converge. And so on.

But at the same time we clearly see correlatifons between various intellectual tasks. On average people that are good with math DO know more about American history than people who aren't. And if you believe the brain is specific, the only logical explanation is that a brain that has been exposed to math is more likely to have been exposed to american history. And the mostly likely reason for that is that this person chose to study.

Also my theory makes vastly different predictions than g. For example g predicts that black people are dumb: every IQ test has blacks about 1 standard deviation below whites. On the other hand my theory predicts that blacks would simply rather play basketball. And in fact IIRC there are studies that black children raised by white parents have comparable IQ scores to white children (don't quote me on that).

I think I understand what you're saying now, but I'm not sure how it applies to IQ testing. IQ tests are not like achievement tests or college placement tests, although parts of the ASVAB (do they still use that?) are similar to IQ tests. They aren't sectioned by discipline or intended to focus on accumulated knowledge. Regardless of your interest focusing on either history or math, you should be increasing your crystallized intelligence.

I don't know of any studies like the one you suggest, though I do know there is a world-wide trend toward higher IQs as books and learning have become more ubiquitous. The tests have been repeatedly adjusted to maintain a median score of 100 with 15 point standard deviations. This would support your assertion that intelligence correlates with exposure to knowledge.

I should probably mention that I am not a true believer in any particular model. I think they're all interesting. What you are proposing is very similar to Gardner's model, which I often think of when I see art or watch dancing and think, "That is brilliant."

[Alawen]

Originally Posted by Splorf22 [You must be logged in to view images. Log in or Register.]

Ah its in the notes. I actually find that a bit puzzling but I seem to recall reading somewhere that test designers do look at other tests. Regardless I don't think it's a critical point; the fact is the tests are correlated and the reason has no meaning on what he is saying.

When I say "all those theories of intelligence are bs" I am referring to all of these factor models that attempt to break up intelligence (and by extension personality) into nice linear combinations. Just because you can do some statistics and find some correlations and make models that may even predict with a moderate degree of accuracy doesn't mean that those models directly reflect reality.



Well then, in your opinion what is the summary of the essay?

It wasn't in the notes, it was early in the essay, and there is absolutely no basis for it. The tests are expensive and in competition. Stanford-Binet and Wechsler are the survivors after the deaths of a number of other tests.

The Big Five, like Myers-Briggs, has problems with test-retest. Comparing intelligence testing to personality profile testing is a stretch.

I already stated my one-sentence summary: he is asserting that exploratory factor analysis cannot prove the existence of g.