Heritability Of Intelligence
Heritability Of IntelligenceFrom GamesEndLiteracy -
"A friend of mine has a precociously intelligent boy. Not surprising, really - she's a world-class astrophysicist with a mind like a Swiss-army knife herself ..."
There is demonstrably little correlation between intelligence of children and parents. The one advantage children of bright/well-educated parents tend to have is a good early education, since the mediocre public fare is usually well supplemented at home.
[Any reference for that assertion? I would have thought there is considerable such correlation. Most bright people I know had at least one bright parent. Not sure whether such correlation is as great for low-intelligence children, but I'd guess it isn't.]
Ok, the original comment was too broad reaching; thanks for calling me on it. I was calling the OP on ‘not surprising, really ...’ which may seem like common sense but does not really prove out. On the other hand, it hasn't been *proven* to be irrelevant either. This is part of the long standing nature vs nurture debate, which is far from over. So it would be unfair of me to find only references that support the above, overly strong statement.
The correlation parent->child has been measured at about 0.4, not huge, but not negligible (this information is easily available on the web. I can dig out specific references if you are really stuck). Unfortunately, there are two problems with using this number in support of anything 1) all intelligence test are fairly poor (mostly because we are testing for something we can't even define properly) 2) there are all sorts of other factors. The 'nature' side will argue that correlation is primarily due to genetics, while the nurture side will argue that it is primarily environment. Neither side has a convincing argument yet! What *should* be common sense (and does prove out) is that if you have a parent who is many standard deviations out in something like intelligence, their descendants will tend to head to the mean fairly quickly. Of course, since we are not instituting breeding programs, this does not invalidate the heredity hypothesis.
Perhaps, but how many children of exceptionally intelligent parents pair with others of such parentage, and so on, for any significant number of generations, and how come such information is available in view of the span of time needed to provide the evidence?
That is exactly what my last comment was referring to. I thought it was obvious, but perhaps I should have been more explicit.
I see, but tending to head towards the mean is what one would expect whether the correlation under discussion is fairly high or not, isn't it? I would appreciate some references as a few Google searches didn't turn up anything quantitative.
By the way, I have been exposed to a very large sample of bright people in all sorts of fields. My non-scientific sampling of people I know would show no bias to having highly intelligent parents...
But are the people you know highly intelligent or just bright?
For the purposes of this discussion, I was considering *only* those I know who are highly intelligent (ignoring the issues of defining this). The majority of my social group are mathematicians, but I know people with outstanding minds in physics, philosophy, fine arts (one composer, two painters), environmental studies, and history. The only reason I commented about this uncontrolled sample was in answer to the comment above, " Most bright people..."
And you also know their parents well enough to assess their intelligence? And you've used your estimates to calculate a correlation coefficient? OK, I haven't done that either, but it's much harder to assess lack of bias correctly than to assess significant correlation correctly.
Well, I was talking about friends and peers, or others where I know the family as well. So yes for the first part, but emphatically no for the second. I did say it was my unscientific sampling. The 0.4 number is not mine, that comes from real research. My claim of lack of bias is just my feeling about it.
Hmm. Hard to imagine what sort of observations would lead to such a feeling. Are you saying you don't notice any bias towards above-average intelligence in your sample of parents of bright people? Yes. Exactly.
If so, I'm somewhat dubious. Do the parents agree with your assessment of their intelligence? Would anybody?
If needed, we can arbitrarily assign "bright", "very bright", "highly intelligent", and "exceptionally intelligent" to scores which are above average by one, two, three and four standard deviations, respectively.
Standard deviations of what? Higher education (by which I mean graduate school, in this case) is highly correlated to good performance in most IQ tests. I would hazard a guess that almost everyone I currently spend any time with is at least 2-3 std. deviations out for these tests, but I wouldn't say that most of them were 'highly' or 'exceptionally' intelligent.
[See note below, and note that three SDs on IQ scores (above average) almost qualifies one for membership of International Mensa.' Hmmm. I had meant 2-3 don't know what happened there. I know at least five of us have been approached/solicited by mensa or mensans, because we were all having a chuckle about it one day'']
This discusses (among many other things) some of the problems with interpreting ‘heritability’: http://www.baylor.edu/~Rick_Duhrkopf/4330/MentalCapacity.pdf.
This has an overview of some of the issues in correlation studies of this nature: http://www.psychol.cam.ac.uk/pages/teaching/NST%20II%20Professor%20Mackintosh%20IQ/NJM%20IQ%20Lec%202.doc.
I'll read them, but note that I didn't claim the correlation is mainly due to heredity. The standard deviations referred to intelligence test scores, but many other tests would correlate quite well with intelligence tests, so would probably do as an alternative. Note that it would be better to use a sample group that doesn't just include people drawn from a small minority.
Okay. I've read them now.
The first has some interesting (and some slightly depressing) facts, but nothing relevant to the subject of the extent to which children's intelligence correlates with their parents' intelligence. Unless I gave you the wrong reference, it quotes a number of those exact statitstics. Will check later. The second is largely non-controversial. In relation to the current issue, it mentions (a) that "intelligence" testing goes back to 1916 (when it was used by the military), but (b) some years earlier, one study (no account given) correlating intelligence of parents and children yielded a correlation of 0.3. So again, nothing useful for the current discussion.
So the requested references don't seem to be forthcoming. I am not at the right place to dig them up at the moment. Tonight, if I have time when I get home. But both those sources support the low corrleation (compared to other factors) of .4, so I don't see exactly what your problem with them is?
I think this page is inappropriately named, since we have not been discussing heritability per se. My point is that when you're considering both nature and nurture together (since no separation of the two is being attempted), the intelligence of the child is considerably related to that of the parents (or at least one of the parents). Agree, we are wandering off the rails.
Of course, the heritability of intelligence is near 100 percent. We are, in fact, measuring the heritability of subtle differences - that is, the relatively unimportant matter of 15 rather than 10 questions answered correctly on an IQ test. If you compare significant differences in intelligence, such as that between humans and donkeys, you will find that humans are consistently much, much more intelligent. You will also find that, again, almost 100 percent of this difference is due to genetics.
Likewise, if you compare a run-of-the-mill physicist with a nobel prize-winning physicist, you can normally claim, correctly, that there is no genetic determinism involved in their differing achievements. Are there any father-son combinations of nobel prize winners? I don't think so. [Think again - William Henry Bragg and William Lawrence Bragg shared the physics prize in 1915.[1]] Thus, if we measure intelligence at the very upper end (nobel prize reception) then the heritability of intelligence can be said to be 0.0. I hope the reader will see the problem with this. You skew the results when you measure at the upper end, and the same phenomenon occurs with IQ tests. Simply being able to READ an IQ test suggests incredible (viz: human) intelligence, and that doesn't measure into whatever correlation we derive.
[1] Duly noted. The point doesn't change.
... mostly because we are testing for something we can't even define properly.
Damn well can we define intelligence! It's what's measured by an IQ test. IQ tests are reproducible (so they don't measure random junk), the results are non-uniformly distributed (so they measure real differences) and there is considerable evidence that intelligence is necessary for a career in the hard sciences (so it's not a test invented for the sake of inventing a test).
This constant bickering that "we can't define intelligence" by people who feel they didn't get enough themselves is annoying. Invent your own language if you want a different definition. Invent your own test, if you want to measure something else, but then also call it something else. If your mother tells you that you're as intelligent as your brother, it's not because it's true, but to get you both to STFU. NuffSaid.
That's a very pragmatic point. But it's a little bit like saying that gravity is what's measured by a spring scale. You have an experiment (spring scale=specific IQ-test), that leads indeed to reproducible results, but that doesn't mean that you have a theory of how gravity works. A definition of gravity allows you to create experiments and predict their outcome. Analogously, a theory of intelligence lets you create IQ-tests and predict their outcome. Also it should explain such phenomena like the http://de.wikipedia.org/wiki/Flynn-Effekt.
We seem to basically agree here. Gravity indeed is what a spring scale shows. If you have a theory that predicts something else, it is a theory of something else (at best, it might also be plain bullshit). In the same fashion you can invent a theory of intelligence and test it against the IQ-test. However, if you invent a theory and a different test, you simply have a theory of something else and should no longer call it intelligence. The word is taken, deal with it.
BTW, what does it mean to "have a theory of how gravity works?" My spring scale says it hurts when I drop a brick on my foot. I don't need to know why it hurts; knowing that it hurts is enough. Likewise, we can collect data about using our measure of intelligence without understanding the mechanisms behind it. In fact, without data no understanding will be possible. But this PC nonsense about "what IQ tests really measure" dilutes even the precious little knowledge we have. There are slow people, small people, weak people, socially inept people and dumb people. That fact isn't changed by debating definitions.
In this context, I expect "a theory about how gravity works" to be an explanation of how the scale and gravity are related. In particular, the force exerted by a spring is proportional to how far the spring is stretched. By attaching a fixed mass to the spring, the spring can be used to measure gravity. (BTW, it's not really a scale, but that isn't all that important.) In terms of IQ it would be an explanation of how IQ relates to the vague, handwavy "thing" we call intelligence. -- MartinShobe?
An artificial distinction for the sake of making a point. "Force" is no less abstract a concept than "skill to solve IQ test problems".
Seems my example was understood wrongly. The point was not that the concept is less abstract, but that the definition is ill-posed. The force is physically there, but where do the IQ-tests come from? "From the definition of IQ" makes it a circular argument.
"Force" is no less abstract a concept than "skill to solve IQ test problems".
The abstractness of the concepts is not the issue. The issue is describing the relationship between the measurement and the actually quality of interest. The concept of intillegence existed before IQ tests, and it is that concept that we are interested in. I'm pretty sure that the people who initially developed the IQ tests had some ideas about that relationship. What I don't know is what those are, and how well they have stood up.
Now I'm not defending the PC nonsense about IQ tests, but I don't know what IQ tests measure (other than the trivial "score on an IQ test on a given day"); I don't know what it's relationship to intelligence is; and without that relationship, I don't have any confidence in an IQ test's ablility to measure intelligence. -- MartinShobe?
How's that any different? "Force" is just "that which makes a spring get longer". Force is in no way "an actual quality", it's just a human made abstraction to describe "that which makes a spring get longer". So after one indirection we're still at "gravity is what is measured by a spring scale", and that's still as precise as "intelligence is what is measured by an IQ test".
That's right, they are equally precise, because they are both dead wrong. Gravity is a concept to explain why things fall, not a displacement on a spring. Intelligence is a concept to explain our problem solving abilities (and a host of other things), not a score an on IQ test. The words are taken. -- MartinShobe?
No, gravity doesn't explain anything. It's just a name for "masses accelerate towards (large) masses". After naming the concept and being able to measure it, we were able to gather more information, especially that gravitational force is proportional to inertial mass. Whatever we know, we needed a name to describe it, then could learn more about it.
Same with intelligence. The only difference is our definition of "intelligence". I maintain it's "that which is measured by the IQ test", while you have a different definition (which is deliberately fuzzy, I might add). So the meat of the matter is not that we "don't know what IQ tests measure", it's rather that no two people agree on a common definition of "intelligence". If we cannot agree what we're talking about (never mind if can measure it), we cannot meaningfully debate whether it's heritable.
So to summarize, you already have a definition of "intelligence". But your concept isn't measurable. What's the point of debating an essentially undefined concept? Heck, even the supposed implication that "intelligence" is a good thing/ separates us from the animals/ makes us more valuable humans/ whatever may be completely wrong. Really, what's the point of a void definition?
See: IntelligenceQuotient
EditText of this page
(last edited August 20, 2014)
or FindPage with title or text search