Amusingly, Razib’s post does not include the word “genetic”. I can’t tell if that was intentional, but in any case, when a trait is highly heritable, that doesn’t mean it’s genetic. One nice example is accent. It’s also a nice example of a trait that a teacher would find really hard to change, unles given huge authority over the kid’s entire life. Maybe basic math aptitude is similar.
ETA: this comment is wrong by the technical definition of heritability, see Vladimir_M’s replies. I should have said something like “has high correlation between parents and children”.
The discussion of accent in that dialog is a neat rhetorical trick, but its main premise is false. If you were to examine the heritability of accent using the standard methods of behavioral genetics, it would turn out to be near zero. (Maybe some confounding factors would yield a small spurious heritability, but there’s no way you’d get a “highly heritable” result.)
Some of the other cited facts are also dubious or exaggerated. For example, while accent of adults is no longer as perfectly plastic as before adolescence, it’s obviously absurd to claim that “nothing that happens after early adolescence makes an impact on it.”
(Also, kids’ accent can be easily influenced if you can just place them into a peer group with the desired accent. No such simple solution exists for traits that are known to be heritable.)
If you were to examine the heritability of accent using the standard methods of behavioral genetics, it would turn out to be near zero.
Can you explain in more detail? I don’t know much about heritability, but would be pretty surprised if Shalizi turned out to be wrong on a question of fact.
You can take any of the usual lines of evidence for heritability, and the result will be negative. Unrelated kids growing up in the same linguistic environment end up with the same accent, while related kids, even identical twins, growing up in different linguistic environments end up with completely different accents—with no more similarity between them compared to the other randomly selected kids from these different environments.
In contrast, with IQ, you get dramatically different results. If you discover a lost twin brother who grew up in Hungary, his accent won’t be any more similar to yours than a random Hungarian’s—whereas his IQ test results would be similar to yours with much more than random chance.
(I am ignoring here some minor factors like e.g. speech impediments due to hereditary conditions. But clearly the context is normal linguistic variation.)
I don’t know that much about heritability, but would be pretty surprised if Shalizi turned out to be wrong on a question of fact.
When it comes to sheer intellectual ability, I admit that I’m not worthy to sharpen Shalizi’s pencils. Unfortunately, he is not reliable on ideologically charged topics; when discussing these, he will not use his abilities to clarify the matter, but rather to make the best lawyerly case for his favored side. I wish it were otherwise—I’d be delighted to see someone as smart as him try to make sense of these controversial and muddled topics—but that’s the way it is.
Thanks! It looks like you’re right. This is bizarre, Shalizi says accent is “highly heritable”, yet in another post he explains that estimated heritability of accent will be quite low if you measure it properly (see the section “Cultural transmission”). Edited my comment.
Shalizi’s usage seems perfectly reasonable to me. People use the phrase “highly heritable” to refer to the experiments they actually do, not to the theoretically best experiments they could do. Shalizi claims that performing the same experiments with accent would yield a conclusion of “heritable.” (though he backs off of “highly” and is generally evasive about quantity)
Also, kids’ accent can be easily influenced if you can just place them into a peer group with the desired accent.
AIUI, that’s where people’s accents are generally set, in school from about six to twelve. (You can self-modify afterwards, but that’s where people start.)
Yeah, the point that genetic != heritable is really important- religion is highly inherited but obviously there’s no gene for being Christian. Some of this inheritance may be due to cultural or socioeconomic effects, but at this point, the data with twin studies seems to suggest that a lot of the heritable differences in intelligence are genuinely genetic.
Yeah. Cultural problems are a tricky one. It’s repeatedly shown you can get 15 IQ points from non-genetic differences. So you do need to pay attention to the apparently stupid or shiftless kids so they can in fact be at least some of all they can be. But not neglect the bright ones.
I’ll note here as well: it’s bloody hard work being a teacher.
(Having met the kids in my daughter’s class, her teachers deserve a medal. And they’re pretty nice kids.)
Amusingly, Razib’s post does not include the word “genetic”. I can’t tell if that was intentional, but in any case, when a trait is highly heritable, that doesn’t mean it’s genetic. One nice example is accent. It’s also a nice example of a trait that a teacher would find really hard to change, unles given huge authority over the kid’s entire life. Maybe basic math aptitude is similar.
ETA: this comment is wrong by the technical definition of heritability, see Vladimir_M’s replies. I should have said something like “has high correlation between parents and children”.
The discussion of accent in that dialog is a neat rhetorical trick, but its main premise is false. If you were to examine the heritability of accent using the standard methods of behavioral genetics, it would turn out to be near zero. (Maybe some confounding factors would yield a small spurious heritability, but there’s no way you’d get a “highly heritable” result.)
Some of the other cited facts are also dubious or exaggerated. For example, while accent of adults is no longer as perfectly plastic as before adolescence, it’s obviously absurd to claim that “nothing that happens after early adolescence makes an impact on it.”
(Also, kids’ accent can be easily influenced if you can just place them into a peer group with the desired accent. No such simple solution exists for traits that are known to be heritable.)
Can you explain in more detail? I don’t know much about heritability, but would be pretty surprised if Shalizi turned out to be wrong on a question of fact.
You can take any of the usual lines of evidence for heritability, and the result will be negative. Unrelated kids growing up in the same linguistic environment end up with the same accent, while related kids, even identical twins, growing up in different linguistic environments end up with completely different accents—with no more similarity between them compared to the other randomly selected kids from these different environments.
In contrast, with IQ, you get dramatically different results. If you discover a lost twin brother who grew up in Hungary, his accent won’t be any more similar to yours than a random Hungarian’s—whereas his IQ test results would be similar to yours with much more than random chance.
(I am ignoring here some minor factors like e.g. speech impediments due to hereditary conditions. But clearly the context is normal linguistic variation.)
When it comes to sheer intellectual ability, I admit that I’m not worthy to sharpen Shalizi’s pencils. Unfortunately, he is not reliable on ideologically charged topics; when discussing these, he will not use his abilities to clarify the matter, but rather to make the best lawyerly case for his favored side. I wish it were otherwise—I’d be delighted to see someone as smart as him try to make sense of these controversial and muddled topics—but that’s the way it is.
Thanks! It looks like you’re right. This is bizarre, Shalizi says accent is “highly heritable”, yet in another post he explains that estimated heritability of accent will be quite low if you measure it properly (see the section “Cultural transmission”). Edited my comment.
Shalizi’s usage seems perfectly reasonable to me. People use the phrase “highly heritable” to refer to the experiments they actually do, not to the theoretically best experiments they could do. Shalizi claims that performing the same experiments with accent would yield a conclusion of “heritable.” (though he backs off of “highly” and is generally evasive about quantity)
AIUI, that’s where people’s accents are generally set, in school from about six to twelve. (You can self-modify afterwards, but that’s where people start.)
Yeah, the point that genetic != heritable is really important- religion is highly inherited but obviously there’s no gene for being Christian. Some of this inheritance may be due to cultural or socioeconomic effects, but at this point, the data with twin studies seems to suggest that a lot of the heritable differences in intelligence are genuinely genetic.
Interestingly, how seriously people take religion appears to be genetic.
Yeah. Cultural problems are a tricky one. It’s repeatedly shown you can get 15 IQ points from non-genetic differences. So you do need to pay attention to the apparently stupid or shiftless kids so they can in fact be at least some of all they can be. But not neglect the bright ones.
I’ll note here as well: it’s bloody hard work being a teacher.
(Having met the kids in my daughter’s class, her teachers deserve a medal. And they’re pretty nice kids.)