Eu acabo de ler um livro que, dentre outros assuntos, aborda esse. Cita um estudo de gêmeos que dá herdabilidade 0, mesmo em gêmeos... não acho a ser tomado "literalmente", porque como essa citação e outros trabalhos sugerem, as expressões dependem do ambiente, mesmo para coisas mais simples, resultados mais realmente determinados... como sementes para tamanos médio, grande e pequenho podem ter o efeito inverso ou embaralhado num ambiente diferente.
Trecho de um post interessante que encontrei:
[...]
Rushton has, over the years, compiled extensive data, much of which is considered bogus, showing that there is a correlation between, essentially, skin color, brain size, intelligence, and various aspects of sexuality.
In Rushton’s model, “White” People have average IQ’s of about 102, are only moderately sexual, and have a gestation period of about 9 months.
In contrast, “Black” people have an IQ averaging about 80, a high and difficult to control sex drive, and a measurably shorter gestation length.
“Oriental” people have an average IQ of somewhat over 100, and a diminished sex drive.
Rushton argues that these differences are deep and genetic, owing to thousands of years of genetic evolution as separate subspecies. Ruston’s research is explicitly the basis for parallel assertions made in Murray and Herrnstein’s book “The Bell Curve.” In the Bell Curve, Murray and Herrnstein make the argument that whatever the cause of these differences, they are immutable or at least economically impossible to really overcome, and thus society must simply adjust to them. For instance, if girls can’t, on average, really learn math, then programs that encourage girls to take math classes and to try harder in math are a waste of money and should not be funded. Similarly, if one of the characteristics (as asserted by Rushton and taken up by Murray and Herrnstein) is that “Blacks” can’t lead, then efforts to recruit “Blacks” into leadership roles are a waste of time and should not be funded with tax dollars.
There are a number of false assumptions and highly suspicious, if not totally bogus, data sets used to develop these arguments. For instance, IQ is known to have a high heritability value. Heritability is a correlational statistic that indicates the likelihood that if one individual has a certain trait (in this case, a certain IQ) that another individual will have that same trait based on degree of relatedness.
But correlation does not necessarily imply causality. This is a cautionary statement hardly ever mentioned when heritability comes up in discussion, although it usually does come up when correlation is mentioned. This indicates to me that most people don’t realize that heritability is about correlation, as well as about variance.
With respect to the latter (variance) it is worht considering heritability in a bit more detail, because it is very often misused, especially in the present discussion.
For instance note that the heritability of how many heads you have … head number across Hom sapiens … is 0. Head number has a heratibility coeficient of zero.
If Heritability is equal to “genetic” or “likelihood that a trait … such as intelligence … is passed on by genes” then Having one head is notinherited, or the number of heads you have is not genetic.
In truth, heritability is too limited of a measure to use it to investigate a trait like intelligence, represented by what is usually called “IQ” or any similar statistic or any nominal characteristic (”can lead” vs. “not good leaders” etc.).
Heritability does not distinguish in and of itself between traits passed on with genes from traits passed on via culture, learning, environment, and so on. For instance, which language a person speaks has a very high heritability value, but this “trait” is entirely, 100% learned with absolutely no genetic component whatsoever. Twin studies have been used to suggest that IQ has a component of heritability that is genetic since it is more correlated in twins than in, say, full sibs. However, non-genetic traits can follow the same pattern. Non-genetic traits can show this pattern because, with respect to environments, full sibs who are not twins do not share the same environment as twins. (And for other reasons.)
Another attempt at showing that IQ is genetically determined by using correlational analysis, carried out by University of Minnesota “Twin” Researchers as part of the famous twin studies, was to examine differences and correlations among twins who were separated at birth. If one can re-unite individuals who were separated at the time they were born and then raised in different environments, one would have an interesting set of data to look at heritability.
These “separated at birth” studies were heavily criticized when it was realized by the consumers of these research reprots that non, or almost none, of the twin sets were really separated at birth. Most were cases of two individuals at some point in their childhood being parceled out into different units of an extended family. For example, I myself was raised, according to the Twin Study criteria, separately from my brother because he lived upstairs and I lived downstairs in a two-family house. But in fact, we were simply a large family spread out across two flats, and at no point during my childhood did we distinguish among these two areas as different “apartments.” But the Twin Studies criteria would have put my brother and I in the “separated” category (… I quickly add, we are not twins, but you get the point).
Rushton’s research has been criticized in other ways as well. His categories are not valid groupings .. for instance, his brain size data for “Orientals” is mainly from Thailand, but is “IQ” data for “Orientals” is mainly from Japan. He uses a number around 70 for the IQ of “Africans” from Africa, as opposed, say, to African Americans. This number is from a single set of tests given several decades ago in South Africa. The test was invalid. It consisted of a non-written (visual) test developed by Yerkes during World War I for use by the US Army but rejected after a short time because it seemed useless. The test included numerous culturally-bound questions, such as one question where two tennis players are shown on a court with no net. The viewer is asked to indicate what is missing in the picture (”The Net” being the correct answer). The test was administered to Zulu middle school age girls who would never have seen a tennis court. The fact that they “scored” a 70 on the test is testament to their intelligence … they did about as well as Irish and Italian Americans did on this test early in the 20th century.
The problems with Rushton’s data are myriad. Brain size information is actually estimated, for most of the data, from skull size, but the skull size measure is actually a hat-size measurement taken while conscripts were entering the army, or something similar, not measured with calibrated instruments. The raw “hat size” data is converted based on a “skull thickness” calibration that Rushton, in turn, based on highly selected data. For instance, even though we know that some of the thinnest-skulled populations in the world are African, Rushton picked some of the thickest-skulled specimens available to create an African “Skull Thickness” fudge factor. In this way, estimated brain size for the “Black” part of his data set were reduced across the board with an invalid adjustment. In essence the brain size calculation involved, variously, something like: Take a hat size from army data, then estimate brain size using a formula that assumes the individual has an extra thick skull if they are “Black,” an extra thin skull if they are “Oriental” and an average skull if they are “White.”
A possibly even greater effect comes from another fudge factor in which Rushton uses an invalid set of assumptions, and associated invalid formulae, to adjust brain size given assumptions about body size. The way this works is too complex to go into here. Suffice it to say that this fudge factor guaranteed an incorrect reduction in brain size for each point in the “Black” data set.
In short, Rushton’s data are bad and his analysis is highly questionable, to the extent that one must question not only his analysis but also his motives; the Bell Curves’ incorporation of these bogus data … in a book that was only published because of the availability of funding from highly conservative organizations with a history of supporting racist doctrine … is highly questionable and generally offensive; and James Watson’s remarks, which clearly state that Africans are an inferior race, based in turn on this information, are outrageous.
http://gregladen.com/wordpress/?p=1536
Fontes excelentes, para quem gostar de estatística e genética e querer gastar algumas horas estudando material bem esclarecedor para a controvérsia:
"g", a statistical myth
http://www.cscs.umich.edu/~crshalizi/weblog/523.html
Yet more on the heritability and malleability of IQ:
http://www.cscs.umich.edu/~crshalizi/weblog/2007/09/
