eric@snark.UUCP (06/13/87)
In article <2172@mmintl.UUCP>, franka@mmintl.UUCP (Frank Adams) writes: > I am curious about why "is intelligence heritable?" was included in this > list. This seems to me to be very much a scientific question, albeit one we > can't really answer yet. Intelligence is *extremely* heritable. The results that show this are solid but not as well known as they might be due to the fact that hereditarianism in general is out of fashion and 'politically incorrect'. Empirical evidence: the I.Q. and aptitudes of identical twins raised apart are quite strongly correlated -- sorry, I don't have numerical statistics handy. 'Intelligence' is a multidimensional composite of aptitudes in several distinct areas. However, factor analysis on the results suggest that about half the variance is due to a single 'hidden variable' which some psychometricians call xxxx's q, where xxxx is a psychometrician's name that I can't recall right now -- something like Jensen or Ivorsen. This q factor is about 80% heritable. Certain specific talents generally thought of as forms of intelligence -- such as mathematical and musical ability -- have long been known to be that heritable. All of this data is discussed, from a strongly environmentarian viewpoint, in Stephen Jay Gould's _The_Mismeasure_of_Man_ -- a book I recommend for facts and style while disagreeing almost totally with its analysis and conclusions. Gould is a fine and lucid writer when his Marxist sympathies are dormant. What seems to be true is that heredity sets a fairly hard upper limit on the capability of the brain to do particular kinds of information processing (things like, say, spatial visualization). Whether that limit is ever pushed depends on the individual's environment. Nearly all psychometricians agree on this much. Go any further and you get into controversies with such heavy political overtones that it's hard to see light through the smoke. Environmentarians believe the inherited limits are generally well above the performance level of typical individuals, so that there's plenty of room for improving the masses by improving things like richness of early environment, education, etc. This tends to correlate a more general beief in human perfectibility and with politics that are statist-liberal, socialist or Marxist. Hereditarians believe that individuals routinely push their inherited limits; you can and should train an individual up to maximum, but unless that person has been lucky in the genetic draw that maximum will be at or below average for the population. This tends to correlate with conservative, religious-fundamentalist, reactionary and fascist politics. Historically, psychometrics has swung like a pendulum between hereditarian and environmentarian extremes, often in a way correlated with the political ideology of the period's dominant culture. The late 1800s saw an extreme of hereditarian dogma (the Nazi master-race bulls**t was a sort of degraded pop version of stuff that was mainstream scholarly anthropology in 1850-1900). The 1960s saw an extreme of environmentarian dogma, more recently corrected by the influence of Wilson and the sociobiological school. My personal opinion: the hereditarian claim is unpleasant, but fits the observed statistical facts better than most versions of the environmentarian thesis. Attempts by Gould and others to dismiss the q-factor as an artifact aren't convincing, particularly since they tend to fall back on ad-hominem attacks on infamous hereditarians of the past when reasoned argument fails. (My politics, you ask? A fair question. Libertarian -- I loathe the left and right with equal intensity...:-)) -- Eric S. Raymond UUCP: {{seismo,ihnp4,rutgers}!cbmvax,sdcrdcf!burdvax}!snark!eric Post: 22 South Warren Avenue, Malvern, PA 19355 Phone: (215)-296-5718
g-rh@cca.UUCP (06/15/87)
[Note -- I have directed followups to sci.bio and sci.misc, since I don't really feel that this is a sci.philosophy.tech subject.] In article <123@snark.UUCP> eric@snark.UUCP (Eric S. Raymond) writes: >In article <2172@mmintl.UUCP>, franka@mmintl.UUCP (Frank Adams) writes: >> I am curious about why "is intelligence heritable?" was included in this >> list. This seems to me to be very much a scientific question, albeit one we >> can't really answer yet. > >Intelligence is *extremely* heritable. The results that show this are solid >but not as well known as they might be due to the fact that hereditarianism >in general is out of fashion and 'politically incorrect'. > >Empirical evidence: the I.Q. and aptitudes of identical twins raised apart are >quite strongly correlated -- sorry, I don't have numerical statistics handy. Well now, the evidence is not nearly so strong as you claim. The principle results on identical twins were those of Cyril Burt's, which is where the conventional figure of 80% heritability comes from. However Burt's data and results were forged. (I don't know if this is mentioned in Gould, but it is well established.) The reason you don't have the statistics at hand for identical twins is because it doesn't exist, apart from Burt's forged data -- there is some data but it is very sparse. >'Intelligence' is a multidimensional composite of aptitudes in several >distinct areas. However, factor analysis on the results suggest that about >half the variance is due to a single 'hidden variable' which some >psychometricians call xxxx's q, where xxxx is a psychometrician's name >that I can't recall right now -- something like Jensen or Ivorsen. This I have a copy of Cattell's classic work on the subject. On the whole one has to be impressed, and I am inclined to believe that there is such a thing as general intelligence, with a reasonably high heredit- ability. However I have a lot of caveats. Firstly, factor analysis is a linear analysis method which can break down if there are nonlinearities in the underlying factors. Secondly the correlation analysis is the isolation of agreement between results on tests (lots of them). The tests themselves, however, are not culture free (if there is such a thing as a culture free test.) This is a notorious problem -- the early workers in IQ testing, Terman, et. al. had no trouble believing that immigrants from Eastern Europe all had abysmally low IQ's because that is what their tests showed. Psychometricians today have a much clearer understanding of these difficulties. Again, the degree of cultural inhomogeneity affects the determination of the heritability figure [which is not a simple fixed number, by the way.] >q factor is about 80% heritable. > >Certain specific talents generally thought of as forms of intelligence -- such >as mathematical and musical ability -- have long been known to be that >heritable. > >All of this data is discussed, from a strongly environmentarian viewpoint, >in Stephen Jay Gould's _The_Mismeasure_of_Man_ -- a book I recommend for >facts and style while disagreeing almost totally with its analysis and >conclusions. Gould is a fine and lucid writer when his Marxist sympathies >are dormant. Marxist sympathies!!?? What an odd thing to say. I will also endorse the book, even though I also have reservations about some of his analysis. > >What seems to be true is that heredity sets a fairly hard upper limit on >the capability of the brain to do particular kinds of information >processing (things like, say, spatial visualization). Whether that >limit is ever pushed depends on the individual's environment. Nearly >all psychometricians agree on this much. Go any further and you get >into controversies with such heavy political overtones that it's hard >to see light through the smoke. > True enough, that there "seem" to be hard limits set by heredity. Since the nature and operation of the underlying factors are unknown it is indeed hard to say much more than that legitimately. The politics arise because people extrapolate wildly from what is actually known and use those extrapolations to buttress preconceived political viewpoints. It's all bad science. We have nothing like a good physical model of general intelligence -- we don't even have a good identification of what is meant by the term. >Environmentarians believe the inherited limits are generally well above >the performance level of typical individuals, so that there's plenty >of room for improving the masses by improving things like richness of >early environment, education, etc. This tends to correlate a more >general beief in human perfectibility and with politics that are >statist-liberal, socialist or Marxist. A relevant point: In Japan there is a 'caste' roughly equivalent to the untouchables of India. Now the segregation of this caste (whose name I don't recall but it starts with an 'h') occurred only a few hundred years ago and was occupationally based. Accordingly there is no good reason to believe that the caste is genetically any different from the population of Japan as a whole. Same gene pool, arbitrary splitting off of an oppressed class. This class, as a group, scores about 15 points lower in IQ tests than the general population of Japan. Oddly enough this is about the same differential as that between whites and blacks in this country. >Hereditarians believe that individuals routinely push their inherited >limits; you can and should train an individual up to maximum, but unless that >person has been lucky in the genetic draw that maximum will be at or below >average for the population. This tends to correlate with conservative, >religious-fundamentalist, reactionary and fascist politics. Your observations on the political correlations are probably correct. One's science in these areas tends to strongly affected by the political views that one holds. More importantly, holders of particular views tend to select "scientific results" that support the views that they hold. Intellectually honest scientists recognize that we don't know enough to legitimately make any such claims. The only correct anwer to "Who are right, the hereditarians or the environmentalists?" is "I don't know". >Historically, psychometrics has swung like a pendulum between hereditarian >and environmentarian extremes, often in a way correlated with the political >ideology of the period's dominant culture. The late 1800s saw an extreme of >hereditarian dogma (the Nazi master-race bulls**t was a sort of degraded pop >version of stuff that was mainstream scholarly anthropology in 1850-1900). >The 1960s saw an extreme of environmentarian dogma, more recently corrected >by the influence of Wilson and the sociobiological school. > If one is inclined to be charitable, the best that can be said for Sociobiology is that it is speculative. In view of the recurrent tendency to set public policy upon ill-founded speculation masquerading as science, I suppose there is something to be said for having a balance of such materials available. -- Richard Harter, SMDS Inc. [Disclaimers not permitted by company policy.]
c60a-4er@tart7.berkeley.edu.UUCP (06/15/87)
A trap to beware of in any discussion of the heritability of IQ:
The definition of "heritability" commonly used by population geneticists
and such is not quite the same as the common meaning of the word. The
definition is this: heritability is the amount of the variation within
the population under consideration (with regard to a particular trait)
that can be attributed to heredity. The words "within the population
under consideration" are tremendously important. Within inbred mouse
strains, for example, size of body has a very low heritability--since the mice
are practically identical genetically, almost all the variation is due to
birth order, feeding, etc. However, in wild mice size is quite heritable.
The point is, that figure of 80% heritability of IQ means that within
the population under consideration, about 80% of the observed variation
appears to be genetic. It DOES NOT mean that, for example, no possible
environment will make more than a 20% difference in IQ--an obviously false
statement (consider massive brain damage as an environmental condition).
I realize that this sounds like a semantic quibble, but it's not.
It is very easy to draw nasty false conclusions from heritability statistics.
An example: mental retardation due to phenylketonuria was, until the
disease was understood, nearly 100% heritable--much more heritable than IQ.
However, appropriate treatment early in life will prevent such retardation.
Can we be sure that there is not some equivalent of "appropriate treatment"
(I don't mean drugs or surgery, I mean maybe better education and home environ-
ment) to benefit those with "heritably" lower IQ?
Mary K. Kuhner
"the dance of the nucleotides, the fourfold sarabande..."werner@aecom.YU.EDU (Craig Werner) (06/15/87)
I'd like to point out the difference between heritable and
genetic (or inheritable).
The number of digits a person has is determined by genetic
factors, but that number has a very low heritability.
Skin color in the United States is very heritable. In Sweden
and Japan, it has a very low heritability.
What gives?
Heritability is not a measure of genetics. It is a measure of
variation. It is roughly defined as:
h = Genetic variation / total phenotypic variation
where total variation is
(genetic variation + environmental variation + error of measure)
Ergo, the number of digits, a highly genetically determined
trait, but with the majority of variation due to loss of digits by
accident, shows low heritability.
Since skin color variation in the US (mixture of African
descent and on the opposite extreme Scandanavian descent, and
everyone in between) is extremely genetically determined, heritability
is high. Whereas in Japan, where the gene pool is much more
limited, skin color is more determined by whether one lives in the
city or works as a farmer. The observed phenotypic variability is
due almost solely to number of hours spent in the sun, hence
heritability very low.
IQ scores are heritable trait. No less so than height or weight.
It is not perfectly heritable, and in fact, it's heritability varies
among population groups: it is higher in whites than in blacks in the
US, presumably because there is more environmental variation between
black communites.
Furthermore, as first pointed out by Jensen, attempts to
improve the environment will, paradoxically to some, make IQ scores
MORE heritable, since they will eliminate environmental variation
from the denominator without touching the numerator.
While many would argue this point, that means in an ideal
system, the best possible environment for every person, that IQ
would be 100% heritable.
Stephen J. Gould makes a very good case against those who
equate heritability with genetic determinism. Unfortunately, he
makes the same mistake in most of his rebuttal arguments.
So, I ask you, before comitting yourself to the argument,
think several times about this definition of heritability. It is
definitely a non-trivial point.
Maybe I should dig out my old IQ series, update it, and
repost it. Maybe not? Feedback is requested.
--
Craig Werner (MD/PhD '91)
!philabs!aecom!werner
(1935-14E Eastchester Rd., Bronx NY 10461, 212-931-2517)
"Why is it that half the calories is twice the price?"aweinste@Diamond.BBN.COM (Anders Weinstein) (06/15/87)
As I understand it, "heritability" is a rather bogus statistic from which to draw political or sociological conclusions. Heritability indicates the extent to which genes account for variance *within a population*. By itself, it doesn't license any meaningful conclusions about the causes of differences *between* populations (which is what is usually at issue in IQ debates). Example: Take two identical samples, A and B, of genetically diverse corn, and grow each in a controlled environment. Deprive one group, B, of some essential nutrient. Both populations will exhibit variation in height. In each group this is due entirely to genetics, so height is highly heritable in each. But the fact that the average height in B is much lower than that of A is *not* due to genetics; it is entirely environmentally caused.
lincoln@randvax.UUCP (06/16/87)
In article <123@snark.UUCP> eric@snark.UUCP (Eric S. Raymond) writes: >In article <2172@mmintl.UUCP>, franka@mmintl.UUCP (Frank Adams) writes: >> I am curious about why "is intelligence heritable?" > >Intelligence is *extremely* heritable. > >Empirical evidence: the I.Q. and aptitudes of identical twins raised apart are >quite strongly correlated -- sorry, I don't have numerical statistics handy. > >'Intelligence' is a multidimensional composite of aptitudes in several >distinct areas. > >All of this data is discussed, from a strongly environmentarian viewpoint, >in Stephen Jay Gould's _The_Mismeasure_of_Man_ -- a book I recommend for >facts and style while disagreeing almost totally with its analysis and >conclusions. Gould is a fine and lucid writer when his Marxist sympathies >are dormant. . ( .) ((). CAUTION ()(( FLAMES >to see light through the smoke. {\@@/} AT WORK >What seems to be true is that heredity sets a fairly hard upper limit on >the capability of the brain to do particular kinds of information >processing (things like, say, spatial visualization). >Environmentarians believe the inherited limits are generally well above >the performance level of typical individuals, so that there's plenty >of room for improv(ement)... It is interesting that the Japanese have competative categories that lie above the present masters. I believe that the top 35 master categories for the game of GO are empty. We, by contrast grade on a sliding scale.. >Hereditarians believe that individuals routinely push their inherited >limits; It simply defies credulity to suppose that todays couch potatoes are pushing their inherited limits - or any other limits. If we try Toynbee (why not add an historian) - there is some evidence that the best challenge is neither too severe or too easy (usually whatever society that those who write about such things - and feel they are pushing their productive limits - live in represents their parochial estimate). See such books as "Why Jonny can't run". >My personal opinion: the hereditarian claim is unpleasant, but fits the >observed statistical facts better than most versions of the environmentarian >thesis. Well, someone with that opinion who tries to read Stephen Gould is probably pushing his/her limits. |\ /| . {O O} . Its a dog's life, eh Sandy? ( " ) lincoln%iris@rand-unix.arpa `U'
eddy@boulder.Colorado.EDU (Sean Eddy) (06/16/87)
In article <1142@aecom.YU.EDU> werner@aecom.YU.EDU (Craig Werner) writes: > I'd like to point out the difference between heritable and >genetic (or inheritable). > The number of digits a person has is determined by genetic >factors, but that number has a very low heritability. > Skin color in the United States is very heritable. In Sweden >and Japan, it has a very low heritability. > Heritability is not a measure of genetics. It is a measure of >variation. It is roughly defined as: > > h = Genetic variation / total phenotypic variation > >where total variation is > (genetic variation + environmental variation + error of measure) Now hold on. Heritability sure is a measure of genetics; otherwise it wouldn't be in every genetics textbook I've ever owned. I'm not being facetious. Heritability is a measure that is very helpful in agriculture (where you can ethically breed organisms to be 'better'). The higher the heritability for a particular trait, the easier it is for one to 'improve' that trait through breeding programs. Such a program not only depends on the trait being genetically determined, but also on the capability of the trait to be varied. This is the reason for the invention of the number called 'heritability'. Choosing examples such as 'number of digits' are unfair, since there is little variation. But IQ scores are distributed on that classical (artifactual?) Gaussian curve. That's why heritability is indeed a legitimate measure to discuss; because if the heritability of IQ is high, one can imagine that the IQ of the race can be improved through careful breeding. Not a pleasant concept, but that's why the issue is controversial. - Sean Eddy - MCD Biology; U. of Colorado at Boulder; Boulder CO 80309 - eddy@boulder.colorado.EDU !{hao,nbires}!boulder!eddy - - "Don't drink and drive. You might hit a bump and spill your drink."
hes@ecsvax.UUCP (06/18/87)
A fairly readable, somewhat technical, and slightly dated reference is "Race Differences in Intelligence" by John C. Loehlin, Gardner Lindzey and J. N. Spuhler. 1975, W. H. Freeman and Co. It gives much background, definitions and surveys the literature. (Chap. 4, 27 pages, is on Heritibility.) Does anyone know if there is a newer edition? --henry schaffer n c state univ
hes@ecsvax.UUCP (06/18/87)
In article <6566@diamond.BBN.COM>, aweinste@Diamond.BBN.COM (Anders Weinstein) writes: > As I understand it, "heritability" is a rather bogus statistic from which to > draw political or sociological conclusions. Heritability indicates the extent > to which genes account for variance *within a population*. By itself, it > doesn't license any meaningful conclusions about the causes of differences > *between* populations (which is what is usually at issue in IQ debates). Many misunderstandings arise because the array of environmental conditions are not stated. The above is not wrong, but it does not deal with the environment array. > Example: Take two identical samples, A and B, of genetically diverse corn, > and grow each in a controlled environment. Deprive one group, B, of some > essential nutrient. Both populations will exhibit variation in height. In each > group this is due entirely to genetics, so height is highly heritable in > each. But the fact that the average height in B is much lower than that of A > is *not* due to genetics; it is entirely environmentally caused. But the two samples of corn are part of the same population. The problem actually is that the heritability statistic applies to a given array of genotypes (ok, a population) *and* a given array of environmental conditions (often confusingly called simply an "environment"). In the example the heritability could have been calculated for an overall environment which included variation in the essential nutrient - and then the herit. would have been smaller than calculated for either of the two environmental situations alone, and it *would* be appropriate to consult it in trying to understand the origin of the difference between the corn plants in the two groups. In this case the herit. value would be intermediate and it would tell one that the difference could be genetic, environmental or some combination (which usually is the case in real life.) (In all the above when I say "array" of genotypes or environmental conditions, I really mean to refer to a frequency distribution, so both the effects and their relative degrees of occurence are both important.) --henry schaffer n c state univ
benson@alcatraz.UUCP (06/18/87)
In article <123@snark.UUCP> eric@snark.UUCP (Eric S. Raymond) writes: >In article <2172@mmintl.UUCP>, franka@mmintl.UUCP (Frank Adams) writes: >> I am curious about why "is intelligence heritable?" was included in this >> list. This seems to me to be very much a scientific question, albeit one we >> can't really answer yet. > >Intelligence is *extremely* heritable. The results that show this are solid >but not as well known as they might be due to the fact that hereditarianism >in general is out of fashion and 'politically incorrect'. > >Empirical evidence: the I.Q. and aptitudes of identical twins raised apart are >quite strongly correlated -- sorry, I don't have numerical statistics handy. > That is, Cyril Burt, who invented his data, as documented in the Gould book and elsewhere. >All of this data is discussed, from a strongly environmentarian viewpoint, >in Stephen Jay Gould's _The_Mismeasure_of_Man_ -- a book I recommend for >facts and style while disagreeing almost totally with its analysis and >conclusions. Gould is a fine and lucid writer when his Marxist sympathies >are dormant. > I'm not a statistician, but I know enough about it to find Gould and a host of other plausible. You might considered submitting a CV for consideration. In any case, I'll throw a different two cents in: I'll believe that some intelligence is inheritable. I won't believe, and am strongly convinced of the pseudo-scientific quality of the "studies" that "prove", that blacks (or any other 'race') as a class are less intelligent. Most of the heat of the heredity versus environment debate is generated by the curious fact that the publishing hereditatians keep "proving" such racial factoids. Herrenstein and Burt were/are grinding axes, so they have no one else to blame if they get the same in return. --benson Benson I. Margulies Kendall Square Research Corp. harvard!ksr!benson All comments the responsibility ksr!benson@harvard.harvard.edu of the author, if anyone.
howard@cpocd2.UUCP (Howard A. Landman) (06/18/87)
Please followup to rec.games.go only, since this isn't really a bio topic. In article <301@iris.randvax.UUCP> lincoln@iris.UUCP (Tom Lincoln) writes: >It is interesting that the Japanese have competative categories that lie >above the present masters. I believe that the top 35 master categories >for the game of GO are empty. We, by contrast grade on a sliding scale.. It would be interesting if it were true, but I've been playing for 20 years and I've never heard this. The top "normal" Japanese rank is 9 dan, and there are many players of this rank. (I've had the privilege of watching my game get ripped apart by a couple of them!) The rank of 10 dan is reserved for the winner of a particular tournament; if you win it enough you can get to be "honorary 10 dan" like Sakata Eio. This doesn't mean that the Japanese believe that they are playing perfect Go. Every professional game has at least one or two known mistakes, and probably many more unknown ones. And if 19x19 Go is ever perfected, there are always larger boards ... -- Howard A. Landman ...!{oliveb,...}!intelca!mipos3!cpocd2!howard howard%cpocd2%sc.intel.com@RELAY.CS.NET "Half a mile from Tuscon, by the morning light"
tim@ism780c.UUCP (Tim Smith) (06/26/87)
In article <16745@cca.CCA.COM> g-rh@CCA.CCA.COM.UUCP (Richard Harter) writes:
< Well now, the evidence is not nearly so strong as you claim.
< The principle results on identical twins were those of Cyril Burt's,
< which is where the conventional figure of 80% heritability comes from.
< However Burt's data and results were forged. (I don't know if this is
There was an article in a U.S. News and World Report a while back ( I
can't be more specific, since I read it in a waiting room somewhere )
on this. It said that a large study of twins and other siblings had
recently been completed ( I think they said that it was the largest
such study ever conducted ), and that the results were that such things
as intelligence and personality were very strongly determined by
heredity. For instance, introvert vs. extrovert was something like
70% determined by heredity, and intelligence was similar.
Does anyone have more information on this? Was the USN&WR article
accurate, or is this another case of a popular magazine screwing
up their science reporting?
Also, what does it mean to say that something is 70% determined
by heredity?
--
Tim Smith, Knowledgian {sdcrdcf,seismo}!ism780c!tim