To the Editor:
R. J. Herrnstein’s article, “In Defense of Intelligence Tests” [February], is the best discussion of the IQ testing controversy that I have yet seen in any popular publication. One technical point he raises struck me; this concerns his paraphrase of Arthur Jensen’s argument that a substantial degree of genetic determination of intelligence implies that “individual differences in scholastic achievement [are] bound to be relatively hard, though not necessarily impossible, to improve by simple remedies.” This is a courageous effort to summarize a very complicated matter in simple language. I would like, if I may, to expand upon this because it is widely misrepresented to the public, and its implications are central to any sensible discussion of IQ and educational policy.
The degree of a trait’s genetic determination is measured by what geneticists call its “heritability,” which is an index of how much of a trait’s deviation from the average value can be attributed to environment. . . . High heritability for IQ means that the spread in IQ’s can be attributed only in a modest degree to the existing spread in environments. There are three vital points to understand about this in drawing implications for policy:
- A finding of high IQ heritability means that overall variations in IQ cannot be significantly reduced by means of narrowing environmental variability to the average, that is, middle-class, environment. Since extending the middle-class cultural milieu to every child is the explicit motive behind most efforts at compensatory education, high IQ heritability does predict the lack of success of this kind of program, though not necessarily of all kinds, in reducing overall IQ and achievement differences. This does not mean that such programs should not be undertaken on other grounds. But it is foolish to justify them in terms of their supposedly beneficial effects on IQ.
- How heritable a trait will appear is critically dependent upon the particular environmental background against which it is measured. The same trait may appear highly heritable in one set of environmental circumstances and only weakly so in another. Thus the judgment that IQ can be influenced only minimally by environmental factors strictly applies only to the kinds of environmental variation that are already common. This does not foreclose the possibility of discovering new types of treatments that could raise IQ. Rather than implying the necessary futility of all efforts aimed at improving IQ, then, high IQ heritability suggests redirecting those efforts to examine novel forms of intervention.
- The heritability statistics are inferred as an average for an entire population and hence may conceal rare instances where IQ is being greatly influenced by existing environmental factors. Thus high overall IQ heritability should not discourage the search among exceptional situations for potent environmental agents that, once identified, could be made the basis of deliberate remedial policies. Conversely, finding such occasional instances (and the popular literature is filled with claims of this nature) should not be interpreted as disproving high IQ heritability; as long as these instances are rare and concern only small numbers of children, they cannot invalidate the overall finding.
In sum, a substantial degree of genetic determination of IQ differences does not quite mean that remedies are bound to be “hard” rather than “simple,” but it does mean that those remedies cannot be based simple-mindedly upon equalizing environments to the middle-class norm. There may be remedies for low IQ, but we will have to look for them among novel innovations or now rare environmental circumstances. I agree it will probably be hard to discover what these might be. But once we understood them, they might be simple to apply.
William R. Havender
To the Editor:
R. J. Herrnstein’s article is considerably more reasoned and more temperate than the usual outpourings of the contemporary hereditarian school of psychologists. Nevertheless, several flaws limit his persuasiveness.
For example, Mr. Herrnstein handles the late Cyril Burt with remarkable gentleness. Kamin and Dorfman have demonstrated that over a period of many years Burt’s work was systematically fraudulent. Therefore, Burt’s results must be suspect, except for those that have been thoroughly and independently checked by other, more reliable investigators.
Mr. Herrnstein handles the statistical and genetic parameters of heritability rather cavalierly. As Lewontin and others have pointed out, heritability is useful under certain conditions in determining the genetic component in the phenotypes of a single population. But using this parameter to compare different populations, or different races, is an invalid procedure.
The example of a verbal analogy that Mr. Herrnstein cites, “Fish is to scales as dog is to: lap; skin; fur; bite,” would frustrate anybody with enough “g” to have picked up some biology in the course of his life. What is supposed to be the correct answer: skin? fur? Actually there is no correct answer among the choices given. Scales, like fur and hair, are outgrowths of the skin. Hair would be the correct answer. But while dogs have hair, they don’t have fur. As Banesh Hoffman has noted, knowing too much can kill you on an intelligence test.
There are other lapses in the article. But these should suffice to suggest that Mr. Herrnstein has been a little careless. Nevertheless, I was glad to see Mr. Herrnstein’s lists of dos and don’ts, especially “Don’t . . . draw firm conclusions about the origins of the average difference between the races in intelligence-score tests.”
Stanley L. Weinberg
R. J. Herrnstein writes:
William R. Havender’s letter enlarges usefully on several points I glossed over in my article. I’m glad to see it published, for what the heritability of IQ means usually gets less attention than the controversy surrounding it. I think Stanley L. Weinberg is saying something similar about heritability, but he also chides me for confusing fur and hair. I would gladly grant my error if I had really made one, but as far as I can tell from various references, fur is the soft hair of certain animals, such as dog, fox, beaver, or cat. The right answer to my made-up item was “fur,” and if Mr. Weinberg got it wrong, it was probably not because he knows too much. In fact, items on the major intelligence tests are validated in standardizing procedures making it unlikely that anyone is penalized for being especially bright or knowledgeable. Finally, I treated Cyril Burt gently because nothing I have seen calls for anything less. The data from his later years are now in doubt, as I said in my article, but he remains a major early contributor to psychological measurement. To deny him that is unjust and unjustified.