Posted on March 2, 2009

Should Scientists Study Race and IQ?

Nature, February 12, 2009

Steven Rose, Nature, February 12, 2009


{snip} Further, given that our society already accepts a number of restrictions to the pursuit of knowledge, it is sensible to require that funded research also addresses questions that either contribute to basic scientific understanding, offer new beneficial technological prospects, or aid sound public policy-making. These criteria are, of course, those used by both public and private funding bodies.

So what should we make of the century-old but regularly-recycled call for research aimed at discovering whether there are group differences in intelligence?

These days the ‘groups’ under consideration are ‘race’ and ‘gender’. But it has not always been so. A hundred and fifty years ago, when Darwin published The Descent of Man, and Selection in Relation to Sex, he regarded it as so self-evident that white Anglo-Saxon upper-class males were the most intelligent as not to need evidence. Half a century ago, at least in Britain, class was the more relevant grouping, leading to eugenic concerns that the genetically inferior workers were outbreeding their superiors. The issue of race and intelligence became prominent in the United States in the late 1960s, perhaps in response to the civil-rights movement. Arthur Jensen’s How Much Can We Boost IQ and Scholastic Achievement? (A. R. Jensen Harvard Educ. Rev. 39, 1-123; 1969) argued that the deficit in black IQ was too great to be explained by deprivation and must be genetic. {snip}

The categories judged relevant to the study of group differences are clearly unstable, dependent on social, cultural and political context. No one, to my knowledge, is arguing for research on group differences in intelligence between north and south Welsh (although there are well-established average genetic differences between people living in the two regions). This calls into question the motivation behind looking for such specific group differences in intelligence, sheds doubt on whether such research is well-founded, and begs whether answers could possibly be put to good use. {snip}

There is a difficulty in the first instance of measuring ‘intelligence’. For around a century, this has been done with the IQ test, originally developed in France as a way of supplementing teachers’ assessments of their pupils. In the hands of later psychometricians, the tests became increasingly reified, and seemingly made more scientific by the development of the term ‘g’ to encapsulate ‘crystallized’ or ‘general intelligence’.

However, except to a small band of dedicated psychometricians, it seems obvious that to try to capture the many forms of socially expressed intelligent behaviour in a single coefficient–and to rank an entire population in a linear mode, like soldiers on parade lined up by height–excludes most richly intelligent human activities. Social intelligence, emotional intelligence, the intelligent hands of the craftsman or the intelligent intuition of the scientist all elude the ‘g’ straightjacket.

The flexibility of IQ

Group comparisons of IQ are even more problematic. Attempts have been made to make ‘culture-fair’ or ‘culture-free’ tests, as if such a thing were possible, to allow comparisons of ‘g’ between people from very different societies. But IQ is clearly a flexible construct–as amply demonstrated by decisions in the 1930s and 1940s in the United States and Britain to ‘adjust’ test questions to equalize the scores of boys and girls, because in previous versions of the tests girls had scored higher. {snip}

As for ‘race’, the problem is whether it is a biologically, as opposed to socially, meaningful category. Among geneticists interested in differences in gene frequencies between populations, there is increasing consensus that the word obscures more than it reveals, and should be replaced by the concept of biogeographic ancestry, which makes possible the study of subpopulations for relevant genetic and phenotypic characteristics. There are some well-recognized, meaningful genetic differences between groups, for instance between Ashkenazi and Sephardi Jews in terms of their risk to Tay-Sachs disease, and the study of such differences may reveal important clues with respect, for instance, to disease propensity. But such groups are not normally considered socially distinct races for the purposes of studies of group differences in intelligence. Broad divisions between ‘white’ or ‘Caucasian’ and ‘black’ or ‘Asian’, the groups generally discussed in the context of the IQ debate, especially in the United States, hide genetically important subpopulation differences within these groups. See box


To conclude: the categories of intelligence, race and gender are not definable within the framework required for natural scientific research, failing my first criterion of being well-founded. They also fail the second criterion of being answerable: we lack the theoretical or technical tools to study them.

The standard approach of population biologists to estimating the potential genetic contribution to a trait is to make a heritability estimate. Whatever the strengths and weaknesses of this measure within a population, it is essentially just that: a within-population measure, only valid for a given environment. The nature of the equations means that if the environment changes, the heritability estimate changes too. Moreover, the measure relates to a randomly interbreeding population–useful for agricultural purposes such as estimating optimal genotypes for crop yields or milk production–but not for people. Even if reliable correlations were found between some intelligence test score and a measure of brain physiology or activity held by a specific group, such a correlation says nothing about the direction of causation.

As for the third and final criterion, if attempts to answer these group-difference questions are fraught with scientific fallacies, might there nonetheless be some public-policy implications making investigation worthwhile? The answer sometimes advanced is that if there were such differences, and their causes were understood, the less well-endowed groups could be ‘compensated’ by some form of differentiated education. But in practice, claims that there are differences in intelligence between blacks and whites, or men and women, have always been used to justify a social hierarchy in which white males continue to occupy the premier positions (whether in the economy in general or natural science in particular). Using pseudoscience, based on concepts as ill-founded as was phlogiston, to justify preordained conclusions should not serve as the basis of sound policy-making.


Stephen Ceci & Wendy M. Williams, Nature, February 12, 2009

The Soviet Union lost a generation of genetics research to the politicization of science when Trofim Lysenko, director of biology under Joseph Stalin, parlayed his rejection of Mendelian genetics into a powerful political scientific movement. By the late 1920s, Lysenko had denounced academics embracing Mendelian genetics, which some said undermined tenets of Soviet society. His efforts to extinguish ‘harmful’ scientific ideas ruined opponents’ careers and delayed scientific progress.

It is difficult to imagine this situation repeating today, when rival views feed the scientific process, and inquiry and debate trump orthodoxy. Yet the spectre of Lysenkoism lurks in current scientific discourse on gender, race and intelligence. Claims that sex- or race-based IQ gaps are partly genetic can offend entire groups, who feel that such work feeds hatred and discrimination. Pressure from professional organizations and university administrators can result in boycotting such research, and even in ending scientific careers.

But hatred and discrimination do not result from allowing scientists to publish their findings, nor does censuring it stamp out hatred. {snip}

There is an emerging consensus about racial and gender equality in genetic determinants of intelligence; most researchers, including ourselves, agree that genes do not explain between-group differences. But some issues remain unresolved, such as identification of mechanisms that bring genetic potential to fruition. Censuring debaters favouring genetic explanations of intelligence differences is not the answer to solving such mysteries.


Nobel prizewinner William Shockley became a subject of controversy in the 1970s, after his work turned to racial differences in intelligence. In recent decades, the writings, statements and teachings of Arthur Jensen, Michael Levin and John Philippe Rushton, also on racial differences in intelligence, have met variously with acclaim, outcries and demands for job termination. So have writings of Richard Herrnstein and Charles Murray on the differential distribution of IQ by race. And Frank Ellis, a lecturer at the University of Leeds, UK, took early retirement in the face of an ethical storm that developed after he suggested in a student newspaper that intelligence levels were related to ethnicity. The list goes on. Many have been dissuaded from even looking at the research topic for fear of condemnation.


Judged too fast?


James Watson is the most illustrious scholar to have his career ended for reckless language. Watson’s downfall was his assertion that “all our social policies are based on the fact that [African] intelligence is the same as ours–whereas all the testing says not really”. Although he hoped everybody was equal, “people who have to deal with black employees find this is not true”. Watson instantly plunged from A-list Nobelist to outcast, and was suspended from his chancellorship of Cold Spring Harbor Laboratory. Watson later clarified in a statement that he does not believe Africans to be genetically inferior, but this had little impact on the controversy.

Watson’s first assertion could be read as scientifically supported: black Africans’ IQ scores are lower than those of white Europeans. But Watson’s use of ‘intelligence’ was interpreted as meaning ‘intrinsic cognitive ability’, ignoring how unfamiliarity with testing format, low quality of schooling, or poor health might depress IQ scores. There have been analyses showing average national IQs for sub-Saharan Africa to be approximately 30 points lower than average IQs for predominantly white European nations, and drawing a racial conclusion from those results. A refutation of these analyses would provide an opportunity to advance understanding. Sadly, although these analyses can be refuted, as we and others have done, most of those who scorned Watson never knew they existed.


Some scientists hold more ‘acceptable’ views, ourselves included. We think racial and gender differences in IQ are not innate but instead reflect environmental challenges. Although we endorse this view, plenty of scholars remain unpersuaded. Whereas our ‘politically correct’ work garners us praise, speaking invitations and book contracts, challengers are demeaned, ostracized and occasionally threatened with tenure revocation.

Acts of censure edge close to Lysenkoism. They also do a disservice to science. When dissenters’ positions are prevented exposure in high-impact journals and excluded from conferences, the dominant side goes unchallenged, and eventually its rationale is forgotten, forestalling the evolution of crucial ideas.

James Flynn, the foremost proponent of the environmental basis of intelligence, notes that when he first rebutted Jensen’s hereditarian claims 30 years ago, he never anticipated later breakthroughs that evolved from the debate. Without Jensen, he has written, “I would never have made any contribution to psychology”. His landmark documentation of the steady rise in IQ scores across generations and nations, known as the Flynn effect, might never have been done.


So, vigorous debate has resulted in great progress in our understanding, and more breakthroughs will come–if we allow free speech in science.

One could argue that some peer-reviewed reports feed racial and gender stereotypes. Perhaps such research should be forced to pass a higher cost-benefit threshold before publication. But this is a slippery slope: philosopher Jean-Jacques Rousseau argued that atheists should be silenced, lest they convince the masses to abandon faith, condemning them to hell. This would now be viewed as a ludicrous violation of free speech. Who is to be impanelled with the wisdom to decide which views can be aired, and which research questions pass muster?

It might also be argued that only primary researchers who are experts in their field, rather than administrators or non-experts, deserve protected speech in these areas. A statement’s validity, however, lies in its congruence with scientific data, not in the role occupied by its speaker.

One powerful argument states that groups need protection against bigotry, and that censuring one side in a debate is necessary to prevent the harm done to victims of race and gender arguments. {snip} Such problems, however, arise not from scientific discourse, but from political applications of those ideas. {snip}


When scientists are silenced by colleagues, administrators, editors and funders who think that simply asking certain questions is inappropriate, the process begins to resemble religion rather than science. Under such a regime, we risk losing a generation of desperately needed research.