The Bell Curve is a controversial book about IQ and, in a broad sense of the expression, success in life from 1994.


In this Wikipedia article, in the introduction, you can read

The authors claimed that average intelligence quotient (IQ) differences between racial and ethnic groups are at least partly genetic in origin, a view that is now considered discredited by mainstream science.

(Please don't change the formatting of the previous paragraph)

Is this really true?

Another article, this times in Nature Genetics, https://www.nature.com/articles/ng.3869 finds that IQ is around 80 % inherited/genetic. If this is true, and it is true that different ethnic groups perform differently on IQ tests (which seems to be true, and independently of how the IQ is tested or measured), the (only) logical conclusion is that there is an inherent difference between groups: if IQ is mostly genetic then the IQ on group level is inherited and if groups perform differently on IQ test it is caused by genetics.

This seems to confirm this thesis in The Bell Curve.

I also note that two the three sources for the claim I quoted above is not scientific (The Guardian and Vox) while the last seems to be mostly a political argument against undesired results of natural science.

  • The edit history shows that there were several attempts to improve this question. The OP reverted them, was unable to see why they were improvements and became belligerent in the comments. There is no point in us trying to make this into a reasonably written question at this point. The question has now been locked to allow the existing answer to be voted upon.
    – Oddthinking
    Commented May 3 at 1:25

1 Answer 1


The 80% heritability was also part of the claims/data by the authors of the The Bell Curve.

As for the latter GWAS paper (Sniekers et al.), it doesn't talk about 'racial or ethnic groups' in its abstract.

So there's little else to say here but repeat the usual refrain:

the heritability of a phenotype within a population does not determine the extent to which genetic variation is responsible for mean differences between populations that can be differentiated from one another with respect to factors that exert significant influence on variation in a phenotype of interest. Furthermore, the genetic variation that influences variation in measures of cognitive performance within a population is not necessarily identical to the genetic variation that influences variation in mean measures of cognitive performance between populations. [...]

The implicit assumption of some participants in the controversy that genes involved in influencing IQ in some putative racial groups are clearly and unalterably superior to those same genes in other racial groups, creating a static hierarchy, is unjustified.

If the genes that influence intelligence have indeed been nailed down (although I have my own skepticism on that, and the degree of variation they explain matters [a lot]--but alas that's not even detailed in the abstract of the Sniekers et al.), then the convincing thing to do is to measure the variation in those precise genes across the racial groups of interest. But I'm not sure if anything like that has been done (in the meantime).

Furthermore, polygenic score studies are non-trivial to interpret even for less controversial issues, e.g.

We show how genetic contributions to traits, as estimated by polygenic scores, combine with environmental contributions so that differences among populations in trait distributions need not reflect corresponding differences in genetic propensity. Under a null model in which phenotypes are selectively neutral, genetic propensity differences contributing to phenotypic differences among populations are predicted to be small. We illustrate this null hypothesis in relation to health disparities between African Americans and European Americans, discussing alternative hypotheses with selective and environmental effects. Close attention to the limitations of research on polygenic phenomena is important for the interpretation of their relationship to human population differences.

And IIRC height was one of the phenotypes that was better explained in that regard [amount of variation] by polygenic scores, but still

Recent studies of height have suggested that polygenic adaptation tests are sensitive to the choice of GWAS data that provide the effect sizes: even if two sets of effect sizes produce correlated polygenic scores, effect sizes estimated from one study can generate erroneously exaggerated signatures of polygenic adaptation when assessing polygenic adaptation in a second dataset [34, 35]. This result, which arises from subtle population differences between study samples, calls into question claims about polygenic adaptation even of traits for which it has been most extensively investigated.

[...] So far, because most genome-wide association studies have been conducted in populations with European ancestry [33], the effect sizes used in calculating polygenic scores have been calibrated on Europeans, and their values might not transfer accurately to other populations. Even among populations with European ancestry, subtle ancestry differences between samples can lead to polygenic scores that overstate between-population differences: small biases in locus-wise effect estimates that arise from the ancestry differences can potentially accumulate across loci [34, 35].

The last of those refs really says something like that

our results imply that typical constructions of polygenic scores are sensitive to population stratification and that population-level differences should be interpreted with caution. [...]

For example, if height were differentiated along a north-south axis because of differences in environment, any variant that is differentiated in frequency along the same axis would have an artificially large effect size estimated in the GWAS.

Even a much more upbeat review (and highly cited for its relatively recent date) says something similar in that respect:

genetic associations may differ across ancestries, complicating direct comparisons between groups of individuals. Some of these limitations hamper drawing unambiguous conclusions about the biological meaning of GWAS results, sometimes limiting their utility to produce mechanistic insights or to serve as starting points for drug development.

But that review cites a slightly more recent paper as a relevant GWAS intelligence study, and this one happens to be open access, so I don't have to go to a library to see how much variation they could explain:

we used our meta-analytic GWAS data to predict almost 7% of the variation in intelligence in one of three independent samples. The range of similar estimates across the three independent samples was 3.6 to 6.8%. Previous estimates of prediction have been ∼5% at most; [16]

Where [16] is Sniekers et al.

So the OP's main issue/claim seems to be

You don't need to know exactly which gene that is responsible to realise that something is genetic.

But that's not really the case. The first paper I linked goes into more detail on that:

The extent of the impact of allelic variation at a particular locus on variation in a trait, such as happiness, cognitive function, height, eye color, or skin color may either correlate or interact with (i.e., depend on) one or more environmental factors such that disentangling the relative importance of genetic versus environmental factors for individuals becomes highly challenging or practically impossible.

So, the heritability value for a trait might be expected to vary depending on the environmental factors operative within a population and the distributions of the impacts of those environmental factors on the members of a population, as well as the precise manner by which the trait is assessed or, if quantitative, measured.

Heritability can vary for a trait depending on when in the developmental process (e.g., neonate vs. child vs. adult) the traits are assessed in members of a population.

The magnitude of heritability for a given trait is influenced by the particular allele frequency distributions at the relevant loci in the population studied and for which other populations may differ substantially.

As implied by the preceding points, changes in the relevant allele frequency distributions (which are not merely possible but over sufficiently long periods of time likely) or in the presence or distribution of environmental factors can lead to alterations in the degree of heritability.

Combining some of the above points, the value of the heritability is not an intrinsic property of a gene, a trait, or a gene-trait pair for a given population or whole species.

And since your pointer to the Sniekers paper was apparently a red herring (in terms of GWAS) and you [apparently] care about 'old school' twin studies for establishing heritability (that's were the 80% heritable intelligence comes from), there's something on caveats on that too; detecting gene-environment correlations is not possible even in that design without extra tricks that have e.g. presuppose a set of correlations (limit env factors examined) or use more complex longitudinal models attempt to suss out such correlations in toto e.g. by an autoregressive model. (I would paste a quote here, but the paper uses a lot of notations with subscripts that will render poorly here--but it's open access, so you can read it for free at the source.)

Some of the more famous twin studies on IQ have in fact been attacked on that angle, of failing to account for the environment, in particular for reporting only the MZ (monozygotic) results and not reporting the DZ (dizygotic) ones, even though some theorists argue that if the latter are the same, then there's not much genetic heritability to talk about. N.B. in a rejoinder the [first] author of that paper points out that same slightly later published twin studies on IQ did follow protocol more strictly.

Anyhow, this mostly parenthetical info because even assuming correct heritability estimated based on twin studies tells us little about implications for racial group differences, per the first quote.

FWTW, even the (itself) somewhat controversial 'Mainstream Science on Intelligence' letter, makes this point of not endorsing the race/ethnicity-related claims in The Bell Curve:

There is no definitive answer as to why bell curves differ across racial-ethnic groups. The reasons for these IQ differences between groups may be markedly different from the reasons for why individuals differ among themselves within any particular group.

The Bell Curve itself claimed these must be partly genetic although it didn't venture an estimate of how much.

I'm not going to delve here into why that letter was itself controversial (Wikipedia covers that in detail), but it suffices to say here that its main signatories that had a connection to the field were among those who published the aforementioned twin studies (Plomin, Bouchard). So, the point that matters here is that even researchers who did publish IQ heritability results themselves (so they're not skeptics of IQ as a construct etc.) doubt the validity of the somewhat stronger claim in that regard in The Bell Curve about racial/ethnic differences.

BTW, that was hardly the most controversial bit in the Bell Curve book, IMHO. Their policy recommendations to effectively conduct a kind of racial immigration profiling, eliminate all antipoverty and affirmative action programs were probably far more unpalatable. Oddly enough, the book argued all this should be done regardless whether the 'cognitive differences between races' prove to be 100% genetic or 100% environmental (pp. 314-315 in the book). One reviewer (previous link) tried to explain this by the fact that the book had two authors, who could not quite agree on the importance of the genetics angle. Murray had apparently argued for those policies before, without the genetics angle.

  • 1
    This paper about polygenic scores predicting 10x as many schizophrenia cases among Africans (than actually diagnosed) is also somewhat interesting. That is a field of much more research, so I'm not sure if it's an issue that's been resolved in the meantime. Commented Apr 30 at 11:42
  • 1
    One obscure but interesting fact is that a single African ethnic group like the Xhosa has much more (intra-group) genetic variation than [the way more numerous] Europeans or Asians science.org/doi/10.1126/science.aay8833 Commented Apr 30 at 12:01
  • 2
    @thegodsfromengineering Because the question concerns inheritability. You don't need to know exactly which gene that is responsible to realise that something is genetic.
    – d-b
    Commented Apr 30 at 18:08
  • 6
    @d-b Except height has the exact same problem; if group A is taller than group B that may be an environmental or genetic difference. The Japanese average height grew by four inches after WWII, with no generic changes.
    – prosfilaes
    Commented May 1 at 10:24
  • 3
    @d-b So you claim differences in intelligence between different racial groups is more akin to differences in heights between males and females than it is to differences in heights between different racial groups. Nope, doesn't work. You're going to need hard evidence, not handwaving, to make that fly.
    – prosfilaes
    Commented May 2 at 20:48

Not the answer you're looking for? Browse other questions tagged .