What is race science?

Scientific racism is the misuse of science to promote racist ideology. Scientific racists hope to take advantage of the wider public’s lack of expertise in matters of genetics and intelligence to advance the view that races can be meaningfully categorised and ranked. If it can be proved that some races are smarter than others, they argue, then governments can justify the segregationist policies. This explainer on scientific racism addresses some of the key claims and misconceptions advanced by scientific racism.

1) Race is not a genetically meaningful concept

An argument popular among scientific racists is that there are differences in IQ between races which are substantially determined by genetics. This claim is not supported by scientific evidence. Scientific understanding of both human genetic variation and of intelligence suggest the reality is much more complex than this false and simplistic account implies.

Contemporary research in genetics shows that race is not a genetically useful concept. There are clearly genetic differences between individuals, and these differences can result in observable physical differences between people, such as in skin colour. But this genetic variation is highly complex and on a continuum; it does not fall into discrete categories which map on to racial groups. The existence of observable physical differences mean that many societies do categorise people into different racial groups, but how these categories are defined varies across cultures: race is a social, not a biological, construct.

In comparison with other species, humans show relatively little genetic variation. This is likely because we went through a genetic “bottleneck” in our recent history, reducing the human population down to relatively few members, and there has been little time for genetic variation to subsequently emerge. The variation we do see tends to be related with to observable physical characteristics – which may mislead us into thinking genetic variation is more substantial that is really is – or is variation in genetic material which relates to nothing at all. Most of our genetic material is non-functional, meaning it is not used during development for building our bodies or brains. It is this genetic material which is most likely to vary, because changes in noncoding gene sequences have little or no impact on the person they’re in. Any mutations (changes to genetic material which occur when cells make copies of themselves, during the formation of egg and sperm cells, for example) in these sequences may therefore be passed on from parents to children and spread through local populations because they have no consequences (this is a process known as “genetic drift” – random changes in gene frequencies between generations). Such material can be used in identifying “genetic ancestry”. For example, for-profit gene-testing companies will tell you if you share genetic similarities with people from various parts of the world.

But “genetic ancestry” is also highly complex, and depends on how deep into history we want to define genetic ancestry. The history of human migration across the globe is complex, with multiple “waves” of migration happening across different regions at different times. Populations in different regions have not remained isolated from one another across long time periods; there’s been constant mixing of genetic material across regions. The genetic ancestry of someone who lived in Europe a few thousand years ago would look different to that of a European today, because of multiple waves of migration which have happened across Europe in the last few thousand years. But their genetic ancestry would also be complex and involve ancestry from multiple regions.

Ultimately, all humans are of African ancestry, because that’s where our species evolved. But “African ancestry” is not a meaningful way of characterising people in the 21st century, because the African continent – as the longest inhabited continent – has more genetic variation within it than there is between Africa and other continents. Two African people (even from the same region of Africa) are likely to be more genetically different from one another than an African and a European. The genetic variation seen in populations on other continents is also not distinct from that seen in Africa, but is nested within the variation seen in Africa. The figure below (by Sasha Gusev) contrasts a false essentialist view of race, in which different racial groups are genetically largely separate and distinct (A) with the genetically accurate view of race, in which variation within world regions is nested within the diversity seen in Africa (C and D).

by Sasha Gusev

There are examples of changes in functional gene variants which are geographically patterned, such as those related to skin colour (which matters because darker skin provides protection against harmful effects of sunlight whereas lighter skin allows better vitamin D synthesis), but again these patterns are complex and don’t fall into socially constructed racial categories. Skin colour, and its genetic underpinning, varies considerably across the African continent: many of the genes related to light skin in Europe are also seen in some African populations, such as relatively light-skinned hunter-gatherer populations in Southern Africa. Lactase persistence – the ability to digest milk in adulthood, which is absent in most mammals – is another example of a physiological trait which appears to have resulted from relatively recent changes in gene frequencies which were selected for because they conferred some kind of advantage. This trait is linked to cultural practices, not race. Populations which have a history of animal herding tend also to have the ability to keep digesting milk in adulthood, because milk products are an important part of the diet. Lactase persistence is relatively common in Europe for this reason, but is also common in several other populations worldwide which also keep animals.

2) Racism has biological consequences, but race does not have biological causes

Though race is not a genetically useful concept, race does exist as a social construct and so there still might be value in recognising race in some contexts. Classifying humans into racial groups has consequences. Some societies have enforced different treatment of socially constructed races through the legal system. The “Jim Crow” laws in some US states enforced segregation of Black and White people in state-funded organisations such as schools and other facilities; the facilities provided to Black populations were often of inferior quality to those provided to White populations. In such systems, race becomes a predictor of educational, health or other life outcomes, not because of biological differences between races but because of structural discrimination against Black people. Such structural racism does not require legal enforcement; there are many other examples of discrimination against African Americans which continue to lead to racial inequalities long after the Jim Crow laws disappeared (and similar examples of structural racism exist in other countries).

Collecting data on how people identify in terms of race may therefore be useful, to document such inequalities in order to develop strategies to counter them (though such data collection exercises are social constructed themselves; people from the Middle East and North Africa are counted as White in the US census, for example). There may also be circumstances in which it’s useful to use racial categories in clinical contexts, because Black and White people will have different experiences which may lead to different health outcomes: higher levels of stress may lead to poorer health amongst discriminated-against groups. Discrimination may also extend to healthcare facilities. An example of the latter includes cases where race is incorporated into clinical practice because of the belief that there are inherent biological differences between races. A widely used algorithm used to assess kidney function assumes there are biological differences in muscle mass between Black and White people. Tracing the origins of this assumption revealed that it was not based on any credible scientific evidence. In clinical contexts, as in all other contexts, it’s not useful to believe that there are inherent biological differences between races, but it is useful to be aware that different racial groups are likely to have had different experiences which have consequences for their health.

3) IQ might measures a narrow range of cognitive abilities, in specific cultural contexts, but cannot be used in research across very different cultures.

Given that genetic variation between humans doesn’t map onto racial categories, it doesn’t make sense to claim that differences in IQ scores between races are substantially determined by genes. IQ itself is also a complex trait to study. IQ is intended to be a measure of intelligence, which is a psychological construct. It’s a property of the brain, in other words, and so is not easily observable or measurable as a physical characteristic such as height might be. There isn’t clear consensus among researchers about what intelligence is, though Western psychologists tend to include in their definitions the ability to learn from, and adapt to, the environment. This ability will involve several sub-components of intelligence, and the way IQ scores are constructed acknowledges this complexity. The results from different types of cognitive tests, such as tests of verbal and visuo-spatial ability, are typically combined to calculate an IQ score. Such tests prioritise skills which are valued and taught in formal education systems, including abstract reasoning, for example. These tests, and Western definitions of intelligence, may therefore miss out other components of intelligence. Western societies tend to be highly individualistic compared to societies throughout most of human history, and their definitions of intelligence may deprioritise aspects of intelligence which are valued in other societies, such as social competence, or other aspects of interacting with other people.

The cognitive tests used to calculate IQ scores require knowledge which is culturally specific, and so have been criticised for being unable to accurately measure intelligence even within the same population (for example, different social classes, ethnicities or religions may have different cultural knowledge). Attempts have been made to try and reduce this problem by reducing the amount of culturally-specific knowledge required, and trying to focus on other abilities. But no cognitive tests exist which can measure intelligence across very different populations. Given that tests of language may require very specific cultural knowledge, it’s been argued that tests of visuo-spatial ability are more “culture-free” and more appropriate for use across different cultures. But even these tests require familiarity with materials which are not universally used across the world (such as paper or building blocks). This is important not least because people may be given limited time to complete cognitive tests: speed of processing ability is often included in the definition and measurement of intelligence. In contexts where there is little or no access to formal education, or where these types of materials are just less frequently used than in Western classrooms, testing visuo-spatial awareness using cognitive tests developed in the West will not provide a good measure of the intelligence of those populations. It certainly won’t provide a measure which is comparable with scores calculated from similar tests given to populations which are familiar with such tests. There is no such thing as a culture-free cognitive test.

So IQ scores are not meaningful in a cross-cultural context, but some psychologists have argued that when used within the same population, IQ scores are useful. For example, they show significant correlations with other life outcomes, such as income or health, and therefore may be a useful predictor for life outcomes. Like most aspects of IQ research, relationships between IQ and other outcomes are not always seen and may be confounded by other factors, such as socioeconomic status or motivation. It is therefore appropriate to interpret research correlating IQs with other outcomes with caution. Any practical use of IQ testing should also be cautious. The first uses of contemporary cognitive tests were to identify children who performed particularly poorly in school settings, in order to provide them with the additional help they needed. This may be one of their few uses today, as this involves using IQ testing in the context in which it was devised - measuring skills valued in formal education – and using them cautiously, as most useful in identifying those at the tails of the distribution, rather than assuming variation within the majority-range of IQ is particularly meaningful (any use should also acknowledge their limitations and potential biases). IQ testing may also have some use for identifying cognitive changes over time within individuals (given data collected repeatedly form the same person will be less subject to bias). A fundamental problem with any widespread use of IQ testing is that, as with all psychological constructs, it isn’t very clear what IQ scores are really measuring. This problem is not specific to IQ: the field of psychology has been reflecting on its research practices recently, as it turns out that many results in psychology cannot be replicated (sometimes referred to as the “replication crisis”). Psychological constructs are just much more difficult to define and measure than are physical traits such as height. Given there is so much uncertainty in the psychological literature on intelligence, the confident claim that there are racial differences in IQ which are substantially genetically determined can’t be supported by current psychological evidence.

4) The heritability of IQ

Whether or not IQ is actually a good, comprehensive measure of intelligence, it might perhaps provide a rough measurement of a narrow range of cognitive abilities in particular cultural contexts. There is research which has attempted to determine how much variation in IQ scores is associated with genetic variation between individuals – the heritability of IQ, in other words. Variation in IQ between individuals does appear to be linked to genetic variation in individuals. This isn’t surprising given all human characteristics have some genetic basis. These links are complex, which also isn’t surprising given IQ scores, at best, measure different aspects of intelligence. There aren’t just a handful of gene variants which are associated with IQ; a very large number of differences in gene sequences are associated with IQ: IQ is a “polygenic” trait. But scientific racists don’t just make the claim that IQ is heritable; they claim that IQ has such substantial genetic underpinnings that any environmental changes – including educational policy – won’t have much impact on IQ. This statement can’t be supported by contemporary scientific evidence: recent research suggests that the genetic contribution to intelligence is small. Differences in intelligence between individuals appear to be particularly strongly affected by the environment.

Early research on this topic did suggest the heritability of IQ might be reasonably high. Such research didn’t directly investigate genetic variants – it was typically done by psychologists or other social scientists rather than geneticists – and used twin studies to estimate genetic and environmental contributions to IQ. Twin studies compare traits such as IQ between identical twins (who are genetically very similar to one another) and non-identical twins (who are only as related to one another as any other siblings), in order to tease apart genetic and environmental contributions to these traits. These methods broadly assume that if identical twins look more similar on a trait than non-identical twins, then the difference must be explained by the genetic contribution to that trait. But they involve a number of assumptions, such as that identical and non-identical twins are exposed to similar environments. If, for example, identical twins are treated more similarly than non-identical twins, then twin studies may overestimate genetic contributions to traits (identical twins may have very similar outcomes not only because of similar genes but also because they experience more similar environments than non-identical twins). More recent research which does involve examining genetic material and attempting to directly correlate gene variants with IQ scores suggests much lower estimates of heritability for IQ, with the contribution from genes shrinking to explain only a small proportion of variation in IQ. Even some of this research may still be confounded by environmental and other factors which may inflate heritability estimates.

Recent research also suggests the degree to which gene variants are associated with IQ may vary substantially according to context (such as socioeconomic position), even within the same population. The heritability of a trait varies between environments, because traits are influenced by the environment, as well as genes, and the interaction between genes and the environment. If everyone were raised in exactly the same environment, there would be high correlations between genetic variation and variation in traits like IQ because there is little environmental variation to influence these traits. When different individuals experience different environments, genetic correlations will typically be lower because variation in the environment will explain a larger proportion of variation in traits. But how much the environment influences traits will depend on a number of factors including how much variation there is in environmental conditions, and also on the trait we’re interested in: IQ appears to be more sensitive to variation in environments than the physical trait of height, for example (even though height is also sensitive to environmental conditions, having increased substantially in recent decades in higher income countries because of improved health). We don’t actually know much about the heritability of IQ across most of the world, as the great majority of research on genetic contributions to IQ has taken place in populations which have majority European recent ancestry.

There is already a lot of evidence that IQ is very malleable. IQ tests involve measuring skills which are taught in formal education, so that people with more exposure to formal education have higher IQ scores. Motivation, experiencing anxiety around test-taking and environmental factors (such as lead exposure) all affect scores as well. Even scores calculated for the same person won’t necessarily always be the same: IQ scores improve with training, for example. Clear evidence of this malleability is that there have been substantial increases in IQ scores in Western populations over the last century or so (a phenomenon known as the “Flynn effect”). These increases cannot be caused by changes in gene frequencies, as there hasn’t been enough time for such changes to happen. Instead, they are explained by environmental changes such as increased exposure to formal education and other cultural changes, as well as better nutrition and healthcare (which improve health, including brain health).

Identifying the genetic contributions to any traits is still relatively new research, and identifying the genetic contributions to psychological traits is particularly difficult, because of the problems of defining and measuring such traits, as well as their substantial environmental malleability. Given the weight of evidence we have that IQ is environmentally malleable, claiming that the much weaker evidence we have on genetic contributions to IQ is reason for wholescale reform of educational policy is hubris (which is not to say understanding how gene variants contribute to variation in intelligence will never be useful; we just need a better understanding of the mechanisms by which genes and the environment contribute to intelligence - which might first require that we extend our definition of intelligence beyond IQ scores). Given all these complexities in heritability, it is a fundamental principle of genetics that evidence of heritability within a population can’t be used to draw conclusions about genetic contributions to traits between groups, such as between races (even if they were genetically meaningful categories).

5) We would not expect there to be substantial genetically-underpinned differences in cognitive function between different world regions

The measurement of intelligence across very different populations is currently impossible given the current tools we have available. It’s not clear that it would even be a useful thing to attempt, given that humans may use different components of intelligence for different lifestyles. Humans have lived for most of our history as hunter-gatherers, but the skills valued by contemporary high income societies, which are those developed in formal education, may be less useful to a hunter-gatherer. There are significant behavioural differences between populations in, for example, subsistence strategies (the way we make a living), so coming up with intelligence tests which are comparable across such behaviourally different populations may never be achievable. These behavioural differences between populations, however, are highly unlikely to be substantially determined by genetic differences. What characterises our species is enormous behavioural flexibility; this is likely the secret of our success, our expansion across the globe and our considerable population growth. Hunter-gatherers don’t all make a living in the same way, but were able to spread out across the globe, extracting resources from many kinds of different environments by adapting their behaviour to local conditions. The development of agriculture, animal-herding and various industrial technologies were also not the result of changes in gene frequencies, but were developed (on many occasions, in different world regions) through our ability to adapt our behaviour to the environment, and then spread new behaviours through social transmission (learning and copying new behaviours from other individuals).

In comparison to other species, all humans have a high degree of intelligence – the ability to respond and adapt to novel environments. This behavioural flexibility means we didn’t need to evolve different cognitive strategies through the process of natural selection. The likelihood of developing locally-specialised cognitive strategies through changes in gene frequencies is any case low, given the complexity of the heritability of cognitive abilities. The physiological adaptations that have arisen in our recent history tend to involve relatively more straightforward links between genetic variation and the trait in question. Highlighting the importance of environmentally-determined behavioural flexibility isn’t “blank-slatism”: the assumption that human behaviour is infinitely malleable and determined only by the environment. Clearly, genes play a role in human behaviour, and there are constraints on what we can do. But equally, the claim that there are significant racial differences in behaviour substantially caused by variation in genes is simply not supported by current scientific evidence on human behaviour and cognition.

6) “Research” claiming to find evidence that there are racial differences in IQ which are substantially influenced by genetics typically does not meet scientific standards

There are published papers in the academic literature which claim there are genetically-determined differences in IQ between races, but they typically do not meet the standards required for rigorous scientific evidence, and are often based on discredited data. The person who has perhaps done most to promote this claim is psychologist Richard Lynn, who produced what he claimed was a database providing average “national IQs” in nation-states worldwide. Given the difficulties of defining and measuring intelligence cross-culturally, sufficient evidence to build such a database does not exist. Even if we suspended disbelief about the measurement of intelligence and assumed it was possible to build such a database, no rigorous scientist would have used the methods Lynn did. Lynn has actually never described the methods he used to produce the database, despite updating the database over some decades (it’s a fundamental principle of research that methods are described in sufficient detail for others to replicate a piece of research). He appears to have searched through the existing literature for studies done in a particular country and assumed the results could be used to calculate the “national IQ” for that country. As a result of very patchy data on cognitive tests in many countries, the database is full of examples where tiny and wholly unrepresentative samples are used to calculate average “national IQs”. Just as some example: the “national IQ” of Angola is calculated from tests given to only 19 people, of Dominican Republic from 34 people, of Sierra Leone from 119 people tested in 1966, of Eritrea from orphans living in orphanages. None of these samples will produce anything close to a representative average IQ of that country.

Other psychologists who have tried to replicate his database have also found evidence that the dataset is systematically biased. Lynn appears to have selected studies based on the IQ scores themselves: he included all published studies which found particularly low IQ scores for nations in sub-Saharan Africa, but excluded some with higher IQ scores. As a result, the average “national IQ” in sub-Saharan Africa is only 70, an exceptionally low value, typically used as the cut-off value for intellectual impairment (carrying the wholly implausible implication that roughly half the population of sub-Saharan Africa is intellectually impaired). Despite this database clearly failing to meet typical standards of scientific rigour, Lynn used this database to argue that differences in intelligence between nations and races are substantially genetically determined, and that these differences in intelligence are the cause of differences in economic development between world regions. He also promoted this database in white supremacist spaces, arguing for the “phasing out” of certain populations and for the “rehabilitation of eugenics”.

Others have used his database to make similar arguments, often providing more evidence that those who make claims about genetically-determined racial differences in intelligence are not conducting rigorous science. One paper used the “national IQ” database to test whether populations which are further away from Africa have higher intelligence. The method used to calculate distance from Africa assumes the earth is flat. When this flaw was pointed out to the editor of the journal which published the paper, no action was taken (typically, the author of a paper about which as concern has been raised would also be notified). The journal is Intelligence, the flagship journal of the intelligence research community and its society the International Society for Intelligence Research (ISIR). Some influential twin research has also been called into question for not meeting scientific standards.

As another example, some researchers have suggested that IQ scores may be declining a little in some higher income populations in recent years. This may be because of environmental changes which are affecting IQ scores, including cultural changes. Given the difficulty of understanding what IQ is measuring, this doesn’t necessarily mean these populations are becoming less intelligent. But a problem with this literature is that it’s contaminated by contemporary eugenicists who want to demonstrate that populations are becoming less intelligent, in order to argue for their eugenic social policies. Eugenics is the belief that human populations can, and should, be ‘improved’ by selectively allowing people to reproduce or migrate (only allowing people with “desirable” traits to have children or immigrate into a particular population). It’s an ideology which is based on a flawed understanding of science, and is essentially an ideology that wants to divert society’s resources towards those who are already privileged, while taking resources away from those who are socioeconomically disadvantaged. Eugenicists have recently claimed that intelligence is declining in high income populations because of “dysgenic fertility”: this is the idea that “wrong” kind of people (those with low intelligence) are having “too many” children and outreproducing the “right” kind of people (those with higher intelligence). They have tried to provide evidence for this argument using flawed data, including flawed studies which claim to show that intelligence is negatively correlated with fertility. Those studies which have used high quality data, from Scandinavia, to investigate whether aspects of intelligence are correlated with fertility in recent decades have shown, if anything, that there’s a positive correlation between cognitive ability and fertility (driven by lower fertility among those with particularly low cognitive ability).

The field of intelligence research has a long history of research being misused for unscientific purposes. The father of eugenics, Francis Galton, began the academic study of intelligence, because of his belief in its importance in distinguishing between inferior and superior types of people. The example of Richard Lynn shows that political motivations have affected the intelligence literature up to the present day. The journal Intelligence has published the largest number of “national IQ” papers both by Lynn and others. ISIR has also repeatedly allowed scientific racists such as Richard Lynn to present at their conference. In 2024, Emil Kirkegaard presented at the conference, despite having been removed from ISIR’s programme in 2022 because another attendee had objected to his presence. The repeated platforming of extremely low quality research, which appears to have a political rather than scientific motivation, in the intelligence research community makes interpreting research on intelligence particularly challenging. We simply cannot be confident enough in existing research on intelligence to make the strong claim that there are genetically determined racial differences in intelligence. And, when combined with what we know about human genetic variation, behavioural variation and how evolution works, we know that this statement simply can’t be supported by scientific evidence.