The Zombie of Genetics

Joe Thornton

I was struck in our readings by how hard it has been to sever the tie between genetics and racism. Like a zombie, racialized thinking keeps coming back in new genetic language and concepts, rising again in the wake of findings that should have put it to rest. Through its relationship to apparently cutting-edge science, racism has continually legitimized itself in society. In this post, I consider how and why racialized thinking in genetics keeps resurrecting itself, examine some ways in which it distorts and undermines the science we do today, and and suggest how an awareness of the field’s historical relationship to racism can help us to improve the science of genetics and construct a more powerful scientific argument against racism.

Contents

1. Racism reanimated
2. Co-production of genetics and eugenics
3. A history-aware scientific argument against genetic racism
4. Scientific vanity: co-producing the genetics of intelligence 
5. The end of the undead: Hope for genetics?
6. Notes and more reading

Racism reanimated

Here are a few examples of the zombie-like persistence of racialized thinking in genetics:

• Early in the 20th Century, geneticists established race for the first time as a genetic concept: human races were conceived in fixed, typological terms as genetically distinct from each other, as if they were separate species. In the 1930s and 1940s, Theodosius Dobzhansky explicitly redefined races as genetically related groups that differ only in the frequency of alleles (different versions of a gene) because they have been partially isolated from each other. Dobzhansky wanted to undermine racism in society and to replace typological concepts in evolutionary biology. But his new definition was easily made compatible with a racist agenda: if races are defined as groups with different allele frequencies, then those alleles might cause the behavioral and cognitive traits that are used to justify racial discrimination, exploitation, and inequality. Despite Dobzhansky’s intent, the new definition ended up providing apparent legitimation for human races as if they are real genetic objects, thus helping to establish a new “modern” form of racism.
• Richard Lewontin showed in 1972 that the traditional races provide a very poor categorization of genetic differences among humans across the globe, accounting for only about 6 percent of all the variation in allele frequencies; the vast majority of genetic differences are between people of the same race. This result, which has been reproduced repeatedly using genomic sequences, shows that categorizing people according to the traditional races is a poor way of describing natural variation among human beings or predicting their characteristics. It’s as if one went to the hardware store, divided everything into two groups – things that are round when seen from below and those that have corners – and then treated them as if these are natural categories, despite the fact that each group would include some hand tools, some metal fasteners and lumber items, some power tools and appliances, and so on. But despite the power of Lewontin’s critique, many population genetic studies today – even those that intend to reveal and honor complexity in human genetic diversity – continue to categorize people using continentally defined “ancestry groups” that are easily translated into traditional racial classifications. Similarly, ancestry services have built a booming business on the fact that the huge number of SNPs in human genomes can be used to probabilistically classify individuals into pre-defined racial groups (or finer scale ethnic groups). Despite their intent, an effect of these practices is to continually reinforce the idea that the traditional races provide a primary and genetically real way of categorizing human beings.
• The population geneticist David Reich recently challenged the mainstream conception of race by using genomic data to show how the history of human populations has been a continuous process of mixing over the last 50 million years. He also condemned racism on ethical grounds. He then told us that, even though races are not historically clear or stable categories, we should accept the progress of science and prepare ourselves for the “likely” discovery that races differ genetically in cognitive capacity.

How can we understand this zombie-like behavior of race concepts in genetics? Can we conceive of our science in a way that will see the world more clearly, and in a way that will not unintentionally perpetuate and reinforce racism? A key obstacle is the fact that we lack a language and analytical lens to understand clearly how science and society interact to constitute the conceptual structure of a scientific discipline. In the case of genetics, we need a better understanding of the history by which racism in society and the development of genetics as a discipline have exerted formative and continuing influences on each other.

Co-production of genetics and eugenics

A concept that has helped me to wrestle with the persistent relationship between genetics and racism is the idea that the social order and the scientific order co-produce each other: that is, the ways in which we understand and represent the world to ourselves and each other via science are inseparable from the ways in which we organize our shared social and political lives in that world (1). Society’s order always reflects ideas about the nature of human beings and the world around us; science is the authoritative source of those ideas in modern society, so scientific knowledge either legitimizes or challenges some aspects of that order. Conversely, scientists and scientific institutions always operate in a social, political and economic context, which shapes the questions that we think are worth asking, the categories we impose on the world when we ask them, the kinds of cause-effect explanations we can imagine, and the various burdens of proof that that we impose on our inferences, depending on whether they are expected or unexpected given our prior beliefs and perspectives. That is, the social order affects how we do science and the knowledge we produce, which in turn affects the social order.  This reciprocal influence is ongoing and inevitable.

Our readings highlight this feedback between science and society in the history of race and genetics. Genetics was born with deep ties to the political and cultural agenda of eugenics – to rationally and scientifically engineer the “quality” of the human population based on genetic principles. Many of the field’s most fundamental concepts and analytical tools for understanding and analyzing genetic variation and causality were developed as part of this effort, and it was in this context that race as a genetic idea was established. Until the late 19^th Century, race was a cultural or geographic way of categorizing people, rather than a biological one; with the rise of genetics and eugenics, race was redefined as a phenomenon of biological heredity, first using the ideas of Mendel and then again in the 20^th Century using the frequency-based framework of population genetics. In turn, putative heritable differences between races became scientific objects that could be studied using new scientific techniques. Genetic science as a whole – including work that had no direct relationship to race – gradually built up a rich body of knowledge and methods that applied concepts developed in the context of eugenics, such as estimating heritability, partitioning genetic variance, and inferring ancestry from allele composition. It gradually became more and more difficult to think about or practice genetics in terms that did not implicitly or indirectly reproduce and entrench racialized science and, indirectly, a racialized social order. When new approaches have emerged in genetics, like Dobzhansky’s reconception of evolutionary genetics or whole-genome sequencing, they have been influenced by and deployed within the context of the overwhelming momentum of racialized thinking, both in society and in genetics.    

Social forces themselves are diverse and have the potential to move science in multiple directions at different times. On occasion, cultural and political developments have propelled scientific critiques of racialized genetics, such as Lewontin’s work in the wake of the U.S. civil rights movement, or the broad retreat from eugenics after World War II. But these developments ultimately became incorporated into the entrenched ways of doing and thinking about genetics and race. Race thus got a makeover each time as state-of-the-art, science-sanctioned reality. The zombie rose again.

A history-aware scientific argument against genetic racism

By understanding the historical interactions between genetics, eugenics, and racism, I hope that we can avoid unintended outcomes like Dobzhansky’s and make our scientific critiques stick more effectively than Lewontin’s argument about variation has. The current mainstream arguments against genetic racism are inadequate, in part because they misunderstand the relationship between science, society, and nature. The statement that race is “just a social construct” with no genetic reality does not seem very persuasive in light of many people’s experience. We can all see that children generally inherit from their parents the characteristics that society uses to categorize them by race, such as skin color or hair texture. And if ancestry services can analyze DNA from your cheek and tell what fraction of your ancestors came from West Africa or from Europe, how can race possibly be just a social construct?

The argument against genetic racism will be much stronger if it explicitly incorporates both an understanding of the history of racialized genetics and a deeper analysis of the methods and findings of contemporary genetics.  This argument will need a lot more work to make it persuasive, but I’ll discuss here some initial thoughts about a few elements that could go into it. 

First, the apparent fact that the visible markers most often used to define race are passed from parents to children does not make it biologically natural or meaningful to classify people based on race. That skin color or hair texture are heritable does not imply that they reliably predict the inheritance of other phenotypes, particularly the cognitive and behavioral traits that are the focus of racist justifications of social inequality. A striking discovery of modern population genetics is that the genes that are most strongly differentiated in frequency between traditional races are those that produce the visible markers by which we classify individuals into races on sight. That is, race as a genetic classifier is best at predicting itself.         

Second, modern sequence-based classification of people into races and ethnic groups reproduces socially imposed categories rather than revealing natural ones. That 23andme can tell from a DNA sample where your great-great-grandparents probably lived – and which of the pre-defined ancestry groups that they probably “belong” to – does not make race a natural category that is “out there” in the world. Once ancestry services have sorted their first customers into pre-defined groups based on self-identified race or geography, it becomes easy to assign others into those same groups based on similarities in their genome sequences. This does not mean that the groups represent the natural way to categorize people or, more importantly, that they tell us anything biologically interesting or important that distinguishes people in one such category from others. Again, the racial categories serve primarily to predict themselves.

Given the fact that racial categories are predominantly self-referential, why do geneticists keep using race as if it were a primary descriptor of distinct genetic groups? The history shows that in a culture with a different political history, we wouldn’t be asking about the categories of Black, White, Asian, Pacific, and indigenous American at all. Because of the long history of human migration, we could classify people more precisely and reliably into other kinds of groups, and the groups that would work best for our purposes would shift with the number of generations back we are interested in, because the porous boundaries of human populations and their locations have changed continuously throughout history. The races that we recognize now were created – and subsequently geneticized – because they correspond to broad continental groups associated with the political/economic relationships produced by European/American colonialism and slavery (and their aftermaths). And the history of genetics shows that our science played a key role in legitimating these categories, just as it continues to do now.  

These ways of categorizing people are so baked into our society and into our science that it is extraordinarily difficult to think and talk about human genetic variation without resorting to them, despite what poor descriptors they are of the actual genetic similarities and differences among people around the world.  The historical interplay between genetics and society created the zombie, so it takes extraordinary effort to think in ways that don’t bring it back to life in a new form.  

This argument – that race is an impoverished way of describing human genetic variation that persists primarily because of social forces rather than scientific value – helps, but it doesn’t go far enough.  It seems to leave room for racists to argue that the small portion of genetic variation that corresponds to differences in allele frequency between racial groups (around 6 percent according to Lewontin’s studies) might cause differences in cognitive and behavioral traits.   

In fact, the 6% statistic hides underlying genetic patterns that are even more inconsistent with the premises of racialized thinking. An oft-repeated version of the argument for genetic racism is that natural selection drove the evolution of increased intelligence in Europeans (and, in some versions, Asians too), but not in other so-called races. If this were true, the alleles that cause increased intelligence would be common in some races but rare or absent in the others. But it turns out that virtually none of the variation is structured in this way. Of the alleles that differ in frequency between the traditional races, the vast majority are rare in one group and essentially absent elsewhere (2). Differences like this could cause conditions that are rare in one group and absent in others, but they don’t have the potential to cause broad phenotypic differences between people of different races.  Moreover, most of the variation in the frequency of alleles between is attributable to alleles that are common in all races but differ slightly in their frequencies, a pattern that is also inconsistent with this version of genetic racism.

Now let’s add two other scientific observations. First, the vast majority of the genome is of no biological significance whatsoever; rather, it is an amassed load of junk between genes that has no known biological function. The fact that some sites in the genome differ in their DNA sequence between so-called races does not imply that they cause biological differences between those groups.  In fact, without specific evidence to the contrary,  differences in gene sequence should be presumed to have no biological effects at all. Second, there is no evidence causally linking any of the sequence variants that differ in frequency among races to cognitive or behavioral differences. The idea that racial categories correspond to genetic differences in the traits used to legitimize social inequality dissolves under scrutiny of its scientific premises. 

Scientific vanity: co-producing the genetics of intelligence

This brings us to other pillar of scientific racism: the idea that there are differences between the races in inherited behavioral and cognitive traits, particularly intelligence, and that it is possible (or will soon be so, as David Reich contends) for these differences to be established scientifically. I think that the idea of co-production can help to address the persistence of this idea and to reveal its fundamental scientific flaws.

Other than wealth, the idea of intelligence is arguably the predominant currency of advancement, status, and virtue in our society. It is also the major way in which social inequality is legitimated, based on the idea that smart people advance to the best schools, that they can handle the most rewarding and challenging jobs, and that they should be remunerated accordingly. Living in this putative meritocracy seems to reinforce the reality that intelligence is genetically determined, and therefore that it might differ among groups: parents who are apparently intelligent seem to produce offspring that are also apparently intelligent, so why shouldn’t such differences in inherited intelligence extend to groups of parents and kids? 

This perception is a legacy of the cultural and scientific work done by genetics and especially eugenics. Parent-offspring correlation is the heart of the genetic concept of heritability, which Galton – the founder of the field of statistical genetics – invented specifically to justify eugenic efforts to produce more “geniuses” and fewer “imbeciles.” The same is true of the work of Pearson and Fisher, who developed the fundamental tools for quantifying the genetic influence on phenotypic variation in the context of their intense personal commitments to eugenics. During the same period, eugenics advocates in the field of psychology developed an apparently scientific concept of intelligence as a distinct phenotype – the intelligence quotient (IQ)– thus establishing intelligence as a measurable “thing” that people have in varying degrees.

The work of the early geneticists established the definition of genetic causality as heritability. But this equivalence is problematic, because traits can be passed from parents to offspring for cultural and social reasons, not only biological ones. The structure of social advantage in our society produces correlations between parents’ and their children’s intelligence, just as it does between parents’ and their children’s financial wealth. Families possessing the social currency of intelligence hand it down in innumerable ways when they socialize and educate their kids. Social advantages have genealogical structure to them – parents’ status and advantages are conferred on their kids, even if there is nothing biological behind them. 

Social heritability is a fundamental problem with respect to race and genetics. Skin color and some other visible markers of race are biologically heritable; being subject to racial discrimination – and all its effects – is therefore heritable, too. Geneticists are well aware that genes’ effects on phenotypes typically depend on the environment, and we have tools to try to tease apart these dependencies (3). But we cannot effectively distinguish between genetic and environmental causes when genes are correlated with environment, as they are because of racism. Even more problematic is the fact that genes associated with race actually cause the environment, because the inherited visible phenotypes associated with race determine how people are treated in society, which in turn affects innumerable aspects of a person’s development and experience. 

The concept of partitioning out the particular effect of genes independent of the social environment becomes not only difficult but fundamentally misconceived, at least for cognition and behavior. The thing that we as geneticists want to isolate – the pure genetic causes of phenotypic variation – does not exist per se, waiting to be unveiled if we are only clever enough to disentangle it from other influences. In a society replete with racism and inherited privilege, any genetic causality is also environmental causality, and vice versa. The object of genetic analysis, in this case at least, is not a natural object at all. The scientific effort to characterize it – to identify the genetic causes of intergroup differences in cognition and behavior – is a vain one, which will never reliably yield the knowledge it seeks to produce.  

This is a hard thing for a geneticist to admit. But it becomes a bit easier if we recognize how and why we were brought up to pursue this kind of agenda, and why it is so hard to abandon. Doing so requires us to question the reality of the fundamental thing that our discipline is meant to discover and the adequacy of our methods to prove its existence. It also forces us to question the system of reward and status in our society and its justification by intellectual “merit” – which can hit pretty close to home for an academic scientist. Our sense of purpose and value as scientists and individuals make us unconsciously but powerfully invested in concepts that the zombie embodies.

The end of the undead: Hope for genetics?

David Reich has said that those who criticize studies that seek to identify putative genetic causes of cognitive differences between racial groups are motivated by a repressive political orthodoxy. He is wrong. The arguments that I made above are driven by an open-eyed analysis of fundamental scientific flaws in the research program that Reich defends and an understanding of the ways in which social forces have helped to create those flaws.  

In contrast, consider what kinds of willful scientific blindness, and what kinds of motivations or political assumptions, conscious or unconscious, must underlie a devotion to the research program that Reich tells us we must be open to pursuing. We have very solid evidence that non-genetic factors are a huge cause of differences in the distribution of IQ scores between races in the United States. It has been shown, again and again, that improving education, nutrition, and other correlates of social status increase IQ, and these factors differ between groups because of social inequality. Improving these factors in the United States has been shown to have an even bigger effect on the scores of Black children than White children. Moreover, there have been pronounced increases across the globe in IQ scores during the 20th Century, which cannot possibly be explained by genetic factors because the genetic composition of human populations has hardly changed over this time.  Within the U.S., the gap in IQ scores between Black and White children narrowed dramatically from the 1970s to the 2000s, which again can only be explained by changes in society, not genetics  (4).

We have overwhelming evidence of environmental causes for intergroup differences in IQ , no persuasive evidence of genetic causes, and no promise that our science could identify such causes to a high standard of rigor.  So why do we continue to seek or expect them? Reich has it exactly backwards about the effect of orthodoxy: the pervasive history of racism in the structure of genetic science hides fundamental scientific flaws in the project of finding the basis of intergroup differences in cognition and behavior. That history has lowered the scientific standards that many geneticists impose on themselves with respect to these questions.

My hope is that we might admit two things: that science is always coproduced with aspects of the social order, and, that genetics in particular has been continually coproduced with a social order that has caused immense suffering for vast numbers of people. As scientists, we are expected to suppress our social awareness in our science – to be, or at least pretend to be, unconscious of the ways that society shapes science and science in turn shapes society. This reflex makes us unwitting agents of coproduction of the existing social order, which in the case of genetics and race has been a mostly exploitative and dehumanizing one.

If we scientists are inevitably going to coproduce social order, why not instead become conscious agents in this process?  As investigators, we can choose the questions that we ask and the assumptions, concepts and burdens of proof that we impose on our observations of nature. So why not do that with awareness of the ways in which those choices affect and are affected by society and history? Scientists are taught to act as if we stand apart from society. It felt in the reading group that we were actively trading in this illusion for a shared hope (5) that we can intentionally create a better science – a discipline of genetics that not only contributes to making a more just society but that also sees the biological world more clearly. With enough work, we ought to be able to recraft genetics as a science that can query and interpret nature without all of the distorting intellectual baggage of the 19th and 20th Centuries.

Notes

1. For more on coproduction, see Sheila Jasanoff. States of Knowledge: The Co-Production of Science and Social Order, Routledge, 2004.
2. This is the work of my colleagues Arjun Biddanda, Daniel Rice, and John Novembre in eLife 9:e60107, 2020. Available here.
3. To appreciate some fundamental flaws in these approaches caused by gene-environment interaction — which pertain even without the deeper problems of  gene-environment correlation and causation — see Richard Lewontin, The analysis of variance and the analysis of causes, Am J Hum Genet 26:400-411, 1974.  Available here.
4. James R. Flynn, What is Intelligence?  Cambridge, 2009.
5. See Richard Rorty, Philosophy and Social Hope, Penguin, 2000; Contingency, Irony, Solidarity, Cambridge, 1989; and Objectivity, Relativism and Truth, Cambridge, 1990.

Return to Contents