Heritability Explains Less About Mental Disorders Than You Think

The paradigm of biological psychiatry—in short, the view that mental disorders are diseases of the brain—is essentially based on three assumptions: First, that the brain is the organ of the mind; second, that psychoactive substances or electrical stimulation of the nervous system alleviate the symptoms of mental disorders; and third, that parents genetically pass on a predisposition to mental disorders to their children.

It is now common knowledge that there is no reliable biological test, or biomarker, for any of the hundreds of disorders in the DSM-5. I assume that our mind is embodied, yes. That is why psychoactive substances affect our feelings, thoughts and behavior. Many people know this from their own experience, even if it is only with caffeine, alcohol or tobacco. However, mental disorders are not concrete things that can be found with a brain scanner or treated with medication like a bacterial infection with antibiotics.

Much has already been written about these points, for example in my book on mental health and substance use (open access). I will therefore focus here on the third pillar of biological psychiatry: genetics.

Our bodies are also biological and the result of a long evolutionary history. Genes are permanently active and regulate the production of proteins from which larger structures are built. In this sense, everything is somehow genetic. And genes also play a role in our behavior. Eric Turkheimer, professor at the University of Virginia, formulated the first law of behavioral genetics 25 years ago: All human behavioral traits are heritable.

Nevertheless, the heritability estimates that are still frequently reported exaggerate the importance of biology for psychiatry. In this essay, I explain why this is the case and why it is not possible to clearly separate the role of genes, environment and psychology.

How to measure the genetic contribution

We have probably all heard statements such as “intelligence, schizophrenia or ADHD are 80 percent heritable” more than once. Scientists also regularly quote such figures. But only few understand that these statements do not prove a strong genetic influence. Leading human geneticists and evolutionary biologists have been explaining this for decades.

However, this message is hardly taken up, probably because it does not fit in with the biological-neuroscientific zeitgeist. In order to explain the criticism, I will not go deep into the mathematical basics here. Instead, I will briefly describe the most common measurement methods and then use concrete examples to expose the misunderstandings about heritability.

Two methods are particularly important for genetics in psychiatry: genome-wide association studies (GWAS) and twin studies. For both, a distinction is made between the “genotype” and the “phenotype.” The latter refers to the appearance of a living being and can be a physical measurement such as height, but also a psychological characteristic such as intelligence or the severity of a depressive disorder.

In the 1990s, genetic studies were still very complicated and expensive. At that time, scientists had to select only a few so-called candidate genes, based on what they thus far knew about genetics, and to investigate their variation between people. Would certain genotypes occur more frequently together with a phenotype? For example, are people diagnosed with depression more likely to have gene variant XYZ than ZYX?

Genome-wide association studies

Because it was already assumed at that time that mental disorders were highly heritable, the small and inconsistent results found in these studies were disappointing. For this reason, genome-wide association studies, known as GWAS, were called for in ever larger samples. These allowed scientists to not just investigate a selected few, but all genes at the same time. There are now studies in which the data of tens, hundreds of thousands or even millions of people have been examined.

Two findings are clear, as shown in the following table: The larger the studies become, the more genes stand out as statistically significant; however, this is due to the mathematical fact that smaller and smaller differences are found with ever larger samples. Therefore, the explained variance remains rather small even in these newer studies. This means that even if 270 risk genes for schizophrenia have now been found, they can only explain a small proportion of the symptoms.

Table: According to Giangrande and colleagues. Although large studies have now been completed for many disorders, only a few percent of the differences in symptoms are usually explained genetically (explained variance). The number of genes represents the number of locations on the genome where statistically significant differences were found.

It is often forgotten that the “A” in “GWAS” stands for “association,” for a common occurrence. However, every introductory course in statistics teaches you that this does not prove a cause-and-effect relationship. For example, in some rural regions there are more storks and the birth rate is higher. However, this is because both storks and people with many children prefer a rural environment to cities—and does not prove that storks bring the babies.

This is also shown by the fact that the contribution of genes found in such studies decreases when the researchers include more psychosocial factors. In a corresponding study led by Jim van Os, a long-standing schizophrenia expert at the University of Utrecht in the Netherlands, differences in genes only explained 3 percent of the symptoms. In contrast, social circumstances and the environment played a much greater role, accounting for 30 and 24 percent respectively.

Interim conclusion: The GWAS that are so often praised today do not confirm any major influence of genes, in particular no cause-and-effect relationship, and are dependent on the inclusion of other factors.

Twin studies: Pragmatism and mistakes

Twin studies are cited as further evidence of the importance of genes for psychiatry. After all, we know that monozygotic (MZ) twins are almost 100 percent identical, whereas fraternal or dizygotic (DZ) twins are only about 50 percent identical. By comparing how often a phenotype such as schizophrenia occurs in monozygotic and dizygotic twins, it is possible to estimate heritability as opposed to environmental influences. This leads to the aforementioned claims such as “80 percent heritable.”

However, it is rarely explained that twin studies use statistical models whose assumptions are controversial. For example, it is not taken into account that identical twins are treated differently by their environment than fraternal twins due to their strong similarity. In addition, errors occur during data collection. The clinical psychologist Jay Joseph has published several books and articles here on Mad in America about the limitations of twin studies.

Many researchers pragmatically ignore such limitations and errors. The twin studies deliver the results that fit the zeitgeist: The aforementioned problems lead to an overestimation of genetic and underestimation of environmental influences. Incidentally, the large differences between the results of the GWAS and the twin studies are now explained by the fact that there must be a “hidden heritability.”

Instead of repeating the criticism that Joseph and other researchers have already formulated in more detail, I would like to discuss a few specific examples here. The so-called concordance rates are easier to understand than the heritability estimates, even without in-depth mathematical knowledge. These describe how often the second twin has a diagnosis such as schizophrenia if the first twin already has it.

Get real: concordance rates

A large Danish study investigated the heritability of schizophrenia. Data from around 32,000 pairs of twins born between 1951 and 2000 were analyzed. From this, the researchers calculated a heritability of 80 percent and interpreted this as a sign of “a substantial genetic risk” of this psychological-psychiatric disorder.

Let’s imagine for a thought experiment that you meet one of two identical twins and find out that he has been diagnosed with schizophrenia. You have just read that the heritability of this is 80 percent. Later you also meet the other twin. In your opinion, what is the probability that this twin will have the same diagnosis? (What is the concordance rate?)

Laypeople (but also many scientists) would probably now think that the second twin was also diagnosed with schizophrenia in 80 percent of cases. However, that is incorrect: according to the study, the concordance rate for schizophrenia is only 33 percent for MZ twins and only 7 percent for DZ twins. Mind you, the former are genetically more or less identical, while the latter are still around 50 percent identical! This should be taken into account when people with psychiatric diagnoses are advised about their family planning in genetic consultations.

Causes: Biological or psychological?

What has been shown here using the example of schizophrenia applies in a similar way to all mental disorders. It should be borne in mind that schizophrenia is already considered one of the most biological disorders, as can be seen in the following illustration.

Illustration: Biology or psychology, which is considered the cause of mental disorders? According to the study by Ahn and colleagues, experts considered the biological component to be particularly high in schizophrenia, autism and Alzheimer’s disease (blue), while they assumed a particularly high psychological component in bereavement, social phobia and post-traumatic stress (red). Image from Schleim (2023), CC BY 4.0.

As a final objection to heritability estimates, I will explain that they themselves depend on the environment.

Understanding heritability

While the linguistic meaning of “heritability” clearly suggests a genetic contribution, this is unfortunately not necessarily the case. To illustrate this, we must understand that researchers try to explain differences in the phenotype due to differences in the genotype.

For example, are taller people or people with more severe depressive symptoms more likely to have genotype XYZ than ZYX? If so, then these differences in phenotype are said to be explained by differences in genotype. Results from large GWAS can be seen in the table above, in the column for the explained variance.

The misunderstanding of heritability estimates is often explained using examples such as the following: Let’s assume a farmer has a wheat field. She sows the wheat, fertilizes the soil and irrigates it. However, due to a period of drought, the farmer’s water supply runs out after a while.

She decides to use the remaining water only for the left half of the field. This will then produce the full yield. The farmer expects a better harvest than if she irrigates both halves insufficiently. The wheat on the right half will then wither, as shown in the following illustration.

The important thing now is that we have the same environment on the left and right halves, apart from the irrigation. Within the left-hand field, there will be differences in the growth of the plants, which must then be (almost) exclusively genetic—the environment is identical. This means that the existing differences are explained genetically, the heritability is high.

The wheat on the right is stunted, but here too there are small differences in growth. These are also (almost) only due to genetic differences. So heritability is also high here. But if we compare the left and right fields, we see huge differences. Despite the high heritability, this is due to irrigation, an environmental characteristic.

In general, it can be said that the more uniform the environment, the greater the estimated heritability. This is because the differences that then still exist must be genetically determined. You can also imagine this in humans: If, for example, everyone is equally poorly or equally well nourished, the differences in body size will be mainly due to genes and the calculated heritability will be high. However, if there are greater differences in nutrition, part of the differences in size will be explained by the environment and part by genetics. The heritability is then lower.

In a study led by the aforementioned psychologist and geneticist Turkheimer, such an effect was demonstrated in reality: For children from poorer backgrounds, differences in IQ were explained primarily by environmental effects, whereas for the wealthy they were more strongly explained by genes. This could mean that the latter already lived in an environment that was ideal for intelligence and that the differences that then still existed were due more to genetic differences—as with the wheat on the left-hand side of the field in the example.

Interim conclusion: Heritability estimates exaggerate the genetic influence due to incorrect assumptions and errors in data collection. The true concordance rates suggest a much smaller genetic contribution, in line with the newer GWAS results. In particular, heritability estimates cannot be applied to individuals and are only meaningful for a specific population in a specific environment. They do not allow us to infer the relationship between genotype and phenotype in another population in a different environment.

Old knowledge

What I have just explained is old knowledge of genetics. The frequent misinterpretation of heritability estimates in psychiatry and the media is based on a linguistic misunderstanding. A few years ago, two researchers even described heritability as “one of the most misleading [terms] in the history of science” in an essay worth reading and warned that “continued use of the term does enormous damage to the public understanding of how human beings develop their individual traits and identities.” The misunderstanding plays into the hands of those who want to research mental disorders genetically.

As early as the 1970s, Stanford biologist Marcus Feldman and Richard Lewontin from Harvard University criticized the concept of heritability in human genetics in a much-cited publication in Science. They concluded that this approach is like trying to explain the mechanism of a clock by observing the hands and their ticking.

In the 1990s, philosophy and psychology professor Ned Block of New York University discussed the widespread misconceptions about heritability and the danger of racist conclusions. The reaction to the then much-discussed book The Bell Curve, which claimed that the heritability of IQ in white people was 60 percent and that black people had poorer genes, was also discussed in the media at the time.

Perhaps such thoughts prompted dozens of renowned ADHD researchers not to quantify the genetic contribution in a widely cited consensus paper. In particular, they do not report a heritability estimate. And rightly so! On the website of his ADHD Evidence Project, Stephen Faraone, one of the authors of the consensus paper and professor at SUNY Upstate Medical University, unfortunately informs the public differently: “Genetics contributes substantially to ADHD, with a heritability of approximately 80%.” We now know that this statement is misleading.

But if it is not possible to calculate the genetic influence in this way, how can it be done? It is generally recognized that many characteristics are based on a gene-environment interaction, including in psychology and psychiatry. Here, however, we are dealing with circular causality: gene, person and environment interact endlessly.

Genes influence how we feel, think and behave, yes; they partly explain why one person has a greater tendency to be impulsive or depressed than another. However, the environment and our experiences in it have a continuous effect on our body—and this in turn regulates gene activity, as is currently being researched in epigenetics. We have to say goodbye to the idea that we are dealing with mathematics in the form 2 + 2 = 4.

In order to obtain reliable results on the genetic influence, we would need experimental interventions: Individual genes or gene segments would have to be switched on and off or reprogrammed. The consequences for people’s health would be dramatic. In addition, the test subjects would have to be kept in controlled environments, like in a zoo. This is all forbidden for ethical reasons. It is done in animal experiments, but their results only apply to humans to a limited extent. In fact, heritability estimates only make sense in animal and plant breeding.

And even genes strongly determine a particular trait, they are not necessarily our fate. Think of rare monogenic diseases such as Huntington’s or spinocerebellar ataxia: If you have the wrong genotype, you are (almost) 100 percent likely to become ill. At the same time, however, this also defines a therapeutic goal.

This means that even in such a case of strongest genetic determination, social factors such as the availability of medical help can be decisive. This brings me to the last point, the social consequences of misunderstood research.

Social consequences

Until the 20th century, the concept of heritability was misused to justify the prevailing social conditions: Those in the upper classes simply had the better genes. Social Darwinism, inequality and colonial oppression were thus presented as a natural order. Even today, researchers criticize “biological essentialism,” which can lead to the justification of exclusion and racism.

This is how misunderstood biology becomes politics: For example, if intelligence were strongly genetically determined, couldn’t education spending be cut? But, as we have seen, even then the remaining percentages could still make a big difference. You could just as well argue the opposite and provide more social support for the naturally disadvantaged in the interests of equal opportunities.

Regarding mental disorders, pharmaceutical treatments appear better if genetic causes are assumed. However, according to current knowledge—apart from rare individual exceptions—this only involves slightly increased risks. Nevertheless, the discourse on diseased brains and genes threatens to obscure the significance of social and environmental influences such as poverty, serious life events, migration, stress and trauma.

The paradigm of biological psychiatry mentioned at the beginning, with its image of disorders as brain diseases, is so persistent because the three basic assumptions each have a true core: Mental processes are embodied and can therefore be influenced pharmacologically or electrically. The risk of a mental disorder is genetic and heritable in a weak but not exactly quantifiable sense. As I once put it myself: Mental disorders are brain disorders—and they are not.

We now know that the many billions of research dollars that have been invested in biological psychiatry since the 1980s have achieved little: There are no reliable biological tests, new revolutionary therapies have failed to materialize and the vaccinations against mental disorders once promised by the former director of the National Institute of Mental Health, Thomas Insel, for the year 2020 are not even on the horizon. At the same time, many countries are facing an unprecedented mental health crisis.

My point of view is that the disorders are only biological in a weak sense. Prevention and therapy should focus primarily on the psychosocial sphere. This is not only a scientific, but essentially a socio-political task. Stopping the spread of misleading heritability estimates would be a step in this direction.