Fixed Personality Traits, Randomly Arrived At
By Arnold Kling
- … variation in how brain circuits develop makes a major contribution to our psychological traits. Crucially, that variation can arise from genetic differences, but also from the processes of development themselves. Stochastic developmental variation will play a large role in determining how the effects of genetic variation are played out in each of us, and will also contribute to innate differences in temperament.
- —Kevin Mitchell, Innate1
If you could corner a professor of neuropsychology at a cocktail party, you probably would want to ask questions like these:
- 1. What sort of progress is likely in the prevention and treatment of mental disorders?
- 2. What is the appropriate perspective to take concerning the controversies pertaining to race and gender?
- 3. What sorts of constraints does human nature place on public policy?
As a researcher, Kevin Mitchell has studied the first question, and that is the primary focus of Innate: How the Wiring of Our Brains Shapes Who We Are. But in that book, he also offers his opinion on the second question, and one can draw inferences concerning the third. I will focus this essay primarily on the second and third questions.
I took away two main lessons from Innate. The bumper-sticker versions of these lessons are:
- a) gestation matters; and
- b) humanity is a set of individual mental disorders.
When we discuss “nature vs. nurture,” we typically think of DNA as representing “nature” and the environment that a person inhabits, especially during childhood, as “nurture.” But Mitchell draws our attention to the process of gestation in utero as a source of differences in human traits. Gestation is a complex, dynamic process of forming cells of various types and organizing them into particular locations in the brain and other organs.
This process is set in motion by our DNA at conception. But once cell division gets underway, many alternative paths are possible, as mutations and other random occurrences take place during gestation. As a result, the relationship of the newborn baby to its DNA is probabilistic rather than determinate.
Because of gestational variation, identical twins are not truly identical. Mitchell points out that if you had been cloned one hundred times, the clones’ brains would all turn out slightly differently from one another.
A metaphor that comes to my mind is the performance of a play, such as Romeo and Juliet. Shakespeare’s script is the DNA. But what the audience observes will depend on the approach taken by the director and the rest of the theater troupe, which we may think of as analogous to gestation. Perhaps the troupe will choose to stage the play using costumes from the Renaissance, or perhaps instead they will act in modern dress. Perhaps Romeo and Juliet will be portrayed as adolescents seized by puppy love, or perhaps instead they will be made to seem wise beyond their years. Any two performances that follow Shakespeare’s script are like identical twins, alike in many respects because of the shared DNA, but different in other respects because they came from different processes of gestation.
Extending this metaphor, we might say that a performance of West Side Story is the fraternal twin of a performance of Romeo and Juliet. The two performances do not have identical DNA, but their DNA comes from the same source, in that they both follow Shakespeare’s plot line. As a result, an audience is likely to sense more similarity between these two performances than would be found in two other theatrical performances chosen at random from the set of all plays ever performed.
This has significant implications for the nature-nurture issue. If only 50 percent of the variance in a trait can be accounted for by variation in DNA, then we should not presume that the other 50 percent comes from the environment. In fact, Mitchell argues that relatively little of the variation in human traits can be explained by the environment. Moreover, environmental influences can turn out to be transitory disturbances without lasting effects.
Instead, much of human variation comes from variation in gestation. In terms of our metaphor, most of the differences that audiences experience between the productions of Romeo and Juliet in two different theaters reflect differences in how the respective acting troupes wound up staging the play, not differences between the theater buildings themselves.
- When people look back on their lives and describe how they turned out, we often speak of formative experiences and other environmental influences. Mitchell questions whether we are correctly understanding the causal relationship.
But, actually, many experiences and environments are actively chosen by the individual. Variation in such experiences may indeed shape our psychological traits, as a mechanism of change, but the source of such variance may ultimately be the innate differences between people… An aptitude for music, for example, may lead people to pursue musical training, which will only serve to increase their musicality.
Thus, the “plasticity” of the brain may not imply that the environment molds people in a direction of similarity.
- Rather than overriding or flattening out their effects, processes of brain plasticity may reinforce and even exaggerate the widespread initial differences that arise due to both genetic and developmental variation… this kind of effect happens as people select and construct their own environments and experiences throughout their early life.
We also may tend to overestimate the influence of parental behavior. Some of the correlation between “parenting style” and the behavior of offspring may be due to common genetic determinants of personality. Moreover, “twin and adoption studies have demonstrated that much of the variation in how parents treat their children is indeed driven by the child’s own behavior and genotype.” Parents are more likely to continue to provide a child with dance lessons if the child shows an interest and an aptitude for dancing.
He suggests that brain plasticity actually leads people to counteract forces in the environment that would otherwise change their personalities.
- Evidence for the importance of this kind of mechanism comes from a striking and consistent finding from twin studies of intelligence: the heritability of this trait increases over time. When it is assessed in young children, about 50% of the variance is associated with genetic differences, while the shared family environment also makes a sizable contribution—30%-40%. However, when assessed in adults, the effect of shared family environment goes to zero, while the heritability increases to 80% or more.
As Mitchell points out, a caveat is in order. When we make the observation that environmental factors do not seem to affect human traits, much depends on what we define as the environment and what we define as a trait. As long as we stay within, say, the American middle class in the twenty-first century, we may find that environmental differences have small effects. But widening the range of environments would expose more impact from the environment. As for defining a trait, we might find that “facility with language” is less affected by the environment than, say, “ability to speak French.”
There are many scales used to measure and classify personality. Mitchell, along with other researchers, is partial to the Big Five factors of Extraversion, Neuroticism, Conscientiousness, Agreeableness, and Openness. One reason to rely on the Big Five is that when people are tested and re-tested for these factors, there is a correlation of .7, which is decent, albeit far from perfect. Another reason is that the mean values for these factors are similar across countries and cultures. Still, he cautions that attempts to find biological support for the Big Five system, meaning genetic or observable brain structure differences that correlate with those traits, have not been successful.
Studying the Big Five and other personality scales, researchers find strong evidence for heritability and little evidence for environmental influences.
- The findings from such studies are remarkably consistent and extremely well replicated… such traits are moderately heritable—on the order of 40-50% of the variation in these traits is attributable to genetic differences between people:
- … there is typically a negligible effect of the shared family environment. Being raised in the same family does not make people more similar in personality traits. Conversely, being raised in different families does not make them more different.
The key conclusion of all of the foregoing is that once we take note that brain development emerging during gestation is part of “nature,” it appears that nature accounts for much more of human differences than is generally thought. We have overestimated the effect of the environment.
Mental disorders and personality traits are only partially instilled at the time of conception, but by the time of birth they are well established. As Mitchell puts it:
- In sum, the way our individual brains get wired depends not just on our genetic makeup, but also on how the program of development happens to play out. This is a key point. It means that even if the variation in many of our traits is only partly genetic, this does not necessarily imply that the rest of the variation is environmental in origin or attributable to nurture—much of it may be developmental. Variation in our individual behavioral tendencies and capacities may be even more innate than genetic effects alone would suggest. [emphasis in original]
Mitchell is warning us not to attribute to the environment variation in psychological traits that he argues instead comes from the process of embryonic development. Hence my bumper sticker that gestation matters.
Humanity is a set of individual mental disorders
Mitchell informs us that human development never goes according to an exact plan.
- On average, we each have about 70 new mutations that were not present in our parents’ genomes. Because these occur at random, and because only ~3% of our DNA actually comprises genes, most of them won’t have any effect. The number of de novo mutations that actually hit a gene in any individual is around 1… But if you’re unlucky, that gene may be one of the several thousand absolutely required (in two working copies) for normal brain development or function…
- Recent data implicate one other important kind of mutation—ones that happen in the developing embryo itself… If these mutations disrupt development then they may result in a neurodevelopmental disease even if they are “mosaic,” or present in only some of the person’s cells.
Most mental disorders, such as schizophrenia or autism, are not “given” to people by their DNA. Instead, certain combinations of genes raise the risk that the disorder will emerge. If your identical twin is autistic, then the probability that you will also be autistic is 80 percent.
Even if there were such a thing as a perfectly normal brain, none of us would have it. Instead, as long as my particular set of mutations does not cause a clinically observable mental disorder, I will be termed “normal.” To put this another way, “normal” people are people whose brain structures are flawed in ways that do not result in mental illness as conventionally diagnosed.
In general, the causality from our DNA to psychiatric diagnosis runs as follows: Our DNA determines the probability that certain flaws in our brain structure will appear. Gestation determines which flaws actually do arise, most of which do not result in a clinically diagnosed mental illness. But there are various combinations of flaws that do lead to clinical observations that cause one to be diagnosed with epilepsy, autism, or what have you. In this scheme, DNA does not absolutely determine whether you will have, say autism, but it strongly affects the probability that your brain will develop in such a way that you will manifest this disorder.
Note that I speak of “flaws” as if there were a single optimal brain structure. That incorrectly presumes that we would agree on a value system that defines “optimal.” It also incorrectly assumes away the likely possibility that in a large population it may be optimal to have a variety of brain structures. For these reasons, I recognize that the use of the term “flaws” is clearly misleading, but nonetheless it provides a convenient shorthand.
Mitchell argues that at the level of DNA, genetic variation is more likely to be adverse than favorable. Genes that increase the likelihood that the individual will survive and reproduce will, over a long evolutionary time frame, spread to the entire population. But genes that modestly reduce survival and reproduction chances can persist in sub-populations. If we would prefer to have more good genes and fewer bad genes, then our best chances are with fewer mutations.
- Evolution has crafted a finely honed machine—the human brain—over hundreds of thousands of years. We have become complicit in this selection program, with the inventions of language and culture making it ever more beneficial to be even a small bit smarter. As a result, the space of possible mutations that would increase intelligence has likely already been exhaustively explored by natural selection. New mutations that did so would have been strongly selected for and most likely rapidly become fixed in the population, replacing the previous version of whatever gene they affected.
- … if a new mutation has any effect on intelligence at all, it is far more likely to reduce it than to increase it. It’s just much easier to mess up a complex system than to improve it.
- … there are lots of genetic variants in the population that can lower intelligence and each of us carries some burden of them. We all have some set of rare variants with possibly large effects, as well as a burden of more common variants with small individual effects, but a considerable collective influence.
Even the healthiest among us have survived in spite of DNA that includes some adverse mutations. Moreover, during gestation, each of us acquired more imperfections as further mutations, transcription errors, and other unfortunate events took place. The result is that each of us has a brain structure with its own unique weaknesses, most of which fall short of causing us to be labeled with a clinical diagnosis. Hence, my bumper sticker: Humanity is a set of individual mental disorders.
Any discussion of human brain development is bound to include the topic of intelligence. Mitchell offers this definition:
- At its core, intelligence is the ability to think in more and more abstract ways—to see a specific instance of something and draw larger lessons from it, which can then be applied to other situations, by analogy.
Elsewhere in my reading, I have come across educators using the term “far transfer,” which is the ability to take a concept that one has learned in one setting and apply it to a different situation. To educators, “far transfer” is the ultimate goal of teaching. Because far transfer is the indicator that a concept is truly understood, the professor is frustrated when students fail to demonstrate far transfer. By the same token, students are frustrated when they cannot meet expectations for far transfer. They may even complain that test questions requiring far transfer are “unfair,” because they present students with circumstances that they have not seen before.
If Mitchell’s characterization of intelligence is correct, and if intelligence is innate, then neither students nor professors can do much to affect the results of tests for far transfer. Instead, aptitude will be the determining factor. Intuitively, we realize this, which is why educators for the most part try to match students to the demands of a course rather than believing that it is possible to enable low-aptitude students to master challenging material.
Mitchell points out that the process for creating an intelligent human brain is complex and fragile.
- There are well over 500 specific genetic conditions that are known to cause substantial cognitive impairment, and more are being identified all the time…
- By itself, the existence of these conditions clearly shows that human intelligence is “genetic,” in the sense that it relies on a complex program encoded in our genomes that can be seriously affected by mutations that compromise that program. The fact that there are so many ways to disrupt that program shows just how complex it is—severe mutation of any one of hundreds of different genes is sufficient to drastically derail brain development or function.
Mitchell writes that although the genetic underpinnings of intelligence are not at all well mapped out, “The genetic and neurobiological findings thus paint a consistent picture—intelligence reflects how well the brain is put together, how robust the genetic program of neural development was, and how efficient the resultant neural networks are.” But if there is strong evidence that intelligence is genetic, then there is also important evidence of environmental influences. Most notably, there is the Flynn Effect, which is the observation that average scores on IQ tests have gone up over the past hundred years or so. Before political scientist James Flynn called attention to it, this improvement had gone unnoticed because of the practice of “re-norming.” IQ testers are focused on relative scores, comparing different people at a single point in time. For this purpose, they find it convenient to assign a score of 100 to the mean score at a point in time. As the population gradually performs better at the underlying tests, researchers have been “re-norming” scores, meaning that on any given test someone has to get more questions right to achieve an IQ of 100 than was required fifty years ago.
Mitchell points out:
- The one thing that is clearly not an explanation of the Flynn effect is changing genetics. There have simply not been enough generations in the time span over which these increases in IQ have been observed. The differences are definitely environmental in nature.
Mitchell notes that although some forms of social progress may serve to increase average IQ, they may amplify the inequality that arises from innate IQ differences. In particular, he writes that:
- … while increasing access to education benefits everyone, it may not do so evenly. Those with higher initial IQ may benefit more from education—they may learn more readily and be able to apply that knowledge more productively…. This means that while more educational opportunities will increase everyone’s intelligence, those who start at the higher end may benefit the most. Rather than simply shifting the whole distribution upward, greater education may actually exaggerate initial differences.
To recap, the brain structures that are shaped partly by genetics and ultimately emerge during gestation are what give us our personality traits, including IQ. Again, if you were cloned one hundred times, each time the gestation process would differ somewhat, leading to different flaws, mostly minor ones, in your brain structure. As a result, each clone could have somewhat different personality traits, including different levels of intelligence.
Mitchell believes that, on average, there are structural differences between female brains and male brains. He provides a variety of evidence that more than cultural factors are at work in shaping the two genders.
He points out that there exist physical sex differences in all higher mammals, including our pre-human ancestors.
- It is thus not only plausible that such innate sex differences would exist in humans (and demonstrable that they do), it is completely implausible that they would not. It would take a particularly virulent form of human exceptionalism to expect that we should differ from every other species of mammal in this regard.
Turning to brain anatomy, he writes:
- There are also sex differences in the number and density of connections between different brain areas. These are independent of the differences in cell number and reflect additional effects of hormones on genes that promote the growth of nerve fibers between various structures. The brains of male and female mammals are thus literally wired differently.
Images of human brains reveal many features where males and females differ.. But these differences are small enough that any individual male can have a feature that is more often found in females, and conversely. For example, even though on average male brains are 10 percent larger than female brains, at the individual level one can find large female brains and small male brains, and one cannot be sure that the brain of any particular male will be larger than that of some particular female.
But if one looks collectively at all of the brain features that vary by gender, one can with a high probability of success infer from a brain scan the gender of the person scanned.
- A “multivariate” classifier that considered the size of all 10 brain regions at once from the same brain scan data was able to classify males and females with over 90 percent accuracy. There thus clearly are male and female brains in the same way that there are male and female faces.
That is, although there is no one facial feature that clearly distinguishes human males from females, if we use a set of features we can classify with high accuracy. The same holds for brain images.
Mitchell examines the possibility that these brain structure differences are culturally determined.
- Perhaps those differences arise over time through the mechanisms of brain plasticity, due to the fact that males and females have different experiences and are treated differently by their parents, their peers, and society in general. In isolation, that is at least conceivable, though we have seen that adult brain structure in general is actually highly heritable, reflecting continued, strong genetic influences on growth and maturation after birth. Moreover, the fact that similar differences are seen in every other mammalian species and drive evolutionarily important behaviors makes the argument that they are purely culturally driven in humans much less likely. In essence, it demands two extra things: first, our evolutionary heritage of sex differences as mammals and primates would have to have been wiped clean, despite those differences being highly adaptive, and second, cultural practices would have to have arisen that effectively re-create the same outcome.
To me, this suggests some difficulties with the agenda of the social justice movement. Given the demonstrable statistical differences in brain structure, it would be unlikely that men and women would turn out to be equally represented in every segment of society. But the social justice movement insists that every high-status position in society should have at least as many women as men. That movement demands that we act as if any shortfall in the number of female mathematics professors or software developers can only reflect discrimination and must be remedied by policies to “improve inclusion.”
Unfortunately, I must note that trying to argue against the social justice movement’s presumption of injustice by invoking the sorts of scientific observations included in Innate is a hazardous enterprise. Doing so cost Lawrence Summers his job as President of Harvard and it cost James Damore his job as a software developer at Google.
Mitchell suggests that variation in sexual orientation can be explained by the mechanism of DNA determining probability distributions and gestation determining actual outcomes.
- It is perfectly understandable that a trait like sexual orientation could be only partly genetic, but still completely innate. A person’s genotype confers a certain probability of one outcome or the other, but the actual outcome that arises in the individual depends on how development happens to play out.
Thus, it could be that sexual orientation is only partly heritable but almost completely innate. But he argues against the notion that sexual orientation is fluid in the manner originally posited by Alfred Kinsey. According to Mitchell, “More recent work argues strongly against this view and indicates instead that sexual preference is much more categorical, for both heterosexuals and homosexuals.” This leads me to think that the currently fashionable viewpoint that gender is fluid deserves to be treated with skepticism.
While Mitchell is implicitly questioning the social justice movement concerning gender, he seems to side with the social justice movement concerning race. He writes, ‘Given the historical and continuing inequities between racial groups in the United States and across the world, it would seem appropriate to exhaust the possible contributions of these cultural factors before inferring any contribution from genetic differences.” This strikes me as an odd position to take. I doubt that he would advocate this research strategy to someone seeking to explain average differences in long-distance running ability.
I would think that one could compare the impact on children’s IQ of parents’ IQ and parents’ cultural distress, with the latter proxied by occupation and other indicators of socioeconomic status. One could then assess empirically if parents’ IQ accounts for much more variation than parents’ stress, or vice-versa. But while Mitchell’s entire book looks at the nature-nurture issue by carefully examining populations that differ along some dimensions but not others, he is all but ruling out applying an empirical strategy when race is involved. If we were to adapt his stance on race to the nature-nurture issue in general, we might say, “Given that we know that children raised in different families experience different environments, it would be appropriate to exhaust all possible contributions of these environmental differences before inferring any contribution of genetic differences.” That would shut down much of neuropsychology altogether.
To bolster his case for presuming that racial differences in IQ are not genetic, Mitchell points out that in the 1970s in Ireland, average IQ scores were 85, but by the mid-1990s they were 95, and they now average 100. “Nothing changed genetically over that time—better circumstances just allowed that latent potential to flourish.”
This made me curious as to what accounts for such a rapid improvement in average IQ. Is it possible that although DNA did not change, the improvement was not due to the effects of the postnatal environment on children? Instead, could the vast difference have emerged during gestation? For example, suppose that extremely poor prenatal nutrition increases the likelihood of adverse features in brain development. Can one document substantial prevalence of malnourishment in the mothers of the population sampled in the 1970s, and can we point to better nourishment in the mothers of the population sampled in the more recent time periods?
Alternatively, perhaps there was a change in the mix of children born, because of the demographic transition. All over the world, as women attain greater affluence and education, they have fewer children. If the demographic transition for high-IQ women took place in Ireland relatively early in the 20th century but it took place later in the 20th-century for low-IQ women there, then perhaps the population in Ireland in the 1970s included a temporary bulge in the proportion of children born to low-IQ women. This would cause the average IQ score to drop, but that average would recover once low-IQ women started having fewer children.
In any case, the Irish example seems like an exceptional event. We have observed some dramatic improvements in economic conditions among other groups. Do we observe such dramatic improvements in IQ scores also in those cases?
Raising another issue with looking at average IQ by race, Mitchell writes,
- Most of the discussion in this area centers on the colloquial idea of “races,” but exactly how many such categories exist and how they are defined are hard to agree on… In terms of shared ancestry, thousands of such groups can be defined across all areas of the globe. Some will be reasonably discrete, based on a history of isolation and restricted breeding, while others are much more mixed reflecting more extensive migration and interbreeding.
- Modern genetics can reveal much of this history and clearly illustrates the complexities of humanity’s global family tree. If you cluster people based on genetic similarity, you can indeed derive several major categories, but you can also just as well go to deeper levels and reveal many, many more. There is no reason to think that any one level should have privileged status—none of these groupings reflects a natural kind, in the way that sex does.
Accordingly, one can argue that the usual racial classifications are based on what I might term Folk Anthropology. It is the way Americans (and many others) have grown accustomed to thinking about race. We could instead adopt a much more nuanced classification system using cluster analysis of DNA databases. We might call this Data-driven Anthropology.
But if Data-driven Anthropology would call into question or invalidate studies that show differences in IQ based on Folk Anthropology ethnic classifications, it would do much more than that. Many well-established surveys utilize Folk Anthropology, as do college applications and government regulators looking for evidence of discrimination by banks and other institutions. The U.S. Census has long used the categories of Folk Anthropology, and a shift to Data-driven Anthropology could prove very challenging to implement.
Above all, promoting Data-driven Anthropology would undermine the social justice movement, which is heavily invested in Folk Anthropology. Diversity and inclusion policies are based on the categories of Folk Anthropology.
The alternative to the social justice movement’s drive to direct outcomes by race and gender is to treat people as individuals, letting the chips fall where they may. Clusters of differences in occupational prevalence or educational attainment could very well emerge, and some of these may produce disproportionate results by gender or by the categories of Folk Anthropology. But we need not presume that discrimination is the sole cause of disproportionate representation. Measured differences in the distribution of IQ and other psychological traits suggest otherwise.
The neuropsychology literature speaks of the “shared environment,” meaning the family, and the “non-shared environment,” meaning other institutions and cultural norms. In this terminology, public policy affects the non-shared environment.
The message of Innate is that human traits are not as malleable as we have believed. As the influence of gestation is recognized, the influence of the environment, including the non-shared environment, declines. This predicts/explains widespread policy failure.
For example, policy makers place strong faith in education as a force that can affect citizens’ lives. Partisans of every political persuasion claim that we need a better education system. They call for educational reforms.
But careful studies of educational interventions almost always confirm what I have called The Null Hypothesis. When an intervention is tested using a treatment sample and a control sample randomly chosen from the same pool of students, when the effects are measured over a long time period that allows for the “fade-out” of transitory improvements, and when attempts are made to replicate positive results in subsequent studies, the failure rate of interventions is essentially 100 percent.
Indeed, there is empirical support for a Null Hypothesis writ large. Mitchell writes,
- Many studies have looked for systematic associations between specific environmental factors or experiences that differ between siblings and specific behavioral outcomes. These typically fall under a number of categories including differential parenting, peer relationships, sibling interaction, teacher relationships, and what is known as “family constellation” (birth order, age difference between siblings, whether or not they are the same gender, etc.). The results from these studies are very clear. They have failed to identify any robust, consistent, or substantial effects on any of a variety of outcomes including adjustment, personality measures, or cognitive ability.
Policies that assume that we can directly manipulate human traits are likely to fail. Policies that change behavior by changing incentives are likely to have unintended consequences that could exacerbate inequality.
As an example, consider the objective of increasing personal saving. People who are impulsive are likely to have difficulty saving. But any attempt to try to change individual temperaments to make them less impulsive is likely to have no substantial permanent effect. The Null Hypothesis tends to hold for attempts to change temperament.
On the other hand, a policy to change the incentive to save, by offering tax-advantaged savings accounts, is likely to achieve some behavioral changes. But I would speculate that the people who are most likely to respond to the tax incentive are people who already have at least some propensity to save. People who are very high in impulsivity will not have the temperament to take advantage of the tax incentives. Thus, while an incentives-oriented policy could work to increase overall savings, it might have the unintended consequence of exacerbating inequality over time.
Neuropsychology is a dynamic field. Those of us with an interest in public policy should try to keep up with it. Innate is the most credible, accessible introduction that I have come across.
 Kevin Mitchell, Innate: How the Wiring of our Brains Shapes Who We Are. Princeton University Press, 2018.
*Arnold Kling has a Ph.D. in economics from the Massachusetts Institute of Technology. He is the author of several books, including Crisis of Abundance: Rethinking How We Pay for Health Care; Invisible Wealth: The Hidden Story of How Markets Work; Unchecked and Unbalanced: How the Discrepancy Between Knowledge and Power Caused the Financial Crisis and Threatens Democracy; and Specialization and Trade: A Re-introduction to Economics. He contributed to EconLog from January 2003 through August 2012.
Read more of what Arnold Kling’s been reading. For more book reviews and articles by Arnold Kling, see the Archive.