Unique

ONE

It Runs in the Family

IN 1952, DIMITRI BELYAEV, A RUSSIAN GENETICIST, had an idea for a creative and audacious experiment. He was interested in the domestication of animals that had been important to human civilization, such as dogs, pigs, horses, sheep, and cattle. Dogs are thought to be the first domesticated species, derived from Eurasian gray wolves by hunter-gatherers over fifteen thousand years ago.¹ Belyaev wanted to know how some wild wolves, which are famously averse to human contact and occasionally aggressive, evolved into the affectionate and loyal companions we know and love. Why, as first described by Charles Darwin, did domesticated mammals often share certain physical characteristics—like rounder, more juvenile-appearing faces, floppier ears, curvier tails, and patches of lighter fur or hide—in contrast to their wild forebears? And why did most wild mammals have a single, brief breeding season every year, but their domesticated counterparts could often breed twice or more per year?

Belyaev believed that the single most important trait selected for in the initial process of domestication was not size or breeding capacity but tameness. He hypothesized that the defining characteristic of all the animal species domesticated by our ancestors was a reduction in aggression toward, and fear of, humans. To test his theory, he went to some of the industrial-scale silver fox farms that had been established for fur production in the Soviet Union and instructed the animal breeders there to select only the tamest foxes, a tiny fraction of the total, and breed them together. He believed that by repeatedly selecting for tameness over many generations, he could ultimately approximate wolf-to-dog domestication and produce a friendly, loyal, doglike fox.

In carrying out these experiments, Belyaev hoped to avoid the fate of his beloved older brother Nikolai, who, in 1937, had been executed by the Soviet government for the crime of performing and publishing genetic experiments. Those were dark days for Soviet biology. Stalin’s Communist government, eager to elevate an uneducated “common man” to a position of authority in the scientific leadership, promoted the charlatan Trofim Lysenko to director of the Institute of Genetics at the Soviet Academy of Sciences.

Lysenko faked his data to show that wheat and barley seeds that had been frozen before planting yielded larger crops when planted in winter, and that the second generation of seeds derived from those crops also acquired enhanced winter growth. He claimed that this method could double food production in the USSR and feed the masses, an assertion extolled in the state-controlled newspaper Pravda as a triumph of Soviet science. His seed-freezing techniques were widely adopted in the country but failed utterly, contributing to periods of widespread starvation. Lysenko rejected genetics, a discipline that had thrived in Russia before his rise to power, because simple genetic experiments could disprove his claims. He called Soviet geneticists “Western saboteurs” and, with Stalin’s backing, sought to dismantle the discipline. Those who resisted were fired and even imprisoned. The strongest supporters of genetics, like Nikolai Belyaev and the great Russian plant geneticist Nikolai Vavilov, were executed—Belyaev with a rifle, and Vavilov by slow starvation in a prison cell.

Dimitri Belyaev was fortunate to have some political support for his work. A decorated hero of the Russian Army during World War II, he had presided over improvements in the farming of wild foxes, sable, and mink for fur production. This effort was key to the Soviet economy because it brought in large amounts of foreign currency. Mindful of his brother’s fate, Belyaev conducted his domestication experiments on remote fox farms, far from the prying eyes of Moscow—first in the forests of Estonia and later in a distant part of Siberia near the Mongolian border. The cover story was that he was studying fox physiology, not genetics. To oversee the endeavor, Belyaev recruited the young scientist Lyudmila Trut, an expert in animal behavior who had been trained at the elite Moscow State University. He gave her explicit instructions: when selecting foxes for breeding, the only trait to be considered was tameness—not appearance, nor size, nor behavior toward other foxes.

There was no guarantee that this fox domestication plan would work. Nonetheless, it was a reasonable supposition. After all, dogs were domesticated from wolves, which are closely related to foxes. Yet previous attempts to domesticate wild zebras—which are so closely related to horses that the two species can sometimes be interbred (a Shetland pony-zebra cross is called a zony)—had repeatedly failed.² The reason appears to be that there is not enough genetic variation underlying the trait of tameness in zebras. You can’t effectively pick the tamest zebras for breeding if there aren’t any slightly tame zebras to start with. Fortunately, that wasn’t the case with Trut and Belyaev’s foxes.

FIGURE 1. Dr. Lyudmila Trut with one of her domesticated foxes. Used with permission of the BBC. Photo by Dan Child.

When Lyudmila Trut first slowly introduced her hand into the fox cage, she wore a thick padded glove and held a small stick. The most common reaction to this gentle intrusion was snarling and biting. Other foxes cowered, highly agitated, in the rear of the cage. But about 10 percent of the foxes stayed calm throughout, observing her intently but not approaching.³ These were the animals that she selected for the first round of breeding. Trut was also careful not to breed closely related foxes and thereby introduce inbreeding artifacts that could confound the experiment. To increase the probability that the observed tameness resulted purely from genetic selection, the foxes were not trained and their interactions with humans were strictly limited.

Trut’s initial finding, that there was some partial tameness to serve as a basis for subsequent breeding, was encouraging. But the experiment could still easily fail in a different way: it simply might take too many generations to see any significant changes in fox behavior. It has been suggested, from analysis of the archeological record, that wolf-to-dog domestication proceeded in fits and starts, beginning thousands of years ago. Trut and Belyaev didn’t have that much time and were limited by the slow pace of fox breeding: one mating season per year. So it was cause for joy when, only four years into the experiment, clear behavioral changes emerged. A few of the fourth-generation foxes showed no aggression or fear, and even displayed doglike tail wagging in response to humans. By the sixth generation, some of the fox pups exhibited whining, licking, and whimpering behavior as they eagerly sought human attention. Today, over 80 percent of the adult foxes derived from these crosses are as loyal and tame as any domesticated dog (figure 1).⁴

If you wish, you can go on the internet and obtain your own tame fox from Trut and Belyaev’s experiment, delivered from Siberia to you for $9,000, shipping included.⁵ But be aware that, while domesticated foxes are much friendlier than those in the wild, they are much harder to train than dogs. “[You can be] sitting there drinking your cup of coffee and turning your head for a second, and then taking a swig and realizing, ‘Yeah, Boris came up here and peed in my coffee cup,’” said domesticated fox expert Amy Bassett. “You can easily train and manage behavioral problems in dogs, but there are a lot of behaviors in foxes… that you will never be able to manage.”⁶

THE ORIGINAL FARMED SILVER foxes looked like wild foxes: they had erect ears, low-slung tails, and uniformly silver-black fur, save for a white tail tip. As breeding for tameness continued through the generations, the foxes often developed floppy ears, shorter, curved tails, and patchy, pale fur, particularly on the face. They reached sexual maturity earlier than wild foxes, and some even bred twice per year. It is important to emphasize that the only criterion used for breeding was tameness; the other physical traits just came along for the ride. The remarkable thing is that these particular bodily changes have emerged in many other domesticated animals—from cattle to pigs to rabbits—at various times in history.

When Trut and Belyaev measured the levels of resting stress hormones produced by the adrenal glands, they found significant reductions in the tame foxes. They also found that levels of the neurotransmitter serotonin and its metabolites were increased in the brains of the tame foxes, which is consistent with a reduction in aggressive behavior. One overarching hypothesis for the biochemical, behavioral, and structural changes seen in domesticated foxes and other animals is that their development is somehow arrested at an earlier state than their wild cousins. Perhaps the variation in genes responsible for developmental timing is what gives rise to variation in tameness. When animals are bred for tameness, the other youthful traits noticed by Darwin—like floppy ears, round faces, and curly tails—follow along.

TRUT AND BELYAEV SHOWED that a behavioral trait (tameness) in foxes is heritable, that it can be changed by selective breeding in just a few generations, and that physical changes will accompany selection for this trait. Can these conclusions about the heritability of behavioral and physical traits from the fox taming experiment be usefully applied to us? After all, we humans are not confined to cages in Siberia. And, for the most part, we choose our own mates, rather than having them forced upon us by alien overlords. We even have OkCupid and Bumble to expand our mating possibilities.

Insights about the heritability of human traits can be gleaned from studies of twins. This type of analysis can be used to estimate the degree of variation in a trait that is heritable within a particular group of people (or foxes), from 0 to 100 percent. The key thing to remember about heritability is that it measures variation across an entire population, not individuals. Just because a particular trait is 70 percent heritable doesn’t mean that, for any individual from that population, genes are responsible for 70 percent and other factors for 30 percent.

Heritability estimates from twin studies may be used for both easily measured physical traits, like height or resting heart rate, and behavioral traits like shyness, generosity, or general intelligence, which are somewhat more subjective and harder to measure. One of the challenges with behavioral traits, which are typically measured by direct observation or with a survey, is that they are culturally constructed. The definition of and necessary criteria for the trait of shyness is probably different in Japan than it is in Italy. Concepts of generosity will not be identical for the city dwellers of Pakistan and the Hadza people of Tanzania. What this means is that the assessment of behavioral traits in individuals will be convolved with cultural factors if the individuals come from different cultural backgrounds, even if they live in the same location.

Here’s how heritability estimates work: Fraternal twins are conceived when two eggs are released during the same ovulatory cycle and each is fertilized by a separate sperm cell. The two fertilized eggs then develop separately into two embryos. Fraternal twins are as genetically similar to each other as any other pair of siblings. On average, they share 50 percent of their genes.⁷ Since fraternal twin embryos inherit their sex-determining X and Y chromosomes independently, fraternal twins are as likely to be the same sex (boy/boy or girl/girl) as the opposite sex (boy/girl or girl/boy).

By contrast, identical twins arise from a single fertilized egg that then divides to form two embryos early in development. Each twin inherits the same version of each gene from their parents, and so they are genetically identical. Because identical twin embryos also inherit the same arrangement of sex-determining X and Y chromosomes, they are always the same sex. This means that if you see mixed-sex twins, they must be fraternal, not identical.

In one simple twin study design, a particular trait, like height, is measured in members of a large number of fraternal and identical twin pairs. The difference in height is calculated for each twin pair, and then the results are compared between the fraternal and identical groups.⁸ One study of this type has shown, for example, that the average height difference between fraternal twins is 4.5 centimeters, whereas it is 1.7 centimeters for identical twins. A crucial assumption in these types of twin studies is that both twins (identical and fraternal) have been raised together, in the same household, at the same time, and will thereby have a highly shared social and physical environment, at least during childhood. Therefore, the smaller average difference between identical twins is attributed to their greater degree of genetic similarity. When these values are plugged into a standard equation, we can estimate the degree of heritability of a trait, which is about 85 percent for adult height, at least in affluent countries where basic nutrition needs are met. One can also estimate the degree of variation in height that is attributable to the twins’ shared environment, which is about 5 percent, and to the twins’ unshared environment, which is about 10 percent. Those interested in the calculation of these values are invited to check this endnote.⁹

For most twins, the shared environment is dominated by experiences in the family (both social, like being read to, and physical, like the particular foods on the dinner table) but can also include certain shared experiences at school and in the community, as well as the shared exposure to foods and infectious diseases. Unshared environment is a sort of grab bag for all of the other types of random experience, both social and biological, that individuals do not share. Importantly, this estimate of non-shared environment will also include the random nature of both fetal and postnatal development of the brain and body, which we shall explore in chapter 2.¹⁰

This type of twin analysis can be applied to any trait, not just those that are continuously variable and easily measured, such as height or weight. For example, it can be used to analyze responses to a survey question like “In the last year, have you ever felt sexual attraction to a member of your own sex?” If sexual attraction had no heritable component, we’d expect that the percentage of twin pairs where both answered yes would be roughly the same for identical and fraternal twins. Conversely, if sexual attraction were entirely heritable, then we’d expect that every homosexual/bisexual identical twin would have a homosexual/bisexual twin sibling (and every straight identical twin would have a straight twin sibling). It turns out that the best estimates to date (from a population of 3,826 randomly selected twin pairs in Sweden) are that, in men, about 40 percent of the variation in sexual orientation is heritable with no detectable effect of shared environment and 60 percent is attributable to unshared environment.¹¹ Forty percent is a significant fraction, but it still leaves room for plenty of other nonheritable factors. We’ll discuss the emerging science of sexual orientation and identity in chapter 4.

There have been critiques of these types of twin studies. Some researchers have claimed that studies comparing identical and fraternal twins raised together overestimate the heritable contribution to a trait because family members, friends, and teachers often treat identical twins more similarly than fraternal twins. This could come about in many ways, from the foods they are served to the ways in which people interact with them. Other researchers have claimed the opposite problem: they argue that since identical twins raised together seek to differentiate themselves socially from each other to a greater degree than fraternal twins, such a comparison underestimates the genetic contribution to a trait (particularly a behavioral one). In either case, the key assumption of equal shared environments between identical and fraternal twins would be violated. There have been passionate arguments for and against the validity of studies of twins raised together, and we won’t engage in a blow-by-blow recap of those brawls here. My own reading of the literature leads me to believe that, in most cases, the unequal shared environment problem in studies of twins raised together is small and rarely invalidates the general estimates of heritability that result.¹² Nonetheless, it would be best to have a twin study design that would cleanly estimate heritability without the muddled assumption of equal shared environments.

ON FEBRUARY 19, 1979, (at which point the tame-fox breeding experiment had been underway in the Soviet Union for over twenty-six years), the local newspaper in Lima, Ohio, reported a fun human-interest story about identical twin brothers who had been adopted by different families and raised completely apart, only to reunite at age thirty-nine. The twins were born in 1939 to a fifteen-year-old unwed mother, who immediately put them up for adoption. They were separated four weeks later, when one was adopted by Ernest and Sarah Springer, who brought him to their home in Piqua, Ohio. The second boy was adopted two weeks later by Jess and Lucille Lewis of Lima, Ohio, a town about forty-five miles away from Piqua. For reasons that have never been explained, both couples were told that their adoptive child had a twin who died at birth.¹³

But when Lucille Lewis was finalizing the legal adoption of her son, by then a toddler, a clerk at the county courthouse let the cat out of the bag. She told her, “They named the other little boy Jim, too.” In an interview with People magazine, Mrs. Lewis said, “I knew all those years that he had a brother, and I worried whether he had a home, and whether he was all right.” She waited until her son turned five before telling him about his twin. Jim Lewis couldn’t explain why, at the age of thirty-nine, he finally contacted the court to put him in touch with his brother. The Lima News reported that Jim Lewis called Jim Springer, took a deep breath, and asked, “Are you my brother?” At the other end of the line, Jim Springer answered, “Yep.” And so, the twins were reunited.¹⁴

When the Jim twins reunited, they were not mirror duplicates in either appearance (figure 2) or temperament. Nonetheless, a series of striking similarities emerged. Both brothers worked in law enforcement and enjoyed carpentry and drafting as hobbies. On vacations, they liked to drive their Chevrolets to Pass-a-Grille Beach in the Florida panhandle. In school, both had excelled in math and struggled with spelling. Both had married women named Linda, only to divorce and remarry women named Betty. Both had sons: James Alan Lewis and James Allan Springer. And, most tellingly, they preferred to wash their hands both before and after peeing.

It’s not surprising that these anecdotes were broadly appealing to readers and that the story of the Jim twins quickly made its way around the world. The day after the first story of their reunion appeared in the Lima News, it was reprinted in the Minneapolis Star Tribune, where it caught the eye of Meg Keyes, a psychology graduate student at the University of Minnesota. Keyes had recently taken a course with Professor Thomas Bouchard Jr. on individual behavioral differences. When she showed Bouchard the article, he immediately recognized how interesting it would be to study the Jim twins, and soon. He was quoted in the New York Times as saying, “[To study the Jim twins] I’m going to beg, borrow and steal and even use some of my own money if I have to. It is important to study them immediately because now that they have gotten together they are, in a sense, contaminating one another.”¹⁵

Bouchard quickly contacted the twins, who agreed to come to the University of Minnesota to spend six days undergoing a battery of psychological and medical tests and interviews. More stories of behavioral and physical similarities emerged. Both crossed their legs in the same way and suffered from chronic headaches and a heart condition. Both were described as “patient, kind, and serious.” Both had rapidly gained ten pounds at exactly the same age. These anecdotal similarities were tantalizing, but analysis of a single identical twin pair, even one as striking as the Jim twins, did not allow Bouchard to reach the holy grail: to estimate the heritability of traits without the potential confound of the equal environment assumption. That would require him to compare a sizeable population of identical twins with an equally sizeable population of fraternal twins raised apart.

FIGURE 2. Jim Springer and Jim Lewis pose for a photo shortly after being reunited in 1979. Photo courtesy of Nancy L. Segal and the Jim twins. Used with permission.

When the study of the Jim twins began, Bouchard assumed that they would be a one-off. Other researchers had tried to analyze twins raised apart but had access to so few twin pairs that their results were statistically weak. Bouchard imagined that he would have the same problem, that the cost of finding many twins raised apart would be prohibitive. What he didn’t count on was the public’s insatiable appetite for Jim twin stories. They appeared in newspapers, magazines, and on all the major television shows of the day. Some newly separated twin pairs emerged after the Jims appeared on The Tonight Show with Johnny Carson, others after seeing them on Dinah Shore.

This unprecedented publicity allowed Bouchard to found the Minnesota Study of Twins Reared Apart (MISTRA), which ran for twenty years and analyzed eighty-one identical and fifty-six same-sex fraternal twin pairs.¹⁶ In collaboration with fellow University of Minnesota psychologist David Lykken, the study also compared twins reared apart with twins reared together. MISTRA was a major advance in twin research. The largest and most productive investigation of this type, it produced good estimates of the heritable contribution to variance in many physical traits, like body mass index (about 75 percent) and resting heart rate (about 50 percent), and behavioral traits, like extraversion (about 50 percent) and schizophrenia (about 85 percent).

One main conclusion of MISTRA and related studies was that most human traits, regardless of whether they are physical or behavioral, have a significant heritable component, usually ranging from 30 to 80 percent. Rarely are traits either entirely heritable or entirely nonheritable (we’ll talk about some notable exceptions to this later). The other main conclusion is that variation in certain traits, like IQ, is weakly heritable (about 22 percent) when tested at age five but becomes strongly heritable once school is well underway at age twelve (about 70 percent), and then remains so across the lifespan. Correspondingly, the variation in IQ explained by the shared environment is about 55 percent at age five (when most experience has been within the family) but falls to undetectable levels by age twelve, at which time children have been exposed to a broader range of experiences.¹⁷ Those of you who are doing the arithmetic will notice that the variations accounted for by heritability and shared environment are not adding up to 100 percent. That difference is the aforementioned term “unshared environment,” which, in addition to unshared social experience, also includes the random process of development. More on this in chapter 2.

For decades, the dominant thinking in the field of psychology, and in society at large, was that the most important determinant of one’s adult personality was the influence of immediate family, particularly the parents. This idea came from the twentieth-century psychological movement called behaviorism, which held that humans come into the world as blank slates, ready to be molded by social experience. As a result, it was quite a shock when the MISTRA experiments showed significantly higher correlations in personality measures in identical twin pairs than in fraternal ones. The main result was that about 50 percent of the variation in personality can be accounted for by heritability. This held for all five major standard scales of personality (openness, conscientiousness, extraversion, agreeableness, and neuroticism; abbreviated as OCEAN) and directly contradicted the blank slate hypothesis of the behaviorists.

Most psychologists were guessing that the remaining 50 percent of the variation would be largely explained by social dynamics within the family. By comparing identical twins raised together with identical twins raised apart, the MISTRA researchers estimated the contribution of “shared environment” to individual personality—a factor that includes social experience in the family as well as things like shared nutrition and shared exposure to communicable diseases. To the psychologists’ surprise, shared environment made little or no contribution to variation in personality measures (typically less than 10 percent). It’s not just identical twin results that support the idea that shared environment plays a tiny role in explaining individual personalities. Fraternal twins who grow up together are no more similar in personality than those raised in different families, and unrelated adoptive siblings raised in the same family are barely alike at all.

The failure of shared environment to affect personality goes against some popular ideas about the influence of parents. But these twin study results don’t say that parental behavior is unimportant. Rather, they show that, beyond some minimum level of parental support and encouragement, extra attention doesn’t produce large effects on personality as measured by questionnaires administered in the lab.

Importantly, personality is not the totality of one’s character. Parents can inculcate work habits and teach specific skills, like weaving or car repair. And they can transmit philosophical, religious, or political opinions that are not measured by the OCEAN personality tests. For example, altruism, sharing, and other prosocial behaviors appear to be influenced by shared environments to a greater degree than other behavioral traits.¹⁸