The Anatomy of Violence

by Adrian Raine

  A third but more unusual study comes to the same conclusion: identical twins who were separated at birth are surprisingly similar with respect to antisocial personality, despite being reared in very different environments.

  These twin and adoption studies tell us that there is a significant genetic loading for aggression, but they do not tell us which specific genes are involved. So we’ll finally turn to research at the molecular level that is now beginning to unmask the mean genes giving rise to aggression.

  DOUBLE TROUBLE

  About 2 percent of us are twins. Almost all of these twins are fraternal, or dizygotic, twins, who have about 50 percent of their genetic material in common. They develop from two separate eggs that are fertilized by two separate sperm, and effectively they are just like normal brothers and sisters. Much rarer—only 8 percent of all twins—are identical, or monozygotic, twins. These twins have virtually 100 percent of their genes in common because they develop from a single egg-sperm pair-up—a zygote—that basically malfunctions and splits into two.2 Behavioral geneticists have used this malfunctioning twist of nature to examine genetic influences on antisocial and aggressive behavior. It’s the perfect natural experiment for exploring the extent to which any behavioral, physical, or psychological characteristic is influenced by genetics.

  Although I’ve mentioned that fraternal twins share on average 50 percent of their genes, I should qualify this. You actually have about 99 percent of your genes in common with me. Both of us share about 98 percent of our genes with chimpanzees, who themselves are genetically more similar to humans than they are to gorillas. Speaking of monkeys, we even have 60 percent of our genes in common with banana trees. So when we talk about fraternal twins having 50 percent of their genes in common, we are referring to 50 percent of just those small genetic differences that separate all human beings. Similarly, identical twins are not absolutely 100 percent genetically identical, but are 99 percent identical in that remaining 1 percent of genetic variation that differentiates all of us.

  How do people go about setting up a twin study in the first place? Laura Baker, my longtime colleague at the University of Southern California, brainstormed with me one lunchtime about a nifty study we could do together. She knew a lot about twins. I knew a fair bit about antisocial behavior in kids. So she thought we could do a twin study on child antisocial behavior. Once we got our grant funded by the National Institute of Mental Health, we set to work. While I was setting up our psychophysiology laboratory, Laura started recruiting the twins. She worked with the Los Angeles Unified School District and sent letters to all parents who had a nine-year-old child in a school in Southern California.3 We ended up getting 1,210 twins to participate. So then we were good to go.

  The caregiver and the twin pair would come in for a full day of assessment—cognitive, psychophysiological, personality, social, and behavioral testing. The parents, the kids, and their teachers would fill out checklists on behavior, including antisocial behaviors. Do they bully other kids? Do they steal? Are they cruel to animals? Do they get into fights? Do they physically attack others? Do they skip school? Do they set fires? All the things that are the hallmark of a troublesome kid and a budding offender-to-be. Now we had our measures of antisocial behavior in 1,210 children.

  So how do we work out if antisocial behavior in nine-year-olds is under genetic control? We look at how similar the identical twins are to each other, and compare that to how similar the fraternal twins are. Remember that identical twins are more genetically similar than fraternal twins. So if genes play some role in shaping antisocial behavior, you’d expect pairs of identical twins to be more similar in their level of antisocial behavior than fraternal twins. We use sophisticated statistical techniques—multivariate genetic analysis using structural equation modeling—to compute estimates of the heritability of this behavior.

  What did Laura and I find? Heritabilities that ranged from .40 to .50. That means that 40 to 50 percent of the variability among us in antisocial behavior is explained by genetics. It did not matter who rated the child’s behavior. If it was the teacher, the heritability was 40 percent. If it was the parent, it was 47 percent. If it was the kids themselves, heritability was 50 percent.4 So no matter who makes the assessment, about half of the variation in antisocial behavior among kids is under genetic control. Half of the answer to why some of us are antisocial while others are not is due to genetics.

  Our findings became even more dramatic when we combined our different measures of antisocial behavior. No measure is perfectly reliable. You know how parents, teachers, and kids can disagree on things. How can we derive a more reliable measure of antisocial behavior? By averaging the three informant sources to get a “common view” of what the child really does. When we did that, we found that 96 percent of the variance in this combined view of antisocial behavior is heritable. There is no contribution at all from the shared environment, and only a 4 percent contribution from the non-shared environment.5 Once we have a more reliable measure of antisocial behavior the genetic influence goes way up. We must be very cautious not to overestimate the importance of genetic factors, but all in all there is no question that antisocial behavior is heritable—and significantly so.6

  Twin studies also tell us that aggression and violence are heritable. In our study we measured reactive and proactive aggression. Reactive aggression is a case of someone hitting you, and you hitting them back—a sort of “defensive” or retaliatory aggression where you stand your ground. That form of aggression had a heritability of 38 percent. Proactive aggression, on the other hand, is meaner and crueler—you use force to get things from others. That had a somewhat higher heritability of 50 percent.7 Again, the influence of the shared environment was minimal for both forms of aggression, and indeed was even nonexistent for boys.

  Dozens of other twin studies have found the same effect in children, adolescents, and adults—males and females alike. Indeed, a meta-analysis of 103 studies compared heritability of aggressive behavior with rule-breaking, nonaggressive behavior.8 Nonaggressive antisocial behavior was 48 percent heritable, while aggressive behavior was 65 percent heritable. Yet again, shared environmental influences were small for nonaggressive antisocial behavior (18 percent) and minimal for aggressive behavior (5 percent). Genetics and the non-shared environmental influences rule the roost when it comes to aggression. We also know that genetic influences are strongest for criminal careers that start early, occur across many settings, are persistent and severe,9 and involve callous, unemotional symptoms like lack of remorse.10 This is exactly the form of antisocial behavior that later gives rise to adult violence.

  PLACING IDENTICAL PEAS INTO DIFFERENT PODS

  One problem with twin studies is something called the equal environments assumption. Identical twins may be treated more equally by parents, teachers, and even peers than fraternal twins are. So an argument can be made that, sure, identical twins may be more alike on antisocial behavior than fraternal twins. But it’s not because they are more genetically similar—it’s because they are more environmentally similar.

  This problem is circumvented in studies of identical twins reared apart. These are powerful studies for establishing heritability. Naturally they are very rare. However, one such study has been conducted on antisocial behavior in children and adults, consisting of thirty-two sets of monozygotic twins who were separated shortly after birth and reared apart.11 The result? Statistically significant heritabilities of 41 percent for children and 28 percent for adults.

  These findings from a large sample are striking. But perhaps more dramatic are findings from a case study of just eight monozygotic twin pairs reared apart where it was known that one twin was convicted of a crime.12 The critical question was this: How many of the other eight twins from these pairs were also criminals? Of the eight, four had also committed one or more crimes, indicating clear evidence for the role of genetic factors. Because they were reared apart, you cannot say that the similarity is due to ha
ving the same upbringing—it’s more to do with genetics.

  One of the four concordant cases consisted of a pair of female Mexican monozygotic twins separated at nine months. They were brought up by parents with very different personalities.13 Their environmental upbringing was also very different. One twin was brought up in a town, while the other twin grew up in the desert. Nevertheless, quite independently and as if by magic, just after reaching puberty both twins left their homes, took to the streets, and started to commit juvenile crimes. Both were separately institutionalized several times for their offenses. Recidivistic female crime is unusual, and when packed in the form of identical twins reared apart, it is even more extraordinary. Here we see the powerful influence of the genes that these two girls shared in common. It’s a case of dark genetic forces overshadowing the power of the environment.

  Studies of twins reared apart represent an important research strategy for understanding the genetics of crime. While the eight case studies are not methodologically strong they do illustrate the usefulness of this twin approach. Together with the finding of a methodologically stronger research study that utilized thirty-two pairs of twins reared apart and observed the same findings,14 they add yet another important strand of support for a genetic predisposition to crime and antisocial behavior.

  BUT WHAT ABOUT THE ENVIRONMENT?

  Well might you ask. If you’re an advocate of the importance of the environment, all this genetic stuff is disconcerting. But here’s some good news. Genetic studies inform us about environmental influences just as much as they tell us about genetic influences. Twin studies tell us that about 50 percent of the variance in antisocial behavior is explained by environmental influences. The genes-versus-environment battle comes out as a tie.

  But you know from when you were a teenager yourself that there are different types of environmental influences. Which one is more important in shaping antisocial behavior in children? The influences that come from within the family? Or the influences that come from outside the family? What would your guess be on who is more important in shaping kids’ behavior—their home and parents? Or influences outside of the home?

  It turns out that parents don’t count as much as you would like to think. When Laura and I examined this, the familial home influences accounted for on average 22 percent of the total variance in antisocial behavior. In contrast, environmental influences outside the family accounted for 33 percent of the variance.15 Even at nine years of age, children are being influenced—even pushed and shoved—in directions dictated by their peers rather than their parents.

  This may sound hard to believe, but our study is no fluke. When you look at the results from overarching reviews of all genetic studies of antisocial behavior—over a hundred of them—you get the same result.16 The same is true for a wide range of behavioral and personality measures, so it doesn’t apply just to antisocial behavior. Indeed, Tom Bouchard, a leading behavioral geneticist at the University of Minnesota, has argued that shared environmental influences on adult personality are almost zero.17 Yes, zero—no influence at all.

  If like me you are a parent, it’s sobering news. Do you want to believe that all your valiant caregiving efforts are worth almost zero? We face enormous cognitive dissonance—we do not want to believe findings like these because it means all our best efforts have been a waste of time.

  It’s frankly very upsetting. Parents want their children to be like them, and they put a lot of work into raising their children. And lo and behold, they turn out just the way their parents had wanted. So parents naturally believe that of course their efforts made a difference. But what if it’s genetics? Parents silently and passively contribute half of their genetic material to the child. They cannot see their DNA and how it influences their child. They can, however, see all their socialization efforts, and if their child turns out well, their conclusion that their efforts really counted is reinforced. In our desire to believe we make a difference, we may not want to believe that our perceptions on how important we are as parents are wrong.

  Taken together, the astonishing depth and breadth of twin studies is one factor that is beginning to change criminologists’ minds about genetics. Slowly but surely, more pirate ships have appeared on the research horizon flying a genetic Jolly Roger. One ship you can ignore, but not an armada. Yet explaining the sea change that is occurring in social scientists’ minds goes far beyond this fleet.

  ADOPTION STUDIES—BACK ON THE LANDRIGAN TRAIL

  Twin studies may underestimate the extent to which genes shape antisocial behavior because the error in the measure of antisocial behavior gets counted as non-shared environmental influences. But as we have seen, they may also overestimate genetic influences due to breakage of the equal-environments assumption. We need a pointer to get back onto the right path.

  We leave environmental territory and get back onto the genetic trail. Recall that Darrel Hill left his son Jeffrey Landrigan at birth, and we saw an eerie likeness in their adult violent behavior. Now let’s magnify this case study hundreds of times—by studying together hundreds of Jeffrey Landrigans—to see scientifically if there is a father-son linkage. A linkage even though the offspring never grew up with his true parent, but was instead brought up in a different home, a different environment, with a different way of living.

  In the adoption design, offspring are separated from their criminal biological parents early in life and fostered out to completely different families. This is the experimental group. The control group consists of babies also fostered out soon after birth, but their biological parents do not have criminal records. If the offspring with criminal parents grow up themselves to become criminals at a higher rate than adopted children whose biological parents were not criminals, this would indicate a genetic influence stemming from their biological criminal parents.

  That is precisely what has been found. In a landmark adoption study of crime, my colleague Sarnoff Mednick demonstrated that the adopted-away offspring of criminal parents in Denmark were more likely to become criminals as adults than the adopted-away offspring of noncriminal biological parents.18 You can see these findings illustrated in Figure 2.1.

  Mednick grouped the adoptees based on the number of criminal convictions of their parents. The adoptee controls, of course, had parents with zero convictions. Some adoptees had parents with one conviction, some two, and so on. What you see plotted in Figure 2.1 is the number of criminal convictions in the adoptees as a function of the degree of criminality in their biological parents. You can clearly see that the more convictions the biological parents had, the more offending there was in their adopted-away offspring. It’s a very clear demonstration that one’s genetic heritage predisposes one to crime. It’s also a reliable finding—almost every other adoption study on crime has observed the same finding, and there are more than a dozen of them.19 The findings are replicated time and time again across independent research laboratories in different countries.

  Figure 2.1 The increase in homicide rates in adoptees as a function of the degree of criminal offending in the biological parents

  Which isn’t to say there aren’t caveats. Adoption agencies, for example, try to place babies into adopting families who are similar to the true biological parents—a process termed “selective placement.” Furthermore, there could be differences in the length of time the baby is with their natural mother. If antisocial mothers are neglectful of their offspring before adoption, this negative bonding experience—an environmental process—might account for the later antisocial behavior. Mednick, however, was careful to control for these factors. His findings could not be explained away by selective placement of adoptees into adoption homes of a similar socioeconomic status, or the age at which the infant was taken away from the mother. Other studies have similarly controlled for methodological confounds like these.20

  Of course, twin and adoption studies, like all other studies, have their methodological weaknesses. Critics of the conclusion that there is a
genetic contribution to crime will eagerly latch onto such limitations. Their objections may seem to disqualify the conclusions, but it’s a false alarm. These studies represent different people, time, places, measures, and designs.21 All these differences should often lead to the expectation of divergent, different results—yet very tellingly they all converge on the same intrinsic finding.22

  Let’s apply this principle to the current context. Participants in more than a hundred genetic studies of antisocial behavior have ranged in age from nineteen months to seventy years. They cover the period from the Great Depression to the present. They represent many different Western nations, including Australia, the Netherlands, Norway, Sweden, the United Kingdom, and the United States. They use a wide variety of measures of antisocial behavior. They are made up of twin studies, adoption studies, and sibling designs. They also include large-scale studies that represent the general population and use advanced quantitative modeling techniques. They include studies conducted in the past fifteen years, and the findings from yesteryear stand up in studies done today.23 Taken together as a whole, these studies converge on a simple truth that even the strongest critics of genetic influences in violence are finding harder to resist—genes give us half the answer to the question of why some of us are criminal, and others are not.24

  ACNE AND XYY

  What specific seeds account for sin? It’s a big question, and it has always been controversial. In the past the most sensationalized link between violence and genes has been the case of XYY.

  Normally we each have twenty-three pairs of chromosomes, with each chromosome being a bundle of many genes. One of these chromosomes is the sex chromosome—X or Y. Each parent gives one chromosome to each pair, which determines if we end up as an XY (male) or an XX (female). But on rare occasions there’s a mistake. Instead of one Y chromosome pairing with one X, two Y chromosomes pair with one X. The result is a male who receives an extra male chromosome—XYY.