Do the MAOA and CDH13 ‘human warrior genes’ make violent criminals — And what should society do?


The warrior gene is back. And he’s brought along a buddy. This new research on a gene long associated with aggressive behavior raises an old question: What can–or should–be done about genetic predispositions that lead to grim social consequences in only some of the people with the predisposing genes?

The usual response, picking holes in individual research projects, denying that genes are ever involved in bad behavior, is just not good enough. We need to get serious about figuring out how to interfere with noxious genetic susceptibilities in ways that are fair and decent for everybody.

The so-called warrior gene comprises particular variations in the X chromosome gene that produces monoamine oxidase A (MAOA), an enzyme that affects the neurotransmitters dopamine, norepinephrine, and serotonin. The variants, known collectively as MAOA-L, produce human MAOA “knockouts” with a low level of the enzyme.

MAOA was the first candidate gene to be linked to antisocial behavior, identified in 1993 in a large Dutch family that was notorious for violence. It has been a media favorite ever since, acquiring the nickname “warrior gene” in 2004 as a result of an article in Science, of all places. This I learned from John Horgan’s fine rant about the exploitation of MAOA genetics at Scientific American, which describes weaknesses in the research.

The most recent appearance of MAOA-L is a paper Molecular Psychiatry published a week ago from a host of researchers based mostly in Finland. It showed that Finnish criminals convicted of several violent crimes frequently possessed either MAOA-L or a mutant version of another gene, CDH13, while the nonviolent controls did not. Find details in John Gever’s piece at MedPage Today.

CDH13 is involved in signaling between cells. Previous research has linked it with attention deficit/hyperactivity disorder (ADHD), autism, schizophrenia, substance abuse or bipolar disorder. So far as I know, this is the first time it has been associated with violent criminality. I will ignore it for the rest of this piece because I want to focus on MAOA and its long history of being connected with aggressive behavior.

Genes vs the environmental factors in violent behavior

Although it’s clear that the Finnish researchers believe their findings unequivocally, they also appear to understand the unhappy history of attempts to find genetic explanations for crime and violence. They also know perfectly well that, even if their findings turn out to be true, other factors besides low MAOA go into the making of violent criminals.

Past research has found relationships between specific environmental factors and genes linked to aggressive violence, including MAOA. A particularly strong connection has been noted among abuse in childhood, MAOA-L gene variants, and violent behavior in adulthood. A recent review declared that several studies have shown that MAOA-L men previously exposed to early life abuse engaged in significantly higher levels of violent behavior than men with high levels of MAOA. The authors assert that this is one of the best-supported “observations in the entire literature of psychiatric genetics.”

Well-supported it may be, and child abuse is certainly a plausible connection. But the Finns found no such link in their studies. They say, “maltreatment did not modify the risk in any way.” They have, however, identified another factor they think is crucial: intoxication, either with alcohol or amphetamines.

Intoxication, they say, is a feature of most of the violent crimes in Finland. They propose that intoxicants interact with MAOA-L to affect brain neurotransmitters and produce impulsive aggression. Their suggestion: when violent criminals are released from prison they should be subject to mandatory treatment with drugs like disulfram or naltrexone that interfere with the effects of intoxicants.

Child abuse and intoxicants by no means exhaust the list of possible  influences on genes and behavior. There are doubtless many others. I ran across a paper proposing a complex relationship with the “male” hormone testosterone and antisocial (and prosocial) behavior. High levels of testosterone in fetal life and childhood, the theory goes, combine with negative or positive early life events to produce either “chronic antisocial lifestyles” or men predisposed to “socially adaptive traits such as a strong achievement motivation, leadership, fair bargaining behaviors, and social assertiveness.” That sounds plausible too. Maybe the Finns should be investigating whether their MAOA-L violent criminals have high levels of testosterone too.

Dealing with the revelations of behavioral genetics

The John Horgan piece I referred to above is a rant–a productive and rational rant that will give you a brief history of what’s been misleading and outright wrong about past attempts to link genes with violence and crime. But I’m coming around to the view that ranting is no longer a satisfactory way of dealing with the discombobulating implications of behavioral genetics. We have to start figuring out how to handle them.

It’s not an adequate response to pick nits with particular papers and so by implication condemn all of behavioral genetics as a hopelessly flawed endeavor. MAOA-L is a prime example, maybe the best one–and a good place to start. The studies on low MAOA activity have piled up. Despite their individual flaws, it’s pretty clear that something really does seem to be going on with that gene variant that is (or can be) in some way related to bad behavior.

I have read that MAOA-L is pretty common–one paper says 40 percent of the population possesses it. It gave no reference, and I haven’t been able to nail that number down for sure, but let’s assume it’s true. Let’s assume that many of us are walking around with low MAOA and that we are not aggressive, don’t commit violent crimes, and are really nice people. You may be one of them. I may be one of them.

Does the fact that most people with low MAOA are not violent criminals mean there should be no attempts to identify and prevent whatever bad behavior is encouraged by MAOA-L? The researchers argue that their findings should not lead to screening for these gene variants, and I agree. But what about their proposal to prevent violent criminals from using alcohol and other intoxicants when they get out?

Applying it across the board would mean that former violent prisoners without MAOA-L would also be denied intoxication. My feeling about that is, so what? We know that alcohol and some other drugs precipitate irresponsibility and nastiness in lots of people. We already have laws that punish bad behavior associated with those drugs. The laws and social pressure even help prevent chemically induced bad behavior.

What’s wrong with applying that logic to criminals with a history of violent–often murderous–behavior? It takes the focus off genes and shifts it to well-known environmental triggers for bad behavior. These are much easier to control than genes–and would probably have more widespread social benefits.

Tabitha M. Powledge is a long-time science journalist. She also writes On Science Blogs for the PLOS Blogs Network. Follow her @tamfecit.

Additional reading:

  • Gerhard Adam

    I’m not sure what’s being suggested beyond the fact that human behavior is much more complex than a singular gene. The proposed social “solutions” will undoubtedly work just as well as keeping guns out of the hands of convicts.

    After all, if the point is that we need more laws to protect us from “violent–often murderous–behavior”, then doesn’t it make more sense to leave these people in prison, rather than to create a convoluted legal system whose sole purpose is to put them back in anyway?

    Isn’t the first question we should be asking is why are violent/murderous criminals being released? We already house far too many drug addicts in our prisons, leading to overcrowding, do we really need another drug-related law to do even more of it?

  • Unsilenced Science

    I don’t know who John Rennie is, but the John Horgan rant to which you linked was not “fine,” “productive,” or “rational.” Like many others, he dismissed the research based on a copy-and-paste error that slanders Chinese people.

    The variants, known collectively as MAOA-L, produce human MAOA “knockouts” with a low level of the enzyme.

    No, you are mistaken. Knockout refers to a missense codon that turns the gene completely off. In the case of MAOA, this is called Brunner syndrome. MAOA-L lumps rare alleles with the two common forms, despite the fact that studies have found MAOA-2R to be far worse than MAOA-3R. This means that studies that only look at MAOA-L in populations with relatively high prevalence of MAOA-2R, like African Americans, underestimate the effects of MAOA. In fact, there are many promoters of MAOA and other interacting genes, hormones, and drugs for which studies have not controlled.

    A particularly strong connection has been noted among abuse in
    childhood, MAOA-L gene variants, and violent behavior in adulthood.

    It’s easier to talk about child abuse than child IQ, but the influence of the latter was stronger in Fergusson et al.

    Maybe the Finns should be investigating whether their MAOA-L
    violent criminals have high levels of testosterone too.

    Yes, in fact, testosterone and other hormones also interact with MAOA-3R. I wonder why scientists would rather study only the gene-“environment” interaction with child abuse.

    I have read that MAOA-L is pretty common–one paper says 40% of the population possesses it. It gave no reference, and I haven’t been able to nail that number down for sure, but let’s assume it’s true.

    Let’s not. I can help you nail the number down, since I have been tabulating the allele frequencies. Unfortunately, that might require that you address the prickly issue of race, since the much greater frequencies of both MAOA-3R and MAOA-2R are found in African Americans. (HapMap also shows that the CDH13 allele identified in that study is most common in Mexicans.) Since Africans and Asians outnumber whites, MAOA-3R is the most common version in the world.

    The studies on low MAOA activity have piled up.

    Despite your errors, I like this line. I just wish the word “meta-analysis” had appeared somewhere.

    • Dennis Goos

      Can African Americans and Mexicans as well as Chinese and Finns be distinguished by genetic analysis? If not, those grouping would have little meaning in determining the value of specific codons. Is there an identifiable genetic combination that predicts violent behavior in all places at all times ?

      • Tabitha Powledge

        Good questions.

      • Unsilenced Science

        Your comment contains a non sequitur. However, scientists did publish a study based on this logic. Widom and Brzustowicz grouped their sample into white and “non-white” categories. The white sample was 24% female. The non-white sample was 43% female. They did not control for gender. Since MAOA is on the X-chromosome and is subject to both hemizygosity in men and greater methylation in women, it has much less effect on women. So, the scientists might have intentionally engineered their conclusion that MAOA only affects whites. The studies Haberstick et al and Foshee et al compared white males to black males and found stronger effects of MAOA on violent crime in the black-male samples. Perhaps this is partly because black people are more likely to have versions of dopamine genes like ANKK1, DAT1, and DRD4 that have been linked to antisocial behavior.

        A study that compared self-identified race to genetically determined groups found almost perfect alignment, but anthropologists and population geneticists still try to deconstruct race, and they deny that any genes are “strong candidates for playing a role in behaviour.” Geneticists are not the leading authority on behavioral genetics. Psychiatrists have done more of this research.

        • Dennis Goos

          “but anthropologists and population geneticists still try to deconstruct race,” Can you identify race by genetic analysis?

          • Dennis for a good discussion of that, you might find two of my books of interest: Taboo, on race and sports, and Abraham’s Children, on race and religious identity. The answer to your question is nuanced.

          • Lee

            Dennis glad Jon brought his books up as both of Jon’s book are pretty much discredited by anyone who studies race. A BA in Philosophy doesn’t get one a degree in genetics. Read how Kevin Macdonald says Jon misquotes him in Abraham’s Children and says flay out Jon doesn’t get it. Jon looks good on paper but dig deeper and it gets ugly. Dude thinks there is a black sports gene…he got laughed out of the race space (see hir article) and now shills for the foundations and trust behind GLP. Guess he has to pay that child support and alimony some how. What gene causes spouse abuse I wonder?



          • Unsilenced Science

            Very well, yes. The exception proves the rule.

          • Dennis Goos

            From your reference, I quote;”Subjects identified themselves as belonging to one of four major racial/ethnic groups (white, African American, East Asian, and Hispanic)” . One expects common artifacts among groups and I expect predictions can be made for groupings. But can a researcher identify an individual’s racial or ethnic group when only the genetic analysis is available in an otherwise blind study?

          • adplatt126

            The answer to your question is yes. Geneticists can definitively determine your race by genetics alone (with 99.99% accuracy).

          • klm68

            I would really appreciate a citation for the research publications that support the statement you make. The great majority of studies I see on genetics and race have difficulty even defining race.

            For social, economic, and historical reasons, we’ve designated arbitrary constellations of phenotypic traits to indicate races. There is no compelling biological reason why those traits and not others should take priority in defining race. And the obvious traits we use to indicate “race” are influenced by a fairly small number of genes, so it’s difficult to assign any larger meaning to the artificial categories we create.

            It’s possible to use genetic markers to make inferences about the geographic origins of an individual’s ancestors. But often these markers are from non-coding regions of the DNA, including single nucleotide polymorphisms and repeat sequences that probably have little or no phenotypic effects. So, again, it’s difficult to assign any social or behavioral significance to that level of genetic variation.

            Still, I’m interested in learning more about the issue and would appreciate links to further scientific information. Thanks.

          • adplatt126

            1) I made one claim here. Cluster analysis allows geneticists to determine an individual’s race with an extremely high level of accuracy. That’s easily researchable.
            2) It certainly is difficult to assign social or behavioral significance to these differences, absent direct data on behavior differences and the genetic differences that underlie them. We can however still engage in educated speculation.
            3) Lastly, there are however some examples of underlying allele frequency differences that almost certainly account for some of the predisposition to violence seen in certain groups (relative to other groups). That is to say there is direct evidence of underlying genetic differences that probably manifest themselves in behavioral disparities between racial/ethnic groups. The MOAO (Monoamine Oxidase A) gene springs to mind…

          • klm68

            Thanks for your response. Nevertheless, educated speculation is still speculation, and we need to be careful. The fact that males, almost universally across cultures, are far more likely than females to be violent suggests a genetic influence, but it does not mean that we know anything about any individual male’s propensity toward violence.

  • John Horgan

    Tabitha, as a commenter points out, that “rational rant” was by me, John Horgan, not my former boss at Scientific American John Rennie. Otherwise, nice piece!

    • Tabitha Powledge

      Aaaaaargh! Many apologies. Apparently I get my Johns mixed up, as my boss, a Jon, has mentioned to me.

  • Into the Endless Begin

    “It’s not an adequate response to pick nits” well like it or not pal that’s part of scientific method.

  • Bibibibibib Blubb

    I just want to add that there are experiments done which did not find a correlation with MAOA genotype(allele variation) and MAOA activity/levels. As in the variant had no distinguishable effect on the function of the gene. Links below.

    There are 3 of them from what I know.
    The first one found a correlation with methylation of the gene not the variant. These studies use some fancy medical methods and technology to get info from the tissues themselves.

    …and opposite activity in this Han Chinese population.

    I’m pretty sure there are more out there but are harder to find than the positive associations because of publication bias against negative finding espeically in psychology.

    If I were you I wouldn’t trust studies such as this even a little bit.

  • robotman2009

    Great discussion…I have MAO. I went ahead and did my 23andme. I wouldn’t say I’m antisocial. In fact, I’m kind of a social butterfly. With all that said I find that most people’s discussions are pretty boring….but I do love it when I can get people to engage in very deep discussions…which btw is pretty often. I do find that I have to patiently endure and listen at times and let others have the reigns in the course of polite and fruitful discussion. I’m definitely violent or anything. In fact, I’m kind of the opposite. I prefer to be at peace and interact with folks even when there is not total agreement. I do find I have strong impulsivities. But never have I had violent impulsivities. It’s more like oooh look at this…wow….wow look at this over here. etc.

  • robotman2009

    why was my previous comment deleted?

  • robotman2009

    2nd post was deleted.

  • robotman2009

    3rd post was deleted.

  • robotman2009

    4th post was deleted

  • robotman2009

    5th post

  • robotman2009

    6th post

  • thinkingforyourself

    Im glad Im passing these genes on to my descent. May they ever use their quicker reflexes, faster ability to analyze and freeze frame time, and rapid action muscles to kick the ass of any who want to socially engineer them. 🙂

  • Anton

    I find it interesting that a study from Finland, where child abuse is quite low, but alcoholism quite high, would shun the MAOA/abuse connection in favor of an MAOA/alcohol connection. The author fails to point out that there are sub-classes of MAOA and that only one of those sub-classes MAOA-3R is associated with child abuse.

  • Matt S

    How I understand the connection of genes influencing behavior and personality, from what I’ve read in scientific literature, is nature and nurture work together in producing an individual. If the warrior gene thesis is true, the gene sets a range of potential behavior, but the person’s upbringing is what determines whether or not they become a violent career criminal or a legit, but ruthless business man.

  • Jeff Sidlosky

    There is a medication for the Warrior Gene, that works (for me). Check it out here: