Criminal Genes and Criminal Brains
Are we there yet?
Posted May 07, 2013
Preventing crime rather than waiting for a crime to be committed is appealing. Identify the guy who will commit a crime, and intervene. What could be more sensible? First though, how do you identify the incipient bad guys? Bumps on the skull were once thought by phrenologists to provide a major clue. The hypothesis fell on its sword because its predictive power could not climb above zero. Such predictive ambitions remain entirely alive, however, though importantly, the methods for identification have been upgraded. Advice: Don’t bother looking at the skull. Look under it. At the brain itself. And also at the genes that make the brain.
In The Wall Street Journal, (Saturday/Sunday April 27-28 2013) psychiatrist/neuroscientist Adrian Raine posits a brain signature of the criminal mind. The suggestion is that there is a link between low levels of activity in the prefrontal regions of the brain and psychopathy. A second result involves not brain activity but brain structure: allegedly the size of the striatum is larger in criminals, on average. Raine also claims that genetics has begun to “pinpoint which specific genes promote [criminal] behavior”.
Buzz-wise, this is the scientific analog of a Kardashian wedding.
So there are two parts to Raine’s account: genes and brains. This blog will focus on criminal genes. The next installment will take a hard look at the brain-scan conclusions.
Here is one simple strategy: if you find associations between a gene and those who are criminals, then you have identified the “criminal gene”. Now all we have to do is screen people to see whether they carry the gene. Folks thus identified could be monitored or given prophylactic therapy. Or shipped to Australia. Oh wait, not anymore. Anyhow, crime could be much reduced through prevention. Hot damn.
Although the “genes-for” talk favored by evolutionary psychologists has long been skewered, drubbed and smacked down by geneticists, it lingers like the proverbial bad smell. Here is the gist of the problem:
That some traits are heritable is obviously true. Because my parents were humans, I am not a badger, I am a human. Moreover, I have my father’s height – he was tall, my mother was short. Is there a “gene-for” height? Well, real genes almost never play by simple “gene-for” rules. It turns out that there are at least fifty known genes that have some role in human height, and even then, those fifty are only part of the story. Whatever I got from my dad, it was not a single gene or two that made me tall. More likely a cluster of genes that interact with other networks of genes that interact with the environment, and lo, I ended up tall. What about behavioral characteristics? Did I inherit my father’s thriftiness? Well, he was a thrifty ethnic Scot, Scots tend to be thrifty, I am thrifty, so I guess I inherited a thrifty gene.
Let’s tread carefully. I may indeed have inherited a network of genes linked to brain microstructure linked to various personality traits and hence to behavior. Perhaps not to thriftiness as such, but perhaps to some more very general trait such as risk-aversion that, in a given context, we call thrift. Twin studies, as Raine rightly notes, demonstrate that genome-behavioral links do exist. Tracing those links for causality, however, has been the merry devil, even in simple animals for highly conserved brain-controlled behaviors such as sleep-wake cycles. The problem is that that the route from my genome to my behavior is more like a route through a ever-changing blackberry thicket than down a straight clear pathway from gene to protein to brain to behavior. Finding the route is not impossible, but it does require a lot of care.
The thrifty gene is a cock-up. Herewith why, in the genre of a parable:
The Parable of Aggression in the Fruitfly 
A long time ago, in the 1990s, a connection between a neuromodulator, serotonin, and aggression was observed in fruit flies and in mice. Experimentally elevating levels of serotonin using drugs or genetic techniques increases aggression in the fruit fly; genetically silencing serotonin circuits decreases aggression. These results are, moreover, consistent with experiments on the mouse, suggesting conservation of mechanisms for aggression through evolutionary change. Given this data, you might predict that the gene that expresses serotonin should be known as the “aggression gene” – or even, if you are bold, the “violent criminal gene”. This got many geneticists very excited. But the scientific worriers wondered if it could be so simple. So two worriers decided to test the idea. 
Geneticists, aka worriers, Herman Dierick and Ralph Greenspan  selectively bred aggressive fruit flies. After twenty-one generations the male fruit flies were 30 times more aggressive than the wild-type flies. Next, they compared the gene-expression profiles of the aggressive flies with those of their more docile cousins using molecular techniques (microarray analysis). If serotonin is the “aggression molecule” and the gene for serotonin the “aggression gene”, this experiment should reveal it to be so.
The surprising result was that no single gene could be specifically associated with heightened aggression. Instead, small expression differences were found in about 80 different genes.  What genes were they? Not any of the genes for regulating expression of serotonin. Many of the genes whose expression had changed were known to play a role in a hodge-podge of phenotypic processes – cuticle formation, muscle contraction, energy metabolism, RNA binding, DNA binding, development of a range of structures including cytoskeleton – as well as many genes having unknown functions. No single gene on its own seemed to make much difference to aggressive behavior.
How can that be, given the earlier experiments showing that elevating serotonin levels enhances aggression? The crucial point is that the relationship between genes and brain structures does not remotely reflect a simple “gene-for” model. Genes are part of networks, and there are interactions between elements of the network and with the environment. This is a huge challenge for a psychologist such as Jonathan Haidt who claims there are genes for liberal and conservatives (I am not making this up) let alone for neuroscientist Adrian Raine who frames his point, perhaps with reluctant simplicity, in terms of genes that promote criminal behavior.
Bear in mind that serotonin is a very ancient molecule. It is important in a motley assortment of brain and body functions: the list includes sleep, mood, gut motility (such stomach and intestinal contractions), bladder functions, cardiovascular function, stress responses, induction of smooth muscle proliferation in the lung during embryological development, and regulating acute and chronic responses to low levels of oxygen (hypoxia). 
The point of this list is to dramatize the diversity of jobs of serotonin, and hence the glaring unsuitability of the label, the gene for aggression. This parable illustrates why association studies must be evaluated with great caution. The diversity in serotonin’s jobs helps explain how it is that changing its levels can have widespread effects all over the brain and body. Including changing aggressive behavior. Because these changes can cascade into other effects, which may in turn exert an influence on aggressive behavior.
The moral of The Parable of Aggression in the Fruit Fly is that it is easy to speculate about a “gene-for” aggression based merely on observation of a behavior and perhaps an intervention such as experimentally altering the level of serotonin. But unless you do the right tests, you have no clue whether your speculation will stand.
If the relation between genes and aggression is that messy in fruitflies, how likely is it that the simple “gene-for-criminal behavior” model applies to humans? Not even marginally likely. This is not to say that genes make no difference to aggressive behavior. They absolutely do, as the Dierick and Greenspan selection results also clearly show. But the causal relationship between a gene and the brain structures involved in aggressive behavior is a vast and elaborate network of interacting elements. Moreover, some of those brain structures are responsive to the reward system, which modulates the likelihood of aggressive behavior towards other humans as a function of sensitivity to cultural norms. Finally, the connction may be better understood not as a link to criminal behavior as such, but to some more general trait, such as susceptibility to impulsiveness in contexts involving fear or rage. As with thrift.
Still in tire-kicking mode, let’s consider the one concrete claim Raine makes about the human genome, namely that subjects with a variant for the enzyme monoamine-oxidase –A (MAOA) gene are apt to be violent, if they are subject to an abusive upbringing. The variants produce lower levels of the enzyme MAOA. The link between the MAOA variants and violence rests on epidemiological research by Caspi, Moffitt and their colleagues (Science 2002). They meticulously tracked from birth to mid-life a highly homogenous population of New Zealand males. Replication studies, however, raised the inevitable complications. For example, the interactive MAOA gene x abuse effect may be specific to Causasians, and is not universal even in that population. In this domain, conclusions are still tentative, though additional data should eventually sort things out.
Other studies suggest that the abuse that increases the risk of violence in the MAOA variants needs to be quite specific — not just, for example, strict parenting but sexual abuse, for example. Additionally, it has long been appreciated that abuse and neglect, whatever your genes, are risk factors for subsequent bad behavior. Abuse is not good for any developing brain. The risk may be higher relative to particular genetic variants, including but not restricted to, the MAOA gene.
The data on MAOA and behavior are unquestionably important and fascinating. But let’s not pretend we have made a drone when we are still nailing struts on a glider. Concluding that the MAOA variant is a gene for criminal behavior is a long stretch, to put it politely.
The ambition to identify those at risk for criminal behavior is laudable, because prevention is generally preferable to cure, at least so long as the prevention is not itself catastrophic. To be sure, the research in this area is both very difficult and very important. Hats off to those who are trying, but the thumbs down on over-blown conclusions that are apt to mislead.
2. For a readable and scientifically strong introduction, see Jonathan Flint, Ralph J. Greenspan, and Kenneth S. Kendler. (2010). How Genes Influence Behavior. New York: Oxford University Press. For other articles rmaking a related point, see Risch, N., and Merikangas, K. (1996). The future of genetic studies of complex human diseases. Science 273, 516–1517; Colhoun, H. M., McKeigue, P. M., and Smith, G. D. (2003). Problems of reporting genetic associations with complex outcomes. Lancet 361, 865–872. Hattersley, A. T., and McCarthy, M. I. (2005). What makes a good genetic association study? Lancet 366, 1315–1323.
3. Herman A. Dierick and Ralph J. Greenspan, “Molecular Analysis of Flies Selected for Aggressive Behavior,” Nature Genetics 38, no. 9 (2006): 1023-31.
4. This does not necessarily mean that all of the 80 genes are related to the behavioral phenotype in question, since the differences in some genes may be due to their "hitchhiking" along with those that were selected.
5. Dennis L. Murphy et al., “How the Serotonin Story Is Being Rewritten by New Gene-Based Discoveries Principally Related to Slc6a4, the Serotonin Transporter Gene, Which Functions to Influence All Cellular Serotonin Systems,” Neuropharmacology 55, no. 6 (2008): 932-60.
Patricia S. Churchland is the author of the forthcoming Touching a Nerve: The Self as Brain published by W.W. Norton.