Brain Scans and Brain Scams

Can we tell yet if a brain is a criminal brain?

Posted May 14, 2013

If I peek at your brain, can I tell whether you are a criminal? In his book, The Anatomy of Violence, Adrian Raine thinks he can make a pretty good bet.[i] The giveway? A bigger bit here or smaller bit there. That is the anatomy his title refers to.

My last blog (Criminal brains and criminal genes) unearthed a slew of problems undermining Raine’s idea that we can already identify ‘criminal genes’. Now let’s look closely at his claims to identify criminal brains. 

Consider what brain scans supposedly reveal about your criminal tendencies. Raine makes two different claims. The first result points to the size of specific structures. Compared to regular law-abiding citizens, those predisposed to crime have a larger striatum (whazzat? more anon).[ii] The second idea is that psychopaths’ scans show lower levels of activity in the prefrontal regions of the brain. With brain scans, I can see both of these features, so I can know if you are a bad apple.

What do size differences in brain structures actually mean? Maybe nothing. Eyeballable variability among brains identified as “normal controls” is a fact of biological life. There is variability in whole brain sizes, and in size of particular structures such as the corpus callosum that connects the cortex of the two hemispheres. There is also some variability in minor folds of the cortex and in the size of subcortical structures such as the striatum and regions such as the primary visual cortex. 

You might assume that someone with a larger brain than yours would be smarter than you. Nope. Or that if their corpus callosum is thicker than yours that they have  --- well, what? -- more integrated mental life? Nope. For example, normal age-matched folks show impressive variability in callosal thickness: as large as 9 and as small as 5 millimeters.[iii] All quite respectably normal. There are subjects with hugely enlarged ventricles (the normal fluid-flushing cavities in the cerebrum) whose cognitive function is well within the normal range. There are folks who had a whole hemisphere removed in infancy, and you would never know from their behavior.

In many instances, natural variability of macrostructure does not predict anything about brain function. (I mean natural, as opposed to caused by gunshot, for example.) More interestingly, structural variability often does not predict anything about microstructure, which is where the action is. Or as a political hack might put it: It is the wiring, stupid. Can brain scans see the microwiring? Nope.

So before we look at Raine’s claim about the striatum sizes, let’s raise a glass to brain variability and brain adaptability and brain wiring. And while we have the chardonnay flowing, let’s drink to how much we do not know about the brain.

Back to scanning the brain for criminals. If you have a small sample size, you might find it tempting to claim a causal link to a behavioral trait. Let’s take a ridiculously small sample – just me and you. I am thrifty; you are a spendthrift. I have a bigger claustrum than you (that is a subcortical long and skinny structure whose function is essentially unknown). So we conclude that a larger claustrum predicts thriftiness. Now I look at Bob’s brain, sadly sporting a smaller claustrum than mine. Hey, Bob must be a spendthrift. Bob, you are toast.

We all know that conclusion is really dumb. Because for all we can tell now, had our sample included, say 25,000 people, the somewhat larger claustrum in a few folks might predict nothing at all about thriftiness. Juts normal variability unrelated to thrift. So how big does the sample have to be to establish a causal connection between a claustrum size and thriftiness? Pretty darn big, owing to anatomical variability within the normal behavioral range. And we do not know how much variability in striatum size is within the normal range. Let me repeat. We do not know.

Raine reports a sample size of 22 psychopaths and 22 normal controls. Did each and every psychopath have a larger striatum than the largest of the normals? Nope. By my count from his data two normals had a left striatum larger than 20 psychopaths; 12 normals had a larger striatum than 2 psychopaths.  Four normals had a right striatum larger than 18 psychopaths. So what can you conclude here?  Raine concludes that as a group, psychopaths have a larger striatum. This is gently termed a group effect. That is, on average, the striatum was larger in the psychopaths.

Wait: if you give me any random two scans, one of which reveals a slightly larger striatum than the other, I could well be wrong if I thought it belonged to a psychopath. It might be a perfectly normal guy with a slightly enlarged striatum.

Suppose you scan your two sons, and one has a somewhat larger striatum than the other. Should you conclude he is probably a psychopath? Crikey – what if he is actually within the normal range of variability? A nice burden for the lad—and you—to carry the rest of his life.

If you have a really big sample size, the law of large numbers tells us that the sample average should be very like the population average. You do not have to be a statistician, or even friends with one, to see that 22 psychopaths is a small sample, given regular brain variability and the fact that some normal folks have a larger striatum than some psychopaths in Raine’s own data.  His sample is scarily small if you want to draw a treatment conclusion, such as “let’s fix this guy before he becomes an actual criminal. 

Incidentally, based on a sample of 15 psychopaths and 25 controls, Raine also claims that a larger corpus callosum is associated with psychopaths.[iv] Did I already mention normal variability and small sample sizes yielding iffy conclusions?

What the heck is the striatum? It is a connected set of structures under the cortex. And the complex is humungus. It is really composed of three different components – the globus pallidus, the putamen and the caudate, each of which has its own subcomponents. (See the figure below) These structures are much more ancient in evolutionary history than the cortex, which appears to be unique to mammals and birds. The diverse functions of these old structures are not well understood, but it is known that they play some role in reinforcement learning, some role in motor control, and some role in integration of information. That is sadly vague characterization of functions, but more precise characterizations will come eventually with further research.  Do we have any reason to think that some aspect of the striatum, if larger, might contribute to the phenomenon we call psychopathy? The confirming data are slim pickings.

Remember, the other claim made by Raine is that psychopaths have lower levels of activity in the prefrontal cortex PFC. Do we have good reason to think that there is a connection between a large striatum and lower levels of activity in the prefrontal cortex (PFC)? Nope.

Variability, that driver of evolution, again bedevils simple answers.[v]  If you are scanned today, and then again tomorrow, there may be substantial differences in the activity of your PFC between the two scans. Many factors could go into this: you had a lot of coffee this morning, little yesterday. You had a bit of bad news yesterday, but some good news today. And so forth. Additionally, your brain and my brain may both be well within the normal range, but we may exhibit some differences in levels of activity of the PFC on any given day.

The PFC part of Raine’s story seems to go all wobbly for many of the same reasons the striatum story goes wobbly. Understanding impulse control is currently a major topic of research, with many different strategies converging on how it works, how it can be impaired, and how differences in neurochemicals and their receptors can make for differences in behavior. Prefrontal cortex (a big area in itself) is generally believed to be important in impulse control. Precisely how is not known.

Seriously though, does lack of impulse control typify criminals generally?  Although poor impulse control may characterize the guys who get involved in bar fights, it likely does not hold for con artists or even of soccer hooligans who just enjoy the brawl. Raine does know this. As he acknowledges, criminals such as Ponzi schemer Bernie Madoff or inside trader Raj Rajaratnum seem well able to control impulses and probably have brains with PFC activity to spare. Is either a psychopath? I do not have the slightest idea. I do know both are jailbirds.

We do need develop techniques that can help us get more precision concerning brain wiring and brain communication at the micro level – the level of individual neuron. Brain scans are a highly useful technology when properly used, but they cannot access individual neurons and their business.  What we need are methods that will allow us to tap into microstructure on a grand scale, along with the computer procedures to analyze the data.

Will we find a signature of the criminal brain with those methods? I very much doubt it.  Because the ways of being a criminal are too diverse. It ranges from those who steal bread to keep a family alive, to counterfeiters, extortionists, cattle rustlers, plagiarists, and tax evaders, to those who rebel against a savage dictator or run a Ponzi scheme or peddle pornography or commit adultery in a very rule-strict state.  Moreover, what counts as a law violation varies as a function of the laws.


[i] Raine, A. The Anatomy of Violence. (2013)

[ii] Glenn et al. Biological Psychiatry 2012

[iii] Herron, Kang, and Woods, Frontiers in Neuroinformatics (2012)

[iv] Raine, A. et al. Archives of General Psychiatry (2003)

[v] On variability in activity within an indivisual across scanning sessions, see the very sobering paper by David McGonigle in NeuroImage (2012) and by Van Horn et al. Brain Imaging and Behavior (2008).