Monday, February 11, 2013

Evil Genes and the Long Dark Tea Time of the Soul

Edit before I hit post:  I wrote this whole post and then decided to google a few more things to make sure I had everything straight....only to come across this article that clarifies that Dr Roth was misquote by the Daily Mail in the article I start with.  I'm still posting this though because I think that the criticisms I make below are a good framework for assessing any study that looks at the biology of criminals, and because assessing it like this had already led me to believe something was amiss (see my footnote I wrote before I found the clarification).  What can I say, when in doubt, blame the journalist.

I saw an interesting link over at Maggie's Farm the other day about  a German scientist who claims to have found a "dark patch" in the brains of people who have committed violent crimes.

In the article, the scientist, Gerhard Roth is quoted as saying this:

When you look at the brain scans of hardened criminals, there are almost always severe shortcomings in the lower forehead part of the brain.... 
...But when I will look at young people, and I see there are developmental disorders in the lower forehead brain, I can say that there is a felon in the making with 66 per cent probability. 

What interested me as I read this was that his credentials and studies were being touted only as they related to his work with the criminal population*.  This can seem insignificant, I mean, if we want to predict criminal behavior, go take a look at the criminals right?

I'm not so sure.  It depends what he means by "66% probability".  At first glance, I'd assume he means that have the "dark patch" in question will go on to become felons.  But that sort of assertion would mean he'd have to start with a young, non-incarcerated population, identify those with this particular brain abnormality, and then see how many of them became felons.  The study the article describes does nothing of the sort.  It's one thing to say that criminals have a different brain from non criminals, but to say something's predictive you have to actually, you know, see if it can predict things.  It's possible he meant that 66% of felons have this patch and something got lost in translation?

Prison populations are really interesting to study, because they're literally captive audiences.  However, finding commonalities between criminals are fairly useless unless you have a sense of how prevalent the same thing is in the general population.  If we know this, we can know exactly what we'd do with those 33% up there who are okay and got swept up in the mess.

Some of this is covered at length in Barbara Oakley's Evil Genes.  There are a lot of issues with trying to predict violent behavior, and while I certainly think that linking genes or brain damage to evil is quite reasonable, we always will have to carefully weigh pros and cons of doing so proactively.  Even if you disregard ethical concerns, violent criminals always need to be compared to something to make sure you're not just picking up on a characteristic lots of people happen to have.

*Other weirdness in the article: Roth's not actually a neurologist, he's a neurobiologist, I'm pretty sure the "scans" Roth's referring to and that are pictured are not X-rays, you don't measure brain waves with either of those, and there's no such thing as a "central lobe".  None of this seems to be Roth's fault, but this reporter could use a little work.


  1. Makes me wonder how often reporters manage to accurately represent what they are told.

    Or how often they even understand what they are reporting on.

    (If you've never heard of the "Gell-Mann amnesia effect", this might be a good time to research it. The phrase was coined by Michael Crichton...)

  2. I saw this as well and couldn't figure out what was meant. There is no one, no how no way, who can predict with 66% accuracy whether someone is going to be a felon. In fact even if you know someone has committed a felony and not gotten caught, there's not a 66% chance he'll commit another one and qualify.

    I don't doubt there are some underperforming regions of the brain among criminals - I'm nominating the portions controlling expressive language.