Can Brain Scans Detect Pedophiles?

By Veronique Greenwood | October 5, 2011 3:45 pm


What’s the News: A new study suggests that watching brain activity when subjects are shown images of naked children can identify which are pedophiles. But what does this really mean in practical terms?

How the Heck:

  • 24 self-identified pedophiles, from a clinic that offers anonymous treatment, and 32 male controls were shown pictures of naked men, women, and children. Blogger Neuroskeptic, who brought this study to the web’s attention, notes in an aside that getting that past a university ethics board is quite a coup.
  • Using fMRI, the researchers recorded their brains’ responses and found that by comparing an individual’s brain to the average of the pedophiles and the average of the controls, they could assign them to the correct group more than 90% of the time. Their handling of the statistics avoids the most obvious pitfalls: they used an analyses technique called leave-one-out cross-validation to avoid comparing a given scan to an average that includes it, a common error in neuro studies.
  • When they plotted how the neural scans lined up along the age and sex axes (see image above), the pedophile and control scans formed two clear, separate clusters.

What’s the Context:

  • First, a caveat: the team wasn’t looking for any specific brain activity, and who knows what exactly the pedophiles were thinking when they saw the images of children. It could be sexual attraction, or it could be any number other things, like shame, which seems like a strong candidate, given that these people were in treatment for their pedophilia. Pedophiles who aren’t in treatment, or pedophiles who’ve never acted on their feelings, might not be so easily clustered with these subjects. The point is that the only thing shown here is that this group of pedophiles’ brains behaved differently than those of controls.
  • But moving along to the philosophy, ever since science brought the revelation that our brains are what make us who we are—rather than something like a soul, for example—there’s been the question of to what extent we can be judged on the basis of our biology. This study, raising as it does the idea that your brain betrays you, and that certain brain profiles could be linked with crimes, recalls a piece neuroscientist David Eagleman wrote for The Atlantic recently:

If you are a carrier of a particular set of genes, the probability that you will commit a violent crime is four times as high as it would be if you lacked those genes. You’re three times as likely to commit robbery, five times as likely to commit aggravated assault, eight times as likely to be arrested for murder, and 13 times as likely to be arrested for a sexual offense. The overwhelming majority of prisoners carry these genes; 98.1 percent of death-row inmates do…As regards that dangerous set of genes, you’ve probably heard of them. They are summarized as the Y chromosome. If you’re a carrier, we call you a male.

The Future Holds: Realistically, we can’t judge people on their biological red flags—whether it’s brain scans or genetics—without entering a Minority Report-style future. That’s because being attracted to children isn’t a crime. Acting on it is. Generally, a brain scan is not required to find out whether the crime has happened—there are other, more reliable forms of evidence, like porn downloads. Given that, it’s unlikely that you’ll be seeing regular scanning for pedophilic tendencies anytime soon. And for that, we should be grateful.

Reference: Ponseti, J., Granert, O., Jansen, O., Wolff, S., Beier, K., Neutze, J., Deuschl, G., Mehdorn, H., Siebner, H., & Bosinski, H. (2011). Assessment of Pedophilia Using Hemodynamic Brain Response to Sexual Stimuli. Archives of General Psychiatry DOI: 10.1001/archgenpsychiatry.2011.130

[via Neuroskeptic]

Image courtesy of Archives of General Psychiatry, Ponseti, et al., via Neuroskeptic

CATEGORIZED UNDER: Mind & Brain, Top Posts
  • Greg

    They actually used images of naked children in the study? I know it was brought up in the article, but that seems massively unethical.

  • John Kwok

    Another important caveat to this study is that there it may not be statistically robust enough to support its conclusions.

  • Chris

    Actually you don’t need a fancy MRI to find this. Just put up an advertisement that you are conducing research which will need volunteers to look at naked children. Whoever shows up is probably a pedophile.

  • PersonalJesus

    I suspect the scientists who collected and disseminated the child pornography are themselves pedophiles or at the very least do not possess any sense of ethics.

  • Jake

    Naked does not mean sexually explicit or pornographic.

  • John Kwok

    I’m not sure whether one can really make much out of a study that doesn’t even have a sample size that is normally distributed for both the subjects as well as the controls. As I noted yesterday I don’t think that it may be sufficiently statistically robust to support its conclusions. At best this should be seen as a preliminary report on a potential methodology that could – but rather remotely for now – lead to a “Minority Report” future.

  • Chris

    Hi I’m Chris Hansen. Would you please step into our MRI.

  • Daniel

    Oh c’mon people, what’s all this talk about unethical?? since when looking at pictures of naked children makes you a pedophile?? you’d have to censor tons of pictures in medical books then. It’s about getting sexually aroused, not just looking at them.

    close your eyes everyone, here’s a massive pedophile conversion link

  • Jeff

    Seriously, if this study can help prevent/stop pedophilia what the hell is unethical about it?

    You guys really think they do a meeting and jerk off in front of porn child?

    WTF with this oversensitive bullshit, I tought readers of this blog were more intelligent than that….

    Oh wait, maybe you guys are pedophiles if you’re against those kinds of studies!

  • Gregg

    Jeff, you have to realize these “tests” are very ‘Joseph Mengele’ in nature of cerebral context.

  • Dave

    With only 24 subjects and 32 controls, how do they figure they can correctly assign them to the correct group “over 90% of the time”? This study doesn’t just raise ethics questions. It also raises questions regarding its scientific methods.

  • Carol

    First it was “The Devil made me do it – forgive me, I am possessed!” Now its “My Brain made me do it – forgive me, my genes are possessed!” Because no one would any longer be considered a criminal or pervert by choice or Free Will, in place of arrests and incarceration we’ll be doing exorcisms and lobotomies.

  • John Kwok

    @ Dave –

    Thanks for noting that this study “raises questions regarding its scientific methods.” I remain quite skeptical that the authors’ reliance on cross-validation was the appropriate statistical test; probably far more robust would have been some kind of nonparametric ranks test. All this study could show (maybe) is that pedophiles may be more inclined to show certain brain activity than those who aren’t; the population sample size is not sufficiently adequate to make the very partitions within the data that they opted to choose. I think this is a perfect example of a flawed study that should not have passed peer review; but instead, the reviewer(s) should have told the authors to collect more data that would either confirm or refute their conclusions based on a more robust statistical analysis.

  • Lee

    Is anyone actually getting harmed by these studies? The kids can get their faces covered. Science has to solve this problem, otherwise other kids may be abused and destroyed for life.

    Yeah, Dave may be right. Those data cannot be treated statistically.

  • John Kwok

    @ Lee –

    Not only Dave, but I may be right too. I raised the issue regarding sufficient sample size for statistical analysis yesterday by noting that the study “may not be statistically robust enough to support its conclusions.” IMHO we are probably both right for the same (or virtually identical) reasons.

  • Hibernia86

    Jeff at #9 is disturbing. “We should punish people for their thoughts rather than their crimes and if you don’t do what I say then damn it you are a commie, er, I mean pedo too!” Jeff, you remind me of the novel 1984 far too much.

  • Dr. J

    Nothing new. Previous research used a penile strain gauge to measure response to different pictures. A partial erectile response sorts out pedophiles. Same testing is used to measure possible results of treatment. Conclusion: Since the brain and penis both respond to naked pictures of children, maybe the penis is connected to the brain after all and a penis does not have its own mind after all.

  • Bob C.

    in my old neighborhood, a car had a bumper sticker saying, “i love kids, but i can’t eat a whole one.” it also had another sticker indicating the owner was a wrestling fan. i agree with mr. Kwok that it’s too small a sample, but on the other hand, it’s difficult to get a large one with people trying to hide those tendencies. how do we weed out the Dahmers from the people that just “jerk off” to photos but would never touch a real kid? we can’t even define porn vs. art photography, but that’s why David Hamilton lives in France now. it’s a tough nut to crack, for sure.

  • BioSlayer

    A third aspect to Sample Size and Study Design, these self-identified pedophiles are undergoing treatment/counseling or something like that, so what if the treatment was interfering with their thoughts, were they coming from the same treatment regimen and hence the consistency in their brain scans? besides, their scans did not zero-in on a particular brain center, now that is so inconclusive on their part to come with such an observation !

    Probably on fMRI they got such a clustering, however, more rigorous statistics will need to be used to classify pedos vs. normal, ROC curves down the way probably or a more intelligent model than cross validation !

    To build a model, you have to have enough sample size, 24 doesn’t seem to get anywhere close to a generalizable observation ! The model accuracy is a function of it receiving thorough training and having a manageable number of parameters or problems with overfitting and stuff can happen.

    The adage that our brains makes us who we are doesn’t seem to hold in this way, give someone a chance (law vacuums) or deprive them of some (as in correctional facilities) and any heterosexual Y-Chromosomer or XX-chromosomer will turn into another category, what do you guys think???

  • John Kwok

    @ BioSlayer –

    Given the categories defined by the researchers, they should have had at least 120 experimental subjects divided evenly by four to have a normal distribution for each category, and that would have allowed them to use parametric univariate and multivariate statistics. It would also probably greatly improve the scatter in the data as evidenced in the plot (see above) that was published with the paper.

  • ron hansing

    I agree with Kwok… The problem is that inconclusive science has a way of becoming exaggerated facts and common knowledge in the main stream media.

    As a physician who reads primarily medical research articles, I see this very often… the interpretation by the main stream media is often completely off base.

    They make General ASSUMPTIONS to the articles and then apply these assumptions as gospel.

    And it is not just the main stream media that makes this fundamental mistake… many doctors likewise make these broad assumptions. And not to mention, many of the original articles have big problems, I estimate at least 50% up to 75% of these articles… have big problems, or are misinterpretation by others as gospel.

    Ron hansing 10.7.11

  • John Kwok

    Thanks Ron for your comments. Not only science writers, but the general public too, should learn something about probability and statistics. As an aside, I greatly appreciate physicist Lisa Randall’s discussions of both in her new book “Knocking on Heaven’s Door” (I would recommend especially Chapters Eleven and Twelve).

    Again, all that anyone should conclude from this pedophile brain scan paper is that pedophiles may exhibit different brain activity than those who aren’t when reacting to certain images. Quite frankly, even such an assessment is rather generous given this paper’s poor experimental design and sample size.

  • phil

    Well I hope the children in the pictures were informed( if they’re adults now) or/and their parents. I don’t think the people doing this study thought this one through.
    and what if someone comes from a country where the legal age limit is somewhat lower? Unlikely , but it’s possible.

  • John Kwok

    With regards to the data indicated in the plot (see above), if one were to remove obvious outliers, then it would be difficult to claim that there are indeed two distinct clusters. Instead, I see a continuum of points stretching from pedophile to non-pedophiles. If there was a strong trend, then that could be predicted via some kind of regression analysis, but merely by visual inspection, there doesn’t seem to be any trend.

  • Dan

    On the topic of ethics, I don’t see much face value in a study like this. There’s a category difference between simply having ‘antisocial’ sexual fantasies/desires and acting on them. The former is not based on free will or harmful to anyone while the latter is.

    You can’t police thoughts or desires, only actions and only those that relate to others.

  • Pat

    What – you mean not every pedophile is a homosexual, and vice versa?
    Bad news cycle for the American Family Association, Focus on the Family and the GOP.

  • oceansand

    What you want for a “pedophile test”, is something that would tell you with 100% accuracy that someone has, or will in the future molest children. There is no such thing. At best, it could show some likelihood that if there were 100 clones of the same person, there could be an increased probability that they’d do something wrong. Also while the test may show activity resembling a pedophiles, that is never going to tell anyone what is going on in someone’s brain. I know the article covers this, but want to say I fully agree with that.

    In other words if there are some odd thoughts going on in an ultra creative person’s head who likes to explore every thought, that doesn’t mean that person is going to be going out looking for kids, or will be looking at kids either. The reason is, this test subject may not have any unacceptable thoughts of children what so ever. They may not have any desire to see naked children either. Yet the test is pushing naked children in front of them. And while there may be no interest in the subject, some weird thoughts could trip results so they appear to be similar to a pedophiles when the person isn’t one. Similar things go wrong with lie detector tests as well. Let’s just say, I’m skeptical that this test could never produce erroneous results…

  • Jennifer

    Read the book,”A Mind To Crime”,chapter 11.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!


80beats is DISCOVER's news aggregator, weaving together the choicest tidbits from the best articles covering the day's most compelling topics.

See More

Collapse bottom bar