Let's Face It: Publishing Weak Data on Face Processing in Pedophiles Is Pointless

Modified from Fig. 2 (Ponseti et al., 2014). Brain areas that selectively respond to faces of the sexually preferred age.


Just when we thought it was safe to bury the dead salmon of uncorrected statistical thresholds in neuroimaging studies, a new and incendiary study on face processing in pedophiles emerges (Ponseti et al., 2014). Even if it were surprising and informative that “Human face processing is tuned to sexual age preferences” (Ponseti et al., 2014), the fMRI data analyses failed to correct for multiple statistical comparisons, which is standard in the field. Therefore, by using a very liberal statistical threshold of p < 0.01 uncorrected for the large number of tests, the results could be a series of untrustworthy false positives.¹

Importantly, the basic pattern of findings, that visual parts of the brain are more responsive to pictures of faces that fall within the broad category of “sexual attractiveness”, does not tell us why someone has a particular sexual orientation, nor does it tell us if this preference is “hard wired” (i.e., innate).

The participants in the study were 56 men, 11 of whom were heterosexual pedophiles (prefer young girls), 13 homosexual pedophiles (prefer young boys), 18 heterosexual teleiophiles (prefer adult women) and 14 homosexual teleiophiles (prefer adult men). These are small groups, but to complicate matters, half of the pedophiles had committed sexual offenses and the other half had not. This is a critical difference, as one might expect differences between men who could refrain from acting on their impulses and those who could not. Yet, activation in the dorsal striatum was interpreted as a potential indicator of “efforts in withholding actions”.

Furthermore, the results presented here were part of a larger study that aimed to classify pedophiles solely on the basis of their brain responses to nude photos showing whole-body frontal views or genitals only (Ponseti et al., 2012). The authors claimed an astounding 95% accuracy in distinguishing between pedophiles and non-pedophiles.²

Overall, the participants viewed 14 different categories of visual stimuli in these two papers, so you can see that the number of potential statistical comparisons is astronomical.

The take-home message is that the participants' subjective attractiveness ratings of each face (completed after the fMRI study) were much more reliable at identifying their sexual preferences (p < 0.001) than the brain imaging data. Neuroscientists working with such controversial populations need to be especially careful in analyzing their data, and aware of how their work may be used in a broader social context.


Footnotes

¹ Thanks to commenter Com__Truise on reddit who alerted me to this paper and who noted:

Published cognitive Neuroscientist here. This should not have gotten through peer review. The fMRI analysis is invalid. The study uses an uncorrected threshold (not corrected for multiple comparisons) of p< 0.01 (considered very liberal!) and all the results are probably false positives. This is not to say that the theory is not correct - however, the statistics are invalid and meaningless. You can read more here: http://neurocritic.blogspot.de/2012/03/how-much-of-neuroimaging-literature.html http://www.danielbor.com/dilemma-weak-neuroimaging/

Here is another cautionary note from Professor James Ogloff, Director of the Centre for Forensic Behavioural Science and Legal Studies at Swinburne University of Technology, Victoria.

² Critical and ethical evaluation of this study is beyond the scope of the present post.


References

Ponseti J, Granert O, Jansen O, Wolff S, Beier K, Neutze J, Deuschl G, Mehdorn H, Siebner H, Bosinski H. (2012). Assessment of pedophilia using hemodynamic brainresponse to sexual stimuli. Arch Gen Psychiatry 69(2):187-94.

Ponseti, J., Granert, O., van Eimeren, T., Jansen, O., Wolff, S., Beier, K., Deuschl, G., Bosinski, H., & Siebner, H. (2014). Human face processing is tuned to sexual age preferences Biology Letters, 10 (5), 20140200-20140200 DOI: 10.1098/rsbl.2014.0200