Can Brain Activity Predict Criminal Reoffending?

Is it possible for a brain scan to predict whether a recently paroled inmate will commit another crime within 4 years? A new study by Aharoni et al. (2013) suggests that the level of activity within the anterior cingulate cortex might provide a clue to whether a given offender will be rearrested.

Dress this up a bit and combine with a miniaturized brain-computer interface that continuously uploads EEG activity to the data center at a maximum security prison. There, machine learning algorithms determine with high accuracy whether a given pattern of neural oscillations signals the imminent intent to reoffend that will trigger deep brain stimulation in customized regions of prefrontal cortex, and you have the plot for a 1990s cyberpunk novel.

But we're getting way ahead of ourselves here...

Dr. Kent Kiehl outside the mobile scanner his group uses to look at the brains of inmates at New Mexico prisons. Credit: Nature News.


The actual study in question used functional MRI to scan the brains of 96 male inmates at two New Mexico state correctional facilities while they performed a cognitive task (Aharoni et al., 2013). The task required responding to a frequent stimulus presented 84% of the time ("X") and inhibiting responses to the rare stimulus ("K").

Fig. S4. (Aharoni et al., 2013). Go/No-Go task.

The major comparison examined brain activity on incorrect responses to "K" (commission errors) vs. correct responses to "X" (hits). This contrast was restricted to a region of interest (ROI) in the dorsal anterior cingulate cortex (dACC), which has been associated with a wide array of cognitive and emotional control functions (Posner et al., 2007).

Results from a separate group of 102 age-matched control participants (mean = 33.9 yrs) from Hartford, CT¹ determined the a priori ROI, with the peak voxel located at coordinates x = −3, y = 24, z = 33 in the center of a 14 mm sphere. One control ROI was chosen in a more ventral and anterior region of medial prefrontal cortex (mPFC) at 0, 51, −6.

The most strongly activated voxel in the offender group for the error vs. hit contrast was remarkably close to the one determined from the independent sample and fell well within the a priori ROI (see blue crosshairs in figure below).

Fig. 2 (modified from Aharoni et al., 2013). (B) Mean hemodynamic response change in offender sample (n = 96) during commission errors vs. correct hits from sagittal (Upper Left), coronal (Right), and axial (Lower Left) orientations. Peak activation located at x = 3, y = 24, z = 33 within the anterior cingulate cortex region of interest (P < 0.00001, FWE).


The dACC has been strongly implicated in error processing (Simons, 2010), and that was no different in the offenders as a group. Other regions significantly activated by commission errors included bilateral inferior frontal cortex/insula, fusiform gyrus, and cerebellum but these were not discussed.

Of greatest interest is whether this dACC activity can predict recidivism. For this the authors did a survival analysis:

First, a Kaplan–Meier survival function was computed to describe the proportion of participants surviving any felony rearrest over the 4-y follow-up period, ignoring the influence of any particular risk factor (Fig. S1). Cox proportional hazards regression was then used to examine (i) the zero-order effects of ACC activity on months to rearrest for any crime, (ii) the shared and unique influence of the ACC and other potential risk factors on months to rearrest for any crime, (iii) for nonviolent crimes, and (iv) the shared and unique influence of the medial prefrontal cortex (mPFC) control region and other potential risk factors on months to rearrest for any crime. ...

... A significant association was found whereby, for every one unit increase in ACC activity, there was a 1.39 (i.e., 1/exp[B]) decrease in the probability of rearrest.

...Meaning that the participants with greater ACC activity were less likely to reoffend. The mPFC  ROI did not show this association. Then a median split divided the offender sample into high ACC and low ACC groups (survival function shown below).

Fig. 1 (Aharoni et al., 2013). Cox survival function showing proportional rearrest survival rates of high (solid green) vs. low (dashed red) ACC response groups for any crime over a 4-y period. Results of this median split analysis were equivalent to that of the parametric model: bootstrapped B = 0.96; SE = 0.40; P < 0.01; 95% CI, 0.29–1.84. The mean survival times to rearrest for the low and high ACC activity groups were 25.27 (2.80) mo and 32.42 (2.73) mo, respectively. The overall probabilities of rearrest were 60% for the low ACC group and 46% for the high ACC group.


So for all felonies (both violent and nonviolent), a substantial percentage of participants were likely to be rearrested within 4 years. The ACC classification scheme would wrongly condemn the 40% of low ACC parolees who did not reoffend, and would miss the 46% of high ACC parolees who did commit crimes after release. When you look at it that way, it's not all that impressive and completely inadmissable as evidence for decision-making purposes. For nonviolent felonies only, the probability of rearrest for high ACC offenders was 31%, compared to 52% for low ACC offenders.

A number of other variables were considered in the regression models (and singly as predictors), including age at release, drug and alcohol use, scores on the Psychopathy Checklist-Revised (PCL-R) (Hare, 2003), and commission errors. The best predictor was still ACC activity, but age and score on Factor 2 of the PCL-R both came in at around p=.05. On the PCL-R, Factor 1 includes callousness and the inability to experience remorse, guilt, and empathy while Factor 2 includes impulsivity, stimulation seeking, and irresponsibility (Ermer et al., 2012). The authors consider low ACC activity to be a manifestation of impulsivity, but it could just as easily be related to a lack of concern about making mistakes (i.e., irresponsibility).


Should functional MRI data be used in parole board hearings?

No, absolutely not. No one is suggesting this, not even Kiehl himself:

Kiehl isn’t convinced either that this type of fMRI test will ever prove useful for assessing the risk to society posed by individual criminals. But his group is collecting more data — lots more — as part of a much larger study in the New Mexico state prisons. “We’ve scanned 3,000 inmates,” he said. “This is just the first 100.”

Nonetheless, I was very impressed that fMRI and behavioral data were collected from 96 prison inmates. That's no easy feat. And the total sample size is now up to a staggering 3,000 inmates!!

Another striking aspect of this paper is that Aharoni and colleagues made their individual subject data available as an Excel spreadsheet that can be downloaded from the PNAS website as supplementary material (Download Dataset_S01, XLSX). It includes the ROI beta weights along with a number of demographic and performance variables.

In my next post, I'll present the results of some analyses that I've conducted, and what they might suggest about behavioral performance in the Go/NoGo task.


Footnote

¹ The median income in Hartford is rather low, and 30% of the population lives in poverty. Although not explicitly stated, these participants might be matched to the criminal offenders for socioeconomic status. The mean years of education was not given for either group. One notable difference, however, is that the control group was 52% female while all the offenders were male.


References

Aharoni, E., Vincent, G., Harenski, C., Calhoun, V., Sinnott-Armstrong, W., Gazzaniga, M., & Kiehl, K. (2013). Neuroprediction of future rearrest Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.1219302110

Ermer E, Cope LM, Nyalakanti PK, Calhoun VD, Kiehl KA. (2012). Aberrant paralimbic gray matter in criminal psychopathy. J Abnorm Psychol. 121(3):649-58.

Kiehl KA, Liddle PF, Hopfinger JB. (2000). Error processing and the rostral anterior cingulate: an event-related fMRI study. Psychophysiology 37(2):216-23.

Posner MI, Rothbart MK, Sheese BE, Tang Y. (2007). The anterior cingulate gyrus and the mechanism of self-regulation. Cogn Affect Behav Neurosci. 7(4):391-5.

Simons RF. (2010). The way of our errors: theme and variations. Psychophysiology 47(1):1-14.