Brain Scans Predict Which Criminals Are Most Likely to Reoffend (Wired)


03.26.13 – 3:40 PM

Photo: Erika Kyte/Getty Images

Brain scans of convicted felons can predict which ones are most likely to get arrested after they get out of prison, scientists have found in a study of 96 male offenders.

“It’s the first time brain scans have been used to predict recidivism,” said neuroscientist Kent Kiehl of the Mind Research Network in Albuquerque, New Mexico, who led the new study. Even so, Kiehl and others caution that the method is nowhere near ready to be used in real-life decisions about sentencing or parole.

Generally speaking, brain scans or other neuromarkers could be useful in the criminal justice system if the benefits in terms of better accuracy outweigh the likely higher costs of the technology compared to conventional pencil-and-paper risk assessments, says Stephen Morse, a legal scholar specializing in criminal law and neuroscience at the University of Pennsylvania. The key questions to ask, Morse says, are: “How much predictive accuracy does the marker add beyond usually less expensive behavioral measures? How subject is it to counter-measures if a subject wishes to ‘defeat’ a scan?”

Those are still open questions with regard to the new method, which Kiehl and colleagues, including postdoctoral fellow Eyal Aharoni, describe in a paper to be published this week in the Proceedings of the National Academy of Sciences.

The test targets impulsivity. In a mobile fMRI scanner the researchers trucked in to two state prisons, they scanned inmates’ brains as they did a simple impulse control task. Inmates were instructed to press a button as quickly as possible whenever they saw the letter X pop up on a screen inside the scanner, but not to press it if they saw the letter K. The task is rigged so that X pops up 84 percent of the time, which predisposes people to hit the button and makes it harder to suppress the impulse to press the button on the rare trials when a K pops up.

Based on previous studies, the researchers focused on the anterior cingulate cortex, one of several brain regions thought to be important for impulse control. Inmates with relatively low activity in the anterior cingulate made more errors on the task, suggesting a correlation with poor impulse control.

They were also more likely to get arrested after they were released. Inmates with relatively low anterior cingulate activity were roughly twice as likely as inmates with high anterior cingulate activity to be rearrested for a felony offense within 4 years of their release, even after controlling for other behavioral and psychological risk factors.

“This is an exciting new finding,” said Essi Viding, a professor of developmental psychopathology at University College London. “Interestingly this brain activity measure appears to be a more robust predictor, in particular of non-violent offending, than psychopathy or drug use scores, which we know to be associated with a risk of reoffending.” However, Viding notes that Kiehl’s team hasn’t yet tried to compare their fMRI test head to head against pencil-and-paper tests specifically designed to assess the risk of recidivism. ”It would be interesting to see how the anterior cingulate cortex activity measure compares against these measures,” she said.

“It’s a great study because it brings neuroimaging into the realm of prediction,” said clinical psychologistDustin Pardini of the University of Pittsburgh. The study’s design is an improvement over previous neuroimaging studies that compared groups of offenders with groups of non-offenders, he says. All the same, he’s skeptical that brain scans could be used to predict the behavior of a given individual. ”In general we’re horrible at predicting human behavior, and I don’t see this as being any different, at least not in the near future.”

Even if the findings hold up in a larger study, there would be limitations, Pardini adds. “In a practical sense, there are just too many ways an offender could get around having an accurate representation of his brain activity taken,” he said. For example, if an offender moves his head while inside the scanner, that would render the scan unreadable. Even more subtle strategies, such as thinking about something unrelated to the task, or making mistakes on purpose, could also thwart the test.

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.”

Kiehl hopes this work will point to new strategies for reducing criminal behavior. If low activity in the anterior cingulate does in fact turn out to be a reliable predictor of recidivism, perhaps therapies that boost activity in this region would improve impulse control and prevent future crimes, Kiehl says. He admits it’s speculative, but his group is already thinking up experiments to test the idea. ”Cognitive exercises is where we’ll start,” he said. “But I wouldn’t rule out pharmaceuticals.”