Premotor functional connectivity predicts impulsivity in juvenile offenders

Abstract

Teenagers are often impulsive. In some cases this is a phase of normal development; in other cases impulsivity contributes to criminal behavior. Using functional magnetic resonance imaging, we examined resting-state functional connectivity among brain systems and behavioral measures of impulsivity in 107 juveniles incarcerated in a high-security facility. In less-impulsive juveniles and normal controls, motor planning regions were correlated with brain networks associated with spatial attention and executive control. In more-impulsive juveniles, these same regions correlated with the default-mode network, a constellation of brain areas associated with spontaneous, unconstrained, self-referential cognition. The strength of these brain-behavior relationships was sufficient to predict impulsivity scores at the individual level. Our data suggest that increased functional connectivity of motor-planning regions with networks subserving unconstrained, self-referential cognition, rather than those subserving executive control, heightens the predisposition to impulsive behavior in juvenile offenders. To further explore the relationship between impulsivity and neural development, we studied functional connectivity in the same motor-planning regions in 95 typically developing individuals across a wide age span. The change in functional connectivity with age mirrored that of impulsivity: younger subjects tended to exhibit functional connectivity similar to the more-impulsive incarcerated juveniles, whereas older subjects exhibited a less-impulsive pattern. This observation suggests that impulsivity in the offender population is a consequence of a delay in typical development, rather than a distinct abnormality.

Fig. 1.

Premotor functional connectivity and impulsivity. (A) Bilateral dorso-rostral premotor regions (PMdr). These two regions showed the greatest correlation between resting state functional connectivity (RS-fcMRI) and impulsivity in the juvenile offender cohort. (B) Predicted vs. measured impulsivity evaluated in juvenile offenders using the leave-one-out procedure. Impulsivity (arbitrarily scaled; SI Text) was predicted on the basis of bilateral PMdr RS-fcMRI. The correlation between predicted and measured impulsivity is highly significant (n = 106, r = 0.39; P < 0.001). (C) Dependence of prediction accuracy on selection of regions of interest (ROIs) used to compute the prediction. The IDEA algorithm identifies many regions sorted by decreasing power to discriminate high vs. low impulsivity. The plot shows the effect of including progressively more ROIs from the sorted list. Predictive accuracy peaked with two regions (left and right PMdr), suggesting that the relation of RS-fcMRI to impulsivity is highly focal.

Fig. 2.

(A) PMdr functional connectivity in typical young adults. Images are thresholded at a value of r = 0.1. PMdr is positively correlated with the dorsal-attention network and the executive-control network; it is anticorrelated with the default-mode network. (B) Correlation between PMdr functional connectivity and impulsivity in juvenile offenders. PMdr functional connectivity in less-impulsive juveniles is similar to that of adults. However, in more-impulsive individuals its relationships reverse, such that it is positively correlated with the default-mode network and anticorrelated with the dorsal-attention and executive-control networks.

Fig. 3.

Relationship between PMdr functional connectivity and chronological age. In younger subjects, PMdr is correlated with the default-mode network and anticorrelated with the dorsal-attention and executive-control networks. In older subjects, these relationships are reversed. Thus, with respect to PMdr RS-fcMRI, the effect of youth in typically developing individuals is strikingly similar to the effect of measured impulsivity in the juvenile offenders.