Cognitive control in adolescence: neural underpinnings and relation to self-report behaviors

Abstract

Background: Adolescence is commonly characterized by impulsivity, poor decision-making, and lack of foresight. However, the developmental neural underpinnings of these characteristics are not well established.

Methodology/principal findings: To test the hypothesis that these adolescent behaviors are linked to under-developed proactive control mechanisms, the present study employed a hybrid block/event-related functional Magnetic Resonance Imaging (fMRI) Stroop paradigm combined with self-report questionnaires in a large sample of adolescents and adults, ranging in age from 14 to 25. Compared to adults, adolescents under-activated a set of brain regions implicated in proactive top-down control across task blocks comprised of difficult and easy trials. Moreover, the magnitude of lateral prefrontal activity in adolescents predicted self-report measures of impulse control, foresight, and resistance to peer pressure. Consistent with reactive compensatory mechanisms to reduced proactive control, older adolescents exhibited elevated transient activity in regions implicated in response-related interference resolution.

Conclusions/significance: Collectively, these results suggest that maturation of cognitive control may be partly mediated by earlier development of neural systems supporting reactive control and delayed development of systems supporting proactive control. Importantly, the development of these mechanisms is associated with cognitive control in real-life behaviors.

Figure 1. Between-group differences in blocked Stroop activity reveal adolescents under-activate lateral prefrontal cortex compared to adults.

Significant clusters of activation for the blocked Stroop contrast of Incongruent blocks - Neutral blocks are displayed separately for A. Adolescents and B. Adults. Voxels in red indicate greater activity for I blocks compared to N blocks and voxels in blue indicate greater activity for N blocks compared to I blocks. While both groups of participants activated a network of fronto-parietal regions implicated in cognitive control, C. adults exhibited significantly greater activity in lateral prefrontal regions, dorsal medial prefrontal cortex, and temporal-occipital regions. Note: A voxel-wise threshold of p<0.005 and a cluster-wise threshold of >103 contiguous voxels were applied to the statistical maps using Monte Carlo permutation simulations (AlphaSim). Results are projected onto a surface template (Caret Software) .

Figure 2. Relationships between blocked Stroop activity and age.

Percent signal change for the contrast of Incongruent (I) - Neutral (N) blocks was extracted from the lateral prefrontal cluster highlighted in the box in Figure 1C. This region corresponds approximately to IFJ/pDLPFC and is shown in a sagittal slice in panel A, where the percent signal change within this region is plotted across age, yielding an inverted J-shaped function. B. The magnitude of activity in the same IFJ/pDLPFC region is plotted for the two adolescent age groups as well as the adult group. Note: *p<0.05; **p<0.01. Error bars reflect standard error of the mean.

Figure 3. Between-group differences in trial-related Stroop activity reveal heterogeneity within the adolescent group.

Trial-related (transient) Stroop contrasts were examined by comparing incongruent (i) and neutral (n_i) trials within incongruent blocks. Since no differences in activity were observed between adults and adolescents, adolescents were subsequently divided into two groups of A. 15 14–15 year olds and B. 17 16–17 year olds. Voxels in red indicate greater activity on i trials compared to n_i trials and voxels in blue indicate greater activity on n_i trials compared to i trials. C. Between-group differences revealed older adolescents exhibited elevated BOLD activity in a number of response-related regions including pre-SMA, ACC, lateral pre-motor, motor, and somatosensory areas. Note: A voxel-wise threshold of p<0.005 and a cluster-wise threshold of >103 contiguous voxels were applied to the statistical maps using Monte Carlo permutation simulations (AlphaSim). Results are projected onto a surface template (Caret Software) .

Figure 4. Trial-related group differences in the ACC/pre-SMA.

Percent signal change for the contrast between incongruent (i) and neutral (n_i) trials within incongruent blocks was extracted from the anterior cingulate cortex (ACC)/ pre-supplementary motor area (pre-SMA) cluster shown in a sagittal slice [see box in Figure 3 ]. Activity within this region is plotted separately for young adolescents (14–15 year olds), older adolescents (16–17 year olds), and adults (18–25 year olds). Older adolescents activated the midline cluster significantly more than younger adolescents and numerically (but non-significantly) more than adults. Note: *p<0.05; **p<0.01. Error bars reflect standard error of the mean.

Figure 5. Relationships between blocked Stroop activity and self-report measures of cognitive/social control.

Percent signal change reflecting the contrast of Incongruent (I) vs. Neutral (N) blocks was extracted from the IFJ/pDLPFC seed that exhibited group differences in Figure 1C (see box in Figure 1C and inset in current figure). Activity within this region was correlated with self-report measures of cognitive/social control separately for A. adolescents and B. adults. The cognitive/social control composite represents participants' average z-scores from individual questionnaires assessing impulse control, planning, and resistance to peer pressure. A. In adolescents, a significant positive relationship between the two variables was observed such that adolescents who activate the a priori region to a greater degree reported greater cognitive/social control. B. The relationship in adults, although positive, was not significant. Note: *p<0.05.