The Medial Prefrontal Cortex and the Emergence of Self-Conscious Emotion in Adolescence

Method

Task

Participants were led to believe that a small, one-way video camera was embedded in the head coil of the fMRI scanner. They were told that this was a novel technology and that the experimenters wished to test it during their experimental session by cycling through its settings (off, warming up, on) several times. Participants were instructed to passively view the screen and, cued by low-level changes in the screen display, monitor the video camera’s status. The supposed video camera cycled among three phases (Fig. 1a). In the resting baseline condition, the display said “SYSTEM OFF.” In the anticipation condition, the display said “STARTING . . .,” and participants were told that the camera could turn on at any moment. In the evaluation condition, the display said “VIDEO ON,” and a red dot appeared.

OPEN IN VIEWER

Fig. 1.

Experimental paradigm and predicted patterns of results. Participants were led to believe that a one-way video camera was embedded in the head coil of the functional MRI scanner. During different blocks of the experiment (a), participants were told that a camera was projecting an image of their face in real time (evaluation condition; left), warming up but not yet projecting (anticipation condition; middle), or off (baseline condition; right). The graphs show predicted linear (b), quadratic (c), and asymptotic (d) patterns as a function of participants’ age. Each dot represents a participant.

Participants were told that a same-sex peer of similar age would monitor the video feed during the participant’s scan and could see the participant’s face in real time whenever the camera was “on.” They were told the camera operated one way, and thus they should not expect to view a peer. Although there was, in fact, no camera, all participants had completed a separate peer-interaction task immediately prior to this study (Jones et al., 2011), which conveniently made the cover story more believable. Though it cannot be ruled out that the prior task influenced the present findings, the two tasks were held consistent across all participants, and each participant had a short break between the two studies.

The task was structured as a block design that pseudorandomly alternated among the rest, anticipation, and evaluation conditions. Participants saw a total of 12 blocks, 4 of each condition. To reduce predictability, we varied block length between 16 and 38 s. Across the task, participants spent an equal total duration in the anticipation and evaluation conditions (total time per condition: 92 s), and the mean duration of anticipation and evaluation blocks was matched (mean: 23 s). Participants viewed the resting baseline “camera off” stimulus for a total of 126 s in blocks averaging 31.5 s each.

Results

The social-evaluation task elicited self-conscious emotion (e.g., increased ratings of embarrassment; Keltner & Haidt, 1999) and physiological arousal in adolescents. Repeated measures ANOVAs including condition (anticipation, evaluation) and each age predictor indicated that the adolescent-emergent age predictor yielded a significant main effect on embarrassment ratings, F(1, 67) = 6.07, p = .0163; η _p ² = .083; Bonferroni-adjusted critical α = .0167; (see the Supplemental Material and Fig. 2), with the adolescent-specific age predictor yielding trend-level prediction of embarrassment ratings, F(1, 67) = 5.52, p = .022; η _p ² = .076; Bonferroni-adjusted critical α = .0167, see the Supplemental Material. The estimated age of peak embarrassment ratings was 17.2 years.

OPEN IN VIEWER

Fig. 2.

Scatter plot showing embarrassment ratings in the evaluation and anticipation conditions (collapsed) as a function of participants’ age. The fit line was derived from the adolescent-emergent predictor.

Given the significant adolescent-emergent effects, embarrassment ratings were further queried for modulation by condition. The adolescent-emergent age predictor yielded a trend-level effect for the anticipation condition, F(1, 67) = 3.57, p = .063; η _p ² = .051, and a significant effect for the evaluation condition, F(1, 67) = 7.14, p = .009; η _p ² = .096, which suggests consistency in age effects on embarrassment in both conditions but a more robust age difference during the evaluation phase. There was no main effect of condition on embarrassment ratings, F(1, 67) = 0.093, p = .76; Bonferroni-adjusted critical α = .0167.

Skin conductance data yielded a quadratic age-by-time interaction, F(1, 53) = 10.34, p = .002; η _p ² = .163. Post hoc analyses isolating the first half of the experiment indicated an adolescent-specific age effect, which can be described as greater autonomic arousal (less habituation of slope) in adolescence than in adulthood, F(1, 53) = 9.40, p = .003; η _p ² = .151 (see Fig. S1 in the Supplemental Material). The estimated age of peak GSR response was 14.38 years. There were no significant effects of condition or age during the second half of the experiment (ps > .2). Significant adolescent-specific age effects during the first half were evident for the anticipation and evaluation conditions when tested separately—anticipation: F(1, 53) = 4.77, p = .033, η _p ² = .083; evaluation: F(1, 53) = 6.95, p = .011, η _p ² = .12. These results suggest that minimal social-evaluative contexts are sufficient to induce heightened self-conscious emotion and physiological arousal that peaks in midadolescence.

A triad of voxel-wise, whole-brain, mixed-model ANOVAs were conducted with the repeated factor of condition (anticipation, evaluation) and each age predictor serving as a continuous covariate of interest. Guided by the behavioral findings, we focused fMRI analyses on revealing neural activations that were similarly engaged by the anticipation and evaluation conditions and differentially active as a function of the age predictors. A single region of the brain located in the MPFC (x = −13, y = 53, z = 6; 72 3 × 3 × 3 mm voxels; Brodmann’s area 32/10), mean cluster statistic: F(1, 67) = 11.84, p < .05 corrected (Fig. 3a), was significantly related to the adolescent-emergent age predictor. The estimated age of peak in MPFC activity was 15.25 years. The identical single region was also identified in an analysis of adolescent-specific age effects at a corrected threshold of p < .05, albeit smaller in size—mean cluster statistic: F(1, 67) = 10.38, 30 voxels. No regions demonstrated significant linear-age effects at whole-brain corrected thresholds. To summarize, mirroring the levels of experienced embarrassment and arousal, the MPFC demonstrated an elevated response in adolescents during both the anticipation and evaluation conditions that was partially matched in activity strength in young adults. Age differences in MPFC activity are plotted in Figure 3b for descriptive purposes.

OPEN IN VIEWER

Fig. 3.

Age differences in task-based functional brain activity. The statistical map identified functional activity showing heightened engagement of the medial prefrontal cortex (MPFC) through adolescence (relative to childhood) that persists into adulthood. The image threshold was p < .05, corrected for acquisition space. The scatter plot (b) shows MPFC responses in the evaluation and anticipation conditions (collapsed) as a function of participants’ age, for descriptive purposes. The solid fit line was derived from the adolescent-emergent predictor, and the dashed line was derived from the adolescent-specific predictor.

Given the powerful influence of social evaluation on motivated and affective behaviors in adolescence (Steinberg, 2008), we conducted PPI analyses to test the extent to which the MPFC demonstrates differentially selective connectivity during evaluation periods as a function of age (see the Supplemental Material for methodological details). Resultant statistical maps revealed MPFC coupling with the dorsal striatum that significantly fits an asymptotic age pattern—left caudate: x = −8, y = 20, z = 6; 35 3 × 3 × 3 voxels; mean cluster statistic: F(1, 67) = 10.24; p < .05, corrected (Fig. 4). No other activations were observed for this analysis. No regions demonstrated significant linear-age or adolescent-specific (quadratic) effects at whole-brain corrected thresholds. Thus, the transition from childhood through adolescence predicts the emergence of MPFC-striatum coupling during evaluation contexts, a pattern that persists into adulthood.

OPEN IN VIEWER

Fig. 4.

Age differences in task-dependent functional connectivity. Whole-brain connectivity analysis identified selective functional coupling between the medial prefrontal cortex (MPFC) and striatum during evaluation, which was greater in adolescence (relative to childhood) and persists into adulthood. The image threshold was p < .05, corrected for acquisition space.

Discussion

Using a simulated social-evaluation task, we observed that being watched by a peer was sufficient to generate nonlinear differences among children, adolescents, and young adults in self-conscious emotion and related physiological indices of emotional arousal. The nonlinear pattern of embarrassment and skin conductance findings supports the hypothesis that even subtle social-evaluative contexts—and anticipation of them—lead to heightened self-conscious emotion and arousal during adolescence. Self-conscious emotion rose in adolescents (relative to children) and stabilized in adults, whereas arousal (skin conductance) levels showed maximal levels in adolescents.

Numerous factors are thought to converge during adolescence and contribute to the central role of social evaluation in adolescents’ everyday experience. The transition to adolescence is typically marked by a rise in the frequency and intensity of peer interaction; this increase has been documented in both humans and animals (Cairns, Leung, Buchanan, & Cairns, 1995; Primus & Kellogg, 1989) and is often manifested among humans by high rates of digital communication (Wang, Iannotti, & Nansel, 2009). Such a shift in investment likely reflects a heightened motivation for peer acceptance, rendering social-evaluation contexts increasingly salient to adolescents.

In the present study, we evaluated whether adolescents instantiate unique neural response patterns to instances of social evaluation by peers that parallel phenomenological shifts in social sensitivity. We observed that behavioral shifts in adolescent social sensitivity were accompanied by nonlinear changes in MPFC response magnitude and selective MPFC-striatum connectivity, which were significantly higher in adolescents (relative to children) and slightly lower in early adulthood (relative to adolescence).

Though the present study was optimized to detect such age differences across the whole brain (except for the occipital lobe), nonlinear age effects were highly circumscribed to the MPFC. The specificity of effects to the MPFC converges with widely supported theories of the MPFC’s key role in social cognition and emotional valuation processes (Amodio & Frith, 2006; Blakemore, 2008). These findings extend existing accounts of MPFC function to suggest that it maintains persistent representation in even the most basic of social-evaluative contexts—being looked at.

Although the MPFC is frequently conceptualized as specialized for social cognition, emerging theoretical viewpoints have noted common recruitment of the MPFC during contexts that draw on affective valuation and assessment of significance to the self (Krienen, Tu, & Buckner, 2010; Roy et al., 2012)—which are often, but not exclusively, “social.” Indeed, the nonlinear age differences we observed were also evident in anticipatory situations during which participants believed they were not being viewed but would be imminently, which indicates that explicit evaluation was not necessary to invoke adolescent self-consciousness and its neural correlates. Following this view, we propose that MPFC activity in the present study served to incorporate salient contextual cues (in this case, imminent or perceived social evaluation) with emotional valuation processes. Thus, emergent heightened magnitude of MPFC activity in adolescence could result in assignment of heightened emotional value and self-relevance to instances of supposed social evaluation. This conceptualization is consistent with prior findings indicating that MPFC response to positive and negative social feedback is exaggerated in individuals for whom social feedback is particularly salient: individuals with low self-esteem (Somerville, Kelley, & Heatherton, 2010).

Robust MPFC signaling paired with increasing connectivity between MPFC and striatal regions could provide a mechanism by which peer-evaluation contexts come to increasingly modulate adolescent motivated behavior. Not only does social concern serve as a motivating force that drives adolescents to seek out social bonds (Steinberg & Morris, 2001), but adolescents are also more prone to engage in suboptimal choice behaviors (e.g., risky driving) when with peers than when alone (Gardner & Steinberg, 2005). The striatum plays a key role in incorporating motivational, control, and contextual signals to facilitate context-dependent learning and behavior (Alexander, DeLong, & Strick, 1986). Though tentative, the observed pattern of MPFC-striatum connectivity might selectively upregulate motivational signaling, effectively compelling adolescent behavior toward action or approach when being evaluated by peers. Although consistent with extant models of peer influence on adolescent decision making (Lenhart, Ling, Campbell, & Purcell, 2010; Somerville & Casey, 2010; Steinberg, 2008), findings of the current study illuminate the key role played by the MPFC in maintaining a representation of peer evaluation and its emotional qualities, whereas selective connectivity with the striatum may provide a means for integrating signals relevant to social context with motivational systems that govern goal-directed behavior.

It should be noted that the age effects observed for behavioral measures (embarrassment and GSR) partially correspond with the age-related changes observed for the brain-imaging measures. All measures showed a robust influx of response from childhood into adolescence, but measures diverged into young adulthood: GSR levels showed lower magnitude in young adults than in adolescents, whereas the other measures demonstrated a partial or no decline in adults relative to adolescents. Future research with a broader age range is needed to determine whether the divergence of age patterns in young adults is reliable.

A second feature of the reported findings is that physiological, MPFC, and self-conscious emotion demonstrate common maximal responding during adolescence, as indicated by analyses solving for the peak age of response using fit-line equations. Future research may assess whether the particular convergence of measures during adolescence plays a functionally significant role in promoting social sensitivity. There are also subtle and intriguing differences among the peak ages for GSR (14.38 years), MPFC activity (15.25 years), and self-reported embarrassment (17.2 years). Each measure contains its own profile of measurement error, so comparing the particular timing differences among variables should be interpreted cautiously. However, these findings provoke speculation that social sensitivity resonates in physiological and neural indices at an earlier age than when these emotions are most strongly labeled as self-conscious per se. Though even young children are capable of understanding embarrassment (Seidner, Stipek, & Feshbach, 1988), the current findings suggest that the process of attributing such physiological patterns as “embarrassment” might not manifest until later in adolescence, perhaps because of perspective-taking skills that continue to improve throughout adolescence (Crone & Dahl, 2012; Dumontheil, Apperly, & Blakemore, 2010) that may scaffold simulation of negative consequences of potential social transgressions that serve as a foundation of embarrassment (Keltner & Haidt, 1999). A complementary explanation is that though social evaluations are arousing across different stages of adolescence, they might be experienced as less specifically embarrassing in early and midadolescence relative to late adolescence. Comprehensive studies with broader emotion measures will be needed to address these possibilities.

In conclusion, waiting to be looked at and believing one is being looked at were sufficient to induce nonlinear changes in self-conscious emotion and related physiological indices among children, adolescents, and young adults. Nonlinear differences in response in the MPFC, and MPFC-striatum connectivity, parallel this behavioral shift, and we propose that these response differences influence adolescent social sensitivity. The functional properties of the MPFC are likely to be influenced by continued structural maturation (Mills, Lalonde, Clasen, Giedd, & Blakemore, 2012; Shaw et al., 2008) and subcortical and cortical connections (e.g., Asato, Terwilliger, Woo, & Luna, 2010) during adolescence. That said, future work will be needed to identify biological and experiential mechanisms that give rise to the functional differences illuminated by the present study. Together with other findings, the results of this study bridge examinations of psychosocial development and neurodevelopmental science to show how the emergent features of the adolescent social life can exert such a powerful influence over motivation, emotion, and well-being.