Temptations of friends: adolescents' neural and behavioral responses to best friends predict risky behavior

Abstract

Adolescents are notorious for engaging in risky, reward-motivated behavior, and this behavior occurs most often in response to social reward, typically in the form of peer contexts involving intense positive affect. A combination of greater neural and behavioral sensitivity to peer positive affect may characterize adolescents who are especially likely to engage in risky behaviors. To test this hypothesis, we examined 50 adolescents' reciprocal positive affect and neural response to a personally relevant, ecologically valid pleasant stimulus: positive affect expressed by their best friend during a conversation about past and future rewarding mutual experiences. Participants were typically developing community adolescents (age 14-18 years, 48.6% female), and risky behavior was defined as a factor including domains such as substance use, sexual behavior and suicidality. Adolescents who engaged in more real-life risk-taking behavior exhibited either a combination of high reciprocal positive affect behavior and high response in the left ventrolateral prefrontal cortex-a region associated with impulsive sensation-seeking-or the opposite combination. Behavioral and neural sensitivity to peer influence could combine to contribute to pathways from peer influence to risky behavior, with implications for healthy development.

Fig. 1.

Design of the Best Friend fMRI task, in which adolescents view video clips of positive and neutral affect displays by a same-gender unfamiliar adolescent or by their own same-gender best friend. Video clips from the best friends were drawn from a laboratory-based dyadic interaction in which the adolescents discussed their most pleasant shared experience.

Fig. 2.

Adolescents’ neural response to videos of their best friends’ positive affect relative to neutral affect, masked for meta-analytic findings on regions activated during social processing.

Fig. 3.

Illustration of conditional effects of adolescents’ contingent positive affect (PA) behavior on their real-life risky behavior at high and low levels of left VLPFC response to the Best Friend Task, which involves peer social reward. The lines’ slopes (−0.07, 0.80) reflect the levels of VLPFC response at which contingent positive affect behavior and risky behavior become negatively and positively correlated, respectively. These illustrate the high and low values beyond which VLPFC is a statistical moderator, with better accuracy than the convention of presenting values reflecting ±1 s.d. of the mean value for the moderator. The larger number of cases for the combination of higher positive affect and higher VLPFC (i.e. higher region of significance; red triangles) than for lower positive affect and lower VLPFC (i.e. lower region of significance; blue circles) likely reflects the combinations evident in a community sample of typically developing adolescents. In other words, having fewer cases in the lower region of significance does not indicate that the interaction effect is driven by outliers. Scatter points represent actual data values. One case was removed because analyses indicated that it was an outlier for VLPFC response.