Rewarding safe choices in peer contexts: Adolescent brain activity during decision making

Highlights

• Rewards for safe choices decrease risky decision making in peer contexts.

• Reward+peer conditions creates motivational conflict to approach/avoid risk.

• Cognitive-emotional hubs in the brain were more active in this condition.

• Neural processing of motivational conflict may be largely “under the surface”.

Introduction

Adolescence is marked by heightened sensitivity to reward (Braams, van Duijvenvoorde, Peper, & Crone, 2015; Chein, Albert, O’Brien, Uckert, & Steinberg, 2011; Galvan, 2010; Luciana, Wahlstrom, Porter, & Collins, 2012; Van Leijenhorst et al., 2010), and during adolescence peers serve as powerful social reward cues to take risks (Blakemore & Robbins, 2012; Chein et al., 2011; Shulman et al., 2016; Smith, Steinberg, Strang, & Chein, 2015). These trends are evident in both demographic and experimental research, indicating that adolescent drivers with a teen passenger are more likely to drive aggressively, be involved in accidents, and make riskier decisions (Allen & Brown, 2008; Chassin et al., 2004; Chein et al., 2011; Curry, Mirman, Kallan, Winston, & Durbin, 2012). This peer effect extends to increased likelihood of substance use and delinquent behaviors in social situations as well (Brown et al., 2009; Casey & Jones, 2010; Chassin et al., 2004; Samek, Goodman, Erath, McGue, & Iacono, 2016). While the peer effect is a robust and well documented phenomenon, the underlying mechanisms and limiting conditions are still poorly understood. The specific goal of this study is to examine effects of peer and reward contexts on adolescent brain activity and behavior to better understand the neural mechanisms underlying decision making in adolescence.

Increased peer influence in combination with neurodevelopmental differences in brain regions related to cognitive control and approach motivation are thought to best explain behavioral trends toward risk-taking in adolescence (Albert, Chein, & Steinberg, 2013; Casey & Jones, 2010; Casey, Jones, & Hare, 2008; Steinberg, 2010). The relative influences of approach-reward and cognitive control systems feature heavily in current theories of adolescent risky decision making and behavior, which posit that in contexts of potential reward, adolescents’ heightened activity in approach-reward areas often “outweighs” or preempts countervailing activity of cognitive control areas and leads to risky decision making and behavior, especially in peer contexts (Shulman et al., 2016; Steinberg, 2010).

While heightened reward sensitivity has typically been thought of as a contributor to risk-taking behavior, there is also evidence to suggest that reward sensitivity can be harnessed and leveraged to augment cognitive control (Casey & Jones, 2010; Telzer, 2016). A few innovative studies have found that when inhibitory control is rewarded (i.e., through performance-based rewards) adolescents exhibit performance similar to adults, and these behavioral effects are paralleled by greater activity in regions of the brain that process reward (Geier & Luna, 2012; Geier, Terwilliger, Teslovich, Velanova, & Luna, 2010; Padmanabhan, Geier, Ordaz, Teslovich, & Luna, 2011; Teslovich et al., 2014). These somewhat counterintuitive findings suggest that extrinsic rewards may be especially useful during adolescence to promote coordination between approach and cognitive control systems to optimize behavioral performance. Studies of age differences in performance-reward contingency effects have been limited to measures of cognitive control, such as antisaccade tasks, and have not considered effects of peer contexts. Whether performance-based rewards can mitigate risky decision making during peer observation and affect the underlying neurocircuitry of risky decision making is, to our knowledge, untested. The experiment in the current study addresses this question by pitting conflicting motivational contexts against one another in a driving decision making task. Consistent with prior literature, peer observation should provide contextual motivation toward riskier decisions while performance-based rewards favoring safe choices should provide the opposite motivation. A novel contribution of the current study is in determining if performance-based rewards favoring safe decision making will mitigate the risk enhancing effects of a peer observation context. It remains to be determined whether performance-based rewards affect adolescents’ neural activity during decision-making or modifies neural activity in peer contexts that are related to increased risk-taking behaviors. Finally, consideration of neurocircuitry and connectivity between brain regions relevant to decision making processes has been emphasized as a critical direction for future research (Casey, Galván, & Somerville, 2016; Pfeifer & Allen, 2012). Thus, another aim of this study was to evaluate functional connectivity between relevant brain regions during peer and performance-based reward contexts that create a motivational conflict.

Regions of interest (ROIs) targeted in this study are known to have varying degrees of structural and functional connectivity and are implicated as important components of a functional corticostriatal network in decision making processes (Everitt & Robbins, 2005; Haber, 2016; Jarbo & Verstynen, 2015). The striatum is traditionally divided into ventral (VS; including the nucleus accumbens) and dorsal (DS; including the caudate and putamen) regions, which are highly interconnected with one another and functionally integrated in decision making and behavior (Hart, Leung, & Balleine, 2014). The VS is most strongly associated with reward processing while the DS is most strongly linked to working memory and motor responses (caudate and putamen, respectively). The striatum has structural connections to the anterior insula (Cho et al., 2013), a region that plays important roles in stimulus salience and attentional regulation (Menon & Uddin, 2010), as well as the cingulate cortex, which is implicated in conflict detection and cognitive control required for conflict resolution (Shenhav, Cohen, & Botvinick, 2016). Both the insula and cingulate cortex likely serve as key hubs or mediators in top-down and bottom-up streams of a decision making corticostriatal circuit. Stimulus valuation and reevaluation from experience are also important in decision making processes, and both the medial prefrontal (MPFC) and orbitofrontal (OFC) cortices have structural connections to the striatum to fulfill these tasks (Cauda et al., 2011; Van Den Bos, Cohen, Kahnt, & Crone, 2012). Finally, higher level decision making processes, including goal construction and strategy, involve links between the dorsomedial and dorsolateral prefrontal cortices (DPFC) and the striatum, the caudate in particular (Haber, 2016). Such a series of structural and functional connections provides a plausible model for a circuit of decision making involving cognitive control and reward networks in motivational contexts (Haber, 2016; Niendam et al., 2012).

Our hypotheses focused on contrasts involving the combination of peer observation and performance-based reward favoring safe decision making, the experimental condition that should create a conflict between motivational states to approach or avoid risk. We posited that the peer and performance-based reward condition (hereafter termed PR) would, relative to a peer observation condition absent performance-based reward (Pr), a performance-based reward condition absent peer observation (pR), or a control condition (pr), result in increased activity in cognitive control areas of the brain and greater functional connectivity between reward, cognitive control, and decision making areas that would be suggestive of a coordinated network working to resolve motivational conflict. Further, we hypothesized that PR condition effects on brain activity in and connectivity of cognitive control areas would be most apparent when contrasting safe against risky decisions. We also expected to see decreased risky decision making in the PR condition, relative to Pr or pr conditions, but that the reward-only pR condition should have the fewest risky decisions.