Mothers Know Best: Redirecting Adolescent Reward Sensitivity PDF

Social Cognitive and Affective Neuroscience Advance Access published April 13, 2015 doi:10.1093/scan/nsv026 SCAN (2015) 1 of 9 Mothers know best: redirecti...

Social Cognitive and Affective Neuroscience Advance Access published April 13, 2015 doi:10.1093/scan/nsv026 SCAN (2015) 1 of 9 Mothers know best: redirecting adolescent reward sensitivity toward safe behavior during risk taking Eva H. Telzer,1,2 Nicholas T. Ichien,1 and Yang Qu1 1 Department of Psychology and 2 Beckman Institute for Advanced Science and Technology, University of Illinois, Urbana-Champaign, IL, USA Despite being one of the healthiest developmental periods, morbidity and mortality rates increase dramatically during adolescence, largely due to preventable, risky behaviors. Heightened reward sensitivity, coupled with ineffective cognitive control, has been proposed to underlie adolescents risk taking. In this study, we test whether reward sensitivity can be redirected to promote safe behavior. Adolescents completed a risk-taking task in the presence of their mother and alone during fMRI. Adolescents demonstrated reduced risk-taking behavior when their mothers were present compared with alone, which was associated with greater recruitment of the ventrolateral prefrontal cortex (VLPFC) when making safe decisions, decreased activation in the ventral striatum following risky decisions and greater functional coupling between the ventral striatum and VLPFC when making safe decisions. Importantly, the very same neural circuitry (i.e. ventral striatum) that has been linked to greater risk-taking can also be redirected toward thoughtful, more deliberative and safe decisions. Keywords: adolescence; risk taking; rewards; fMRI; influence; family Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 INTRODUCTION risky behavior is less rewarding) and redirected to promote cognitive Despite being one of the healthiest developmental periods, morbidity control? and mortality rates increase 300% from childhood to adolescence Because neural regions involved in motivation and cognitive pro- (National Center for Health Statistics, 2013; Kann et al., 2014) with cesses undergo significant reorganization during adolescence (Nelson over 70% of adolescent deaths each year due to risk-taking behaviors, et al., 2005), the adolescent brain is thought to be highly flexible and such as reckless driving (Centers for Disease Control and Prevention, malleable (Crone and Dahl, 2012) and therefore particularly sensitive 2012). Evidence from developmental neuroscience suggests that risk- to social influences. While prior work has focused on the social con- taking behavior increases during adolescence partly due to relatively texts which may increase risk taking, for instance peer presence (Chein early functional development of brain regions involved in reward sen- et al., 2011), social contexts can also decrease risk taking, such as par- sitivity (e.g. ventral striatum) compared with more protracted func- ental presence. Indeed, parents represent one of the most direct and tional maturation of brain regions supporting cognitive control (e.g. proximal sources of influence over teenagers. Although parents tend to the prefrontal cortex [PFC]; Somerville et al., 2010). Heightened adjust their supervisory practices to allow their adolescent children to reward sensitivity, coupled with ineffective cognitive control, has lar- be more independent, adolescents tend to engage in more maladaptive, gely been thought to underlie adolescents’ orientation toward risky risky behaviors during unsupervised time (Richardson et al., 1993; behavior (Steinberg, 2008; Casey et al., 2011). While most prior Beck et al., 2001; Borawski et al., 2003), highlighting the important work has focused on how this heightened reward sensitivity creates role of parents in decreasing adolescent risk taking. While parents may vulnerabilities for adolescents, leading to increases in risk taking, scho- decrease adolescent risk-taking merely by serving as gatekeepers and lars have begun to question these dual systems models of adolescent limiting adolescents’ opportunities to make poor decisions, we propose brain development (Crone and Dahl, 2012; Pfeifer and Allen, 2012). that parents may actually change the ways in which adolescents think Recent evidence suggests that rewards can play an adaptive role, facil- and reason about risks during both active/deliberative as well as more itating cognitive control. For example, rewarding cues (e.g. monetary automatic decision-making processes. rewards; happy faces) enhance inhibitory control (Hardin et al., 2009; Parental influence on adolescent risk taking may occur in several Kohls et al., 2009; Teslovich et al., 2014), perhaps by serving to mo- ways. First, due to the relative immaturity of their prefrontal cortex, tivate adolescents to engage in effortful control. Moreover, the ventral adolescents may not yet have the cognitive resources to effectively striatum may itself serve a regulatory function (Pfeifer et al., 2011). avoid risky behaviors and so parents may play an important scaffolding Thus, heightened reward sensitivity, which has largely been proposed role, helping their children to regulate their behaviors and engage in to underlie adolescent risk taking (Steinberg, 2008; Casey et al., 2011), more adaptive decision making. Thus, parents may facilitate improved may also shape adolescents’ motivations to engage in greater cognitive cognitive control and thereby reduce their adolescents’ risk taking. control. In this study, we test whether adolescents’ heightened reward Second, parents may reduce the affective and rewarding nature of sensitivity can be redirected to promote cognitive control and safe engaging in risky behavior. That is, risk taking may be comparatively behavior during risk-taking. In other words, can the heightened less rewarding and more aversive in the presence of family, and so the typical heightened reward response following risky behaviors may be reward response we typically see during risk taking be reduced (i.e. attenuated. Lastly, parents may serve to redirect adolescents’ reward sensitivity toward safe behavior, such that reward and cognitive control Received 6 August 2014; Revised 14 February 2015; Accepted 4 March 2015 systems interact to promote safe choices. Most prior work has focused on the negative contexts (e.g. peer presence) in which We greatly appreciate the assistance of the Biomedical Imaging Center. This research was supported by a grant heightened reward sensitivity and immature cognitive control interact from the National Science Foundation (SES 1459719; Telzer) and generous funds from the Department of Psychology at the University of Illinois. to lead to maladaptive behavior (Chein et al., 2011). However, several Correspondence should be addressed to Eva H. Telzer, 603 East Daniel St., Champaign, IL 61820, USA. E-mail: studies have demonstrated that rewards can also lead to improvements [email protected]. in cognitive control through bottom-up processes that increase ß The Author (2015). Published by Oxford University Press. For Permissions, please email: [email protected] 2 of 9 SCAN (2015) E. H.Telzer et al. activation in brain regions involved in regulation (e.g. ventrolateral monetary incentive was used to encourage risk taking (Chein et al., prefrontal cortex [VLPFC]; Geier et al., 2010; Smith et al., 2011). We 2011). therefore examined whether parental presence functions through Adolescents completed two rounds of the stoplight task during two bottom-up processing, whereby the parent elicits a reward-related re- functional brain scans. Similar to the manipulation used by Steinberg sponse that boosts adolescents’ motivation to regulate their and colleagues with peers (Chein et al., 2011), adolescents played one behavior. Thus, we tested whether parents increase neural coupling round of the game while their mother was watching and one round in between the ventral striatum and PFC to facilitate safe behavior which they were alone. During the mother condition, the participant’s during risk taking. mother was instructed to speak into the intercom naturally and au- thentically, informing their child that they would be watching during METHODS the whole scan. Instructions were given to avoid any comments that might explicitly or intentionally bias their child’s behavior. During the Participants alone condition, the researcher spoke into the microphone and in- Thirty adolescent-mother dyads participated. Two participants were formed the participant that nobody would be watching during the excluded due to excessive movement (>3 mm), two participants did round. Run order was counterbalanced across participants. not complete the scan, and one participant’s behavioral responses were During each round of the task, participants were presented with 26 not recorded due to technical issues. Our final sample included 25 intersections. The probability of crashing was kept constant at 30% fourteen-year-old adolescents and their mother (adolescents: (i.e. eight intersections out of the 26 total intersections had cars ap- Mage ¼ 14.43 years; 15 males; mothers: Mage ¼ 43.89 years). proaching on the cross street, resulting in a crash if the participant Adolescent participants were predominantly from European- made a ‘go’ decision), but this was not explicitly revealed to partici- American (n ¼ 18) or African-American (n ¼ 5) backgrounds with pants. The timing of traffic signals and the presence of a car on the Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 the remaining from Asian-American (n ¼ 1) or Central-American cross street varied so as to be unpredictable by the participant and to (n ¼ 1) backgrounds. Mothers reported their highest levels of educa- introduce variable ITIs. At the end of each round, participants were tion as high school (n ¼ 1), some college (n ¼ 8), college (n ¼ 3) and presented with their overall time and the number of crashes. Due to graduate, medical or law school (n ¼ 13). All participants provided learning effects, such that participants show higher behavioral risk written assent and consent in accordance with the Institutional performance during initial rounds of the task followed by decreased Review Board. and more stable risk patterns thereafter (Peake et al., 2013), partici- pants played two practice rounds of the task, each with 26 intersec- Risk-taking task tions, prior to entering the scanner. The stoplight task measures risk taking at the behavioral and neural Behaviorally, we examined the percent of decisions that are risky and level (Gardner and Steinberg, 2005; Chein et al., 2011). Adolescents safe and examined differences based on maternal presence. In terms of completed a simulated driving course in which they encountered a reaction time differences, the task is not optimized to collect reaction number of stoplights and had to decide whether to ‘stop’ or ‘go’ by time. Because of the long inter-trial interval (12 s between stoplights), pressing one of two buttons (see Figure 1). A decision to ‘go’ through participants often make their decision prior to seeing the yellow light. the intersection is the fastest option, but participants risk the possibil- Although they are instructed to make the decision and subsequent ity of crashing, which causes a 6 s delay. If they choose to ‘stop’, par- button response only after seeing the yellow light, many participants ticipants do not risk crashing, but it results in a short, 3 s delay. press the button before the yellow light appears. Such button responses Participants were told that the goal is to get through the driving are not recorded. Unfortunately, this precludes our ability to examine course in as short of a time as possible to win more money. The reaction time differences. Fig. 1 The stoplight task. Mothers know best SCAN (2015) 3 of 9 fMRI data acquisition and analysis subject contrasts were then submitted to random-effects, group-level fMRI data acquisition analyses. The following analyses were run at each voxel across the Imaging data were collected using a 3 Tesla Siemens Trio MRI scanner. entire brain volume: StopMother-StopAlone, GoMother-GoAlone The stoplight task included T2*-weighted echoplanar images (EPI) and PassMother-PassAlone. Because of the low frequency of crash (slice thickness ¼ 3 mm; 38 slices; TR ¼ 2 s; TE ¼ 25 ms; ma- events (i.e. there were only eight possible crash events per run), we trix ¼ 92  92; FOV ¼ 230 mm; voxel size 2.5  2.5  3 mm3). had restricted power to test for CrashMother-CrashAlone. We present Structural scans consisted of a T2*weighted, matched-bandwidth this contrast for the 17 participants who had at least four crash trials (MBW), high-resolution, anatomical scan (TR ¼ 4 s; TE ¼ 64 ms; per condition. FOV ¼ 230; matrix ¼ 192  192; slice thickness ¼ 3 mm; 38 slices) We conducted psychophysiological interaction (PPI) analyses and a T1* magnetization-prepared rapid-acquisition gradient echo (Friston et al., 1997) to examine functional coupling between the ven- (MPRAGE; TR ¼ 1.9 s; TE ¼ 2.3 ms; FOV ¼ 230; matrix ¼ 256  256; tral striatum and cognitive control regions. We used the ventral stri- sagittal plane; slice thickness ¼ 1 mm; 192 slices). The orientation atum as the seed region, as the striatum has been consistently linked for the MBW and EPI scans was oblique axial to maximize brain with the experience of rewards. The striatum was defined structurally coverage. using the WFUpickatlas (Tzourio-Mazoyer, et al., 2002; Maldjian et al., 2003, 2004). PPI analyses were run using a generalized form of con- text-dependent PPI. Specifically, the automated gPPI toolbox in SPM fMRI data preprocessing and analysis (gPPI; McLaren et al., 2012) was used to i) extract the deconvolved Neuroimaging data were preprocessed and analyzed using Statistical times series from the ventral striatum region of interest (ROI) for each Parametric Mapping (SPM8; Wellcome Department of Cognitive participant to create the physiological variables; ii) convolve each trial Neurology, Institute of Neurology, London, UK). Preprocessing for type with the canonical HRF, creating the psychological regressor; and Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 each participant’s images included spatial realignment to correct for iii) multiply the time series from the psychological regressors with the head motion (no participant exceeded 2 mm of maximum image-to- physiological variable to create the PPI interaction terms. This inter- image motion in any direction). The realigned functional data were action term identified regions that covaried in a task-dependent coregistered to the high resolution MPRAGE, which was then seg- manner with the striatum. For the first-level model, one regressor mented into cerebrospinal fluid, grey matter and white matter. The representing the deconvolved BOLD signal was included alongside normalization transformation matrix from the segmentation step was each psychological and PPI interaction terms for each condition type then applied to the functional and T2 structural images, thus trans- to create a gPPI model. These first levels models created gPPI models forming them into standard stereotactic space as defined by the for the following contrasts: Stop Decisions, Go Decisions, and Pass Montreal Neurological Institute and the International Consortium Outcomes for the Mother Present and Alone conditions separately. for Brain Mapping. The normalized functional data were smoothed At the group level, we conducted random effect analyses to compare using an 8 mm Gaussian kernel, full-width-at-half maximum, to in- functional coupling between conditions of interest. crease the signal-to-noise ratio. To correct for multiple comparisons, we conducted a Monte Carlo Statistical analyses were performed using the general linear model simulation implemented using 3dClustSim in the software package (GLM) in SPM8. Each trial was convolved with the canonical hemo- AFNI (Ward, 2000). We used our group-level brain mask, which dynamic response function. High-pass temporal filtering with a cutoff included only gray matter. Results of the simulation indicated a of 128 s was applied to remove low-frequency drift in the time series. voxel-wise threshold of P < 0.005 combined with a minimum cluster Serial autocorrelations were estimated with a restricted maximum like- size of 35 voxels for the whole brain, corresponding to P < 0.05, false lihood algorithm with an autoregressive model order of 1. wise error (FWE) corrected. We used the MarsBaR toolbox to extract In each participant’s fixed-effects analysis, a GLM was created with parameter estimates from significant clusters in the group-level four regressors of interest each for the alone condition and mother analyses. To ensure that the neural effects are not driven by differences condition, modeled as events: two decision regressors (Stop and Go) in the number of trials in the analyses (e.g. more stop decisions and two outcome regressors (Crash and Pass). In addition, the wait when mothers are present then when alone), all fMRI analyses time after stop decisions were modeled as well as the final ‘Game Over’ covary for the number of decisions to stop during Alone and period at the end of each run, to remove these from the implicit Mother conditions. baseline. This resulted in 12 conditions, six each for alone and mother: StopMother, GoMother, PassMother, CrashMother, RESULTS WaitMother, GameOverMother, StopAlone, GoAlone, PassAlone, Behavioral results CrashAlone, WaitAlone and GameOverAlone. Adolescents made significantly more risky decisions (i.e. ‘go’) when Because the task was self-paced, the duration of the decision trials alone than when their mothers were present [F(1, 24) ¼ 3.9, P < 0.001; (stop or go) represented the time from which the traffic light appeared Figure 2]. In prior studies using this same task, the presence of peers until the participant made a response, and the duration for the out- increased risk-taking behavior of adolescent participants (Gardner and come (pass or crash) was 1 s. The onset of the crash event corres- Steinberg, 2005; Chein et al., 2011). In contrast, we find that the pres- ponded to another car crashing into the participant’s car. The pass ence of mothers reduces risk-taking behavior. and wait events had no specific onset time. However, because the crash events happened at most 2 s after the yellow light, we modeled the pass and wait events as being 2 s after the yellow light, the point at which the fMRI results outcome of the risky decision was clear. Each was modeled with a 1 s Main effects duration. Null events, consisting of the jittered intertrial intervals, were Our first fMRI analyses collapsed across conditions to test what regions not explicitly modeled and therefore constituted an implicit baseline. were activated when making safe and risky choices irrespective of The parameter estimates resulting from the GLM were used to create maternal presence. In whole brain t-tests, we examined neural activa- linear contrast images comparing each of four event conditions tion during Stop decisions, Go decisions and Passes. When adolescents (decisions: Go, Stop; outcomes: Crash, Pass) during the Alone round made decisions to Stop (compared with baseline), they did not show to the corresponding conditions in the Mother round. The individual heightened activation in any region. When making Go decisions 4 of 9 SCAN (2015) E. H.Telzer et al. Fig. 2 Behavioral performance. Adolescents made significantly fewer risky choices when their mothers were present than when alone. Error bars represent SEM. (compared with baseline), adolescents showed heightened activation in the ventral striatum, insula, dACC and SMA. In addition, we com- Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 pared activation between Stop and Go decisions collapsed across conditions. Adolescents demonstrated greater activation in the insula, dACC and midbrain when making Go decisions relative to Stop decisions. The only region that was more active during Stop compared with Go decisions was the precentral gyrus. Finally, when adolescents successfully passed through an intersection without crashing following a risky decision, they demonstrated heightened ac- tivation in a large cluster encompassing the superior frontal gyrus, Fig. 3 (a) Neural activation during decisions to ‘stop’ when adolescents’ mothers were present precuneous and extending into the left temporoparietal junction compared with alone. For descriptive purposes, parameters estimates of signal intensity were (TPJ) (Table 1). extracted from the VLPFC and MPFC cluster separately for Stop decisions when Alone and Mother Present (relative to baseline). Error bars represent SEM. Interactions between Alone and Mother present condition Next, we examined differences in neural activation during Stop deci- sions, Go decisions and Passes when adolescents were alone compared only in the Mother condition [t(24) ¼ 2.1, P < 0.05] when making safe with mother present (Alone vs Mother). decisions. For the MPFC, we find a significant main effect of context [Mother vs Alone; F(1,23) ¼ 19.44, P < 0.0001], but do not find an Decisions interaction. When adolescents made decisions to stop when their mothers were present compared with alone (StopMother-StopAlone), they showed Outcomes increased activation in the left VLPFC and the MPFC (Figure 3a, When adolescents successfully passed through an intersection without Table 2). For descriptive purposes, we extracted parameter estimates crashing following a risky decision (PassMother-PassAlone), they of signal intensity from the VLPFC and MPFC clusters when showed significantly less activation in the amygdala and ventral adolescents made Stop decisions when alone (StopAlone-baseline) striatum when their mothers were present compared with alone and when their mother was present (StopMother-baseline). As (Figure 4a; Table 2). For descriptive purposes, we extracted parameter shown in Figure 3b, adolescents demonstrated significantly greater ac- estimates of signal intensity from the ventral striatum and amygdala tivation in the VLPFC and MPFC when making safe decisions in the clusters for successful passes when adolescents were alone (StopAlone- presence of their mother compared with alone. When adolescents baseline) and when their mother was present (StopMom-baseline). As made decisions to go through the intersection when their mothers shown in Figure 4b, adolescents demonstrated significantly greater ac- were present compared with alone (GoMother-GoAlone), they also tivation in the ventral striatum and amygdala following a successful showed increased activation in the MPFC (Table 2). risky decision when alone than when their mother was present. We do To further probe these effects, we extracted parameter estimates of not find any significant effects for Crash outcomes (CrashMother- signal intensity from the clusters in the MPFC and left VLPFC for the CrashAlone). However, because most participants had too few crash four conditions relative to baseline (StopAlone, GoAlone, StopMother events for statistical tests, we have limited power for this analysis. and GoMother). We used these parameters in SPSS to conduct inter- Therefore this null result should be interpreted with caution. action analyses. For the VLPFC, we find a significant main effect of context [Mother vs Alone; F(1,23) ¼ 20.66, P < 0.0001] as well as a significant interaction [F(1,23) ¼ 6.5, P < 0.01], such that the VLPFC PPI analysis is recruited significantly more within the Mother condition for Stop Using the ventral striatum as the seed region, we first extracted par- (M ¼ 2.14, SEM ¼ 0.96) than Go (M ¼ 0.41, SEM ¼ 0.54) decisions ameter estimates of signal intensity during Stop and Go decisions and [t(24) ¼ 2.2, P < 0.05]. In contrast, when alone, adolescents tend to Pass outcomes during the Alone and Mother Present conditions and recruit the VLPFC more when making Go (M ¼  1.64, SEM ¼ 0.43) ran paired-samples t-tests. Adolescents demonstrated greater ventral than Stop (M ¼ 3.06, SEM ¼ 0.62) decisions [t(24) ¼ 2.07, P ¼ 0.05]. striatum activation when making stop decisions when their mother Importantly, the VLPFC is activated significantly more than baseline was present (M ¼ 2.09, SEM ¼ 1.19) than when alone [M ¼ 0.76, Mothers know best SCAN (2015) 5 of 9 Table 1 Brain regions activated during the stoplight task irrespective of maternal presence Contrast Anatomical region BA x y z t k Stop Decisions – Go Decisions Ventral striatum 12 5 2 4.96 191 R Insula 33 29 2 3.48 110 L Insula 36 14 7 3.84 93 dACC 32/24 3 14 43 4.21 621 Precentral gyrus 33 13 67 4.24 472 SMA 15 2 54 3.35 57 Stop-Go Decisions Precentral gyrus 42 22 61 4.67 47 Go-Stop Decisions Insula 39 5 10 5.94 266 dACC 9 8 43 4.73 1278a a SMA 9 20 48 4.26 a Postcentral gyrus 45 19 49 5.53 Midbrain 6 28 14 4.18 229 R Precuneus 18 76 40 5.32 165 L Precuneus 9 73 46 5.39 392 Pass Outcome Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 Superior frontal gyrus 24 11 55 5.97 2957b b R Precuneus 9 73 43 4.99 b L Precuneus 12 73 46 4.07 b TPJ 54 42 34 5.49 Cerebellum 6 58 26 3.93 122 Note. BA refers to putative Broadman’s areas. x, y, and z refer to MNI coordinates; t refers to the t-score at those coordinates (local maxima); k refers to the number of voxels in each significant cluster. TPJ, temporoparietal junction; SMA, supplementary motor area; dACC, dorsoanterior cingulate cortex. a,b regions are part of the same cluster; – indicates no significant clusters were identified. Fig. 5 (a) PPI analysis during stop decisions when mothers were present compared with alone. Fig. 4 (a) Neural activation during successful passes when adolescents were alone compared with Significant activation represents positive coupling with the ventral striatum. (b) For descrip- mother present. For descriptive purposes, parameters estimates of signal intensity were extracted tive purposes, parameters estimates of signal intensity were extracted from the VLPFC cluster from the ventral striatum and amygdala clusters separately for Passes when Alone and Mother separately for Stop decisions when Alone and Mother Present (relative to baseline). Error bars Present (relative to baseline). Error bars represent SEM. represent SEM. 6 of 9 SCAN (2015) E. H.Telzer et al. Table 2 Brain regions activated during the stoplight task when adolescents’ were alone compared with mother present Contrast Anatomical region BA x y z t k Decisions StopMother-StopAlone R MPFC 32/10 9 47 10 4.34 53 L VLPFC 47/11 42 35 8 3.01 46 Precuneus 0 61 19 4.10 60 L Postcentral gyrus 42 19 34 4.19 75 R Postcentral gyrus 51 7 34 3.95 44 StopAlone-StopMother – – StopMother VLPFC 47 39 23 8 3.33 36 StopAlone – GoMother-GoAlone R MPFC 32/10 9 41 8 4.15 62 L MPFC 10 9 44 5 3.97 44 GoAlone-GoMother – GoMother – – Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 Go Alone Midbrain 6 22 5 4.02 117 R ACC 24/32 12 1 46 3.52 417 R Precentral gyrus 45 19 49 3.90 157 Outcomes PassAlone-PassMother VS 3 5 11 4.50 52 L Amygdala 26 5 24 3.26 36 PassMother-PassAlone – – PassMother R Precentral gyrus 27 22 52 3.70 54 PassAlone L Amygdala 27 5 24 3.56 38 VS 18 17 5 3.78 40 R Insula 44 20 8 3.64 53 R TPJ 40 48 40 43 4.14 186 L Middle frontal gyrus 33 20 40 4.97 41 R Middle frontal gyrus 6 27 11 55 4.70 114 Posterior Cingulate Cortex 3 28 34 4.54 439 R Caudate 16 3 16 3.55 43 PPI with VS StopMother-StopAlone R VLPFC 47/11 30 29 14 4.64 47 StopAlone-StopMother – StopMother R VLPFC 47 30 29 14 4.20 37 L VLPFC 20 62 0 5.61 52 R DLPFC 46 45 41 22 3.76 54 MPFC 32/10 0 50 10 3.76 55 Precentral gyrus 51 2 16 3.45 47 SMA 0 25 52 3.45 103 Caudate 12 5 10 5.45 199 Insula 30 20 11 3.28 37 StopAlone MPFC 32/10 3 59 1 5.79 63 Caudate 6 22 6 4.46 78 GoMother-GoAlone – GoAlone-GoMother – – GoMother – – GoAlone dACC 3 41 13 3.17 41 L Amygdala 22 0 14 3.88 53 STS 51 22 14 3.95 40 (continued) Mothers know best SCAN (2015) 7 of 9 Table 2 Continued Contrast Anatomical region BA x y z t k PassMother-PassAlone – – PassAlone-PassMother – – PassMother Temporal pole 21 51 4 23 4.53 99 Posterior cingulate cortex 6 40 1 4.14 175 PassAlone R Insula 54 14 8 4.06 54 R Amygdala 27 4 14 5.29 40 R TPJ 51 28 16 4.25 132 L TPJ 51 26 23 3.86 104 Posterior cingulate cortex 12 40 4 4.38 530 SMA 3 7 49 3.55 38 Note. BA refers to putative Broadman’s areas. x, y, and z refer to MNI coordinates; t refers to the t-score at those coordinates (local maxima); k refers to the number of voxels in each significant cluster. MPFC, medial prefrontal cortex; VLPFC, ventrolateral prefrontal cortex; DMPFC, dorsomedial prefrontal cortex; VS, ventral striatum; SMA, supplementary motor area; TPJ, temporoparietal junction; STS, superior temporal sulcus; dACC, dorsoanterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; – indicates no significant clusters were identified. Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 SEM ¼ 0.46; t(24) ¼ 2.44, P < 0.05], but did not differ when making go decrease their adolescent’s risk taking. First, adolescents showed greater decisions when their mother was present (M ¼ 0.99, SEM ¼ 0.83) com- recruitment of the VLPFC when making safe decisions in the presence pared with alone [M ¼ 0.23, SEM ¼ 0.79; t(24) ¼ 0.6, ns]. For Pass of their mothers. The VLPFC is involved in self-control and the ability outcomes, adolescents showed significantly greater activation when to regulate and control one’s prepotent thoughts and behaviors (Baker alone (M ¼ 0.58, SEM ¼ 0.38) compared with when their mother was et al., 1997; Gray et al., 2002) and plays a causal role in goal-directed present [M ¼ 1.10, SEM ¼ 0.40; t(24) ¼ 4.56, P < 0.0001]. inhibitory control, including the braking of motor responses (Wessel Next, we conducted PPI analyses to test for functional connectivity et al., 2013). These results suggest that mothers boost self-control by between the ventral striatum and cognitive control regions. When increasing PFC activation, facilitating more deliberative and safe deci- adolescents made a decision to stop (StopMother-StopAlone), we sions. In addition, adolescents showed heightened activation in the found a significant interaction between the ventral striatum and MPFC when making both safe and risky decisions when their VLPFC. For descriptive purposes, we extracted parameter estimates mother was present. The MPFC is involved in monitoring and com- of signal intensity from the VLPFC cluster that showed a significant puting the reward value of ongoing cognitive tasks (Pochon et al., interaction with the VS and plotted the effects separately for 2002), as well as the subjective value of rewards (Kable and StopMother and StopAlone (vs baseline). As shown in Figure 5, ado- Glimcher, 2007; Levy et al., 2010). Thus, when their mothers are pre- lescents show VS-VLPFC coupling significantly greater when their sent, adolescents may evaluate the relative reward of running the stop- mothers are present than when alone (Table 1). To further probe light vs stopping. The MPFC is also involved in self-related processing this effect, we ran PPI analyses separately for the Mother and Alone and thinking about close others (Mitchell et al., 2006), and so adoles- condition for Stop decisions. Adolescents do not show significant cents may be incorporating their mothers’ perspective to inform their functional coupling between the VS and VLPFC when alone but do own behavior and decide whether to engage in risky decisions. show significant coupling between these regions when their mother Second, adolescents showed decreased activation in the ventral stri- was present (Table 1). No brain regions differentially functionally atum and amygdala following a risky decision in the presence of their interacted with the VS during decisions to Go (GoMother-GoAlone) mothers. The ventral striatum has been consistently linked with he- or during successful passes (PassMother-PassAlone; see Table 1 for donic rewards, tends to be more active in adolescents than children or Mother and Alone conditions separately). adults during reward and risk-taking tasks and predicts greater engage- ment in real-life risk taking behavior (Galvan et al. 2006, 2007). The amygdala codes for emotionally salient stimuli and tends to be more DISCUSSION active in adolescents than adults during emotion (particularly positive Adolescence is a developmental period marked by exquisite changes in emotions) and reward processing (Ernst et al., 2005; Hare et al. 2008; functional brain development. Heightened reward sensitivity, coupled for reviews see Somerville et al., 2011, Nelson et al., 2014). Together, with relatively less developed cognitive control, has largely been our findings suggest that mothers reduce the rewarding and salient thought to underlie adolescents’ engagement in risky behavior nature of engaging in risk taking, perhaps taking the fun out of (Steinberg, 2008; Casey et al., 2011). In this study, we test whether being risky. Thus, when adolescents make risky decisions in the pres- the presence of parents alters adolescent risk taking via changes in ence of their mothers, they respond to risky outcomes with decreased reward sensitivity and cognitive control. amygdala and ventral striatum activation than when making the same Our findings suggest that maternal presence has a large impact on risky decisions alone. adolescents’ risky behavior and brain function. When mothers were Finally, and most novel, maternal presence facilitated functional present, adolescents engaged in significantly less risky behavior. coupling between neural regions involved in reward and cognitive Surprisingly, prior research showing that parental supervision is asso- control processing when adolescents made safe but not risky decisions, ciated with lower adolescent risk taking has all been correlational suggesting that mothers increased the rewarding nature of engaging in (Richardson et al., 1993; Beck et al., 2001; Borawski et al., 2003). cognitive control. These functional connectivity results suggest that This is the first study to experimentally demonstrate the protective within adolescents, those who engage the ventral striatum to a greater role that parental presence plays on reducing adolescent risk taking. extent also engage the VLPFC more when making safe decisions, but Importantly, we identified the neural mechanisms by which mothers this only occurs when adolescents’ mothers are present. Theories of 8 of 9 SCAN (2015) E. H.Telzer et al. adolescent risk-taking propose that heightened reward sensitivity lar- their teenagers’ decision making capacities in ways that allow their gely underlies increased risk-taking during adolescence (Steinberg, children to engage in more mature cognitive processes. Importantly, 2008), and most prior work has focused on the contexts in which our findings suggest that parents do not just serve as gatekeepers but reward sensitivity leads to maladaptive, risky behavior, for example, actually change the ways in which adolescents think and reason about in the presence of peers (Chein et al., 2011). Here, we demonstrate that risks. safe behavior occurs when the ventral striatum is functionally coupled with the VLPFC when mothers are present. Thus, heightened coupling Conflict of Interest between these regions is associated with safer behavior, suggesting that None declared. maternal presence may promote a reward response that boosts adoles- cents’ engagement of cognitive control and promotes more safe deci- sion making. Importantly, the very same neural circuitry (i.e. ventral REFERENCES striatum) that has been linked to greater risk taking may also be related Baker, S.C., Frith, C.D., Dolan, R.J. (1997). The interaction between mood and cognitive to thoughtful, more deliberative and safe decisions when mothers are function studied with PET. Psychological Medicine, 27, 565–78. present. Therefore, maternal presence may serve to redirect adoles- Beck, K.H., Shattuck, T., Raleigh, R. (2001). Parental predictors of teen driving risk. American Journal of Health and Behavior, 25(1), 10–20. cents’ reward sensitivity away from risk (i.e. reduced ventral striatum Borawski, E.A., Ievers-Landis, C.E., Lovegreen, L.D., Trail, E.S. (2003). Parental monitor- activation during risky outcomes) and toward more safe behavior (i.e. ing, negotiated unsupervised time, and parental trust: the role of perceived parenting increased VS-PFC coupling during safe decisions). Because PPI ana- practices in adolescent health risk behaviors. Journal of Adolescent Health, 33(2), 60–70. lyses are not directional, an alternative explanation for functional cou- Casey, B.J., Somerville, L.H., Gotlib, I.H., et al. (2011). Behavioral and neural correlates of delay of gratification 40 years later. Proceedings of the National Academy of Science of the pling between the ventral striatum and VLPFC is that engaging in United States of America, 108(36), 14998–5003. cognitive control in the presence of their mother is a rewarding pro- Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016 Chein, J., Albert, D., O’Brien, L., Uckert, K., Steinberg, L. (2011). Peers increase adolescent cess. In other words, recruitment of cognitive control facilitates safe risk taking by enhancing activity in the brain’s reward circuitry. Developmental Science, decisions in the presence of their mother which results in a heighted 14, F1–10. sense of reward. Centers for Disease Control and Prevention Web-based Injury Statistics Query and Reporting System (WISQARS) (2012), National Center for Injury Prevention and Our findings contribute to conceptualizations of adolescent neuro- Control, Centers for Disease Control and Prevention. development. While dual systems models (Steinberg, 2008; Somerville Crone, E.A., Dahl, R.E. (2012). Understanding adolescence as a period of social-affective et al., 2010; Casey et al., 2011) have gained considerable attention, engagement and goal flexibility. Nature Reviews Neuroscience, 13, 636–50. several alternative perspectives have been proposed (Crone and Dahl, Ernst, M., Nelson, E.E., Jazbec, S., et al. (2005). Amygdala and nucleus accumbens in 2012; Pfeifer and Allen, 2012). For example, striatal reactivity is not responses to receipt and omission of gains in adults and adolescents. Neuroimage, 25(4), 1279–91. always associated with greater adolescent risk taking, as suggested by Friston, K.J., Buechel, C., Fink, G.R., Morris, J., Rolls, E., Dolan, R.J. (1997). dual system models, and, depending on the context, striatal reactivity Psychophysiological and modulatory interactions in neuroimaging. NeuroImage, 6(3), can result in lower risk taking behavior (Pfeifer et al., 2011; Telzer 218–29. et al., 2013). We build upon these findings and show that ventral Galvan, A., Hare, T., Voss, H., Glover, G., Casey, B.J. (2007). Risk-taking and the adolescent brain: Who is at risk? Developmental Science, 10, F8–14. striatum activation is associated with safe decision making when it Galvan, A., Hare, T.A., Parra, C.E., et al. (2006). Earlier development of the accumbens occurs in tandem with cognitive control-related neural activation in relative to orbitofrontal cortex might underlie risk-taking behavior in adolescents. the presence of mothers. Together, our findings underscore the im- Journal of Neuroscience, 26, 6885–92. portance of understanding the social context in which adolescent de- Gardner, M., Steinberg, L. (2005). Peer influence on risk taking, risk preference, and risky cision making occurs and suggests that the striatum plays a diverse decision making in adolescence and adulthood: an experimental study. Developmental Psychology, 41(4), 625–35. function, sometimes promoting risky behavior but at other times Geier, C.F., Terwilliger, R., Teslovich, T., Velanova, K., Luna, B. (2010). Immaturities in deterring risky behavior depending upon the motivational context. reward processing and its influence on inhibitory control in adolescence. Cerebral While our task did not have a condition to test for the effect of the Cortex, 20(7), 1613–29. person on adolescent risk taking and neural reactivity, our findings are Gray, J.R., Braver, T.S., Raichle, M.E. (2002). Integration of emotion and cognition in the lateral prefrontal cortex. Proceedings of the National Academy of Sciences of the United the opposite of those found by Chein et al. (2010) who used the same States of America, 99, 4115–20. manipulation with peer presence and found that peers increase risk Hardin, M.G., Mandell, D., Mueller, S.C., Dahl, R.E., Pine, D.S., Ernst, M. (2009). taking and ventral striatum activation during risky decisions. Inhibitory control in anxious and healthy adolescents is modulated by incentive and Therefore, our effects are likely not due to someone watching vs not incidental affective stimuli. Journal of Child Psychology and Psychiatry, 50(12), 1550–8. watching, as peers’ presence alters risk taking and related neural pat- Hare, T.A., Tottenham, N., Galvan, A., Voss, H.U., Glover, G.H., Casey, B.J. (2008). Biological substrates of emotional reactivity and regulation in adolescence during an terns in a different way than does parents’ presence. However, without emotional go-nogo task. Biological Psychology, 63(10), 927–34. directly comparing peer and parent influence in the same paradigm, we Kable, J.W., Glimcher, P.W. (2007). The neural correlates of subjective value during inter- cannot be sure that these contexts have opposite effects on adolescents, temporal choice. Nature neuroscience, 10(12), 1625–33. and our findings could be due to other factors that differ between the Kann, L., Kinchen, S., Shanklin, S.L., Flint, K.H., Kawkins, J., Harris, W.A. (2014). Youth risk behavior surveillance – United States, 2013. Morbidity and Mortality Weekly Report. studies (e.g. characteristics of the samples, age, etc.). Thus, future work Surveillance Summaries (Washington, D.C.: 2002), 63(Suppl 4), 1–168. should examine how both parents and peers impact adolescent deci- Kohls, G., Peltzer, J., Herpertz-Dahlmann, B., Konrad, K. (2009). Differential effects of sion making and neural processing during risk taking as well as the social and non-social reward on response inhibition in children and adolescents. differential role of parental and authority figures to test whether adults Developmental Science, 12(4), 614–25. decrease adolescent risk taking or whether our patterns are specific to Levy, I., Snell, J., Nelson, A.J., Rustichini, A., Glimcher, P.W. (2010). Neural representation of subjective value under risk and ambiguity. Journal of neurophysiology, 103(2), parents. 1036–47. Mothers in our study did not vocally encourage or discourage any Maldijian, J.A., Laurienti, P.J., Burdette, J.H. (2004). Precentral gyrus discrepancy in elec- kind of behavior, actively regulate their child’s decisions or provide tronic versions of the Talairach atlas. NeuroImage, 21(1), 450–5. feedback. Instead, mothers simply told their child they would be pre- Maldjian, J.A., Laurienti, P.J., Kraft, R.A., Burdette, J.H. (2003). An automated method for sent and watching. Our findings suggest that parental influence does neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. NeuroImage, 19(3), 1233–39. not have to be explicit. The mere presence of a parent in the room or McLaren, D.G., Ries, M.L., Xu, G., Johnson, S.C. (2012). A generalized form of context car may help adolescents to slow down, think twice and make more dependent psychophysiological interactions (gPPI): A comparison to standard deliberative and safe decisions. Parents are in a position to regulate approaches. NeuroImage, 61(4), 1277–1286. Mothers know best SCAN (2015) 9 of 9 Mitchell, J.P., Macrae., C.N., Banaji, M.R. (2006). Dissociable medial prefrontal contribu- Smith, A.B., Halari, R., Giampetro, V., Brammer, M., Rubia, K. (2011). Developmental tions to judgments of similar and dissimilar others. Neuron, 50, 655–63. effects of reward on sustained attention networks. NeuroImage, 56, 1693–704. National Center for Health Statistics (2013), Health, United States, 2013. US Department of Steinberg, L. (2008). A social neuroscience perspective on adolescent risk-taking. Health and Human Services, Centers for Disease Control and Prevention. Hyattsville, Developmental Review, 28(1), 78–106. MD. Somerville, L.H., Jones, R.M., Casey, B.J. (2010). A time of change: behavioral and neural Nelson, E.E., Lau, J.Y., Jarcho, J.M. (2014). Growing pains and pleasures: how emotional correlates of adolescent sensitivity to appetitive and aversive environmental cues. Brain learning guides development. Trends in Cognitive Sciences, 18(2), 99–108. and Cognition, 72, 124–33. Nelson, E.E., Leibenluft, E., McClure, E.B., Pine, D.S. (2005). The social re-orientation of Somerville, L.H., Fani, N., McClure-Tone, E.B. (2011). Behavioral and neural representa- adolescence: a neuroscience perspective on the process and its relation to psychopath- tion of emotional facial expressions across the lifespan. Developmental neuropsychology, ology. Psychological Medicine, 35(2), 163–74. 36(4), 408–28. Peake, S., Dishion, T.J., Stormshak, E.A., Moore, W.E., Pfeifer, J.H. (2013). Risk-taking and Telzer, E.H., Fuligni, A.J., Lieberman, M.D., Gálvan, A. (2013). Ventral striatum activation social exclusion in adolescence: neural mechanisms underlying peer influences on deci- to prosocial rewards predicts longitudinal declines in adolescent risk taking. sion-making. NeuroImage, 82, 23–4. Developmental Cognitive Neuroscience, 3, 45–52. Pfeifer, J.H., Allen, N.B. (2012). Arrested development? Reconsidering dual-systems models of Teslovich, T., Mulder, M., Franklin, N.T., et al. (2014). Adolescents let sufficient evidence brain function in adolescence and disorders. Trends in Cognitive Sciences, 16(6), 322–9. accumulate before making a decision when large incentives are at stake. Developmental Pfeifer, J.H., Masten, C.L., Moore, W.E., III, et al. (2011). Entering adolescence: resistance Science, 17(1), 59–70. to peer influence, risky behavior, and neural changes in emotion reactivity. Neuron, Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., et al. (2002). Automated anatom- 69(5), 1029–36. ical labeling of activations in spm using a macroscopic anatomical parcellation of the Pochon, J.B., Levy, R., Fossati, P., et al. (2002). The neural system that bridges reward and mni mri single-subject brain. NeuroImage, 15, 273–89. cognition in humans: an fMRI study. Proceedings of the National Academy of Sciences, Ward, B.D. (2000). Simultaneous inference for fMRI data. Available: http://afni.nimh.nih. 99(8), 5669–74. gov/pub/dist/doc/manuals/AlphaSim.pdf (accessed 24 March 2015). Richardson, J.L., Radziszewska, B., Dent, C.W., Flay, B.R. (1993). Relationship between Wessel, J.R., Conner, C.R., Aron, A.R., Tandon, N. (2013). Chronometric electrical stimu- after-school care of adolescents and substance use, risk taking, depressed mood, and lation of right inferior frontal cortex increases motor braking. Journal of Neuroscience, academic achievement. Pediatrics, 92(1), 32–8. 33(50), 19611–9. Downloaded from http://scan.oxfordjournals.org/ by guest on July 7, 2016