Adolescents’ neural responses to their parents’ emotions: associations with emotion regulation, internalizing symptoms, and substance use

Abstract

Parental emotion expression has been linked to adolescent emotional and psychopathology development. However, neural responses to parental emotion are not well characterized. The present study examined associations between adolescents’ neural responses to parent emotion and adolescents’ emotion regulation (ER) difficulties, depressive and anxiety symptoms, and substance use (SU). One hundred seventy-five 12- to 14-year-olds and their parent(s) participated in the study. In a lab session, families completed a parent–adolescent interaction task. In an MRI session, adolescents viewed videos of their own parents and unfamiliar parents expressing positive, negative, and neutral emotions from the interaction task. Higher salience region responses to own parent negative emotion (versus neutral) in ventral anterior cingulate cortex (vACC), anterior insula (AI), and nucleus accumbens (NAc) were associated with adolescent ER difficulties and depressive and anxiety symptoms. Higher vACC and AI responses to parent positive emotion (versus neutral) were associated with anxiety symptoms only. Higher salience region responses to own parent negative emotion (versus other parent negative) were associated with ER difficulties. Responses to own parent positive emotion (versus other parent positive) were associated with ER and anxiety symptoms for boys. Findings suggest that adolescents’ salience system sensitivity to parental emotion may be important in the development of ER and internalizing symptoms.

The family environment is critical for adolescent emotional development. Parent–youth relationships are rapidly changing during early adolescence, but are an important influence as youth enter into a risk period for psychopathology and substance use (SU) (Steinberg 2001). Early adolescence is also a period of change in social–emotional neurobiology and of sensitivity to environmental influence (Casey et al. 2019). Parent–adolescent interactions characterized by high parent negative emotion and altered parent positive emotion (and by adolescent’s reactivity to parent emotion) have been linked to adolescent emotion regulation (ER) difficulties, depressive and anxiety symptoms, and SU (Chaplin et al. 2014, Schwartz et al. 2017, O’Connor et al. 2020). However, the neural mechanisms by which parent negative and positive emotion impact adolescents are not well known. The present study examined associations between early adolescents’ neural responses to their own parents’ emotion expressions (during a lab interaction task) and adolescent ER difficulties, depressive and anxiety symptoms, and SU.

Parent emotion

Higher negative parenting and high levels of parent negative emotion expression (particularly high negative emotion expression that is inappropriate to context and unmodulated) have been linked to greater ER problems and internalizing symptoms and SU in childhood and adolescence (e.g. Trucco 2020, Clayborne et al. 2021). High parent negative emotion expression towards youth and harsh parenting can lead youth to feel heightened negative emotional arousal, leading to difficulty in ER development (Morris et al. 2017), altered emotion-related neural function (Cosgrove et al. 2020), rumination and depressive symptoms (Morris et al. 2017), and using substances to cope with negative emotion (Chaplin et al. 2018). Additionally, when parents experience high negative emotional arousal in challenging interactions with youth, they may over-accommodate youth to reduce their own discomfort, which can inadvertently reinforce youth anxiety (O’Connor et al. 2020).

Higher positive parenting and parent positive emotion expression have been associated with adolescents developing more adaptive ER skills and showing lower depressive symptoms and SU (Morris et al. 2017, Trucco 2020). For anxiety, the role of parental positive emotion is less clear. Parents of youth with anxiety may show higher positive emotion as part of parental accommodation and reassurance, which can reinforce youth anxiety (Gladstone et al. 2023).

fMRI studies

It is important to understand adolescent’s neural processing of parent emotion and adolescent ER and symptoms. This can help understand how youth respond to the parenting environment and how that impacts adolescents and can provide targets for neurobiologically informed interventions. Some studies have examined parent emotion and youth physiological arousal, finding that adolescents’ heightened cortisol and heart rate reactivity to parent negative emotional behaviors predict depressive symptoms and SU (Chaplin et al. 2014, Wei et al. 2023).

A few initial studies have examined adolescent neural responses to parent emotion. These studies find that adolescents show activation to negative and positive parent emotional stimuli in salience regions involved in emotional reactivity and reward processing, including anterior insula (AI), anterior cingulate cortex (ACC), and nucleus accumbens (NAc), and in social cognition regions including dorsomedial prefrontal cortex (dmPFC), precuneus, and posterior cingulate cortex (PCC) (van Houtum et al. 2022). This is consistent with research finding that parent–child bonding is facilitated by parental responses to child emotion in salience, reward, and social cognition networks (Swain et al. 2012).

Further, initial studies have found associations between adolescents’ neural responses to parental emotion and adolescents’ depressive and anxiety symptoms. In one study, Aupperle et al. (2016), with 12- to 16-year-old girls (n = 18), found that greater R amygdala responses to hearing maternal criticism were associated with greater adolescent depressive and anxiety symptoms. They also found that lower R amygdala responses to maternal praise were associated with greater adolescent depressive and anxiety symptoms. However, anxiety symptoms were associated with mixed neural responses to praise in social cognition region superior medial PFC, with higher activation to anticipating praise and lower activation to hearing praise associated with anxiety. In a second study, Silk et al. (2017), with 9- to 17-year-olds (N = 48), found that adolescents with depression showed greater responses in a salience region (hippocampus) to maternal criticism than control youth. They also found that adolescents with depression showed lower responses to maternal praise in salience/reward regions [caudate, ventromedial PFC (vmPFC)], and in a social cognition region (precuneus). In a third study, Whittle et al. (2012), with adolescents (mean age 17, n = 30), presented video clips of adolescents’ parents showing emotions in a lab parent–adolescent interaction (the approach taken in the present study). They found that adolescents’ lower responses to own mother positive emotion in salience/reward regions (rostral ACC, striatum/putamen) were associated with adolescent depressive symptoms. Saxbe et al. (2016), in 17.5-year olds (n = 21), found that adolescents’ lower responses in salience regions (insula, ACC, amygdala, and striatum) to videos of parental negative and positive emotions were associated with adolescent angry/aggressive behavior. Finally, Cosgrove et al. (2019), in 14- to 16-year-olds (n = 27), found that AI responses to a negative parent behavior (parent making a costly error in a game) were associated with parental anxiety.

Taken together, these studies found adolescent salience and reward regions (ACC, AI, amygdala, and NAc) and social cognition region (e.g. dmPFC) responses to parental emotion, with higher salience region responses to parental negative emotion associated with depressive and anxiety symptoms, and lower salience/reward region responses associated with depressive symptoms. They found mixed evidence for neural responses to parental positive emotion associated with anxiety symptoms. Notably, these studies are limited by small samples and did not examine associations with adolescent SU.

The present study examined responses to parent emotion and adolescent ER difficulties, depressive and anxiety symptoms, and SU in a priori regions of interest (ROIs)—ventral ACC (vACC), AI, amygdala, and NAc. These ROIs were selected because they are part of the salience network and are involved in processing negative and positive emotions and reward (Seeley et al. 2007), have been found in studies of neural correlates of ER, depression, anxiety, and SU (e.g. Young et al. 2019), and have been found in studies of neural responses to parent emotion (noted earlier) and in research on parent–child bonding (Swain et al. 2012). To capture activation outside of these ROIs, including in social cognition regions that respond to parent emotional stimuli, we conducted secondary whole-brain analyses.

Sex differences

It is important to consider adolescent sex in associations between responses to parent emotion and psychopathology and SU. There are sex differences in depressive and anxiety symptoms and emotion expression in adolescence (Lewinsohn et al. 1998, Hankin et al. 2007). Further, girls and boys show different predictors of depressive symptoms and SU, with girls potentially taking a pathway characterized by heightened negative emotional reactivity and boys via altered reward responses (Hankin et al. 2007, Hammerslag and Gulley 2016). For example, one study found that negative parenting predicted greater AI and ACC responses to negative emotional stimuli for girls and that these neural responses predicted depressive symptoms and SU for girls but not boys (Chaplin et al. 2019). These sex differences could be due to biological differences and/or gender socialization (with girls socialized to more freely express negative emotions and boys to seek out new sensations) (Chaplin et al. 2018).

The present study

This study examined associations between neural responses to negative and positive parent emotion and ER difficulties, depressive and anxiety symptoms, and SU in early adolescents, in a large sample of families over sampled for maladaptive parenting. This study adds to the literature by using a large sample, a naturalistic parent–adolescent interaction task, and by examining associations with ER and SU and sex differences. It also tests adolescents’ neural responses to their own parent’s emotions versus an unfamiliar parent. Responses to one’s own parent may be particularly important for psychopathology, given that parents are significant figures and carry a history for youth (Whittle et al. 2012).

Hypotheses

• Adolescents’ heightened neural responses to their parents’ negative emotions (versus neutral) will be associated with greater ER difficulties, depressive symptoms, anxiety symptoms, and SU.

• Adolescents’ neural responses to their parents’ positive emotions (versus neutral) will be associated with ER difficulties, depressive symptoms, anxiety symptoms, and SU. We anticipate that lower responses to parent positive emotions will be associated with ER difficulties and depressive symptoms, higher responses (particularly in NAc) may be associated with SU (given reward system responses to other positive stimuli predicting SU; Bjork 2020), and either higher or lower responses will be associated with anxiety symptoms.

• We explored whether these neural responses are specific to emotion from one’s own parent by examining associations between neural responses to own parent negative emotion (versus other parent negative) and own parent positive emotion (versus other parent positive) and ER, symptoms, and SU.

• We explored whether adolescent sex moderated the above associations.

Methods

Participants

Participants were 175 early adolescents and their parent(s) (see Table 1 for demographics), with 81 boys and 94 girls based on parent-reported sex. Gender identity was 81 boys, 84 girls, and 10 gender nonconforming. Parents were primary caregivers—96% were biological parents and 4% were nonbiological parents (e.g. step-parents). Families were part of a longitudinal study of parenting and SU, with 38% of the families in the sample having high maladaptive parenting. Cut-off scores to indicate high maladaptive parenting were at least two of these: ≥ 9 on inconsistent discipline, ≥6 on poor supervision, and/or ≤ 11 on positive parenting scales of the APQ-SF (Elgar et al. 2007). Cutoffs were based on the top/bottom 10% of scores in two large community samples, following clinical cut-off recommendations (Dadds et al. 2003).

Families were recruited from a suburban mid-Atlantic area and were similar to the local area in terms of race/ethnicity and income. We included adolescents who were 11- to 14-years old, English speaking, and MRI safety-eligible (e.g. no metal in body) and excluded youth with intellectual disability, psychosis, history of congenital brain defect or severe brain injury, or use of dental retainers if the retainer caused artifacts in the localizer scan. Adolescents on medications were included to allow a range of symptoms. The study was approved by the George Mason University IRB, and informed parental consent and adolescent assent were obtained.

The overall study included 228 adolescents. Of those, 175 had MRI data on both runs of the parent emotion task and passed MRI data quality and motion checks and comprise the present sample. This paper focused on baseline data, collected from 2019 to 2022. At baseline, adolescents participated in a behavioral session and, about 2 weeks later, a MRI session (11 youth completed the MRI session 9 months late due to COVID shutdown).

Behavioral session

Adolescents and parent(s) completed questionnaires assessing ER, symptoms, and SU and the parent–adolescent interaction (PAIT) task.

PAIT task

Adolescents and their parent(s) had two 10-min discussions (Discussion 1, Discussion 2) about highly mutually rated conflict topics (Pruim et al. 2015, Chaplin et al. 2012). Adolescents completed one discussion with each parent if two parents attended or two discussions with one parent if only one parent attended. Adolescents and parent(s) then also completed two 2-min positive discussions about something fun they did together. Discussions were recorded and video clips were used in the fMRI parent emotion task. Thirteen families had two parents (a mother and father) participate in PAIT. A total of 162 families had one parent participate (143 mothers, 19 fathers), mostly due to COVID restrictions. Given potential parent sex differences, analyses covary father participation.

MRI session

Adolescents were acclimated to the environment and then underwent MRI, including a T1-weighted structural scan and functional scans.

Parent emotion task

The parent emotion task (based on Whittle et al. 2012) involved viewing 16-s video clips of adolescent’s own mother and/or father (and an unfamiliar mother and/or father) showing negative, positive, and neutral emotions in the PAIT task. For the unfamiliar parents, families were recruited from a nearby community, completed the PAIT, and were informed that videos would be shown in a research study. Unfamiliar parent clips were matched to the participant by parent and child sex and white/non-white race. Adolescents within a category (e.g. white girls with a mother video) all saw the same unfamiliar parent videos. Youth confirmed that they did not recognize the unfamiliar parents.

The task had two runs. Each run included two 16-s clips of each type of stimuli (e.g. own parent negative), for 12 clips per run. Clip order was random with the constraint that two clips of the same condition (e.g. own parent negative clip 1, own parent negative clip 2) were not shown consecutively. The 16-s videos were presented in a block design interspersed with 8 s rests. One run had clips from the first discussion and the other run had clips from the second discussion, with run order randomized. Thus, for families with two parents participating, each run had a different parent. For families with one parent, the second run used clips from the same parent in the second discussion. For two participants, one clip did not have pure negative emotion for 16 s, so an 8-s clip was looped (consistent with Whittle et al. 2012). Participants with looped videos did not differ on study variables.

Clips showed the parent’s face and upper body with a neutral background. Audio contained parent’s speech and, in some cases, pieces of the child’s speech, similar to Whittle et al. (2012). Coders (with 10–12 h of training) watched the discussion videos and identified the 16-s clips that were most intense for the emotion (and did not contain the other type of emotion). They identified emotion clips based on facial, vocal, postural, and verbal content cues, following a coding system (based on Morgan et al. 2015). Negative emotion cues included furrowed brow, eye roll, lowered voice, tense posture, and aggressive statements. Positive emotion cues included smiling and laughter. Neutral clips did not include cues for negative or positive emotion. Coders rated the chosen clips for emotion intensity from 0 to 3. Fifteen percent of intensity ratings were double-coded and checked for reliability, which was high (ICC = 0.78).

MRI acquisition

Functional and structural images were acquired on a Siemens 3 T Prisma MR scanner and a 32-channel phased array headcoil. T1-weighted MPRAGE anatomical images (TR/TE = 2400/2 ms; GRAPPA = 2; 216 0.8 × 0.8 × 0.8 mm slices) were collected. Images of the blood oxygen-level-dependent (BOLD) responses were collected using T2*weighted gradient-echo echoplanar images (EPI) [TR/TE: 1200/33 ms; flip =45°; MB = 4; field of view (FOV): 210 mm; matrix size: 84 × 84; 40 axial 2.5 mm thick slices].

Outcome measures

ER. Adolescents completed the Difficulties in Emotion Regulation Scale (DERS; Gratz and Roemer 2004), a widely used 36-item self-report of ER problems in the past year. The DERS has demonstrated validity and reliability with adolescents (Neumann et al. 2010) and had high internal consistency and range in this study (α = 0.94, Table 1).

Depressive symptoms

Adolescents completed the Children’s Depression Inventory (CDI; Kovacs and Preiss 1992), a 27-item self-report of depressive symptoms in past 2 weeks. The CDI is valid and reliable with adolescents (Kovacs and Staff 2003), with high internal consistency and range in this study (α = 0.89, Table 1).

Anxiety symptoms

Adolescents completed the Revised Child Manifest Anxiety Scale (RCMAS; Reynolds and Richmond 1985), a 37-item self-report of anxiety symptoms. The RCMAS is valid and reliable with adolescents (Lee et al. 1988), with high internal consistency and range in this study (α = 0.90, Table 1).

SU

Adolescents self-reported on SU on the Youth Risk Behavior Surveillance Survey (YRBS; Brener et al. 2004) and parents reported on adolescent SU using parent report YRBS items. The YRBS asks how many days in lifetime youth used eight substances (e.g. alcohol, marijuana). SU was scored as a binary variable (1 if either self or parent YRBS was positive for use, 0 if both reports were negative). We collected urine screens; however, these did not identify additional users. SU rates were 17.7%, consistent with studies of early adolescents (Kristjansson et al. 2018).

Analyses

Neuroimaging data was preprocessed using fMRIPrep 20.0.5 (Esteban et al. 2019). Preprocessing included susceptibility distortion and motion correction, registration of EPI data to the anatomical T1w and MNI template, automatic removal of motion artifacts using independent component analysis (ICA-AROMA, Pruim et al. 2015, and smoothing with 6 mm FWHM, with a temporal filter at 96 s. FSL’s fMRI Expert Analysis Tool (FEAT) was used for first-level analyses with a temporal and spacially regularized auto-correlation model. Coefficient of parameter estimate (COPE) values for each explanatory value were used to create contrasts of interest. Data were excluded if one or both runs had significant motion, specifically >20% of TRs in run with > 0.5 mm FD or >5% of TRs in run with > 3 mm FD. For ROIs [right (R) and left (L) amygdala, AI, vACC, and NAc], cortical masks were multiplied by the MNI152 gray matter mask to remove white matter. Masks for amygdala and NAc were derived from the Harvard–Oxford Atlas. Masks for AI and vACC were created by identifying peak activation coordinates from Neurosynth and creating 10 mm radius spheres. Mean activation across voxels in the masks/spheres was extracted as ROI activation.

Analyses covaried adolescent sex (boy = 1, girl = 0), age, and race/ethnicity (Non-Hispanic white = 1, Other = 0) and father participation. We considered emotion intensity of clips and psychiatric medication use (14% used psychiatric medications) as covariates, but these were not associated with ROI responses/outcomes and so were not covaried. We examined sex moderation, using biological sex (we re-ran moderation excluding non-cisgender youth and findings were similar). Of note, sex was correlated with pubertal level (measured on the Pubertal Development Scale; Petersen et al. 1988), r = −0.57, P < .001, girls > boys. To test the role of sex controlling for pubertal development, significant sex main effects or interactions were re-run covarying pubertal level.

For hypothesis 1a, eight separate regressions were conducted with extracted ROI responses to own parent negative–own parent neutral in the eight ROIs predicting each outcome (ER, depressive symptoms, anxiety symptoms, SU), with FDR correction for the eight analyses within each outcome, with covariates included. For the other hypotheses, similar regressions were conducted with ROI activation to the other contrasts (e.g. own parent positive–own parent neutral) predicting outcomes. Regular regression was used for continuous outcomes and logistic regression for SU. We explored sex moderation by adding ROI response × sex interactions to the regressions. Main findings from ROI analyses are shown in the tables. Detailed ROI regression results are shown in Supplement 1.

We conducted secondary whole-brain analyses using FSL’s randomize (Winkler et al. 2014) for nonparametric permutation testing. These models tested ER/symptoms/SU predicting whole-brain activation to the relevant contrast (e.g. own parent negative–neutral) using 5000 permutations, a .05 significance threshold, and correction for multiple comparisons using threshold-free cluster enhancement.

Results

Continuous variables were examined for outliers (values at least 3 s.d. from the mean). Several ROI responses and depressive symptoms had 1–3 outlier cases and were winsorized.

Sex, age, race, and father participation differences

Older youth showed greater R amygdala responses to own parent negative (–own parent neutral), r = 0.17, P = .02. Race and father participation were not associated with neural responses. Girls showed higher responses in all ROIs to own parent positive (–other parent positive) than boys (t’s > 2.34, P’s < .01). When covarying pubertal level, the effect of sex remained significant for six of the eight ROIs (L amygdala, R and L AI and vACC, and L NAc) (for the other two, neither sex nor pubertal level were significant predictors).

Non-white adolescents showed greater depressive symptoms and SU than white adolescents (t’s > 2.01, P’s < .05). Adolescents with father participation had lower depressive and anxiety symptoms than those without (t’s > 2.70, P’s < .01). Girls showed greater ER difficulties and depressive and anxiety symptoms than boys (t’s > 2.3, P’s < .01). When covarying pubertal level, the effect of sex remained for anxiety, but not ER or depressive symptoms.

Parent emotion ratings

The mean intensity rating (0–3 scale) was 1.73 (SD = 0.51) for negative clips and 1.91 (SD = 0.51) for positive clips. Emotion intensity ratings did not correlate with ROI responses or outcomes. Parents’ overall negative and positive emotion and parental warmth were also coded from the conflict discussions. Parents’ overall emotion was not significantly associated with ROI responses. Parental warmth was correlated with higher R and L NAc responses to own parent positive (–own parent neutral), r’s = 0.21, 0.25, FDR P’s = .02, .01).

Responses to own parent negative versus neutral

Greater R & L vACC, R AI, and L NAc responses to own parent negative emotion (–own parent neutral) predicted greater adolescent ER difficulties (Table 2). There were no significant ROI response × sex interactions or whole-brain findings for ER.

Greater R & L vACC and R AI responses predicted greater adolescent depressive symptoms (Table 2). There were no significant ROI response × sex interactions. Whole-brain analyses supported ROI findings, with depressive symptoms associated with higher activation in insula, dACC, and striatum, and in social cognition region dmPFC [Supplement 2, Table and Figure A (STF 2A)].

Greater R & L vACC, R & L AI, and L NAc responses predicted greater adolescent anxiety symptoms (Table 2). There were no significant ROI response × sex interactions or whole-brain findings for anxiety.

There were no significant ROI response, response × sex interactions, or whole-brain findings for SU.

Responses to own parent positive versus neutral

There were no significant ROI response, response × sex interactions, or whole-brain findings for ER difficulties or depressive symptoms.

Greater R & L vACC and R & L AI responses to own parent positive emotion (–own parent neutral) predicted greater adolescent anxiety symptoms (Table 3). There were no significant ROI response × sex interactions. Whole-brain analyses supported the ROI findings, with higher anxiety symptoms associated with higher activation in insula and ACC and dmPFC (STF 2B).

There were no significant ROI response or response × sex interactions predicting SU, although before FDR correction, higher R NAc responses were associated with SU (Table 3).

Responses to own parent negative versus other parent negative

Greater R & L amygdala, vACC, AI, and NAc responses to own parent negative emotion (–other parent negative) predicted greater adolescent ER difficulties (Table 4). There were no significant ROI response × sex interactions. Whole-brain analyses supported the ROI findings, with higher ER difficulties associated with higher activation in insula, ACC, vmPFC, caudate, and putamen, and in social cognition regions PCC, dmPFC, and precuneus (STF 2C).

There were no significant ROI response, response × sex interactions, or whole-brain findings for depressive or anxiety symptoms or SU.

Responses to own parent positive versus other parent positive

There was a significant R amygdala response × sex interaction to own parent positive emotion (–other parent positive) predicting ER difficulties (Table 5), but no significant ROI response main effects. For boys, higher R amygdala responses to own parent positive (–other parent positive) predicted greater ER difficulties (β = 0.32, P = .01). For girls, R amygdala responses did not predict ER (β = −0.19, P = .08). Whole-brain analyses supported ROI findings, with ER difficulties × sex predicting activation to own parent positive (–other positive) in R amygdala (STF 2D). These findings remained significant when covarying for pubertal level.

There were no significant ROI response or response × sex interactions predicting anxiety symptoms. In whole-brain analyses, anxiety symptoms × sex-predicted responses to own parent positive emotion (–other parent positive) in amygdala, insula, caudate/putamen, and thalamus (STF 2E). For boys, anxiety symptoms were related to greater activation. This finding remained significant when covarying for pubertal level.

There were no significant ROI response, response × sex interactions, or whole-brain findings for depressive symptoms or SU. There was a R NAc × sex interaction predicting SU that fell out of significance with FDR correction (with higher R NAc predicting SU for boys).

Discussion

The present study was the first to examine adolescents’ neural responses to parental emotion expressions from naturalistic parent–adolescent interactions in association with internalizing symptoms and also ER and SU. Findings indicated that early adolescents’ neural responses to parental negative emotion were associated with ER difficulties and anxiety and depressive symptoms, responses to parental positive emotion were associated with anxiety, and responses to own parent-specific positive emotion showed sex differentiated associations. Findings provide a neurobiological basis for how adolescents’ responses to parental emotion are associated with ER and symptoms.

Responses to parent negative emotion

Early adolescents’ heightened reactivity to their parents’ negative emotional expressions (versus neutral) in salience regions, including vACC, AI, and NAc/striatum, were associated with greater ER difficulties and higher depressive and anxiety symptoms. Higher responses in a social cognition region (dmPFC) were also associated with depressive symptoms. Notably, responses to parental negative emotion were not associated with SU, possibly because our sample included early adolescents with low-level use.

Our findings for ER and internalizing symptoms provide replication of prior studies with small samples that found higher salience region responses to negative parent emotional stimuli predicting depressive and anxiety symptoms (Aupperle et al. 2016, Silk et al. 2017). Taken together, this work suggests that adolescents’ increased neural sensitivity/reactivity to negative parent emotions may play a role in developing ER difficulties and internalizing symptoms, although longitudinal studies are needed to determine direction of effects. If neural reactivity to parental emotion predicts ER difficulties and symptoms, prevention programs could identify early adolescents with high sensitivity to parental negative emotion and support adolescents in coping with these arousal responses. Parents of early adolescents with high sensitivity to emotion can be supported in expressing their negative emotions in modulated ways that are sensitive to their child’s reactivity levels.

Given that the present study is cross-sectional, it is also possible that youth may first develop ER difficulties and internalizing symptoms (perhaps due to a genetic predisposition), which leads to a bias to respond to negative parent emotions with high reactivity. Prior work has shown that youth with internalizing symptoms show higher salience system reactivity to general negative emotional stimuli (e.g. Hall et al. 2014) and parental negative emotion may represent just one (potentially important) form of negative emotional stimuli. This suggests that interventions may help adolescents with internalizing symptoms to cope with heightened reactivity to negative emotional stimuli generally, including parental negative emotion.

Reactivity to parent positive emotion

Adolescents’ greater reactivity to their own parents’ positive emotion (versus neutral) in salience regions, including vACC and AI, and in dmPFC, were associated with greater anxiety symptoms, but not with ER, depressive symptoms, or SU. This pattern is consistent with one finding that anxiety symptoms were associated with heightened activation in smPFC to anticipating maternal praise (though anxiety symptoms were associated with blunted activation to receiving praise; Aupperle et al. 2016). Adolescents with anxiety may look out for parent positive emotion (and possible praise or reassurance) during parent–youth interactions and respond strongly to parent positive emotions when they occur. Parents of youth with anxiety have been shown to use accommodation, providing reassurance and rescuing for youth when they feel uncomfortable (e.g. Gladstone et al. 2023), which negatively reinforces anxiety. Our finding suggests that youth with higher anxiety symptoms may be particularly sensitized to positive parent emotions, possibly including reassurance, from parents (in addition to being sensitized to negative parent emotion). Parent-focused interventions to reduce adolescent anxiety symptoms (e.g. Lebowitz et al. 2014) can use the present findings to bolster (and provide “brain”-based evidence for) their approach of coaching parents to use positive emotion and praise strategically to reinforce adolescents’ brave/adaptive behaviors and not anxious behaviors. Of note, the present study did not assess the context of parental positive emotion and future work can examine neural reactivity to parental positive emotion that is in the context of reassurance versus other contexts.

In terms of depressive symptoms, three prior studies (Whittle et al. 2012, Aupperle et al. 2016, Silk et al. 2017) found that adolescent depressive symptoms were associated with blunted neural responses in salience regions to parent positive emotional stimuli. The present study did not find associations between depressive symptoms and responses to parent positive emotion. The prior studies examined responses to maternal praise (Aupperle et al. 2016) and included youth with clinical depression (Silk et al. 2017) and older adolescents (Whittle et al. 2012). Thus, links between blunted neural responses to parent positive emotion and depressive symptoms may be stronger for praise, older adolescents, or more severe depression.

For SU, there was a nonsignificant trend for higher R NAc responses to parent positive emotion to be associated with SU. This could indicate greater reward system response to parent positive emotion as a correlate of SU, consistent with prior findings that striatal responses to other rewards predict SU (Bjork 2020). The finding may not have reached significant due to low rates of SU in this early adolescent sample.

Responses to own parent-specific emotion

The present study was able to isolate neural responses to emotion specifically from one’s own parent, controlling for responses to emotion by an unfamiliar parent. This is important in that these responses do not just indicate salience responses to negative emotion generally, but are personalized to the unique parent–adolescent relationship history. Adolescents’ greater reactivity to own parent negative emotion (versus other parent) in amygdala, vACC, AI, and NAc and in social cognition regions (e.g. dmPFC, precuneus) was associated with greater ER difficulties, but not depressive or anxiety symptoms or SU. The only past study that compared own parent negative emotion to other parent negative emotion was Whittle et al. (2012) and they similarly did not find associations with depression (they did not examine ER, anxiety, or SU). Taken together, these findings suggest that adolescent’s responses to negative emotion specific to their own family dynamic are more related to their ER than symptoms, at least during early adolescence.

Some youth may develop salience and social cognition region sensitization to their own parents’ negative emotions in particular, possibly due to a predisposition to high reactivity plus a history of parental negative emotion in the family environment. Over time, this vigilance to own parent negative emotion may limit youth’s engagement with negative emotions and learning of ER. Vigilance to own parent negative emotion and altered ER may be adaptive short term during early adolescence to manage challenges with parents (and thus may not show up as depressive/anxiety symptoms yet), but this heightened sensitivity to others’ negative emotions could lead to symptoms later in development as youth enter interpersonal challenges outside of the family. Given that our findings were cross-sectional, it is also possible that adolescents first develop ER difficulties, which leads them to respond with higher neural reactivity to negative emotional stimuli, including particularly to their own parent due to the personal nature of this relationship.

Sex moderation

For neural responses to own parent-specific positive emotion there were sex differences. Overall, girls showed higher salience responses than boys to own parent positive emotion (versus other parent), suggesting that girls respond more to parental positive emotion specific to their own family. Notably, our study included mostly mothers and so this may be driven by mother–daughter dynamics. However, higher R amygdala responses to own parent positive emotion (versus other parent) were associated with greater ER difficulties and greater anxiety symptoms for boys but not girls. And there was a nonsignificant pattern for NAc responses to own parent positive (versus other parent) to be associated with SU for boys.

This suggests that girls show high responsivity to own parent positive emotion, but when boys do show responses they are associated with more negative ER and anxiety. For anxiety, the pattern of sensitization to parent positive emotion linking to anxiety symptoms may be more common for boys, when it is in the context of own family history (such as, potentially, a history of reassurance by parents). For ER, heightened amygdala responses to own parent positive emotion may indicate heightened sensitivity to the rewarding quality of own parent positive emotion (given the amygdala’s role in reward/positive emotion) and heightened reward sensitivity has been linked to some forms of ER difficulties, such as impulsivity/sensation-seeking, possibly particularly for boys (Hammerslag and Gulley 2016).

Limitations/future directions

This study had several strengths, including a larger sample than prior work, community families oversampled for maladaptive parenting, the use of naturalistic stimuli from in vivo interactions, complementary ROI and whole-brain analyses, and examination of sex differences. However, the study also had limits. First, our interaction task was lab-based and may not reflect parent emotion in real-life contexts. Future studies could address this by combining fMRI with ecological momentary assessment. Second, our study collapsed across different types of parental negative emotions (e.g. anger, sadness), which have different implications for adolescents. Third, due to COVID restrictions, our study had a low number of two parent families and fathers, so we were unable to examine parent sex differences. Fourth, the study was cross-sectional with early adolescents. Future longitudinal studies should examine prediction into later adolescence, particularly for SU. Fifth, our sample was largely white and upper middle income, which limits generalizability and our ability to test for cultural differences in neural responses to parental emotion (Qu et al. 2021).

Overall, this study adds to the literature by finding that neural responses to parent emotion expressions were associated with adolescent ER and internalizing symptoms. Higher salience region responses to parent negative emotion were associated with ER and depressive and anxiety symptoms and higher responses to parent positive emotion were associated with anxiety symptoms. If replicated in future studies, this work can inform tailoring of interventions, suggesting a focus on reducing adolescents’ reactivity to parent negative emotion expression to prevent depressive and anxiety symptoms and a focus on parents using positive emotion in targeted ways for adolescents with anxiety symptoms.

Acknowledgements

The authors gratefully acknowledge the study sponsor, the participating families, and the study research assistants, particularly Claire Niehaus, Irene Regalario, Christopher Lance Hinagpis, Alec Gamboa, and Rachel Upton.

Supplementary data

Supplementary data is available at SCAN online.

Conflict of interest

None declared.

Funding

This work was supported by the National Institutes of Health (2-R01-DA-033431 to T.M.C. and F31-DA051154 to S.F.G.).

Data availability statement

De-identified data will be available upon request. Given the sensitive nature of the data (adolescent substance use), data will be provided with a data sharing agreement to use data for IRB-approved research purposes, store data securely, and destroy/return data after analyses.

References