Being praised and criticized in adolescents and adults: Age-related changes in neural responses to social evaluations

Abstract

Adolescence is a developmental period marked by heightened sensitivity to social evaluation due to the ongoing maturation of brain regions involved in emotion regulation and social cognition. Adolescents may be more vulnerable to socially evaluative feedback, particularly criticism, than adults. Although previous studies have documented age-related differences in neural responses to social evaluation, it remains unclear how the dynamic interactions among brain regions differ between these two age groups during social evaluation. In this study, sixty-four adolescents (14–17 years) and fifty-nine adults (20–40 years) were exposed to criticism, praise, and neutral comments during fMRI scanning. Dynamic causal modeling (DCM) was used to examine functional interactions among key brain regions involved in social evaluation: the right pregenual anterior cingulate cortex (pgACC), left dorsolateral prefrontal cortex (DLPFC), and left precuneus. Compared to adults, adolescents were more sensitive to criticism as they reported higher perceived criticism and more task-induced mood disturbance following criticism. DCM results showed that adolescents had stronger excitatory DLPFC-to-pgACC connectivity, and weaker excitatory pgACC-to-DLPFC and DLPFC-to-precuneus connectivity during social evaluation in general, compared to adults. Moreover, when being criticized, adolescents showed less excitation of the DLPFC-to-pgACC connectivity, whereas when being praised, they showed less inhibition of the pgACC-to-DLPFC connectivity. These findings highlight age-related differences in the neural dynamics underlying social evaluation processing, providing insight into adolescents’ heightened sensitivity to social evaluation and potential risk for developing affective disorders.

Graphical abstract

Highlights

• •Adolescents are more sensitive to criticism and showed greater mood disturbance induced by social evaluation.
• •fMRI reveals age-related differences in brain activation when processing social evaluative feedback.
• •Adolescents exhibit altered effective connectivity among pgACC, DLPFC, and precuneus.
• •The connectivity between pgACC and DLPFC were modulated by praise and criticism differently in adolescents compared to adults.

1. Introduction

Adolescence is a critical developmental period marked by significant changes in cognitive, affective, and social capacities, which are crucial for emotion generation and regulation (Crone and Dahl, 2012). As adolescents transition into adulthood, they face heightened demands on their emotion regulation systems (Pozzi et al., 2021, Steinberg, 2005), especially in socially evaluative contexts, where acceptance and rejection can significantly impact their emotional responses (Guyer et al., 2014, Heerdink et al., 2015). Positive social evaluation (e.g., praise) functions as a social reward, reinforcing socially desirable behaviors and fostering feelings of competence, acceptance, and belonging (Brudner et al., 2023, Izuma et al., 2010). In contrast, negative social evaluation (e.g., criticism) can signal social threat, rejection, or disapproval, and often elicits strong affective and physiological responses (Eisenberger et al., 2003, Gunther Moor et al., 2010, Sebastian et al., 2010). Examining both praise and criticism is essential for understanding adolescent sensitivity to social evaluation.

Neurodevelopmental models propose that socio-affective and cognitive functions follow a nonlinear developmental trajectory during adolescence (Blakemore, 2008, Casey et al., 2008). Regions of the prefrontal cortex, which are critical for behavior regulation and emotion control, continue to mature structurally and functionally during this phase. Simultaneously, connections between cortical and subcortical structures, such as the amygdala and limbic system, which are involved in emotional responsiveness, are still strengthening. This ongoing maturation limits adolescents' ability to exert cognitive control over their emotions, especially in emotionally challenging situations such as exposure to social evaluation (Somerville, 2013). Combined with the hormonal changes during puberty that also influence neurotransmitter systems and brain circuits (Casey et al., 2008, Steinberg, 2010), this mismatch in maturation makes adolescents’ emotional responses and behavior more contingent on social evaluation (Forbes and Dahl, 2010).

Therefore, social evaluations pose unique challenges for adolescents whose emotion regulation mechanisms are still developing. Importantly, during adolescence, evaluative feedback from close others, including parents and peers, can be especially emotionally meaningful, given the central role of both parent and peer relationships in adolescents’ social and emotional development (Morris et al., 2021, Morris et al., 2007, Schacter and Margolin, 2019). Although peer relationships become increasingly important during this developmental period, parental feedback continues to exert substantial influence on the emotional development and well-being of adolescents (Baumrind, 1991, Butterfield et al., 2021, Dobbelaar et al., 2023a). Being praised or criticized by close others (i.e., parents and peers) can trigger changes in mood states and/or self-referential or ruminative thoughts (Bonduelle et al., 2023, Chen et al., 2024, De Raedt et al., 2017, Dedoncker et al., 2019, van Houtum et al., 2022), which may need to be regulated to prevent maladaptive behavioral responses (Butterfield et al., 2021, Vanderhasselt et al., 2015, Vandermeer et al., 2022). Unlike adults, who typically have well-developed coping mechanisms, adolescents are neurobiologically prone to experiencing heightened emotional reactivity to social evaluation, making it more difficult to regulate their (more intense) emotions (Guyer et al., 2016, Rodman et al., 2017). This vulnerability may help explain why adolescence is a critical period for the development of affective disorders, such as depression and anxiety disorders (Rapee et al., 2019).

Although praise and criticism differ in valence, both forms of feedback within social context can implicate similar neural processes. Neuroimaging studies have revealed the engagement of brain regions associated with affective salience (e.g., anterior cingulate cortex [ACC], insula), self-referential processing (e.g., posterior cingulate cortex [PCC]/precuneus), and cognitive control (e.g., dorsolateral prefrontal cortex [DLPFC]) when people received either positive or negative social feedback (Aupperle et al., 2016, Butterfield et al., 2021, Lee et al., 2015). For example, compared to neutral feedback, both positive and negative feedback elicit stronger ACC and insula activity in healthy adolescents and adults (Achterberg et al., 2016, Achterberg et al., 2018, Miedl et al., 2016). Negative feedback elicits stronger and longer lasting activity in these regions compared to positive feedback (Achterberg et al., 2016, Hooley et al., 2012, Servaas et al., 2013). Previous studies have also suggested that social evaluative feedback, which commonly involves self-relevant information, relates to increased precuneus activation, especially during negative feedback, indicating heightened self-referential processing (van Houtum et al., 2021, van Schie et al., 2018, 2020). At the same time, receiving social evaluative feedback often requires the engagement of coping strategies such as mentalizing, perspective taking (Andrews et al., 2021, Sebastian et al., 2010), and cognitive control to regulate any excessive emotions that may have been triggered and to prevent behavioral maladaptive responses (Ahmed et al., 2015, Silvers et al., 2012). The DLPFC plays an important role in this process because it supports cognitive control and emotion regulation (Etkin et al., 2011, Ochsner and Gross, 2005). Achterberg et al. (2020) found that the DLPFC was more active after positive than negative feedback in minors. Increased DLPFC activity following negative feedback was associated with less subsequent behavioral aggression. Hooley et al. (2005) showed that exposure to criticism elicited higher DLPFC activity in healthy adults. Those who activated the DLPFC to a lesser extent during exposure to criticism reported a greater increase in negative affect. These findings have linked greater DLPFC engagement during social evaluation to better cognitive control and emotion regulation.

Developmentally, adolescents show heightened affective reactivity and less mature regulatory recruitment than adults during social evaluation. When being criticized, adolescents more strongly activate affective salience regions (e.g., ACC, insula) but engage cognitive control (e.g., DLPFC) and self-referential processing/social cognition (e.g., PCC, precuneus) regions to a lesser extent than during neutral feedback (Lee et al., 2015, van Houtum et al., 2022). During adolescence, positive feedback often elicits relatively strong responses in these brain regions, compared with negative feedback (Dobbelaar et al., 2023b, Gunther Moor et al., 2010, Guyer et al., 2012, van Houtum et al., 2022). This is consistent with the increased motivational salience of social acceptance during this period, including both peer (Crone and Dahl, 2012, Foulkes and Blakemore, 2016, Somerville, 2013) and parental acceptance (Barber et al., 2005, Sequeira et al., 2019). Research has indicated that adolescents can recruit the DLPFC during exposure to both positive and negative feedback (relative to neutral feedback), suggesting an emerging capacity to recruit cognitive control as they attempt to regulate inappropriate or impulsive responses to social feedback, although this control is often less effective than in adulthood (Achterberg et al., 2020, Dobbelaar et al., 2023b). Adults, on the other hand, show attenuated affective responses and greater engagement of cognitive control and self-referential processing/social cognition in response to negative feedback, compared with adolescents (Burnett et al., 2011, Crone et al., 2020, Peters et al., 2014, Zanolie et al., 2008). These differences may reflect more mature social-affective and cognitive processes that support the regulation of emotions and the ability to disengage from negative self-relevant information in adulthood. During positive feedback, adults tend to engage these neural processes to a lesser extent compared with adolescents, which may indicate that social approval/acceptance is less important to adults (Crone et al., 2020, Foulkes and Blakemore, 2016, van der Cruijsen et al., 2018).

Although prior research in adolescents and adults has shown engagement of several brain regions involved in the processing of social evaluative information, such as the ACC, DLPFC, and precuneus, no study to date has examined the functional interactions among these regions across adolescent and adult groups. It has been suggested that the ACC (specifically the pregenual subdivision) plays a role in salience detection, evaluating and integrating external stimuli (such as social evaluation) to guide regulation through its connected cortical regions, including the DLPFC and precuneus (Shenhav et al., 2013, Tang et al., 2019). Developmentally, connectivity of the ACC has been identified as a key neural substrate of adolescent neurodevelopment and supports the emergence of adult regulatory capability (Lichenstein et al., 2016). The protracted maturation of DLPFC during adolescence may limit adolescents’ ability to effectively regulate evaluation-evoked emotions, contributing to heightened sensitivity to social evaluation (Casey et al., 2008, Crone et al., 2020, Crone and Konijn, 2018). Previous studies have shown that DLPFC-precuneus connectivity is engaged during social evaluation, particularly criticism, in adolescents and adults (Chen et al., 2024, Lee et al., 2015, Servaas et al., 2013). The precuneus, a hub of the default mode network (DMN), is associated with self-referential processing and social cognition, including adopting another’s perspective to adjust one’s self-view (Menon, 2023, Whitfield-Gabrieli and Ford, 2012). The connectivity between DLPFC and precuneus may support one’s ability to recruit cognitive control over self-referential processing (Bertocci et al., 2023). In addition, adolescents at high risk for affective disorders exhibit altered ACC connectivity, including greater coupling with DMN regions and reduced coupling with the DLPFC (Chai et al., 2016), which may reflect an over engagement of negative affective systems together with inefficient top-down regulation and difficulty shifting away from self-referential processing (Lichenstein et al., 2016). Taken together, age-related changes in the interactions among the ACC, DLPFC, and precuneus are likely to support the maturation of higher-level affective and social-cognitive functions, and may help explain heightened sensitivity to social evaluation during adolescence.

One way to investigate these interactions is by using Dynamic Causal Modeling (DCM), which estimates effective connectivity – the directional influences that one region exerts over another – and how external inputs (i.e., experimental manipulations) modulate these influences (Friston et al., 2003, Stephan and Friston, 2010). DCM overcomes limitations in alternative approaches, such as psychophysiological interaction (PPI), which estimates context-dependent changes (i.e., through experimental conditions) in functional connectivity between brain regions and does not infer directionality of that connectivity (Friston et al., 1997). Using DCM, we investigated how three key brain regions (i.e., ACC, DLPFC, and precuneus) interacted during social evaluation in adolescents in a previous study (Chen et al., 2024); however, it remains unclear whether the interplay between these regions is different in (young) adulthood.

The present study aimed to compare adolescents to (young) adults in terms of neural activity in response to social evaluation from close others, and to study how the brain regions involved in processing social evaluative information interact with one another when exposed to it, using functional magnetic resonance imaging (fMRI) with DCM. We used an ecologically validated paradigm developed by Hooley et al. (2009) to mimic the experience of being evaluated by exposing participants to a series of audio clips containing critical, praising, and neutral comments during fMRI scans (Fig. 1). We sought to investigate differences in brain responses to different valences of comments (i.e., positive, negative, and neutral) between the adolescent and adult groups. Another major aim was to assess the changes in the effective connectivity between these brain regions – where we focus on the ACC, DLPFC, and precuneus – and the differences in modulatory effect of criticism and praise on this connectivity between the two age groups. We also examined whether perceived criticism and mood changes affected neural connectivity patterns in either group.

Fig. 1.

Experimental design. Participants were exposed to a series of self-referential auditory comments made by a female voice during fMRI scans. All participants first heard two neutral comments followed by two positive comments (i.e., praise), and then another two neutral comments followed by two negative comments (i.e., criticism). The negative auditory comment always started with “One of the things that bothers me about you is…”, while the positive auditory comment always started with “One of the things I really like about you is…”, and the neutral auditory comment always started with “One of the things you did…”. Each comment lasted 30 s and was interspersed with 30 s of silence. NEU, neutral comment; POS, positive comment; NEG, negative comment.

We hypothesized that adolescents, compared to adults, would be more sensitive to criticism as they would perceive more criticism and experience a greater mood disturbance (i.e., a higher increase in negative mood state) after being exposed to criticism. We also expected that the two age groups would differentially engage affective (e.g., ACC), control (e.g., DLPFC) and self-referential (e.g., precuneus) regions in response to social evaluation. Based on our prior findings in adolescents (Chen et al., 2024), we further hypothesized that the adults would exhibit stronger modulatory effect of criticism on DLPFC-ACC connectivity and DLPFC-precuneus connectivity, but a weaker modulatory effect of praise on DLPFC-ACC connectivity, compared to adolescents.

2. Methods

2.1. Participants

One hundred twenty-three healthy individuals (aged 14 – 40 years) were included in this study. The adolescent cohort (N = 96; aged 14 – 17 years) and the adult cohort (N = 77; aged 20 – 40 years) were obtained from two separate studies conducted in our laboratory. The adolescent cohort has been reported previously (Chen et al., 2024). The studies were approved by the Ghent University Hospital (UZ Gent) Medical Ethics Committee, and were carried out in accordance with the Declaration of Helsinki (World Medical Association, 2004). Written informed consent was obtained from all the participants and the parents of the adolescents. All the participants were required to be proficient in Dutch and were financially compensated for participating in the study.

Participants were selected for the current study if there was (i) no current or history of psychiatric disorders (adolescents were screened with the Structured Clinical Interview for DSM-5 Junior Edition, SCID-5-Junior; First et al., 2016; adults were screened with the Structured Mini-International Neuropsychiatric Interview, M.I.N.I.; Sheehan et al., 1998); (ii) no current use of psychotropic medication; (iii) no current or history of neurological problems or implanted metal objects in/over the head. Fifty participants were excluded due to technical issues during MRI data acquisition (N = 3), excessive head motion (mean framewise displacement greater than 0.2 mm) during fMRI scanning (N = 25), excessive distortion of images because of wearing braces (N = 20), and having a diagnosis of major depressive disorder (N = 2). In total, sixty-four participants in the “adolescent group” (42 girls/22 boys; aged 14–17 years), and fifty-nine participants in the “adult group” (34 female/25 male; aged 20–40 years) were retained for data analyses.

2.2. Experimental paradigm

We used an experimental paradigm designed to be as ecologically valid as possible to evoke participants’ emotional experiences through exposure to critical, praising, and neutral comments (Hooley et al., 2009). Specifically, participants were asked to listen attentively to a series of self-referential auditory comments made by a female voice actor during fMRI scanning (Fig. 1). The adults were instructed to imagine someone who was really close to them talking directly to them, while the adolescents were told to imagine that one of their parents was talking to them. The comments included positive (praising), negative (critical), and neutral comments. These were adapted from Hooley et al.’s original experiment (2009; 2005) and standardized to be broadly applicable and likely to be personally relevant for a wide range of participants. The critical and praising comments always started with a sentence that indicated how one felt about the participants (critical comments started with “One of the things that bothers me about you is …”; praising comments started with “One of the things I really like about you is …”). The neutral comments described something common in daily life (e.g., grocery shopping, weather, …), and always started with “One of the things you did [time: last week, this week, yesterday, …] is …”. The praising and critical comments were identical for adolescents and adults. The neutral comments used for adolescents were slightly different from the ones used for the adults, which were adapted to be applicable to the adolescent audience. The full list of available comments can be found in the Supplementary Materials.

All participants first listened to two neutral comments followed by two praising comments, and then two neutral comments followed by two critical comments (Fig. 1). This specific order was chosen according to the affective contrast theory (Manstead et al., 1983), which suggests that the impact of an emotional experience (i.e., being criticized) is contingent on the level of contrast with a preceding emotional state (i.e., being praised). Each comment lasted 30 s and was interspersed with 30 s of silence. The total duration of the paradigm was 8.5 min.

2.3. Behavioral measurements

Perceived criticism. The Perceived Criticism Measurement (PCM, Hooley et al., 2012) was administered to measure the participants’ perception of criticism, prior to the scan session. Participants were asked to rate the question “How critical are your significant others - such as family, friends, … - of you?” on a scale ranging from 0 (i.e., not critical at all) to 10 (i.e., very critical). Higher scores on the PCM indicate higher perceived criticism (range: 0–10).

Momentary mood assessment. Momentary mood states were only assessed before and after the entire experimental paradigm (but not after each set of comments with a specific valence), using six Momentary Mood State (MMS) scales (McNair et al., 1992) measuring fatigue, vigor, anger, tension, sadness, and cheerfulness. Participants were asked how they felt “at this moment” for each of these mood states, on a scale ranging from 0 (“not at all”) to 10 (“very much”). The Total Mood Disturbance Score (TMDS) was calculated by summing all MMS scales after reverse-scoring the “vigor” and “cheerfulness” scales. Higher scores on the TMDS indicate more negative affect (range: 0–60).

2.4. Demographic and behavioral data analysis

Demographic characteristics and behavioral measurements were analyzed using JASP (version 0.16.3). We applied a Chi-Square test to detect significant differences regarding sex distribution between the adolescent and the adult groups. Two-sample t-tests were performed to compare between-group differences in PCM scores and TMDS delta scores (i.e., after the experimental paradigm minus before the paradigm). To investigate the mood state changes after the experiment (based on the Total Mood Disturbance Score [TMDS] measured before and after the paradigm), we used a mixed ANOVA for TMDS with the pre- and post-TMDS as a repeated measures factor and group as a between-subject factor. In follow-up analyses, we used paired-sample t-tests to compare TMDS before and after the experimental paradigm within each group. We used p < 0.05, two-tailed, as the criterion for significance for all our analyses.

2.5. MRI data acquisition

Neuroimaging data were collected at Ghent University Hospital in Belgium using a Siemens 3 T Prisma-fit MRI scanner (Siemens, Erlangen, Germany) with a 64-channel head coil. The functional data were obtained using a single-shot gradient-echo echo-planar imaging (EPI) sequence with the following parameters: repetition time (TR) = 2500 ms, echo time (TE) = 35 ms, voxel size = 3.5 × 3.5 × 3 mm³, flip angle = 80°, data matrix = 64 × 64, field of view (FOV) = 224 mm × 224 mm, and 42 slices in interleaved ascending order. The functional MRI scan lasted around 9 min (∼220 volumes were acquired). High-resolution brain structural images were acquired with a T1-weighted Magnetization-Prepared Rapid Gradient Echo (MPRAGE) sequence (TE/TR = 4.18/2250 ms; flip angle = 9°; slice thickness = 1 mm; FOV = 256 mm × 256 mm; data matrix = 256 × 256; voxel size = 1 × 1 × 1 mm³; 176 sagittal slices).

2.6. MRI data preprocessing

MRI data were preprocessed using fMRIPrep v20.2.1 (Esteban et al., 2020) and preprocessing steps were identical to Chen et al. (2024). The T1-weighted images were corrected for intensity non-uniformity, skull-stripped, segmented, and spatially normalized to MNI standard space. Gray matter volumes and cortical thickness values were automatically calculated. Functional data were subsequently preprocessed with the following steps: (1) estimation of head-motion parameters and motion correction; (2) slice-timing correction; (3) co-registration to the T1-weighted structural images; (4) spatial normalization to the MNI standard space; and (5) estimation of confounding factors, including framewise displacement (FD) and global signal. Details of the preprocessing are presented in the Supplementary Materials. Further details on the fMRIPrep pipeline can be found in the fMRIPrep documentation (https://fmriprep.org/).

2.7. General linear model (GLM) analysis for fMRI data

In order to explore the altered brain responses to different emotional valences of auditory comments between adolescents and adults, data analyses were performed by using Nilearn (https://nilearn.github.io/) and SPM12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12/). Data analyses were focused on the block of time during the presentation of the auditory comments. A band-pass filtering of 1/128 Hz and a spatial smoothing with a Gaussian kernel at full-width-at-half-maximum (FWHM) of 6 mm were applied to the preprocessed fMRI data. Then, a first-level GLM design matrix was generated including the three experimental conditions (i.e., positive, negative, and neutral) as regressors of interest. Six head motion parameters, FD, white matter, and cerebrospinal fluid (CSF) signals were additionally added as nuisance regressors. Individual contrast images between each condition (i.e., negative vs neutral, positive vs neutral, and negative vs positive) were created respectively.

In the second-level, a flexible factorial design (Gläscher and Gitelman, 2008) was applied to compute the effect of the task. The design matrix included three factors: a between-subject factor (i.e., subject) that was specified as independent and with equal variance; a second between-subject factor (i.e., adolescent group or adult group) that was specified as independent and with equal or unequal variance; and a within-subject factor (i.e., negative, positive, and neutral comments) that was specified as dependent and with equal variance. Sex, mean FD, and total gray matter volumes were included as covariates. Contrast images for the interaction effect of group (i.e., adolescent, adult) and valence (i.e., negative, positive, and neutral comments) were computed, and corrected with Gaussian random field (GRF) correction (voxel-level p < 0.005, cluster-level p < 0.05). In addition, two-sample t-tests were applied for individual contrast images to compare group differences in the brain activation between each condition (i.e., negative vs neutral, positive vs neutral, and negative vs positive), thresholded with Gaussian random field (GRF) correction (voxel-level p < 0.005, cluster-level p < 0.05). Sex, mean FD, and total gray matter volumes were added as covariates.

2.8. Dynamic causal modeling (DCM) analysis

To investigate the differences between adolescents and adults regarding interactions within the neural network involved in social evaluation, and how this network was modulated by our experimental manipulations (i.e., being criticized and praised), we performed a DCM analysis (using SPM12) to estimate the effective connectivity between the ROIs based on task-related brain responses (Stephan et al., 2010). In this study, we included the pgACC, the DLPFC, and the precuneus as ROIs, in alignment with our previous study (Chen et al., 2024), as these brain regions were consistently reported to be involved in emotion regulation processing, and exhibited significant activation in the interaction effect between group and valence in the GLM analysis. Specifically, we sought to examine the differences between adolescents and adults in (i) effective connectivity among the right pgACC, left DLPFC, and left precuneus, and (ii) the modulatory effects of our experimental manipulations (i.e., being criticized and being praised) on the between-region effective connectivity. In addition, we also examined whether the individual differences in sensitivity to criticism (i.e., PCM scores), and mood changes induced by task (i.e., delta scores of TMDS) would influence the modulatory effects of our experimental manipulations.

First, we identified the peak coordinates within the three ROIs based on the interaction effect of group and valence. The peak coordinates at the group level of each ROI were located in the right pgACC [12, 46, 2], left DLPFC [-16, 32, 41], and left precuneus [-9,−80, 48]. Therefore, these three brain regions were included as ROIs in the DCM model (Fig. 2a). It should be noted that in the present study the pgACC node was right lateralized, whereas in our previous adolescent study (Chen et al., 2024) the peak was located in the left pgACC. This shift from left to right may reflect developmental changes in pgACC involvement in emotion processing, consistent with evidence that the right pgACC is more strongly engaged in affective appraisal and emotion reactivity (Brügger et al., 2011, Watanabe et al., 2015). For each participant, the individual peak coordinates of each ROI were constrained to be a maximum of 10 mm from the group-level peak. Each ROI for each participant was identified using a sphere with a 5 mm radius around the individual peak, ensuring that the selected voxels were engaged in task-related activity. Then, the individual time series of each ROI was extracted, summarized by the first eigenvariate, and adjusted using an F-contrast involving all experimental effects.

Fig. 2.

DCM model architecture. a), ROIs included in the DCM: right pgACC, left DLPFC, and left PreCUN. b), Schematic of the DCM model specification. The light blue circles represent the nodes (ROIs) comprising the model. The model includes: (i) averaged connections over experimental conditions (bidirectional connections between ROIs and self-connections) in A matrix, (ii) modulation by criticism and praise in B matrix, and (iii) driving input by task in C matrix. Abbreviations: pgACC, pregenual anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; PreCUN, precuneus; L (R), left (right) hemisphere.

A full DCM model for each participant was set up (Fig. 2b) to include three parameters: (i) bidirectional connections between the ROIs and their self-connections over experimental conditions in matrix A; (ii) modulatory effects of experimental manipulations (i.e., being criticized and praised) on the between-region connections in matrix B; and (iii) driving input (i.e., audio input condition: the combination of negative, positive, and neutral conditions) into the model on both regions in matrix C. We did not specify neutral conditions as modulatory effects in the DCM model because the negative (being criticized) and positive (being praised) conditions were our main focus, and adding no-interest conditions would have increased the model’s complexity. Then, the full model for each participant was individually estimated. Finally, the group-level DCM inference was performed using a Parametric Empirical Bayes (PEB), which integrates the individual-level DCM to a group-level GLM (Zeidman et al., 2019). We specified a PEB model which included two regressors encoding (i) the group mean of connectivity parameters across all subjects, and (ii) between-group differences in connectivity parameters, which were of primary interest in this study. Additionally, to investigate whether individual differences influenced the modulatory effect, the behavioral measurements (i.e., PCM scores and TMDS delta scores) were added as regressors in the design matrix. Sex was added as a covariate in the PEB model. In the DCM analysis, we excluded seven participants due to their lack of PCM and/or TMDS scores. After specifying and estimating the PEB model, we conducted an automatic search procedure called ‘Bayesian model reduction (BMR)’ to search over nested PEB models, pruning away any parameters (i.e., connections) that did not contribute to overall model evidence, and to compare model evidence among the full and reduced models. A Bayesian model average (BMA) was then used to compute the average of connectivity parameters across all models, weighted by the models’ posterior probability (Pp). The Pp was calculated using the free energy option, which compares the evidence for all models in which a particular parameter is switched on with the evidence for models in which it is switched off. The parameters were considered as significant when their Pp > 0.95 (i.e., strong evidence).

3. Results

3.1. Demographic and behavioral measurements

Sixty-four healthy adolescents (aged 14–17) and 59 (young) adults (aged 20–40) participated in this experiment (see Table 1). We used the Perceived Criticism Measurement (PCM) to assess the participants’ sensitivity to criticism before the experiment, and the Total Mood Disturbance Score (TMDS) to assess their mood state before and after the experiment. Seven participants in the adult group were excluded from behavioral data analysis due to the lack of a PCM and/or TMDS scores. There were no missing data in the adolescent group.

Table 1.

Demographic characteristics of adolescent and adult groups.

Characteristics	Adolescents (N = 64)	Adults (N = 59)	p values
Age (years)	16.00 ± 1.05	30.28 ± 4.74	< 0.001
Sex (female/male)	42/22	34/25	0.36

Note: A two sample t-test was used to detect differences in age between adolescents and adults. A χ2-test was used to detect differences in gender between adolescents and adults. Values are mean ± SD or n.

There was no significant difference regarding sex distribution between the adolescent and the adult groups (χ²₁ = 0.83, p = 0.36). The PCM and TMDS delta scores were significantly higher in the adolescent group than in the adult group (PCM: t₁₁₄ = 4.48, p < 0.001, Cohen’s d = 0.837; TMDS delta scores: t₁₁₄ = 5.14, p < 0.001, Cohen’s d = 0.96), suggesting that the adolescents had a higher tendency to perceive their significant others as critical of them, and that they experienced more mood state disturbance after the experiment (i.e., being exposed to evaluative feedback) than the adults (Fig. 3a and Fig. 3b).

Fig. 3.

Behavioral measurement results. Two-sample t-tests were used to detect differences in sensitivity to criticism (PCM scores), and changes in negative mood state (delta TMDS scores) between adolescents and adults. A paired-sample t-test was used to detect mood disturbance (TMDS scores) before and after exposure to criticism for each group respectively. a), PCM scores for both adolescent and adult groups. b), Delta TMDS scores (i.e., post- minus pre-experimental paradigm measurement) for both adolescent and adult groups. c), Pre- and post-experimental paradigm measurement scores for TMDS in adolescent and adult groups. Abbreviations: PCM, Perceived Criticism Measurement; TMDS, Total Mood Disturbance Score; Pre, before experimental paradigm measurement; Post, after experimental paradigm measurement; ***, p < 0.001.

Regarding the TMDS scores, the mixed ANOVA revealed a main effect of time (pre- and post-experiment), F((1, 114)= 27.79, p < 0.001, η² = 0.03; a main effect of group, F((1, 114)= 4.44, p = 0.04, η² = 0.03; and an interaction effect between time and group, F((1, 114)= 26.45, p < 0.001, η² = 0.03. Post-hoc paired-sample t-tests revealed that only the adolescent group showed a significantly increased TMDS after the experiment (t₆₃ = 6.39, p < 0.001, Cohen’s d = 0.798), which indicates that the adolescents experienced more mood state disturbance after the experiment. This was not the case for the adult group (t₅₁ = 0.14, p = 0.89, Cohen’s d = 0.019) (Fig. 3c).

3.2. Brain activation in response to social evaluation between age groups

The contrast image for the interaction effect of group (i.e., adolescent, adult) and valence (i.e., negative, positive, and neutral comments) was computed and corrected with Gaussian random field (GRF) correction (voxel-level p < 0.005, cluster-level p < 0.05). In alignment with our previous study on the adolescent sample (Chen et al., 2024), similar brain regions (such as bilateral DLPFC, bilateral precuneus, and the right pregenual ACC) showed significant activation in the group (i.e., adolescent, adult) and valence (i.e., negative, positive and neutral comments) interaction (Fig. 4). A detailed list of significantly activated brain regions can be found in Table 2.

Fig. 4.

GLM analysis results. Brain activations of the interaction effect of group (adolescents, adults) and valence (positive, negative, neutral condition). Brain activations are thresholded with Gaussian random field correction (voxel-level p < 0.005, cluster-level p < 0.05). Abbreviations: pgACC, pregenual anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; L (R), left (right) hemisphere.

Table 2.

Brain activations interaction effect of group (adolescents, adults) and valence (negative, positive, neutral comments).

Location	MNI coordinates of peak voxel			F-value	Cluster size
	X	Y	Z		(#voxels)
Interaction effect: group x valence
Occipital_Inf_L	-51	-70	-15	14.94	262
Cerebellum_6_L	-9	-66	-22	11.18
Lingual_L	-22	-62	-1	10.02
Fusiform_L	-30	62	-8	8.03
Occipital_Sup_R	22	-84	27	14.31	1301
Occipital_Sup_L	-20	-80	27	13.41
Lingual_R	26	-48	-5	13.01
Occipital_Mid_L	-34	-73	27	12.79
Cuneus_R	16	-90	23	12.69
Parietal_Sup_L	-16	-80	51	11.04
Temporal_Mid_R	44	-70	23	10.51
Cuneus_L	-12	-87	37	10.51
Precuneus_L	-9	-80	48	9.39
Temporal_Sup_R	54	-10	2	13.93	186
Precentral_R	40	-20	41	13.05	61
Temporal_Sup_L	-40	-38	16	12.51	81
Frontal_Sup_L	-16	32	41	10.41	36
Postcentral_L	-37	-20	48	9.72	59
Parietal_Sup_R	22	-59	65	8.69	42
Frontal_Sup_R	5	50	34	8.33	33
ACC_pre_R	12	46	2	7.16

Note: L (R), left (right) hemisphere

To identify how the adolescents’ brain regions reacted differently to each contrast compared to adults, two-sample t-tests were applied to individual contrast images (i.e., negative vs neutral, positive vs neutral, and negative vs positive). Brain regions with significant group differences in response to different conditions are shown in Supplementary Fig. S1 and listed in Supplementary Table S1.

3.3. DCM results

3.3.1. Differences in effective connectivity between age groups

Our DCM analyses on effective (intrinsic) connectivity showed age-related differences in how the three brain regions (i.e., pgACC, DLPFC, and precuneus) interacted during social evaluation. Compared to adolescents, adults exhibited stronger excitatory connectivity from the pgACC to the DLPFC (0.08 Hz, Pp = 0.95), but weaker excitatory connectivity in the reverse direction – from DLPFC to the pgACC (-0.21 Hz, Pp = 1.00). Moreover, adults exhibited stronger excitatory connectivity from the DLPFC to the precuneus (0.26 Hz, Pp = 1.00), and stronger self-inhibition of the precuneus (-0.16 Hz, Pp = 0.97), when compared to adolescents.

3.3.2. Differences in task-related modulation (criticism and praise) between age groups

The primary interest of the current study focuses on how these connections were modulated when being exposed to criticism or praise between adolescents and adults. The DCM results of modulatory effects showed that compared to the adolescents, adults exhibited stronger inhibitory modulation of the pgACC-to-DLPFC connectivity when being praised (-0.3 Hz, Pp = 0.99), and stronger excitatory modulation of the DLPFC-to-pgACC connectivity when being criticized (0.31 Hz, Pp = 0.99).

The between-group differences in effective (intrinsic) connectivity and modulation by experimental conditions (i.e., praise, criticism) are shown in Fig. 5. Positive values indicate stronger effects in the adult group, whereas negative values indicate weaker effects in the adult group. Group-averaged parameters for each group are shown in Fig. 6. For effective (intrinsic) connectivity, positive values indicate excitation of the connectivity and negative values indicate inhibition of the connectivity. For modulatory effects, positive values indicate that the condition (i.e., being praised or criticized) increases the strength of the connection (i.e., excites) and negative values indicate that the condition decreases the strength of the connection (i.e., inhibits).

Fig. 5.

Differences in DCM between adolescents and adults (adults > adolescents). a), Schematic illustration of the effective connectivity and modulatory effects of experimental conditions that show significant group differences (adults > adolescents) in the DCM model (Pp > 0.95). For the effective connectivity (A matrix), the black arrow, highlighted in orange, represents connections with significant group differences (with a Pp > 0.95), while the gray arrow represents connections that do not show significant group differences (with a Pp < 0.95). For the modulatory connectivity (B matrix), only those with a Pp > 0.95 in group comparisons are shown. The full line represents increased connectivity, while the dotted line represents decreased connectivity in adults when compared to adolescents. b), The matrices corresponding to connectivity parameters of between-group differences in effective connectivity (top) and modulatory effect of praise/criticism (down). The asterisks indicate statistically significant connectivity parameters in between-group comparisons. Positive parameters indicate stronger connectivity in the adult group than in the adolescent group, while negative parameters indicate weaker connectivity in the adult group than in the adolescent group. Abbreviations: pgACC, pregenual anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; PreCUN, precuneus; L (R), left (right) hemisphere; N.A., non applicable.

Fig. 6.

Group-averaged DCM parameters of adult and adolescent groups. a), The effective connectivity (left), modulatory effects of praise (middle) and criticism (right) in the adult group. b), The effective connectivity (left), modulatory effects of praise (middle) and criticism (right) in the adolescent group. Abbreviations: pgACC, pregenual anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; PreCUN, precuneus; L (R), left (right) hemisphere; N.A., non applicable.

3.3.3. Associations with individual differences

No significant associations were found between the modulatory effects and individual differences in perceived criticism or mood state changes (see Supplementary Table S2) across all participants. Exploratory analyses of associations between effective (intrinsic) connectivity and these behavior indices are reported in the Supplementary Table S3.

4. Discussion

In this study, adults and adolescents were exposed to a series of audio clips that included positive (praising), negative (critical) and neutral comments to elicit their experience of being evaluated. Using Dynamic Causal Modeling (DCM), we investigated how three key brain regions involved in affective, cognitive, and self-referential processing (i.e., pgACC, DLPFC, and precuneus) interacted during the processing of social evaluation (i.e., criticism and praise) and how these interactions differed between age groups (i.e., adolescents and adults). Our findings shed more light on age-related differences in social evaluative processing and emotion regulation. Because the evaluator cue differed between groups (i.e., adolescents imagining being evaluated by a parent, and adults imagining being evaluated by a close other), we interpret these findings in the context of personally meaningful close relationship evaluation.

The adolescent group reported a higher tendency to perceive their significant others as critical of them than the adult group, as their significantly higher PCM scores indicate. This was also reflected in the greater mood state disturbance they experienced after the experimental paradigm (which ended with critical remarks). These observations are consistent with our hypothesis that adolescents are more sensitive to criticism and struggle more with negative social evaluation, highlighting the developmental psychological vulnerabilities of adolescents (Powers and Casey, 2015, Rodman et al., 2017).

Our DCM analyses revealed age-related group differences in functional interactions among the three brain regions (i.e., pgACC, DLPFC, and precuneus) during the processing of social evaluation. Specifically, adolescents demonstrated stronger excitatory connectivity from the DLPFC to the pgACC, but weaker excitatory connectivity in the reverse direction during social evaluation in general, when compared to adults. The DLPFC, as a key region of the Central Executive Network (CEN), is implicated in higher-level cognitive control (Barch, 2013, Crone and Steinbeis, 2017) and emotion regulation (Kohn et al., 2014, Ochsner and Gross, 2005). The pgACC, in turn, integrates emotional and cognitive information with regulatory demands and signaling the need for control in response to emotionally salient stimuli (Bush et al., 2000, Etkin et al., 2011, Palomero-Gallagher et al., 2019, Shafritz et al., 2006). As the prefrontal cortex and its associated executive functions continue to mature during adolescence, such regulatory mechanisms may be less developed than in adulthood, causing adolescents to struggle more to regulate their affective responses (Pöpplau and Hanganu-Opatz, 2024, Pozzi et al., 2021). The stronger excitatory DLPFC-to-pgACC connectivity observed in adolescents may reflect a compensatory reliance on top-down control from the DLPFC to the affective regions (i.e., pgACC) during social evaluation (Booth et al., 2003, Ordaz et al., 2013, Willinger et al., 2022). This aligns with previous studies suggesting that adolescents require more effort to inhibit responses to emotional stimuli than adults (Cohen-Gilbert and Thomas, 2013, Pozzi et al., 2021), reflecting less effective emotion regulation. Conversely, adolescents showed weaker excitatory pgACC-to-DLPFC connectivity, which indicates a reduced signaling of affective appraisal on cognitive control systems during social evaluation. Research has indicated the importance of pgACC input in guiding executive function based on social-contextual cues (Schriber and Guyer, 2016, Todd et al., 2012). Inadequate signaling from the pgACC may limit adolescents’ abilities to integrate emotional information into cognitive control, which could contribute to challenges in emotion regulation during adolescence (Schriber and Guyer, 2016).

Importantly, the modulatory effects of criticism and praise on this DLPFC-pgACC pathway differed between the two age groups. Adolescents showed weaker excitatory effects of criticism on the DLPFC-to-pgACC connectivity, and weaker inhibitory effects of praise on the pgACC-to-DLPFC connectivity, compared to adults. The weaker excitation of DLPFC-to-pgACC connectivity by criticism suggests that adolescents recruit less top-down regulatory control over affective regions when confronted with negative feedback, consistent with evidence that youth exposed to maternal criticism show reduced recruitment of cognitive control regions including the DLPFC (Lee et al., 2015, van Houtum et al., 2022). In other words, when receiving emotionally salient stimuli (i.e., criticism), adults seem to upregulate prefrontal control toward the pgACC, whereas adolescents do so to a lesser extent. This reduced engagement of prefrontal control may sustain affective appraisal in the pgACC, allowing negative stimuli to continue to drive emotional responding (Blakemore and Mills, 2014, Somerville, 2013). Consistent with studies linking insufficient top-down prefrontal regulation of affective regions to emotional dysregulation (Etkin et al., 2015, Kaiser et al., 2015, Ochsner and Gross, 2005), this age-related difference suggests that adolescents have a reduced capacity to regulate their affective response to criticism within personally meaningful close relationship contexts. Such reduced regulatory control over affective salience of negative input may contribute to the increased vulnerability to affective disorders that emerges in adolescence (Butterfield et al., 2021, Casey et al., 2008, Somerville, 2013).

During praise, adolescents showed weaker inhibitory modulation of the reverse pathway (i.e., pgACC-to-DLPFC) compared to adults. This pattern implies that the pgACC signals a greater need for cognitive control when receiving positive feedback in adolescents. One possible interpretation is that adults downregulate DLPFC engagement by inhibiting pgACC signaling to DLPFC when receiving positive feedback, which aligns with prior study suggesting that positive feedback is less cognitively demanding to manage in adulthood, and therefore requires less active regulatory engagement (Achterberg et al., 2017, van Duijvenvoorde et al., 2010, Vanderhasselt et al., 2015). By contrast, adolescents may experience praise as more emotionally salient due to their heightened need for social acceptance compared with children and adults (Crone and Dahl, 2012, Kerestes et al., 2016). Developmental evidence supporting this notion has often emphasized peer evaluations, but it is also consistent with the evidence that parental acceptance remains salient during adolescence (Morris et al., 2021, Sequeira et al., 2019). Therefore, they may less downregulate DLPFC engagement in response to positive feedback, as such feedback remains highly salient during this developmental period. Prior studies also found increased activation in the ACC and the DLPFC in response to positive compared to negative feedback from peers in adolescents, suggesting the affective salience of positive feedback and greater need for cognitive control during this developmental stage (Achterberg et al., 2017, Achterberg et al., 2020, Crone and Dahl, 2012). Taken together, our findings suggest that adolescents engage affective and regulatory circuitry differently than adults during social evaluation: they appear to recruit less top-down prefrontal control over affective regions when confronted with negative social evaluation, while signaling a greater need for regulatory control during positive social evaluation. These patterns – difficulty dampening negative input and sustained engagement with positive input – may help explain why social evaluation is impactful during adolescence (Blakemore and Mills, 2014, James et al., 2023, Nelson et al., 2005, Somerville, 2013, van Houtum et al., 2022). This may be particularly relevant in salient close relationship contexts (i.e., parents and peers), where evaluative feedback is closely linked to youths’ affective well-being and vulnerability to depression and anxiety (Schacter and Margolin, 2019, Telzer et al., 2018).

Of note, although adolescents reported higher perceived criticism and greater increase in mood disturbance after the social evaluation task, we did not observe any significant correlation between these behavioral indices (i.e., PCM scores, and TMDS delta scores) and the modulatory effects of criticism or praise on DLPFC-pgACC connectivity. This may seem unexpected, because this pathway was modulated by experimental conditions (i.e., being criticized and praised) and differed between the two age groups. One plausible explanation is that perceived criticism is primarily associated with the interaction between cognitive control and self-referential processing when being criticized (but not being praised). This is consistent with our prior study in adolescents (Chen et al., 2024), in which PCM scores were related to the modulatory effects of criticism on the precuneus-to-DLPFC connection, but not to other pathways. In addition, the timing and frequency of our mood state assessments may not have been optimal. Mood states were assessed only before and after the entire social evaluation task, so we likely missed the momentary mood changes that occurred during or immediately after critical and praising comments were presented, which is when the fMRI data were being acquired. This temporal mismatch between behavioral sampling and neural responding can attenuate observable brain-behavior correlations.

Furthermore, age-related differences were also found in the effective connectivity between the DLPFC and the precuneus—regions implicated in cognitive control and social cognition, respectively. The weaker excitatory DLPFC-to-precuneus connectivity that we observed in adolescents (compared to adults) may indicate that the interaction between the brains’ cognitive control and self-referential processing during social evaluation is not yet fully mature in this age group, reflecting the protracted development of long-range cortico-cortical connections that support cognitive-social integration (Hoshi, 2006, Petrides, 2005, Váša et al., 2020). In our prior study of adolescents, we found that the excitatory precuneus-to-DLPFC connectivity would be inhibited when being criticized, which may reflect an emerging capacity to disengage from the self-referential processing in response to negative feedback (Chen et al., 2024). However, in the current study, adolescents showed weaker excitatory connectivity in the opposite direction (i.e., from the DLPFC to the precuneus) during social evaluation in general. These may indicate that although adolescents can limit self-referential influence during criticism, they are less able to exert cognitive control from the DLPFC over self-referential processing regions. As a result, self-referential thoughts triggered by evaluative feedback may remain more persistent in adolescents. The reduced self-inhibition of the precuneus we observed in the adolescent group provides further insights into these persistent self-referential thoughts. Self-inhibition reflects a region’s ability to regulate its own activity, which is essential for efficient information processing and the prevention of overactivation (Almgren et al., 2018, Snyder et al., 2021). In the context of social evaluative feedback, weaker precuneus self-inhibition may bias adolescents toward prolonged engagement of self-referential processing when receiving social evaluation (Rappaport and Barch, 2020, Smith et al., 2020, Zhou et al., 2020). Taken together, weaker excitatory DLPFC-to-precuneus connectivity and reduced precuneus self-inhibition may jointly contribute to adolescents’ difficulties in disengaging from self-referential thoughts following social evaluative feedback. This pattern aligns with evidence that the adolescent brain is particularly attuned to social evaluation, a key aspect of their developmental stage (Blakemore, 2008, Burnett et al., 2011). These interpretations remain speculative given the lack of direct measurement of participants’ self-referential thoughts.

Although our study reveals distinct functional interaction patterns during social evaluation processing between adolescents and adults, certain methodological limitations must be considered. First, participants were exposed to audio clips that were designed to be broadly applicable and likely to be personally relevant for a wide range of participants, rather than individually tailored praising and critical feedback from the participants’ close others (e.g., their own mothers). Although these comments were very similar across both groups, there were some small differences in the neutral comments (see Supplemental Materials). Second, the adolescents were told to imagine that one of their parents was giving them feedback, whereas the adults were told to imagine that it was someone they were close to. This difference in evaluator cueing may influence the direct comparison of neural responses between groups. At the same time, prior work examining neural correlates of both parental and peer influence in adolescents suggests that socially meaningful input from different sources of close relationships can recruit overlapping neural patterns (Welborn et al., 2016). In addition, the emotional salience of such evaluative cues may nevertheless differ across different life stages. Although adolescence is widely considered as a time when peer influence on thoughts, feelings and behavior becomes particularly salient (Guyer and Jarcho, 2018), parents continue to play an important role in adolescents’ socioemotional development during this stage of life (Morris et al., 2021, Steinberg and Morris, 2001). Despite spending more time with their peers than with their parents (Venticinque et al., 2024), adolescents are usually still much more dependent on their parents than (young) adults (Szwedo et al., 2022). As they continue to live under the same roof, they are likely to be exposed to parental feedback on a daily basis. For (young) adults who no longer live with their parents, the number of interactions becomes smaller, and receiving parental feedback may become more exceptional. In both groups, the audio clips were recorded with an adult female voice actor who intended to mimic a maternal voice, but given the experimental setup, the adult participants may well have imagined that someone other than their mother was giving them feedback. We cannot fully disentangle developmental effects from source effects, and future studies could assess subjective salience of the comments across groups, or use recordings by the participants’ real mothers (as in the original experiment by Hooley et al., 2009), fathers or peers, to directly test potential source-specific effects of evaluative feedback. A further limitation is that we did not collect demographic variables such as race/ethnicity and socioeconomic status, which limits our ability to characterize sample representativeness.

Participants’ mood states were only assessed before and after the experiment. Since the two critical comments were always presented at the end, any observed mood state changes are presumed to result (mostly) from exposure to these negative comments. We chose this fixed order to maximize the impact of the critical comments (prior to measuring their impact on the participants’ self-reported mood states) based on the well-established affective contrast theory (Manstead et al., 1983). However, we were not able to disentangle the contributions of the neutral and/or positive comments that the participants heard prior to the negative comments from the overall effect of the experiment. Indeed, this design prevents a direct comparison of mood state changes across the different experimental conditions. Future studies could include additional mood state assessments after positive (and neutral) feedback. Additionally, the use of a counterbalanced design with randomized allocation to a subgroup (e.g., one group with neutral - positive - neutral - negative feedback and another group with neutral - negative - neutral - positive feedback) would be useful to rule out effects related to the order of the comments.

Although some frameworks extend adolescence into the (early) twenties (Sawyer et al., 2018), we defined the adolescent group using the legal age at which adulthood starts (i.e., one is legally considered an adult at 18 years of age). The (young) adult group was defined in line with a prior large-scale lifespan brain development study that characterizes (young) adulthood from 20 to 40 years (Bethlehem et al., 2022). We note that important social transitions continue between ages 17 and 20 (e.g., in Belgium, most people go to secondary school until they are 18 years old). Additionally, the adult group covered a relatively wide age range. Future research could benefit from using more specific subgroups—such as older adolescents/younger adults (e.g., 18–25 years) and (older) adults (e.g., 25–40 years)—to better examine developmental trajectories. Furthermore, we compared two distinct groups to get an idea of how neural and behavioral responses to social feedback change over time. However, a longitudinal design with repeated assessments in different life stages would have been even more suited to study these developmental trajectories.

5. Conclusions

The present study offers valuable insights into the age-related differences in neural dynamics of social evaluation processing. By comparing healthy adolescent and adult groups, we demonstrate that adolescents exhibit heightened sensitivity to criticism, as reflected in greater mood disturbances, higher perceived criticism, and distinct dynamic interactions among brain regions involved in affective, cognitive, and self-referential processing. Specifically, DCM revealed that adolescents show less efficient bidirectional connectivity and reduced modulatory effects among the pgACC, DLPFC, and precuneus—highlighting the ongoing maturation of neural circuits responsible for integrating cognitive, emotional, and social information. These findings underscore adolescence as a period of neurobiological vulnerability, during which the capacity for emotion regulation in socially evaluative contexts is still developing. The differential connectivity patterns observed in adolescents suggest that their heightened sensitivity to social evaluation may stem from immature regulatory networks in response to negative feedback and heightened affective signaling in response to positive feedback. This work contributes to a growing body of literature elucidating the neural basis of adolescent development and may inform the development of early interventions during this formative stage.

CRediT authorship contribution statement

Guo-Rong Wu: Writing – review & editing, Supervision, Conceptualization. Rudi De Raedt: Writing – review & editing, Supervision, Conceptualization. Qinyuan Chen: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Formal analysis, Data curation, Conceptualization. Sam Luc Bart Bonduelle: Writing – review & editing, Investigation, Conceptualization. Chris Baeken: Writing – review & editing, Supervision, Resources, Conceptualization.

Compliance with ethical standards

The study protocol involving human participants was approved by UZ Gent Medical Ethics Committee. Written informed consent was obtained from all participants and parents/legal guardians of adolescents before the study.

Funding sources

Q.C. was supported by the Chinese Scholarship Council. S.B. was supported by the Research Foundation-Flanders (FWO)/Red Nose (Rode Neuzen) fund (G.0F46.17N). This work was supported by the National Natural Science Foundation of China awarded to G.W. (Grant No. 62271415, 61876156). This work was funded by a Concerted Research Action grant from Ghent University Special Research Fund awarded to R.D.R. & C.B. (Grant number BOF 16/GOA/017); R.D.R. is supported by the Research Foundations FWO and F.R.S.-FNRS under the Excellence of Science (EOS) program (EOS 40007528. - G0I2422N). This work was also supported by the Queen Elisabeth Medical Foundation for Neurosciences and by an Applied Biomedical (TBM) grant of the Agency for Innovation through Science and Technology (IWT), part of the Research Foundation - Flanders (FWO) PrevenD Project 2.0 (T000720N), the BRAINN HORIZON-WIDERA-2021-ACCESS-03 – Twinning project (Grant No. 101079001), BeNeFIT No 10390082210018, and SOA FRA 2024 0023.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

Supplementary material

Data Availability

I have shared the link to my data and code at the Attached File step. The demographic characteristics, behavioral data, thresholded and unthresholded whole-brain maps for between group comparisons in each contrasts, individual-level DCMs, and codes used in this study are available online on Open Science Framework at https://osf.io/fr8mn/. The raw neuroimaging data used in this study are available upon request from the corresponding authors.