Associations between perceived parenting, brain activity and connectivity, and depression symptoms in Brazilian adolescents

ABSTRACT

In adolescence, parental care is associated with lower depression symptoms whereas parental overprotection is associated with greater depression symptoms, effects which may be mediated by adolescent brain activity and connectivity. The present study examined associations between perceived parenting, brain activity and connectivity, and depression symptoms in adolescents from Brazil, a middle-income country (MIC). Analyses included 100 adolescents who underwent functional magnetic resonance imaging scanning while completing a face matching task. Parental care and overprotection were associated with adolescent depression symptoms in expected directions. We also found that parental care and overprotection were associated with amygdala connectivity with several brain regions; however, amygdala activity was not associated with parenting and neither activity or connectivity mediated the association between parenting and depression symptoms. Results identify how parenting influences brain function and depression symptoms in youth from a MIC.

KEYWORDS:

• Adolescence
• parenting
• depression
• amygdala
• functional MRI (fMRI)

Introduction

Though adolescence marks a stage of increased independence from parents, parents continue to influence adolescent emotional reactivity and regulation (Kerr et al., 2019; Tan et al., 2020) and risk for depression (Pinquart, 2017; Schwartz et al., 2017). One pathway through which this may occur is through the influence of parenting on adolescent brain function (Kerr et al., 2019; McLaughlin et al., 2015; Romund et al., 2016; Tan et al., 2020). However, most research has been conducted in countries classified by the World Bank as high-income countries (HICs) (i.e. with gross national income per capita of $13,846 or more (The World Bank, 2022)). To date, it is unclear if these findings will generalize to adolescents in low- or middle-income countries (LMICs) (classified as having a gross national income between $1,136 and $4,465 (The World Bank, 2022)).

Although the majority of youth (9 out of 10) are from LMICs, most health-related research is conducted in HICs (Kieling et al., 2011). Even further, depression is a leading cause of disability in adolescents globally (GBD 2019 Mental Disorders Collaborators, 2022) and this is especially true for youth between the ages of 15–19 for whom depressive disorders are the second leading cause of disability (Kieling et al., 2024). Moreover, 10–20% of youth from LMICs deal with mental health problems (Kieling et al., 2011) and most neuroimaging studies examining associations between brain activity and connectivity and parenting (Tan et al., 2020) or, brain activity and connectivity and depression (Battel et al., 2021), have been conducted in HIC samples.

With this regard, it is unclear how current neuroimaging findings generalize to adolescents in LMICs. On the one hand, as we review below, studies have found that parenting predictors of depression are often similar across different countries and settings. Thus, it is possible that associations between parenting and brain function would generalize from high-income countries to other settings. On the other hand, however, differences in cultural expectations and family processes regarding parenting in different settings (Lansford, 2022) and increased economic pressures on parents and adolescents in LMIC settings may lead to different associations between parenting, brain activity and connectivity, and depression symptoms in LMICs than have been observed in HICs.

Because the majority of adolescents in the world live in LMICs, it is important to examine whether findings obtained in HIC samples generalize to youth in LMICs. This could provide evidence to support novel strategies for intervention and prevention in LMIC settings, and help inform global mental health policy work on a broader scale. In the present study, we aim to address this gap in the existing literature by examining associations between parenting, brain activity and connectivity, and depression symptoms in Brazilian youth. In doing so, we are able to extend the generalizability of this research outside of high-income settings.

Parenting and depression

Across multiple countries, parental care and warmth (i.e. emotional support, affection and acceptance) are generally associated with positive developmental outcomes in adolescents, whereas parental overprotection (i.e. not encouraging independence or controlling what the child does) and overcontrol are associated with worse outcomes (Bahreini et al., 2011; Garcia et al., 2019; Khalid et al., 2018; Pinquart, 2017). Garcia et al. (2019), for example, examined adolescents from Brazil, Spain, Germany, and the United States (US) and found that high parental warmth and low parental strictness were predictive of greater personal well-being (Garcia et al., 2019). A separate study of Chinese adolescents found that, when controlling for maternal care, higher paternal care was associated with less depression (Wang et al., 2020). Another study of low-income youth from the US found that increased parent–child closeness at age nine was associated with lower depression symptoms at age 15, whereas harsh parenting and neglect were associated with higher depression symptoms (Fagan, 2022). Then, in a separate study of Iranian youth, researchers found that depression was most common for adolescents whose parents exhibited low parental care (Bahreini et al., 2011; Khalid et al., 2018) and high excessive control (Bahreini et al., 2011). Findings suggest that parental care/warmth mitigates the development of depression and parental overprotection/control may increase risk, as this has been observed across several different countries and cultural contexts.

An open question in this research is whether maternal and paternal parenting have similar effects on adolescent development. Previous work suggests that maternal warmth and support is associated with lower internalizing symptoms, increased emotion regulation, and more positive adjustment in adolescence (Manuele et al., 2023). However, in contrast to the number of studies examining the influence of maternal parenting in adolescence, especially within the context of brain development, relatively less work has examined the influence of paternal parenting. Current work suggests that paternal figures influence adolescent development in ways similar to maternal parenting – such that warm-sensitive parenting from fathers bolsters positive adjustment in adolescents whereas overcontrol has the opposite effect (Manuele et al., 2023). This suggests that paternal parenting may have a similar effect as maternal parenting on adolescent development – with maladaptive parenting styles (i.e. overcontrol) appearing to be strong contributors to adolescent emotional reactivity (Kerr et al., 2019; Tan et al., 2020) and risk for depression (Pinquart, 2017; Schwartz et al., 2017). By including an examination of both maternal and paternal figures in our research we can start to develop a more thorough understanding of the role each parent plays on brain development and psychopathology in adolescence.

Parenting and brain activity and connectivity

Evidence also suggests that parenting is associated with the function of adolescent neural circuitry involved in emotional reactivity and control including the amygdala, responsible for threat detection and evaluation of emotional content (Rasia-Filho et al., 2000; Ressler, 2010), and the medial prefrontal cortex (MPFC), which plays a role in higher order emotional functioning and emotion regulation through its connectivity with the amygdala (Etkin et al., 2011; Hariri et al., 2003).

Youth who experience childhood maltreatment or other forms of negative caregiving such as overprotection display heightened amygdala reactivity to negative facial expressions (angry, sad, and fearful) in adulthood (Dannlowski et al., 2013; Farber et al., 2019). On the other hand, adolescents (aged 13 to 16 years) whose mothers were high on maternal warmth and support exhibited lower left amygdala activation to fearful faces during an emotional face-matching task (Romund et al., 2016). Similarly, another study found that adolescents whose parents demonstrated high levels of warmth also exhibited lower activation in the left amygdala, bilateral insula, anterior cingulate cortex (ACC), subgenual ACC, and right ventrolateral prefrontal cortex (PFC) when responding to criticism versus neutral statements (Butterfield et al., 2021). Together, findings suggest that positive parenting reduces neural reactivity to negative stimuli whereas negative parenting enhances it.

In addition to parenting effects on brain activity, research shows that parenting is associated with brain connectivity, though findings appear to be mixed. Maternal hostility, harshness, or overprotection has been associated with decreased amygdala-MPFC connectivity to sad faces (Kopala‐Sibley et al., 2020), decreased structural connectivity in the uncinate fasciculus (a white matter tract that connects the amygdala and MPFC) (Farber et al., 2019), and increased left amygdala-left ventrolateral PFC connectivity (Jiang et al., 2021). Maternal support, on the other hand, has been associated with less positive amygdala-ventromedial PFC (vMPFC) connectivity to angry faces (Chen et al., 2020). Results suggest that parenting influences the neural circuitry underlying emotion reactivity and regulation, but the direction of these associations remains to be clarified.

Brain activity and connectivity and depression

Patterns of brain activity and connectivity described above have also been associated with depression in other studies, and may help explain the correlation between parenting and depression symptoms. For instance, adolescents at risk for or with current depression exhibit enhanced or heightened amygdala activity when viewing fearful faces (Swartz et al., 2015; Yang et al., 2010). Adolescents with depression symptoms also exhibit blunted PFC activity (Kaya & McCabe, 2019) and reduced connectivity between the amygdala and PFC, which has been associated with greater depression symptoms (Connolly et al., 2017; Yoon et al., 2021) and clinical depression (Groenewold et al., 2013).

Study aims

The goal of the present study is to extend our understanding of associations between parenting, emotion-related neural function, and depression symptoms to adolescents outside of HICs. To do so, this study examines the association between perceived parenting, brain activity and connectivity, and depression symptoms in adolescents from Brazil, a MIC. Guided by previous research, we developed four hypotheses. First, we expected adolescent-perceived parental care to be negatively associated with depression symptoms and parental overprotection to be positively associated with depression symptoms (Pinquart, 2017). Second, we expected higher maternal care to be associated with reduced amygdala activity to negative faces (fearful, angry, sad; Romund et al., 2016) and maternal overprotection to be associated with increased amygdala activity (Farber et al., 2019), but did not have a priori hypotheses for paternal care or overprotection and amygdala activity due to limited research that currently exists. Third, given that maternal care is associated with decreased amygdala activity (Romund et al., 2016) and enhances more mature amygdala-vMPFC connectivity (Chen et al., 2020) in our study, we expected maternal care to predict similar patterns of increased neural connectivity between the amygdala and PFC. We also expected that maternal overprotection would be associated with decreased amygdala-PFC connectivity to negative faces based on prior studies finding these effects (Farber et al., 2019; Kopala‐Sibley et al., 2020) but did not have a priori hypotheses on neural connectivity for paternal care or overprotection due to the lack of prior research on fathers. Lastly, we expected these patterns of brain activity and connectivity to mediate the association between parenting and depression symptoms (Figure 1).

Figure 1. Planned mediation model.

Mediation model that will be used to test whether brain activity and connectivity mediates association between perceived parenting and depression symptoms.

We also conducted post hoc exploratory analyses to examine associations between perceived parenting and whole-brain activity and amygdala-whole-brain connectivity with regions outside of the PFC. Given the limited research that currently exists on fathers, analyses examining amygdala activity and amygdala-PFC connectivity for paternal parenting (care and overprotection) should also be considered exploratory. Exploratory study aims were motivated by previous research examining parenting in adolescents, which has found significant neural activity and connectivity in other brain regions associated with emotion processing and regulation (Butterfield et al., 2021; Jiang et al., 2021; Kopala‐Sibley et al., 2020; Pozzi et al., 2020).

Methods

Participants

Hypotheses were tested with the Identifying Depression Early in Adolescence Risk Stratified Cohort (IDEA-RiSCo) dataset. As described in our previous work (Kieling et al., 2021; Yoon et al., 2021, 2022), 7,720 adolescents from Porto Alegre, Brazil were screened for risk for depression and separated into three groups—(1) high-risk and clinically depressed (DEP), (2) high-risk but had no prior or current diagnosis of depressive disorders (HR), and (3) low-risk and had no prior or current diagnosis of depressive disorders (LR). The risk groups were stratified using a multivariable prognostic model, the Identifying Depression Early in Adolescence Risk Score (IDEA-RS; Rocha et al., 2021). This model integrated 11 socio-demographic variables which included biological sex, skin color, drug use, school failure, social isolation, fight involvement, poor relationship with father, mother, and between parents, childhood maltreatment, and ran away from home (full details regarding the procedures can also be found in the published protocol for the IDEA-RiSCo study; Kieling et al., 2021). A total of 150 adolescents (ages 14 to 17 years, 50% female, 44% African descent) completed fMRI scanning from August 2018 to December 2019. Of these 150 participants, 50 were DEP, 50 were HR, and 50 were LR. Written informed consent/assent was obtained from adolescents and their caregivers after the procedures had been fully explained.

Participant exclusion criteria

Participants who could only report on one parental figure were excluded from the sample for analyses, resulting in a loss of 24 participants—22 adolescents did not report on a paternal parent figure and 2 did not report on a maternal parent figure. For fMRI quality control, 26 participants were excluded—10 for excessive head movement, 12 for low behavioural accuracy, and 4 for missing face task data. After exclusion for missing questionnaire data (n = 24) and fMRI data quality control (n = 26), the sample size for analyses was 100 (47% female, Age M = 15.61, SD = .82).

Within this sample, 33 were DEP (45.9% females, Age M = 15.34, SD = .82), 30 were HR (50% female, Age M = 15.93, SD = .78), and 37 were LR (45.5% females, Age M = 15.62, SD = .78). Participant characteristics are reported in Table 1 and bivariate correlations are reported in Table 2. Further details regarding depression diagnosis, fMRI inclusion/exclusion criteria, and the study protocol have been reported in prior research (Kieling et al., 2021; Yoon et al., 2021). We have previously published results in this sample examining differences in the face matching task between the three risk status groups and examining associations between neural function and self-reported depression symptoms (Yoon et al., 2021), as well as examining the association between resting state functional connectivity and self-reported depression symptoms (Yoon et al., 2022).

Table 1. Participant characteristics.

Total Sample (n = 100)	Mean	SD	Min	Max
Adolescent	15.61	.82	14	17
Age
Maternal	26.84	8.05	5	36
Care
Maternal	16.35	7.64	4	34
Overprotection
Paternal	22.10	10.16	1	36
Care
Paternal Overprotection	14.25	8.35	0	39
CDRS-R	30.76	15.96	17	72

Notes: SD =Standard deviation; Min=Minimum; Max=Maximum.

Table 2. Bivariate correlations.

	Age	Sex	Maternal Care	Maternal Over-protection	Paternal Care	Paternal Over-protection	Depression Symptoms	Left Amyg Fear	Right Amyg Fear	Left Amyg Angry	Right Amyg Angry	Left Amyg Sad	Right Amyg Sad
Age	1
Sex	−.144	1
Maternal Care	−.123	−.044	1
Maternal Overprotection	.041	−.049	−.573**	1
Paternal Care	−.236*	.-142	.450**	−.164	1
Paternal Overprotection	.123	.189	−.324**	.460**	−.390**	1
Depression Symptoms	−.027	.022	−.540**	.457**	−.532**	.448**	1
Left Amyg Fearful	.150	−.138	.116	−.026	.017	−.002	.018	1
Right Amyg Fearful	.137	−.210*	.085	.061	.002	.002	.075	.783**	1
Left Amyg Angry	−.008	−.218*	−.128	.253*	.109	−.010	.039	.054	.194	1
Right Amyg Angry	−.058	−.141	−.086	.110	.056	−.102	.048	.070	.154	.738**	1
Left Amyg	.049	−.150	.011	.110	.056	.125	−.085	.067	.161	.370**	.022	1
Sad
Right Amyg	.013	−.153	.036	.081	.103	.069	−.047	.211*	.327**	.352**	.116	.817**	1
Sad

Notes: *Correlation is significant at the 0.05 level (2-tailed). **Correlation is significant at the 0.01 level (2-tailed). Note. Sex was coded as Male = 0, Female = 1.

Perceived parenting

Adolescents completed the validated Brazilian Portuguese translation of the Parental Bonding Instrument questionnaire (Hauck et al., 2006) to assess dimensions of perceived parenting such as parental care and overprotection and reported on both mothers and fathers (or an equivalent maternal/paternal figure). The Parental Bonding Instrument (PBI) assessed the likelihood of a particular event occurring in the household (e.g. understood their problems/worries, tried to control everything they did), on a scale ranging from ‘Very unlikely’ (1) to ‘Very likely’ (4). The parental care subscale included 12 items assessing level of parental warmth and support within their children’s lives, from enjoying talking with their child to offering emotional support. This measure demonstrated good reliability in this sample, ɑ=.89. The parental overprotection subscale included 13 items assessing the use of control and overprotection such as not encouraging independence or controlling what the child does. This measure demonstrated good reliability in this sample, ɑ=.83.

Depression symptoms

Depression symptoms were obtained with the Children’s Depression Rating Scale – Revised questionnaire (CDRS-R; Poznanski, 1985) which is a well-validated measure of global depression in adolescents (Mayes et al., 2010; Plener et al., 2012). In this measure, depression symptoms are rated by a clinician during a 15–20-min interview with the adolescent to assess depression diagnosis and severity of depression symptomology. More specifically, the CDRS-R questionnaire includes 17 items that assess psychological adjustments as it relates to social functioning (i.e. difficulty having fun, social withdrawal, impaired schoolwork), negative thoughts and feelings (i.e. suicidality, self-esteem, guilt, irritability), as well as somatic symptoms (i.e. physical complaints, fatigue, sleep disturbances). Each item ranges from 1 to 5 or 1 to 7 with a possible total score that ranges from 17 to 113. Scores ≥ 40 indicate a presence of depression symptomology, and scores ≤ 28 indicate minimal to no depression symptomology. The CDRS-R demonstrated good reliability in this sample, ɑ=.90. We used a measure of clinician-rated depression symptoms in these analyses, rather than adolescent-reported depression symptoms, to avoid shared reporter variance, as adolescents reported on perceived parenting.

Emotional face-matching task

Neural function was measured using an emotional face-matching task (Hariri et al., 2002) that contained two face matching blocks each for 5 facial expressions – fearful, angry, happy, sad, and neutral – and control blocks consisting of matching geometric shapes. In a face matching trial, participants viewed a trio of facial expressions that included a target facial expression, and participants were instructed to select the facial expression that matched the target facial expression. Whereas in the control trial, participants were instructed to match geometric shapes instead of facial expressions. Control blocks were interleaved between the face matching blocks. Face matching and control blocks had 6 trials within each block presented. In each trial, facial expressions or geometric shapes were presented for 2000 ms, which was followed by a fixation cross intertrial interval of 2000 ms. Further details regarding the face matching task can be found in Yoon et al. (2021).

fMRI data acquisition, preprocessing, quality control criteria, and first-level analysis

All images were acquired on a 3T Ingenia scanner (Koninklijke Philips N.V., The Netherlands), software version 5.3.1. Full details on acquisition parameters have been reported in published research (Battel et al., 2020; Kieling et al., 2021; Yoon et al., 2021). A standard preprocessing pipeline was conducted in SPM12 (see Yoon et al., 2021 for full details) and quality control criteria required the exclusion of participants with ≥ 10% volumes censored for excessive head movement (i.e. mean signal exceeds 4 standard deviations of the global mean or the scan-to-scan movement exceeds 2 mm translation or 2° rotation in any direction), ≤90% signal coverage within the amygdala ROI, and ≤ 75% behavioural accuracy during the face matching task. For first-level analysis, face and shape blocks were modelled with boxcar regressors and regressors for head movement-outlier volumes detected from the Artifact detection toolbox (http://www.nitrc.org/projects/artifact_detect) were included in the model as nuisance covariates. We generated contrast images for fearful, angry, and sad facial expressions versus shapes to examine emotion-related neural activity to negative faces for each individual. These contrast images were then entered into second-level random effects models for group analyses.

Amygdala activity

Based on previous research that has found effects for amygdala activity and parenting (McLaughlin et al., 2015; Romund et al., 2016), we used an anatomical region-of-interest (ROI) approach to test our hypotheses that amygdala activity to fearful, angry, and sad faces would be associated with parental care and overprotection. ROIs were defined as left and right amygdala from the Wake Forest University (WFU) Pickatlas using the Automated Anatomical Labeling Atlas (Maldjian et al., 2003). Left and right amygdala activity was extracted from the anatomically-defined ROIs for each contrast of interest and submitted for further statistical analyses.

Amygdala-PFC connectivity

To examine amygdala connectivity, subject-level functional data were modelled with a generalized linear model (GLM) using a generalized psychophysiological interaction (gPPI) toolbox (Mclaren et al., 2012). We conducted gPPI analyses using the anatomical amygdala as our seed ROI. Time series were extracted from the left and right amygdala seed regions which served as the physiological variable in our analysis. To create the psychophysiological variable (the interaction terms of the physiological and psychological variables) for each participant, BOLD signal time courses from the conditions of interest (fearful > shapes, angry > shapes, and sad > shapes) were extracted from the amygdala, and we created an interaction term with the contrasts of interest (the psychological variable), resulting in PPI connectivity maps. The psychophysiological variable was then used to identify regions that covary with either the left or right amygdala seed region in a task-dependent manner. The contrast images for the PPI of fearful vs. shapes, angry vs. shapes, and sad vs. shapes were each submitted to a second-level group analysis.

We then created ROIs for the PFC, given our hypotheses that parenting would be associated with amygdala-PFC connectivity (Chen et al., 2020; Kopala‐Sibley et al., 2020). Two ROIs were created using MarsBaR in SPM based on prior work. One ROI in the MPFC (−24, 54, 30) was created based on peak voxel coordinates from Kopala‐Sibley et al. (2020) and a second ROI in the vMPFC (−2, 42, −10) was created based on peak voxel coordinates from Chen et al. (2020). Using these ROIs, left and right amygdala-PFC connectivity was extracted for each contrast of interest and submitted for further statistical analyses.

Statistical analyses

All analyses were conducted with SPM12 and SPSS26 (IBM) software, and we used multiple regression to test associations between parenting and depression symptoms, associations between parenting and neural activity and connectivity, and associations between neural activity and connectivity and depression symptoms. We included covariates for adolescent age and adolescent sex in our analyses to control for any covariance with both the PBI and CDRS-R measures and brain activity. In addition, given prior work suggesting that children and adolescents from lower socioeconomic backgrounds tend to exhibit altered brain activity within emotion regulatory brain regions (Javanbakht et al., 2015), we followed up any significant results with secondary analyses to control for socioeconomic status (SES), in addition to age and sex. The Associação Brasileira de Empresas de Pesquisa socioeconomic stratification index was used to measure SES (Kamakura & Mazzon, 2016). SES was examined as a covariate in secondary analyses because two participants were missing the SES measure, and we aimed to maximize sample size in the main analyses. When only including participants with parenting data available for both maternal and paternal figures, 24 participants were excluded from analyses. Therefore, to increase sample size, we also conducted sensitivity analyses to examine any participants who had available data for maternal parenting (n = 122) and any participants who had available data for paternal parenting (n = 102) separately.

First, to test for associations between parenting and depression symptoms, we conducted four separate regressions in SPSS: (1) maternal care predicting depression symptoms, (2) maternal overprotection predicting depression symptoms, (3) paternal care predicting depression symptoms, and (4) paternal overprotection predicting depression symptoms, adjusting for the effect of age and sex.

Second, to test for associations between parenting and amygdala activity, we submitted extracted contrast values for amygdala activity into SPSS and conducted separate regressions for maternal care predicting left and right amygdala activity, maternal overprotection predicting left and right amygdala activity, paternal care predicting left and right amygdala activity, and paternal overprotection predicting left and right amygdala activity for each face condition (fearful > shapes, angry > shapes, and sad > shapes), adjusting for the effect of age and sex.

Third, we examined the association between perceived parenting and amygdala-PFC connectivity to negative faces. To test for associations between parenting and amygdala-PFC connectivity, we submitted extracted contrast values for amygdala-PFC connectivity into SPSS. We conducted separate multiple regressions in SPSS to test for associations between maternal care predicting left and right amygdala-PFC connectivity, maternal overprotection predicting left and right amygdala-PFC connectivity, paternal care predicting left and right amygdala-PFC connectivity, and paternal overprotection predicting left and right amygdala-PFC connectivity for the contrasts of fearful > shapes, angry > shapes, and sad > shapes, adjusting for age and sex. We tested for associations between parenting and amygdala-PFC connectivity using the MPFC and vMPFC ROIs described in the methods section above.

Fourth, we built upon our previous analyses to test for associations between brain activity and connectivity and depression symptoms, in order to inform the mediation models. After running multiple regressions on associations between parenting and amygdala activity in SPSS, any significant associations were submitted for further statistical analyses to test associations between amygdala activity and depression symptoms, including age, sex, and parenting as covariates. Likewise, after running multiple regressions in SPSS to examine whether any patterns of amygdala-PFC connectivity were associated with parenting, any significant associations were submitted for additional analyses to test associations between amygdala-PFC connectivity and depression symptoms, controlling for age, sex, and parenting. If any patterns of brain activity or connectivity were associated with depression symptoms while controlling for parenting, then mediation analyses were conducted.

Lastly, we conducted exploratory analyses to test associations between parenting, whole-brain activity and amygdala connectivity with the whole brain, and depression symptoms (using the same methods as described above). A further description of exploratory analyses can be found in the supplemental material under Appendix S1 and S2.

Correction for multiple comparisons

We had a priori hypotheses regarding the associations between maternal/paternal care and overprotection and depression symptoms, between maternal care and overprotection and amygdala activity, and between maternal care and overprotection and amygdala-PFC connectivity. We conducted 4 regressions to test for associations between maternal/paternal care and overprotection and depression symptoms, and therefore we set a Bonferroni-corrected threshold of p < .0125 for these analyses. We conducted 12 regressions to test for associations between maternal care and overprotection and amygdala activity (2 parenting variables x 3 emotional contrasts x 2 hemispheres), therefore we set a Bonferroni-corrected threshold of p < .004 for these analyses. We conducted 24 regressions to test for associations between maternal care and overprotection and amygdala-PFC connectivity (2 parenting variables x 3 emotional contrasts x 2 hemispheres x 2 PFC ROIs), therefore we set a Bonferroni-corrected threshold of p < .002 for these analyses. Analyses examining associations between paternal care and overprotection and brain activity and connectivity were exploratory, as were all whole-brain exploratory analyses, so we did not correct for multiple comparisons for these exploratory analyses.

Results

Associations between parenting and depression symptoms

In support of hypotheses, we found significant associations between adolescent-perceived parenting and clinician-rated depression symptoms. Maternal care was negatively associated with depression symptoms ($\mathit{\beta }$ = −.56, R² = .30, p < .001) (Figure 2) and maternal overprotection was positively associated with depression symptoms ($\mathit{\beta }$ = .46, R² = .21, p < .001) (Figure 2). Paternal care was negatively associated with depression symptoms ($\mathit{\beta }$ = −.59, R² = .32, p < .001) (Figure 2) and paternal overprotection was positively associated with depression symptoms ($\mathit{\beta }$ = .48, R² = .21, p < .001) (Figure 2). Results remained significant when controlling for SES and when examining all participants who had available data for maternal or paternal parenting separately (Table S2).

Figure 2. Associations between adolescent-reported parental care, parental overprotection, and clinician-reported depression symptoms.

Regression plots demonstrate associations between parental care, parental overprotection, and depression symptoms. (a) Association between maternal care and depression symptoms. (b) Association between maternal overprotection and depression symptoms. (c) Association between paternal care and depression symptoms. (d) Association between paternal overprotection and depression symptoms.

Associations between parenting and amygdala activity

In support of hypotheses, we found a positive association between maternal overprotection and left amygdala activity during angry faces vs. shapes ($\mathit{\beta }$ = .24, R² = .11, p = .013; Figure 3), although this did not survive correction for multiple comparisons and was not significant when we included participants who had been excluded for missing paternal parenting data (p > .05). Contrary to expectations, there were no other significant associations between the parenting variables and amygdala activity (all p’s > .05).

Figure 3. Association between adolescent-reported maternal overprotection and left amygdala activity to angry faces.

ROI analysis revealed a positive association with maternal overprotection such that increased activity of the left amygdala to angry faces is associated with higher maternal overprotection. The regression plot describes residualized values of this association between maternal overprotection and left amygdala activity.

Associations between parenting and gPPI amygdala-PFC connectivity

Counter to hypotheses, we did not observe any significant associations between maternal parenting and amygdala-PFC connectivity. However, in the exploratory analyses with paternal parenting there was a positive association between adolescent-reported paternal care and left amygdala-vMPFC connectivity during angry vs. shapes ($\mathit{\beta }$ = .21, R² = .21, p = .043; Figure 4) and these results remained marginally significant when including participants who had been excluded for missing maternal parenting data ($\mathit{\beta }$ = .21, R² = .210, p = .050). We also found a negative association between adolescent-reported paternal overprotection and left amygdala-vMPFC connectivity ($\mathit{\beta }$ = −.24, R² = .24, p = .021; Figure 5) and amygdala-MPFC connectivity ($\mathit{\beta }$ = −.23, R² = .28, p = .025; Figure 6) during angry vs. shapes. Results remained significant when examining any participants with available data for paternal overprotection (amygdala-vMPFC connectivity results: ($\mathit{\beta }$ = −.23, R² = .23, p = .026); amygdala-MPFC connectivity results: ($\mathit{\beta }$ = −.26, R² = .27, p = .012)) and when controlling for SES; however, this was not the case for our results examining paternal care. No other associations tested for amygdala-PFC connectivity were significant (all p’s > .05).

Figure 4. Association between adolescent-reported paternal care and gPPI left amygdala-vMPFC connectivity to angry faces.

There was a positive association with paternal care such that reduced connectivity between the left amygdala and vMPFC to angry faces is associated with lower paternal care. The regression plot describes residualized values of this association between paternal care and left amygdala-vMPFC connectivity.

Figure 5. Association between adolescent-reported paternal overprotection and left amygdala-vMPFC connectivity to angry faces.

There was a negative association with paternal overprotection such that reduced connectivity between the left amygdala and vMPFC to angry faces is associated with higher paternal overprotection. The regression plot describes residualized values of this association between paternal overprotection and left amygdala-vMPFC connectivity.

Figure 6. Association between adolescent-reported paternal overprotection and left amygdala-MPFC connectivity to angry faces.

There was a negative association with paternal overprotection such that reduced connectivity between the left amygdala and MPFC to angry faces is associated with higher paternal overprotection. The regression plot describes residualized values of this association between paternal overprotection and left amygdala-MPFC connectivity.

Associations between parenting, amygdala activity/gPPI amygdala-PFC connectivity, and depression symptoms

When further examining neural function that had been associated with parenting, left amygdala activity and connectivity to angry faces was not significantly associated with adolescent depression symptoms while controlling for parenting (p > .05), and so we did not run mediation analyses.

Exploratory analysis: associations between parenting and whole-brain activity

Although we did not have specific a priori predictions about whole-brain activity or amygdala whole-brain connectivity analyses, we found a significant positive association between maternal care and right cerebellum exterior activity (cerebellar vermal lobules, right cerebellum white matter) to fearful faces and these results remained significant when we included participants who had data available for maternal parenting, but not paternal parenting. Results of the association are reported in Appendix S3, Figure S1.

Exploratory analysis: associations between parenting and gPPI amygdala-whole brain connectivity

Exploratory analyses revealed there were associations between parenting and amygdala connectivity (Appendix S3, Table S3). Maternal care was positively associated with left amygdala connectivity to occipital regions during fearful faces vs. shapes (Table S3, Figure S2); however, these results did not remain significant when controlling for SES nor when we examined all participants with data available for maternal parenting. Similarly, paternal care was positively associated with left amygdala connectivity to occipital and parietal regions, as well as frontal regions during fearful faces vs. shapes (Table S3, Figure S2) and these results remained significant when we included all participants with data available for paternal parenting (Table S3) and when controlling for SES. For paternal overprotection, we also found a negative association for left amygdala-right posterior cingulate gyrus (PCG) connectivity during angry vs. shapes (Table S3, Figure S2) and these results remained significant when we examined all participants with data available for paternal parenting (Table S3) and when controlling for SES. There was also a positive association for right amygdala-cerebellum connectivity during sad vs. shapes for paternal overprotection (Table S3, Figure S2) and this result remained significant when we examined all participants with data available for paternal parenting (Table S3) but this result was not significant when controlling for SES.

Exploratory analysis: associations between parenting, whole brain activity, gPPI amygdala-whole brain connectivity, and depression symptoms

Any patterns of brain activity or connectivity significantly associated with parenting were extracted in order to examine whether they were also associated with clinician-rated depression symptoms, while controlling for parenting. However, no patterns of activity or connectivity predicted depression symptoms while controlling for parenting (p > .05), and therefore we did not run mediation analyses.

Discussion

The aim of the present study was to examine associations between perceived parental care and overprotection, emotion-related brain function (activity and connectivity), and clinician-rated depression symptoms in a sample of Brazilian adolescents. In support of our hypotheses, we found that perceived parental care was negatively associated with depression symptoms and perceived parental overprotection was positively associated with depression symptoms. These findings are in line with previous research (Fagan, 2022; Pinquart, 2017) and suggest that aspects of both positive and negative parenting continue to play an important role in adolescents’ lives, despite adolescence being defined as a period of increased autonomy from primary caregivers and the family. Another novel component to these findings is that adolescents reported on perceived parenting for both maternal and paternal figures, and we found that both mothers and fathers (when information was available for both figures) are important contributors in Brazilian adolescents’ mental health and well-being. Due to culture-related factors surrounding collectivism, in which the self is integrated as part of the family, but relationships between family members tend to be equal (Martinez et al., 2020), it is possible that Brazilian adolescents form strong emotional connections with both maternal and paternal figures equally (Novianti et al., 2023) thus, perceived paternal parenting may have a similar influence on adolescent development as maternal parenting.

Our hypotheses examining the association between perceived parenting and brain activity and connectivity were only partially supported. For neural activity, we found that for angry faces vs. shapes, there was a positive association between adolescent-reported maternal overprotection and left amygdala activation, although it should be noted this effect did not survive correction for multiple comparisons and was not significant when including all participants with maternal parenting data. Then in exploratory analyses when examining paternal care and amygdala-PFC neural connectivity, we found a positive association between paternal care and left amygdala-vMPFC connectivity during angry vs. shapes, and a negative association between paternal overprotection and left amygdala-vMPFC and left amygdala-MPFC connectivity during angry vs. shapes. Findings are in line with previous research examining parental overprotection (Kopala‐Sibley et al., 2020) but provide novel insight into how paternal care and overprotection may influence connectivity of this emotion regulation circuit. Because previous research has focused primarily on mothers, and findings examining parenting and neural connectivity are mixed, more research is necessary to make stronger inferences. Contrary to hypotheses, amygdala-PFC and amygdala-whole brain connectivity did not mediate the association between parenting and depression symptoms. This is perhaps not surprising, as clusters associated with parenting were extracted, and parenting and depression symptoms were strongly correlated, and so controlling for parenting while examining associations with depression symptoms could have removed a large amount of overlapping variance. In addition, it is important to note that findings for amygdala-PFC connectivity should be considered preliminary given that we did not have strong a priori hypotheses for effects regarding paternal parenting.

Interestingly, we did not find significant associations when examining maternal parenting and amygdala-PFC connectivity in this sample. It could be the case that our results differed because adolescents in this study reported on parental figures (i.e. perceived parenting behaviours) whereas in previous research, parents (mostly mothers) are often instructed to report on their parenting behaviours themselves (Romund et al., 2016) or their parenting behaviours were observed in a laboratory setting (Kopala‐Sibley et al., 2020). Perhaps there are also additional parenting factors, related to mothers (i.e. maternal mental health, stress, parental monitoring, harsh discipline), that are important for Brazilian adolescents that we did not fully capture which could have consequently affected our ability to detect associations with individual differences in neural function. Given that the majority of research has focused on youth from HICs, more research is needed to understand why we may not have seen similar patterns of associations for some of our analyses between parenting, brain activity and connectivity, and depression symptoms in Brazilian adolescents.

In our exploratory whole-brain activity analyses we found that maternal care was associated with increased right cerebellum activity. Then, in our exploratory analyses for amygdala-whole-brain connectivity, we found perceived parental care for both mothers and fathers was positively associated with amygdala connectivity to cortical and occipital regions. In prior research, the cortical regions (LG, PCG) identified have been associated with the processing of faces (Fusar-Poli et al., 2009). Speculating on our exploratory findings between parenting and amygdala-whole-brain connectivity, maybe the observation that lower maternal/paternal care is associated with decreased amygdala-LG connectivity suggests that these aspects of parenting influence the neural circuitry involved in emotional face processing, given the potential role of the LG in perceptual processing of faces (Fusar-Poli et al., 2009). This could suggest that parental care may be associated with the ability of the LG and amygdala to integrate affective signals from emotional faces. Higher paternal overprotection was also associated with decreased amygdala connectivity with the PCG, an area of the brain thought to be important for affective mentalizing (Takahashi et al., 2015), which could suggest that negative aspects of parenting may be associated with the ability of the PCG to integrate social signals from emotional face expressions. In other studies examining parenting, maternal harshness and hostility has also been associated with lowered intrinsic functional amygdala connectivity with temporal, frontal, and precentral gyrus brain regions in adolescents (Jiang et al., 2021) and more negative maternal behaviour (frequency and intensity of aggressive and dysphoric affect, anger) during an event-planning interaction has been associated with decreased activity in the LG when girls processed angry and fearful faces (Pozzi et al., 2020), suggesting some overlap with results observed here. However, given that these effects were not hypothesized, further research is necessary to interpret whole-brain analyses. Additionally, some of our results did not remain significant when we included SES as an additional covariate in our analyses. This could be the case because parenting may be confounded with SES, and we cannot tease apart whether it is parenting or SES predicting brain activity or connectivity.

It is important to consider the results of this study in light of several limitations. First, there were a high proportion of adolescents who did not have a paternal parent (or an equivalent figure) to report on for the PBI questionnaire, which resulted in a smaller sample size than would have been available had we only focused on maternal figures. This is a limitation of the study given recent evidence that suggests associations examining individual differences with neuroimaging data, like those investigated in this current study, are likely underpowered (Marek et al., 2022) which may have reduced our ability to detect significant effects. However, we did address this limitation by running sensitivity analyses in any participants that had available maternal parenting data. Second, and given the prior explanation regarding sample size, we were unable to examine these associations separately by risk group for this study because the sample sizes for each risk group were too small. Third, we want to note that adolescents’ poor relationship quality with or between parents was one of the risk factors used to determine the adolescent’s risk score for developing depression and ultimately, how participants were selected for the study. This could limit the generalizability of results to other samples, as participants in this study were selected based on their depression risk score which included (1) the presence of risk (which took into consideration poor relationship quality with parents), and (2) presence or absence of a major depressive disorder diagnosis. Fourth, this study does not employ a longitudinal framework so we cannot be certain of the direction of effects of our findings. For instance, we cannot be sure whether neural function is predicting adolescent reports of parenting or the other way around – parenting is predicting neural function. Follow-up studies of these same participants would therefore be beneficial to more strongly determine direction of effects. Lastly, we were not able to conduct mediation analyses to examine whether brain activity or connectivity mediated associations between parenting and depression symptoms possibly because of high overlapping variance between the parenting and depression measures and the measures of brain activity and connectivity. Future research using longitudinal designs to disentangle the order of effects will be helpful to test whether the patterns of brain function (activity and connectivity) identified here mediate the association between parenting and later depression symptoms. Despite the aforementioned limitations, strengths of this study include the careful study design, examination of both maternal and paternal parenting, use of adolescent-reported parenting and clinician-reported depression in order to reduce shared variance between assessments, and recruitment of youth from LMIC who have been relatively understudied in neuroscience research, filling an important gap in existing literature.

Our study extends previous work that has primarily focused on youth from HICs by examining associations between parenting and depression symptoms as well as the neural correlates associated with both parenting and depression symptoms in Brazilian youth from a MIC. Moreover, while country classifications are based on the gross national income (GNI) per capita in United States dollars (The World Bank, 2022), our analyses also consider SES as a covariate, which is measured through assessments other than income alone (i.e. basic sanitation, possessions, educational attainment, occupation), and might be able to better capture nuances within the adolescents’ realities that both influence day-to-day parenting practices and go beyond country-wide estimates. This enables the opportunity to further understand how neurobiological mechanisms (i.e. brain activity and connectivity) and social factors (i.e. parenting) are associated with depression in Brazilian youth while controlling for SES. Thus, our results sum to a body of evidence that can (1) provide evidence to support novel strategies for parenting interventions in LMICs, (2) prevent negative long-term effects (i.e. social and educational losses) from developing for these youth, and (3) help inform global mental health policy work on a broader scale. In addition, by including an examination of both maternal and paternal parenting influences on brain activity and connectivity and depression symptoms, our research adds to a better understanding of the role each parent plays in LMIC settings and across cultural contexts. Our finding that paternal parenting was associated with both depression symptoms and patterns of brain connectivity suggests that paternal parenting could be an important target of interventions in addition to maternal parenting.

To conclude, very limited research has examined adolescent brain development in youth from LMICs (Battel et al., 2021), and even less research has examined the influence of parenting on brain development and depression symptoms in Brazilian adolescents. The study herein provides novel insight into how parenting is associated with both brain activity and connectivity and depression symptoms in adolescents from Brazil, a MIC. In addition, given that the majority of research on brain development and depression in adolescents comes from HICs, the findings from this study extend previous research to LMIC settings and can be utilized to solve mental health problems within a global context to develop novel strategies to combat negative mental health outcomes. Key insights from this study include that both maternal and paternal parenting are significantly associated with adolescent brain function and depression symptoms, highlighting the important role that parents continue to play in adolescent development and mental health. Our findings suggest that parental care may be particularly important in mitigating the development of depression symptoms. Further research will be necessary to continue to investigate the associations between parenting and adolescent brain activity and connectivity, as well as how parenting and brain activity and connectivity may influence risk for the development of depression.

Author contribution statement

All authors agree to the authorship order and content of the manuscript.

Human subjects ethical statement

Authors certify compliance with APA ethical standards. This study was approved by the Brazilian National Ethics in Research Commission (CAAE 50,473,015.90000.5327). Written informed consent/assent was obtained from adolescents and their caregivers.

Acknowledgments

The authors would like to express their gratitude to the schools and individuals who participated in this study, and to all members of the IDEA team for their dedication, hard work, and insights. A.C. also thanks Christine Westall for her generous donation in support of graduate students in the UC Davis Human Development Graduate Group.

Disclosure statement

V.M. has received in the past research funding from Johnson & Johnson as part of a research program on depression and inflammation, but the research described in this paper is unrelated to this funding. The remaining authors have declared that they have no competing or potential conflicts of interest.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/02673843.2024.2354910

Additional information

Funding

A.C. was supported by the Westall Fellowship from UC Davis. The IDEA (Identifying Depression Early in Adolescence) project of which this research belongs to is funded by an MQ Brighter Futures grant (MQBF/1 IDEA). Additional support was provided by the Medical Research Council (MC_PC_MR/R019460/1) and the Academy of Medical Sciences (GCRFNG\100281) under the Global Challenges Research Fund. This work is also supported by research grants from Brazilian public funding agencies Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq; 477129/2012-9 and 445828/2014-5), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES; 62/2014), and Fundação de Amparo à Pesquisa do Estado do Rio Grande do Sul (FAPERGS; 17/2551-0001009- 4). C.K. is a CNPq researcher and an Academy of Medical Sciences Newton Advanced Fellow. V.M. is supported by the Medical Research Foundation [MRF160-0005-ELPMONDE] and by the National Institute for Health Research (NIHR) Mental Health Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London. J.S. and C.K. were also supported by the National Institute of Mental Health (NIMH) [R21MH124072]. The views expressed are those of the authors and not necessarily those of the National Health Service, the NIHR, the NIMH, or the Department of Health and Social Care.