Neuroanatomical Correlates of Perceived Stress Controllability in Adolescents and Emerging Adults

Abstract

Stressful life events predict changes in brain structure and increases in psychopathology, but not everyone is equally affected by life stress. The learned helplessness theory posits that perceiving life stressors as uncontrollable leads to depression. Evidence supports this theory for youth, but the impact of perceived control diverges based on stressor type: perceived lack of control over dependent (self-generated) stressors is associated with greater depression symptoms when controlling for the frequency of stress exposure, but perceived control over independent (non-self-generated) stressors is not. However, it is unknown how perceived control over these stressor types is associated with brain structure. We tested whether perceived lack of control over dependent and independent life stressors, controlling for stressor exposure, is associated with gray matter (GM) in a priori regions of interest (ROIs; mPFC, hippocampus, amygdala) and across the cortex in a sample of 108 adolescents and emerging adults ages 14–22. There were no associations across the full sample between perceived control over either stressor type and GM in the ROIs. However, less perceived control over dependent stressors was associated with greater amygdala gray matter volume in female youth and greater medial prefrontal cortex thickness in male youth. Further, whole-cortex analyses revealed less perceived control over dependent stressors was associated with greater GM thickness in cortical regions involved in cognitive and emotional regulation. Thus, appraisals of control have distinct associations with brain morphology while controlling for stressor frequency, highlighting the importance of differentiating between these aspects of the stress experience in future research.

Introduction

Stress is increased during adolescence and emerging adulthood compared to childhood, driven by increased academic workload and the transition to independence (Hankin et al., 2016). Stressful life events can affect brain structure and predict increases in psychopathology (Kendler & Gardner, 2016; Radley et al., 2015), which could be particularly harmful during adolescence, a time of ongoing brain development coinciding with onset of internalizing disorders (Andersen & Teicher, 2008). However, life stress does not affect everyone equally: perceived control over stressors is important in shaping stress outcomes, including depression (e.g., Fassett-Carman et al., 2019, 2020). In rodent models, the ability to exert control over stressors modulates their behavioral and neural effects (Maier & Seligman, 2016). However, it is unknown how perceived control over life stressors and brain structure are associated in humans. The current study investigates this relation in a sample of adolescents and emerging adults.

Neural Correlates of Stress Exposure

Stress causes structural changes in brain regions that are associated with psychopathological states (e.g., Radley et al., 2015; Teicher & Samson, 2016). In rodents, chronic stress leads to dendritic atrophy in the medial prefrontal cortex (mPFC) and hippocampus, areas high in glucocorticoid receptors and critical to stress and mood regulation, and dendritic expansion in the amygdala, a region crucial for fear, emotion, and salience processing (McEwen et al., 2015; Radley et al., 2015). In humans, similar patterns are observed: stressful life events are associated with less mPFC, insula, and hippocampal gray matter volume (GMV) in adults (Ansell et al., 2012; Gianaros et al., 2007; Papagni et al., 2010; H. Li et al., 2014). However, these areas are not consistently found across all studies. Life stress is additionally associated with altered GM in the parahippocampal gyrus (H. Li et al., 2014; Papagni et al., 2010) anterior cingulate cortex (ACC; Kuhn et al., 2015; Y. Li et al., 2017; Papagni et al., 2010), and amygdala (Gerritsen et al., 2014; Ganzel, Kim, Glover, & Temple, 2008; Lotze et al., 2020; Sublette et al., 2016), but with mixed findings of either increased or decreased GM.

Learned Helplessness Theory

These inconsistencies could stem from variation between studies in severity and recency of life stressors, type of life stressor, or differences in how the stressors are perceived. Appraisals of stressful events are important in shaping psychopathological outcomes and could shape, and/or be shaped by, neural structure as well. Learned helplessness theory posits that perceived lack of control over stressors leads to helplessness and subsequent depression symptoms (Maier & Seligman, 1976; Seligman, 1975). Supporting this theory, perceived lack of control over acute, in-lab stressors leads to greater emotional arousal (Maier & Seligman, 1976; Seligman, 1975; Hartley et al., 2014; Geer, Davison, & Gatchel, 1970; Glass, Singer, & Friedman, 1969) and reduced working memory (Wanke & Schwab, 2020) compared to when control is detected. Further, perceived lack of control over life stressors is associated with depression symptoms (Fassett-Carman et al., 2019; 2020). However, no studies have tested the structural neural correlates of perceived control over life stressors in humans.

Neural Processing of Stress Controllability

Relations between GM structure and perceived lack of control over life stressors could reflect a variety of mechanisms. One possibility is that the degree to which one perceives control over their stressors will affect neural processing when faced with stress, leading to differences in attributes of GM structure. Rodent models provide clues as to brain regions involved in processing controllable and uncontrollable stressors. When rodents detect control over a stressor, the vmPFC inhibits the typical widespread serotonergic stress response that projects to areas including the amygdala (Amat, Matus-Amat, Watkins, & Maier, 1998a) and hippocampus (Amat, Matus-Amat, Watkins, & Maier, 1998b). This mPFC inhibition protects against the helpless, depression-like behavior exhibited by rats lacking control over the stressors (Amat et al., 2008). If a similar mechanism occurs in humans, perceived control over life stressors may affect the neural processing of life stress, leading to a relation between perceived control and GM. The relation between perceived control and GM has not been previously tested and is addressed in the current study.

GM May Affect Stress Appraisals

It is additionally possible that gray matter structure affects appraisals of control over life stressors. First, much of the evidence that life stress is associated with GM in humans uses the Perceived Stress Scale (Cohen et al., 1983), a measure that includes items about perceived severity and control over life stressors, and the findings of these studies may be partially due to perceptions rather than occurrence of stressors (e.g., H. Li et al., 2014; Zimmerman et al., 2016; Moreno et al., 2017). Further, the amygdala and hippocampus are thought to be involved in perception of emotion (McEwen & Gianaros, 2010), and there is some evidence for relations between GM structure and stressors appraisals (e.g., lower hippocampal GMV is associated with threat appraisals; Grupe et al., 2019). Although these studies are cross-sectional and directionality of effects cannot be determined, they raise the possibility that individual differences in GM structure play a role in shaping stressor appraisals. Thus, there are possible bidirectional relations between GM and appraisals of control over life stressors. The current study cannot determine directionality of results, but addresses how perceived control over stressors relates to GM structure.

Dependent vs Independent Stressors

One factor that may play a role in the relation between perceived control over stressors and GM is the dependence of a stressor. Only perceived control over dependent stressors, those that individuals play a role in causing (e.g., fights with a friend, getting bad grades), are associated with depression symptoms (Fassett-Carman et al., 2019). This finding is consistent with research showing that dependent stressors are more likely to lead to depression symptoms than independent, or fateful, stressors (e.g., crime in the neighborhood, illness in the family; Kendler et al., 1999; Technow et al., 2015). Individuals may understand they play a role in causing dependent stressors, leading to depression symptoms such as low self-esteem and guilt, which may be exacerbated by feelings of being unable to control these stressors (Fassett-Carman et al., 2019). Although this association has not been tested, it is consistent with work demonstrating that attributing negative events to internal causes is associated with depression symptoms (Huang et al., 2015). Further, these stressors may be more salient in the lives of adolescents and emerging adults, making exposure to, and appraisals of, these stressors more likely to lead to depression. Stressor type is thus important to consider when investigating the relation between perceived control over stressful experiences and neural structure, but this has not previously been tested.

Adolescent and Emerging Adult Development

Development timing is also important to consider when testing the relation between perceived control over life stressors and neural structure. The time period of adolescence to emerging adulthood is a time of heightened life stress, depression symptom onset, and intense GM development, including in areas canonically associated with stress (i.e., mPFC, hippocampus, and amygdala; Wierenga et al., 2014a, b): thus, this period may be characterized by heightened plasticity in these regions allowing for greater impacts of stressors and stressor appraisals, or changes in appraisals stemming from GM alterations.

Whole-brain GMV takes an inverted-U shape trajectory across development, increasing during childhood due to synaptic proliferation and decreasing across adolescence and emerging adulthood (Gogtay & Thompson, 2010; Giedd et al., 2015; Matsuzawa et al., 2001) likely due to synaptic pruning and increased myelination (among other mechanisms) as regions and networks become more efficient based on environmental input (Johnson, Blum, & Giedd, 2009; Gogtay & Thompson, 2010). However, specific regions take distinct trajectories. For example, GMV in the hippocampus and amygdala increase until late adolescence with a subsequent decline (Wierenga et al., 2014a).

Development of cortical regions involves change in both cortical thickness and surface area, characterized by distinct genetic influences, neurobiological mechanisms, and neurodevelopmental trajectories. It is thought that cortical thickness is influenced by synaptic proliferation and pruning (Rakic, 1988; Tau & Peterson, 2010), and then thins as networks are being pruned and becoming more efficient during adolescent and emerging adulthood, including in areas associated with stress such as the mPFC (Tau & Peterson, 2010; Tamnes et al., 2017; Ducharme et al., 2016; Wierenga et al., 2014b). However, this scenario is debated (see Paus et al., 2008), and other mechanisms are likely involved, including changes in glial cell number (Mills & Tamnes, 2014), glial support of GM structure (Vidal-Pineiro et al., 2020), myelination, and axon caliber (Paus et al., 2008). Surface area development is likely affected by mechanisms such as sulcul widening and cortical stretching from increased myelination rather than pruning (Aleman-Gomez et al., 2013, Seldon, 2005). There is conflicting evidence for gradual decrease (Tamnes et al., 2017; Weirenga et al., 2014) and increase (Vijayakumar et al, 2016) in surface area during adolescence and emerging adulthood. Thus, although changes in surface area in development are well documented, more research is needed to better understand the mechanisms underlying those changes.

Importantly, protracted GM development during adolescence and emerging adulthood may affect links between appraisals and GM structure. First, ongoing GM development is posited to create sensitive periods in which stressors exert long-lasting effects on brain structure and behavior (Andersen & Teicher, 2008; Fuhrmann, Knoll, & Blakemore, 2015). Appraisals may thus be particularly important in shaping stress outcomes at this time. Further, as attributes of GM structure might be associated with cognitive processes that contribute to shaping the degree to which an individual perceives control over their stressors, GM development may affect stressor appraisals. For example, GM development in regions involved in detecting threat (McEwen & Gianaros, 2010; Wierenga et al., 2014a) could affect appraisals of severity and perceived control over these stressors. However, it is unknown how GM and appraisals are associated during adolescence and emerging adulthood.

Current Study

The current study investigates the relation between perceived control over dependent and independent life stressors and GM morphometry in an adolescent and emerging adult sample (age 14–22). We tentatively hypothesized that perceived lack of control over dependent stressors would be associated with decreased mPFC and hippocampal GM measures (thickness and surface area for mPFC, and volume for hippocampus), and increased amygdala GMV. While these structural effects are associated with life stress exposure in both rodent models and humans, it would not be surprising for appraisals of control to have distinct neural correlates (Fassett-Carman et al., 2020). We additionally ran whole-brain exploratory analyses to look for clusters where GM thickness or area is associated with perceived control over life stressors. We tested for age and gender moderation due to GM development and gender differences in this developmental across the relatively wide age range of the sample (Giedd et al., 1999; Lenroot et al., 2007), and gender differences in the relation between stress and psychopathology (Hankin et al., 2007). These analyses were preregistered prior to conducting analyses¹.

Method

Participants

All participants (N=144, 50.1% Female) were adolescents and emerging adults ages 14–22 from the greater Denver/Boulder metro area in the Colorado Cognitive Neuroimaging Family Emotion Research (CoNiFER) study. Participants were drawn from an unselected community sample originally recruited to participate in the GEM (genes, environment and mood) study (NIH Grant R01 MH077195) and an associated follow-up study (R21MH102210). For details of the two samples and studies, see Hankin et al. (2015) and Snyder et al. (2019). These community samples were recruited from the Denver metro area via public schools and using direct mail to target zip codes to maximize demographic and socioeconomic diversity. All participants were screened to be free of a history of neurological insult, spoke English as their first language, and did not report having dyslexia or difficulty reading. Participants gave written informed consent (if age 18–22) or assent (if age 14–17) and parents gave informed consent for minors. All procedures were approved by the University of Colorado Institutional Review Board.

Participants were included in the current analysis if they completed the neuroimaging portion of the study (n=126) and the ALEQ stress measure (n=129). Of the participants who completed the ALEQ measure, six endorsed zero dependent stressors and fifteen endorsed zero independent stressors, preventing the assessment of their perceived control over dependent and independent stressors, respectively. The final sample thus included 108 participants for dependent stressor analyses, and 101 participants for independent stressor analyses.

Procedure

Analyses presented here use data from Timepoint 1 of the Colorado CONIFER longitudinal study. Timepoint 1 participation consisted of two 1-hour neuroimaging sessions to collect structural and functional data, and one session for behavioral tasks and assessment of psychopathology.

Measures

Stressful life events.

The Adolescent Life Event Questionnaire – Revised (ALEQ-R; Fassett-Carman et al., 2019) assesses 24 dependent (i.e., self-generated) and 26 independent (i.e., fateful) negative life events typically experienced by youth, occurring within the past 6 months. It has been shown to be a valid measure of stress in youth that consistently predicts psychopathology (e.g., Calvete et al., 2017; Young & Dietrich, 2015; Young et al., 2012). For nine items (eight independent stressors, one dependent stressor) that typically occur infrequently (e.g., “Parents getting divorced”), participants indicated whether each even occurred within the past 6 months (“This event did not happen” [0] or “This event happened” [4]). For the rest of the items, participants rated how often each event occurred within the past 6 months on a Likert scale from 0 (“Never”) to 4 (“Always”). For all items that were endorsed, participants rated perceived control over the stressor (“How much control did you feel like you had during that time? e.g., How much did you feel like you could make things better or less stressful?”) from 1 (“No control/Completely out of my control”) to 5 (“Completely in my control”), and perceived severity of the stressor (“How stressful was it for you?”) from 1 (“Not very stressful”) to 5 (“Very stressful”). Frequency ratings were summed across items to calculate a total frequency score for each participant. Mean controllability scores were calculated across stressors experienced for each participant to ensure they were statistically independent of frequency.

Structural MRI acquisition and surface-based morphometry.

Whole brain high-resolution structural MRI data were acquired at the Intermountain Neuroimaging Consortium located at the University of Colorado Boulder. Scans were conducted using a research-dedicated Siemens 3-Tesla PRISMA MRI scanner for all but 18 participants, for whom data were acquired on the pre-upgrade version of the same magnet (TIM TRIO). A 32-channel headcoil was used for radiofrequency transmission and reception. Data pertaining to GM structure were acquired via a T1-weighted Magnetization Prepared Gradient Echo sequence in 224 sagittal slices, with a repetition time (TR)= 2400 ms, echo time (TE)= 2.07 ms, flip angle= 8 degrees, field of view (FoV)= 256 mm, and voxel size of .8 mm³.

T1-weighted structural images were brain extracted using a hybrid watershed/surface deformation procedure (Ségonne et al., 2004), followed by a transformation into Talairach space, intensity normalization (Sled et al., 1998), tessellation of the gray/white matter boundary (Fischl et al., 2001), and surface deformation along intensity gradients to optimally differentiate gray matter, white matter, and cerebral spinal fluid (CSF) boundaries (Dale et al., 1999; Fischl and Dale, 2000). The resulting segmented surfaces were registered to a standard spherical inflated brain template (Fischl et al., 1999a, b), parcellated according to gyral and sulcal structure (Desikan et al., 2006; Fischl et al., 2004), and then used to compute a range of surface-based measurements, including cortical volume, surface area, and thickness. Prior to running surface-based analyses, data quality assurance was checked using FreeSurfer’s standard quality assurance tools (https://surfer.nmr.mgh.harvard.edu/fswiki/QATools), including visual inspection of the relevant parcellations and checking for outliers in morphometry measures, signal-to-noise ratio, and white matter intensity. ROI gray matter computations and whole cortex exploratory analyses were conducted in fsaverage7 space, with 163,842 vertices per hemisphere.

Analysis

To test whether perceived lack of control over life stressors was associated with gray matter structure in ROIs, we used linear regressions with GMV as the outcome measure for subcortical ROIs (left and right amygdala and hippocampus), and surface area and thickness as outcome measures for cortical ROIs (left and right mPFC). Additional linear regressions tested for gender and age moderation of these relations. Whole-brain exploratory analyses were conducted with permutation testing, employing the Permutation Analysis of Linear Models (PALM) tool (Winkler et al., 2014). Data at each vertex were shuffled 10,000 times, using sign-flipping for main effect and age-moderation models, and regrouping for tests of gender moderation. We used exchangeability blocks to account for dependence between data points created by 14 sibling-pairs and one sibling-trio present in the dataset.

All analyses controlled for stress frequency, age, and gender, consistent with past analyses demonstrating a relation between perceived lack of control over dependent stress and depression (Fassett-Carman et al., 2019; 2020). Cortical surface area and subcortical GMV analyses were conducted both with and without controlling for average surface area and total brain volume excluding ventricles, respectively, because these gray matter measures are associated with total head size, whereas cortical thickness is not (Barnes et al., 2010).

To account for the large number of analyses conducted, but balance this with the exploratory nature of the study as no previous study has tested how perceived lack of control over stress is associated with brain structure, we analyzed and present the data at multiple levels of correction. For ROIs analyses, we present all uncorrected analyses at an alpha of .05 in the tables, but focus the results and discussion on analyses that pass FDR correction for multiple comparisons at the .05 level. For whole-brain analyses, we present uncorrected clusters with a vertex-wise threshold of p<.01 and a maximum vertex of p<.001 in tables and figures, but focus the results and discussion sections on clusters that pass FDR and FWE correction at the .05 level.

Results

For descriptive statistics, bivariate correlations, and tests for gender differences in stress variables, see Supplemental Materials (Tables S1 & S2).

ROI Results

Datapoints that were outliers (+/− 3 SD from the mean) on the ALEQ or ROI measures were excluded from analysis (n=2 for dependent stress frequency, n=1 for left hippocampus GMV, n=1 for right hippocampus GMV, n=1 for left amygdala GMV, n=1 for right amygdala GMV).

Main effect ROI results.

No relations between perceived lack of control over dependent or independent stressors and ROIs, controlling for age, gender and stress frequency, with and without whole-brain volume and area controls, passed FDR correction (Table 1).

Table 1.

Associations between Perceived Lack of Control over Life Stressors and ROIs Controlling for Stress Frequency, Age and Gender, for Dependent and Independent Stressors, with and without Whole Brain Control

		With whole brain control					Without whole-brain control
ROI		b	β	SE	t	p	b	β	SE	t	p
Dependent Stressors	L. Amygdala	12.924	0.052	18.402	0.702	.484	18.360	0.074	23.456	0.783	.436
	R. Amygdala	5.786	0.023	18.049	0.321	.749	11.309	0.046	22.792	0.496	.621
	L. Hippocampus	58.176	0.114	43.737	1.330	.187	76.013	0.149	53.298	1.426	.157
	R. Hippocampus	49.941	0.105	35.949	1.389	.168	67.763	0.142	47.188	1.436	.154
	L. mPFC
	Area	34.953	0.134	24.384	1.433	.155	25.047	0.096	27.472	0.912	.364
	Thickness	-	-	-	-	-	0.040	0.196	0.022	1.773	.079
	R. mPFC
	Area	33.984	0.144	19.864	1.711	.090	22.607	0.096	24.736	0.914	.363
	Thickness	-	-	-	-	-	0.046	0.213	0.024	1.934	.056
Independent Stressors	L. Amygdala	−5.523	−0.030	13.267	−0.416	.678	−20.919	−0.114	16.288	−1.284	.202
	R. Amygdala	−22.532	−0.115	13.003	−1.733	.086	−35.256	−0.180	15.826	−2.228	.028
	L. Hippocampus	27.426	0.069	32.657	0.840	.403	−3.203	−0.008	39.293	−0.082	.935
	R. Hippocampus	−13.719	−0.037	27.125	−0.506	.614	−43.189	−0.116	34.426	−1.255	.213
	L. mPFC
	Area	5.492	0.026	18.905	0.291	.772	−11.809	−0.057	20.446	−0.578	.565
	Thickness	-	-	-	-	-	0.027	0.175	0.016	1.719	.089
	R. mPFC
	Area	24.459	0.134	14.625	1.672	.098	2.610	0.014	18.108	0.144	.886
	Thickness	-	-	-	-	-	0.017	0.107	0.016	1.047	.298

Note. Bold=p<.05

^*=passes FDR correction at the .05 level.

ROI gender moderation.

Gender moderated the relation between perceived lack of control over dependent stress and right amygdala GMV, with (β=0.152, p=.022) and without (β=0.252, p=.002) controlling for whole brain volume, but only passed FDR correction without controlling for whole brain volume (Table 2, Figure 1). In particular, female youth (β=0.355, p=.013), but not male youth (β=-0.277, p=.107), who perceived their dependent life stressors to be more uncontrollable showed greater right amygdala GMV (Table 3). The same pattern of gender moderation was found for the left amygdala without the whole brain control (β=0.172, p=.042), but this relation did not pass the FDR threshold.

Table 2.

Gender Moderation of Associations between Perceived Lack of Control over Life Stressors and ROIs Controlling for Age and Stress Frequency, for Dependent and Independent Stressors, with and without Whole Brain Control

		With whole brain control					Without whole-brain control
ROI		b	β	SE	t	p	b	β	SE	t	p
Dependent Stressors	L. Amygdala	19.309	0.077	16.863	1.145	.255	42.941	0.172	20.842	2.060	.042
	R. Amygdala	38.020	0.152	16.305	2.332	.022	62.772	0.252	19.722	3.183	.002 *
	L. Hippocampus	−0.626	−0.001	40.328	−0.016	.988	45.890	0.089	48.319	0.950	.345
	R. Hippocampus	−17.223	−0.036	33.102	−0.520	.604	29.711	0.062	42.870	0.693	.490
	L. mPFC
	Area	−10.998	−0.042	22.853	−0.481	.631	21.090	0.080	24.835	0.849	.398
	Thickness	-	-	-	-	-	-0.035	-0.168	0.020	−1.716	.089
	R. mPFC
	Area	4.509	0.019	18.633	0.242	.809	40.158	0.169	22.080	1.819	.072
	Thickness	-	-	-	-	-	−0.072	−0.330	0.020	−3.560	.001 *
Independent Stressors	L. Amygdala	21.590	0.115	14.093	1.532	.129	38.049	0.203	17.049	2.232	.028
	R. Amygdala	14.261	0.071	13.899	1.026	.308	2.751	0.149	16.663	1.785	.077
	L. Hippocampus	−1.337	−0.003	34.592	−0.039	.969	35.267	0.088	41.331	0.853	.396
	R. Hippocampus	0.978	0.003	28.733	0.034	.973	36.141	0.096	36.160	0.999	.320
	L. mPFC
	Area	−4.569	−0.022	20.071	−0.228	.820	7.313	0.034	21.947	0.333	.740
	Thickness	-	-	-	-	-	<.001	0.001	0.017	0.009	.993
	R. mPFC
	Area	−28.824	−0.154	15.244	−1.891	.062	−13.442	−0.072	19.399	−0.693	.490
	Thickness	-	-	-	-	-	−0.019	−0.117	0.017	−1.090	.278

Note. Gender coded as −1=male, 1=female. Bold=p<.05

^*=passes FDR correction at the .05 level.

Figure 1.

Gender Moderation of the Relation between Perceived Lack of Control Over Life Stressors and Gray Matter in ROIs (Hippocampus, Amygdala, and mPFC)

Note. Only significant gender moderation relations shown. *=passes FDR correction. Orange=mPFC, green=amygdala, blue=hippocampus.

Table 3.

Associations between Perceived Lack of Control over Life Stressors and ROIs Controlling for Stress Frequency and Age, for Dependent and Independent Stressors, with and without Whole Brain Control, Split by Gender

			With whole brain control					Without whole-brain control
ROI			b	β	SE	t	p	b	β	SE	t	p
Girls	Dependent Stressors	L. Amygdala	31.226	0.177	20.274	1.540	.130	57.510	0.325	25.158	2.286	.026
		R. Amygdala	36.659	0.198	19.712	1.860	.069	65.635	0.355	25.632	2.561	.013
		L. Hippocampus	61.057	0.131	59.784	1.021	.312	118.515	0.254	66.980	1.769	.083
		R. Hippocampus	44.589	0.118	46.900	0.951	.346	95.470	0.253	54.379	1.756	.085
		L. mPFC
		Area	10.293	0.044	28.009	0.367	.715	28.205	0.122	32.718	0.862	.393
		Thickness	-	-	-	-	-	0.015	0.073	0.029	0.522	.604
		R. mPFC
		Area	31.324	0.137	25.586	1.224	.226	51.112	0.223	31.861	1.604	.115
		Thickness	-	-	-	-	-	−0.015	−0.067	0.032	−0.458	.649
	Independent Stressors	L. Amygdala	24.547	0.155	18.539	1.324	.192	29.015	0.183	23.679	1.225	.227
		R. Amygdala	−3.273	−0.020	18.893	−0.173	.863	1.112	0.007	23.613	0.047	.963
		L. Hippocampus	34.047	0.078	58.303	0.584	.562	44.067	0.101	66.415	0.664	.510
		R. Hippocampus	−6.195	−0.018	44.790	-0.138	.891	2.034	0.006	51.915	0.039	.969
		L. mPFC
		Area	−4.642	−0.021	28.259	−0.164	.870	−9.472	−0.042	33.884	−0.280	.781
		Thickness	-	-	-	-	-	0.025	0.134	0.027	0.909	.368
		R. mPFC
		Area	−5.357	−0.025	24.731	−0.217	.829	−10.694	−0.050	32.353	−0.331	.742
		Thickness	-	-	-	-	-	−0.006	−0.033	−0.028	−0.214	.831
Boys	Dependent Stressors	L. Amygdala	−13.710	−0.056	34.749	−0.395	.695	−36.232	−0.149	42.738	−0.848	.401
		R. Amygdala	−43.802	−0.191	33.155	−1.321	.193	−63.611	−0.277	38.658	−1.645	.107
		L. Hippocampus	42.475	0.084	70.389	0.603	.549	5.741	0.011	89.479	0.064	.949
		R. Hippocampus	64.533	0.134	60.752	1.062	.294	24.402	0.051	86.039	0.284	.778
		L. mPFC
		Area	61.779	0.231	44.834	1.378	.175	25.702	0.096	46.908	0.548	.586
		Thickness	-	-	-	-	-	0.056	0.282	0.034	1.663	.103
		R. mPFC
		Area	31.598	0.142	33.673	0.938	.353	−10.324	-0.046	39.169	−0.264	.793
		Thickness	-	-	-	-	-	.125	0.582	0.033	3.785	<.001
	Independent Stressors	L. Amygdala	-21.114	−0.129	20.102	−1.050	.299	−47.394	−0.290	23.091	−2.052	.046
		R. Amygdala	−33.256	−0.204	19.724	−1.686	.099	−57.104	−0.350	22.471	−2.541	.015
		L. Hippocampus	30.990	0.090	40.799	0.760	.452	−28.179	−0.082	50.984	−0.553	.583
		R. Hippocampus	−8.985	−0.027	37.196	−0.242	.810	−70.169	−0.207	49.172	−1.427	.160
		L. mPFC
		Area	−0.417	−0.002	27.481	−0.015	.988	−19.422	−0.104	27.497	−0.706	.484
		Thickness	-	-	-	-	-	0.026	0.182	0.021	1.250	.218
		R. mPFC
		Area	44.077	0.283	18.646	2.364	.022	14.136	0.091	22.665	0.624	.536
		Thickness	-	-	-	-	-	0.031	0.214	0.021	1.470	.148

Note. Bold=p<.05

^*=passes FDR correction at the .05 level.

Additionally passing FDR correction, gender moderated the relation between perceived lack of control over dependent stressors and right mPFC thickness (β=-0.330, p=.001), with a trend in the same direction in the left hemisphere. In particular, male youth (β=0.582, p<.001), but not female youth (β=-0.067, p=.649), who perceived their stressors to be more uncontrollable showed greater right mPFC thickness (Table 3).

ROI age moderation.

No age moderation passed FDR correction at the .05 level (Table 4).

Table 4.

Age Moderation of Associations between Perceived Lack of Control over Life Stressors and ROIs Controlling for Stress Frequency and Gender, for Dependent and Independent Stressors, with and without Whole Brain Control

		With whole brain control					Without whole-brain control
ROI		b	β	SE	t	p	b	β	SE	t	p
Dependent Stressors	L. Amygdala	−2.329	−0.012	12.661	−0.184	.854	−22.620	−0.120	15.636	−1.447	.151
	R. Amygdala	−13.721	−0.073	12.335	−1.112	.269	−32.270	−0.172	15.013	−2.149	.034
	L. Hippocampus	−17.258	−0.045	29.662	−0.582	.562	−60.292	−0.158	34.899	−1.728	.087
	R. Hippocampus	−3.506	−0.010	24.420	−0.144	.886	−47.138	−0.132	31.001	−1.521	.132
	L. mPFC
	Area	1.669	0.008	16.484	.101	.920	−8.523	−0.043	18.480	−0.461	.646
	Thickness	-	-	-	-	-	0.031	0.199	.015	2.072	.041
	R. mPFC
	Area	−18.375	−0.103	13.302	-1.381	.170	−29.809	−0.167	16.388	−1.819	.072
	Thickness	-	-	-	-	-	0.013	0.080	0.016	0.825	.411
Independent Stressors	L. Amygdala	−2.914	−0.022	9.856	−0.296	.768	−13.969	−0.105	12.023	−1.162	.248
	R. Amygdala	−5.816	−0.041	9.776	−0.595	.553	−19.026	−0.134	11.629	−1.636	.105
	L. Hippocampus	−46.606	−0.167	23.434	−1.989	.050	−71.901	−0.258	27.683	−2.597	.011
	R. Hippocampus	−13.406	−0.051	19.826	−0.676	.501	−38.618	−0.147	24.792	−1.558	.123
	L. mPFC
	Area	5.530	0.037	13.796	0.401	.689	0.512	0.003	15.182	0.034	.973
	Thickness	-	-	-	-	-	0.025	0.218	0.012	2.146	.034
	R. mPFC
	Area	−6.569	−0.050	10.661	−0.616	.539	−12.825	−0.097	13.381	−0.958	.340
	Thickness	-	-	-	-	-	0.010	0.088	0.012	0.841	.403

Note. Bold=p<.05

^*=passes FDR correction at the .05 level.

Whole Brain Exploratory Results

Main effect whole brain results.

Perceived lack of control over dependent stressors was associated with cortical thickness in 10 clusters across the cortex that passed FDR correction (rostral middle frontal gyrus, two clusters in the lateral orbitofrontal cortex, rostral anterior cingulate cortex, postcentral gyrus, paracentral lobule, middle temporal gyrus, banks superior temporal sulcus, precuneus, and inferior parietal lobule), and three that passed FWE correction (left superior frontal gyrus, right precentral gyrus, and right superior temporal gyrus) at the .05 level (Table 5, Figure 2). There were no clusters where perceived lack of control over dependent stress was associated with surface area that passed correction for multiple comparisons, with or without controlling for whole-brain average area (Figure S1).

Table 5.

Vertex-Wise Cortical Neuroanatomical Correlates of Perceived Lack of Control over Dependent and Independent Stressors.

	Cortical Measure	Hemi	+/−	Region	Size (mm2)	MaxP (log(10)p)	X	Y	Z	Passes FDR	Passes FWE
Dependent Stressors	Thickness	l	+	MFG	1066.89	4	−39.8	15.3	44.9	n	n
	Thickness	l	+	SFG	509.73	4	−20.0	−8.1	52.5	n	y
	Thickness	l	+	paracentral	263.43	4	−8.8	−22.1	52.0	n	n
	Thickness	l	+	MTG	120.98	4	−57.2	−23.0	−10.7	n	n
	Thickness	l	+	MTG	271.97	3.699	−58.1	−9.5	−18.7	n	n
	Thickness	l	+	precentral	115.24	3.046	−52.1	−2.4	41.7	n	n
	Thickness	l	+	postcentral	191.86	3.000	−40.5	−26.1	60.2	n	n
	Thickness	r	+	rMFG	533.84	4	36.7	36.8	14.3	y	n
	Thickness	r	+	paracentral	300.89	4	7.9	−15.4	66.7	y	y
	Thickness	r	+	rACC	294.52	4	12.5	38.9	2.3	y	n
	Thickness	r	+	banksSTS	220.87	4	65.2	−31.5	9.2	y	y
	Thickness	r	+	precuneus	175.51	4	8.4	−45.4	45.8	y	n
	Thickness	r	+	IPL	153.20	4	52.9	−47.1	38.0	y	n
	Thickness	r	+	MTG	138.92	4	62.0	−46.7	−0.2	y	n
	Thickness	r	+	lOFC	670.75	3.699	25.7	19.8	−18.2	y	n
	Thickness	r	+	lOFC	166.24	3.398	16.3	51.0	−15.2	y	n
	Thickness	r	+	postcentral	88.92	3.301	37.8	−17.5	39.9	y	n
	Thickness	r	+	IPL	69.74	3	45.8	−70.3	14.5	n	n
	Area	l	−	precuneus	266.83	4	−5.1	−63.4	37.2	n	n
	Area	l	−	PCC	125.81	3.398	−14.7	−14.8	37.1	n	n
	Area w/ WBC	l	−	precuneus	185.20	4	−4.6	−63.6	36.1	n	n
	Area w/ WBC	l	+	SFG	644.91	3.097	−16.6	53.7	16.9	n	n
	Area w/ WBC	l	+	STG	81.33	3.046	−62.1	−31.2	9.8	n	n
	Area	r	−	lingual	369.61	3.523	24.3	−71.0	−1.4	n	n
	Area	r	−	SMG	187.08	3.523	49.0	−21.2	19.2	n	n
	Area	r	−	MTG	267.25	3.222	49.3	−58.5	5.9	n	n
	Area	r	+	STG	223.98	3.097	43.4	9.5	−26.1	n	n
	Area w/ WBC	r	+	MTG/STG	687.35	4	47.2	6.8	−29.2	n	n
	Area w/ WBC	r	+	precuneus	126.77	3.699	23.0	−56.4	18.2	n	n
	Area w/ WBC	r	−	SMG	126.60	3.046	51.0	−20.6	18.4	n	n
Independent Stressors	Thickness	l	+	MTG	111.68	3.523	−55.5	−10.0	−18.7	n	n
	Thickness	l	+	MFG	192.39	3.398	−23.2	44	14.3	n	n
	Thickness	l	+	lOFC	150.15	3.222	−15.1	22.4	−19.3	n	n
	Thickness	l	+	lOFC	71.24	3.222	−23.0	34.6	−10.3	n	n
	Thickness	r	+	lOFC	278.50	4	33.4	22.2	−15.9	n	y
	Thickness	r	+	OC	190.41	4	46.6	−73.8	11.7	n	n
	Thickness	r	+	SFG	72.51	3.301	14.1	16.8	54.3	n	n
	Thickness	r	+	postcentral	65.65	3.155	28.5	−31.3	56.0	n	n
	Thickness	r	+	rMFG	114.65	3	34.1	33.6	24.3	n	n
	Area	l	−	SFG	1400.83	4	−19.2	3.4	50.4	n	n
	Area	l	−	lOFC	534.29	3.097	−15.1	44.0	−16.1	n	n
	Area	l	−	STG	364.25	3.046	−49.1	−14.7	−13.5	n	n
	Area	l	−	precuneus	104.72	3.046	−5.3	−66.1	36.2	n	n
	Area w/ WBC	l	−	SFG	665.64	3.699	−19.6	6.9	54.6	n	n
	Area w/ WBC	l	−	lOFC	295.62	3.155	−21.4	43.9	−12.8	n	n
	Area	r	−	lOFC	490.92	4	29.1	21.6	−18.2	n	n
	Area	r	−	SFG	224.28	3.097	16.5	14.7	54.6	n	n
	Area w/ WBC	r	−	lOFC	278.01	3.699	30.1	21.3	−18.1	n	n
	Area w/ WBC	r	+	postcentral	145.83	3.222	62.8	−7.2	19.9	n	n

Note. All results passed a vertex-wise threshold of .01 with a max p less than .001. Clusters with sub-clusters passing FDR and FEW corrections are indicated. WBC=whole brain control (average area); y=yes; n=no; hemi=hemisphere; r=right; l=left; MFG=middle frontal gyrus; rMFG= rostral middle frontal gyrus; SFG=superior frontal gyrus; paracentral=paracentral lobule; MTG=middle temporal gyrus; postcentral=postcentral gyrus; PCC=posterior cingulate cortex; STG=superior temporal gyrus; lOFC=lateral orbitofrontal cortex; banksSTS= Banks superior temporal sulcus; IPL=inferior parietal lobule; rACC=rostral anterior cingulate cortex; lingual=lingual gyrus; SMG=supramarginal gyrus.

Figure 2.

Cortical Thickness Correlates of Perceived Lack of Control over Dependent and Independent Stressors

Note. RH=right hemisphere; LH=left hemisphere; banksSTS= Banks superior temporal sulcus; IPL=inferior parietal lobule; lOC=lateral occipital cortex; lOFC=lateral orbitofrontal cortex; MFG=middle frontal gyrus; rMFG= rostral middle frontal gyrus; MTG=middle temporal gyrus; paracentral=paracentral gyrus; postcentral=postcentral gyrus; rACC=rostral anterior cingulate cortex; SFG=superior frontal gyrus; STG=superior temporal gyrus.

Perceived lack of control over independent stressors was positively associated with cortical thickness for one cluster that passed correction for multiple comparisons (right lateral orbitofrontal cortex; Figure 2). There were no clusters where perceived lack of control over independent stressors was associated with surface area that passed correction for multiple comparisons, with or without controlling for whole-brain average area (Figure S1).

Gender moderation.

There were no clusters for which gender moderated the relation between perceived lack of control over stressors and cortical thickness or area that passed correction for multiple comparisons (Table S3, Figures S2 & S3).

Age moderation.

There were two clusters for which age moderated the relation between perceived lack of control over dependent stressors and cortical thickness that passed correction for multiple comparisons: one in the left precuneus, and one in the right supramarginal gyrus (Table S4, Figure S4 & S5). These clusters do not overlap with clusters for which relations between perceived lack of control over dependent or independent stressors and cortical thickness were observed.

Discussion

The current study tested whether perceived lack of control over dependent and independent life stressors is associated with brain structure in specific ROIs (amygdala, hippocampus, mPFC) and across the cortex in a sample of adolescents and emerging adults. Contrary to our hypotheses, adolescents’ and emerging adults’ perceptions of stressor controllability were not associated with ROI neuroanatomy across the whole sample. However, less perceived control over dependent life stressors was associated with greater amygdala GMV in female youth and greater mPFC thickness in male youth. There were no significant gender differences in the hippocampus, but for girls, greater left and right hippocampus GMV were marginally associated with greater perceived lack of control over dependent stressors (ps<.09). We thus cannot rule out relations between perceived lack of control over life stressors and hippocampal GMV, which could be explored in larger samples, especially given the hippocampus’ role in stress regulation and its dense connections with the amygdala (Tottenham & Sheridan, 2010). Additionally, whole-brain exploratory analyses revealed a pattern of greater cortical thickness in numerous regions involved in cognitive and emotional regulation with less perceived control over dependent stressors.

The mechanisms that underlie the observed associations between perceived lack of control over dependent life stressors and GM morphometry in youth are not clear, but there are some potential explanations that should be explored further in future work. All regions in which GM morphometry was associated with perceived lack of control over life stressors, either across the whole sample (frontoparietal and temporal regions), or for male (mPFC) or female (amygdala) youth specifically, are involved in processing of stressful experiences. The amygdala is involved in detection of threats (McEwen & Gianaros, 2010). The mPFC regulates neural reactivity to stressful situations depending on situational factors such as stressor controllability (Maier & Seligman, 2016). The frontoparietal and temporal regions identified in the whole-brain exploratory analyses are active during cognitive reappraisal, an emotion regulation strategy that can downregulate one’s negative emotional response to stressful events (Ochsner et al., 2012; Banich et al., 2009). Thus, these regions may all be associated with cognitive and neural processes that help shape an individual’s perceived control over stressors, and/or their activity may be affected by appraisals of control when faced with stressors.

Further, GM structure of these regions is developing across this adolescent and emerging adult period, which may contribute to bidirectional relations between GM and perceived control over stressor. Frontoparietal and mPFC gray matter is thinning (e.g., Tau & Peterson, 2010), which is associated with increased efficiency of processing during this developmental period. Amygdala GMV is continuing to grow through late adolescence (Wierenga et al., 2014a). These GM structural changes may be associated with changes in cognitive processing (e.g., Ahmed et al., 2015; Larsen & Luna, 2018) that could affect perceptions of control over stressors. As an example, as GM in regions involved in cognitive aspects of emotion regulation thins, emotion regulation ability improves (Ahmed et al., 2015). This improvement may contribute to making stressors feel more manageable and controllable. Likewise, developmental GM changes in the mPFC and amygdala in male and female youth, respectively, may affect cognitive processes that contribute to perceived control over stressors. On the other hand, GM development during adolescence and emerging adulthood may allow environmental input such as stressful life events to have a greater impact on GM (Andersen & Teicher, 2008; Fuhrmann, Knoll, & Blakemore, 2015). If appraisals affect neural processing during these stressful events, appraisals could affect GM development and form relations with GM structure. In sum, GM structure during development could be associated with cognitive processes that contribute to shaping individual’s perceptions of control over their stressors, and/or appraising stressors as more or less controllable could affect neural processing of life stress, contributing to sculping of GM during development.

The gender differences found in the current study are interesting given sex differences seen in the processing of stressors in rodents and humans. Perceived lack of control over life stress was associated with thicker GM in the mPFC for male youth. Rodent models exhibit a sex difference in mPFC engagement when faced with controllable stress, such that male but not female rats initiate mPFC inhibition of widespread serotonergic reactivity to a stressor when control is detected (Baratta et al., 2018; 2019). It is possible that similar gender differences exist in humans, but this has not been tested. Future work should thus further investigate gender differences in mPFC activity associated with appraisals of control.

The relation between amygdala GMV and greater perceived lack of control over life stressors was specific to female youth. This result is paralleled by work demonstrating gender differences in patterns of acute amygdala reactivity. Women demonstrate greater (Stevens & Hammen, 2012) and more prolonged (Andreano et al., 2014) amygdala reactivity in response to negative stimuli, while men show greater amygdala reactivity to positive emotional stimuli (Stevens & Hammen, 2012). Perceived lack of control over a stressor may increase its subjective severity (Fassett-Carman et al., 2019; 2020) and thus may make it feel more negative or aversive, potentially leading to greater amygdala reactivity in female youth. However, whether or not there are gender differences in amygdala reactivity as a function of perceived control, or in how amygdala GMV relates to perceived severity and control of life stressors, remain unknown.

Whole brain exploratory analyses revealed relations between perceived lack of control over life stressors and cortical thickness, but not surface area, that passed correction for multiple comparison. As cortical thickness is being shaped largely by synaptic pruning during adolescence and emerging adulthood, it is more sensitive to experience-dependent changes, such as those caused by stressful events, than cortical surface area (Tau & Peterson, 2010). Most clusters that passed correction were associated with perceived lack of control over dependent stressors, with only one cluster associated with independent stressor controllability. These results are consistent with past work highlighting the importance of perceived lack of control over dependent stressors specifically in regard to psychopathology in adolescents and emerging adults (Fassett-Carman et al., 2019). In the current study, perceived lack of control over dependent stressors was associated with greater cortical thickness in a network of regions involved in cognitive control and cognitive aspects of emotion regulation, including the dlPFC, posterior superior frontal gyrus, rACC, superior temporal sulcus, inferior parietal lobule, and middle temporal gyrus.

We do not know the mechanisms underlying the relation between GM thickness in these frontoparietal and temporal regions and perceived lack of control over dependent stressors. Yet, it is possible that perceived lack of control could relate to the emotional and/or non-emotional aspects of cognitive control subserved by this network of regions. First, it is possible that these brain regions are engaged to regulate emotions and inhibit stress reactivity when control is detected, or aid in making stressors feel more controllable. Indeed, during cognitive reappraisal, the most cognitively demanding and highly studied emotion regulation strategy, frontoparietal regions involved in working memory, selective attention, and inhibitory processes are all active (Ochsner et al., 2012; Banich et al., 2009), and superior and middle temporal regions are potentially involved in the reworking of meaning or interpretation of stimuli (Ochsner et al., 2012). As this network is both important for downregulating negative emotional responses to situations, and thinking through how the stressor can be changed or managed (McRae et al., 2012), it could have bidirectional relations with perceived control which should be investigated in the future.

Further, perceived lack of control over dependent stressors may also relate to gray matter structure of cognitive control regions via a non-emotional mechanism. For example, better cognitive control could increase an individual’s actual control over life stressors, thus increasing their perceived control. Indeed, there is evidence that poor cognitive control increases generation of dependent stressors (Snyder & Hankin, 2016, Snyder et al., 2019). Once these stressors are generated, poor cognitive control may further prevent individuals from controlling these negative situations. Each of these possible mechanisms through which perceived lack of control over dependent stressors relates to GM thickness in cognitive control regions may play a role, and should be further investigated.

The relations between cortical thickness and perceived lack of control over independent stressors were in similar regions to those of perceived lack of control over dependent stressors, although the clusters associated with perceived lack of control over independent stressors did not pass correction for multiple comparisons (with the exception of one cluster in the lateral orbitofrontal cortex also associated with dependent stress) and were mostly smaller than those identified in relation to perceived lack of control over dependent stressors. This pattern suggests that effects for perceived lack of control over independent stressors may be similar but less strong than those for dependent stressors. Dependent stressors may cause greater stress reactivity due to the understanding that one contributed to causing these negative events. This knowledge may lead to negative cognitive responses to the stressors such as self-blame and rumination, which can increase and prolong biological stress reactivity (LeMoult, 2020). Appraisals that modulate reactivity to these dependent stressors could therefore have a greater impact on neural development than for independent stressors. Another possible reason for the stronger relation between dependent stressors and GM could relate to their relative frequency and salience during adolescence and emerging adulthood. Dependent stressors include peer stressors, such as fights with a friend or break-ups, and academic stressors, such as getting bad grades. As adolescence and emerging adulthood is a time of increased complexity and salience of peer relations and greater achievement-related pressures, these stressors and their appraisals may have a greater impact than independent stressors on brain development during this period (e.g., Fuhrman et al., 2015), as they do on psychopathology (Hankin et al., 2016).

Limitations and Future Directions

This study has limitations that are important areas for future research. First, the cross-sectional nature of the data prevents conclusions about directionality of effects. Future work should investigate longitudinal links between perceived control over dependent stressors and cortical thickness. These longitudinal relations may be especially important to investigate in adolescence and emerging adults, as this developmental period may be crucial for shaping controllability appraisals. During this developmental period, autonomy is increasing, creating more of an opportunity to exert control over life events. Additionally, cognitive control is improving, mediated by development of cognitive control networks (Larsen & Luna, 2018), increasing ability to exert control over life stressors. Thus, this period may be a time when adolescents and emerging adults are honing their controllability appraisals, creating bidirectional relations between gray matter development and perceived controllability: GM development could both contribute to shaping perceived control, and be shaped by it.

Another limitation of this study is that despite its relatively large sample size, it is not big enough to investigate three-way interactions between age, gender, and perceived lack of control in relation to gray matter. Evidence suggests that gray matter trajectories are sexually dimorphic, with female youth reaching peak cortical GM and subsequent decline before male youth (Lenroot et al., 2007; Giedd et al., 2015; Mills et al., 2016). Investigating these interactions would allow a better understanding of how perceived lack of control and GM relate across adolescence and early adulthood. An important future direction will thus be to investigate these relations across development with a bigger sample.

Further, because these analyses only investigated neuroanatomy, we cannot make conclusions regarding how perceived lack of control over life stressors relates to neural activity and functional connectivity. While there is detailed rodent research in this area, there is much less work exploring how perceived control over stress affects brain activity in humans. Future work should investigate whether perceived lack of control over life stressors affects resting state functional connectivity of brain networks involved in cognitive aspects of emotion regulation as well as brain activity in these areas in response to an acute stressor.

Finally, these results suggest broader future directions for stress research. We demonstrated that appraisals of controllability have specific structural neural correlates in regions involved in cognitive emotion regulation, distinct from areas where GM morphology is typically thought to correlate with life stress (e.g., hippocampus, amygdala, mPFC). These findings build on psychopathology research demonstrating relations between controllability appraisals and depression over and above stress frequency (Fassett-Carman et al., 2019; 2020) and emphasize the importance of stressor appraisals in fully understanding the neural correlates of life stress. Most research on the effects of life stress on brain structure compares “stressed” versus “non-stressed groups” or measures stress in a way that combines stress frequency and appraisals, along with other factors such as coping ability, anger, and nervousness. While these methods have been valuable for demonstrating that stress does relate to GM morphology, our results highlight the importance of disentangling the neural correlates of distinct aspects of the stress experience moving forward.

Our results further demonstrate that the dependence of stressors may be important in influencing their neural correlates, as GM correlates were stronger for dependent than independent stressors. Future work aimed at understanding the neural effects of stress should thus investigate stressor dependence. One method to accomplish this is to distinguish between these stressor types, as we did in the current study. However, because individual’s perceptions of how much they caused negative events may be driving these effects, potentially via relations with negative cognitions such as low self-esteem, worthlessness, and guilt, measuring appraisals of stressor dependence may further elucidate underlying mechanisms driving the relation between stressor dependence and GM morphometry.

Conclusion

In sum, perceived lack of control over dependent stressors was associated with greater amygdala GMV in female youth, and greater mPFC thickness in male youth across adolescence and early adulthood. Exploratory whole brain analyses additionally revealed a network of cognitive emotion regulation regions where greater thickness was associated with perceived lack of control over dependent stressors. These results provide insight into relations between stress and the developing brain, which is crucial to understand as adolescence and emerging adulthood is a time of heightened onset for stress-related disorders such as depression. Future work should investigate these effects longitudinally to understand the directionality of effects, and functionally to determine how perceived control affects emotion regulation network activity. These results also suggest that future research should investigate distinct aspects of the stress experience (e.g., stress frequency, appraisals, and dependence) to better understand the mechanisms driving relations between life stress on GM morphometry.