Witnessing their mother's acute and prolonged stress affects executive functioning in children

The study investigates whether and how witnessing their mothers’ stress affects children’s stress resilience and executive functioning. The authors conceptualize resilience as the measurable adaptive outcome when facing stress and consider stress from environmental, psychological, and biological perspectives. Prior work shows that children can empathically resonate with caregivers’ stress, potentially transmitting stress within families. The research question asks whether children’s acute psychophysiological stress response (subjective stress, heart rate, cortisol), elicited by observing their mothers under stress, impairs cognitive flexibility and working memory, and whether maternal stress (acute and prolonged) predicts children’s cognitive performance beyond children’s own stress responses. Hypotheses: H1) Children’s acute stress response and resonance with maternal stress (interactions of mother–child acute stress markers) predict poorer performance (slower reaction times and reduced accuracy). H2) Prolonged everyday stress (past month for mothers; past week for children) modulates the effect of the acute condition on children’s performance. H3) Children’s trait empathy (cognitive empathy) and relationship closeness to the mother moderate the association between maternal and child prolonged stress.

The authors summarize extensive evidence that stress in childhood can impair physical, emotional, and cognitive development, and that close relationships can transmit stress within families. Empathic stress has been demonstrated among adults, with physiological correlates (cortisol, heart rate, HF-HRV) and is influenced by empathic ability and relational closeness. Naturalistic mother–child studies have shown cortisol covariation across the day, particularly during negative affect and with more time together, suggesting immediate stress transmission pathways. Experimental work with toddlers shows maternal sympathetic activation can be transferred to children, especially with touch. In a prior study using the same sample, children (8–12 years) who observed their mothers in a TSST showed empathic stress responses (higher cortisol likelihood, subjective stress, state empathy, HF-HRV changes contingent on trait cognitive empathy). Literature also links both acute and chronic stress to impairments in executive functions (cognitive flexibility and working memory), with mechanisms in adults including sympathetic arousal and cognitive interference, and stress-sensitive PFC alterations under prolonged psychosocial stress. However, prior chronic stress studies often involve severe adversity (e.g., maltreatment, poverty), leaving a gap regarding everyday stressors’ impact on children’s executive functioning.

Design: Experimental, between-groups design with mother–child dyads. Mothers were assigned to an acute psychosocial stress condition (TSST) or a stress-free control task; children observed their mothers through a one-way mirror and then completed executive function tasks. Repeated measures of subjective stress, heart rate, and salivary cortisol were collected from both mothers and children at eight time points from −20 to +70 minutes relative to task onset. Prolonged stress perceptions were assessed via questionnaires at the end of the session. Cognitive tasks followed immediately after mothers’ tasks. Participants: 76 healthy mother–child dyads (children 8–12 years: M=9.95, SD=1.42; 37 girls, 39 boys; mothers 31–45 years: M=40.3, SD=3.33) recruited in Leipzig, Germany (2018–2020). Inclusion/exclusion criteria aimed to minimize hormonal confounds and health factors influencing stress responses (e.g., excluded pregnancy, daily smoking, BMI <18.5 or >30, psychotropic interference, prior TSST, non-fluency, dyslexia; children past menarche excluded; mothers post-menopause excluded). Dyads were pseudo-randomized to TSST (n=39) or control (n=37), balanced by age and sex. Ethics approval (Leipzig University 084/18-ek); informed consent obtained. Procedure: Sessions (~2.5 h) scheduled 3–7 pm to control for diurnal cortisol. After arrival and brief rest, mothers consented and provided a drug test; children were familiarized with tasks. Children then observed their mothers performing the assigned task via one-way mirror (mothers knew children were observing). After maternal task, children performed two computerized tasks (order randomized): category-switching (cognitive flexibility) and n-back (working memory). Questionnaires completed before reunion. Stress induction: TSST (5-min anticipation; 5-min speech; 5-min mental arithmetic) in front of a non-responsive panel; audio/video recorded. Control: 5-min silent reading (anticipation) and 10-min reading aloud from a nature book; mothers performed alone. Mothers in both conditions were aware of child observation. Cognitive tasks: Implemented in Presentation (2018). Category-switching: three blocks × 64 trials; categorize words by alternating rules (animate/inanimate; bigger/smaller than a soccer ball); CTI short (200 ms) or long (1000 ms); 3000 ms response window; analyzed error rate and RT; first trial per block and post-error trials excluded; personalized Tukey far-out RT outliers (>3 SD from individual mean) removed. N-back: letters presented 500 ms; 1-back block (65 trials), 2-back block (65 trials); 1500 ms response window; RT computed on correct trials with outliers removed; accuracy as d′ (hits vs false alarms). Acute stress measures: Subjective stress (single item, 1–7 Likert). Heart rate (bpm) via OMRON RS2 wrist monitor, concurrent with subjective stress and cortisol timepoints. Salivary cortisol via Salivettes; time-resolved fluorescence immunoassay; intra/inter-assay <10% and 12%. Preprocessing: ln-transform HR and cortisol; winsorize cortisol at 3 SD; compute AUC with respect to increase (AUCi) per Pruessner et al. across eight timepoints for cortisol, HR, and subjective stress. Questionnaires (prolonged stress and moderators): Mothers: Perceived Stress Scale (PSS, 10 items, 0–40; α=0.76) for past month. Children: PSS-C (13 items, past week, 0–39; α=0.5). Trait empathy: EmQue-CA (14 items; total α=0.67; cognitive empathy α=0.76); only cognitive empathy used in moderation H3a. Relationship closeness: Inclusion of Other in the Self Scale (IOS; 1–7). Data quality and exclusions: Children lacking task understanding excluded (category-switching: n=1; n-back: n=2). For stress markers, dyads with ≥2 consecutive missing timepoints per marker excluded from respective analyses; single missing points imputed as mean of adjacent values. Sporadic missingness: subjective stress (2 individuals missing 1 timepoint), HR (1 child missing 1 timepoint), cortisol (5 individuals missing 1 timepoint; 1 mother missing 2 consecutive timepoints). Continuous ECG recorded but not analyzed due to missingness. Statistical analyses: Conducted in R 4.2.0. Repeated-measures ANOVAs tested group (TSST vs control) effects on cognitive outcomes and on prolonged stress reports (PSS/PSS-C). Primary analyses used linear mixed models (lme4; p-values via lmerTest) with random intercepts for subjects. Two models per hypothesis and task (accuracy or RT as DV). Fixed effects in all models: child age, sex, task factors (category-switch: CTI, Switch vs Repeat; n-back: block), and condition. Predictors of interest: child AUCi markers; mother–child AUCi interactions (H1); prolonged stress three-way interactions PSS×PSS-C×condition (H2). Numeric predictors z-standardized; VIF ≤5. Bootstrap (1,000 resamples) for parameter estimates. Quadratic terms (squared AUCi or squared PSS) added in exploratory second step to test nonlinearity. Moderation H3: multiple regression predicting child PSS-C from maternal PSS with interactions with child cognitive empathy (H3a) or IOS closeness (H3b), controlling age and sex. Alpha=0.05, two-sided; multiplicity correction not applied per García-Pérez due to non-omnibus hypotheses and potential speed–accuracy tradeoffs.

Manipulation check (from prior report): More cortisol responders in stress vs control (mothers: 28 vs 6; children: 7 vs 1), higher subjective stress in both mothers and children, and higher maternal HR reactivity in the stress group, indicating successful stress induction. Group differences in cognition: No significant condition effects on children’s RT or accuracy in category-switching or n-back tasks (e.g., category-switching RT F(1,73)=0.28, p=0.599; accuracy F(1,73)=1.37, p=0.245; n-back RT F(1,72)=0.03, p=0.853; d′ F(1,70)=0.33, p=0.565). Acute stress effects:

• N-back RT: Maternal subjective stress (AUCi) predicted faster RT (t(67)=-2.41, p=0.017, β=-53.64, 95% CI [-100.15, -11.48]). No other child or maternal acute markers predicted n-back RT or accuracy.
• Category-switching accuracy (error rate): Linear models showed higher child error rates with stronger maternal cortisol reactivity (Mother AUCi cortisol: β=0.04, t=3.35, p=0.001). No significant child–mother AUCi interactions (no evidence for acute empathic resonance affecting accuracy).
• Quadratic models revealed U-shaped relationships: • Child subjective stress AUCi^2 predicted error rate (β=0.02, t=3.58, p<0.001), with optimal accuracy around average child subjective stress; • Child cortisol AUCi positively related to errors (β=0.03, t=2.25, p=0.025); • Maternal stress markers (subjective stress, HR, cortisol) each showed significant quadratic effects on child error rates: Mother AUCi subjective stress^2 β=0.02, t=3.31, p=0.001; Mother AUCi HR^2 β=0.02, t=2.13, p=0.034; Mother AUCi cortisol^2 β=0.02, t=2.55, p=0.011. Across markers, children’s optimal accuracy occurred near slightly below-average maternal stress reactivity (≈-1.0 to -0.5 SD), with higher errors at very low and high maternal reactivity. Prolonged stress effects:
• Category-switching RT: No significant associations with PSS or PSS-C (linear or quadratic).
• Category-switching accuracy: Maternal PSS exhibited a quadratic effect predicting error rate (PSS^2 β=0.03, t=2.74, p=0.007), with optimal performance around ~0.25 SD above average maternal PSS. Interaction between condition and maternal PSS indicated that lower maternal past-month stress related to better accuracy in control but worse accuracy under TSST; this pattern reversed at higher maternal PSS (linear model Condition×PSS β=-0.05, t=-2.00, p=0.046; quadratic model β=-0.06, t=-2.50, p=0.013). No significant effects of child PSS-C or interactions involving PSS-C.
• N-back outcomes: No significant effects of prolonged stress on RT or accuracy. Relationship between maternal and child prolonged stress and moderators: Multiple regression models showed no significant prediction of child PSS-C from maternal PSS and no moderation by child cognitive empathy or dyad closeness (R^2≈0.06–0.085; p≥0.29). H3 rejected. Overall: Maternal acute and prolonged stress were stronger and more consistent predictors of children’s cognitive flexibility accuracy than children’s own stress responses. Working memory effects were limited to faster RT with higher maternal subjective stress. No evidence supported acute empathic resonance (mother×child AUCi interactions) or prolonged stress covariance as mechanisms in this dataset.

The findings indicate that children’s executive functioning, particularly cognitive flexibility accuracy, is sensitive to mothers’ stress experiences. While group-level performance did not differ between TSST and control, individual-level analyses showed that maternal acute stress reactivity (subjective stress, heart rate, cortisol) and maternal perceived stress over the past month predicted children’s error rates in a nonlinear manner: very low and very high maternal stress related to more errors, with optimal performance at moderate levels. Children’s own acute stress linked to cognitive flexibility only sporadically and mostly in quadratic patterns, suggesting a bell-shaped relation consistent with notions akin to Yerkes–Dodson for mild stress optimizing performance. Working memory showed limited effects, with faster RT when maternal subjective stress was higher. The hypothesized empathic resonance mechanism (mother–child acute stress interactions) did not predict cognitive outcomes, nor did prolonged stress covariance between mothers and children, suggesting that maternal stress may affect children’s cognition through pathways other than direct synchrony at the time scale tested. Potential mechanisms include general family stress climates driven by maternal stress reactivity/appraisals, alterations in parenting behavior, or children’s adaptive processes. Notably, the interaction between maternal past-month stress and acute condition suggests possible adaptation: when mothers reported higher prolonged stress, children’s cognitive flexibility under acute maternal stress was less impaired or even relatively better than in control, potentially reflecting coping or emotional detachment strategies in higher-stress environments. These results highlight that everyday maternal stress, even at moderate levels, can shape children’s cognitive outcomes in daily life and that the family environment remains a critical context for child resilience and vulnerability.

This study links mothers’ acute and past-month stress to their children’s executive functioning, showing that maternal stress robustly predicts children’s cognitive flexibility accuracy, often in a nonlinear fashion, with impairments at both very low and very high maternal stress. Children’s own acute stress responses played a smaller role, and no evidence supported empathic resonance or prolonged stress covariance as explanatory mechanisms for cognitive effects. These findings underscore that close relationships can both buffer and transmit stress, with implications for everyday child outcomes. Future research should: (1) examine diverse populations and wider stress ranges, including different caregiver dyads (e.g., fathers); (2) incorporate more reliable and multimethod measures of child prolonged stress (e.g., physiological indices such as hair cortisol); (3) optimize paradigms to maximize ecological validity and potential for empathic transmission (e.g., allow proximity/touch where appropriate); and (4) delineate mediating pathways (parenting behaviors, appraisals, attentional/motivational mechanisms) linking parental stress to child cognition to inform targeted interventions that support parents in promoting positive child outcomes.

• Sample generalizability: Healthy, German, predominantly highly educated dyads; overall low prolonged stress levels limit generalization and may constrain detection of effects across the full stress spectrum.
• Child prolonged stress measure: PSS-C showed poor internal consistency (α=0.5), likely compromising validity; few children reported high stress, limiting tests of interactions and associations with cognition.
• Procedural factors: Mother–child separation may have reduced empathic stress contagion; laboratory context and cognitive tasks themselves may have induced stress, potentially confounding effects.
• Power and design: Exploratory analyses and modest sample size may limit detection of smaller interaction effects (e.g., empathic resonance). Continuous ECG was collected but not analyzed due to missingness, reducing autonomic specificity.
• External stressors and mechanisms: The study did not directly assess broader family stressors (e.g., socioeconomic adversity, discrimination) or parenting behaviors that may mediate effects of maternal stress on child cognition.