The effects of rumination and reappraisal on repeated exposure to stress

Abstract

Adequate recovery and habituation to acute stressors in daily life are essential for mental health. One potential moderator might be the use of different emotion regulation (ER) strategies contributing to interindividual differences in vulnerability to chronic stress. Rumination has been linked to impaired endocrine adaptation, whereas reappraisal tendencies have been associated with boosted habituation to psychosocial stress. Yet, no experimental study has directly compared the causal effects of these strategies on psychoneuroendocrine responses to repeated stress. To address this gap, 91 healthy participants (47 women) underwent a short Trier Social Stress Test (TSST) twice on two consecutive days and were randomly assigned to a rumination, reappraisal, or control intervention in between. Cognitive-affective ratings indexed psychological stress, while salivary cortisol, alpha-amylase (sAA), and heart rate (HR) served as biomarkers. Across the entire sample, reduced increases in negative affect and cortisol in response to the second TSST confirmed successful habituation. As expected, rumination immediately increased negative affect, reduced positive affect, and lowered perceived coping abilities indicating successful induction of ruminative thinking. Moreover, it prevented physiological habituation to repeated stress, evidenced by stable cortisol and HR responses. Unexpectedly, reappraisal also lowered perceived coping abilities in men, followed by prolonged sAA reactivity to the first stress exposure, hinting at a sex-specific impairing effect of reappraisal on noradrenergic recovery. However, reappraisal neither affected cortisol recovery nor habituation, which may result from generally poor reappraisal performance. Together, these findings provide initial evidence for differential effects of rumination and reappraisal on psychophysiological adaptations to repeated TSST exposure.

Highlights

• •Rumination and reappraisal both hindered cognitive-affective stress recovery.
• •Rumination prevented physiological habituation to repeated stress.
• •Participants showed poor reappraisal performance immediately after stress.
• •Reappraisal neither accelerated stress recovery nor enhanced habituation.
• •Immediately after stress, reappraisal prolonged noradrenergic recovery in men.

1. Introduction

Stress is an inevitable part of everyday life, and the way in which the body and mind respond to and recover from it is central to long-term health. Acute stress responses are generally adaptive, mobilizing biological and psychological resources to meet situational demands [1]. However, insufficient recovery or a lack of habituation to repeated stress can promote chronic stress [2], representing one major risk factor for the development and maintenance of mental and somatic disorders, including depression, anxiety, and cardiovascular disease [3]. Elucidating the mechanisms that enable some individuals to adapt and gain resilience to stress, while others remain vulnerable to its harmful consequences, is therefore of considerable scientific and clinical relevance. Physiological responses to stress are primarily mediated by two systems: the sympathetic-adrenal-medullary (SAM) system and the hypothalamic-pituitary-adrenocortical (HPA) axis. The SAM system is characterized by the rapid release of catecholamines, primarily adrenaline and noradrenaline, from the adrenal medulla. These hormones elicit widespread physiological adaptations, including increases in heart rate, blood pressure, and glucose availability, thereby preparing the organism for immediate fight-or-flight responses [4]. In contrast, activation of the HPA axis results in a somewhat delayed secretion of cortisol from the adrenal cortex, which, e.g., mobilizes energy and suppresses inflammatory processes [5]. These two systems interact and work together to enable an optimal adaptation to the stressor [6]. Yet, prolonged activation or inadequate recovery can increase allostatic load, thereby posing long-term health risks [3]. A hallmark of adaptive stress regulation is habituation, which refers to the attenuation of stress responses upon repeated exposure to the same or similar stressors. Typically, individuals show a marked reduction in cortisol reactivity when being repeatedly stressed via the Trier Social Stress Test (TSST; [7]), while evidence for habituation of the SAM system is less consistent [8]. This decline in cortisol responses is thought to result from reduced novelty and unpredictability, which are key drivers of HPA axis activation [9]. In this way, habituation conserves organic resources and protects against chronic overactivation of the stress systems [10]. However, it is not a universal phenomenon: some individuals fail to recover, habituate, or even show a sensitization, maintaining heightened physiological reactivity (for a review, see Ref. [8]). This raises the question of which factors drive variability in stress recovery and habituation. According to the transactional stress theory [11], appraisal processes - evaluating threat and coping abilities - shape physiological stress reactions and can be modified by cognitive regulatory interventions [12]. As a definition, emotion regulation (ER) encompasses all cognitive and behavioral processes by which individuals influence the type, intensity, duration, or expression of an emotion [13]. People differ in their ER competencies and in their preference for specific strategies [14,15]. Given their clinical relevance, two cognitive strategies have received particular attention: reappraisal and rumination. Reappraisal involves reinterpreting events to alter the valence of their emotional meaning and is widely regarded as one of the most effective and beneficial long-term strategies for managing (low to moderate) negative emotions [[16], [17], [18]]. In contrast, rumination involves a repetitive and passive focus on distressing thoughts and feelings and has frequently been associated with heightened negative mood and various psychopathologies [19]. Previous research provides initial evidence for distinct associations of reappraisal and rumination with physiological stress reactivity, recovery and habituation. Specifically, trait rumination has consistently been linked to heightened cortisol responses [20] and delayed cortisol recovery from acute stress [21,22]. In addition, the amount of spontaneous ruminative thoughts following the TSST predicted increased cortisol responses to re-exposure, suggesting that rumination blocks habituation processes [23]. Moreover, rumination tendencies have repeatedly been linked to lower resting heart rate variability [24,25] and attenuated vagally mediated cardiac recovery following acute stress [for a review, see [26]]. Existing findings on reappraisal effects on physiological stress reactivity are quite heterogeneous (for meta-analyses, see Ref. [27]). For instance, reappraisal decreased HRV and increased cortisol levels during stress, but also accelerated HRV recovery following TSST exposure [28]. This reappraisal-induced cardiac-vagal-flexibility was particularly evident among habitual reappraisers. In contrast, trait reappraisal has been linked to enhanced cortisol habituation to the TSST [29], whereas no association to cardiovascular habituation has been found [30]. Consistently, Finke and colleagues [31] demonstrated that positive post hoc framing of performances in the isometric handgrip task (a psychosocial stressor) facilitated cortisol recovery and reduced baseline cortisol levels prior to a second stress exposure. Taken together, previous research suggests that these two strategies have different or even opposing effects on stress recovery and habituation, particularly regarding cortisol. However, most existing studies have either assessed habitual ER tendencies via self-report or captured automatic regulatory thoughts (i.e., rumination) in correlational designs, rather than experimentally comparing the impact of different ER interventions on stress. To address this gap, the present pre-registered study investigated whether rumination and reappraisal influence recovery and habituation of psychophysiological responses to repeated psychosocial stress relative to an active non-regulatory control condition. Therefore, 91 healthy participants (47 women) were repeatedly stressed via a short version of the TSST on two consecutive days and randomly assigned to one of three different ER conditions (rumination, reappraisal, control). Salivary cortisol indexed HPA axis activity, whereas salivary alpha-amylase (sAA; [32]) and heart rate (HR) served as markers of the SAM axis. Additionally, repeated assessments of participants’ cognitive-affective state provide information about emotional and cognitive experiences following the ER and stress interventions. As pre-registered, we hypothesized that rumination prolongs physiological stress recovery and prevents habituation to repeated stress, particularly with respect to cortisol [21,23]. In contrast, we expected reappraisal to accelerate cortisol recovery and boost habituation [29,31]. Given findings showing that stress improves ER performance in men but not in women [33], we predicted that the effects of ER (especially reappraisal) on stress recovery and habituation are particularly pronounced in men.

2. Materials and methods

2.1. Participants

To determine the required sample size, we conducted an a priori power analysis using G∗Power 3.1 [34]. Given that ER × Sex interactions on stress habituation have not previously been reported, no empirical estimates were available for the expected Day × Condition × Sex interaction effect. We therefore based our assumptions on prior research showing small-to-medium effects of cognitive interventions on stress reactivity and recovery in men and women [28,31] as well as large associations between spontaneous ruminative thoughts and cortisol reactivity to repeated psychosocial stress in men [23]. Based on this evidence, we conservatively assumed a medium-sized interaction effect (d = .5) for the combined effects of ER condition, Day, and Sex. With an assumed correlation of r = .5 for repeated measurements, an alpha-level set to .05, and a power of 1-β ≥ .95, analyses revealed a total sample size of 90 participants required to detect a significant interaction between the within-subject factor Day (first vs. second), and the between-subjects factors Condition (reappraisal vs. rumination vs. control) and Sex (male vs. female). Thus, 91 healthy participants (47 women and 44 men) aged between 18 and 35 years (M = 23.91, SD = 3.81) and a mean Body Mass Index (BMI) between 18 and 30 (M = 23.14 kg/m²; SD = 2.82 kg/m²) were recruited via online advertisements, mailing lists and notice boards throughout the Ruhr University Bochum. Volunteers were excluded from participation if they reported any chronic or acute illnesses, history or current psychological treatment, hormonal contraception, irregular menstrual cycle (within the past six month), medication or drug use, including smoking, or previous experiences with the current stress protocol. All naturally cycling women were tested outside of their period, with scheduling based on self-reported menstrual cycle. A similar number of male and female participants was randomly assigned to three ER conditions (reappraisal, rumination, control), which did not differ in BMI, age, habitual use of reappraisal and rumination as assessed via the Cognitive Emotion Regulation Questionnaire (CERQ; [35]) or flexibility in the use of different ER strategies in daily life (FlexER; all ps > .05). In addition, a chi-squared test confirmed that ER groups did not significantly differ in the distribution of self-reported menstrual cycle phase at the time of testing (χ²(2) = 1.06, p = .589, Cramer V = .152). The present study was pre-registered at the Open Science Framework (OSF; https://doi.org/10.17605/OSF.IO/NGK4M). The experimental procedure was conducted in accordance with the Declaration of Helsinki and approved by the ethics committee (No. 801) of the faculty of psychology at Ruhr University Bochum.

2.2. Experimental procedure

All participants were instructed to refrain from sports, drugs, and alcohol 24h prior to both experimental testing as well as food and any drinks except for water 2 h before. To control for diurnal fluctuations in endogenous cortisol secretion [5], all testing took place between 12:30 p.m. and 6:30 p.m., with the same testing time on each of the two testing days. Importantly, regression analyses showed no significant associations between time of day and baseline cortisol levels (day 1: β = .066, p = .536; day 2: β = .132, p = .211) or cortisol reactivity to stress (Δ cortisol; day 1: β = .129, p = .223; day 2: β = .089, p = .402). In addition, ER conditions did not significantly differ in the time of day for both testing sessions, and no interactions with sex occurred (no main effect of condition: p = .610; no condition × sex interaction: p = .266). The experimental procedure comprised two testing days. On the first day, participants completed several questionnaires and were then exposed to the Trier Social Stress Test before working on one of three questionnaires for a fixed period of 15 min, designed to induce either rumination, reappraisal, or to serve as a neutral control task (ER intervention). After a 45-min waiting period with repeated stress assessments, the first testing day concluded. On the second day, participants again completed the same ER questionnaire as on the previous day (reinstatement of the ER process) and were subsequently exposed to the same stressor once again. The second testing day also concluded after a 45-min waiting period. At the end of day two, participants were debriefed and compensated with either € 35 or course credits. At several time points across the two testing days, saliva samples, cognitive-affective ratings (in detail described below), and heart rate (HR) measures were collected (for exact sampling times, see Fig. 1).

Fig. 1.

Study procedure. Participants provided five saliva samples per testing day, along with affective state ratings (Differential Affect Scale; DAS) at multiple time points across the experiment (baseline, t+15, t+30, t+45, t+60; marked with dark blue boxes). In addition, participants repeatedly completed appraisal ratings (Primary and Secondary Appraisal Scale; PASA) and a visual analogue scale (VAS). They underwent the Trier Social Stress Test (TSST) twice, once on the first day and again on the following day. After the first stress exposure, participants completed one of three ER questionnaires (rumination, reappraisal, control) which they filled in again at the beginning of testing day two.

2.3. Repeated stress exposure

All participants completed a short version of the Trier Social Stress Test (TSST; [7]) to ensure a 15-min interval between salivary samples on two consecutive days at the same time of day. This short version of the TSST, was repeatedly shown to elicit significant increases in salivary cortisol, sAA, and negative affect, verifying successful stress induction (e.g., Ref. [33]). Except for timing deviations, the TSST followed the same standardized principles as described in the original format. On each day, the TSST procedure began with a 2-min preparation period, followed by a 5-min free speech in which participants described personal characteristics qualifying them for a desired job. Subsequently, participants performed a 3-min arithmetic task (i.e., counting backwards from 2043 in steps of 17 on day one and from 1022 in steps of 13 on day two). During the task, participants were videotaped and observed themselves on a television screen in front of them, while being monitored and evaluated by a panel consisting of one woman and one man wearing white lab coats. Both panel members acted neutrally and reservedly using standardized phrases (as described in the original TSST protocol) to interact with the participant. To prevent familiarity effects, the panel members were replaced on the second testing day.

2.4. Emotion regulation conditions

Each participant was randomly assigned to either a rumination, reappraisal, or control condition with an almost equal number of men (rumination: 14, reappraisal: 13, control: 17) and women (rumination: 15, reappraisal: 18, control: 14) in each group. Analyses verified that the distribution of sex did not differ significantly across the three groups (χ²(2, N = 91) = 1.03, p = .596). Immediately after the first TSST and before the second, participants completed one of three paper-and-pencil questionnaires. Each questionnaire consisted of 10 items, either inducing rumination or reappraisal through questions related to the stressful situation that had been experienced previously (regulation conditions) or addressing general job-related topics (control condition). Participants were instructed to read each question carefully, write down all thoughts and emotional responses evoked by the items to ensure proper task execution, and continue working on the questionnaires until the experimenter told them to stop. In the rumination condition, the questionnaire consisted of items inducing (circular) thoughts on negative/socially threatening aspects of the TSST. Here, the participants were asked to think of negative aspects of their performance and its meaning for their self-perception and evaluation by others, to ruminate on their emotional experiences, to think of the experience of uncontrollability and to write down fearful thoughts if they were exposed to the situation again (for original questions, see OSF: https://doi.org/10.17605/OSF.IO/NGK4M). These items were adapted from established rumination questionnaires (Ruminative Response Scale; Post-Event Rumination Scale). In the reappraisal condition, the questionnaire consisted of items that stimulate thoughts for a positive reinterpretation of the situation experienced before (i.e., questions related to positive aspects of the performance; evaluating the current stressor as a preparation for future similar situations; alternative explanations for the reserved reaction of the panel; thoughts that might help to perceive the situation as a challenge rather than a threat) and the stressful bodily experiences (e.g., focus on positive aspects of the stress response: reframing an increase of HR as an adaptive response which may help coping with the stressor; for original questions, see OSF: https://doi.org/10.17605/OSF.IO/NGK4M), as have already been shown to successfully influence stress responsivity and recovery [28]. The non-regulation active control condition included questions about the job context and the application process in general. For example, participants were asked to indicate which jobs they like or dislike, which characteristics they believe an ideal boss should possess, and their general attitude toward the use of assessment centers (for original questions, see OSF: https://doi.org/10.17605/OSF.IO/NGK4M). Thus, the items in this questionnaire were designed to stimulate thoughts of similar content as those in the regulation conditions, but without inducing specific regulatory processes.

2.5. Measures and indices

2.5.1. Subjective measures

Three rating scales were used to check for successful manipulation of the cognitive-affective state following stress and the ER intervention. First, as a measure of stress-related appraisals, participants filled in the Primary and Secondary Appraisal Scale [36] prior to and after exposure to both TSSTs, as well as at the end of testing day one. The PASA was developed in accordance with Lazarus and Folkman's [11] transactional stress theory and assesses the following cognitive appraisal processes related to the TSST: stress, threat, challenge, self-concept of one's own abilities, control expectancy, building the primary, and secondary appraisal scale. Previous research has demonstrated its sensitivity to experimentally induced stress and its associations with affective and endocrine stress responses, supporting its construct and criterion validity [36]. Internal consistency indexed by Cronbach's α was calculated based on the present baseline measurement, as this time point was unaffected by stress exposure or intervention (primary appraisal scale: α = .774; secondary appraisal scale: α = .760). The PASA was administered three times on day 1 (during anticipation of TSST 1, immediately after TSST 1, and following the ER intervention) and twice on day 2 (during anticipation of TSST 2 and immediately after TSST 2), as the ER intervention did not take place after the TSST on day 2 and a third assessment was therefore not warranted. Second, the Differential Affective Scale (DAS) was used to measure changes in affective state via mean summary scores for negative (sadness, anger, disgust, contempt, anxiety, shame, guilt) and positive (joy, surprise, interest) affective factors. The DAS is sensitive to affective changes in response to acute stress, as previous studies have repeatedly shown reduced positive factor values and increased negative factor values following the TSST compared to the P-TSST [33,37]. Internal consistency of each factor at baseline verified good reliability (positive affect: α = .819; negative affect: α = .884). The DAS was assessed in parallel with each saliva sample at five time points across both experimental testing days. Third, participants repeatedly rated how unpleasant (from “not at all unpleasant” to “very unpleasant”) and stressful (from “not at all stressful” to “very stressful”) they experienced the TSST, and provided information on the frequency (from “not at all” to “very often”) and valence (from “negative” to “positive”) of thoughts related to the TSST on a visual analogue scale (VAS) ranging from 0 to 100. As the VAS items captured single, situation-specific subjective ratings, internal consistency estimates were not applicable. VAS ratings were obtained at the end of each testing day and additionally at the beginning of day 2 to assess potential differences between ER conditions prior to repeated stress exposure. Of note, timing differences across psychological measurements and between the days (PASA & VAS) reflect measure-specific analytical aims rather than procedural inconsistencies.

2.5.2. Endocrine measures

Five saliva samples were collected on each testing day (10 in total per participant) to assess levels of the enzyme alpha-amylase (sAA), an indirect marker of noradrenergic activity [32], as well as salivary cortisol. On each testing day, saliva samples were collected immediately before the TSST (t_baseline), after TSST exposure (t₊₁₅), after the ER manipulation (t₊₃₀), and 45 min (t₊₄₅) as well as 60 min (t₊₆₀) post-stress to capture the recovery period. Samples were obtained using Salivette® sampling devices (Sarstedt, Nümbrecht, Germany), stored at −20 °C, and subsequently analyzed in the on-site biochemical laboratory of the Departments of Cognitive Psychology and Genetic Psychology at Ruhr University Bochum. Free, unbound salivary cortisol concentrations were determined using a time-resolved fluorescence immunoassay (IBL, Hamburg, Germany). Due to insufficient saliva volume, cortisol concentrations could not be determined for seven samples. In addition, a colorimetric test using 2-chloro-4-nitrophenyl-α-maltrotriosoide (CNP-G3) as a substrate reagent was applied to assess sAA concentrations [38]. Intra- and inter-assay coefficients of variance of both assays were below 8.6 %.

2.5.3. Cardiovascular measures

As an additional biomarker of the SAM system, electrocardiogram-derived interbeat intervals (IBI) were recorded at a sampling rate of 1000Hz at both testing days using a Polar watch device (Polar V800, Polar® Electro, Finland) connected wirelessly to an elastic chest strap (Polar H10 Heart Rate Sensor, Polar® Electro, Finland). The chest strap was positioned just below the chest muscles and recorded interbeat interval (IBI) data derived from internal R-peak detection algorithms. On both experimental testing days, recordings were obtained during a 6-min resting period immediately prior to the TSSTs (baseline), throughout the entire duration of both TSSTs (stress), and during a 6-min post-intervention period (post; i.e., 17 min after the end of the TSSTs; see Fig. 1). Due to technical issues, IBI data could not be obtained from seven participants. IBI data were exported using device-specific software (Polar Flow; Polar® Electro, Finland) and further processed with Kubios HRV Scientific 4.1.2.1 in accordance with established HRV guidelines as reported in Jentsch and Wolf [28]. Before analysis, IBI data were detrended using smoothness priors (λ = 500), visually inspected for abnormal or biologically implausible beats, and corrected using a threshold-based artifact correction algorithm (mean percentage of corrected IBIs: 3.9 %) with cubic spline interpolation adjusted individually for every participant. HR and HRV indices were both computed from the same processed IBI data. Mean HR values were calculated for the last 5 min of the baseline and post recordings and across the 12-min TSST period. HRV was calculated for exploratory purposes reflecting fluctuations in the time intervals between consecutive heartbeats and indexes the dynamic balance between sympathetic and parasympathetic influences on cardiac activity. In this study, we used the root mean square of successive differences (RMSSD), a widely established HRV measure and robust proxy of cardiac vagal tone and parasympathetic activity [39].

2.5.4. Indices

Similar to previous research on ER and stress habituation [23,29], delta scores between peak and baseline levels of cortisol (i.e., Δ cortisol = t₊₃₀ – t_baseline), alpha-amylase (i.e., Δ sAA = t₊₁₅ – t_baseline), heart rate (i.e., Δ HR = stress – baseline), HRV (i.e., Δ RMSSD = stress – baseline) as well as the affective state (i.e., Δ negative affect and Δ positive affect = t₊₁₅ – t_baseline) were calculated separately for day one and day two as measures of stress reactivity. Additionally, we calculated the area under the curve with respect to increase [40] for cortisol and sAA from baseline (t_baseline) to the end of experimental testing (t₊₆₀) on each testing day. To capture the total stress response, including anticipatory responses and the recovery period, the area under the curve with respect to ground [40] was calculated for cortisol and sAA from baseline to the end of the testing on each day. As a separate index of stress recovery, we computed the area under the curve with respect to decrease (AUCd), defined as the area from the peak value to the final measurement [41]. Note that this index was added in the review process of this manuscript to complement the preregistered AUCg/AUCi indices. Finally, difference scores of all indices (Δ, AUCi, AUCg, AUCd) between the two days (i.e., day one – day two) were computed to assess group differences and enhance comparability with prior research [42].

2.6. Statistical analysis

To analyze differences in stress recovery and habituation between the ER conditions, we employed a 3 × 2 × 2 mixed design with the between-subjects factors Condition (rumination vs. reappraisal vs. control) and Sex (men vs. women), and the within-subjects factor Day (first vs. second). Statistical analyses were conducted in IBM SPSS Statistics 21 (Armonk, USA) using α = .05. As preregistered, physiological data were corrected for outliers (>±3.5 SD). Normality was assessed using Kolmogorov-Smirnov tests and skewed variables (raw salivary cortisol & negative affect data) were log-transformed. Homogeneity of variance was checked using Levene's tests, and Greenhouse-Geisser corrections were applied when sphericity was violated. Partial eta squared (η²) is reported as an estimate of effect size. Before testing the main hypotheses, we checked whether the TSST and the ER manipulation had been successful. To this end, general stress reactivity was examined using mixed ANOVAs on all outcome measures, with Sex as a between-subjects factor and Time (affect, cortisol, sAA = baseline, t+15, t+30, t+45, t+60; HR/RMSSD = baseline, stress, post) and Day (first vs. second) as within-subjects factors. To assess general habituation, univariate ANOVAs on delta scores (Δ) served to compare stress reactivity between the two days. To evaluate whether the ER intervention successfully altered stress-induced changes in the cognitive–affective state, we analyzed group differences in DAS and PASA separately for each day using mixed ANOVAs with Time as the repeated measure (DAS: baseline to t+60; PASA: Day 1 = t0, t+15, t+30; Day 2 = t0, t+15). A multivariate ANOVA additionally tested group differences in perceived unpleasantness and stressfulness of the TSST, as well as TSST-related thought frequency and valence (items of the VAS) on each day. To test the main hypotheses, we ran mixed ANOVAs with Day as the within-subjects factor and physiological outcomes (Δ, AUCg, AUCi, AUCd for cortisol and sAA; Δ HR; Δ RMSSD) as dependent variables. Given that AUCg reflects both baseline levels and reactivity, baseline cortisol, sAA, and HR levels across both days were analyzed separately as an additional indicator of habituation [31]. Furthermore, univariate ANOVAs were conducted on day-to-day differences in Δ, AUCg, AUCi, AUCd (cortisol, sAA), Δ HR, and Δ RMSSD. Significant main effects of Condition and interaction effects were followed by Bonferroni–Holm–corrected post-hoc tests.

3. Results

3.1. Stress induction and habituation check

3.1.1. Subjective measures

Analyses of the DAS showed that negative affect significantly increased (main effect of Time: F(2.50,217.28) = 21.14, p < .001, η² = .195) and positive affect decreased in response to stress on both days (main effect of Time: F(2.94,255.55) = 32.44, p < .001, η² = .272) verifying that the TSST worsened participant's affective state. Importantly, negative affect was overall lower on day two compared to day one (main effect of Day: F(1,87) = 61.63, p < .001, η² = .415) and increased less strongly after re-exposure to the TSST on day two (Δ negative affect, main effect of Day: F(1, 87) = 11.91, p = .001, η² = .120) indicating successful affective habituation. Similarly, positive affect was lower on day two relative to day one (main effect of Day: F(1,87) = 33.28, p < .001, η² = .277) but increased somewhat more strongly to the second stress exposure (Δ positive affect, main effect of Day: F(1,87) = 2.80, p = .098, η² = .031). Overall, women reported feeling more negative and less positive than men (main effects of Sex: F ≥ 5.19, p ≤ .025, η² ≥ .056).

3.1.2. Endocrine measures

The TSST induced significant activation of both stress axes on both days, as evidenced by increases in sAA (main effect of Time: F(2.28,197.97) = 44.22, p < .001, η² = .337), and cortisol (main effect of Time: F(1.72,141.12) = 38.96, p < .001, η² = .322; Fig. 2a and b). As expected, analyses revealed reduced endocrine responding to the second compared to the first TSST, with reduced overall cortisol levels (main effect of Day: F(1,82) = 21.64, p < .001, η² = .209) and reactivity (Δ cortisol, main effect of Day: F(1,89) = 8.49, p = .005, η² = .087) as well as reduced overall sAA concentrations (main effect of Day: F(1,87) = 7.33, p = .008, η² = .078). However, sAA reactivity to stress (Δ) did not differ significantly between the days (no main effect of Day; p = .638). No sex differences occurred (no main effects or interactions with Sex: all ps > .111).

Fig. 2.

Endocrine and cardiovascular responses to repeated Trier Social Stress Test (TSST) exposure. Mean (±SEM) levels of salivary cortisol (in nmol/l, a), alpha-amylase (in U/l, b), heart rate (in beats/minute, c), and heart rate variability indexed by RMSDD (in ms, d) at each time point of measurement on day one and day two. Exposure to the TSST caused significant increases in salivary cortisol, alpha-amylase, heart rate, and decreases in RMSSD on both days. Moreover, physiological responses to the second TSST were lower on day two relative to day one, especially with respect to cortisol. Significant differences in any time point of measurement between the days resulting from Bonferroni-Holm-corrected post hoc t-tests are marked as follows: ∗∗∗p< .001; ∗∗p< .01, ∗p < .05.

3.1.3. Cardiovascular measures

The TSST significantly enhanced cardiovascular activity on both days as reflected by increases in HR (main effect of Time: F(1.41,115.50) = 275.10, p < .001, η² = .770) and decreases in RMSSD (main effect of Time: F(1.77,145.10) = 23.59, p < .001, η² = .223; Fig. 2c and d). Significant Time × Day interactions for HR (F(2,164) = 6.86, p = .001, η² = .077) and RMSSD (F(2,164) = 4.61, p = .011, η² = .053) indicated lowered cardiovascular responses during the second compared to the first stress exposure, with the RMSSD effect mostly being driven by women (Time × Day × Sex: F(2,164) = 3.71, p = .026, η² = .043). However, there were no significant differences in Δ HR and Δ RMSSD between the days (both ps ≥ .382).

3.2. Emotion regulation manipulation check

3.2.1. Differential affective scale (DAS)

A mixed ANOVA for negative affect on day one resulted in a significant Time × Condition interaction (F(5.57,236.86) = 2.62, p = .008, η² = .058), indicating that participants in the rumination group felt significantly more negative than controls immediately after ER manipulation (main effect of Condition at t₊₃₀: F(2,86) = 5.50, p = .006, η² = .113; pairwise t-test: rumination > control, p = .009; Fig. 3a). Unexpectedly, there was also a trend toward increased negative affect in the reappraisal group compared to controls (p = .097). Consistently, a trend for a Time × Condition interaction on day two (F(4.29,186.58) = 2.30, p = .056, η² = .050) again indicated that the reappraisal group reported to feel somewhat more negative than controls immediately after re-exposure to the ER intervention (main effect of Condition at t_baseline: F(2,88) = 5.22, p = .007, η² = .106, pairwise t-test: reappraisal > control, p = .006). With respect to positive affect, both ER groups overall felt less positive than controls on the first day (main effect of Condition: F(2,83) = 4.28, p = .017, η² = .093, reappraisal < control, p = .026; rumination < control, p = .069). Importantly, these differences occurred especially 15 min and 30 min after the ER intervention (Time × Condition: F(8,332) = 1.73, p = .091, η² = .040; main effects of Condition at t₊₄₅ and t₊₆₀: F(2,88) ≥ 4.54, p ≤ .013; η² ≥ .095; rumination and reappraisal < control, p ≤ .018; Fig. 3b). On day two, participants of the rumination group overall tended to feel less positive than controls (main effect of Condition: F(2,85) = 2.96, p = .057, η² = .065; rumination < control, p = .054), whereas no differences emerged between the reappraisal and the control group (p = .428). For both days, no interactions with Sex occurred (all ps ≥ .134). Together, rumination and reappraisal led to more negative and less positive affective states following stress compared to controls.

Fig. 3.

Changes in the affective state following different ER conditions on day one.

Mean (±SEM) negative (a) and positive (b) affective state ratings of the Differential Affective Scale (DAS) are shown for each ER condition at all measurement time points on day one. The ER interventions resulted in increased negative affect and decreased positive affect ratings. Significant differences and trends between ER conditions at any time point of measurement, based on Bonferroni–Holm–corrected pairwise t-tests, are marked as follows: ∗∗p< .01, ∗p < .05, ⁺p < .10.

3.2.2. Primary and secondary appraisal scale (PASA)

Analyses of cognitive appraisals revealed a Time × Condition interaction for self-concept of abilities (F(4,150) = 2.51, p = .044, η² = .063), indicating lower perceived abilities in the rumination group than in controls immediately after the ER manipulation on day one (main effect of Condition at t₊₃₀: F(2,88) = 3.10, p = .050, η² = .067). This effect was mainly driven by men, as reflected by a Time × Condition × Sex interaction (F(4,150) = 2.60, p = .038, η² = .065). Consistently, secondary appraisals showed a significant Time × Condition × Sex interaction (F(3.51,131.50) = 5.53, p = .001, η² = .129), demonstrating that men in the rumination group reported markedly lower coping expectations than controls (Time × Condition: F(4,66) = 5.78, p < .001, η²= .259; main effects of Condition at t+15 and t+30: F ≥ 3.87, p ≤ .029, η²≥ .166). Men in the reappraisal group also showed slightly reduced secondary appraisals at t+30 relative to controls (p = .053). A similar pattern emerged for control expectancy: both rumination and reappraisal reduced perceived controllability in men (Time × Condition × Sex: F(3.34,125.21) = 4.31, p = .005, η² = .103; Time × Condition in men: F(4,66) = 3.08, p = .022; η² = .157; main effect of Condition at t₊₃₀: F(2,41) = 5.39, p = .008, η² = .216). No group differences occurred at any other time point or on the second testing day (all ps ≥ .385). Multivariate analyses for pre-post differences of PASA values including ER condition and sex revealed similar results (for details, see Supplementary Information A). These findings indicate that the rumination but also the reappraisal intervention reduced awareness of one's own coping abilities and controllability of the stressor, particularly in men.

3.2.3. Visual analogue scale (VAS)

Multivariate analyses of VAS ratings revealed significant Condition × Sex interactions for stressfulness on both days (F(2, 85) ≥ 4.33, p ≤ .016, η² ≥ .092), with men in the reappraisal group rating the TSST as less stressful than controls (F(2,40) ≥ 3.40, p ≤ .043, η² ≥ .145; reappraisal < control, p ≤ .041). In addition, MANOVA resulted in a significant Condition × Sex interaction for thought frequency on day one (F(2,85) = 6.46, p = .014, η² = .096), with a trend for women in the reappraisal group to report more frequent TSST-related thoughts than controls (F(2,44) = 2.72, p = .077, η² = .110). Neither were there any group differences in unpleasantness nor in valence of thoughts (all ps ≥ .393).

3.3. Effects of emotion regulation on biomarkers of stress recovery and habituation

3.3.1. Endocrine measures

Analysis of differences in cortisol reactivity (Δ) to repeated stress revealed a trend toward a Day × Condition interaction (F(2,85) = 2.45, p = .092, η² = .054). Exploratory post hoc analysis within each ER condition revealed reduced cortisol responses to the second compared to the first TSST in the reappraisal (main effect of Day: F(1,30) = 7.25, p = .012, η² = .195) and control group (main effect of Day: F(1,30) = 4.52, p = .042, η² = .131) verifying successful habituation. In contrast, no such reduction in cortisol reactivity to the second stress exposure was observed in the rumination group, indicating no habituation at all (no main effect of Day: p = .763; Fig. 4a). No significant differences between ER conditions emerged for AUCg, AUCi, AUCd or difference scores (all ps ≥ .120). Regarding sAA, no differences in Δ or AUCg were observed between ER conditions across days (all ps ≥ .135; Fig. 4b). However, analyses revealed significant larger AUCi of sAA on the first compared to the second day in male participants of the reappraisal group (Day x Condition in men: F(1,39) = 4.28, p = .021, η² = .180, main effect of Day in the male reappraisal group: F(1,12) = 5.17, p = .042, η² = .301), whereas no day-to-day differences occurred within the rumination and the control group (both ps ≥ .442). Given comparable delta scores on both days, the observed differences in AUCi were most likely due to prolonged sAA reactivity to stress on the first, relative to the second day (see Fig. 5b for sAA curves). In line with this interpretation, further analyses on the recovery index revealed a trend towards larger AUCd of sAA on the first compared to the second day in male participants of the reappraisal group (Day x Condition in men: F(1,41) = 2.56, p = .089, η² = .111, main effect of Day in the male reappraisal group: F(1,12) = 4.11, p = .066, η² = .255; Fig. 5a), whereas no day-to-day differences occurred within the rumination and the control group (both ps ≥ .612). Separate analyses of baseline values revealed that cortisol and sAA did not differ between ER conditions on either day (no Day × Condition interactions: both ps ≥ .322). However, a significant main effect of Condition (F(2, 83) = 4.42, p = .015, η² = .096) showed that across days, the reappraisal group exhibited higher baseline sAA levels than the rumination group (p = .012).

Fig. 4.

Physiological responses to stress on each day in each ER condition. Mean (±SEM) increases (Δ; peak – baseline) in cortisol (a), sAA (b), HR (c) and RMSSD (d) in response to the TSST on day one (dark blue bars) and day two (light blue bars). Whereas cortisol responses to the second TSST decreased in the reappraisal group and controls (a) HR responses decreased in the control group only (c) and no day-to-day differences were found in the rumination group. Significant Bonferroni-Holm-corrected post hoc comparisons are marked as follows: ∗p < .05.

Fig. 5.

Alpha-amylase decreases following repeated stress exposure in both men and women. Mean (±SEM) area under the curve with respect to decrease (AUCd) from peak value to the final measurement of sAA in men (a) and women (c) following the TSST on day one (dark blue bars) and day two (light blue bars) with respect to each ER condition. In addition, sAA levels (±SEM) of the reappraisal group per time point of measurement on each day in men (b) and women (d) are depicted. Whereas no day-to-day differences occurred in the rumination and control group, men of the reappraisal group showed reduced AUCd sAA at the first relative to the second day (a). These results indicate that reappraisal prolonged the recovery period following the first TSST exposure in men. Bonferroni-Holm-corrected post hoc comparisons at a trend level are marked as follows: ⁺p < .10.

3.3.2. Cardiovascular measures

Analysis of Δ HR revealed a significant Day × Condition interaction (F(2,78) = 3.62, p = .031, η² = .085) showing reduced HR increases to the second relative to the first stress exposure in the control group (F(1,28) = 8.03, p = .008, η² = .223) indicating successful habituation of the cardiovascular system, whereas no day-to-day differences were observed in both ER groups (both ps ≥ .397; Fig. 4c). Consistently, analysis of difference scores between the days resulted in a main effect of Condition (F(2,78) = 3.62, p = .031, η² = .085), with the rumination group exhibiting significantly reduced HR difference scores compared to controls (p = .035). No such effects of the ER condition were found for RMSSD (all ps ≥ .166; Fig. 4d).

To account for the nested data structure, we additionally analyzed physiological data with a multilevel modeling approach that controls for within-person variation in cortisol, sAA, HR and RMSSD data. We estimated random intercepts for participants as well as random slopes for Time and Day to account for interindividual differences in baseline levels and within-person changes across the testing procedure and across testing days. These linear mixed-effects models yielded results consistent with those obtained from the reported mixed-design ANOVAs (for details, see Supplementary Information B).

3.4. Exploratory analyses: evaluation of emotion regulation task performances

Given the partially unexpected effects of the reappraisal intervention, we explored whether participants of this condition had performed the task as successfully as the other two groups. Two independent raters (blind to hypotheses, analyses and results) evaluated all questionnaire responses on a scale ranging from 1 (responses did not correspond to the questions at all) to 10 (responses fully corresponded to the questions). Since each participant completed two questionnaires per session, mean scores were calculated. Interrater reliability, assessed via intraclass correlation coefficients (ICCs), was moderate for single ratings (ICC [1,3] = .623, 95% CI [.477, .735]) and further increased for mean ratings (ICC [2,3] = .767, 95% CI [.646, .847]), with agreement between the raters exceeding chance (F(88,88) = 4.30, p < .001). As averaging improved reliability, mean ratings were used for subsequent analyses. A univariate ANOVA revealed significantly lower performance evaluations for the reappraisal task than for both other tasks, independent of sex (F(2,83) = 19.30, p < .001, η² = .317), indicating substantially poorer regulatory performance when working on the reappraisal relative to the rumination and control questionnaire. In addition, a main effect of Sex showed that ER task performance in women was generally rated as better than in men (F(1,83) = 5.85, p = .018, η² = .066).

4. Discussion

In the present study, we examined how rumination and reappraisal affect recovery from and habituation to repeated stress in men and women compared to an active control condition. The short version of the TSST successfully induced psychological and physiological stress responses that gradually diminished over time. This response was attenuated upon re-exposure, indicating successful habituation, particularly in terms of negative affect and cortisol. As expected, rumination impaired affective recovery after the first stressor and lowered perceived coping abilities and controllability of the stressor, confirming the successful induction of ruminative thinking. Consistent with our hypothesis, control participants showed a significant reduction in cortisol and HR responses to the second stress exposure, whereas no such reductions were found in the rumination group. These findings indicate that rumination prevented physiological habituation to acute psychosocial stress. Unexpectedly, reappraisal also hindered affective recovery and reduced perceived coping abilities and controllability in men. These cognitive–affective effects were accompanied by prolonged sAA reactivity to the first stressor, suggesting a sex-specific impairment in noradrenergic recovery. However, reappraisal neither accelerated cortisol recovery nor enhanced habituation. The present findings provide initial causal evidence that rumination disrupts physiological habituation to repeated stress independent of sex.

This experimental evidence extends previous correlational work showing that spontaneous rumination following the TSST predicts non-habituation of the HPA axis upon re-exposure [23] and aligns with findings linking trait rumination to increased cortisol reactivity and delayed recovery [20]. These effects might be driven by altered evaluations of threat and perceived coping abilities following rumination, which are critical cognitive determinants of physiological stress responses [11]. Consistently, this study found that rumination reduced participants' perceived coping abilities and their sense of control over the stressor. These effects likely impeded affective recovery after first stress exposure and thus contributed to sustained perceived social-evaluative threat over time, which in turn may have disrupted the cognitive conditions under which physiological habituation typically occurs. In line with this idea, rumination has already been described as maintaining threat perceptions and interfering with the downregulation of the HPA axis (for reviews, see Refs. [[22], [43]]). Prior research, however, has primarily focused on the role of rumination in endogenous cortisol secretion [[20], [21], [22]], cortisol reactivity to stress [20], and failure to habituate at the HPA level [23], without addressing whether and how rumination acts on the SAM axis. The present study incorporates sympathetic biomarkers and demonstrates that rumination-induced disruption of habituation may extend beyond the HPA axis, as it additionally hampered reductions in cardiovascular reactivity to re-occurring stress. This finding again complements existing correlational work showing that high trait rumination is linked to reduced decreases in reactivity of systolic blood pressure to repeated stress exposure [44]. Accordingly, rumination impairs habituation processes in both major stress systems, which may compromise an individual's capacity to adequately adapt to stressors faced multiple times.

Importantly, failure of physiological stress habituation has been characterized as one of the major risk factors for chronic stress, thereby increasing vulnerability to mental and physical health problems (for a review, see Ref. [8]). Hence, rumination-driven disruptions in habituation contribute to increases in allostatic load, which reflects the cumulative “wear and tear” on the body's central regulatory systems [[2], [10]]. Over time, this may overburden organic functioning, e.g., reflected in immune system impairments, cardiovascular dysregulation, and heightened risk for stress-related mental disorders, such as depression and anxiety (for a review, see Ref. [3]). Crucially, such psychopathologies are often accompanied by heightened cognitive rigidity and a tendency toward ruminative thinking [[19], [43]]. Thus, rumination not only contributes to dysregulated stress physiology but may also be reinforced by its psychological and physiological consequences, establishing a vicious cycle in which maladaptive cognitive patterns and stress responses mutually intensify over time. Importantly, future research is needed to determine whether and how the reported effects in laboratory research generalize to everyday life. Longitudinal studies using ecological momentary assessment may help to clarify how naturally occurring rumination shapes physiological stress habituation in daily contexts and whether repeated episodes of elevated physiological responding accumulate and contribute to clinically meaningful outcomes.

Unlike previous research that linked reappraisal to increased cortisol habituation [29] and faster HRV recovery [28], the reappraisal intervention in this study did not strengthen stress recovery or habituation. Notably, both studies emphasize the critical role of habitual reappraisal use in daily life for beneficial effects on endocrine stress responding and, in particular, cardiac vagal flexibility. Evidence for improved HRV recovery following reappraisal [28] has been interpreted in light of the match–mismatch hypothesis, which posits that deliberate emotion regulation via reappraisal is more automatic, less effortful, and ultimately more effective under stress when it aligns with an individual's habitual regulatory preferences. Accordingly, it is reasonable to assume that a certain degree of practice or habitual proficiency in reappraisal is necessary for this strategy to exert beneficial effects in stressful contexts. Future research implementing a reappraisal training (compared to a control training) before stress exposure may clarify whether its beneficial effects on stress reactivity, recovery, and habituation require a more automatized mode of application.

However, the present findings not only failed to demonstrate beneficial effects of reappraisal but even pointed in the opposite direction. Reappraisal caused a slower recovery of the noradrenergic system after initial stress exposure in men, a pattern that appears counterintuitive at first glance. Importantly, participants of the reappraisal group reported somewhat higher negative and lower positive affect following the ER task compared to controls, suggesting ineffective regulatory performance in the intended direction. This regulatory failure may potentially have concealed beneficial reappraisal effects on stress physiology. Two important methodological aspects should be considered when discussing why reappraisal failed, while the rumination task worked as expected. First, the ER task was scheduled immediately after the TSST, a timing that likely coincided with peak noradrenergic activity and the initial rise of glucocorticoid effects [6]. This dominance of sympathetic-driven responses has been associated with increased activation in emotion-related brain regions (e.g., amygdala) and reduced prefrontal control capacities [45]. Of note, the rumination task encouraged participants to sustain the repetitive elaboration of negative thoughts intentionally elicited by the TSST, whereas the reappraisal task required them to generate alternative, positively reframed interpretations. It is therefore reasonable to assume that reappraisal was cognitively more demanding and probably recruited prefrontal control resources to a greater extent than rumination. In line with this assumption, instructed reappraisal has repeatedly been shown to rely on intact prefrontal control functioning [[46], [47]]. By contrast, rumination has primarily been associated with activation of the default mode network and comparatively reduced engagement of prefrontal control regions [48]. Consistently, rumination is positively associated with psychopathology (e.g., depression [49]) and may even be facilitated by diminished prefrontal control capacity [50]. Accordingly, the stress-induced temporary depletion of prefrontal resources may have particularly limited participants' ability to work on the reappraisal but not the rumination task in this specific time window after stress. A recent theoretical model postulates that rapid stress-induced impairments in reappraisal are counteracted by somewhat delayed beneficial effects of cortisol on higher-order cognitive functioning [51]. Accordingly, future studies may benefit from scheduling reappraisal interventions at least 30 min after stress onset when regulatory resources are likely to be restored. Second, the emotional intensity of the TSST may have undermined the effectiveness of reappraisal. Evidence suggests that engagement-based strategies such as reappraisal become less effective as the intensity of negative emotional states increases [18]. Given that the TSST reliably elicits strong stress responses [52], its emotional intensity may have exceeded the threshold at which reappraisal can be successfully implemented. In favor of this idea, intervention studies in which participants were instructed to apply reappraisal during the TSST or a similar psychosocial stress paradigm reported null findings [30] or even paradoxical effects, i.e., increased cortisol reactivity [28].

Despite these methodological considerations, it is still an open question why reappraisal impaired noradrenergic recovery from stress, particularly in men. Of note, men reported higher negative affect and lower positive affect across testing sessions compared to women, suggesting that exposure to and anticipation of the TSST may have exerted a stronger impact on their affective state. This interpretation is consistent with findings showing that men are particularly sensitive to achievement-related challenges - such as mathematical and verbal performance tasks - which represent key components of the TSST [53]. Importantly, such evaluative performance situations pose a substantial threat to the social self and are therefore highly salient stressors [9]. It is therefore reasonable to assume that the TSST represents a more potent threat to self-esteem for men, increasing their susceptibility to adverse consequences when regulatory strategies fail. Supporting this idea, men in the present study reported reduced perceived coping ability and controllability of the stressor immediately after the reappraisal intervention, suggesting that insufficient regulatory success may have had a particularly detrimental impact on their stress appraisals. Since reappraisal requires cognitive engagement with the emotional stimulus [18], it inherently involves re-experiencing aspects of the stressful situation, potentially sustaining or amplifying stressor-related appraisals. Given that such appraisals play a central role in shaping physiological stress responses [11], cognitively re-engaging with a highly salient stressor in the absence of effective regulation may have prolonged SAM-axis activation specifically in men. However, these interpretations remain somewhat speculative and call for future studies examining sex differences in the effects of reappraisal (and reappraisal failure) on stress recovery.

Some limitations need to be mentioned. First, we relied on participants' self-reported cognitive-affective ratings to assess the success of the ER intervention. While these measures provide valuable insight into subjective experience, they are susceptible to self-report biases, such as social desirability or individual differences in awareness of one's own thoughts and feelings [54]. In light of evidence pointing to potential sex differences in emotional awareness [55], it cannot be ruled out that such differences in susceptibility to self-report biases may have contributed to sex-specific effects in the present cognitive–affective data. Moreover, although the questionnaire instructions were identical across all three ER conditions, we cannot fully exclude the possibility that the content of the ER items implicitly suggested an expected regulatory direction, thereby inducing expectancy effects in self-report outcomes. Importantly, however, the primary outcomes of this study are physiological measures, which are less susceptible to such effects. Second, given that participants worked on the ER task twice (immediately after the first TSST and immediately before the second TSST), it remains unclear whether the observed effects on habituation reflect regulatory processes occurring on one of the testing days or represent the cumulative influence of both interventions. This raises the question of whether successful cognitive-affective recovery following the initial stress exposure is a necessary prerequisite for physiological habituation upon re-exposure, or whether regulatory influences exerted immediately before the second stressor are sufficient to shape habituation dynamics. Third, while the present study demonstrates that rumination influences habituation to a psychosocial stressor administered on two consecutive days, the temporal dynamics of this effect remain unclear. Given that both the number of stress exposures and the interval between them may shape habituation processes [[56], [57]], further research is needed to clarify whether and how the impact of cognitive ER interventions on habituation changes when stressors are spaced further apart or occur repeatedly across longer time periods. Fourth, even though we exclusively tested women outside of their menses, we did not account for fluctuations of sex hormones across the menstrual cycle. There is, however, multiple evidence for sex differences in cognitive ER skills [58] as well as in their availability under stress [51]. These effects at least in part result from variations in sex hormones that interact with HPA axis reactivity and neural actions of its end-product cortisol [59]. Importantly, based on self-reported cycle phase, the distribution of menstrual cycle phases did not differ significantly between ER groups, reducing the likelihood of systematic group bias. Nevertheless, self-report does not allow precise determination of hormonal status. Future studies should therefore include direct assessments of sex hormones (e.g., estradiol and progesterone) to more accurately examine whether and how hormonal fluctuations moderate the effects of cognitive ER on stress recovery and habituation. Fifth, the TSST slightly varied from the original protocol in terms of timing (shortened preparation and arithmetic task), which might have reduced comparability across studies. Sixth, RMSSD was calculated from varying IBI recording lengths (5-min baseline/post-TSST and 12-min during the TSST), which may reduce strict comparability. In conclusion, the present study revealed that rumination prohibits physiological habituation to the TSST across both major stress systems, impairing individuals' ability to adapt adequately to repeated psychosocial stress exposure. Unexpectedly, the reappraisal intervention did not help to promote stress recovery or habituation when applied immediately after stress, i.e., a period in which cognitive resources are likely limited due to noradrenergic disruptions of cognitive control. Positive reappraisal impaired noradrenergic recovery in men, implying that regulatory interventions requiring substantial cognitive flexibility may be even maladaptive for some individuals under certain conditions. Taken together, this study identifies rumination as a key intervention target for reducing chronic stress risk and highlights the need to clarify the conditions under which reappraisal facilitates - rather than impedes - physiological stress adaptations. In doing so, it provides important insights to guide the development of effective preventive and therapeutic interventions for stress-related (psycho)pathologies.

CRediT authorship contribution statement

Katja Langer: Writing – review & editing, Writing – original draft, Visualization, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Oliver T. Wolf: Writing – review & editing, Resources. Valerie L. Jentsch: Writing – review & editing, Supervision, Methodology, Conceptualization.

Funding and disclosure

This work was supported by the German Research Foundation (DFG) grant LA 5599/2-1 to Katja Langer (grant number: 551671001). The DFG has no role in study design, collection, analysis and interpretation of data, writing of the manuscript or in the decision to submit the paper for publication. All authors reported no biomedical financial interests or potential conflicts of interest.

Declaration of competing interest

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

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

The data supporting the findings of this study are available at the Open Science Framework (OSF) under https://doi.org/10.17605/OSF.IO/NGK4M.