Introduction

Stress reactivity represents an important mechanism to cope with environmental challenges and threats. Faced with a challenging situation, both physiological and psychological processes are triggered to different degrees and interact in order to enable the individual to cope with acute and chronic stress [1].

With respect to psychological mechanisms influencing stress reactivity, emotion regulation plays an important role. Gross [2] defined emotion regulation as encompassing “the processes by which individuals influence which emotions they have, when they have them, and how they experience and express these emotions” (pp. 275). Adaptive emotion regulation has been summarized in a multidimensional model by Gratz and Roemer [3] and emphasizes four dimensions: the awareness and understanding of emotional states, the acceptance of emotions, the ability to inhibit impulsive behavior and to maintain goal-directed behavior, and the flexible use of adaptive and appropriate strategies to modulate emotional responses. Impairments in any of these dimensions stand for a rigid control, for suppression of emotions or for an inadequate use of strategies reflecting trait-like emotion regulation difficulties (ERD). ERD are widely accepted as a transdiagnostic core feature involved in the onset and maintenance of psychopathology [4–6]. Prior research showed evidence that ERD are related to a broad range of mental disorders [7] such as depressive disorders and anxiety disorders [4,8] or eating disorders such as anorexia nervosa (AN) and bulimia nervosa (BN) [9,10]. In these studies, ERD were measured by well-validated self-report questionnaires to assess different facets of emotion regulation and its dysfunction such as the Difficulties in Emotion Regulation Scale (DERS) [3].

Lately, there has been increasing interest in the interaction between ERD and biological mechanisms when faced with situational challenges. The capacity of the organism to adequately react to the environment depends on the flexible neural network system called autonomic nervous system (ANS) which is connected to the amygdala and the medial prefrontal cortex and is indexed by heart rate variability (HRV) [11,12]. Based on Thayer and Lane’s Model of Neurovisceral Integration [13], high frequency heart rate variability (HF-HRV) represents parasympathetic activation (PNS) and has been identified as a marker of emotion regulation [13]. In other words, higher HF-HRV in resting state and greater HF-HRV changes when facing situational challenges are related to emotion regulation attempts and stand for a more adaptive response and subsequent recovery from a stressor [11]. In contrast, lower HF-HRV, during resting conditions, reflects impairments of the regulatory role of the ANS. Increased ERD are associated with limited inhibition capacity of the PNS [1,14], in healthy individuals [15] and in mental disorders such as eating disorders [16]. In contrast, data on HRV change is limited and little is known about the link between ERD and HRV change [16]. Those studies investigating the relation between ERD and ANS responses relied on standardized stress tasks [17], while only few applied ecologically valid paradigms eliciting daily moderate stress conditions by personal recall or picture viewing methods [18]. Berna, Ott and Nandrino [19] investigated HF-HRV changes during an emotion-elicitation video clip-based stress task in a sample of 63 healthy undergraduate students at the age of 18 to 27. Prior to the exposition to three consecutively presented anger-inducing short video clips (FilmStim database), Berna and colleagues [19] assessed ERD and divided participants into lower and upper quartiles depicting well-developed and limited emotion regulation capacities respectively according to the DERS. They further measured skin conductance and HF-HRV. The result was that, if controlled for the influence of depression, anxiety and age, the group in the upper quartile of ERD showed less change in HF-HRV between an elicitation task and recovery than the low ERD group and thus took longer to recover after emotion elicitation. In contrast to prior research, there was no difference between groups at rest. According to Berna and colleagues [19], enduring ERD causes prolonged HRV recovery and thus a decrease of resting HRV level over time due to an extended overload of the stress system.

Up to now, data on the association between ERD and physiological stress responses focused on the assessment of HRV as a correlate of the autonomic activation and neglected the stress responses of the hypothalamic-pituitary-adrenal (HPA) axis indexed by the increased release of salivary cortisol in response to a challenge [20]. The HPA axis is especially important in terms of interactions of ERD and biological stress response [21], as the HPA system is influenced by stress and by higher order cognitive processes such as the evaluation of the perceived information [22]. We found only one study by Lam and colleagues [23] of healthy male and female students, where higher scores of maladaptive trait emotion regulation strategies predicted heightened cortisol release after a social speech task.

Data on the interrelation between ERD and biological mechanisms of individuals facing standardized stress situations is very partial and knowledge on the processes during moderately stressful daily events remains scarce. Apart from emotion eliciting video or picture viewing [19,24], engagement in aversive or in worrisome mental imagery has been shown to be an important daily stressor [25]. Therefore, this laboratory study seeks out to examine the effect of exposure to thin ideals in media and subsequent vivid imagery of thin ideals in young females with different levels of ERD. Based on prior research regarding the stress inducing effect of body image exposure [26,27], we expected exposure to thin ideals in media to be a moderate daily stressor, fulfilling criteria of ego-involvement in young females in the age of risk between 18 to 35 years [28–31]. To intensify the effect of exposure, we instructed participants to a subsequent vivid imagination of the most attractive female bodies previously seen in the magazines [32]. Our assumption was confirmed in our previous study where combined exposition and vivid imagery of thin ideals led to self-reported stress responses in terms of impaired mood, body satisfaction and body-related cognitive distortions in healthy women [33]. Therefore, in the current study, we considered vivid imagination of thin ideals to be a moderate daily stressor triggering stress responses in our female sample and favored ecological validity over a fully standardized exposure. In order to control for effects of quality of exposure, we additionally asked for self-report of the vividness of imagery and carried out an implicit memory test on the material seen during the exposure.

In this framework, the aim of the present study is fourfold: First, we investigated whether self-reported stress response in terms of mood is related to increased trait-like ERD measured by DERS. Second, we examined, whether increased ERD are associated with impaired PNS functioning. Third, we aimed at extending knowledge on the relation between ERD and the dysregulation of the HPA axis [34,35]. We included measures of HRV as well as of salivary cortisol in order to detect activation of both branches of the stress response system. Fourth, we investigated the association between ERD and psychological and biological stress responses to moderately stressful daily stressors in a large sample. This consisted of healthy and young women suffering from a mental disorder (eating disorders such as AN and BN, depressive or anxiety disorder) thereby potentially reflecting a wide range of ERD. The interest of the current study is therefore based on the impact of ERD and not on the categories of mental disorder.

Specifically, we hypothesized that participants with high levels of self-reported ERD are characterized by a more pronounced decrease of mood and impaired PNS and HPA functioning at both rest and during stress. This should be observed in lower resting HRV level and by smaller HRV changes from baseline to vivid imagination and from vivid imagination to recovery. We should also see higher salivary cortisol levels at rest and smaller salivary cortisol differences between the periods before (baseline) to after the vivid imagination (recovery) than participants with low ERD compared to a neutral condition, where no stress response independent of ERD is expected.

As advised in previous relevant studies [36,37], we controlled for the effects of age, BMI, use of hormonal contraceptives and medication with effects on the central nervous system (pharmaceuticals and psychopharmaceuticals) on biological parameters. We further considered the effect of the vividness of imagery and the implicit recognition of pictures presented during the exposure in order to approximate the strength of stress induction [33,38].

Methods

Procedure

The study is part of the multi-site cross and longitudinal experimental trial “Psychological and physiological consequences of exposure to mass media in young women–the role of moderators”. A full description of the overall study sample, protocol, all measurements and procedure is provided in Munsch [32].

In all study units, the protocol was administered as follows: After signed informed consent, all participants were randomly assigned to a thin ideal exposure condition (EC: exposure to a fashion magazine followed by vivid imagination of the thin ideals) or neutral exposure condition (CC: exposure to a nature magazine followed by vivid imagination of landscapes). Participants received written information about the procedure and were told that the study examines mental well-being and psychophysiological stress reactivity relating to body image satisfaction in young adults. They were only informed about the objectives of the study and the experimental manipulation after participation. Participation included a diagnostic interview, the completion of a set of online questionnaires and an experimental testing day. Diagnostic interviews were conducted either by telephone or face-to-face. Participants were interviewed shortly after agreeing to participate and within the first weeks of beginning treatment for the patient group. One week after the interview, self-report questionnaires, such as DERS [3], were sent to all participants using an open source online survey tool. Healthy participants were invited to an individual testing afternoon the following week at the Laboratory of the University of Fribourg and patients to an individual testing at the corresponding clinical institution. Participants received oral and written information on physiological assessment prior to the experimental procedure. This included the following restrictions: participants were instructed not to drink anything else but water, eat, brush their teeth, use mouth-rinse or smoke in the hour preceding the experimental procedure. The experimental procedure took place between 2 and 4.30 p.m. with respect to circadian rhythm of the HPA axis and the ANS system. Time stamps of each step of the experimental procedure were computer-based. Fig 1 highlights the parts of the experimental procedure after exposure to the thin ideals used for the purpose of this substudy.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Experimental procedure with exact saliva sampling time points.

Q = Questionnaires 1 to 5 including mood questionnaire; HRV segmentation: B = Baseline, VI = Vivid imagination, R = Recovery. The part of the experimental procedure used for this substudy is highlighted.

https://doi.org/10.1371/journal.pone.0199769.g001

Physiological measurements and repeated administration of self-report questionnaires on body image, eating behavior and mood (Brief Mood Scale) [43] were assessed repeatedly before and after media exposure and subsequent vivid imagination. Participants were asked to put on the Electrocardiogram (ECG) breast belt for heart rate (HR) measurement at 10 minutes after the beginning of the experimental procedure and were told not to touch or refasten the belt once adjusted. Heart rate was recorded continuously during the whole experimental procedure. Nine saliva samples for salivary cortisol were collected in order to capture peak and recovery phases over exposure periods [44], and three measurements were included in this study (Fig 1). For each saliva sampling, participants were told to keep the cotton roll in the mouth for 2 minutes, avoiding any contact with other body parts.

Media exposure with fashion (EC) respectively nature magazines (CC) started 40 minutes after the beginning of the experiment. Media exposure was inspired by the waiting room study of Turner et al. [30]. Participants were asked to wait for 10 minutes in a second room, being told that baseline HR measures were assessed during this period, and were explicitly told to look at the pictures in the magazine they were given. During recovery time between magazine viewing and vivid imagination an 8 minute nature movie was presented. Vivid imagination took place 40 minutes after media exposure. The vivid imagination procedure was conducted in accordance with Radomsky et al. [45], Shafran et al. [46] and Coelho et al. [47] aiming to induce the cognitive distortion ‘Thought-Shape Fusion’ (TSF). Participants were given standardized instructions related to either the thin ideal exposure condition during vivid imagination (Vogue Germany) or neutral exposure condition during this time period (Geo Special Germany). Participants were asked to close their eyes and were instructed to remember and vividly imagine for 4 minutes the most attractive thin ideals (EC) or impressive landscapes (CC) referring to the magazine previously viewed. Instructions were given to deepen the detailed imagination. Following vivid imagination, participants were asked to write down a short description of the thin ideals or the landscape they just imagined. They were told to start the description with “I imagine…”. Shortly after vivid imagination, participants were given the opportunity to neutralize their feelings during 5 minutes and were provided with examples of neutralization activities such as exercising, drawing pictures, body checking and mental neutralization. Participants then completed a picture recognition task to check whether they viewed the images presented during the exposure. At the end of the experimental procedure, weight and height were measured for BMI assessment and participants were asked to rate the vividness of the imagination of the pictures viewed during exposition.

Results

Sample characteristics

The 273 participants (172 patients and 101 healthy controls) were randomly assigned to EC and CC. Table 1 displays characteristics of the entire sample and, for clarity and only in this table, by subsamples low or high ERD (according to median split of the DERS). The low DERS group consisted of 68.6% healthy controls and 31.4% patients. The high DERS group of 3.8% healthy controls and 96.2% patients.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sample characteristics.

https://doi.org/10.1371/journal.pone.0199769.t001

Effects on self-reported mood

Linear mixed models revealed a significant time x condition interaction (F(1,247) = 67.518, p < .001/ < .001), indicating an overall increase in self-reported mood in the CC and a decrease in the EC after vivid imagination as expected. There was no significant time x DERS interaction (F(1,247) = 1.494, p = .223/.287). Furthermore, there was a significant three-way interaction of time x condition x DERS (F(1,247) = 13.202, p < .001/ < .001). Thus, during vivid imagination, participants with higher DERS values showed a stronger decline in self-reported mood compared to participants with lower DERS values in the thin ideal condition but a stronger increase with increasing DERS values in the CC (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mood across conditions and levels of DERS over the course of vivid imagination.

VI = Vivid imagination. Lines denote predicted values for mixed model including standard errors. To meaningfully cover the range of ERD values encountered in the sample, trajectories of the median, 5^th and 95^th percentiles of the DERS are shown.

https://doi.org/10.1371/journal.pone.0199769.g002

Discussion

The aim of the present study was to examine whether changes occurring in mood, HRV and salivary cortisol following a moderate daily stress induction (laboratory thin ideal exposure and vivid imagination) are associated with different range of ERD in a large sample consisting of healthy and mentally ill young women.

We hypothesized that higher self-reported trait ERD are linked to a higher decrease in self-reported mood after vivid imagination of thin ideals. Higher trait ERD should be observed in a lower HF-HRV at rest and higher baseline cortisol levels, as well as in smaller HRV changes during vivid imagination of thin ideals but reduced cortisol responses and slower recovery of salivary cortisol levels after vivid imagination of thin ideals compared to landscapes.

In line with our previous research [33], vivid imagination of the thin ideal had a negative impact on self-reported mood in young women. Further, results showed that there was an increase of positive mood in the CC, whereas a stronger increase of negative mood was revealed in the EC with increasing ERD. It is reasonable to assume that exposure to thin ideals and vividly imagining them, impose a negative affect. Especially for participants who tend to engage in dysfunctional emotional processing [53]. However, in our sample the imagination of landscapes might have led to a shift of attention towards more positive stimuli, especially in those participants characterized by high ERD. This is an interesting finding as it illustrates that engaging in vivid imagination of non-idiosyncratic problem-related topics, such as pictures of landscapes, might intervene with ruminative processes [38], as demonstrated by the subsequent increase in mood.

Our results corroborate the role of ERD regarding the psychological stress response to vivid imagination of the thin ideal. A decrease in mood after vivid imagination was observed in a similar study by Wyssen et al. [33] among healthy women, where results depended on eating disorder symptomatology. In fact, the decrease in mood in the experimental condition could only be found in subjects with more pronounced eating disorder symptomatology. Our results should therefore be interpreted with caution. Although ERD contribute to the better understanding of the effect of vivid imagination on mood, the effect of other factors related to psychopathology or to the specificity of the stressor cannot be ruled out.

Regarding biological measurements, trait ERD were not linked to baseline HF-HRV. Further, our assumption that low HF-HRV is linked to low emotion regulation abilities in challenging situations was not confirmed. These findings are partly in line with the results of Berna et al. [19], who did not find differences in smaller study samples with low ERD (n = 18) and high ERD (n = 19) concerning HF-HRV at rest. However, in their study high ERD showed a slower recovery after stress than low ERD. In contrast to our study, Berna et al. [19] did not control for medication or use of contraceptives although they excluded subjects with substance use disorder from their sample. In our study, alcohol or drug consumption was not an exclusion criterion in order to preserve clinical reality of the sample. The comparability of these studies is therefore limited, due to varying sample size and characteristics.

Our assumption that ERD were related to higher baseline salivary cortisol levels was also not confirmed. Further, our results showed a general decline of salivary cortisol independently of ERD and condition between baseline and recovery. As outlined by Gilbert et al. [35], there is a lack of studies assessing the impact of ERD on the HPA axis and comparable findings are therefore scarce.

To sum up, the hypothesis that thin ideal exposure and vivid imagination would trigger a biological stress response was not confirmed, even though results on self-reported mood underline the negative impact of our stress induction paradigm [29,30]. This lack of a biological stress response to a daily stressor applied in the laboratory is noteworthy considering that our study fulfilled several of the suggestions of Campbell and Ehlert [54]. That is to increase the validity of results by focusing on a dynamic pattern of mood change, using precise time and context records, including large samples and considering the effect of mental illness or related biological factors such as menstrual cycle or medication. Our findings of discordant self-report and biological stress response patterns are in line with existing literature which generally showed that stress inducing paradigms using thin ideal or body confrontation have failed to reveal consistent results in ANS or HPA axis activation [27,54]. Even following mirror confrontation, where pronounced negative emotions and cognitions were reported in patients with eating disorders, both HRV and salivary cortisol remained stable [27]. A similar dissonance between self-report and physiological response has also been reported in AN patients [55].

Additionally, the discrepancy between the influence of emotion regulation mechanisms on the psychological level and the absence of physiological arousal may be due to the direct impact of varying state-related emotion regulation strategies not captured by the DERS. Another line of argument assumes, as in Gross and Levenson [53], that physiological activation depends on whether a stimulus sufficiently triggers emotional expression or not. This assumption has been confirmed by Mauss et al. [56], who showed that the type and intensity of a stimulus determine the degree of concordance between psychological and physiological outcomes. According to the authors, specific emotions with a cognitive component such as anxiety may lead to a more pronounced imbalance between the extent of the psychological and the physiological reaction compared to other emotions such as fear. Furthermore, the imbalance is related to the intensity of elicitation of a target emotion, as physiological response may have a higher activation threshold. We therefore think that future studies should investigate in more detail the immediate effect of thin ideal exposure and vivid imagery on cognitions and affect. The effects of frequency and intensity of exposure of a moderate daily stressor on biological stress responses also merits further study.

Several limitations must be considered in our study. First, the complexity of the biological processes involved may be insufficiently captured by the specific measurement time points in our design. This has been recently demonstrated by Goodman et al. [57] for salivary cortisol. Additionally, it has to be mentioned, that the length of periods to assess HF-HRV differed from baseline compared to during stress exposure and thereafter. In general, spectral values should be less affected by time length than time domain values, but this potential limitation has to be kept in mind. As previously shown in Wyssen et al. [33], a general anticipatory arousal to the experiment and a subsequent habituation to the experimental manipulation along with the combined effects of media exposure/vivid imagination, is to be expected. Furthermore, self-reported intensity of vividness of imagery was assessed, but additional analyses would be necessary to define the valence of its core content. Second, our design did not involve the assessment of state-related emotion regulation while being exposed to thin-ideals or while vividly imagining them compared to the CC. Instead, we examined, whether and how self-reported trait-like ERD influenced stress responses to a laboratory stress task. In other words, the design of our study does not allow drawing conclusions about how state-related emotion regulation influences the response when facing a stressor. We also did not assess the impact of the type of emotion experienced during thin ideal imagination. Even though it is known that, certain physiological responses are associated to specific emotions [18]. In addition, we cannot rule out the effect of other psychopathological features or attentional processes during the exposure with our specific stressor. In sum, our findings point to the role of ERD, but also accentuate that ERD does not fully explain the impact of vivid imagination of the thin ideal on mood. Third, the healthy study sample was solely recruited in Switzerland, while the clinical sample included participants of both countries, which might limit the comparability of the findings. Nor did the study include male participants or a broader age range, although age may influence the impact of ERD on the evolution of different pathologies.

Apart from these limitations, the large sample size and the inclusion of mentally healthy and participants suffering from a range of mental disorders add to the strength of the study. Furthermore, the study contributes to the exploration of physiological processes during a moderate daily stressor exposure.

In sum, this study showed a clear discrepancy between psychological and physiological responses as previously detailed out by Campbell and Ehlert [54]. The discrepancy between psychological and biological measures of stress responses is a well-known phenomenon in anxiety patients during treatment and especially in phobic patients showing heightened cortisol responses in height exposure despite improved subjective and behavioral responses [58]. We applied a moderate daily stressor with salience for our young female population in the laboratory and found a subsequent deterioration in mood, relative to the CC, where an increase in mood was observed, especially in women with high ERD. This is an important finding, as first, there might be additive negative psychological and biological effects of repeated exposure and related information processing in a longitudinal perspective. Second, enduring ERD are known to be associated with impaired inhibition of irrelevant emotional information, less goal-oriented behavior and impaired ability to reappraise stressful situations [59]. These mechanisms are per se relevant as they predict mood, anxiety, and EDs [4,59]. Added knowledge on the relation between ERD and biological correlates of stress response is of high interest, as impaired stress reactivity is associated with impaired mental health and increased symptom severity in a broad range of mental disorders [60–63]. If future studies confirm that ERD relate to impaired stress responses to challenges, emotion regulation trainings could be implemented in either prevention or treatment trials; similar to what has already been shown to be effective for depressive and anxiety disorders [64,65].

Our findings highlight the necessity to investigate ecological valid moderate stressors, occurring repeatedly on various levels of intensity, in order to investigate the repercussions of daily stressors on psychological and physiological levels.

Acknowledgments

We would like to thank all the participants and clinical centers that contributed data to TSF_Body/ media exposure. We also thank all students and the research team for their valuable contribution.