Emotions are multifaceted phenomena involving both experiential, behavioral and somatic domains (Gross, 2015a). Thus, a person that is experiencing an emotion like anger or sadness, will experience rapid changes in how he or she feels and behaves, in addition to bodily reactions. There has been increasing attention on how individuals orchestrate emotional reactions in order to adapt to an ever changing environment. Emotion regulation is now considered to play a central role in psychological health and well-being, and has been proposed as a transdiagnostic factor in mental health (Gross and Muñoz, 1995; Fernandez et al., 2016). Although the concept of emotion regulation have gained increased popularity, controversy and debate regarding the operationalization of the concept is still ongoing, as is evident by the many definitions of the concept (Bloch et al., 2010). In clinical research, the definition of emotion regulation by Gratz and Roemer (2004) is frequently used. These authors propose that emotion regulation involves multiple processes. First, emotion regulation consists of the ability to be aware of emotions as they arise, and to experience them clearly. Emotional awareness and clarity thus enable us to monitor and evaluate emotional experience, and to modify the experience when needed. It has been suggested by several authors that deficiencies in emotional awareness and clarity may lead to emotion regulation failure (Gross, 2015b). Second, emotion regulation involves the ability to accept emotions. The inability to accept emotional reactions, by suppressing or avoiding them, is associated with increased emotional valence, cognitive load and physiological arousal (Hayes et al., 2006). Further, acceptance represent a psychological flexibility which is increasingly being incorporated into modern psychotherapy (Dryden and Still, 2006). Third, emotion regulation is also associated with the ability to tolerate emotions to meet situational demands and personal goals without being overwhelmed by them. For example, when experiencing negative emotions, behaving in accordance with ones goals, selecting appropriate emotion regulation strategies, and modulating emotional responses to meet goals and situational demands is of high importance to adapt to the environment.

To measure emotion regulation and the above mentioned facets, Gratz and Roemer (2004) constructed the Difficulties in Emotion Regulation Scale (DERS). Converging evidence show that self-reported difficulties with emotion regulation as measured by the DERS are associated with psychopathology including borderline personality disorder (Rajappa et al., 2011), post-traumatic stress disorder (Tull et al., 2007) as well as mood and anxiety disorders (Mennin et al., 2009; Fowler et al., 2014; Menezes and Bizarro, 2015).

When assessing complex, multifaceted processes like emotion regulation, several methods of assessment is of interest to gain more insight into different aspects of emotion regulation, such as psychophysiological assessments (Seeley et al., 2015). As self-report questionnaires may measure the subjective experience of emotion regulation, psychophysiological assessment may be valid assessment of the somatic processes underlying emotional processing. Furthermore, psychophysiological assessments can provide data that are not prone to biases that underlies the use of self-report measures, including social desirability, misunderstanding of items and deficient self-knowledge or psychological mindedness. In addition, some populations may not be able to provide self-reports, making it important to assess validity of other modes of assessments. For example, previous research have utilized the measurement of cortisol levels in toddlers to assess stress (Drugli et al., 2017), skin conductance in postoperative patients to assess pain perception (Storm, 2008) and autonomic flexibility in patients with alcohol related brain damage to assess quality of life (Steinmetz et al., 2016). Further, given the usage of the DERS in clinical populations, it is also of importance to validate this self-report measure to other assessments that propose to measure similar or related concepts.

Heart rate variability (HRV) is a proposed psychophysiological marker of emotion regulation capacity (Appelhans and Luecken, 2006; Thayer et al., 2012). The heart is innervated by the central autonomic network (Benarroch, 1993), through both postganglion sympathetic fibers and vagally mediated parasympathetic nervous system (PNS). This network allows the individual to adapt in a flexibly manner to a continuously changing environmental demands (Kemp and Quintana, 2013). The vagal innervation slows down the heart rate, whereas the sympathetic nervous system (SNS) increases the heart rate – often in situations with perceived threats or danger. According to the Theory of Neurovisceral Integration the variability between heart beats is affected by several brain areas such as the prefrontal cortex, a brain area also involved in regulating emotions (Thayer and Lane, 2009; Thayer et al., 2012). With emotional arousal, e.g., emotional stress responses, the SNS produces higher heart rates when there are difficulties in emotion regulation, the PNS may not slow down this heart rate and the physiological stress responses. The dominance of the SNS on heart rate will therefore lead to a lower HRV (Appelhans and Luecken, 2006). High HRV in a resting state indicates a flexible PNS, and conversely, low HRV indicates sympathetic and more inflexible influence of the heart rate. The ability to adjust the physiological systems to meet environmental demands thereby reflects a general ability to adapt to situational contingencies. There is emerging evidence that lower HRV is associated with psychological disorders, including anxiety (Chalmers et al., 2014), depression (Kemp et al., 2010; Kemp et al., 2012), schizophrenia (Montaquila et al., 2015), and borderline personality disorder (Koenig et al., 2016). Vagus mediated HRV (vmHRV) is thus used as a peripheral measure of prefrontal inhibition of the amygdala (Thayer et al., 2012). As the prefrontal cortex is crucial in the inhibition of the amygdala and is considered a neural basis of emotion regulation, vmHRV can be defined as a peripheral measure of emotion regulation (Johnstone and Walter, 2014). Evidence also indicates that general emotion regulation capacity is associated with vmHRV, including acceptance of emotions (O’Connor et al., 2002; Cristea et al., 2014), and accessibility of emotion regulation strategies (Aldao and Mennin, 2012; Gillie et al., 2015). In addition, we have previously shown that higher self-reported usage of the emotion regulation strategy self-compassion is associated with higher vmHRV (Svendsen et al., 2016). In sum, available research points to vmHRV being a marker of emotion regulation capacity (Appelhans and Luecken, 2006; Beauchaine and Thayer, 2015). However, the relation between vmHRV and self-report measures of emotion regulation difficulties is inconclusive, and few studies have yet explored this association.

To our knowledge, only two studies have investigated the association between self-reported difficulties in emotion regulation and HRV. One of these studies showed that a subsample of participants reporting higher emotion regulation difficulties had lower HRV and recovered slower in their heart rates after a negative emotion elicitation using film clips, compared to participants reporting lower emotion regulation difficulties (Berna et al., 2014). Thus, participants reporting high emotion regulation difficulties also had lower HRV, and needed more time to recover after an emotional provocation. However, the study did not find a significant correlation between self-reported emotion regulation and HRV when using the whole sample. The study used a shorter, non-validated version of the DERS, which may explain the insignificant finding. In contrast, Williams et al. (2015) used the full and validated version of the DERS and found an association between HRV and several facets of emotion regulation difficulties as measured by DERS. More specifically, they found that lower resting HRV was negatively correlated with all but one facet of emotion regulation difficulties (emotional awareness). Multiple hierarchial regression analyses showed that only impulse control difficulties and emotional clarity were associated with resting HRV. The latter analysis controlled for the effect of emotional stress such as trait anxiety and rumination. Emotional dysregulation is related to higher stress reactions and as such, controlling for trait anxiety and rumination may remove variance that is due to difficulties in emotion regulation. As a recent studies have shown, worry is highly associated with self-reported difficulties in emotion regulation (Allan et al., 2015), as well as lower HRV and lower usage of the adaptive emotion regulation strategy reappraisal (Knepp et al., 2015). Further, higher usage of the maladaptive emotion regulation strategy rumination may reflect limited access to emotion regulation strategies. In healthy adults, it can therefore be argued that it is important to measure self-reported emotion dysregulation with the DERS without adjusting for overlapping emotional traits. Further, both of the aforementioned studies exclusively relied on laboratory derived data, raising questions about the ecological validity of the results.

The main aim of the present study was to further investigate the association between self-reported difficulties in emotion regulation and vmHRV. We also wished to investigate the validity of the findings using 24-h ambulatory HRV data. Assessing vmHRV throughout 24-h using ambulatory equipment in naturalistic settings may validate data retrieved from shorter laboratory settings (Ewing et al., 1984). These long-term assessments may also shed light on the importance of diurnal fluctuations in vagal activity (i.e., Brosschot et al., 2007). Our first hypothesis was that higher self-reported emotion regulation difficulties would be associated with HRV. Given that there are few studies that have explored this association, our study was exploratory without specific hypotheses on which facets of emotion regulation that would be associated with HRV. Our second hypothesis was that similar results will be found using 24-h ambulatory HRV data.

No difference between males and females were found in respect to all HRV variables. A paired t-test showed that nighttime HRV was significant higher than daytime HRV (t₂₇ = -4.01, p < 0.000). Descriptive data from all participants (n = 60) and subsample (n = 28) including means (M), standard deviations (SD) can be found in Table 1. Bivariate correlations showed that only the DERS Total (r = -0.273, p = 0.035) and the DERS Non-accept (r = -0.345, p = 0.007) correlated significantly with resting logRMSSD of the DERS scores. Thus, participants that had lower resting vmHRV reported higher difficulties in emotion regulation and higher difficulties in accepting negative emotions. In addition, resting logRMSSD correlated with 24-h logRMSSD (r = 0.491, p = 0.008), daytime logRMSSD (r = 0.374, p = 0.05) and marginally significantly with nighttime logRMSSD (r = 0.372, p = 0.051). None of the DERS scales correlated significantly with the long-term logRMSSD data when not controlling for covariates of sex, gender and BMI (see Table 2).

Similar results were evident using the HF HRV component (see Supplemental Table 1).

In order to assess the relative contributions of self-reported emotion regulation difficulties on resting logRMSSD, we conducted independent multiple hierarchial regression analyses using DERS Total in addition to the six subscales as dependent variables and resting logRMSSD as independent variable. The analyses shared a first step, including age, gender, BMI and respiration index values (HF Peak) as covariates. The second step included DERS Total and the subscales as predictor variables respectively. Multiple hierarchial regression analyses showed that variance in resting logRMSSD was explained by 9% from the DERS Total score (β = -0.31, t = -2.48, p = 0.016). Thus, there is a significant association between lower vmHRV and higher self-reported emotion regulation difficulties when covariates are controlled for.

Multiple hierchial regression analyses using DERS Non-accept as a predictor variable in the second step showed that DERS Non-accept explained 12.6% of the variance (β = -0.37, t = -3.02, p = 0.004). Thus, having difficulties with accepting negative emotional reactions is associated with lower vmHRV, even when controlled for covariates, and alpha corrigated for number of analyses. DERS Goals explained 9.1% of the variance in resting logRMSSD (β = -0.31, t = -2.5, p = 0.015). The remaining subscales showed no significant models using multiple hierarchial regression analyses. In all analyses, no significant effect of the covariates gender, BMI and age on the resting logRMSSD were found. When applying the HF HRV into the model as dependent variable, similar results emerged on DERS Total (β = -0.25, t = -1.96, p = 0.055), DERS Non-accept (β = -0.34, t = -2.7, p = 0.008) and DERS Goals (β = -0.23, t = -1.8, p = 0.081). Thus, when utilizing HF HRV into the model, only DERS Non-accept remains significant. Model summaries of the multiple hierarchial regression analysis, including model change statistics are shown in Table 3. The complete regression matrix is enclosed in Supplementary Table 2.

Finally, to examine the ecological validity of our findings, bivariate correlation analyses using long-term logRMSSD controlling for age, gender and BMI, showed that the DERS total score correlated significantly with 24-h logRMSSD (r = -0.38, p = 0.049) and daytime logRMSSD (r = -0.40, p = 0.33), but not nighttime logRMSSD (r = -0.24, p = 0.218). In addition DERS Non-accept correlated with daytime logRMSSD (r = -0.45, p = 0.017). The 24-h, daytime and nighttime logRMSSD variables did not correlate significantly with the remaining DERS subscales.

The purpose of the present study was to explore the associations between a well-established measurement of emotion regulation (DERS) and HRV, a proposed psychophysiological marker of emotion regulation capacity. Results showed that higher self-reported difficulties in emotion regulation were negatively associated with resting vmHRV in our bivariate correlation and multiple hierarchical regression analyses. When exploring the individual subscales of DERS, difficulties with accepting negative emotions, in addition to the inability to act in accordance with personal goals when experiencing negative emotions were associated with resting vmHRV. Not accepting negative emotions had the strongest association with resting vmHRV, and this association remained significant after Bonferroni adjusting the alpha level for number of analysis. Higher self-reported difficulties in emotion regulation were also negatively associated with vmHRV measured in a 24-h ambulatory setting. More specifically, we found that daytime HRV were negatively associated with both the total score of the DERS, as well as the non-acceptance subscale when controlling for the covariates of gender, sex and BMI. In contrast, no associations were found between nighttime vmHRV and the DERS in the same analyses. In sum, our hypotheses of vmHRV to reflect self-reported emotion regulation capacity was supported both in resting laboratory derived data and in the ambulatory data collected through 24-h.

The results indicate that self-reported difficulties with emotion regulation, and especially not accepting emotions, can be detected by measuring the variability between heart beats. The results give support to the model of Neurovisceral Integration (Thayer and Lane, 2009), that higher self-reported difficulties in emotion regulation and acceptance of negative emotions are associated lower variability between consequent heart beats. Thus, the findings indicate that emotion regulation capacity is related to the ability to flexibly modulate physiological processes in response to changing contexts.

Our findings confirms the hypothesis that self-reported emotion regulation difficulties are associated with autonomic flexibility as measured by HRV, and thus partially replicate earlier results (Williams et al., 2015). In addition, our findings from the 24-h measurement of HRV indicate that the results obtained from the short term HRV are ecologically valid. However, our findings were different in terms of associations between facets of emotion regulation difficulties and vmHRV. Whereas Williams et al. (2015) found that particularly lack of emotional clarity and impulse control were associated with resting vmHRV, we found non-acceptance of negative emotional reactions to be the strongest predictor of vmHRV. From a Neurovisceral Integration theoretical perspective (Thayer and Lane, 2000), the difficulty of accepting negative emotional reactions is associated with limited cortical inhibition of the amygdala. The ability to attain an attitude of acceptance and willingness to experience negative emotions is considered preventive in regards to development of psychopathology. Aldao et al. (2010) showed in their meta-analysis that acceptance of emotions is negatively associated with symptoms of anxiety and depression. In contrast, avoiding negative feelings and pushing them away is associated with psychopathology (Hayes et al., 2006) and decreased positive emotional experience and increased amygdalae and sympathetic activation (Gross, 2015a). Interestingly, acceptance as an emotion regulation strategy can be considered as an antecedent emotion regulation strategy that can be applied before an emotional reaction (i.e., accepting fear-related aspects of a situation) (Werner and Gross, 2010), in addition to a response-focused emotion regulation strategy that can be applied after an emotional reaction has occurred (i.e., accepting feeling afraid). In contrast, suppression as an emotion regulation strategy is considered as purely response focused emotion regulation strategy (Gross and John, 2003). The literature suggest that acceptance requires less cognitive resources to regulate emotions compared to suppression (Chambers et al., 2009; Lutz et al., 2014). Thus, by being both an antecedent and a response-focused emotion regulation strategy, acceptance may enable the person be more flexible in meeting potential emotionally charged situations. It is interesting that this type of psychological flexibility is associated with flexibility of the autonomic nervous system, here represented by HRV. Thus, acceptance of emotional responses may be an important focus for psychological treatment for psychopathology. Interestingly, the ability to accept negative emotions is addressed through third-wave cognitive therapies like Mindfulness-Based Cognitive Therapy for depression (Segal et al., 2013), and Compassion-Focused Therapy (Gilbert, 2009).

Although the findings of the present study support the hypothesis that self-reported difficulties in emotion regulation is associated with vmHRV, several limitations should be mentioned. First, our sample was consisting of relatively young healthy students with a majority of females, which may limit the ecological validity of the results. Although the present study adds support to the notion that increased emotion dysregulation is reflected in parasympathetic flexibility, the underlying facets of emotion dysregulation may differ across age. It could be hypothesized that hard working students that want to be included in a study offering a mindfulness-based intervention are less accepting of their emotional reactions. Also considering the sample, we did not assess the participants’ mental health, i.e., conducting semi structured interviews to confirm that the sample was a non-clinical sample. Another limitation is the use of one single questionnaire to assess emotion regulation. Although the DERS is a common inventory in assessing emotion regulation it may not capture the whole complexity of the concept. Finally, we did not perform follow-up studies to assess the reliability of our findings in this sample.

Consequently, future research should use other samples across age and clinical diagnoses to further expand our understanding of self-reported emotion dysregulation and vmHRV. Furthermore, longitudinal experimental studies should assess self-reported difficulties in emotion regulation and vmHRV to investigate temporal changes in these variables as result of psychological interventions. Also, studies should continue to use 24-h recordings of vmHRV on larger samples to obtain statistical significant results. Finally, the association between acceptance and HRV should further be investigated, for example in experimental settings, where participants are instructed to accept or suppress emotions.

Despite the limitations, the present study is a valuable contribution to the literature on emotion regulation difficulties, as it indicates that self-report measurement of these difficulties have reverberations into our bodily reactions, including central and peripheral somatic markers.

The following authors have contributed to the enclosed manuscript: EV. Project planning, collection of data, analyses of data, main author of manuscript: written the abstract, introduction, methods and materials and discussion. LS: Project planning, collection of data, analyses of data, supervision of statistical analyses, co-author and supervisor in writing process. BO: Project planning, collection of data, analyses of data (HRV), supervision of statistical analyses, supervision in writing process. JS: Project planning, collection of data. P-EB: Project planning, collection of data. ES: Project planning, collection of data, supervision in writing process.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.