Edited by: Lee E. Eiden, National Institutes of Health (NIH), United States
Reviewed by: Ben Nephew, Worcester Polytechnic Institute, United States; Gábor B. Makara, Hungarian Academy of Sciences (MTA), Hungary
This article was submitted to Neuroendocrine Science, a section of the journal Frontiers in Neuroscience
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This study aimed to investigate the impact of mental stress on salivary cytokines and attention to emotional stimuli, as well as associations between stress-induced changes of immune and cognitive parameters. In a randomized order a total of 60 young adults were assigned to one of two stress conditions with varying stress intensity. High stress was induced by a socially evaluated Paced Auditory Serial Addition Test (PASAT). As a low stress task a paper-and-pencil version of PASAT was administered. Salivary cytokines were measured before, 5 min after, and 45 min after completion of the stress task, and were assayed for pro- and anti-inflammatory cytokines. Three distinct types of attention – alerting, orienting, and executive control – were measured by the modified Emotional Attention Network Test Integration (E-ANTI). IL-1β and IL-6 increased only in the high-stress group. Significant increases in IFN-α, IFN-γ, TNF-α, and IL-10 at 45 min after stress induction (all
Considerable evidence indicates that psychological stress can lead to alterations of the immune system (
One factor hypothesized to be associated with cytokine stress responses is attention. As a major pathway of emotion regulation, attention processes are obviously implicated in stress responses. For example, by attentional engagement an individual can focus on potentially threatening aspects of the stressful event, thus intensifying the experience of stress, or ignore them and thereby possibly dampen the stress response (
To date, there is little evidence on the relationship between cytokine stress responses and attention processes. One recent attempt to fill this gap is a study by
In the current study, we varied the intensity of a laboratory mental stressor between two groups of participants, before and after which we assessed levels of saliva cytokines. As recommended by
Participants were 60 healthy adults (30 men; 30 women) aged 19–35 years (
Participants were tested individually. They arrived at the laboratory between 2 p.m. and 3.30 p.m. After signing the informed consent, saliva samples were taken. Saliva was sampled using the Salivette® Cortisol with synthetic swab (Sarstedt). The swab was placed in the mouth of the participant for 1 min with gentle movement. The saliva-soaked swab was immediately stored on ice. Samples were centrifuged and saliva was aliquoted and stored at -80°C until analysis.
Next, participants filled out questionnaires on their actual stress and anxiety level. Each participant was then randomly assigned to one of the stress conditions. Then, they were asked to perform the corresponding version of the PASAT (cf. Stress protocol), which took about 25 min. After that, the same questionnaires on stress and anxiety level were filled out, which was followed by the second sampling of saliva. Next, participants completed the E-ANT task. Finally, a third saliva sample was drawn and participants were debriefed.
Subjective stress experience was measured by means of a visual analog scale (VAS). The scale comprised a continuous horizontal line 20 centimeters in length, anchored by 2 verbal descriptors (
Anxiety was assessed by the state anxiety scale of the State-Trait Anxiety Inventory (STAI-S;
Cytokine levels in saliva were measured using the LEGENDplex Human Inflammation Panel (13-plex) and the LEGENDplex Human Th1/Th2 Panel (8-plex) (both from BioLegend) according to the manufacturer’s instructions with slight modifications.
Briefly, 10 μl saliva or standard was added to a V-bottom 96 well plate and mixed with 30 μl assay buffer, 10 μl beads and 10 μl biotinylated detection antibody mix and incubated for 2 h at RT on a shaker at 600 rpm. Then, 10 μl PE-conjugated Streptavidin was added, followed by an additional incubation for 30 min at RT on a shaker at 600 rpm. After two washes with wash buffer (provided in the kit), PE fluorescence intensity of the beads was measured on a LSRFortessa flow cytometer (BD Biosciences). Beads populations were identified by Fsc/Ssc features and fluorescence intensity in the APC channel. Approx. 400 beads per analyte were acquired. Data were analyzed using the LEGENDplexTM Data Analysis Software (VigeneTech). All samples were analyzed at the same day to avoid inter-assay variation. According to the manufacturer, intra-assay variation is between 3 and 16% CV.
In the high-stress condition, the PASAT was applied (
In the low-stress condition, the PASAT was replaced by a paper-and-pencil version of the task. Participants were required to add the same series of single digits for approximately the same duration as the PASAT. During this time the experimenter sat 1 m diagonally opposite to the participant measuring time and monitoring task performance. In this condition, participants were not exposed to socially evaluative stress. No visual feedback, no instruction about alleged analyses of body language were given. Mental stress was also reduced by keeping the difficulty level constant. A low-stress condition was preferred to a basal rest condition in order to ensure that cytokine responses were elicited by variations in stress reaction and not by secondary characteristics of the experimental setting, such as, e.g., performing mental arithmetic, differences in body posture etc. (
To assess attention to emotional stimuli, we used the modified version of the ANT based on Cohen and colleagues (emotional attention network test integration E-ANTI,
The alerting stimulus was a 84 dB (2000 Hz) tone. Orienting cues were 10 positive and 10 negative pictures (5.42 cm × 4.06 cm) selected from International Affective Picture System (IAPS,
Each trial started with the presentation of a fixation plus sign for 1000 ms in the middle of the computer screen. In tone trials alerting signal was delivered for 50 ms. No alerting tone was presented in no-tone trials. Next, an asterisk was presented for 400 ms. After the asterisk disappeared, a cue was presented for 100 ms. The cue was presented horizontally centered, with the center of the cue being positioned 2.03 cm above or below the center of the screen with equal frequency. After a cuing interval of 50 ms, a target and distractors replaced the cue either at the same position (valid cue condition), or at the opposite position (invalid cue condition). The target remained on the screen for 2,000 ms or until the participants’ response. This sequence of events is depicted in
Experimental procedure of E-ANTI.
Apart from subjecting the data of the E-ANTI to full-factorial analyses of variance (cf. Results), several psychometric scores based on main effects were derived from these data. Alerting efficiency was defined by subtracting reaction time (RT) of tone trials from RT of no tone trials (i.e., alerting efficiency). Larger numbers of difference between tone and no tone trials indicate that participants benefit more from tone trials reflecting more efficient performance. Orienting efficiency was measured by subtracting RT of trials with the valid cue from those with the invalid cue (i.e., orienting efficiency). Larger differences of RTs are assumed to arise because of a difficulty to relocate attention after an invalid cue was presented. The measure for Executive efficiency was the difference between RT of target congruent trials and RT of target incongruent trials (i.e., executive efficiency). The greater the difference, the greater the difficulty to resolve cognitive conflict which reflects less efficient performance. In sum, in contrast to Alerting, higher scores represent less efficient processing in case of Orienting (i.e., less efficient reorienting after invalid cues) and Executive efficiency (i.e., more distraction by incongruent distracters).
All statistical analyses were conducted using IBM SPSS Statistics 25.0. The raw data underlying the analyses can be found in
Similarly, anxiety levels did not significantly differ between high-stress group (
We subjected mean individual reaction times (RTs) and error rates (ERs) into 2 × 2 × 2 × 2 × 2 × 2 × 2 analyses of variance (ANOVAs) with the between-subjects factor Condition (high vs. low stress) and the within-subjects factors Tone (tone or no-tone), Cue Valence (positive or negative), Cue Validity (valid or invalid), Target Congruity (congruent or incongruent), Target Valence (positive or negative), and Cue-Target Congruity (congruent or incongruent). Trials from the very first block were regarded as practice trials and excluded from the analyses. Error trials (2.9%) and trials which were preceded by error trials were also excluded. Significant interactions were followed by least significant difference (LSD)
Mean reaction times and proportions of error rates for each experimental condition.
Alerting cue condition |
Orienting cue condition |
Orienting cue valence |
Target valence |
Cue-target valence congruity | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Group and congruity condition | ||||||||||
Low-stress | ||||||||||
Incongruent | 766 (144) | 765 (138) | 765 (142) | 766 (139) | 768 (146) | 762 (136) | 766 (141) | 764 (140) | 761 (141) | 769 (141) |
Congruent | 721 (137) | 718 (134) | 710 (135) | 729 (136) | 725 (140) | 714 (132) | 727 (138) | 713 (133) | 722 (136) | 718 (136) |
High-stress | ||||||||||
Incongruent | 735 (133) | 736 (125) | 729 (129) | 742 (129) | 735 (130) | 735 (128) | 726 (125) | 745 (131) | 733 (129) | 737 (129) |
Congruent | 694 (124) | 689 (121) | 681 (121) | 702 (123) | 694 (122) | 690 (123) | 694 (121) | 690 (122) | 693 (125) | 690 (121) |
Low-stress | ||||||||||
Incongruent | 2.88 (0.15) | 3.55 (0.18) | 3.90 (0.18) | 2.55 (0.14) | 2.62 (0.15) | 3.81 (0.18) | 2.96 (0.15) | 3.48 (0.17) | 3.03 (0.15) | 3.42 (0.17) |
Congruent | 1.60 (0.11) | 1.50 (0.11) | 1.40 (0.10) | 1.70 (0.11) | 1.59 (0.11) | 1.51 (0.11) | 1.65 (0.11) | 1.44 (0.10) | 1.74 (0.12) | 1.35 (0.10) |
High-stress | ||||||||||
Incongruent | 4.91 (0.20) | 4.12 (0.18) | 5.31 (0.21) | 3.72 (0.17) | 4.12 (0.18) | 4.91 (0.20) | 3.07 (0.16) | 5.96 (0.21) | 4.16 (0.18) | 4.87 (0.20) |
Congruent | 1.78 (0.12) | 1.76 (0.12) | 1.91 (0.12) | 1.64 (0.11) | 2.08 (0.13) | 1.49 (0.11) | 1.83 (0.11) | 1.72 (0.12) | 1.65 (0.11) | 1.90 (0.12) |
In the analysis of RTs, significant main effects of Cue Validity [
In the following, we will restrict the report of significant interactions to those involving the between-subjects factor Condition. The analysis yielded a significant three-way interaction of Cue Validity, Target Congruity, and Condition [
Cue Valence interacted with Condition [
We also observed a significant interaction of Target Valence and Condition [
Finally, we observed a significant three-way Cue-Target Congruity-by-Target Congruity-by-Condition interaction [
The corresponding analyses of ERs revealed an almost identical pattern of results. For each main effect and interaction we found for RTs, we scrutinized the data for eventual speed-accuracy trade-offs. We observed a speed-accuracy trade-off only for the main effect of Cue Validity in that faster RTs in valid trials went along with increased ERs [
For the ERs, we will only report effects of Condition which go beyond those reported in the analysis of RTs. The analysis yielded a significant interaction of Tone and Condition [
We also observed a significant three-way Tone-by-Target Congruity-by-Condition interaction [
In order to explore if subjective stress was related to the efficiency of any of the attention networks, we conducted additional correlation analysis separately for the high- and the low-stress group. Separate analyses for the two groups were based on the rationale that different levels of stress may be associated with different functional relationships between subjective stress and attention. Efficiency measures of each of the networks were correlated with changes of subjective stress levels (stress levels after stress induction or low-stress activity minus stress levels at baseline). The same was done with the state anxiety measures. No significant correlations were observed in the low-stress group (all
Prior to analyses the normality assumption was checked for all continuous variables. Cytokine data showed left skewed distributions requiring log-transformation. We assessed several control variables thought to be associated with cytokine levels, and hence might provide alternative explanations for any observed relationships between stress, attention and cytokine levels. These were age, sex, and body mass index (weight/height2). Each control variable was considered as a covariate. To check for randomization, low-stress and high-stress group were compared on control variables and baseline cytokine levels using t-tests (all
To test if cytokine levels increased with increasing levels of stress and if attention would moderate the relationships between stress and cytokine responses, we conducted a series of linear mixed models with repeated measures. Each mixed model included fixed effects of Time (5 min post-stress, 45 min. post-stress), Condition (low-stress, high-stress), the efficiency scores of the particular attention network (Alerting, Orienting, and Executive efficiency as continuous predictor), their respective two-way interaction terms (Time-by-Condition, Time-by-network efficiency, Condition-by-network efficiency), and the three-way interaction term (Time-by-Condition-by-network efficiency). The predictor variables were calculated as described in section E-ANTI with higher scores reflecting more efficient Alerting and less efficient Orienting and Executive control (cf. E-ANTI). Age, sex, body mass index were entered into the model as fixed factors. To control for baseline imbalance, each post-stress cytokine score was adjusted for its baseline score by including it as an additional control variable in each model. This procedure was chosen due to its higher efficiency gains (e.g., power) in randomized controlled studies, as compared to the analyses of change scores (
Each cytokine type was analyzed as a dependent variable. ANCOVA and regression models assume uncorrelated residuals. Since we used repeated measures design, multiple observations on cytokines originated from the same individuals. In this case, residuals from measurements next to each other might be correlated. For this reason, we set covariance structure type to autoregressive (AR1). For convenience, continuous explanatory variables were centered (i.e., by subtracting the respective sample mean) prior to analyses. Categorical explanatory variables (Condition, Time) were dummy-coded. Interaction effects were tested by simple effects tests which were adjusted for multiple comparisons (Bonferroni). Pairwise comparisons were based on estimated marginal means (EM-means, adjusted for age, sex, and body mass index). In order to interpret significant interactions including attention network efficiency variables, we divided the sample by median split into participants with low and high efficiency of each of the attention networks.
Since each cytokine type was treated as a separate dependent variable, the analyses produced multiple hypothesis tests. Sequential Bonferroni-Holm procedure was used to control for the family-wise error rate and thus reduce the probability of Type I error (
Levels of psychological stress had differential effects on IL-1β and IL-6. Mixed model analyses showed a significant effect of Time [
Effects of acute laboratory stress level on IL-1β
Repeated measures mixed models showed instead that both high and low stress led to an increase of cytokines. As
Descriptive statistics of cytokine concentrations in samples obtained 5 and 45 min after stress induction.
Cytokine type | 5 min |
45 min |
|||
---|---|---|---|---|---|
Mean (SE) | Mean (SE) | F (df1, df2)a | p | padj | |
IFN-α | 1.43 (0.16) | 1.82 (0.18) | 13.07 (1,56) | ||
IFN-γ | 5.25 (0.82) | 6.83 (0.81) | 11.74 (1,56) | ||
TNF-α | 1.95 (0.19) | 2.7 (0.29) | 12.34 (1,56) | ||
IL-2 | 27.1 (4.32) | 37.8 (5.56) | 4.59 (1,56) | 0.036 | 0.01 |
IL-4 | 43.2 (5.96) | 53.6 (8.01) | 1.86 (1,56) | 0.178 | n.s. |
IL-5 | 18.5 (3.23) | 25.6 (4.14) | 9.84 (1,56) | ||
IL-8 | 1914.4 (164.92) | 2254.42 (230.54) | 5.63 (1,56) | 0.02 | 0.01 |
IL-10 | 6.5 (1.01) | 9.7 (1.5) | 16.78 (1,56) | ||
IL-12p70 | 4.9 (0.95) | 6.06 (1.02) | 4.32 (1,56) | 0.04 | 0.01 |
IL-13 | 13.3 (1.71) | 17.1 (2.07) | 6.55 (1,56) | 0.013 | 0.008 |
IL-17A | 55.7 (10.25) | 60.4 (10.85) | 2.28 (1,56) | 0.136 | n.s. |
The analyses revealed a significant effect of gender on IL-8 [
Mixed linear models analyses revealed a significant sample Time-by-Alerting network efficiency effect on IL-1β [
Moderation of IL-β
Similarly, Alerting efficiency was associated with IFN-α levels and their stress-induced changes [
The Time-by-Orienting interaction was statistically significant for IFN-α [
Moderation of IFN-α
The analyses further revealed significant Condition-by-Orienting efficiency effects on IFN-α [
Finally, we observed a significant three-way interaction effect of Time-by-Condition-by-Executive efficiency on IL-17A levels [
No further significant Time-by-Condition-by-attention network efficiency interaction effects on cytokine reactivity were observed, indicating that modulating effects of attention processes on cytokine reactivity over time were similar for both high-stress and low-stress groups.
So far, the attention network measures were collapsed across positive and negative stimuli. In order to tap eventual valence-specific effects, we tested modulating effects of stress and attention on cytokine reactivity depending on stimulus valence (positive or negative). To this end, we first calculated change scores (second post-stress sample – first post-stress sample) for each cytokine type. We further calculated Orienting and Executive network efficiency separately for positive and negative stimuli. Positive Orienting efficiency was calculated by subtracting RTs in trials with positive valid cues from RTs in trials with positive invalid cues. Negative Orienting efficiency was calculated in the same way using negatively cued trials. For Executive network, efficiency was computed by subtracting congruent from incongruent trials separately for positive and negative targets. We applied partial correlation analyses to test for relationships between cytokine changes and attention networks efficiency for emotional stimuli. Age, sex, and body mass index were included into analyses as control variables.
The analyses revealed negative correlations between negative Orienting efficiency scores and changes of IFN-α (
For emotional Executive network efficiency, we detected a mutual dissociation between low-stress and high-stress group and positive and negative stimuli. Higher increases of IL-17A in the low-stress group were negatively correlated with (inversely scored) negative Executive network efficiency (
Correlations between (emotional) Executive efficiency and change scores of IL-17A in low-stress
In order to test for associations between stress and stress-induced cytokine changes, we conducted additional correlation analysis separately for the high- and low-stress groups. To this end, we first calculated cytokine change scores by subtracting sample 2 (5 min. post-stress) from sample 3 (45 min. post-stress). We correlated cytokine change scores of each cytokine type with changes of subjective stress levels (stress levels after stress induction minus stress levels at baseline). The same was done with the state anxiety measures. We observed positive correlation between subjective stress levels and levels of INF-α (
The purpose of our research was to examine the impact of acute mental stress on levels of salivary cytokines and different dimensions of attention to emotional information (alerting, orienting, and executive network) measured by emotional attention network test integration (E-ANTI). Furthermore, we examined if efficiency of each of the attention networks would be associated with cytokine changes induced by psychological stress. Cytokines were determined from saliva with saliva samples having been collected before, 5 min after, and 45 min after stress exposure. The stress protocol applied in our study was designed to induce different levels of mental stress, high and low stress. The high-stress condition consisted of a socially evaluated PASAT. In the low-stress condition a paper-and-pencil arithmetic task without socially evaluation was used.
The data showed greater increase of IL-1β and a trend toward a greater increase of IL-6 in high-stress condition. Increases of further cytokines did not vary with varying levels of stress. Instead, IFN-α, IFN-γ, TNF-α, and IL-10 increased in both conditions. The second question was if levels of stress would predict efficiency of emotional attention networks. The main results were interactions between experimental condition and emotional valence of the stimuli. These interactions indicated that participants who were exposed to PASAT reacted faster to negative stimuli. In contrast, participants who underwent low-stress activity showed improved performance in positive cue and positive target trials. The third hypothesis predicted that emotional network attention efficiency would be related to cytokine responses. Higher Alerting Efficiency was positively associated with higher increases in IFN-α and IL-10. Higher Orienting efficiency was positively associated with higher increases in IFN- α, TNF- α, IL-2, IL-5, and IL-10.
At the outset, we point out that our study is limited by the lack of a basal rest condition, which was not possible to obtain within the parameters of our study design, and our analysis and discussion of the results obtained, below, reflects this limitation. Thus, apart from distinctive effects of the stress manipulation on IL-1β, IL-6 and stimulus valence in the attention task, contrasting cytokine and attention measures between the high- and low-stress groups failed to yield significant effects of stress intensity. The low-stress condition may be an inappropriate reference measure to establish potential effects of stress, given potentially stressful features of the experimental situation, therefore decreasing differences in stress experience between the high- and the low-stress group. Second, the speeded attention task which was carried out before the last saliva collection might have contributed to increased stress levels in both conditions. Timing of the sampling is another critical variable that may require further attention in future studies. Nevertheless, the data as obtained offer significant insights into the association of cytokine responses to psychological stress with attention to emotional information, as discussed in detail below.
Apart from an impact of stress on error rates regarding Alerting network efficiency, there were no overall effects of condition on any of the attention networks. However, emotional stimulus valence had differential effect on participants’ performance depending on stress level. Participants in the high-stress condition reacted faster after the presentation of a negative cue. In the same vein, negative targets also improved performance in this group. In contrast, the low-stress group took longer to respond when they were faced with negative cues or targets. In other words, negative information disrupted information processing in the low-stress group and facilitated performance in the high-stress group. The interference of negative cues with performance in the low-stress group may reflect a relatively automatic increase of the salience of negative stimuli that enhances their processing at the cost of task-related cognitive processes. As research demonstrates, negative stimuli have a greater disruptive effect on other cognitive processes than positive stimuli (
Our results demonstrated that increases in IL1-β and IL-6 varied with levels of stress (see
With regard to the primary concern of our study, we found evidence that attention processes modulated the impact of stress on cytokine responses. Increases in cytokines after stressful task varied depending on the efficiency of attention networks. High efficiency of the Alerting network was associated with stronger responses of IL-1β, IFN-α, and IL-10. In both the low-stress and the high-stress group, increases of cytokines were more pronounced in participants with higher Alerting efficiency (see
Higher efficiency of the Orienting network was furthermore related to greater elevations of IFN-α, TNF-α, IL-2, IL-5, and IL-10 (see
Contrary to our expectation, we observed little evidence for the relationship between executive control and cytokines. On the descriptive level, higher efficiency of the Executive network was associated with lower levels of IL-17A in the low-stress group. No relationship was detected in the high-stress group, suggesting that the modulating effects of Executive efficiency take place only if stress is low. To our knowledge, no study has linked IL-17A to psychological stress and Executive attention before. There is one study which examined the associations between cognitive control and pro-inflammatory cytokines. Specifically,
Increases of pro- and anti-inflammatory cytokines observed in our study can be interpreted as a part of an adaptive stress response. Facing a stressful challenge, an organism prepares for a fight-or-flight reaction and its possible consequences (e.g., injury, infection). As mediators of inflammation, cytokines play a key role in wound healing and defending the body against infections. That is, higher levels of cytokines along with an activated HPA-Axis imply protective effects of acute stress (
However, potential modulation of cytokine responses through attention processes should also be considered. Shifting attention toward or away from certain aspects of a situation is one of the strategies to regulate emotions (
Finally, psychological stress and cytokine responses may produce synergistic effects on attention by influencing cognition and behavior through functionally similar pathways. For example, in the study of
Several limitations of our study have been noted earlier in the discussion. Another limitation of our study is that the first saliva sample was collected shortly after participants arrived at the laboratory. Thus, it cannot be ruled out that this first sampling was affected by the novelty of the laboratory setting, which could have contaminated the baseline measurement especially in the low-stress group. Some relaxation period before collecting the first sample or including a sample taken at home should be considered in future studies. Furthermore, the last saliva sample was collected 45 min. after completion of the stress task. Since levels of the most cytokines were elevated at that time of measurement, it is unclear whether the cytokine stress response already achieved its maximum at this time. Further research with additional samples taken at later time points after the stressor is needed. Second, there was no rest condition in our study. As indicated by the results, our stress manipulation failed to induce differences in responses of several cytokines between the high-stress and low-stress groups. Therefore, it may be beneficial to include a no-stress (i.e., rest) control condition to determine the associations of immune and cognitive parameters unaffected by stress. However, it should be noted that any laboratory procedure will induce some amount of stress in naïve participants, making it hard to realize a no-stress condition in a strict sense. One way to deal with this problem is to implement a habituation phase to make participants familiar with the laboratory setting and saliva sampling. Finally, our results on the relationships between cytokine responses and attention measures are correlational. Future studies should thus examine, if e.g., experimental stimulation of cytokine production would affect attention networks, or, vice versa, if manipulation of attention networks would have impact on cytokine reactivity.
Taken together, higher intensity of acute laboratory mental stress induced an increase in IL-β and a marginal increase in IL-6. Levels of IFN-α, TNF-α, IFN-γ, and IL-10 increased in response to both high and low intensity of the stressor. Stress-induced cytokine responses were associated with the efficiency of attention networks. Furthermore, this relation partially depended on the valence of stimuli used in the attention task such that cytokine increases were closer related to attention to negative information than to positive information. The associations between acute stress, attention to emotional information, and cytokine responses are relevant for understanding individual differences in stress reactivity. In addition to studies examining the modulatory role of personality traits in immunity, research on cognitive processes such as attention promises to shed new light on the relationship between psychological and immunological factors, e.g., mechanisms underlying the link between stress and mental and physical health outcomes. To date, attention processes and stress-induced inflammation have been closely related to pathology of depression and anxiety disorders by two independent lines of research. Biased attention to emotional information plays a key role in stress reactivity and in the development and persistence of symptoms of affective disorders (
All immunological and psychological data obtained and analyzed for the present study are available in the
This study was carried out in accordance with the recommendations of the local Ethics Committee of the Leibniz Research Centre for Working Environment and Human Factors. The protocol was approved by the local Ethics Committee of the Leibniz Research Centre for Working Environment and Human Factors. All subjects gave written informed consent in accordance with the Declaration of Helsinki.
VM and TK designed the study. MC and VM collected the data. All authors analyzed and interpreted the data. VM prepared the manuscript, which was revised by TK, MC, and CW. All authors agreed to be accountable for the content of the work.
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.
We thank all volunteers for participating in this study. The publication of this article was funded by the Open Access Fund of the Leibniz Association.
The Supplementary Material for this article can be found online at: