Participants

The study was approved by the Ethics Committee of the Rhineland-Palatinate State Board of Physicians (Germany). All participants provided their written informed consent to participate in this study. The sample consists of n=22 DPD patients and n=15 patient controls (Table 1). The diagnosis of DPD was established by M.M. according the German version of the Structured Clinical Interview for Dissociative Disorders [31]. Participants fulfilled the criteria of DPD according to DSM-IV (300.6) as well as the criteria of the depersonalization-derealization-syndrome according to ICD-10 (F48.1). Patients were recruited from the DPD clinic of the Department of Psychosomatic Medicine and Psychotherapy (Mainz, Germany). All DPD patients experienced persistent depersonalization. Persons with a lifetime diagnosis of a psychotic disorder or brain damage were not eligible. The mean duration of DPD was 11.7 years (standard deviation (SD)=7.5 years) with a range from 1-24.5 years. Current mental disorders other than DPD and psychotropic medication were comparable between the groups (see Table S1, supplemental material).

Questionnaires

The Cambridge Depersonalization Scale (CDS) [32,33] consists of 29 items and measures frequency and duration of depersonalization over the last 6 months. The state version of the CDS (S-CDS) comprises 22 items and reflects intensity of depersonalization right now. The Dissociative Experiences Scale (DES) [34] is a 28-item self-report scale that asks respondents to indicate the frequency of dissociative experiences on a 11-point-Likert-scale from 0%-100%. The scale contains several subscales. For this study we report the scores for the total scale, and the subscales amnesia and depersonalization. Dispositional mindfulness was assessed by the Mindful Attention Awareness Scale (MAAS) [35]. The Childhood Trauma Questionnaire (CTQ) [36] is a self-report scale of childhood interpersonal trauma. The Beck Depression Inventory-II (BDI-II) [37] is a 21-item inventory that measures depression. The State-Trait Anxiety Inventory intends to tap trait and state anxiety (STAI-T/-S) [38]. The trait questionnaires were administered within 3 days prior to the experiment (further information about the questionnaires is given in the supplemental material, Text S1).

Instruments

Physiological responses were recorded with a Biopac MP100 system (Biopac Systems, Inc.) and stored at 200 Hz for further offline processing. Skin conductance was measured by a constant voltage system (0.5 V) using a bipolar recording with two Ag/AgCl electrodes (0.8 cm diameter) filled with 0.05 M NaCl electrolyte. The electrodes were placed at the palmar surface of the medial phalanx of the index and second finger of the non-dominant hand. An electrocardiogram (ECG) was measured with pre-gelled electrodes placed at the right arm and left leg.

Stimulus material

A set of acoustic emotional stimuli from the International Affective Digitized Sounds (IADS) [30] was used for emotional stimulation. These sounds have a duration of 6 seconds each. They evoke scenarios of different emotional valence and arousal. Based on the normative ratings for valence and arousal [30], we selected a set of 20 sounds from the IADS: Four neutral sounds (neutral valence and low arousal: #132, #262, #720, #725) and four sets with four emotional sounds each that differed on the dimensions valence and arousal. This 2 × 2 classification included sounds with negative valence and medium arousal (#280, #319, #380, #706), negative valence and high arousal (#276, #277, #278, #279), positive valence and medium arousal (#221, #351, #602, #721); and positive valence and high arousal (#810, #815, #816, #820). Additionally, 2 low arousing positive sounds (#151, #812) were selected and used as buffers to familiarize subjects with the sound stimulation and to adjust the volume individually. These sounds were always presented at the beginning of the stimulation period and physiological responses to these sounds were not analyzed. Before running the experiment with patients, all sounds were tested in an independent sample of healthy controls to obtain normative ratings for a German sample and to examine the influence of valence and arousal on electrodermal and heart rate responses (see supplemental online material, Text S2).

Experimental design

The experiment started with a 5 min rest period. Patients were instructed to sit quietly on a comfortable chair with their eyes closed. Electrodermal and heart rate data were acquired continuously during this rest period to allow for analyzing tonic physiological activation in DPD patients as compared to patient controls.

Subsequently, participants were informed about the upcoming stimulation period. They were asked to keep their eyes closed and listen to the sound presentation via circumaural earphones. The acoustic stimuli were presented in a random order (with the exception of the two buffer sounds that were always presented as first and second stimulus) while continuously measuring autonomic physiological responses. For presentation we used a Dell Latitude E6500 laptop with an Intel IDT 92HD71B7 HD Audio (ICH9) soundcard and AKG K141 MKII circumaural headphones. The volume was adjusted individually, so that the patients felt comfortable with the volume. Each sound of 6 seconds duration was followed by a jittered pause that lasted between 10 and 20 s (M=11.3 s, SD=1.8 s). Measurements were obtained in two conditions, whose sequential order was counterbalanced across subjects: a) Listen to the sounds and be attentive to the emotional scenarios b) breathe mindfully while listening to the sounds, i.e. direct your the attention to the bodily sensations while letting yourself sink in the sound scenarios. Participants had no formal training in mindfulness exercises.

Subjective ratings of valence and arousal for all sounds were obtained in a separate session by using the Self-Assessment Manikin [39]. The Self-Assessment Manikin measures valence and arousal on a pictorial 9-point-Likert-scale (valence: 1=extremely negative, 9=extremely positive; arousal: 1=extremely calm, 9=extremely excited). Further, participants rated on a similar, self-constructed, 9-point-Likert-scale the vividness of the imagination that was triggered by the sounds (1=not vivid at all, 9=extremely vivid). Vividness of imagination was assessed, because impaired ability to generate visual images is common among DPD patients [40]. The above ratings represent the cognitive evaluation of the affective stimuli, i.e. the evaluation of the emotional valence of the stimuli, the awareness of internal bodily and mental processes such as arousal and imagination. These ratings were not accomplished by two patients from the DPD group. Thus, corresponding analyses rely on n=20 DPD patients and n=15 patient controls.

In order to capture subjective changes induced by the mindful breathing manipulation, participants rated the degree of feeling grounded in their body on a self-constructed 21-point-Likert-scale (-10=extremely detached (no-body), 0=normal, +10=grounded very much) and the overall intensity of the sounds (-10=lowest intensity, 0=normal, +10=highest intensity). Further, participants rated present depersonalization intensity by the state CDS. The participants rated their degree of feeling grounded in their body and the severity of present depersonalization directly after each condition (normal, mindful breathing). The degree of feeling grounded was not rated by two patients from the DPD group and the overall intensity rating and S-CDS was missing from one patient of the DPD group. Thus, corresponding analyses rely on n=20 or n=21 DPD patients, respectively, and n=15 patient controls.

Quantification of physiological responses

From the electrodermal data that were recorded during the rest period, we calculated the skin conductance level (SCL) as the average low pass filtered skin conductance signal (cutoff frequency 0.05 Hz) as well as the number of nonspecific skin conductance responses with an amplitude of at least 0.02 µS [41]. Furthermore, we quantified heart rate (HR) and heart rate variability (HRV) from the ECG recordings. To this aim, R-waves were detected in the ECG recordings and visually inspected to identify artifacts. Afterwards, interbeat-intervals (IBIs) were calculated and used to compute the mean heart rate during the rest period. Statistical parameters of HRV were calculated using the software artiifact [42]. The two time-domain measures calculated from the IBIs were the standard deviation (SDNN), and the square root of the mean squared differences between successive IBIs (RMSSD). Additionally, frequency domain measures were derived using fast Fourier transformation. According to previous recommendations [43], frequency bands were labeled as high-frequency (HF, 0.15–0.4 Hz) and low-frequency (LF, 0.04–0.15 Hz) and expressed in power (ms²) and normalized units (n.u.). We specifically analyzed HF and time domain measures as indicators for cardiac-vagal tone and additionally calculated LF/HF that can be interpreted as a measure of sympathovagal balance. The quality of the ECG recording of one DPD patient was insufficient for further analyses. Therefore, HR measures were analyzed for n=21 DPD patients and n=15 patient controls.

To quantify phasic physiological responses to the emotional sounds, we measured skin conductance response (SCR) amplitudes and heart rate changes. For SCR quantification, we measured the largest increase in skin conductance during a period from 1 to 8 s after stimulus onset as change in μS. These values were log-transformed to reduce the skew of the amplitude distribution [44]. For the analysis of heart rate responses, R-waves were detected in the ECG data. The automatically detected R-waves were visually inspected and R-R intervals were converted to HR (in beats per minute). Afterward, a second-by-second sampling was applied [45] resulting in one HR value for each of 6 s following stimulus onset (the whole stimulus duration). The HR in the last second prior to stimulus onset represented the prestimulus baseline and was subtracted from all HR values during stimulus presentation. Finally, the average of all these ΔHR values was calculated. The data sets of three DPD patients had to be discarded due to very low quality of the ECG recording. Thus, HR analyses were based on n=19 DPD patients and n=15 patient controls.

Statistical analyses

T-tests were used to assess differences between DPD patients and patient controls in tonic physiological activity during the 5 min rest period. Ratings of valence, arousal and vividness were compared between patient groups using t-tests (for neutral sounds) or 2×2×2 analyses of variance (ANOVAs) with patient group as between-subject factor and valence category (positive vs. negative) and arousal category (medium vs. high arousal) as within-subject factors, respectively. To assess discrepancies between affective ratings and IADS norm ratings, we calculated differences between each participant’s ratings and gender-specific norm ratings. These differences were compared to 0 as well as between groups using a one-way ANOVA (for neutral sounds) with patient group as between-subject factor or a 2×2×2 ANOVAs (for emotional sounds) with the additional within-subject factors valence category (positive vs. negative) and arousal category (medium vs. high arousal). Ratings of mindful breathing effects were subjected to a 2×2 ANOVA using patient group as between-subject factor and experimental condition (normal listening vs. mindful breathing) as within-subject factor. Finally, SCR amplitudes and HR changes that were elicited by the neutral sounds were analyzed using 2×2 ANOVAs with patient group as between-subject factor and experimental condition (normal listening vs. mindful breathing) as within-subject factor; corresponding responses to the emotional sounds were analyzed using 2×2×2× 2 ANOVAs with the additional within-subject factors valence category (positive vs. negative) and arousal category (medium vs. high arousal). Significant interaction effects in the ANOVAs were followed by post-hoc t-tests with Bonferroni correction. Finally, we correlated affective ratings and physiological responses with anxiety (STAI-T/-S) and depression scores (BDI-II) using Pearson’s product-moment correlation coefficient.

An a priori significance threshold of α=.05 was used for all analyses. Partial η² (for main and interaction effects in the ANOVAs) or Cohens’s d (for comparisons of two groups) are reported as effect size estimates.

Psychometric data of DPD and control patients

There were no differences between both groups for age, sex, years of education, severity of current depression, anxiety, symptoms of dissociative amnesia and childhood traumatic experiences. However, severity of depersonalization and self-rated mindfulness differed strongly between the groups (see Table 1).

Tonic physiological activity

As shown in Table 2, physiological activity during the 5 min rest period was very similar between DPD patients and patient controls. A significant difference between groups was only observed for the number of NSRs with DPD patients showing more NSRs than patient controls.

Sound ratings

DPD patients and patient controls did not differ in their valence, t(33)<1, d=0.07, arousal, t(33)<1, d=0.06, or vividness ratings for neutral sounds, t(33)=1.21, p=.24, d=0.41. With respect to the emotional sounds, the ANOVA on the valence ratings revealed a significant main effect of valence category, F(1,33)=132.53, p<.001, η²=.80, indicating that patients of both groups rated positive sounds as pleasant and negative ones as unpleasant. Moreover, as revealed by a significant valence × arousal category interaction, F(1,33)=20.44, p<.001, η²=.38, this difference was more pronounced for highly arousing, t(34)=17.58, p<.001, d=3.00, as compared to medium arousing sounds, t(34)=5.37, p<.001, d=0.91. Finally, a significant group × valence category interaction was obtained, F(1,33)=7.13, p<.05, η²=.18. DPD patients rated unpleasant sounds as less unpleasant than patient controls, t(33)=2.36, p<.05, d=0.81, whereas no such difference was observed for positive sounds, t(33)=1.72, p=.19, d=0.59 (see Figure 1A, right panel).

Download:

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

Figure 1. Valence and arousal ratings for neutral and emotional sounds.

A) Valence and arousal ratings for neutral and emotional sounds as a function of group (DPD patients vs. patient controls). The center of the ellipses in the left and middle column represent the mean rating and the radii correspond to the standard error of mean (SEM). The values next to the ellipses depict mean and SEM for vividness ratings. The right column shows valence and arousal ratings collapsed across arousal categories as a function of group with error bars indicating SEM.

B) Average differences to the norm ratings (Bradley & Lang, 2007) [30]. The dots in the left and middle column represent mean norm ratings and the crosses depict average ratings of each group. The bars in the right column depict differences to norm valence and arousal ratings collapsed across arousal categories as a function of group with error bars indicating SEM. neut = neutral, neg = negative, and pos = positive sounds; -/+ indicate medium and high arousal, respectively.

https://doi.org/10.1371/journal.pone.0074331.g001

The ANOVA on the arousal ratings of emotional sounds revealed significant main effects of arousal category, F(1,33)=7.85, p<.01, η²=.19, and valence category, F(1,33)=67.23, p<.001, η²=.67, indicating that highly arousing as well as negative sounds were rated more arousing. Furthermore, similar to the valence ratings, a significant interaction of valence × arousal category was obtained, F(1,33)=11.51, p<.01, η²=.26. Thus, arousal ratings differed significantly between medium and highly arousing negative sounds, t(34)=6.10, p<.001, d=1.03, whereas no such difference was obtained for positive sounds, t(34)=1.55, p=.26, d=0.26. Finally, a significant group × valence category interaction, F(1,33)=8.13, p<.01, η²=.20, indicates that differences in arousal ratings between negative and positive sounds varied between groups. However, corresponding post-hoc tests within each valence category did not reveal significant differences between groups.

Finally, the ANOVA on the vividness ratings only revealed a significant valence × arousal category interaction, F(1,33)=9.44, p<.01, η²=.22. Across groups, vividness ratings only differed between medium and highly arousing positive sounds, t(34)=4.62, p<.001, d=0.78, but not for negative sounds, t(34)=1.75, p=.18, d=0.30 (see values in Figure 1A).

With respect to the comparison to IADS norm ratings of neutral sounds, we did not observe significant effects for the intercept or the patient group, neither for valence, nor for arousal ratings. Thus, affective ratings of neutral sounds were very similar to the norm sample and did not differ between patient groups (Figure 1B)

With respect to differences in valence ratings of emotional sounds, we obtained a significant intercept, F(1,33)=13.64, p<.001, η²=.29, significant main effects of valence category, F(1,33)=9.33, p<.01, η²=.22, and arousal category, F(1,33)=4.81, p<.05, η²=.13, as well as a significant interaction of patient group × valence category, F(1,33)=7.31, p<.05, η²=.18. As can be seen from Figure 1B, valence ratings differed substantially from IADS norm ratings. This difference was larger for positive than for negative and to medium arousing as compared to highly arousing sounds. Whereas the valence of negative sounds was rated relatively similar between patient controls and the norm sample, t(33)=1.70, p=.20, d=0.58, these sounds were rated significantly more neutral in DPD patients, t(33)=2.42, p<.05, d=0.83.

A comparable analysis on the differences in arousal ratings of emotional sounds revealed a significant intercept, F(1,33)=6.95, p<.05, η²=.17, as well as significant main effects of valence, F(1,33)=13.41, p<.001, η²=.29, and arousal category, F(1,33)=4.25, p<.05, η²=.11. Overall, arousal ratings were lower than in the norm sample (Figure 1B). Moreover, larger discrepancies were observed for positive as compared to negative sounds as well as for highly arousing as contrasted with medium arousing sounds. The ANOVA additionally yielded a significant group × valence category interaction, F(1,33)=8.04, p<.01, η²=.20, indicating that differences in arousal ratings for negative sounds were more pronounced for DPD patients (Figure 1B, right panel). However, post-hoc tests within each valence category did not reveal significant differences between groups.

Ratings of mindful breathing effects

The mindful breathing manipulation enhanced the degree of feeling grounded in both groups, F(1,33)=8.85, p<.01, η²=.21. DPD patients felt significantly less grounded than patient controls in both conditions, F(1,33)=32.14, p<.001, η²=.49 (see Figure 2). State depersonalization as assessed by the S-CDS was higher in DPD patients than in patient controls, F(1,34)=61.40, p<.001, η²=.64. As indicated by a significant group × experimental condition interaction, F(1,34)=5.61, p<.01, η²=.14, mindful breathing significantly reduced DP intensity in the DPD group, t(20)=2.99, p<.01, d=0.65, but not in patient controls, t(14)<1, d=0.19 (see Figure 2). With respect to the subjective intensity of the sounds, no significant effect was obtained in the ANOVA. Thus, ratings did not differ between groups and the mindful breathing manipulation did not modulate self-rated intensity ratings.

Download:

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

Figure 2. Effects of mindful breathing.

Average ratings of the degree of grounding, the overall sound intensity, and results of the state version of the Cambridge Depersonalization Scale (S-CDS) as a function of group (DPD patients vs. patient controls) and mindfulness manipulation (normal listening vs. mindful breathing). Error bars indicate SEM.

https://doi.org/10.1371/journal.pone.0074331.g002

Physiological responses to emotional sounds

The ANOVA on the SCR amplitudes for neutral sounds yielded significant main effects of the mindful breathing manipulation, F(1,35)=8.82, p<.01, η²=.20, and the patient group, F(1,35)=5.22, p<.05, η²=.13. Thus, SCR amplitudes were higher for DPD patients and decreased under mindful breathing in both groups (see Figures 3A and 4). The ANOVA for emotional sounds revealed significant main effects of group, F(1,35)=8.44, p<.01, η²=.19, valence category, F(1,35)=14.92, p<.001, η²=.30, and arousal category, F(1,35)=11.19, p<.01, η²=.24. However, these main effects were qualified by significant interactions of group × valence category, F(1,35)=6.21, p<.05, η²=.15, group × arousal category, F(1,35)=4.93, p<.05, η²=.12, and arousal × valence category, F(1,35)=10.02, p<.01, η²=.22. As shown in Figure 4A, DPD patients responded more strongly to the emotional sounds than patients controls. Furthermore, DPD patients showed larger SCR amplitudes to negative as compared to positive sounds, t(21)=3.05, p<.05, d=0.65, and to highly as compared to medium arousing sounds, t(21)=2.84, p<.05, d=0.61. Such differential response was not observed for patient controls, neither for a valence modulation, t(14)<1, d=0.04, nor an arousal modulation, t(14)<1, d=0.05. Across groups, SCR amplitudes were larger for highly as compared to medium arousing negative sounds, t(36)=2.82, p<.05, d=0.64, whereas no such difference was observed for positive sounds, t(36)<1, d=0.07.

Download:

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

Figure 3. Physiological responses elicited by neutral and emotional sounds.

Physiological responses elicited by neutral and emotional sounds as a function of group (DPD patients vs. patient controls) and mindfulness manipulation (normal listening vs. mindful breathing). A) Skin conductance response (SCR) amplitudes, B) phasic heart rate (HR) responses. Error bars indicate SEM, neut = neutral, neg = negative, and pos = positive sounds; -/+ indicate medium and high arousal, respectively.

https://doi.org/10.1371/journal.pone.0074331.g003

Download:

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

Figure 4. Illustration of significant effects in the electrodermal responses.

A) Average skin conductance responses aggregated across valence or arousal categories as a function of group (DPD patients vs. patient controls). B) Average electrodermal responses aggregated across arousal categories as a function of the mindfulness manipulation (normal listening vs. mindful breathing). Error bars indicate SEM, neut = neutral, neg = negative, and pos = positive sounds.

https://doi.org/10.1371/journal.pone.0074331.g004

Finally, we obtained a significant mindful breathing × arousal category interaction, F(1,35)=8.95, p<.01, η²=.20, indicating that highly arousing sounds elicited larger SCR amplitudes than medium arousing sounds in the mindful breathing condition, t(36)=3.22, p<.01, d=0.53, whereas no such global effect was observed under normal listening conditions, t(36)<1, d=0.07 (see Figure 4B).

In the ANOVA on the heart rate responses elicited by neutral sounds, no significant effect was obtained. Thus, HR responses were comparable between groups and unaffected by the mindful breathing manipulation. In the ANOVA on emotional sounds, only a significant main effect of arousal condition was obtained, F(1,32)=4.51, p<.05, η²=.12, indicating that highly arousing sounds were more likely to elicit a HR deceleration than medium arousing sounds (Figure 3B).

Correlation analyses did not reveal significant associations of anxiety or depression scores with electrodermal or heart rate responses, neither in the whole sample nor in the two groups.