Accepted Manuscript Title: Autonomic effects of cognitive reappraisal and acceptance in social anxiety: evidence for common and distinct pathways for parasympathetic reactivity Author: Ioana A. Cristea Gaetano Valenza Enzo Pasquale Scilingo Aurora Szentagotai Tatar Claudio Gentili Daniel David PII: DOI: Reference:
S0887-6185(14)00133-9 http://dx.doi.org/doi:10.1016/j.janxdis.2014.09.009 ANXDIS 1644
To appear in:
Journal of Anxiety Disorders
Received date: Revised date: Accepted date:
21-11-2013 9-7-2014 8-9-2014
Please cite this article as: Cristea, I. A., Valenza, G., Scilingo, E. P., Tatar, A. S., Gentili, C., and David, D.,Autonomic effects of cognitive reappraisal and acceptance in social anxiety: evidence for common and distinct pathways for parasympathetic reactivity, Journal of Anxiety Disorders (2014), http://dx.doi.org/10.1016/j.janxdis.2014.09.009 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Running head: AUTONOMIC EFFECTS OF EMOTION REGULATION IN SOCIAL ANXIETY
ip t
Autonomic effects of cognitive reappraisal and acceptance in social anxiety: evidence for common and distinct pathways for parasympathetic reactivity
Department of Clinical Psychology and Psychotherapy, Babes-Bolyai University, Cluj-Napoca,
an
1
us
Claudio Gentili2, Daniel David1,5
Romania
Clinical Psychology Branch, Department of Surgical, Medical, Molecular and Critical
M
2
d
Pathology, University of Pisa, Italy
Interdepartmental Research Centre "E. Piaggio", Faculty of Engineering, University of Pisa,
te
3
Italy
5
Department of Psychology, Babes-Bolyai University, Cluj-Napoca, Romania
Ac ce p
4
cr
Ioana A. Cristea1,2, Gaetano Valenza3, Enzo Pasquale Scilingo3, Aurora Szentagotai Tatar4,
Department of Oncological Sciences, Icahn School of Medicine at Mount Sinai, New York,
USA
Corresponding author: Ioana A. Cristea
Babes-Bolyai University, Department of Clinical Psychology and Psychotherapy No.37, Republicii St., 400015, Cluj-Napoca, Romania E-mail address: [email protected]; Tel/fax: +40 264 43414
1
Page 1 of 37
Abstract Few studies investigated the effects of emotion regulation strategies on autonomic
ip t
parameters in socially anxious individuals. We asked 99 socially anxious participants to give an impromptu speech in front of an audience in a virtual reality environment. In the anticipation
cr
phase, they practiced an emotion regulation strategy: negative functional reappraisal, acceptance,
us
negative dysfunctional reappraisal. All strategies led to decreases in parasympathetic activity and increases in heart rate during anticipation. Parasympathetic activity remained low in the recovery
an
phase, while heart rate increased, indicating a possible rebound effect of social performance. Exploratory moderation analysis revealed that for subjects with higher social anxiety, acceptance
M
led to increased parasympathetic activity in the anticipation and recovery phases than negative
d
functional reappraisal. Our results indicate that although globally parasympathetic reactivity
te
seems to be a more general marker of simply attempting to regulate emotions, it could help
Ac ce p
distinguish between emotion regulation strategies for some participant subgroups.
Keywords: heart rate variability; parasympathetic reactivity; social anxiety; emotion regulation; acceptance; cognitive reappraisal
2
Page 2 of 37
Highlights
ip t
We compared three emotion regulation strategies for individuals with social anxiety.
cr
We looked at autonomic parameters during a public speaking task in virtual reality.
us
The recovery period after the speech proved a stressful one across subjects.
an
Reappraisal and acceptance diverged for highly socially anxious subjects.
Ac ce p
te
d
M
Adaptive emotion regulation strategies may function differently for social anxiety.
3
Page 3 of 37
ip t cr us
1. Introduction
an
1.1. Emotion regulation and parasympathetic reactivity
The links between regulated emotional responding, psychopathology and the functioning
M
of the human heart represent an important and contentious topic in affective research. The normal variability in the activity of the human heart, as expressed in the fluctuations
d
of the heart rate signal, is the result of the joint actions at the sinoatrial node of two branches of
te
the autonomic nervous system (ANS), the sympathetic and the parasympathetic branches
Ac ce p
(Berntson et al., 1997). At rest, the parasympathetic one dominates, ensuring that energy balance is preserved. When physical or psychological stress ensues, the sympathetic branch takes over, producing physiological arousal to sustain the adaptation to the stressor. Parasympathetic influences over the heart (in the order of milliseconds) are faster than sympathetic ones (in the order of seconds) and thus more capable of producing rapid adaptive changes in the beat-to-beat timing (Berntson et al., 1997).
The activity of the parasympathetic branch can be indexed by variations in the high frequency bandwidth of the heart rate signal, which is also the frequency range corresponding to respiration (0.15 to 0.4. Hz). As such, high frequency heart rate variability (HF-HRV) is 4
Page 4 of 37
considered a reliable marker of the reactivity of the parasympathetic nervous wing of the ANS (Berntson et al., 1997; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). Moreover, a recent review that
ip t
comprehensively re-analyses previous HRV studies argues that “the suitability of HRV analysis is restricted to the estimation of parasympathetic influences on HR, whereas further
us
del Paso, Langewitz, Mulder, van Roon, & Duschek, 2013, p.484).
cr
interpretations of spectral components as extravagal have to be regarded as misleading” (Reyes
.
an
It has been argued that HF-HRV could be an important resource in coping with emotions, supporting flexible adaptation to changing environmental pressures (Thayer & Lane, 2009) and
M
an important counterpart of emotion regulation (Appelhans & Luecken, 2006). Moreover, the autonomic influences over the heart rate reflect cortical control. Neuroimaging studies in healthy
d
subjects offered evidence that the activity of the medial prefrontal cortex is associated to the
te
vagal function (Gianaros, Van Der Veen, & Jennings, 2004; Lane et al., 2009). In a meta-
Ac ce p
analysis, Thayer, Ahs, Fredrikson, Sollers, & Wager (2012) showed that several brain areas such as the amygdala and the ventromedial prefrontal cortex, key structures in emotion regulation (Urry et al., 2006; Wager, Davidson, Hughes, Lindquist, & Ochsner, 2008), are also associated with HF-HRV.
Tonic/trait and phasic/state HF-HRV have been considered in relationship to both healthy, regulated emotional responding, and psychopathology. A broad consensus exists regarding tonic HF-HRV, with low resting levels being associated with poor psychological health, reflected in a variety of mood and anxiety disorders, and high resting levels with context appropriate and regulated emotional responses (Libby, Worhunsky, Pilver, & Brewer, 2012; 5
Page 5 of 37
Thayer & Brosschot, 2005). However, things are far less clear regarding the association between phasic modifications in HF-HRV and emotion regulation. Interestingly, recent studies reported that for subjects at risk or affected by anxiety,
ip t
regulation strategies typically considered adaptive (e.g., reappraisal) seem to be accompanied by a decrease of parasympathetic activity during exposure or recovery from stressor, a reverse
cr
pattern from what is observed in normal, healthy controls. Di Simplicio et al. (2012) showed that
us
subjects high on neuroticism presented a pattern of decrease in HF-HRV during reappraisal of negative emotional pictures, contrary to what was reported for those low on neuroticism. Using
an
another index of parasympathetic activity (MSD- mean square difference of the successive interbeat intervals) Aldao and Mennin (2012) showed that both cognitive reappraisal and
M
acceptance were accompanied by decreases in autonomic activity during the recovery phase of
d
exposure to an emotion eliciting stimuli for individuals with generalized anxiety, an opposite
te
pattern of what was observed in healthy subjects. Another surprising observation comes from studies looking at another marker of
Ac ce p
parasympathetic activity, closely related to HF-HRV, namely respiratory sinus arrhythmia. Putatively adaptive regulation strategies (see Aldao & Nolen-Hoeksema, 2012 for a discussion) seem to not differ from the ones considered maladaptive in their impact on parasympathetic reactivity. Butler, Wilhelm, and Gross (2006) showed that both cognitive reappraisal and suppression were accompanied by increases in RSA in healthy individuals. In a similar study on participants with anxiety and mood disorders, Campbell-Sills, Barlow, Brown, and Hofmann (2006) again found no differences between another adaptive strategy (acceptance) and suppression on state changes in RSA. 1.2. Study overview 6
Page 6 of 37
Starting from these few but intriguing results, the goal of our study was to investigate the effects of putatively adaptive regulation strategies on autonomic activity in a clinical condition. Social anxiety (SA) provided an interesting psychopathological model for these comparisons.
ip t
Firstly, it was closely linked to difficulties in emotional regulation (Mennin, McLaughlin, &
Flanagan, 2009; Salovey, Stroud, Woolery, & Epel, 2002; Turk, Heimberg, Luterek, Mennin, &
cr
Fresco, 2005). Secondly, studies looking at the efficiency of emotion regulation strategies per se
us
(i.e. not as integrated parts of therapy protocols) for socially anxious subjects have been scarce. The two studies conducted on this topic (Goldin, Manber, Hakimi, Canli, & Gross, 2009; Goldin,
an
Manber-Ball, Werner, Heimberg, & Gross, 2009), both fMRI studies, indicated that while SA patients are capable of using cognitive reappraisal to reduce negative emotions, their underlying
M
brain patterns are different from those of normal controls performing the same task. Emotion
d
regulation following social threat in these subjects resulted in reduced activation, later and fewer
te
brain responses in regions considered key to cognitive control and reappraisal (dorsomedial and dorsolateral prefrontal cortex, anterior cingulate). Thirdly, we note that brain regions found to
Ac ce p
display reduced or deficient activity during emotion regulation in SA patients were also associated with autonomic reactivity (Ahs, Sollers, Furmark, Fredrikson, & Thayer, 2009). As adaptive emotion regulation strategies, we focused on cognitive reappraisal and acceptance (Aldao & Nolen-Hoeksema, 2012), which have both been also associated with widespread, major therapeutic approaches. Reappraisal involves the modification of dysfunctional cognitions that sustain psychological distress (Clark, 1999) and is recognized as one of the main active ingredients of classic cognitive-behavioral therapy/CBT (Hofmann & Asmundson, 2008). Acceptance refers to paying attention to one’s experiences in the present moment in an open, nonreactive, accepting, and nonjudgmental way (Kabat-Zinn, 1994). It is considered central in 7
Page 7 of 37
more recent cognitive-behavioral therapeutic approaches such as acceptance and commitment therapy/ACT (S. C. Hayes, Strosahl, & Wilson, 2012). We set to contrast these two strategies for socially anxious individuals in a public speaking situation, as well as compare them with a
ip t
control condition mimicking the typical dysfunctional thoughts socially anxious people have in this situation.
cr
To increase the clinical validity of the study, regulation instructions were constructed to
us
represent beliefs about the public speaking situation, similar to what is done in sessions of CBT. We also used an ecological form of reappraisal, closely informed by cognitive-behavioral
an
therapy protocols: negative functional reappraisal (Cristea, Szentagotai Tatar, Nagy, & David, 2012; Ellis, 1994). In this framework, the emotional situation maintains its negative characters,
M
which is reformulated in more adaptive- albeit still negative- terms. Our previous work has
d
shown this strategy to be efficient in reducing momentary distress in healthy individuals (Cristea
te
et al., 2012; Cristea, Matu, Szentagotai Tatar, & David, 2012) and it could be closer to how people tend to respond to negative events than detachment or positive thinking.
Ac ce p
For emotion induction, we focused on a public speaking task, which, along with its discrete phases (i.e., anticipation, speech, recovery), has demonstrated high relevance for SA. We modified the procedure to use a virtual reality (VR) environment in which subjects gave an impromptu speech in front of a virtual audience. Previous studies showed this particular VR environment reliably induced anxiety in individuals vulnerable to socially evaluative situations (Cornwell, Heller, Biggs, Pine, & Grillon, 2011; Cornwell, Johnson, Berardi, & Grillon, 2006). Given recent reconsiderations of spectral analysis of heart rate variability as reliably reflecting solely parasympathetic influences, we looked at two markers of autonomic activity: parasympathetic reactivity (indexed by HF-HRV), and heart rate- the resultant of the joint action 8
Page 8 of 37
of the sympathetic and parasympathetic branches. As the focus of our study was on autonomic changes associated with emotion regulation strategies in social anxiety, we did not include selfreport measures of subjective anxiety before and after each phase. This was also done in order to
ip t
minimize demand characteristics, present in most emotion regulation studies and unavoidable with repeated applications of the same measure, as well as to avoid the distraction of our subjects
cr
by completing a scale between the segments of the task.
us
The three emotion regulation strategies proposed have not been compared before for social anxiety. Also, the negative dysfunctional reappraisal strategy (Dysfunctional) we
an
introduced has not studied before in the emotion regulation literature. Based on its experimental conceptualization and on the available literature on cognitive reappraisal and acceptance, we
M
expected the Dysfunctional group to have worse autonomic results than both the Reappraisal and
te
parasympathetic activity.
d
Acceptance group, more specifically to be associated with reduced heart rate and increased
We were able to find only four studies that compared cognitive reappraisal and
Ac ce p
acceptance (Cristea, Matu, et al., 2012; Hofmann, Heering, Sawyer, & Asnaani, 2009; Rood, Roelofs, Bögels, & Arntz, 2012; Wolgast, Lundh, & Viborg, 2011) for and only two of these used autonomic outcomes. Hofmann et al. (2009) found no differences between the strategies of Reappraisal and Acceptance on HR changes from baseline in a public speaking task. Wolgast et al. (2011) found higher skin conductance levels (an index of sympathetic activity) in the Acceptance condition as contrasted to the Reappraisal one in response to negative film clips. Therefore, we did not expect to find differences between Reappraisal and Acceptance on heart rate. We did not formulate specific hypotheses regarding their impact on parasympathetic activity, since this marker was not approached in previous studies. 9
Page 9 of 37
2. Method 2.1. Participants
ip t
Ninety nine participants (7 men; 92 women; Mean age= 20.19, SD= 2.25) participated in the experiment. They were recruited through online ads and e-mails from the student population
cr
at the university of the first author. Out of the 191 respondents to the invitation to take part in
us
the study, we selected those scoring over the clinical threshold of 30 on the Liebowitz Social Anxiety Scale.
an
All of them were undergraduate students and of Romanian nationality. Their ethnicity was unanimously White Caucasian. None of the participants had had any previous experience
M
with therapy. Participants were screened regarding the presence of diagnosed cardiac problems
d
and the use of cardiac medication. None of them reported either of these.
te
Participation was voluntary and subjects received course credit in compensation for their involvement. Written informed consent was obtained from all the participants after the procedure
Ac ce p
had been explained. The study was approved by the Institutional Review Board of the first author’s university. 2.2. Measures
Liebowitz Social Anxiety Scale- Self-Report (LSAS-SR; Liebowitz, 1987; Fresco et al., 2001) is an instrument designed to measure social anxiety by assessing the fear and avoidance individuals might experience in social interaction and performance situations. It consists of 24 items, each of them rated separately on two 0 to 3 Likert scales for the intensity of fear and the frequency of avoidant behaviors. An overall score is calculated by summing the fear and avoidance scores, and this index is the one most commonly employed in clinical trials of social 10
Page 10 of 37
phobia (Heimberg et al., 1999). Excellent reliability and validity were reported for the LSAS, as well as sensitivity to pharmacological treatments over time (Heimberg et al., 1999). Results of validation studies of the LSAS-SR found little difference in psychometric indexes with the
ip t
clinician-based version of the instrument, both on scale or subscale scores. A cut-off point of 30 was shown to be indicative of a diagnosis of social phobia, and a cut-off point of 60 for the
cr
generalized sub-type of social phobia (Mennin et al., 2002; Rytwinski et al., 2009). We used the
us
self-report version of the LSAS, which was translated into Romanian. Data indicate excellent reliability (Cronbach’s alpha of 0.93).
an
The State Trait Anxiety Scale (Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983) asks participants to evaluate how they feel “right now”, and respectively how they feel “in
M
general”, by rating twenty statements regarding mood in terms of their perceived intensity (not at
d
all, a little, pretty much, a lot). We used these scales to assess the subjects’ general anxiety, as
te
well as their anxiety level at baseline, before starting the experiment. The STAI was adapted on the Romanian population (Pitariu & Peleasa, 2007).
Ac ce p
Physiological measures
Cardiac data was acquired using the ECG100C Electrocardiogram Amplifier (MP150: Biopac Systems Inc., US), at a sampling rate of 1 kHz, which recorded the D2 lead ECG signal (bandwidth: 0.05-35 Hz), connected with pre-gelled Ag/AgCl electrodes placed according to the Einthoven triangle configuration. The Matlab package was used for data analysis. ECG was pre-filtered through a Moving Average Filter (MAF) in order to extract and subtract the baseline. The ECG signal was used to extract the HRV (Task Force, 1996). PanTompkins’ automatic algorithm (Pan & Tompkins, 1985) was used for R-peak detection. We corrected for technical artefacts using a proper piecewise cubic spline interpolation method 11
Page 11 of 37
(Lippman, Stein, & Lerman, 1993). We also manually checked for physiological artefacts (e.g. ectopic beats) and only artefact-free sections were included in the analysis. In order to ensure
m 2sd range, thus excluding 9 HRV signals.
ip t
reliability, we rejected those HRV signals which presented more than 20% of samples out of the
Regarding the time domain, we computed the heart rate (HR), which was estimated as:
cr
HR 60 / t RR , where t RR represents the interval between two R-peaks.
us
For the frequency domain, given a reliable signal, the Power Spectral Density (PSD) of the HRV was estimated by means of an Auto-Regressive (AR) model. This non-parametric
an
approach provides a better frequency resolution than the other methods (e.g., Fourier
M
transformation). Furthermore, conventional frequency transformation based on the Fourier transform technique is not very suitable for analyzing non-stationary signals. We used the Burg
d
method to get the AR model parameters, according to the results presented by Akaike (1969).
te
This method provided high resolution in frequency and yielded a stable AR model. The HF power was calculated by integrating the spectral power density in the bandwidth 0.15–0.4 Hz.
Ac ce p
2.3. Procedure
Participants were told the purpose of the study was to see how individuals reacted in a task performed in a virtual reality environment. No details regarding the task or the instruction were offered beforehand. They were randomly assigned to one of the three groups, corresponding to each emotion regulation instruction. Participants underwent four stages within a single experimental session. In the baseline phase, physiological data were recorded for 5 minutes after a short habituation period with the sensors. In the anticipation-instruction phase (5 minutes), subjects were told they had to give a speech in front of a virtual audience on a topic that would be announced to them just before the 12
Page 12 of 37
actual speech. In the meanwhile they were given a written instruction corresponding to their experimental condition and were told to read it carefully, reflect on it and practice it in expectation of giving the speech. At the end of this phase, the head-mounted display was
ip t
installed and subjects were immersed in the virtual reality environment. The VR environment (Virtual Classroom; Grapp, 2012) consisted of a virtual audience arranged in a medium sized
cr
classroom (15-20 individuals), in which the participant took the position of the speaker at the
us
podium, in front of the audience. The audience was oriented so that it was looking at the speaker. Subjects were given a short habituation period with the head-mounted display, after which they
an
were announced a topic and were asked to speak on it for 3 minutes in front of the virtual audience (speech phase). A list of speech topics on controversial social, economic, political
M
issues (e.g., violent computer games should be banned) was constructed and each subject got a
d
different topic. Afterwards, subjects were let to rest for 3 minutes, with electrophysiological
te
sensors still attached (recovery phase). Subjects were then debriefed. Subjects were sitting down all throughout the experiment for reasons related to HRV measurement.
Ac ce p
Each of the emotion regulation instructions (see Appendix A) consisted of beliefs about
the public speaking situation. In the Dysfunctional (negative dysfunctional reappraisal) group, participants were given a set of irrational beliefs, which mimicked the ones a socially anxious person could hold in a socially evaluative situation, such as the one awaiting (e.g., “I might not have any idea what to talk about and say stupid things. This would be a terrible and catastrophic thing.”). In the Reappraisal (negative functional reappraisal) group, they received a set of rational beliefs as to how to evaluate the situation, which were similar to the ones a socially anxious client would be presented with during therapy, to replace the dysfunctional negative thoughts about the situation (e.g., I might not have an idea what to talk about and say stupid 13
Page 13 of 37
things. This would be bad, but I would not be so terrible and catastrophic.). In the Acceptance group, they got an acceptance-based rationale, stressing on the idea of accepting to remain in the present moment experiencing the thoughts and fears that arise, without trying to modify them
cr
accept to just experience them and stay with them here and now.).
ip t
(e.g., I have thoughts about how I will not do well and not be able to handle the situation, but I
us
3. Results 3.1. Baseline differences
an
The sample had mean social anxiety values of 48.58 (SD= 14.54) and mean general anxiety values of 44.14 (SD= 10.58). Means and standard deviations for the cardiac parameters
M
extracted are displayed in Table 1. At baseline there were no significant differences (all ps >.05)
d
between the groups on measures of social anxiety, self reported anxiety or any of the
te
physiological parameters extracted. Non-parametric correlation analysis (Spearman’s rho) between self-report variables at baseline and autonomic activity revealed a significant positive
Ac ce p
correlation across groups between trait anxiety and parasympathetic activity in the recovery phase (r=.220, p= .037). All other correlations were close to 0 and non-significant. We excluded the speech phase from the subsequent analysis as it most likely elicited substantial differences in respiration as compared to the other phases, which may have in turn impacted the HRV measures considered (see Ritz & Dahme, 2006 for a discussion). Table 1. Mean HRV measures during different task conditions. Measure
Group
Dysfunctional a
Reappraisal a (n=
Acceptancea
14
Page 14 of 37
/Study
(n= 30)
33)
(n=27)
Baseline
47.61 (11.43)
47.62 (17.23)
50.59 (14.53)
Baseline
43.87 (10.43)
44.97 (11.95)
Baseline
37.24 (11.96)
41.14 (12.15)
39.40 (11.49)
Baseline
82.26 (13.44)
87.75 (10.57)
83.45 (12.97)
Anticipation
93.24 (13.81)
97.7 (9.83)
96.05 (16.13)
103.11 (14.62)
103.67 (18.47)
864.88 (874.35)
976.49 (815.85)
615.18 (605.75)
449 (408.68)
584.05 (570.62)
651.95 (516.13)
411.29 (314.41)
549.77 (603.36)
Social Anxiety (LSAS) Trait Anxiety
Instruction
M
HR (bpm)
an
(STAI-S)
99.2 (16.8)
d
Recovery Baseline
1264.6 (1427.55)
te
HF (ms2)
us
State Anxiety
Anticipation
43.53 (9.41)
cr
(STAI-T)
ip t
Phase
Ac ce p
Instruction Recovery
Note. LSAS, Liebowitz Social Anxiety Scale; STAI, State Trait Anxiety Scale; HRV, heart rate variability; HR, Heart rate (bpm; beats/minute); HF, high frequency (ms2). Values represent means and standard deviations (in parantheses). a For self-report values n= 33 for the Dysfunctional Group, n= 34 for Reappraisal and n= 32 for Acceptance 3.2. Power analysis 15
Page 15 of 37
We conducted a post-hoc power analysis using the G*Power software. Our focus was repeated measures ANOVA looking for within between interactions. We assumed a p of 0.05, N= 90 (the number of subjects that had data for HF-HRV and HR) and 3 repeated measures. We
ip t
assumed a medium correlation among repeated measures of 0.50 (in reality this was a
conservative estimation since correlations among repeated measures were much higher for both
cr
HF-HRV and HR, ranging from 0.67 to 0.84). Power analysis indicated we would be able to
us
detect interaction effects as small as 0.18. 3.3. Group differences
an
Parasympathetic activity
We conducted a 3 (Group: Dysfunctional, Reappraisal, Acceptance) by 3 (Time:
M
baseline, anticipation-instruction, recovery) repeated measures ANOVA on HF-HRV,. Multivariate tests (Wilks’ Lambda) indicated a significant effect of Time, F(2,86)= 20.42,
te
d
p
S0887-6185(14)00133-9 http://dx.doi.org/doi:10.1016/j.janxdis.2014.09.009 ANXDIS 1644
To appear in:
Journal of Anxiety Disorders
Received date: Revised date: Accepted date:
21-11-2013 9-7-2014 8-9-2014
Please cite this article as: Cristea, I. A., Valenza, G., Scilingo, E. P., Tatar, A. S., Gentili, C., and David, D.,Autonomic effects of cognitive reappraisal and acceptance in social anxiety: evidence for common and distinct pathways for parasympathetic reactivity, Journal of Anxiety Disorders (2014), http://dx.doi.org/10.1016/j.janxdis.2014.09.009 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Running head: AUTONOMIC EFFECTS OF EMOTION REGULATION IN SOCIAL ANXIETY
ip t
Autonomic effects of cognitive reappraisal and acceptance in social anxiety: evidence for common and distinct pathways for parasympathetic reactivity
Department of Clinical Psychology and Psychotherapy, Babes-Bolyai University, Cluj-Napoca,
an
1
us
Claudio Gentili2, Daniel David1,5
Romania
Clinical Psychology Branch, Department of Surgical, Medical, Molecular and Critical
M
2
d
Pathology, University of Pisa, Italy
Interdepartmental Research Centre "E. Piaggio", Faculty of Engineering, University of Pisa,
te
3
Italy
5
Department of Psychology, Babes-Bolyai University, Cluj-Napoca, Romania
Ac ce p
4
cr
Ioana A. Cristea1,2, Gaetano Valenza3, Enzo Pasquale Scilingo3, Aurora Szentagotai Tatar4,
Department of Oncological Sciences, Icahn School of Medicine at Mount Sinai, New York,
USA
Corresponding author: Ioana A. Cristea
Babes-Bolyai University, Department of Clinical Psychology and Psychotherapy No.37, Republicii St., 400015, Cluj-Napoca, Romania E-mail address: [email protected]; Tel/fax: +40 264 43414
1
Page 1 of 37
Abstract Few studies investigated the effects of emotion regulation strategies on autonomic
ip t
parameters in socially anxious individuals. We asked 99 socially anxious participants to give an impromptu speech in front of an audience in a virtual reality environment. In the anticipation
cr
phase, they practiced an emotion regulation strategy: negative functional reappraisal, acceptance,
us
negative dysfunctional reappraisal. All strategies led to decreases in parasympathetic activity and increases in heart rate during anticipation. Parasympathetic activity remained low in the recovery
an
phase, while heart rate increased, indicating a possible rebound effect of social performance. Exploratory moderation analysis revealed that for subjects with higher social anxiety, acceptance
M
led to increased parasympathetic activity in the anticipation and recovery phases than negative
d
functional reappraisal. Our results indicate that although globally parasympathetic reactivity
te
seems to be a more general marker of simply attempting to regulate emotions, it could help
Ac ce p
distinguish between emotion regulation strategies for some participant subgroups.
Keywords: heart rate variability; parasympathetic reactivity; social anxiety; emotion regulation; acceptance; cognitive reappraisal
2
Page 2 of 37
Highlights
ip t
We compared three emotion regulation strategies for individuals with social anxiety.
cr
We looked at autonomic parameters during a public speaking task in virtual reality.
us
The recovery period after the speech proved a stressful one across subjects.
an
Reappraisal and acceptance diverged for highly socially anxious subjects.
Ac ce p
te
d
M
Adaptive emotion regulation strategies may function differently for social anxiety.
3
Page 3 of 37
ip t cr us
1. Introduction
an
1.1. Emotion regulation and parasympathetic reactivity
The links between regulated emotional responding, psychopathology and the functioning
M
of the human heart represent an important and contentious topic in affective research. The normal variability in the activity of the human heart, as expressed in the fluctuations
d
of the heart rate signal, is the result of the joint actions at the sinoatrial node of two branches of
te
the autonomic nervous system (ANS), the sympathetic and the parasympathetic branches
Ac ce p
(Berntson et al., 1997). At rest, the parasympathetic one dominates, ensuring that energy balance is preserved. When physical or psychological stress ensues, the sympathetic branch takes over, producing physiological arousal to sustain the adaptation to the stressor. Parasympathetic influences over the heart (in the order of milliseconds) are faster than sympathetic ones (in the order of seconds) and thus more capable of producing rapid adaptive changes in the beat-to-beat timing (Berntson et al., 1997).
The activity of the parasympathetic branch can be indexed by variations in the high frequency bandwidth of the heart rate signal, which is also the frequency range corresponding to respiration (0.15 to 0.4. Hz). As such, high frequency heart rate variability (HF-HRV) is 4
Page 4 of 37
considered a reliable marker of the reactivity of the parasympathetic nervous wing of the ANS (Berntson et al., 1997; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). Moreover, a recent review that
ip t
comprehensively re-analyses previous HRV studies argues that “the suitability of HRV analysis is restricted to the estimation of parasympathetic influences on HR, whereas further
us
del Paso, Langewitz, Mulder, van Roon, & Duschek, 2013, p.484).
cr
interpretations of spectral components as extravagal have to be regarded as misleading” (Reyes
.
an
It has been argued that HF-HRV could be an important resource in coping with emotions, supporting flexible adaptation to changing environmental pressures (Thayer & Lane, 2009) and
M
an important counterpart of emotion regulation (Appelhans & Luecken, 2006). Moreover, the autonomic influences over the heart rate reflect cortical control. Neuroimaging studies in healthy
d
subjects offered evidence that the activity of the medial prefrontal cortex is associated to the
te
vagal function (Gianaros, Van Der Veen, & Jennings, 2004; Lane et al., 2009). In a meta-
Ac ce p
analysis, Thayer, Ahs, Fredrikson, Sollers, & Wager (2012) showed that several brain areas such as the amygdala and the ventromedial prefrontal cortex, key structures in emotion regulation (Urry et al., 2006; Wager, Davidson, Hughes, Lindquist, & Ochsner, 2008), are also associated with HF-HRV.
Tonic/trait and phasic/state HF-HRV have been considered in relationship to both healthy, regulated emotional responding, and psychopathology. A broad consensus exists regarding tonic HF-HRV, with low resting levels being associated with poor psychological health, reflected in a variety of mood and anxiety disorders, and high resting levels with context appropriate and regulated emotional responses (Libby, Worhunsky, Pilver, & Brewer, 2012; 5
Page 5 of 37
Thayer & Brosschot, 2005). However, things are far less clear regarding the association between phasic modifications in HF-HRV and emotion regulation. Interestingly, recent studies reported that for subjects at risk or affected by anxiety,
ip t
regulation strategies typically considered adaptive (e.g., reappraisal) seem to be accompanied by a decrease of parasympathetic activity during exposure or recovery from stressor, a reverse
cr
pattern from what is observed in normal, healthy controls. Di Simplicio et al. (2012) showed that
us
subjects high on neuroticism presented a pattern of decrease in HF-HRV during reappraisal of negative emotional pictures, contrary to what was reported for those low on neuroticism. Using
an
another index of parasympathetic activity (MSD- mean square difference of the successive interbeat intervals) Aldao and Mennin (2012) showed that both cognitive reappraisal and
M
acceptance were accompanied by decreases in autonomic activity during the recovery phase of
d
exposure to an emotion eliciting stimuli for individuals with generalized anxiety, an opposite
te
pattern of what was observed in healthy subjects. Another surprising observation comes from studies looking at another marker of
Ac ce p
parasympathetic activity, closely related to HF-HRV, namely respiratory sinus arrhythmia. Putatively adaptive regulation strategies (see Aldao & Nolen-Hoeksema, 2012 for a discussion) seem to not differ from the ones considered maladaptive in their impact on parasympathetic reactivity. Butler, Wilhelm, and Gross (2006) showed that both cognitive reappraisal and suppression were accompanied by increases in RSA in healthy individuals. In a similar study on participants with anxiety and mood disorders, Campbell-Sills, Barlow, Brown, and Hofmann (2006) again found no differences between another adaptive strategy (acceptance) and suppression on state changes in RSA. 1.2. Study overview 6
Page 6 of 37
Starting from these few but intriguing results, the goal of our study was to investigate the effects of putatively adaptive regulation strategies on autonomic activity in a clinical condition. Social anxiety (SA) provided an interesting psychopathological model for these comparisons.
ip t
Firstly, it was closely linked to difficulties in emotional regulation (Mennin, McLaughlin, &
Flanagan, 2009; Salovey, Stroud, Woolery, & Epel, 2002; Turk, Heimberg, Luterek, Mennin, &
cr
Fresco, 2005). Secondly, studies looking at the efficiency of emotion regulation strategies per se
us
(i.e. not as integrated parts of therapy protocols) for socially anxious subjects have been scarce. The two studies conducted on this topic (Goldin, Manber, Hakimi, Canli, & Gross, 2009; Goldin,
an
Manber-Ball, Werner, Heimberg, & Gross, 2009), both fMRI studies, indicated that while SA patients are capable of using cognitive reappraisal to reduce negative emotions, their underlying
M
brain patterns are different from those of normal controls performing the same task. Emotion
d
regulation following social threat in these subjects resulted in reduced activation, later and fewer
te
brain responses in regions considered key to cognitive control and reappraisal (dorsomedial and dorsolateral prefrontal cortex, anterior cingulate). Thirdly, we note that brain regions found to
Ac ce p
display reduced or deficient activity during emotion regulation in SA patients were also associated with autonomic reactivity (Ahs, Sollers, Furmark, Fredrikson, & Thayer, 2009). As adaptive emotion regulation strategies, we focused on cognitive reappraisal and acceptance (Aldao & Nolen-Hoeksema, 2012), which have both been also associated with widespread, major therapeutic approaches. Reappraisal involves the modification of dysfunctional cognitions that sustain psychological distress (Clark, 1999) and is recognized as one of the main active ingredients of classic cognitive-behavioral therapy/CBT (Hofmann & Asmundson, 2008). Acceptance refers to paying attention to one’s experiences in the present moment in an open, nonreactive, accepting, and nonjudgmental way (Kabat-Zinn, 1994). It is considered central in 7
Page 7 of 37
more recent cognitive-behavioral therapeutic approaches such as acceptance and commitment therapy/ACT (S. C. Hayes, Strosahl, & Wilson, 2012). We set to contrast these two strategies for socially anxious individuals in a public speaking situation, as well as compare them with a
ip t
control condition mimicking the typical dysfunctional thoughts socially anxious people have in this situation.
cr
To increase the clinical validity of the study, regulation instructions were constructed to
us
represent beliefs about the public speaking situation, similar to what is done in sessions of CBT. We also used an ecological form of reappraisal, closely informed by cognitive-behavioral
an
therapy protocols: negative functional reappraisal (Cristea, Szentagotai Tatar, Nagy, & David, 2012; Ellis, 1994). In this framework, the emotional situation maintains its negative characters,
M
which is reformulated in more adaptive- albeit still negative- terms. Our previous work has
d
shown this strategy to be efficient in reducing momentary distress in healthy individuals (Cristea
te
et al., 2012; Cristea, Matu, Szentagotai Tatar, & David, 2012) and it could be closer to how people tend to respond to negative events than detachment or positive thinking.
Ac ce p
For emotion induction, we focused on a public speaking task, which, along with its discrete phases (i.e., anticipation, speech, recovery), has demonstrated high relevance for SA. We modified the procedure to use a virtual reality (VR) environment in which subjects gave an impromptu speech in front of a virtual audience. Previous studies showed this particular VR environment reliably induced anxiety in individuals vulnerable to socially evaluative situations (Cornwell, Heller, Biggs, Pine, & Grillon, 2011; Cornwell, Johnson, Berardi, & Grillon, 2006). Given recent reconsiderations of spectral analysis of heart rate variability as reliably reflecting solely parasympathetic influences, we looked at two markers of autonomic activity: parasympathetic reactivity (indexed by HF-HRV), and heart rate- the resultant of the joint action 8
Page 8 of 37
of the sympathetic and parasympathetic branches. As the focus of our study was on autonomic changes associated with emotion regulation strategies in social anxiety, we did not include selfreport measures of subjective anxiety before and after each phase. This was also done in order to
ip t
minimize demand characteristics, present in most emotion regulation studies and unavoidable with repeated applications of the same measure, as well as to avoid the distraction of our subjects
cr
by completing a scale between the segments of the task.
us
The three emotion regulation strategies proposed have not been compared before for social anxiety. Also, the negative dysfunctional reappraisal strategy (Dysfunctional) we
an
introduced has not studied before in the emotion regulation literature. Based on its experimental conceptualization and on the available literature on cognitive reappraisal and acceptance, we
M
expected the Dysfunctional group to have worse autonomic results than both the Reappraisal and
te
parasympathetic activity.
d
Acceptance group, more specifically to be associated with reduced heart rate and increased
We were able to find only four studies that compared cognitive reappraisal and
Ac ce p
acceptance (Cristea, Matu, et al., 2012; Hofmann, Heering, Sawyer, & Asnaani, 2009; Rood, Roelofs, Bögels, & Arntz, 2012; Wolgast, Lundh, & Viborg, 2011) for and only two of these used autonomic outcomes. Hofmann et al. (2009) found no differences between the strategies of Reappraisal and Acceptance on HR changes from baseline in a public speaking task. Wolgast et al. (2011) found higher skin conductance levels (an index of sympathetic activity) in the Acceptance condition as contrasted to the Reappraisal one in response to negative film clips. Therefore, we did not expect to find differences between Reappraisal and Acceptance on heart rate. We did not formulate specific hypotheses regarding their impact on parasympathetic activity, since this marker was not approached in previous studies. 9
Page 9 of 37
2. Method 2.1. Participants
ip t
Ninety nine participants (7 men; 92 women; Mean age= 20.19, SD= 2.25) participated in the experiment. They were recruited through online ads and e-mails from the student population
cr
at the university of the first author. Out of the 191 respondents to the invitation to take part in
us
the study, we selected those scoring over the clinical threshold of 30 on the Liebowitz Social Anxiety Scale.
an
All of them were undergraduate students and of Romanian nationality. Their ethnicity was unanimously White Caucasian. None of the participants had had any previous experience
M
with therapy. Participants were screened regarding the presence of diagnosed cardiac problems
d
and the use of cardiac medication. None of them reported either of these.
te
Participation was voluntary and subjects received course credit in compensation for their involvement. Written informed consent was obtained from all the participants after the procedure
Ac ce p
had been explained. The study was approved by the Institutional Review Board of the first author’s university. 2.2. Measures
Liebowitz Social Anxiety Scale- Self-Report (LSAS-SR; Liebowitz, 1987; Fresco et al., 2001) is an instrument designed to measure social anxiety by assessing the fear and avoidance individuals might experience in social interaction and performance situations. It consists of 24 items, each of them rated separately on two 0 to 3 Likert scales for the intensity of fear and the frequency of avoidant behaviors. An overall score is calculated by summing the fear and avoidance scores, and this index is the one most commonly employed in clinical trials of social 10
Page 10 of 37
phobia (Heimberg et al., 1999). Excellent reliability and validity were reported for the LSAS, as well as sensitivity to pharmacological treatments over time (Heimberg et al., 1999). Results of validation studies of the LSAS-SR found little difference in psychometric indexes with the
ip t
clinician-based version of the instrument, both on scale or subscale scores. A cut-off point of 30 was shown to be indicative of a diagnosis of social phobia, and a cut-off point of 60 for the
cr
generalized sub-type of social phobia (Mennin et al., 2002; Rytwinski et al., 2009). We used the
us
self-report version of the LSAS, which was translated into Romanian. Data indicate excellent reliability (Cronbach’s alpha of 0.93).
an
The State Trait Anxiety Scale (Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983) asks participants to evaluate how they feel “right now”, and respectively how they feel “in
M
general”, by rating twenty statements regarding mood in terms of their perceived intensity (not at
d
all, a little, pretty much, a lot). We used these scales to assess the subjects’ general anxiety, as
te
well as their anxiety level at baseline, before starting the experiment. The STAI was adapted on the Romanian population (Pitariu & Peleasa, 2007).
Ac ce p
Physiological measures
Cardiac data was acquired using the ECG100C Electrocardiogram Amplifier (MP150: Biopac Systems Inc., US), at a sampling rate of 1 kHz, which recorded the D2 lead ECG signal (bandwidth: 0.05-35 Hz), connected with pre-gelled Ag/AgCl electrodes placed according to the Einthoven triangle configuration. The Matlab package was used for data analysis. ECG was pre-filtered through a Moving Average Filter (MAF) in order to extract and subtract the baseline. The ECG signal was used to extract the HRV (Task Force, 1996). PanTompkins’ automatic algorithm (Pan & Tompkins, 1985) was used for R-peak detection. We corrected for technical artefacts using a proper piecewise cubic spline interpolation method 11
Page 11 of 37
(Lippman, Stein, & Lerman, 1993). We also manually checked for physiological artefacts (e.g. ectopic beats) and only artefact-free sections were included in the analysis. In order to ensure
m 2sd range, thus excluding 9 HRV signals.
ip t
reliability, we rejected those HRV signals which presented more than 20% of samples out of the
Regarding the time domain, we computed the heart rate (HR), which was estimated as:
cr
HR 60 / t RR , where t RR represents the interval between two R-peaks.
us
For the frequency domain, given a reliable signal, the Power Spectral Density (PSD) of the HRV was estimated by means of an Auto-Regressive (AR) model. This non-parametric
an
approach provides a better frequency resolution than the other methods (e.g., Fourier
M
transformation). Furthermore, conventional frequency transformation based on the Fourier transform technique is not very suitable for analyzing non-stationary signals. We used the Burg
d
method to get the AR model parameters, according to the results presented by Akaike (1969).
te
This method provided high resolution in frequency and yielded a stable AR model. The HF power was calculated by integrating the spectral power density in the bandwidth 0.15–0.4 Hz.
Ac ce p
2.3. Procedure
Participants were told the purpose of the study was to see how individuals reacted in a task performed in a virtual reality environment. No details regarding the task or the instruction were offered beforehand. They were randomly assigned to one of the three groups, corresponding to each emotion regulation instruction. Participants underwent four stages within a single experimental session. In the baseline phase, physiological data were recorded for 5 minutes after a short habituation period with the sensors. In the anticipation-instruction phase (5 minutes), subjects were told they had to give a speech in front of a virtual audience on a topic that would be announced to them just before the 12
Page 12 of 37
actual speech. In the meanwhile they were given a written instruction corresponding to their experimental condition and were told to read it carefully, reflect on it and practice it in expectation of giving the speech. At the end of this phase, the head-mounted display was
ip t
installed and subjects were immersed in the virtual reality environment. The VR environment (Virtual Classroom; Grapp, 2012) consisted of a virtual audience arranged in a medium sized
cr
classroom (15-20 individuals), in which the participant took the position of the speaker at the
us
podium, in front of the audience. The audience was oriented so that it was looking at the speaker. Subjects were given a short habituation period with the head-mounted display, after which they
an
were announced a topic and were asked to speak on it for 3 minutes in front of the virtual audience (speech phase). A list of speech topics on controversial social, economic, political
M
issues (e.g., violent computer games should be banned) was constructed and each subject got a
d
different topic. Afterwards, subjects were let to rest for 3 minutes, with electrophysiological
te
sensors still attached (recovery phase). Subjects were then debriefed. Subjects were sitting down all throughout the experiment for reasons related to HRV measurement.
Ac ce p
Each of the emotion regulation instructions (see Appendix A) consisted of beliefs about
the public speaking situation. In the Dysfunctional (negative dysfunctional reappraisal) group, participants were given a set of irrational beliefs, which mimicked the ones a socially anxious person could hold in a socially evaluative situation, such as the one awaiting (e.g., “I might not have any idea what to talk about and say stupid things. This would be a terrible and catastrophic thing.”). In the Reappraisal (negative functional reappraisal) group, they received a set of rational beliefs as to how to evaluate the situation, which were similar to the ones a socially anxious client would be presented with during therapy, to replace the dysfunctional negative thoughts about the situation (e.g., I might not have an idea what to talk about and say stupid 13
Page 13 of 37
things. This would be bad, but I would not be so terrible and catastrophic.). In the Acceptance group, they got an acceptance-based rationale, stressing on the idea of accepting to remain in the present moment experiencing the thoughts and fears that arise, without trying to modify them
cr
accept to just experience them and stay with them here and now.).
ip t
(e.g., I have thoughts about how I will not do well and not be able to handle the situation, but I
us
3. Results 3.1. Baseline differences
an
The sample had mean social anxiety values of 48.58 (SD= 14.54) and mean general anxiety values of 44.14 (SD= 10.58). Means and standard deviations for the cardiac parameters
M
extracted are displayed in Table 1. At baseline there were no significant differences (all ps >.05)
d
between the groups on measures of social anxiety, self reported anxiety or any of the
te
physiological parameters extracted. Non-parametric correlation analysis (Spearman’s rho) between self-report variables at baseline and autonomic activity revealed a significant positive
Ac ce p
correlation across groups between trait anxiety and parasympathetic activity in the recovery phase (r=.220, p= .037). All other correlations were close to 0 and non-significant. We excluded the speech phase from the subsequent analysis as it most likely elicited substantial differences in respiration as compared to the other phases, which may have in turn impacted the HRV measures considered (see Ritz & Dahme, 2006 for a discussion). Table 1. Mean HRV measures during different task conditions. Measure
Group
Dysfunctional a
Reappraisal a (n=
Acceptancea
14
Page 14 of 37
/Study
(n= 30)
33)
(n=27)
Baseline
47.61 (11.43)
47.62 (17.23)
50.59 (14.53)
Baseline
43.87 (10.43)
44.97 (11.95)
Baseline
37.24 (11.96)
41.14 (12.15)
39.40 (11.49)
Baseline
82.26 (13.44)
87.75 (10.57)
83.45 (12.97)
Anticipation
93.24 (13.81)
97.7 (9.83)
96.05 (16.13)
103.11 (14.62)
103.67 (18.47)
864.88 (874.35)
976.49 (815.85)
615.18 (605.75)
449 (408.68)
584.05 (570.62)
651.95 (516.13)
411.29 (314.41)
549.77 (603.36)
Social Anxiety (LSAS) Trait Anxiety
Instruction
M
HR (bpm)
an
(STAI-S)
99.2 (16.8)
d
Recovery Baseline
1264.6 (1427.55)
te
HF (ms2)
us
State Anxiety
Anticipation
43.53 (9.41)
cr
(STAI-T)
ip t
Phase
Ac ce p
Instruction Recovery
Note. LSAS, Liebowitz Social Anxiety Scale; STAI, State Trait Anxiety Scale; HRV, heart rate variability; HR, Heart rate (bpm; beats/minute); HF, high frequency (ms2). Values represent means and standard deviations (in parantheses). a For self-report values n= 33 for the Dysfunctional Group, n= 34 for Reappraisal and n= 32 for Acceptance 3.2. Power analysis 15
Page 15 of 37
We conducted a post-hoc power analysis using the G*Power software. Our focus was repeated measures ANOVA looking for within between interactions. We assumed a p of 0.05, N= 90 (the number of subjects that had data for HF-HRV and HR) and 3 repeated measures. We
ip t
assumed a medium correlation among repeated measures of 0.50 (in reality this was a
conservative estimation since correlations among repeated measures were much higher for both
cr
HF-HRV and HR, ranging from 0.67 to 0.84). Power analysis indicated we would be able to
us
detect interaction effects as small as 0.18. 3.3. Group differences
an
Parasympathetic activity
We conducted a 3 (Group: Dysfunctional, Reappraisal, Acceptance) by 3 (Time:
M
baseline, anticipation-instruction, recovery) repeated measures ANOVA on HF-HRV,. Multivariate tests (Wilks’ Lambda) indicated a significant effect of Time, F(2,86)= 20.42,
te
d
p
