Can Virtual Reality Effectively Elicit Distress Associated with Social ...

J Psychopathol Behav Assess (2015) 37:296–305 DOI 10.1007/s10862-014-9454-x

Can Virtual Reality Effectively Elicit Distress Associated with Social Anxiety Disorder? Maryann E. Owens & Deborah C. Beidel

Published online: 5 September 2014 # Springer Science+Business Media New York 2014

Abstract This study examined the ability of a Virtual Reality (VR) environment to elicit the physiological and subjective arousal typically associated with public speaking. Using 21 adults with Social Anxiety Disorder (SAD) and 24 adults with no disorder, this study had three objectives: (a) to determine whether speaking to a virtual audience elicited significant increases in physiological response (e.g., heart rate, electrodermal activity, and respiratory sinus arrhythmia) and subjective distress over baseline resting conditions (b) to determine if individuals with SAD had a greater increase in physiological arousal and subjective distress when speaking in front of a live audience vs. the virtual environment and (c) to determine whether individuals with SAD had greater changes in physiological and self-reported arousal during each speech task compared to controls. All participants gave an impromptu speech in front of an in vivo and VR audience while measures of physiological arousal and self-reported distress were obtained. Results demonstrated that the VR task elicited significant increases in heart rate, electrodermal activity, and respiratory sinus arrhythmia, and self-reported distress over baseline conditions but was less anxiety-producing than the in vivo speech task. In addition, participants reported a moderate level of presence in the VR task, but significantly less than in the in vivo task. No group differences were found on physiological measures. Clinical implications of these findings and the role of VR in the treatment of SAD are discussed. Keywords Social anxiety disorder . Public speaking . Virtual reality . Exposure therapy . Physiological arousal . Subjective distress M. E. Owens (*) : D. C. Beidel Department of Psychology, University of Central Florida, 4000 Central Florida Blvd. Bldg 99, Suite 320, Orlando, FL 32816, USA e-mail: [email protected]

Introduction Social Anxiety Disorder (SAD) is characterized by significantly impairing and burdensome social distress, avoidance, and in some cases a deficit in social skill (Liebowitz et al. 1985; Patel et al. 2002; Turner et al. 1986; Wittchen and Beloch 1996; Zhang et al. 2004). Among the most commonly avoided situations (e.g., attending parties, meeting new people, using public restrooms, and speaking up in class or at meetings), the most prevalent is public speaking (Mannuzza et al. 1995; Stein et al. 1996). Currently, exposure therapy (EXP) utilizing in vivo stimuli is considered the most effective treatment for anxiety disorders such as SAD (Craske et al. 2008). However, in vivo exposure is not always practical or ethical, especially when the stimuli are dangerous, difficult to recreate or repeat (e.g., giving a speech in front of a full auditorium), prohibitively expensive (e.g., flight phobia), or elicit such intense fear that a patient is unwilling to enter therapy. To address these limitations, virtual reality exposure therapy (VRET) was developed as a potential alternative to imaginal or in vivo exposure (Krijn et al. 2004; Rothbaum and Hodges 1999) and data suggest that VR may be a viable tool for the treatment of specific phobias and SAD, including public speaking anxiety (Anderson et al. 2003, 2005; Harris et al. 2002; Klinger et al. 2005; Wiederhold and Wiederhold 1998). However, efficacious methods of exposure therapy must meet certain underlying tenets and to date, few data have examined the match between those conditions and VRET. Foa and Kozak (1986) proposed the theory of emotional processing as the mechanism by which exposure therapy produces emotional and behavioral change. Emotional processing theory postulates that fear occurs when neural networks, which produce information about a stimulus (a situation or event), the meaning of that stimulus (threat or danger)

J Psychopathol Behav Assess (2015) 37:296–305

and behavior (escape or avoidance, are activated. Exposure therapy alters these relationships by producing new and incompatible information/networks (see also Craske et al. 2008). Thus, to be consistent with the principles of emotional processing theory, VRET must meet three conditions. First, the virtual environment must be generalizable to real-life situations so that when extinction occurs in the virtual environment, there is a decrease in distress and avoidance in the corresponding real-life situation. Second, the patient must feel immersed (presence) in the environment as opposed to a passive observer (Slater et al. 1999). Finally, the virtual environment should elicit physiological arousal, which indicate that the core elements of the fear are being addressed (North et al. 1998; Regenbrecht et al. 1998; Schuemie et al. 2000). The second and third conditions are closely related as presence or engagement in a “fear-eliciting” environment should result in physiological arousal and subjective distress (Lee 2004; Schubert et al. 2001). However, previous trials, while suggesting that VR elicited physiological arousal (e.g., blood pressure and heart rate), self-reported distress and a sense of immersion were limited by small sample sizes, lack of a clinical population, and the lack of a comparable control task (i.e., the VR condition was not compared to an in vivo speech (Hartanto et al. 2014; Kotlyar et al. 2008; Pertaub et al. 2002; Slater et al. 2006)). One recent trial did demonstrate the ability of a VR conversation task to elicit significant levels of subjective distress and immersion when compared to a similar in vivo task (Powers et al. 2013); however, this study lacked the inclusion of a clinical population and objective measures of anxiety, thus its relevance to the treatment of a phobic population is unclear. VRET has the potential to become a cost-effective, practical, and efficacious treatment for SAD, as it appears to appeal to a significant proportion of individuals with SAD (76 %) who chose it over in vivo exposure (Garcia-Palacios et al. 2007). Given the high prevalence of SAD, low rates of treatment seeking, and difficulty in constructing appropriate in vivo exposure conditions for people with SAD (e.g., difficulty finding audience members), the development of VR has the potential to alleviate the burden this disorder places on the individual, clinicians, and the economy. However, prior to its wholesale adoption, questions about VR’s ability to satisfy Foa and Kozak (1986) basic requirements for emotional processing theory remain. This study will extend previous research in two ways. First, in order to examine VR’s ability to elicit physiological arousal, subjective distress, and feelings of presence similar to in vivo exposure, this study compared a VR public speaking environment and a comparable in vivo speech task in individuals with SAD and individuals with no disorder. We hypothesized that when placed in a virtual environment and asked to give a speech, all participants would experience significantly increased physiological response and subjective distress over baseline resting

297

conditions. Additionally, we hypothesized that individuals with SAD would have a greater increase in physiological and self-reported arousal during the in vivo speech task than the VR speech task. Finally, we predicted that individuals with SAD would experience greater changes in physiological and self-reported arousal during each of the speech tasks in comparison to healthy controls. In regards to presence, we hypothesized that all participants would feel present during both speech tasks, but more so during the in vivo speech task.

Method Participants Participants were recruited via community advertisement and through the university’s undergraduate research pool. Participants were admitted into the study if they met diagnostic criteria for SAD or had no psychological disorder and met additional inclusion and exclusion criteria. Specifically, participants meeting criteria for a primary diagnosis of SAD were also required to have a Clinician Severity Rating (CSR) of at least 4 on the Anxiety Disorders Interview Schedule for the DSM-IV (ADIS-IV; Brown et al. 1994). Participants with additional Axis I disorders (e.g., depression) were included if they were secondary to their primary diagnosis of SAD. However, presence of a lifetime diagnosis of current bipolar disorder, suicidal ideation, current alcohol or substance abuse, or psychosis was exclusionary. For both groups, additional exclusion criteria included any unstable or serious medical conditions or taking any medications that, in the opinion of the researcher, might have interfered with the measures being assessed (e.g. psychoactive medications, anti-hypertensives). Fifty-nine (59) adults responded to recruitment efforts. Nine participants met exclusion criteria for comorbid disorders, (1) primary Substance Dependence, (1) primary Generalized Anxiety Disorder, (1) no public speaking anxiety, and (1) currently taking beta blockers. Five participants did not return to participate in the study following the initial interview. An additional 5 participants were removed due to incomplete physiological assessment data. The final sample consisted of 45 adults representing two groups: 21 adults with SAD (10 males; 11 females) and 24 adults without any psychiatric disorder (11 males; 13 females). Adults ranged in age from 18 to 25 years (MSAD =20.90 and MControls =19.42 years old). Most were Non-Hispanic White (51, 13 % African American, 20 % Hispanic, 11 % Native American/Pacific Islanders, 4 % Other). Demographic characteristics, comorbid diagnoses, and mean assessment scores for the sample are shown in Table 1. Two of the participants were currently receiving talk therapy to address anxiety-related difficulties; however, none of the participants had prior experience with exposure therapy. Three of the participants endorsed taking an SSRI (e.g.,

298


Table 1 Demographic and assessment data for participants

Age M(SD) Gender Males Females Race/Ethnicity Caucasian African American/Black Hispanic/Latino Asian/Pacific Islander Other Assessment measures M(SD) ADIS-IV SAD CSR SPAI difference score LSAS total score HAMD total score Comorbidity (N) MDD Depressive disorder NOS GAD Panic disorder Bipolar disorder Provisional OCD

SAD (n=21)

Controls (n=24)

20.90(2.19)

19.42(1.72)

10 11

11 13

9 3 7 2 0

14 3 2 3 2

5.48(1.12) 97.40(29.63) 76.00(22.55) 14.29(7.07)

– 24.77(16.48) 27.96(14.08) 4.83(4.27)

3

–

1 2 1 1 1

– – – – –

ADIS-IV SAD CSR Anxiety Disorders Interview Schedule for the DSMIV Social Anxiety Disorder Clinician Severity Rating, SPAI Social Phobia and Anxiety Inventory, LSAS Liebowitz Social Anxiety Scale, HAMD Hamilton Depression Scale, MDD Major Depressive Disorder, NOS Not Otherwise Specified, GAD Generalized Anxiety Disorder, OCD Obsessive Compulsive Disorder

Citalopram, Escitalopram, and Buproprion) and one endorsed taking a tri-cyclic antidepressant (e.g., Doxepin). Measures At the clinic, potential participants were interviewed by doctoral students in clinical psychology using the ADIS-IV (Brown et al. 1994). As part of the ADIS-IV diagnostic interview, a CSR was assigned to each diagnosis, using a nine-point scale (0–8) where higher numbers were indicative of greater perceived distress. To calculate inter-rater reliability, 20-percent of the interviews were scored by a second blinded evaluator (e.g., a doctoral student within the clinical psychology program). For the diagnosis of SAD, the kappa coefficient was k=1.00. Inter-rater agreement for the CSR intraclass correlation coefficient was ICC(2,2)=.970 and the reliability was r=.967. Axis II Disorders were not assessed as part of the initial assessment. In addition to the diagnostic interview, the following measures were included to determine study eligibility and SAD symptom severity:

Participants completed the Social Phobia and Anxiety Inventory (SPAI; Turner et al. 1989) to assess the range and severity of their social fears. The SPAI has high test-retest reliability of .86, good concurrent and external validity (Beidel et al. 1989; Turner et al. 1989), and differentiates patients with social phobia from people with no psychiatric disorder or from patients with other anxiety disorders (Turner et al. 1989). To be included in the study, participants with SAD were required to endorse at least at four on item five indicating frequent anxiety when giving a speech in front of a small audience. Similarly, to be included as a control, scores of four or higher on this item were exclusionary. Doctoral students who conducted the diagnostic interview also administered the The Liebowitz Social Anxiety Scale (LSAS; Liebowitz 1987) to quantify the degree of fear and frequency of avoidance behavior across different social situations. The LSAS has strong convergent and discriminate validity (Heimberg et al. 1999). To calculate inter-rater reliability and agreement, 20-percent of the interviews were scored by a second blinded evaluator. Inter-rater agreement for the LSAS was ICC(2,2)=.999 and reliability r=.998. To assess potential depressive symptoms and rule out participants who may be suffering from significant depression, participants were administered the Hamilton Rating Scale for Depression (HAM-D; Hamilton 1960). Mean scores ranged from 0 to 27 and in each case, depressive symptoms were determined to be secondary to SAD.

Self-Report Measures Participants completed the following battery of self-report measures: The Self-Statements During Public Speaking (SSPS; Hofmann and Dibartolo 2000) is a ten-item questionnaire designed to assess fearful thoughts experienced during public speaking. The SSPS consists of two five-item subscales, the “Positive Self-Statements” (SSPS-P) and the “Negative SelfStatements” subscale (SSPS-N). The SSPS has demonstrated good levels of internal consistency and test-retest reliability and differentiates high and low levels of public speaking anxiety (Hofmann and Dibartolo 2000). The Subjective Units of Distress Scale (SUDS) asks the participant to rate their own level of anxiety using a nine-point likert type rating scale (0–8; no distress to extreme distress). Two Visual Analogue Scales (VAS) were included after each speech task to assess the degree to which the participant felt engaged/involved with the speech task environment and separately, how strong was their sense of “being there.” Participants asked to indicate their response by drawing a vertical mark on a 100 mm line, which was anchored by labels representing the extremes of the continuum (e.g., Not Engaged/Involved At All to Completely Engaged /Involved


and Completely Detached/No Sense of Being There to Complete Sense of “Being There”).

Behavioral Measures The behavioral assessment consisted of (a) an impromptu speech in front of a live five-person audience and (b) a VR five-person audience. The order of task administration was counterbalanced. The VR speech task utilized the conference room scene from Virtually Better’s© software package. In this virtual environment, the participant viewed a waiting room through a head mounted display (HMD), and the researcher moved the participant into a virtual conference room in which a five person audience was seated around a conference table. As illustrated in Fig. 1, the audience members consisted of two men and three females of varying ethnicities (3 Caucasians, 1 African American, and 1 Asian) wearing business attire. For the in vivo speech task, the virtual conference room was recreated in a conference room in UCF’s Psychology Building. Four to five undergraduate volunteers, instructed to wear business attire, were seated around the conference table, consistent with the virtual task. As a measure of avoidance/escape, participants were given the option to escape once they completed 3 min of the task. Speech duration in minutes was recorded as a behavioral measure. Physiological Measures During the behavioral assessment (described below), heart rate (HR), electrodermal activity (EDA), and respiratory sinus arrhythmia (RSA) were measured. HR, a measure of sympathetic and parasympathetic responses to external stimuli, was measured via EKG at 30 s intervals. EDA, as measured by skin conductance level (SCL) and response (SCR) provides a measure of sympathetic activity on the autonomic nervous system (ANS). Finally, RSA is a measure of vagal cardiac control related to respiration (Berntson et al. 2007) that offers a direct examination of parasympathetic activity within the ANS. During the tasks, the MindWare Psychophysiological Ambulatory system was

299

used to continuously measure these variables. Continuous recording allowed assessment of physiological arousal over time and in relation to the two speech tasks. MindWare Version 3.0 allows the conversion of physiological data into meaningful statistical data to be analyzed using Mindware analysis software version 3.0.9. Procedure After receiving the informed consent, participants completed the ADIS-IV, HAM-D, LSAS, and self-report measures. If eligible, participants were fitted with electrodes and a respiration belt to measure RSA. Participants were asked to sit quietly during a 10-min adaptation/baseline period. At the end of the baseline period, participants provided a SUDS rating. They were informed that they would participate in two 10-min speech tasks. Participants were provided with five topics (a different set was provided for each task with topic areas including: the legal drinking age, qualities of a good president, and whether the public should be informed of alien life) and instructed to choose up to three topics to use during their first speech. Participants were given 3 min to prepare this speech and were allowed to reference their topic cards during the tasks. After delivering their speech, participants recorded their SUDS rating and completed the SSPS and VAS. The participant then sat quietly for 5 min before the next task to allow the participant’s physiological response to return to approximately baseline levels. Participants were then provided a new set of topics, and then the same procedure was repeated for the second speech task. Design This study utilized a 2x2 factorial design including a manipulated within-subjects factor (Task: in vivo vs. VR) and a nonmanipulated between-subjects factor (Group: SAD vs. Control). Repeated measures of HR, SCL, SCR, RSA, and SUDS were obtained during the baseline and experimental tasks. Measures of cognitive distortions (SSPS) and presence (VAS) were also obtained after the experimental tasks. Analytical Strategy

Fig. 1 Image courtesy of virtually better, Inc

One participant’s HR data was determined to be an outlier, as the values fell above the 3rd quartile, and was removed. SCRs were counted if the fluctuation exceeded .05 μS. HR data were edited for artifact following data collection. The mean change score of each channel during the first 3 min of each speech task was calculated and used as the overall task mean change. The mean of the final 60 s of the initial baseline period was used in the calculation of change scores. One-tailed, paired samples t-tests were used to examine change in physiological arousal during the VR speech task. A

300


series of 2×2 Mixed Subjects Repeated Measures ANOVAs with planned comparisons of simple effects were used to analyze physiological arousal (e.g., HR, RSA, EDA, SCR), self-reported distress (e.g., SUDS), self-reported perception of task performance (e.g., SSPS), and degree of engagement with the task environments (e.g., VAS) (Hypothesis 2 & 3), with one-tailed analyses to follow-up on overall significant effects.

in vivo environment (M(SD)Engagement =69.28(29.04) and M(SD)“Being There”=74.31(29.64)) than the VR environment (M(SD)Engagement =52.63(25.52) and M(SD)“Being There”= 52.22(25.94)); however, the group and interaction effects for these variables were not significant. Scores for each group for each task are depicted in Table 2.

Does the VR Elicit Subjective Distress and Cognitive Distortions? Results Does the VR Environment Produce Feelings of Immersion? Prior to determining whether the VR environment produced physiological arousal and subjective distress, VAS “Engagement” and “Being There” scores were analyzed to determine whether the VR environment was a valid manipulation. The main effect for task on the VAS Engagement and VAS “Being There” scales was significant, with all participants reporting more engagement and a stronger sense of “being there” in the

The main effect for the SSPS Negative subscale was significant, with all participants reporting more negative selfstatements in the in vivo environment (M(SD)=13.22(4.06)) than in the VR environment (M(SD)=11.87(3.88)). There was no significant main effect for the SSPS Positive Subscale. The main effect for group on the SSPS Positive and SSPS Negative subscales was not significant. The interaction effects for the SPSS Positive and Negative subscales were not significant. However, the interaction effect on SUDS Ratings was significant; indicating that while both groups had a

Table 2 RMANOVA results and descriptive statistics for self-report measures SAD (N=21) M(SD) Variable SUDS (Change score) Main effect: task Main effect: group GXT interaction Simple effects VAS “engagement” Main effect: task Main effect: group GXT interaction VAS “being there” Main effect: task Main effect: group GXT interaction SSPS positive subscale Main effect: task Main effect: group GXT interaction SSPS negative subscale Main effect: task Main effect: group GXT interaction

Controls N=24 M(SD)

In Vivo

VR

a

In Vivo

3.00(1.79)

5.43(1.57)a

3.43(1.99)b

66.74(24.19) 46.17(27.68)

60.91(30.11) 70.88(33.71)

65.33(25.93) 50.93(26.51)

14.36(3.27) 13.95(3.44)

13.81(2.02) 14.76(3.10)

a

14.81(3.75) 15.76(3.22)

VR

3.42(1.82)b

69.28(29.04)a 3.35 75.19(25.83)

58.29(22.55) 74.31(29.64)a .428

77.31(25.92)

50.93(25.94) 13.96(2.82) .645

13.96(2.22)

2.98(1.90)b .041

F

p

26.24