Background

During EMDR therapy, rhythmic eye movements are induced while the patient focuses on memories of traumatic events and related emotions and cognitions (Seidler & Wagner, 2006). EMDR is considered an efficient treatment tool (Shapiro & Forrest, 2016), especially for post-traumatic stress disorder (PTSD; Bisson & Andrew, 2007), for depression (Wood, Ricketts, & Parry, 2018), and even for specific phobias (De Jongh, Holmshaw, Carswell, & van Wijk, 2011; Lapsekili & Yelboga, 2014; Yunitri et al., 2020). Yet, there is no validated integrated model regarding the working mechanisms that could explain the effectiveness of EMDR (Jowett et al., 2016). One broadly accepted theory proposes that rhythmic eye movements (REM)—or other bilateral stimulations—strengthen the communication between the right and the left cerebral hemisphere and thus improve the subject’s ability to remember distressing events without being negatively aroused (Gunter & Bodner, 2008; Shapiro & Solomon, 2010). Yet, there is evidence questioning this theory by indicating that not only horizontal but also vertical REM can reduce the vividness of traumatic memories (Gunter & Bodner, 2008).

Andrade, Kavanagh, and Baddeley (1997) hypothesized that taxation of working memory could be an explanation for the effectiveness of EMDR. In that regard, dual tasks seem to be effective as long as they demand the working memory in an appropriate manner (van Schie, van Veen, Engelhard, Klugkist, & van den Hout, 2016). Eye movements provide an example for an effective dual task (Lee & Cuijpers, 2013; van Schie et al., 2016): Simultaneously recalling stressful memories and undergoing a visual task like performing REM causes an overload on the working memory as these processes are competing for working memory resources. Consequently, the reported distressing memories and images become less vivid and hence less emotionally intense (van Schie et al., 2016). This in turn could facilitate a subsequent adequate processing of traumatic memories. This hypothesis has found empirical support (Gunter & Bodner, 2008; Kavanagh, Freese, Andrade, & May, 2001; van den Hout & Engelhard, 2012).

Furthermore, several studies report that REM, or other dual-attention stimuli, provoke a psychophysiological de-arousal effect (see Schubert, Lee, & Drummond, 2016). The first to show this effect of REM were Wilson, Silver, Covi, and Foster (1996): Their eye movement group showed a physiological pattern of de-arousal; that is the respiration rate synchronized with the rhythm of the eye movements in a regular and shallow pattern, with heart rate (HR), systolic blood pressure, and skin conductance (SC) decreasing during the session. According to Wilson et al. (1996), this physiological relaxation pattern is incompatible with the perceived stress reaction during the recall of disturbing memories and thus results in desensitization as described in Wolpe’s (1995) concept of counter conditioning through reciprocal inhibition. Several studies on the physiological effects of eye movements support the view that eye movements induce a physiological de-arousal pattern (for review see Schubert et al., 2016; Söndergaard & Elofsson, 2008); for instance, an HR reduction, an increase in HR variability, a lowered respiratory rate, and a decrease in SC were found in patients with PTSD (Elofsson, von Schèele, Theorell, & Söndergaard, 2008; Sack, Lempa, Steinmetz, Lamprecht, & Hofmann, 2008). With arousal being reduced during exposure, distress might be attenuated to a tolerable level, making it more bearable for patients (Sack et al., 2008; Schubert, Lee, & Drummond, 2011) and facilitating effective processing during exposure.

Considering the findings mentioned above, we expected the outlined physiological changes induced by eye movements to also be beneficial during exposure therapy for specific phobias.

Furthermore, we aimed to combine rhythmic eye movements with another recent trend in exposure therapy: the use of virtual reality (VR; Kim & Kim, 2020). VR is a form of human–computer interface that allows the user to actively engage in a three-dimensional computer-generated environment, enabling the simulation of complex real-world environments, creating highly immersive naturalistic scenarios (Schultheis & Rizzo, 2001). Due to its ability to create situations that provoke real anxiety (Owens & Beidel, 2015), VR is of especially good fit for treating anxiety disorders (Boeldt, McMahon, McFaul, & Greenleaf, 2019) such as phobias (Botella, Fernández-Álvarez, Guillén, García-Palacios, & Baños, 2017). VR can enhance and accelerate the treatment process, offering several advantages over in vivo exposure, for example flexibility, controllability, and high acceptance (see Boeldt et al., 2019; García-Palacios, Botella, Hoffman, & Fabregat, 2007; Gorini & Riva, 2008.)

We examined whether adding REM could enhance the efficacy of a VR exposure therapy paradigm for spider-phobic patients. We hypothesized that the physiological reaction provoked by the confrontation with the virtual spider would decrease to a greater extent if REM were performed during exposure. Similarly, lower fear ratings throughout the course of the exposure were expected for participants conducting eye movements. Furthermore, REM were expected to not only have a positive influence on the treatment process, but also on overall treatment outcomes and effects.

Method Participants

Fifty-three volunteers (5 male, 48 female, Mage = 24.14, SDage = 6.00, age range: 18–40) were recruited through internet and advertisements at the University of Regensburg. All participants underwent the German version (Wittchen, Wunderlich, Gruschwitz, & Zaudig, 1997) of the Structured Clinical Interview for DSM-IV (SCID-I; First, Spitzer, Gibbon, & Williams, 2002) to ensure that all met the DSM-IV criteria for spider phobia. Exclusion criteria included current psychopharmacological medication, current involvement in psychiatric or psychotherapeutic treatment, cardiovascular or neurologically related diseases, colour blindness, hearing disorders, and age < 18 years. Participants were randomly assigned to either the experimental group (n = 26) or the control group (n = 27).

Design

As the goal of this randomized controlled trial was to investigate treatment effects, our primary outcomes are the subjective fear ratings 1 min before and during exposure in the behavioural avoidance test (BAT) in vivo and in VR measured by the subjective units of disturbance scale. The level reached for both BAT in VR and in vivo represents an additional dependent variable. Furthermore, the German versions of the Fear of Spider Questionnaire (FSQ; Rinck et al., 2002) and the State-Trait Anxiety Inventory (STAI; Laux, Glanzmann, Schaffner, & Spielberger, 1981) are used to determine the fear of spiders and anxiety level in general at the beginning of the therapy session and at follow-up.

The following physiological correlates of anxiety were measured as secondary outcomes: skin conductance response (SCR) during the VR session; HR; and eye movements (EOD). Besides, to assess participants' sense of presence in VR, a German version of the Igroup Presence Questionnaire (IPQ; Schubert, Friedmann, & Regenbrecht, 2001) was used.

Procedure

Testing took place at the University of Regensburg from July 2015 to September 2016 and was approved by the Ethics Committee of the University of Regensburg as well as pre-registered in ClinicalTrials (number: NCT02973919).

The study consisted of a therapy session, lasting about three hours (50 min in VR), and one follow-up session, taking place approximately two weeks after the first session, lasting about one hour (10 min in VR).

The concrete procedure of the therapy session is described below (see also Figure 1).

CONSORT flowchart of the procedure.

After the spider phobia had been diagnosed using the SCID-I, participants filled in the STAI as a measure of their current anxiety level and the FSQ as a measure of their fear of spiders. The investigators then briefed participants about the procedure of the BAT and guided them to the room with the spider where the BAT in vivo was conducted to assess the avoidance behaviour towards spiders (see Figure 2a). The experimental setup of the BAT in vivo and in VR was adopted from Shiban, Schelhorn, Pauli, and Mühlberger (2015) and Shiban, Pauli, and Mühlberger (2013). The real spider (a female Grammostola rosea) was placed in a transparent plastic box (7 × 14 × 10 cm) with a removable lid. This box was placed on a wooden slide 3 m away from the participant who was instructed to sit down on a chair behind a crank and pull the slide with the spider towards himself/herself as far as his/her fear allowed (level 2–5, every meter from 3 to 0 corresponding to one level). Refusing to enter the corresponding room was counted as level 1. If participants reached level 5, they had to touch the box in the next step, then move the box from the slide to a nearby table, open the lid, and touch the spider with a 30-cm long ruler (level 6–9). The investigator was not in the room while the participant performed the BAT in order to minimize any potential impact on the participant’s behaviour.

Behavioural avoidance tests in vivo (a) and in virtual reality (b).

As our procedure is strongly inspired by exposure therapy procedures aimed at traumatized patients (see Shapiro, 2017), the participants were asked to imagine their worst spider-related scenario before the start of the VR session. Before each exposure block, they received instructions via headphones to imagine this worst-case scenario.

The VR session consisted of an initial vision test and an orientation phase, a diagnostic block followed by four identical exposure blocks, and another diagnostic block at the end of the exposure (see Figure 3).

Chronology of virtual reality procedure for both sessions. In the therapy session, the exposure block was repeated four times after initial vision test, orientation, and the first diagnostic block. A second diagnostic block took place after the four exposure blocks.

Having put on the Head-mounted display, the vision test was presented. Once participants had confirmed good vision, the programme started with a welcome message and initial instructions to explore the orientation room. Having confirmed familiarization with the handling of the joystick without feeling any signs of simulator sickness, the orientation room was replaced by the diagnostic room where participants were told to relax (neither concrete instructions nor particular relaxation exercises were used). They were also told to move as little as possible for a two-minute baseline measurement of physiological parameters (relaxing phase). Then, the appearance of the spider was announced, expected fear was inquired, and participants were instructed to look at the spider. Next, the diagnostic room with the spider appeared for 30 s and participants reported their current fear level towards the end of this confrontation test. Afterwards, they were instructed to move towards the spider as close as possible, the spider appeared again for 30 s, and the joystick was activated so that they could conduct the virtual BAT (see Figure 2b). The participant was asked to move towards the virtual spider, placed in a distance of three metres, using a joystick. The BAT in VR consisted of five (distance) levels: no movement towards the spider (level 0), distance of 3.00–2.01 m (level 1), distance of 2.00–1.01 m (level 2), distance of 1.00–0.01 m (level 3), and distance of 0 m (level 4). The BAT lasted 30 s, and the level that the participant reached in the end (not necessarily the highest one) was registered.

After having completed the BAT, the first exposure block started with another two-minute relaxing phase in the empty exposure room. Instructions to look at the spider, which would be in the subsequent room, followed for the control condition. Participants of the eye movement group were told to follow an oscillating ball with their eyes while in the room with the spider. Based on the recommendation of Shapiro (2017) that the appropriate speed should comply with the patient, each participant could choose her/ his preferred frequency.

The ball was present during exposure in both groups but remained still in the control group. Before the exposure room with the spider appeared, participants were first told to imagine their personal ‘worst-case scenario’ (15 s) and then to communicate their expected fear on a 0–100 score. Thereafter, the exposure started, and the current fear level was registered every minute. Considering that the fear network can be accessed very rapidly with minimal stimulus input (Globisch, Hamm, Esteves, & Öhman, 1999), a duration of 5 min was chosen for the exposure block. The exposure block was repeated three more times. The VR session ended with a second diagnostic block consisting of relaxing, expected fear rating, confrontation test, real fear rating, and virtual BAT. Immediately after the VR exposure, participants filled in the IPQ questionnaire on presence in VR. Finally, another in vivo BAT was conducted and fear ratings were documented.

The follow-up session consisted of filling in the STAI-State, the STAI-Trait and the FSQ, followed by a BAT in VR (one diagnostic block after vision test and orientation room). Then, participants filled in the IPQ and went through one last BAT in vivo.

Materials

In order to achieve a reliable diagnosis of spider phobia or the presence of any other Axis I condition, the German version (Wittchen et al., 1997) of the SCID-I for DSM-IV (First et al., 2002) was conducted with each participant. The SCID reportedly has a relatively high reliability of .83 for specific phobias (Lobbestael, Leurgans, & Arntz, 2011).

During exposure, the participants indicated their expected and actual fear level verbally on a scale ranging from 0 (no fear at all) to 100 (extreme fear). Instructions were provided acoustically and visually. The expected fear was registered just before the exposure room with the spider appeared, and the actual fear level was registered 30 s after the appearance of the spider and later in intervals of one minute (in total five times per room).

Electrodermal activity (EDA), HR, and eye movements (EOG) were recorded using a V-Amp 16 and the BrainVision Recorder software, version 1.20, (both: Brain Products GmbH, Gilching, Germany) with a sampling rate of 1000 Hz. Two surface electrodes (Ag/AgCl, Ø = 6/14 mm), filled with an isotonic electrode gel (TD-246, PAR Medizintechnik GmbH), were placed next to each other onto the Thenar muscle of the non-dominant hand to measure EDA. HR was registered with the help of two adhesive pre-gelled surface electrodes (Ag/AgCl, Ø = 50 mm) that were attached to the middle of the upper chest and on the rib tip of the left half of the body. Reference (right) and ground (left) electrodes (Ag/AgCl, Ø = 6/14 mm) were fixed behind the ears. To measure eye movements, one more electrode of the same type as the reference electrodes was placed under the participants’ right eye.

In order to measure the fear of spiders, participants filled in the German version of the Fear of Spiders Questionnaire (FSQ; Rinck et al., 2002), an 18-item self-report questionnaire which is sensitive subtle differences due to a 7-point-Likert scale (Muris & Merckelbach, 1996). It has very good retest reliability (rtt = .95 after one week) and internal consistency (Cronbach’s α = .96; Rinck et al., 2002).

To measure anxiety, the German version of the STAI (Laux et al., 1981) was used. The STAI consists of a state and trait anxiety subscale, each consisting of 20 items rated on a 4-point-Likert scale. The psychometric properties of the STAI proved to be good with an internal consistency of Cronbach’s α = .90, retest reliability of r = .77 to r = .90 after 63 days for the trait scale and, as expected, lower retest reliability for the state scale (r = .22 to r = .53; Laux et al., 1981).

Participants’ sense of presence experienced in a virtual environment (VE) was assessed with a German version of the Igroup Presence Questionnaire (IPQ; Schubert et al., 2001) consisting of 14 items rated on a 7-point-Likert scale with the subscales spatial presence, involvement, and experienced realism (Schubert, 2003).

A high overall reliability (α = .85 to .87) could be demonstrated (Schubert, Friedmann, & Regenbrecht, 1999). An immersive VR environment was generated using the Steam Source Engine (Valve Corporation, Bellevue, WA, USA) and controlled by Cybersession software (CS-Research, VTplus GmbH, Würzburg, Germany). The virtual scenario was presented via a Head-mounted display (Oculus Rift DK2, Menlo Park, CA, USA) with a 360-degree view in combination with a positional tracking device (Oculus Positional Tracker DK2, Menlo Park, CA, USA).

The VR environment consisted of a diagnostic room (Figure 4a) and an exposure room (Figure 4b), both plain office rooms with only a door, a shelf, and heaters inside. The spider was a big, hairy tarantula, sitting at the opposite end of the room, always remaining in the same spot at a 3 m distance from the participant, but moving its four front legs in the air and taking defence (Figure 4c) and attack (Figure 4d) positions.

Virtual reality environment and stimulus material. (a) Diagnostic room. (b) Exposure room. (c) Virtual spider in defence position. (d) Virtual spider in attack position.

Data reduction and statistical analysis

Physiological data were pre-processed with the Brain Vision Analyzer 2.0 Software (Brain Products GmbH, Gilching, Germany). The EDA data were rectified by a 1 Hz high cut-off filter (12 dB) and segmented with the help of markers. For the four five-minute exposure blocks, the skin conductance level (SCL) of the first 30 s and the last 60 s were exported in order to analyse habituation to the spider. For the three BATs, the period from 1,000 ms before to 8,000 ms after confrontation with the spider were taken and the first 1,000 ms of this sequence was searched for its minimum to receive a baseline value; the last 7,000 ms starting after confrontation with the spider were searched for its maximum to receive the peak value of the event-related SCR, both with the MinMax-Macro. These two values were exported, and the difference calculated in order to receive the baseline-corrected SCR. For HR, the difference values between the ECG electrodes were computed, a 1.59 Hz (12 dB) high cut-off filter, a 30 Hz (12 dB) low cut-off filter and a 50 Hz notch filter were administered. After that, R-spikes were automatically detected and counted, manually controlled and corrected if necessary. HR per minute was exported for the first and fifth minute of every exposure block and for the 8 s after confrontation with the spider in the BAT.

Further analyses were conducted with SPSS 23.0 (IBM Corp., Armonk, NY, USA). Repeated-measures ANOVAs were applied for the exposure process analysis of the dependent variables (subjective fear ratings, SC, and HR) with exposure block (1–4) and time of measurement (minute 1 vs. minute 5 within one exposure session) as within-subject factors and group (eye movement vs. control) as between-subject factor. Post-hoc t-tests were performed to analyse differences between block 1 and block 4 (overall fear reduction during therapy) and between minutes 1 and 5 of every block (within session habituation). Therapy effect was analysed for all dependent variables (SC and HR for BAT in VR, fear ratings and level reached for both BAT in VR and in vivo, and FSQ and STAI for questionnaire data) with repeated-measures ANOVAs with time (pre-exposure vs. follow-up) as within-subject factor and group (eye movement vs. control) as between-subject factor. Significant differences in the ANOVAs were followed up using t-tests. Greenhouse–Geisser correction was used whenever the assumption of sphericity was violated. The significance level was set at p < .05. Effect sizes of the ANOVAs are reported as . Data from the questionnaires were analysed for group difference at baseline measurement and to evaluate how present participants felt in the VR environment. Seven participants (five in the eye movement group and two in the control group; see Figure 1) were excluded from the entire analysis because they reported a subjective fear lower than 40 in the first fear rating during VR exposure. With such a low fear level, the fear activation necessary for a significant habituation and therefore a relevant treatment effect cannot be expected (Benito & Walther, 2015; Foa & Kozak, 1986). In total, the data of 46 participants were included in the analysis (see Figure 1).

Results Manipulation check of the eye movements

The power spectral density plots in Figure 5 show the frequencies of the eye movements for the baseline and the exposure condition in both groups. The dominant frequencies are 0.45, 0.65, and 1 Hz which indicates that participants in the experimental group performed REM during the virtual exposure in the speed of one of the three frequencies we offered.

Eye movement frequencies. Averaged power spectral density estimate of vertical EOG electrode showing distinct peaks at the given frequencies during exposure for eye movement group but no detectable periodicity during baselines of both groups and exposure of the control group.

Process analysis Subjective fear ratings

Figure 6a shows a reduction in subjectively perceived fear throughout the exposures for both the eye movement group and the control group, which was confirmed by main effects of exposure block, F(1.73, 76.27) = 154.70, p < .001,  = .78, and of time of measurement, F(1, 44) = 136.37, p < .001,  = .76. There was a significant interaction effect for exposure block × time of measurement, F(2.35, 103.42) = 27.20, p < .001,  = .38. No other significant effects were found, all p > .05. Post-hoc t-tests for the significant main effect of exposure block revealed that the differences in the ratings between exposure 1 and 4, t(45) = 7.60, p < .001, d = 1.47 were significant with fear ratings being significantly lower for exposure 4. Apart from that, post-hoc t-tests for the main effect of time of measurement revealed significant differences in the ratings of the first and the last minute of every exposure throughout every exposure block in both groups. For the control group, the following values resulted: exposure 1, t(24) = 8.50, p < .001, d = 1.25, exposure 2, t(24) = 6.47, p < .001, d = 0.74, exposure 3, t(24) = 5.28, p < .001, d = 0.48, exposure 4, t(24) = 3.02, p < .006, d = .31. For the eye movement group, the analysis showed the following results: exposure 1, t(20) = 7.85, p < .001, d = 1.19, exposure 2, t(20) = 5.63, p < .001, d = .82, exposure 3, t(20) = 4.42, p < .001, d = 0.54, exposure 4, t(20) = 3.50, p < .002, d = 0.27. These results indicate a strong decrease in subjectively perceived fear during VRET in both groups whether eye movements had been performed or not.

Process analysis of (a) subjective fear ratings, (b) skin conductance response (SCR), and (c) heart rate during virtual reality exposure. (a) First and last fear rating of every exposure cycle for both groups. (b) Average T-scores of the SCR for both groups at minute one and minute five of every exposure cycle. (c) Average HR for both groups at minute one and minute five for every exposure cycle.

Skin Conductance Response (SCR)

SCR of both groups decreased throughout every exposure cycle and between exposures, resulting in a significant main effect of exposure block, F(1.89, 79.22) = 9.30, p < .001, = .18, and a significant main effect of time of measurement, F(1, 42) = 20.19, p < .001,  = .33 (Figure 6b). No other significant effects were found, all p > .05. Post-hoc t-tests comparing the SCR measured at the beginning of the exposure blocks revealed a significantly lower SCR for block 4 compared with block 1, t(43) = 4.06, p < .001, d = 1.01. Further post-hoc t-tests for the main effect of time of measurement separated by groups yielded significant differences for the control group throughout exposure 1, t(23) = 4.46, p < .001, d = 0.50, exposure 2, t(23) = 4.42, p < .001, d = 1.03, and exposure 3, t(23) = 2.35, p = .028, d = 0.35, and significant differences for the eye movement group throughout exposure 1, t(19) = 4.33, p < .001, d = 1.3, exposure 3, t(19) = 2.44, p = .025, d = 0.72, and exposure 4, t(19) = 2.84, p = .010, d = 0.67. These results indicate that SCR decreased during exposure for both groups. No other significant effects were found.

HR

Figure 6c depicts an almost constant HR for both groups with a lower HR for the eye movement group during all exposure blocks which was verified by a main effect of group, F(1, 42) = 5.56, p = .023, = .12. No other significant effects were found, all p > .05.

Therapy effect BAT in VR

Figure 7a–d show a reduction in subjectively perceived and physiological fear and a reduction in avoidance behaviour from pre-exposure to follow-up without a difference between groups. ANOVAs revealed a significant main effect of time for fear ratings, F(1, 43) = 195.26, p < .001, = .82, SCR, F(1, 38) = 19.45, p < .001,  = .34, HR, F(1, 37) = 11.53, p = .002,  = .24, and the level reached, F(1, 41) = 77.78, p < .001,  = .66. There was neither a significant main effect of group nor a significant interaction effect for any of these outcome variables (all p > .05).

(a) Fear ratings, (b) skin conductance response (SCR), (c) heart rate, and (d) level of approach for both groups during behavioural avoidance test in virtual reality before exposure and at follow-up. Note. Definitions of levels reached: Level 0: no approach to the spider (virtual distance to participant = 3 m). Level 1: distance between virtual spider and participant = 2.99–2.01 m. Level 2: distance between virtual spider and participant = 2.00–1.01 m. Level 3: distance between virtual spider and participant = 1.00–0.01 m. Level 4: full approach to the spider (distance to participant = 0 m).

BAT in vivo

A decrease in fear and avoidance behaviour could be observed in both groups from pre-exposure to follow-up, which was confirmed by a main effect of time for perceived fear, F(1, 42) = 43.04, p < .001,  = .51, and the level reached, F(1, 42) = 77.13, p < .001, = .65, indicating a success of the intervention regardless of whether eye movements had been performed (Figure 8a, b). No group differences or interaction effects were found.

(a) Fear ratings and (b) level of approach for both groups during behavioural avoidance test in vivo before exposure and at follow-up. Note. Definitions of levels reached: Level 8: Participant did not move the spider towards him- or herself (distance between spider and participant = 3 m). Level 7: Participant uses the crank to move the spider towards him- or herself. Distance between spider and participant = 2 m. Level 6: Distance between spider and participant = 1 m. Level 5: Distance between spider and participant = 0 m. Level 4: Participant touches the box with the spider sitting in it. Level 3: Participant holds the box and puts it on a table. Level 2: Participant opens the lid of the box. Level 1: Participant removes the lid and touches the spider with a pen.

FSQ

The analysis of the FSQ showed an improvement from pre-exposure to follow-up for the eye movement group, t(19) = 4.24, p < .001, d = 0.99, as well as for the control group, t(23) = 4.75, p < .001. d = 0.79, as indicated by significantly decreased scores of phobic symptoms.

Questionnaires

See Table 1 for differences in the FSQ, STAI, and IPQ scores between eye movement and control group at pre-exposure and at follow-up. The groups did not differ significantly in any of these measurements, neither at pre-exposure nor at follow-up.

Table 1. FSQ, STAI, and IPQ scores before and after exposure in the control group and the eye movement (EM) group Control group EM group Difference M SD M SD t df p FSQ Pre 73.84 17.81 81.62 11.70 −1.71 44 .09 Post 56.46 24.76 64.45 19.13 −1.18 42 .25 STAI Pre 47.80 11.35 48.33 10.50 −0.16 44 .87 Post 38.46 6.59 39.33 5.26 −0.49 43 .63 IPQ Pre 49.24 11.35 45.76 13.52 0.95 44 .35 Post 45.96 12.58 41.90 16.31 0.94 43 .35 Note FSQ = Fear of Spiders Questionnaire; IPQ = IGroup Presence Questionnaire; pre, before exposure; post, at follow-up; STAI = State-Trait Anxiety Inventory. Discussion

The aim of this study was to examine whether the induction of rhythmic eye movements would initiate a change in physiological parameters and subjective fear ratings and whether these changes could improve the effectiveness of an exposure therapy. A VR treatment paradigm for spider phobia was used to test a potential beneficial effect of horizontal REM on VRET. EDA and subjective fear ratings decreased significantly during the exposure to the virtual spider, regardless of whether participants performed eye movements or not. The HR of the group performing eye movements was significantly lower than the HR of the control group during the entire virtual exposure. The examined therapy effect from pre-exposure to follow-up showed a significant improvement for both groups in the VR and the in vivo BAT, as well as in the FSQ. The subjectively perceived fear of the spider and the avoidance behaviour related to the spider as well as HR and SCR were significantly reduced, regardless of the performance of the eye movements. This (physiological) fear reduction indicates that the intervention in VR was successful and that the effect of the virtual exposure paradigm was generalized to life by the phobic patients. This generalization was also found in a study of García-Palacios, Hoffman, Carlin, Furness, and Botella (2002), in which phobic patients showed better results in a BAT in vivo after a virtual exposure therapy than before therapy, and better results when compared to a control group without any intervention.

Contrary to our expectations, the implementation of REM during virtual exposure did not enhance the effectiveness of the chosen paradigm. Nonetheless, the significantly lower HR in the eye movements group during the virtual exposure and the similarly strong therapeutic effect in both groups suggest additional physiological relaxation due to the eye movements. Besides, the lower HR of the eye movement group in comparison with the control group could also be interpreted as an indicator for a less stressful experience for patients (Taelman, Vandeput, Spaepen, & Van Huffel, 2009). However, the addition of REM does not seem to have a beneficial effect on the subjective fear ratings. So even if patients are in fact more relaxed, as hinted by a reduced HR, they may still not feel less distressed. In contrast to the present study, Schubert et al. (2011) reported that eye movements further increased the benefit of exposure regarding the self-reported level of distress right after exposure. Future research will have to investigate whether the effect of REM on HR found in the present study reflects a meaningful benefit of REM during exposure and therefore increase treatment effectiveness.

The lack of an additional positive effect of the REM on the VR paradigm could be due to distraction. Researchers such as Rachman (1980) assumed that a distraction during the exposure would inhibit emotional processing because the fear activation, which is necessary for a successful treatment, would be insufficiently initiated. Thus, it is still advised to prevent distraction from the phobic stimulus or cognitive avoidance during an exposure in vivo (Drechsel, 2002; Paul, Kathmann, & Riesel,