Is virtual reality always an effective stressors for exposure treatments? some insights from a controlled trial

Different research studies have shown that exposure-based therapies are more suitable and effective than others in the treatment of anxiety disorders [1‐9].

Exposure is a process in which the patient is progressively exposed to the feared stimulus or the situation that provokes anxiety. Traditionally, exposure is experienced in two ways: by imagery in the person’s imagination, or in vivo, with direct contact to the stimulus.

Nevertheless, all roses have thorns, and this is true for both types of exposure, too. On one side, some patients report difficulties when asked to imagine the feared situation; furthermore, emotions have been shown to modulate visual imagery and perception [10], impairing visualization of detailed scenes. On the other side in vivo exposure is not fully under the control of the therapist and requires a high effort in terms of money and time expenditure.

To overcome these issues an emerging approach is the use of Virtual Reality Exposure Treatments (VRET) [11‐15]. Compared to traditional methods, VR has the advantage to elicit a sense of presence in users, defined as the “feeling of being in a world that exists outside of the self” [16, 17]. Another advantage of VR is that in such a setting, as in the real world, the subject can manipulate and interact with the environment differently from all the other techniques in which users are passive and not active [13, 18]. It is also possible to grade the intensity of the stimulus following the personal needs of each patient. In this way, the patient feels less uncomfortable about the treatment and his/her motivation increases [19]. Recently different quantitative meta-analyses, have been conducted on studies reporting VRET treatments. Parson and Rizzo [20] analyzed data from 21 studies who have evaluated anxiety and/or phobia before and after VRET. They conclude that VRET has a statistically large effect on all affective domains, and thus it is a relevant approach to reduce anxiety-related symptoms. They conclude that VRET has a statistically large effect on all affective domains, and thus it is a relevant approach to reduce anxiety-related symptoms. Similarly, Powers and Emmelkamp [21] provide effect size estimates for virtual reality treatment in comparison to in vivo exposure and other control conditions. They found a predictable larger effect of VRET compared to the control conditions; but more interestingly, VRET outperformed in vivo exposure.

But is VR always an effective stressor? Are the technological breakdowns that may appear during such an experience a possible risk for its effectiveness?

Unfortunately only a few studies compared the efficacy of VR in inducing an emotional response with other media and/or with real-life reactions [22‐24]. These studies also have the limitation to investigate very simple emotive reactions, not so complicated as might be a response related to a real-life stressor. Recently, Hartanto and colleagues investigated the effects of three different interruption mechanisms on subjective experience during a VRET session for social phobia with scripted avatar-patient dialogues [25]. Specifically, the study examined three interruption mechanisms: (1) dialogue dependent, that occurs freely in the flow of the dialogue; (2) speech dependent, that occurs during the pauses in the dialogue; and (3) context independent, that is randomly. Although, the dialogue dependent interruption seems superiors on the perceived dialogue flow, on the user preference and on the dialogue replies, participants rated presence highest for the context independent mechanism. According to Riva and Mantovani [26] indeed, we are present in an environment - real and/or virtual - when we are able, inside it, to intuitively transform our intentions in actions.

Based on the current status of presence research, much uncertainty remains about its usefulness for emotional responses induction and VRET. Only weak evidence for a relationship between presence and emotional responses has been found [22, 27], and no study has yet addressed this relationship. In other words, it is still unclear whether VR with low measured presence is characterized by a reduced effectiveness in inducing negative emotional responses compared to other traditional procedures.

According to these theoretical premises, we manipulated the level of presence by creating a virtual environment with two technical breakdowns: the head tracking was randomly reversed for 20 seconds during the experience and the avatars’ lips didn't move at the same time as their voice. We argued that these two kinds of technical breakdowns reduce the capability of VR in supporting the intentions of the user and its capability to elicit emotive reactions.

Hence, we compared changes following the exposure to an academic examination, one of the most universal examples of real-life stressors, in a sample of 39 undergraduate students. The same experience was offered using text, audio, video, and “impoverished” VR.

Data were entered into Microsoft Excel and analyzed using SPSS and STATA. Change in psychometric and physiological measures within subjects were calculated using repeated measure ANOVAs.

Virtual Reality Exposure Treatment (VRET) is an emerging approach for the treatment of anxiety disorders that offers different advantages over classical imaginative and in vivo exposure. Recently, different quantitative meta-analyses have been conducted on studies reporting VRET treatments supporting their efficacy in reducing anxiety-related symptoms [11, 12].

But is VR always an effective stressor? Are the technological breakdowns that may appear during such an experience a possible risk for its effectiveness? Different studies suggested a possible link between emotional response and the level of presence experienced in a virtual environment. For example, Michaud and colleagues [42] experimentally manipulated presence in a sample of heights phobics who had to take an elevator and perform tasks on a scaffold outside of a 15-story building. When the immersion in the virtual environment was conducted in a high-presence setting, the level of anxiety was significantly higher than when the immersion was conducted in a low-presence setting.

Riva and colleagues [43] explored, too, the relationship between presence and emotions experienced in VR. Their data showed a circular interaction between presence and emotions: on one side, the feeling of presence was greater in the "emotional" environments; on the other side, the emotional state was influenced by the level of presence.

Taken together these results underline the existence of a bi-directional relationship between presence and emotions. However, no previous studies investigated if the reduction of presence induced by technological breakdowns may be accompanied by a reduced results efficacy of VR as a stressor for exposure treatments.

To answer this question we compared changes following the exposure to an academic examination in a sample of 39 undergraduate students. The same experience was offered using text, audio, video, and VR. However, in the virtual environment we manipulated the experience introducing two technological breakdowns: the head tracking was randomly reversed for 20 seconds during the experience and the avatars’ lips didn't move at the same time.

In the study we first investigated if our emotional induction procedures were really effective in eliciting an emotional stress-related response. Results, in accordance with our hypothesis, showed that conditions were effective in inducing a stress-related response. We found a significant increase in anxiety scores measured through PMQ and a mirror decrease in relax scores after all our emotional procedures.

In addition, we assumed that the four emotive induction procedures elicits different intensity of responses and, in particular, we hypothesized that VR, when experienced with breakdowns, was less effective than traditional procedures. As hypothesized, results showed significative differences in the psychometric scores and psychophysiological correlates between the four experimental conditions. In particular, results on anxiety scores assessed through the PMQ showed that negative emotional response elicited through VR was less effective than AU procedure. Psychophysiological indexes confirmed this interesting result and, moreover, showed that VR was less effective in inducing a stress-related response compared to the other conditions. Results, in fact, showed an increased level from baseline in EMG_RMS index, a well-known index of emotional valence, in AU, VD and TX conditions, clearly indicate a higher negative emotion for the participants. On the other hand, VR condition showed no increased level from baseline. Also RSP_Rate showed similar results, indicating an accelerated respiration pattern in AU, VD and TX condition and a decelerated one in the VR condition. This confirms that Virtual Reality containing technological breakdowns has not been able to induce stress, besides the negative script. Finally, a further confirmation arises from HF, an important index of Heart Rate Variability. Decreases in HF is a well-recognized index of stress, more it has been recognized as an index that works well with different kind of stress, from a more cognitive working load to a situational stress due to psychological individual perception of environmental demands that exceed one's own adaptive ability to meet them [44]. In our study, lower level of HF for Audio, Video, and Text indicate a stress measured by the means of this cardiovascular index. Virtual Reality, on the opposite, showed an increased level. Thus also for this index we can confirm that technological breakdowns in Virtual Reality worked well.

Finally, results confirmed our fourth hypothesis. As we expected, “impoverished” VR induced a sense of presence similar to that experienced during the exposition to other media. This data confirms the hypothesis that technological breakdowns impacted the VR efficacy in inducing a sense of presence in the user, creating break in presence [26, 44]. Moreover, this fact was associated in our study with less effectiveness of VR in eliciting a negative emotional response compared to other media.

In conclusion, even if using VR it is possible to experience more presence than in real life [24, 45], or to generate a body transfer illusion [46], it is also possible to experience less presence than reading a text or watching a video. Therefore, if the focus is on designing applications capable of eliciting emotions with the goal of reducing or modifying them (as in psychological therapy), the environments must be able to induce a high feeling of presence. For this reason an important goal of VR developer is to guarantee that the virtual experience offered to their user is smooth and transparent enough to provide a high level of presence through a full support to the intentions of the user. Without it, the significant advantages offered by VR disappear and its emotional induction abilities are even lower than the ones provided by much cheaper media.