Specific phobia is a common anxiety disorder characterized by an excessive fear response to specific objects or situations, fears that are greater than justified, and accompanying avoidance that causes daily impairment (Oosterlink et al., 2009). A specific phobia for spiders (arachnophobia) has a prevalence rate of 2.7%, but research additionally shows that fear of spiders is experienced by 23.3% of the general sample (approximately 35.8% in women and 10.9% in men) and fears are particularly common in early adulthood (Oosterlink et al., 2009). Although many individuals do not seek treatment, success rates of exposure therapy in this population are high (Choy et al., 2007; Gadermann et al., 2012). Exposure therapy is an intervention from the field of cognitive behavioral therapy (CBT) which involves confronting individuals with the subject of their fears in a controlled and hierarchical manner. Single-session exposure treatment has demonstrated longstanding positive effects, with a significantly reduced burden of anxiety in 90% of treatment participants at 4-year follow-up (Öst, 1989).

Digital mental health (DMH) services can nevertheless provide a useful addition to this treatment. Extended Reality (XR), which includes Virtual Reality (VR) and Augmented Reality (AR), has gathered a considerable evidence base in anxiety treatment (De Witte et al., 2021). The main difference between VR and AR lies in the degree of contact with the real world. While VR aims to immerse you in a computer-generated environment, AR allows you to maintain your sense of presence in the actual world but adds virtual elements (e.g., computer models of animals) to it. Research has supported the therapeutic effectiveness of exposure therapy in VR, and the first studies exploring Augmented Reality Exposure Therapy (ARET) have also been promising (Botella et al., 2017; Suso-Ribera et al., 2019; Albakri et al., 2022). By viewing your own body and behaviors in interaction with a virtual element, AR can provide an ecologically valid exposure experience that can enhance therapy engagement (Baus & Bouchard, 2014). ARET can be defined as “an add-on to or type of exposure therapy in which clients can interact with a virtual representation of an object or situation in the actual world to help them reduce fear or avoidance” (De Witte et al., 2020).

Benefits of DMH interventions include the provision of objects and contexts that are readily available for therapy, independent of seasonal variations (e.g., spiders) or local availability (e.g., snakes). Therapists can easily direct these animals and elicit behaviors that can activate individual fear responses (e.g., locomotion, vocalization), facilitating gradual and hierarchical exposure. Exposure experiences can easily be provided in multiple contexts and through homework exercises (to diminish the risk of renewal of fear; Bandarian-Balooch et al., 2015). Additionally, new technologies can promote acceptability of anxiety treatment in participants, since previous work has indicated a higher willingness to engage in exposure treatment with virtual spiders as opposed to their real counterparts (Garcia-Palacios et al., 2001,2007).

Similar to VR though, AR experiences need to be perceived as realistic (i.e., there needs to be a convergence between expectations and experiences of the user) and promote “co-presence” with the virtual elements (Baus & Bouchard, 2014). Additionally, the activation of the fear response is the key element of exposure therapy, as proposed by several theoretical frameworks, such as emotional processing theory (EPT; Foa & Kozak, 1986) and inhibitory learning theory (Craske et al., 2014). Therefore, the virtual representation in AR needs to be able to evoke discomfort and anxiety to be useful for exposure therapy in clinical samples. Immersive experiences can be offered through multiple technological modalities, which could influence AR experiences and resulting co-presence and activation of the fear response. Previous research on AR in specific phobia has employed several technology-based modalities, such as a projection-based display (Wrzesien et al., 2011), a smartphone application (Botella et al., 2011; Zimmer et al., 2021), or a Head-Mounted Display (HMD) (Juan et al., 2004; Botella et al., 2005). These modalities differ in technological embodiment, visual presentation, and user experience (UX). Technological embodiment refers to the level to which experiences are integrated with the human body and extend natural bodily capacities (Flavián et al., 2019). When looking at the technological embodiment continuum, smartphones are viewed as external portable devices while wearable HMDs are situated closer to the internal technologies (promoting higher levels of technological embodiment). In this capacity, HMDs can strengthen sense of presence and promote visual appeal in customer experience (Orús et al., 2021), which could potentially translate to better quality of the therapeutic AR experience. In terms of visual presentation, a smartphone displays the object superimposed over the real-world surroundings while the optical lenses of an AR headset (e.g., HoloLens 2) result in a more transparent object appearance. Finally, UX has been deemed important for DMH adoption and behavioral program engagement (Ramtohul, 2015; Hentati et al., 2021), and different AR devices vary in user interfaces and interactions. These three important factors could shape modality suitability for ARET, but no studies have compared AR modalities directly (to our knowledge).

The current study aims to investigate the potential of the Microsoft HoloLens 2 headset for future use in exposure therapy by comparing its performance and personal preferences in comparison with a smartphone AR application. A within-subjects counterbalanced study design with two hierarchical Behavioral Approach Tasks (BATs) was implemented in a general sample to gain insight into 1) the extent to which a virtual spider viewed through an AR headset or smartphone application evokes discomfort and anxiety (as shown by self-report and physiological measures) and 2) participants’ personal experiences with and preferences for both technological modalities. The more continuous and handsfree visual experience of a headset can contribute to technological embodiment, which could improve the quality of the AR experience as compared to a handheld device. Therefore, our primary hypothesis is that exposure to virtual spiders through the HoloLens 2 will result in higher anxiety levels compared to the smartphone condition. Since technological embodiment has been linked to immersion and higher levels of experienced realism can lead to a stronger increase in anxiety in a Behavioral Approach Task (BAT) with virtual animals (De Witte et al., 2020), our second hypothesis is that the HoloLens 2 will lead to higher levels of subjective experienced realism compared to the smartphone condition and that experienced realism is related to individual differences in anxiety levels in the BAT task. Finally, we explored personal preferences regarding the implemented technology modality.

In both AR modalities self-reported discomfort increased during the BAT task, indicating that increasing proximity and movement of a virtual spider can induce subjective anxiety. This was also supported by an increase in SC in the HoloLens 2 modality. This finding is in line with previous work stating that virtual animals can implicate a physiological challenge and evoke associated subjective anxiety (Diemer et al., 2014; De Witte et al., 2020). Self-reported anxiety levels during the BAT tasks indicate that movement of the spider itself induces a stronger rise in fear as compared to movement of the participant in the room, which is in line with findings from Van Uijen et al. (2015) suggesting that active approach adds to exposure effects.

About half of participants reported the spider being more frightening in the HoloLens 2 modality while the other half thought the spider in the smartphone condition was more frightening and, while anxiety increased with difficulty in both conditions, there was no general effect of modality on the SUDs. However, anxiety levels did differ between conditions when the spider started walking towards the participant for the first time. This spider locomotion was perceived as more threatening in the continuous headset experience. Higher levels of technological embodiment could make this activity a stronger sensory stimulation, contributing to an elevated fear response. However, a difference in perceived level of control could also influence the experience since participants in the smartphone condition were asked to press a button that stopped the movement of the spider when it reached the 50 cm mark, as opposed to the spider stopping automatically in the HoloLens 2 application. Experiencing control over stressful situations can diminish distress, although studies manipulating level of control during exposure experience have not been able to replicate this finding (Maier & Watkins, 2005; Healey et al., 2017). Nevertheless, participants with greater control during a computerized exposure task were able to approach closer towards a spider (Healey et al., 2017). Perceptual control theory suggests that exposure is more effective when clients have increased control over the process, for example in self-guided virtual reality exposure therapy (Healey et al., 2017; Premkumar et al., 2021).

In contrast to our hypothesis, modalities did not differ in experienced realism. While higher technological embodiment could increase immersion in VR applications (Orús et al., 2021), qualitative data suggest that transparency of the spider was a disadvantage for the HoloLens 2 in the current study. Nevertheless, higher experienced realism was associated with increased anxiety within the HoloLens 2 condition but not in the smartphone condition, contrary to our previous work (De Witte et al., 2020). While reviews and meta-analyses emphasize the importance of realism to attain effective exposure in XR, empirical research into perceived realism in ARET is lacking (Powers & Emmelkamp, 2008; Baus & Bouchard, 2014). We also need better instruments to measure realism in AR since the IPQ is not tailored an AR context and internal consistency was suboptimal in the current sample.

Participants’ preferences regarding modality were mixed. The continuous handsfree integration of the spider in the physical surrounding resulted in approximately half the sample preferring the HoloLens 2 headset. However, the translucency of the spider in this modality made the remainder of the sample prefer the more realistic representation of the spider in the smartphone application. In practice, the optimal ARET modality should therefore depend on the specific patient’s preference. However, the future of XR lies in merging VR and AR functionalities in one headset allowing for passthrough VR (Hillmann, 2021), a new technology that could accommodate the drawbacks of both modalities. Passthrough VR refers to the implementation of a VR headset which captures the physical surroundings through a camera and superimposes the virtual object. Therefore, the spider could look like the current smartphone application but be presented through a continuous HMD experience. Other potential future avenues for designers of XR exposure are the integration of haptic feedback, embedding it in a self-help application, personalization (based on physiological responses), and gamification (Wiederhold & Bouchard, 2014; Kurscheidt et al., 2019; Mahmoudi-Nejad et al., 2021; Dibbets, & Schruers, 2022; Toffolo et al., 2022).

The current study has several limitations. Some participants went through the experimental procedure more quickly than expected, which reduced physiology settling in periods and resulted in exclusion of some participants from physiological analyses to guarantee a stable measurement signal. While cybersickness can also be present in AR (yet generally to a lesser extent than VR, especially with short exposure times) and lead to increased SC, effects have mostly been found on forehead SC measurements (Gavgani et al., 2017; Hughes et al., 2020). Peripheral measures, such as the current one, are generally linked to task-based arousal. Using the same virtual spider in both sequentially administered BAT tasks could attenuate anxiety levels in the second condition, but counterbalancing the order prevented bias in the overall findings. Findings in a general population cannot be directly transferred to phobic individuals. There is no gender parity in the sample. Future controlled research (RCT) should provide insight into the active ingredients promoting the anxiety-inducing potential of animals presented through different AR modalities and fully harmonize applications running on different modalities.

The current study assessed whether virtual spiders can induce anxiety and could consequently be a good candidate for inclusion in ARET in line with existing theory (Foa & Kozak, 1986; Craske et al., 2014). Virtual spiders induced equivalent levels of subjective anxiety in an AR smartphone application or HMD. Perceived realism did not differ between conditions and preferences regarding the technology of choice varied strongly. As it stands, the AR smartphone application has a larger clinical potential since smartphones are more ubiquitous and affordable, while the HoloLens 2 is uncommon and situated in a higher price range. Future research should further clarify the role of perceived realism and assess whether new XR modalities and features could promote the therapeutic potential of ARET.