Introduction
The use of virtual reality (VR) in psychosocial interventions has expanded rapidly over the past decade, giving rise to an increasing number of VR‐based therapies (VRT) (Wiebe et al. 2022; Zeka et al. 2025).
Growing evidence supports the effectiveness of VRT across a wide range of mental health conditions, highlighting its potential to improve psychological care (Ghorayeb et al. 2026; Pedram and Piatkowski 2025). VR has been implemented in VR exposure therapy (VRET) for anxiety‐related disorders, obsessive‐compulsive disorder, post‐traumatic stress disorder and substance use disorders (Carl et al.
2019; Dellazizzo et al. 2020; Glavak‐Tkalić et al. 2025; Low et al.
2021; Van Loenen et al. 2022). The primary objective of VRET is to provide gradual, controlled exposure to disorder‐specific triggers in order to reduce fear and avoidance through emotional processing and habituation (Rothbaum 2010).
VR has also been integrated with cognitive behavioural therapy (CBT) to address psychotic symptoms, depression and substance use disorders (Baghaei et al. 2021; Bell et al. 2024; Spark et al.
2025; Thaysen‐Petersen et al. 2024). In addition, relational approaches involving dialogue with an avatar in VR have been developed for treatment‐resistant depression (TRD), auditory hallucinations in treatment‐resistant schizophrenia and cannabis use disorders (Dellazizzo et al.
2021; Giguere et al. 2024; Giguère et al. 2026; Liang et al.
2022; Smith et al. 2025). These psychotherapies aim to modify maladaptive intrapersonal and interpersonal relationships, improve emotion regulation and help patients regain control over their symptoms (Dellazizzo et al.
2021; Giguere et al. 2024; Giguère et al. 2026).
Among these approaches, VRT was specifically developed by our team for individuals with TRD to address the limitations of existing treatments for this population, given that major depressive disorder is among the most prevalent mental health disorders and a leading cause of disability worldwide. Indeed, approximately 30% to 55% of individuals with major depressive disorder remain symptomatic despite two adequate antidepressant trials or two psychotherapies (Cuijpers 2023; Cuijpers et al. 2023; Cuijpers et al.
2020; Li 2023; McIntyre et al. 2023; Zhdanava et al. 2021).
Given the substantial burden of TRD, including impaired functioning, reduced quality of life, increased suicide risk, greater healthcare utilization and medical and psychiatric comorbidities, as well as its broader impact on families and society (Arnaud et al. 2022; Cai et al. 2021; Kessler et al.
2003; Marx et al. 2023; Pan et al. 2019).
There was a critical need for novel, effective therapeutic interventions. VRT for TRD is a relational, dialogical and experiential psychotherapy in which patients engage in dialogue with avatars representing significant individuals or aspects of the self‐associated with the onset or maintenance of their depressive symptoms (Giguère et al. 2026).
Through these interactions, patients practice therapeutic strategies in real time, including emotion regulation, stress management, behavioural activation, cognitive restructuring, adaptive coping, self‐compassion and self‐esteem enhancement. By allowing patients to directly engage with emotionally challenging interpersonal situations and negative self‐representations, VRT may offer a relevant therapeutic approach for individuals with TRD who have experienced insufficient benefits from previous treatments.
Although VRT represents a promising approach for individuals with TRD and other psychiatric conditions, the therapeutic processes through which VR and avatar‐based dialogue may provide benefits beyond those of conventional psychotherapy without VR remain poorly understood. One hypothesis is that it enhances treatment by increasing immersion, allowing patients to experience simulated real‐life situations that closely replicate the emotional intensity of daily life, thereby facilitating the transfer of skills learned in VR to real‐life contexts (Bell et al. 2024; Colombo et al.
2021; Morina et al. 2015). Previous studies have proposed a few hypotheses to explain the added value of VR and/or avatar‐based dialogue in VRT; however, these have largely emerged from theoretical and clinical perspectives rather than patients' lived experiences.
A few studies have explored patients' perspectives across different forms of VRT. In social anxiety disorder, patients receiving group CBT incorporating exercises in VRET reported that VR enabled them to observe feared situations ‘from the outside’, without the associated emotional activation, facilitating a better understanding of their anxiety (Arnfred et al. 2021).
The study did not specifically aim to compare the VR and non‐VR components of the intervention. Other studies have explicitly asked patients about their perceptions of VRT compared with traditional psychotherapy. However, participants' previous experience with conventional psychotherapy without VR was either not assessed or was limited to only a subset of the sample (Levy et al.
2023; Matsangidou et al. 2022). In adolescents and young adults with depression or anxiety, a standardized VR‐based cognitive defusion exercise was perceived as more concrete, engaging, and easier to focus on compared to an equivalent audio‐based exercise (Bell et al.
2025). Although these studies have provided important insights into patients' experiences with VRT, questions remain regarding the therapeutic processes and perceived benefits involved in a complete course of therapist‐delivered VRT, particularly in relational approaches involving dialogue with an avatar in VR, and how these compare with those associated with conventional psychotherapy. To the best of our knowledge, no study has invited patients who have experienced both an entire course of VRT and conventional psychotherapy without VR to compare the two approaches.
In addition, no study has specifically examined therapist‐delivered VRT involving dialogue with an avatar.
Thus, this study aimed to explore patients' lived experiences of VRT to better understand the therapeutic processes and perceived benefits associated with VR and avatar‐based dialogue, and how these experiences differed from those associated with conventional psychotherapies. To this end, a qualitative analysis was conducted among individuals with treatment‐resistant major depressive disorder (TRD) who had completed VRT after previously receiving at least two conventional psychotherapies, allowing for a within‐subject comparison of their therapeutic experiences.
Methods
Participants
A total of 12 participants with TRD were randomly selected from a pilot clinical trial evaluating the efficacy of VRT for TRD (Giguère et al. 2026). Participants were required to be aged 18 years or older, as the intervention was developed and evaluated within clinical services specializing in adult mental health care.
Participants met diagnostic criteria for major depressive disorder, with symptoms persisting despite two or more pharmacological treatment trials. Diagnosis was confirmed using the Structured Clinical Interview for the DSM‐5 (Shankman et al. 2018).
Specifically, participants in this study had previously received at least two conventional psychotherapies (i.e., interventions without VR or avatar‐based dialogue), providing an experiential basis for comparing VRT for TRD with traditional psychotherapeutic approaches. Exclusion criteria for the main trial included the presence of any other severe mental disorder (e.g., psychotic or bipolar disorder), a neurological disease, a substance disorder or an unstable serious physical illness. In this qualitative study, no participants declined to take part.
The trial was conducted in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of the Centre Intégré Universitaire de Santé et de Services Sociaux de L'est‐de‐L'île‐de‐Montréal (CER‐CEMTL; approval No. 2024–3661). Written informed consent was obtained from all participants prior to study participation.
Avatar Therapy for TRD
Participants attended at least eight weekly sessions, each lasting 60–90 min. VRT for TRD immersed participants in emotions and contexts that precipitated and maintained depressive symptoms to develop strategies to improve mood and functioning. To achieve this, participants created several avatars representing significant individuals or internal self‐representations associated with their depressive symptoms (e.g., a family member or romantic partner involved in an unresolved interpersonal conflict, or maladaptive aspects of the self‐characterized by self‐critical, ruminative, demanding and disappointed).
The therapist embodied these avatars in VR from the patient's perspective during the intervention. The first two sessions are devoted to reviewing the patient's clinical history and creating the avatars. Each subsequent session follows a three‐phase structure: a review of the previous week and preparation for the immersion scenario, followed by the VR immersion phase (lasting approximately 10–20 min), and a post‐immersion debriefing.
The setup of the VR immersion phase is illustrated in Figure 1. For further details on the therapeutic protocol, see Giguère et al. (2026).
Qualitative Interview
Following completion of the therapy, a semi‐structured qualitative interview (i.e., guided by a predefined set of themes and questions, while allowing the interviewer flexibility to adapt the order and wording of questions and to use follow‐up prompts based on participants' responses) was conducted by a research nurse, either in person or by telephone (DeJonckheere and Vaughn 2019). The research nurse who conducted the interviews was not involved in the qualitative analyses or manuscript preparation. Participants were invited to discuss their overall experience of the therapy (VRT for TRD) and compare VRT for TRD with their previous experience of conventional psychotherapy.
In addition, participants were asked about the more difficult and most helpful aspects of VRT compared with other therapies they had previously received. They were also invited to describe their experience of interacting with an avatar and what this interaction brought them. The semi‐structured interview guide used with participants is provided in the Supporting Information.
All interviews have been audio‐recorded.
Qualitative Analysis
A total of 12 interviews (ranging from 9 to 24 min) were transcribed verbatim using BUZZ for Windows (Version 0.7.1), a closed transcription application, and subsequently verified by A.Da., S.G. or S.N. An inductive content analysis approach was adopted. This study was grounded in a realist ontology, assuming that therapeutic processes exist independently of how participants perceive or describe them (Maxwell 2012).
Epistemologically, the study adopted an intersubjectivist perspective, whereby interpretations are established through shared understanding and agreement among multiple individuals regarding the meaning of the text (Mascolo 2017). Accordingly, the researchers sought to describe participants' experiences as faithfully as possible while acknowledging that their own interpretations could influence the findings. First, three members of the research team (A.Da., S.G. and S.N.) read and annotated each transcript to identify emerging themes.
To reduce subjective bias from relying solely on written transcripts, all coders simultaneously listened to the corresponding audio recordings while reviewing the transcripts. Links between the emerging themes were then identified to organize them into broader, meaningful categories. During this iterative process, the themes clustered around three overarching categories: therapeutic processes, perceived benefits and confrontation.
Because therapeutic processes, perceived benefits, and confrontational experiences were examined as distinct dimensions of VRT, they were coded separately. However, these dimensions were not mutually exclusive. The same interview excerpt could therefore receive codes from more than one category when, for example, a participant described both a therapeutic process and its perceived benefit within the same statement.
Consequently, the coding process was conducted separately within each of the three overarching categories while allowing codes from different categories to overlap within the same idea segment when appropriate. This process enabled the creation of a preliminary theme grid and a dictionary that includes all themes, their definitions and associated examples. Before formal coding began, each transcript was segmented into idea units, with each verbatim defined as expressing a single idea.
Using QDA Miner (Version 2024.0.7), a qualitative data analysis software designed to organize, annotate, retrieve and analyse qualitative data (QDA Miner, n.d.), the first coding team (A.Da., S.G. and S.N.) coded all idea units. Coding continued until data saturation, defined as the absence of new themes and a stable coding framework across three consecutive transcripts. Following this initial coding, an independent coder (F.V.), who had not participated in developing the theme grid, coded all idea units using only the final codebook, which contained the themes and their definitions.
The first coding team and the independent coder reviewed disagreements jointly. After the discussion, the researchers determined whether the coding framework needed modifications, including adding new codes, revising code definitions, reorganizing the coding structure or refining the segmentation of idea units. Inter‐rater reliability was calculated before and after disagreement resolution using Scott's Pi coefficient generated by QDA Miner.
Scott's Pi measures agreement beyond chance and typically ranges from 0 to 1, with higher values indicating stronger agreement. Negative values are also possible and indicate agreement lower than expected by chance. Idea units that were not assigned any code by either coder (common absence) were considered agreements in the calculation of Scott's Pi.
Agreement was interpreted according to the following benchmarks: poor (< 0.00), slight (0.00–0.20), fair (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80) and almost perfect (0.81–1.00) (Wombacher 2017). When the two coding groups could not reach consensus, disagreements were discussed with a third independent team member (K.P.) until consensus was achieved. The number of participants for whom each code was present was extracted using QDA Miner.
In addition, code co‐occurrence across the three overarching categories was examined by calculating the number of participants for whom codes from different categories overlapped within the same idea segment. Themes and verbatim were translated from French into English for this article.
The four coders had varying levels of research and clinical experience. Three were research volunteers, while one was a doctoral student involved in the VRT project. Two coders had a clinical background: one was a first‐year medical student and the other a second‐year psychiatry resident.
Neither had delivered VRT, and their experience with psychotherapy was limited to conventional approaches. Recognizing that joining a research laboratory specializing in VRT may reflect a pre‐existing interest in these interventions, the coders remained aware of the potential influence of such interests on their interpretations throughout the analysis. Particular attention was paid to ensuring that interpretations remained grounded in participants' accounts and reflected both positive and critical experiences of VRT.
Results
Patient Characteristics
Participants were evenly distributed by sex. Age ranged from 27 to 71 years. The mean age of depression onset was 16.3 years (SD = 10.8).
Regarding previous treatments, participants had undergone multiple pharmacotherapy trials (mean = 6.8, SD = 5.0) and an average of four psychotherapies, with a range of two to 10 prior psychotherapy experiences. Table 1 presents detailed sample characteristics. Participants completed an average of 10.75 therapy sessions (SD = 1.54), including 6.33 immersion sessions (SD = 1.83), which lasted 11.8 (SD = 3.4).
Participants encountered 2.4 avatars representing significant others (e.g., spouses, parents or children), including deceased individuals and people involved in unresolved relationships. Participants also interacted with an average of 3.8 sessions with an avatar representing different aspects of themselves related to depression, such as demanding, ashamed, pessimistic, self‐critical, incompetent, dissatisfied or low self‐esteem.
Themes
Analysis of the 12 interviews identified six themes related to therapeutic processes and three themes related to their associated benefits, with confrontation emerging as both a therapeutic process and a benefit. Below is a description of each theme, supported by illustrative excerpts from participants.
Therapeutic Processes
Externalization of internal psychological processes refers to representing a mental process (e.g., a cognitive schema) in a tangible, embodied form by creating and interacting with an avatar. Participants reported that this process helped them identify and name the problem, explore its influence on their lives and, eventually (through various therapeutic processes), deconstruct the narrative that had developed around it. Some participants also reported that interacting with an avatar representing themselves facilitated psychological distancing from the problem and/or reduced opportunities for avoidance associated with their habitual thinking patterns.
P3: ‘It was as if I could distance myself enough from my depressed self and see myself as someone trying to help another person with depression. It made me realize things like that more easily. I know my escape routes, such as the excuses I tend to make, and I didn't let myself fall back on those escape routes because I could anticipate them’.
Emotional investment is the process through which participants reported being emotionally engaged in the therapeutic experience, particularly through emotional openness and genuine involvement in dialogues with the avatars.
P2: ‘When he embodied the characters who confronted me, everything was in place for me to fully experience it. I was able to relive the emotions and work through them because, with the confrontation, it was as if I was reliving them, resolving them, and understanding them better. In other therapies, I understood those emotions, but I didn't actually feel them.
That's really the biggest difference’.
Atypical therapeutic stance refers to the therapist adopting a character through the avatar, allowing them to express statements and address topics in ways consistent with the personality being portrayed. According to participants, this was important for maintaining the immersive nature of the role‐play and resulted in interactions that differed from those typically encountered in conventional psychotherapy. Participants also reported that the therapist‐controlled avatar offered a perspective distinct from the therapist's.
In addition, some participants perceived the therapist's nonverbal cues as inaccessible during avatar‐mediated interactions.
P6: ‘It was the fact that Dr. Dumais was able to confront me with my own thoughts. He played my role, but he was able to challenge my dysfunctional thinking.
Before the therapy, I would tell him what I thought about myself, how stupid I felt, how ignorant I felt, how unproductive I felt. Later, when we did the avatar exercise, he repeated those very words back to me. That's when it really hit me: [when I was] hearing someone else say the words I used to tear myself down’.
Therapy structure refers to the organization of therapy as a personalized, concrete, goal‐oriented approach focused on a predefined issue and conducted within a safe emotional framework. Participants described several structural processes, including a preparation phase (pre‐immersion), during which the patient and therapist collaboratively developed the scenario. They also emphasized the importance of the therapist exploring the targeted issues in depth to capture the essence of the character to be embodied by the avatar from the patient's perspective.
Other structural elements included the progression of avatars based on a gradual increase in emotional intensity and the systematic debriefing of each immersion experience (post‐immersion).
P11: ‘But, obviously, beyond the use of VR and the avatar tool itself, it was different because we prepared beforehand. We talked about the difficulties with the psychiatrist, and we discussed the issues we wanted to address using the avatar. So it was different because you don't do that in traditional therapy.
I mean, you also talk about the difficulties you've experienced during the day, but it's more… It has a different structure. The goals are a bit more concrete and a bit less abstract than they would be in traditional therapy’.
Immersive role‐play refers to how engaged participants felt in the role‐play experience. Participants described immersion along a continuum, ranging from experiences characterized by breaks in immersion to those requiring conscious imaginative effort, to moments in which the interaction felt real. The avatar and VR environment can facilitate engagement in the role‐play regardless of the avatar's visual realism.
P7: ‘Once again, I'll come back to the role‐play. Having done it before with other therapists, I can say that avatar therapy makes it much more immersive’.
Putting theory into practice involves participants applying theoretical knowledge or therapeutic skills (e.g., changing self‐talk, enhancing self‐worth, practicing assertiveness, improving self‐expression, engaging in behavioural activation or resolving conflicts) through communication with the avatar. The avatar provided opportunities to communicate with oneself or others in ways not attempted in real life or that differed from usual interaction patterns.
P1: ‘I think what is interesting, for example, if I go back to CBT and psychoanalytic approaches, is that we work a lot on thought patterns. We are often focused inwards, which is also very good and well established in terms of evidence. But sometimes, in mental health, the issue is that we tend to ruminate, so it often stays in the mind.
We ruminate, we keep ruminating, along with all the internal work we are supposed to do. There was something interesting in the fact that, even if it's not a real person and it's interactive and semi‐fictional, it already allows you to put things into practice. For example, when I leave a session, I'm not sitting there internally ruminating about it afterwards.
I know that some work has been done with someone’.
Benefits of the Therapeutic Process
Participants reported developing self‐determination, which is the ability or right of an individual to make their own decisions, control their own actions and determine their own destiny without external pressure.
P3: ‘Whereas by speaking to my past depressed self and the version of myself who is possibly not depressed, it gave me the opportunity to see that it was possible: that it depended on what I was doing, how I was, whether it was something I could change, what I could change. It's more of an opportunity, I suppose’.
Development of mentalization capacities refers to any statement in which participants demonstrated an emerging or enhanced ability to reflect on their own mental states, including thoughts, emotions, beliefs, intentions, and attitudes. This includes recognizing, evaluating and restructuring mental states, including increased self‐awareness, shifts in perspective and self‐reflection. The development of mentalization capacities was described as occurring during therapy sessions.
P6: ‘Hearing someone else express the same thoughts I had made me realize how I can be toxic toward myself’.
Integration of theory involves transferring techniques practiced with the avatar into participants' daily lives. Participants described applying cognitive or behavioral changes developed during therapy, including changes in beliefs or internal dialogue about themselves, others, as well as changes in assertiveness, self‐esteem and behavioral activation.
P12: ‘The thing is that after each session, I would come in because, for me, my avatar was mainly myself, I was able to convince myself a lot that I deserved to be loved, that I deserved to set boundaries. It was as if, after the session, it kept working afterward, and I was like: well, no. Even though I know the theory, after the session I can really apply it and actually feel that indeed, I have the right to set boundaries, I have the right to be loved, I have the right to deserve better than this.
So yeah, I think that's one of the most striking things: you actually feel it afterwards, you're not just holding it on a cognitive level’.
Confrontation involves being exposed to and/or interacting with an avatar representing a person who triggers negative emotions or a part of oneself associated with negative internal dialogue. Participants may experience this interaction as challenging or emotionally confronting and may respond by actively facing, questioning or opposing the avatar and what it represents.
P9: ‘It was more confrontational; the reflection was deeper, so it was definitely more confrontational with myself. […] There was confrontation, and it raised questions, because in Avatar therapy it was basically me, so I could somewhat control the other person's responses, but here I couldn't’.
Frequency of Themes Reported by Each Patient
As shown in Figure 2, all therapeutic process themes were endorsed by more than half of the participants. Immersive role‐play was the most frequently reported theme (n = 11), followed by put into practice (n = 10), structure of therapy (n = 9), emotional investment (n = 8) and both atypical therapeutic stance and externalization of an internal process (n = 7). Regarding the benefits associated with therapeutic processes, integration of theory was the most frequently reported theme (n = 8), followed by development of mentalization capacities (n = 7).
One‐third of participants reported self‐determination (n = 4). Finally, three‐quarters of participants reported confrontation (n = 9).
Frequency of Code Co‐Occurrence Across Three Theme Categories
Table 2 presents the relationships between therapeutic process themes and the benefits reported by participants. Overall, putting into practice was the therapeutic process most frequently associated with reported benefits, particularly integration of theory (n = 4), self‐determination (n = 3) and development of mentalization capacities (n = 2). Externalization of an internal process was associated with the development of mentalization capacities (n = 2) and integration of theory (n = 2), and with self‐determination (n = 1).
Emotional investment was associated with both the development of mentalization capacities (n = 2) and the integration of theory (n = 1), whereas an atypical therapeutic stance was linked to either the development of mentalization capacities and the integration of theory (n = 1). No direct associations were identified between immersive role‐play or the structure of therapy and the reported benefits.
In addition, confrontation emerged as a transversal theme linking therapeutic processes and therapeutic benefits. Patients associated the confrontational aspect of therapy with all identified therapeutic processes. Confrontation was also linked to therapeutic benefits, most frequently associated with the development of mentalization capacities (n = 5) (Table 3).
Coding Agreement
The percentage agreement for classifying verbatim excerpts by the two independent coding teams was 91.5% for therapeutic processes (Scott's Pi = 0.69) and 86.3% for benefits associated with therapeutic processes (Scott's Pi = 0.69). Following a consensus meeting during which discrepancies in coding were discussed and resolved, inter‐coder agreement increased. Specifically, agreement for therapeutic processes increased to 99.3% (Scott's Pi = 0.98), while agreement for benefits associated with therapeutic processes reached 100% (Scott's Pi = 1.00).
Table 4 presents the agreement for each subtheme.
Coding Challenges and Reflexivity
The relatively lower inter‐coder agreement for Putting theory into practice and Integration of theory reflected initial conceptual overlaps with other themes. Refining theme definitions and discussing them between the two coding teams contributed to greater conceptual clarity and more consistent coding decisions. For Putting theory into practice, discrepancies arose because this central therapeutic process overlapped with elements in other themes.
The second coder, who was not involved in developing the coding framework, tended to prioritize other themes without consistently favouring any one theme. During discrepancy resolution, discussions focused on identifying the predominant therapeutic process described in each participant's account. For integration of theory, discrepancies primarily reflected its conceptual overlap with Development of mentalization capacities, as both themes involved cognitive processes.
To distinguish them, the definitions were refined to emphasize the difference between intrapsychic reflective processes occurring during therapy (development of mentalization capacities) and the transfer and application of therapeutic insights and strategies to everyday life (integration of theory). These discussions clarified conceptual boundaries between themes and resolved coding discrepancies through consensus.
Discussion
This study aimed to explore the therapeutic processes and perceived benefits of a VRT for TRD from the perspective of patients who had completed both a full course of VRT and at least two prior courses of conventional psychotherapy. Drawing on direct personal experience with both approaches, the study sought to identify features that distinguished VRT from conventional psychotherapy without VR and avatar dialogue. The analysis of participants' interviews identified six therapeutic processes.
The most frequently reported theme was immersive role‐play, followed by putting theory into practice, therapy structure, emotional investment, an atypical therapeutic stance and externalization of internal psychological processes. In addition, participants associated three benefits with these processes: integration of theory, development of mentalization capacities and self‐determination. Finally, confrontation emerged as a transversal theme across both the therapeutic processes and perceived benefits.
The following sections discuss each theme as perceived by participants, highlighting the therapeutic processes that distinguish VRT from, or are shared with, conventional psychotherapies.
Experiential Engagement: Immersive Role‐Play and Emotional Investment
Immersive role‐play was the most frequently reported therapeutic process. Although role‐play is an established component of interpersonal psychotherapy (IPT) and emotion‐focused therapy (EFT) (Cuijpers et al. 2016; Goldman and Goldstein 2023), participants described its immersive nature as distinctive in VRT.
Unlike conventional techniques such as the empty‐chair exercise, in which patients generate both sides of the interaction (Goldman and Goldstein 2023), VRT involves a therapist‐controlled avatar that can respond in ways the patient neither anticipates nor controls. This may create a more dynamic and realistic interpersonal exchange that more closely approximates emotionally meaningful real‐life interactions.
Participants frequently framed the contrast with prior psychotherapy in terms of feeling versus understanding: emotions previously understood were, in VRT, described as relived, worked through, and resolved. This distinction aligns with the initial hypothesis that VRT recreates emotional intensity approximating real‐life situations [14, 20, 21], extending it by grounding it in patients' own accounts.
From Enactment to Integration
Putting theory into practice was the second most frequently reported process and the one most consistently associated with reported benefits, particularly integration of theory, self‐determination, and the development of mentalization capacities. Participants described VRT as enabling them to actively apply knowledge gained through conventional psychotherapies, suggesting its perceived benefits may extend beyond cognitive understanding to experiential application. This pattern suggests that the perceived benefits of VRT may lie not only in facilitating cognitive understanding, but also in providing an opportunity to experience and actively apply previously acquired therapeutic knowledge.
Participants often described already understanding certain emotions, behaviours or therapeutic concepts through previous psychotherapies, while emphasizing that VRT enabled them to put this understanding into practice, which in turn allowed for proper integration. This distinction is consistent with evidence suggesting that incorporating experiential therapeutic strategies, such as behavioural experiments, into cognitive therapy may provide additional benefits compared with relying predominantly on verbal cognitive interventions (Ma and Lo 2022; Yilmaz et al. 2025).
Building on previously acquired cognitive understanding, VRT may provide an immersive, emotionally engaging context in which active engagement with an avatar enables participants to put therapeutic knowledge into practice. Thus, rather than replacing the cognitive work accomplished in previous therapies, VRT could provide a context in which previously acquired therapeutic knowledge can be enacted, emotionally processed and more fully integrated.
The structural features participants valued partly overlap with established psychotherapies for depression, which are often structured, collaborative, goal‐oriented and organized around identifying specific problems and selecting therapeutic strategies to address them (Gautam et al. 2020; Parikh et al. 2016).
However, the preparation–immersion–debriefing sequence may make therapeutic goals more concrete while providing a structured, predictable transition from preparation to emotionally salient experiential work and subsequent reflection, potentially allowing participants to approach difficult interpersonal material while maintaining a sense of emotional safety. This interpretation aligns with the literature, which highlights VR's capacity to provide individualized, controlled therapeutic experiences in which the intensity of emotionally salient stimuli can be adjusted (Bell et al. 2020; North and North 2016).
Externalization of an Internal Process and the Atypical Therapeutic Stance
Participants reported an average of 3.8 sessions with an avatar representing an aspect of themselves associated with their depression. Externalizing an internal process in this way was described as producing psychological distance from self‐critical cognition, allowing participants to relate to their own depressive discourse as something observable and addressable rather than as an ambient truth. Emerging evidence supports these findings, notably that self‐compassion training via a ‘self‐to‐self’ VR simulation paradigm is significantly more effective at reducing self‐criticism than expressing compassion to a suffering avatar in VR that does not represent the self and is also more effective than computer‐based self‐compassion training (Xia et al.
2026). VRT thus appears to enable one to more concretely face a projection of their internal world, allowing for growth that could not otherwise be easily attained through more traditional mediums. This is of particular interest for TRD, where entrenched negative self‐schemas and rumination are prominent features and where introspective work may itself become part of the ruminative process it seeks to address (Ehret et al.
2015; Visted et al. 2018).
The atypical therapeutic stance appears to be closely related to the externalization of an internal process, primarily through the therapist's embodiment of the different avatars. Within these embodiments, they may voice positions and adopt a directness incongruent with a conventional therapeutic posture. Participants experienced this as offering a perspective distinct from the therapist's own.
This resembles the “disciplined personal involvement” described in the Cognitive‐Behavioral Analysis System of Psychotherapy, in which the therapist departs from neutrality deliberately and strategically to disconfirm maladaptive interpersonal expectations (Diehm et al. 2025). VRT may achieve a comparable effect through a different route: the avatar functions as a third position in a therapeutic triad, which appears to permit the exploration of more sensitive material while preserving the therapeutic relationship itself.
Development of Mentalization Capacities and Self‐Determination
Enhanced self‐determination and the development of mentalization capacities were reported as therapeutic benefits of VRT by more than half and approximately one‐third of participants, respectively. Both represent clinically relevant dimensions in the treatment of depression and are not unique to VRT, as mentalization is explicitly targeted in mentalization‐based and psychodynamic therapies (Fonagy et al. 2020; Luyten et al.
2024), whereas self‐determination is emphasized in Cognitive‐Behavioral Analysis System of Psychotherapy (Locke et al. 2017).
However, our findings suggest that VRT may offer a distinctive experiential approach to fostering these capacities. Putting theory into practice was the therapeutic process most frequently associated with enhanced self‐determination, suggesting that actively engaging in dialogue with avatars and practicing new responses in real time may help patients move beyond cognitive understanding towards expressing their needs, making decisions, and exercising greater control over their actions. In contrast, the development of mentalization capacities appeared to emerge from a broader range of therapeutic processes, with confrontation being the most prominent transversal theme.
Specifically, being confronted with avatars reproducing maladaptive interpersonal patterns or negative self‐representations may have encouraged patients to recognize, question, and reconsider their habitual thoughts, emotional reactions, and beliefs about themselves. Thus, while these therapeutic benefits are also addressed in conventional psychotherapies, VRT may provide a distinctive context in which patients can experience and reflect on these patterns through direct interactions.
Confrontation as a Transversal Process
Confrontation was associated with all six therapeutic processes and, among benefits, most frequently with the development of mentalization capacities, suggesting that it may represent a central component of the therapeutic experience of VRT. Participants characterized it as uncomfortable but useful, and as generating a depth of reflection that they had not encountered in previous therapies. Confrontation is not unique to VRT and is embedded in other therapeutic approaches.
Exposure‐based therapies involve sustained engagement with emotionally aversive stimuli rather than avoidance (Carl et al. 2019; Rothbaum 2010; Van Loenen et al. 2022), while experiential techniques such as Chairwork use role‐play to directly confront and explore internal conflicts, significant relationships and unresolved emotional experiences (Ottingerová et al.
2026). However, the transversal nature of confrontation observed in the present study suggests that its role in VRT may extend beyond confronting a difficult experience itself. Rather, confrontation occurred alongside several distinctive therapeutic processes, including immersive role‐play, putting theory into practice, emotional investment, externalization of internal psychological processes and the atypical therapeutic stance.
The convergence of these processes may strengthen the confrontational experience by making internal or relational difficulties more concrete, emotionally salient and difficult to avoid while requiring patients to actively apply therapeutic strategies in real time.
The association of confrontation and the development of mentalization capacities merits particular attention. A recent systematic review found that mentalizing may act as both a mediator and a moderator of psychotherapeutic outcomes (Luyten et al. 2024).
In VRT, the confrontational encounter appears to create conditions in which the patient must be exposed to a mental state that challenges their own or recognize their own mental state as one perspective among others, which is an important aspect of mentalization. The fact that this benefit was linked to confrontation more often than any other process suggests that the discomfort reported by participants may be functional rather than incidental.
This study deepened understanding of patients' perspectives on therapeutic processes and the perceived benefits of VRT, and examined how these relate to those associated with conventional psychotherapeutic approaches. To our knowledge, this is the first study to specifically examine therapist‐delivered VRT involving dialogue with an avatar among TRD patients who had completed both courses of VRT and previous courses of conventional psychotherapy without VR, enabling a direct comparison of the two approaches. Nevertheless, a few limitations must be acknowledged.
First, the sample included only participants who completed VRT and agreed to be interviewed, potentially favouring more positive experiences; participants who discontinued the parent trial were not represented. Future studies should include participants who withdrew from or responded poorly to VRT. Second, the relationship between the identified themes and treatment outcomes was beyond the scope of this study.
Examining whether specific themes are associated with outcomes could help identify processes related to treatment response. Third, one therapist treated all participants, limiting generalizability of therapist‐dependent processes, particularly the atypical therapeutic stance. Multi‐therapist studies are needed, especially given that standardized training and competency requirements for avatar‐based therapy remain to be established.
Fourth, the use of semi‐structured interviews presents both advantages and potential limitations. While the flexibility inherent in this approach enabled richer and more detailed responses, it may also have introduced some variability across interviews and influenced participants' responses through differences in probing or question formulation. Finally, although the patient perspective is essential for understanding the experience of therapeutic value, it may not capture all mechanisms underlying VRT.
Integrating patient and therapist perspectives with analyses of avatar–patient dialogues could provide a more comprehensive understanding of these mechanisms.
Conclusions
Exploring the therapeutic processes and perceived benefits of VRT, and comparing them with those associated with conventional psychotherapeutic approaches, revealed a shift from understanding to enactment. Across accounts, participants described a transition from analysing relational difficulties to directly encountering them in the presence of avatars, under conditions emotionally close to real life, so newly acquired strategies could be practiced rather than merely planned. Several of these processes have recognized analogs in IPT, EFT, and CBT, which situated VRT within established psychotherapeutic principles rather than apart from them.
However, this study identified the features of VRT that patients perceived as distinct from established psychotherapeutic approaches. Specifically, the immersive role‐play component, combined with the therapist's atypical therapeutic stance through dialogue with avatars, enhanced emotional engagement, facilitated the externalization of internal processes and enabled patients to translate therapeutic concepts into practice, thereby fostering their integration into everyday life. The processes described by our participants suggest that this shift, from anticipatory preparation to accompanied enactment, may be a substantial part of what patients experience as the benefits of VR in psychotherapy.
These findings also highlight a limitation of conventional psychotherapy: It often cannot provide therapist support within the real‐world situations in which emotional difficulties are experienced. Finally, although the present analysis concerns TRD, several identified processes may not be disorder‐specific and could extend to other forms of dialogical VRT. Further studies in other clinical populations are needed to determine the generalizability of these findings.
Funding
This research was funded by Fondation J.‐ Louis Lévesque et le Centre de recherche de l'Institut Universitaire en Santé Mentale de Montréal.
Ethics Statement
The trial was conducted in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of the Centre Intégré Universitaire de Santé et de Services Sociaux de L'est‐de‐l'Île‐de‐Montréal (CER‐CEMTL; Approval No. 2024‐3661). Written informed consent was obtained from all participants prior to study participation.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information