The MVP platform was developed to support avatar-based psychotherapy sessions involving real-time facial expression and nonverbal cue synchronization. Participants could access the platform using a PC, smartphone, or tablet. Camera activation was required throughout all sessions to enable accurate reflection of nonverbal expressions through anonymous avatars. The platform included multiple virtual environments, namely a pretherapy preparation space, a private therapy room, a posttherapy reflection area, and a community cafeteria for indirect social interaction.

Anonymous avatars were designed to prioritize behavioral realism over visual similarity to participants’ physical appearances. Participants were able to select from a range of predefined avatar options; however, fully free-form customization of physical appearance was not supported. The avatar system captured and synchronized real-time facial expressions, eye movements, gestures, and head movements, while maintaining participant anonymity. This design approach was based on prior research suggesting that behavioral authenticity facilitates therapeutic communication, whereas anonymity reduces barriers to self-disclosure[37]. Real-time nonverbal expression synchronization was incorporated to preserve emotional communication while maintaining participant anonymity. Figure 2 illustrates an overview of the metaverse platform and avatar design.

Professional psychotherapists were randomly assigned to participants, with the same therapist conducting all 4 sessions for each participant to ensure therapeutic continuity. Four licensed psychotherapists participated in this study, all with over 10 years of clinical experience; 2 had prior telepsychotherapy experience. All therapists received specialized training in metaverse platform navigation, avatar-mediated communication techniques, and management of technical difficulties during virtual sessions. Therapists delivered individual supportive psychotherapy based on established supportive therapy principles—including empathic listening, emotional validation, and therapeutic alliance building—adapted for the avatar-mediated virtual environment.

Each participant attended 4 weekly psychotherapy sessions over a one-month period, with each session lasting approximately 50 minutes. The sessions were scheduled on weekdays using an integrated reservation system. Participants could join sessions from any location, although guidance was provided regarding optimal environmental conditions for therapeutic engagement. When technical issues arose, therapists used the platform’s text chat as a backup communication channel, and affected sessions were rescheduled where necessary.

Following completion of the 4-session intervention, the 30 selected participants completed individual semistructured interviews lasting 30‐50 minutes. Interviews were conducted remotely via Zoom (Zoom Communications, Inc) by trained researchers experienced in qualitative health research. The interview guide explored six key domains: (1) reflection of avatar facial expressions and emotional responses; (2) sense of intimacy and communication effectiveness with avatars; (3) concentration levels during sessions; (4) significant aspects of metaverse therapy experiences; (5) perceptions of virtual-space functionality; and (6) suggestions for platform improvement.

All interviews were audio-recorded with participants’ consent and transcribed verbatim by professional transcription services. The transcripts were checked for accuracy and anonymized prior to analysis. The interviews were conducted by trained researchers who were not the treating psychotherapists and had no prior therapeutic relationship with the participants. No repeat interviews were conducted. No non-participants were present during the interviews.

We used Braun and Clarke’s thematic analysis framework [38], using ATLAS.ti software (version 23; ATLAS.ti Scientific Software Development GmbH). The analysis process included (1) familiarization with the data through repeated transcript reading; (2) systematic generation of initial codes; (3) searching for themes by collating related codes; (4) reviewing and refining themes; (5) defining and naming final themes; and (6) producing the analytical report. This 6-stage process is illustrated in Figure 3.

Four researchers (HL, JHH, CHC, and JH) with backgrounds in psychiatry, digital health, and qualitative research participated in the analysis. Two researchers had been involved in the platform development process, whereas the other 2 were not involved in platform development and contributed to analysis from a more independent analytic position. This study adopted an inductive coding approach to derive themes from the participants’ experiences. During the initial familiarization phase, all team members independently reviewed transcripts and established their own coding systems. The coding system refinement process proceeded in 3 stages. Each researcher proposed individual themes (4-5 per researcher), and coding discrepancies and differences of opinion regarding themes were resolved through iterative discussions until a consensus was reached. After integrating and coordinating the themes through team consensus, 4 main themes were finally derived. Furthermore, the progress of the research was documented throughout the entire process by recording the coding decisions, theme development, and analysis stages.

Data saturation was assessed through ongoing evaluation of new themes emerging during the analysis process. No new themes emerged after approximately 25 interviews, suggesting that saturation was achieved within the 30-interview sample. When coding discrepancies arose among the 4 researchers, they were discussed in team meetings until consensus was reached. Researcher reflexivity was maintained through regular team discussions and documentation of analytical decisions. Member checking was not used due to the anonymous nature of platform usage, but thick description and detailed quotation ensured authentic representation of participant experiences.

In AMT, avatars ensured anonymity, which encouraged active expressions and notably shaped participants’ experiences throughout the therapy sessions. This effect was particularly pronounced in the initial stages of the therapy sessions. Participants reported feeling less burdened when their actual appearances were concealed, which allowed them to share more comfortably with therapists.

A male hospital employee explained:

Anonymity allowed the participants to speak more candidly, fostering intimacy with therapists. These experiential findings align with evidence that higher avatar identifiability tends to reduce self-disclosure, whereas lower identifiability facilitates it [39]. The avatars used in this study did not resemble the participants’ physical appearances, suggesting that they could be regarded as having low identifiability, thereby encouraging greater self-disclosure.

By using anonymous avatars, the participants felt a sense of psychological safety and expressed themselves authentically during therapy sessions. Authentic self-expression was expected to contribute substantially to emotional connectedness and trust in therapeutic relationships.

The findings demonstrate that strong self-identification may emerge even without visual similarity to the avatar’s appearance. Many participants described the avatar as “feeling like myself,” attributing this perception not to the avatar’s appearance but to the synchronization of nonverbal expressions, such as facial expressions, gestures, and movements.

A male hospital employee noted:

Behavioral realism, combined with the anonymizing effect on self-disclosure, enabled more natural emotional transmission and effortless emotional expression during psychotherapy. Previous studies emphasized the importance of visual similarity in enhancing identification between users and their avatars [33-35]. In contrast, it has also been suggested that behavioral realism plays a more critical role than visual realism in fostering user-avatar identification [36].

Furthermore, highly realistic avatar appearances can paradoxically inhibit self-disclosure, whereas simplified avatars facilitate more comfortable self-expression [34,40]. In contexts such as psychotherapy, where individuals may be reluctant to disclose personal information, the anonymity afforded by avatars, combined with nonverbal expressive features that support emotional communication, has emerged as a key factor in fostering participants’ psychological safety and promoting self-disclosure.

This study analyzed the interplay between the degree of nonverbal expressiveness and anonymity as determinants of self-disclosure, suggesting that such mechanisms may enable the participants to freely express their authentic emotions and states, thereby potentially supporting therapeutic engagement and communication.

The prepsychotherapy space served as a psychological transitional area, helping participants ease into the therapeutic process. Participants mentioned that they became aware they were about to enter psychotherapy and of their presence in the “virtual space,” engaging in psychological preparation such as alleviating tension.

A female university student explained:

A male university student (KU_23, age 25 years) emphasized the preparatory value, “Yes, because it feels like you're waiting for some real psychotherapy.”

Participants perceived this space as helping alleviate tension and reduce anticipatory anxiety before sessions. Within the AMT in a metaverse-informed virtual environment, participants expressed views on both the background setting and screen configuration. Regarding the background, the majority indicated that it “did not matter,” while others emphasized that it should “not be distracting” to maintain focus during psychotherapy.

The waiting area helped participants transition smoothly, alleviated tension, and prepared them psychologically for psychotherapy. In the proof-of-concept AMT in a metaverse-informed platform, a separate lobby or waiting room was provided for the participants to prepare themselves before the sessions. These spaces served a psychological transitional role, similar to waiting rooms in the real world.

Participants expressed differing opinions about the screen configuration. While some participants preferred to view their own avatars to continuously monitor their state, others reported that seeing their avatars disrupted their immersion during psychotherapy. Designing a system that accommodates user preferences regarding the visibility of their own avatars during psychotherapy is essential.

A female hospital employee appreciated self-observation.

Conversely, a male hospital employee found self-viewing disruptive:

Fatigue from repeatedly viewing one’s own face is moderated by public self-consciousness, producing divergent user experiences [41]. This finding helps account for the differing responses of participants to avatar visibility in the psychotherapy interface and underscores the importance of personalization in the design of psychotherapy platforms.

Supporting activities, such as journaling or writing reflections after psychotherapy sessions, can help participants revisit their experiences and record their thoughts or memorable expressions. Participants expressed interest not only in keeping psychotherapy diaries but also in leaving brief daily comments or saving images of their avatar’s facial expressions.

A female hospital employee suggested the following:

Journaling is an accessible and cost-effective adjunctive therapeutic method [42,43]. It is effective in alleviating mental health symptoms, enhancing self-reflection, and promoting continued engagement in psychotherapy.

These spaces may serve as meaningful “psychological safe zones” where participants can continually reflect on their psychotherapy experiences and deepen their self-understanding. Moreover, participants indicated a strong demand for spaces that facilitate interaction with others. Beyond simply leaving comments, this reflects a need for “psychological community spaces” where participants could read others’ experiences, empathize, and exchange encouragement.

This finding suggests that empathy and solidarity are vital resources for psychological recovery. Such community spaces would allow participants to interact with and support one another while maintaining their anonymity. Participants expressed interest in activities such as clubs, lectures, and small-group classes where they could learn, communicate, and strengthen their personal competence.

Moreover, ensuring anonymity through avatars constitutes a fundamental requirement in the design of such communal spaces. Participants sought to minimize anxiety about disclosing personal information, even in open or participatory activities. Accordingly, community designs and activity spaces should incorporate structures and functions that ensure anonymity.

These findings suggest that postpsychotherapy metaverse spaces may be designed not merely as waiting areas but as optional spaces for reflection and anonymous peer connection. Such platforms should integrate self-recording and reflection, community-based empathy and solidarity, expressive social activities, and anonymity through avatars.

AMT in a metaverse-informed virtual environment should be designed to support participants’ motivation for behavioral change. Behavior change requires sufficient motivation, the ability to perform the behavior, and a triggering cue, and it is therefore important to develop content that participants can independently practice in their daily lives [44].

Participants described a need for cyclic feedback mechanisms that could help connect session experiences with everyday reflection and intended behavior.

A female hospital employee explained the following:

Prior research indicates that such cyclic interventions have achieved adherence in roughly half of participants [45]; participants in this study expressed similar motivations for continued engagement. The methods mentioned for cyclic feedback include diaries, feedback, inner healing content, and regular psychological assessments.

Specific examples include sharing psychological status data with therapists through regular psychological assessments to facilitate psychotherapy or recommending inner-healing content based on the psychological state.

A male hospital employee described desired personalization:

Participants suggested that feedback through mobile notifications, checklists, or other brief prompts could support continued engagement and between-session reflection; however, the clinical effects of such mechanisms require empirical validation.

There are differences between the roles of therapists in F2FT and metaverse-informed psychotherapy. Communication training for therapists delivering psychotherapy via videoconferencing has been emphasized as an important consideration [6]. Unlike F2FT, metaverse-informed therapy requires therapists to understand and use avatar-mediated nonverbal communication.

A female university student emphasized the need for specialized training:

Therapists should consider nonverbal cues (eg, nodding and hand gestures) that encourage participant responses [4]. In virtual psychotherapy, a therapist’s lack of response may make participants feel abandoned [46].

When the therapist’s avatar displayed no change in nonverbal expressions, some participants questioned their attentiveness, leading to psychological uncertainty. This psychological uncertainty adversely affects trust in the therapist.

A female hospital employee expressed the following:

Therefore, therapists must respond actively and provide continuous verbal and nonverbal feedback.

The analysis revealed that avatars in the metaverse do more than facilitate therapist communication; they also provide social connection and a sense of perceived co-presence within a community. Avatar customization may serve a purpose beyond mere self-expression, acting as a form of social connection.

Based on these findings, spaces can be designed to convey a metaphorical sense of community presence.

A female hospital employee described this preference:

Participants preferred indirect communication, confirming their own presence and that of others. Displaying consultation reviews may also raise concerns about maintaining anonymity. Therefore, it is necessary to consider strategies that exert a positive influence on participants while preserving their anonymity.

Unlike the metaverse spaces, the psychotherapy environment was private and closed. Discussions took place regarding the psychotherapy background, avatar size, and screen composition. The most common responses regarding the psychotherapy background were “I don’t care” or “I wish the background weren’t so noticeable,” indicating a desire to focus on the psychotherapy session.

Participants were generally satisfied with the avatar size, though some wished for slightly larger faces. In avatar psychotherapy, observing the avatar’s facial expressions plays a crucial role in effective communication, as it facilitates the recognition of another person’s nonverbal cues. Clear facial expression display is another factor to be considered when setting avatar size.

Careful observation and nonverbal responsiveness are crucial in metaverse-informed psychotherapy, as participants interpret limited cues from avatars and voices. Both therapists and participants should be trained in the appropriate and competent use of VCT [47].

They need a thorough grasp of the communication characteristics, pros and cons of this treatment method, as well as strategies to enhance the therapeutic environment. This also applies to the metaverse-informed psychotherapy environment. As KU_10 previously pointed out, therapists should possess specialized communication competencies specific to the metaverse platform, with careful consideration of the participants’ psychological states and emotions. Participants generally reported positive experiences with the platform; however, several described technical challenges—including network instability and interface-related difficulties—that occasionally disrupted the flow of communication and reduced their sense of immersion in the virtual environment. These findings underscore the need for a stable, cross-platform system design that ensures reliable access across both desktop environments (Windows or macOS) and mobile devices (iOS or Android).

A male hospital employee noted:

A female university student experienced server-related problems:

The system should be easily accessible to participants with limited digital experience. Guidance systems, such as onboarding tutorials and voice guidance, can lower the initial entry barriers for participants.

A female hospital employee described the adaptation challenge:

A female hospital employee encountered interface issues:

In a metaverse-informed psychotherapy, therapists’ platform-specific competencies, technical stability, and guidance systems should be integrated and function organically. Effectiveness and credibility can be enhanced by providing training and guidance for therapists and participants, ensuring devices and network stability, and implementing user-friendly interface design.

The 4 themes identified in this analysis demonstrate the multifaceted nature of user experiences with AMT in a metaverse-informed virtual environment, revealing both the perceived psychological affordances and practical requirements for an effective platform design. Table 2 summarizes the key findings, participants’ perspectives, and theoretical implications of the themes.

Our findings challenge conventional assumptions about avatar design in therapeutic contexts. While previous studies have emphasized the visual similarity between users and avatars as crucial for identification [33-35], this study revealed that behavioral realism may be more important than visual resemblance for therapeutic applications. This aligns with evidence that behavioral authenticity plays a more critical role than visual realism in fostering user-avatar identification [36]. The finding that participants reported anonymous avatars as facilitating self-disclosure supports the conclusion that lower avatar identifiability promotes greater self-disclosure [39]. Our study extends this understanding by showing that anonymity can coexist with authentic emotional expression through synchronized nonverbal behaviors. This differs from previous findings, suggesting that highly realistic avatars may inhibit self-disclosure [34,40]. The level and modality of avatars’ nonverbal expressions produced varying effects depending on the participants’ individual traits. Responses elicited by avatars’ nonverbal expressions differ according to individual dispositions [48]. For some participants, detailed reflection of facial expressions and gestures facilitated deeper immersion in psychotherapy and enhanced emotional communication. However, for others, excessive nonverbal expressions became burdensome and hindered self-disclosure.

Four key motivational factors in avatar design have been identified, including virtual exploration, social navigation, contextual adaptation, and identity representation [49]. They emphasized the importance of customization features that reflect users’ traits and motivations [50,51]. These results indicate that avatars’ expressive modalities and customization features can serve as critical variables that influence the effectiveness of psychotherapy depending on the participants’ dispositions and expectations. These patterns are consistent with the growing body of evidence supporting the Proteus Effect Theory [29] and suggest that, in therapeutic applications, the avatar’s capacity for authentic behavioral expression may be more influential than its visual characteristics in shaping perceived user engagement and platform usability. These results suggest that anonymous avatars equipped with nonverbal expressions may serve as a promising tool for fostering participants’ self-disclosure and satisfying the dual demands of emotional communication and anonymity in psychotherapy. Taken together, these findings are also consistent with broader evidence that computer-mediated communication can reduce social anxiety and lower inhibition compared to face-to-face interaction, thereby facilitating greater openness and participation in digital therapeutic settings [52].

Spatial design in metaverse-informed psychotherapy extends beyond accessibility, functioning as a psychological tool that may support emotional engagement and psychological comfort throughout the therapeutic process. The functions and meanings of spaces were examined by delineating 3 categories, including spaces prior to psychotherapy, spaces during psychotherapy, and supplementary interaction spaces following psychotherapy.

The design of the prepsychotherapy space should consider its role as a transitional space that helps participants achieve psychological stability and immersion before a session begins. Similar to the provision of lobbies and waiting rooms in real-world settings, metaverse-informed psychotherapy platforms require spaces that establish psychological boundaries, considering the unique characteristics of virtual environments. Furthermore, the design of the space may enhance accessibility [53], going beyond mere implementation of physical settings and functioning as psychological tools to promote emotional stability and immersion.

During psychotherapy sessions, these spaces should ensure private and focused interactions between therapists and participants. However, divergent opinions were observed regarding the design element of avatar self-view in psychotherapy spaces, a divergence also evident in previous research. While Self-Perception Theory posits that self-observation aids in inferring one’s attitudes and emotions [54], a persistent self-view may diminish concentration and contribute to fatigue [41,55]. Hence, flexible interface configurations that allow participants to selectively manage the visibility of their avatars are critical for preserving emotional concentration during sessions. The postpsychotherapy space supports emotional processing, self-reflection, and indirect social interaction with others. Participants preferred indirect, anonymous interactions. This suggests that the metaverse, unlike traditional psychotherapy rooms, can create an environment that fosters social interaction and support. Social interaction within the metaverse has been reported to contribute to improved mental health, including reduction in loneliness and risk of depression [56].

The perceived credibility and usability of AMT may depend on the integration of 2 elements—therapists’ interactional competence and the platform’s technological sophistication. Metaverse-informed psychotherapy therapists must not only possess psychotherapy skills but also undergo retraining to adapt to the unique modalities of digital environments and manage potential technical difficulties. Retraining should cover nonverbal communication skills, strategies for handling technical issues such as network instabilities, and the use of alternative channels, such as emoticons and chat functions [13,14,57,58]. To prevent disruptions in psychotherapy caused by technical problems during sessions, it is essential to ensure system-level safeguards and prompt response from therapists. Furthermore, the technological foundation of the metaverse-informed psychotherapy platform must be stable and secure enough to adequately support therapists’ efforts and participants’ engagement. Failure of key platform functions correctly, such as avatar facial expressions, real-time interaction, and data transmission, can negatively affect the overall psychotherapy experience.

However, the distinctive nature of psychotherapy requires careful consideration of privacy and security. Psychotherapy data are highly sensitive, comparable to medical information, making it challenging to protect them securely at multiple levels [59]. Moreover, because the metaverse is a composite environment that integrates hardware and software components, ensuring the security and consistency of information is essential [60]. Given that this platform processes real-time facial expression data—a form of biometric information—future implementation will require robust security architecture, including end-to-end encryption specifications, breach-response protocols, and data minimization procedures that limit the retention of facial and nonverbal data to what is strictly necessary for session delivery. Informed consent procedures should explicitly address the nature and scope of facial expression processing, distinguishing it from general audiovisual data, so that participants can make fully informed decisions about their involvement. These findings suggest that therapist competency and technological reliability should be integrated when developing AMT as a clinically cautious and technically robust adjunct or complement to existing care pathways. Future deployment should incorporate transparent informed consent, data minimization principles, end-to-end encryption, and incident response protocols regarding the collection, processing, and retention of facial expression signals.

Participants’ reports suggesting that behavioral realism may facilitate communication perceived as therapeutically meaningful, while anonymity lowers disclosure barriers, suggest avatar-based therapies should prioritize expressive authenticity over visual similarity. This implies that cost-effective systems should focus computational resources on behavioral synchronization rather than photorealistic rendering. Future research should refine these design principles and develop adaptive mechanisms that adjust expressive intensity to individual users, broadening potential applications across psychotherapeutic contexts. The finding that avatars’ expressive modalities may influence perceived therapeutic engagement suggests the need for personalized systems that adapt to participants’ characteristics and preferences.

Beyond psychotherapy sessions, metaverse-informed psychotherapy design should incorporate mechanisms that promote motivation for behavioral changes in the participants’ everyday lives. An example is providing self-guided content that can be enacted in daily routines. The participants’ accounts suggest that engagement may be strengthened when the emotional and functional design of the metaverse environment is closely linked to real-life user contexts. This indicates a design orientation toward a cyclic feedback system whereby changes in participants’ emotions and behaviors in real life emerge from metaverse-informed psychotherapy and, in turn, feed back into subsequent sessions. Users need motivation, the ability to perform the behavior, and triggers to act [44]. The cyclic feedback system design is necessary to foster participants’ meta-connection, potentially supporting between-session reflection and behavioral carryover, which require empirical validation in future studies. Prior studies show that such cyclic interventions, integrating system, therapist, and peer interactions, achieve approximately 50% adherence without attrition [45]. The integration of various feedback mechanisms, including mobile notifications, progress tracking, and Internet of Things (IoT)–linked systems, may increase the frequency of connection with metaverse-informed psychotherapy platforms, potentially supporting perceived therapeutic engagement and platform use, though further investigation is needed.

To institutionalize diverse telehealth approaches, several components are necessary, including training skilled professionals, empowering participants, restructuring funding mechanisms, improving digital infrastructure, and integrating telehealth into routine clinical practice [57]. This study revealed that metaverse environments have the potential to function as comprehensive therapeutic ecosystems rather than simple treatment delivery platforms. This finding has potential implications for psychotherapy delivery, suggesting that virtual environments can address multiple therapeutic needs simultaneously by reducing access barriers, providing preparation and transition support, enabling ongoing reflection and peer connection, and supporting motivation for real-world behavioral changes. Emphasizing community spaces while maintaining anonymity presents unique opportunities for addressing the social isolation and stigma associated with mental health treatment. Participants’ preference for anonymous community spaces suggests that metaverse environments may offer peer connection while preserving privacy, a feature that warrants further study in populations who avoid traditional psychotherapy.

For users with social anxiety or shame-prone presentations, sustained avatar concealment carries the clinically meaningful risk of functioning as a safety behavior—fostering an implicit belief that “I can only speak when I am hidden”—which, while reducing short-term distress, may inadvertently maintain longer-term avoidance and prevent the disconfirmation of feared outcomes [61,62]. This concern becomes particularly salient when AMT is considered for real-world deployment beyond a circumscribed research setting. The reduced visibility and altered self-presentation that facilitate disclosure in our findings reflect a phenomenon with both adaptive and maladaptive faces; the same conditions that lower the threshold for initial engagement can, over time, become a protective mechanism that forecloses corrective experiences [52]. We therefore regard this not as a structural flaw of AMT itself, but as a clinically meaningful consideration to be addressed through the way in which the modality is operationally embedded.

Accordingly, AMT, on its own, should not be regarded as constituting psychological care in its entirety. Rather, it is most appropriately understood as one component within a broader continuum of mental health resources—including in-person psychotherapy, psychiatric assessment, and pharmacotherapy when indicated, and the wider public mental health infrastructure—in which modalities of differing intensity are matched to differing levels of need [63]. Positioned within this kind of stepped, integrated flow, AMT may be most useful as an early-access entry point, a bridge during waiting periods, or an adjunct between in-person sessions, rather than as a standalone long-term modality. From this orientation, several modest design considerations follow: platforms could provide gentle prompts when platform use remains exclusively online over an extended period, signaling the option to consider in-person care; clinicians could be supported with structured monitoring for sustained concealment patterns that may indicate avoidance; and clear referral pathways could be embedded so that users who would benefit from a wider range of interventions can be transitioned with minimal friction. We propose these considerations cautiously and as design directions for further investigation rather than as established prescriptions.