Recent Publications

Abstract: In social virtual reality (VR), harassment persists as a pervasive and critical issue. Prior work emphasizes its perceived realness and emotional impact. However, the influence of avatar design, particularly the role of self-similarity, remains underexplored. Self-similar avatars can enhance user identification and engagement, yet potentially intensify the psychological and physiological effects of harassment. Existing studies often rely on interviews or user-generated content, lacking systematic analysis and controlled comparisons. To address these gaps, we present a process for creating realistic VR harassment scenarios. We built a scenario based on existing literature and validated it with expert reviews and user feedback. We propose a 2 x 2 between-subjects design to systematically examine users' emotional and physiological states, their identification with avatars, and the effects of avatar self-similarity. The study design will deepen the understanding of harassment dynamics in VR. Additionally, it can provide actionable insights for designing safer, more inclusive virtual environments that promote user well-being and foster equitable communities.

Abstract: In this work, we explored asymmetric collaboration under two distinct tasks: collaborative sorting and conversational talking tasks. We answer the research question of how different tasks impact the user experience in asymmetric interaction. Our mixed design compared one symmetric and one asymmetric interaction and two tasks, assessing self-perception (presence, embodiment), other-perception (co-presence, social presence, plausibility), and task perception (task load, enjoyment). 52 participants collaborated in dyads on the two tasks, either using head-mounted displays (HMDs) or one participant using an HMD and the other a desktop setup. Results indicate that differences in social presence diminished or disappeared during the purely conversational talking task in comparison to the sorting task. This indicates that differences in how we perceive a social interaction, which is caused by asymmetric interaction, only occur during specific use cases. These findings underscore the critical role of task characteristics in shaping users’ social XR experiences and highlight that asymmetric collaboration can be effective across different use cases and is even on par with symmetric interaction during conversations.

Abstract: This paper investigates if and how self-similarity and having agency impact sense of embodiment, self-identification, and body weight estimation in Augmented Reality (AR). We conducted a 2x2 mixed design experiment involving 60 participants who interacted with either synchronously moving virtual humans or independently moving ones, each with self-similar or generic appearances, across two consecutive AR sessions. Participants evaluated their sense of embodiment, self-identification, and body weight perception of the virtual human. Our results show that self-similarity significantly enhanced sense of embodiment, self-identification, and the accuracy of body weight estimates with the virtual human. However, the effects of having agency over virtual human movements were notably weaker in these measures than in similar VR studies. Further analysis indicated that not only the virtual human itself but also the participants' body weight, self-esteem, and body shape concerns predict body weight estimates across all conditions. Our work advances the understanding of virtual human body weight perception in AR systems, emphasizing the importance of factors such as coherence with the real-world environment.

Abstract: Interpupillary distance (IPD) is a crucial characteristic of head-mounted displays (HMDs) because it defines an important property for generating a stereoscopic parallax, which is essential for correct depth perception. This is why contemporary HMDs offer adjustable lenses to adapt to users' individual IPDs. However, today's Video See-Through Augmented Reality (VST AR) HMDs use fixed camera placements to reconstruct the stereoscopic view of a user's environment. This leads to a potential mismatch between individual IPD settings and the fixed Inter-Camera Distances (ICD), which in turn can lead to perceptual incongruencies, limiting the usability and potentially the applicability of VST AR in depth-sensitive use cases. To investigate this incongruency between IPD and ICD, we conducted a 2x3 mixed-factor design empirical evaluation using a near-field, open-loop reaching task comparing distance judgments of Virtual Reality (VR) and VST AR. We also explored improvements in reaching performance via perceptual calibration by incorporating a feedback phase between pre- and post-phase conditions, with a particular focus on the influence of IPD-ICD differences. Our Linear Mixed Model (LMM) analysis showed a significant difference between VR and VST AR, a significant effect of IPD-ICD mismatch, as well as a combined effect of both factors. This novel insight and its consequences are discussed specifically for depth perception tasks in AR, eXtended Reality (XR), and potential use cases.

Abstract: With recent developments in spatial computing, work contexts might shift to augmented reality. Embodied AI - virtual conversa- tional agents backed by AI systems - have the potential to enhance these contexts and open up more communication channels than just text. To support knowledge transfer from virtual agent research to the general populace, we developed My CoWorker ChatGPT - an interactive demo where employees can try out various embodied AIs in a virtual office or their own using augmented reality. We use state-of-the-art speech synthesis and body-scanning technology to create believable and trustworthy AI assistants. The demo was shown at multiple events throughout Germany, where it was well received and sparked fruitful conversations about the possibilities of embodied AI in work contexts.

Abstract: Virtual Reality Exposure Therapy (VRET) has become an effective, customizable, and affordable treatment for various psychological and physiological disorders. Specifically, it is used to treat specific anxiety disorders, such as acrophobia or arachnophobia, for decades. However, to ensure a positive outcome for patients, we must understand and control the effects potentially caused by the technology and medium of Virtual Reality (VR) itself. This article specifically investigates the impact of the Plausibility illusion (Psi), as one of the two theorized presence components, on the fear of heights. In two experiments, 30 participants each experienced two different heights with congruent and incongruent object behaviors in a 2 x 2 within-subject design. Results show that the strength of the congruence manipulation plays a significant role. Only when incongruencies are strong enough will they be recognized by users, specifically in high fear conditions, as triggered by exposure to increased heights. If incongruencies are too subtle, they seem to be overshadowed by the stronger fear reactions. Our evidence contributes to recent theories of VR effects and emphasizes the importance of understanding and controlling factors potentially assumed to be incidental, specifically during VRET designs. Incongruencies should be controlled so that they do not have an unwanted influence on the patient's fear response.

Abstract: Facial expressions are crucial for many eXtended Reality (XR) use cases, from mirrored self exposures to social XR, where users interact via their avatars as digital alter egos. However, current XR devices differ in sensor coverage of the face region. Hence, a faithful reconstruction of facial expressions either has to exclude these areas or synthesize missing animation data with model-based approaches, potentially leading to perceivable mismatches between executed and perceived expression. This paper investigates potential effects of the coverage of facial animations (none, partial, or whole) on important factors of self-perception. We exposed 83 participants to their mirrored personalized avatar. They were shown their mirrored avatar face with upper and lower face animation, upper face animation only, lower face animation only, or no face animation. Whole animations were rated higher in virtual embodiment and slightly lower in uncanniness. Missing animations did not differ from partial ones in terms of virtual embodiment. Contrasts showed significantly lower humanness, lower eeriness, and lower attractiveness for the partial conditions. For questions related to self-identification, effects were mixed. We discuss participants' shift in body part attention across conditions. Qualitative results show participants perceived their virtual representation as fascinating yet uncanny.

Abstract: Media bias is a multifaceted problem, leading to one-sided views and impacting decision-making. A way to address digital media bias is to detect and indicate it automatically through machine-learning methods. However, such detection is limited due to the difficulty of obtaining reliable training data. Human-in-the-loop-based feedback mechanisms have proven an effective way to facilitate the data-gathering process. Therefore, we introduce and test feedback mechanisms for the media bias domain, which we then implement on NewsUnfold, a news-reading web application to collect reader feedback on machine-generated bias highlights within online news articles. Our approach augments dataset quality by significantly increasing inter-annotator agreement by 26.31% and improving classifier performance by 2.49%. As the first human-in-the-loop application for media bias, the feedback mechanism shows that a user-centric approach to media bias data collection can return reliable data while being scalable and evaluated as easy to use. NewsUnfold demonstrates that feedback mechanisms are a promising strategy to reduce data collection expenses and continuously update datasets to changes in context.