Virtual reality meetings increasingly represent users through avatars. Self-views, such as virtual mirrors or miniature avatars, allow users to observe how their virtual body appears and behaves during interaction. Prior work shows that such self-views can support users in monitoring their avatar and can influence how they experience communication in multi-user VR.
However, self-views do not have to remain passive representations of what a user is already doing. They could instead become private interfaces that actively guide behavior during interaction. Imagine, for example, a user who wants to gesture more while presenting. Their self-view could display virtual bubbles that only they can see and that they are encouraged to touch. Reaching toward these targets would induce arm movements that appear to other meeting participants as natural gestures, while the underlying guidance remains private. Similar mechanisms could potentially support posture, gaze behavior, expressiveness, or other communicative behaviors.
In this thesis, you will investigate how behavior-shaping self-views can be designed and how they influence users’ behavior and perception in VR. The goal is to move beyond conventional feedback interfaces that explicitly tell users what to do and instead explore how elements embedded directly into a self-view can scaffold desired behavior. This builds on work such as GestureCoach, which demonstrates that targeted recommendations can encourage speakers to gesture more effectively during presentation rehearsal. It also connects to research on the Proteus effect, which shows that characteristics of virtual self-representations can influence users’ subsequent behavior.
The thesis will involve reviewing related work on self-views, avatar embodiment, behavioral feedback, and virtual coaching; designing and implementing one or more self-view mechanisms in VR; and evaluating them in a user study. Depending on the scope of the thesis, the study could focus specifically on gesture behavior or compare several forms of behavioral guidance.
Potential research questions include how self-views can provide implicit visual guidance for specific behaviors; whether such guidance changes users’ observable gestures, posture, or communication behavior; how implicit self-view guidance compares with explicit instructions; and how these mechanisms affect perceived naturalness, agency, embodiment, cognitive load, and awareness of the intervention. An additional question is when such guidance is experienced as supportive and when it becomes distracting or overly controlling.
The thesis is suitable for students interested in Human-Computer Interaction, Virtual Reality, avatar-mediated communication, and behavioral research. Experience with Unity/C# or empirical user studies is helpful, but can also be developed during the thesis.
Main References
Krug, K., Fink, D. I., Ellenberg, M. O., Reinschluessel, A. V., Büschel, W., Feuchtner, T., & Dachselt, R. (2025). Mirror Me: Exploring Avatar Self-Views in Multi-User Virtual Reality. Proceedings of the ACM on Human-Computer Interaction, 9(8).
The paper investigates different avatar self-view representations in multi-user VR and provides the most direct foundation for understanding self-views as a design space for communication.
Ram, A., Suresh, V., Abadian, A. S., Demberg, V., & Steimle, J. (2025). GestureCoach: Rehearsing for Engaging Talks with LLM-Driven Gesture Recommendations. UIST 2025.
GestureCoach demonstrates how targeted, proactive guidance can encourage speakers to use more appropriate and diverse gestures, providing a useful reference for the behavioral-guidance side of the thesis.
Yee, N., & Bailenson, J. N. (2007). The Proteus Effect: The Effect of Transformed Self-Representation on Behavior. Human Communication Research, 33(3), 271–290.
This foundational paper shows that properties of virtual self-representations can influence users’ behavior, providing the theoretical link between how users see themselves and how they subsequently act.