Project overview
Our recent VR dyad study showed that at a 1.22 m viewing distance, observers decode where an avatar is attending almost entirely from gross body orientation, while the avatar’s tracked eye direction adds no decodable information. That distance sits at the boundary of the personal and social proxemic zones, and display geometry suggests the balance between eye and body cues should shift with distance: up close, the rendered iris spans many more pixels and eye direction could in principle become readable; far away, even the head may lose value and the torso should carry everything. This project maps the cue hierarchy across interpersonal distance.
Project motivation
Social VR platforms place avatars anywhere from intimate (under 0.45 m) to public (over 3.6 m) distances, and empirical studies show dyads spontaneously adopt 1.3 to 1.8 m [1, 2]. If the value of eye rendering depends on distance, avatar systems could budget rendering fidelity dynamically: high-fidelity eyes only when someone is close enough for them to matter. Psychophysics predicts such a crossover, since head and body orientation remain readable far into the visual periphery while eye direction is resolvable only near fixation [3].
Project goals
- Distance-sweep experiment. Replicate the shared-attention decoding task (Sender attends a target, Receiver estimates it from the avatar) at four to five interpersonal distances between roughly 0.5 m and 4 m, within participants.
- Cue-hierarchy modeling. For each distance, decompose the Receiver’s estimation error into eye, head, and body contributions using the existing GLMM variance-decomposition pipeline, and identify the distance (if any) at which eye direction gains predictive value.
You will
- Perform a literature review on proxemics in social VR and gaze perception at a distance
- Modify the existing Unity VR environment (distance conditions, logging)
- Run a dyad study reusing the existing OptiTrack + eye-tracking pipeline
- Analyze the data with the existing Python/R variance-decomposition tooling
- Summarize your findings in a thesis and present them to an audience
- (Optional) co-write a research paper
You need
- Strong communication skills in English
- Good knowledge of Unity
- Basic statistics (mixed models are a plus)
References
- Miller, M. R., DeVeaux, C., Han, E., Ram, N., & Bailenson, J. N. (2023). A Large-Scale Study of Proxemics and Gaze in Groups. IEEE VR 2023. https://doi.org/10.1109/VR55154.2023.00056
- Rivu, R., Zhou, Y., Welsch, R., Mäkelä, V., & Alt, F. (2021). When Friends Become Strangers: Understanding the Influence of Avatar Gender on Interpersonal Distance in Virtual Reality. INTERACT 2021. https://doi.org/10.1007/978-3-030-85607-6_16
- Loomis, J. M., Kelly, J. W., Pusch, M., Bailenson, J. N., & Beall, A. C. (2008). Psychophysics of perceiving eye-gaze and head direction with peripheral vision. Perception, 37(9). https://doi.org/10.1068/p5896
- The parent study on avatar-mediated shared attention in virtual reality; details are available from the advisor on request.
Keywords: VR, Proxemics, Gaze, Body Orientation, OptiTrack