Study: MR window previews speed layout, but users still want manual control
24 VR users tested 3 window-placement visualizations; fast previews lost to manual control.
Adaptive mixed reality (MR) systems often auto-optimize window layouts, ignoring individual preferences. A new arXiv study (2608.00403) from Abdelrahman Zaky and Tiare Feuchtner explores a middle ground: showing users placement proposals they can accept or reject. The researchers tested three visualization techniques—Situated Icon Preview, Situated Window Preview, and 3D Preview—against a Manual Positioning baseline in a within-subjects VR experiment with 24 participants.
Each participant completed a multi-stage trip-planning task, individually placing seven sequentially introduced windows. Results showed that Situated Icon Preview and Situated Window Preview significantly reduced layouting time compared to manual positioning, and Situated Window Preview was also faster than 3D Preview. However, when asked, users overwhelmingly preferred direct manual control. Interviews revealed why: perceived control, cognitive cost, familiarity, and how informative the proposal felt. The authors suggest hybrid approaches—proposal-based suggestions combined with manual refinement—as a promising path for future adaptive MR interfaces.
- Situated Window Preview was faster than 3D Preview and reduced layouting time vs. manual positioning
- 24 participants placed 7 windows in VR across 4 techniques (3 proposals + baseline)
- Despite efficiency gains, users preferred manual control due to perceived control and cognitive cost
Why It Matters
Balancing AI-generated layout suggestions with user control will shape the next generation of adaptive mixed-reality productivity tools.