Evaluating Layout Dimensionalities in PC+VR Asymmetric Collaborative Decision Making

TL;DR

This study investigates how layout dimensionality affects asymmetric collaboration between PC and VR in data-driven decision-making. Using a within-subject design, researchers compared three conditions—PC2D+VR2D, PC2D+VR3D, and PC3D+VR3D—applied to a hotel-search task with multi-scale navigation and real-time awareness cues. Key findings show that optimizing for individual effectiveness (PC2D+VR3D) increases user satisfaction and perceived productivity, while consistent dimensionality across devices reduces task time; however, 3D navigation on PC imposes higher mental demand. The work provides practical guidance for cross-device visualization design, balancing personal comfort with collaboration costs, and highlights directions for future work on richer 3D content and adaptive interface switching.

Abstract

With the commercialization of virtual/augmented reality (VR/AR) devices, there is an increasing interest in combining immersive and non-immersive devices (e.g., desktop computers) for asymmetric collaborations. While such asymmetric settings have been examined in social platforms, significant questions around layout dimensionality in data-driven decision-making remain underexplored. A crucial inquiry arises: although presenting a consistent 3D virtual world on both immersive and non-immersive platforms has been a common practice in social applications, does the same guideline apply to lay out data? Or should data placement be optimized locally according to each device's display capacity? This study aims to provide empirical insights into the user experience of asymmetric collaboration in data-driven decision-making. We tested practical dimensionality combinations between PC and VR, resulting in three conditions: PC2D+VR2D, PC2D+VR3D, and PC3D+VR3D. The results revealed a preference for PC2D+VR3D, and PC2D+VR2D led to the quickest task completion. Our investigation facilitates an in-depth discussion of the trade-offs associated with different layout dimensionalities in asymmetric collaborations.

Figures (6)

• Figure 1: Demonstration of multiple-level of the hotel information and the amount of information presented on PC2D at each level.
• Figure 2: Illustration of bidirectional multiscale zooming in VR. Inspired by proxemic interaction ballendat2010proxemicbadam2016supporting, the presented level of detail is determined by the distance between the user and the views. Three levels of detail were provided in our study: far distance results in an overview with hotel images; medium distance adds hotel names, prices, and ratings; close distance further provides breakdown ratings and amenities. The same levels of detail were provided on PC, and users used the mouse scroll to switch between them.
• Figure 3: Demonstration of real-time awareness cues across PC and VR. Leveraging the depth-adaptive cursor technique zhou2022depthtong2023towards, we are able to provide real-time awareness cues across platforms and dimensions (i.e., in PC2D+VR3D). On PC (left), a moving indicator shows which window the VR collaborator is looking at. At the same time, in VR (right), an icon is rendered to indicate the PC collaborator's cursor position.
• Figure 4: Three tested conditions in the user study. (a) PC2D+VR2D, where the layout of views is 2D in both PC and VR; (b) PC2D+VR3D, which involves a 2D layout for the PC collaborator and a curved 3D layout for the VR collaborator; and (c) PC3D+VR3D, where both PC and VR have a 3D layout. The PC collaborator uses pan&zoom to navigate in 2D environments (i.e., PC2D+VR2D and PC2D+VR3D), while employing a combination of WASD keys and the mouse to navigate in 3D environments (i.e., PC3D+VR3D). This navigation method is similar to playing a first-person shooter (FPS) game and is commonly provided by commercial PC+VR social platforms. The VR collaborator walks in the space for both 2D and 3D layouts.
• Figure 5: The four different views from the perspective of a user.