[論文レビュー] The Eye-Head Mover Spectrum: Modelling Individual and Population Head Movement Tendencies in Virtual Reality

Jinghui Hu, Ludwig Sidenmark+2 more|arXiv (Cornell University)|Feb 5, 2026

Gaze Tracking and Assistive Technology被引用数 0

ひとこと要約

この論文は、VRにおける視線シフトに対する個々の頭部寄与をモデル化する連続的な eye-head mover spectrum を導入し、集団レベルのばらつきを示し、ソフトヒンジモデルを複数タスクで検証する。

ABSTRACT

People differ in how much they move their head versus their eyes when shifting gaze, yet such tendencies remain largely unexplored in HCI. We introduce head movement tendencies as a fundamental dimension of individual difference in VR and provide a quantitative account of their population-level distribution. Using a 360° video free-viewing dataset (N=87), we model head contributions to gaze shifts with a hinge-based parametric function, revealing a spectrum of strategies from eye-movers to head-movers. We then conduct a user study (N=28) combining 360° video viewing with a short controlled task using gaze targets. While parameter values differ across tasks, individuals show partial alignment in their relative positions within the population, indicating that tendencies are meaningful but shaped by context. Our findings establish head movement tendencies as an important concept for VR and highlight implications for adaptive systems such as foveated rendering, viewport alignment, and multi-user experience design.

研究の動機と目的

• Characterize how individuals differ in head contribution to gaze shifts in VR.
• Provide a continuous, interpretable per-user profile of head movement tendencies.
• Quantify population-level distribution of head movement strategies using a large open dataset.
• Assess cross-task stability of head contribution tendencies between free viewing and controlled tasks.

提案手法

• Analyze a large 360° free-viewing VR dataset (N=87; later filtered to N=80 monoscopic) with head and gaze tracking.
• Define horizontal head contribution to gaze shifts as a function of target eccentricity within [0, 50] degrees.
• Fit three parametric models (linear baseline, hinge, soft hinge) to individual data and compare using R^2, RMSE, and AIC.
• Select the soft hinge model as the preferred per-participant profile based on fit quality and interpretability.
• Apply functional PCA to obtain a population distribution of eye-head strategies from participant profiles.

Figure 1. Example fits of three model formulations to one participant (P05). Each plot shows horizontal head contribution ( $\Delta$ Head) against target eccentricity, with grey dots as data points. (a) Linear + EOR baseline assumes a fixed eye-only range followed by a linear increase. (b) Hinge mod

実験結果

リサーチクエスチョン

• RQ1 RQ1: How does head contribution to gaze shifts vary between individuals?
• RQ2 RQ2: How prevalent are different head-movement tendencies in the population?
• RQ3 RQ3: Are individual head-contribution tendencies consistent across tasks?

主な発見

• Head contribution to gaze shifts increases with target eccentricity, forming a spectrum from eye-dominant to head-dominant strategies.
• The soft hinge model explains more variance and achieves lower RMSE than the hinge and linear models, making it the preferred per-participant fit.
• The population exhibits a continuous distribution of head-movement tendencies rather than discrete head movers/non-head movers.
• Tendencies show partial alignment across tasks, suggesting context shapes but does not fully override intrinsic coordination.
• Across participants, head movement patterns are strongly symmetric left–right, supporting data mirroring for analysis.
• The approach provides interpretable per-user profiles (beta, tau, s) that can inform VR system adaptations like foveated rendering and viewport alignment.

Figure 2. Comparison of model fits for gaze shift data. Boxplots show the distribution of R² (left) and RMSE (right) across participants for three models: Linear baseline, two-parameter hinge, and three-parameter soft hinge. Boxes represent the interquartile range (IQR) with medians; black dots indi