Evaluating the Effects of Saccade Types and Directions on Eye Pointing Tasks

Title of the Paper

Evaluating the Effects of Saccade Types and Directions on Eye Pointing Tasks

Paper Information

• Research Area: Human-Computer Interaction (HCI), gaze interaction, and eye movement modeling.
• Keywords: eye pointing, saccade types and directions, 2D targets, modeling, eye-tracking technology, user interface design.

Research Background and Issues

• Problems and Challenges:

  1. Existing eye movement interaction models (e.g., blink-stay-based eye pointing models) are only applicable to circular targets and lack the ability to predict performance for traditional 2D rectangular targets.
  2. Previous studies generally neglect to explain the impact of eye movement behavior on task performance from an anatomical perspective (eye structure).
  3. Although Fitts' Law is widely used in gesture interaction modeling, its applicability to eye movement interaction remains controversial.

• Research Significance: Given the potential of gaze interaction devices in daily life, developing performance prediction models for eye pointing is of great value for interface evaluation and design optimization.

• Motivation and Related Work:

  1. Address the limitations of existing models in adapting to 2D targets, including parameters like target width and height.
  2. Investigate the specific impact of saccade directions (horizontal/vertical) on eye movement performance.
  3. Explore new patterns based on the anatomical characteristics of the eye (e.g., speed differences between centripetal and centrifugal saccades).

Solution

• Methods and Innovations:

  1. Propose a new index of difficulty model for 2D targets (ID_eye), based on a weighted squared norm approach, integrating the effects of target width and height to overcome the limitations of previous models.
  2. Systematically validate the model's performance through experiments involving three types of saccades (centripetal, centrifugal, and symmetric saccades) and multiple directions (12 angles).
  3. Investigate the impact of target shape, position, and saccade path on eye movement efficiency to provide guidance for user interface design.

• Implementation Steps and Techniques:

  1. Construct a 2D Euclidean Model: [ ID_{eye} = \lambda \cdot \frac{A}{\sqrt{\omega/(W-\mu)^2 + (1-\omega)/(H-\mu)^2}} ] where ( \mu ) represents the jitter range caused by eye instability.
  2. Design two experiments:
     • Experiment 1: Test the effects of different saccade directions on eye movement time (EMT) and pointing time (EPT) and validate the accuracy of the new model.
     • Experiment 2: Examine the generalizability and performance differences of the model under different saccade types (centripetal vs. centrifugal).

Research Findings

• Key Experimental Results:

  1. Experiments demonstrated that the improved 2D ID_eye model outperforms the original Fitts' model, offering better predictive capabilities for gaze interaction.
  2. Horizontal saccades are more efficient than vertical ones, and target height (H) has a greater impact on completion time than target width (W).
  3. In symmetric saccade tasks, the parameter ( \omega ) exhibits periodic variation, revealing the relationship between target direction and eye movement performance.
  4. Centripetal saccades outperform centrifugal saccades in both pointing time (EPT) and movement time (EMT).

• Advantages:

  • The new model integrates multiple factors, such as direction, target size, and distance, using the 2D geometric properties of targets, effectively capturing the efficiency patterns of eye movements.
  • Introduced an alternative factor ( \mu ) based on eye instability (jitter), which is strongly correlated with experimental data and demonstrates practical utility.

• Limitations and Future Directions:

  1. The current model does not delve into complex 3D saccade scenarios beyond the initial position.
  2. Further development is needed to efficiently handle dynamic targets (non-static UI elements).
  3. Future studies could explore the adaptability of model parameters to different experimental devices (e.g., other eye-tracker models).

Conclusion and Design Recommendations

1. User Interface Design:
   • For vertical layouts, increasing target height (H) can significantly enhance pointing efficiency.
   • Dynamic target designs with asymmetric extensions (e.g., prioritizing vertical extensions) are recommended to optimize screen space.
2. Best Practices for Experimental Design:
   • Use monocular data collection to avoid experimental bias caused by differences in control between the left and right eyes.
   • Avoid designs where multiple target heights and distances change simultaneously to reduce inconsistencies.

This study proposes a set of symbolic design principles and modeling improvements for gaze interaction, providing valuable academic references for the widespread adoption of eye-tracking technology.