Less is More! Visual Suppression for Bottom-up and Top-down Attention in Dynamic Environments

Paper Title

Less is More! Visual Suppression for Bottom-up and Top-down Attention in Dynamic Environments

Publication Info

• Topic area: Visual attention mechanisms in dynamic virtual environments
• Keywords: Visual suppression, salience-relevance, attention redistribution, dimming, blurring, virtual environments, cognitive load, visual search, sustained monitoring, immersive systems

Background and Problem

• Problem / challenge: Dynamic virtual environments create visual competition between salient and relevant objects, impairing user attention. Existing methods to enhance salience introduce visual clutter and fail to address salience-relevance interactions comprehensively.
• Significance: Effective attention management in virtual environments is critical for applications like training, navigation, and monitoring, where distractions can compromise performance and safety.
• Motivation and related work: Prior research has focused on enhancing relevant object salience or suppressing irrelevant distractions but has not systematically explored salience-relevance interactions or the effects of suppression intensity. This paper addresses these gaps by introducing suppression-based visual filters.

Solution

• Proposed approach: Suppression-based visual filtering mechanisms, implemented as Dim (brightness modulation) and Blur (foveated depth-of-field blur), at Strong and Weak intensity levels.
• Novelty:
  1. Development of validated suppression-based visual filters across different intensities in dynamic virtual environments.
  2. Empirical evidence from user studies showing the effects of filters on attention performance.
  3. Introduction of the Attention Redistribution Principle, explaining how suppressed irrelevant salience reallocates attentional resources.
  4. Systematic exploration of salience-relevance interactions and their impact on attention.
• Procedure and key techniques:
  • External object attribute modulation (Dim): Reduces brightness of irrelevant objects to create feature contrast.
  • Internal vision simulation (Blur): Mimics human visual acuity fall-off using Gaussian blur centered on the target.
  • Controlled virtual environment with abstract moving objects to manipulate salience-relevance configurations.
  • Evaluation through visual search and sustained monitoring tasks using metrics like response time, accuracy, and cognitive load.

Results

• Concrete findings:
  • Dim-Strong achieved the fastest response time (6.59s) and highest accuracy (99.2%) in visual search tasks.
  • Blur conditions improved accuracy but slowed response times during sustained monitoring.
  • Strong intensity outperformed Weak intensity across both Dim and Blur filters.
  • Cognitive load reduced by up to 41.4% (Dim-Strong), while motion sickness effects were minimal.
• Advantage over baselines:
  • All visual filters improved attention performance compared to Baseline, with Dim-Strong showing the most significant gains.
  • Dim filters demonstrated faster response times and higher accuracy than Blur filters.
• Experiments / evaluation:
  • Within-subjects study with 38 participants, testing five conditions (Dim-Strong, Dim-Weak, Blur-Strong, Blur-Weak, Baseline) across nine salience-relevance configurations.
  • Metrics included response time, accuracy, distance perception, eye gaze angle, discriminability (d′), and response criterion (C).
• Limitations and future work:
  • Limited ecological validity due to abstract virtual environment design.
  • Salience defined solely by spatial distance, ignoring multi-feature interactions.
  • Need for systematic validation of intensity levels and salience differences.
  • Open questions on intermediate intensities, dynamic relevance changes, and unexpected relevant information.

Summary

This study introduced suppression-based visual filters (Dim and Blur) to manage attention in dynamic virtual environments. Dim-Strong demonstrated superior performance in visual search and monitoring tasks, validating the Attention Redistribution Principle, which explains how suppressing irrelevant salience reallocates attentional resources to relevant information. Findings revealed salience-relevance challenges and offered design insights for immersive systems, such as driving displays and monitoring tools. While the abstract experimental setup limits direct real-world application, the results provide a foundation for future research and practical implementations in complex environments.