Control of attention and gaze in complex environments

Area of Science:

• Human-Computer Interaction
• Cognitive Psychology
• Virtual Reality

Background:

• Task-driven attention is crucial for natural behavior.
• Handling unexpected stimuli is a challenge for autonomous systems.
• Gaze patterns are typically linked to ongoing tasks.

Purpose of the Study:

• Investigate the impact of unexpected pedestrian collision risks on gaze distribution.
• Understand attention allocation when facing potential hazards in virtual environments.
• Examine how task load influences responses to unexpected events.

Main Methods:

• Subjects walked in a virtual reality environment.
• Unexpected pedestrian collision courses were introduced.
• Gaze distribution and fixation patterns were recorded.
• Task load was manipulated to assess its effect on attention.

Main Results:

• Potential collisions were ineffective in eliciting fixations, particularly under high task load.
• Prior fixations on pedestrians enhanced collision detection performance.
• Collision detection induced short-term alterations in subsequent visual search strategies.
• Results support a top-down, learned search strategy for detecting unexpected events.

Conclusions:

• Attention to unexpected events like collisions is not automatic but relies on learned, top-down mechanisms.
• Effective detection of environmental hazards requires experience-based scheduling of visual search.
• Virtual reality provides a controlled environment to study human attention and hazard perception.