Contour Integration in Dynamic Scenes: Impaired Detection Performance in Extended Presentations.

Area of Science:

• Visual perception
• Cognitive neuroscience
• Computational vision

Background:

• Visual scenes are dynamic, requiring continuous integration of local information into global representations.
• Contour integration (CI) links oriented line segments into contours, aiding object boundary detection.
• Previous CI research predominantly used static stimuli, leaving dynamic CI poorly understood.

Purpose of the Study:

• To investigate contour detection performance in dynamic visual scenes.
• To determine if the robustness of CI observed in static stimuli extends to dynamic environments.
• To explore the role of stimulus duration and top-down control in dynamic CI.

Main Methods:

• Dynamic visual stimuli composed of Gabor elements forming contours were presented.
• Stimulus presentation times varied, including brief (230 ms) and extended (1.9-3.8 s) durations.
• Cueing strategies (location, shape, combined) were employed to assess their impact on performance.

Main Results:

• Detection performance was high (87%) in brief dynamic displays but significantly decreased (67%) in extended presentations.
• Reduced performance in extended displays mirrored effects seen with much shorter (50 ms) brief presentations.
• Cueing, particularly combined location and shape cues in extended presentations, partially restored performance.

Conclusions:

• Dynamic contour integration is less robust than previously assumed and not solely stimulus-driven.
• Top-down processes play a significant role in supporting visual integration in dynamic scenes.
• The findings challenge the notion of effortless, stimulus-driven contour perception in naturalistic, dynamic environments.