Investigating Visual Crowding of Objects in Complex Real-World Scenes.

Area of Science:

• Vision Science
• Cognitive Psychology
• Perceptual Science

Background:

• Visual crowding, the difficulty in recognizing objects in peripheral vision due to nearby items, is typically studied with simple stimuli.
• Its occurrence in natural scenes, where context can aid recognition, remains unproven.
• Prior research suggests crowding is distance-dependent, occurring when flankers are close to the target relative to eccentricity.

Purpose of the Study:

• To rigorously investigate visual crowding in computer-generated, semantically consistent real-world scenes.
• To determine if crowding effects persist in complex naturalistic environments.
• To examine the impact of varying target-flanker spacing on object recognition accuracy within scenes.

Main Methods:

• Two experiments were conducted using computer-generated real-world scenes with target objects surrounded by flankers.
• Target-flanker spacing was manipulated around a critical threshold (0.5 eccentricity) to modulate crowding.
• Experiment 1 used cued target locations and brief scene presentations; Experiment 2 employed eye-tracking for central fixation and sustained presentation.

Main Results:

• Object recognition accuracy consistently decreased as the spacing between target and flanker objects diminished.
• This effect was observed despite the presence of complex scene context.
• The findings provide robust evidence for visual crowding in naturalistic settings.

Conclusions:

• Visual crowding significantly impairs object recognition in peripheral vision, even within complex, semantically consistent real-world scenes.
• The study confirms that crowding is not limited to simplified laboratory stimuli.
• Spatial proximity of objects remains a critical factor influencing object identification in naturalistic visual environments.