How do humans group non-rigid objects in multiple object tracking?: Evidence from grouping by self-rotation.

Area of Science:

• Cognitive Psychology
• Visual Perception
• Motion Perception

Background:

• Previous research on perceptual grouping focused on spatiotemporal and featural cues for rigid objects.
• The role of self-motion, particularly self-rotation, in grouping spatially separated objects remains under-explored.
• Naturalistic object motion involves simultaneous translation and rotation, complicating rigidity and potentially affecting perceptual grouping.

Purpose of the Study:

• To investigate the specific role of self-rotation information in the perceptual grouping of spatially separated, non-rigid objects.
• To determine if self-rotation is used strategically or automatically for grouping.
• To differentiate grouping based on self-rotation from other cues like surface-level motion or shape changes.

Main Methods:

• A modified multiple object tracking (MOT) paradigm was employed using self-rotating objects.
• Experiments systematically varied object properties and motion cues to isolate the effect of self-rotation.
• Behavioral data were analyzed to assess grouping performance and strategic use of self-rotation information.

Main Results:

• Participants successfully used self-rotation information to group non-rigid objects, despite it being detrimental to attentive tracking.
• The grouping advantage was attributed to self-rotation itself, not surface-level cues like 2D motion or shape.
• Grouping by self-rotation was comparable to grouping by changing shape but additive, suggesting distinct information sources.

Conclusions:

• Self-motion, specifically self-rotation, serves as a novel feature for perceptual grouping of dynamic objects.
• Object grouping in dynamic scenes can leverage distinct information sources, including motion flow fields and deformation.
• This study advances our understanding of how the visual system forms object units during motion perception.