The Veiled Virgin illustrates visual segmentation of shape by cause

Area of Science:

• Visual Perception
• Cognitive Neuroscience
• Computational Vision

Background:

• Three-dimensional (3D) shape perception is vital for tasks like object recognition and tool use.
• Current theories often emphasize geometric computations, neglecting causal inference in shape representation.
• Understanding how the brain represents complex shapes, especially those with layered surfaces, remains a challenge.

Purpose of the Study:

• To investigate whether visual shape perception involves inferring causal origins of surface features.
• To determine if the brain can segment shapes based on whether features arise from an object or its covering.
• To explore the role of causal inference in understanding composite shapes, inspired by draped sculptures.

Main Methods:

• Created composite 3D shapes by wrapping unfamiliar objects in textile.
• Used 3D scans to obtain ground-truth measures of underlying object and fabric surface reliefs.
• Employed a virtual painting task where participants identified the causal origin of surface ridges.
• Conducted experiments measuring perceived depth profiles of both object and textile layers.

Main Results:

• Participants could robustly distinguish surface features caused by the underlying object from those caused by the overlying textile.
• Judgments of surface relief and depth profiles reflected an ability to parse shape based on inferred causal origins.
• The visual system successfully segmented features related to the hidden object versus the drapery.

Conclusions:

• Visual shape perception relies on sophisticated causal inferences, not solely geometric calculations.
• The brain employs shape-segmentation processes that parse visual input based on distinct causal origins.
• This research reveals a deeper understanding of how the brain constructs 3D shape representations from complex visual scenes.