Quantifying density cues in grouping displays.

Area of Science:

• Visual Perception
• Cognitive Psychology
• Computational Neuroscience

Background:

• Perceptual grouping research often uses sparse displays.
• Ensuring the absence of proximity cues is crucial for non-proximity grouping studies.
• Existing methods for controlling density lack empirical validation.

Purpose of the Study:

• To validate local density metrics for stimulus selection in perceptual grouping research.
• To identify the most reliable metric for ensuring the absence of proximity cues.
• To compare computational models with human performance data.

Main Methods:

• Testing local density metrics as constrained ideal observer models.
• Comparing metric performance against a large dataset of human visual detection trials.
• Evaluating metrics based on their ability to control for proximity cues.

Main Results:

• The Voronoi density metric, particularly with nearest neighbor distance, is recommended.
• Human observers showed reduced sensitivity to sparse target groupings and regularity cues.
• Observers were more sensitive to local clusterings of target elements.

Conclusions:

• The Voronoi density metric is the preferred method for selecting stimuli without density cues.
• The study provides a benchmark dataset for future metric development.
• Human visual grouping by proximity is influenced by element density and clustering.