Perceptual Asymmetry Between Hue Categories: Evidence from Human Color Categorization

Researchers extend the COLIBRI fuzzy color model to reveal that human color categories exhibit significant perceptual asymmetry, with yellow forming a narrow, sharply-defined region while green spans a broader interval. This finding challenges computational models that assume uniformly distributed color representations and suggests color naming follows non-uniform geometric organization in perceptual space.

This research addresses a fundamental gap between how humans naturally categorize colors and how computational systems model them. Traditional color models assume evenly distributed, symmetric categories across perceptual space, but empirical evidence from large-scale human color naming datasets reveals a strikingly different picture. Yellow functions as a highly constrained, specific perceptual label with clear boundaries, while green operates as a broad, tolerant region that accommodates greater variation in hue perception.

The findings stem from cognitive science and linguistics research demonstrating that color categories are fuzzy—lacking crisp boundaries—rather than discrete. The COLIBRI framework already incorporated fuzziness through membership functions, but this extension quantifies the asymmetry using metrics like Wideness and Boundary Width at the 0.5 membership level. This reveals that human color cognition doesn't distribute naming categories uniformly; instead, some colors receive highly specific lexical treatment while others serve as catch-all categories.

For machine learning and computer vision applications, this asymmetry matters significantly. Systems trained on categorical data without accounting for perceptual non-uniformity may perform inconsistently across different color regions. Color-based classification tasks, image segmentation, and human-computer interaction systems relying on color semantics could benefit from models that reflect actual human perceptual organization rather than assuming symmetric distributions.

Future work should examine whether this asymmetry extends across different languages and cultures, as color categorization varies significantly in linguistic systems worldwide. Additionally, understanding the cognitive basis for asymmetry—whether driven by biological factors, environmental statistics, or linguistic conventions—could improve both AI color perception and theories of human categorization more broadly.

• →Human color categories exhibit significant non-uniform distribution, contradicting assumptions in most computational color models
• →Yellow forms a narrow, sharply-bounded perceptual category while green spans a broader, more tolerant region
• →Fuzzy membership functions reveal that some colors function as specific labels while others serve as broad catch-all categories
• →Perceptually grounded color models must account for asymmetry to improve machine learning performance on color-based tasks
• →Cross-linguistic and cross-cultural studies needed to determine whether perceptual asymmetry is universal or culturally dependent