Can Language Models Encode Perceptual Structure Without Grounding? A Case Study in Color

TL;DR

This study investigates whether pretrained language models implicitly encode perceptual color structure by analyzing the alignment between text-derived color representations and perceptual color spaces, revealing notable correlations especially for warmer colors.

Contribution

The paper provides a detailed case study demonstrating that language models encode perceptual color structure, with insights into how color usage and collocationality influence this encoding.

Findings

Significant correspondence between language model representations and perceptual color space.

Warmer colors are better aligned with perceptual structure than cooler colors.

Color usage and syntactic context influence the encoding of perceptual color information.

Abstract

Pretrained language models have been shown to encode relational information, such as the relations between entities or concepts in knowledge-bases -- (Paris, Capital, France). However, simple relations of this type can often be recovered heuristically and the extent to which models implicitly reflect topological structure that is grounded in world, such as perceptual structure, is unknown. To explore this question, we conduct a thorough case study on color. Namely, we employ a dataset of monolexemic color terms and color chips represented in CIELAB, a color space with a perceptually meaningful distance metric. Using two methods of evaluating the structural alignment of colors in this space with text-derived color term representations, we find significant correspondence. Analyzing the differences in alignment across the color spectrum, we find that warmer colors are, on average, better aligned to the perceptual color space than cooler ones, suggesting an intriguing connection to findings from recent work on efficient communication in color naming. Further analysis suggests that differences in alignment are, in part, mediated by collocationality and differences in syntactic usage, posing questions as to the relationship between color perception and usage and context.