@inproceedings{chen-etal-2026-canonical,
title = "Canonical Color as a Lens into Concept Decodability in Vision Encoders and {VLM}s",
author = "Chen, Xiaofu and
Frank, Stella and
Kementchedjhieva, Yova",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1291/",
pages = "29413--29424",
ISBN = "979-8-89176-459-0",
abstract = "Vision encoders construct the image representations used by vision-language models. How much conceptual information, rather than immediately visible image evidence, do these representations contain? We use canonical color as a controlled test case: can a vision encoder represent the typical color of an object concept, such as yellow for bananas, when visible color is removed from the input image? We construct a dataset of objects with canonical color labels and probe vision encoders for both color and object identity using RGB and grayscale inputs. We find that canonical color remains decodable from grayscale images and is tied to predicted object identity, suggesting a conceptual link between object and color information. Extending this analysis to full VLMs, we find that VLM post-training can have a surprisingly large effect on color decodability in the vision encoder. Overall, canonical color provides a controllable lens for tracing object-level semantic information in vision encoders and VLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2026-canonical">
<titleInfo>
<title>Canonical Color as a Lens into Concept Decodability in Vision Encoders and VLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xiaofu</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Stella</namePart>
<namePart type="family">Frank</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yova</namePart>
<namePart type="family">Kementchedjhieva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Vision encoders construct the image representations used by vision-language models. How much conceptual information, rather than immediately visible image evidence, do these representations contain? We use canonical color as a controlled test case: can a vision encoder represent the typical color of an object concept, such as yellow for bananas, when visible color is removed from the input image? We construct a dataset of objects with canonical color labels and probe vision encoders for both color and object identity using RGB and grayscale inputs. We find that canonical color remains decodable from grayscale images and is tied to predicted object identity, suggesting a conceptual link between object and color information. Extending this analysis to full VLMs, we find that VLM post-training can have a surprisingly large effect on color decodability in the vision encoder. Overall, canonical color provides a controllable lens for tracing object-level semantic information in vision encoders and VLMs.</abstract>
<identifier type="citekey">chen-etal-2026-canonical</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1291/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>29413</start>
<end>29424</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Canonical Color as a Lens into Concept Decodability in Vision Encoders and VLMs
%A Chen, Xiaofu
%A Frank, Stella
%A Kementchedjhieva, Yova
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F chen-etal-2026-canonical
%X Vision encoders construct the image representations used by vision-language models. How much conceptual information, rather than immediately visible image evidence, do these representations contain? We use canonical color as a controlled test case: can a vision encoder represent the typical color of an object concept, such as yellow for bananas, when visible color is removed from the input image? We construct a dataset of objects with canonical color labels and probe vision encoders for both color and object identity using RGB and grayscale inputs. We find that canonical color remains decodable from grayscale images and is tied to predicted object identity, suggesting a conceptual link between object and color information. Extending this analysis to full VLMs, we find that VLM post-training can have a surprisingly large effect on color decodability in the vision encoder. Overall, canonical color provides a controllable lens for tracing object-level semantic information in vision encoders and VLMs.
%U https://aclanthology.org/2026.emnlp-main.1291/
%P 29413-29424
Markdown (Informal)
[Canonical Color as a Lens into Concept Decodability in Vision Encoders and VLMs](https://aclanthology.org/2026.emnlp-main.1291/) (Chen et al., EMNLP 2026)
ACL