mlsanityData to 7 Oct 2026

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

arXiv 2109.06129 ↗First posted 13 Sep 2021cs.CV cs.CLcited by 118

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. Cited by 118 papers in the map (data through 7 Oct 2026). Builds on 22 of its 61 references that are in the map. First posted on arXiv 13 Sep 2021.

Builds on

22 of 61 references.

References · 22

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
  3. 2019Language Models as Knowledge Bases?cited by 1,922
  4. 2016Enriching Word Vectors with Subword Informationcited by 1,578
  5. 2019VisualBERT: A Simple and Performant Baseline for Vision and Languagecited by 1,291
  6. 2019BERT Rediscovers the Classical NLP Pipelinecited by 1,046
  7. 2019VideoBERT: A Joint Model for Video and Language Representation Learningcited by 702
  8. 2019Designing and Interpreting Probes with Control Taskscited by 608
  9. 2020Universal Dependencies v2: An Evergrowing Multilingual Treebank Collectioncited by 269
  10. 2020Experience Grounds Languagecited by 241
  11. 2019Visualizing and Measuring the Geometry of BERTcited by 215
  12. 2019Commonsense Knowledge Mining from Pretrained Modelscited by 189
10 more
  1. 2014SimLex-999: Evaluating Semantic Models with (Genuine) Similarity Estimationcited by 170
  2. 2020Probing Pretrained Language Models for Lexical Semanticscited by 156
  3. 2020Analyzing Individual Neurons in Pre-trained Language Modelscited by 97
  4. 2019Do Neural Language Representations Learn Physical Commonsense?cited by 62
  5. 2020Pareto Probing: Trading Off Accuracy for Complexitycited by 55
  6. 2021Provable Limitations of Acquiring Meaning from Ungrounded Form: What Will Future Language Models Understand?cited by 55
  7. 2021The Low-Dimensional Linear Geometry of Contextualized Word Representationscited by 39
  8. 2019Neural Naturalist: Generating Fine-Grained Image Comparisonscited by 36
  9. 2017Are distributional representations ready for the real world? Evaluating word vectors for grounded perceptual meaningcited by 19
  10. 2020Probing Contextual Language Models for Common Ground with Visual Representationscited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 1; 2022: 10; 2023: 29; 2024: 30; 2025: 35; 2026: 13.

118 papers.

Sort:

By area:

  1. 2022Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Themcited by 1,767
  2. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 673
  3. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 417
  4. 2024The Platonic Representation Hypothesiscited by 362
  5. 2023Language Models Represent Space and Timecited by 306
  6. 2023Reasoning or Reciting? Exploring the Capabilities and Limitations of Language Models Through Counterfactual Taskscited by 243
  7. 2022Pre-Trained Language Models for Interactive Decision-Makingcited by 220
  8. 2023Grounding Large Language Models in Interactive Environments with Online Reinforcement Learningcited by 193
  9. 2023Inspecting and Editing Knowledge Representations in Language Modelscited by 140
  10. 2024Is A Picture Worth A Thousand Words? Delving Into Spatial Reasoning for Vision Language Modelscited by 111
  11. 2024Evaluating the World Model Implicit in a Generative Modelcited by 107
  12. 2024Towards Principled Evaluations of Sparse Autoencoders for Interpretability and Controlcited by 83
38 more
  1. 2024Robust agents learn causal world modelscited by 81
  2. 2022Meaning without reference in large language modelscited by 76
  3. 2023Eight Things to Know about Large Language Modelscited by 73
  4. 2023Could a Large Language Model be Conscious?cited by 70
  5. 2023Evaluating Spatial Understanding of Large Language Modelscited by 65
  6. 2024Mind's Eye of LLMs: Visualization-of-Thought Elicits Spatial Reasoning in Large Language Modelscited by 65
  7. 2022Do Large Language Models know what humans know?cited by 60
  8. 2024One fish, two fish, but not the whole sea: Alignment reduces language models' conceptual diversitycited by 57
  9. 2024Elements of World Knowledge (EWoK): A Cognition-Inspired Framework for Evaluating Basic World Knowledge in Language Modelscited by 55
  10. 2023Can Large Language Models be Good Path Planners? A Benchmark and Investigation on Spatial-temporal Reasoningcited by 49
  11. 2024ICLR: In-Context Learning of Representationscited by 49
  12. 2023Large language models predict human sensory judgments across six modalitiescited by 47
  13. 2023From task structures to world models: What do LLMs know?cited by 46
  14. 2025What Has a Foundation Model Found? Using Inductive Bias to Probe for World Modelscited by 46
  15. 2022Event knowledge in large language models: the gap between the impossible and the unlikelycited by 44
  16. 2024LLM-A*: Large Language Model Enhanced Incremental Heuristic Search on Path Planningcited by 44
  17. 2024Monitoring Latent World States in Language Models with Propositional Probescited by 39
  18. 2024A Vision Check-up for Language Modelscited by 36
  19. 2024Standards for Belief Representations in LLMscited by 30
  20. 2023Emergent Representations of Program Semantics in Language Models Trained on Programscited by 28
  21. 2023Dissociating language and thought in large language modelscited by 27
  22. 2023Do Vision and Language Models Share Concepts? A Vector Space Alignment Studycited by 27
  23. 2023Are Large Language Models Temporally Grounded?cited by 25
  24. 2025iVISPAR -- An Interactive Visual-Spatial Reasoning Benchmark for VLMscited by 24
  25. 2024KiVA: Kid-inspired Visual Analogies for Testing Large Multimodal Modelscited by 22
  26. 2023Large Language Models: The Need for Nuance in Current Debates and a Pragmatic Perspective on Understandingcited by 20
  27. 2023Can Language Models Understand Physical Concepts?cited by 19
  28. 2021Do Language Models Know the Way to Rome?cited by 18
  29. 2023MEWL: Few-shot multimodal word learning with referential uncertaintycited by 17
  30. 2022COMPS: Conceptual Minimal Pair Sentences for testing Robust Property Knowledge and its Inheritance in Pre-trained Language Modelscited by 15
  31. 2024What Do Language Models Hear? Probing for Auditory Representations in Language Modelscited by 15
  32. 2024Vision-Language Models Create Cross-Modal Task Representationscited by 14
  33. 2025The Geometry of Prompting: Unveiling Distinct Mechanisms of Task Adaptation in Language Modelscited by 14
  34. 2025General agents contain world modelscited by 14
  35. 2023What makes a language easy to deep-learn? Deep neural networks and humans similarly benefit from compositional structurecited by 13
  36. 2024Does ChatGPT Have a Mind?cited by 13
  37. 2023The System Model and the User Model: Exploring AI Dashboard Designcited by 12
  38. 2024Do Music Generation Models Encode Music Theory?cited by 12

50 most cited of 118 All 118 (JSON)