mlsanityData to 7 Oct 2026

Context Is King: How In-Context Specification Shapes the Geometry of Concepts

arXiv 2607.24425 ↗First posted 27 Jul 2026cs.LGcited by 0

On the map: World models → World models for language agents

Abstract (arXiv)

Large language models place structured concepts on geometrically faithful manifolds: weekdays lie on a circle, months on another, usually taken to be a fixed world-model the network stores and looks up. We show that context is king: the structure a model actually uses is set by the in-context specification. A declarative rule fixes not only which relations the geometry encodes but its topology type: the same tokens form a cycle or a branching tree on command, built even on arbitrary, meaning-free tokens with no prior to inherit, which a relabeled stored shape cannot do. When the specification conflicts with a strong pretrained prior, the context-set geometry dominates it in capable models, read from the same activations (representational similarity 0.6--0.9 to the imposed structure versus near-zero to the prior), across the priors we test and both families we study (Gemma, Qwen). Activation patching shows the map is causally used, not a probe correlate: swapping one entity's activation for another's makes the model answer with the other entity's successor under the imposed order. A rough map forms readily, present even in small and base models; what scale gates is using it cleanly: clean dominance and the causal crossover emerge only in the larger models (up to Gemma-31B and Qwen-27B) and weaken or reverse below, so a mechanism present in a large model can be absent in a smaller one of the same family. Whether the model builds this geometry anew or reconfigures a stored one we leave open; operationally, the geometry it uses is the one the context specifies.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in World models for language agents, one of 11 directions in World models. Builds on 18 of its 23 references that are in the map. First posted on arXiv 27 Jul 2026.

Builds on

18 of 23 references.

Closest ancestors · 2

Same direction, newest first

  1. 2026Language Models Struggle to Use Representations Learned In-Context4 Feb
  2. 2023Emergent Linear Representations in World Models of Self-Supervised Sequence Models2 Sep

Other references · 16

  1. 2022Locating and Editing Factual Associations in GPTcited by 2,609
  2. 2022Emergent Abilities of Large Language Modelscited by 2,483
  3. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 960
  4. 2019How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddingscited by 770
  5. 2023Are Emergent Abilities of Large Language Models a Mirage?cited by 515
  6. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 417
  7. 2023Language Models Represent Space and Timecited by 306
  8. 2024Not All Language Model Features Are One-Dimensionally Linearcited by 201
  9. 2024The Geometry of Categorical and Hierarchical Concepts in Large Language Modelscited by 151
  10. 2023Does Circuit Analysis Interpretability Scale? Evidence from Multiple Choice Capabilities in Chinchillacited by 145
  11. 2023Improving Activation Steering in Language Models with Mean-Centringcited by 64
  12. 2026Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behaviorcited by 22
4 more
  1. 2023The Topology and Geometry of Neural Representationscited by 13
  2. 2026Context Structure Reshapes the Representational Geometry of Language Modelscited by 6
  3. 2025Task Schema and Binding: A Double Dissociation Study of In-Context Learningcited by 2
  4. 2026Large language models reorganize representational geometry during in-context learningcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026The Geometric Price of Discrete Logic: Context-driven Manifold Dynamics of Number Representations
  2. 2026The Geometry of Knowledge Accessibility in Large Language Models
  3. 2025The Geometry of Reasoning: Flowing Logics in Representation Space
  4. 2026Context Cartography: Toward Structured Governance of Contextual Space in Large Language Model Systems
  5. 2025Hypothesis-Driven Feature Manifold Analysis in LLMs via Supervised Multi-Dimensional Scaling
  6. 2026The Lattice Representation Hypothesis of Large Language Models
  7. 2026Beneath the Surface of Chains-of-Thought: A Mechanistic Interpretation of Reasoning Operations in LLMs
  8. 2026Encoded but Not Actionable: Auditing the Decode-Generate-Steer Gap in Frozen LLMs for Geometric Constraints