mlsanityData to 14 Oct 2026

State over Tokens: Characterizing the Role of Reasoning Tokens

arXiv 2512.12777 ↗First posted 14 Dec 2025cs.CL cs.AIcited by 5

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

Large Language Models (LLMs) can generate reasoning tokens before their final answer to boost performance on complex tasks. While these sequences seem like human thought processes, empirical evidence reveals that they are not a faithful explanation of the model's actual reasoning process. To address this gap between appearance and function, we introduce the State over Tokens (SoT) conceptual framework. SoT reframes reasoning tokens not as a linguistic narrative, but as an externalized computational state -- the sole persistent information carrier across the model's stateless generation cycles. This explains how the tokens can drive correct reasoning without being a faithful explanation when read as text and surfaces previously overlooked research questions on these tokens. We argue that to truly understand the process that LLMs do, research must move beyond reading the reasoning tokens as text and focus on decoding them as state.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 38 of its 56 references that are in the map. First posted on arXiv 14 Dec 2025.

Builds on

38 of 56 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025Latent Reasoning via Sentence Embedding Prediction28 May
  2. 2025s1: Simple test-time scaling31 Jan

Other references · 36

  1. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  2. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
  3. 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,122
  4. 2024Tulu 3: Pushing Frontiers in Open Language Model Post-Trainingcited by 1,053
  5. 2021Show Your Work: Scratchpads for Intermediate Computation with Language Modelscited by 886
  6. 2024Training Large Language Models to Reason in a Continuous Latent Spacecited by 797
  7. 2023Measuring Faithfulness in Chain-of-Thought Reasoningcited by 703
  8. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 422
  9. 2025Reasoning Models Don't Always Say What They Thinkcited by 415
  10. 2024Chain of Thought Empowers Transformers to Solve Inherently Serial Problemscited by 260
  11. 2024Let's Think Dot by Dot: Hidden Computation in Transformer Language Modelscited by 191
  12. 2025Chain-of-Thought Reasoning In The Wild Is Not Always Faithfulcited by 170
24 more
  1. 2025Thought Anchors: Which LLM Reasoning Steps Matter?cited by 129
  2. 2024Making Reasoning Matter: Measuring and Improving Faithfulness of Chain-of-Thought Reasoningcited by 110
  3. 2024Faithfulness vs. Plausibility: On the (Un)Reliability of Explanations from Large Language Modelscited by 105
  4. 2025Subliminal Learning: Language models transmit behavioral traits via hidden signals in datacited by 95
  5. 2025When Chain of Thought is Necessary, Language Models Struggle to Evade Monitorscited by 88
  6. 2025Are DeepSeek R1 And Other Reasoning Models More Faithful?cited by 72
  7. 2025Artificial Intelligence Index Report 2025cited by 64
  8. 2024How to think step-by-step: A mechanistic understanding of chain-of-thought reasoningcited by 53
  9. 2025Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokenscited by 41
  10. 2025Position: Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces!cited by 37
  11. 2025Soft Tokens, Hard Truthscited by 35
  12. 2025Large language models can learn and generalize steganographic chain-of-thought under process supervisioncited by 32
  13. 2025Base Models Know How to Reason, Thinking Models Learn Whencited by 27
  14. 2024Dissociation of Faithful and Unfaithful Reasoning in LLMscited by 23
  15. 2025Verifying Chain-of-Thought Reasoning via Its Computational Graphcited by 20
  16. 2025Early Signs of Steganographic Capabilities in Frontier LLMscited by 19
  17. 2025How does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencodingcited by 19
  18. 2025LLMs Can Covertly Sandbag on Capability Evaluations Against Chain-of-Thought Monitoringcited by 18
  19. 2024Large Language Models Cannot Explain Themselvescited by 11
  20. 2025Interpretable Traces, Unexpected Outcomes: Investigating the Disconnect in Trace-Based Knowledge Distillationcited by 8
  21. 2025Tensor Logic: The Language of AIcited by 8
  22. 2025Humans Perceive Wrong Narratives from AI Reasoning Textscited by 6
  23. 2025Unspoken Hints: Accuracy Without Acknowledgement in LLM Reasoningcited by 4
  24. 2025Typed Chain-of-Thought: A Curry-Howard Framework for Verifying LLM Reasoningcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026Quantifying the Necessity of Chain of Thought through Opaque Serial Depthcited by 8
  2. 2026Reasoning Primitives in Hybrid and Non-Hybrid LLMs: Do Architectural Differences Yield Advantages in State-Tracking and Recall?cited by 1
  3. 2026Trace as State: Reasoning Traces as Conditional States for Long-Context Transformerscited by 1
  4. 2026Forecasting Future Behavior as a Learning Taskcited by 0
  5. 2026Legibility is Not Interpretability: Comparing Judged and Actual Importance in Chain-Of-Thought Reasoningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2023Large Language Models are In-Context Semantic Reasoners rather than Symbolic Reasoners
  2. 2026State of Thought Enables Endogenous Reasoning
  3. 2026Towards a Mechanistic Understanding of Propositional Logical Reasoning in Large Language Models
  4. 2026LLM Reasoning Is Latent, Not the Chain of Thought
  5. 2025Do Latent Tokens Think? A Causal and Adversarial Analysis of Chain-of-Continuous-Thought
  6. 2026Shorthand for Thought: Compressing LLM Reasoning via Entropy-Guided Supertokens
  7. 2026Reasoning as State Transition: A Representational Analysis of Reasoning Evolution in Large Language Models
  8. 2025The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity