mlsanityData to 7 Oct 2026

Implicit Reasoning Steering via Concept Chaining

arXiv 2607.14242 ↗First posted 15 Jul 2026cs.CLcited by 0

On the map: LLM reasoning → Prompting and chain-of-thought reasoning

Abstract (arXiv)

Large language models often appear to reason reliably, yet on many questions repeated sampling yields both correct and incorrect answers, revealing an underlying fragility in how final decisions are formed. We study whether this fragility can be exploited through implicit reasoning steering: using natural-language text to bias a model toward a designated answer without explicit instructions, triggers, or direct answer cues. Our approach, Concept Chaining, generates a short connection paragraph that links question entities to a target option through one or two intermediate concepts. We then continue pretraining a victim model on these connection paragraphs and evaluate whether its answer preference shifts on the original multiple-choice questions. Our results show that indirect, natural-looking text can systematically steer model predictions while remaining substantially less inferable than direct paraphrases, which shows that reasoning brittleness is not merely an evaluation artifact: it creates a practical channel through which latent biases can be amplified by ordinary-looking text to covertly redirect model decisions.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning. Builds on 60 of its 81 references that are in the map. First posted on arXiv 15 Jul 2026.

Builds on

60 of 81 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024Can Language Models Perform Robust Reasoning in Chain-of-thought Prompting with Noisy Rationales?31 Oct
  2. 2023Measuring Faithfulness in Chain-of-Thought Reasoning17 Jul

Other references · 58

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  2. 2025Qwen3 Technical Reportcited by 8,195
  3. 2024Qwen2.5 Technical Reportcited by 4,713
  4. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,710
  5. 2022Locating and Editing Factual Associations in GPTcited by 2,609
  6. 2017Understanding Black-box Predictions via Influence Functionscited by 1,714
  7. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,258
  8. 2022Mass-Editing Memory in a Transformercited by 976
  9. 2024GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Modelscited by 533
  10. 2021Fast Model Editing at Scalecited by 491
  11. 2022Memory-Based Model Editing at Scalecited by 443
  12. 2023Scalable Extraction of Training Data from (Production) Language Modelscited by 415
38 more
  1. 2023Detecting Pretraining Data from Large Language Modelscited by 375
  2. 2023Making Retrieval-Augmented Language Models Robust to Irrelevant Contextcited by 364
  3. 2023MQuAKE: Assessing Knowledge Editing in Language Models via Multi-Hop Questionscited by 292
  4. 2024AlphaEdit: Null-Space Constrained Knowledge Editing for Language Modelscited by 224
  5. 2023Knowledge Editing for Large Language Models: A Surveycited by 189
  6. 2024Do Membership Inference Attacks Work on Large Language Models?cited by 188
  7. 2024A Comprehensive Study of Knowledge Editing for Large Language Modelscited by 157
  8. 2024GSM-Plus: A Comprehensive Benchmark for Evaluating the Robustness of LLMs as Mathematical Problem Solverscited by 155
  9. 2023Backdooring Instruction-Tuned Large Language Models with Virtual Prompt Injectioncited by 143
  10. 2024Min-K%++: Improved Baseline for Detecting Pre-Training Data from Large Language Modelscited by 102
  11. 2024WISE: Rethinking the Knowledge Memory for Lifelong Model Editing of Large Language Modelscited by 90
  12. 2025AnyEdit: Edit Any Knowledge Encoded in Language Modelscited by 74
  13. 2024A Peek into Token Bias: Large Language Models Are Not Yet Genuine Reasonerscited by 73
  14. 2024Copyright Traps for Large Language Modelscited by 36
  15. 2024AKEW: Assessing Knowledge Editing in the Wildcited by 35
  16. 2025Benchmarking Reasoning Robustness in Large Language Modelscited by 30
  17. 2024Undesirable Memorization in Large Language Models: A Surveycited by 29
  18. 2024Calibrating Reasoning in Language Models with Internal Consistencycited by 27
  19. 2023Deceptive Semantic Shortcuts on Reasoning Chains: How Far Can Models Go without Hallucination?cited by 26
  20. 2025How Is LLM Reasoning Distracted by Irrelevant Context? An Analysis Using a Controlled Benchmarkcited by 26
  21. 2024Scaling Up Membership Inference: When and How Attacks Succeed on Large Language Modelscited by 21
  22. 2025A Survey on Privacy Risks and Protection in Large Language Modelscited by 20
  23. 2025Exploring the limits of strong membership inference attacks on large language modelscited by 20
  24. 2026Large Language Model Reasoning Failurescited by 20
  25. 2024Special Characters Attack: Toward Scalable Training Data Extraction From Large Language Modelscited by 16
  26. 2024Has this Fact been Edited? Detecting Knowledge Edits in Language Modelscited by 14
  27. 2024RUPBench: Benchmarking Reasoning Under Perturbations for Robustness Evaluation in Large Language Modelscited by 13
  28. 2024How to Make LLMs Forget: On Reversing In-Context Knowledge Editscited by 9
  29. 2024FamiCom: Further Demystifying Prompts for Language Models with Task-Agnostic Performance Estimationcited by 8
  30. 2024Stepwise Reasoning Error Disruption Attack of LLMscited by 8
  31. 2024Identifying Knowledge Editing Types in Large Language Modelscited by 7
  32. 2025Tracing and Reversing Edits in LLMscited by 7
  33. 2025Multidimensional Consistency Improves Reasoning in Language Modelscited by 6
  34. 2025Is Fine-Tuning an Effective Solution? Reassessing Knowledge Editing for Unstructured Datacited by 6
  35. 2024A Statistical and Multi-Perspective Revisiting of the Membership Inference Attack in Large Language Modelscited by 5
  36. 2025Chain-of-Defensive-Thought: Structured Reasoning Elicits Robustness in Large Language Models against Reference Corruptioncited by 5
  37. 2025CC-LEARN: Cohort-based Consistency Learningcited by 5
  38. 2025Automated Detection of Pre-training Text in Black-box LLMscited by 5

50 most cited of 58 All 58 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Answer-Centric or Reasoning-Driven? Uncovering the Latent Memory Anchor in LLMs
  2. 2024Making Reasoning Matter: Measuring and Improving Faithfulness of Chain-of-Thought Reasoning
  3. 2025More Bias, Less Bias: BiasPrompting for Enhanced Multiple-Choice Question Answering
  4. 2025Implicit Reasoning in Large Language Models: A Comprehensive Survey
  5. 2026On the Generalization of Steering Vectors for Chain-of-Thought Faithfulness
  6. 2025Mitigating Content Effects on Reasoning in Language Models through Fine-Grained Activation Steering
  7. 2025Understanding Reasoning in Thinking Language Models via Steering Vectors
  8. 2024Conceptual and Unbiased Reasoning in Language Models