# When Reasoning Goes Astray: Attention Dynamics of Uncontrolled Reasoning

arXiv: https://arxiv.org/abs/2609.38817 · HTML: https://mlsanity.com/paper/2609.38817/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning.
- Builds on 32 of its 40 references that are in the map.
- First posted on arXiv 30 Sep 2026.

## Placed in

- LLM reasoning / [Test-time compute and efficient thinking](https://mlsanity.com/lr/test-time-compute-and-efficient-thinking/)

## Builds on (most cited first, 30 of 32)

- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://mlsanity.com/paper/2305.10601/) (2023)
- [GPQA: A Graduate-Level Google-Proof Q&A Benchmark](https://mlsanity.com/paper/2311.12022/) (2023)
- [s1: Simple test-time scaling](https://mlsanity.com/paper/2501.19393/) (2025)
- [STaR: Bootstrapping Reasoning With Reasoning](https://mlsanity.com/paper/2203.14465/) (2022)
- [A Deep Reinforced Model for Abstractive Summarization](https://mlsanity.com/paper/1705.04304/) (2017)
- [O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning](https://mlsanity.com/paper/2501.12570/) (2025)
- [TokenSkip: Controllable Chain-of-Thought Compression in LLMs](https://mlsanity.com/paper/2502.12067/) (2025)
- [Token-Budget-Aware LLM Reasoning](https://mlsanity.com/paper/2412.18547/) (2024)
- [Dynamic Early Exit in Reasoning Models](https://mlsanity.com/paper/2504.15895/) (2025)
- [ThinkPrune: Pruning Long Chain-of-Thought of LLMs via Reinforcement Learning](https://mlsanity.com/paper/2504.01296/) (2025)
- [Sponge Examples: Energy-Latency Attacks on Neural Networks](https://mlsanity.com/paper/2006.03463/) (2020)
- [OverThink: Slowdown Attacks on Reasoning LLMs](https://mlsanity.com/paper/2502.02542/) (2025)
- [S-GRPO: Early Exit via Reinforcement Learning in Reasoning Models](https://mlsanity.com/paper/2505.07686/) (2025)
- [Missing Premise exacerbates Overthinking: Are Reasoning Models losing Critical Thinking Skill?](https://mlsanity.com/paper/2504.06514/) (2025)
- [Inverse Scaling in Test-Time Compute](https://mlsanity.com/paper/2507.14417/) (2025)
- [Denial-of-Service Poisoning Attacks against Large Language Models](https://mlsanity.com/paper/2410.10760/) (2024)
- [An Engorgio Prompt Makes Large Language Model Babble on](https://mlsanity.com/paper/2412.19394/) (2024)
- [LoopLLM: Transferable Energy-Latency Attacks in LLMs via Repetitive Generation](https://mlsanity.com/paper/2511.07876/) (2025)
- [Fine-Grained Activation Steering: Steering Less, Achieving More](https://mlsanity.com/paper/2602.04428/) (2026)
- [Early Stopping for Large Reasoning Models via Confidence Dynamics](https://mlsanity.com/paper/2604.04930/) (2026)
- [RECUR: Resource Exhaustion Attack via Recursive-Entropy Guided Counterfactual Utilization and Reflection](https://mlsanity.com/paper/2602.08214/) (2026)
- [Mitigating the Learning Bias towards Repetition by Self-Contrastive Training for Open-Ended Generation](https://mlsanity.com/paper/2307.01542/) (2023)
- [CODE: A Contradiction-Based Deliberation Extension Framework for Overthinking Attacks on Retrieval-Augmented Generation](https://mlsanity.com/paper/2601.13112/) (2026)

## Abstract (arXiv)

Large reasoning models (LRMs) improve performance on complex tasks through extended reasoning, yet the same process can degenerate into redundant verification and persistent generation loops. Such uncontrolled reasoning increases inference cost and creates risks of resource exhaustion and service degradation. However, existing mitigations largely truncate long outputs or react to surface repetition, and thus fail to distinguish normal thinking from uncontrolled reasoning or explain how benign reasoning degenerates into harmful behavior. In this paper, we operationalize LRM generation as four states and further introduce Reasoning-state Analysis via Dynamic Attention Responses (RADAR), which identifies the current reasoning state in real time and characterizes how effective reflection can develop into uncontrolled generation. Guided by RADAR's analysis, we further realign abnormal attention distributions toward patterns observed in normal requests and examine how this correction affects excessive reflection and persistent looping. Temporal analyses show that uncontrolled reasoning is characterized by attention distributions that deviate from normal generation, with abnormal trends becoming detectable before repetition begins. Correcting these deviations through Attention Realignment consistently reduces looping while largely preserving benign performance. Together, RADAR provide a mechanistic account of how reasoning becomes uncontrolled, offering actionable guidance for identifying critical failure stages and designing targeted runtime interventions.