Identifying and Transferring Reasoning-Critical Neurons: Improving LLM Inference Reliability via Activation Steering

• URL: http://arxiv.org/abs/2601.19847v1
• Date: Tue, 27 Jan 2026 17:53:01 GMT
• Title: Identifying and Transferring Reasoning-Critical Neurons: Improving LLM Inference Reliability via Activation Steering
• Authors: Fangan Dong, Zuming Yan, Xuri Ge, Zhiwei Xu, Mengqi Zhang, Xuanang Chen, Ben He, Xin Xin, Zhumin Chen, Ying Zhou,
• Abstract summary: We propose AdaRAS, a lightweight test-time framework that improves reasoning reliability by selectively intervening on neuron activations.<n>AdaRAS identifies Reasoning-Critical Neurons (RCNs) via a polarity-aware mean-difference criterion and adaptively steers their activations during inference.<n> Experiments on 10 mathematics and coding benchmarks demonstrate consistent improvements, including over 13% gains on AIME-24 and AIME-25.
• Score: 50.63386303357225
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Despite the strong reasoning capabilities of recent large language models (LLMs), achieving reliable performance on challenging tasks often requires post-training or computationally expensive sampling strategies, limiting their practical efficiency. In this work, we first show that a small subset of neurons in LLMs exhibits strong predictive correlations with reasoning correctness. Based on this observation, we propose AdaRAS (Adaptive Reasoning Activation Steering), a lightweight test-time framework that improves reasoning reliability by selectively intervening on neuron activations. AdaRAS identifies Reasoning-Critical Neurons (RCNs) via a polarity-aware mean-difference criterion and adaptively steers their activations during inference, enhancing incorrect reasoning traces while avoiding degradation on already-correct cases. Experiments on 10 mathematics and coding benchmarks demonstrate consistent improvements, including over 13% gains on AIME-24 and AIME-25. Moreover, AdaRAS exhibits strong transferability across datasets and scalability to stronger models, outperforming post-training methods without additional training or sampling cost.

Related papers

• Conditional Advantage Estimation for Reinforcement Learning in Large Reasoning Models [50.84995206660551]
  We introduce Conditional advANtage estimatiON (CANON) to amplify the impact of a target metric without presuming its direction.<n>CANON based on entropy consistently outperforms prior methods on both math reasoning and high-complexity logic tasks.
  arXiv  Detail & Related papers  (2025-09-28T16:33:07Z)
• Sycophancy Mitigation Through Reinforcement Learning with Uncertainty-Aware Adaptive Reasoning Trajectories [58.988535279557546]
  We introduce textbf sycophancy Mitigation through Adaptive Reasoning Trajectories.<n>We show that SMART significantly reduces sycophantic behavior while preserving strong performance on out-of-distribution inputs.
  arXiv  Detail & Related papers  (2025-09-20T17:09:14Z)
• Uncertain but Useful: Leveraging CNN Variability into Data Augmentation [1.137457877869062]
  We investigate the training-time variability using FastSurfer, a CNN-based whole-brain segmentation pipeline.<n>We find that FastSurfer exhibits higher variability compared to that of a traditional neuroimaging pipeline.<n>As a proof of concept, we demonstrate that numerical ensembles can be used as a data augmentation strategy for brain age regression.
  arXiv  Detail & Related papers  (2025-09-05T16:54:26Z)
• Exploring and Exploiting the Inherent Efficiency within Large Reasoning Models for Self-Guided Efficiency Enhancement [101.77467538102924]
  Large reasoning models (LRMs) exhibit overthinking, which hinders efficiency and inflates inference cost.<n>We propose two lightweight methods to enhance LRM efficiency.<n>First, we introduce Efficiency Steering, a training-free activation steering technique that modulates reasoning behavior via a single direction.<n>Second, we develop Self-Rewarded Efficiency RL, a reinforcement learning framework that dynamically balances task accuracy and brevity.
  arXiv  Detail & Related papers  (2025-06-18T17:18:12Z)
• Reasoning Models Hallucinate More: Factuality-Aware Reinforcement Learning for Large Reasoning Models [83.24079543652253]
  Large language models (LLMs) have significantly advanced in reasoning tasks through reinforcement learning (RL) optimization.<n>However, reasoning-oriented RL fine-tuning significantly increases the prevalence of hallucinations.<n>We propose Factuality-aware Step-wise Policy Optimization (FSPO), an innovative RL fine-tuning algorithm incorporating explicit factuality verification.
  arXiv  Detail & Related papers  (2025-05-30T14:23:32Z)
• SEAL: Steerable Reasoning Calibration of Large Language Models for Free [58.931194824519935]
  Large Language Models (LLMs) have demonstrated compelling capabilities for complex reasoning tasks via the extended chain-of-thought (CoT) reasoning mechanism.<n>Recent studies reveal substantial redundancy in the CoT reasoning traces, which negatively impacts model performance.<n>We introduce SEAL, a training-free approach that seamlessly calibrates the CoT process, improving accuracy while demonstrating significant efficiency gains.
  arXiv  Detail & Related papers  (2025-04-07T02:42:07Z)
• SD-LoRA: Scalable Decoupled Low-Rank Adaptation for Class Incremental Learning [73.93639228235622]
  Continual Learning with foundation models has emerged as a promising paradigm to exploit abundant knowledge acquired during pre-training for tackling sequential tasks.<n>Existing prompt-based and Low-Rank Adaptation-based (LoRA-based) methods often require expanding a prompt/LoRA pool or retaining samples of previous tasks.<n>We propose Scalable Decoupled LoRA (SD-LoRA) for class incremental learning, which continually separates the learning of the magnitude and direction of LoRA components without rehearsal.
  arXiv  Detail & Related papers  (2025-01-22T20:00:41Z)
• Self-Normalized Resets for Plasticity in Continual Learning [5.57186754179308]
  Self-Normalized Resets (SNR) is an adaptive algorithm that mitigates plasticity loss by resetting a neuron's weights when evidence suggests its firing rate has effectively dropped to zero.<n>We show that SNR consistently attains superior performance compared to its competitor algorithms.
  arXiv  Detail & Related papers  (2024-10-26T06:47:13Z)
• CARE: Certifiably Robust Learning with Reasoning via Variational Inference [26.210129662748862]
  We propose a certifiably robust learning with reasoning pipeline (CARE) CARE achieves significantly higher certified robustness compared with the state-of-the-art baselines. We additionally conducted different ablation studies to demonstrate the empirical robustness of CARE and the effectiveness of different knowledge integration.
  arXiv  Detail & Related papers  (2022-09-12T07:15:52Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.