Two Experts Are All You Need for Steering Thinking: Reinforcing Cognitive Effort in MoE Reasoning Models Without Additional Training

• URL: http://arxiv.org/abs/2505.14681v2
• Date: Tue, 27 May 2025 09:35:12 GMT
• Title: Two Experts Are All You Need for Steering Thinking: Reinforcing Cognitive Effort in MoE Reasoning Models Without Additional Training
• Authors: Mengru Wang, Xingyu Chen, Yue Wang, Zhiwei He, Jiahao Xu, Tian Liang, Qiuzhi Liu, Yunzhi Yao, Wenxuan Wang, Ruotian Ma, Haitao Mi, Ningyu Zhang, Zhaopeng Tu, Xiaolong Li, Dong Yu,
• Abstract summary: We introduce a novel inference-time steering methodology called Reinforcing Cognitive Experts (RICE)<n>RICE aims to improve reasoning performance without additional training or complexs.<n> Empirical evaluations with leading MoE-based LRMs demonstrate noticeable and consistent improvements in reasoning accuracy, cognitive efficiency, and cross-domain generalization.
• Score: 86.70255651945602
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Mixture-of-Experts (MoE) architectures within Large Reasoning Models (LRMs) have achieved impressive reasoning capabilities by selectively activating experts to facilitate structured cognitive processes. Despite notable advances, existing reasoning models often suffer from cognitive inefficiencies like overthinking and underthinking. To address these limitations, we introduce a novel inference-time steering methodology called Reinforcing Cognitive Experts (RICE), designed to improve reasoning performance without additional training or complex heuristics. Leveraging normalized Pointwise Mutual Information (nPMI), we systematically identify specialized experts, termed ''cognitive experts'' that orchestrate meta-level reasoning operations characterized by tokens like ''<think>''. Empirical evaluations with leading MoE-based LRMs (DeepSeek-R1 and Qwen3-235B) on rigorous quantitative and scientific reasoning benchmarks demonstrate noticeable and consistent improvements in reasoning accuracy, cognitive efficiency, and cross-domain generalization. Crucially, our lightweight approach substantially outperforms prevalent reasoning-steering techniques, such as prompt design and decoding constraints, while preserving the model's general instruction-following skills. These results highlight reinforcing cognitive experts as a promising, practical, and interpretable direction to enhance cognitive efficiency within advanced reasoning models.

Related papers

• 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)
• Incentivizing Dual Process Thinking for Efficient Large Language Model Reasoning [75.04643265875072]
  Large reasoning models (LRMs) have demonstrated strong performance on complex reasoning tasks, but often suffer from overthinking.<n>Inspired by the dual process theory in cognitive science, we propose Adaptive Cognition Policy Optimization.<n>ACPO enables LRMs to achieve efficient reasoning through adaptive cognitive allocation and dynamic system switch.
  arXiv  Detail & Related papers  (2025-05-22T07:15:08Z)
• RARE: Retrieval-Augmented Reasoning Modeling [41.24577920467858]
  We propose Retrieval-Augmented Reasoning Modeling (RARE), a novel paradigm that decouples knowledge storage from reasoning optimization.<n>RARE externalizes domain knowledge to retrievable sources and internalizes domain-specific reasoning patterns during training.<n>Experiments demonstrate that lightweight-trained models (e.g., Llama-3.1-8B) could achieve state-of-the-art performance, surpassing retrieval-augmented GPT-4 and DeepSeek-R1 up to approximately 20% accuracy.
  arXiv  Detail & Related papers  (2025-03-30T16:49:44Z)
• Trade-offs in Large Reasoning Models: An Empirical Analysis of Deliberative and Adaptive Reasoning over Foundational Capabilities [101.77467538102924]
  Recent advancements in Large Reasoning Models (LRMs) have demonstrated remarkable performance in specialized reasoning tasks.<n>We show that acquiring deliberative reasoning capabilities significantly reduces the foundational capabilities of LRMs.<n>We demonstrate that adaptive reasoning -- employing modes like Zero-Thinking, Less-Thinking, and Summary-Thinking -- can effectively alleviate these drawbacks.
  arXiv  Detail & Related papers  (2025-03-23T08:18:51Z)
• From Chaos to Order: The Atomic Reasoner Framework for Fine-grained Reasoning in Large Language Models [46.02816479205161]
  We present textbfAtomic Reasoner (textbfAR), a cognitive inference strategy that enables fine-grained reasoning.<n>AR decomposes the reasoning process into atomic cognitive units, employing a cognitive routing mechanism.<n>Results show AR's superior reasoning capabilities without the computational burden of exhaustive solution searches.
  arXiv  Detail & Related papers  (2025-03-20T08:34:53Z)
• Unlocking Structured Thinking in Language Models with Cognitive Prompting [0.0]
  We propose cognitive prompting as a novel approach to guide problem-solving in large language models (LLMs)<n>We introduce three variants: a deterministic sequence of cognitive operations, a self-adaptive variant, and a hybrid variant.<n>Experiments with LLaMA, Gemma2, and Qwen models in each two sizes on the arithmetic reasoning benchmark GSM8K demonstrate that cognitive prompting significantly improves performance compared to standard question answering.
  arXiv  Detail & Related papers  (2024-10-03T19:53:47Z)
• Mimicking Human Intuition: Cognitive Belief-Driven Reinforcement Learning [0.0]
  We propose an innovative framework inspired by cognitive principles: Cognitive Belief-Driven Reinforcement Learning (CBD-RL)<n>CBD-RL transforms conventional trial-and-error learning into a more structured and guided learning paradigm, simulating the human reasoning process.<n>The concrete implementations of this framework, CBDQ, CBDPPO, and CBDSAC, demonstrate superior performance in discrete and continuous action spaces.
  arXiv  Detail & Related papers  (2024-10-02T16:50:29Z)
• MR-Ben: A Meta-Reasoning Benchmark for Evaluating System-2 Thinking in LLMs [55.20845457594977]
  Large language models (LLMs) have shown increasing capability in problem-solving and decision-making.<n>We present a process-based benchmark MR-Ben that demands a meta-reasoning skill.<n>Our meta-reasoning paradigm is especially suited for system-2 slow thinking.
  arXiv  Detail & Related papers  (2024-06-20T03:50:23Z)
• From Heuristic to Analytic: Cognitively Motivated Strategies for Coherent Physical Commonsense Reasoning [66.98861219674039]
  Heuristic-Analytic Reasoning (HAR) strategies drastically improve the coherence of rationalizations for model decisions. Our findings suggest that human-like reasoning strategies can effectively improve the coherence and reliability of PLM reasoning.
  arXiv  Detail & Related papers  (2023-10-24T19:46:04Z)
• Leveraging Expert Consistency to Improve Algorithmic Decision Support [62.61153549123407]
  We explore the use of historical expert decisions as a rich source of information that can be combined with observed outcomes to narrow the construct gap. We propose an influence function-based methodology to estimate expert consistency indirectly when each case in the data is assessed by a single expert. Our empirical evaluation, using simulations in a clinical setting and real-world data from the child welfare domain, indicates that the proposed approach successfully narrows the construct gap.
  arXiv  Detail & Related papers  (2021-01-24T05:40:29Z)

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.