Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs

2026/03/10 by Zorik Gekhman, Roee Aharoni, Eran Ofek +3 · 2 voices
Computer Science · #cs.CL

paper · pdf · doi:10.48550/arxiv.2603.09906

Abstract

While reasoning in LLMs plays a natural role in math, code generation, and multi-hop factual questions, its effect on simple, single-hop factual questions remains unclear. Such questions do not require step-by-step logical decomposition, making the utility of reasoning highly counterintuitive. Nevertheless, we find that enabling reasoning substantially expands the capability boundary of the model's parametric knowledge recall, unlocking correct answers that are otherwise effectively unreachable. Why does reasoning aid parametric knowledge recall when there are no complex reasoning steps to be done? To answer this, we design a series of hypothesis-driven controlled experiments, and identify two key driving mechanisms: (1) a computational buffer effect, where the model uses the generated reasoning tokens to perform latent computation independent of their semantic content; and (2) factual priming, where generating topically related facts acts as a semantic bridge that facilitates correct answer retrieval. Importantly, this latter generative self-retrieval mechanism carries inherent risks: we demonstrate that hallucinating intermediate facts during reasoning increases the likelihood of hallucinations in the final answer. Finally, we show that our insights can be harnessed to directly improve model accuracy by prioritizing reasoning trajectories that contain hallucination-free factual statements.

Citations

• What Characterizes Effective Reasoning? Revisiting Length, Review, and Structure of CoT
• SimpleQA Verified: A Reliable Factuality Benchmark to Measure Parametric Knowledge
• Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
• How Well Can Reasoning Models Identify and Recover from Unhelpful Thoughts?
• Rewarding the Unlikely: Lifting GRPO Beyond Distribution Sharpening
• Don't Overthink it. Preferring Shorter Thinking Chains for Improved LLM Reasoning
• Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens
• Qwen3 Technical Report
• Reasoning Models Don't Always Say What They Think
• Chain-of-Thought Reasoning In The Wild Is Not Always Faithful
• Cognitive Behaviors that Enable Self-Improving Reasoners, or, Four Habits of Highly Effective STaRs
• Towards Thinking-Optimal Scaling of Test-Time Compute for LLM Reasoning
• When More is Less: Understanding Chain-of-Thought Length in LLMs
• s1: Simple test-time scaling
• Tulu 3: Pushing Frontiers in Open Language Model Post-Training
• Measuring short-form factuality in large language models
• Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models
• Large Language Monkeys: Scaling Inference Compute with Repeated Sampling
• Let's Think Dot by Dot: Hidden Computation in Transformer Language Models
• BioXP-0.5B: Explainable Medical-AI via RL-GRPO
• The Impact of Reasoning Step Length on Large Language Models
• Think before you speak: Training Language Models With Pause Tokens
• Measuring Faithfulness in Chain-of-Thought Reasoning
• Let's Verify Step by Step
• Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions
• Measuring and Narrowing the Compositionality Gap in Language Models
• Distributing Accountability, Not Capability: Phase Separation and the LLM Workflow Quadrant in Autonomous AI Agent Architectures
• BNAI, NO-TOKEN, and MIND-UNITY: Pillars of a Systemic Revolution in Artificial Intelligence
• Simple Entity-Centric Questions Challenge Dense Retrievers
• Evaluating Large Language Models Trained on Code
• A spreading-activation theory of semantic processing.

Discussions

• New Preprint From Google Research: "Thinking to Recall: How #Reasoning Unlocks Parametric Knowledge in #LLMs" (via #arxiv) arxiv.org/abs/2603.09906 [bsky, 1 points, 0 comments]
• 面白いのは理由で、意味のない "Let me think" の反復でも少し効く一方、関連事実を先に出すともっと効くらしい。計算の助走と、事実の呼び水。推論って、論理だけではなく記憶の索引でもあるんですね。 https://research.google/blog/thinking-to-recall-how-reasoning-unlocks-parametric-knowledge-in-llm [bsky, 0 points, 1 comments]

Related

• Unveiling Factual Recall Behaviors of Large Language Models through Knowledge Neurons
• Mitigating Factual Hallucination in Large Reasoning Models via Mixed-Mode Advantage Regularization
• Reasoning Error from Known Fact: Step-Level Self-Consistency Group Relative Policy Optimization for LLM
• The Reasoning Trap: How Enhancing LLM Reasoning Amplifies Tool Hallucination
• The Reasoning-Memorization Interplay in Language Models Is Mediated by a Single Direction