Implicit Reasoning in Language Models

Follow this topic Ask about this topic

The field asks how hidden states, recurrent or looped computation, latent tokens, and memory can support multi-step inference without generating full rationales. Approaches include distilling from explicit reasoning traces, adaptively allocating test-time computation, and using mechanistic tools such as activation patching and probes; evaluations span math, symbolic and multi-hop tasks, retrieval, commonsense, and hidden constraints. These methods can reduce token use or latency and sometimes match explicit reasoning, but results depend on supervision, model, and task, while extra computation can cause shortcuts or overthinking. A central open question is whether latent states carry causally useful, generalizable reasoning—and how to verify, interpret, and secure it when visible answers or rationales may be misleading.

Latest papers

99 papers20 papers in the last four weeks, up 122% on the four weeks before. 0.1% of all new papers.

Ask Before You Answer: A Question-Driven Evidence Reasoning Architecture (QERA) for In-Context Learning
Oct 26, 2026Jinhu Qi, Minda Hu, Wentao Zhang +2Evidence-Grounded QA Implicit Reasoning in Language Models
The Chinese University of Hong Kong Hong Kong, China · Macao Polytechnic University Macao, China · The Hong Kong Polytechnic University Hong Kong, China
Reasoning or Just Prompting? Probing the Mechanism of Latent Chemical Reasoning
Oct 25, 2026Kaiyuan Lu, Jiatong Li, Yatao BianChemical Reasoning in LLMs Implicit Reasoning in Language Models
National University of Singapore
Curriculum Learning Teaches LLMs Long-Horizon Implicit Reasoning, Up To A Limit
Oct 19, 2026Jimson Huang, Maxwell Nickell, Ruowang Zhang +1Long-Horizon LLM Reasoning Implicit Reasoning in Language Models
Purdue University West Lafayette, IN 47907, USA
The One-Pixel of Reasoning: One Truthful Relation Is Enough to Mislead LLM Agents
Oct 17, 2026Jinwei Hu, Youcheng Sun, Yi Dong +1Adversarial Attacks on LLMs LLM Agent Reliability
School of Computer Science and Informatics, University of Liverpool · Department of Computer Science, MBZUAI
Beyond Large Language Models: A Mental World Model Framework for Mental Modeling, Simulation, Evolution, and Interpretability
Oct 16, 2026Qingshi Wang, Huiqi TongTheory of Mind Implicit Reasoning in Language Models
School of Computer Science, University of Bristol, Bristol, United Kingdom · Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, CA, United States
Sep-Latent: Separator Tokens as Natural Compression Anchors for Implicit Reasoning
Oct 16, 2026Yinlong Xu, Jianfeng Huang, Zhengdong Zhou +5Efficient Language Model Reasoning Implicit Reasoning in Language Models
College of Computer Science and Technology, Zhejiang University, Hangzhou, China · Alibaba Group, Hangzhou, China
Reasoning Reranking with Semantic IDs as the Sole Item Representation
Oct 16, 2026Zhenwei Tang, Jintao Liu, Di Bai +2LLM Reranking Semantic ID Recommendation
Google · University of Toronto
ToMStories: Interpreting Theory-of-Mind in Synthetic Narratives
Oct 15, 2026Melanie Subbiah, Chia-En Jacklyn Tsai, Kathleen McKeownTheory of Mind Implicit Reasoning in Language Models
Columbia University
INTACT: Inducing Latent Reasoning on a Frozen Backbone
Oct 14, 2026Kutay Tire, Atula Tejaswi, Haris Vikalo +1Efficient Language Model Reasoning Implicit Reasoning in Language Models
The University of Texas at Austin
Monitoring latent reasoning in looped language models
Oct 13, 2026Pawel Olszowiec, Paulina Tomaszewska, Maciej Żelaszczyk +2Implicit Reasoning in Language Models Recursive Language Models
Samsung AI Center, Warsaw, Poland · IFTR, Polish Academy of Sciences, Warsaw, Poland · University of Warsaw, Poland
Facts Don't Speak Louder than Words: The Behavioral Essence of Long-CoT Distillation
Oct 13, 2026Yongcan Yu, Jian Liang, Lingxiao He +5CoT Distillation Language Model Distillation
NLPR & MAIS, Institute of Automation, Chinese Academy of Sciences · School of Artificial Intelligence, University of Chinese Academy of Sciences · Meituan +1
Latent Taxonomies: How Property Induction Develops in Language Model Pre-training
Oct 11, 2026Jan Sobotka, Ekaterina Goliakova, Marcin Detyniecki +2Inductive Reasoning in Language Models Inductive Inference
EPFL · AXA · Sorbonne University, CNRS, LIP6 +1
The Thinking Leakage in Reinforcement Learning for LLMs
Oct 9, 2026Wenxuan Zhu, Shuming Liu, Bang An +4RL for Language Model Reasoning RL Post-Training
King Abdullah University of Science and Technology · Shanghai Jiao Tong University
SanSi: A Looped Typed Decision Model for System 1.5 Thinking
Oct 6, 2026Shuyu Gan, Young-Jun Lee, Dongyeop KangEfficient Language Model Reasoning Implicit Reasoning in Language Models
University of Minnesota
CALR: Continuous Anchored Latent Reasoning via Render-of-Thought Compression
Oct 5, 2026Zhaoyang Wei, Bowen Jiang, Yanchao Hao +5Efficient Language Model Reasoning Latent Visual Reasoning
University of the Chinese Academy of Sciences · Tencent
Base Models Can Reason By Taking a Cue From Training Data
Oct 5, 2026Sophie L. Wang, Amil Dravid, Rulin Shao +3LLM Prompting RL for Language Model Reasoning
MIT · UC Berkeley · University of Washington +1
HyperThink: Text-to-Parameter Hypernetworks for Efficient Reasoning
Oct 2, 2026Donggyun Kim, Jack Lu, Chanwoo Kim +2Efficient Language Model Reasoning LLM Inference Acceleration
KAIST School of Computing · NYU Courant Institute School of Mathematics, Computing, and Data Science
AURAL: Adaptive Latent Reasoning with Joint Chunk for Speech Language Models
Oct 1, 2026Yuxiang Wang, Kunyu Feng, Yuancheng Wang +12RL for Language Model Reasoning Audio Reasoning
The Chinese University of Hong Kong, Shenzhen · Tencent Hunyuan · Tsinghua University +2
Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard
Sep 30, 2026Julian Schulz, Lukas Fülle, Rieke FruengelLLM Security Implicit Reasoning in Language Models
Meridian Cambridge · SAIGE
Bongard: Training Machine Intuition
Sep 30, 2026Li Ding, Haidi Jin, Chen JiImplicit Reasoning in Language Models AI-Assisted Decision Making
AgentBull Pte Ltd
DatalogBench: Evaluating Large Language Models on Text-to-Datalog Synthesis
Sep 29, 2026Yuan Li, Hanyun Jiang, Guowei Tian +2LLM Evaluation LLM-Based Program Synthesis
The State Key Laboratory of Blockchain and Data Security, Zhejiang University · National University of Singapore
Capable yet Parsimonious: Extracting and Characterizing Hidden Chain-of-Thought in Frontier Models
Sep 22, 2026Xiaoyu Luo, Tao Ren, Wenrui Yu +3LLM Interpretability CoT Reasoning
Department of Computer Science Aalborg University, Copenhagen, Denmark · Department of Electronic Systems Aalborg University, Copenhagen, Denmark · Seafill
CogGym: Towards Large-Scale Comparative Evaluation of Human and Machine Cognition
Sep 18, 2026Lance Ying, Jinzhou Wu, Yingshan Susan Wang +53LLM Evaluation Cognitive Modeling
State of Thought Enables Endogenous Reasoning
Sep 13, 2026Zhiren Gong, Yikun Hou, Zihao Zeng +3Efficient Language Model Reasoning Implicit Reasoning in Language Models
College of Computing and Data Science, Nanyang Technological University, Singapore · Interdisciplinary Graduate Programme, Nanyang Technological University, Singapore · Information Science & Engineering, KTH Royal Institute of Technology, Sweden +1
Training Specialist Models without Reasoning Trajectories for Domain Expert Distillation
Sep 12, 2026Yilei Tu, Zihao Li, Shaoxiong Ji +2Implicit Reasoning in Language Models Language Model Distillation
University of British Columbia · Shanghai Artificial Intelligence Laboratory · University of Helsinki +2
How Should Reasoning Be Organized in a Transformer's Latent Space?
Sep 12, 2026Hongyu Gu, Chang Liu, Jingwen FuTransformer Attention Implicit Reasoning in Language Models
University of Science and Technology of China Hefei, China · Zhongguancun Academy Beijing, China
RFCLLM: Evaluating LLMs' Reasoning Ability of Network Protocol State Machines
Sep 11, 2026Anqi Chen, Dan Goldwasser, Cristina Nita-RotaruImplicit Reasoning in Language Models
Northeastern University · Purdue University
Revisiting Complete Reasoning Traces for Post-Training
Sep 7, 2026Jaehui Hwang, Sangdoo Yun, Byeongho Heo +1LLM Post-Training Implicit Reasoning in Language Models
NAVER AI Lab
Assessing Adversarial Robustness of Latent Reasoning Models
Sep 3, 2026Shaolong Chen, Ang Li, Mingjie Li +1Adversarial Robustness Implicit Reasoning in Language Models
Yuanpei College, Peking University · State Key Lab of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University · CISPA Helmholtz Center for Information Security
Wrong Prediction, Right Answer: Recovering Evidence from Collapsed LLM Sequence Scores
Aug 31, 2026Qiyao Yan, Chenpeng Wang, Liangming PanLanguage Modeling Language Model Calibration
State Key Laboratory of Multimedia Information Processing, Peking University · School of Computer Science, Peking University · YiXin-AILab, YIXIN, Beijing, China +1

Oct 26, 2026J/K move · Enter open · S save

Ask Before You Answer: A Question-Driven Evidence Reasoning Architecture (QERA) for In-Context Learning

Jinhu Qi, Minda Hu, Wentao Zhang, Zixing Song+1

The Chinese University of Hong Kong Hong Kong, China · Macao Polytechnic University Macao, China · The Hong Kong Polytechnic University Hong Kong, China

Evidence-Grounded QA Implicit Reasoning in Language Models LLM Answer Verification

Open paper AskPDF arXiv

AI assistants are increasingly asked to reason over a user's own web traces: group chats, community threads, activity logs and revision histories. On CL-bench Life, a benchmark of such real-life context learning, frontier LLMs solve fewer than one task in five. We find that the request is not the specification: over 80% of the benchmark's rubrics share less than 60% of their content words with the final request, as for a rule set two turns earlier or every record that meets the user's criterion, and LLMs answering directly pass only 67% of these implicit rubrics against 81% of the others. We propose QERA, a Question-driven Evidence Reasoning Architecture that asks before it answers. QERA first turns the request into a requirement checklist and typed information needs, filtered by code for faithfulness to the user's words; it then answers from the whole conversation, asks the same questions again with quotes verified by code, and applies a guarded local repair. It needs no training or rubrics and makes about four LLM calls per task. On all 405 tasks with four frontier LLMs, QERA attains the highest solving rate on every LLM (21.0% on average, against 18.2% for the strongest of three inference-time baselines and 14.6% for direct answering) and leads every baseline significantly when pooled across LLMs, while using fewer tokens than PSCI and 4-5% of the tokens of Ctx2Skill. Ablations on GPT-6-Luna show that the gain is settled before the answer, by a fixed answer protocol (+3.8 rubric points) and the request-specific questions (+1.6), and that it is concentrated on implicit rubrics, with checking afterwards refining mainly numeric requirements.