The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes

Периодическая таблица рассуждений больших языковых моделей: структурированный обзор парадигм, методов и режимов отказа
Avinash Anand, Mahisha Ramesh, Avni Mittal, Ashutosh Kumar, Rishitej Reddy Vyalla, Erik Cambria, Zhengkui Wang, Timothy Liu, Aik Beng Ng, Simon See, Rajiv Ratn Shah
2026-06-09

Chain-of-ThoughtLLM reasoningmulti-hop question answeringretrieval-augmented reasoningverifier-guided inference
Reasoning has become central to how Large Language Models (LLMs) are evaluated and interpreted, spanning Chain-of-Thought (CoT), mathematical problem-solving, multi-hop question answering, code generation, retrieval-augmented reasoning, tool use, and multimodal decision-making. In this survey, we introduce the Periodic Table of LLM Reasoning, a framework organizing 300+ recent papers by reasoning paradigm, methodological mechanism, evaluation setting, and failure mode. We classify LLM reasoning into nine paradigms: Chain-of-Thought, Multi-Hop, Mathematical, Commonsense, Visual and Temporal, Code and Algorithmic, Retrieval-Augmented, Tool-Augmented or Agentic, and Reinforcement Learning-based reasoning. For each, we review approaches, including prompting, architectural interventions, supervised fine-tuning, verifier-guided inference, reward modeling, retrieval, tool interfaces, agentic workflows, and benchmark design. We argue that LLM reasoning is not a single emergent capability but a family of scaffolded behaviors shaped by model scale, task structure, external memory, supervision, and evaluation protocols. We synthesize recurring failure modes, including hallucinated reasoning, brittle multi-step inference, spurious rationales, weak causal grounding, poor out-of-distribution generalization, benchmark contamination, and unreliable self-verification. Progress is difficult to compare across paradigms because gains may arise from prompting, retrieval, verifier design, or benchmark structure rather than general reasoning ability. The survey connects methods to their assumptions, strengths, and failure modes, providing a reference map of the field and a diagnostic framework for future work. We conclude that robust LLM reasoning will require meta-reasoning, multimodal and temporal grounding, adaptive tool use, and principled evaluation under distribution shift.
1
Across paradigms, recurring failures include hallucinated reasoning, brittle multi-step inference, spurious rationales, weak causal grounding, poor out-of-distribution generalization, contamination, and unreliable self-verification.
2
Comparisons of reasoning progress are confounded because improvements may result from prompting, retrieval, verifier design, or benchmark structure rather than general reasoning ability.
3
It argues that LLM reasoning is a family of scaffolded behaviors shaped by model scale, task structure, external memory, supervision, and evaluation protocols.
4
The survey identifies meta-reasoning, multimodal and temporal grounding, adaptive tool use, and principled distribution-shift evaluation as requirements for robust LLM reasoning.
5
The survey organizes 300+ studies into a structured framework covering nine LLM reasoning paradigms, methods, evaluation settings, and failure modes.

Large Language Model (LLM) reasoning

reasoning paradigms, methodological mechanisms, evaluation settings, and recurring failure modes of LLMs

Publication Details
Publication Date
2026-06-09
Journal
Publisher
ISSN
Access Type
Author Information
Authors
Avinash Anand
Mahisha Ramesh
Avni Mittal
Ashutosh Kumar
Rishitej Reddy Vyalla
Erik Cambria
Zhengkui Wang
Timothy Liu
Aik Beng Ng
Simon See
Rajiv Ratn Shah
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%