Reasoning Beyond Language: A Comprehensive Survey on Latent Chain-of-Thought Reasoning

TL;DR

This survey categorizes latent chain-of-thought (CoT) reasoning in LLMs into token-wise horizontal and layer-wise vertical frameworks, detailing initialization, optimization, and inference strategies for latent reasoning. It contrasts latent CoT with explicit CoT, discusses training paradigms (pre-training, post-training with SFT and RL), and analyzes external behavior versus internal mechanisms. The work highlights applications across textual and multimodal reasoning, information retrieval, and recommendations, and identifies major challenges such as training instability, generalization gaps, and interpretability. By providing a unified taxonomy and synthesis of methods, the paper sets a foundation for advancing latent CoT research and practical, efficient reasoning in AI systems.

Abstract

Large Language Models (LLMs) have shown impressive performance on complex tasks through Chain-of-Thought (CoT) reasoning. However, conventional CoT relies on explicitly verbalized intermediate steps, which constrains its broader applicability, particularly in abstract reasoning tasks beyond language. To address this, there has been growing research interest in \textit{latent CoT reasoning}, where the reasoning process is embedded within latent spaces. By decoupling reasoning from explicit language generation, latent CoT offers the promise of richer cognitive representations and facilitates more flexible, faster inference. This paper aims to present a comprehensive overview of this emerging paradigm and establish a systematic taxonomy. We analyze recent advances in methods, categorizing them from token-wise horizontal approaches to layer-wise vertical strategies. We then provide in-depth discussions of these methods, highlighting their design principles, applications, and remaining challenges. We hope that our survey provides a structured foundation for advancing this promising direction in LLM reasoning. The relevant papers will be regularly updated at https://github.com/EIT-NLP/Awesome-Latent-CoT.

Figures (5)

• Figure 1: Explicit CoT (left) generates reasoning steps with natural language, while latent CoT (right) allows the model to reason internally in latent spaces.
• Figure 2: Taxonomy of latent Chain-of-Thought (CoT) reasoning.
• Figure 3: Illustration of three representative categories for latent representation initialization.
• Figure 4: Illustration of inference exploration (both parallel and sequential scaling) for latent CoT reasoning.
• Figure 5: Illustration of layer-wise vertical-level latent CoT reasoning.