Fast, Slow, and Tool-augmented Thinking for LLMs: A Review

2025-08-17

AI Summary

Current large language models (LLMs) lack a systematic framework for selecting appropriate reasoning strategies—such as intuitive response, stepwise deduction, or tool invocation—based on task requirements. To address this, we propose the first LLM reasoning strategy taxonomy grounded in cognitive psychology’s dual-process theory, jointly characterizing “fast/slow thinking” and “internal/external knowledge boundaries.” This framework explicitly defines three reasoning modes: intuitive, incremental, and tool-augmented. Through systematic literature review and categorical analysis, we model existing technical approaches, identify key decision factors—including task complexity, knowledge accessibility, and real-time constraints—and synthesize corresponding implementation mechanisms. Our framework establishes a theoretical foundation and practical guidance for interpretable strategy selection, dynamic scheduling, and controllable optimization of LLM reasoning, thereby advancing the development of more efficient and robust adaptive reasoning systems.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Natural Language Processing: (Large) Language Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Large Language Models (LLMs) have demonstrated remarkable progress in reasoning across diverse domains. However, effective reasoning in real-world tasks requires adapting the reasoning strategy to the demands of the problem, ranging from fast, intuitive responses to deliberate, step-by-step reasoning and tool-augmented thinking. Drawing inspiration from cognitive psychology, we propose a novel taxonomy of LLM reasoning strategies along two knowledge boundaries: a fast/slow boundary separating intuitive from deliberative processes, and an internal/external boundary distinguishing reasoning grounded in the model's parameters from reasoning augmented by external tools. We systematically survey recent work on adaptive reasoning in LLMs and categorize methods based on key decision factors. We conclude by highlighting open challenges and future directions toward more adaptive, efficient, and reliable LLMs.

Problem

• Adapting LLM reasoning strategies to task demands
• Classifying reasoning by fast/slow and internal/external boundaries
• Surveying adaptive reasoning methods for efficient LLMs

Innovation

• Fast and slow reasoning strategies for LLMs
• Tool-augmented thinking for enhanced reasoning
• Taxonomy based on cognitive psychology principles