Beyond Fast and Slow: Cognitive-Inspired Elastic Reasoning for Large Language Models

2025-12-17

AI Summary

Large language models (LLMs) struggle to dynamically balance reasoning efficiency and accuracy in response to query complexity. Method: This paper proposes CogER, a cognition-inspired elastic reasoning framework that (1) establishes an elastic reasoning paradigm grounded in human hierarchical cognition; (2) formulates strategy selection as a reinforcement learning–driven Markov decision process for complexity-aware, dynamic mode switching; and (3) introduces a cognitive tool coordination mechanism enabling LLMs to autonomously invoke external tools for reasoning augmentation. Results: On in-domain and out-of-domain tasks, CogER improves average exact match accuracy by 13% and 8%, respectively—outperforming state-of-the-art test-time scaling methods. The framework achieves adaptive reasoning without compromising fidelity, demonstrating robust generalization across diverse query complexities and task distributions.

Technology Category

• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large language models (LLMs) have demonstrated impressive performance across various language tasks. However, existing LLM reasoning strategies mainly rely on the LLM itself with fast or slow mode (like o1 thinking) and thus struggle to balance reasoning efficiency and accuracy across queries of varying difficulties. In this paper, we propose Cognitive-Inspired Elastic Reasoning (CogER), a framework inspired by human hierarchical reasoning that dynamically selects the most suitable reasoning strategy for each query. Specifically, CogER first assesses the complexity of incoming queries and assigns them to one of several predefined levels, each corresponding to a tailored processing strategy, thereby addressing the challenge of unobservable query difficulty. To achieve automatic strategy selection, we model the process as a Markov Decision Process and train a CogER-Agent using reinforcement learning. The agent is guided by a reward function that balances solution quality and computational cost, ensuring resource-efficient reasoning. Moreover, for queries requiring external tools, we introduce Cognitive Tool-Assisted Reasoning, which enables the LLM to autonomously invoke external tools within its chain-of-thought. Extensive experiments demonstrate that CogER outperforms state-of-the-art Test-Time scaling methods, achieving at least a 13% relative improvement in average exact match on In-Domain tasks and an 8% relative gain on Out-of-Domain tasks.

Problem

• Balancing reasoning efficiency and accuracy for varying query difficulties
• Dynamically selecting optimal reasoning strategies for each query
• Enabling autonomous external tool invocation for complex queries

Innovation

• Dynamic strategy selection based on query complexity assessment
• Reinforcement learning agent balancing solution quality and cost
• Autonomous tool invocation within chain-of-thought reasoning