Chain of Mindset: Reasoning with Adaptive Cognitive Modes

TL;DR

The paper tackles the limitation of fixed-mindset prompting in large language models by introducing Chain of Mindset (CoM), a training-free framework that orchestrates four heterogeneous mindsets (Spatial, Convergent, Divergent, Algorithmic) via a Meta-Agent and a bidirectional Context Gate. By modeling reasoning as a sequential decision process with state-dependent mindset dispatch and information filtering, CoM achieves state-of-the-art accuracy across six diverse benchmarks while maintaining reasonable efficiency and cross-model generalization. Key contributions include formalizing the mindset framework, detailing the three-layer architecture, and demonstrating substantial gains on math, coding, science QA, and spatial tasks, with insightful ablations and analyses of mindset invocation patterns. The work advances AI reasoning toward human-like cognitive flexibility, offering a transparent, training-free paradigm that can inform future research in metacognition, multimodal reasoning, and safe, auditable AI systems.

Abstract

Human problem-solving is never the repetition of a single mindset, by which we mean a distinct mode of cognitive processing. When tackling a specific task, we do not rely on a single mindset; instead, we integrate multiple mindsets within the single solution process. However, existing LLM reasoning methods fall into a common trap: they apply the same fixed mindset across all steps, overlooking that different stages of solving the same problem require fundamentally different mindsets. This single-minded assumption prevents models from reaching the next level of intelligence. To address this limitation, we propose Chain of Mindset (CoM), a training-free agentic framework that enables step-level adaptive mindset orchestration. CoM decomposes reasoning into four functionally heterogeneous mindsets: Spatial, Convergent, Divergent, and Algorithmic. A Meta-Agent dynamically selects the optimal mindset based on the evolving reasoning state, while a bidirectional Context Gate filters cross-module information flow to maintain effectiveness and efficiency. Experiments across six challenging benchmarks spanning mathematics, code generation, scientific QA, and spatial reasoning demonstrate that CoM achieves state-of-the-art performance, outperforming the strongest baseline by 4.96\% and 4.72\% in overall accuracy on Qwen3-VL-32B-Instruct and Gemini-2.0-Flash, while balancing reasoning efficiency. Our code is publicly available at \href{https://github.com/QuantaAlpha/chain-of-mindset}{https://github.com/QuantaAlpha/chain-of-mindset}.

Figures (5)

• Figure 1: Performance comparison on Qwen3-VL-32B-Instruct across six reasoning benchmarks.
• Figure 2: Comparison of reasoning paradigms. (a) Single-mode reasoning applies a single mindset throughout, failing to address heterogeneous sub-task demands. (b) Static reasoning strategy selection chooses a strategy at task onset but cannot adapt to intermediate states. (c) Chain of Mindset dynamically switches mindsets at subtask boundaries based on the progress of reasoning.
• Figure 3: Overview of the Chain of Mindset framework. Left: The Meta-Agent operates as a meta-cognitive orchestrator, iteratively generating cognitive decisions (<cognitive_decision>), dispatching subtasks to specialized mindsets via call instructions (<call_mindset>), receiving summarized results (<mindset_result>), and internalizing key insights (<Insight>) before producing the final answer. The agent may revise its plan when intermediate results warrant replanning. Right: The Mindset Experts comprise four heterogeneous modules—Divergent, Algorithmic, Convergent, and Spatial—each providing distinct cognitive capabilities. The bidirectional Context Gate mediates information flow: the Input Gate filters relevant history for mindset execution, while the Output Gate distills verbose reasoning traces into concise results for the main chain.
• Figure 4: Fermi problem (#494) demonstrating the Spatial Mindset. The Spatial Mindset generates an anatomy diagram to visually ground the abstract proportion and extract the head-to-arm ratio ($\approx 3.5\times$). The subsequent Convergent call resolves an ambiguity: "head size" maps to the Sun's radius rather than diameter.
• Figure 5: Our method achieves state-of-the-art performance while balancing reasoning efficiency.