Meta-R1: Empowering Large Reasoning Models with Metacognition

TL;DR

Meta-R1 introduces a meta-cognitive framework to large reasoning models, significantly improving their performance, efficiency, and adaptability by enabling explicit thinking about their reasoning process.

Contribution

It is the first systematic framework that incorporates explicit metacognitive capabilities into large reasoning models, inspired by cognitive science principles.

Findings

Surpasses state-of-the-art by up to 27.3% on benchmarks.

Reduces token consumption by 15.7% to 32.7%.

Maintains robust performance across datasets and models.

Abstract

Large Reasoning Models (LRMs) demonstrate remarkable capabilities on complex tasks, exhibiting emergent, human-like thinking patterns. Despite their advances, we identify a fundamental limitation: current LRMs lack a dedicated meta-level cognitive system-an essential faculty in human cognition that enables "thinking about thinking". This absence leaves their emergent abilities uncontrollable (non-adaptive reasoning), unreliable (intermediate error), and inflexible (lack of a clear methodology). To address this gap, we introduce Meta-R1, a systematic and generic framework that endows LRMs with explicit metacognitive capabilities. Drawing on principles from cognitive science, Meta-R1 decomposes the reasoning process into distinct object-level and meta-level components, orchestrating proactive planning, online regulation, and adaptive early stopping within a cascaded framework. Experiments on three challenging benchmarks and against eight competitive baselines demonstrate that Meta-R1 is: (I) high-performing, surpassing state-of-the-art methods by up to 27.3%; (II) token-efficient, reducing token consumption to 15.7% ~ 32.7% and improving efficiency by up to 14.8% when compared to its vanilla counterparts; and (III) transferable, maintaining robust performance across datasets and model backbones.