REAT: A Reflective Experience-Augmented Tutoring Framework for Multi-turn Mathematical Instruction

2026-09-24

AI Summary

This study addresses the limitations of large language models in multi-turn pedagogical interactions, specifically their inability to accumulate and reuse instructional experience or adapt to diverse student needs. To this end, we propose the REAT framework, which introduces a novel Observer-Critic-Mentor multi-agent distillation pipeline that structures teaching experiences from historical dialogues into transferable knowledge capable of generalizing across models. Furthermore, by integrating state-aware retrieval techniques, REAT enables real-time experience invocation and adaptive tutoring. Experimental results demonstrate that REAT significantly outperforms prompt engineering and supervised fine-tuning baselines on mathematical tutoring tasks, exhibiting particularly strong performance in complex, low-proficiency scenarios.

Abstract

Current Large Language Models (LLMs) excel at solving complex mathematical problems, yet this proficiency does not inherently translate into effective tutoring. While advanced LLM tutors may leverage multi-agent frameworks or fine-tuning, most still lack a mechanism to systematically accumulate and reuse pedagogical experience over time, limiting their adaptability to diverse student needs during fluid, multi-turn interactions. To bridge this gap, we propose the Reflective Experience-Augmented Tutoring (REAT) framework, which couples experience distillation from historical dialogues with real-time adaptive retrieval. Driven by a multi-agent Observer-Critic-Mentor (OCM) distillation pipeline, REAT reviews past conversational trajectories and distills raw interactions into structured, problem-agnostic pedagogical experiences. During live tutoring, a state-aware retrieval module injects these curated experiences to provide adaptive scaffolding based on the student's cognitive state. Experiments demonstrate that the proposed framework significantly outperforms both prompt-only and supervised fine-tuning (SFT) baselines, particularly in improving complex, low-scoring tutoring scenarios. Crucially, the distilled experiences exhibit robust generalization across diverse model architectures and mathematical datasets.

Problem

• Large Language Models
• Mathematical Tutoring
• Multi-turn Interaction
• Pedagogical Experience
• Adaptive Scaffolding

Innovation

• Reflective Experience-Augmented Tutoring
• Multi-agent Distillation Pipeline
• State-aware Retrieval
• Pedagogical Experience Distillation
• Cross-architecture Generalization