Evidence-Decision-Feedback: Theory-Driven Adaptive Scaffolding for LLM Agents

2026-02-01

AI Summary

This study addresses the limitation of existing large language model (LLM)-based multi-agent tutoring systems, which often lack dynamic awareness of students’ comprehension levels and thus struggle to deliver personalized scaffolding. To overcome this, the authors propose the Evidence-Decision-Feedback (EDF) framework—a novel approach that integrates evidential reasoning with pedagogical decision-making to enable an interpretable, progressively fading, and over-reliance-averse adaptive scaffolding mechanism. Built upon an LLM multi-agent architecture and grounded in intelligent tutoring system theory and collaborative agent design, EDF was evaluated in authentic high school STEM+C classrooms. Results demonstrate that the framework effectively aligns with students’ cognitive states, delivers timely and meaningful feedback, and significantly enhances knowledge construction and critical thinking development.

Technology Category

• Multiagent Systems: Multiagent Learning
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Humans and AI: Planning and Decision Support for Human-Machine Teams

Application Category

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

Abstract

Multi-agent LLM architectures offer opportunities for pedagogical agents to help students construct domain knowledge and develop critical-thinking skills, yet many operate on a"one-size-fits-all"basis, limiting their ability to provide personalized support. To address this, we introduce Evidence-Decision-Feedback (EDF), a theoretical framework for adaptive scaffolding using LLMs. EDF integrates elements of intelligent tutoring systems and agentic behavior by organizing interactions around evidentiary inference, pedagogical decision-making, and adaptive feedback. We instantiate EDF through Copa, an agentic collaborative peer agent for STEM+C problem-solving. In an authentic high school classroom study, we show that EDF-aligned interactions align feedback with students'demonstrated understanding and task mastery; promote gradual scaffold fading; and support interpretable, evidence-grounded explanations without fostering overreliance.

Problem

• personalized support
• multi-agent LLM
• adaptive scaffolding
• intelligent tutoring
• one-size-fits-all

Innovation

• adaptive scaffolding
• evidentiary inference
• LLM agents
• pedagogical decision-making
• multi-agent LLM