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

TL;DR

This paper introduces Evidence-Decision-Feedback (EDF), a theoretical framework for adaptive scaffolding in LLM-based pedagogical agents, demonstrated through a STEM+C problem-solving agent in high school classrooms.

Contribution

It presents a novel, theory-driven approach to personalized LLM tutoring that aligns feedback with student understanding and promotes scaffold fading.

Findings

EDF-guided interactions improve feedback relevance

Promotes scaffold fading in classroom settings

Supports evidence-grounded, interpretable explanations

Abstract

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