[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85508-en":3,"doc-seo-85508-105":29,"detail-sidebar-cat-0-en-105":83},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},85508,34359740700684,"Finn","https://ap-avatar.wpscdn.com/avatar/1f400023980c374ae676?_k=1777273430885731487",8,"Research & Report","Evidence-Decision-Feedback: Theory-Driven Adaptive Scaffolding for LLM Agents","LLMs enable pedagogical agents that can support learners, but many systems provide generic, one-size-fits-all assistance that limits personalization and explanatory clarity. Evidence-Decision-Feedback (EDF) is proposed as a theory-driven framework for adaptive scaffolding, unifying intelligent tutoring ideas with agentic behavior. EDF structures interactions around evidentiary inference, pedagogical decision-making, and adaptive feedback. EDF is instantiated in Copa, a collaborative peer agent, and evaluated in an authentic high school study showing alignment with demonstrated understanding, scaffold fading, and evidence-grounded explanations that avoid encouraging overreliance.","arXiv :2602 .01415v5 [ cs .MA] 11 Jul 2026  \nEvidence-Decision-Feedback: Theory-Driven Adaptive Scaffolding for LLM Agents  \nClayton Cohn [0000−0003−0856−9587], Siyuan Guo [0009−0001−4305−9147], Surya Rayala [0009−0005−8192−8138], Hanchen David Wang [0000−0001−5990−5865], Naveeduddin Mohammed [0000−0002−3706−2884], Umesh Timalsina [0000−0002−5430−3993], Shruti Jain [0009−0000−7853−0560], Angela Eeds,  \nMenton Deweese [0000−0001−7361−3826], Pamela J. Osborn  \nPopp [0009−0007−5782−8895], Rebekah Stanton [0009−0003−6562−8018], Shakeera Walker, Meiyi Ma [0000−0001−6916−8774], and Gautam Biswas [0000−0002−2752−3878]  \nVanderbilt University, Nashville, TN, USA {clayton.a.cohn, siyuan.guo, surya.chand.rayala, [hanchen.wang.1](hanchen.wang.1) , naveeduddin.mohammed, umesh.timalsina, shruti.jain, angela.eeds, menton.deweese, pamela.popp, rebekah.stanton, shakeera.walker, [meiyi.ma](meiyi.ma) , [gautam.biswas}@vanderbilt.edu](gautam.biswas}@vanderbilt.edu)  \nAbstract. 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.  \nKeywords: Adaptive Scaffolding · Pedagogical Agents · LLMs · Agentic AI · Multi-Agent Architectures  \n1 Introduction  \nLarge language model (LLM) pedagogical agents afford dialogic support that aligns with learning theories emphasizing knowledge construction through social interaction, including Social Cognitive Theory [2] (SCT) and Social Constructivism [35] . However, recent studies have raised concerns: Kosmyna et al.(2025) [21] reported 83% of students receiving ChatGPT assistance while writing were unable to quote their own work, and Zhou et al. (2025) [36] found GPT-4 feedback on compiler errors benefited students during assisted tasks but  \n2 Cohn et al.  \nnot after feedback was withdrawn. While they can be helpful, these agents can also encourage overreliance [30], thereby impeding sustained learning. As such, Borchers et al. (2025) [4] concluded that LLM tutoring in its current form is “unlikely to produce learning benefits rivaling known-to-be-effective ITS tutoring.”  \nPrior research also shows students often attempt to “game” learning systems by prioritizing task completion and performance metrics over deeper conceptual understanding [7] . Such behaviors can create the appearance of mastery without corresponding gains in conceptual knowledge, diverging from learning-theoretic aims that emphasize sense-making and self-regulation. These concerns highlight the need for pedagogical agents that do more than optimize task performance, motivating a growing call within the AIED community for LLM-based agents to support students while remaining grounded in learning theory [32,10] .  \nSuch systems must be personalized to students’ diverse needs while generating feedback interpretable to stakeholders, who may be wary of adopting systems whose reasoning is opaque [13] . The AI community has recently embraced agentic AI, where agents autonomously plan actions, reason over data, and execute tasks to achieve defined goals, often within multi-agent architectures [33] . These systems are well-suited to educational contexts, as agent reasoning can support adaptation an","cbCaibDa6FJ7HJNa","https://ap.wps.com/l/cbCaibDa6FJ7HJNa","pdf",1993447,1,15,"English","en",105,"# Introduction\n## Motivation and Research Need\n## Evidence-Decision-Feedback (EDF) Framework\n## Copa and Classroom Study\n# Background\n## OELE and Learner Modeling","[{\"question\":\"What does the Copa instantiation and classroom evaluation show?\",\"answer\":\"In an authentic high school classroom study, EDF-guided interactions provided feedback aligned to student understanding, enabled scaffold fading, and produced interpretable, evidence-grounded explanations without increasing learner overreliance.\"}]",1784204075,38,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":78,"head_meta":80,"extra_data":82,"updated_unix":27},"evidence-decision-feedback-theory-driven-adaptive-scaffolding-for-llm-agents","",{"@graph":35,"@context":77},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/evidence-decision-feedback-theory-driven-adaptive-scaffolding-for-llm-agents/85508/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71],{"name":72,"@type":73,"acceptedAnswer":74},"What does the Copa instantiation and classroom evaluation show?","Question",{"text":75,"@type":76},"In an authentic high school classroom study, EDF-guided interactions provided feedback aligned to student understanding, enabled scaffold fading, and produced interpretable, evidence-grounded explanations without increasing learner overreliance.","Answer","https://schema.org",{"og:url":51,"og:type":79,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":81,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":84},[85,89,93,97,102,107,112,115,120,123,127],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":45,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":103,"doc_module":4,"doc_module_name":45,"category_name":104,"show_sort_weight":105,"slug":106},6,"Technology",50,"technology",{"id":108,"doc_module":4,"doc_module_name":45,"category_name":109,"show_sort_weight":110,"slug":111},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":113,"slug":114},30,"research-report",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":45,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":98,"slug":130},19,"General","general"]