Evidence-Decision-Feedback: Theory-Driven Adaptive Scaffolding for LLM Agents
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910090868555776 |
|---|---|
| author | Cohn, Clayton Guo, Siyuan Rayala, Surya Wang, Hanchen David Mohammed, Naveeduddin Timalsina, Umesh Jain, Shruti Eeds, Angela Deweese, Menton Popp, Pamela J. Osborn Stanton, Rebekah Walker, Shakeera Ma, Meiyi Biswas, Gautam |
| author_facet | Cohn, Clayton Guo, Siyuan Rayala, Surya Wang, Hanchen David Mohammed, Naveeduddin Timalsina, Umesh Jain, Shruti Eeds, Angela Deweese, Menton Popp, Pamela J. Osborn Stanton, Rebekah Walker, Shakeera Ma, Meiyi Biswas, Gautam |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_01415 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Evidence-Decision-Feedback: Theory-Driven Adaptive Scaffolding for LLM Agents Cohn, Clayton Guo, Siyuan Rayala, Surya Wang, Hanchen David Mohammed, Naveeduddin Timalsina, Umesh Jain, Shruti Eeds, Angela Deweese, Menton Popp, Pamela J. Osborn Stanton, Rebekah Walker, Shakeera Ma, Meiyi Biswas, Gautam Multiagent Systems 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. |
| title | Evidence-Decision-Feedback: Theory-Driven Adaptive Scaffolding for LLM Agents |
| topic | Multiagent Systems |
| url | https://arxiv.org/abs/2602.01415 |