Emotion features improve tutor performance only when grounded in education-theory-based pedagogical strategies; emotion information alone misguides a smaller model¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? In an ablation on Qwen2.5-7B-Socratic-LM, prompting with pedagogical strategies derived from detected emotion yields gains of 9 points (simple prompt) and 2 points (complex prompt), while emotion-only prompting hurts performance. → Debanjana Kar 2025

Evidence¶

Debanjana Kar 2025¶

Debanjana Kar, Leopold Böss, Dacia Braca, Sebastian Dennerlein, Nina Hubig, Philipp Wintersberger, and Yufang Hou. (2025). MathBuddy: A Multimodal System for Affective Math Tutoring. https://arxiv.org/abs/2508.19993 doi:10.48550/arxiv.2508.19993

q2 · i? · design · r2

Ablation study (Table 3) on the Qwen2.5-7B-Socratic-LM model across simple and complex prompts, comparing baseline, emotion-only, and emotion-plus-education-theory configurations. The authors note that "providing only the emotional information misguides the model".

"prompting the model with relevant pedagogical strategies based on the detected emotion of the student results in a steep performance gain (by 9 points and 2 points through simple and complex prompts respectively)"

Discussion¶

Related Claims¶

• Adding student-emotion information to LLM tutor prompting produces more desirable pedagogical qualities across most evaluated dimensions, with larger models benefiting most — related
• MathBuddy's multimodal emotion aggregation matched participant-reviewed emotions at 60% overall accuracy, with sharply reduced recall for the Neutral class — related
• Model scale does not consistently improve pedagogical safety: within the Qwen2.5 family the 72B model beats the 7B on only 17 of 33 subject-dimension pairs — related