Design adaptive teachable agents that promote sustained knowledge-building responses so learners of any prior knowledge can benefit¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The article's design recommendation is to build teachable agents that pose persistent, contextually relevant follow-up questions adapted to the tutor's responses, "scaffolding tutors' elaborations and prompting reflection". Because %KBR predicted post-test scores and was not predicted by prior knowledge, such systems "may help all learners––regardless of prior knowledge––improve their conceptual and procedural knowledge." The strategy targets the knowledge-telling default by mirroring the tutor's confusions and eliciting elaboration.

Design Implications¶

Context¶

Requirements¶

• The agent must sustain productive dialog with persistent, contextually relevant follow-up questions that adapt to the tutor's responses over time
• Follow-up questions should target conceptual elements of an expected elaborated response that the tutor has not yet conveyed

Constraints¶

• The observed relationships are correlational, not causal; the authors plan a randomized controlled trial before stronger design claims

Target Learners¶

• middle school students with low prior knowledge in the taught domain
• middle school students generally (grades 6–8)

Target Learning Goals¶

• conceptual knowledge
• procedural skill
• knowledge-building over knowledge-telling

Related Strategies¶

Examples¶

-

Key Sources¶

• Fathima Ameen, Tasmia Shahriar, Aditi Mallavarapu, Shiyan Jiang, Noboru Matsuda. (2026). Beyond prior knowledge: the predictive role of knowledge-building in tutor learning. Behaviormetrika. https://doi.org/10.1007/s41237-026-00294-9