[論文レビュー] Scaffolding Metacognition with GenAI: Exploring Design Opportunities to Support Task Management for University Students with ADHD

Zihao Zhu, Junnan Yu+1 more|arXiv (Cornell University)|Feb 10, 2026

Attention Deficit Hyperactivity Disorder被引用数 0

ひとこと要約

研究は ADHD を抱える大学生とエキスパートインタビューを用いた共創を通じ、タスク管理におけるメタ認知の GenAI デザイン機会を特定し、スキャフォールディング、反省、情動調整に焦点を当てた三つのデザイン指向を導出した。

ABSTRACT

For university students transitioning to an independent and flexible lifestyle, having ADHD poses multiple challenges to their academic task management, which are closely tied to their metacognitive struggles--difficulties in awareness and regulation of one's own thinking processes. The recently surged Generative AI shows promise to mitigate these gaps with its advanced information understanding and generation capabilities. As an exploratory step, we conducted co-design sessions with 20 university students diagnosed with ADHD, followed by interviews with five experts specialized in ADHD intervention. Adopting a metacognitive lens, we examined participants' ideas on GenAI-based task management support and experts' assessments, which led to three design directions: providing cognitive scaffolding to enhance task and self-awareness, promoting reflective task execution for building metacognitive abilities, and facilitating emotional regulation to sustain task engagement. Drawing on these findings, we discuss opportunities for GenAI to support the metacognitive needs of neurodivergent populations, offering future directions for both research and practice.

研究の動機と目的

• Understand how ADHD university students experience metacognition in academic task management.
• Explore GenAI design ideas through co-design sessions with ADHD students to address metacognitive gaps.
• Validate and refine student ideas with ADHD intervention experts to assess feasibility and risks.
• Synthesize findings into design directions that scaffold cognition and support engagement for neurodivergent learners.
• Propose a practical research agenda for GenAI-enabled metacognition support in education.

提案手法

• Conducted 20 one-on-one co-design sessions with university students diagnosed with ADHD to generate GenAI-based task management ideas.
• Performed in-depth interviews with five ADHD intervention experts to evaluate feasibility and clinical relevance of student ideas.
• Applied a bottom-up thematic analysis (Braun & Clarke) to transcripts and design artifacts to identify metacognitive challenges and design themes.
• Coded 764 initial concepts from co-design data and 245 codes from expert interviews, then integrated them to derive themes.
• Identified four student metacognitive challenges and three design directions for GenAI support.
• Ensured an open-ended co-design approach to avoid framing ideas around metacognition.

実験結果

リサーチクエスチョン

• RQ1 How can GenAI be leveraged to support the metacognitive processes (knowledge, monitoring, control) of university students with ADHD in task management?
• RQ2 What design opportunities do students with ADHD identify for GenAI to assist with organizing, executing, and adjusting academic tasks?
• RQ3 What are the perceived risks of relying on GenAI for metacognitive tasks, and how can design mitigate them?
• RQ4 What concrete design directions emerge when combining student-led ideas with expert feedback on feasibility and clinical relevance?

主な発見

• Three design directions emerged: Metacognitive scaffolding to enhance task and self-awareness, Reflective task execution to develop metacognitive abilities, and Emotional regulation to sustain task engagement.
• Participants reported four metacognitive challenges: lack of awareness, barriers to task initiation, difficulty controlling attention, and emotion regulation struggles.
• Experts highlighted risks such as cognition outsourcing and unhealthy dependence, informing design with cautions against over-automation.
• GenAI can help integrate fragmented task information, calibrate time allocation, and decompose tasks, while emphasizing reflection over full automation.
• There is potential for GenAI to support neurodivergent students, but designs should promote self-regulation and emotional growth rather than dependence.