Kashika Khurana, Ally Liew (2026) — arXiv cs.HC / cs.CY 📄 Full text (arXiv) 📄 PDF
This study investigates how different modes of AI interaction affect cognitive engagement and learning outcomes in high school students. Using a within-subjects design with 24 students, the researchers compared three conditions: Auto mode (AI solves problems independently), Interactive mode (student-AI collaboration with scaffolding), and Manual mode (no AI assistance). The Interactive mode produced the highest cognitive engagement and task accuracy, while the Auto mode led to reduced engagement and potential over-reliance.
The use of electroencephalography (EEG) provides a neurophysiological dimension to understanding AI's impact on learning. Though EEG results did not reach statistical significance, descriptive patterns suggested differences in neural activity across the three AI interaction modes. This connects to broader debates about student-AI interaction design in k-12 classrooms.
The finding that full automation reduces cognitive engagement echoes the over-reliance concerns documented in prior research on AI tutoring systems. The study's framework for categorizing AI interaction modes (Auto, Interactive, Manual) provides a replicable methodology for future human-AI interaction research in educational settings.
Related Pages
- k-12 — K-12 education contexts
- student-experience — Student experiences with AI in education
- over-reliance — Risks of over-reliance on AI systems
- engagement-metrics — Measuring student engagement in AI-mediated learning
- active-learning — Active learning pedagogies
- ai-literacy — AI literacy concepts and programs
- learning-behavior-background-advantage-ai-ed — Behavioral mechanisms of AI learning (2026-07-14)
Citation
APA: Kashika Khurana, Ally Liew (2026). An exploratory behavioral and electroencephalographic study of artificial intelligence-assisted learning modes in high school students. arXiv:2606.26579. arXiv cs.HC / cs.CY.