The Neural Bridge: From Haptic Cords to Python Code — A Multi-Abstraction Instructional Sequence for Teaching Neural Networks in K-12

Abstract

Teaching neural networks at the secondary level is frequently impeded by the conceptual gap between everyday intuition and implementation-level code. This paper presents The Neural Bridge, a four-lesson instructional sequence for K-12 students that traverses three levels of abstraction: an embodied unplugged simulation using physical cords and carabiners (''7 Little Neurons''), a manual computation phase in Google Sheets, and a professional implementation in Google Colab. The sequence integrates Productive Failure [2] and the PRIMM framework [9] as its pedagogical scaffolding, and is grounded in Long and Magerko's [8] seventeen AI literacy competencies. Delivered to 27 students across Grades 10–12 in December 2025, a pre-/post-study showed that correct identification of AI as a data-processing system increased from 63% to 82%. Notably, students' self-assessed AI understanding decreased (M=2.70 to M=2.00), a positive outcome interpreted as metacognitive growth consistent with the Dunning-Kruger effect. Qualitative feedback identified the unplugged activity as the most impactful element.