Preventing Cognitive Offloading in Generative AI: Scaffolding Complex Design Reasoning for Transfer Beyond AI Assistance

This repository provides the de-identified replication dataset, instruments, and the frozen prompt-scaffolding package for the study: "Preventing Cognitive Offloading in Generative AI: Scaffolding Complex Design Reasoning for Transfer Beyond AI Assistance" Authors:** Rui Chen, Junqing Hao, Wei Zong* - Affiliation: School of Architecture and Design, China University of Mining and Technology - Package Contents 1. De-Identified Trial Dataset:** Baseline demographics, blinded BARS scores for Task A & Task B, inter-rater reliability (ICC), and questionnaire scales (N = 64). 2. Frozen LLM Skill Package (`v1.1-frozen`): Executable prompt specifications, progression gates (0–12), and the H0–H6 help hierarchy with archived SHA-256 digests. 3. Research Instruments & Rubrics: English versions of Task A/B briefs, parallel FBS knowledge tests (Forms X & Y), and scoring rubrics. All data are de-identified for peer review, scientific auditability, and replication.