User-Responsive Solutions for Cognitive Load Reduction in CAD Platforms

Abstract

Amid the evolution of cloud-based Computer-Aided Design (CAD) platforms, traditional educational approaches fail to address the diversity of cognitive obstacles that users across expertise levels and learning behaviors fac. This thesis investigates whether behavior-adaptive CAD tools can reduce friction as hypothesized by Cognitive Load Theory (CLT) while enhancing skill development in modern engineering environments.A two-phase mixed-methods approach was employed that combined large-scale behavioral persona identification with controlled user testing. TF-IDF and PCA on the results of an MIT-wide survey identified four distinct behavioral archetypes corresponding to unique tool usage patterns and learning preferences independent of technical proficiency. A/B testing of three behavior-adaptive custom tools which addressed workflow optimization, parametric knowledge retention, and contextual-aware passive modelling guidance was done with novice and advanced users. Command logging captured behavioral features and analysis discovered significant cognitive load reduction, improved workflow efficiency, and higher-retained skill development. NLP of post-session survey responses revealed deeper conceptual engagement. From these results, a three-stage model progressing from friction reduction through behavioral analytics to continuous personalization optimization was developed to inform business applications. The findings demonstrate that effective CAD education requires addressing individual behavioral patterns rather than traditionally uniform skill-based approaches. Behavior-adaptive tools enhance learning pathways and workflows by preserving user agency over creative and parametric decisions during modelling while reducing cognitive friction.

Keywords: Computer-Aided Design (CAD), Cognitive Load Theory (CLT), Behavioral Analytics, Behavior-Adaptive Learning, Engineering Education