Beyond the Grey Area: Exploring the Effectiveness of Scaffolding as a Learning Measure

In this paper, we aim to explore students’ help-seeking patterns, and learning performance using a computational model of the Zone of Proximal Development (ZPD). We used student traces in four courses supported by Intelligent Tutoring Systems (ITSs) to assess whether a student may be in their ZPD. Then, we analyzed students’ help-seeking behavior and their performance after receiving scaffolding to investigate if and under which circumstances our assessment may act as a proxy of the ZPD. Results suggest that the computational model may offer an acceptable approximation of the student’s ZPD. However, several factors, such as the difficulty of the learning task, the student’s age, and their attitude regarding hints, may affect modeling students’ cognitive states. We discuss the implications of this research on the design of computational student models for providing personalized and adaptive feedback and scaffolding. We envision that this work contributes toward bridging the gap between theory and practice for AI-assisted tutoring and scaffolding, and it provides insights regarding evaluation techniques of theory-driven, computational approaches.