| dc.description.abstract | Self-regulated learning (SRL), which involves the regulation of cognitive, affective, metacognitive, and motivational processes, is an important construct for analyzing effective student learning in open-ended learning environments (OELEs). Novice K-12 students working on complex learning tasks in OELEs often engage in sub-optimal use of SRL strategies and, therefore, encounter learning and problem-solving difficulties. These students are likely to benefit from in-the-moment scaffolding, but designing adaptive scaffolds to support SRL strategies in OELEs is a challenging task.
This dissertation presents a design, development, and evaluation framework for adaptive scaffolding to support the application of task-oriented SRL strategies in Betty’s Brain, an OELE where K-12 students learn about scientific processes by building causal models to teach a virtual agent named Betty. Our scaffold design and development process was based on tracking students’ activity sequences to detect their use of strategies related to their cognitive and metacognitive processes. Ineffective strategy use triggered the adaptive feedback, implemented as conversation trees and delivered by the virtual mentor agent present in the system. The conversations were contextualized to the learner’s current task and recent activities, and supported the development of strategies to resolve learner difficulties. Sensor-free interaction-based affect detectors were also employed to track students’ affective states.
A classroom study was conducted to evaluate the impact of adaptive scaffolding on middle school students. Exploratory data analysis revealed four student groups with differing behavioral profiles. More targeted temporal analyses showed how each group received, responded to, and used the strategy scaffolds in their learning process. The results suggest how certain adaptive scaffolds were more effective in helping groups of students become more strategic in their learning and problem-solving behaviors. We discuss the implications of our findings on the future design of adaptive scaffolding to support cognitive, metacognitive, and affective aspects of SRL in OELEs. | |
