Shallow strategy development in a teachable agent environment designed to support self-regulated learning

► We examine feature use in a teachable-agent system - Betty's Brain. ► Middle school students learn about science processes by constructing concept maps. ► Learning gains are associated with use of SRL-supportive features of the system. ► Promising strategy patterns weaken over time due to shallow strategy development. ► Students focus on a limited set of strategies at the cost of other beneficial strategies.