Parent-child and AI-child interaction in facilitating children’s scientific reasoning

Everyday interaction offers rich experiences for children’s STEM learning. Children grasp fundamental chemistry while cooking with parents, mathematics during grocery shopping, physics on the playground, and biology by simply exploring the outdoors. But learning information about science is not the same as learning the process of scientific reasoning. To make novel innovations, and to engage in creative problem-solving, children must develop and use their scientific reasoning capacities. This project examines how children's emerging scientific reasoning is facilitated by parent-child interaction, describing how parents support children's emerging scientific reasoning. It also examines whether children's scientific reasoning can be facilitated similarly by interaction with AI (artificial intelligence) agents as it is through parent-child interaction, with implications for how educators structure science learning and what role AI tools may appropriately play in that process. In two lines of studies, 4- to 7-year-olds will be introduced to sets of scientific reasoning problems previously demonstrated to be age-appropriate. Prior work has also shown that performance on these problems relates to children’s performance on standardized measures of scientific reasoning. In the first line of studies, children will be administered these problems on their own or with a parent under different experimental conditions focused upon what parents know about the problem, and how they are instructed to help their child. The project will measure children's ability to solve the problems, the way in which they reason about similar problems on their own, and how parents and children interact during the problem-solving process to describe what parents do to facilitate children's reasoning. In the second line of studies, children will be given similar problems as they interact with an AI agent (a robot, controlled by the experimenter), who will guide children through the problem-solving measure in a similar manner to the parent-child interaction. The second line of research uses findings from the first line to investigate whether children learn similarly from parent-child and AI-child interaction, with implications for how AI tools might be thoughtfully integrated into early science learning without displacing the unique benefits of parent-child engagement. This project was funded by the STEM K-12 program, which supports fundamental, applied, and translational research on STEM learning in both formal K-12 education and informal learning environments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.