Helping children become qualitative modelers

Early science education is essentially qualitative. Children must learn causal theories: what kinds of things happen, when they happen, and what their consequences are. Although modeling is a key skill in scientific reasoning, traditional modeling techniques do not express causal theories explicitly, and rely on higher mathematics (such as algebra) that children do not learn until later. We believe that qualitative modeling techniques developed in AI can become an important tool for early science education. This paper describes a notational system we are building that enables middle-school students to create and use qualitative models. We use a restricted concept map language based on qualitative process theory to provide a student-friendly visual modeling language. Interactive qualitative reasoning techniques and analogical coaching will be used to help students understand and refine their models. The basic design of the system will be outlined, as well as experiences with its use with students in the Chicago Public School system.