WPI Receives $1.5 Million NSF Award to Develop Intelligent Tutoring System for Improved Science Education
WORCESTER, Mass. – A new computerized tutoring system being developed by researchers at Worcester Polytechnic Institute (WPI) aims to help middle school students develop a deep understanding of scientific concepts by teaching them to “think like scientists” (for example, by conducting experiments in the way scientists do). The project is aligned with the content and skills standards set forth by the Massachusetts Curricular Frameworks.
The development of the tutoring system is being funded by a five-year, $1.5 million award from the National Science Foundation to principal investigator Janice Gobert, research associate professor in computer science and social sciences, and a team that includes WPI computer science professors Neil Heffernan and Carolina Ruiz, and mathematics professor Ryung Kim. The award will support the development of a system that uses computer simulations of scientific experiments (to be developed by the WPI team) and a sophisticated logging system for systematically tracking how students tackle experiments to both guide students’ learning and assess their progress.
“Simulated experiments are an excellent and practical way for students to conduct scientific inquiry,” Gobert says. “Through this approach, a student can pose questions, develop a plan, conduct a virtual ‘experiment,’ run the simulation to collect data, analyze the data, and then communicate what they’ve learned in the form of a scientifically warranted explanation. By using these simulations and the technological infrastructure underlying them, we can seamlessly integrate assessment with instruction and provide feedback automatically to teachers about their students.”
The work will build upon previous research by Gobert and Heffernan. Specifically, ASSISTments, a middle school mathematics tutoring and assessment system developed by Heffernan over the past four years, will be adapted to guide, tutor, and assess science skills. The project will also draw from research by Gobert that focused on the design of scientific simulations and methods for tracking the decisions students make while they interact with them.
Gobert, a cognitive scientist and former senior research scientist at the Concord Consortium, a nonprofit educational research and development organization based in Concord, Mass., will bring to the new system her extensive experience in developing science learning systems that help students develop inquiry skills as they engaging in realistic computer simulations of physical, chemical, geological, or biological phenomena. She will also develop and test hypotheses about how students with different levels of prior knowledge and skills levels will interact with the simulations, and conduct research about the various pathways students take to skill development.
National science education standards place considerable emphasis on teaching students the skills of inquiry, such as the systematic process of asking questions and posing and testing hypotheses, Gobert says. But in practice, she notes, inquiry skills are difficult to teach and to assess. In addition, the standardized tests typically used by states to assess student learning emphasize knowledge of vocabulary, facts, and formulas, which place a premium on rote learning, rather than learning through inquiry.
According to Gobert, learning through inquiry leads to a deeper understanding of science and to an ability to apply scientific knowledge in everyday life—what is often called scientific literacy. “We need ways to educate the populace so that once they leave high school, they can use what they’ve learned to make good decisions that affect their everyday life,” she says.
She says the system will also benefit students who may be inclined to pursue careers in science and engineering. “They can use the skills and deep knowledge fostered by this tutoring system to drive themselves forward, whereas students taught in the rote method will find that once they are out of class, their knowledge is not transferable to other science topics and they may not be able to apply it.”
The new tutoring system will incorporate methodology and algorithms developed for the math ASSISTments project (while also adding new approaches needed to teach science). These include “scaffolding,” in which the intelligent tutor guides students by breaking down a task into steps, providing tailored help to students who need it. It will also use the detailed reports ASSISTments generates on students’ knowledge and skills, which help teachers adjust class content and work more productively with students who need help, alert parents when their children need assistance, and enable principals and school systems to assess student learning in real time.
As was done during the development of the math ASSISTments system, the researchers will work closely with teachers and students in the Worcester public school system to develop and pilot test the system, which will be implemented first in middle school science classes in the city before being rolled out more widely. Gobert says the city, which is ethnically and economically diverse, will provide an excellent testbed for the system, which needs to be able to address the needs of a broad range of learners. The system will be aligned with the Massachusetts science curriculum frameworks to ensure that students using the system will understand the concepts included in the state’s standard assessment test (the MCAS, or Massachusetts Comprehensive Assessment System).
“And we mean understand deeply,” Gobert says. “If you show students the rich underlying causal structure of science phenomena, and give them the skills to learn how those systems work, you will teach them a great deal more than just content; you will teach them habits of mind and skills of science inquiry that they can apply to the next science topic they encounter, as well as to decision-making in everyday life.”