Much research in education now points to the importance of both cognitive and motivational processes in determining learning outcomes. Yet we do not yet understand exactly how these processes interact within the individual learner over time to produce specific learning outcomes, and how different forms of instruction and tutoring can influence the interaction of cognition and motivation to facilitate learning. Advanced learning technologies such as intelligent tutoring systems typically focus on cognitive issues, and do not yet address motivational issues. Thus there is a major gap between educational practice, as supported by learning technologies, and research in cognitive and motivational aspects of human learning. The goal of the proposed project is to investigate the relations of learner engagement, cognitive processes, and scaffolding strategies in producing specific learning outcomes in mathematics. It brings together an interdisciplinary team of specialists in psychology, education, and information technology to pursue this work. The project will employ educational software designed to enable us to address and measure cognitive and motivational factors in systematic ways. It will significantly extend previous work on intelligent tutoring systems and computer-based learning assistants to assess in real time the attention and motivation of individual learners, infer their level of engagement, and estimate their domain knowledge. The research results will be obtained in realistic learning conditions: public high school classrooms in Los Angeles, with student populations reflecting the ethnic, linguistic and economic diversity of urban California. The software will make instructional decisions based on decision rules used by expert human tutors who consider motivational and cognitive goals to optimize learning outcomes. Instructional strategies will also be instantiated in the form of a responsive animated virtual character, to learn if the presence of a human-like pedagogical agent influences the learner.s engagement as well as learning outcomes in mathematics. Classroom teachers will be able to customize the instruction through innovative tools to view continually updated assessments and refine pedagogical decisions made by the software. Rich learner data sets will be automatically collected as students work on mathematics problems, and will provide an integrated evaluation linked to specific learning outcomes, including word problem solving and transfer to novel, challenging problems embedded in real-world STEM contexts. To accomplish this goal, the researchers will build on the results of two projects developed with prior NSF support: an empirically validated focus of attention tracking system, and an intelligent tutoring system (ITS) designed specifically to enhance math skills in students who have been traditionally under-represented in STEM fields of study. The results of the project will have broad societal impact through providing individualized, effective web-based instruction in STEM-related mathematics that will be freely accessible to students in any high school with access to the Internet. We anticipate that the impact will be greatest among students who currently have motivational difficulties with mathematics, including women and underrepresented groups. The intellectual merit of the project will be a deeper understanding of the very nature of learning processes through the ability to model in detail and in real time the interaction of cognition and engagement in the individual learner, and to observe changes in the learner.s knowledge as a function of instructional history.
