The purpose of this proposal is to explore the extent to which timely emotional, cognitive, and metacognitive interventions in tutoring software will have positive effects on students' emotions, attitudes, and achievements in mathematics. The intelligent tutor, Wayang Outpost, a high school mathematics tutoring system, is being enhanced to leverage automatic detection of emotions to guide cognitive, metacognitive, and affective forms of learning support.

The PIs are conducting a set of experiments to understand the interplay of observed emotional states, emotion assessments, student behavior within tutors, and student achievement. In particular, the experiments are testing the effects of the tutoring system when it assesses the emotions of a student and then responds with instructional support appropriate to that student's affect and content knowledge.

In this project, an interdisciplinary team of researchers in learning technologies and mathematics education are working together to investigate issues related to motivation in learning mathematics. They are taking the results of lab-based studies into classrooms. The novel technology and approaches developed in the lab were tested with a small population of learners; in their classroom-based investigations, they are testing feasibility of the approach with a more diverse population and refining the technology for use in a broad range of classroom learning environments. This translational research project will not only make significant contributions to the field of learning technologies, but will also contribute to our understanding of issues related to motivation in mathematics learning.

Project Report

This project focused on two research agendas: 1) validation of models for automatic detection of student emotion and 2) responding to several emotional states within an intelligent tutoring system for mathematics. One of the few emotional theories grounded in education is the control-value theory of emotions in education by Pekrun and colleagues (2006), which describes several emotions related to achievement in learning situations, and instruments for measurement. Achievement emotions may be classified according to their valence (positive/negative), and their arousal (activating/deactivating), and their focus (activity/outcome). Positive activating achievement emotions (enjoyment of the activity, hopeful/confident to achieve the outcome) exert a positive impact on achievement, while negative deactivating emotions (e.g. boredom, hopelessness) have a negative impact. We identified correlation strengths for several pairs of constructs. We used Pekrun's enumeration of student emotions corresponding to the AEQ-M (Achievement Emotions Questionnaire for math) and found that they were highly correlated to our measures of emotion in general (e.g., Arroyo et al, 2009). From the results we have also learned that activating emotions (e.g., math-liking) are highly correlated among each other, but not with deactivating emotions (e.g., boredom, anxiety). We have achieved several accomplishments in modeling and understanding how to address student emotion in smart digital learning environments and tutoring systems. The first research question was whether emotions truly represent our own student emotion reports. We verified that a tight relationship exists between our constructs and the more traditional ones in the literature. We conclude that our emotion self-reports capture similar achievement emotion constructs to those of Pekrun and colleagues, and similar math value and self-concept constructs as Eccles and Wigfield. The primary level of emotional repair in the MathSpring system is support in the form of meta-cognitive messages (audio and/or in a text bubble on screen) that the learning companions provide in appropriate moments in order to encourage students to engage in meta-cognitive behaviors and adopt attitudes that have been shown to be beneficial for learning. For instance, two meta-cognitive messages advanced include the following phrasing: "Hmm, I think you may be looking through the hints quickly for the answer. (Am I right?) If so, here's a tip: The goal is not so much to get the answer right as it is to learn new ideas and new ways to solve problems, thanks to the hints." and "Even this computer is impressed with your dedication and your hard work. Did you know that your work is causing the nerves to get stronger in your brain?" These messages are tailored to students' progress, e.g., generated when the system detects students rushing through hints instead of trying to learn from them. Students are drawn to the level of control provided, and are interested in what the system says about their progress and how they are doing. Another emotional repair tool tested and proven effective was the Student Progress page, which provided a way for students to keep track of their progress and, particularly, to be reminded of their progress by the tutor when emotions such as boredom were being detected. Student progress pages, which attempt to meta-cognitively support students in a self-regulation process and that are offered at precise moments of student boredom or lack of excitement, revealed improved levels of interest and decreased boredom compared to control groups for which the progress page was available but not offered at those key moments. Another repair mechanism tested was the program feature of being able to share problems with neighboring students, encouraging student pair and group work. Our results were in support of other research that has found that collaboration between friends fosters greater development of scientific reasoning. The ultimate research question being addressed was whether or not it is useful and beneficial to students’ affective states to provide interventions based on student emotion. We conducted studies involving ~40 students, refined analysis of our entire data set and enhanced our student models and assessment of the impact of affective interventions. In these efforts, we showed that the repair efforts of the system based on emotional reports and behavior prediction were effective in improving student performance. At the same time, we further advanced our design of meta-cognitive support in the form of progress pages, which provide many of the benefits predicted for open student models by encouraging students to reflect on their progress, supporting learners to take greater control and responsibility over their learning, and increasing learner trust in the environment.

Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Arizona State University
United States
Zip Code