Primary Participants: Alok Baikadi (Computer Science), Kirby Culbertson (Art & Design), Lori Dolezal (Education), Julius Goth (Computer Science), EunYoung Ha (Computer Science), Sarah Hegler (Art & Design), Ashley Hoffman (Psychology), Seung Lee (Computer Science), Samuel Leeman-Munk (Computer Science), Eleni Lobene (Psychology), Erin Lyjak (Education), Karoon McDowell (Art & Design), Stacie McGowan (Art & Design), Angela Meluso (Psychology), Bradford Mott (Computer Science), Adam Osgood (Art & Design), Justin Phillips (Art & Design), Jonathan Rowe (Computer Science), Marc Russo (Art & Design), Robert Taylor (Computer Science), Jen Sabourin (Computer Science), Lucy Shores (Education), Andy Smith (Computer Science), Kim Turner (Education), Donnie Wrights (Art & Design), Meixun Zheng (Education)

Sponsor: National Science Foundation – Discovery Research K-12 Program (2008-2012)

Objectives: The Crystal Island project addresses the challenge of assuring that all students have the opportunity to learn significant science content by investigating the following research question: How can intelligent game-based environments promote problem solving and engagement in science learning for upper elementary students? The project investigates problem solving, engagement, and science learning by targeting the following two objectives:

1. Design a suite of intelligent game-based learning environment technologies for elementary science education. To promote effective science learning, we are creating intelligent game-based learning environment technologies that leverage the rich interactive 3D game environments provided by commercial game engines and the inferential capabilities of intelligent tutoring systems. Building on our experience in these two areas, we are creating an engaging intelligent game-based learning environment for 5th grade science.

2. Implement an empirically-based research program to provide a comprehensive account of elementary students’ problem-solving processes and engagement with STEM content as they interact with intelligent game-based learning environments. To understand the cognitive mechanisms by which learning occurs, we are taking a mixed method approach to investigating science learning with an intelligent game-based learning environment for 5th grade science. These studies are investigating the central issues of problem solving (strategy use, divergent thinking, and collaboration), and engagement (motivation, situational interest, presence) with respect to achievement as measured by both science content knowledge and transfer. With diverse student populations drawn from both urban and rural settings, the studies will determine precisely which technologies and conditions contribute most effectively to learning processes and outcomes.

Primary Participants: Alok Baikadi (Computer Science), EunYoung Ha (Computer Science), Joe Grafsgaard (Computer Science), Megan Hardy (Psychology), Chris Mitchell (Computer Science), Bradford Mott (Computer Science), Rob Phillips (Computer Science), Mladen Vouk (Computer Science)

Sponsor: National Science Foundation – Research and Evaluation on Education in Science and Engineering Program (2010-2014)

Objectives: Providing dialogue systems with the ability to engage users in rich natural language dialogue has been a long-term goal of the computational linguistics community. Tutorial dialogue systems, which are designed to support students completing a learning task, are an increasingly active area of research. Though effective, these systems have not yet attained learning gains on par with those observed in expert human tutoring. Additionally, handcrafted tutorial dialogue systems require significant development time. Even with the emergence of authoring tools for rapid development, the current generation of tutorial dialogue systems requires significant manual authoring.

In response to these challenges, the JavaTutor project investigates human-human natural language tutorial dialogue as a model for human-computer tutorial dialogue. With a curricular focus of first-year post-secondary computer science education and a task focus of problem-solving dialogues, we collect corpora of human-human tutorial dialogues, annotate them with rich dialogue act tags, and then use machine learning techniques to automatically acquire the structure of effective tutorial dialogue.

The JavaTutor project is concerned with 1) designing rich dialogue act coding schemes for coding cognitive and affective dimensions of task-oriented tutorial dialogue interactions, and 2) learning hidden Markov models to discover the structure of task-oriented tutorial dialogue. Our studies to date have found that learner characteristics influence tutorial dialogue structure in significant ways, even when these characteristics are not revealed to the tutors, and that human tutors naturally attempt to strike a balance between cognitive and motivational scaffolding: in response to a mistake, positive cognitive feedback may better facilitate student learning, and may provide equally beneficial affective outcomes, compared with overt encouragement. The long-term objective of this line of investigation is to create computational models of tutorial strategies, design intelligent tutoring systems that utilize these models, and study the differential impact of alternative strategies on a broad scale.

Primary Participants: Alok Baikadi (Computer Science), John Bedward (Education), Courtney Behrle (Education), Elysa Corin (Education), Kirby Culbertson (Art & Design), Julius Goth (Computer Science), Sarah Hegler (Art & Design), Chris Kelley (Psychology), Jennifer London (Psychology), Seung Lee (Computer Science), Samuel Leeman-Munk (Computer Science), Eleni Lobene (Psychology), Chris Mitchell (Computer Science), Adam Osgood (Art & Design), Lindsay Patterson (Education), Justin Phillips (Art & Design), Cathy Ringstaff (Evaluation), Jonathan Rowe (Computer Science), Marc Russo (Art & Design), Wayne Sheffield (Education), Andy Smith (Computer Science), Robert Taylor (Computer Science), Michael Timms (Evaluation)

Sponsor: National Science Foundation – Discovery Research K-12 Program (2010-2014)

Objectives: Central to elementary science education is the development of conceptual understanding through the modeling of scientific phenomena. The process of inquiry, focused on observed and described phenomena, requires students be equipped with epistemic tools that allow them to construct extended discursive threads—synthesizing concrete observations and abstract concepts through multiple modes of representation. Students need scaffolded support in this process from sophisticated and powerful modeling tools applied along a learning progression of scientific understanding. The objective of the Leonardo project is to develop and evaluate an intelligent cyberlearning system for interactive scientific modeling in elementary science education. Students in Grades 4 and 5 will use Leonardo’s intelligent virtual science notebooks to create and experiment with interactive models of physical phenomena. The project has three major thrusts:

1. Develop CyberPads, intelligent virtual science notebooks with sketch-based multimodal interfaces. To promote effective science learning, we will design and develop CyberPads, intelligent virtual science notebooks. Leonardo’s CyberPads will be artificial intelligence-based software systems that enable students to create graphical representations to model physical phenomena. Students’ models will “come alive” as interactive media artifacts that combine animation, sound, and narration. With a curricular focus on the physical and earth sciences, CyberPads will support multimodal interactive scientific modeling for four curricular units: forces and motion, magnetism and electricity, landforms, and weather and climate.

2. Develop PadMates, intelligent virtual tutors to support interactive scientific modeling. Intelligent virtual tutors are “embodied” artificial intelligence-driven characters that interact with students to provide engaging, personalized tutorial support. We will develop Leonardo’s PadMates, who will provide customized advice and explanations during students’ interactive scientific modeling experiences with their CyberPads. Students will engage in rich interactions with their PadMates as they construct a deep understanding of science concepts and how to think scientifically.

3. Evaluate the synergistic impact of intelligent virtual science notebooks and intelligent virtual tutors on elementary science learning. To evaluate the impact of the Leonardo system on science learning in Grades 4 and 5, the evaluation will investigate the central issues of interactive scientific modeling (strategy use, divergent thinking, and collaboration) and engagement (motivation, situational interest) with respect to achievement as measured by both science content knowledge and transfer, and metacognitive aspects of scientific ways of thinking. Emphasizing connections to the National Science Education Standards and AAAS’s Benchmarks and building on our experience with school-based research, all studies will be conducted onsite at the project’s partner elementary schools in North Carolina, Texas, and California. With diverse student populations, the studies will determine precisely which technologies and conditions contribute most effectively to learning processes and outcomes.

Primary Participants: Kristy Boyer (Computer Science), Julius Goth (Computer Science), Joe Grafsgaard (Computer Science), EunYoung Ha (Computer Science), Kristin Hoffman (Psychology), Seung Lee (Computer Science), Sunyoung Lee (Computer Science), Eleni Lobene (Psychology), Scott McQuiggan (Computer Science), Bradford Mott (Computer Science), Jennifer Robison (Computer Science), Jonathan Rowe (Computer Science), Lucy Shores (Computer Science), Kim Turner (Psychology)

Sponsors: National Science Foundation – Research and Evaluation on Education in Science and Engineering Program (2007-2009) & Human-Centered Computing Program (2008-2011)

Objectives: Recent years have seen a growing recognition of the importance and challenges of creating learning environments that promote motivating, inquiry-based science learning. Pedagogical agents are embodied software agents that have emerged as a promising vehicle for promoting effective learning. They provide customized problem-solving experiences and advice that are precisely tailored to individual learners in specific contexts. By co-habiting a rich inquiry-based learning environment with learners, pedagogical agents can meticulously observe learners’ problem solving activities, offer situated advice, and actively support learners’ iterating through cycles of questioning, hypothesis generation, data collection, and hypothesis testing. However, inquiry-based learning also presents a significant challenge: the very “openness” of the learning environment introduces multiple sources of complexity into tutorial planning. To address the complexities associated with scaffolding inquiry-based learning, this project explores Bayesian pedagogical agents that leverage recent advances in Bayesian and decision-theoretic computational models of reasoning to promote self-regulated learning experiences that are both effective and engaging.

The project has two complementary technology and learning thrusts:

1. It will develop a full suite of Bayesian pedagogical agent technologies for inquiry-based science learning environments. To promote effective and engaging learning processes and outcomes, the research team is creating Bayesian pedagogical agents that leverage probabilistic computational models that systematically reason about the multitude of factors that bear on decision making to infer learners’ beliefs, goals, and plans, including strategy use, from their problem-solving actions. By introducing pedagogical agents into the visually engaging environments that typify high-end game platforms and embedding them in dynamically generated science narratives, we are addressing the complementary goals of achievement and engagement.

2. It will provide a comprehensive account of the cognitive processes and results of interacting with Bayesian pedagogical agents in inquiry-based science learning by conducting extensive empirical studies. To understand the cognitive mechanisms by which self-regulated inquiry-based science learning occurs with middle school students interacting with Bayesian pedagogical agents, the research team is taking a multi-method approach to investigating the use and effectiveness of Bayesian pedagogical agents. In both controlled laboratory and classroom-based field settings, these studies are investigating the central issues of self-regulation with respect to both achievement (science content knowledge, transfer, and effective strategy use, including strategy selection and strategy shifting) and engagement (self-efficacy, situational interest, and mastery orientation with an emphasis on persistence) to determine precisely which technologies and conditions contribute most effectively to learning processes and outcomes.