Gaming to learn: A new interactive model developed by
Professor Dan Schwartz

Teaching fourth graders about eco-systems usually involves assigning dry readings and asking them to memorize facts and relations; the lessons culminate in a paper-and-pencil test. But a new computer game under development at the CAT2 Lab based in Wallenberg Hall is using virtual, on-line worlds and the concept of learning-by-teaching to offer a unique and promising approach that could transform educational computer gaming.

The Triple A Game Show

Headed by Professor of Education Dan Schwartz and a team of graduate students, the CAT2 Lab (formerly AAA Lab) has created The Triple A Game Show, modeled after popular Internet games. The project is funded by the National Science Foundation and is part of the research portfolio of the Stanford Center for Innovations in Learning, housed in Wallenberg Hall.

The Triple A Game Show's Waiting Room.

Based on the fact that students who prepare to teach someone learn more than students who prepare to take a test themselves, the CAT2 Lab's Teachable Agent studies have expanded into the world of Internet games as a means to augment learning. The newly developed game show allows students to collaborate on homework, interact with a teachable agent they create themselves, and have their agents play games that require knowledge of a particular topic area in order to win.

Not only can the new game be utilized to teach any subject - from pond ecology to math and literature - but it is so engaging and effective that students are willing to play for extended periods of time. The Internet component, which allows players to collaborate or compete with each other and even chat along the way, creates a social environment that more closely replicates the experience students might have were they to gather for the afternoon to do homework together.

"We think this environment is unique," said Schwartz. "First, students get to study together on-line in an engaging and research-informed environment. Second, the students need to teach a computer agent about their homework so it can also answer questions in the game show. Our research has indicated that this "learning by teaching" is highly effective. Third, the homework environment prepares students to learn when they get to school, instead of just to practice what they have already been taught."

How Does It Work?

Students can teach their agents by making a concept map that sets out relations between concepts visually, such as the relation between water, sun and plant growth, or oxygen, bacteria and algae. Teachable agents can answer questions based on how they have been taught, and even engage in friendly social interaction. The agents have already been shown to improve student learning and engagement. The new CAT2 project, however, is the first time they have been used in a virtual homework environment.

Users of the unique games log on from home or school to create and teach their agents. The first step is to name their agent and choose what it will look like, from hair color and style to clothing and skin color. The second step is to teach the agent. Each individually constructed agent then participates in a game show using the knowledge the student taught by creating the concept map.

Teachable agents use artificial intelligence techniques to reason with what they have been taught. An agent, for example, can reason through causal relations such as an increase in algae causes more oxygen and bacteria, and an increase in bacteria causes a very big decrease in oxygen, which reduces animal life. Once the student has made the map and the game host poses a question, each agent gives an answer and shows the way it got to an answer by highlighting its reasoning step-by-step through the concept map. The process, which makes reasoning visible, is called an "animation of thought." Students can correct mistakes and replay the game to improve their score and increase their knowledge.

What's Next?

During the last year the games were prototyped and tested with older students to see if the players enjoyed and readily grasped the game show format. In the coming year, researchers will conduct further studies within schools and debug the games for broader distribution.

Back to News