Virtual co-learners may provide keys to faster, deeper learning

When communication professor Clifford Nass began his work with OMRON Corporation examining how virtual co-learners could influence other students, he didn’t know exactly how much impact the computer-generated co-learners, or “avatars,” as he calls them, might have. Now several years into the Media X project exploring whether avatars – the computer-generated characters that act as co-learners in a virtual classroom – hurt or help students learn, Nass and his team have uncovered surprising findings that have the potential to improve future online teaching in both academic and corporate settings.

“We were most interested in to what extent could this co-learner provide help or harm,” says Nass, whose research project is called “Cross Cultural Characters and Avatars: eLearning Analysis.” The project is one of dozens supported through Media X, Stanford’s campus-wide research network for the innovative design and study of technologies of the future. Media X coordinates the development and implementation of active collaborative partnerships with industry, foundations and governments with Stanford’s labs and classrooms.

“We wanted to ask to what extent should a virtual co-learner be supportive or not supportive, and to what extent could it play an important role,” explains Nass, who is collaborating with Ph.D. students Roselyn Lee and Heidy Maldonado and OMRON researchers Kinihiko Iwamura, Hiroshi Nakajima, Ritsuko Nishide, and Ryota Yamada.

To explore the notion of the virtual co-learner, Nass’ group set out to study the two functions the co-learner might play – the social function in which the co-learner provides support (“Hey, good job answering that question”); and a “release-the-heat” function in which the co-learner takes some of the instructor’s questions and provides his own answers.

Three scenarios helped researchers evaluate the best format for co-learners within a class teaching English as a second language:

• Virtual teacher and student interact
• Virtual teacher, student and virtual co-learner who only served to “take some of the heat off,” i.e. answer occasional questions
• Virtual teacher, student and supportive co-learner who made supportive comments, such as, “That was a hard question,” or “Good job!”

“The findings were that people learned much more with a supportive agent,” observes Nass. ”Some of our other findings indicate that a smarter co-learner (or agent), the one who gets the answer right, helps people learn more than dumber agents. It is clear that in any teaching or learning situation it is worthwhile to have a co-learner – someone else who appears interested.”

Future research will explore questions about many different possible types of co-learners and their impact on the primary student, says Nass. Consider, for example, the impact of a “teacher’s pet,” or a “good cop/bad cop” scenario.

“We are very excited about this research,” Nass says. “This issue of the co-learner is new and very little is understood even about how important the co-learner is. There is a lot of work on collaboration, but the idea of having a co-learner that is like me or not like me is unexplored territory.”

____________________________________________________________________

For more information: http://chime.stanford.edu/

Back to News