Learning By Teaching
Daniel Schwartz (TC ’88, ’92) has created software that puts students in the driver’s seat
By Barbara Finkelstein
Dan Schwartz likes to recall a graduate student at Stanford who wanted to offer teenagers in his native India an alternative to the one history textbook available to them. “He got five teachers, each with a very different perspective, to give a lecture based on the same chapter,” says Schwartz, a professor in the Stanford University School of Education. “He put the lectures online, and the students in his test group were able to choose the lecture they wanted to hear.”
Schwartz tells this story to illustrate the power of technology to expand choice in the classroom. Ever since the mid-1980s, when, as a teacher in an Alaskan village school, he encountered the Apple IIe and saw its possibilities for educational instruction, Schwartz has been riveted by one question: What is the best way to teach?
One answer, he believes, is to empower students to teach others—even, or sometimes especially, virtual others. To that end, he has developed a series of software programs called Teachable Agents. A Teachable Agent called Betty, for example, will wait for the student to impart knowledge to her. The student might ask Betty to tell him what happens when, say, algae are added to a fish pond. During much sophisticated give-and-take between human and computer, the student creates a “concept map” in Betty’s “brain.” By periodically asking Betty what she now knows, the student ultimately teaches her that bacteria increase with an increase in algae, which decreases the oxygen that fish need to live.
Another Teachable Agent, called Milo, generates more than one interpretation of the data that the student gives him. Information about a rectangle, for example, whose length times width equals 8 will prompt Milo to produce a 1 x 8 rectangle, a 2 x 4 rectangle or a parallelogram with an area of 8.
Do these pedagogical video games really work? To test the effectiveness of Teachable Agents, Schwartz and his colleagues designed an experiment in which half of the students-turned-teachers in a test group used conventional science kits and Teachable Agents to conduct their classes. The other half used only the science kits. Schwartz found that students who had the advantage of both learning materials acquired greater understanding of scientific concepts. Moreover, when Teachable Agents were removed from the classroom, the students were able to apply what they learned about causal relations to the next unit in the science curriculum.
While Teachable Agents are effective for all students, they have been particularly effective with low-achieving students. Schwartz believes that the technology makes up for learning experiences that the students didn’t have when they were younger.
Schwartz has found similar results in studies of a game called Stats Invaders, which he developed with a Stanford colleague, Dylan Arena. In the game, which is modeled after the arcade classic Space Invaders, aliens drop from the sky according to one of two probability distributions. In addition to shooting down the aliens, players must determine which of the two displayed distributions is generating the alien attack. Schwartz and Arena found that students who played Stats Invaders were better prepared for a classroom lecture on probability distribution than those who did not play the game.
“This is a nice example of using technology to do what technology can do well,” Schwartz says. “Technology can give you a set of experiences that will prepare you to understand a more formal treatment that you get from a textbook or a class.”
Schwartz himself got much the same kind of preparation from Teachers College, where he was the first recipient of TC’s Ben D. Wood Fellowship. A single father at the time, he credits the support for helping him to raise his son.
“The faculty I found at TC were profound and inspiring,” Schwartz says. “John Black, my adviser, provided a model and gave me the freedom, gentle steering and support to find my own interests. Herb Ginsburg opened the very precise world of children learning mathematics. Robbie McClintock showed how to see the ideas du jour within the sweeping themes and changing conditions of cultural history.”
Schwartz also was inspired by his fellow students. “I can still remember my intellectual conversational partners 20 years later. We provided each other cognitive and emotional sustenance, nourishing big ideas out of rich experience.”
Perhaps technology’s ultimate value, as Schwartz sees it, is that it can identify the choices that students make as they learn. “How do students deal with failure?” he asks. “Do they try to resolve contradictions, or do they slide over them? Technology is helping us teach and understand 21st-century skills, which aren’t so much about facts and procedures but about making good choices when you need to learn and adapt.”