Virtual Students Are Used to Train Teachers
The student-teacher faces a rowdy class.
“We’re not going to have that kind of behavior in here,” she says. “It’s too loud in here to move on.”
The students don’t pay much attention. A boy in the back row, wearing a sleeveless T-shirt, slumps his shoulders. Another student waves his hand aimlessly.
“Nah, just stretching,” he replies, when the teacher asks if he needs something.
Scenes such as that aren’t uncommon in urban classrooms, but in this case there is one critical difference: These students are avatars—computer-generated characters whose movements and speech are controlled by a professional actor.
Each of the five characters—all with distinct abilities, personalities, and psychological profiles, and even names like “Maria” and “Marcus”—were created as part of the TeachME initiative at the University of Central Florida, in Orlando. There, teacher-candidates can practice in a virtual classroom before ever entering a real one.
Real-time classroom simulations like TeachME, supporters say, offer promise for a host of teacher-training applications. Through them, candidates could gain hands-on practice with urban students, or practice a discrete skill such as classroom management.
Most of all, such simulations give teachers in training the ability to experiment—and make mistakes—without the worry of doing harm to an actual child’s learning.
“It allows the teacher to fail in a safe environment,” said Lisa Dieker, a professor of education at the University of Central Florida and one of the designers of TeachME. “Real kids, trust me, will remember in May what you said to them in August. You can’t reset children.”
The Florida project is among the most sophisticated experiments with classroom simulations to date, but other projects offer similar benefits. The computer program simSchool, which mimics a classroom setting, can be populated with up to 18 students with different features and emotional characteristics—all of whom will respond differently to stimulus from a teacher-candidate in charge of the virtual classroom.
Though not yet widespread in teacher education, the idea of classroom simulations could receive more attention in coming years, especially with the student-teaching aspect of teacher preparation now receiving scrutiny. Groups such as the National Council for Accreditation of Teacher Education, or NCATE, are pushing for teacher education programs to step up the variety, length, and quality of their field-based experiences.
“There’s a realization that we have to be able to ensure that we can prepare teachers well for the demands of practice,” said Pamela L. Grossman, a professor of education at Stanford University who has written about the place of simulations, among other methods, for practicing teaching skills.
Both TeachME and simSchool are the product of unusual partnerships linking teacher-educators, researchers, and experts in simulations or immersive media.
Though widely used to train professionals in medicine, nursing, and aviation, simulations are uncommon in the preparation of teachers. Many aspiring educators get about 10 to 14 weeks of student-teaching in local schools.
The idea behind the simulations isn’t to replace traditional face-to-face student-teaching, but to give teacher-candidates the ability to experience specific skill-building lessons, explains David C. Gibson, an associate research professor at the School of Social Transformation at Arizona State University in Tempe.
The principal designer of simSchool, Mr. Gibson was inspired to undertake the project by the work of researchers at Western Oregon University. There, researchers created a computer program allowing teachers to examine student profiles, come up with a preassessment for a lesson, administer it, view student-achievement results, and reflect on them.
Like a board game, it was “turn-based,” contingent on teachers’ inputting an action before generating information. But Mr. Gibson wanted to pursue a program that got deeper into classroom practice—how teachers respond minute by minute in a dynamic environment.
“What happens when you shut the door and you’re teaching? What are you going to say? What are you going to do? What happens when the kids change?” Mr. Gibson said. “I wanted to take apart the black box that was actually the classroom.”
The simSchool program essentially puts the Oregon project into real time. While a teacher-candidate uses the system, it tracks how student performance rises and falls. Such variation depends on the complexity of the tasks the user has assigned, the personalities and abilities of the students, and students’ engagement with each activity.
As in a real classroom, these children don’t sit statically in place. Depending on how engaged they are by the teacher’s actions, students will pay close attention—or distract their peers. That aspect, Mr. Gibson says, gives teacher-candidates insights into features like differentiation. A task that one student masters may be too complicated or frustrating for another, and teachers must analyze student data to respond appropriately.
A complex emotional, perceptive, and cognitive profile determines how each child responds to an assigned task. Emotional factors include openness to learning new things, conscientiousness, degrees of extroversion, agreeableness, and sensitivity. The perception variables include visual, auditory, and kinesthetic abilities. Finally, the cognitive dimension represents the student’s general capacity to learn.
Thanks to the research and development efforts of a team housed at the University of North Texas, in Denton, the program has expanded to allow users to customize individual simulated students, with 2 million discrete combinations possible.
At the University of Central Florida, the TeachME program expands upon the artificial-intelligence programming often used in simulations. It draws on a professional actor to represent all five characters created for the project.
The characters are based on a psychological framework different from simSchool’s, but are no less diverse in their attributes. Each avatar has a individual profile, and every actor who participates in the program is trained on all five students’ personalities and profile. Many of the actors, in fact, hail from a fine-arts program at UCF combining aspects of psychology, improvisation, performance, and philosophy.
To use TeachME, a teacher-candidate stands in front of a large screen depicting the five students at their desks and begins to teach. In a separate lab across campus connected by camera and microphone, the actor—who’s already received a copy of the lesson a teacher-candidate will be using that day—responds to the teacher’s direction, operating each character in turn. Motion-capture instruments replicate the actor’s gestures for each character on the screen.
“The interactors know the personality of the character they’re playing, but that’s it,” said Charles E. Hughes, a professor of electrical engineering and computer science and a director of the Synthetic Reality Lab at UCF, where the TeachME application was created. “They know the backstory, they know the family life, but they have no script associated with what that person has to say; what they do is totally driven by the circumstances.”
Jacqueline A. Rodriguez, now the program director of TeachME and a doctoral student at the Florida university, was among those initially skeptical that the student-avatars could resemble students in an urban classroom. It took just minutes in the simulation to change her mind, she said.
“My question was, were they going to respond like students I’ve taught before, or were they going to be like kids in suburbia—very vanilla and easygoing?” said Ms. Rodriguez, a Teach For America veteran from the District of Columbia public schools. “They weren’t.”
A project like simSchool lacks live actors, but the basic program could be expanded to include more-complex emotional profiles for the characters, Mr. Gibson said. He would also like to see the cognitive variables in the program multiply to allow for more-nuanced achievement patterns among the virtual students.
Right now, he said, “it’s like saying this student is going to be an A student or a B student. It’s highly inadequate, but unless we have a national impetus to build on the model and move it forward, we’re kind of stuck there.”
The projects are limited in the number of teacher-candidates that experience them as part of their initial preparation, but both are starting to receive recognition.
During its last accreditation review at the University of North Texas, NCATE approved use of simSchool for up to 10 of the 40-plus required hours teacher-candidates must spend observing classrooms before they begin formal student-teaching.
TeachMe is available for all teacher education instructors at the University of Central Florida, though it’s not a mandatory part of the teacher education curriculum.
There is still much to learn about the ideal uses for the systems. Ms. Grossman of Stanford Unversity would like to see additional attention paid to a framework for breaking down and analyzing specific teaching skills—getting the proper grain size of the individual practices that novice teachers need to master in the simulated setting.
Such frameworks for underpinning the simulations are more advanced in medicine and nursing, she said. But a school classroom requires more-complex interactions than the comparatively easier-to-simulate, one-on-one doctor-patient relationship.
“The work of teaching always requires, in essence, the crowd,” Ms. Grossman said. “What’s complex about it is the ability of the person to manage multiple interactions at once.”
Research on both systems is ongoing, and partner universities are extending the projects in new directions.
Old Dominion University, in Virginia, Utah State University, and West Virginia University, among others, are in various phases of incorporating TeachME into their own programming; the Utah university has developed two new student-avatars.
Cost is potentially an issue for wide-scale use of the TeachME system. It costs university partners about $6,000 for equipment, plus $120 per hour of use.
The research base for simSchool is gradually expanding as well. In spring 2007, researchers at the University of North Texas conducted a small study on two classes of preservice candidates; those who used the system reported, on average, higher levels of “instructional self-efficacy.” In lay terms, they felt more resilient in the face of instructional setbacks and more confident in trying alternative approaches.
Tandra L. Tyler-Wood, an associate professor of learning technology at UNT who has conducted much of the research on the simSchool program, says the finding indicates that the system could help improve retention rates among newly minted teachers. On-the-job-frustrations are particularly likely to cause novice teachers to change schools or leave the profession.
Ms. Tyler-Wood is now using the programmable aspects of simSchool’s students to approximate the features of students with disabilities—reducing auditory perception approximates deafness, for instance.
Whether the appetite for simulation in the field of teacher education will match the growing sophistication of the systems remains to be seen.
The appeal of TeachME might be less immediate for today’s teacher-educators, who grew up without much computer technology, than for a group of budding Gen Y teachers who grew up playing World of Warcraft and tinkering with electronic gadgets, Ms. Rodriguez postulated.
Still, as more teacher-educators experience simulations, Ms. Dieker expects the new tools to gain legitimacy. Among her triumphs was the discovery that a group of middle school students themselves found the student-avatars to be lifelike approximations of their peers.
Before using the system, during a field trip to the TeachME lab, many of the middle schoolers expected to easily play the boss of their own classrooms. Ten minutes with Maria, Marcus, and their virtual peers changed their minds, Ms. Dieker says.
“When they leave the simulator, they say, ‘I’ll never be mean to my teacher again.’ ”
Vol. 30, Issue 15