The RoBallet project run by MIT's Future of Learning Group doesn't look or sound like hard research: Nine children, dressed in sweats fitted with flexible sensor strips, stomp on pressure sensors to trigger changes in the ambient lighting and sound. The performances are choreographed by a professional ballet dancer in collaboration with the kids. The idea is to give students the experience of controlling technology to realize the stuff of imagination. Other projects use software, robotics and sensors as tools with which children can design environmental exploration projects, such as water-quality studies.
It's what Future of Learning co-director David Cavallo calls "hard fun" creative yet disciplined and purposeful uses for technology. One surprise of the computer revolution has been children's affinity for what adults have generally regarded as a potentially dehumanizing technology. Popular computer games and educational software are only the beginning. In projects conducted at MIT and in classrooms around the world, Cavallo and his Future of Learning colleagues are using technology in ways they believe will revolutionize our understanding of the human mind.
EE Times: What was your first encounter with computers and digital technology, and how did it influence your intellectual development?
David Cavallo: The first was in the '60s, when I was in high school. I grew up in Cleveland, and our math class had a connection to Case Western Reserve. We were able to do some work, things around Fortran, to think about math and computers. I thought programming was just a blast, a different way of thinking about problems.
That led to thinking about how you could use computers for learning first thinking about artificial intelligence and intelligent tutoring systems. A professor at Rutgers, Ken Kaplan, introduced me to Logo [a programming environment widely used as a classroom tool], and that's when my interest really took off.
EET: And that was actually an MIT innovation.
Cavallo: Logo was developed by Seymour Papert here at MIT. That's when I decided, "Wow, I really want to work with Seymour," and I moved to Boston.
EET: So you encountered Papert's ideas via computer technology, rather than by picking one of his books off the shelf?
Cavallo: Yes, exactly.
EET: What is it about his ideas that really attracted you?
Cavallo: It tied a lot stuff together that I had kind of been feeling, but in different pieces. It's like a little network of certain ideas. One would be, what's deeply there in the intelligence of kids? And then, what's the role of the computer, not as a teaching machine but as a machine that you could explore with? Exploring this with kids could yield a very liberating, empowering form of learning.
EET: The computer and AI have been compared to the mind in some ways, but they are also very different from how the mind works. Is the computer the appropriate instrument for that type of work?
Cavallo: What's really been rich in AI, what's really rich in the computer and what has helped us to understand minds better was trying to build models of minds. Let's say you are trying to figure out a way of [understanding] intelligent behavior. You can observe people, you can do other things; but a model can be much more complex, and then you can explore how this [aspect under study] might actually work. And this kind of exploration, being really rich, gives you a better way to think about what you're studying. The model is just this open space to try and make things work.
EET: So it's like an external measuring rod that you can use to calibrate theories of the mind?
Cavallo: In some ways, I suppose, but what is rich in working with kids is model building. Let's say you build a model boat [by hand]; it will look like a boat. But if you build the model on the computer, you can think about it in all these ways. You can run tests a million times, vary parameters, and it can wind up rich.
Early AI was logic-driven, and it was good for some kinds of problems but failed miserably on others. Without the computer, we had a much more simplistic view of what was really going on. We found we couldn't model [intelligence] in simple ways, we found all these things that we couldn't simply do. The computer has given us a much greater feel for complexity and better models for thinking about those things.
EET: What would you say is a seminal idea that has come out of this that was not known before?
Cavallo: A lot of visionary people from way back [such as Frederick] Bartlett [who advanced the thinking] on memory and [John] Dewey on learning didn't have the chance to work with this stuff, but they developed a lot of great ideas. We're exploring a lot of those again now.
If you go back 50 years, the view of what developed minds did was mostly limited to just planning, reasoning, logic. We now realize the richness of thought that there are many ways of thinking.
For example, [MIT's Marvin] Minsky is doing work on common-sense reasoning. [Earlier], people put so much work into building expert systems, and then we discovered that [building an expert knowledge base] was much easier to do than thinking about how you could cross the street safely, which a 3-year-old could kind of figure out. Intelligence is really mixed; there are tons of stuff going on that work together, and we learn from it [all]. What we've tried to do on the computer has helped break the more-restricted view of what intelligence really is.
EET: The computer has evolved quite dramatically into areas such as wireless computing and ubiquitous computing. Has that expanded your research as well?
Cavallo: Yes, definitely, in different ways. What we find is fun to do with kids is to take something simple, like, "How do ants collaborate?" You have a chance now, as things have gotten smaller and cheaper and wireless, to actually build a [model in which the students themselves, equipped with ubiquitous-computing tools, move about the space and interact as ants would] build your own "ant colony" and make the "ants" walk around and start to do things.
You can do these simulations on the screen, but when you do them physically, there is a different feeling to them. And that has started us thinking about how school could be very different.
School has been organized around the dissemination of information: You have classrooms, and you give information to kids. But I think people have known for a long time that active learning is better. Now, with ubiquitous computing, you are not stuck in the classroom.
EET: In some of your writing, you have indicated that the arts side of education is more receptive to this [method of learning] than the hard technical people. Is that a trend that you see?
Cavallo: Absolutely. When we start to work with a school system, they usually send us the science teacher, the computer teacher, the math teacher. Our response is, "Send us the art teachers." They're used to working with a variety of materials, and they want to know "What is the idea, what is the vision?" It's been really successful in terms of the feel for these projects.
For example, we've run a class in which the students build computational models of how they would like to improve life in their community. You get the kind of aesthetics from the arts class that just naturally fits in and that you would like engineers to have. The art teachers are great colleagues because they just jump into the spirit of, "Wow, new stuff to work with, new ideas to express let's make it." And that's important because we look at computation as a new type of material to work with, and we want people to be fluent with it.
EET: Are you looking at traditional areas? Suppose you wanted to enhance learning calculus?
Cavallo: What should kids learn now? One thing we like to do is to talk to an audience and ask for a show of hands on how many of them are working in the field they studied for. It's usually fewer than 50 percent. Then you say, "How many of you are working in fields that didn't exist when you were in school?" That's growing. It gives you a different sense of what is important.
Everybody talks superficially about learning to learn, but we don't talk to kids much about learning, or work with kids on learning; it's just supposed to happen. And much of the stuff in the curriculum reflects the views of a hundred years ago. In math, we spend so much time with kids on calculations, and everything gets based on arithmetic. But mathematical thinking is more important than faster calculation or knowing how to bisect an angle.
EET: What would you want to present as a really successful and exciting class that you have done at MIT?
Cavallo: The project that was designed around building the kind of city you would want to live in, and then using the project to get the students, teachers and administrators to think about how learning could be different. It was a rich experience in a lot of different ways.
[When we took the project to Manaus, Brazil] one kid said at the end of the week that he'd had a spectacular experience but that when he went back to his regular class on Monday, he wouldn't be able to look at it the same way. That made me feel happy and sad at the same time happy that he'd had such a good experience as a learner, seeing what he could do and how school could be; but, then again, he was going back to a regular school on Monday, and he was going to be disappointed.
It was the same thing for everybody in the class to be able to see how much they could accomplish, in a relatively short time, in areas where they didn't really have a lot of experience.
EET: How did you get them to use computers to model the city? That sounds like a very high-level, difficult task.
Cavallo: We start with a brainstorming session: "Let's look at your community. What do you like, what don't you like, what might you dream that you have never seen?" We start with that, and they choose what their own project is going to be. So you don't try to model the whole city; you just choose what aspect you want to work on, like transportation or recycling. And some of the projects are quite fanciful, like one kid who wanted to shoot all the garbage out into space. So we explored whether we would be helping or hurting the environment by doing that.
And as you build a model, you start to think through the problems. It starts out as fun, but you have to get pretty serious about it. How much energy would it take [to launch trash into space]? What would happen to all this garbage out in space?
EET: How about here in the United States? The Internet has been introducing changes such as remote learning. Can your methods be incorporated into existing schools?
Cavallo: I'll go back to my own experience. When I was taking geography, we had to memorize all the capitals. But why would anyone care about geography in that kind of way? We tend to emphasize the information-retrieval aspect of education, but we should get rid of that. I think too much of the use of the computer [in schools] is still in a kind of trivial way.
EET: Do you foresee a preponderance of these computer-based approaches eventually creating a revolutionary change in educational institutions?
Cavallo: I do, because people find really rich uses for computers. There are enough of these kinds of experiences where you see what kids do, what adults do, how they have adopted this technology in their lives and what they really want from schools so, yeah, I'm an optimist.
EET: So what is the future of learning?
Cavallo: What I haven't talked about is the new thing not just developing the content of education and the new tools that help you think about that, but, really, the computational side. The biggest hope for us is how we can use the computer as something that you create with, something that you think with and dream with and imagine with and then try to realize what you imagine. We're not just looking at the computer as an information-delivery device or a communication device: It's a dreaming and making device. That's the hugest potential.
We are really talking about the development of thinking, not just the acquisition of information. When we do these classes, we end up learning more than the kids. They always seem to come up with better ways to do things. It's a blast perhaps the greatest job in the world.
Born: Nov. 6, 1951
- BS, computer science, Rutgers University
- MS, PhD, media arts and sciences Massachusetts Institute of Technology
Co-director, MIT's Future of Learning Group
- Research scientist, MIT Media Lab
- Designed and implemented medical informatics system for Harvard Health Services
- Consulting software engineer at Digital Equipment Corp.'s Artificial Intelligence Technology Center
- Founded Digital Equipment's Latin American and Caribbean Advanced Technology Group
- Designed knowledge-based systems for industry
- Has advised heads of state and ministries of education around the world on the adoption of advanced technologies for learning and educational reform