Cynthia Breazeal: Expert Q&A
Engineer Cynthia Breazeal answered questions about her work with sociable robots, offered her career advice, and more on November 28, 2006.
Q: Hi Cynthia,
I'm a psychology student, and I was wondering what theories/theorists you look to when designing robots' learning processes and social interaction? Thanks!!
Cynthia Breazeal: Hi Lara,
With respect to learning processes and social interaction, I tend to look at the developmental psychology literature, although the comparative psychology literature is also very inspiring. Forms of early social learning that bootstrap mind-reading (theory of other minds) are a favorite topic of mine. Psychologists and neuroscientists who have inspired my work are Andrew Meltzoff, Linda Smith, Jean Decety, Vittorio Gallese, Antonio Damasio, and many others. My book, Designing Sociable Robots, discusses many of the theories that inspired the design of Kismet. Much of my work with robots investigates how these various theories and ideas can be integrated into a coherent, functioning whole. It is a process of science by synthesis.
Q: Cynthia,
Your childhood stories were truly inspiring. Therefore, I would like to know what you would say to girls of today who might still think that science and inventing is something for boys.
Best, Cris Scorza, Cultural and Community Programs Coordinator at the Brooklyn Children's Museum, New York
Breazeal: Cris,
I say that the future welfare of our nation depends on plenty of girls and boys pursuing careers in science and technology. If we leave it only to one gender, then we've cut our young intellectual talent pool roughly in half. I think the bigger issue is communicating to girls that science and engineering is something they should really consider. I highly recommend the book series Adventures of Women in Science (a project of the National Academy of Sciences). Each book is a biography of a woman scientist in a particular field. The one thing that all of the profiled women scientists have in common is that they love their work, are making valuable contributions to society, and have fulfilling personal lives. I want to see more girls feel the love.
Q: How has having children yourself changed your perspective on your work, and in particular, has it made you think about ways in which your work might benefit ill children in the future? Ryan Hall, Silver Lake, Ohio
Breazeal: Hi Ryan,
My work has always been strongly inspired by the cognitive and social development of children and the crucial role that social interaction plays in that process. I wouldn't say that having children has changed my perspective as much as broadened it and made my research and its application much more personal. My parents always told me that you can't appreciate what it's like to be a parent until you are one—and then you're one for life. Having children of your own, nurturing them, experiencing the joy of seeing them learn and grow—well, it adds depth and meaning to everything. Being a parent has certainly influenced my work to pursue projects like the Huggable—to try to make a more direct connection between my academic life and helping real people.
Q: I'm just starting my B.S. in Computer Science. What educational path should I take to get into social robotics and AI [artificial intelligence]? Jim Meeker, Normal, Illinois
Breazeal: Hi Jim,
Most people working in social robotics have a computer science background, so you're at a good starting point. I recommend taking a class or two in AI and rounding out your CS curriculum with some courses in psychology or brain and cognitive science. I'd also recommend a couple of courses in electronics and mechanical design to appreciate the hardware side. Your undergraduate degree will give you breadth. Graduate school is when you start to engage in research and can pursue more specialized topics like social robotics and human-robot interaction.
Q: Will anthropomorphic robots ever be able to "think" in the same way and to the same degree as humans? Ray, USA
Breazeal: Hi Ray,
There are many opinions on this question. My interest is not how to make robots indistinguishable from humans, but rather how to make robots that are compatible and synergistic with people. Robots are interesting and bring value in part because they are different from us. For me, the more interesting question is how to design robots for a more effective human-robot team. After all, if we want to make more people, well, that's a solved problem.
Q: I wish I were rich so that I could buy a robot like yours. How long do you think it will be until affordable interactive "toy" robots are available? Nathan Haney, Waipahu, Hawaii
Breazeal: Hi Nathan,
You can buy interactive toy robots now. They're just not as much fun as we'd like. But toy companies keep trying every year. A lot of people enjoy their Roomba vacuum cleaners. That's interesting because the Roomba was designed to be functional, but people also find amusement in using it. I think for personal robots to succeed as a commercial product, they need to be useful, appealing, and fun over the long-term.
Q: What combination of "bottoms-up" programming techniques do you find works best to create self-learning autonomous robots? (Examples are genetic algorithms, neural nets, Bayesian nets, etc.) Tom, Vienna, Virginia
Breazeal: Tom,
I view all of these techniques as useful tools in the toolbox. I'm interested in building robots that can learn opportunistically in the human environment (like my own kids!). This involves being able to learn on their own from their own self-guided exploration, as well as being able to learn from and with others. You need an end-to-end, cognitive-affective-social architecture to support this—with processing "loops" as you find in biological brains. So, as you might imagine, this involves bottom-up processing as well as top-down. And, of course, interpersonal interactions and coupling with the world (outside-in and inside-out).