Kids love robots. I have a three-year-old friend who can identify the 1950s cult icon Robbie the Robot at 20 paces. My own son Jim could do an impressive multi-voiced impression of R2D2 by age five. Now that real robots are beginning to be everyday household items (when I was a kid, if I’d known I’d be able to buy a vacuum-cleaner robot from Sears when I was a grown-up, I’d be ashamed to learn that I never actually bought one!), one wonders how real kids will respond to them.
When, for example, might a child begin to believe that a robot has a conscious mind, and that humans might communicate with robots the same way they talk with each other? The photo above (source: IRC) depicts a child interacting with Robovie, a robot designed to make human gestures, speak natural languages, and establish eye contact, just like real people. Perhaps these interactive features are the key, but perhaps a mere humanoid shape is all that is necessary to convince a child that something is “human.”
A team led by Akiko Arita developed a test to see how 10-month-old infants reacted to the Robovie. They showed these babies a movie of a person talking with the robot. Some babies saw the robot responding and interacting in a natural human way, another group saw a human talking to an unresponsive robot, and a third group saw the robot interacting with an unresponsive human. Next, They were shown movies of a human talking to someone hidden behind a curtain. The curtain was removed and either a robot or a human was revealed.
Most research with infants is conducted in a similar manner: since babies can’t tell us what surprises or interests them, researchers show them a stimulus of some type, then measure how long they look at the stimulus. The longer the stimulus keeps their attention, the more surprising it’s surmised to be. In this case, the researchers measured how long babies looked at the newly revealed robot. Here are the results:
When babies had seen the robot interacting with the human previously, they appeared to be equally interested in a human-human conversation and a human-robot conversation later. But when the babies initially saw the human talking with an immobile robot, they looked at the hidden robot significantly longer than the hidden human. It appears that they were surprised to see another human trying to talk with it later. Perhaps more surprisingly, when the robot had tried to interact with an unresponsive human in the movie, babies again appeared to be surprised when a human tried to talk with a robot later.
So even as early as 10 months of age—well before they are able to talk themselves—it seems that infants consider interactivity to be the key factor in deciding whether a robot is something to talk to, whether it has a human mind. Perhaps it’s no wonder, then, that this was my son’s reaction to the scene in Star Wars when Luke and Han were awarded medals for destroying the Death Star: “Why didn’t R2D2 get a medal, too?”
Arita, A., Hiraki, K., Kanda, T., & Ishiguro, H. (2005). Can we talk to robots? Ten-month-old infants expected humanoid robots to be talked to by humans. Cognition, 95, B49-B57.