Keepon Helps Kids Learn to Argue Better

Social robots (especially cute ones) can help children resolve conflicts without all the screaming

Social robot Keepon
Photo: Randi Klett/IEEE Spectrum
With Keepon helping kids work through their disagreements, conflicts were more often resolved constructively through sharing, turn taking, trading, and playing together.

Kids are not well known for their conflict resolution skills. That’s part of being a kid, I guess, but they’ve got to learn these skills at some point, or they turn into teens without conflict resolution skills. And then you end up with adults that only know how to solve problems by throwing tantrums of one sort or another: We’ve all met people like that.

It would be great if there was a way to teach children how to handle disagreements equitably, and there is: It’s called teachers (or adults in general). But having adults around all the time gets expensive. As always, whenever human time is at a premium, some people are quick to suggest that robots should take over, whether it’s necessarily a good idea or not. In order to figure out whether it is a good idea or not, Solace Shen and two colleagues from Cornell University and University College London decided to do a series of experiments. The researchers set up a social robot (Keepon, one of our favorite robots of all time!) in a playroom and tested whether or not it could successfully help small children resolve conflicts over toys.

Most conflicts that young children have with each other are settled without adult intervention in under a minute, but usually this happens in such a way that means one kid “wins” and the other “loses.” Ideally, both children would end up as winners, or at least as co-not-losers, but it takes some skills to make that happen. Depending on how good your own conflict resolution skills are, you may or may not be familiar with the following six-step integrative negotiation procedure:

  1. Recognizing that a conflict exists and expressing a desire to resolve it constructively (“Stop. We have a conflict. Let’s work it out.”).
  2. Stating what you want and giving your underlying reasons (“I want…because….”).
  3. Expressing how you feel (“I feel mad or sad.”).
  4. Communicating your understanding of what the other wants and why (“You want…because….”).
  5. Inventing options that maximize mutual gain (“Some ideas are….”).
  6. Reaching an agreement by selecting one of the options (“We agree on….”).

Yay, a list of rules! Seems like something robots could probably handle, right?

The first step to teaching kids how to resolve a conflict is to, well, create a conflict. To do this, the researchers put a pair of children (3-6 years old) in a playroom with a bunch of toys. The toys were somewhat nefariously chosen to prompt conflicts through resource scarcity, as the researchers explain in a paper presented this week at the ACM/IEEE International Conference on Human Robot Interaction, in Chicago:

We arranged children’s play material so that there were only one of certain key pieces (e.g., only one ice cream cone in a Lego Duplo ice cream set, only one controller for a remote control car). If the children asked for additional resources, they would be told that there was only one.

Keepon was set up in Wizard-of-Oz (WoZ) mode, with researchers remotely controlling the robot’s motions and speech. With some reinforcement from an adult, the children were introduced to Keepon in a way that established it as the authority in the room, in charge of the play session—Keepon tells the children that it’ll decide when they should switch the kinds of toys they’re playing with, and also tells them that if they have a disagreement, it will play a loud noise (like a referee whistle) to get their attention and then help them resolve it:

The video shows a conflict over a certain magnetic tile the girl wanted, which the boy was using as a wing on his plane. Both children paused and reacted to Keepon’s mediation intervention but weren’t able to resolve the conflict right away. However, upon Keepon’s fourth prompt, the boy flew his plane one last time and gave the girl both of his wing pieces, and as a trade, asked for two other pieces from the girl.The video also shows a subsequent conflict between the same pair 13 minutes later over the remote control car, in which right after Keepon signaled the conflict, the boy did what he thought was appropriate and gave the girl a turn at the toy.

Keepon isn’t really don’t anything that an adult couldn’t do, and it’s not autonomous, with two people remotely controlling everything that it says and all the movements that it makes. But the point of this research was to see whether a social robot could be used beneficially in situations like these: Would the children resolve conflicts more effectively with the help of the robot than they would by themselves?

The short answer is, yes. The study recorded 89 object possession conflicts, 39 of which happened while the robot was in mediation mode (the rest were a control). With Keepon trying to be helpful, 26 of those 39 conflicts were resolved constructively, through sharing, turn taking, trading, playing together, or other consensual settlements. Keepon got directly involved in 20 of those 26 conflicts, and 13 of those 20 ended well. Compare this to the control condition, where only 9 of the 50 conflicts had constructive resolutions.

Children were about 4 times more likely to resolve object possession conflicts constructively when they played with a robot that not only facilitated and directed the play session, but also signaled the onset of conflicts and offered prompts for constructive conflict resolution whenever possible, compared to when they played with a robot that only facilitated and directed the play session.

It’s important to note here that Keepon seems to be effective even when it’s not directly responsible for resolving a given conflict. The robot is also useful indirectly, possibly because the children recognize its presence as a mediator and act differently just because of its presence. Even if Keepon just got the kids to pause for a second during their conflict, that made them three times more likely to resolve it without bloodshed.

For this kind of thing to be effective in practice, Keepon (or another social robot platform) would need to act autonomously, which will likely be a challenge. Every conflict is different, and conflicts evolve rapidly, making it tricky for an autonomous system to adapt quickly enough to be effective. This study also only looked at two children at once, and conflict in larger groups can be much more complex—seems like you might need a robot that’s a bit bigger than Keepon to get in the middle of something like that.

“‘Stop. I See a Conflict Happening.’ A Robot Mediator for Young Children’s Interpersonal Conflict Resolution,” by Solace Shen, Petr Slovak, and Malte F. Jung from Cornell University and University College London, was presented this week at HRI 2018 in Chicago.

Robotics News

Biweekly newsletter on advances and news in robotics, automation, control systems, interviews with leading roboticists, and more.

About the Automaton blog

IEEE Spectrum’s award-winning robotics blog, featuring news, articles, and videos on robots, humanoids, automation, artificial intelligence, and more.