November 21, 2017

Robots as Dogs?

In this experiment, scientists tested the perceptions of kids on the robotic dog AIBO vs. a real Australian Shepherd dog. 

Two previous studies have been conducted on people’s perceptions of robotic dogs and were used to set up this experiment. The first study gathered adults’ perceptions of AIBO over the course of three months. Researchers concluded that the adults felt AIBO possessed technological essences, life-like essences, mental states and social rapport. Moral standing was not attributed to AIBO by the adults. The second study questioned 80 preschoolers on their behavioral interactions with AIBO in comparison with a stuffed dog. Children thought about the two objects similarly. Two-thirds of children attributed mental states, social rapport AND moral standing to both objects. However, behavioral differences did exist when children engaged with the two toys, showing that they could distinguish between the two toys.

              Both previous experiments failed to compare perceptions of AIBO with a living dog, so in this experiment, that’s what they did. They compared the perceptions of 72 children, ages 7-15 (even male and female per age group). The experiment consisted of two segments. Children would be allowed to play with the dog and AIBO for five minutes, and then would be monitored while they were asked questions about their perceptions of the dogs. The questions were categorically divided into physical essences, mental states, sociality, and moral standing. The children were monitored for how long they interacted with AIBO/Canis, how long they were in arms distance, and their relative position while interacting (squatting, standing, lying down, chair, etc..). The second segment consisted of a card comparison task, where the children were asked to compare whether the card of AIBO was more similar to a card of live dog, a humanoid robot, a stuffed dog, or a desktop computer.

The results showed that more children were in arms distance of Canis over the 30 ten-second intervals of the interview. However, 80% of the children were within arms distance of both over the majority of the intervals. Children spent the majority of their time in contact with Canis (192.8/300 seconds) while  not many spent time in contact with AIBO (30.7/300 seconds). Every single child had more physical contact with Canis than AIBO. The majority of the time, children were kneeling or squatting with both AIBO and Canis.

More children gave Canis moral standing, sociality, mental states, and physical essences than AIBO. However, the majority of  children did give AIBO moral standing, sociality, and mental states. And a small few gave AIBO physical essences.

              Majority of children viewed AIBO as a robot rather than a desktop computer, stuffed dog, or live dog. Children were evenly split on the comparison of AIBO to a desktop computer and a live dog comparison.

              Did children treat AIBO as if it were a live dog? No. All the data shows otherwise (Behavioral monitoring, questions and card comparison). HOWEVER, children did treat AIBO in ways that were dog-like. While the differences in how the children viewd AIBO and Canis were statistically significant, the majority of children believed that AIBO did have moral standing, sociality and mental states. Children, based on the card comparison, did not merely associate AIBO with the digital world.

              It may be possible that there exists a combination of perceptions between animate and inanimate. In essence, there is an in-between. 

https://dl.acm.org/citation.cfm?id=1056988

Melson, G., Kahn, P., Beck, A., Friedman, B., Roberts, T. and Garrett, E. (2005). Robots as dogs?. CHI '05 extended abstracts on Human factors in computing systems - CHI '05.

Reviewed by: Anneliese Ceisel