Experiments to clarify whether the "gain and loss of esteem" could be observed in human-robot interaction

摘要

This paper describes the psychological experiments to clarify whether the gain and loss of esteem could be observed in human-robot interaction. The gain and loss of esteem is defined as follows: changes in rewarding behavior from another have a greater impact on an individual than consistent, invariant reward. Specifically, person A prefers person B who changes her/his bad evaluation about person A into friendly evaluation than person C who shows the friendly evaluation consistently (gain effect), and person A does not prefer person B who changes her/his friendly evaluation into bad evaluation than person C who shows the bad evaluation consistently (loss effect). We then conducted psychological experiments to clarify whether this "gain and loss of esteem" could be observed in the interaction between human participants and the AIBO robot. Actually, the experimental conditions were set as similar to Aronson and Linder's experiment. As the results of the experiments, we could not observe the gain and loss of esteem between the participants and the AIBO robot. The reason of this phenomenon would be rooted on the fact that the robot's reaction was too simple to the participants compared to the Aronson and Linder's human-human experiment.