Recent studies have utilized self-disclosure as a strategy to foster intimacy in child-robot interaction (cHRI). However, fictional self-disclosure, in which a robot expresses fabricated experiences and emotions as if it were human, may lead to ethica...
Recent studies have utilized self-disclosure as a strategy to foster intimacy in child-robot interaction (cHRI). However, fictional self-disclosure, in which a robot expresses fabricated experiences and emotions as if it were human, may lead to ethical deception by instilling false beliefs in children. As an alternative, factual self-disclosure, where a robot explicitly reveals its mechanical identity and limitations, has been proposed. Nevertheless, empirical evidence is needed to determine whether factual self-disclosure can sufficiently promote intimacy with children while preventing ethical deception.
Accordingly, this study examined differences in children’s self-disclosure (amount and intimacy level) and robot perceptions (anthropomorphism, likeability, and perceived safety) as a function of the robot’s self-disclosure condition (no self-disclosure, fictional self-disclosure, and factual self-disclosure) among 5- to 6-year-old children. Furthermore, the study investigated the structural relationship through which the robot’s self-disclosure affects the child’s self-disclosure via likeability, and explored the correlation between children’s actual self-disclosure behaviors and their subjective evaluations of the conversation experience.
Seventy-three children aged 5 to 6 years from four public kindergartens in Goyang-si, Pyeongtaek-si, and Hwaseong-si, Gyeonggi-do, Korea, engaged in one-on-one conversations with a social robot. Following the interaction, robot perceptions and conversation experience were assessed through individual interviews where a researcher read each question aloud and recorded the child’s responses. All robot-child conversations were recorded, transcribed verbatim, and coded to derive behavioral measures of the amount and intimacy level of the children’s self-disclosure. Data were analyzed using IBM SPSS Statistics 29.0, employing descriptive statistics, Kruskal-Wallis tests, Mann-Whitney U tests, mediation analysis using PROCESS Macro, and point-biserial correlation analyses.
The main findings were as follows. First, while there were no statistically significant differences in the amount or intimacy of children’s self-disclosure across conditions, the factual self-disclosure condition showed a tendency toward the highest level of intimacy. Second, regarding robot perception, likeability was significantly higher in the two self-disclosure conditions (fictional and factual) than in the no-disclosure condition. Item-level analyses further revealed that children in the factual self-disclosure condition attributed significantly lower levels of anthropomorphism to the robot on specific items related to autonomy and naturalness of movement. Third, mediation analyses showed that the robot’s self-disclosure influenced the amount of children’s self-disclosure indirectly through increased likeability, with this indirect effect reaching statistical significance in both the fictional and factual self-disclosure conditions. Fourth, children’s positive evaluations of the conversation experience were significantly correlated with higher levels of both the amount and intimacy of their actual self-disclosure behaviors.
In conclusion, this study demonstrates that factual self-disclosure, in which a robot transparently communicates its mechanical nature, can effectively promote children’s self-disclosure via enhanced likeability while simultaneously mitigating excessive anthropomorphism. These findings suggest that children do not need to mistake a robot for a human in order to form positive relationships; rather, they are capable of experiencing likeability and engaging in meaningful interactions while understanding the robot’s non-human nature. By empirically showing that ethical transparency and interaction effectiveness can coexist, this study provides foundational evidence for the design of child-centered and ethically responsible child-robot interactions.