The dynamics of human–robot trust attitude and behavior — Exploring the effects of anthropomorphism and type of failure

Abstract

In two experiments centered around trust dynamics in human–robot interaction, we explored the interplay between attribute-based characteristic of robots (i.e., anthropomorphism and type of failure) and trust attitude, perceived reliability, as well as trust behavior. In general, trust attitude is considered to be one of the most important factors for successful collaboration. However, although previous research has shown that performance and attribute-based characteristics of robots lead to changes in trust attitudes, the question remains whether this is also linked to behavioral adaptations. To investigate this relationship, participants collaborated with robots via voice commands that failed. In the first experiment, participants worked with either an anthropomorphic or technical robot. In the second experiment, all participants worked with the technical robot, which made a failure in either information acquisition/processing or action implementation. Participants trusted the technical robot significantly more than the anthropomorphic one. Moreover, failures in information acquisition/processing seemed to lead to more trust dissolution than failures in action implementation. In both experiments, the dynamics of trust attitudes as a function of interaction experience were shown, in particular, trust dissolution due to failure. However, there was no evidence of behavioral adaptation following a failure experience. In general, the effects on the attitudinal level did not clearly translate into behavioral adaptations. For trust attitude, purely decorative anthropomorphic robot design and failures related to the cognitive abilities of robots should be avoided.