Trust is a crucial guide in interpersonal interactions, helping people to navigate through social decision-making problems and cooperate with others. In human-computer interaction (HCI), trustworthy computer agents foster appropriate trust by supporting a match between their perceived and actual characteristics. As computers are increasingly endowed with capabilities for cooperation and intelligent problem-solving, it is critical to ask under which conditions people discern and distinguish trustworthy from untrustworthy technology. We present an interactive cooperation game framework allowing us to capture human social attributions that indicate trust in continued and interdependent human-agent cooperation. Within this framework, we experimentally examine the impact of two key dimensions of social cognition, warmth and competence, as antecedents of behavioral trust and self-reported trustworthiness attributions of intelligent computers. Our findings suggest that, first, people infer warmth attributions from unselfish vs. selfish behavior and competence attributions from competent vs. incompetent problem-solving. Second, warmth statistically mediates the relation between unselfishness and behavioral trust as well as between unselfishness and perceived trustworthiness. We discuss the possible role of human social cognition for human-computer trust.

No abstract

No abstract

No abstract

Abstract. Success in extended human-agent interaction depends on the ability of the agent to cooperate over repeated tasks. Yet, it is not clear how cooperation and trust change over the course of such interactions, and how this is interlinked with the developing perception of competence of the agent or its social appearance. We report findings from a humanagent experiment designed to measure trust in task-oriented cooperation with agents that vary in competence and embodiment. Results in terms of behavioral and subjective measures demonstrate an initial effect of embodiment, changing over time to a relatively higher importance of agent competence.