Although some research has incorporated human cognition into the design of intelligent agents, few studies examine how agents' understanding of intentions and emotions affects collaboration. This study employed a scenario-based questionnaire with 109 participants, using a structural equation model to analyse how intelligent agents' intention and emotion understanding ability (literal vs. perspective-taking) influence trust and use intention. Results showed that intelligent agents with a higher understanding ability increased user perceptions of their intelligence and empathy, which led to greater trust and use intention. Unlike the 'uncanny valley' effect, perspective-taking did not increase negative attitudes towards the intelligent agents. The agent's enhanced understanding ability increased users' use intention via two routes: by improving perceptions of intelligence and hence increasing use intention, and by enhancing perceptions of empathy and hence improving trust and further boosting use intention. These findings provided practical guidance for designing intelligent agents with improved cognitive capabilities.
Peng et al. (Fri,) studied this question.