When Digital Human Avatars Emote, Users Pull Back: The Hidden Cost of Expressiveness in Trust and App Engagement

ABSTRACT Although virtual agents are being used more and more in our everyday lives, one barrier to effective interaction with digital human avatars is the way they express emotions. While emotions such as happiness, sadness, anger, fear, and disgust are central to human communication, when displayed by avatars they may undermine rather than enhance trust. Researchers have thus been exploring ways to endow virtual agents with human characteristics, such as the ability to express humanlike emotions. However, very little research has been done on how humans react to virtual agents displaying emotions. This paper investigates how emotional expressiveness in digital human avatars affects users' trust and subsequent intention to use apps across health and tourism contexts. We conducted three experimental studies to test the effect of the expression of emotions on trust toward virtual agents: specifically digital human avatars. We find that—contrary to generalized belief—the expression of emotions has a negative effect on the intention to use an app featuring a digital human avatar. Moreover, we demonstrate that this negative impact on the intention to use such an app is mediated by the reduced trust in the emotional digital human avatar. Finally, accounting for consumers' individual differences, we provide evidence that this mediated relationship is further moderated by some individuals' tendency to anthropomorphize nonhuman agents. Specifically, the negative indirect effect of digital human avatars' emotionality display on intention to use an app through experience and trust drives intentionality only for individuals with lower anthropomorphism tendencies.