Virtual humans as social actors: Investigating user perceptions of virtual humans’ emotional expression on social media

Virtual influencers, digitally-created characters with realistic human-like appearances, have emerged as a distinctive type of virtual humans prominently featured in various virtual environments, with a notable presence on social media. They emulate human behaviors, including displaying emotions, to enrich their narratives. Yet, limited research has assessed how users interpret these virtual influencers' emotions as well as whether these emotional cues affect users' perceptions and attitude towards these virtual entities. To address these research gaps this research draws upon the Computers as Social Actors (CASA) paradigm. In study 1, sentiment analysis of a life-like virtual influencer's Instagram posts (N = 1196) conveyed a higher frequency, proportion, and intensity of positive (vs. negative) emotions. Study 2 utilized a U.S.-based Generation Z and Millennial sample from a large online platform to conduct a between-subjects online experiment (N = 142), examining whether Instagram posts conveying a virtual influencer's emotion (vs. no emotion) impacted users' attitude towards the virtual influencer. Findings indicate that the virtual influencer's perceived emotional intelligence, anthropomorphism, and authenticity are significant and serial mediators of the relationship between the virtual influencer's display of emotions in the social media posts and users' attitude towards the virtual influencer. Further, when looking at individual emotional effects (happiness vs. sadness vs. love vs. lust vs. no emotion), the expression of happiness in the virtual influencer's social media posts is greatly recommended for high perceptions and a positive attitude towards the virtual influencer while the display of sadness and lust are recommended to be avoided.