Warmth and competence in affective companion virtual humans: differential effects on behavioral intention via hedonic benefits.

Generation Z increasingly engages with affective companion virtual humans not as tools but as partners. However, it remains unclear how social perceptions-specifically warmth and competence-privacy-related attitudes jointly influence their behavioral intention, and whether hedonic benefits mediate these effects. Drawing on the hedonic motivation system acceptance model and the stereotype content model, this study proposes an integrated model linking perceived warmth, perceived competence, perceived privacy risk and privacy cynicism to behavioral intention toward affective companion virtual humans, with hedonic benefits as a mediator. Using partial least squares structural equation modeling (PLS-SEM) on 348 valid responses from a Generation Z sample, the results show that: (1) perceived warmth, perceived competence and privacy cynicism are directly and positively associated with behavioral intention; (2) hedonic benefits partially mediate the effect of perceived warmth on behavioral intention (VAF = 36%), but play only a negligible mediating role for perceived competence (VAF = 15%); and (3) perceived privacy risk has no significant direct or indirect effect. These findings clarify the differential pathways of warmth and competence in affective companion virtual humans acceptance: warmth is associated with both direct and hedonic routes, whereas competence is predominantly associated with intention through direct instrumental pathways. The results also indicate that privacy cynicism can co-exist with continued use intention without diminishing hedonic experience. Practically, the findings suggest that enhancing emotional warmth and functional competence, while acknowledging users' privacy fatigue, is more effective than solely focusing on reducing privacy risk perceptions.