Can avatar affective valence determine whether virtual reality embodiment reduces implicit workplace ageism?

Workplace ageism disproportionately hinders older workers' employment access and organizational wellbeing. Existing interventions produce limited effects on implicit age bias, particularly among young men. VR embodiment of older adult avatars offers a promising alternative, though the role of avatar affective valence in shaping these outcomes remains poorly understood. We conducted a randomized, pre-post, between-subjects experiment (N = 107 young men, age range 20-30) in which participants embodied one of four VR avatars: positive (Albert Einstein), negative (Hassan Nasrallah), neutral adult, or neutral young. Implicit ageism (Age IAT) and self-esteem were assessed 1 week before and immediately after the VR session; willingness to hire a 60-year-old job candidate and explicit ageism were assessed post-intervention. Einstein's embodiment produced a significant medium-to-large reduction in implicit ageism (d = 0.65, p < 0.001); a smaller but significant reduction followed neutral adult embodiment (d = 0.36, p = 0.040). Contrary to hypotheses, negative embodiment did not increase ageism, but exploratory within-condition analyses suggested a significant decrease in self-esteem (d = 0.59) and a significant increase in implicit gender-career bias (d = 0.46). Willingness to hire was higher in the Einstein and neutral adult conditions than in the neutral young condition. Self-esteem did not moderate implicit bias change, and post-VR explicit ageism did not differ significantly across conditions, consistent with the implicit-explicit dissociation. These findings advance a meaning-based model of VR embodiment, provide controlled evidence that embodiment in a positively valenced avatar can reduce implicit workplace ageism, and document novel collateral risks of negatively valenced avatar exposure.