Do Synthetic Avatars Need to Match Learners? Testing Belonging and Perceived Effectiveness

This experiment tested the impacts of demographic congruence between synthetic humanlike spokesperson avatars and learners. Among 437 U.S. working adults, congruence did not affect knowledge transfer, and effects on perceived effectiveness and belonging were confined to one subgroup: Heterosexual White men rated incongruent avatars as more effective yet reported lower belonging with avatars of the same race but different gender. No other subgroup showed significant effects. Behavioral avatar choices mirrored this pattern. These findings identify a precise intersectional boundary condition, suggesting that effectiveness ratings alone may mask inclusion gaps. Implications for corporate training and business communication pedagogy are discussed.