The spread of cultural variants, such as dress or speech patterns, may be promoted or inhibited by different types of bias. In model-based bias, variants are differentially adopted according to characteristics of individuals exhibiting them. Based on a hypothesis from sociolinguistic fieldwork, we posit two types of model-based bias: bias associated with alienable traits such as “tough” and inalienable traits such as “male.” We tested this by conducting a laboratory experiment in which participants played a computer game using an artificial language with two different dialects. Players were significantly more likely to borrow features of the other dialect when the variation was explicitly associated with toughness than when it was associated with another alien race. This suggests that cultural variants linked to alienable traits are more likely to be adopted than those linked to inalienable ones, even if the practical implications of the two traits are very similar.