In this paper I argue that the animacy scale proposed by Silverstein (1976) is one of the factors that explains some contrasts among the arbitrary se constructions in Romance and Slavic. I propose an Optimality-Theoretical implementation of this hypothesis, following Aissen's (1999, 2000) approach to Silverstein's scale. To account for a reversal of the normal markedness implications in the arbitrary se constructions I extend OT's notion of faithfulness, proposing a universal ranking of faithfulness constraints that is sensitive to the animacy scale.