New evidence from cognitive development suggests that action selection is contingent not only on ones own mental states (e.g. beliefs, etc.), but may also be affected by ones representation of the beliefs of others (Kovacs et al. 2010). Herein, I propose a preliminary computational model accounting for the general pattern of results in the recent study by Kovacs et al. that is broadly simulation-theoretic in nature, and briefly discuss its implications for computational cognitive architecture.