|Abstract:||There is a well-known asymmetry in children's pattern of acquisition of the binding principles A and B. Children correctly interpret reflexives from the age of 3:0, but they continue to perform poorly on the interpretation of pronouns even up to the age of 6;6 (Chien & Wexler, 1990; Grimshaw & Rosen, 1990). At the same time, data from child language production shows that children correctly produce both pronouns and reflexives from the age of 2-3 (Bloom et al., 1994). Current explanations of this delay in comprehension have either rejected the comprehension data outright (e.g., Bloom et al.) or have tried to argue that the problems should be accounted for outside the grammar (e.g., Reinhart, 1983; Chien & Wexler, 1990; Grodzinsky & Reinhart, 1993). We account for the experimental evidence by instead arguing that children acquire the ability to optimize bidirectionally relatively late. It is this type of optimization, which involves reasoning about alternative forms, that is necessary for correctly interpreting pronouns.