["Mind &Language, EarlyView. ", "\nSome philosophers have recently proposed an error detection condition (EDC) for representation, such that for R$$ R $$ to be a representation for system S$$ S $$, S$$ S $$ must be capable of detecting errors in tokenings of R$$ R $$. We argue that this condition is unmotivated, and that it is too strong. We show that theories of representation that are committed to the EDC will fail to capture cases of representation proposed to account for visual illusions.\n"]