This paper explores a distinction between two types of response- dependence (RD) account (shallow vs. deep). This distinction is inherent in much of the existing literature, however it is neither widely nor well understood, and has never been drawn explicitly. The distinction is often taken to be a metaphysical, or ‘realism-relevant’ one—i.e. deep RD accounts entail qualified realism (or perhaps anti-realism), while shallow RD accounts are metaphysically neutral. I argue that the distinction is not reliably realism-relevant. I formulate a weaker (...) version of the distinction that may help prevent some common and understandable confusion about RD biconditionals and their relationship to realism. The weaker distinction rests on the different roles assigned to RD biconditionals by the two types of account. (shrink)