A rationally inattentive agent processes information by balancing benefits and costs of attention (Sims [1998, 2003]). Predicting the agent’s behavior therefore requires a measure of her attention costs, but these costs can be dicult to identify as they incorporate hidden factors - such as time, eort and cognitive resources - that are not directly observable. In this paper, we identify attention costs by characterizing the implications of rational inattention for choices over opportunity sets, where the trade-o between benefits and costs are revealed by attitudes towards flexibility (Kreps [1979]) and temporal resolution of uncertainty (Kreps and Porteus [1978]). Exploiting this connection, we provide a procedure to elicit attention costs using willingness-to-pay data, and indicate how such data could be obtained in dynamic choice environments (e.g., consumption-saving problems) or generated in experimental settings.