While optimization frictions have been shown to attenuate earnings responses to financial incentives, less is understood about the individual factors shaping the response. The main contribution of this paper is to separately quantify the role of learning the tax and benefit schedule versus other kinds of frictions. A unique combination of notches in the tax and benefit schedule and an information policy in a Norwegian welfare reform facilitate our study. The presence of notches allows us to measure overall frictions. Quasi-random assignment of a letter targeting misperceptions about the slope and locations of benefit phase-out regions allows us to pin down the role of information. Our analysis delivers two main findings. First, about 50% do not behave as predicted by standard labor supply models, and optimization frictions are particularly prevalent when financial incentives change. Without adjusting for these overall frictions, estimated elasticities would be attenuated by at least 70%. Second, the observed elasticity among those who receive the information letter is at least twice as large as among the non-informed, suggesting governments can partly offset the attenuation with information policy. Our calculations suggest misperceptions of the tax and benefit schedule account for two-thirds of the attenuation in earnings responses to financial work incentives. The findings have important implications for the effectiveness of tax and transfer policy.