Implicit bias is generally seen as something we should be worried about. However, what’s the problem exactly? It is often argued that in cases of implicit bias there is a conflict between two attitudes, one attitude that is conscious and reflectively endorsed, and another that is non-conscious and automatic. In the talk I argue that this presentation seriously misrepresents what attitudes are and what is really discordant about having divergent results on implicit and explicit measures of attitudes. Instead, I propose that one should take a dispositionalist account towards attitudes and explain discordance in terms of the absence of certain expected manifestations of the attitude of belief. However, existing dispositionalist accounts do not explain why we expect beliefs to manifest themselves in certain specific ways. In the talk I will attempt to fill this gap, and defend a view on what is problematic about implicit bias.