Abstract: Our best formal theories of belief are committed to logical omniscience, whereby agents are characterized as believing all logical consequences of what they believe. Logical omniscience is a problem if we want to reason and make predictions about the doxastic lives of agents who---much like ordinary human beings---are not logically omniscient but nevertheless logically competent. Intuitively, while we cannot expect such agents to believe all logical consequences of what they believe, we can at least expect them to believe the obvious, immediate, or easy ones. In the paper, we argue, first, that standard responses to the problem of logical omniscience fail. Second, we argue that there is no "conservative fix" to the standard responses. Rather, we need to rethink what a proper solution to the problem of logical omniscience should look like. Finally, we present a new formal framework that gives a proper solution to the problem.