<p>The paper addresses the issue of “epistemic permissivism”: does epistemic rationality ever permit more than one doxastic attitude to some proposition, given some body of evidence? One approach has taken up William James’s idea that there are different ways of weighing our two central cognitive goals as believers: Believe truth! Shun error! This motivates an affirmative answer to the question: agents with the same evidence can rationally come to different conclusions about some proposition because they weigh the importance of the goals differently. I examine the consequences of formulating the “Jamesian" argument using epistemic utility theory. This more precise formulation turns out to predict the permissibility of different incompatible doxastic attitudes not just across agents, but for a single agent. That is, we get the possibility of intrapersonally permissive situations, where more than one attitude is rationally open to a single agent. The apparent arbitrariness of rational belief on this picture calls, or so I argue, for serious consideration. I suggest that looking to practical rationality and the practical aspects of belief can explain (away) the arbitrariness worries.</p>