According to what I will call 'the disanalogy thesis,' beliefs differ from actions in at least the following important way: while cognitively healthy people often exhibit direct control over their actions, there is no possible scenario where a cognitively healthy person exhibits direct control over her beliefs. Recent arguments against the disanalogy thesis maintain that, if you find yourself in what I will call a 'permissive situation' with respect to p, then you can have direct control over whether you believe p, and you can do so without manifesting any cognitive defect. These arguments focus primarily on the idea that we can have direct doxastic control in permissive situations, but they provide insufficient reason for thinking that permissive situations are actually possible, since they pay inadequate attention to the following worries: permissive situations seem inconsistent with the uniqueness thesis, permissive situations seem inconsistent with natural thoughts about epistemic akrasia, and vagueness threatens even if we push these worries aside. In this paper I argue that, on the understanding of permissive situations that is most useful for evaluating the disanalogy thesis, permissive situations clearly are possible. Epistemologists have grown increasingly interested in the question how epistemic rationality compares to practical rationality (Berker 2013, Cohen 2016, Rinard 2017, etc.), but epistemologists have been asking the more general question how belief relates to action for a very long time-for at least as long as they've wondered whether, and to what extent, we can control our beliefs. According to the currently dominant view, beliefs differ from actions in at least this way: while we often have direct control over our actions, we never have direct control over our beliefs. On this view, just as we might cause ourselves to blush by thinking about something embarrassing, we might cause ourselves to believe (e.g.) that the lights are on by looking at the lights and turning them on (Feldman 2001). But on this view, we can't form the belief that the lights are on, or any other belief, by simply deciding to form it, the way we can (for example) raise our arms by simply deciding to raise them. On this view, if direct control over our beliefs isn't fully conceptually impossible, it's at least impossible for cognitively healthy people like you and me. Perhaps Bennett's Credamites can do it (1990), but they aren't functioning properly, and we can't do it without getting ourselves into a defective cognitive state like theirs. Thus, while cognitively healthy people often exhibit direct control over their actions, there is no possible scenario where a cognitively healthy person exhibits direct control over her beliefs. Call this thesis about belief and action the 'disanalogy thesis.'