Napoleon lost at Waterloo. Or so you believe. You believe it because, although not on-hand to witness the event, you suppose that you stand at the end of a long chain of events stretching back to Napoleon's actually losing at Waterloo. But consider how you would respond to the following: Imagine you become convinced that as a child you were secretly given a pill that selectively caused you to believe that Napoleon lost at Waterloo. Your parents, your pediatrician, FDA officials, all attest to administering this Napoleon pill.
Now as someone committed to maintaining true beliefs, what would be the responsible thing for you to do? Should you continue believing that Napoleon lost at Waterloo? Surely not. Your belief that Napoleon lost at Waterloo, it turns out, was not caused by the (alleged) fact that Napoleon lost at Waterloo, but by something altogether unrelated to that putative event. There's a reason that you believe it. There's no reason for you to believe it. That is, there's an explanation, but no justification. You should be agnostic about Napoleon's fate at Waterloo.
This little thought experiment, first introduced by the philosopher Richard Joyce, is meant to teach a lesson about the rationality of belief. Take any object of belief (call it P)--Napoleon lost at Waterloo; vaccines cause autism; angels exist. If we discover that the belief that P was not caused by the fact that P (but instead by some other means unrelated to the fact that P), then we should give up the belief that P, for we no longer have a reason to believe that P is true. Believing that P is irrational.
It's difficult to overstate the significance of this lesson. Joyce employs it in an attempt to show that believing that actions are objectively right or wrong is unjustified. According to Joyce, our moral beliefs were not caused by moral facts. Rather, they were caused by powerful selective pressures acting on early humans. Individual survival depended critically on cooperation, and cooperation depended critically on resisting cheating our neighbor, and such resistance depended critically on believing that cheating our neighbor was objectively prohibited. We can explain our moral judgments without once appealing to moral facts--just as we can explain your belief about Napoleon without once appealing to facts about Napoleon at Waterloo. Such facts might hold; they might not. But responsible "believers" should be agnostic.
But the lesson extends to practically any domain. Belief in God, the afterlife, our preferred scientific hypothesis; beliefs about what will make us happy, about who's trustworthy and who's not, about who'll make a good mate, and who'll not.
We think our beliefs stand at the end of a long justificatory chain extending back to the facts. But if contemporary social psychology has revealed anything, it's that very many of our beliefs (let alone our desires) are the result of Napoleon pills. Freud may have been comically mistaken about the nature of our underlying drives, but he was spectacularly prescient about how much of our psychology is driven by sub-rational forces. From the work of Daniel Kahneman & Amos Tversky to Jonathan Haidt to evolutionary psychologists, modern cognitive psychology has revealed a startling range of Napoleon pills. We believe what we believe--and do what we do--very often for reasons that have nothing to do with our internal justifications. We are, to borrow Haidt's apt metaphor, tiny riders atop giant elephants. We believe we're in control. But, really, most of our control is over the stories we tell ourselves.