How should we talk someone out of faith and superstition and into reason? These are two distinct matters. Superstitions are beliefs largely to do with how we make sense of the world, and psychological research has shown pretty plainly that most people under most circumstances favor what are called natural attributions to superstitious ones. (Exceptions to these circumstances go beyond the scope of this question to address here, and religious fundamentalism is only one of them.) This implies that superstitions are erroneous conclusions about the world held mostly for lack of better ones, and overcoming them is often a straightforward matter of providing a better explanation for the phenomena in question. 'Better,' here, might mean more accurate, less presumptive, more elegant, or a number of other things that make an explanation deserve to be called 'good,' and usually this approach, which we call education, is sufficient for the job.
This, however, is not always the case; education is not always enough. Some superstitious beliefs are maintained by faith, and faith complicates the matter by making its victim less willing to replace certain superstitious beliefs about the world with natural ones. Since many superstitions are easily overcome with education, and the particularly sticky ones are glued down with faith, talking someone out of faith, not superstition, becomes the central objective in nearly all the cases where it really matters. This can be very hard to do.
There is only one way that I am aware of that someone can be talked out of faith and into reason: introducing doubt. Doubt is the feeling that accompanies becoming aware of a lack of conviction, and faith and conviction walk hand-in-hand. Thus, to break conviction is to crack the edifice of faith, and once that wall admits gaps, the superstitions guarded by it stand a chance of being revised or replaced, as is warranted by the best available knowledge that we have. The question, then, as we should be concerned with practicality here, is how we crack open the tough nut of faith and introduce doubt.
There are times where the world, or clever arguments of certain kinds that challenge someone's assumptions, buffet someone's faith enough to break it open and awaken doubt from within. These, though, represent a special case of a very general answer to the question of how we cultivate doubts in the otherwise certain, by raising questions. Doubt follows questions that the belief system at hand cannot answer. Unanswered, yet interesting, questions instill doubt, and doubt corrodes faith. At that point even superstitious beliefs maintained by faith can be replaced with better ones.
When the world, or even clever arguments, split someone's faith, it is because the experience at hand, be it a tragedy like a death or the sudden introduction of an idea that had never before been considered, raises questions in the mind of the newly curious doubter. We should not rely upon these methods alone, though. A better approach, if we hope to talk someone in the direction of better beliefs, is simply to ask them questions. We should ask honest questions, whose answers may sway us too, until the people with whom we speak are left in want of an answer they do not have. Then there is doubt, and then there is the chance to help someone break the spell of faith and begin the process of belief revision. 'That's an interesting thought, but how might it go wrong?' is a general model of an honest question that encourages critical thinking and intellectual honesty.
We must, however, be responsible with this method and be aware of how it can be misused. Questions, simply, are dangerous things. The reason is familiar to us all: doubt evokes vulnerability. The same space that allows belief revision leaves us vulnerable to it, and that vulnerability is uncomfortable and easily exploited. We see this manipulation frequently from people wishing to spread their beliefs—they may ask, 'how do you explain this or that without these particular beliefs, myths, or superstitions?' At the moment of doubt, an evangelist swoops in with an answer, exploiting the vulnerability, pretending to be educating, and this is despicable. To help talk someone into reason, then, we must avoid this kind of manipulation, spurning the desire to answer questions for which we only pretend to know the answers, and trying to help the people we question with their vulnerability, not stuff it with what we want them to believe. Modeling intellectual honesty, especially when it means saying that we don't know something, is a critical part of this process.
How, then, do we talk someone out of faith and superstition and into reason? First, we must ask honest, fair questions that introduce doubts by taking someone who is sure to where they are unsure. Second, we must be ethical with that vulnerable space of doubt and guard it as if it were something that deserves to be treated as if it were sacred. Third, we must aim to help educate, which requires the honesty not to try to teach things we, ourselves, only pretend to know. This requires authenticity and integrity, so we should cultivate those if helping people this way is our goal. We must also be patient and understanding that this process, though it can sometimes be quick, is often very slow and even painful, as many kinds of beliefs held in faith are held that way for powerful emotional, psychological, and cultural reasons that will take time to be dealt with. All throughout, though, honest questions asked authentically and honest answers given when appropriate are the surest way to talk someone from faith to reason. —James A Lindsay