Among the current crop of atheist philosophers, Graham Oppy is one of the best they've got. So it's useful to see him summarize his case against the credibility of reported supernatural encounters. The argument doesn't get any better than this:
First, there is no question that the history of reports of encounters with supernatural beings and forces is, at least in very large part, a history of fraud, gullibility, deception, stupidity, ignorance, and so forth. Second, there is no serious doubt that there is at least good prima facie reason to believe that there is a huge panoply of supernatural beings whose existence would be vindicated by the recorded supernatural experience of humanity if the existence of any supernatural beings was vindicated by that recorded supernatural experience. Third, it is quite clear that the joint effect of these first two points is to raise serious questions about the evidential worth of any reports of experiences that are claimed to be of, or directly caused by, supernatural agents. Fourth, it may well be that, in the absence of defeating considerations, it is the case that p (cf. Swinburne 1979). But, as we have just noted, there is no serious doubt that there are very weighty candidate defeating considerations in the case of "seemings" that are tied to the supernatural.
In the absence of any independent support for belief in gods–i.e., support founded in something other that reports of experiences that have been taken to be of, or directly caused by, gods–there is clearly reason to prefer the uniform treatment of reports of supernatural experiences that naturalism affords to the non-uniform treatment of reports of supernatural experiences that is required by any developed version of theism. Graham Oppy, "Arguments for Atheism," S. Bullivant & M. Ruse, eds. The Oxford Handbook of Atheism (OUP, 2013), 67.
By defining "encounters with supernatural beings" as "experiences that are claimed to be of, or directly caused by, supernatural agents," I take it that he's using a definition broad enough to cover diverse phenomena like apparitions, miracles, precognition, and answered prayer. With that in mind, let's run back through his objections:
1. Regarding the first claim:
i) What does he mean by "in very large part"? Does he mean most reported supernatural encounters reflect a history of "fraud, gullibility, deception, stupidity, ignorance, and so forth"? If so, what's his quantitative evidence for that assessment? What's the sample group? How representative is the sample group?
ii) What's the intended distinction, if any, between "fraud" and "deception"? Does Oppy uses those as synonyms?
iii) Wouldn't the primary motivation for fraud be cases where appeal to supernatural encounters is used for personal or institutional gain? To lend credence to a new religion or a new dogma? Maybe a career booster (e.g. faith-healer)?
iv) Even when supernatural encounters are invoked to attest the message or the messenger, the ostensible witness may have something to lose rather than something to gain. What if his claim exposes him to predictable persecution? Wouldn't that be a disincentive to make fraudulent claims about supernatural encounters? So we need to draw that distinction in assessing the credibility of the witness.
v) Just to play it safe, suppose, for the sake of argument, that we discount the subset of reported supernatural encounters where there might be an incentive to deceive or commit fraud. That leaves "gullibility, stupidity, and ignorance." However, gullibility, stupidity, and ignorance aren't distinctive to reported supernatural encounters. That, therefore, would not be a specific reason to doubt reported supernatural encounters. There are gullible, stupid, and/or ignorant witnesses to everything under the sun. That, however, is not a reason to doubt testimonial evidence in general. Indeed, Oppy's claim about "a history of fraud, gullibility, deception, stupidity, ignorance, and so forth" is, in itself, dependent on historical testimony. Therefore, his objection would be self-refuting if he were propounding skepticism about testimonial evidence in general. So, at best, his skepticism is only warranted in reference to cases where we might suspect fraud and deception, even assuming that fraud and deception are more prevalent in reported supernatural encounters.
vi) Of course, fraudulent and deceptive reports are hardly unique to reported supernatural encounters, so Oppy needs more discriminating criteria to justify his skepticism about reported supernatural encounters, in contradistinction to other kinds of fraudulent reports.
vi) What about reported supernatural encounters where personal or institutional gain is not a plausible motive? Take answered prayer or a miraculous healing. A witness might share that experience with a small circle of friends and family. He (or she) doesn't do that for social advancement. Doesn't do that to start a new religion. He simply wants to share his marvelous experience with friends and relatives. He's so thankful and overawed by his experience that he wants other people to know how wonderful God is. He can't contain himself. He doesn't do it to become the founder of a new religion, or kickstart a career as a prophet or faith-healer. It may be a once in a lifetime experience.
This isn't just hypothetical. Rather, it's commonplace if you move in religious circles.
vii) Perhaps Oppy would object that a primary function of reported supernatural encounters is to authenticate religious claims. But even though that's true, we can, for the sake of argument, take those examples off the table because we don't need to include them to test Oppy's claim. Oppy rejects reported supernatural encounters in toto. So even if we bracket the subset of reported supernatural encounters that serve to validate religious claims, that leaves us with an enormous margin for error, given the remaining reports that don't fall under that rubric.
2. Regarding the second claim:
i) It's hard to see how that's supposed to be an argument for atheism. For Oppy fails to explain why that would be an unacceptable consequence. On the face of it, his objection seems to be circular: once you credit reported supernatural experiences, you open the door for the existence of supernatural beings! Okay, but how does that consequence undercut the credibility of reported supernatural experiences?
ii) Perhaps, though, he's attempting to cast a dilemma for supernaturalists. Perhaps he means that swings the door wide open for every supernatural claimant. For instance, Christians have no problem with supernatural beings like God, angels, and demons. Some might also make allowance for the existence of ghosts or poltergeists. But perhaps he means that once you open the door a crack, you can't shut out the whole "panoply" of candidates, viz. Zeus, Thor, jinn, genies, elves, wood nymphs and water nymphs, trolls, leprechauns, fairies. If you credit any supernatural being, you must credit them all. If that's what he means, I'd say the following:
a) We need to distinguish between the ostensible experience and how that's interpreted. For instance, suppose pagans experience supernatural beings. However, they then create a backstory about the supernatural being. A story about the origin, abode, and social life of the supernatural beings in question. That backstory is not a part of the encounter. They didn't experience the backstory. Rather, they created a narrative to explain where the supernatural being fits in their world. Likewise, once a society has developed a mythology for experiences of this type, people in that society automatically classify their experience according to the available cultural categories.
To credit the underlying experience doesn't commit you to the backstory, since that's not given in the experience itself. That's a cultural overlay.
b) Apropos (a), this means you can have a multiplication of categories for the same thing. Different cultures will have different names, categories, and narratives. That doesn't imply that there's actually a different supernatural being for each cultural classification. Suppose for the sake of argument that there are really just six different kinds of supernatural beings. Yet there might be a "panoply" of supernatural beings in comparative mythology and comparative folklore, even though these are actually reducible to our half-dozen different kinds of supernatural beings. Different cultures will develop their own folkloric classifications. That gives the appearance of a "panoply" of supernatural beings, yet that's not because we're combining different entities, but because we're combining different cultural descriptions of the same kinds of entities. A poltergeist in one culture might be a goblin or gremlin in another culture. That doesn't mean there's a corresponding entity for each category. To take a comparison, different cultures have different mythologies for the same animals.
c) It's not even possible for some candidates to exist. Thor is a barely disguised personification of thunderstorms. Moreover, pagan deities like Thor are physical beings. It isn't possible for a finite physical being like Thor, even if he did exist, to control the weather. Likewise, there is no palace of the gods on the summit of Mt. Olympus. By the same token, Greek mythology treats wood nymphs and water nymphs as visible, physical beings. If they did exist, there'd be abundant evidence for their existence.
3. Regarding the third claim:
He uses the first two claims to support the third claim. The first two claims, in conjunction, constitute "defeating considerations". But having critiqued the first two claims, the third claim is unwarranted, while his fourth claim piggybacks on his third claim, which piggybacks on the first two claims.
4. Regarding his conclusion:
i) He acts as though his first two claims are sufficient to discredit any and every reported supernatural experience, without regard to the specific evidence in any particular case. But even if you grant his first claim, at best that's just a generalization. It hardly preempts exceptions.
And it's illicit for him to insist that you mustn't credit any supernatural being unless you credit every supernatural being. That's like saying you can't give credence to any reported seamonsters unless you give credence to all reports seamonsters. By that logic, you can't believe in giant squid unless you believe in Scylla and Charybdis.
ii) To say we should always prefer a uniform treatment begs the question. That's like saying we should automatically dismiss any and all reports of water flowing uphill. But sometimes water does flow uphill, because humans build water pumps. To demand a uniform treatment ignores the evidence in any particular case.
iii) The basic problem with Oppy's overall argument is that he's laboring to sidestep the duty to examine specific evidence on a case-by-case basis by invoking general considerations. Yet general considerations are, at best, inductive abstractions, based on a sampling of particular cases. You can't rightly use them to prejudge any particular case on pain of vicious circularity. Your generalization is only as good as your sample.
Oppy's entire argument becomes an exercise in intellectual evasion. He doesn't need to consult the evidence because he's concluded in advance that supernatural encounters lack credibility. But that's premature. That forecloses further investigation in spite of counterevidence.