I began writing this as a comment on this blog post, but it quickly achieved a length that I am pretty sure would seriously violate blog comment etiquette. So, I am posting it here. But, as it is a reply to the aforelinked post, it may be a good idea to read that before checking out what I have to say.
To very briefly summarize, Andrew Bremmer suggests that perhaps Hume’s various claims/positions in “Of Miracles” are contradictory in the sense that Hume equivocates or otherwise alternates between different ‘central’ theses about miracles. While I think it is fair to charge the discussion with being somewhat slippery, I don’t see it as an act of interpretive acrobatics to avoid treating Hume as contradicting himself. And to frame my response in terms of Andrew’s central questions: I interpret Hume’s argument in “Of Miracles (Part 1)” to be an argument against the possibility of any testimony (no matter how strong) serving to justify belief in a miracle, and also establishing the condition which would need to be met for testimony to justify belief in a miracle: namely, for it to be sufficiently more contrary to one’s experience that the testimony is innaccurate than that the attested event occurred. As I read him, Hume clearly does not think that any testimonial evidence for a miracle can be stronger than the experiential evidence against it (evidence one is guaranteed to have in virtue of the event in question being a miracle). Hume then proceeds, in “Of Miracles (Part 2)” to argue that the actual situation for purported miracles is much worse than it may have seemed from the discussion in part 1, since no purported miracle attested in actual human history enjoys testimonial evidence anywhere near as strong as that which was supposed for purposes of part 1. Let me also be clear up front that I am not (at least not in this post) defending the substance of Hume’s arguments; I don’t intend to defend the principles Hume invokes in offering his argument, but simply to draw out what I take to be his argument.
First, we get some Humean epistemological preliminaries (pertaining to experiential reasoning):
1) A certain (relatively high degree) of uniformity in experience is required to justify belief in an unobserved matter of fact.
2) The evidential value of testimony is based on past experience of testimony being correct.
3) Most testimony provides only probability (rather than proof).
4) In addition to circumstantial factors that weaken the evidential value of a particular piece of testimony (such as when there are conflicting reports, when the testifier has an interest in what they report, the tone of their voice, etc.), there are content-based factors that undermine the evidential value of a particular piece of testimony. For instance, a normal case would be where a generally reliable witness tells me that they are serving hamburgers at the campus dining hall. Since, in my experience, campus dining halls frequently sell hamburgers, this testimony has pretty good evidential value. If, on the other hand, a generally reliable witness were to tell me that they were serving foie gras in the campus dining hall, the (moderate) incredibility of a campus dining hall serving foie gras would serve to undermine the evidential value of their testimony.
It will be important for Hume later (in the actual discussion of miraculous testmiony) that the very same thing which underwrites the evidential value of any testimonial evidence (namely, experiential evidence of a constant – or frequent – connection between the evidence and the fact) is the thing underwriting the incredibility of the proposition being attested.
We can extrapolate this definition of Marvelous testimony:
MAR: Testimony that P is marvelous (for S) just in case P is neither contrary nor conformable to (S’s) experience.
As Hume applies this to the Indian prince and the marvelous testimony of frozen water, we get something like the following:
Hume maintains that, holding fixed circumstantial factors, Marvelous Testimony has lower evidential value than Mundane testimony (where “mundane” is a term I am introducing to pick out the category of testimony that is neither miraculous nor marvelous).
So, supposing that there is a witness who is generally reliable, it is possible that, were he to report to the prince that there is a river located X miles to the north and Y miles to the west, this testimony would be good evidence of the existence of such a river (meriting some degree of assent to the proposition attested), while, were he to instead report to the prince that the water in that river was sometimes solid and could be walked across, his testimony would not be good evidence of the existence of a river whose water was sometimes solid and could be walked on.
However, if there were additional witnesses, etc., it would be possible to get sufficient evidence to merit assent, since the proposition is simply beyond the prince’s experience, rather than at odds with it.
It is after this discussion that Hume turns to miracles. We can define miraculous testimony and a maximally credible witness:
MIR) Testimony that P is miraculous (for S) just in case P is entirely contrary to (S’s) experience; that is, just in case (S’s) experience is itself a proof against P.
WIT) A witness W is maximally credible (for S) just in case W’s testimony that P is a proof (for S) of P.
The case Hume is interested in, in part 1 is this:
MT) Suppose that there is a witness W, such that:
A) W is (to S) a maximally credible witness.
B) W testifies (to S) that M and M is miraculous (for S).
Hume’s positions on the case seem to be the following:
1) Since M is contrary to (S’s) experience, there is a maximally strong proof against M.
2) Since W is maximally credible, there is a maximally strong testimonial proof of M.
3) It is impossible that the the maximum strength of a testimonial proof is greater than the maximum strength of proofs, so, at best, the strength of the proof for M would be equal to the strength of the proof against M.
4) For the proof of M to be as strong as the proof against M would require that M being wrong is so contrary to one’s experience that one’s experience provides a proof against M testifying to something that is not true.*
5) The strength of support for M or for ~M is equal to the difference in strength of the proof for M and the proof against M.
6) So, the strength of support for M on the basis of W’s testimony is never positive (given 3).
Thus, Hume reasons, there is no testimony strong enough to warrant belief in a miracle.
*Note that this is not simply a situation where W being wrong would be extraordinary or marvelous, but a situation in which it would be miraculous (on Hume’s definition) for W to be wrong. It is possible that Hume is thinking, due to the subordination of testimonial evidence to experiential evidence more generally, that no witness is so credible that their inaccuracy would legitimately be a miracle.
Now, there are a lot of potential places to criticize Hume’s position (for instance, it is not at all clear how Hume can really cash out the marvel/miracle distinction so that it legitimately does the work he needs it to). I do not mean to subscribe to the arguments/principles Hume is advocating (though obviously some of them seem to be at least prima facie plausible), but I do think that it is clear that Hume’s is (in part 1) arguing against the very possibility of establishing a miracle on the basis of testimonial evidence. The end result is that the best support testimony can offer for a miracle is quite weak, and the main point of part 2 is to drive home the empirical claim that no miracle attested by any religion whatsoever approaches the strength of support hypothetically granted in part 1.