Saturday, March 07, 2009

Elliot Sober reviews Hume's Abject Failure

Earman's critique of Hume is highly regarded, please note that Earman is not a Christian apologist. (Neither was Patrick Maher, my philosophy of science teacher from University of Illinois who advised my papers on miracles). Sober suggests a more modest employment for Hume's argument.

In spite of William Lane Craig's enthusiastic employment of Earman's critique of Hume against Ehrmann (two similar names!), a Bayesian version of Hume's argument could be used, it seems to me, to undermine some of the more audacious claims made by Christian apologists.


exapologist said...

You might also be interested in seeing Michael Jacovides' (Purdue) review of three recent books on Hume on miracles in The Philosophical Review117 (2008):

David Johnson, Hume, Holism, and Miracles
Ithaca: Cornell University Press, 2002. ix + 106 pp.

John Earman, Hume’s Abject Failure: The Argument Against Miracles
Oxford: Oxford University Press, 2000. xi + 217 pp.

Robert J. Fogelin, A Defense of Hume on Miracles
Princeton: Princeton University Press, 2003. xii + 101 pp.

Brandon said...

In general I've found Peter Millican's work on this to be very good. His "Hume, Miracles, and Probability" is fairly decent as a reconstruction of Hume that handles much of Earman's complaint.

A fundamental problem, really, is that Hume himself is not, in fact, a Bayesian, and almost certainly does not have what we would call a standard probability theory; virtually no discussion of miracles before the nineteenth century makes use of a standard theory of probability (in part because it hadn't completely settled down yet). The paper that I think probably comes closest to accurately capturing Hume's view of probability in this context is:

Barry Gower, David Hume and the Probability of Miracles


Thanks for linking Jacovides' review; I seem to have missed it when it came out.

Perezoso said...

That's a subtle Ad Hom. It's not so much about Bayes--which is just a fairly trivial equation about conditional probability (and Hume did touch on probability)--but about the evidence itself; indeed in most normal applications, the evidence exists, has been confirmed (like in the Cookie example. google er). Hume's point would preclude the unconfirmed, inconsistent ancient testimony from being inputted. The New Testament is a book of parables and wisdom (at least on occasion)--it's not a case study.

Ancient, inconsistent reports of supposed supernatural events do not count as data, except to theologians. The same would hold for say UFO sightings: are all reports of UFO sightings to be considered reliable, and thus can be inputted into BT, and that shows likelihood? No. There's a verification issue as well (which Hume was aware of): given bare bones positivism, ancient historical events (say in Tacitus) cannot be verified, and thus we cannot say they occurred with any degree of confidence--that's the case a fortiori with religious dogma.

Used correctly (ie theology-free hypothesis) Bayes shows miracles or the Resurrection to be highly unlikely anyway (probably like under .001%), given say 2000 years or so, and that's even if we allow a bit of padding (say, allowing zombification along with Rez.: the power of Truth cannot be denied!)

Brandon said...

Hume did touch on probability; but as I said the probability is nonstandard, like Bernoulli's: for instance, it allows probabilities to sum to greater than 1, and there was no standardized conception of conditional probability. No one would have been using what we consider standard probability theory at the time. Bayesian rewrites of Hume are fine, but they are precisely that: different arguments, not Hume's.

Hume's point does not prevent unconfirmed, inconsistent testimony of any sort from being input; precisely his point is that it doesn't matter even if you put it in, even if you count it as data. And he explicitly denies your historical claim; not surprisingly, since he was among other things a historian and regularly said that historical events had occurred, with considerable confidence.

I am glad you have managed to figure out that resurrection from the dead is an improbable event.

Brandon said...

By the way, I think your percentage is probably not conveying the idea you intend it to convey. Saying that it is "probably under .001%" makes it sound like .001% is at least in the ballpark, even given padding; but even .0001% probability (one more decimal place) for resurrection of the dead would mean that any given person is massively more likely to rise from the dead than hold a ticket that wins the Powerball jackpot, and that it is almost certain that tens of thousands of people have risen from the dead in recorded human history, which I don't think is what you intended to convey. In the sort of context you're talking about, .001% is a very large probability.

Perezoso said...

The Sober sorts miss the point (usually intentionally): Hume's arguing for secularism, and wants to show the fallibility of scripture (as others had, even Spinoza). To do that, he does make use of what we might call common-sense empiricism or Jefferson's "public reason." (Jeff. btw also denies miracles, the Rez., and excised the Book of Rev. from his abridged good book).

So Hume's argument more Newtonian, and deterministic (even if not fleshed out that way) than probabilistic. The uniformity of experience--our perception of nature, which is Newtonian, and regular, by and large--precludes the dead coming back from life, or Beasts with 7 heads (even if some hottie Jezebel's on its scaly back), or Moses' Cecil B Demille stuff. It doesn't really matter if there are 10 reports, or 1000 (anyway, there were only a few reports, but those were transmitted over decades, and there are inconsistencies, and mistakes. So reliability still an issue, Hume or not).

Anyway, as some pagan pointed out on another thread on DI, why not also calculate the probability of other ancient reports of miracles from other religions-- say whether the Buddha levitated? Many ancient reports, and tradition, hearsay evidence, etc. Indeed, that's what this is in legalistic terms: hearsay, not even circumstantial (as would be say, an apparently legitimate film of Maria appearing over the waves, or something like that. ).

Perezoso said...

even .0001% probability (one more decimal place) for resurrection of the dead would mean that any given person is massively more likely to rise from the dead than hold a ticket that wins the Powerball jackpot,

No. Nyet. That's where the confusion starts. Powerball exists. Powerball winners exist, even if one in 10 million or something. Probability applies to powerball: someone will win. The "event evidence" exists (people with tickets). Probability doesn't apply, at least in same way, to reports of events (especially supernatural events) which have not been proven to have occurred. You are assuming any likelihood means it occurred at least once in a million years or something, when it doesn't--that's a misuse of probability and BT.