Miracles and Antecedent Probabilities
Victor Reppert responded succinctly but thoughtfully to my posting on ECREE (the principle that “extraordinary claims require extraordinary evidence”):
There is a sense in which I agree with the ECREE thesis, it is just that I don’t believe that there is any objective way of proving that one set of antecedent probabilities is rational and another is not. So what is “extraordinary” is just what your antecedent probabilities tell you is improbable.
I think that as you pull at the story of the founding of Christianity, as you play out the various scenarios all the way through, you end up thinking that none of the scenarios for what might have really happened fit the facts very well. They run into factual brick walls of one kind or another. The Christian story, IF you can get over the initial antecedent improbability of the miraculous, makes more sense of it that any other story does. I think there is no logical proof that a miracle cannot happen, since it is possible that God exists, and God is omnipotent. Further, I think that this miracle is one that God would have a fairly understandable reason to perform. So, given my prior probabilities, the evidence lifts the case for the resurrection at least over 90%. But I can’t prove [to] someone else that they shouldn’t have priors so low for the Resurrection that it never gets above 10 for them. Both can be rational, and that’s just life in the big city. (I also don’t think salvation is a matter of passing a theology test). If you don’t believe in the Resurrection, then I think there are a bunch of inconvenient facts out there that are hard to make sense of. But I think every philosophy has to deal with inconvenient facts.
Keith Parsons would probably call this Bayesian Balkanization.”
An advantage of the Bayesian approach to rationality is that it plausibly accounts for how evidence can lead to rational consensus from very different starting points. Since prior probabilities are conditioned upon background beliefs, and since it appears that different persons and communities might reasonably possess very different sets of background commitments, Bayesianism countenances a very considerable degree of latitude in the positing of antecedent probabilities. This feels right. Surely, it is reasonable to expect that people confronted with a claim C will estimate C’s prior probability in accordance with what the background of their entrenched beliefs says about claims like C. Thus, if I heard a report that Victor had become an ardent University of Arizona Wildcats fan, I would antecedently rate the probability of that report as very low, since I know the depth of Victor’s commitment to the Arizona State Sun Devils. On the other hand, someone ignorant of Victor’s sports loyalties would not rate the prior probability of such a report nearly so low. The same applies when the topic is not sports loyalties but, say, the Resurrection. Those with different sets of background beliefs will—completely reasonably—assign very different priors to such a claim. As Victor colorfully puts it, that’s life in the big city.
Yet the latitude the Bayesian approach allows does not mean license. You are not licensed to assign prior probabilities in a biased or arbitrary way. For instance, you cannot rate the antecedent probability of a given miracle claim as high and another as low just because you want to believe the one and not the other. To do so would be to engage in special pleading by imposing a much lighter burden of proof for a favored claim over one not so favored. Thus, if C and C* are both miracle claims, and if the evidence for C* is as good, or better, than that for C, it would not be right to accept C and dismiss C* just because we have arbitrarily set the antecedent probability of C* lower than that for C. I hope it is clear that the latitude that a Bayesian conception of rationality gives us in setting up our priors does not extend so far as to allow such blatant special pleading.
That some Christians might in fact be involved in such special pleading is the point of T.H. Huxley’s brilliant and now largely neglected essay “The Value of Witness to the Miraculous (1889).” Huxley points to the testimony of Eginhard, a high official in the court of Charlemagne and a personal friend and biographer of the Emperor. Huxley says of him:
There is excellent contemporary testimony not only of Eginhard’s existence, but to his abilities, and to the place which he occupied in the circle of the intimate friends of the great ruler whose life he subsequently wrote. In fact, there is as good evidence of Eginhard’s existence, of his official position, and of his being the author of the chief works attributed to him, as can reasonably be expected in the case of a man who lived more than a thousand years ago, and was neither a great kind nor a great warrior.
In his work “Translation of the Blessed Martyrs of Christ, Saints Marcellinus and Petrus,” Eginhard reports his first-hand observation of the miracles performed by the relics of these saints. He reports the cure of a paralytic nun who spent the night praying besides the relics. He also tells of the complete cures of two lame persons, one whose body was made straight rather than twisted and the other who
was also deaf and regained his hearing. Eginhard had to return to court, but he continued to receive reports from his staff about further miracles. They reported that the power of the saints’ remains had exorcised a demon from a girl. Prior to the exorcism, they reported, the demon would converse in Latin, but after the expulsion of the demon, the girl could not speak Latin, but only “the tongue of the barbarians.”
Would a modern, educated Protestant accept such stories of the healing efficacy of relics? Surely most would dismiss such stories out of hand. To these Huxley addresses a most embarrassing question:
If you do not believe in these miracles recounted by a witness whose character and competency are firmly established, whose sincerity cannot be doubted, and who appeals to his sovereign and other contemporaries as witness of the truth of what he says, in a document of which a MS. Copy exists, probably dating within a century of the author’s death, why do you profess to believe in stories of a like character, which are found in documents of the dates and of the authorship of which nothing is certainly determined, and no know copies of which come within two or three centuries of the events they record? If it be true that the four Gospels and the Acts were written by Matthew, Mark, Luke, and John, all we know of these persons comes to nothing compared with our knowledge of Eginhard; and not only is there no proof that the traditional authors of those works wrote them, but very strong reasons to the contrary may be alleged.
Perhaps, then, our modern, educated Protestant would accept that Eginhard did indeed observe apparent healings, but would then rush to explain these away. After all, paralysis and lameness are often psychosomatically caused, and the placebo effect would explain the apparent efficacy of the relics in alleviating such ills. Also, one of the lame persons was a professional beggar, so his claim to be lame and deformed of body may have just been an act (recall the Court of Miracles scene from Hugo’s Hunchback of Notre Dame). As for the exorcism, recall that Eginhard was reporting this one second hand, and who knows what some over-excited and superstitious 9th Century zealot might have seen?
Why, though, try to explain these reports away? Why not take them at face value? Surely, the reason is that the hearer of the reports rates the prior probability of their truth quite low. In that case, why not strive, pari pasu, to give naturalistic explanations of the Gospel miracles? Why, for instance, not explore a hallucination account of the purported post-mortem appearances of Jesus? The reason why the one set of miracle-claims would be immediately explained away and the other not seems, inescapably, to be that bias is at work. The Eginhard miracle claims are arbitrarily assigned a lower antecedent probability than the Gospel claims.
Alternatively, our Modern Protestant might accept the Eginhard miracle-claims at face value. Richard Swinburne argues in The Concept of Miracle (1970) that there is no reason why a compassionate God would not perform miracles of healing at various times and places, and so extra-biblical healing miracle reports, even those performed by practitioners of non-Christian religions, need not constitute a problem for believers. So, perhaps the way to remove the inconsistency is to accept all miracle claims if they are at least as well supported as the Biblical ones. However, such a broad-minded approach seems to be a wholesale repudiation of ECREE. Innumerable ancient texts report miracles with seemingly as much reliability as the Biblical reports. Are we really to accept all the miracles attributed to the saints or their remains in the vast hagiographical literature? Why not the report of Herodotus that a female apparition urged the Greeks to attack at Salamis “in a voice so loud that the whole fleet could hear it?” Why not Herodotus’ report that the God Pan accosted Philippides as he was running from Athens to Sparta before the Battle of Marathon? Innumerable such examples could be cited from the world’s literature. To accept all of them would be credulous indeed; to accept only a subset raises the question of arbitrariness in the assigning of priors.
Perhaps, though, there is something special about THE miracle—the Resurrection—that makes it particularly hard to doubt. Victor says:
I think that as you pull at the story of the founding of Christianity, as you play out the various scenarios all the way through, you end up thinking that none of the scenarios for what might have really happened fit the facts very well. They run into factual brick walls of one kind or another. The Christian story, IF you can get over the initial antecedent improbability of the miraculous, makes more sense of it that any other story does.
WHICH Christian Story? The Gospels give very divergent accounts of what happened at the empty tomb, and, as G.A. Wells points out in The Jesus Legend (1996), even the Synoptic Gospels widely disagree about the post-mortem appearances of Jesus. OK, well let’s just say that the “Christian story” is what all th
e accounts have in common, namely that Jesus rose from the grave, that the tomb was empty, and that Jesus was seen after his death by various persons (actually, the original text of Mark probably ended before recounting any postmortem appearances). Does this agreed-upon core square with the facts, leaving only the skeptic’s antecedent probabilities to stand the way of acceptance?
Well, if you remove nearly all of a story’s content it is pretty easy to get it to square with the facts. There isn’t much left for it to contradict. Remember too that the accounts of the “facts” were written by persons who already accepted the truth of the resurrection, and so, unless they were stupid or incompetent, they would tell the tale so that the “facts” would not conflict with the message, namely, that Christ is risen. Finally, Christians, even sophisticated ones like Victor, tend to take as “facts” what we skeptics regard as hearsay. What are the indisputable facts that, supposedly, all skeptical scenarios crash upon?