Understanding the Spirit vs. the Letter of Probability

My apologies for there being so little activity on Civitas Humana for a long while now. I have been insanely busy and stressed lately (both professionally and personally), and I have only recently been able to catch up with blogging.

I’ll begin pumping some fresh blood into Civ by discussing Bayes’ theorem, the resurrection of Jesus, and why I think that, even with a non-zero prior (which is still very, very low) for the resurrection event, Paul’s letters and the Gospels are far too weak of consequent evidence to offset more probable (naturalistic or mundane) explanations for the same data.

Over a decade ago (March 2006) secular New Testament scholar Bart D. Ehrman debated Christian philosopher and theologian William Lane Craig about the evidence for the resurrection of Jesus (the transcript of the debate can be read here). Overall, the debate left me with the impression that Ehrman made a better case for naturalistic or non-paranormal explanations being more probable than a veridical resurrection event, with regards to the origin of the resurrection belief among Jesus’ disciples and the first generation of Christians. But there was one area where I think Craig scored a technical, though relatively minor point against Ehrman (as will be discussed below), and this was with regards to how Ehrman was defining a miracle event and conflating prior probability with posterior probability.

Lowder’s Summary of Craig’s First Rebuttal

Jeff Lowder on the Secular Outpost has made a useful summary of Craig’s critique of how Ehrman was defining a miracle event, and the arguments Ehrman had presented (in earlier publications) for why historians cannot argue that a miracle is the most probable explanation for a past event. This summary can be read in Lowder’s post “William Lane Craig’s Critique of Bart Ehrman on the Probability of Miracles.” Since the ensuing discussion involves Bayesian reasoning, you can read my essay “History, Probability, and Miracles” for a basic overview of how Bayesian logic works, if you are unfamiliar with the theorem.

Lowder begins by listing two published statements by Ehrman, which were quoted by Craig during the debate (bolding is my own):

(1) “Because historians can only establish what probably happened, and a miracle of this nature is highly improbable, the historian cannot say it probably occurred.”
(The Historical Jesus, part 2, page 50)
(2) “Since historians can establish only what probably happened in the past, they cannot show that miracles happened, since this would involve a contradiction — that the most improbable event is the most probable.”
(The New Testament: A Historical Introduction, page 229)
In his response to these statements of Ehrman, Craig critiques his line of reasoning by arguing that Ehrman is conflating prior probability with posterior probability. The odds that a given individual may resurrect from the dead could, indeed, be very, very low. But if there is very, very good evidence that such a resurrection event has occurred, it may offset the low prior, and even outweigh alternative explanations, to degree such that Pr (R/B & E) > 0.5 (perhaps even by a wide margin, e.g., +0.9).
Here is what Craig states in his own words (bolding is my own):
“In other words, in calculating the probability of Jesus’ resurrection, the only factor he [Ehrman] considers is the intrinsic probability of the resurrection alone [Pr(R/B)]. He just ignores all of the other factors. And that’s just mathematically fallacious. The probability of the resurrection could still be very high even though the Pr(R/B) alone is terribly low.”
Lowder likewise offers his own interpretation of Ehrman’s two quotations, and here is what he states regarding the first:
“I am inclined to interpret (1) as the following claim: (1′) Pr(R/B) is so low that it is impossible, even in theory, for there to be sufficient evidence to confer a high final epistemic probability on R, i.e., Pr(R/B & E) > 0.5.”

I would argue that to describe a miracle as an event that cannot be probable, even in theory, one would need to assign its prior probability a value of zero. And this is the same conclusion that Lowder reaches, when he states (bolding is my own):

The only way to reconcile (1′) with BT would be to assign Pr(R/B) a value of zero. If Pr(R/B) = 0, then it follows from BT that Pr(R/B&E;)=0. So, on the basis of (1) alone, as Craig has quoted Ehrman, I think it is premature to assume that Ehrman ‘just ignores all of the other factors.’ Maybe he does do that, but the quotation provided in (1) doesn’t show that. What I can say is that either Ehrman ignores all of the other factors or Ehrman assumes that historians must assign Pr(R/B) a value of zero. If the latter, then I think that is false.”

With regards to Ehrman’s second quotation, Lowder briefly states:

“Turning to (2), I don’t have much to say, other than I think Craig is 100% correct when he says that Ehrman ‘Confuses Pr (R/B & E) with Pr (R/B).’”

My Thoughts on Ehrman’s Quotes

At the end of this essay, I will make some suggestions for improving Ehrman’s arguments (which in spirit I think are correct, even if they may be formally invalid at parts). That said, I agree with Lowder’s conclusions on both accounts, for at least three reasons:

1) Nothing is Completely Impossible (Aside from Logical Contradictions)

The first is that, inductively speaking, virtually no event in the empirical world has a prior value of zero, beyond logical contradictions. There is even a slight prior probability, however small, that the sun may not rise tomorrow, or that fire may suddenly cease to burn and be safe to put your hand in, or that we are all living in a matrix designed by extraterrestrials. I would assign all of these types of events very, very low priors, but none are impossible, in theory. As stated, the only things I that would argue are theoretically impossible are logical contradictions. There cannot be things like square circles, pink gold, or one plus one equaling three.

2) Miracles Need Not Be “The Most Improbable Event,” Even Intrinsically

The second reason I disagree with Ehrman deals with his definition of miracles in the latter quotation — namely, that a “miracle” is “the most improbable event.” This is because one could imagine a number of possible worlds in which miracles happened quite frequently (or at least not on an unprecedented basis). We could live in a possible world such as described in the Book of Exodus, in which the Red Sea suddenly parted to allow a group of people to pass through, food rained down from the sky, rivers spontaneously turned into blood, and darkness covered the earth for a period of multiple days (without any solar explanation). If these events further happened within some sort of religious context — such as a prophet predicting their occurrence, or their development favoring a specific religious group of people — I think we could further extrapolate that they were miracles that were happening as part of some sort of purposeful design, perhaps as even pointing to the truth of one exclusive religious faith.

On this point, I will provide what I think is a more helpful definition of a miracle event. The Cambridge Introduction to the Philosophy of Religion (pg. 208) defines a “miracle” as follows:

“An event (ultimately) caused by God that cannot be accounted for by the natural powers of natural substances alone. Conceived of this way, miracles don’t violate the laws of nature but rather involve the occurrence of events which cannot be explained by the powers of nature alone. When dead bodies come back to life it is a miracle because the molecules that make up the corpse lack the powers necessary to generate life.”

Usefully, this definition even discusses dead bodies miraculously returning to life, which pertains to the present discussion. I would add, however, that I think other supernatural agencies (or even teleologies), acting from outside of ordinary physical causation, could likewise cause miracles. One could certainly imagine a plurality of gods, or angels, or demons performing miracles. Furthermore, agents who are located within the physical world, such as a miracle worker (e.g., Jesus), could also hypothetically perform miracles, even if their miraculous effects depart from ordinary patterns of physical cause and effect. We’ll return to this definition later on as I offer input for how to strengthen some of Ehrman’s arguments against miracles.

3) Events with the Lowest Prior Can Still Have the Highest Posterior

But to return to both Craig’s and Lowder’s critiques of Ehrman’s arguments from probability, my third point is that I agree Ehrman appears to be conflating prior probability with posterior probability. This leap seems to occur when Ehrman argues that historians “cannot show that miracles [have] happened” since it would involve “the most improbable event” being “the most probable.” Now, since in the quotes provided Ehrman does not speak in terms of priors or posteriors, it’s a little difficult to ascertain which kind of probability he is referring to, so he may not be making this leap.

But I will say that, even if a miracle is the most intrinsically improbable event out of a set of explanations (prior probability), there may be enough evidence in favor of its occurrence (consequent probability), to outweigh more intrinsically probable alternative explanations (posterior probability).

The Most Improbable Prior Becoming the Most Probable Posterior

Consider the following example: Say that one day I am walking on an abandoned beach somewhere in California. The sand is largely undisturbed and smooth. Then, somewhere along my journey, I come across hoofprints that resemble those of a horse. I spot no animal in the near vicinity that may have caused the hoofprints. As such, I am left to probability when interpreting what caused the evidence of the hoofprints. Was it 1) a horse?, 2) a zebra?, or 3) a unicorn? Out of these three explanations I would certainly argue that the unicorn is intrinsically the “the most improbable event,” followed by the zebra, then the horse, so that:

Unicorn: Pr (H3/B) < Zebra: Pr (H2/B) < Horse: Pr (H1/B)

But now let’s say that I return home and turn on the news, and to my surprise, learn that a creature resembling a unicorn had been taken into captivity, on that very same beach. Zoologists had likewise done DNA tests on the animal and discovered that they did not match any known Equidae. Likewise, X-rays had revealed that the distinctive horn on its head was an authentic part of its skeleton, and not a fake. Eventually, a public showing of the creature was made available, and I saw it for myself. At the same time, there was no report or evidence of horses or zebras in the same area.

I may, after all of this, even still suspect the whole thing to be an elaborate hoax. I may still suspect that the hoofprints, which I had seen, to have been caused by an unknown horse in the same area. But evidence of this kind would nevertheless drastically increase the probability of Pr (H3/B & E), perhaps to the extent that it outweighed Pr (H1/B & E) + Pr (H2/B & E). My point is that even the most intrinsically improbable explanation, out of a set of explanations, can still be the most probable posterior explanation, when factoring in consequent probability and weighing it against alternative explanations of the same data.

The Analogy of Winning the Lottery

The example of the unicorn may be a tad extreme (though I still think it’s perfectly suitable for something as extreme as a resurrection event), but there are also more mundane examples in which the lowest prior explanation still turns out to be the highest posterior. Winning the lottery is often used as an example to illustrate this kind of scenario.

Suppose that I purchased a ticket in a 1/1,000,000 odds lottery from a state vendor. The lottery only has two options:

H1 = Winner

H2 = Loser

The prior probability that I will win the lottery is vastly lower than the prior probability that I will lose, so that:

Winner: Pr (H1/B) < Loser: Pr (H2/B)

Suppose that when the winning numbers are announced, however, the numbers on my ticket match them, and I win the lottery. The ticket further contains all of the marks and signs of a valid lottery ticket. I return to the vendor to collect my winnings. The vendor carefully inspects the ticket and finds no signs of tampering or forgery. We have all of the evidence we would expect if I had genuinely won the lottery, and none of the evidence we would expect if I lost the lottery:

Winner: Pr (E/H1 & B) > Loser: Pr (E/H2 & B)

The prior probability that I would win is 1/1,000,000, and the prior probability that I would lose is 999,999/1,000,000. But let’s say that the consequent probability that I have a genuine lottery ticket is 999,999,999/1,000,000,000, versus the consequent probability that I have managed to perfectly forge a winning ticket and cheat the system is 1/1,000,000,000. Our Bayes’ formula for the probability of me genuinely winning the would be:

Pic 3

In this example, despite a very low prior probability for winning the lottery, the expected evidence is so strong, and the alternative hypothesis of the same data is so weak, as to yield a posterior probability of 99.9% that I am a genuine winner. Accordingly, even though a lottery win was originally the most intrinsically improbable event, a posterior consideration of the evidence was still able to overturn this low prior and allow for this event to become the most probable outcome.

Now, I have used the analogy of a lottery win to show how intrinsically improbable events can still become the most probable interpretation of a set of data, even in the mundane world. But in truth, it is an absolutely terrible analogy for the resurrection of Jesus, despite Craig’s attempt to appeal to it in some of his other debates. The reason why is because a lottery win is only an unlikely particular proposition, whereas a resurrection from the dead is an unlikely general proposition.

Lottery wins are not unprecedented events, and are recorded to happen in the world all the time. In fact, if ten million people entered a 1/1,000,000 lottery, and then nobody won, I would be rather surprised. This is nothing like a person returning to life after three days of brain death (especially in the manner depicted in the Gospels, as will be discussed below). Such an event would have no precedent whatsoever documented by medicine, forensic science, or parapsychology. To propose that such an event has occurred, therefore, relies on unprecedented general assumptions about the world, as opposed to a particular case of an improbable lottery win. As such, the resurrection of Jesus is far more similar to the example of the unicorn above, which would likewise be an unprecedented general proposition, than it is to the example of an unlikely lottery win, which would still have a documented reference class of occurrences in the world.

With this appreciation for the unprecedented nature of Jesus’ resurrection, and how it makes general and not just particular claims about the world, we can now look towards some of the considerations for how to calculate its prior probability.

The Prior Probability of a Miraculous Resuscitation or Resurrection

To begin with, I will dispel with any notion that assigning a low prior probability to miraculous resurrection from the dead has to do with “presuppositions against miracles” or “biases of worldview.” That billions upon billions of people die and do not resurrect after three days of brain death is empirically observed. In fact, this would be true, even in a possible world in which God exists. Bayesian expert Robert Cavin (slide 108) shows how this can be demonstrated through little more than a simple statistical syllogism, such as:

  1. 99%+ of Xs are Ys
  2. A is an X
  3. Therefore, A is probably a Y

In the case of Jesus’ resurrection, Cavin computes the prior probability using “The Anti-Resurrection Prior Probability Statistical Syllogism”:

  1. 99.999…999% of the dead are not supernaturally interfered with by God, and, thus, not raised by Him.
  2. Jesus was dead.
  3. Therefore, it is 99.999…999% probable that Jesus was not supernaturally interfered with by God, and, thus, not raised by Him.

As this syllogism shows, the prior probability for the resurrection of Jesus must be, at the very least, extremely low. And really, we aren’t out of the woods yet, because this same syllogism could be applied first to a miraculous resuscitation from the dead. Resuscitation only involves the more common process of a dead person returning to ordinary life (only to later die again). If Jesus had merely resuscitated from the dead, then such a process would entail that he had died at around age 30, returned to life, and then died again at around age 70 (allowing for a late lifespan in antiquity). Such an event could still be deemed “miraculous,” if it had happened on the third day after brain death (a medically unattested event), but it would still fall short of the teachings of orthodox Christian theology.

Orthodox Christian theology teaches that Jesus did not merely resuscitate from death, but that resurrected to immortal life. As Craig (Knowing the Truth about the Resurrection, pg. 15) explains about the nature of Jesus’ resurrection:

“Resurrection is not resuscitation. The mere restoration of life to a corpse is not a resurrection. A person who has resuscitated returns only to this earthly life and will die again.”

Instead, Craig (pg. 127) explains the actual transformation that is entailed by the Christian claim that Jesus resurrected from the dead:

“Jesus rose to eternal life in a radically transformed body that can be described as immortal, glorious, powerful, and supernatural. In this new mode of existence, he was not bound by the physical limitations of the universe, but possessed superhuman powers.”

The prior probability of this kind of event must, indeed, be even lower than Cavin’s statistical syllogism above. This is due to the fact that we are now not just discussing bodies that return to life after a period of death which exceeds the limits of resuscitation documented by medicine. We are now discussing further additional phenomena like “transformed bodies” that can exceed “the physical limitations of the universe” and possess “superhuman powers.” Such a body would be an unprecedented medical phenomena, even if it hadn’t previously returned to life on the third day after brain death.

If billions upon billions of people do not return to life on the third day after brain death (a miraculous resuscitation), and billions upon billions of people do not possess bodies with superhuman abilities exceeding physical limitations (transformed bodies), then surely the combination of both (a resurrection) would involve even more kinds of unprecedented phenomena than either alone. That which is unprecedented, particularly when we have a large pool of data to establish a precedent (in this case, billions upon billions of humans, whose bodies and deaths we can study), is almost by its very nature an event with a low prior probability.

Evidence that Could Render a Resurrection Event as Probable

Now, those points being made, I have nevertheless stated above that very strong evidence can still overturn a very low prior. With this in mind, let’s now turn to the evidence that Craig offers for the resurrection of Jesus:

  1. Jesus’ burial
  2. The discovery of his empty tomb
  3. His post-mortem appearances
  4. The origin of the disciples’ belief in his resurrection

Craig elaborates further on these points later in his opening statement (drawing primarily on information in Paul’s letters and the Gospels), but I think this preliminary sketch will serve well to discuss, abstractly, what kind of evidence could be offered to argue that a given person had resuscitated or resurrected from the dead.

To begin with, I think it will be helpful to discuss another individual, besides Jesus of Nazareth, whose death and (hypothetical) resuscitation or resurrection we could investigate in a similar manner. Another executed individual, whose death was personally witnessed by dozens of individuals and electronically witnessed by hundreds (whose testimony we could actually corroborate in modern times), is Oklahoma city bomber Timothy McVeigh. As The Guardian (“A Glance, a Nod, Silence, and Death”) explains about the witnesses for McVeigh’s death:

“The execution was watched by 10 journalists, McVeigh’s lawyers, government officials, 10 survivors and victims’ relatives in Terre Haute and another 232 in Oklahoma City, linked by satellite feed.”

We can even plot on a reliable timeline the precise seconds of McVeigh’s death. Now, the Gospels claim that Jesus’ execution started around “noon,” with the Synoptics adding a (historically uncorroborated) three-hour midday darkness (likely invented in imitation of OT verses like Jl. 2:1-2; Am. 5:18-20; Zp. 1:14-15), and the ripping of the curtain in the Jewish Temple (likely invented for theological reasons, discussed here). Those theological embellishments aside, we can certainly say that we have nothing like the evidence below for Jesus’ execution (even if understandably so):

Following his death, McVeigh was cremated at Mattox Ryan Funeral Home in Terre Haute, IN. Now, let’s imagine an alternate sequence of events:

Let’s imagine, following his placement in the incinerator, that following the 1-3 hours of cremation, those arriving to pick up the ashes suddenly found an angelic being standing in front of the furnace. This angelic being then opened the incinerator to reveal McVeigh perfectly alive and well. McVeigh then revealed himself to all present at the funeral home, and walked out, miraculously thrusting away any attempt to constrain him.

In the days and weeks following, McVeigh made many public appearances before whole crowds. Any police or military personal who attempted to re-arrest him found themselves constrained by invisible forcefields. When they fired guns at him, the bullets penetrated his body, and yet did not kill or injure him (a superhuman feat, indeed). Eventually, forensic scientists arrived who confirmed that McVeigh had the exact same DNA and finger prints as the individual who had been executed. Multiple journalists from multiple news networks documented his appearances, complete with live video broadcasting. McVeigh took up the “One Million Dollar Paranormal Challange” offered by the James Randi Educational Foundation, and demonstrated the ability to take a gunshot directly to the head, at point-blank range, and yet to walk away unscathed (demonstrating that his body was truly immortal). Eventually, McVeigh ascended into space in New York Central Park, before hundreds of thousands of people, filming on home video cameras, as well as multiple film networks.

Now, as I have stated above, the prior probability for a resurrection event is very low, indeed. But in an example such as above, I would argue that its posterior probability is very good. This is because in such a hypothetical example we are dealing with very extraordinary evidence. And future historians, I would think, could argue that this event probably took place, based on the evidence that I have discussed.

The Meaning of “Extraordinary” Evidence

Something should be said at this point about what is meant by the term “extraordinary.” As forensic expert Richard Packham (“The Man with No Heart: Miracles and Evidence”) explains, the term “extraordinary” does not mean that the type of evidence itself has to be remarkable. Video tapes, DNA and finger print testing, medical records, and so on are all part of ordinary life experience. What is meant by “extraordinary” in this case is that the evidence in question cannot be equally explained by a wide range of causes, but is only rendered probable under a very specific hypothesis.

Extraordinary evidence is especially necessary when dealing with events that have extremely low priors, since a higher consequent probability will be necessary, and a lower consequent probability for alternative hypotheses with higher priors, in order to offset the disparity between the priors of those alternative hypotheses. If the consequent probability is roughly equal for an event with a low prior, as it is for an event with a higher prior, then the event with the higher prior will have a greater posterior value, by virtue of its prior alone. This is especially relevant for the study of Jesus’ resurrection. As has been emphasized, the prior for such a resurrection event is very, very low.

It is also fair to say that alternative scenarios — such as Jesus’ body being obscurely buried or moved/stolen, and the disciples first conceiving of the resurrection through a combination of visionary experiences and theological rationalizations — very likely have higher priors. Phenomena like bodies being obscurely buried, or moved/stolen from their burial sites are documented occurrences (unlike immortal resurrections). Furthermore, especially in cases when a burial site is found empty, the prior probability for a body being moved or stolen would increase substantially.

Sometimes apologists try to argue that naturalistic accounts of the group appearances in Paul’s letters (such as to the 12 and the 500) necessitate things like mass hallucinations (a phenomenon to which they assign a very low prior). But this is based on a misconstrual of the evidence. To argue that 12 people, for example, had all seen and heard the same thing at the same time, we would need something like 12 independent accounts from different individuals all testifying to the same kind of appearance. In contrast, Paul’s testimony only provides a single account of an anonymous group of 12 people that Jesus (allegedly) “was seen” by, without providing the details of this appearance. For an in depth response to apologetic assertions of this sort, see my reply to Vincent Torley on the probability of Jesus’ resurrection. For a further response to the “mass hallucination” objection for the appearances in the Gospels, see Keith Parsons’ article “Kreeft and Tacelli on the Hallucination Theory.”

If the consequent evidence found in Paul’s letters and the Gospels is equally probable for such alternative scenarios (which I will argue is roughly the case), as they are for a veridical resurrection event, then these alternatives to the resurrection will have a greater posterior value, by virtue of their higher prior value alone. (I will also make some additional arguments at the end for why the consequent probability of these texts may even yield a higher value for naturalistic or non-paranormal explanations.)

In the case of the hypothetical example of McVeigh above, however, there are not too many alternative explanations that we can turn to (with higher priors), without resorting to ones that are likewise paranormal (which also have low priors). One could posit, for example, that the apparently resurrected individual was not actually the McVeigh who had died, but instead a clone with his same DNA and fingerprints, who had been implanted with extraordinary technology that allowed him to perform the feats described. But since such technology would exceed all known technology to have existed in the early 2000’s (or even today), it would still be an appeal to a highly paranormal explanation (which likewise had a very, very low prior).

I suppose the weakest link in the chain of events above is what happened to McVeigh’s corpse. One could, of course, appeal to the eyewitness testimony of those at the funeral home. But, at least as someone who has worked at a funeral home in the past (Evergreen Mortuary and Cemetery, Tucson, AZ, 2007-2008), I wasn’t aware of any video cameras or recording devices at the facility. It could be confirmed that McVeigh’s body was checked in (with a timestamp), and that certain eyewitnesses had seen it enter the cremation furnace and then come out alive. But I’m not sure much more could be offered (though, if anyone knows about the specifics of McVeigh’s situation, please provide me with the details). But even this would be more than can be offered with Jesus. We only have anonymous literature (the Gospels) for the specifics of Jesus’ burial, to state that Joseph of Arimathea buried Jesus (perhaps with Nicodemus) and that women (who vary in their role call) found the tomb empty (perhaps with Peter and an unnamed discipled). At least journalists could investigate those at McVeigh’s funeral home in ways that cannot be done with the Gospels.

Why the Evidence for Jesus’ Resurrection is Not Extraordinary

So, if the hypothetical resurrection event of McVeigh above could be enough to overturn a low prior, why do I think the evidence for Jesus’ resurrection is insufficient? Let’s turn now to the evidence that Craig presents. Using a variation of the “minimal facts” approach, Craig presents the following “facts” that supposedly must be explained by the historians, surrounding death and burial of Jesus:

  1. After his crucifixion Jesus was buried by Joseph of Arimathea in a tomb.
  2. On the Sunday after the crucifixion, Jesus’ tomb was found empty by a group of his women followers.
  3. On different occasions and under various circumstances different individuals and groups of people experienced appearances of Jesus alive from the dead.
  4. The original disciples suddenly and sincerely came to believe that Jesus was risen from the dead despite their having every predisposition to the contrary.

Craig describes these alleged circumstances as “facts” in order to treat them as non-negotiable, but the reality is that there is a wealth of NT scholars who do not accept all of them (nor how they are worded), or who are at least agnostic about whether they can be verified. Ehrman himself, for example, has recently cast doubt on Jesus’ burial by Joseph of Arimathea, which he elaborates on in his blog series “Did the Romans Allow Decent Burials?.” Likewise, Ehrman also doubts the discovery of the empty tomb by women, which he discusses in his blog “The Women and the Empty Tomb.” Below is an additional list of NT scholars who do not accept the “fact” of the empty tomb:

James Crossley, Marcus Borg, Günther Bornkamm, Gerald Boldock Bostock, Rudolf Bultmann, Peter Carnley, John Dominic Crossan, Steven Davies, Maurice Goguel, Michael Goulder, Robert Price, Hans Grass, Charles Guignebert, Uta Ranke-Heinemann, Herman Hendrickx, Roy Hoover, Helmut Koester, Hans Küng, Alfred Loisy, Burton Mack, Willi Marxsen, Gerd Lüdemann, Norman Perrin, Marianne Sawicki, John Shelby Spong, Howard M. Teeple, and Rev. John T. Theodore.

I’ve critiqued Craig’s variation of the minimal facts argument before in my essay “Knocking Out the Pillars of the “Minimal Facts” Apologetic,” in which I discuss some of the statistical problems with trying to gauge scholarly consensus for his alleged “facts” (the first two of which do not even have a consensus of scholarly support), so I won’t elaborate too much further here on the points that I have already made in that essay.

There are a couple points that I will reiterate, however, the first of which pertains to Craig’s third “fact” about Jesus’ post-mortem appearances. Even if there are multiple post-mortem appearances of Jesus described in Paul’s letters and the Gospels, that doesn’t mean that every encounter has a historical basis, nor that, historically speaking, every person who thought that Jesus appeared to them saw the same thing. Even the Gospels/Acts describe Jesus’ appearance to Paul, for example, differently than how he appeared to the original disciples. On one occasion Jesus appears without being recognizable to his followers (Lk. 24:15-16). On another he walks through doors (Jn. 20:26). Jesus’ appearances in 1 Cor. 15:3-7 are entirely vague, and we cannot assume that what “was seen” by Peter, for example, was the same as what “was seen” by the 500. Neither can we assume that such appearances (especially in Paul’s letters) amounted to corporeal experiences. With regards to the group appearance to the 500, for example, scholar Stephen Patterson (The God of Jesus, pg. 236) points out:

“It is not inconceivable that an early Christian group might have interpreted an ecstatic worship experience as an appearance of the risen Jesus.”

We lack the ability to investigate any of these specifics, and Paul’s account itself is too vague. In such a circumstance, this allows for a wider range of causes to produce the same evidence. This is precisely the opposite of extraordinary evidence, since its consequent probability is roughly equally expected for alternative hypotheses to a corporeal experience (if not greater, as will be argued below).

My second point is that many scholars would take issue with the wording in Craig’s fourth “fact,” particularly with regards to the original disciples “suddenly” coming to believe he was risen, and “having every predisposition to the contrary” of forming such a belief. Both of these descriptions are vaguely worded and can be picked at to add more nuance.

To begin with, although Matthew, Luke, and John depict Jesus appearing to his original disciples quite shortly after his death (causing them them to believe in his resurrection), many scholars would argue that they may have formed the resurrection belief after a substantial gap of time later. Since the creed in 1 Cor. 15 is generally agreed to have been composed 2-5 years after Jesus’ death, they must have formed this belief within a few years after the crucifixion, but it can be debated whether this constitutes a “sudden” belief. It may have taken the original disciples months or even a year to form such a view. As Ehrman (How Jesus Became God, pg. 175) has recently argued:

“If it is true that the disciples fled from Jerusalem to Galilee when Jesus was arrested, and that it was there that some of them ‘saw’ him, they could not have seen him the Sunday morning after his death. If they fled on Friday, they would not have been able to travel on Saturday, the Sabbath; and since it was about 120 miles from Jerusalem to Capernaum, their former home base, it would have taken a week at least for them to get there on foot. Maybe some of them, or one of them, had a vision of Jesus in Galilee soon after he was crucified — possibly the following week? The week after that? The next month? We simply don’t have sources of information available that make this kind of judgement possible.”

And so I think many scholars would dispute whether it is a “fact” that the original disciples “suddenly” came to believe that Jesus was risen. But I think even more would dispute that they had “every predisposition to the contrary” of forming such a belief. I think that Ehrman scored a very solid substantive (not just technical) point against Craig, when he argued that this belief may not have been improbable at all, if the disciples were searching for a rationalization for how Jesus could have been crucified, and yet still be the Messiah. As Ehrman argued during the debate:

“The one thing we know about the Christians after the death of Jesus is that they turned to their scriptures to try and make sense of it. They had believed Jesus was the Messiah, but then he got crucified, and so he couldn’t be the Messiah. No Jew, prior to Christianity, thought that the Messiah was to be crucified. The Messiah was to be a great warrior or a great king or a great judge. He was to be a figure of grandeur and power, not somebody who’s squashed by the enemy like a mosquito. How could Jesus, the Messiah, have been killed as a common criminal? Christians turned to their scriptures to try and understand it, and they found passages that refer to the Righteous One of God’s suffering death. But in these passages, such as Isaiah 53 and Psalm 22 and Psalm 69, the one who is punished or who is killed is also vindicated by God. Christians came to believe their scriptures that Jesus was the Righteous One and that God must have vindicated him. And so Christians came to think of Jesus as one who, even though he had been crucified, came to be exalted to heaven, much as Elijah and Enoch had in the Hebrew scriptures. How can he be Jesus the Messiah though, if he’s been exalted to heaven? Well, Jesus must be coming back soon to establish the kingdom. He wasn’t an earthly Messiah; he’s a spiritual Messiah. That’s why the early Christians thought the end was coming right away in their own lifetime.”

And so, we can hardly say that the disciples had “every predisposition to the contrary” of forming the resurrection belief. If such a rationalization provided a doctrine that, in the face of grief, allowed them to hold on to their faith in Jesus as the Messiah, as well as his apocalyptic teachings, then such a development could not be improbable at all. It’s important to remember that most historians do not regard Jesus as a militaristic Messianic figure who was leading a resistance movement. Had he been one, perhaps his movement might have died out, following his death. But had Jesus already been preaching about an apocalyptic event in the near future, then his death (and expected return) could fit easily within such an eschatology. All it would take would be an additional rationalization that the Messiah had only temporarily suffered (as would his followers), but would soon return, to usher in the tribulation that he had predicted.

It should be noted that this belief in Jesus’ return could have very likely preceded the experiences of his post-mortem appearances. If the early Christians came to believe that Jesus was raised to heaven, and would soon return, then it may be this precise view that triggered them to have visions or experiences of him raised from the dead. In that case, Craig may have “facts” three and four backwards.

Craig is laying far too much emphasis on the post-mortem appearances and acting as if it would be impossible for them to be produced by human imagination (necessitating something like a veridical resurrection event). Craig wants to then use this as the launch pad for what formed the resurrection belief. But, it would certainly not be hard for human imagination to theologically rationalize that Jesus had been raised to heaven, and would soon return. And, once one had come to such a belief, having visions or experiences of Jesus in his risen state certainly seems less improbable. The Gospels, of course, depict the disciples as being surprised by the resurrection, but it’s important to remember that these are later sources (likely influenced by later storytelling, embellishment, and apologetic agendas). Paul’s testimony in 1 Cor. 15 emphasizes the resurrection before the appearances, and it should be noted that the vocabulary ἐγήγερται (“he rose”) can apply equally to ascension to heaven as it does to rising on earth. As Ehrman (How Jesus Became God, pgs. 205-206) interprets the earliest resurrection belief:

“They believed that Jesus had come back from the dead — but he was not still living among them as one of them. He was nowhere to be found. He did not resume his teaching activities in the hills of Galilee … The disciples, knowing that Jesus was raised that he was no longer among them, concluded that he had been exalted to heaven. When Jesus came back to life, it was not merely that his body had been reanimated. God had taken Jesus up to himself in the heavenly realm, to be with him … This is why the disciples told the stories of Jesus’ post-resurrection appearances the way they did. Jesus did not resume his earthly body. He had a heavenly body. When he appeared to his disciples, in the earliest traditions, he appeared from heaven.”

Either way, Paul’s testimony is too vague to furnish any narrative for a corporeal appearance of Jesus on earth. The vaguer the evidence, the more alternative hypotheses will have roughly equal consequent probability for explaining the same the same data. And if an alternative hypothesis with a higher prior value — such as an obscure burial of Jesus with no empty tomb, followed by the disciples coming to rationalize that Jesus had been raised to heaven (and would soon return) in the weeks and months following his death, triggering them to have post-mortem experiences, and culminating in the testimony found in 1 Cor. 15 — has equal consequent probability to a veridical resurrection event, then its posterior probability will be greater, by virtue of the prior alone.

Qualifiers that Often Go Ignored in Resurrection Debates

One thing that often irks me is that a number qualifiers often go unmentioned by apologists in debates when they discuss the evidence for Jesus’ resurrection. I’ll discuss a few below:

1) A Lone Enterprise

One is that, outside of the field of New Testament Studies, I am unaware of any ancient miracle claim in antiquity (analogous to something as remarkable as Jesus’ resurrection) that Classicists, scholars of Near Eastern Studies, or ancient historians argue can be proved using ancient texts. This circumstance needs to be stated at the beginning of any opening statement that Craig or any other apologist makes trying to prove Jesus’ resurrection using ancient texts like Paul’s letters and the Gospels. They are practically embarking on a lone enterprise. Such an enterprise becomes particularly suspicious when one looks into the universities that employ these apologists, and realizes that the vast majority come from faith-based institutions that dogmatically require adherence to Christian theological beliefs. There is no parallel for such institutions in either Classics or among atheists. I know of no institution of higher education that has a policy requiring faculty to sign doctrinal statements requiring adherence to Pagan theological beliefs, or of an atheist university requiring all faculty to be atheists. One has to question where and how an apologist like Craig fits anywhere into ordinary trends within secular academia, when has pretenses to holding “academic debates” about Jesus’ resurrection.

2) The Limitations of Ancient Literature

Second, I have repeatedly emphasized that the locus of evidence that we are dealing with, when discussing Jesus’ resurrection, is ancient texts. And really, since we aren’t dealing with documentary evidence (like inscriptions or papyri notes), it is better described as ancient literature. This kind of evidence needs to be contrasted heavily with the kinds of evidence that I discussed in the hypothetical cases of the unicorn and McVeigh above — medicine, forensic science, parapsychology, and journalism — precisely because the latter form of evidence is far better at furnishing evidence that can be considered “extraordinary.” Forensic science, for example, is designed to weed out alternative explanations for a piece of evidence (e.g., blood) beyond a very specific hypothesis, through things like DNA testing. Ancient literature, in contrast, is very bad at giving such specifics.

Ancient literature was produced in a context from which we are distanced by thousands of years (permitting us to interview no original witnesses). We do not possess the autograph manuscript for any ancient literary work, meaning that we must piece together every ancient text from copies of copies, passing through countless unknown hands. The standards for what constituted even historiography and historical biography in antiquity differed drastically from modern standards, meaning that even the best historians frequently did not identify their sources, explain how they interacted with them, or to what degree they were reliable. Frequently, whole speeches were reconstructed or even invented/imagined, by historians of the likes of Tacitus. Even historians like Polybius and Livy could contradict in the details of important events, such as Hannibal’s crossing of the Alps.

When we consider that the Gospels do not even meet the literary standards of the ancient historiography and historical biography of their period, that their authors are highly anonymous and do not identify their sources, and that they fashion their narratives in imitation of previous mythical literature, the reliability of these texts is called even further into question. This coupled with the vague nature of Paul’s testimony makes the evidence for Jesus’ resurrection (primarily restricted to this body of evidence) quite problematic. Craig tries to shortcut this issue by appealing to scholarly consensus. But as discussed above, the first two of Craig’s “facts” do not even have a consensus of scholarly support, and his framing of the latter two can be highly disputed.

The reality is that for most obscure claims in the ancient world (excluding highly general claims, e.g., Jerusalem was a real city), especially when we only have ancient literature to rely on, we are going to have far less certainty than for modern events, which we can investigate with more reliable methods. This reality pertains to the resurrection of Jesus, for which we have no archaeological or documentary evidence, but only ancient literature composed decades after his death [1]. Such evidence, by its very nature, will be less probative and more obscure in its origins. The more obscure the evidence is, the greater its consequent probability is for a wider range of explanations. In such a circumstance, such evidence is far less extraordinary.

Elaborating on this limitation of ancient literary evidence, even Christian scholar Dale Allison (Resurrecting Jesus, pp. 337-339) makes the following observation about appealing to the post-mortem appearances as “evidence” for Jesus’ resurrection:

“Most of the past – surely far more than 99 percent, if we could quantify it – is irretrievably lost; it cannot be recovered. This should instill some modesty in us. Consider the weeks following the crucifixion. We have only minuscule fragments of what actually transpired. What, for instance, do we really know about the resurrection experience of James? First Corinthians 15:7 says that he saw the risen Jesus. And that is it. What Jesus looked like, what he said, if anything, where the encounter took place, when precisely it happened, how James responded, what state of mind he was in, how the experience began, how it ended – all of this had failed to enter the record. Almost every question that we might ask goes unanswered … Yet they are the sorts of questions historians often ask of old texts. The fact that we cannot begin to answer them shows how emaciated historically – as opposed to theologically – the Gospel narratives really are. Even if we naively think them to be historically accurate down to the minutest detail, we are still left with precious little. The accounts of the resurrection, like the past in general, come to us as phantoms. Most of the reality is gone … Even if history served us much better than it does, it would still not take us to promised land of theological certainty.”

If far more than 99% of the circumstances behind Jesus’ resurrection are lost, then we are dealing with unknowns that leave open a lot of room for alternative hypotheses besides a veridical resurrection event. Such ambiguity of the evidence certainly diminishes Pr (R/B & E), particularly when it raises the consequent probability of those alternative hypotheses for explaining the same data. This is part of a broader point that is implied when Classicists, for example, say that we “know less” about the ancient world than today. We know less, because often times the evidence is too limited to yield a high posterior value for specific hypotheses, than what we can do for many modern events. It’s simply a reality of working with more limited evidence, to which Jesus’ resurrection belongs.

That said, I don’t wish to exclude ancient literary evidence entirely from the table. One could certainly imagine scenarios in which ancient literature furnished extraordinary evidence. Say, perhaps, that we had eleven independent accounts written by the original disciples (excluding Judas), in which they identified their names and discussed seeing Jesus in a corporeal setting. Say that even the original autograph manuscripts of these texts had been preserved, and that based on a variety of considerations — such as manuscript testing and early external quotations — we could reliably date these texts to only a few years after Jesus’ death. Say that Paul’s testimony was more extensive, and that he identified more of his sources and described Jesus’ appearances in a corporeal manner, which aligned with the other hypothetical texts described above. Say that Caiaphas, Pilate, and Tiberius Caesar had likewise written about Jesus’ resurrection in literary texts that were preserved. Say that early-1st century inscriptions were found in Jerusalem that described Jesus’ resurrection in detail. All of this combined evidence would certainly raise the probability of Pr (R/B & E) drastically.

But when we are only dealing with vague testimony in Paul and anonymous literature (composed decades later) in the case of the Gospels, the case becomes weaker. It becomes further weaker, when we consider that the Synoptics show heavy inter-dependence, and that John shows inter-dependence (either through Markan-dependence or a common Passion narrative) with the story of the empty tomb (one of the most crucial pieces of evidence). Where the Gospels are not textually dependent, such as the post-mortem appearances, they substantially diverge (with Matthew placing Jesus’ appearance to his disciples in Galilee, but Luke and John placing it in Jerusalem, with highly different details in the latter two cases). When agreement can only be reached through collusion, and discrepancies abound when there is a lack of collusion, this is precisely the opposite of strong evidence. Then there is the fact that, when the evidence is lined up diachronically (Paul -> Mark -> Matthew -> Luke -> John), it grows in the telling, exactly the sort of thing you would expect with embellishment and legendary development.

Such evidence can hardly be described as “extraordinary.” It’s consequent probability fits nicely into legendary development, perhaps even more nicely than into a veridical resurrection event, as will be discussed further below.

3) Where Is Jesus Today?

A final qualifier that I think often gets ignored in debates about the resurrection is an additional “minimal fact” that I think we can almost all agree upon: Jesus is not on earth today. In Luke/Acts (as well as what I think is the correct interpretation of Paul), this circumstance is explained by Jesus’ ascension to heaven. Given our current understanding of cosmology, it’s a little difficult to understand how flying into the sky would do little more than take Jesus into space. But I suppose one could conjecture that Jesus perhaps teleported into another dimension of reality after he was hidden up in the clouds somewhere. (Or maybe he is out somewhere beyond Voyager?)

Now, this additional fact (which is an actual minimal fact) requires that Bill Craig will need to appeal to more than just the resurrection to explain all the data. Simply rising from the dead doesn’t explain Jesus’ absence from earth today. He needs to give an explanation for how Jesus could rise to immortal life, and yet disappear entirely from earth. (Such evidence, in fact, would be unexpected for the resurrection hypothesis, without appealing to further explanations of the data.) Suggesting that Jesus is now in another plane of existence (or in outer space), however, would require him to appeal to even more general assumptions than the resurrection hypothesis. And such general assumptions would plunge his case into even further depths of a lower prior probability.

But here, naturalistic or non-paranormal hypotheses have complete explanatory ease. They can appeal to a very easy explanation: Jesus is dead. Such an explanation requires that we make no general assumptions, and it would have a very, very high prior probability. In fact, it’s literally what happened to everyone else from the ancient world. The evidence of Jesus’ absence is likewise completely expected on this hypothesis. Considering that Craig has to distort his appeal to scholarly consensus to marshal his other alleged “facts,” I’m surprised that he seems to never bring this one up. It certainly comes off as sweeping the elephant under the rug, while doing one’s best to turn the molehills into mountains.

Considerations that May Lower Pr (E/R & B)

Above I have taken the approach of arguing that the consequent probability of the resurrection hypothesis is roughly equal for alternative explanations, whereas those explanations have a much higher prior probability than the resurrection. As such, alternative explanations to the resurrection will inevitably have a greater posterior probability to the resurrection. I have also argued that assigning the resurrection a low prior probability has nothing to do with “naturalist presuppositions” or “worldview bias.” That billions upon billions of people die, and do not supernaturally resurrect in the manner described with Jesus, is an observed medical fact. But I don’t wish to leave this debate a hostage to prior probability. As such, I will further make a couple additional arguments for why a veridical resurrection event does not even have an equal consequent probability to alternative hypotheses, but in fact a value that is lower.

1) A Story that Grew in the Telling 

The first argument touches on a point that is already noted above about how the story of the resurrection grew in the telling. This kind of trend renders a high consequent probability for legendary development, but it fits awkwardly with the resurrection hypothesis.

We can start with the earliest testimony in Paul’s letters, particularly 1 Corinthians. First off, unlike the Gospels, Paul makes no mention of an empty burial place being discovered after Jesus’ death, the first mention of which appears in Mark. It’s also an odd detail in Mark that the women who discover the empty tomb run away and tell no one. Many scholars suggest that this detail bears traces of the author of Mark inventing the story. As James Crossley (“Against the Historical Plausibility of the Empty Tomb Story and the Bodily Resurrection of Jesus,” pg. 186) argues:

“The earliest empty tomb story we have (Mk 16.1-8) suspiciously makes it clear that the only witnesses to the empty tomb told no one (16.8). All the other Gospel narratives make good sense in the context of creative storytelling, including the grounding of present beliefs in the life, death, and resurrection of Jesus.”

Paul’s silence on the empty tomb, if he had actually heard of such a thing from Peter, or anyone else, is likewise an awkward fit of the evidence. As G.W.H. Lampe (The Resurrection, pg. 43) points out:

“In this case I think that the argument from silence has unusual force. For the situation in which Paul wrote 1 Corinthians 15 was that some of the Corinthians were denying there is a resurrection of the dead (1 Cor. 15:12). In answer to them Paul marshals every possible argument…If Jesus’ resurrection is denied, he says, the bottom drops out of the Christian gospel. And the evidence that he was raised consists in the appearances to himself and to others. Had he known a tomb was found empty it seems inconceivable that he should not have adduced this here as a telling piece of objective evidence.”

But what is even further noteworthy is that Paul not only describes the appearance to Peter and the other early Christians in this passage, but also Jesus’ appearance to himself. In Acts, however, Paul does not have a corporeal experience with Jesus, but rather a celestial vision. What is strange, however, if Jesus had actually appeared to Peter or the other Christians in a corporeal manner, is that Paul in 1 Cor. 15 describes his own celestial appearance with the same vocabulary. As even Christian scholar Raymond Brown (An Introduction to the New Testament, pg. 534) points out:

“The concluding reference to himself is extremely important since Paul is the only NT writer who claims personally to have witnessed an appearance of the risen Jesus … Paul places the appearance to himself, even if it was last, on the same level as the appearance to all the other listed witnesses.”

And so, we get an odd reading of the earliest testimony, if we are to interpret it as Jesus appearing to Peter and the other Christians in a corporeal manner, but to Paul in a different manner. Instead, Crossley (pg. 186) argues:

“The list of eyewitnesses in 1 Cor. 15.5-8 gives no evidence pointing in the direction of the bodily resurrection as an historical event, except in the sense of a visionary experience.”

Decades after Paul, however, the Gospels depict Jesus’ resurrection in an increasingly bodily (and embellished) manner, with it growing between each Gospel. Diachronically, the evidence can be lined up as follows:

  • Paul (c. 50’s CE), the earliest source, has no empty tomb and just vague “appearances” of Jesus (likely interpreted as visionary experiences).
  • Mark (c. 70’s CE), half a century after Jesus’ death, then has an empty tomb.
  • Matthew (c. 80’s CE), after Mark, then has Jesus appear to his disciples in Galilee.
  • Luke (c. 90’s CE), even later, instead has Jesus appear to his disciple in Jerusalem (a different story than Matthew’s), and likewise this Jesus can teleport and is not at first recognizable to his followers.
  • Finally, John (c. 90-100’s CE) has Thomas be able to touch Jesus’ wounds.
  • If you go even later into the Gospel of Peter (2nd century CE), Jesus emerges as a giant from the tomb with giant angels accompanying him (verses 39-40).

The fish gets bigger

This sequence of event reads somewhat oddly, under the hypothesis that its origin started with a veridical resurrection event. Why wouldn’t Paul and the Gospel of Mark explain things more clearly and with more detail in the earliest accounts? Why do we only see the more corporeal elements in later accounts, building on earlier texts, from which they are absent?

Under the hypothesis of legendary development (beginning with visionary experiences and theological rationalization), however, they fit quite nicely. Under this interpretation, though, the consequent probability is actually greater for an alternative hypothesis to the resurrection event. And so, it’s not just that the resurrection has a low prior, but also that the evidence is not even the most expected under this hypothesis.

2) A World with Pre-Jesus Models for Resurrection Fabulation

My second argument deals with the broader Mediterranean context in which the resurrection belief emerged. If we were dealing with a world in which no one was ever thought to resurrect from the dead, and the belief in Jesus’ resurrection had just come out of nowhere, then this might be evidence that rendered the consequent probability quite high for the resurrection hypothesis. But recent scholarship is finding this more and more to be the exact opposite case.

Richard Miller in his monograph Resurrection and Reception in Early Christianity, for example, has recently performed an exhaustive thematic and historical study of how Jesus fits into the divine translation of other ancient resurrected figures, such as Hercules, Romulus, Julius Caesar, and Augustus.

MillerInteracting with the testimony of Justin Martyr (the earliest exegete of the Gospels), Miller concludes in the monograph: “[T]he earliest Christians would not have considered the New Testament accounts of Jesus’ resurrection to be literal or historical, but instead would have recognized this narrative as an instance of the trope of divine translation, common within the Hellenistic and Roman mythic traditions.” In that case, the consequent probability for a veridical resurrection event is lowered substantially, since the earliest textual evidence does not even point in that direction.

Then there is the circumstance that much of the mythology surrounding Jesus (including his resurrection, ascension, and attribution of a divine name) fits well into previous Mediterranean concepts of divinity. If the evidence for Jesus’ resurrection and deification fits into a cultural cookie-cutter shape, one has to question whether it is better explained by an unprecedented veridical resurrection, or whether the evidence is more expected by such cultural fashioning.

M. David Litwa has recently argued in Iesus Deus (pg. 4), through a synchronic analysis of the mythology about Jesus’ life — pertaining to his divine conception (ch. 1), his childhood zeal for honor (ch. 2), his miraculous benefactions (ch. 3), his epiphanic transfiguration (ch. 4), his immortalizing resurrection (ch. 5), and his reception of a divine name after his ascension (ch. 6) — that the early Christian depiction of Jesus fits quite nicely into common Mediterranean tropes of godhood and divinity. In that case, the evidence for Jesus’ resurrection, ascension, and deification may have a greater consequent probability under the hypothesis of culturally (re)fashioning his image, than it does for a veridical resurrection event.

Considerations like those noted above show that the case against Jesus’ resurrection need not hinge solely on factors of prior probability alone (something that apologists will often try to chalk up to “worldview bias,” despite the counter-points I’ve already mentioned). Even with issues of prior probability laid aside (so that all hypotheses are on equal footing), the evidence can still be read as rendering a higher consequent probability for alternative hypotheses. And in that case, the posterior probability would still yield a higher value for alternatives to the resurrection. And it need hardly be pointed out that, if both the prior and the consequent is lower for the resurrection than for alternative hypotheses, then its posterior doesn’t have a chance.

Suggestions for Strengthening Ehrman’s Case

In his debate with Craig, Ehrman noted many of the other alternative explanations that I have discussed above — such as the body being moved/stolen, the original disciples having only visions of Jesus, the resurrection belief being formed out of theological rationalization, etc. As I have said, I think that Ehrman overall made a stronger case that these alternative hypotheses are more probable than the resurrection. For this reason, I think that the “spirit” of Ehrman’s probability argument was correct.

But if I can identify a weak point in Ehrman’s presentation, I don’t think he clearly explained the formal validity of why. That does not entail that he made a weaker case. Someone can argue a case, which has a few invalid points, that nevertheless is the stronger position. To contend otherwise is to commit the fallacy fallacy. But Ehrman never made a response to Craig’s Bayesian critique. Instead he remarked:

“Thank you, Bill, for that impressive refutation! I do have to tell you that if you think I’m going to change my mind because you have mathematical proof for the existence of God, I’m sorry, but it ain’t gonna happen!”

Jokes aside, though, this is not what Craig was arguing. Craig was not providing a mathematical proof of God, but a critique of Ehrman’s probabilistic argument against miracles. Later on his blog, Ehrman (“Carrier, Bayes Theorem, and Jesus’ Existence“) did remark:

“After that debate I got a bunch of emails from mathematicians and statisticians who also thought Craig’s argument was outrageously funny — not to say outrageous — and explained to me mathematically why Craig had absolutely botched the ‘proof.'”

I’m not aware of anywhere online, however, where Ehrman has made those emails public or discussed the substance of their content. As such, I’ve written this (rather lengthy) essay for what I think is the correct way to respond to Craig’s argument. I think that Craig presented the weaker case, but that he also did score a technical point on this issue. To add further strength to Ehrman’s case, therefore, I shall make the following suggestions:

1) Revising Ehrman’s Definition of a Miracle Event

I don’t think that Ehrman presents the strongest case against miracles (including the resurrection) when he defines them, from the get go, as “the most improbable event.” This kind of definition is too question-begging and it opens the door to the stock “naturalist presupposition” apologetic slogan. The reason we are looking at stuff like the texts that discuss Jesus’ resurrection is precisely to see whether such a miracle could ever be probable.

As I stated at the beginning of this essay, in certain possible worlds miracles may not even be intrinsically improbable. One could imagine a possible world in which miracles happened quite frequently, if we define them as departures from ordinary patterns of physical cause and effect, which are brought about by miracle-working agents. What matters most is that such events would be empirically observable. If someone has changed water into wine, or risen from brain death into an immortal body, then this is a demonstrable and observable effect. And science could hypothetically study such a change, which could even be documented in peer-reviewed literature. I am just aware of no such peer-reviewed literature in the fields of medicine, forensic science, or parapsychology in the actual world that we live in.

But to discuss the resurrection of Jesus, I don’t think that one even needs to mention the word “miracle.” All that needs to be done is to define “resurrection” as Craig has previously in his own published literature. Then, the next step is to point out that billion upon billions of people have been recorded to die, without there being any documentation for such an event in medical literature. That’s all it takes to assign a very, very low prior probability to the resurrection hypothesis.

2) Noting that Pr (R/B) is Non-Zero, But So Low as to Only Be Rescued by Extraordinary Evidence

I’ve noted at the top of this essay that no event in the empirical world (aside from logical contradictions), strictly speaking, has an inductive prior of zero. I think Ehrman should make this simple qualification, in order to be fair in his treatment of Craig. But some events, such as the resurrection, have an extremely low prior (based on documented empirical observation) for the reasons that I have discussed. Ehrman only needs to note this, in order to demand some very extraordinary evidence from Craig.

I’ve also discussed how “extraordinary” evidence should be defined as that which renders a high consequent probability only for a specific hypothesis, and not for other hypotheses. Extraordinary evidence is necessary for events with low priors, particularly when they are weighed against alternatives with substantially higher priors, precisely for the reason that a high consequent will be necessary to offset the higher priors of those alternative hypotheses. Short of this, if the consequent is roughly equal for those alternatives (or even less than the relative disparity between the priors), then it won’t be enough to offset their priors, and the alternatives will still yield a higher value for their posterior probability.

Ehrman only needs to demonstrate 1) that alternative hypotheses have a higher prior than the resurrection, and 2) that their consequent is roughly equal (or, if less, still not by a ratio greater than the disparity between the priors). If Ehrman can further demonstrate that alternative hypotheses even have a greater consequent probability than the resurrection (which I think he can), then that is gravy.

3) Emphasizing that such Ancient Literature is Not Extraordinary Evidence

I would recommend to anyone who argues contrary to Craig’s variation of the “minimal facts” apologetic to promptly point out that his “facts” are not facts. His first two don’t even have scholarly consensus, and he’s framed the latter two in misleading ways. But I think it’s perhaps even more important to bring people’s attention back to the actual evidence under consideration.

In framing the debate as a discussion of “facts,” I think Craig wants to make the audience imagine that they are staring at Jesus’ burial, the empty tomb, and Jesus appearing to his disciples as the “evidence.” But that’s hardly the case. What we are actually looking at is ancient texts, particularly Paul’s letters and the Gospels. It should further be pointed out that nowhere in Classics, Near Eastern Studies, or ancient history is there a precedent for scholars using such ancient texts to prove a kind of event as unprecedented as Jesus’ resurrection. Craig is embarking on a lone venture, one that suspiciously aligns with his religious and apologetic interests.

Once focus is correctly brought back to the ancient literature that is our actual evidence, the next step is to emphasize that this is a very weak kind of evidence, when compared to other methods of inquiry. We are not talking about medical evidence, forensic science, parapsychology, or even journalism. We are dealing with literature, the origins of which we are distanced from by thousands of years. We can speak to none of the original witnesses, nor directly investigate any of the original circumstances. Even if this is an understandable circumstance, it still means that we are very, very limited in what we can know. In situations that involve more unknowns, the consequent probability of the same evidence will very likely rise for a broader range of explanations. This is particularly bad for the posterior probability of events with extremely low priors, such as the resurrection.

Once all of this context has been laid out, Ehrman can then lay out the wide range of alternative explanations. I think it is particularly helpful to discuss diachronically how the story of the resurrection grew over time, exactly as you would expect for legendary development. It’s also helpful to lay out all of the pre-Jesus models of resurrection, ascension, and divinization in the ancient Mediterranean world. It’s not like the idea of Jesus’ resurrection came out of nowhere, and the fact that it fits so well into cultural cookie-cutter models makes a good case for how the same evidence can be equally explained (or explained better) by non-resurrection hypotheses. Those factors coupled with the observations that Ehrman already made for why the disciples would have sought theological rationalizations to justify Jesus’ crucifixion, and expected his return (a context in which his resurrection to heaven makes perfect sense), wraps up the case against the resurrection hypothesis quite nicely.

I don’t think in making these suggestions that I have changed the spirit of Ehrman’s case too much, but I do think that I have tightened the formal validity of his arguments from probability, and I think that this helps in responding to the technical point that he missed in his debate against Craig.

-Matthew Ferguson

[1] For a discussion on the issue of archaeological evidence and the resurrection, including the Church of the Holy Sepulchre, you can read a comment thread that I engaged in on the matter, which started here.


25 thoughts on “Understanding the Spirit vs. the Letter of Probability

  1. Nice work Matthew ! I can remember that debate and others since involving William Lane Craig and Gary Habermas, my pet peeve with these debates is how their opponent’s give them too easy a ride concerning their “Minimal Facts.”

  2. Pingback: Vridar » How to Improve Bart Ehrman’s Argument Against the Resurrection of Jesus

  3. I wonder why essays such as this are written.

    After all is said and done they offer nothing new, Origen dealt with such questions
    in his reply to Celsus.

    Any argument via Probability has to have as its foundation what all possible events can be
    and we cannot know that in terms of History.

    If we lost all the records of the 20th Century and were looking back in time and came across a paragraph that said a one time homeless man led Germany into World War II…who would
    believe it as a Historical Fact ? Or that Cortez conquered the mighty Aztec Empire with so
    few men or that Gavrilo Princip happened to be eating a sandwich when ArchDuke Ferdinand
    and his wife Sophia were driven by and their driver took the wrong turn and thus World War I
    began…what are we to make of Saint Joan of Arc or Einstein’s Miracle year of 1905 ?

    You can, if you wish, just believe that Saints Peter, Stephen, Paul, the James all just
    created an elaborate fiction to allow them to run a Ponzi scheme – or were just wholly
    deluded about Jesus – the “Last prophet of the Jews” – but then why did they die for
    Christianity when they could have easily taken their ill gotten gains and avoided preaching about
    Jesus ever again, once the Jews demanded their deaths and the Roman began to
    persecute them.

    [ By the way, why the late dating of the Gospels ? Good Heavens man, Mark/Matthew
    and Luke were written before the Fall of Jerusalem in 70 A.D. John either before
    or soon after. ]

    I admire the clarity of the writing but to attempt to constrain God to human expectations
    or to expect History, as partial as it is, to provide complete answers is naive.

    As for the Theologians cited:

    Hans Kung is not a New Testament Scholar.

    Ute Ranke-Heinemann has so many axes to grind against the Church and
    Patriarchy she cannot think straight.

    John Dominic Crossan is a populizer of his own views which are rather shallow.

    Alfred Loisy had agendas against the Traditions of the Church and thought it was
    his vocation to bring the Church into what he considered “Modern Times”.

    John Shelby Spong was another version of Loisy but without the intellectual equipment.

    Rudolf Bultman changed his mind so many times, you never knew where he stood.

    None of the above compare to Saint Paul in intelligence and lived experience, none
    of them come even close to the intelligence, lived experience and expertise of Origen.

    Let us suppose that the Heavens parted and God in all His Heavenly Glory appeared
    in the sky to all humans and the blind could see Him and the Deaf could Hear Him
    and all those who had died in the past year were resurrected and testified to
    God being God and their being a Heaven and a Hell and Satan the accuser of all

    All is this is filmed by millions of cameras.

    Then the Heavens close.

    Yes, humans might stop sinning for a few days and then we would go on sinning.
    Some might join monasteries, others might become missionaries.

    In time new generations would grow up and sooner or later pseudo-scholars
    would declare it all to be a massive hoax. Just an attempt by your grandparents
    to get you to believe in their God…and essays like the above would be written.

    In the end it is a matter of faith.

    What is sad is that those who write these blogs don’t understand what
    faith is and how it freely defines your whole being unlike knowledge claims.

    • Well, the main reason I wrote this essay is because apologists like William Craig, Gary Habermas, and Mike Licona try to take the resurrection to a level beyond faith, and to argue that it is a probable historical fact. Since they first brought it to that level, I offered a rebuttal to their arguments from probability (or inference to the best explanation, which is really just the same thing). If no apologist had ever made such argument, then I probably would not have written this essay.

      With the Hitler analogy, even if we lost all records of the 20th century, presumably we would have evidence of wars and political leaders from other periods of history. Hitler would be a false analogy, then, to something like the resurrection. A political leader starting a war would have a reference class documented in other periods of history, whereas something like the resurrection would not. We could grant it, therefore, a much higher prior probability. And if we only had the paragraph to work with (and no evidence to the contrary), it would probably yield a higher posterior than doubting the paragraph.

      As for the whole persecution and martyrdom of the apostles argument, it’s built on even flimsier evidence than the resurrection (mostly Christian apocrypha or later patristic propaganda dating to centuries later). And none of the Jewish or Roman sources say that any Christians were persecuted for belief in the resurrection. Pliny says Christians were killed for refusing to sacrifice to Pagan idols (something that non-Christian Jews were already persecuted for). Josephus says James was killed for a dispute over the Law (something Jews frequently killed other Jews for). Tacitus says Christians were killed on charges of arson. I go through the evidence of each apostle’s death, and show how they don’t strengthen the case for the resurrection, here:


      The Gospels can’t be granted earlier dates before 70 CE with any certainty. External patristic quotations of them don’t begin to appear until around the early- to mid-2nd century and we have no manuscript evidence dating that soon. They don’t state when they were written (unlike historians like Dionysius of Halicarnassus), and internal clues are often speculative. I’m open to early dates, but really we have to be open to a wider range of dates, which makes the later dates entirely plausible (and really, most scholars would agree with the estimates I gave). I debated evangelical scholar Craig Evans on the dating of the Gospels a little over a year ago, and the video can be viewed here:


      Finally, I highly doubt that if the video footage you describe of God opening the heavens, filmed by millions of cameras, were preserved that there would be nearly as many people apostatizing as you claim. It wouldn’t be an issue of generational gaps at that point, as you imply, since at that point there would be documented evidence that could carry over across generations. Sure there would be a few loons and pseudo-scholars, but such evidence would win over vastly more people than there would be without the millions of cameras.

      I think I get the concept of faith quite well, and apologists like Craig want to step beyond it. That’s why this essay was written in response to Craig. It wasn’t written to target faith in the resurrection.

  4. I agree that Ehrman’s understanding of probability leaves much to be desired, but I question whether we can ever have evidence of a miracle.

    Evidence is an effect from which we infer a cause. If we come across a body with a knife sticking out of its back and that knife has little swirly patterns that match those on a particular person’s fingers, we deem this to be evidence of who put the knife there. We can do this because we understand the natural processes of cause and effect that lead to those swirly patterns appearing on objects other than fingers. We also believe that they act consistently, if not invariably. If we thought those patterns appeared randomly or by divine fiat, we couldn’t infer anything from them. We couldn’t say that the murderer was Professor Plum in the library with the knife.

    Miracles may not be the least probable events by definition, but I think that they are, by definition, events that occur contrary to natural processes. I don’t think we could ever infer a miracle as the cause of any particular collection of evidence because we don’t know what effects are produced by supernatural causes.

    At best, I think we might say that no known cause accounts for the evidence.

    • I think it depends on the event in question, and how you define the supernatural. The example you gave with the knife is one of abductive reasoning or inference to the best explanation. One way of defining the supernatural is through a distinction between physical vs. non-physical. So, let’s say in the McVeigh example above, if his body acted in ways that were highly contrary to physical processes (let’s say by violating the conservation of energy), we might abduce that some force outside of physics is giving him “supernatural” abilities. One could still perhaps argue that there is a physical cause outside of known physics (possibly energy being transferred from another dimension) that is causing the same effect, but I think a miracle could still be on the table for a possible explanation. I don’t think we would have to claim it was completely unknowable.

  5. I think C&C’s statistical syllogism needs a bit of readjustment. It’s not really correct that 99% of dead people stay dead. As far as we know, the number is zero, so a statistical syllogism would give us a 0% probability of Jesus rising. But this would be question begging, so the proper procedure would be to use laplace’s rule of succession: (0+1/100 billion + 2).

    But this low probability assumes that a resurrection is even logically possible. It’s possible if God exists, not possible if naturalism is true, and possible if some other form of supernaturlaism is true. So we use the total theorem of probability to divide up the possibilities:

    P(R|medical data)=P(theism)xP(R|theism&medical data)+P(naturalism)xP(R|naturalism&medical data)+P(nontheistic supernaturalism)xP(R|nontheostic supernaturalism & medical data)

    Let’s say P(theism)=.1, P(naturalism)=.8, P(nontheistic supernaturalism)=.1 *just for the sake of illustration)

    In this case, P(R|medical data)=.1x(1/100 billion+2)+.8×0+.1x(1/100 billion+2))

    So in actuality, the prior of the resurrection is much lower than just one person out of everyone who has ever died

    • That’s all true. But I don’t see how it changes anything. the C&C premise is already the vague “99.999…999%” which includes even the result you would aim to get. So I don’t see how this point changes any premise in their argument. (And indeed I’d bet they’d agree with you and specifically said “99.999…999%” precisely on account of reasoning just like his.) So what wording is there to change?

      • I recall reading C&C saying that the prior is 1 in 100 billion. My argument shows that it is actually much lower than that. But perhaps I’m misremembering, or remembering someone else’s mischaracterization of their argument

  6. I wonder if what is called the resurrection should instead be called the recreation. We have a human brain before the crucifixion, a decayed batch of watery pulp after three days and then a Jesus walking around again after the “resurrection”. Would it not be more reasonable to consider the 2nd Jesus as a perfect copy, one with full recreated memories, etc?

    I come to this question from the discussions about uploading one’s mind to the internet. The idea is that someday it may be possible to scan my brain thoroughly so that it can be transferred to the web. But what goes to the web is a copy even though it will have a full sense of my past. It isn’t actually me.

    Of course, there are differences between these two examples but I don’t see a compelling reason to think that the Jesus who died would be the same Jesus who rose. It’s moot though since the whole thing is extraordinarily unlikely.

  7. Matthew,

    This Brian again, the one who discussed archaeological evidence with you on Celsus. I am in agreement with much of what you say, in particular your criticism of both Craig and Ehrman. By and large, your arguments are clear, sensible, and well motivated.

    That said, I am not a big fan of applying Bayesian-type arguments to cases like this. The reason is that, as you say, the consequent probabilities for the various options are likely to be similar, so the posterior probability is essentially determined by the prior probability of each. Unfortunately, the prior probabilities are set by whatever the arguer thinks sounds right and really by any evidence. That leaves the entire process ripe for confirmation bias. It’s not really a good approach to critical thinking. If anything, I think in situations like this, the Bayesian approach is really only useful in reverse. That is, one can use it to test one’s skill at setting prior probabilities.

    Here are two ways to test whether your own Bayesian arguments about the resurrection stand up. Rather than consider unicorns and Timothy McVeigh (fanciful indeed), think about, for example, the Shroud of Turin. We know, of course, that it’s a medieval artifact, thanks to the radiocarbon dating. But suppose that test hadn’t been done yet. What would you personally require to be discovered about the shroud to raise the posterior probability of Jesus’ resurrection to a reasonable, even probable, level? And what are the probabilities of those things being discovered about the shroud? For example, how would a dating to the 1st century have changed your (and scientists’) view of the likelihood of the resurrection? How likely could such a dating be before performing it? And what do those results imply about you prior probability?

    Once you’ve done that example in the comfort of your own opinions, consider a second example that requires you to put a little skin in the game. Let’s consider an event that, like the resurrection, involves 1) multiple people seeing a dead historical person, 2) a unique event unattested in all of human history, and 3) an essentially impossible occurrence. It has the advantage of having occurred in modern times under full media scrutiny. I am speaking, of course, of the apparitions of the Virgin Mary at Fatima, and in particular the “Miracle of the Sun” on October 13, 1917. Using the same arguments you applied to the resurrection, what is the prior probability of such an event? What are the consequent probabilities for various explanations, given the evidence? What do you then calculate as the posterior probability that the “Miracle of the Sun” was in fact supernatural?

    • Hey Brian,

      Thank you for your thoughtful comment. I am in the middle of working on revisions to a peer reviewed paper right now, and so I will only offer some preliminary thoughts, before I have time for the rest of your question.

      When it comes to prior probability, I agree that calculating the prior can be quite subjective in cases of new, unknown phenomena, or for events that involve a large number of complex variables. (In the latter case, this is why I am not really behind Richard Carrier’s approach to the historical Jesus; I think there are too many variables at stake for everyone to agree upon the input data and methodology that Carrier uses to calculate the prior for Jesus’ existence.)

      But in situations involving a well-documented phenomenon–one for which we have an observed frequency based on a large pool of data–the prior becomes less subjective, and can instead be objectively based on known empirical evidence. For example, we are able to calculate the prior probability of someone having cancer and the consequent probability of still receiving a false negative test, based on observed medical evidence.

      When it comes to resurrection from the dead, we are dealing with a kind of phenomenon for which we have a large pool of data and an observed frequency—namely that applying to death. Billions upon billions of people die, and yet there has never been an example, documented by medicine, of a specific individual resurrecting from the dead (in the manner Craig describes). Given this universal precedent of remaining dead and not resurrecting, based on billions of observed cases in which humans die, the prior probability of a specific individual resurrecting would thus be extremely small.

      Note too, that this medical evidence of billions of deaths and no resurrections is a “minimal fact” relevant to minimal facts arguments for the resurrection. It’s not normally stated as such, since when apologists use these arguments they usually try to gauge scholarly opinion among NT scholars and historians (not medical doctors), but I think that, when responding to resurrection apologetics, it is fair to bring to the table general background knowledge (universally agreed upon by experts in other relevant disciplines, or even among ordinary reasonable people). It also fair, for example, to state as a “minimal fact” that bodies can and have been stolen from burial sites in documented instances. I hate to use the term “common sense,” but it is.

      When calculating the prior for resurrection from the dead, therefore, one is basing the calculation on an observed, empirical set of data. I think confirmation bias plays far less of a role in this kind of calculation, because the prior is not being reached based on subjective speculation, but objective data.

      Where confirmation bias does play more of a role, however, is when apologists attempt to bring subjective or controversial theological and philosophical premises into the calculation, in order to raise the prior for Jesus’ death. Craig in the debate, for example, argued that the prior for Jesus’ resurrection is not low, if God wants to raise Jesus from the dead. But, it is not a “minimal fact” that God wants to raise Jesus from the dead, or that God even exists. It is solely Craig’s theological beliefs (which would not be agreed upon by everyone engaging the argument).

      Other considerations fall into this category, such as arguments that the prior is raised if there is a religious context, or if Jesus was known as a healer, or if Jesus demonstrated divine moral perfection. It is not a “minimal fact” agreed upon by everyone engaging the argument that any of these considerations would raise the prior for the resurrection. But I think that everyone can agree that billions upon billions of people have been documented to die, and yet there is no medical evidence of one given person resurrecting from the dead.

      The low prior is thus being reached based on far less disputed premises, derived from observed empirical data, whereas the higher prior that apologists try to argue for is being based on subjective or non-empirical theological and philosophical conjecture, which is highly disputable. I think that confirmation bias has a much greater effect when trying to assign a high prior, therefore, then it does when trying to assign a low prior.

      I think Bayes’ theorem becomes less helpful when there is less clarity about which inputs to factor into the equation. But on the other hand, I think that the theorem is very useful for illustrating, in extreme cases, the maxim “extraordinary claims demand extraordinary evidence.” If a phenomenon has a low prior, which is reached based on observed data of past cases involving the same kind of phenomenon, then mathematically the claim does require extraordinary evidence (defined as evidence for which the consequent especially favors the extraordinary prior, and only furnishes a low consequent for alternative hypotheses with higher priors).

      The maxim of extraordinary claims and evidence becomes important to assessing the weight of the evidence in the case of Jesus’ resurrection, when it is primarily drawn from ancient literature (and certain considerations of archeological evidence, which I think we agreed last time is still in murky waters; for example, I think we agreed that the Church of the Holy Sepulchre still involves a lot of unknowns, even if there are certain arguments in favor of identifying it as Jesus’ burial site.)

      In the past, some apologists have compared the evidence for Jesus’ resurrection to that of Julius Caesar crossing the Rubicon. The logic goes that, if we have X number of sources dating to X number of years after the event, and the sources for Jesus measure up to those for Caesar, then we should accept the resurrection as a historical fact as much as we accept the Rubicon crossing as a historical fact. But this kind of approach only factors in the consequent evidence, without considering how the prior increases the posterior for Caesar’s case (that generals cross rivers during times of war is an observed phenomenon), while lowering the posterior for Jesus’ case (the prior odds of a person resurrecting in that manner is extremely low, based on observed medical fact). It’s a false analogy, therefore, to claim that these two events demand the same quality of evidence. Bayes’ theorem is very useful for demonstrating this fact, and the calculations can be based on observed empirical data.

      Say, for example, that you tell me that you have a car. What kind of evidence would I demand for you to prove it? I would probably just take your word, and wouldn’t even ask to see the keys or something. Why? Well, because I observe that most adults in my time and region do, in fact, possess cars and they seldom lie about it. But what if you told me that you personally possessed a nuclear weapon? Well, I would demand much more evidence for that, because private individuals virtually never possess nuclear weapons, based on observed data.

      Even the nuclear weapon, though, would be a case of an extremely rare *particular* phenomenon, but it would still be *general* background knowledge that nuclear weapons exist. The case of Jesus’ resurrection is more like you claiming that you possess an inter-stellar spaceship. That kind of claim does not even belong to *general* background knowledge of existing technology, not to mention that it would be an extraordinary *particular* case.

      For non-paranormal accounts of what happened to Jesus after he died, however, it is *general* background knowledge that billions of people die without resurrecting, and there is likewise *particular* evidence for Jesus’ death. The question is whether the ancient evidence available, claiming that he resurrected, is capable of overcoming this steep prior calculation that Jesus probably remained dead. Considering that I do not even consider Paul’s letters and the Gospels to be of the same quality as an ancient historical author like Suetonius, and I do not even believe in Suetonius’ miracles, I wouldn’t regard the NT accounts to be that extraordinary.

      For me then, I would thus say that the resurrection has a low posterior. The event is intrinsically very unlikely, and the NT accounts are unpersuasive evidence. Interestingly, however, where I think there is more subjectivity in reaching this conclusion (and perhaps confirmation bias) actually pertains to the consequent, more so than the prior. I am very confident that the prior for Jesus’ resurrection is astronomically low, since it’s based on observed medical evidence. But, assessing the historical reliability of the NT accounts is a far more nuanced, since the historical reliability of ancient narratives is a complex question, involving many variables.

      That’s why I spend a lot of time studying Paul and the Gospels in the context of other Jewish and Classical literature from the same time period. I’m interested in learning more about how to assess the evidence–and I must admit that it is an ever-growing process, since the scholarship, data, and methodologies are very complex.

  8. Matthew,

    Thanks for the reply. With regard to your calculation of the prior of Jesus’ resurrection, I agree that the prior is likely “low.” I do question whether it is “astronomically low.” The example I mentioned that you should analyze might help you see why. I understand that you’re busy. All I ask is that you give it a try when you have a chance. Since that might be a while, if you e-mail me when you’ve looked at them, I’ll be happy to chat about it.

    Aside from those examples, one reason to doubt your low prior is that it depends on the category you’ve created. It’s true that we have few, if any, examples of resurrection as you describe it. But there are many, many reports of miracles of various kinds by religious faithful. The Catholic Church has documented boatloads of them. So do we count the resurrection as a unique type, or do we count it as another possible miracle? How that category is chosen can dramatically change the prior. What about the essence of the resurrection data, which is that early Christians claimed to have seen him? Apparitions and visions are a dime-a-dozen among the religious. There were 386 publicly claimed apparitions of Mary alone during the 20th century. That’s one every 3 months. Taking into account that the resurrection can be viewed as largely an apparition event, that greatly increases the prior also.

    I certainly agree with you that it’s easy for believers to greatly overestimate the priors. That’s because they find certain scenarios more realistic than non-believers do. But non-believers are bedeviled by similar biases–they just cut the other way. Coming up with an unbiased estimate is nearly impossible when the thing being estimated cuts close to peoples’ values. In that sense, I think Bayes’ rule can be more useful when used to calculate the prior based on an unbiased estimate of the posterior (i.e., before measurements/observations are available). The posterior can often be further removed from peoples’ values than the prior. I think you’ll see what I mean if you work out the examples.

    Finally, the issue of “extraordinary evidence” is a funny one. It’s really not necessary for the evidence to be extraordinary at all, in the sense of being special. All that’s needed is for 1) the conditional probability of the dubious claim to be higher than all the other conditional probabilities combined, and 2) the resultant posterior to be observed. The priors then get automatically updated to favor the unexpected one. As applied to the resurrection, while you are correct that resurrections are nearly unknown, claims of resurrections of actual historical persons (as opposed to resurrection myths) are (I believe) also very rare. When considering the various causes of such claims–hallucination, fraud, wishful thinking, actual resurrection, etc.–I suspect that the predicted likelihood of the claim would be seen as very small as long as the resurrection prior is small (because hallucinations, fraud, etc. in this context are rare). But it then follows that the observation of such claims greatly increases the prior of the resurrection. Again, you’ll see what I mean if you work through the calculation yourself.

    • Hey Brian,

      Sorry for the delay on this. I just had my paper accepted for publication, and I also am catching up on dissertation work. I’ll get to this discussion when I have the time. As a note, long comments like this can often absorb hours of time in responding. It’s a lot to commit to when you are a graduate student, and so I often invite people to make a small donation for my responses to long comments like this on PayPal:


      It’s not a prerequisite for my response. I just tend to distinguish long comments from shorter ones, on time commitments like this.

      • I have no criticism of your arguments, Matt, except that you could probably facilitate more powerful rebuttals to the apologists if you took their detailed points and made them the subject of singular blog posts. I think you cover so much material all at once that while a few intellectuals and academics can stay put through it all, you might be losing many of the doubting fundamentalist Christians who are like most other average people and simply tune out in sight of so much text.


      • Yes, I do have a bad habit of writing long blog posts. I’m going to work this year on breaking up my discussions into smaller posts, now that I have a base set of essays which lay out my topics, theories, positions, and arguments in fuller detail.

    • Alright, I have dissertation work today, but I also can’t fall asleep at the moment, so I figure that I will use this time to get back to you:

      “Aside from those examples, one reason to doubt your low prior is that it depends on the category you’ve created. It’s true that we have few, if any, examples of resurrection as you describe it. But there are many, many reports of miracles of various kinds by religious faithful. The Catholic Church has documented boatloads of them. So do we count the resurrection as a unique type, or do we count it as another possible miracle? How that category is chosen can dramatically change the prior.”

      First off, the definition of a “miracle” (defined as God or a supernatural agency intervening in the physical order to cause events that are ordinarily not physically possible) is so broad that hypothetically a lot of conceivably miraculous events could have their prior raised, simply because of a wide abundance of miracle claims around the world, along the line of reasoning that you are arguing. We could say, for example, that God spontaneously transforming me into a pickle was a type of miracle. One could argue that the prior of this event is low, because we never scientifically observe people suddenly being transformed into pickles. But, what if we don’t consider this miracle to be a “unique type,” and simply lump it in the same category with all the other miracle claims around the world? Does the prior probability of me being turned into a pickle suddenly increase? In other words, I don’t see how having a bunch of apples increases your probability of getting an orange.

      Second, let’s turn this argument in reverse, and consider how a scientifically observed resurrection event would increase the prior for Jesus’ resurrection. Let’s say that we did have a medically documented case of someone dying, returning to life on the third day, and also having a transformed body with super-human abilities. Certainly this kind of event would drastically raise the prior for Jesus’ resurrection, much more so than having a bunch of unverified claims around the world about prayers for money being answered, sightings of the Virgin Mary, or even people resuscitating from the dead after a few minutes of death. In other words, that the correct reference class is in fact resurrection events, and not just miracles defined broadly, can be shown by how much an documented resurrection would raise the prior for Jesus’ resurrection, as opposed to a hodgepodge of miscellaneous miracle claims.

      Third, the circulation of tons of miracle reports, unverified by science and often falsified by investigation, would actually lower the prior for Jesus’ resurrection, since it shows that stories about miracles easily circulate without being based on a veridical event. So simply having a bunch of unverified miracle reports is not the same thing as a scientifically verified miracle event, let alone a scientifically verified resurrection event.

      “What about the essence of the resurrection data, which is that early Christians claimed to have seen him? Apparitions and visions are a dime-a-dozen among the religious. There were 386 publicly claimed apparitions of Mary alone during the 20th century. That’s one every 3 months. Taking into account that the resurrection can be viewed as largely an apparition event, that greatly increases the prior also.”

      We have tons of apparition events around the world, but none that I am aware of that we could verify by an actual person returning to life. In contrast, there are a bunch of verified natural explanations for these events, such as hallucinations, visionary experiences, and false rumors. So how would this data increase the prior of a veridical resurrection event? If anything, it would lower the value of the consequent, since it would show that the stories of Jesus’ post-mortem appearances could have actually emerged from purely natural causes, like the ones described above.

      “I certainly agree with you that it’s easy for believers to greatly overestimate the priors. That’s because they find certain scenarios more realistic than non-believers do. But non-believers are bedeviled by similar biases–they just cut the other way. Coming up with an unbiased estimate is nearly impossible when the thing being estimated cuts close to peoples’ values.”

      I don’t see how this consideration cuts both ways, when believers and skeptics are working with two very different kinds of evidence. Theologians like Richard Swinburne, for example, have argued that God existing would raise the prior of resurrection, or that Jesus living a perfect moral life and teaching moral truths would do so. But, for one, there is no academic consensus that these premises are true at all. Nor are these conclusions reached through verifiable scientific studies. In this case, Swinburne’s data is far more subjective and speculative.

      But the skeptical position that the prior for the resurrection is low, because we medically observe that billions upon billions of people die with no resurrections, and that people can hallucinate or have visions of dead people, is actually based on scientific facts. This kind of consideration is far less subjective, and can be empirically demonstrated. So I think the skeptical view is actually working with much more concrete data than the arguments put forward by theologians like Swinburne.

      “As applied to the resurrection, while you are correct that resurrections are nearly unknown, claims of resurrections of actual historical persons (as opposed to resurrection myths) are (I believe) also very rare.”

      This depends on the reference class. Richard Miller has shown that divine translation, resurrection, and post-mortem appearance claims were actually quite common in antiquity for king-like figures such as Roman emperors, and it is quite relevant in this context that Jesus was considered to be “King of the Jews.” Likewise, Miller has shown that demigods such as Romulus and Hercules were commonly said to have been resurrected. Even if these are mythical figures, it’s important to remember that Jesus was said to have been conceived by a deity with a virgin, and so the adaptation of mythical concepts to a historical figure is not really that unusual.

      Another phenomenon, related to the case of Jesus, even if a slightly different phenomenon, is the post-mortem appearances of popular and controversial figures after their death. Elvis Presley has had multiple people claim that they saw him alive after his death, and there were likewise similar stories about the Roman emperor Nero and the Russian princess Anastasia. Curiously, what all of these individuals have in common is that they died relatively young, and so many people asked the “what if” question of whether they could have survived. The fact that Jesus supposedly died around age 30 aligns him quite closely with figures like this, even if the stories of his post-mortem appearances may have taken a somewhat different shape in casting him as a resurrected figure.

      “When considering the various causes of such claims–hallucination, fraud, wishful thinking, actual resurrection, etc.–I suspect that the predicted likelihood of the claim would be seen as very small as long as the resurrection prior is small (because hallucinations, fraud, etc. in this context are rare). But it then follows that the observation of such claims greatly increases the prior of the resurrection.”

      I would argue that the story of the resurrection increases the prior of hallucinations, fraud, body theft, etc. If we have never verified that a genuine resurrection has taken place, but we have verified that people frequently have hallucinations of dead people, that divine translation was a common literary trope in antiquity, and that false rumors can circulate that someone is still alive on earth after death (cf. Nero), then all of these kinds of phenomena would have a higher prior than the resurrection. And no, I disagree that “in this context [these things] are rare.” They’re actually the kind of phenomena that surround false resurrection stories and stories of people appearing alive after their death.

      “Let’s consider an event that, like the resurrection, involves 1) multiple people seeing a dead historical person, 2) a unique event unattested in all of human history, and 3) an essentially impossible occurrence. It has the advantage of having occurred in modern times under full media scrutiny. I am speaking, of course, of the apparitions of the Virgin Mary at Fatima, and in particular the “Miracle of the Sun” on October 13, 1917. Using the same arguments you applied to the resurrection, what is the prior probability of such an event? What are the consequent probabilities for various explanations, given the evidence? What do you then calculate as the posterior probability that the “Miracle of the Sun” was in fact supernatural?”

      Well, let’s see. The important thing to stress about this kind of event is that all the phenomena you describe is *interconnected*. It’s not like people having mass hallucinations/delusions and a story about a revered dead figure appearing to her followers are two unlikely things that we have to assume happened at the same time, and independently of each other. One happened because of the other, and vice versa. If we consider that a group of people holds to beliefs in the supernatural, believes that their saints appear to them after death, and are looking for an example of it, they’ll find what they want. So a group of people who wanted to see the Virgin Mary relied on a solar illusion to provide a basis for their expectations. Also, this is not a “a unique event unattested in all of human history.” I know that there was a similar solar miracle in Conyers, Georgia that occured in more recent times. It’s an attested phenomenon of mass delusions. If we grant the conditions of 1) prior belief in the supernatural, 2) looking for an example of a saint, and 3) relying on a solar vision, I would argue that the prior probability of this kind of event is quite high, the consequent is quite low, since we don’t have anything like a video recording of the Virgin Mary appearing, or special evidence like that. We have the kind of evidence we would expect for ordinary activity, mixed in with people wanting to see a miracle. I would argue that the posterior is quite low.

      “Finally, the issue of “extraordinary evidence” is a funny one. It’s really not necessary for the evidence to be extraordinary at all, in the sense of being special. All that’s needed is for 1) the conditional probability of the dubious claim to be higher than all the other conditional probabilities combined, and 2) the resultant posterior to be observed.”

      This is similar to the definition of an extraordinary event that I gave above. In the case of the solar miracle, we have all the evidence we would expect for an imagined miracle event among a group of people who were looking for a miracle. There was no solar abnormality, documented by science, which would suggest a departure from physical cause and effect. We know that people can have mass sightings of a dead figure (especially when they are looking for it). In contrast, the hypothesis of a miracle would require the sun doing abnormal things (like dancing in the sky), never before observed. That gives it a low prior. The people witnessing the “miracle” gave inconsistent reports and can’t back up the miracle with any hard evidence. That’s a low consequent, and combined we get a low posterior. In this case, therefore, we don’t have “extraordinary evidence.” The evidence in question is not rendered probable, only in the case of a genuine solar miracle. It’s outweighed by more common naturalistic hypothesis that explain the same evidence just as well (if not better).

      “Rather than consider unicorns and Timothy McVeigh (fanciful indeed), think about, for example, the Shroud of Turin. We know, of course, that it’s a medieval artifact, thanks to the radiocarbon dating. But suppose that test hadn’t been done yet. What would you personally require to be discovered about the shroud to raise the posterior probability of Jesus’ resurrection to a reasonable, even probable, level? And what are the probabilities of those things being discovered about the shroud? For example, how would a dating to the 1st century have changed your (and scientists’) view of the likelihood of the resurrection? How likely could such a dating be before performing it? And what do those results imply about you prior probability?”

      The fact that the Shroud of Turin first emerged in Medieval times already assigns it a low prior of authenticity. We know that enough fake crosses were circulating in that time to build Noah’s Ark. The only thing unique about the shroud is that it is extremely well-made, which one might argue raises its consequent a bit. But the fact that we have no ancient testimony for it, and that we can’t even explicitly connect the image with Jesus, even if it were genuine, lowers the consequent. What I would personally require as evidence for the shroud (in the absence of radiocarbon dating) is some kind of ancient testimony or physical evidence that connected it with Jesus. If the radiocarbon dating had dated it to the 1st century CE, I think that would raise the prior for it being authentic, but it would still be speculative whether the image is that of Jesus. Of course, we also have to ask whether wrapping a shroud around someone would even create such an image (as opposed to a stretched out face), but that’s another question…

  9. mr ferguson

    i quote :
    18 “Truly, truly, I say to you, when you were younger, you used to gird yourself and walk wherever you wished; but when you grow old, you will stretch out your hands and someone else will gird you, and bring you where you do not wish to go.”
    19 Now this He said, signifying by what kind of death he would glorify God. And when He had spoken this, He said to him, “Follow Me!” (Jn. 21:18-19 )

    does this text show that jesus predicts that peter is not willing to die for his beliefs?
    it says “where you do not wish to go”

    if he does not wish to go to some place that would cause his death, then does that mean he does not wish to die?

    • I doubt that it’s diminishing Peter’s resolve. The passage is probably just saying that Peter will be led to a fate that is undesirable in general, but it notes that Peter’s death would glorify God, and thus would still be honorable. While this passage vaguely predicts Peter’s death, it doesn’t give any of the specifics. And, since it is part of the appendix to the Gospel of John, it’s quite late in composition, and was probably written after Peter had died. The author may have inferred Peter’s death, based on the date of composition, but not known the specifics of what happened. It’s hard to say.

      • But it remains that if Peter did not wish to endure this death, he cannot be included among those whom modern apologists say were willing to die for their beliefs.

        17 He said to him the third time, “Simon, son of John, do you love Me?” Peter was grieved because He said to him the third time, “Do you love Me?” And he said to Him, “Lord, You know all things; You know that I love You.” Jesus said to him, “Tend My sheep.
        18 “Truly, truly, I say to you, when you were younger, you used to gird yourself and walk wherever you wished; but when you grow old, you will stretch out your hands and someone else will gird you, and bring you where you do not wish to go.”
        19 Now this He said, signifying by what kind of death he would glorify God. And when He had spoken this, He said to him, “Follow Me!” (Jn. 21:17-19 NAU)

        In the text, it is the author, not Jesus, who interprets Peter’s death as glorifying God, and in either case, this text seems to establish, under Christian assumptions, that the death of even an unwilling martyr would yet glorify god, so that the original Christian belief would have viewed the death of disciples in persecution as glorifying god even if any particular disciple was less than willing to die. Hence this passage does legitimate violence to the popular apologetics argument that the disciples were willing to die at the point at which they were put to death, and this in turn strikes at the apologetics claim that the disciples were mightily transformed. There are other reasons in the NT why the transformation of the disciples at seeing the alleged risen Jesus was something less than “mighty”.


  10. Mr Ferguson
    Bart Ehrman believes that Jesus hung on the cross for several days. Why do you think that Paul narrates that Jesus rose on the third day ? If we assume that Paul was not dependent on Jewish scripture for the “third day” what else could be the possible reason?

    • Third day motifs were common in ancient literature. In the Certamen of Homer and Hesiod, for example, Hesiod has his dead body thrown into the sea, which is then carried to the shore by dolphins on the “third day” after his death. When it comes to Jewish use of the motif, the early Christians may have been inspired by Jonah in the whale. One thing worth noting is that my dissertation advisor, Andrew Zissos, thinks it is highly unlikely that the Romans would have removed the body for burial so quickly, if they really had given Jesus a crown of thorns and labeled him “King of the Jews.” The use of decorations like that implies that Jesus was meant to hang on the cross for a while, to spread a message against sedition. This doesn’t mean that he was never removed for burial, but it does support Ehrman’s view that Jesus hung on the cross for several days.

  11. When a Christian starts using complex mathematical formulas and philosophical theories to defend his belief in first century corpse reanimation-transformation (aka:
    resurrections)…I yawn.

    I yawn because it is soooo silly.

    I know for a fact that if a Muslim attempted to use these same ploys to defend the veracity of Islam’s claim that Mohammad flew to heaven on a winged horse, the very same Christians would snicker and hand-wave away these arguments without giving them a second thought, believing that these tactics are nothing more than an obvious, desperate attempt to dress up a superstition as believable reality.

  12. Pingback: links – Site Title

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s