Why you shouldn’t believe the resurrection happened

The 12th (and final) part in “20 Atheist answers to questions they supposedly can’t.”

  1. What accounts for the empty tomb, resurrection appearances and growth of the church?

Short answer: We shouldn’t be that confident of these facts, but in any case the base rate fallacy and selection bias nixes the confirmatory power.

Longer answer: The argument implied in the question is that the historical record of Jesus provides strong evidence to believe he actually died and rose again, which provides evidence that Christianity’s central claims (e.g. God exists, Jesus is the son of god) are true. The question neatly summarizes the three main ‘planks’ of evidence usually offered:

  1. The Empty Tomb. When Jesus died, his body was placed in a tomb. Not only was a stone rolled in front of it, but also the authorities posted sentries outside the tomb to stop anyone stealing the body. Despite this, the stone was discovered to be rolled away, and the body had gone. (e.g. Mark 16:4, Luke 24:2-3)
  2. Resurrection appearances. Several different groups of people (the disciples, some women, etc.) are reported to have seen Jesus after he died. (e.g. Luke 24:15-31, 36-48; Matthew 28:9-10)
  3. The growth of the church. After Jesus died, his apostles (and figures like Paul) were committed to the message of Jesus, and helped the church spread rapidly. (cf. Acts, but also the historical record re. the Holy Roman empire, etc.)

The idea is this data is very hard to explain via purely atheistic means. Maybe Jesus didn’t really die, but is it plausible he could have got up and escaped the guarded tomb after being crucified and speared for good measure? Maybe the disciples managed to steal the body, but how did they manage to get that past the guards? (And what was in it for them? Why would many of them go on to die for a belief they knew to be false?) Maybe the appearances of the resurrection were just hallucinations, but how could there have been so many hallucinations, of so many different people, and why didn’t the authorities just squash the story by presenting the public with Jesus’s corpse?

So, it’s argued, the best explanation for the historical data is the Christian one: Jesus rose from the dead and left the tomb miraculously, and then appeared to people like the apostles and women who visited the tomb, and these people, convinced by the truth, to go on and grow the church.


Something seems to be going wrong with this sort of argument. We could imagine a ‘Magicianist’ making a similar sort of case in defence of the claim they  could perform supernatural feats of magic:

“You saw me pull that rabbit out of the hat! How do you explain that naturalistically? I showed you it wasn’t in there already, I showed you the table didn’t have a hidden compartment, I showed you there was nothing up my sleeves. So what other explanation can you offer that explains what happened better than me really making the bunny appear by magic?”

However, offering parodies doesn’t really get to grips with what (if anything) is going wrong with the historical argument from the resurrection. So let’s pick firstly at the fact pattern, and secondly at the pattern of inference we are making.

Minimal facts?

For many of the events surrounding the resurrection depicted in the bible, we only have the Bible’s word for it. Extrabiblical attestation (e.g. TacitusJosephus) note the beliefs of Christians rather than substantiating their content, and archeological finds support only peripheral details (e.g. that there was a guy called Pilate alive and in the right position of power). For example, there is no support besides the gospel’s say so that the tomb was guarded, or even that it was empty.

So a lot hinges on how credible the resurrection accounts are, and there are some factors that count against their credibility:

One: the authorship of both Matthew and Luke is disputed, with the balance of scholarly opinion is that they were not written by the disciples they were claimed to be, and only one (Mark), does the balance of scholarly opinion agree with the ‘canonical’ claim of who wrote it.[ref]Aside: I’m neglecting for current discussion the Gospel of John, a later ‘non-synoptic’ gospel. For what it is worth, no one really knows who wrote that either.[/ref] Pseudoepigraphy is not a problem per se (the writer could still be correct even if he is not who we thought he was), but it does throw into doubt what information the author (whoever they were) had access to when compiling their account: were they recording eyewitness testimony, or second or third hand accounts? If they were to write something mistaken, were they near any original eyewitnesses who would correct them? How exactly did the oral traditions percolate for the 20-60ish years until the synoptic gospels were written?

Two: the authors of the synoptic gospels are not impartial observers. These are members of (at the time) a new religious movement, and the gospels were written, at least in part, to persuade others. So the chances of them omitting or distorting or briefing against adverse evidence is substantial. (And that is arguably what we observe, e.g. Matt 28:11-15)

Three: The synoptic gospels show signs of being copied from one another, and yet still have resurrection accounts that significantly contradict. The considerable textual similarity between the three synoptic gospels suggests that there was copying going on, with the best bet that Matthew and Luke used Mark and a lost ‘Q’ document in their formation. So the synoptic gospels are not ‘independent witnesses’ giving the same story.

In any case (and despite this) they disagree with respect about what happened at the resurrection. Different groups of people visit the tomb at different times to be told different things by different angels. As Ehrman surmises:

The way to see differences in the Gospels is to read them horizontally. Read one story in Matthew, then the same story in Mark, and compare your two stories and see what you come up with. You come up with major differences. Just take the death of Jesus. What day did Jesus die on and what time of day? Did he die on the day before the Passover meal was eaten, as John explicitly says, or did he die after it was eaten, as Mark explicitly says? Did he die at noon, as in John, or at 9 a.m., as in Mark? Did Jesus carry his cross the entire way himself or did Simon of Cyrene carry his cross? It depends which Gospel you read. Did both robbers mock Jesus on the cross or did only one of them mock him and the other come to his defense? It depends which Gospel you read. Did the curtain in the temple rip in half before Jesus died or after he died? It depends which Gospel you read.

Or take the accounts of the resurrection. Who went to the tomb on the third day? Was it Mary alone or was it Mary with other women? If it was Mary with other women, how many other women were there, which ones were they, and what were their names? Was the stone rolled away before they got there or not? What did they see in the tomb? Did they see a man, did they see two men, or did they see an angel? It depends which account you read. What were they told to tell the disciples? Were the disciples supposed to stay in Jerusalem and see Jesus there or were they to go to Galilee and see Jesus there? Did the women tell anyone or not? It depends which Gospel you read. Did the disciples never leave Jerusalem or did they immediately leave Jerusalem and go to Galilee? All of these depend on which account you read.

One can offer ‘harmonisations’ of these accounts, but these tend to be a bit implausible: why believe there were several groups of people all doing these things, and each gospel bizarrely reported the goings on of one group but no others, as opposed to one sequence of events they disagree upon?[ref]This also undermines how much we can trust the traditions that formed the gospels to be error correcting, as obviously some of these accounts must be erroneous, and yet they persisted, despite the gospel writers (allegedly) having access to eyewitness accounts.[/ref]

Those presenting these reasons are sensitive to these sorts of concerns, and hence the focus on minimal facts. Even if the textual evidence is a bit ropey by some lights, it still lends pretty strong support to the main planks of the minimal fact account: that there was an empty tomb, that people reported seeing Jesus after he died, and the church grew after his death. How can atheism explain that?

Particular and general explanations

It may not be so hard to come up with a plausible atheistic account for the ‘minimal facts’ after all. Take this one from Ehrman’s debate with Craig:

Jesus gets buried by Joseph of Arimathea. Two of Jesus’ family members are upset that an unknown Jewish leader has buried the body. In the dead of night, these two family members raid the tomb, taking the body off to bury it for themselves. But Roman soldiers on the lookout see them carrying the shrouded corpse through the streets, they confront them, and they kill them on the spot. They throw all three bodies into a common burial plot, where within three days these bodies are decomposed beyond recognition. The tomb then is empty. People go to the tomb, they find it empty, they come to think that Jesus was raised from the dead, and they start thinking they’ve seen him because they know he’s been raised because his tomb is empty.

This account has problems: if there was a failed tomb robbery, why didn’t the Jewish authorities make a big deal of it to crush this nascent cult? Wouldn’t Jesus’s death be a ‘big enough’ event for Roman soldiers to do more than kill the robbers and throw Jesus’s body away too? But it doesn’t seem too hard to patch this account up to vague plausibility (maybe the soldiers were incompetent, maybe they wanted a cover up, etc.)

To return to our parody, though, it seems mistaken to demand atheist give a particular plausible account for the resurrection on pain of conceding it was the work of god. Rather we should consider the collection of possible naturalistic counter-explanations. So if for the bunny rabbit sleight-of-hand explanations seem not too implausible (despite the ability of magician to knock down particular sleight-of-hand explanations offered), perhaps, so long as candidate naturalistic explanations are not too implausible (even if they do not match the facts as well as divine resurrection), the ‘not-god’ explanations do not collectively fare worse than the Christian one.

Priors and Miracles

One elephant in the room is that resurrection (or magic) is not another mundane explanation of a fact pattern. Like magic, or teleportation, an act like the resurrection violates all sorts of strongly supported beliefs we hold about the world (people don’t come back to life after being killed). In short, it would be a miracle.

It is hard to see how we should consider miraculous explanations. Hume’s famous statement that – for any miracle – it is always more plausible the account is mistaken than the miracle occurred seems to go too far: if God appeared to everyone at once in an inter-subjective way, that would seem excellent reason to think God existed, even if it broke our then-understanding of how nature worked.[ref]There’s a discussion to be had about whether – if God existed – his action in the world would be truly supernatural, instead of just ‘facts about nature we are not well aware of’, but bracket that[/ref]

Yet it seems fair to give miraculous explanations a pretty low prior probability – given what we know about the world around us, someone rising from the dead seems extremely unlikely, so we would need very powerful evidence to persuade us it really having happened.

Picking the right reference class is hard, though. One could say (as Craig occasionally does), that although it is really unlikely that Jesus rose naturalistically from the dead, the Christian case is that Jesus rose supernaturally from the dead, and so saying supernatural occurrences are improbable just begs the question against Christianity. This doesn’t seem right – the same words transposed into the Magician’s mouth doesn’t seem plausible; in the same way we shouldn’t describe certain objects (like God) as supernatural, and thus impossible to substantiate, we shouldn’t ring fence them and demand different rules apply. Although naturalism being false should not have a really low prior probability (and asserting it does probably approaches question-begging), it seems fair to say the particular sort of supernaturalism that involves bodily resurrection is strongly discomfirmed by our inductive experience of the dead staying dead.

Given our experience of the world around us, resurrection (divine or not) is not a common occurrence, so it is under a big prior penalty – we would want really compelling evidence against a possible naturalistic explanation of a resurrection before believing the dead might not always stay dead, in the same way we would want really compelling evidence Magicians tricks have no naturalistic explanation before suggesting magic. So perhaps Hume’s dictum applies in this case: it is more probable the resurrection accounts are mistaken, than there really was a resurrection, and the ancient texts and scholarly opinion do not provide convincing enough evidence to overturn this presumption.[ref]Of course, you may hold much more hospitable priors about bodily resurrection – religious experiences of a christian nature might make you think it is quite likely Jesus was resurrected before encountering the historical case. But this sort of historical argument is generally made to convince people whose priors aren’t slanted in a Christian-friendly direction.[/ref] Hume also cites in his discussion of miracles the Roman saying that “I would not believe it even if were told to me by Tacitus”. A similar principle should apply here: bodily resurrection is so outlandish a claim that, even if a panel of historians of antiquity went to Judea and verified the resurrection accounts in person, we should still favour some mistake being made.

Base rates and prediction

Even if naturalism was true, and there were no supernatural events, that does not mean that it would always seem to be true. There might be occasions where it really does look like supernatural things occurred, and that the naturalistic explanations seem inadequate. True theories may have inclement observations. An example from Stephen Law:

Let me tell you a story from 1967. It’s a UFO story. There were reports of a strange object appearing nightly over a nuclear power site in Wake County. The police investigated; a police officer confirmed it was about half the size of the moon, and it just hung there over the plant. The next night the same thing happened. The deputy sheriff described a large lighted object; the county magistrate saw—and I quote—:

A rectangular object, looked like it was on fire, we figured it about the size of a football field; it was huge and very bright.

There was, in addition, hard data, a curious radar blip reported by local air traffic control.

Now what’s the best explanation of these reports? We have multiple attestation, we have trained eyewitnesses, police officers’ putting their reputations on the line by reporting a UFO. We have hard, independent confirmation, that blip on the radar scope. Surely, then, it’s highly unlikely these witnesses were, say, all hallucinating, or lying, or just saw a planet. Clearly, by far the best explanation, you might think, is that they really did see a large lighted object hovering close to the plant.

But here’s the thing: we know pretty much for sure that what was seen by those police officers was the planet Venus! Journalists arrived upon the scene, were shown the object, and chased it in a car. They found they couldn’t approach it. Finally, they looked at it through a long lens and saw it was the planet Venus. That radar blip was just a coincidence.

What does this show? Every year there are countless amazing reports of religious miracles, alien abductions, ghosts, and so on. In most cases it’s not easy to come up with plausible, mundane explanations for them. Sorry, it is easy to come up with plausible, mundane explanations for them! But not all, right? Some remain deeply baffling. So should we believe in such things, then? No, for as my UFO illustrates… story illustrates, we all know that some hard to explain reports of miracles, flying saucers, and so on are likely to crop up anyway, whether or not there is truth to these claims. That 1967 case could easily have been such a baffling case.

Even if we did not have the later explanation of what the officers saw, we should still not have believed UFOs existed based on that account. So in a similar way the occasional baffling account of a UFO does not lead us to believe UFOs (or that one in particular), a baffling resurrection account, if that is what the gospels are, should not lead us to believe resurrections (or that one in particular).

Abandoning a theory in light of a single (or a few) bizarre observations may not be the best approach. However, most of us suffer from the opposite problem – being too eager to ‘stick to our guns’ as counter-veiling evidence mounts against our view. If we kept having accounts of UFOs, which were generally baffling, it seems after a while we should be inclined to dumping the ‘no UFO’ theory. Most scientific progress is heralded by baffling observations; we would not get very far if we discounted observations we could not explain with our current theories.

There’s a happy medium to be struck as to when to dump the theory, and when to dump the observations, and it is likely governed by the relative strengths of the theory and observations in question. If the theory has proven itself to have general explanatory power, and usually does a good job explaining the sort of inclement observations, and the inclement observations themselves are sporadic, not repeated, and dubious, it is probably better to dump the observations – and vice versa. It seems the former account better matches the resurrection, however.

Summing up

We can formalize all of the foregoing concerns into bayes rule:

P(Resurrection|E&k)   = P(E|Resurrection&k)  * P(Resurrection|k)
P(¬Resurrection|E&k)    P(E|¬Resurrection&k)   P(¬Resurrection|k)

In english, the ratio to the left is our posterior odds: given our general background knowledge (k); once we add in the evidence (E) of the gospels, the growth of the early church, and whatever else, what are the odds of there being a resurrection? By probability, this is the product of two quotients: the one on the far right is the prior probability: given our background knowledge, how likely is a resurrection to have taken place? The middle quotient is the likelihood ratio, how likely is likely is the evidence to have occured if there was a resurrection, rather than there not being a resurrection. The mathematical mechanics imitate common-sense reasoning well: if an evident is really improbable (so the prior probability is low), we need really good evidence (so a very high likelihood ratio) to persuade us it happened.[ref]Aside: This – and only this – is how the ‘extraordinary claims require extraordinary evidence‘ canard should be read, although a more accurate phrase would be ‘Really unlikely events should only be believed due to really hard-to-explain-away evidence’.[/ref]

So we have already shown that the prior probability for the resurrection should be set pretty low. Even if we are not born again atheists or naturalists, the sort of supernaturalism/divinity picked out by ‘Jesus, son of God, died and rose again for our salvation at c. 30CE in Judea’ is extremely specific, and so the sheer weight of counterpossibles necessitate the prior being low. Our background knowledge lowers it still further: our experience of life and death, as well as our knowledge of science, suggest bodily resurrection never happens – so our confidence in this induction counts against the prior probability. In short, the evidentiary bar is set pretty high.

So what about the likelihood ratio? It seems clear that the likelihood ratio should be greater than 1. The evidence of gospels surely makes the resurrection more plausible than if it was not there (consider the hypothetical where there was no account of Jesus’s life or death or resurrection – would that make you assign a higher probability to ‘Jesus, son of God, died and rose again for our salvation at c.30CE in Judea’?) The question is how much more plausible: given the prior is small (1 in 10000? 1 in a million? worse?), we need a really powerful likelihood ratio to make the resurrection plausible.

We can see the evidence is going to fall comfortably short of this mark. Although it is hard to explain how the resurrection accounts would occur without a resurrection, it is not that hard: good sensible people deluding themselves is not that uncommon, and seems a safer bet that bodily resurrection, even if a particular account of how that occurred is not easy to find. Further, although the likelihood of this particular resurrection account is very unlikely on resurrection, the likelihood of some account similarly baffling is not so surprising, as one would expect hard-to-explain events to occur once-in-a-while.[ref]This move is not open if one has free-standing reasons to suppose the resurrection must have occurred at a particular place and time. Yet there’s no particularly compelling story why Jesus had to arrive at 30CE in Judea – indeed, there are at least a few counterveiling reasons not to, given modern society’s superior ability in recording and transmitting information.[/ref] So the likelihood ratio is not that high, and combined with the prior being very low, the resulting posterior is not that high (certainly <0.5).

Frustratingly, there is no clear way of vindicating any of these assignments: Christians will have the same sign for all of their assignments in this formula, but just of different degrees, and without clear numerical measures, it is hard to put a bound on ‘reasonableness’ – plausible offers for the prior probability likely range over many orders of magnitude. Despite this, the above shows one need not have inhospitable or question-begging prior convictions to find the resurrection uncompelling proof of God.

Further reading

The Ehrman-Craig debate is a good start on the general discussion about whether the resurrection is good evidence for Christianity.

For a different probabilistic gloss supportive of the argument, see this paper by the McGrews. For another sceptical take, see here.

Besides Wikipedia, Early Christian Writings is a good intro for those interested in the early textual data.

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 )

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