Davis on the Rationality of Belief in the Resurrection

Davis on the Rationality of Belief in the Resurrection (2005)

Michael Martin


From 1998 to 2000, I engaged Christian apologist Steven Davis in an exchange on the rationality of belief in the Resurrection in the pages of Philo.[1] Subsequently, I assumed that the debate had ended, but to my surprise Davis recently published an article in Philosophia Christi that revisits our exchange and criticizes many of the arguments I raised there. Why Davis has returned to the fray after a three-year hiatus is not completely clear. In the introduction to “The Rationality of the Belief in the Resurrection: A Reply to Michael Martin”[2] (RMM), Davis complains that I will always be given the last word in an exchange between the two of us in Philo given the journal’s ideological orientation. Are we to suppose that Philosophia Christi, a publication of the Evangelical Philosophical Society, does not have a distinct ideological orientation? Or that in the pages of Philosophia Christi Davis would not be given the last word in an exchange between us? He also complains that he had more to say when the debate ended in Philo.

I am happy to obligeDavisand will therefore continue the exchange in this paper, although whether it is fruitful to do so is an open question. As we shall see, in RMM Davis fails to address my main argument, and when he does take up subsidiary issues he often raises points that I rebutted earlier.

Readers of the Secular Web will need some background to understand the Davis/Martin exchange on the Resurrection. I did not argue that belief in the Resurrection was irrational for non-Christians. Had I done that, my task would have been relatively easy. Rather, I argued that belief in the Resurrection was irrational for Christians. To be sure, asDavissays in RMM, a Christian is someone who believes “virtually by definition” that Jesus was resurrected. The crucial question is whether this belief is rational for Christians. My claim was, and still is, that it is not. The gist of my argument is that the initial probability of the Resurrection is low even for Christians. This would not be a problem if the historical evidence for the Resurrection were very strong. But it is not. It is mediocre at best, which is not nearly strong enough to overcome the initial improbability of the Resurrection.

Since a significant part of the exchange relied on Bayes’ Theorem, let me say a few words here about this theorem.


R = The Resurrection
K = Background knowledge
~R = Alternative theories to R
EH = Historical evidence

Then, according to Bayes’ Theorem:

P(R/K&EH)    =                        P(R/K) x P(EH/R&K)                          
[P(R/K) x P(EH/R&K)] + [P(~R/K) x P(EH/~R&K)]

This formula can be used to show that given a low initial probability, the probability of the Resurrection is less than 50% even if the Resurrection explains all existing historical evidence. Suppose that the initial probability of the Resurrection is 15%; that is, suppose that the probability of the Resurrection R, given background knowledge K, is .15–i.e., P(R/K) = .15. Suppose further that the Resurrection completely explains the historical evidence. Here the historical evidence is entailed by R and K, and consequently P(EH/K&R) = 1. The probability of the Resurrection could still be less than 50% since any value for P(EH/K&~R) more than 0.185 will give P(R/K&EH) a value less than 0.5. Even if the probability of the historical evidence given alternative explanations and the relevant background knowledge is merely 20%, belief in the Resurrection is still not rational [P(R/K & EH) < .5]

The Initial Improbability of R

Why assume that P(R/K) is low for Christians? Suppose that the background knowledge K of the Resurrection consists of three of the four items that Davissuggests in RMM (p. 503): (1) God exists and sometimes performs miracles; (2) God wants to redeem human beings; (3) God in the OT promises life after death. The truth of (1)-(3) is compatible with the initial improbability of the resurrection of Jesus as described in the NT. As I pointed out in my replies to Davis, God could have redeemed human beings in many different ways. The bodily resurrection of Jesus on the third day after having died on the cross is merely one way. Since Davisgives no reason to prefer one way to any other a priori, he tacitly seems to assume that all ways are equally probable. So if there were, for example, just eight possible ways in which God could redeem humanity, only one of which was the one God in fact used, Davis must suppose that the a priori probability of this way would be 1/8 or .125. Thus, given the background knowledge (1)-(3), the belief that Jesus died on the cross and arose on the third day and so forth is initially improbable. This scenario is just one of many that God could have chosen. Of course, strong historical evidence could overcome this initial improbability.

Davis’ latest reply to my challenge misses the point. For example, he maintains that someone who accepts (1)-(3) as background knowledge will find R more initially probable than someone who does not. I have never denied this, for it is irrelevant to my main point. Acceptance of (1)-(3) is compatible with P(R/K) being very low. The fact that P(R/K&(1&(2)&(3)) is greater than P(R/K) does not mean that P(R/K) is not very low.Davisdoes not explain why he supposes otherwise.

However, Davisbelieves that a fourth assumption is both part of a Christian’s background knowledge and would raise the initial probability of the Resurrection: (4) Jesus predicted his own death and resurrection. Supposing that (4) is part of the background knowledge that Christians should rationally accept as true, it is not clear how (4) should be understood. It can be understood as a historical claim solely about what Jesus preached: Jesus preached Q (that Jesus will die and be resurrected). On this interpretation the truth of Q is not assumed. However, it can be also interpreted so that Q is true.

On the first interpretation, (4) has no force in showing that P(R/K) is not low. Q can be entailed by K and yet P(R/K) can still be low. There are still multiple ways that God could have redeemed humanity. Jesus’ alleged prediction does not change that. But on the second interpretation, since R is entailed by Q, and Q is entailed by K, P(R/K) = 1. On this interpretation, the rationality of R is decided trivially since it is entailed by K. Consequently, there is no need to bring in historical evidence and the debate between Davis and me should have been over before it began.Davisseems to suppose that a trivial outcome in our debate is not a defect. Yet at the beginning of his paper he says that the question of the rationality of belief in the Resurrection is important, and implies that the answer is neither easy to come by nor trivial. Indeed, he spends over twenty pages trying to answer my arguments. All of this would be unnecessary if P(R/K) = 1. So I will assume here that (4) does not entail the truth of R.

In RMM (p. 503)Davissuggests that my intent was to argue that (1)-(4) are themselves improbable and that R should be judged on the basis of ordinary secular beliefs. But this is not the case. As should be clear from the above, I assume for the sake of the argument that (1)-(4) are true and maintain that this assumption is compatible with P(R/K) being low. The only qualification I make is to exclude any interpretation of (4) in which P(R/K) = 1, which I am convinced thatDaviswould agree with on reflection.

In my earlier paper (CRR, p. 54), I considered a historically abstract definition of Christianity in which the son of God could have been incarnated in other times and places. Note that I do not assume this definition of Christianity here, although it would be easier to make my case if I did. Rather, I assume (withDavis) a historically concrete definition of Christianity (CRR, p. 54). In other words, I do not assume that God could have redeemed humanity by the Incarnation of the son of God (or in other ways) in any times and places other than first-centuryPalestineand still have Christianity as it is usually understood.

The Particular Time and Place Argument

In my exchange withDavisI used what I called the particular time and place argument (PT&PA). In RMM (p. 505)Daviscalls it my weakest argument and is surprised that I keep defending it. Before I explain and defend it again, it is important to understand that I really do not need this argument to make my case. Even if PT&PA is rejected, the initially probability of P(R/K) is still low because there are many a priori equally likely ways that God could have redeemed humankind. Different times and places need play no important role in making this point.

Nevertheless, PT&PA is really quite simple and worth defending. Just as God can redeem humanity in many different ways–only one way being through the death and resurrection of his incarnated son–so God can redeem humanity in many different times and places. He does not have to perform the redeeming act at the particular time and place he does either by the resurrection of his son or in other ways specified by Scripture. Consequently, the initial probability of the Resurrection occurring when and where it is alleged to have occurred is low.

Note that this is true even if one adopts a historically concrete definition of Christianity. Let us suppose, as Davisdoes, that Christianity by definition originated in first-century Palestine. So any religion that could have been founded on a redeeming act by God in other times and place, such as third-century B.C. China, would not be Christianity and thus out of place in the present discussion. But God still had many different times and places to perform his redeeming act in first-century Palestine. Indeed, even if one restricts the location to a more limited temporal-spatial location within first-century Palestine there will be many alternatives. For example, suppose Jesus arose from the dead at place P1 and at time T1 in first-century Palestine. Surely, it is possible that God could have arranged for Jesus to arise from the dead at any number of different spatial locations within a forty-mile radius of P1 and at any number of different times within 20 years plus or minus T1. Consequently, the initial probability that Jesus arose at place P1 and at time T1 is low. Historical evidence may be able to overcome this low initial probability, but that it will remains to be seen.

It is important to see that PT&PA is compatible with the assumption that by definition Christianity arose in first-centuryPalestine; that is, it is compatible with a historically concrete definition of Christianity. Second, even though Christians maintain that the Resurrection occurred in or nearJerusalemaround 30 A.D., the initial probability of this time and place could still be low. If God could have chosen other times and places in Palestine and there is no a priori reason that we know of for him to pick the place and time that he did, the initial probability of this place and time is low.Davisgives no argument to the contrary and merely asserts that I am mistaken.

Atonement and Arbitrariness

Throughout our exchange Davis and I argued about the arbitrariness of the Atonement. I claimed that existing theories of the Atonement either do not explain why God sacrificed his son in order to save humankind or else make God’s choice arbitrary. This is clearly relevant to establishing the low initial probability of the Resurrection. If I am correct, then the lack of plausible explanations of the Resurrection or the seeming arbitrariness of such accounts strengthens my thesis that the probability of the Resurrection is low. If there are indeed no plausible explanations of the Resurrection, or if the explanations of it are arbitrary, then we have no good reason for supposing that God had grounds for choosing the Resurrection over other options that would have redeemed humankind. But if this is so, the initial probability of the Resurrection is low.[3] There would be no a priori reason to expect God to bring about the Resurrection rather than other redeeming scenarios.

Davis’ responses to my criticism of the Atonement are unclear (to say the least) and may be inconsistent. At one point Davis says that God’s actions with respect to the Atonement are arbitrary and that he finds nothing wrong with this (TROR, p. 44). But on the same page he also seems to suggest that the Atonement (as it is understood by Christians) is plausible. One might be tempted to infer that the Atonement is not arbitrary, at least to Christians. In his latest responseDavistakes the position that although the Atonement seems arbitrary, it is not. After reviewing the details of my arguments and his responses in our past exchanges, in the end he says:

Clearly, the issue between Martin and me depends on what is meant by the term arbitrary. Apparently…. he meant, ‘not based on any reasons at all.’ And I meant, ‘not based on any reasons that are clear to us.’…. I insist that God’s actions could well have been arbitrary in my sense. That is, God’s reasons for choosing the death and resurrection of Jesus of Nazareth in first centuryPalestine as the crucial atoning acts are obscure to us, and perhaps will always remain so. But of course God had reasons for the divine choices, and they were and are good reasons” (RMM, p. 509).

So in the latest installment of his defense of the rationality of belief in the Resurrection, Davis admits that the Resurrection seems as if it was arbitrary (not based on good reasons), but still asserts that it was based on good reasons despite this appearance. Notice thatDavisprovides no justification for discounting this appearance except his mere assertion that God does have a reason. Surely a justification of his claim that a reason exists is called for; its absence adversely affectsDavis’ thesis that belief in the Resurrection is rational for Christians. This is especially true givenDavis’ early claim that God’s choices are sometimes arbitrary.

First, if the most important event in the history of the world (according to Christians) seems to be without reason, Christians have grounds for complaint. If God cannot explain why redemption must be through the resurrection of Jesus, he could at least assure Christians that good reasons exist even though he cannot reveal them. But althoughDavishas assured us, God has not. Second, this appeal to unknown reasons suggests, contraDavis, that belief in the Resurrection is not a rational belief for Christians. This is true for at least two reasons.

First, as already argued, belief in the Resurrection is not rational for Christians because the initial probability of the Resurrection is low. Moreover, the choice of the Resurrection is arbitrary unless one makes the gratuitous assumption that God has good undisclosed reasons for the Resurrection. So, in the end, the alleged rationality of belief in the Resurrection is dependent on faith in God’s nonarbitrariness despite the appearance of arbitrariness. But why should that faith be maintained? Even if it should be maintained, it could hardy be called rational.

Second, if it is permissible to fall back on the ad hoc hypothesis that God has undisclosed reasons wheneverDavisis pressed with the apparent arbitrariness of the Resurrection, a similar ploy is possible whenever the very probability of the Resurrection itself seems in doubt. Suppose that I show that in the light of evidence and arguments it seems as if P(R/EH&K) < .5. Then couldn’tDavisuse the same sort of argument he used to defend the non-arbitrariness of explanations of the Resurrection? Could he not press that reasons known only to God make P(R/EH&K) > .5? If so, the exchange between Davis and me has no point. No matter what I say,Daviscould claim that reasons known only to God support the opposite conclusion.

Strong and Neutral Resurrection

In RI Davis maintained that Jesus was resurrected in a strong sense (SR). By this he meant that God raised Jesus from the dead, and that the raised Jesus could do supernatural things such as materializing in rooms and moving instantaneously from one place to another. But in our long exchange in Philo,Davis refused to defend SR and defended instead what he called the neutral sense of resurrection (NR)–the mere fact of Jesus being resurrected. To be sure, it is very much toDavis’ advantage to defend NR rather than SR. The initial probability of SR is lower than NR: a priori it is the case that P(NR/K) > P(SR/K). So it is perfectly understandable whyDavis refused to defend SR. But is it justified?

As before, in his latest paper Davisattempts to defend his refusal. But I am as unimpressed with his latest rationalizations as I was with his earlier ones. He argues that the context of our exchange in Philo was different from the context of his book RI and that this justifies his refusal to defend SR. Yes, there are differences, one of which is that in the context of Philo his arguments are subject to a higher standard of scrutiny that they cannot meet. As I said in CRR (pp. 55-56):

Davisbelieves there is nothing wrong with his defending the rationality of Neutral R rather than the rationality of Strong R. I would have thought there is a lot wrong with it, however. In my original Philo paper which started this debate I cited Davis’ advocacy of Strong R. Now that his views are threatened it seems all too convenient for him to say that he only wishes to defend Neutral R. After all, Davis is the one who asserted Strong R–I am not the one who brought it up. Moreover, it seems fair to assume that Strong R is what Christians generally understand by R. So again, if this exchange concerns the question of the rationality of R for Christians, thenDavis’ refusal to defend the rationality of R thus interpreted is difficult to understand.

In my Philo papers I argued that since the initial probability of the Resurrection is very low, Davis must show that the historical evidence for alternative accounts is extremely weak. Davis rejects this by maintaining that for Christians the initial probability of the Resurrection is not nearly as low as I allege. How Davis can continue to claim this is hard to understand. As I have shown in Philo and again here, there is every reason to suppose that even for Christians P(R/K) is very low. To review: First, God has many ways of redeeming humanity and salvation through the resurrection of Jesus is only one way. Second, even if one restricts salvation scenarios to first-century Palestine, there are many times and places that God could have chosen for his redemptive purposes. Third, Davis admits that there are no known acceptable theories of the Atonement and that the Atonement seems arbitrary. Davis insists that there is an unknown explanation for the necessity of the Resurrection to redeem mankind and that the Resurrection only seems arbitrary. But he gives us no reason to believe that this claim is true. Fourth, if one takes the Resurrection to entail what it does in Davis’ Risen Indeed–strong resurrection (SR)–then the initial probability is even lower than otherwise. ThatDavis refuses to defend SR gives one confidence that P(SR/K) is indeed very low.

What this means is that Davishas his work cut out for him. He must show that the probability of alternative accounts of the Resurrection is very low. It is not enough, asDavis supposes, to raise criticisms against these alternatives; but this is exactly whatDavis does. In any case, these criticisms are easily challenged, as I shall proceed to show.

1. The Empty Tomb

In RTD (pp. 67-68) and CRR (pp. 56-57) I raised several arguments against the traditional account of the empty tomb. (As an alternative to the traditional account, I suggested that the location of Jesus’ tomb was not known.) For example, I questioned the historical accuracy of the traditional account in terms of Roman and Jewish burial customs. These customs suggest that either Jesus was not buried at all, or was ignominiously buried by his enemies in an unmarked grave. Surprisingly, Davisdoes little to combat this evidence in his Philo papers, and does nothing at all to combat it in RMM.

As I argued in Philo, many Biblical scholars disagree with the traditional account. Despite Davis’ repeated characterization, these are not radical scholars, they are simply not conservative ones. With Davis, I concur that the truth about the Resurrection is not decided by majority rule; I never implied as much. But the profound disagreement of New Testament scholars over it suggests that the traditional account is questionable and that alternative accounts are not easily dismissed. I also argued in Philo that if the location of Jesus’ tomb was known it would have been venerated, as were the tombs of ancient saints.Davis questions my analogy with ancient saints and says that it is “far too weak” to support my idea (RMM, p. 512). But it does not seem weak to me, and in any case my arguments were based on more than this analogy. See, for example, my citation of the Holy Sepulchre Church (CRR, p. 57).

Legend Explanations

In “Christianity and the Rationality of the Resurrection” (CRR) (Philo 3, 2000) I gave five reasons why the Resurrection can be plausibly understood as a legend (pp. 57-58). For example, I pointed out that the accretion of details as the Resurrection story got older is typical of legends. Davis points out that this is sometimes true of real historical events. While that’s true, it doesn’t invalidate the point that the Resurrection could be a legend since it has many salient features of legends. Remember that Davis must show that the probability of the evidence for the legend explanation is very low relative to the Resurrection account. Davis has not done this. As I repeatedly stressed in the past, it is not necessary for me to show that the legend account is very probable–only that it is not improbable (CRR, p. 56).

In addition to giving reasons to doubt that the legend account is improbable, I give reasons to suppose that the probability of the Resurrection account is low. For example, in Philo I also argued that neither New Testament sources such as Paul and other early letter writers, nor Jewish and Pagan sources, confirm many of the details of the Resurrection story as it appears in the Gospels. Strangely, in RMM (p. 512) Davis cites me (CRR, p. 57) to rebut this. But there I pointed out thatDavis admits that Paul does not cite the details of the Resurrection.

Moreover, I also noted that there have been many cases of religious people who hold irrational beliefs no matter what the evidence, and that this fact weakens the probability of the Resurrection. Davisreplies that I have not shown that the earliest Christians were like this. But barring some relevant difference between the earliest Christians and the irrational religious believers that I give as examples, there is a prima facie case that these cases can be treated similarly.Davis gives no relevant dissimilarity.

Davis also argues that there is big difference “in feel” between reading pagan stories and reading New Testament accounts, which suggests that pagan stories are mythical whereas New Testament stories are not. This appeal to how the stories “feel” is a very subjective basis for such an important judgment. Moreover, some of the cases that I cited are not pagan, but Jewish and later Christian cases (RTD, p. 67). Finally,Davisfaults me for not telling my readers the origins of the Resurrection legend. But just because one does not know the origin of an alleged legend does not mean it is not a legend. For example, if scholars lacked knowledge of the origin of the William Tell legend, they could still have good reasons to suppose that it is a legend. To be sure, the legend explanation would be strengthened if the origins of the legend were known. But ignorance of the origins of a story does not make a legend explanation improbable. And as I have already pointed out, I do not need to show that a legend explanation is probable.


In CRR I pointed out that given the low initial probability of the Resurrection,Davismust show that it is extremely improbable that the post-Resurrection appearances can be partly explained by hallucinations. In RMM (p. 513)Davisthinks that he does this by citing certain factors that were present in the context of these appearances. For example, he maintains that there is no evidence that the people who claimed to see Jesus were on drugs or had a high fever. But I showed in CRR and RTD that in analogous historical cases of hallucination, drugs and fever were also absent.Davisis also impressed by the claim that some people recognized Jesus with difficulty and even doubted that he was Jesus. However, it is well known that the skepticism of characters in myths is used as a literary device to stress the reality of miracles performed by the hero. Given this background, it is not surprising that if the Resurrection story were based on legend and hallucinations it would develop complete with skeptical characters (See RTD, p. 69.)

2. Inconsistencies

In my Philo papers I pointed out a number of significant inconsistencies among the Gospels regarding the empty tomb story.Davis does not deny these but claims that there are also significant agreements in the story. While this is true, my main point is that the inconsistencies would lower P(EH/R&K) to some extent, which in turn would lower P(R/EH&K). How much they would lower P(EH/R&K) is uncertain, butDavis offers no reason to think that they would not lower the probability by a significant amount. After all, in order for P(EH/R&K) to be high if EH is inconsistent, EH must be probable relative to R&K. It is certainly not obvious that this is true. Other things being equal, this in turn lowers P(R/EH&K). But since P(R/EH&K) < .5 even if P(EH/R& K) = 1, this cannot be good news forDavis.

3. Eyewitness Testimony

In my Philo articles I argued that the eyewitness testimonyDavis depends upon is unreliable.Davis admits that this evidence is unreliable with respect to the details but not with respect to the essential facts.Davis is apparently unaware of the great errors that have come to light in eyewitness testimony in the investigation of UFOs and other paranormal phenomena. For example, the planet Venus has been identified hundreds of times by eyewitnesses as an alien space ship. Clever conjurers have induced eyewitnesses to believe that they saw people walk through solid walls and make various objects (including an elephant) disappear into thin air. Eyewitnesses have reported the sighting of fairies, little green men, ghosts and monsters. It is an abuse of language and the facts to say that these eyewitnesses were wrong just about the details.

Two Kinds of Rationality

In TROR (pp, 49-50)Davisintroduced a different concept of rationality into the debate. In RMM (pp. 513-516) he denies this but, as far as I can determine, up to this point in the exchange Davis and I had both assumed that the Resurrection R was rational only if P(R/EH&K) > .5. In the conclusion of TRORDavismaintained for the first time that even if P(R/EH&K) < .5, R is still rational so long as P(R/EH&K) > the probability of any specific alternative to R. In CRR (pp. 59-60) I rebutted this idea by raising two critical points, neither of which are answered byDavis. First, I showed that this new concept of what is rational has absurd consequences. So long as this probability is higher than any specific alternative to R, it would entail that R is rational even if P(R/EH&K) =.000000000000000001. Second, I questioned whetherDavishas shown that none of the specific alternatives has a higher probability than R.


Once again Davis and I come to the end of a chapter of our exchange, and I remain unimpressed with his arguments. Once again he concludes that unless I am able to spell out the details of some alternatives to R the probability of any alternative is less than the probability of R. But in CRR (p. 60) I argued that lack of detail is not fatal to maintaining that one alternative has a higher probability than another. The probability of alternatives to R is a function of many things. Detail is only one of them.

In any case, Davis’ criticism that alternative explanations lack detail while the traditional account provides it can be considered a subtle form of begging the question.[4] Alternative explanations lack details because their proponents recognize that the Easter narratives do not provide accurate reports. Davis is in the grip of an illusion when he believes that the stories of the burial, the empty tomb and so on provide a detailed narrative. Davis scorns the appropriate historical vagueness of a good historian, who understands that when accurate details are missing one must be vague. Any other stance is dogmatism. Imagine the level of vagueness of the Easter narrative if scholars only used facts that Biblical scholars–not just conservative ones–agreed were true. Would the level of vagueness between this honest Easter narrative and the alternative accounts be any different? This is to be doubted.


[1] In Philo 1 (Spring-Summer 1998), pp. 63-73, I published the essay “Why the Resurrection Is Initially Improbable” (WTR). WTR criticized the case Davis had made in favor of the Resurrection in his Risen Indeed: Making Sense of the Resurrection (RI) (Grand Rapids, MI: Eerdmans, 1993). Davis replied in “Is Belief in the Resurrection Rational? A Response to Michael Martin” (IB) in Philo 2 (1999), pp. 51-61. I answered Davis in the same issue of Philo in my “Reply to Davis” (RTD), pp. 62-76. Davis replied to that essay in “The Rationality of Resurrection for Christians: A Rejoinder” (TRDR) Philo 3 (2000), pp. 41-51, and I answer in the same issue in my “Christianity and the Rationality of the Resurrection” (CRR), pp. 52-63.

[2] Steven Davis, “The Rationality of the Belief in the Resurrection: A Reply to Michael Martin.” Philosophia Christi 3 (2003), pp. 501-517, hereafter RMM.

[3] Low because there are other ways of saving humans besides the Resurrection. So, for example, if there are ten ways to save humans, the initial probability of the Resurrection would be 10%.

[4] I owe this point to Robert Price in personal correspondence.

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 )

Google+ photo

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

Connecting to %s

%d bloggers like this: