Home » Library » Modern Library » Review of The Blackwell Companion to Natural Theology

Review of The Blackwell Companion to Natural Theology


Review: William Lane Craig and J. P. Moreland (Eds.). 2012. The Blackwell Companion to Natural Theology. West Sussex, UK: Blackwell Publishing Ltd. 697 pp.


Are there any good reasons to think that God exists? The Blackwell Companion to Natural Theology, edited by William Lane Craig and J. P. Moreland, seeks to answer this question in the affirmative. This massive anthology (nearly 700 pages) assembles the world’s foremost experts on the various arguments for the existence of God in an effort to give the greatest defense of theism to date. In Craig’s words, “Our hope is that this will become the standard reference work in the area where, if you are going to dispute the existence of God or you are going to affirm it, you’ve got to go to this volume and take account of what these authors say.”[1]

The volume contains chapters on the kalam cosmological argument, the Leibnizian cosmological argument, the fine-tuning argument, the moral argument, the argument from consciousness, the argument from reason, the argument from religious experience, the ontological argument, the argument from miracles, and a defensive chapter on the problem of evil. Each of these chapters is the culmination of a career’s worth of research and reflection; consequently, The Blackwell Companion to Natural Theology is not recommended for the uninitiated. Readers without some background knowledge in the philosophy of religion, as well as some training in analytic tools such as modal logic and Bayesian confirmation theory, will have a difficult time following the arguments. However, more advanced readers will surely benefit from this work, and it simply cannot be ignored by anyone who wishes to argue that belief in God is necessarily irrational or intellectual bankrupt.

Given its breadth and depth, I cannot possibly cover every chapter in one review, so I will focus on just four chapters: the kalam cosmological argument (Chapter 3), the fine-tuning argument (Chapter 4), the moral argument (Chapter 7), and the final chapter on the argument from miracles (Chapter 11).

The Kalam Cosmological Argument

In William Lane Craig’s chapter, he teams up with physicist James Sinclair to give us the most rigorous and technical defense of the kalam cosmological argument to date. The argument goes as follows:

  1. Everything that begins to exist has a cause.
  2. The universe began to exist.
  3. Therefore the universe has a cause.
  4. If the universe has a cause, then an uncaused, personal Creator of the universe exists, who sans the universe is beginningless, changeless, immaterial, timeless, spaceless, and enormously powerful.
  5. Therefore, an uncaused, personal Creator of the universe exists, who sans the universe is beginningless, changeless, immaterial, timeless, spaceless, and enormously powerful.[2]

Much of the chapter is simply copied-and-pasted from Craig’s previous work. However, the chapter does contain some noteworthy new material, particularly an extended defense of premise 2 in which Craig and Sinclair engage with the attempts of cosmologists such as Stephen Hawking, Sean Carroll, and Anthony Aguirre to develop models of beginningless universes. In contrast to the defense premise 2, which takes up roughly 80 pages, the defenses premises 1 and 4 get less than 15 pages at the end of the chapter. This asymmetry is a weakness that runs throughout the book. In several chapters, crucial premises are defended merely in passing.

Perhaps one of the most common objections to the Kalam cosmological argument is that causes must always temporally precede their effects, and because time began with the universe, it is misconceived to ask for a cause of the universe. Craig responds only briefly, suggesting that the cause of the universe can be simultaneous with the Big Bang.[3] However, Craig cannot dodge this objection so easily. Craig says, “Given that time had a beginning, the cause of the beginning of time must be timeless,”[4] but cannot affirm this statement while also affirming the statement that the cause of the universe is also simultaneous to the beginning of time, for simultaneity is a temporal relation. On its face, it is a flat out contradiction to say that a cause can be both timeless and simultaneous to the Big Bang.

Craig tries to elucidate on the idea of a timeless cause being simultaneous to its effect by drawing an analogy from cosmology. He writes:

The initial Big Bang singularity is not considered to be part of physical time, but to constitute a boundary to time. Nevertheless, it is causally connected to the universe. In an analogous way, we could say that God’s timeless eternity is, as it were, a boundary of time which is causally, but not temporally, prior to the origin of the universe.”[5]

It is not clear how this analogy sheds any light on how a timeless cause could be simultaneous with its effect. Craig appears to be saying that although the singularity is timeless, it is still causally related to the universe. However, if the singularity is not a part of physical time, then surely it cannot be simultaneous with any point in physical time. Really, then, this analogy is an example of timeless causation, rather than simultaneous causation. The idea seems to be that if a timeless singularity can be causally related to the universe, then so can a timeless God. But if Craig is willing to say that the Big Bang singularity is outside of space-time, and is therefore spaceless, timeless, changeless, and beginningless, then why doesn’t he just say that the singularity, rather than God, is the cause of the universe? He cannot argue that the singularity is also in need of a cause, because according to his definition of “begins to exist,” the singularity did not begin to exist. So if the singularity causes the rest of the universe, and it doesn’t need a cause itself, what room is left for God?

I suspect that Craig would respond by saying that the singularity cannot be the cause of the universe, because the singularity does not have any real ontological status. Elsewhere, Craig has stated his opinion that the singularity is a mere mathematical fiction that appears only in cosmological models. It is a “mere conceptualization ontologically equivalent to nothing.”[6] If this is how Craig understands the singularity, then he should reject the idea that the singularity is capable of causing the universe, because the singularity is not real. However, then Craig must also retract his analogy to the singularity. The purpose of the analogy was to give an example of a timeless entity causing a temporal effect, and this analogy loses its force once we admit that such this situation is merely a “conceptualization,” not meant to be taken literally.

A large part of Craig’s defense of the causal principle that “everything that begins to exist has a cause” is an appeal to empirical confirmation.[7] However, as Wes Morriston has pointed out, our experience also confirms more specific versions of the causal principle that are incompatible with the existence of an immaterial creator.[8] For example, all the evidence suggests that nothing can begin to exist with an efficient cause, but not a material cause. A material cause is the stuff out of which something is made, and an efficient cause is that which actually brings about the effect. For example, the material cause of a chair is the wood, and the efficient cause is the carpenter. If God created the universe, it would mean that the universe came into being with an efficient cause, but no material cause. Never in our experience do we witness this sort of causation. This would be like a lumberjack building a log cabin without any wood.[9]

As a defeater to the claim there can be no efficient causation without material causation, Craig appeals to abstract objects. He writes:

There are also many abstract objects which seem to exist contingently and noneternally, for example the equator, the center of mass of the solar system, Beethoven’s Fifth Symphony, Leo Tolstoy’s Anna Karenina, and so forth…. Now these things all began to exist: the equator, for example, did not exist before the earth did. But if they began to exist, did they have a cause or did they come into being out of just nothing?… Many philosophers would say that these objects did indeed have causes: it was Tolstoy, for example, who created Anna Karenina. So in cases such as these (and they are legion), we do, indeed, have instances of efficient causation without material causation.[10]

For this response to work, one must adhere to Platonism, “the view that there exist such things as abstract objects—where an abstract object is an object that does not exist in space or time and which is therefore entirely non-physical and non-mental.”[11] Tolstoy’s creation of Anna Karenina would only qualify as an example of efficient causation without material causation if there really is such a thing as Anna Karenina, above and apart from its physical copies, which popped into existence in some Platonic heaven when Tolstoy thought it up. If one adopts an antirealist approach to abstract objects, then this counterexample does not work.

It is a bit disingenuous for Craig to make this appeal, given that earlier in this very chapter he denies the reality of abstract objects.[12] Indeed, he notes that the truth of Platonism would mean the downfall of his argument that an actual infinity cannot exist, because if Platonism is true then sets such as the set of all real numbers would constitute an actual infinity.[13] Craig has even gone so far as to criticize Platonists for being “cavalier and uncritical when it comes to embracing the sweeping metaphysical commitments of Platonism.”[14]

So how can Craig raise this defeater is he doesn’t even believe that abstract objects exists? His answer is that although he believes Platonism to be false, it is at least coherent.[15] But if coherence is all it takes for something to constitute a counterexample to a proposed principle, then an atheist could just as easily say that it is at least coherent to suppose that things could begin to exist uncaused, and that this counts as a counterexample to premise 1.

There appears to be some unwarranted burden shifting going on here. Earlier in the chapter, Craig addresses the Platonist challenge to his anti-infinity argument the following way:

The Realist, then, if he is to maintain that mathematical objects furnish a decisive counterexample to the denial of the existence of the actual infinite, must provide some overriding argument for the reality of mathematical objects, as well as rebutting defeaters of all the alternatives consistent with classical mathematics—a task whose prospects for success are dim, indeed.[16]

Thus, when Craig’s critics appeal to abstract objects as a counterexample to his principle that actual infinites cannot exist, Craig demands that they establish the truth of Platonism. However, when Craig appeals to abstract objects as a counterexample to the principle that everything that begins to exist has a material cause, Craig does not feel he needs to go any further than establishing the mere logical possibility of Platonism. It is not clear how he can justify this double standard.

The Fine-Tuning Argument

In his chapter on the fine-tuning argument, Robin Collins uses the tools of confirmation theory to argue that the fine-tuning of the laws, constants, and initial conditions of the universe are evidence for theism. His argument goes as follows:

  1. A life-permitting universe is very improbable if the naturalistic single universe hypothesis is true.
  2. A life-permitting universe is not very improbable if theism is true.
  3. Theism is not an ad hoc explanation.
  4. Therefore, the existence of a life-permitting universe significantly raises the probability of theism relative to naturalism.[17]

It is important to note the modesty of this conclusion. As Collins himself is careful to explain, this argument merely increases the probability of theism relative to that of naturalism. Collins writes:

[My argument] neither shows that, everything considered, T [theism] is probably true, nor that it is the most plausible explanation of existence of the universe, nor even that it is more probable than NSU [a naturalistic single universe]. In order to show that any hypothesis is likely to be true using a likelihood approach, we would have to assess the prior epistemic probability of the hypothesis, something I shall not attempt to do for T. (p. 208)

The overwhelming majority of the chapter is dedicated to establishing premise 1. However, when it comes time to defend premise 2, we are given very little to work with; a mere three pages.[18] Before looking at Collins’ argument, we must introduce a few of the terms and abbreviations that he uses:

EMA = the universe contains embodied moral agents.
TSU = the theistic single universe hypothesis.
k′ = our total background information minus the fact that a particular constant C has a life-permitting value.
Lpc = a constant C has a life-permitting value.
Wh = whatever else God must do over and above creating the universe with the right laws, constants, and initial conditions to ensure that it contains embodied moral agents, such as God’s intervening in the unfolding of the universe.[19]

What Collins wants to establish is that it is not improbable that God would create a universe in which the physical constants have values that fall within the narrow life-permitting range. That is to say, he wants to defend the claim that, for any given constant, ~P(Lpc/TSU & k′)<<1. Collins argues that P(EMA|TSU & k′) = P(Lpc & Wh|TSU & k′) = P(Wh|Lpc & k′ & TSU) × P (Lpc|TSU & k′).[20] Because P(EMA|TSU & k′) is necessarily less than or equal to P(Lpc/TSU & k′), Collins argues that if he can show that the former is not very improbable, then neither is the latter.[21]

Why should we accept that P(EMA|TSU & k′) = P(Lpc & Wh|TSU & k′)? Collins does not say, but I suspect his reasoning goes something like this[22]: if God wanted to create embodied moral agents, he had two options. He could either fine-tune all the constants to fall within the life-permitting range and do whatever else was necessary to bring about these agents, or he could not do this. Thus, by the theorem of total probability[23]:

P(EMA|TSU & k′) = P(Lpc & Wh|TSU & k′) × P(EMA/Lpc&Wh&TSU&k′) +P(~(Lpc & Wh)|TSU & k′) × P(EMA/~(Lpc&Wh)&TSU&k′)

From here, Collins argues that P(EMA/~(Lpc&Wh)&TSU&k′)=0, because Lpc&Wh are necessary conditions for EMA.[24] Thus, the equation reduces to this:

P(EMA|TSU & k′) = P(Lpc & Wh|TSU & k′) × P(EMA/Lpc&Wh&TSU&k′)

From here, I suspect Collins would go on to argue that P(EMA/Lpc&Wh&TSU&k′)=1, because Lpc&Wh&TSU&k′ entails the truth of EMA. Thus, this term drops out of the equation and this is how Collins arrives at the conclusion that P(EMA|TSU & k′) = P(Lpc & Wh|TSU & k′).
But why should we think that P(EMA/~(Lpc&Wh)&TSU&k′)=0. Collins’ answer is that “k′ includes the laws of nature as part of the background information. Hence, P(EMA/~(Lpc&Wh)&TSU&k′)=0, since God could only bring about EMA if he changed the laws of nature.”[25] However, this would only be the case if God could only bring about outcomes that were consistent with the laws of nature, and therefore Lpc&Wh would be necessary steps that God must take if He wanted to create embodied moral agents. But why should we think Lpc&Wh are necessary conditions for EMA? Granted, they are physically necessary conditions for EMA. Without Lpc&Wh, it would be physically impossible for embodied moral agents to exist. But this is no problem for an omnipotent God. If God wants to create embodied moral agents, he is completely unrestrained by the laws of nature. He doesn’t need to assign the constants life-permitting values. He can bring about life no matter what values those pesky little constants have. Even if the cosmological constant is so large that the universe would blow apart if left on its own, there is nothing stopping God from supernaturally holding galaxies together in order to support life. Perhaps if Collins believed in a limited god who needed to operate within the laws of nature and couldn’t perform miracles, then P(EMA/~(Lpc&Wh)&TSU&k′)=0. But that is not the God we are talking about. Omnipotence does not mean that God is like a very powerful human who can do everything that physics allows. Rather, omnipotence means that God can do everything that logic allows. Cosmologist Sean Carroll makes this point:

God doesn’t need to fine-tune anything. We talk about the parameters of physics and cosmology: the mass of the election, the strength of gravity. And we say if they weren’t the numbers that they were then life itself could not exist. That really underestimates God by a lot, which is surprising from theists, I think. In theism, life is not purely physical. It’s not purely a collection of atoms doing things like it is in naturalism. I would think that no matter what the atoms were doing God could still create life. God doesn’t care what the mass of the electron is. He can do what he wants. The only framework in which you can honestly say that the physical parameters of the universe must take on certain values in order for life to exist is naturalism.[26]

Consider an analogy. Imagine that Ted wants to buy a computer. Because Ted cannot get to the computer store without taking 5th street, the probability that he will take 5th street, given that he wants to buy a computer, is very high—P(5th/wants computer)≈1. But now imagine that Ted has a jetpack. Now that Ted has a way of getting to the store without taking 5th street, P(5th/wants computer) is lower. Now let’s imagine that Ted can also teleport. This lowers P(5th/wants computer) even further, because Ted’s methods of getting to the store have increased. Now let’s imagine that Ted has the ability to create computers ex nihilo. Because Ted doesn’t even need to go to the store to get a computer, P(5th/wants computer) has gone down even further.

The point of this analogy is that as a person’s methods of achieving a goal increases, unless we have some antecedent reason to think they would prefer one method over another, we should be less confident that they will use any given method. Just as the probability that Ted will take 5th street was high when this was the only way he could get a computer, the probability that God would fine-tune the constants would be high if this was the only way he could create embodied moral agents. However, just as P(5th/wants computer) goes down in light of the fact that taking 5th is only one of many ways Ted could achieve his goal, P(Lpc/TSU & k′) goes down in light of the fact that fine-tuning the constants is only one of many ways God could achieve his goal.

For any given constant, there are an infinite number of possible values that God could assign it. Let’s label these values V1, V2, V3,…V. For the sake of simplicity, let’s assume that only V1 is life-permitting, and all other values are life-prohibiting. For any given constant, P(Lpc/TSU & k′) will not be low if there is some reason to think God would assign the constant V1. But what reason is there to think God would prefer V1 over V2-V? We cannot simply assume, as Collins assumes, that God would prefer V1 because V1 is a physical requirement for life, because physical requirements do not constrain God. Nor could we argue that it is easier for God to assign the constant V1, rather than perpetually perform a miracle by sustaining life in a non-fine-tuned universe. For an omnipotent God, no action is harder than any other. Perhaps a theist may argue that God would prefer to assign V1 because He finds a fine-tuned universe more aesthetically pleasing than a non-fine-tuned universe. But how could we possibly know what God would find aesthetically pleasing, especially in regards to something as oddly specific as which values he likes best for the physical constants?

Maybe Collins could argue that it would be irrational for God to give the constants life-prohibiting values if he planned on just intervening to make life possible anyway. Isn’t this inefficient? Why not just set the constants to the right values in the first place? The answer to these questions depends largely on God’s personality. Is he like a manufacturer or an artist? Imagine a machine that can automatically churn out a sculpture in a few minutes. Although using this machine would be attractive to a manufacturer who values efficiency, an artist would prefer to carve the sculpture herself, because she enjoys the creative process. If God is like the artist, then he would want to assign the constants values between V1 and V, because then He could actively participate in the universe, rather than just give the constants the correct values and let the universe create life on its own. Craig has no problem imagining why God might want to create a life-prohibiting universe and then intervene to make it life-permitting:

Mitigating factors pertinent to one’s desired ends easily override the importance of the aesthetic value of efficiency. Would we dare to call an artist wanting in aesthetic value for preferring the creative labor of executing his oil on canvas rather than simply having, if he could, the finished painting? I suggested that the Creator likewise perhaps delights in the work of creation…. [The noninterventionist view of God] condemns artists, chefs, and boys building model airplanes as persons who “‘delight’ in doing something inefficient, irrational or aesthetically disvaluable.” The point is that the delight of engaging in creative activity can itself be a justification for what the rationalist deems inefficient and aesthetically disvaluable activity…. [W]hat if His goals include, not merely the having of a created order, but the divine pleasure of fashioning a creation? By focusing too narrowly on the end product, [deists] fails to see the wider purposes which God may have in view.[27]

Additionally, God might prefer to assign constants life-prohibiting values between V2-V, because it would be a way for God to reveal himself to His creation. To discover that we live in a universe in which life cannot possibly exist would be like discovering God’s fingerprint in the cosmos.

Ultimately, there is no principled way to discern which value God would want to assign any given constant because His omnipotence completely frees up His options and we have no direct access to His preferences on the matter. Thus, it seems that we must appeal to Collins’ “restricted principle of indifference.”[28] Because we have no reason to think God would prefer any one value for the constants, we must assign them all equal epistemic probabilities. Thus, the probability of God choosing V1 for any given constant is 1/∞. Thus, P(Lpc/TSU & k′) is just as low as P(Lpc/NSU & k′), and therefore Lpc is not evidence for theism over naturalism. To defeat this argument, Collins would need to present some reason to think that God would favor V1 over all other possible values, which so far he has not done.

This same logic applies to the initial conditions of the universe. Even if stars and galaxies could not form if the expansion rate was too high or too low, this is irrelevant for an omnipotent God who can simply hold stars and galaxies together by a miracle. Additionally, although the universe must have begun in a state of low entropy in order for life to arise if naturalism is true, any old state of entropy would do if theism is true. Collins approvingly quotes Roger Penrose, who says, “In order to produce a universe resembling the one in which we live, the Creator would have to aim for an absurdly tiny volume of the phase space of possible universes.”[29] But this is quite obviously false. An omnipotent God wouldn’t need to aim at anything. He could have started the universe in a state of extraordinarily high entropy, and then stepped in at any point in time to miraculously lower the entropy. He was under no obligation to set the universe up in the beginning such that it would naturally evolve to its present state. He could have started it off in a state that, left alone, would lead to a lifeless universe, but then intervene to set the universe on a path towards life. Christian philosopher Chad Meister argues that God might have reasons to start the universe off in a life-prohibiting state:

[I]t could be the case that God intended to intervene in the early stages of the universe in order to ensure that living organisms, including human beings, would eventually evolve. It is not, necessarily, a sign of poor or irrational planning on God’s part to do so. It could be that, unlike the clockmaker universe posited by the deists, God is creatively involved in the universe at different stages of its development. While this may not be the most efficient way to create a universe, the God of theistic religions is not primarily concerned with running out of energy … if such a transcendent reality does exist, then it seems likely that it would have a purpose for the universe, and that it would be able to accomplish this purpose with or without the utilization of natural laws.[30]

Thus, there is no particular reason why God would need or want to fine-tune the initial conditions of the universe. Since we have no reason to think that God would favor any particular initial state of the universe, we can apply the principle of indifference and conclude that the probability of God choosing a low entropy state is 1 in 10^10^123—the same probability of the universe starting in a low entropy state on naturalism.

Collins’ argument for the necessity of gravity is similarly misguided. He writes:

If no such force existed, then there would be no stars, since the force of gravity is what holds the matter in stars together against the outward forces caused by the high internal temperatures inside the stars. This means that there would be no long-term energy sources to sustain the evolution (or even existence) of highly complex life. Moreover, there probably would be no planets, since there would be nothing to bring material particles together, and even if there were planets (say because planet-sized objects always existed in the universe and were held together by cohesion), any beings of significant size could not move around without floating off the planet with no way of returning.”[31]

All of this is completely irrelevant to an omnipotent God. God would have no need to create gravity, because He Himself could hold the matter in stars together. He Himself could bring the material particles together to form planets. He Himself could prevent us from floating off of planets. Collins’ argument for the necessity of gravity is only correct if God cannot, or will not, perform miracles.

Furthermore, many of the examples of fine tuning that Collin’s says are necessary for EMA seem to come dangerously close to endorsing a physicalist theory of mind. Collins writes that “a necessary requirement for the evolution of embodied moral agents is that there exist material systems that can sustain a high level of self-reproducing complexity—something comparable to that of a human brain,”[32] and “only organisms with brains of a size comparable to our own have significant moral agency.”[33] If theism is true, then persons are essentially immaterial minds, and therefore consciousness and agency is emphatically not dependent on brains. Presumably Collins thinks God has significant moral agency, and His mind is certainly not dependent on a complex brain. So if Collins thinks that premise 2 is true because God would need brains for EMA, then this is not a good reason.[34]

In sum, the second premise of Collins’ argument only works if we make the theological assumption that God would only choose to accomplish his goals through the use of natural laws, rather than direct intervention. But this is a massively controversial assumption, one which ignores several centuries’ worth of in-house theological debate among theists.

The Moral Argument

In Mark Linville’s chapter on the moral argument, he presents two separate arguments: one against the truth of evolutionary naturalism based on the existence of moral facts/knowledge, and one for the truth of theism based on human dignity. Like the other chapters, Linville’s defense of the moral argument is uneven. The bulk of the chapter is devoted to refuting the ability of secular moral theories to explain moral facts, but little attention is paid to whether or not there really are moral facts in the first place.

On what grounds do moral realists defend the existence of objective moral facts? As J.L. Mackie noted, they are typically defended by an appeal to intuition:

When we ask the awkward question, how can we be aware of this authoritative prescriptivity, of the truth of these distinctively ethical premises or of the cogency of this distinctively ethical pattern of reasoning, none of our ordinary accounts of sensory perception or introspection or the framing and confirming of explanatory hypotheses or inference or logical construction or conceptual analysis, or any combination of these, will provide a satisfactory answer; ‘a special sort of intuition’ is a lame answer, but it is the one to which the clear-headed objectivist is compelled to resort.[35]

Are we committing intellectual suicide by affirming moral realism simply because it “feels” true? Linville says no. He takes a page from Reformed Epistemologists, such as Alvin Plantinga, and points out that many of our beliefs cannot be justified by any evidence or argument, such as our beliefs in other minds, the external world, and induction.[36] Nonetheless, we are justified in believing in these things in a noninferential way, because our belief systems need to start somewhere. To avoid an infinite regress of justification, we need a set of beliefs to serve as the foundation for all our other beliefs. We call such beliefs “properly basic beliefs.”[37] Linville argues that belief in moral facts is one such properly basic belief.[38]

Is Linville justified in declaring moral beliefs as properly basic? To answer this, we need to ask under what circumstances a belief qualifies as properly basic. There are two general approaches here: the particularist approach and the methodist approach. Under the particularist approach, we identify a few paradigm cases that are obviously properly basic, and then use these examples to develop a criterion for identifying other basic beliefs. In contrast, the methodist approach says that we cannot identify the obvious paradigm cases unless we already have a criterion of proper basicality to identify them with.

I would argue that under either approach, belief in moral realism does not qualify as properly basic. First, using the particularist approach, can we really say that moral realism is “obviously properly basic”? One survey of philosophers found that only 56.4% accept moral realism, which does not bode well for the idea that moral realism is obviously true.[39] True, philosophers are a notoriously skeptical bunch, but moral antirealism is fairly widespread, even among the general population. According to one college professor, “professors with whom I have spoken suggest that the overwhelming majority of college freshmen in their classrooms view moral claims as mere opinions that are not true or are true only relative to a culture.”[40] A survey conducted by the Barna Group found that 64% of adults affirm moral relativism, and the number is even higher—83%—for teenagers.[41] Of course, this does nothing to show that moral realism is false, but it does count against the idea that moral realism is as an “obvious” paradigm case of proper basicality.

A theist may object to this survey data by saying that although people claim to be moral antirealists, they don’t believe this deep down, because this is not how they live their lives. William Lane Craig, for example, writes: “Every day that you get up you answer the question of whether there are objective moral values and duties by how you live. It’s unavoidable.”[42] However, the mere fact that people behave as though there are objective moral facts does not mean that, despite their claims, they really do believe in moral realism “deep down.”[43] After all, I live my life as though $100 bills are more valuable than $1 bills. Life would be unlivable if we did not behave this way. Without an agreed upon system of currency and economics, society would crumble. Nonetheless, I do not actually believe that $100 bills are imbued with some kind of metaphysical quality that $1 bills lack. I know that all paper money is equally worthless, and we only ascribe higher value to higher bills as a matter of social convention. The mere fact that I do not live as though this is true does not mean that I secretly believe that $100 bills really do have more intrinsic value than $1 bills.
Plantinga seeks to avoid the problem of controversy regarding which beliefs constitute “obvious” paradigm cases by relativizing proper basicality to particular communities. He writes:

The Christian will of course suppose that belief in God is entirely proper and rational; if he doesn’t accept this belief on the basis of other propositions, he will conclude that it is basic for him and quite properly so. Followers of Bertrand Russell and Madelyn Murray O’Hare may disagree, but how is that relevant? Must my criteria, or those of the Christian community, conform to their examples? Surely not. The Christian community is responsible to its set of examples, not to theirs.[44]

Thus, maybe Linville would argue that although it is not obvious to everyone that moral realism is properly basic, some communities of people with think it is obvious. Within those communities, members are entitled to regard moral realism as properly basic. However, this relativizes proper basicality in such a way that all sorts of wild and contradictory beliefs can all be deemed rational despite a total lack of evidence. Under this approach, virtually anything can be deemed properly basic. Further, once we go down this road, proper basicality has little to do with whether a belief is true, but only whether holding a belief can be rational. At most, this argument shows that if a person belongs to a community that believes moral realism is properly basic, then her belief can be rational. However, outside of this community, where the status of moral realism is a matter of controversy and therefore not a paradigm case of proper basicality, the moral argument cannot get off the ground. And given that there is likely a considerable overlap between theistic communities and communities that regard moral realism as an obviously properly basic belief, the moral argument doesn’t have much apologetic value in converting atheists.

Can the methodist approach do a better job of securing the proper basicality of moral realism? The Christian philosopher James Sennett has suggested a criterion for proper basicality that he calls “universal sanction.”[45] A belief has universal sanction if it is believed by virtually everyone under normal circumstances, we routinely hold beliefs of such a kind, and the wholesale denial of these types of beliefs would make life practically unlivable.[46] This is an attractive criterion for proper basicality because it accommodates all the classic examples of properly basic beliefs. Take belief in the external world. Clearly, nearly everyone believes in the existence of that things outside themselves, and to deny this would make life impossible. How could we motivate ourselves to go to work if we thought our job didn’t really exist? How could we interact with our friends and family if we thought they were illusions? Our beliefs in other minds, the reality of the past, memories, and induction also enjoy universal sanction.

Belief in objective moral facts, on the other hand, does not have universal sanction. As noted above, there is widespread controversy regarding the truth of moral realism. Further, contra Craig, the denial of moral facts does not make life unlivable. We must ask, do the intuitions and beliefs that we have about morality—the beliefs that guide our everyday behavior—imply that we believe in objective, mind-independent moral facts? As Stephen Finlay has demonstrated, the way that we commonly use moral terms in ordinary moral discourse does not actually presuppose any kind of moral absolutism, and is in fact perfectly compatible with an antirealist view of morality in which moral claims are always conditional on certain goals or desires.[47]

When someone asks how I feel about baby torture, I have the following types of reactions:

  • I find it to be personally upsetting and disgusting.
  • I would very much hate to live in a world where such a thing wasn’t discouraged and prohibited.
  • I wouldn’t want this to happen to my child.
  • I’m glad it didn’t happen to me as a child.
  • I would feel bad for the baby who suffered and for the parents who lost their child.
  • Allowing or encouraging baby torture would result in a tremendous increase in sadness, anger, revenge, violence, and fear, which is something that is not conducive to human well-being.

None of these reactions commit me to the view that baby torture is bad in an absolute, objective, transcendent, and mind-independent way. They merely commit me to the view that torturing babies is bad relative to the wants and needs of humanity. This belief is enough to capture what my intuitions about morality tell me. It is therefore false to say that, in my moral experience, I apprehend a realm of moral facts. This would go well beyond my intuitions, and is instead a metaphysical extrapolation from those intuitions.

Christian philosopher Paul Copan writes that if we reject the existence of objective moral facts, then we “reject something fundamental about our humanness.”[48] I agree that we reject something about our humanness if we condone or advocate things like rape and torture. This is because people who do such things have divorced themselves from the well-being of their fellow humans. They have shown a disregard for the wants and needs of humanity, and therefor they have literally isolated themselves from other people and rejected something about their humanness. But believing that morality is relative to the wants and needs of humans is not the same thing as divorcing yourself from the well-being of humanity. Rather, it is precisely because you are conditioning morality on the wants and needs of humans that you are embracing your humanness. This does not require that we go so far as to embrace full-blown moral realism. We merely need to recognize that certain things are objectively better at leading to the states of affairs that humans value, even if the states of affairs that humans value do not have intrinsic value on their own.

If I deny the existence of moral facts, my ability to navigate the world does not evaporate in the same way that it does when I deny the existence of the external world or that of my memories. I am still able to hold my moral intuitions, trust my senses, trust my memories, make plans and carry them out, etc. Some people may find moral antirealism depressing, but this is hardly the same as making ordinary life unthinkable or impossible. Thus, belief in moral realism is not properly basic under the criterion of universal sanction.

Linville endorses Plantinga’s criteria for warranted belief, which says that “a belief is warranted just in case it is the product of a belief-producing mechanism that is truth-aimed and functioning properly in the environment for which it was designed.”[49] Linville says that if God exists, then our moral beliefs meet these criteria because then “human moral faculties are designed [by God] to guide human conduct in light of moral truth.”[50] On the other hand, if naturalism is true, then Linville argues that we have no reason to think that our moral beliefs are the result of a truth-aimed mechanism, and therefore naturalism “presents an undercutting defeater for our moral beliefs taken as a whole.”[51] If Linville is correct, then if God exists, then belief in moral realism is warranted. Conversely, if naturalism is true, then belief in moral realism is not warranted.

This conclusion is similar to Plantinga’s conclusion that basic belief in God can be warranted, if there is a God who endowed us with an immediate awareness of His existence. Because this conclusion is conditional, one cannot know whether their noninferential belief in God is warranted unless they also know that God exists. As philosopher Herman Philipse writes, people who are unsure whether they should believe in God “cannot reassure themselves that their beliefs may be warranted even if they have no arguments to support them. For their beliefs may be warranted in the basic way only if God exists, and that was precisely what they were calling into question.”[52] Anthony Kenny makes the similar point: “The doubting believer in God cannot reassure himself that his belief is warranted; for only if there is a God is his belief warranted, and that is what he was beginning to doubt.”[53] Likewise, Keith Parsons explains, “[Plantinga] argues that theistic belief is very likely warranted and properly basic, in the externalist sense, but only if theism is in fact true. This means that believers are in no position to argue that their belief in God is warrant basic unless they can adduce reasons, arguments, or evidence for the existence of God.”[54]

Like Plantinga’s argument for the warrant of basic belief in God, Linville’s argument for the warrant of basic belief in moral realism is conditional on the truth of theism. Thus, under Linville’s view, moral realists cannot reassure themselves that their beliefs are warrant basic unless they believe in God. But this is precisely the conclusion that Linville’s moral argument seeks to prove. Thus, because a premise in Linville’s argument can only be rationally affirmed by people who already accept the conclusion, his moral argument can have no persuasive force. It can only reinforce the belief of people who are already theists.

The Argument from Miracles

In their chapter on the resurrection of Jesus, Timothy and Lydia McGrew present what is certainly the most rigorous probabilistic case for Jesus’ resurrection to date. Unlike previous Bayesian resurrection arguments, such as that of Richard Swinburne, the McGrews’ argument is far less ambitious. They make it clear that they are not arguing that the resurrection probably happened (though they certainly believe that it did). They are merely arguing that certain pieces of evidence make the resurrection more probable than it otherwise would have been.[55] In this sense, their resurrection argument has the same logical structure as Collins’ fine-tuning argument. Both arguments appeal to the likelihood principle to argue that some evidence (E) raises the probability of some hypothesis (H).

In the McGrews’ chapter, their “H” is “R”—the hypothesis that Jesus rose bodily from the dead. Their “E” is comprised of three separate pieces of evidence:

W: the discovery of the empty tomb by Jesus’ women followers.
D: the appearances to the disciples and their willingness to die for their beliefs.
P: the conversion of Paul.

They argue that in light of E, the probability of R is significantly increased. One strong point of the McGrews’ chapter is that it considers the evidential force of each piece of evidence on its own, rather than lumping them all together and roughly estimating their combined force. Swinburne, for example, simply lumps all of the facts in his “E” together and estimates that E is 100 times more likely on R than ~R.[56] Thus:


this ratio of 100/1, known as the Bayes’ factor, tells us that E makes the R 100 times more probable relative to ~R than it otherwise would have been. Unlike Swinburne, the McGrews are careful to separate out the different facts that comprise E into W, D, and P, and then they determine the separate Bayes’ factors for each piece of evidence. For the discovery of the empty tomb, they estimate that it is 100 times more likely on R than ~R.[57] That is:


For the appearances to the disciples (D), the McGrews divide this evidence even further into 13 separate appearances, one to represent the experience of each individual disciple. They estimate that the experience of each individual disciple was 1,000 times more likely on R than ~R, and the conjunction of all 13 disciples having these experiences is thus extremely more likely on R than ~R.[58] Thus:

P(D1/R)/P(D1/~R) × P(D2/R)/P(D2/~R) × … × P(D13/R)/P(D/13/~R)=1039/1

For the conversion of Paul, the McGrews estimate that this was 1,000 times more likely on R than ~R[59], giving us:


The McGrews then go on to multiply these Bayes factors together, and conclude that the conjunction of W, D, and P makes R 1044 times more probable relative to ~R than it otherwise would have been.[60] Formally, their argument can be represented this way:

  1. P(W/R) / P(W/~R) = 100/1
  2. P(D/R) / P(D/~R) = 1039/1
  3. P(P/R) / P(P/~R) = 1,000/1
  4. P(W&D&P/R) / P(W&D&P/~R) = P(W/R) / P(W/~R) × P(D/R) / P(D/~R) × P(P/R) / P(P/~R)
  5. P(W&D&P/R) / P(W&D&P/~R) = 1044/1

How might a skeptic challenge this argument? Most of the argument’s heavy lifting is done by the large Bayes’ factor assigned to D, so we must ask, what exactly is D? What sort of appearances of Jesus did the disciples see? Most Christian apologists today take a “minimal facts” approach, in which they make as few assumptions as possible about the reliability of the Gospel accounts.[61] They don’t assume that the disciples claimed to eat fish with Jesus, or to have interacted with him for 40 days, or to have had integrated polymodal interactions with Jesus. Minimal facts apologists merely say that, from a historical perspective, we can be sure that the disciples had experiences in which they believed they saw Jesus, though the details of these experiences are unclear. The McGrews distinguish themselves and take the entirety of the Gospels at face value. They write:

Our argument will proceed on the assumption that we have a substantially accurate text of the four gospels, Acts, and several of the undisputed Pauline epistles (most significantly Galatians and I Corinthians); that the gospels were written, if not by the authors whose names they now bear, at least by disciples of Jesus or people who knew those disciples—people who knew at first hand the details of his life and teaching or people who spoke with those eyewitnesses—and that the narratives, at least where not explicitly asserting the occurrence of a miracle, deserve as much credence as similarly attested documents would be accorded if they reported strictly secular matters. Where the texts do assert something miraculous—for example, Jesus’ post-resurrection appearances—we take it, given the basic assumption of authenticity, that the narrative represents what someone relatively close to the situation claimed.[62]

If a skeptic is willing to grant this much, then much of the McGrews’ chapter is unnecessary. If the disciples really did have extended, detailed, integrated, polymodal sightings of Jesus in which they all ate and lived together, then of course these appearances cannot be explained by something like a hallucination. But as far as I know, no advocate of the hallucination theory has ever seriously suggested that the appearances as described in the Gospels are explained by hallucinations. Those who advocate a hallucination hypothesis are using hallucinations to explain the “minimal facts” version of the appearances, not the “maximal facts” version the McGrews use.

If the skeptic is willing to grant these “maximal facts,” then the game is already over. There is no need to go through this exercise in Bayesian reasoning, because the case for the resurrection is already made at this point. Assigning Bayes factors to these “maximal facts” is just decorative overkill. To their credit, the McGrews do attempt to justify their textual assumptions, but as is the case with the rest of this anthology, this premise, which is by far the argument’s most important, receives the least amount of attention. The bulk of the chapter proceeds from this assumptions, and so nearly all of the chapter will be unconvincing to everyone who doesn’t already accept the New Testament accounts.

Putting this issue aside, there a few other ways for skeptics to challenge the McGrews’ argument. First, they can challenge the actual numbers. For example, the McGrews argue that P(P/R) is 1,000 times greater than P(P/~R). They do offer an argument for why P(P/~R) is low, but offer no argument as to why P(P/R) should be so much greater. Given that they define the hypothesis R as “the bodily resurrection of Jesus of Nazareth circa A.D. 33,”[63] there is nothing in R that predicts P would occur. In order for R to predict P, we need to add the auxiliary hypothesis (A) that Jesus chose to appear to Paul. Thus, P(P/R)=P(A/R)xP(P/A&R) + P(~A/R)xP(P/~A&R).

In order for P(P/R) to be high, the prior probability of the added auxiliary hypothesis—P(A/R)—needs to be much higher than P(~A/R). But we have every reason to think P(A/R) is low. What reason would Jesus have to appear to Paul? Of all the persecutor’s of the church, why him? Why not some other member of the Jewish community in Jerusalem? If Jesus’ goal was to stop the persecution and help spread Christianity, then why would he only appear to Paul? Why not appear to the entire Sanhedrin? Why didn’t he also appear to more influential Christian persecutors, like Mao or Stalin? The fact is, if someone knew that Jesus rose from the dead, but had no knowledge of Paul’s conversion, that person simply wouldn’t be able to infer the latter from the former. Paul’s conversion was probably just as much of a surprise to the disciples as it was to the Jews. In the absence of any particular reason to think that Jesus would single out Paul from of all the church persecutors throughout history, we should regard the probability of such a specific event to be quite low.

Another strategy would be to challenge the McGrews’ assumption that W, D, and P are independent of each other. For example, the probability of someone coming to believe that Jesus rose from the dead, given that he knows that the tomb is empty, is higher than the probability of developing this belief if he does not know that the tomb is empty. As Paul Draper has noted, surely the number of people who believed Elvis was alive would increase if it was discovered that his grave was empty.[64] Thus, premise 4 in the McGrew’s argument is false. The formula that they use is:

P(W&D&P/R) / P(W&D&P/~R)=P(W/R) / P(W/~R) × P(D/R) / P(D/~R) × P(P/R) / P(P/~R)

But instead of using that formula, they should have used this one:

P(W&D&P/R) / P(W&D&P/~R) = P(W/R) / P(W/~R) × P(D/R&W) / P(D/~R&W) × P(P/R&W&D) / P(P/~R&W&D).

Not only do the McGrews treat W, D, and P as independent, they treat the experiences of all the disciples as independent. However, it is well known among psychologists that people can influence each other’s’ actions and beliefs. They can create expectation effects that induce hallucinations in others, they can contaminate each other’s’ memories or even implant false ones, they can reinforce each other’s’ biases and help each other rationalize away cognitive dissonance, they can pressure each other into adopting beliefs, they can influence each other through the power of suggestion, they can fall victim to groupthink and mass hysteria, etc. The McGrews calculate the evidential force of the experiences of the thirteen disciples as follows:

P(D1-D13/R) / P(D1-D13/~R) = P(D1/R)/P(D1/~R) × … × P(D13/R)×P(D13/~R)

But instead, they should have calculated it this way:

P(D1-D13/R&W) / P(D1-D13/~R&W) = P(D1/R&W) / P(D1/~R&W) × P(D2/R&W&D1) / P(D2/~R&W&D1) × … × P(D13/R&W&D1&…&D12) / P(D13/~R&W&D1&…&D12)

The McGrews are aware that their argument assumes independence between the disciples’ experiences. However, they argue for this assumption by saying that, if anything, this assumption hurts, rather than helps, their argument. Thus, they argue that they are actually being charitable to the skeptic by assuming independence. The reason, they say, is that if one disciple knows that another disciple died for publically proclaiming the resurrection, then this would dissuade him from doing the same. Thus, the disciples’ martyrdoms are negatively relevant to each other—the occurrence of one lowers the probability of the occurrence of another.

To see what is wrong with this reasoning, imagine the following scenario: you are the pit boss in a casino watching over a line of 13 slot machines. These machines are all rigged so that one jackpot makes subsequent jackpots less likely. Thus, while the probability of winning on any given machine starts off at 1 in a million, it lowers to 1 in 2 million if someone hits a jackpot. And if a second person hits a jackpot, then the probability lowers to 1 in 3 million, and so on. Now imagine that someone hits the jackpot on machine #1. Because you know the machines are supposed to follow this rigging system, you should think that the probability of second jackpot is now 1 in 2 million. But now imagine that someone hits the jackpot on machine #2 as well. If the machines are functioning properly, then the probability of a third jackpot, given the previous 2 jackpot, is 1 in 3 million. But you are not so sure that the machines are functioning properly. You are beginning to suspect that maybe someone has tampered with the machines and is conning the casino. Now imagine that machines #3 and #4 also hit the jackpot. At this point you have received very strong evidence that the machines have been tampered with. Now machine #5 wins, and then machine #6. Every time an additional machine hits the jackpot, this should increase your confidence that the next machine will hit the jackpot as well. By the time machines #1 though #12 have all won, it is all but certain that the #13 is going to hit the jackpot as well. The point of this analogy is that, although we should start off thinking that the jackpots are negatively relevant to each other, as more machines win, we quickly reach a point where it becomes clear that the jackpots are positively relevant to each other.[65]

The disciples’ willingness to die is analogous to the slot machines hitting the jackpot. Let D1=disciple #1 was willing to die, D2=disciple #2 was willing to die, etc. The McGrews are arguing that P(D2/~R)>P(D2|D1&~R), because if disciple #2 knew that disciple #1 died for his beliefs, this would scare disciple #2 into dropping his claims. We can grant this for the sake of argument. But what if we learn that disciples #1-3 all testified to the resurrection? At this point, a pattern is starting to emerge. We know that three of the disciples were already willing to die for their beliefs, so this gives us some insight into the beliefs and resiliency of the disciples. Thus, upon learning D1&D2&D3, it should come as less of a surprise to learn D4. P(D4|D1&D2&D3~R) is therefore greater than P(D3|D1&D2&~R). Now what about P(D5|D1&D2&D3&D4&~R)? Here, we have even more evidence that we are dealing with a group of people who strongly believe in the resurrection and do not give in to intimidation, so P(D5|D1&D2&D3&D4&~R) is greater than P(D4|D1&D2&D3&~R). Now what about P(D13|D1& … &D12&~R)? Here, we know that all the other disciples have already been willing to die for their beliefs, so this is very strong evidence that disciple #13 will do the same.

Notice the pattern here. Every time we learn that another disciple was willing to die for his faith, this should make us more confident that the next disciple was also willing to die for their faith. Thus, P(D13|D1&…&D12&~R)>P(D12|D1&…&D11&~R)>P(D11|D1&…&D10&~R) > P(D10|D1&…&D9&~R)>P(D9|D1&…&D8&~R)>P(D8|D1&…&D7&~R) > … etc. Perhaps the McGrews are correct that we should initially think that the disciples are negatively relevant to each other. However, as we learn that more and more disciples were willing to die for their beliefs, we very quickly reach a tipping point where they become positively relevant, and the addition of every new disciple who was willing to die his beliefs should make us more confident that the next disciple would do the same. The disciples are just like the slot machines. At first, we think they are negatively relevant to each other, but our expectations should change as we learn that more and more disciples were willing to die. For this reason, the McGrews’ independence assumption causes them to seriously overestimate the evidential force of D1-D13. It doesn’t take long before the addition of more disciples who were willing to die start to provide diminishing returns to the case for R.[66]

Furthermore, by underestimating the effects of group behavior, the McGrews’ overestimate the extent to which even the first few disciples are negatively relevant to each other. The McGrews’ analysis looks at each of the disciples and asks how they individually would behave in response to what happened to their peers. However, it is well known among psychologists that people in groups can think and behave very differently than they would on their own. In group settings, factors involving peer pressure, mutual reinforcement, and diffusion of responsibility can lead groups to take far riskier actions than they would on their own. As the social scientists Clark McCauley and Sophia Moskalenko explain, very strong cohesion develops among small groups facing common threats, and “high cohesion brings high pressures for both behavioral compliance and for internalized value consensus.”[67] Furthermore, when group members are cut off from other social groups, cohesion within the group is at its strongest, and this in turn maximizes the value of that group’s “social reality.” This social reality includes the answers to questions like “What is good and what is evil? What is worth working for, worth dying for? What does it mean that I am going to die?”[68] For this reason, the risky behavior exhibited by the disciples becomes more likely, given ~R, when we take the dependence between the disciples into account.

Additionally, the McGrews seems to underestimate the importance placed on dying a noble death in the ancient world. The McGrews write that “human beings naturally fear death and are horrified by even the account of torture”[69] and therefore the disciples would be unlikely to put themselves in danger after seeing their fellow disciples receive punishment. But as Professor of New Testament and Early Christianity Candida Moss writes:

[M]ost people in the ancient world accepted suicide, and even saw it as noble and courageous … Many people in the ancient world were unfazed by and even admired this kind of behavior. Although later Christians would condemn volunteering oneself for martyrdom precisely because it was a kind of suicide, at the time many Christians, including bishops and priests, actually sought out suffering and death.”[70]

Tertullian even wrote that, in the second century, thousands of Christians lined up at the Roman proconsul’s house and begged to be martyred, only to be turned away.[71] Considering that many people in ancient times did not fear death the way moderns do, and often even sought it out, it is not particularly surprising that the disciples would place themselves in danger, even after seeing other disciples get killed. Indeed, it is easy to see how watching their friends die for their beliefs could embolden and inspire the remaining apostles. After seeing your fellow disciple die for his beliefs, this fortifies your own belief that this must be a true and worthy cause. Seeing someone else bravely go to their grave could embolden and inspire you to do the same if the situation came to it. Further, in a shame culture, you wouldn’t want to betray your followers and back out after one of the other disciples has already been killed. This would make you look like a coward and a fraud. This shame factor plays a large role in preventing Islamic terrorists from backing out of their suicide attacks.[72] In this sense, seeing a fellow disciple die for his beliefs could actually increase the probability that you would be willing to do the same.

Further, the evidence they provide that the disciples really were killed for their beliefs is quite weak. Their “best attested”[73] example is the death of James bar Zebedee, documented in Acts 12, which merely says that Herod “killed James the brother of John with the sword.” This passage says nothing about why Herod killed James, so we can hardly argue that it was because he was unapologetically preaching the strongly detailed story of the resurrection depicted in the Gospels. The McGrews write that the early Christian martyrs were “claiming not simply that Christianity is true but rather concretely, to have seen the risen Jesus.”[74] But how do they know this? The McGrews seem to think that before the disciples were killed, something like the following conversation took place:

“Admit you never saw Jesus!”
“Are you sure you don’t want to change your story? We’ll let you go if you admit you’re wrong.”
“No, I saw him! Go ahead and kill me!”
“You really ate with him?”
“You really touched his wounds?”
“Of course.”
“For forty whole days?”

But we cannot possibly know if this was the case for James. He could have been killed just for being blasphemous, like Stephen. In Acts 7, Stephen is killed because he blasphemes the laws of Moses, not because he preaches the resurrection. Further, we don’t know if James was ever given the chance to recant, or whether it would have many any difference if he did.

Similarly, Josephus’ depiction of James’ execution does not say that he was killed for resolutely standing by his detailed testimony regarding Jesus’ resurrection. It says he was killed simply for being a breaker of the law. Nor does the McGrews’ citation of Hegisippus have James providing any testimony about the risen Jesus.

Additionally, the alleged martyrdom of Peter does not lend much to the McGrews’ argument. Peter was allegedly killed during the reign of Nero, who was scapegoating Christians in order to quash the rumor that he had started the Great Fire of Rome. If Nero’s goal was just to find a scapegoat, then there wouldn’t be any reason to offer Peter a chance to recant his story about seeing Jesus, and it wouldn’t have made much of a difference if Peter did. Nero was just looking for a fall man. Furthermore, some have expressed skepticism that the Neronian persecution even took place. Professor Moss writes: “It’s highly unlikely that, at the time the Great Fire occurred, anyone recognized Jesus followers as a distinct and separate group…. If followers of Jesus weren’t even identified as Christians, it’s highly improbable that Christians were well known and disliked enough that Nero could single them out as scapegoats.”[75]

Lastly, a skeptic could respond to the McGrews’ argument by granting that even if (W&D&P/R) is much greater than P(W&D&P/~R), this does not mean that P(R/W&D&P) is much greater than P(~R|W&D&P). To show this, we would also need to know the prior probability for R—P(R)—which is the probability of R apart from W, D, and P. The skeptic could argue that for various reasons, such as the improbability of theism or historical criticisms of the Old Testament, P(R) is so low that R is still probably false, even after taking W, D, and P into consideration.

The McGrews are quite clear that they are not arguing that the resurrection probably happened, and that their aim is simply to identify some pieces of evidence that raises the probability of R.[76] As long as this is all they are trying to prove, then they are perfectly within their rights to refrain from considering the prior probability of R. However, throughout their chapter the McGrews routinely reject competing naturalistic hypotheses on the grounds that their prior probabilities are so low. It seems a bit unfair to assign priors the competing hypotheses and reject them for being low, while at the same refusing to assign a prior to your own theory.

Are the McGrews guilty of a double standard here? Not really. The reason is that they are not comparing R to a specific competing theory. Rather, they are comparing R to ~R, which is simply the disjunction of every logically possible competing theory. The trouble with evaluating the likelihood of some observation conditional on the negation of a hypothesis—P(E/~H)—is that ~H is made up of infinitely many subhypotheses, all of which confer different probabilities on E. So how do we assess P(E/~H)? We consider the predictive power of each subhypotheses (SH) and weigh them by their priors using the theorem on total probability. Thus:

P(E/~H)=P(SH1/~H) × P(E/SH1&~H) + P(SH2/~H) × P(E/SH2&~H) + … +P(SHn/~H) × P(E/SHn&~H)

In the case of ~R, we have multiple different subhypotheses: theft theory (T), hallucination theory (H), swoon theory (S), etc. Thus:

P(W&D&P/~R)=P(T/~R) × P(W&D&P/T&~R) + P(H/~R) × P(W&D&P/H&~R) + P(S/~R) × P(W&D&P/S&~R) + … etc.

Notice that in order to calculate P(W&D&P/~R), we need to consider the prior probabilities of all the subhypotheses under consideration. In contrast, calculating P(W&D&P/R) does not require us to consider the prior probability of R at all. Thus, the McGrews aren’t doing anything wrong by considering the priors of competing theories, but not the prior of their own theory. However, while there is nothing logically invalid about this approach, one can still argue that, by comparing R to ~R, rather than specific subhypotheses, the McGrews have formulated their argument in such a way that the resurrection explanation is subjected to less scrutiny than competing explanations.

To demonstrate this point, think up the most ad hoc and implausible explanation for W, D, and P that you can. Perhaps some kind of hybrid Leprechaun-Alien-Voodoo explanation (LAV). Now pack this explanation with as many contrived auxiliary assumptions that it takes for LAV to confer a probability of 100% on W&D&P. Thus, P(W&D&P/LAV)=1.0. Let’s be generous and assume that P(W&D&P/R) is also 1.0. When we compare LAV and R in terms of predictive power, they are the same with respect to W, D, and P, so these facts do not confirm either hypothesis. But if we compare LAV to ~LAV, then we must treat R as a subhypothesis within ~LAV, and therefore R must be weighted by its prior probability. This gives us the following formula:

P(W&D&P/~LAV)=P(R/~LAV) × P(W&D&P/R&~LAV) + … + P(subhypothesisn/~LAV) × P(W&D&P/subhypothesisn&~LAV)

Earlier, when we were comparing the predictive powers of LAV and R, we considered the prior probability of neither. But now that we are comparing the predictive powers of LAV and ~LAV, the predictive power of R is weighed down by its prior probability, while LAV gets a free pass. Even the McGrews agree that the prior probability of R is low, and therefore P(W&D&P/LAV) is much lower than P(W&D&P/~LAV). However, because LAV has been given a free pass while the competing hypotheses were forced to be weighed down by their priors, this conclusion is not particularly significant. The only way to remedy this asymmetry in scrutiny is to also consider the prior probability of LAV, and when we do, the advantage LAV has over ~LAV disappears. I would argue that the same is true for the resurrection. Even if R has an advantage over ~R in terms of predictive power, this is only because R does not have to reckon with its low prior probability. Once R is subjected to the same scrutiny as its competitors, I would argue that the advantage R has over ~R would vanish. To reiterate, there is nothing logically invalid about comparing the predictive powers of a hypothesis and its negation, so the McGrews are not incorrect or misleading by doing this. Nonetheless, the conclusion of these types of arguments are just very unambitious in what they seek to prove, and therefore their conclusions are very weak.

A better approach would have been to compare R directly to ~R’s subhypotheses. This way, if the McGrews wanted to critique a subhypothesis on the basis of its prior probability, they could only do so if they considered the prior probability of their own hypothesis as well. If the McGrews want to hide in the comfort of likelihoods and avoid discussing priors, then the naturalist can do the same. For example, if we compare P(W&D&P/R) to a sufficiently detailed naturalistic explanation (N), then P(W&D&P/R)=P(W&D&P/N). The only way, then, to argue that N is a worse explanation than R would be to argue that it has a lower prior. But the naturalist can simply respond, “Hey, I thought we were only comparing likelihoods here? All of a sudden you want to talk about priors?”


Despite its flaws, the contributors to The Blackwell Companion to Natural Theology are able to score some major points throughout the book. For example, Robin Collins manages to correct some confusions among people who deny the fine-tuning of the universe, Alexander Pruss goes well beyond an appeal to intuition and makes a rigorous attempt to defend the principle of sufficient reason, and Timothy and Lydia McGrew present a takedown of Alvin Plantinga’s “Problem of Dwindling Probabilities” that is so decisive that Plantinga himself has been convinced. Nonetheless, the volume still suffers from the primary defect of focusing too heavily on some premises, while tending to ignore the most theologically important ones.

A final word is in order about the very nature of The Blackwell Companion to Natural Theology. As I previously noted, this is a highly technical volume. Even readers with PhD’s could have trouble keeping up with some of the material. Robert Maydole’s “Temporal-Contingency Argument” is 87 steps long and is presented entirely in symbolic logic; Craig and Sinclair’s chapter is filled with diagrams depicting complex cosmological models; the chapters by Collins and the McGrews contain long strings of probabilistic notation; Alexander Pruss’ chapter is rife with modal and symbolic logic; etc. If God really does exist and wants us to know Him, then we wouldn’t expect arguing for His existence to take so much work. What reason would God have for only making the evidence for His existence accessible to professional academics? If God really did exist, it is inconceivable that He would make his existence so hard to prove that it could not be done until the 21st century, and only by a select handful of people who have reached a level of educational achievement that surpasses nearly everyone who ever lived. And it certainly seems inconceivable that God would punish people with eternal torment because they misunderstood the implications of the most recent cosmological speculations, or because they failed to properly apply Bayesian reasoning to little known historical data. The very fact that this book needs to dig so deep in order to make its case is, in a way, evidence against the existence of God.


[1] William Lane Craig & Kevin Harris, “Blackwell Companion Book” (July 27, 2008). Reasonable Faith website. <https://www.reasonablefaith.org/blackwell-companion-book>.

[2] William Lane Craig & James Sinclair, “The Kalam Cosmological Argument” in The Blackwell Companion to Natural Theology ed. William Lane Craig and J. P Moreland (Malden, MA: Blackwell, 2009): 101-201, p. 102, 194.

[3] Craig & Sinclair, “The Kalam Cosmological Argument,” p. 196.

[4] Craig & Sinclair, “The Kalam Cosmological Argument,” p. 192.

[5] Craig & Sinclair, “The Kalam Cosmological Argument,” p. 196.

[6] William Lane Craig, “God and the Initial Cosmological Singularity: A Reply to Quentin Smith.” Faith and Philosophy Vol. 9, No. 2 (April 1992): 238-248. See also William Lane Craig, “Theism and Big Bang Cosmology.” Australian Journal of Philosophy Vol. 69, No. 4 (1991): 492-503, pp. 496-499.

[7] Craig & Sinclair, “The Kalam Cosmological Argument,” pp. 187-189.

[8] Wes Morriston, “Must the Beginning of the Universe Have a Personal Cause? A Critical Examination of the Kalam Cosmological Argument.” Faith and Philosophy Vol. 17, No. 2 (April 2000): 149-169, p. 162.

[9] Wes Morriston, “Creation ex Nihilo and the Big Bang.” Philo Vol. 5, No. 1 (Spring-Summer 2002): 23-33, p. 29.

[10] Craig & Sinclair, “The Kalam Cosmological Argument,” p. 188.

[11] Mark Balaguer, “Platonism in Metaphysics” in The Stanford Encyclopedia of Philosophy (Fall 2009 edn.) ed. E. N. Zalta (Stanford, CA: Stanford University, 2009). <https://plato.stanford.edu/archives/fall2009/entries/platonism/>.

[12] Craig & Sinclair, “The Kalam Cosmological Argument,” pp. 106-107.

[13] Craig & Sinclair, “The Kalam Cosmological Argument,” pp. 106-107.

[14] William Lane Craig, “J. Howard Sobel on the Kalam Cosmological Argument.” Canadian Journal of Philosophy Vol. 36, No. 4 (December 2006): 565-584, p. 570.

[15] Craig & Sinclair, “The Kalam Cosmological Argument,” p. 188.

[16] Craig & Sinclair, “The Kalam Cosmological Argument,” p. 108.

[17] Robin Collins, “The Teleological Argument: An Exploration of the Fine-Tuning of the Universe” in The Blackwell Companion to Natural Theology ed. William Lane Craig and J. P. Moreland (Malden, MA: Blackwell, 2009): 202-281, p. 207.

[18] Collins, “The Teleological Argument,” pp. 254-256.

[19] Collins, “The Teleological Argument,” pp. 203-205, 254.

[20] Collins, “The Teleological Argument,” p. 254.

[21] Collins, “The Teleological Argument,” p. 254.

[22] Collins has confirmed my suspicions via personal communication.

[23] Or, as Paul Draper calls it, the “Weighted Average Principle.” See Paul Draper, “Pain and Pleasure: An Evidential Problem for Theists.” Noûs Vol. 23, No. 3 (June 1989): 331-350, p. 340.

[24] Collins, personal communication.

[25] Collins, personal communication.

[26] Sean Carroll & William Lane Craig, “The Existence of God in Light of Contemporary Cosmology” (September 14, 2015). Reasonable Faith website. <https://www.reasonablefaith.org/media/debates/god-and-cosmology-the-existence-of-god-in-light-of-contemporary-cosmol/>

[27] William Lane Craig, “God and the Initial Cosmological Singularity.”

[28] Collins, “The Teleological Argument,” p. 234.

[29] Collins, “The Teleological Argument,” p. 220.

[30] Chad Meister, Philosophy of Religion (New York, NY: Palgrave Macmillan, 2014), p. 57.

[31] Meister, Philosophy of Religion, p. 212.

[32] Meister, Philosophy of Religion, p. 211.

[33] Meister, Philosophy of Religion, p. 214.

[34] Similar arguments are made in Neil Sinhababu: “Divine Fine-Tuning vs. Electrons in Love.” American Philosophical Quarterly, Vol. 54, No. 1 (2017); and M. C. Bradley, “The Fine-Tuning Argument: The Bayesian Version.” Religious Studies, Vol. 38, No. 4 (December 2002): 375-404.

[35] J. L. Mackie, Ethics: Inventing Right and Wrong (New York, NY: Penguin Books, 1977), pp. 38-39.

[36] Mark Linville, “The Moral Argument” in The Blackwell Companion to Natural Theology ed. William Lane Craig and J. P. Moreland (Malden, MA: Blackwell, 2009): 391-448, p. 415.

[37] Linville, “The Moral Argument,” p. 416.

[38] Linville, “The Moral Argument,” p. 416.

[39] David Bourget & David Chalmers, “PhilPapers Survey: Results” (November 2009). <http://philpapers.org/surveys/results.pl>.

[40] Justin P. McBrayer, “Why Our Children Don’t Think There are Moral Facts.” The New York Times (March 2, 2015). <https://opinionator.blogs.nytimes.com/2015/03/02/why-our-children-dont-think-there-are-moral-facts/>.

[41] Barna Group, “Americans are Most Likely to Base Truth on Feelings” (February 12, 2002). Barna Research Releases in Culture & Media. <https://www.barna.com/research/americans-are-most-likely-to-base-truth-on-feelings/>.

[42] William Lane Craig, Reasonable Faith: Christian Truth and Apologetics (3rd ed.) (Wheaton, IL: Crossway Books, 2008) p. 194.

[43] William Lane Craig & Walter Sinnott-Armstrong, God? A Debate between a Christian and an Atheist (New York, NY: Oxford University, 2004), p. 18.

[44] Alvin Plantinga, “Is Belief in God Properly Basic?” Noûs Vol. 15, No. 1 (March 1981): 41-51, p. 50.

[45] James F. Sennett, “Direct Justification and Universal Sanction.” Journal of Philosophical Research Vol. 23 (1998): 257-287.

[46] Sennett, “Direct Justification and Universal Sanction.”

[47] Stephen Finlay, “The Error in Error Theory.” Australasian Journal of Philosophy Vol. 86, No. 3 (2008): 347-369.

[48] Paul Copan, “The Moral Argument” in The Rationality of Theism ed. Paul Copan and Paul Moser (New York, NY: Routledge, 2003): 149-174, p. 150.

[49] Linville, “The Moral Argument,” p. 416.

[50] Linville, “The Moral Argument,” p. 414.

[51] Linville, “The Moral Argument,” p. 392.

[52] Herman Philipse, God in the Age of Science? A Critique of Religious Reason (Oxford, UK: Oxford University Press, 2012), p. 41.

[53] Anthony Kenny, quoted in Philipse, God in the Age of Science? p. 41.

[54] Keith Parsons, “Some Contemporary Theistic Arguments” in The Cambridge Companion to Atheism ed. Michael Martin (Cambridge, UK: Cambridge University Press, 2007): 102-117, p. 110.

[55] Timothy & Lydia McGrew, “The Argument From Miracles: A Cumulative Case for the Resurrection of Jesus of Nazareth” in The Blackwell Companion to Natural Theology ed. William Lane Craig and J. P. Moreland (Malden, MA: Blackwell, 2009): 593-662, p. 595.

[56] Richard Swinburne, The Resurrection of God Incarnate (Oxford, UK: Oxford University Press, 2003), pp. 212-214.

[57] McGrew & McGrew, “The Argument From Miracles,” p. 622.

[58] McGrew & McGrew, “The Argument From Miracles,” p. 628.

[59] McGrew & McGrew, “The Argument From Miracles,” p. 630.

[60] McGrew & McGrew, “The Argument From Miracles,” p. 630.

[61] See, for example: Michael R. Licona, The Resurrection of Jesus: A New Historiographical Approach (Downers Grove, IL: InterVarsity Press, 2010); and William Lane Craig, Reasonable Faith: Christian Truth and Apologetics (3rd ed.) (Wheaton, IL: Crossway Books, 2008).

[62] McGrew & McGrew, “The Argument From Miracles,” p. 597.

[63] McGrew & McGrew, “The Argument From Miracles,” p. 594.

[64] Paul Draper, “Craig’s Case for God’s Existence” in Does God Exist? The Craig-Flew Debate ed. Stan W. Wallace (Burlington, VT: Ashgate, 2003): 141-154, p. 150.

[65] I owe this analogy to Jeffrey Amos Heavener (personal communication).

[66] To demonstrate this point more technically, imagine the following possibilities:

R: The resurrection occurred.
A: The resurrection didn’t occur, and the disciples would be afraid to testify that it occurred if previous disciples suffered negative consequences, so their testimonies are negatively correlated.
B: The resurrection didn’t occur, but due to one or more of the previously mentioned psychological mechanisms, the disciples came to believe that it occurred in such a way that their testimonies are positively correlated.

The McGrews argue that A takes up most of ~R’s probability space, and we may grant for the sake of argument that this is a reasonable assumption at the outset. However, the distribution of A and B within ~R’s probability space is not fixed. Rather, each piece of testimony will shift some confidence from A to B. For the first witness, P(D1|~R) is low, because while B predicts D1, B only takes up a small portion of ~R’s probability space. The majority of ~R’s probability space is occupied by A, which does not predict D1 at all. Thus, P(D1|R)>P(D1|~R), so D1 raises the probability of R. However, D1 also raises the probability of B, because while B predicts D1, ~B does not. This is because most of ~B’s probability space is occupied by A, which doesn’t predict D1. Thus, D1 raises the probability of R and B, and lowers the probability of A. When we learn of the second disciple’s testimony, this is less surprising under ~R than the first disciple’s testimony was; for after learning D1, B has taken over a larger portion of ~R’s probability space. And with each new witness, B takes over an even greater portion of ~R’s probability space relative to A, and therefore ~R predicts each new testimony more strongly than the last. Therefore, the disciples begin to provide diminishing returns as the ratio P(Di|R)/P(Di|~R) shrinks with each additional disciple.

A similar argument is made in Arif Ahmed, “Hume and the Independent Witnesses.” Mind Vol. 124, Issue 496 (October 2015): 1013-1044.

[67] Clark McCauley & Sophia Moskalenko, “Mechanisms of Political Radicalization: Pathways Toward Terrorism.” Terrorism and Political Violence, Vol. 20 (2008): 415-433, p. 423.

[68] McCauley & Moskalenko, “Mechanisms of Political Radicalization,” p. 423.

[69] McGrew & McGrew, “The Argument From Miracles,” p. 634.

[70] Candida Moss, The Myth of Persecution: How Early Christians Invented a Story of Martyrdom (New York, NY: HarperCollins, 2013), pp. 194-195.

[71] Moss, The Myth of Persecution, p. 192.

[72] Jerrold M. Post, “Hamas: The Islamic Resistance Movement” in The World’s Most Threatening Terrorist Networks and Criminal Gangs ed. Barry R. Schneider, Jerrold M. Post, and Michael T. Kindt (New York, NY: Palgrave Macmillan, 2009): 145-164, p. 153.

[73] McGrew & McGrew, “The Argument From Miracles,” p. 614.

[74] McGrew & McGrew, “The Argument From Miracles,” p. 635.

[75] Moss, The Myth of Persecution, p. 139.

[76] McGrew & McGrew, “The Argument From Miracles,” p. 595.

Copyright ©2018 Aron Lucas. The electronic version is copyright ©2018 by Internet Infidels, Inc. with the written permission of Aron Lucas. All rights reserved.