People have been asking for a while why I don’t find the kalam cosmological argument very convincing, so I thought it would be useful to have this to refer them to. For those who are unfamiliar with this argument, see my outline of it here.
1) I don’t think objections based on A/B-theory of time are very good. It seems to me that a KCA could just as easily be made using a B-theory of time.
2) Most objections to the KCA are terrible. If you’re objecting because you think the KCA commits the fallacy of composition, or if you’re objecting because you’re a mereological nihilist, then you are not objecting to it properly.
3) I am enormously grateful to William Lane Craig for his work in Christian apologetics, and his work has greatly inspired my own. This shouldn’t be taken as a personal attack!
Now that those indefensibly dismissive disclaimers are out of the way, I can turn to my own concerns.
P1 Everything which begins to exist has a cause
Craig’s three arguments for this premise are as follows:
1.1 Ex nihilo, nihil fit (out of nothing, nothing comes)
1.2 The non-discrimination of nothingness
1.3 Experiential confirmation
1.1 Ex nihilo, nihil fit
It is hard to see exactly what 1.1 means in this context, and how it supports P1. There are two options: either “out of nothing, nothing comes” is semantically equivalent to “everything which begins to exist has a cause”, or it is not. If it is, then this clearly cannot be used as a non-question-begging premise in support of P1. If not, then we will have to give a clearer analysis of what it means. If Craig means the following:
For any time t, if at t nothing exists, then for any positive n, at time t + n nothing exists.
Then the principle seems very probably correct (ignoring concerns about whether “nothing exists at t” is even possible, and the relevant semantics and truth conditions for counterpossibles). But then it is difficult to see how the principle supports P1. For surely no reasonable (note the emphasis – clearly many atheists actually think this) atheist is going to think that there really was a time when nothing existed, and that there was then a later a time when something existed. It seems to me that most atheists will say either that time extends infinitely in the past, and that at each past time something has existed (and so they will reject P2), or that time itself had a beginning, and so there was no time when nothing existed. This latter option is perfectly compatible with P2 and with ex nihilo, nihil fit (as I have construed it), but also seems perfectly consistent with the rejection of P1.
1.2 The non-discrimination of nothingness
A similar objection can be brought against 1.2. Craig seems to want to say that if an atheist denies P1, they will have to think that nothingness can give rise to universes, but not to anything else. Even if it were true that an atheist would have to agree that universes were unique in the relevant respect, it would not be true that an atheist has to think that nothingness can give rise to universes. The reason for this is given above: that no reasonable atheist will really think that nothingness gives rises to universes in any intelligible sense.
1.3 Experiential confirmation
Craig’s third argument for P1 is an inductive one: Craig alleges that we have overwhelming inductive confirmation of the principle, since everything we have observed (other than the universe – this caveat will be assumed henceforth) which began to exist has had a cause. But there are a few problems here: firstly, it’s not clear that everything we have observed which began to exist had a cause. Secondly, it’s not clear that we observe causes at all, so it’s difficult to see that we have overwhelming experiential/observational confirmation of this principle. Thirdly, it’s hard to see exactly what kind of inductive inference is used to reach P1 from this ostensible fact. Even if it was known to be the case that everything we have observed which began to exist had a cause, we would have to explain what kind of inductive inference we are using to reach P1. I am firmly of the opinion that statistical syllogisms are invalid, since we can easily come up with parodies:
A1) Everything we have observed so far (except the universe) has not been a universe.
Therefore (with high probability),
A2) The universe is not a universe.
This is obviously absurd. Even weaker inductive principles saying that P(the universe is not a universe) > 0.5 clearly fail. Consider this parody too:
B1) Everything we have observed so far has had a cause.
Therefore (with high probability):
B2) Everything has a cause.
Theists will obviously want to deny this move, since they deny that God has a cause. So the inference is not necessarily a good one.
It seems to me that Bayesianism provides the best account of valid inductive inference. For 1.3 to be useful, we will want an argument that P(P1|E1 & E2) is pretty high, where E1 is the information that everything else we have observed which began to exist had a cause, and E2 is everything else we know (again, this is only useful supposing we grant E1). This depends on P(P1|E2), P(E1|P1 & E2) and P(E1|E2). But it hasn’t been shown that P(P1|E2) is particularly high, and we might have very good reason to think it is low to the extent that we think it is intrinsically probable that just one thing began to exist without a cause. To overcome this intrinsic improbability, the ratio P(E1|P1 & E2)/P(E1|E2) must be sufficiently large. But this does not seem demonstrably true. P(E1|P1 & E2) is 1, to be sure. But P(E1|E2) does not seem to me to be very low. Our whole reason for thinking P1 is false might be that while it seems plausible that most things which begin to exist have causes, it seems implausible that the universe as a whole should have a cause. This is a perfectly intelligible idea: we readily accept that B2 is implausible because there is at least one exception (viz. God), even though we expect (almost) everything else we observe to have a cause. Similarly, P1 might be implausible because there is one exception (the universe), even though we expect (almost) everything else we observe which began to exist to have a cause.
So at the very least, the inductive step will have to be made a lot clearer for us to consider 1.3 to be a good justification for P1.
Aside from these problems, there is some ambiguity over exactly what is meant by a cause. In particular, we will have to find out if “cause” is understood in the classical sense where causes necessitate their effects, or if causes do not necessitate their effects. If causation is understood classically and P1 is granted, then we run into a problem: there must be intra-universal determinism. But even Craig doesn’t grant this: Craig holds that humans have libertarian free will, and therefore are not compelled by anything in certain circumstances. But then certain things (free intentions to perform an action) do not have causes in the classical sense.
Of course, it is possible to reject the classical understanding of causation, but this brings its own attendant problems: in the first place, it is difficult to see what it means to say that some event is caused if the cause isn’t sufficient for the event. I would love to defend this in greater detail, but here is not the place. Perhaps this concern can be alleviated. But allowing that causes do not have to be sufficient for their effects may create further trouble for Craig’s arguments for the personhood of the cause – we will see this later.
P2 The universe began to exist
I don’t have much objection to this premise, but it’s worth noting a few issues. Firstly, it is unclear whether the universe mentioned here is what physicists often mean by a universe (that is, just one local region of space-time, possibly as part of larger multiverse), or whether “all of space-time” (including other space-times minimally or not at all related to ours) is meant. For Craig’s later arguments to work, he needs to mean “all of space-time”, but then it seems to me that the scientific arguments for the universe having a beginning are not as powerful as we might think. While the impression I get is that most physicists would agree that this local region of space-time had a beginning, it is not so clear that physicists would agree that all space-time in existence had a beginning. Many physicists will be hesitant, saying things like, “the Big Bang was the beginning of the universe as we know it”, indicating that there may be more to the story. Perhaps I am overestimating the level of dissent to this thesis within the scientific community. I’m not well-read enough to know, but I simply note this here as a point I am yet to be convinced of, whether that is because of my own ignorance or something else.
Meanwhile, I do not find the philosophical arguments to be persuasive. It is difficult to see how one can hold to the impossibility of an actually infinite number of things existing, while holding to some kind of realism about abstract objects, or even while holding to theism. If one is a realist about, for example, propositions, one will surely have to hold that there are an actually infinite number of propositions. And if one holds to theism, plausibly (though not necessarily) one would agree that God has an actually infinite number of true beliefs. So any argument relying on the impossibility of actual infinities is not going to be too favourable to theism. And I’m personally not convinced that the arguments against the possibility of actual infinities are successful ones. Craig formalises the intuitive argument behind Hilbert’s hotel thus:
(i) There are not more things in a multitude M than there are in a multitude M’ if there is a one-to-one correspondence of their members.
(ii) There are more things in M than there are in M’ if M’ is a proper submultitude of M.
(iii) An infinite multitude exists.
These three propositions cannot all be true. Since (i) and (ii) seem the most innocuous, we should reject (iii).
But (i) and (ii) are far from innocuous, especially when put together. After all, the typical motivation for (i) is an acceptance of standard Cantorian set theory. But anyone who accepts Cantor’s system as a way of understanding infinities will surely reject (ii). (i) and (ii) are far from innocuous when put together. So it doesn’t seem like Craig’s primary philosophical argument for P2 is successful.
A further brief concern about the philosophical justifications for P2: it seems that if something is necessarily false (and certainly known to be so), then the probability of it given any set of propositions is either undefined or 0. If this is the case for P2, then P(P2|theism) = P(P2|atheism) = 0, and so P2 cannot confirm theism over atheism or vice versa. So it is then hard to see how it can be used as a datum in support of theism.
The nature of the cause
Craig holds that the cause of the universe must be immaterial, timeless, uncaused and personal. As I mentioned earlier, for the arguments for an immaterial and timeless cause to work, Craig has to argue that all space-time has a cause – and even then it’s not obvious that the cause must be timeless. For example, perhaps something within time but non-spatial caused all space-time to exist – to object to this, one has to show that it is sufficiently improbable, in light of our knowledge, that there should be something which is inside time but outside of space-time. But the real problem, for me, is with Craig’s arguments for the personhood of the cause. Craig offers three arguments here:
2.1 There are two candidates for timeless, immaterial beings: abstract objects and unembodied minds. Abstract objects cannot cause anything. So the cause of the universe must be an unembodied mind.
2.2 There are two types of causal explanation: personal explanation and scientific explanation. Scientific explanations involve laws acting on initial conditions and therefore cannot explain the universe, since there were no initial conditions before the universe. So the explanation must be personal.
2.3 A mechanistic cause provides sufficient conditions for its effect. But this means that if a cause of the universe existed timelessly, then the universe would exist timelessly too. But since the universe doesn’t exist timelessly, the cause cannot be timeless. So only a personal agent with real freedom is capable of causing a universe to exist.
It seems difficult to justify the claim that 2.1 exhausts all the possible causes of the universe. Defenders of the KCA will typically ask for alternative possibilities – but it should be obvious that we don’t need to come up with another detailed possibility in order to rationally resist the idea that abstract objects and minds are the only alternatives. In any case, we should be immediately suspicious of any argument based on such a fluid expression as “materialism”. Our physics has changed enough over the past century to show us that plenty of non-personal things exist which might not previously have been considered “material” – many of these things will not have been conceived of at some time in the past. But they nevertheless existed as conceivable, non-personal things with causal power, and which eventually came to stand under the umbrella of the material. So it seems far too quick to assume that the lack of a presently conceived detailed alternative would show that a personal agent must have done it.
Moreover, it is surely difficult to defend the idea that a timeless unembodied mind is much more prima facie coherent than a timeless material object. I have about as much reason to think one is impossible as the other (aside from religious commitments), since both are relatively inconceivable to me. The only reason I have for thinking that a timeless non-personal cause is incoherent is that I cannot conceive of one – but I have an exactly parallel reason for thinking that a timeless personal cause is incoherent. So this argument will not do much to persuade me that the cause is more plausibly personal than impersonal.
I cannot quite see the force behind 2.2. Craig relies on Swinburne’s distinction between personal and inanimate explanations, and the reason inanimate explanations fail in this case is because, on Swinburne’s characterisation, they involve laws acting on initial conditions to bring about later conditions. That is, the action of the law necessarily involves a change in time. But Craig thinks that simultaneous causation is possible, and gives examples of inanimate causes causing a simultaneous effect to support this, which undermines Swinburne’s characterisation of inanimate explanation completely. Craig denies the very proposition which is required for inanimate explanations to fail in explaining the universe, and so I fail to be persuaded by this argument.
2.3 seems to me to equivocate on “timeless”, seems to rely on some very dubious premises, doesn’t seem to exclude every option other than personal agents, and seems to present a comparable difficulty for theism. In the first place, it relies on the premise that if a mechanistic cause existed timelessly/eternally, then so would its effect. But I don’t see much reason to think that this should be the case. This can be illustrated by considering a similar difficulty for theism: if God’s intention is sufficient for the universe to exist, and if God’s intention existed timelessly, then the universe should exist timelessly.
Perhaps the premise would have some plausibility if it were something more akin to this: “if a mechanistic cause existed atemporally, then so would its effect”. But Craig’s premise is more like, “if a mechanistic cause existed “timelessly” in at least one sense, then its effect would also exist “timelessly” in another sense.” I simply don’t see why any premise of this kind should be persuasive. There doesn’t seem to me to be anything in the metaphysics of causation to suggest that causes and effects should have an extremely similar relation to time, and indeed, the theist will want to deny that causes and effects necessarily have such similar relations. So I can’t see that this is a good justification for thinking that the cause of the universe is personal.
Finally, 2.3 doesn’t seem to exclude non-mechanistic non-personal causes. Since Craig seems to think that some x can cause some y even though x is not sufficient for y (that is, with a rejection of the classical understanding of causation), this seems to be a perfectly plausible option. I have little more to say on this point, but I note it for completeness.
One final problem
Again, for completeness, I should briefly detail one or two final concerns. These are to do with the framework of the argument as a whole. The first is that in general, I find many deductive arguments unsatisfactory. It is not always clear how the plausibility of the premises is supposed to relate to the plausibility of the conclusion. To secure a conclusion which, given our evidence, is more probable than not, the conjunction of the premises of a deductive argument must have a probability of greater than 0.5. But the fact that each premise is more likely than not, given our evidence, is certainly not sufficient to establish this.
This raises a further problem, something of a “dwindling probabilities” argument: we can only guarantee a probability of the conclusion insofar as the conjunction of the argument’s premises is probable. But if each premise is only somewhat more probable than not, the probability of the conclusion we can derive drops off pretty quickly. Suppose we think that P(P1) = P(P2|P1) = 0.7: then P(P1 & P2) = 0.49. Let P3 be the premise that if the universe has a cause, the cause of the universe is personal. Suppose P(P3|P1 & P2) = 0.7. Then P(P1 & P2 & P3) = 0.343. Of course, some might question these probabilities: but it is clear that the more uncertain steps there are, the more the probability we can guarantee for the conclusion decreases. Since a lot of steps in the KCA are far from certain, and since some seem only slightly more probable than their negations, this presents a serious problem for the KCA.
In conclusion, there are a number of problems with the KCA. I know that Craig has addressed many of these, and I have read some of Craig’s responses – I invite my reader to assume that I have found them unconvincing, but I have not had enough time so far to read the voluminous literature Craig has produced on the subject. I am happy to concede that I would not be surprised if Craig addressed at least some of these concerns adequately (particularly regarding the scientific arguments). I’m happy to be proven wrong – it’s always nice to have another compelling argument for theism – but I have to be honest in saying that I find the KCA uncompelling, at least as a deductive argument.*
* I have recently written a paper explicating a Bayesian approach to the kalam cosmological argument. I am hoping to publish it so cannot say much here – but watch this space!