Oppca 1

Download as pdf or txt
Download as pdf or txt
You are on page 1of 37

COSMOLOGICAL ARGUMENTS

Suppose that arguments are identified with—or, at any rate, characterized by—sets of

sentences. If a set of sentences C is a cosmological argument, then there is a c∈C that is

identified as the conclusion of the argument; and the remaining c’∈C, c’≠c, are the

premises of the argument. (For later purposes, let Δ = {c’∈C: c’≠c}.)

There are various ways in which the conclusion of a cosmological argument might be—

or might be intended to be—related to the premises of that argument.

For instance, it might be—or it might be intended to be—that it is necessarily that case

that, if Δ then c. Or it might be—or it might be intended to be—that it is obvious a priori

that, if Δ then c. Or it might be—or it might be intended to be—that it is

uncontroversially a truth of logic that, if Δ then c. And so on. Speaking roughly, in these

kinds of cases, we might say that the argument is a deductive cosmological argument.

Alternatively, it might be—or it might be intended to be—that it is highly probable that,

if Δ then c. Or it might be—or be intended to be—that the best explanation of why Δ—or

of why Δ’, for some Δ’⊆ Δ—is that c. And so on. Speaking roughly, in these kinds of

cases, we might say that the argument is an evidential cosmological argument.


2

Of course, the above considerations—concerning kinds of relationships between premises

and conclusions—are not unique to cosmological arguments. What distinguishes

cosmological arguments from other arguments is the nature of c (the conclusion of a

given cosmological argument), and the nature of Δ (the set of premises of that given

cosmological argument).

In principle, it could be that philosophers offered cosmological arguments for specifically

Christian doctrine, e.g. the claim that there is an immaterial, omnipotent, omniscient,

wholly good creator (ex nihilo) and sustainer of all things who is three persons in one

substance, with one of these three persons being numerically identical to a human being

who died to atone for human sins; who exercises providential control over free human

beings; who will bring about the bodily resurrection of all to eternal life; who allows

some lives to lead to eternal bliss and other lives to lead to eternal torment; and who is

the ultimate author of authoritative (and perhaps inerrant) scripture, viz. the Christian

Bible.

In practice, however, what Christian philosophers typically do is to offer cosmological

arguments for the conclusion that some proper part of the role that they take to be

occupied by the Christian God is actually occupied. Exactly how these sub-roles are to be

characterized is a matter of controversy, but it seems to me to be right to say that they all
3

arise from particular interpretations of the claim that the Christian God is the sole creator

of ‘the world’ ex nihilo.

We can think of Christian philosophers as beginning with the following picture:

GOD ⇒ THE WORLD

where both ‘THE WORLD’ and ‘⇒’ are items in need of further explanation. In typical

cosmological arguments, what Christian philosophers then aim to show is that there is a *

such that

* ⇒ THE WORLD

where ‘THE WORLD’ and ‘⇒’ are given that particular further explanation.

Given that this is an accurate characterization of the way in which typical cosmological

arguments proceed, it is clear that there are several different points at which the

conclusions of arguments of this kind might be attacked. In particular, on the one hand, it

is possible for opponents to claim that, while there is a * that occupies the role in

question, that * is not the Christian God. And, on the other hand, it is possible for

opponents to claim that there is no occupant of the role in question. Of course, it is also
4

possible for opponents to adopt a mixed strategy, i.e. to claim that, on some

interpretations of ‘⇒’ and ‘THE WORLD’, there is no occupant of the role while, on

other interpretations of ‘⇒’ and ‘THE WORLD’, there is an occupant of the role but that

occupant is not the Christian God.

Given that we are talking about arguments for a conclusion of the form ‘there is a * such

that * ⇒ THE WORLD’, there are numerous ways in which these arguments are

susceptible to criticism. In some cases, the relationship that is alleged to hold between the

premises and the conclusion will be contested. In some cases, the premises will be

claimed to be inadequate: false, or rationally rejectable, or evidently controversial, or the

like. In some cases, both of these criticisms will be made together. (Some philosophers

may think that there is another class of criticisms that I have overlooked: criticisms that

hold that certain arguments are circular, or question-begging, or the like. I take it that

these kinds of criticisms may be subsumed under criticisms of the premises: given that a

claim is plainly controversial, any circular or question-begging argument with that claim

as conclusion will also have premises that are plainly controversial, and hence unsuited to

the persuasive task that I take to be the principle aim of cosmological arguments. 1 )

One way to approach the task of criticizing cosmological arguments is to examine them

one by one, scrutinizing the premises and inferential steps in each argument. If one

adopts this approach, one will find that there are many defective cosmological arguments
5

out there. Moreover, in my opinion, one will also find that, to date, there is no persuasive

cosmological argument that has been produced. 2 (That is, there is no cosmological

argument which is such that it ought to persuade reasonable people like me who are not

already persuaded of the conclusion that there is a * such that * ⇒ THE WORLD, that

this claim is indeed true, on some appropriate interpretation of ‘⇒’ and ‘THE WORLD’.)

But, even if I am right in claiming that no persuasive cosmological arguments have yet

been produced, it is clear that the truth of this claim does not in any way foreclose on the

possibility that there is a persuasive cosmological argument out there, awaiting

formulation. In particular, it seems, those of us who deny that there is a * such that * ⇒

THE WORLD—on any suitable interpretation of ‘⇒’ and ‘THE WORLD’—are required

to have some justification for our contention that there is no such *.

As I have argued elsewhere 3 , I do not suppose that it is incumbent upon opponents of

cosmological arguments to come up with arguments that ought to persuade Christian

philosophers that there is no God of the kind in which they believe. Indeed, I see no

reason to suppose that such arguments will ever be within the reach of non-Christians.

However, it should be possible for reasonable non-Christians to explain to reasonable

Christians how things look from their point of view, in such a way that reasonable

Christians can see how it is possible for reasonable people to see things that way, even

though, of course, they suppose that those who see things that way hold false beliefs in

consequence of seeing things that way. (And vice versa: it should be possible for
6

reasonable Christians to explain to reasonable non-Christians how things look from their

point of view, in such a way that reasonable non-Christians can see how it is possible for

reasonable people to see things that way, even though, of course, they suppose that those

who see things that way hold false beliefs in consequence of seeing things that way. But

that’s not my present topic.)

So, what I propose to do in this paper, is to set out my reasons for thinking that there is no

* such that that * ⇒ THE WORLD, on any interpretation of ‘⇒’ and ‘THE WORLD’

that would set this claim at odds with anti-supernaturalism. As just indicated, my aim is

not to convince Christians to embrace anti-supernaturalism. Rather, my aim is to

convince Christians that it is highly unlikely that there is a convincing cosmological

argument that ought to persuade reasonable anti-supernaturalists like me to give up their

anti-supernaturalism. (Remember: I’m taking for granted that cosmological arguments

that have been presented so far are representative of the cosmological arguments that

there are; and I’m also taking for granted that all cosmological arguments that have been

produced so far have been shown to be unpersuasive.)

I shall divide the argumentative strategies that I consider into three classes, which I shall

call TOTALITY, CHAIN, and FIRST. I shall try to say more about the nature of these

classes later. For now, a rough initial characterization will suffice.


7

First, arguments that belong to TOTALITY aim to show that there is a certain totality—

the conjunction of all contingently obtaining states of affairs, the aggregate of all

contingently existing objects, the aggregate of all contingently occurring events, or the

like—to which God stands in an appropriate relation, as cause, or sufficient reason, or

explanation, or the like. Of course, in the notation introduced above, the totality in

question is THE WORLD.

Second, arguments that belong to CHAIN aim to show that, if one starts with

uncontroversial instances of pairs that stand in appropriate relations—of cause and effect,

or explainer and that which is explained, or sufficient reason and that which is provided

with sufficient reason, or the like—and then traces back the ancestry of pairs that stand in

that relation, one will eventually come to a pair in which God is—or is crucially

implicated in—the first member of the pair, i.e. the cause, or the explainer, or the

sufficient reason, or the like, of the second member of the pair. (Moreover, of course,

these arguments will also aim to show—or will take for granted—that there is no pair in

which God is—or is crucially implicated in—the second member of that pair; i.e., there is

nothing that is a cause, or an explainer, or a sufficient reason, for God, or God’s

existence, or God’s intentions, or God’s reasons, or the like.)

Third, arguments that belong to FIRST aim to show that, if one considers the temporal

organization of THE WORLD, then one finds that there is an initial PART (or STAGE,
8

or SEGMENT) of the WORLD that requires a cause, or an explanation, or a sufficient

reason, or the like.

The class of paradigmatic instances of TOTALITY includes the cosmological arguments

recently propounded by Robert Koons 4 , and by Richard Gale and Alexander Pruss 5 ; the

class of paradigmatic instances of CHAIN includes at least the first three of Aquinas’ five

ways; the class of paradigmatic instances of FIRST includes numerous versions of the

Kalām cosmological arguments 6 .

There are some arguments that have been called ‘cosmological arguments’ that do not fit

neatly into this system of classification. However, it seems to me that these arguments are

much less worthy of consideration than the familiar, standard arguments that do clearly

fit neatly into this scheme. Consequently, I shall have nothing more to say about these

other so-called ‘cosmological arguments’ here. (One possible example, I think, is the

family of cosmological arguments that have been defended by Barry Miller in a steady

stream of publications over the past thirty years or so. 7 )

1. TOTALITY

It is uncontroversial that it is part of orthodox Christian doctrine to claim that God is the

maker of heaven and earth. The intuition that guides the construction of arguments that
9

belong to TOTALITY is that it is possible to show that it must be the case that—or that it

is highly probable that it is the case that, or that it is most reasonable to believe that—

there is a maker of heaven and earth. (There is a * such that * MAKES HEAVEN AND

EARTH.)

In order to construct a particular argument that belongs to TOTALITY, we need to decide

what we mean by HEAVEN AND EARTH. There are at least four potential candidates

here.

First, we might mean (roughly) the mereological aggregate of particular individuals, or

objects, or things. (An alternative possibility is that one might mean the set, or class, of

particular individuals, objects, or things. But I do not think that there are any advantages

to be gained by framing cosmological arguments in terms of set theory rather than in

terms of mereology.) This formulation is rough because there are some particular

individuals—most notably, God—that should not be thought to be part of HEAVEN

AND EARTH (at least for the purposes of the present exercise). However, it does not

seem particularly problematic to repair the formulation so as to overcome this difficulty.

Second, we might mean (roughly) the maximal actual state of affairs, i.e. the global way

that things are. This formulation is rough because we need to ‘subtract’ the way that God

is from the global way that things are in order to arrive at a characterization of the way
10

that heaven and earth are. Moreover—as we shall soon see—it is no easy matter to see

how to repair the initial formulation in a way that effects the needed ‘subtraction’.

Third, we might mean (roughly) everything that happens, i.e. the sum total of history, or

the maximal actual event. Again, this formulation is rough because we need to ‘subtract’

God-events from the global total of events in order to arrive at a characterization of the

total heaven-and-earth event; and, again, it is no easy matter to see how to repair the

initial formulation in a way that effects the needed ‘subtraction’.

Fourth, we might mean (roughly) everything that is true, i.e. the maximal actual truth, or

the maximal actual fact. Once again, this formulation is rough because we need to

‘subtract’ God-facts from the maximal actual fact in order to arrive at a characterization

of the total heaven-and-earth fact. But the formulation is also rough because we need to

‘subtract’ necessary facts as well. (Perhaps there is a similar need to ‘subtract’ necessary

states of affairs in the previous case as well. Whether there is such a need depends upon

the theory of states of affairs that one adopts.) There are even more formidable

complexities here than there are in the case of the maximal actual state of affairs and the

maximal actual event.

In order to construct a particular argument that belongs to TOTALITY, we also need to

decide on a suitable interpretation of ‘⇒’ in our conclusion ‘there is a * such that * ⇒


11

THE WORLD’. There are several potential candidates here. For example, we might think

that HEAVEN AND EARTH requires a cause; or we might think that HEAVEN AND

EARTH requires a sufficient reason; or we might think that HEAVEN AND EARTH

requires an explanation; and so forth. (Strictly, the formulations here are required to vary

depending upon which interpretation of HEAVEN AND EARTH is adopted. So, for

example, we might think that the existence of a particular mereological aggregate requires

explanation; or that the coming into existence of that particular mereological aggregate

requires a cause, or an explanation, or a sufficient reason. Or we might think that the

obtaining of a particular near-maximal actual state of affairs requires explanation; or that

the coming to obtain of that particular near maximal actual state of affairs requires a

cause, or an explanation, or a sufficient reason. Or we might think that the occurring of a

particular near-maximal event requires an explanation; or that the coming to occur of that

particular near-maximal event requires a cause, or an explanation, or a sufficient reason.

Or we might think that the obtaining of a particular near-maximal fact requires

explanation; or that the coming to obtain of a particular near-maximal fact requires an

explanation or a sufficient reason, or the like. Perhaps we shan’t need to fuss too much

about these grammatical distinctions in what follows.)

1.1 Near-Maximal Mereological Aggregate


12

Let U be the maximal mereological aggregate that is HEAVEN AND EARTH (on the

mereological interpretation). Then, according to Christian doctrine, there is an entity *

such that * ⇒ U, on one of the above described interpretations of ‘⇒’; but it is not the

case that there is an entity $ such that $ ⇒ *+U, where, of course, *+U is the

mereological fusion of * and U. So, if there is to be a successful cosmological argument

constructed in these terms, then we need a compelling reason for thinking that there is a *

such that *⇒U, even though there is no $ such that $ ⇒ *+U. What could this

compelling reason be?

The first point to note is that, on each view—i.e. the Christian view according to which

there is a * such that * ⇒ U, and the anti-supernaturalist view according to which there is

no such *—there is a mereological aggregate A for which it is true that there is no # such

that #⇒A. So it isn’t true that the Christian view should be favoured because it avoids the

existence of mereological aggregates A for which there is no # such that #⇒A.

The second point to note is that there is no evident, neutral reason for supposing that one

of U and *+U is a better candidate for being the maximal mereological aggregate A for

which there is no # such that #⇒A. Of course, that’s not to say that there aren’t putative

reasons that might be suggested. For instance, it might be said that, while U is an entity

that only exists contingently, * is an entity that exists necessarily (and, of course,
13

whatever is necessary has, and requires, no explanation, since there is no alternative to

the necessary that could be set aside by adverting to considerations of some kind). But the

difficulty with this suggestion is that, even if we suppose that * exists necessarily, we

need to examine our ⇒ in order to decide whether there is a genuine advantage here.

There are two doxastic possibilities. On the one hand, it might be necessary that * ⇒ U.

But, in that case, since * exists necessarily, it will surely follow that U exists necessarily

(contradicting the assumption that U exists only contingently, which was the reason

advanced for supposing that there is genuine advantage in postulating *). On the other

hand, it might be contingent that * ⇒ U. But, in that case, what is the clear advantage in

postulating the existence of a (necessarily existent) * such that *⇒ U, given that there

remains unexplained contingency in the relationship between * and U? Why not settle

instead for unexplained contingency in the existence of U? (There is more to say here; I

shall return to these considerations below.)

1.2 Near-Maximal Actual Event

Suppose that N is the near-maximal actual event that is required by the interpretation in

terms of events. Then, according to Christian doctrine, there is a * such that *⇒N, for

some suitable interpretation of ‘⇒’; but there is no $ such that $ ⇒ (*’s ⇒–ing N),

where, of course, *’s ⇒-ing N is a further event.


14

This is trouble. Recall that the aim of the cosmological arguer is (roughly) to show that

there must be a * such that *⇒N. But, if there must be a * such that *⇒N, how can we

avoid the conclusion that there must be a $ such that $ ⇒ (*’s ⇒–ing N)?

We might try arguing that *’s ⇒–ing N is necessary; but, as we have already seen, that

avenue of argument leads to a dead end. We might try arguing that *’s ⇒–ing N requires

no $ such that $ ⇒ (*’s ⇒–ing N) on the grounds that *’s ⇒–ing N is self-explanatory,

or the like. But that route too is closed: there is nothing that is self-explanatory in the

relevant sense. ‘A because A’ is always an explanatory solecism; ordinary talk of ‘self-

explanation’ is never properly intended to be interpreted in this flat-footed way. (When,

for example, I say that the instructions are ‘self-explanatory’, what I mean is that you

won’t need to refer to anything other then the instructions in order to figure out what to

do. But I take for granted a whole range of competencies which it is perfectly possible

that cognitive agents might lack; and I can happily acknowledge that agents who lacked

those competencies would need to refer to more than the instructions in order to figure

out what to do.)

Perhaps one might try arguing that * marks the appropriate initial point for ⇒ because it

is the intentions and reasons of * that are the engine of ⇒, and intentions and reasons are
15

not the kinds of things for which there can be an ⇒. This claim seems to me to be

evidently false. It is true that, in the ordinary business of giving explanations, we often

treat intentions and reasons as suitable ‘stopping points’ for explanation. (‘Why did x do

A? Because x had beliefs {B} and desires {D}.’) But this is not because it is the case that,

or because we think it is the case that, intentions and reasons are not the kinds of things

for which there can be an ⇒. Rather, it’s because we know that we typically lack both

theory and information needed in order to identify appropriate ⇒ ’s. Compare with the

way that we typically think about coin tosses. In everyday life, we treat coin tosses as

effectively indeterministic events; but that’s not because we think that coin tosses really

are indeterministic events (except, perhaps, insofar as coin tosses are subject to quantum

indeterminacy—but that’s an irrelevant side issue). We have lots of evidence—

concerning, for example, the distribution of intentions and reasons across human groups

throughout history—that supports the claim that intentions and reasons are the kinds of

things for which there are ⇒ ’s. At the very least, I think that it is clear that there is no

evident, neutral reason for thinking that * marks the appropriate initial point for ⇒

because it is the intentions and reasons of * that are the engine of ⇒: the required claim

about intentions and reasons is at least as controversial as the claim that there is a * such

that *⇒N.

1.3 Near-Maximal Actual State of Affairs


16

Suppose that N is the near-maximal actual state of affairs that is required by the

interpretation in terms of states of affairs. Then, according to Christian doctrine, there is a

* such that *⇒N, for some suitable interpretation of ‘⇒’; but there is no $ such that $ ⇒

(* ⇒’s N), where, of course, * ⇒’s N is a further state of affairs.

As in the case of the near maximal actual event, we now have serious trouble. The

cosmological arguer wishes to insist that a certain near-maximal actual state of affairs

requires an ⇒, while an even more nearly maximal actual state of affairs need not have

an ⇒, even though there is no evident, neutral reason for this differential treatment. (The

more nearly maximal actual state of affairs is the ‘conjunction’ of N with * ⇒’s N. I

assume here that this ‘conjunction’ lacks an ⇒ if one of its ‘conjuncts’ lacks an ⇒.)

As in the case of the near maximal actual event, we can perhaps think of the disagreement

here in ‘economic’ terms. On the one hand, the anti-supernaturalist supposes that N is the

maximal actual state of affairs, and that it has no ⇒. On the other hand, the Christian

supposes that N is a near-maximal state of affairs, for which there is a * such that * ⇒’s

N. On the Christian view, there is no ⇒ for the * such that * ⇒’s N (or, at any rate, there

is no ⇒ for the intentions and reasons that are possessed by this *). From the anti-

supernaturalist standpoint, the Christian proposal is not progress: there is a cost in terms
17

of ontological commitment to *, but there is no advance in reduction of commitment to

states of affairs that lack ⇒’s (since those events that lacked ⇒’s on the initial scheme

are now matched with reasons and intentions that lack ⇒’s on the Christian scheme).

However, from the Christian standpoint—as least as envisaged by proponents of the

cosmological argument now under consideration—the Christian proposal is progress,

since the ontological payment does buy a reduction of commitment to states of affairs

that lack ⇒’s: for, from this standpoint, reasons and intentions are not the kinds of things

that require, or perhaps even can have, ⇒’s.

1.4 Near-Maximal Actual Fact

Suppose that N is the near-maximal actual fact/truth that is required by the interpretation

in terms of facts/truths. Then, according to Christian doctrine, there is a * such that *⇒N,

for some suitable interpretation of ‘⇒’, but there is no $ such that $ ⇒ that (*⇒N),

where, of course, that (*⇒N) is a further fact/truth.

It is clear enough that the same kinds of difficulties that arose in the case of the

interpretation in terms of events and the interpretation in terms of states of affairs also

arise in this case. However, as I noted earlier, there is another kind of difficulty that also

arises in this case, involving the treatment of necessary facts/truths.


18

Given the kinds of interpretations that are appropriate for ‘⇒’ when we are talking about

truths, it is a live question whether ‘⇒’ must at least involve entailment. Certainly, if ‘⇒’

is interpreted in terms of ‘sufficient reason’, then it seems quite clear that ‘⇒’ must at

least involve entailment. But even if ‘⇒’ is interpreted merely in terms of ‘explanation’,

it is plausible that the explanation that is required will involve entailment. (Of course, it is

not appropriate to interpret ‘⇒’ in terms of ‘cause’, when we are talking about truths.)

If ‘⇒’ does involve entailment, and if entailment is given a classical interpretation, then

it cannot be that * is taken to obtain necessarily while N only obtains contingently. Either

N obtains necessarily—in which case it can be plausibly maintained that there is no

explanation of the obtaining of N to be provided—or * obtains contingently, in which

case it is hard to see what kind of theoretical advantage could be obtained by the

postulation of *. Perhaps it might be suggested that entailment should be given a non-

classical interpretation; but, at the very least, that suggestion surely requires weighty

independent support.

1.5 Last Thoughts on TOTALITY

In the preceding discussion, I have focused on the difficulty that is raised for TOTALITY

cosmological arguments by the creative reasons and intentions of the postulated creator.
19

The supposed advantage of the postulation of a creator—viz. the provision of an ⇒ for

the TOTALITY in question—is arguably brought into question by the standing of these

reasons and intentions with respect to the relevant ⇒. Given that the claim that these

reasons and intentions either do not or need not have an ⇒ is no less controversial than

the claim that there is a * such that * ⇒ THE WORLD, we find here compelling reason

for thinking that TOTALITY cosmological arguments are unlikely to achieve their

intended persuasive aim.

Of course, it should not be thought that there are no other difficulties that could be raised

for TOTALITY cosmological arguments. In particular, there are interesting questions to

be raised about indeterminism, and about libertarian free actions. Note that these further

concerns press directly on those versions of TOTALITY that advert to events, or states of

affairs, or facts or truths; if they press on versions of TOTALITY that advert to objects,

or individuals, or things, they do so only indirectly.

If there is indeterminism in some part of THE WORLD, and if there is no * such that *

⇒ the indeterministic part of THE WORLD, then it surely follows that there is no * such

that * ⇒ THE WORLD. Since there is strong reason to suppose that there is

indeterminism in our world, there is strong reason to suppose that there is no * such that *

⇒ THE WORLD. Of course, it is at least doxastically possible that, the empirical


20

successes of quantum mechanics notwithstanding, our world is deterministic—but it is

very hard to believe that there is evident, neutral reason to believe that the world is

deterministic.

Similarly, if there are libertarian free actions in some part of THE WORLD, then—at

least in the eyes of many philosophers—it follows that there is no * such that * ⇒ that

part of THE WORLD. Whence it follows that, if there are libertarian free actions, then

there is no * such that * ⇒ THE WORLD. Of course, it doesn’t follow from these

considerations that there is no * such that * ⇒ THE WORLD; rather, what follows is

that, if there is a * such that * ⇒ THE WORLD, then there are no libertarian free actions.

This result seems to me to be bad news for proponents of TOTALITY cosmological

arguments. On the one hand, those proponents of these arguments who are strongly

wedded to the idea that we have libertarian freedom have to fold their cards. On the other

hand, those proponents of these arguments who are prepared to give up the idea that we

have libertarian freedom have various battles to fight without falling back on that familiar

trump: no free-will defense against logical arguments from moral evil; no free-will

theodicy to match against evidential arguments from evil; and so forth. (It is, I think, not

very controversial to claim that it is much harder to find arguments for the claim that, if

there is a * such that * ⇒ THE WORLD, then that * such that * ⇒ THE WORLD is a

(morally) perfect being, if one adopts a compatibilist conception of freedom. So,


21

plausibly, the ultimate aim of the Christian TOTALITY arguer will be hindered by the

mooted concession in connection with libertarian freedom.)

Of course, some philosophers have denied the claim that if there are libertarian free

actions in some part of THE WORLD, then it follows that there is no * such that * ⇒ that

part of THE WORLD. In particular, some proponents of agent causation insist—contrary

to my earlier assertion—that there can be full explanation without (classical) entailment.

(See, for example, O’Connor (2003).) On this view, there can be a full explanation of

why an agent made a particular choice C rather than any other choice C’ that it was open

to that agent to make in the circumstances of the choice, even though the choice C was

not determined by the global state of the world at any point prior to the time of the

choice. Against this, I’m inclined to say that what we have here is partial explanation—

albeit all of the explanation that there is to be had—of the choice in question. And,

certainly, that is what everyone is inclined to say in all other non-deterministic cases.

(Suppose that the decay of an atom is non-deterministic. Then, while there is a partial

explanation of the decay of the atom—the atom had a certain objective chance of

decaying when it did—it is also true that there is no explanation of why the atom decayed

at the particular time that it did, rather than at any other time at which it might have

decayed. 8 ) However, for present purposes, I don’t need to insist that one could not

reasonably accept the agent causation account. Rather, it suffices to point out that this

agent causation account is no less controversial than the claim that there is a * such that *
22

⇒ THE WORLD. At the very least, it is not a secure foundation upon which to rest a

convincing argument for the existence of God. 9

2. CHAIN

The intuition that guides the construction of arguments that belong to CHAIN is perhaps

slightly less obviously central to orthodox Christian doctrine than is the intuition that

guides the construction of arguments that belong to TOTALITY.

Arguments that belong to CHAIN start with an observation about elements of THE

WORLD—e.g. that there are elements a and b in THE WORLD such that a ⇒ b. The

central intuition is that, where there are such elements a and b, they will belong to an ⇒-

CHAIN for which there is a * such that * is the first element of that ⇒-CHAIN. More

exactly, the intuitive picture is that there is a * such that, for any elements a and b such

that a ⇒ b, there is an ⇒-CHAIN in which * is the first element, and which has a and b

as members.

As in the case of TOTALITY, there are interpretative choices that need to be made in

order to arrive at a particular member of CHAIN. There is a choice to be made about the

elements of THE WORLD: are we talking about objects, or events, or states of affairs, or
23

facts, or what? And there is a choice to be made about ⇒: are we talking about causation,

or explanation, or sufficient reason, or something else?

There are difficulties in getting arguments that belong to CHAIN to turn out to be valid.

If we allow ourselves assumptions about the absence (or impossibility) of circles and

regresses, we can get the result that any given ⇒-CHAIN has a first element, i.e. an

element that is the first member of at least one ⇒-relation, but is not itself the second

element in any ⇒-relations. (Even if we don’t allow ourselves assumptions about the

absence (or impossibility) of regresses, we can still get the result that any given element

has an ancestral first element, provided that we insist that regressive ⇒-CHAINS must

themselves be elements in ⇒-relations, and provided that we can help ourselves to the

axiom of choice. 10 ) But the aim of the argument is not merely to establish that, for any

elements a and b such that a⇒b, there is a * such that * is the first element in an ⇒-

CHAIN that includes a⇒b; rather, the aim of the argument is to show that there is a *

such that, for any elements a and b such that a⇒b, * is the first element in an ⇒-CHAIN

that includes a⇒b.

Among the evident difficulties that arise for arguments that belong to CHAIN, there are

the difficulties that are raised by intentional action and indeterminism.


24

If there is indeterminism in THE WORLD, then different ⇒-CHAINS will terminate in

different first members that lack ⇒’s. If there is indeterminism in THE WORLD, it may

be that not all ⇒-CHAINS that commence with contemporary elements begin with

members that lack ⇒’s; but, if we help ourselves to standard Big Bang Cosmology, in

which the initial stages of our universe are confined entirely to the quantum realm, then it

is not clear that it is even ruled out that all ⇒-CHAINS that commence with

contemporary elements begin with members that lack ⇒’s, where these members are

simply elements of the quantum domain. (More about these kinds of considerations in a

moment.)

Unless we are eliminativists about either folk psychology or causation, we shall suppose

that there are ⇒-CHAINS that contain contemporary elements, and that have the like of

the intentions and reasons of human agents ‘upstream’ from those contemporary

elements. If we suppose that intentions and reasons do not have ⇒’s, then we cannot get

the conclusion that there is a * such that, for any elements a and b such that a⇒b, * is the

first element in an ⇒-CHAIN that includes a⇒b. On the other hand, if we suppose that

intentions and reasons do have ⇒’s, at least in the case of human agents, then we surely

do not have evident, neutral reason for supposing that the intentions and reasons of God

neither have nor require ⇒’s. Since it is plausible that there are contemporary elements

that lie on ⇒-CHAINS that do not contain the like of the intentions and reasons of human
25

agents ‘upstream’ from those contemporary elements, it should not be thought that the

objection being presented here makes an insurmountable objection to arguments

belonging to CHAIN. For the cosmological arguer could restrict the argument to those

⇒-CHAINS that do not contain the like of the intentions and reasons of human agents.

Suppose, then, that we restrict our attention to those ⇒-CHAINS that do not contain the

like of the intentions and reasons of human agents as elements, and suppose that there is

no indeterminism in THE WORLD that has any consequences for these ⇒-CHAINS. At

least very roughly, in the context of cosmological theorizing, these suppositions amount

to supposing that there is a standard general relativistic Big Bang model that accurately

represents the nature of THE WORLD. If we suppose that the initial singularity is a point

that belongs to the spatiotemporal manifold, then we shall suppose that all of the relevant

⇒-CHAINS lead back to that singularity. On the other hand, it we suppose that the initial

singularity is a surface that belongs to the spatiotemporal manifold, then we shall suppose

that the relevant ⇒-CHAINS lead back to different initial elements. So, it seems, if we

make enough controversial assumptions, even the anti-supernaturalist can get out the

conclusion that there is a * such that, for any elements a and b such that a⇒b, * is the

first element in an ⇒-CHAIN that includes a⇒b. But, even with all of these controversial

assumptions, we don’t get to the conclusion that the CHAIN cosmological arguer seeks;
26

for, of course, even if there were such an initial cosmological singularity, it would not be

the Christian God.

A little reflection suggests that the prospects for finding a persuasive cosmological

argument that belongs to CHAIN are very dim. The anti-supernaturalist supposes that

there are none but naturalistic ⇒’s. Hence, the anti-supernaturalist supposes that either

there are ⇒-CHAINS that regress—as they do, for example, in standard general

relativistic Big Bang models in which the initial singularity is supposed not to be part of

the spatiotemporal manifold—or else that ⇒-CHAINS have naturalistic first elements

that lack ⇒’s—as they do, for example, in standard Big Bang models in which the initial

singularity is supposed to be part of the spatiotemporal manifold, on the further

(naturalistic) assumption that there are no ⇒’s that involve elements that do not

themselves belong to that spatiotemporal manifold.

Of course, none of the above is intended to suggest that supernaturalists—and, in

particular, Christians—are irrational, or otherwise evidently cognitively deficient for

thinking that there are ⇒’s where anti-supernaturalists deny that there are ⇒’s. To help

fix ideas, let’s go back again to the pretence that THE WORLD is accurately

characterized by a standard general relativistic Big Bang model in which a global time

function can be defined. In this case, we can think of THE WORLD as consisting of a

sum of global time slices with ⇒’s ‘governing’ the evolution of these time slices. While
27

the anti-supernaturalist will insist that these ⇒’s all conform to naturalistic scruples, there

have been many Christians who have thought that among these ⇒’s are such things as

God’s conserving THE WORLD in existence. In criticizing the arguments belonging to

CHAIN, I have not been trying to argue that Christian’s cannot reasonably believe that

there are such ⇒’s; rather, my target is the much more modest objective of securing

agreement that there are no evident, neutral reasons for believing that there are such ⇒’s.

If there are no such evident, neutral reasons, then, I think, there is no prospect for finding

a persuasive argument that belongs to CHAIN.

There is perhaps one final point worth making in connection with arguments that belong

to CHAIN before we turn our attention to the final class of arguments that I wish to

examine. It is sometimes the case that defenders of arguments that belong to CHAIN

choose to mix their ⇒’s. So, for example, when the prospect of indeterminism is raised in

connection with arguments that are initially couched in terms of efficient causation, some

cosmological arguers then fall back on the observation that, even if certain kinds of

quantum events lack efficient causes, they nonetheless have material causes. While this

might look like progress, it should then be borne in mind that all of the intentions,

reasons, decisions, and so forth that we meet with in our everyday experience have

material causes. Hence, by the lights of the anti-supernaturalist, this move does not

increase the persuasiveness of the CHAIN argument that is being presented: nowhere in

our everyday experience do we meet with intentions, reasons, decisions, and so forth that
28

do not have material causes. Perhaps we might go so far as to say that it is a ‘defeasible

rule’ that all intentions, reasons, decisions, and the like have material causes; even if we

don’t go this far, we should, at the very least, note that substance dualism is no less

controversial than the claim that the Christian God exists.

It is also perhaps worth thinking a little more about the way in which considerations

about the nature of freedom impact on TOTALITY and CHAIN. I argued above that, if

there is a * such that * ⇒ THE WORLD, on the TOTALITY interpretation, then there is

no libertarian freedom; and I argued that, if there is libertarian freedom, then it is not the

case that there is a * such that, for any elements a and b such that a⇒b, * is the first

element in an ⇒-CHAIN that includes a⇒b (or, in other words: if there is a * such that,

for any elements a and b such that a⇒b, * is the first element in an ⇒-CHAIN that

includes a⇒b, then there is no libertarian freedom). However, as I suggested above, it

seems to me that there is some ‘wiggle room’ in the case of CHAIN: the CHAIN

cosmological arguer can simply set aside all of the ⇒-CHAINS that originate in the

libertarian free choices of natural agents, and argue the case in connection with this

restricted set of ⇒-CHAINS.

Is there a similar strategy that is available to the TOTALITY cosmological arguer? Can

we proceed by ignoring those parts of TOTALITY that involve the libertarian free
29

choices of natural agents, and arguing in terms of what remains? It would seem so.

However, if we do proceed in this way, then it seems to me that we shall be taking up an

argument that can quite properly be said to belong to FIRST (since the intuitive idea is

simply to consider THE WORLD as it was before there were any agents with libertarian

freedom, and to make the TOTALITY cosmological argument in connection with that

stage of THE WORLD). Perhaps this is not the right way to think about these matters;

however, I shall only revisit this assumption if it is shown to have unfortunate

consequences.

3. FIRST

The intuition that guides the construction of arguments that belong to FIRST is that, in

the beginning, the Christian God made heaven and earth. More exactly, what these

arguments seek to show is that there is a * such that * ⇒ INITIAL HEAVEN AND

EARTH (where it is a question for interpretation exactly what is meant by ‘INITIAL

HEAVEN AND EARTH’).

I think that the most natural way to interpret the conclusion of arguments that belong to

FIRST is something like this: There is a time t=0 that marks the beginning of HEAVEN

AND EARTH. Moreover, there is a * such that * ⇒ HEAVEN AND EARTH at t=0.

Given this interpretation, there are two different questions that can be raised. First, is it
30

true that there is a time t=0 at which HEAVEN AND EARTH begins. Second, given that

there is a time t=0 at which HEAVEN AND EARTH begins, is it true that there is a *

such that * ⇒ HEAVEN AND EARTH at t=0.

If THE WORLD were accurately described by a standard general relativistic Big Bang

model, then there is an important sense in which it would not be true that there is a time

t=0 at which HEAVEN AND EARTH begins. For, as I noted earlier, in standard general

relativistic Big Bang models, the time t=0 is a boundary ‘time’ that is not attached to the

spatiotemporal manifold. Consequently, in these models, it is not true that HEAVEN

AND EARTH exists at t=0. Rather, what is true is that HEAVEN AND EARTH exists at

all times t’>0. Of course, in these models, it is also true that HEAVEN AND EARTH

exists at all times; part of what follows from saying that t=0 is a boundary time is that

there is no time t=0 if these models are true.

Now, of course, we shouldn’t suppose that THE WORLD is accurately described by a

standard general relativistic Big Bang model: a full model of the early universe needs to

incorporate quantum considerations as well. However, we don’t yet have an accepted,

detailed model of the early universe. So it remains an open question whether our best

theory will ultimately tell us that there is a time t=0 at which HEAVEN AND EARTH

began. (A further complication here is that, if we are treating standard general relativistic

Big Bang models with full ontological seriousness, then—as John Earman has
31

argued 11 —it is a live question whether we should think that trajectories can be extended

‘through’ the initial singularity. Even if this were not so, surely it would be a live

question whether, in better theories than those that we currently have, it is possible for

there to be trajectories that extend ‘through’ initial singularities. At the very least,

speculations about these kinds of ‘possibilities’ abound in the recent literature, as in, for

example, the Smith/Smolin multiverse theory.)

Suppose, nonetheless, that it is the case that there is a time t=0 at which HEAVEN AND

EARTH began (and that the domain of contingency coincides with the spatiotemporal

manifold S that lies on our side of that initial singular point). Are reasonable anti-

supernaturalists then rationally obliged to go on to say, with the proponents of FIRST

cosmological arguments, that there is a * such that * ⇒ HEAVEN AND EARTH at t=0?

I don’t think so. Let Si = HEAVEN AND EARTH at t=0. Then, on the one hand,

proponents of arguments from FIRST are Christians who suppose that there is a * such

that * ⇒ Si even though there is no $ such that $⇒*. And, on the other hand, anti-

supernaturalists suppose that there is no * such that * ⇒ Si. Structurally, at least, we are

back to the same kind of position that we examined in the case of near-maximal

mereological aggregates: both camps suppose that there is something that lacks an ⇒, but

proponents of arguments in FIRST suppose that there is theoretical advantage to be

gained by postulating an additional supernatural agent, so that it is the intentions and

reasons of that supernatural agent that lack ⇒’s, and so that the intentions and reasons of
32

that supernatural agent can provide an ⇒ for Si. As I have already argued, it is very hard

to believe that there is evident, neutral reason for thinking that the postulation of a

supernatural agent in this case is theoretical progress. (Perhaps it is worth noting here that

the theoretical costs involved may run further than the postulation of the supernatural

agent. Does it need a domain to inhabit? Must this domain have some kind of ‘time’ or

‘spacetime’ associated with it? If so, how should we suppose that this ‘time’ or

‘spacetime’ relates to our spacetime? And so forth. But the key critical point can be made

without adverting to the nature of these possible further costs.)

It seems to me that it is quite proper for anti-supernaturalists to be undecided on the

question whether there is a time t=0 at which HEAVEN AND EARTH began. Current

scientific cosmological theorizing has not reached a stable consensus about the very

earliest history of our universe; consequently, there is no evident, neutral reason for

deciding one way or the other on this question. (Moreover, I think that current scientific

cosmological theorizing has not reached a stable consensus on the question whether our

universe exhausts HEAVEN AND EARTH, so that there is no evident, neutral reason for

deciding one way or the other on that question either.) However, even if it turns out that

there is a time t=0 at which HEAVEN AND EARTH began, it seems pretty clear that

there is no evident, neutral reason for supposing that there is a * such that * ⇒ HEAVEN

AND EARTH at t=0. By the lights of anti-supernaturalists, the postulation of a * such


33

that * ⇒ HEAVEN AND EARTH at t=0 is not justified by the theoretical returns that can

be obtained from this postulation.

CONCLUDING REMARKS

It is highly plausible to suppose that a reasonable Christian and a reasonable anti-

supernaturalist will disagree about a great many things. While they will certainly disagree

on the question of the existence of God, it is also highly likely that they will reasonably

disagree on such matters as the nature of free action, the ontology of mind, the

relationships that hold between reasons and causes, the metaphysics of quantum-vacuum

pair production, and so on.

At any rate, we can certainly suppose that there is a set (or class) of propositions P ={pi}

that are reasonably believed by the Christian, but that are reasonably not believed by the

anti-supernaturalist; and that there is a set (or class) of propositions Q={qj} that are

reasonably believed by the anti-supernaturalist, but that are reasonably not believed by

the Christian.

Let G be the proposition that God exists. Then, as I have just claimed, it seems to me that

we can reasonably suppose that the set of propositions {G, pi} is such that it can be

reasonably believed by the Christian; and it seems to me that we can reasonably suppose
34

that the set of propositions {not-G, qj} is such that it can be reasonably believed by the

anti-supernaturalist. Furthermore, it seems to me that we can reasonably suppose that

there are sets of propositions P’⊆P for which P’⇒G, where A⇒B just in case A entails

B, or A provides strong probabilistic support for B, or the like; and it seems to me that we

can reasonably suppose that there are sets of propositions Q’⊆Q for which Q’⇒not-G,

under the same interpretation of A⇒B. Finally, it seems to me that we can reasonably

suppose that there are sets of propositions {P’, Q’, not-G} and sets of propositions {P”,

Q”, G}, where P’, P”⊆P and Q’, Q”⊆Q, which are logically inconsistent, or

probabilistically inconsistent, or the like.

If this is how things are, then we should not make the mistake of thinking that arguments

P’, Q’ ⇒ G and/or P”, Q” ⇒ not-G are successful arguments for the Christian and/or the

anti-supernaturalist. However, if I am right, it is not plausible to suppose that there are

better cosmological arguments—either for the conclusion that God exists or for the

conclusion that God does not exist—than cosmological arguments of the P’, Q’ ⇒ G

form. Consequently, if I am right, it is not plausible to suppose that there are successful

cosmological arguments.

References

Craig, William. (1979) The Kalām Cosmological Argument London: Macmillan


35

Earman, John. (1995) Bangs, Crunches, Whimpers and Shrieks Oxford: Oxford

University Press

Gale, Richard. and Pruss, Alexander. (1999) “A New Cosmological Argument” Religious

Studies 35, 461-76

Gale, Richard. and Pruss, Alexander. (2003) “A Response to Oppy, and Davey and

Clifton” Religious Studies 38, 89-99

Koons, Robert. (1997) “A New Look at the Cosmological Argument” American

Philosophical Quarterly 34, 193-211

Koons, Robert. (2001) “Defeasible Reasoning, Special Pleading and the Cosmological

Argument: A Reply to Oppy” Faith and Philosophy 18, 192-203

Meyer, Robert. (1987) “God Exists!” Noûs 21, 345-61

Miller, Barry. (1992) From Existence to God London: Routledge

Nowacki, Mark. (2006) The Kalam Cosmological Argument for God Amherst, NY:

Prometheus Books

O’Connor, Timothy. (2003) “Agent Causation” in G. Watson (ed.) (2003) Free Will,

second edition, Oxford: Oxford University Press

Oppy, Graham. (2006a) Philosophical Perspectives on Infinity Cambridge: Cambridge

University Press

Oppy, Graham. (2006b) Arguing about Gods Cambridge: Cambridge University Press

Endnotes
36

1
For further discussion of this point, and for more general discussion of my views about

rationality, justification and argumentation, see Oppy (2006b), Chapters One and Eight.
2
I argue for this claim at length in Oppy (2006b), Chapter One.
3
See, in particular, Oppy (2006b), Chapter One.
4
Koons (1997)(2001)
5
Gale and Pruss (1999)(2003)
6
See, for example, Craig (1979) and Nowacki (2006).
7
See, for example, Miller (1992).
8
While there is more to argue here, it does seem to me that it is worth noting that this just

is how we ordinarily talk about indeterminism in physical systems: if quantum mechanics

is true, then there are things that have no explanation, we say. I reckon that there is no

relevant different in the case of non-deterministic agent causation.


9
We might think to run another “economic” argument at this point. Proponents of non-

deterministic agent causation claim that their view accords better with certain pre-

theoretical intuitions. Opponents of non-deterministic agent causation may disagree about

even this; but they will certainly insist that there are theoretical costs involved in the

adoption of a sui generis category of non-deterministic causation. There seems to be little

prospect of a resolution to this stand-off any time soon.


10
See Meyer (1987) for an extended discussion of this point.
37

11
Earman (1995), cited in Oppy (2006a), Chapter Four.

You might also like