"Bertrand Russell's defence of the cosmological argument."
"Misery," Trinculo tells us, "acquaints a man with strange bedfellows." Sometimes philosophy does, too: the natural theologian, for example, does not ordinarily regard Bertrand Russell as an ally, but Russell has in fact defended the cosmological argument. I do not mean the un-interesting observation that, as Russell changed his mind so many times in his career, he must have accepted the cosmological argument at some time or other. (Indeed, he did accept the cosmological argument when he was seventeen years old, but when he was eighteen he read Mill's Autobiography, where, he reported, "I found a sentence to the effect that his father taught him that the question `Who made meT cannot be answered, since it immediately suggests the further question `Who made God?' This led me to abandon the "First Cause" argument, and to become an atheist."(1)) No, I mean that the mature Russell, the unbeliever who claimed that "All religions are both harmful and untrue," has made an important contribution to the defence of the cosmological argument. At least, this is what I shall argue.
I shall begin by presenting what I take to be the strongest version of the argument, one which avoids those objections most commonly associated with the cosmological argument. I shall then consider a new, but potentially more damaging, objection to this version of the argument. Finally, I shall consider various strategies for replying to that objection, drawing on Russell's account of facts to provide materials for the strongest reply.
There have been many versions of the cosmological argument, but in my view the one presented by Samuel Clarke and elaborated by William Rowe is the most powerful deductive version, as well as the most elegant.(2) Presenting this version requires only one distinction, one principle, and two premises.
The distinction is that between a "dependent being," a being whose existence is explained by the causal activity of other things, and a "self-existent being," a being whose existence is explained by itself, i.e., by its own nature. (As Clarke puts it, to be a self-existent being is ". . . to Exist by an Absolute Necessity originally in the Nature of the Thing itself."(3)) It would beg the question to presuppose that there are any self-existent beings, but the distinction can be drawn even without this presupposition. Yet it is one thing to draw a distinction and another to understand it, and one may wonder what it means for the existence (or, indeed, any aspect) of a thing to be explained by that thing's nature. Here, an example of Rowe's may help:
Suppose on a cold night we come upon a
blazing campfire. We note that a rock a few
feet from the campfire is warm. If we ask
what explains this fact about the rock (that
it is warm), it would he absurd to suggest
that the explanation is to be found within the
rock itself, that there is something about the
nature of the rock that makes it warm. The
fire and the nearness of the rock to the fire
explain the rock's warmth. Suppose we then
note that the fire is also warm. What accounts
for the fact that the fire is warm? Here it does
not seem absurd to suggest that the explanation
lies within the fire itself. It is the nature
of a fire to be warm just as it is the nature of
a triangle to have three angles.(4)
By analogy, it is the nature of a self-existent being to exist, just as it is the nature of fire to be warm, or of a triangle to have three angles.
The principle is the Principle of Sufficient Reason (hereafter PSR), which can be stated in two parts:(5)
PSR1. For every being that exists or ever existed, there is an explanation of the existence of that being.
PSR2. For every positive fact, there is an explanation of that fact.(6)
PSR employs an intuitive distinction between "facts" and "beings": I am a being, I exist; and there are certain facts about me, such as the fact that I was born in Massachusetts. I, the being, and this fact about me, belong to different ontological categories, but PSR requires that both have explanations. It does not require that anyone know, or even could know, these explanations; it requires only that there be such an explanation. PSR also employs a notion of explanation which is itself unexplained, but we may assume that Clarke (who uses the word "reason") takes it in the general and intuitive sense of "a correct answer to a `why?' question."
With these preliminaries in place, the argument can now be stated:(7)
C1. Every being (that exists or ever did exist) is either a dependent being or a self-existent being.
C2. Not every being can be a dependent being.
C3. Therefore, there exists a self-existent being.
Premise (C1) is a corollary of PSR1, since it states that every being's existence has an explanation, either in its own nature or in the causal activity of other things. Premise (C2) is supposed to be a corollary of PSR2, since if every being were a dependent being, at least one positive fact would remain unexplained: namely, the fact that there are any dependent beings at all, rather than nothing (hereafter, "the cosmic fact").(8) For, on the hypothesis that there has only ever been:
an Infinite Succession of changeable and
dependent Beings, produced one from
another in an endless Progression without any
Original Cause at all, . . . [it would be] . . .
equally possible that from Eternity there
should never have Existed any Things at all;
as that there should from Eternity have
Existed a Succession of changeable and
dependent Beings. Which being supposed;
then, What is it that has from Eternity
determined such a Succession of Beings to Exist,
rather than that from Eternity there should
never have Existed any Thing at all?(9)
This is the cosmological argument at its leanest and best; certainly, it compares favourably with any of St. Thomas's Five Ways. It has only two premises, compared with the eighteen in, e.g., Thomas's second way.(10) It requires neither the notoriously sticky disproof of the possibility of an actually infinite series of dependent beings, nor the truth of Aristotelian physics, nor any distinction between "essentially (per se) ordered" causes and "accidentally ordered" causes.(11) And its conclusion describes a being--a self-existent, necessary being that is the explanation of all the dependent beings--that is closer to the God of traditional theism than an unmoved first mover or an uncaused first efficient cause.(12) But this is not to say that Clarke's argument has no problems of its own.
The conclusion of Clarke's argument has itself been criticised for not yielding a being with the singularity, power, knowledge, goodness, or personality that theists ascribe to God. And the premises of Clarke's argument have been criticised, mainly because of their reliance on PSR, which has been criticised as unjustified, or even demonstrably false.(13) These criticisms need to be taken seriously, but I will not consider them in this paper. For one thing, one can mount passable replies to all these points. (The criticism of the Clarke's conclusion ignores the second part of his argument [in propositions IV-XII], which purports to show that the self-existent being must also be the one and only God. Even if direct justifications of PSR prove elusive, indirect justifications may be possible, as I have tried to show elsewhere. And even if this version of PSR does turn out to be false, there may well be weaker versions which will suffice.)(14) For another, the criticism I will consider would still arise even if PSR were true and the second part of Clarke's overall argument were sound.
A NEW OBJECTION?
This new objection can be approached by imagining a discussion of Clarke's argument in an undergraduate seminar on philosophy of religion. You, the teacher, explain the operative concepts of "dependent being" and "self-existent being," you introduce PSR1 and PSR2, and you present the argument as sketched above, praising it for the reasons already cited. Your students nod appreciatively, and after a brief discussion of the evidence for and against PSR, you prepare to move on to your next theistic argument. Things are going smoothly, when Jeremy, one of your smarter students, asks, "Why can't there be self-explanatory facts? Why can't the fact there are some dependent beings at all, rather than nothing, be a self-explanatory fact?" You try to decide whether he's a budding reformed epistemologist or a budding natural atheologian, but he isn't finished: "And if there could be self-explanatory facts, why bother with a self-existent being?" Good, old, Jeremy ...
The point behind Jeremy's questions is not hard to grasp: If Clarke allows that there could be a self-existent being, why would he not also allow that there could be self-explanatory facts? And if he allows that there could be self-explanatory facts, then the supporting argument for premise (C2), "Not every being can be a dependent being," is undermined. Clarke's argument for (C2) states that not every being can be a dependent being, because if every being were a dependent being, we would lack an explanation for at least one positive fact, the cosmic fact that there are any dependent beings at all, rather than nothing.(15) But if there are self-explanatory facts, then there may be an explanation of the cosmic fact, even if dependent beings are the only things ever to have existed.
Now, how is the defender of Clarke's argument to reply to this objection? Several possibilities suggest themselves:
A. Facts Do Not Exist?
First, one could try to nip the objection in the bud by denying that there are such things as "facts," much less "self-explanatory" ones. That is, we could adopt an extreme form of nominalism, which countenances only concrete particulars, and holds that talk about facts is, if not an illicit appeal to abstract entities on a par with sets, propositions, or states of affairs, merely a misleading way of referring to things.(16)
Such a move would be a deferential nod to the minimalistic wisdom of this age, but it faces several problems. First, as neither Clarke nor Rowe ever state clearly what they mean by "fact," it is unfair to lumber them with an account according to which facts are a dubious sort of abstract object. Other accounts are possible, according to which facts are, say, concrete (complex) particulars. At this stage of the argument, we need not commit ourselves to any particular account of facts; we can hold facts to be whatever it is that makes true sentences (or utterances) true; in the simplest case, the fact that makes "S is P" true is merely S's being P. In this limited sense, we can affirm that facts "exist," without committing ourselves to any particular account of them, whereas denying that facts (in this limited sense) exist goes far beyond simple nominalism. More importantly, rejecting facts will not do as a defence of this cosmological argument, for the simple reason this version of the argument requires facts. (If there are no facts, then only beings, and not facts, need to be explained. And if only the existence of dependent beings, either individually or as a series, needs to be explained, then an explanation in terms of "an Infinite Succession of changeable and dependent Beings, produced one from another in an endless Progression" would, as Hume argued, explain everything that needed to be explained.(17))
This first objection to self-explanatory facts is inadequate, but before considering a second objection, we would do well to collect our findings on facts so far, since all of the other objections will centre on the notion of facts as well. What has emerged from our discussion so far is this: facts as they figure in Clarke's argument are best understood as truth-makers, i.e., more similar to states of affairs than to propositions. And what needs to be explained about them is their obtaining rather than their existence. (For, on the assumptions that facts were similar to states of affairs, and that states of affairs could be actual or not, what PSR would demand is not an explanation of the existence of some actual or non-actual state of affairs--which may after all be a necessary being--but the obtaining of that state of affairs.(18)
B. Facts Cannot Explain Anything?
A second strategy for replying to Jeremy's question would be to concede that facts may "exist"--in whatever attenuated sense of the word the nominalist likes--but to insist that there can be no self-explanatory facts, on the grounds that facts cannot explain anything, let alone themselves. And we can imagine someone taking this view after reflecting on what it is to be a fact. As we have suggested above, one might conceive of a fact as similar to a "state of affairs," which is roughly "a way the world can be." But this suggests that facts are (in seventeenth century terminology) "modes," rather than substances. As modes are "ontologically secondary" to the substances that they modify, we might infer further that they are "causally secondary" as well. After all, when my chicken loses its head, it is because I chopped it off, not because of the fact that I chopped it off. I, the substance, am the agent here; the fact is a causal/explanatory epiphenomenon; to suppose otherwise is to commit what philosophers quaintly used to call a "category mistake."
I concede that it is tempting to understand facts as being a sort of "mode"; and there is textual evidence that Clarke accepts this characterisation of the substance/ mode distinction.(19) I also concede that it would sound odd to say, as in our chicken-chopping example, that a fact could be an agent. Nevertheless, this line of argument is mistaken. The chicken-chopping example is misleading, because it confuses being an explanation with being an agent or being a cause when it is only explanation that concerns us here. And the general claim that "facts cannot explain anything" is demonstrably false, as it is not hard to show that some facts do explain. Consider the creaky, old "covering-law" theory of explanation. According to that theory:
... the occurrence of an event is explained
when it is subsumed under or covered by a
law of nature .... [and] the logical form of
an explanation can be exhibited as follows:
[C.sub.1], ..., [C.sub.n]
[L.sub.1], ..., [L.sub.m] / E
The explanans consists of two sets of premises:
(1) a set of singular statements, [C.sub.1],
..., [C.sub.n], describing relevant initial conditions
and (2) a set of general laws, [L.sub.1], ..., [L.sub.m]
The explanandum ..., E, which describes
the phenomenon to be explained, is logically
deduced from the explanans.(20)
With this model in mind, suppose we conjoin the two sets of premises from the explanans, "C.sub.1], ..., [C.sub.n]" and "[L.sub.1], ..., [L.sub.m]." In that case, we could reasonably presume that whatever it is that makes the conjunction "[C.sub.1], ..., [C.sub.n] & [L.sub.1], ..., [L.sub.m]" true is a fact, and it is precisely this fact that explains E. For example, it is the fact that the paint contained lead carbonate; and sulfur was contained in the gas used to light the room; and sulfur and lead carbonate combine to form lead sulfide, which is black, that explains why the painted white walls of a room turned black.
Indeed, these last two points together suggest (though they do not prove) not only that facts can explain, but that only facts can explain, and that things, i.e., substances or particulars considered in isolation, cannot. When asked for an explanation of why Nelson's chicken lost its head, we answer that Nelson chopped it off with an hatchet. We do not just recite the names of a bunch of particulars: "Nelson. Hatchet. Chicken."(21) It is worth nothing that, if all explanation via appeal to things (i.e., substances or particulars) gives way to explanation via appeal to facts, then, in the context of Clarke's argument, explanations of the existence of beings via appeal to the causal activity of things would give way to explanation via appeal to facts about those things, and PSR2 would subsume PSR1.(22)
C. Facts Cannot Explain Themselves?
A third strategy for replying to Jeremy's question would be to concede that facts exist and that facts can explain, but to deny that facts can explain themselves. The justification for this would seem to be straightforward: facts cannot explain themselves for the simple reason that nothing can explain itself. Indeed, explanations of the form
P The explanans
P The explanandum
are so inadequate that we hesitate to call them explanations at all.
It may be objected straightaway that this argument trades on a false comparison, as the explanans and explanandum in the above explanation are truth-bearers, while facts are truth-makers, as stated in section A. But the main insight of this argument can easily be recast: in an adequate explanation, the truth-maker(s) of the explanans should not be identical with the truth-maker of the explanandum. Yet even this recast argument will not do as a defence of Clarke, for if nothing were self-explaining, Clarke could not claim--as he does--that God was in some sense self-explaining. Of course, describing Clarke's God as self-explaining is somewhat misleading: a self-existent being is not simply a thing that explains itself. As Rowe's example made clear, it is a thing whose nature explains its existence, making the "explainer" and the "explained" at least notionally distinct. And if self-explanatory facts are to be understood as parallel to self-existent beings, it is not the cosmic fact that explains itself. Instead, it must be the fact that the cosmic fact has a certain nature that explains the cosmic fact that there are any dependent beings at all rather than nothing. And it would not be absurd to think that there could be self-explanatory facts in this sense. For example, if asked for an explanation of the fact that T, a triangle, has three sides we may intelligibly reply that it is a self-explanatory fact, the nature of the fact that T has three sides explains the obtaining of the fact that T has three sides.
D. The Cosmic Fact Cannot Explain Itself?
This suggests a fourth, more powerful, reply to Jeremy's question. We can allow that facts exist, that facts can explain, and that there could, in principle, be self-explanatory facts, at least in the sense sketched above; but this is not enough to undermine Clarke's argument. It would undermine Clarke's argument only if the cosmic fact were self-explanatory. But is there any reason to suppose that it could not be?
First, one might suppose that the nature of the cosmic fact could not explain the obtaining of the cosmic fact, on the grounds that facts do not have natures at all. To hold that they do have natures would again be a sort of category mistake: facts do not have natures; they are natures, in the already mentioned sense that a fact is a way the world can be. But this reply will not work for at least two reasons. First it is not obvious that facts really are natures, except in some very loose sense. We have not yet characterised facts as anything more than truth-makers, suggesting only that they might be understood as resembling one sort of truth-maker, i.e., states of affairs; but even so, states of affairs are not ordinarily thought of as natures. Second, even if facts were natures, it would not follow that they could not have natures. We are already familiar with properties having properties ("redness is exemplifiable"); relations having relations ("material implication is weaker than strict implication"); and propositions about propositions ("I once had a teacher who denied Leibniz's law"), so why should we not add to the list including second-order properties, relations, and propositions, second-order natures?
We may concede for the sake of argument, then, that the cosmic fact has a nature, but we may still wonder whether it has the right sort of nature, the sort of nature which would explain why the cosmic fact obtains. Why might one doubt that the cosmic fact, the fact that there are any dependent beings at all rather than nothing, has the right sort of nature? This requires an account of facts which characterises them more specifically than merely as "truth-makers." It is here that Bertrand Russell becomes useful to the proponent of the cosmological argument, as the account of facts he offers in "The Philosophy of Logical Atomism" is among the most detailed and sophisticated available.(23)
E. Detour: Russellian Facts
Whole books could be written about Russell's account of facts, but for our purposes the outlines of his account can be sketched as follows;
1. Facts are truth-makers.
2. Facts are objective, i.e., mind-independent.
3. There are general facts as well as particular facts.
4. There are negative facts as well as positive facts.
5. Facts can be completely characterised individually.
6. Facts are complexes, i.e., have constituents and an internal structure.
7. A particular may be a constituent of a fact, but a fact is not identical with any of its constituents.
8. The internal structure of a fact is isomorphic with the structure of a proposition.
9. Typically, the internal structure of a fact consists in an object exemplifying a property or objects standing in relation.
10. The constituents of facts must be known via direct acquaintance.
Of course, detail may be purchased at the expense of plausibility, and we may wonder what there is to recommend Russell's characterisation of facts. I will briefly note several things: First, Russell's account is by no means idiosyncratic, and has served as one of the main inspirations for subsequent treatments of facts by mainstream analytic metaphysicians. Herbert Hochberg, for instance, echoes Russell when he characterises facts as "complex entities . . . consisting of particulars in an arrangement, of entities exemplifying properties or standing in relation."(24) Second, we can retain the useful elements of Russell's account of facts without thereby committing ourselves to some of its dubious elements, such as its atomistic semantics or phenomenalist epistemology. Third, we have good reason to countenance facts characterised thus in our ontology, as they cannot be reduced without remainder to the concrete particulars that figure in them. As Hochberg argues, "The fact that something is yellow is not a mere collection of the object and the colour, or of the object, the colour, and the exemplification relation. It must be construed as the object's having or exemplifying the colour."(25) In a similar vein, C. A. Baylis argues further that:
We have the same kind of evidence in experience
for facts that we do for particulars,
namely direct awareness of them or of certain
aspects of them.
To take a concrete example, let us suppose
that . . . we are aware of two sense-data, a
large blue patch and a small red patch. Their
discrepancy in size is such that we know with
certainty in this case that the blue patch is
larger than the red one. Our ground for this
knowledge seems to be that we observe directly
that it is larger. Now what we are aware
of here is not merely a string of particulars
but a fact, a relation holding between particulars.
We are not aware merely of two
patches and conjunctively adding to them a
relation. We are aware of a relation of a certain
kind, larger than, holding between the
two patches. This relation has a sense or direction
and it has terms. The complex whole
of the relation holding between its terms is a
fact, a particular fact, if you like, but not a
particular in the sense of a substance, like
either of the patches.(26)
F. Again: The Cosmic Fact Cannot Explain Itself?
If Russell's (and Hochberg's) characterisation of facts is correct, the nature of facts in general is "to consist of particulars in an arrangement, or of entities exemplifying properties or standing in relation." The nature of a particular positive fact would consist, we might suppose, in those particular entities being arranged in that particular way, or exemplifying that particular property. That is, the nature of a particular positive fact would at the very least depend on the nature of its constituent entities and the relations or properties they exemplify. Let us call this the "Russellian Insight." In the light of this insight, the question now becomes, "Does the cosmic fact have the right sort of constituent entities, properties, or relations to explain why it obtains?". But what would it mean for a fact's having certain constituent entities, properties, or relations to explain why that fact obtains? It is not clear, but we have a clue about what Clarke would have thought it meant.
As we have already seen, Clarke argues against the hypothesis that every being could be a dependent being on the grounds that, if all beings were part of "One entire Endless Series of Dependent Beings":
`tis plain this whole Series of Beings can
have no Cause from without, of its Existence;
because in it are supposed to be all Things
that are or ever were in the Universe: And
`tis plain it can have no Reason within itself,
of its Existence; because no one Being
in this Infinite Succession is supposed to be
Self-Existent or Necessary, (which is the only
Ground or Reason of Existence of any Thing,
that can be imagined within the Thing itself
Here Clarke implies that self-existent beings are necessary, as necessity is the only possible "internal" ground or reason for the existence of a being.(28) That is, a being's nature explains that being's existence only if the nature renders the existence necessary. While Clarke makes this point in terms of the existence of beings, by parity of reasoning we may extend it to the obtaining of facts as well: the nature of a fact would explain the obtaining of the fact only if the nature of the fact rendered the obtaining of the fact necessary. Perhaps the question which concerns us in this section should therefore be recast as: "Does the cosmic fact have constituent entities, properties, or relations such that it would obtain necessarily?" The defender of the cosmological argument might have several reasons to think it does not.
First, what are the constituents of the cosmic fact? The constituent entities corresponding to the subject would appear to be "dependent beings." But what is the property that corresponds to the predicate? For now, let us take the suggested isomorphism at face value, and suppose that it is "existence." If we grant that existence is a property or a relation, we might be tempted to deny that the constituents of the cosmic fact could render the obtaining of the cosmic fact necessary. If the constituents of the cosmic fact are "dependent beings" and "existence," and if neither one of them is necessary, then the fact which they constitute cannot be necessary, either. Clarke never makes such a claim about the obtaining of facts, but it resonates with his claim about the existence of beings, that, ". . . absolute necessity of existence, not being an outward, relative, and accidental determination . . . [is] an inward and essential property of the nature of the thing that so exists."(29) Unfortunately, this argument is not obviously adequate, as it may trade on the fallacy of composition, as a whole need not, indeed cannot, have all of the properties of all of its parts.(30)
Moreover, Russell would certainly reject this characterisation of the nature and structure of the fact that there are any dependent beings at all. To see this, we need only consider the sentence which this fact is supposed to make true. It can't be the sentence, "There are any dependent beings at all," for, as Zeno Vendler observes, the word "any" cannot be used in simple indicative sentences in the past or present tense.(31) We must reformulate the sentence in some other way, such as "There are some dependent beings." But formulated thus, it does not express an ordinary proposition. According to Russell, it expresses a "propositional function," which is simply an "expression containing an undetermined constituent [i.e., a variable], or several undetermined constituents, and becoming a proposition as soon as the undetermined constituents are determined."(32) According to Russell, the logical structure of this propositional function is nothing more than, "There exists an x, such that x is a dependent being," which is either necessary (true for all values of x), possible (true for at least one value of x), or impossible (true for no values of x). But having clarified it thus, we can now see that its logical subject is not "dependent beings" at all! In the fact which makes this sentence true, dependent beings cannot be the constituent objects, and existence cannot be the constituent property, so any account of the nature of the cosmic fact which treats these as isolable constituents cannot be correct.(33) But, then, how are we to understand the nature of the cosmic fact?
Let us ask once more: what is the truth-maker for the sentence, `1here exists an x, such that x is a dependent being"? According to Russell, this sentence is possible just in case there is at least one value for x which makes it true. That is, there must be at least one true sentence like "Nelson is a dependent being." Perhaps, then, we can understand the nature of cosmic fact if we can understand the nature of an individual instantiation of that fact. Now, what is the truth-maker for this instantiation? Two possibilities suggest themselves: a) Nelson; and b) Nelson's being a dependent being. Let us consider each, asking what its nature is, and whether its nature could explain its obtaining. Is Nelson himself the truth-maker of the sentence "Nelson is a dependent being"? It is tempting to suppose that he is, but on this point Russell is emphatic:
... [W]hen I speak of a fact I do not mean a
particular existing thing, such as Socrates or
the rain or the sun. Socrates himself does not
render any statement true or false. You might
be inclined to suppose that all by himself he
would give truth to the statement `Socrates
existed," but as a matter of fact that is a mistake
.... Socrates himself, or any particular
thing just by itself, does not make any proposition
true or false. "Socrates is dead" and
"Socrates is alive" are both of them statements
about Socrates. One is true and the
other is false. What I call a fact is the sort of
thing that is expressed by a whole sentence,
not by a single name like Socrates.(34)
The second possibility is that Nelson's being a dependent being is the truth-maker of "Nelson is a dependent being," and apart from its apparent vacuity, it is a plausible suggestion. What is the nature of this fact? It is not clear, but whatever it is, according to the Russellian Insight, it depends on the nature of the entity which figures in the fact. In this case, the entity is Nelson, a dependent being, so the defender of the Clarke's argument must argue that Nelson's dependent nature prevents that fact from being self-explanatory. One strategy would be to argue, as above, that the nature of a fact about dependent beings is somehow insufficient to account for its necessity. Another strategy would be to argue that, given the Russellian Insight, the nature of a fact about dependent beings is otherwise incompatible with its being self-explanatory. The first strategy is always in danger of committing the fallacy of composition (as I explained above), so my final reply to Jeremy's question will follow the second strategy.
G. The Final Reply
S1. If there are some dependent beings, then their existence is explained by the causal activity of some beings or other, [B.sub.1] ... [B.sub.n].
S2. If the existence of these dependent beings is explained by the causal activity of [B.sub.1] ... [B.sub.n], then the fact that some dependent beings exist is explained by the fact that [B.sub.1] ... [B.sub.n] were causally active.
S3. If the fact that some dependent beings exist is self-explanatory, then the fact that some dependent beings exist is explained by the nature of that fact.
S4. But it is not the case that the fact that some dependent beings exist is (sufficiently) explained by two distinct facts.
S5. The fact that [B.sub.1] ... [B.sub.n]. was causally active is distinct from the fact that that some dependent beings exist is self-explanatory.
S6. Therefore, that some dependent beings exist is not a self-explanatory fact.
COMMENTARY AND CONCLUSION
(S1) is entailed by the definition of "dependent being." (S2) is strongly suggested --I do not say entailed--by the conjunction of (S1) and the Russellian Insight. (S3) is entailed by the definition of "self-explanatory fact." (S4) denies "explanatory over-determination" regarding the existence of beings. I cannot now argue for this point, so I will merely report that it seems obvious to me. In case it is not obvious to anyone else, here is another illustration: if the fact that baby Lydia exists is sufficiently and correctly explained by the fact that Pauline and Mark were causally active in such-and-such a way, then it is it is not the case that that fact is also sufficiently and correctly explained by the fact that Camilla and Charles were causally active in such-and-such a way--or by any other distinct fact.(35) (S5) is implied by the apparent mutual exclusiveness of dependent facts and self-explanatory facts. (S6) follows from (S2) through (S5).
Having taken from Russell the insight that the nature of a fact will depend on the entities which figure in the fact, we must note that Russell would have objected to my final reply on one small point, viz., that its wording appears to treat existence as a predicate of individuals in the world. And on this point, Russell is equally emphatic:
It is of propositional functions that you can
assert or deny existence .... If I say "The
things that there are in the world exist," that
is a perfectly correct statement, because I am
saying something about a certain class of
things I say it in the same sense in which I
say "Men exist." But I must not go on to
"This is a thing in the world, and therefore
this exists." It is there the fallacy comes in,
and it is simply, as you see, a fallacy of transferring
to the individual that satisfies a
propositional function a predicate which only
applies to a propositional function.(36)
We may wonder at Russell's quickness to abandon linguistic intuitions to a theory of philosophical logic, but we must not quarrel with him here. It would be discourteous to quibble with a co-laborer in the service of the cosmological argument; far better to follow the courteous Dr. Clarke, who advises:
I think it is not the best way for anyone to
recommend his own performance, by endeavouring
to discover the Imperfections of
Others who are engaged in the same design
with himself .... But every man ought to use
such arguments only, as appear to him to be
clear and strong; and the reader must judge
whether they truly prove the conclusion.(37)
Received May 6, 1997
APPENDIX: ST. THOMAS'S SECOND WAY
(formulated in sentence logic)
1. There exist sensible things which have efficient causes.
2. The cause of a thing is always prior to that thing.
3. Nothing is prior to itself.
4. If (2) and (3), then nothing is the cause of itself.
5. Nothing is the cause of itself. [(2)-(4), add., m.p.]
6. If there are sensible things which have causes, and nothing is the cause of itself, then there are sensible things which are caused by another.
7. There are sensible things which are caused by another. [(1), (5), (6), add., m.p.]
8. If there are sensible things which are caused by another, then either there is an infinite regress of caused efficient causes, or there is an uncaused first efficient cause.
9. Either there is an infinite regress of caused efficient causes, or there is an uncaused first efficient cause. [(7), (8), m p.]
10. If there is an infinite regress of caused efficient causes, then there is no first efficient cause.
11. If there is no first efficient cause, then there is no intermediate cause.
12. If there is no intermediate cause, then there is no ultimate cause.
13. If there is no ultimate cause, then there is no effect.
14. If there are no effects, then there are no sensible things (which have efficient causes).
15. There is a first efficient cause [(1), (10)-(14), h.s., m t.]
16. There is no infinite regress of caused efficient causes. [(10), (15), m.t.]
17. There is an uncaused first efficient cause. [(9), (16), d.e.]
18. "God" means "uncaused, first efficient cause."
19. God exists. [(17), (18), def.]
Earlier versions of this paper were presented at the University of Leeds and King's College, London. I thank Roger Crisp, John Divers, Steven French, Robin Le Poidevin, Anna Maidens, Peter Millican, William Rowe, Scott Shalkowski, and Peter Simons, as well as my audience at King's, for helpful discussion of earlier drafts of this paper, and Prof. J. Bobik for comments on my formulation of Thomas's "second way" in the Appendix.
(1.) See The Autobiography of Bertrand Russell, Vol. I (London: George Allen and Unwin, 1967), p. 41. See also extracts from his diary, recorded on p. 48.
(2.) Samuel Clarke, A Demonstration of the Being and Attributes of Cod, delivered as his Boyle Lectures in 1704, and later published as the first part of A Discourse Concerning the Being and Attributes of God, The Obligations of Natural Religion, and the Truth and Certainty of the Christian Revelation. References here are to the tenth edition, printed in London for H. Woodfall, J. Beecroft, et al., 1766. William Rowe treats Clarke's argument in The Cosmological Argument (Princeton, NJ: Princeton University Press, 1975), esp. pp. 60-269, and in Philosophy of Religion (Belmont, CA: Wadsworth, 1977), pp. 16-30.
(3.) Clarke, A Discourse Concerning the Being and Attributes of Cod, The Obligations of Natural Religion, and the Truth and Certainty of the Christian Revelation, p. 15.
(4.) Rowe, Philosophy of Religion, pp. 10-11.
(5.) This formulation of PSR is due to Rowe. See his Philosophy of Religion, pp. 18-19.
(6.) A "positive fact" is a fact the obtaining of which entails the existence of at least one contingent being. For a similar formulation, using slightly different wording, see Rowe, The Cosmological Argument, p. 103.
(7.) See Clarke, pp. 11-14, and Rowe, pp. 18-19.
(8.) The argument for (C2) might be formulated as follows:
D1. If every being could be a dependent being, then the explanation for the fact that any dependent beings exist at all (rather than nothing) could be only in terms of the existence of either the totality of dependent beings or some subset of that totality.
D2. Explanation is not possible, in terms of the existence of either the totality of dependent beings or of some subset of that totality, for why any dependent beings exist at all (rather than nothing).
D3. There is an explanation of every positive fact.
D4. That any dependent beings exist at all (rather than nothing) is a positive fact.
D5. Therefore, not every being can be a dependent being.
Also, it is worth nothing that, at times, Clarke seems to treat (C2) as a corollary of PSR1, as he occasionally talks as if the issue is the explanation of a certain being, namely the Series of dependent beings. See Clarke, p. 12.
(9.) Ibid., pp. 13-14 (italics Clarke's). For brief commentary, see Rowe, pp. 25-27. For more detailed commentary, see Rowe, The Cosmological Argument, pp. 115-167.
(10.) For a complete formulation of Thomas's second way, see the appendix.
(11.) For a discussion of this distinction, see Rowe, pp. 23-32.
(12.) It must be acknowledged, however, that Thomas's Five Ways taken together, with the funkier argument in ST Ia, Q's 2-11, may yield a being that is much closer to the God of traditional theism than do any of the Five Ways taken individually.
(13.) For a criticism of the rational support for PSR, see "The Existence of God, A Debate between Bertrand Russell and Fr. F. C. Copleston," broadcast by the BBC in 1948, reprinted in Bertrand Russell, Why I Am Not A Christian (London: George Allen and Unwin, 1957), and Rowe, Philosophy of Religion, pp. 27-29. For criticisms of PSR itself, especially PSR2, see Rowe, The Cosmological Argument, pp. 97-114, and James Ross, Philosophical Theology (Indianapolis, IN: Bobbs-Merrill, 1969), pp. 259-304.
(14.) I have sketched an indirect justification for PSR in "The Principle of Sufficient Reason: A Moral Argument," Religious Studies, vol. 32, pp. 15-26, and Rowe suggests a weaker principle which appears to survive his criticism of PSR2:
PSR4. For every set whose members are existing beings (which
can be caused to exist or which can cause the existence of
other beings) there must be an explanation of the fact that
it has members.
(15.) For a complete formulation of Clarke's argument for (C2), see footnote 8. In terms of that formulation, Jeremy's question can be seen as an attack on premise (D1).
(16.) For a lucid account of the several uses of the term "nominalism," see Peter M. Simons, "Nominalism in Poland," Poznan Studies in the Philosophy of the Sciences and the Humanities, vol. 28, 1993, pp. 207-231, esp. pp. 207-209.
(17.) Hume's Cleanthes remarks, "Did I show you the particular causes of each individual in a collection of twenty particles of matter, I should think it very unreasonable should you afterwards ask me what was the cause of the whole twenty. This is sufficiently explained in explaining the cause of the pans." See Hume, pp. 190-191. It is also worth noting that thoroughgoing nominalism is not a natural ally to theism, unless God is conceived of as a concrete particular.
(18.) John Divers has helped me to understand these points more clearly. (19.) In a discussion of whether there can be "infinity" and "eternity" in the universe without some being which is infinite or eternal, Clarke denies that there can be, because "infinity" and "eternity" are modes and " . . . Modes and Attributes exist only by the Existence of the Substance to which they belong." Clarke, p. 15.
(20.) J. Kim, "Explanation in Science," in P. Edwards, ed., The Encyclopedia of Philosophy (New York: Macmillan, 1967), Vol. 3, pp. 159-163, p. 159. The subsequent example about the paint is Kim's, too. The covering law model of explanation has generally been rejected in recent years, often in favour of "semantic" conceptions of explanation, but even on most semantic conceptions of explanation, the sort of example Kim cites would be recognised as a successful explanation.
(21.) This is not a unique feature of explanations; it is entailed by a feature of facts. See the passage by Russell cited in footnote 34. Also, H. D. Mellor has recently argued that facts, not events, play a primary role in causation. See his "The Singularly Affecting Facts of Causation," in Matters of Metaphysics (Cambridge: Cambridge University Press, 1991).
(22.) William Rowe has confirmed this in private correspondence.
(23.) See Bertrand Russell, "The Philosophy of Logical Atomism," The Monist, 1918, reprinted in The Collected Papers of Bertrand Russell, Vol. 8, John G. Slater, ed., The Philosophy of Logical Atomism and Other Essays: 1914-1919 (London: George Allen and Unwin, 1986), pp. 163-164.
(24.) Herbert Hochberg, "Fact," in J. Kim and E. Sosa, eds., A Companion to Metaphysics (Oxford: Basil Blackwell, 1995), pp. 164-165. See also, C. A. Baylis, "Facts, Propositions, Exemplification and Truth," Mind, vol. 57 (1948), pp. 459-479.
(25.) Hochberg, pp. 164-165.
(26.) Baylis, p. 461.
(27.) Clarke, p. 12 (italics mine).
(28.) Clarke does not spell out clearly his reason for thinking that a self-existent being must be a necessary being, but Rowe attributes to him the following argument:
N1. There exists an independent being (a being that has the reason for its existence within its own nature).
N2. If a being has the reason for its existence in its own nature that being exists in every possible world.
N3. If a being exists in every possible world, then it is a necessary being.
N4. Therefore. there exists a necessary being.
See Rowe, pp. 201-202.
(29.) Clarke, p. 12.
(30.) For a brief discussion of why facts are not merely sets of entities, properties, and some connection between entities and properties, see Hochberg, p. 165. For an illuminating treatment of the fallacy of composition, see William Rowe, "The Fallacy of Composition," Mind, Vol. LXXI, no. 281,1964.
(31.) See Zeno Vendler, "Any and All," in P. Edwards, ed., The Encyclopedia of Philosophy (New York: Macmillan, 1967), Vol. 1, pp. 131-133.
(32.) Bertrand Russell, p. 202.
(33.) See especially ibid., pp. 211-213.
(34.) lbid., pp. 163-164.
(35.) Importantly, I am not denying that there may be overdetermination of the causation of some events, if the events are described at a sufficiently general level. For example, if two stones strike a pane of glass at separate places, that may causally overdetermine the event of the glass's shattering. But this is not the same as overdetermination of the explanation of the existence of a being.
(36.) Ibid., p. 205.
(37.) Clarke, Preface, p. ii.