05 / 06

Timothy O’Connor on Contingency: A Review essay on Theism and Ultimate Explanation: The Necessary Shape of Contingency (Oxford: Blackwell, 2008)

William Lane Craig

Used by permission of Philosophia Christi 12 (2010): 181-8.


In the first part of Theism and Ultimate Explanation Timothy O’Connor provides a compact survey of the metaphysics and epistemology of modality, defending modal realism and a priorism. In the book’s second half he defends a Leibnizian-style cosmological argument for an absolutely necessary being. He seeks to answer four questions: (1) Is the idea of a necessary being coherent? (2) In what way is the postulation of such a being explanatory? (3) Does the assumption of necessary being commit us to denying the very contingency of mundane things which it is meant to explain? (4) What are the implications of necessary being for theology? In this review I highlight a few of the obscurities and apparent weaknesses of this otherwise commendable book.

This slim but pithy volume is an exercise in natural theology, defending a Leibnizian-style cosmological argument for an absolutely necessary being. O’Connor seeks to answer the fundamental question, “Are there contingently existing objects and, if there are, why do these particular contingent objects exist and undergo the events they do?” The only viable answer, he argues, is there are and that they are the causal product of a purposive, transcendent, necessary being.

The book is divided into two nearly equal parts. The first part “The Explanatory Role of Necessity” lays the modal foundations for the cosmological argument from contingency in the second half. This first part offers a compact survey of the metaphysics and epistemology of modality. O’Connor champions modal realism (not to be confused with David Lewis’ concretism with respect to possible worlds) against an array of reductionists, deflationists, and conventionalists regarding modal notions. O’Connor defends the objective reality of irreducible modal properties principally on the basis of their explanatory fecundity, for example, in explaining why induction is reliable, as well as the self-defeating character of the alternatives.

Somewhat confusing in this part of the book is O’Connor’s critique of fictionalism with respect to possible worlds (pp. 23-26). While Gideon Rosen’s own suggestion that we prefix Lewis’ concretist modal statements with the operator “According to the fiction of possible worlds” is doubtless defective, the fictionalist who takes possible worlds as a mere heuristic device is by no means committed to anti-realism about modality as O’Connor intimates. Indeed, the reader, aware of O’Connor’s rejection of Lewis’ concretism, wonders how O’Connor will reconcile a realist account of possible worlds as abstract objects with divine aseity and necessity, which he is so anxious to safeguard. It is not until the book’s end that one comes to realize that O’Connor himself is an anti-Platonist with respect to possible worlds and other abstracta. He explicitly denies “that there are self-subsisting Platonic entities that are the ground of logical truths. The being of such truths consists, rather, in the intellectual activity of the necessarily unique necessary being (the only being that exits of itself)” (p. 128). Modal truths are grounded in God’s understanding the range of His power. O’Connor thus endorses divine conceptualism as the ultimate basis for modal truths and “as the only plausible way to steer a middle course between the excesses of Platonism and views which try, problematically, to tie the basis for logical truths to the nature of the human mind” (p. 129). Such a view is quite consistent with taking possible worlds to be useful fictions.

Apart from the metaphysics of modality, the other main concern of the book’s first part is whether and how we come to a knowledge of modal truths. O’Connor defends an a priorism which holds that we begin our theorizing with a stock of pre-theoretical modal beliefs which are subsequently refined through a process of reflective equilibrium. Thus, a priorism is quite compatible with fallibilism with respect to our modal beliefs. Still, the reader might wonder what justification there is for thinking any of our pre-theoretical modal beliefs to be reliable. It is, again, not until the book’s end that we learn that, on O’Connor’s view, God has so created us that “human cognitive abilities are disposed to modalize reliably in accordance with modal fact. In this way, there can be a tie between modal fact and human modal judgments that plausibly accords with the proper function of human cognition while preserving the integrity of the purely natural causal processes generating such judgments” (p. 129).

Part two of the book “The Necessary Shape of Contingency” is devoted to answering four hostile questions from those who find the explanation of contingent being in a necessary being problematic: (1) Is the idea of a necessary being even coherent? (2) In what way is the posit of such a being truly explanatory? (3) Does not the assumption of necessary being commit us in the end to denying the very contingency of mundane things it is meant to explain? (4) What are the implications of necessary being for theology?

It is remarkable that in his developing answers, O’Connor nowhere explicitly lists the premises of his argument for a necessary being. So doing would have served to crystallize his wide-ranging discussion and so abetted its assessment.

O’Connor considers it unreasonable to hold that while the question of why the contingent universe exists is perfectly coherent, nevertheless the universe’s existence is a brute, inexplicable fact, for there could have been an equally contingent being that caused it. Neither are adequate answers to be found in the postulate of a beginningless past or an abstract metaphysical principle. Rather the best explanation will be the free intentionality of a personal, necessarily existing agent. O’Connor defends the interesting claim that his argument does not assume anything like the Principle of Sufficient Reason, the truth of which, in O’Connor’s view, would imply the necessity of all truths and so destroy contingency (p. 80). His key move here is to claim that a fully adequate explanation need not be contrastive. In order to achieve some purpose, a necessary being may freely create some contingent order C, so that C is fully explained as a result of that being’s intentions. But there need be no explanation of why C obtains rather than C*. Still, to mollify those who do insist on contrastive explanations, O’Connor offers a reformulated PSR to the effect that there is an explanation for every fact or else an explanation for why that fact can have no explanation. The philosopher who asserts that the universe is inexplicable cannot explain why the universe can have no explanation.

This leads to the question whether the universe itself might not be the necessary being. O’Connor argues against this hypothesis on the grounds that the universe exhibits a complexity which is incompatible with the unity of nature which a necessary being must have. In O’Connor’s view the essential properties of a necessary being must form a tight unity:

They must be so bound up with one another that any one could not exist in the absence of the others . . . . A fully adequate grasp of such a being’s nature would involve a recognition of the intimate interdependency or interconnectedness of these attributes in broadly the way we recognize the internal connection between triangularity and trilaterality in two-dimensional Euclidean geometry . . . .” (p. 132).

I found the argument of this section very strange and among the least perspicuous of the book. It needs to be kept in mind that when O’Connor speaks of a necessary being, he means a being which exists not only necessarily but a se (p. 155, n. 7), so that beings that are necessary ab alio do not enter the picture. He asks, “Is the property of necessary existence something that results from the rest of its essential nature, or from some part of its nature?” (p. 87). It’s hard to know even what to make of this question. I suppose it might mean, “Is there some other property which is explanatorily prior to necessary existence?” Perhaps not; but one could imagine a plausible Scotist or Anselmian answer that the property of supreme perfection or of maximal greatness is even more fundamental than necessary existence. O’Connor’s response to his question is negative, because “then there being ‘in the first place,’ so to speak, a thing having the ‘base’ set of properties giving rise to necessary existence would itself be a contingent fact, which contradicts the assumption that NB is truly necessary” (p. 87). The grammatical subject of this obscure sentence is “there being a thing” and the predicate is “would be a contingent fact.” So the complaint is that the fact that this thing exists would be a contingent fact. Now the thing seems to be the object having all the properties of a necessary being once necessary existence is abstracted away. In the explanatorily prior moment (“in the first place”), we have the nature without necessary existence, e.g., maximal greatness and all that follows from it apart from necessary existence. Now obviously there is no suggestion that such an object could be instantiated as such, without necessary existence, so that it would be a contingent being; au contraire, necessary existence necessarily follows from its other properties. But O’Connor’s claim is that the fact that there would in that explanatorily prior moment be such an entity is a contingent fact.

But that seems wrong: abstract necessary existence from the nature of a necessary being, and necessarily what is left over is a thing with all the remaining properties. O’Connor adds this explanatory gloss to his sentence: “That is, there would be a problematic explanatory/ontological priority of these base properties relative to the property of necessary existence—the problem being that the existence of the putatively necessary being would be only conditionally necessary on the instantiation of some more basic features” (p. 87). Here the problem of the explanatory priority of other properties to the property of necessary existence is that the necessary being would be only conditionally necessary, which presumably is incompatible with aseity. But that untoward consequence does not at all seem to follow. A being which actually has such base properties has essentially necessary existence and exists independently. Of course, a Platonist could run a familiar argument against aseity based on the dependence of a necessary being on its properties; but such an argument would also apply to necessary existence itself even if it were fundamental. The concrete maximally great being depends on nothing, certainly not on the fictitious object existing explanatorily prior to its exemplifying the property of necessary existence.

O’Connor then asks whether necessary existence (N) could have been conjoined to some other nature and not to E, the conjunction of certain other essential properties of a necessary being. That is, might N be only contingently connected to E? Although O’Connor says, “It seems not” (p. 87), the two questions do not come to the same thing. Consider the properties of being concrete and being personal. These are essential properties of God, the uniquely necessary being. But obviously, there are contingent persons possessing these properties. So these properties are not necessarily connected to necessary existence. Nonetheless, that fact does not imply that necessary existence could have been conjoined to some other nature than E. For necessary existence, in O’Connor’s usage, involves aseity, and nothing but God can exist a se (pace Platonists). So it is impossible for aseity to be an essential property of any being not possessing the divine nature.

Of course, these are theological convictions, and we do not yet know at this stage of the argument what properties E comprises. So who can say what the answers to O’Connor’s questions are? O’Connor thinks to answer the questions without further specification of E by claiming that, given affirmative answers, “there would be no explanation for the fact that N is actually connected to E” (p. 87). But that seems a clear non sequitur. The answer might be that there is some more fundamental property like maximal greatness, from which E and necessary existence each follow. Of course, we do not at this stage know that a necessary being is maximally great. For all we now, abstract mathematical objects might exist necessarily and a se. In that case E might be comprised of invisibility and timelessness; but it would be rash to infer that N is necessarily conjoined with E, or that E’s conjunction with N in the actual world has no explanation (such as the mathematical nature of the objects).

O’Connor proceeds to argue that N entails the nature E. For if there could be a plurality of natures each entailing but not entailed by N, then these natures would be explanatorily prior to N.

If we asked, why do the natures E1 and E2 necessarily exist, and not some others (E20, say)? the answer would be, Because E1 and E2 —apart from N—are the sorts of nature that simply must be (whereas E20 is not). But this (contrary to intention) can only allow for the entities’ existing by a kind of conditional necessity; given that there is an E1, it exists ‘of necessity.’ And this is inconsistent with the claim that the necessity is absolute (p. 88).

This argument seems multiply confused. O’Connor appears to confuse the existence of natures with the existence of those natures’ instantiations. Presumably the natures themselves (which, recall, are just conjunctions of properties) all exist necessarily, even if some of the objects which exemplify or instantiate them exist contingently. The natures themselves do not, ex hypothesi, encode the property of necessary existence, but they do exemplify the property of necessary existence. So when O’Connor concludes that “the entities” exist with conditional necessity, he cannot mean the natures themselves, since it has not been shown that they depend on anything, nor can he justifiably make this claim of the objects exemplifying or instantiating those natures, for in cases in which E entails N, these objects exist necessarily through their own essences. Hence, I do not see how O’Connor has warranted his conclusion that “there must be an internal connection between necessary existence, N, and any other essential features of NB” (p. 90).

O’Connor now applies these considerations to the hypothesis that the universe is a necessary being:

How can N be a basic property of the universe, given that it has enormous mereological complexity? Apart from topological features, its other basic properties are properties of its parts (the elementary particles and fields), not the thing as a whole (p. 90).

I’m puzzled by this argument. By basic properties, O’Connor does not apparently mean essential properties, not only because the universe’s topological properties are contingent, but also because any argument to the effect that E in this case is too thin to be entailed by necessary existence is a non-starter. So O’Connor must mean that the universe has too few properties as a whole to be considered a necessarily existing thing. But such a claim seems to underestimate the properties the universe possess as a whole: a certain size, density, pressure, expansion speed, age, geometry, not to mention metaphysical properties like concreteness, spatio-temporality, and so on. And even if its basic properties were few, why could it not be a necessary being? Why cannot the universe be necessarily existent even though the stuff which constitutes it may take varying configurations over time? Granted, few non-theists have, as O’Connor notes, been willing to embrace this hypothesis, but his case strikes me as weak at this point. He has not shown either that a necessary being must possess a highly unified nature or that the universe’s nature is too thin for the universe to exist necessarily.

Having concluded that there can be only one kind of necessary being, O’Connor proceeds to argue that its nature is Logos rather than Chaos, that is, a personal agent who freely creates a contingent universe. But O’Connor takes divine freedom to be compatible with God’s acting necessarily, and, in interaction with William Rowe’s challenging argument that God must create the best world that He can, O’Connor argues that God not only creates necessarily, but that He creates necessarily an actually infinite number of universes without a finite upper bound to their goodness (ch. 5). Since O’Connor also argues that God is essentially temporal (p. 139), his view has the implication that God necessarily creates from infinity past, so that O’Connor must reject a biblical doctrine of creatio ex nihilo, according to which the actual world includes a state of affairs of God existing alone without a universe (cf. p. 160). It is ironic—and all too characteristically Leibnizian!—that someone bent on defending an argument from contingency to an absolutely necessary being should wind up making the creation of contingent beings necessary!

Ironically, despite these unorthodox concessions, it is far from clear that O’Connor has blunted the force of Rowe’s argument. According to O’Connor, “The natural object of a perfect Creator’s consideration will be any infinitely membered, partially ordered super-universe for which there is no finite upper bound on the organic value of its members (and perhaps intensive value of its member’s [sic] constituent objects), all of which exceed threshold τ” (p. 120). But then the question still presses, why did God not set the threshold higher? No matter where the threshold lies, it will always be the case that God could have done better. In the end, O’Connor seems forced to reject the intuition driving Rowe’s argument, in which case O’Connor’s bloated ontology may appear superfluous.

It seems to me that the orthodox theologian will challenge O’Connor’s fundamental assumption that God cannot have a resistible reason to create. After all, since God is already the locus of infinite value, God + the universe is no more valuable a world than a world comprised of God alone. Why, then, is creation irresistible for God? O’Connor answers, “God’s very being, which is goodness, necessarily diffuses itself. Perfect goodness will naturally communicate itself outwardly; God who is perfect goodness will naturally create . . .” (p. 112). O’Connor then demands what kind of reason could God have for not creating? He notes that some theologians hold that the inter-Trinitarian relationships suffice as an expression of God’s goodness. O’Connor’s response is that it is equally plausible that a natural impetus toward inward expression of divine goodness would be matched by an impetus to outward sharing of goodness with nondivine creatures” (p. 112). This plausibility judgement on O’Connor’s part might seem a frail reed on which to hang so momentous and unorthodox a doctrine as God’s necessary creation of infinite worlds! As to reasons for resisting the impetus toward sharing goodness with created persons, how about this: God knew that a significant proportion of the population of any world involving such creatures would freely reject His overtures and damn themselves forever? Some philosophers have considered such a prospect so horrifying that they argue that God in such a case is morally constrained not to create! Balancing the blessing of eternal salvation of some persons with the tragedy of the eternal damnation of others is a delicate business, and it is surely the better part of epistemic humility to recognize that we are in no position to make confident judgements that God could not have resistible reasons to create or or good reasons to refrain from creating a world of free persons. This possibility seems especially open when one reflects that which counterfactuals of creaturely freedom confront God is a contingent matter varing from world to world.

O’Connor would not be open to my suggestion because he goes on to argue against the possibility of middle knowledge (pp. 135-38). He denies that there are true counterfactuals of creaturely freedom because they would involve brute facts that defy ultimate explanation. Here again we see the Leibnizian proclivity to nullify human freedom in the name of the Principle of Sufficient Reason. For the problem with such brute facts is not that they lack truth-makers—Molinist accounts have been provided for those who espouse truth-maker maximalism—, but that no ultimate, contrastive explanation can be given for a libertarian free choice. Since O’Connor has already rejected the demand for contrastive explanations in cases of free choice, why should that conclusion not hold in the case of counterfactuals of creaturely freedom? In any case, to return to the question of the necessity of creation, God, in the absence of middle knowledge, has no idea at all logically prior to creation whether a world involving great moral goods would ensue as a result of His creating free persons and whether such a world would not also feature the eternal damnation of a significant proportion of created persons. That might well suffice to make the impetus to create resistible!

Although I have expressed criticisms of O’Connor’s argument at certain places, I remain nonetheless enthused at his defense of the argument from contingency and welcome his entry into the lists of natural theology.