This is an impressive book, a truly extraordinary achievement. I can think of no other treatment of theism, whether by theist or non-theist, comparable to it. Its combination of wide-ranging scope and penetrating analysis makes it a unique contribution to philosophical theology. The product of a lifetime of study, Logic and Theism is testimony to Howard Sobel’s remarkable mastery of his subject, for very few contemporary philosophers could have written so comprehensive and incisive a treatment.
The book is divided into five parts: “Divinity,” a short introductory section that is surprisingly interesting on what it means to say that God exists; “Arguments for the Existence of God,” which treats ontological, cosmological, teleological, and miracle-based arguments for God over the space of 370 pages; “On Two Parts of the Common Conception [of God],” dealing with the coherence of omnipotence and omniscience; “Arguments against the Existence of God,” a defense of both evidential and logical versions of the problem of evil; and, finally, “Practical Arguments for and against Theistic Beliefs,” a sympathetic and wide-ranging look at all sorts of Pascalian wagers that a would-be theist might confront.
As the title of the book intimates, Sobel’s strong suit is logic, mathematics, and probability theory, and these are brought heavily to bear in his analysis of the various arguments he considers. Fortunately for the sake of readability much of the heavy-duty work is reserved for the multiple appendices that cluster at the close of nearly every chapter. Another nearly hundred pages of detailed commentary appear in the book’s endnotes. The book is thus marked by tremendous rigor, helping to clarify exactly which premises are at stake in the debate between theists and atheists.
Sobel shows himself to be a bold and independent thinker, unafraid to challenge consensus views. For example, in reaction to the prevailing view that the logical version of the problem of evil is no longer a problem, he snaps, “I do not believe it” (437) and proceeds to press the problem at length. He presents a surprisingly persuasive argument on behalf of the idea that God must create a best of all possible (or, alternatively, of all feasible) worlds. He thinks that it can be proved that no concrete object can be necessarily existent. Moreover, he holds that no being worthy of worship can be necessarily existent. This willingness to defy conventional views adds to the interest of his discussion.
Sad to say, Sobel’s verdict on theism is almost uniformly negative. He judges all the theistic arguments to be unsound and the concept of God as a perfect being to be incoherent, while arguing that the problem of evil succeeds in proving that God as traditionally conceived does not exist. This is not to say that Sobel’s treatment always goes against the theist: he defends the coherence of an omniscient being, providing a masterful defense of God’s knowledge of all true propositions against Patrick Grimm’s anti-omniscience arguments and offering a (somewhat kooky) defense of the coherence of foreknowledge of future contingents; moreover, he argues for the viability of at least some Pascalian wagers in favor of theistic belief. This show of even-handedness bespeaks Sobel’s overall fairness and objectivity in his assessments. I could not help but notice that it is chiefly in responding to Plantinga’s arguments, when one senses that Sobel’s back is to the wall, that his tone becomes somewhat shrill and his punctuation marks multiply themselves (e.g., ?! or !!). It is at such moments that one senses that Sobel really is doing his best, not simply to assess, but to avoid theism.
Although Sobel’s book may be without peer, this is not to say that his critique succeeds. For the company of contemporary theists have offered separate treatments of the respective issues handled in the book which do answer effectively Sobel’s criticisms. That theism thus survives the acid bath of Sobel’s trenchant critique increases one’s confidence in the arguments of natural theology and the rationality of theistic belief.
A disproportionate part of the book is devoted to an examination of various ontological arguments. I found it odd that Sobel discusses Descartes, Spinoza, and Leibniz’s arguments prior to Anselm’s, as though they had not profited from the latter’s mistakes. His claim that an understanding of modern symbolic logic would have prevented their formulations of the ontological argument strikes one as perhaps too easy a criticism, based on an unsympathetic formulation of those arguments. In any case, his critique of modern modal ontological arguments like Plantinga’s is, I think, clearly indecisive. His central argument against the crucial premiss “Possibly, a maximally great being exists” is that the intuitive possibility of a maximally great being ought to carry no weight in our assessment of its metaphysical possibility. For, he points out, a concept need not be prima facie incoherent, as in the case of a married bachelor, in order to be impossible. Thus, we cannot know a priori whether maximal greatness is possibly instantiated. But the illustration Sobel uses to undermine our confidence in our modal intuitions concerning the possibility of a maximally great being—namely, a dragoon, a thing which, if it is possible, is a dragon in whichever world is the actual world—is not at all analogous to the concept of a maximally great being. On the contrary, while a maximally great being seems intuitively possible, the concept of a dragoon, a contingent thing which mysteriously manages to track whichever world is actual so as to exist in the actual world while not existing in all possible worlds, seems intuitively impossible (cf. Sobel’s own misgivings about what could make a possibly true counterfactual of creaturely freedom false in the actual world [457-8]). Sobel’s dragoon illustration therefore does little to undermine the force of our modal intuitions concerning the possibility of maximal greatness.
Moreover, Sobel’s objection, even if successful, would only show that the crucial premiss of these modern modal ontological arguments is not knowable a priori. He fails even to discuss whether we might not have good reasons a posteriori for thinking that maximal greatness is possibly exemplified. Thus prematurely terminated, Sobel’s long and fascinating discussion of the ontological argument proves to be abortive.
In his briefer discussion of the cosmological argument, Sobel raises against the Leibnizian version of the argument the customary objection to the Principle of Sufficient Reason that not every contingent fact can have an explanation. As a result of reading Alexander Pruss’s new book The Principle of Sufficient Reason, I have come to doubt the probativeness of this objection. It assumes, in Sobel’s words, that “If there is a true contingent proposition, then there is a true contingent proposition that entails every true contingent proposition” (219), what Pruss calls the Big Contingent Conjunctive Fact (BCCF). But, as Sobel’s own extensive discussion of Grimm’s objection to omniscience suggests, it is not obviously true that there is such a totality or proposition as the BCCF. If there is, then it is alleged that its explanation cannot be necessary, since anything entailed by a necessary truth is itself necessary. But why must explanations be entailments rather than material implications? Perhaps the explanation of the BCCF is that necessarily, God has weighed the reasons for creating a world and freely chosen the world He desires. If we do insist that explanations be entailments, then why could not the explanation of the BCCF be itself contingent? The assumption here is that no contingent truth can be self-explained. But that is not obvious. William Vallicella has offered the disarming suggestion that the reason the BCCF is true is that all of its conjuncts are true. Given that a conjunction’s truth value is a function of the truth values of its conjuncts, no further explanation is needed. Or perhaps the explanation of the BCCF is the contingent fact that God wills the BCCF. As a contingent fact, that fact will itself be a conjunct of the BCCF, leading to either an explanatory loop or a benign infinite regress.
This debate is, in any case, somewhat academic, since the cosmological argument does not depend for its success on anything so strong as Leibniz’s own version of the PSR. For example, in their discussion of Hartry Field’s anti-Platonist claim that it is an inexplicable contingency whether mathematical objects exist, Wright and Hale (Journal of Philosophy 89 : 111-35), while rejecting the demand for an explanation of something like the BCCF, nevertheless maintain that explicability is the default position and that exceptions to this rule have to be explicable exceptions—some explanation is needed for why no explanation is possible. For example, they claim that if physical existence is at issue, Leibniz’s question, “Why is there something rather than nothing?” is an unanswerable question if a satisfactory explanation of why a physical state of affairs obtains has to advert to a causally prior situation in which it does not obtain, since a physically empty world would not cause anything. As non-theists, they believe that the demand for an explanation of the contingency of physical existence is suitably pre-empted by the restrictive principle that the explanation of the obtaining of a (physical) state of affairs must advert to a causally prior state of affairs in which it does not obtain. Such a principle will be seen by the theist, however, as not at all restrictive, since the explanation of why the physical world exists can and should be provided in terms of a causally prior non-physical state of affairs involving God’s existence and will.
The proponent of the Leibnizian cosmological argument could generate his argument by holding, in conjunction with the above principle, that the obtaining of any physical state of affairs has an explanation. Or he could claim that for any contingently existing thing, there is an explanation why that thing exists. Or again, he could assert that everything that exists has an explanation of its existence, either in the necessity of its own nature or in an external cause. Or, more broadly, he might maintain that in the case of any contingent state of affairs, there is either an explanation for why that state of affairs obtains or else an explanation of why no explanation is needed. All of these are more modest, non-paradoxical, and seemingly plausible versions of the PSR.
Sobel’s response to such more modest versions of PSR is curious. Rather than rebut them, he brushes them aside:
My work on Leibniz’s reasoning has a different objective. Instead of seeking ways in which, by curtailing his Principle of Sufficient Reason, modal collapse can be avoided, and much of his conclusion retained, I seek to bring out problems made by the principles that actually inform cosmological reasoning, such as the Principle of Deductive Reasons . . . and the Principle of Sufficient Reason. . . . I want to advertise problems with the ‘ambition’ of cosmological reasoning, that everything should have a reason and the related idea that contingencies . . . needs to be explained and can be explained by necessities (221-2).
But this sounds for all the world as though Sobel is less interested in finding out whether God exists than in advertising philosophical problems. If more modest and more plausible principles serve to prove God’s existence, then why fuss about overly-ambitious versions of those principles?
It is noteworthy that in response to Robert Koons’s cosmological argument, Sobel chooses to deny that every contingent being has an external cause rather than to surrender his claim that every cause of a contingent being is itself contingent. Why?—because a necessary being would exist no matter what and would thus be irrelevant to the existence of something contingent. This response reverts, in effect, to the advertised problems of Leibniz’s argument based on a more ambitious version of the Principle of Sufficient Reason rather than offers grounds for denying the very plausible principle underlying Koons’s argument. In the end Sobel’s view is that the universe exists contingently and inexplicably and in this sense “has ‘sprung from nowhere’” (225).
That conclusion is hard enough to believe if the universe is eternal, but it becomes unbearable if the universe began to exist and so literally sprang from nowhere. Sobel’s brief response to the kalam cosmological argument, with its central premiss that the universe began to exist, is therefore not without interest. Against philosophical arguments for the finitude of the past Sobel gratuitously assumes a Platonistic view of mathematical existence and asserts that the problems with an actually infinite multitude illustrated by Hilbert’s Hotel serve to bring out merely its physical impossibility, not its logical impossibility. He thereby ignores both his own distinction between logical and real impossibility as well as the contradictions arising from inverse operations like subtraction performed with infinite multitudes. Sobel also reiterates Aquinas’ riposte to the impossibility of traversing an infinite past that from any point in the past there is only a finite and, hence, traversable distance to the present. This response does nothing to explain how the entire infinite past, as opposed to any part of it, could have been traversed. Sobel’s response to the scientific evidence for the beginning of the universe is really quite desperate: he hypothesizes that temporally prior to the universe’s coming into existence there was a time at which nothing existed and after which the universe came into being through a beginningless series of “fasting-starting” causes on the pattern of the series: . . . , 1/8, 1/4, 1/2, 1. (For a critique see my "J. Howard Sobel on the Kalam Cosmological Argument," Canadian Journal of Philosophy 36 : 565-84.)
Sobel’s treatment of the teleological argument is largely a discussion of the objections of David Hume, who emerges in the present book as a sort of philosophical hero of Sobel. Conspicuously absent is any discussion of the natural theology of William Paley, whose argument was formulated with full knowledge of Hume’s objections and fares very well in light of them. It is interesting that on Sobel’s read, Hume himself turns out to have been a proponent of Intelligent Design:
This was partly because he was convinced that, as an explanation of appearances of design, philosophical theism . . . had no serious competitor. He would have preferred a theory that ascribed ‘an inherent principle of order to the world’. . . and has Philo sketch with some enthusiasm such a theory in Part 8. But though Hume would have liked such a theory, he could not in honesty find one ‘more plausible’ . . . than theories of directing intelligence. In the end he has Philo confess that it is implausible that all the appearances of design should be the products of blind, unguided by intelligence, natural forces and that . . . he . . . employed disingenuously ‘objections [that] appear (what I believe they really are) mere cavils and sophisms’ on which no weight can be justly reposed . . . . (271)
Sobel’s Hume thus belongs to the ranks of Dembski, Denton, Behe, and other I.D. theorists!
But, alas, Sobel himself does not. He thinks that “new facts and new theories” undermine the inference to intelligent design (272). Here Sobel’s relatively weak grasp of contemporary science, already visible in his brief remarks on Big Bang cosmology, manifests itself plainly. The “new facts” that now subvert the design inference include the five billion year age of the earth, homologous structures, make-do adaptations (the Panda’s “thumb”), imperfections in “natural designs” (the human eye), and “gill” slits in the embryos of advanced species! Anyone familiar with even Michael Denton’s work alone will realize that these facts do very little to block a design inference. As for “new theories,” the reader is not surprised to hear that evolutionary theory has shown how “all forms of life . . . evolved from very simple ones that were little more than overgrown molecules . . . [in] a primordial soup” (273). Sobel passes over any problems of the theory is silence.
Of course, on the contemporary scene teleological arguments based on the fine-tuning of the cosmos have done an end run around the bogey of biological evolution, since fine-tuning is a pre-requisite for the evolution of intelligent life. Sobel thinks that Lee Smolin’s evolutionary many worlds scenario matches design hypotheses in terms of explanatory power and bests them “without contest” in terms of intrinsic plausibilities (283), apparently unaware of the many utterly implausible assumptions underlying Smolin’s scenario as well as its demonstrated explanatory inadequacy (universes producing primordial black holes would have a selective advantage in cosmic evolution, so that star-producing universes would tend to be weeded out). But never mind; Sobel does not really believe in these extravagant many worlds scenarios anyway. Rather if the fine-tuning cannot be explained in terms of a deeper theory, “I would put down the appearances of fine-tuning to our dumb luck” (284). It is at places in his book like this that we see how deeply resistant to theism Sobel is, for given the incomprehensibly small odds of the fine-tuning’s being due to chance alone in the absence of a World Ensemble, such a response is worse than Philo’s most disingenuous objections, as if Philo had asserted that the present ordered universe is due to the dumb luck of a single roll of the cosmic dice, a hypothesis not even Hume had the temerity to suggest.
In his chapter on miracles, Sobel surprisingly seeks to salvage what he can of Hume’s argument against the identification of a miracle. This involves him in a very generous re-interpretation of several of Hume’s claims. But central to the chapter is Sobel’s defense of the first part of Hume’s maxim “that no testimony . . . is sufficient to establish a miracle, unless this testimony is of such a kind that . . . its falsehood would be more miraculous, than the fact which it endeavours to establish.” For a miracle M and testimony to it t(M), Sobel takes Hume’s maxim to assert that Pr[M/t(M)] > 1/2 only if Pr(M) > Pr[t(M)& ¬ M]. This rendering of the falsehood of the testimony is controverted (Earman takes it more plausibly to be Pr[¬ M/t(M)] or Pr[t(M)/ ¬ M] [Hume’s Abject Failure, 40]. But Sobel’s formulation does state a necessary condition of Pr[M/t(M)] > 1/2. The problem is that this alone is just not significant in determining whether a miracle has occurred. Sobel earlier indicted Swinburne for failing to consider the full probability calculus in his design argument; but that also goes for arguments against miracles. Sobel also claims that everyone has a strong presumption against the identification of an event as a miracle, but his grounds for saying this seem strikingly question-begging, namely, “the principle that events in nature have causes in nature” (311). In any case, as Earman points out, even assigning a low value to Pr(M) will still not suffice to prove that miracles are unidentifiable. Hence, it is not clear to me that Sobel has shown that belief in God could not be warranted on the basis of powerful evidence for a miracle.
It is noteworthy that Sobel leaves entirely out of account moral arguments for God’s existence. He lets it be known that he does not believe that objective moral values exist (444). But would he agree with Mackie that if moral values did exist, then we would have a defensible argument from morality to the existence of a God? If so, then the implausibility of moral nihilism gives us good grounds to believe in God.
Sobel’s two chapters on the coherence of theism treat omnipotence and omniscience respectively. I have already remarked on his impressive handling of Grimm’s objections to a person’s knowing all truths. His chapter on omnipotence is hamstrung by his defining that notion in terms of ability to do tasks, rather than ability to actualize states of affairs. He then has an easy time demonstrating the incoherence of omnipotence so defined. But he takes no cognizance of Flint and Freddoso’s “Maximal Power” (Existence and Nature of God, 1983), the most important defense of omnipotence to date, so that this chapter is of only passing interest.
We come at length to Sobel’s anti-theistic arguments from evil. Sobel’s treatment of the logical version of the problem of evil is interesting because it is not posed in the typical way as an internal problem for theism. Rather he argues that evil exists, that evil’s existence is incompatible with the existence of a perfect being, and that, therefore, a perfect being does not exist. This is thus an evidential argument employing a modal premiss. Now since the argument is evidential, the first premiss is not so uncontroversial as it might appear. For what sense can the non-theist make of the notion of evil? For Sobel moral evils do not exist because there are no objective values. As for natural evils, how can the non-theist make sense of the claim that these are truly evil, that they “ought not to be”? As my colleague Douglas Geivett is wont to point out, a tsunami that washes over an island in the South Pacific may be disastrous for the population of the island, but it may be a great boon for the marine life offshore. Apart from God how can the non-theist even justifiably assert that such acts of nature ought not to occur?
Perhaps Sobel could recast his argument in terms of pain and suffering’s being logically incompatible with a perfect being. But then this claim becomes extraordinarily difficult to prove. With a bravado reminiscent of “Baghdad Bob” (Saddam Hussein’s notorious Information Minister), Sobel beats a steady retreat from one fall back position to another throughout the course of his chapter, insisting all the while that he has the upper hand. In the end his argument does not even assert the incompatibility of evil with God’s existence, and its premisses are mere speculations.
Sobel indicts Plantinga’s Free Will defense because Plantinga does not argue for the possibility of Transworld Depravity. But while that may be an indictment of Plantinga’s positive attempt to prove the compatibility of God and evil, all he needs to undercut the atheist’s argument that they are incompatible is the epistemic possibility of this hypothesis. For that alone reveals that the atheist has unjustifiably assumed that God can actualize just any possible world. Since Sobel admits its epistemic possibility, his sarcastic remarks on the possibility of attributing natural evils to demons only shows that he has failed to grasp the thrust of a Defense.
Suppose, then, we accept, as I think we must, a differentiation between possible worlds and feasible worlds. Sobel reformulates his argument: the world is not the best feasible world; that fact is incompatible with the existence of a perfect being; therefore, a perfect being does not exist. He just takes it for granted that the first premiss is true. But given a God endowed with middle knowledge, I think the first premiss is mere conjecture. Given the inscrutable complexity of a world of free creatures providentially governed by a God with middle knowledge, we have no idea whether better possible worlds are feasible for God. Thus, even given the second premiss, which is moot, Sobel’s argument has little force against the Molinist.
In addition to logical versions of the problem of evil, Sobel also presents probabilistic versions of the problem. But a major shortcoming of those arguments is that Sobel takes for granted that everyone should agree that this is not the sort of world we should expect beforehand from an all-good deity. This does not seem to me at all the case. If God’s purpose is not to “render the whole world happy” (405) but some other purpose, e.g., to bring people freely to a knowledge of Himself, then the world might well be filled with pain and suffering. How can anyone pretend to know beforehand what God’s purpose for creation is?
Sobel’s probabilistic argument from evil (422-6) consists of a single premiss:
Q*. E1 and E2 [certain evils exist] and either there is not a perfect being or there is no x such that there is a reason for thinking that x is a fact, that it is good that x, and that x justifies a perfect being’s permitting E1 or permitting E2.
Sobel’s argument presupposes that we have no reasons for thinking that a perfect being exists, a presupposition which most theists will not be prepared to concede. Yet Sobel’s Q* relies crucially on the assumption that any reason we have for thinking that x is a fact, etc. is not a reason for thinking that there is a perfect being. A curious feature of Q* is that the perfect being theist could seem quite ready to agree with what Q* affirms, holding that although there is a justifying good for E1 & E2, still it is not such that we have a reason for thinking it is what justifies God’s permission of these evils. Acknowledging the absence of such a reason seems all the more incumbent upon us if we must disallow any reasons that imply that a perfect being exists. Thus, the theist can affirm Q*. But Sobel assumes that the truth of Q* renders it certain that there is no perfect being. On behalf of that assumption he says, “if there is a perfect being, then there is a good that justifies its permitting E1 and E2; and . . . if something is a fact, then there is a reason for thinking that it is a fact, if this reason is only the fact itself” (423). But if the reason can just be the fact itself, then we do not need any evidence for the fact. The fact can be something like “Evils occur only to bring about greater goods.” I think that is a fact, and hence I have a reason for thinking it to be a fact, namely, the fact itself. So Q* is false, after all.
These comments only scratch the surface of this rich and complex book. It merits being read slowly, with ample time for reflection. In fact, it is hard not to read the book slowly, for Sobel’s style of writing is incredibly convoluted, so that sentences often need to be read and re-read in order even to understand them.
Finally, one of the imperfections of Logic and Theism is that it has been very poorly edited, being utterly riddled with printer’s errors from start to finish. Some of these errors are mildly amusing, as when we read of an “omniscient diving being” (483). But in most cases they are just annoying. There are omitted words, misspellings, redundant words, missing syllables, spacing and line errors, apostrophes instead of prime signs, and, most seriously, even mistakes in the logical symbolizations of premises (see, e.g., 37, 141, 174, 180, etc.). There are other minor mistakes, too: Sobel consistently refers to John Hick as Hicks (444), misquotes Mackie (458, 460), and confuses maximal greatness and maximal excellence (521). These mar the face of an otherwise magnificent work.