When approached about participating in a debate on the implications of “religious uncertainty” for religious commitment, I was asked to represent the theistic side. This I was happy to do. My opponent was asked to represent the atheistic side. Little did I imagine then that the debate would devolve into quibbles about faith commitments, with John Schellenberg, like a prophet founding a new religion, advancing “ultimism” and variously suggesting “beliefless religious faith,” “naturalistic faith,” “ultimistic faith,” and “rational faith” as the proper attitudes to adopt toward his fledgling faith. The fuzziness of these New-Age-sounding attitudes is enough to make one nostalgic for that old-time atheism of J. L. Mackie and Antony Flew. Schellenberg suggests that his ultimism is preferable to the old-fashioned theistic commitment I advanced since, he claims, theistic commitment is incurably tainted with self-deception, while ultimistic faith (whatever it is) is not. But let’s see about that.
Is it morally and rationally problematic to engage in pragmatic acceptance, or pragmatic belief-formation, insofar as these involve self-deception? While self-deception may be a serious problem with regard to inculcating a belief which one takes to be false, it does not seem to be a serious threat with the inculcation of a belief which one thinks has as much evidence in its favor as against, nor does it seem to be a threat when one takes the probability of the proposition to be indeterminate, since one could form the belief knowing full well the evidential situation. Even if it is true that believing that p is being disposed to feel that p is probably the case, it does not follow that believing that p involves being disposed to feel that p is probably the case based on the evidence at hand. Second, this is an objection not to pragmatic belief-formation per se, but an objection to pragmatic belief-formation that involves self-deception. Suppose it is true that the employment of self-deceptive belief-inducing technologies is morally and rationally problematic. This says nothing about belief-inducing technologies that do not involve self-deception. If there are belief-inducing technologies free of self-deception and which generate a belief on the basis of a pragmatic reason, then Schellenberg’s objection fails.
Is there a belief-inducing technology available that does not involve self-deception? Consider a technology consisting of two components, the first of which is the acceptance of a proposition, while the second is a behavioral regimen of acting on that acceptance. Keep in mind that accepting a proposition, unlike believing, is an action that is characterized, in part, by one’s assenting to the proposition, whether one believes it or not. One accepts a proposition, when she assents to its truth and employs it as a premise in her deliberations. One can accept a proposition that one does not believe; and one can believe a proposition that one does not accept. One might be disposed to believe that the next toss of the fair coin must come up tails, since it has been heads on the previous seven tosses. Nevertheless, one ought not to accept that the next toss must come up tails, or that the probability that it will is greater than one-half. Acceptance, we should remember, unlike believing, is an action that is under our direct control.
If one accepts a proposition, then one can also act upon the proposition. Acting upon a proposition is behaving as though it were true. The two-step regimen of accepting a proposition and then acting upon it is a common way of generating belief in that proposition. And, importantly, there is no hint of self-deception coloring the process. So, it is false that self-deception must taint pragmatic belief-formation; and false that a Pascalian theistic commitment is incurably infected with self-deception.
Perhaps Schellenberg holds that to accept or knowingly cultivate belief in a proposition not supported by the evidence is an intrinsic evil so bad as to always fall on the far side of impermissibility. We might understand this idea by employing the notion of a basic evil. A basic evil, or what is sometimes called ‘evil per se,’ is an action that is always wrong for an agent intentionally to do, no matter what instrumental benefits may follow from it. Suppose that lying were a basic evil. It would be wrong to lie even if the heavens should otherwise fall. Likewise, one might suppose, to knowingly cultivate belief in a proposition unsupported by the evidence is to expose oneself to the great wrong of basic evil.
It’s hard to take this idea seriously. Consider the Alpine Hiker case again. Is it really plausible to hold that the hiker commits a basic evil by maximizing his chance for survival? Or recall the “ET case.” If this objection were valid, you’ve impermissibly dirtied your hands in the service of humankind, by committing a basic evil. But that conclusion is too wildly implausible to take seriously. It is clear enough that there are occasions in which it would not be a basic evil to knowingly cultivate belief in a proposition that’s just as likely as not.
Schellenberg also argues that there’s good reason to think that the expected value of F5 exceeds F2, when we consider only the “N” column. Schellenberg thinks that adopting naturalism means that one will no longer have to cart around “those very heavy ideas involved in theism…” Ideas that include, among others not found within naturalism, that one’s life has an eternal meaning, that justice will be satisfied, and that the vital goods of this world are not extinguishable.
Consider just those three “very heavy ideas” of theism. Whatever their weight, they are clearly very optimistic and very hopeful ideas. When we talk about optimism and hope, with regard to theism and naturalism, it is important to keep in mind that one worldview asserts that this world is all there is; while the other asserts that there is more to this world than meets the eye, and that human life and relationships have a permanency that extends beyond the grave. So, within a context of “religious uncertainty” in which there is no decisive evidence which worldview is false, the palm must be awarded to theism rather than naturalism with regard to hopefulness and optimism. The opportunity for hope and optimism is far greater with theism than with naturalism, and it is silly to assert otherwise.
In addition, there is an interesting possible postmortem asymmetry noted by Pascal:
Who has the most reason to fear hell: he who does not know whether there is such a thing as hell and who is sure of damnation if there is, or he who is certainly convinced that hell exists, but hopes nevertheless to be saved?
If the theist dies and theism is false, it is likely that she will never have occasion to regret her theistic commitment. The atheist, however, who dies could very well have the occasion to regret his lack of theistic commitment if theism were to prove true. The exposure to regret is far greater on the atheistic side than on the theistic.
As if whistling past a graveyard one can invoke as many visions of “undulating hills” and the “biting wind” as he wants, but it will still be the case that a grasping for a vague ultimistic faith betrays the atheist cause and is itself an acknowledgment that the expected benefits associated with a religious commitment swamp those of naturalism, even if religious uncertainty obtains. And not realizing this is perhaps itself an interesting case of self-deception.
Well, what about F2 exceeding F8? Schellenberg holds that there’s no reason to think that the expected value of F2 exceeds that of F8, despite the paucity of studies involving nontheistic religions. Perhaps there is a similar empirical benefit to be had with non-Western religions as with Christian theism. If there is it is hard to see how this would comfort the naturalist. I addressed this issue in an endnote, but Schellenberg complains that the note was too short, so allow me to expand. We are discussing a forced decision. A decision is forced whenever failing to decide is practically equivalent to having chosen one of the alternatives. So, suppose you were making an important medical decision for a loved one, some one under age perhaps, or incapacitated. You must decide which therapy, if any, to choose. Suppose therapy X has some experimental support. It would be irrelevant to point out that there are therapies no one has yet thought of. It would also be irrelevant to point out that there are alternative therapies, of which little study has been done. Clearly, it would be irresponsible to forego therapy X, which has some experimental support, in favor of an alternative therapy with no experimental support. It is true that if the situation is desperate enough, you may consider an alternative therapy, but part of the desperation will be that the conventional therapies have all been tried and have failed. Likewise, those religions lacking the sort of social science support enjoyed by theism would have a value comparable to what’s found in cells F5 or F8, and, hence, less than cell F2. Of course, if more information were to become available, assignments may change. As exploration proceeds what had been designated terra incognita becomes a region now known. But until the exploration is done, terra incognita it remains.
Schellenberg also attempts to revive the many-gods objection. This objection is, of course, an arrow found in nearly every quiver of those attacking Pascal’s wager since the 1700s. According to the many-gods objection, Pascal’s wager proves too much as a wager similar to Pascal’s is possible for any number of incompatible religious options. Let’s quash this revival by driving a stake into the heart of the many-gods objection by noting that a principle found in nearly all versions of that objection is false:
F. every logically possible proposition has a probability greater than zero. According to (F) logical possibility is sufficient for an assignment of positive probability.
Why is (F) false? There are propositions that are both logically possible and yet plausible candidates for a zero probability assignment. For instance, when I consider the statement that:
J. there is not at present a living human body which is mine
and call to mind that I enjoy neither necessary existence nor self-existence, and for that matter, with just a tiny change in the cosmological constants of the universe, or a slight revision of history, I would not exist, it would be absurd (and unduly modest) for me to assign (J) anything other than zero. And, of course, there is nothing unique about me with regard to (J); it would be absurd for anyone to assign (J) a positive probability. Or think of the proposition that:
K. human beings exist.
It would be absurd for any human to assign (K) a value less than one, and its denial anything greater than zero, even though (K) is not a necessary truth and its denial is not a contradiction, since there are possible worlds that contain no humans. Or consider the proposition that:
L. I had parents.
While it is logically possible that I had no parents, it would be madness for me to assign any value less than unity to the proposition that I in fact had parents. But if I assign unity to (L), then the denial of that proposition, although logically possible, receives a zero value. Even without listing examples, it is clear enough that there are many propositions in addition to (J), (K) and (L) which are logically possible but deserve a zero assignment. As there are logically possible propositions that deserve a zero probability assignment, (F) is false.
One might object that rejecting (F) leads to a semi-Dutch book situation (a situation in which one is open to a series of bets that one cannot win, but there is a possibility of loss). Perhaps it does. Of course, semi-Dutch books, or even a strict Dutch book for that matter (a situation in which one cannot win and can only lose) are threatening only if there are Dutch bookies about. And since a level of knowledge approaching omniscience would be required to gain a Dutch book, one can rest assured that no Dutch bookie will ever be encountered. So even if a semi-Dutch book situation arises from the rejection of (F), this is insufficient to show the rejection irrational.
Another objection might run as follows. Only contradictions and other necessarily false propositions have an objective probability value of zero. Further, it is absurd for one’s assignment of subjective probability to a proposition not to reflect the objective probability value of that proposition. Since a possible proposition is not necessarily false, it follows that (F) must be taken as true as regards subjective probability (by subjective probability is meant, roughly, one’s personal assignment of probability values to various uncertain propositions, with the assignments reflecting the strength of one’s belief in those propositions).
The problem with this objection can be seen by generalizing one of its premises:
M. if the objective probability of a proposition p is not equal to n (and one knows that), then it would be irrational for one’s subjective probability of p to be n.
Goldbach’s conjecture asserts that every even integer, save two, can be expressed as the sum of two prime numbers. This conjecture, despite many attempts, has never been proved or disproved. If true, the conjecture is necessarily true; if false, then, necessarily false. According to the probability calculus, a necessarily true proposition is assigned probability one, a necessarily false one, zero. So, the objective probability of Goldbach’s conjecture is either one or zero. Nevertheless, it is perfectly reasonable to assign it a subjective probability that falls somewhere between one and zero. I, for one, suppose it to be true, based on the authority of others. I would not, however, take a bet with much at stake that it is true, and certainly not a bet with everything to lose and nothing to gain. So, I do not assign it probability one–and this seems perfectly reasonable. Given that one does not know whether the conjecture is true, one is not required to assign it either one or zero, even though its objective probability is either one or zero.
The same point holds, clearly enough, if we switch from subjective probability to epistemic probability (by epistemic probability is meant, roughly, the likelihood of a proposition relative to one’s evidence). Relative to what one knows or what one justifiably believes, there are many logically possible propositions that are properly assigned a zero probability. In any case, given the falsity of (F) the revival of the many-gods objection is doomed to fail.
Let me end by way of a point made earlier, a point worth repeating as it has been widely neglected: the divine hiddenness argument rests on the shaky foundation of absolute evidentialism. Absolute evidentialism, recall, implies that one should refrain from believing or accepting any proposition that is not rendered more likely than not by the evidence. Quite apart from quibbles about theistic faith and ultimistic faith, the vulnerability of absolute evidentialism to easily constructed counterexamples is the bane of the divine hiddenness argument. With the collapse of absolute evidentialism, the divine hiddenness argument topples into irrelevancy, as there is overwhelming reason supporting theistic faith as compared to atheism, naturalism, or ultimistic faith, even in the fog of religious uncertainty.
 Additionally, it is far from clear that every case of self-deception is morally or rationally problematic. It seems plausible enough that just as it is sometimes permissible to deceive other people (for instance, leaving the lights on when one goes out for the evening is a permissible way of deceiving would-be burglars), it is likewise sometimes permissible to deceive oneself. For a useful discussion of this, see Mike W. Martin, Self-Deception and Morality (Lawrence, KS: University of Kansas Press, 1986).
 The figures F2, F5, and F8 and N all refer to the three-columned matrix found in my initial installment in which “D” represented the existence of a nonstandard deity, “N” represented the world with no deity of any sort, and “G” represented the god of theism.
Believe in G
Believe in Neither
Believe in D
With regard to that evidence, one might worry about the old bugaboo of statistical studies showing correlations. Do they show merely a correlation, or causality, and if there is a causal connection, which way does it flow? For instance, it may be true that the depressed drink more than those not depressed, but is the depression caused by the excessive drinking, or the drinking by the depression, or do both flow from some other factor? Typically, the flow of causality is shown by a counterfactual dependency of an event or phenomenon upon another. In any case a Pascalian response to this worry builds upon our ignorance. In the absence of an answer that settles the correlation question, the prudential response is to proceed as if the religious commitment produces the benefits, or plays a significant role in their production (which in fact the studies tend to support). There is little if any harm in doing so, and much that might be gained.
 A proponent of the many-gods objection can avoid employing (F) if she uses only actual religions (Christianity, Judaism, Islam..) in the construction of the objection. But constructing the many-gods objection this way provides no comfort for the naturalist, since, the infinite value in cell F6 is thereby converted into a finite value.
Copyright ©2008 Jeffrey Jordan. The electronic version is copyright ©2008 by Internet Infidels, Inc. with the written permission of Jeffrey Jordan. All rights reserved.