B. Draper’s Argument

Paul Draper presents an argument of quite a different sort.⁵⁸⁴⁵⁸⁴   See his “Pain and Pleasure: An Evidential Problem for Theists,” Noûs 23 (1989), pp. 331 ff. (This work is reprinted in EAESL; page references in the text are to this work). See also his ”Evil and the Proper Basicality of Belief in God,” Faith and Philosophy 8 (April 1991), pp. 135ff.; “Probabilistic Arguments from Evil,” Religious Studies 28, no. 3 (September 1992), pp. 285ff.; and “Evolution and the Problem of Evil,” in Louis Pojman, ed., Philosophy of Religion, 3d ed. (Belmont, Calif.: Wadsworth, 1997). He asks us to consider the pattern of pain and pleasure in the world: the amount and distribution of each and the sorts of conditions under which each is found. Draper then claims two things: first, this pattern of pain and pleasure is much less probable on theism than on a certain other hypothesis h inconsistent with theism; and second, this fact poses a serious problem for theistic belief. A way in which Draper’s argument is superior to the Rowe variety is that it doesn’t require that we be in a position to judge, with respect to any kinds of evils, the likelihood that an omniscient, omnipotent, and wholly good being would permit them. Nevertheless, he says, “Our knowledge about pain and pleasure creates an epistemic problem for theists” (p. 12). Why so, exactly?

1. Draper’s Argument Initially Stated

The problem is not that some proposition about pain and pleasure can be shown to be both true and logically inconsistent with theism. Rather, the problem is evidential. A statement reporting the observations and testimony upon which our knowledge about pain and pleasure is based bears a certain significant negative evidential relation to theism. And because of this, we have a prima facie good epistemic reason to reject theism—that is, a reason that is sufficient for rejecting theism unless overridden by other reasons for not rejecting theism. (p. 12)

What is that statement, and what is the significant negative evidential relation it bears to theism? As for the statement:

Now let “O” stand for a statement reporting both the observations one has made of humans and animals experiencing pain or pleasure and the testimony one has encountered concerning the observations others have made of sentient beings experiencing pain and pleasure. By “pain” I mean physical or mental suffering of any sort. (pp. 13–14)

So O is the statement that bears a “significant negative evidential relation to theism.” Note that O is person relative: each of us will have her own O, and my O may differ from yours. My O, we might say, sets out the facts about the magnitude, variety, distribution, duration, and the like (for short, the ‘disposition’) of pleasure and pain as I know them; yours does the same for you.

But what is this significant negative evidential relation in which O stands to theism? Here Draper bows in the direction of David Hume: most contemporary philosophers of religion (unlike Hume) “fail to recognize that one cannot determine what facts about evil theism needs to explain or how well it needs to explain them without considering alternatives to theism” (p. 13). The important question is “whether or not any serious hypothesis that is logically inconsistent with theism explains some significant set of facts about evil or about good and evil much better than theism does” (p. 13). And the answer to this important question, says Draper, is that indeed there is such a serious hypothesis, one that is both inconsistent with theism and explains some significant facts about good and evil much better than theism does. This is the “hypothesis of indifference” (HI, for short):

HI: Neither the nature nor the condition of sentient beings on earth is the result of benevolent or malevolent actions performed by non-human persons. (p. 13)⁵⁸⁵⁵⁸⁵   In “Evolution and the Problem of Evil,” he takes metaphysical naturalism—substantially, the view that there is no such person as God or anything much like God—to be the serious alternative hypothesis. My evaluation of Draper’s approach does not depend on a choice between these two candidates for the post of serious alternative hypothesis.

HI, of course, is inconsistent with theism (taking the latter to entail that the world has been created by a person who is wholly good as well as omnipotent and omniscient). Draper’s claim is that:

C: HI explains the facts O reports much better than theism does. (p. 14)

He claims furthermore that if one could show that there is a serious hypothesis that is incompatible with theism and explains O much better than theism does, then “one would have a prima facie good reason to believe that this alternative hypothesis is more probable than theism and hence that theism is probably false.”⁵⁸⁶⁵⁸⁶   “The Skeptical Theist” in EAESL, p. 178. What is it for a proposition to ‘explain’ something like the facts that O reports?

I will reformulate C as the claim that the facts O reports are much more surprising on theism than they are on HI, or, more precisely, that the antecedent probability of O is much greater on the assumption that HI is true than on the assumption that theism is true. (p. 14)

I take it the more precise formulation is the operative one here; we aren’t really talking about explanation⁵⁸⁷⁵⁸⁷   See William Alston’s reply to Draper in “Some (Temporarily) Final Thoughts on Evidential Arguments from Evil,” in EAESL, pp. 328–30. but just about the antecedent probabilities of O on theism and HI. Accordingly, we must ask what this ‘antecedent probability’ is. “By the ‘antecedent’ probability of O,” says Draper, “I mean O’s probability, independent of (rather than prior to) the observations and testimony it reports” (p. 14). So the antecedent probability of O is the probability of O on something like the rest of what I know.⁵⁸⁸⁵⁸⁸   Or perhaps on a noetic structure as similar as possible to mine that does not contain or entail O. This still isn’t quite right: the noetic structure in question also can’t contain or entail some proposition almost as strong as O. Perhaps we should think, then, of a noetic structure that contains no propositions about the distribution of pain and pleasure and is otherwise as similar as possible to mine. For possible difficulties with this notion, see Peter van Inwagen, “Reflections on the Chapters by Draper, Russell, and Gale” in EAESL, p. 222.

Finally, the probability in question is epistemic probability, not (for example) logical, statistical, or physical probability. And what is epistemic probability?

The concept of epistemic probability is an ordinary concept of probability for which no adequate philosophical analysis has, in my opinion, been proposed. As a first approximation, however, perhaps the following analysis will do:

Relative to K, p is epistemically more probable than q, where K is an epistemic situation and p and q are propositions, just in case any fully rational person in K would have a higher degree of belief in p than in q. (p. 27, footnote 2)

As Draper says, epistemic probability is an ordinary concept that is difficult to analyze or explain; suppose we provisionally accept his proposed first approximation.⁵⁸⁹⁵⁸⁹   For a fuller account of a closely related notion (epistemic conditional probability), see chapters 8 and 9 of Warrant and Proper Function (hereafter WPF). (I take it there is an implicit restriction to human persons; how things might go with other rational creatures is not our present concern.) What does K include? What goes into an epistemic situation? We shall have to return to this question later; for now, let’s say initially that K, for a given person S, would include at least some of the other propositions S believes, as well as the experiences S is undergoing and perhaps has undergone; it would also include what S remembers, possibly a specification of S’s epistemic environment, and no doubt more besides.

Now we see the general shape of the argument: the first premise is C, the claim that the antecedent epistemic probability of O given HI is much greater than the antecedent probability of O given theism. And second, if C is true, says Draper, then “we have a prima facie good epistemic reason to reject theism—that is, a reason that is sufficient for rejecting theism unless overridden by other reasons for not rejecting theism” (p. 12). Here he is apparently relying on a general principle, perhaps something like

(1) For any propositions P and Q and person S, if S believes P and Q and there is a serious hypothesis R that is incompatible with P and such that the antecedent epistemic probability of Q with respect to R for S is much greater than the antecedent epistemic probability of Q with respect to P for S, then S has a prima facie good epistemic reason to reject P.

Draper’s claim is that the antecedent epistemic probability of O on HI is much greater than on theism, and because HI is a serious hypothesis and is inconsistent with theism, we have a prima facie good reason for rejecting theism:

Now suppose I succeed in showing that C is true (relative to our own and my reader’s epistemic situations.) Then the truth of C is (for us) a prima facie good (epistemic) reason to believe that theism is less probable than HI. Thus, since the denial of theism is obviously 473entailed by HI and so is at least as probable as HI, the truth of C is a prima facie good reason to believe that theism is less probable than not. And since it is epistemically irrational to believe both that theism is true and that it is less probable than not, the truth of C is also a prima facie good reason to reject (i.e., to cease or refrain from believing) theism. (p. 14)

The claim, then, is that the truth of C gives me a “prima facie good reason to believe that theism is less probable than not”—that is, that its probability is less than .5. Less probable than not with respect to what? The answer must be K. The idea is that the truth of C gives me a prima facie good reason for thinking that theism is improbable with respect to my noetic situation; hence, unless I can find some reasons for theism, the rational thing to do is to give it up.⁵⁹⁰⁵⁹⁰   In “Evolution and the Problem of Evil,” Draper puts the same thought slightly differently; speaking of a similar argument, he says, “This is why my case against theism is a prima facie one. I am entitled to conclude only that other evidence held equal . . . it is highly probable that theism is false.” What is it to hold other evidence equal? Here’s a suggestion: it would be to consider the probability of theism with respect to an evidential situation that was as similar as possible to mine, given that it contained no evidence for or against theistic belief, or given that the evidence it contained for theistic belief was precisely balanced by the evidence it contained against theistic belief. We could put this by saying that, according to Draper, my knowledge of the truth of C gives me a defeater for theism, unless I can find some reasons for it; alternatively, it gives me a potential defeater for theism, a potential defeater that will be actual unless I can find those reasons for theism.

2. On Being Evidentially Challenged

This is a subtle challenge and a fascinating new entry into the lists; Draper deploys it with power and sophistication. Nevertheless I think the argument utterly fails to show that traditional Christian theism is threatened by a defeater or epistemologically threatened in some other way. Suppose we take a closer look. Now Draper’s argument really has two premises, C and (1). I have argued elsewhere⁵⁹¹⁵⁹¹   “On Being Evidentially Challenged,” EAESL, pp. 250ff. that in fact C is false: it is not the case that the amount, duration, and distribution of pain and pleasure, as I understand it, are more probable on HI than on theism. Here I want to focus on the other premise, the claim that if, in fact, O is much more likely on a serious alternative hypothesis like HI than on theism, then the theist has a prima facie reason to reject theism. Why think a thing like that? Suppose (contrary to fact, as I see it) C were true: what kind and how much of a challenge to theistic belief would this be? How widespread 474is this alleged evidential disability? Before we can answer this question, however, we must ask another: what, exactly, is a serious alternative hypothesis? Draper’s answer: “Specifically, one hypothesis is a ‘serious’ alternative to another only if (i) it is not ad hoc—the facts to be explained are not arbitrarily built into it—and (ii) it is at least as plausible initially as the other hypothesis.”⁵⁹²⁵⁹²   “Probabilistic Arguments from Evil,” pp. 315–16. Condition (i) requires no present comment; what about condition (ii)? How are we to understand ‘plausibility’ here? I think Draper means to abstract from specific epistemic situations: we are to think of the plausibility of a hypothesis as depending not on considerations such as the specific evidence (propositional and nonpropositional) I may have for or against it, but on more general considerations such as its scope and specificity, and perhaps how it fits in with what is generally known (a hypothesis entailing that the world is flat wouldn’t be plausible). Thus, for example, he defends the plausibility of HI as follows:

And it [HI] is at least as plausible initially as G [i.e., theism]. After all, G is a very specific supernaturalist hypothesis with strong ontological commitments. If, on the other hand, we take the Indifference Hypothesis to be the hypothesis that the first causes of the universe, if there are any, are neither benevolent nor malevolent, then the Indifference Hypothesis is consistent with naturalism as well as with many supernaturalist hypotheses and its ontological commitments are much weaker than G’s.⁵⁹³⁵⁹³   Ibid., p. 316.

So what counts for plausibility are these general facts about relative scope and strength. Still further, if I had to consider the specific evidence I have (propositional or otherwise) for HI to evaluate its plausibility, I would have to take any reasons I have for theism as evidence against HI; HI might then be very implausible for me. So plausibility must abstract from such specific evidence.

Suppose we say that a proposition P is evidentially challenged for S if it satisfies the antecedent of (1): P is evidentially challenged for a person S if and only if S believes P and there are propositions Q and R such that S believes Q, R is a serious hypothesis incompatible with P, and Q is much more probable with respect to R than with respect to P. What (1) claims, therefore, is that if a proposition P is evidentially challenged for S, then S has a prima facie good epistemic reason for rejecting P—for being agnostic with respect to it or believing its denial. Is this really true? Is being evidentially challenged a serious handicap?

Well, how widespread is it? How many of my beliefs are evidentially challenged, for me? More, perhaps, than we might initially think. For example, here are three more propositions related, for me, as are theism, O and HI:

(2) George is a non-Catholic academic,

(3) George is a professor at Notre Dame,

and

(4) George is a Catholic academic.

First, I believe both (2) and (3). Second, (3) is vastly more likely on (4) (relative to K) than it is on (2). (After all, the proportion of Catholic academics who are professors at Notre Dame is many times greater than that of non-Catholic academics who are professors there.) Further, (4) is incompatible with (2). Still further, (4) is a serious hypothesis: it is not ad hoc, and it is as plausible as (2). (True, I have a lot of evidence for (2)—the fact, e.g., that George is an elder in the Christian Reformed Church, which is non-Catholic, the fact that George has always claimed to be a Protestant, and so on—but as we saw above, this specific evidence isn’t relevant to the plausibility of (4).) So (2) is evidentially challenged for me. Does this fact give me a good reason to reject (1)? (Should I reconsider: George is a professor at Notre Dame, after all, and that is much more likely on (4) than on (2); so maybe he’s really a Catholic?) Not clearly.

A similar trio of propositions:

(2*) I am in my study,

(3*) I am within four feet of a dog,

and

(4*) I am at the dog pound.

Again, I believe (2*) and (3*); (4*) is a serious (in Draper’s sense) alternative to (2*), and (3*) is much more likely on (4*) than it is on (2*) (usually there aren’t any dogs in my study); therefore (2*) is evidentially challenged for me. So, incidentally, is (3*) itself:

(3*) I am within four feet of a dog,

(5) I hear no doggie sounds such as barking, growling, panting, or jingling of tags,

and

(6) I am not within earshot of any dogs.

Again, (6) is a serious alternative hypothesis to (3*), and (5) is much more likely with respect to (6) than it is with respect to (3*). (3*), therefore, is evidentially challenged for me. A couple of more examples: my friend has a cat named Maynard; I believe that Maynard is a cat and 476also (as my friend reports) that Maynard likes cooked green beans; the latter, however, is much more likely on the serious (in Draper’s sense) alternative hypothesis that Maynard is a Frisian, or possibly a Frenchman; so the belief that Maynard is a cat is evidentially challenged for me. I believe (naturally enough) that you are a human being; you and I are on a walk in the woods, however, so I also believe that you are in a forest; of course that proposition is vastly more likely on the serious alternative hypothesis that you are a tree; so the belief that you are a human being is evidentially challenged for me. (As far as that goes, so is the belief that I am a human being.)

I think you get the picture. It seems likely that most of what we believe—at any rate for propositions that are contingent in the broadly logical sense—is also evidentially challenged. I don’t know how to give a proof of this claim (it probably isn’t worth spending a whole lot of time trying to find a proof); but it certainly seems likely to be the case. And this suggests that a challenge of this sort is not very significant by itself or in the general case. If most of the propositions I believe face an evidential challenge, then I don’t learn much of interest about theism by learning that it, too, faces such a challenge.

Under what conditions (if any) would a challenge of this sort be significant? What sorts of beliefs are such that their being subject to an evidential challenge gives us serious reason to doubt them? Here we think first of scientific hypotheses. I propose a hypothesis H* to explain the behavior of gases: you point out that certain data are more probable with respect to another hypothesis H' incompatible with mine; that certainly seems to be a strong prima facie reason to doubt my hypothesis. Of course the data must be relevant data, the sort of data H* is in the business of explaining. Suppose Sam presently feels a mild pain in his left knee. That is much less probable with respect to H* than with respect to the hypothesis H': Overcome by astonishment at learning that H* is false, Sam fell and injured his knee; still, that is nothing whatever against H*. For the typical scientific hypothesis H, there will be a body of relevant data (past and future as well as present) such that the success of H depends on how well it explains that data; and many scientific hypotheses (at least on the most usual stories) get all or nearly all of their warrant from the fact that they account for the relevant data.⁵⁹⁴⁵⁹⁴   At any rate nearly all of its original warrant. Special relativity, for example, gets its warrant for me, not from the fact that it properly accounts for those data, but from the fact that I have been told and believe that it does (in such away as to satisfy the conditions for warrant). But if those conditions are indeed satisfied, then there must be someone at the other end of the testimonial chain for whom these beliefs have warrant in some way other than by testimony. See WPF, chapter 4. A proposition of that sort is seriously threatened by a relevant evidential challenge. If I discover that a belief of this sort is subject to an evidential challenge, then I do have substantial evidence against it and a strong prima facie reason to give it up.

As I have argued throughout this book, however, it is an enormous assumption to think that belief in God or, more broadly, the larger set of Christian (or Jewish or Muslim) beliefs of which belief in God is a part, is in this respect like a scientific hypothesis. Not only is this assumption enormous: it is also false. The warrant for these beliefs, if they have warrant, does not derive from the fact (if it is a fact) that they properly explain some body of data. For most believers, theistic belief is part of a larger whole (a Christian or Muslim or Jewish whole); it is accepted as part of that larger whole and is not ordinarily accepted because it is an explanation of anything; hence its rationality or warrant, if it has some, does not depend on its nicely explaining some body of data.⁵⁹⁵⁵⁹⁵   See my “Is Theism Really a Miracle?” and see above, pp. 330ff. I don’t mean to deny, of course, that Christian or theistic belief can get more warrant by nicely explaining something else one believes.

Still, does this fact, crucially important as it is, deliver theism from Draper’s evidential challenge? Is it only scientific hypotheses for which (relevant) evidential challenges are serious? No. Suppose you are under the impression that your friend Paul has been vacationing on Cape Cod for the last couple of weeks (you have a rather weak memory that this is where he said he was going), but the postcards you get from him were mailed from Grand Teton National Park; he doesn’t say in the postcards where he is, but he does note the remarkably dry air, as well as the great differences between day and night temperatures. Then I think your belief that he is vacationing at Cape Cod is seriously challenged (a relevant alternative hypothesis being that he is vacationing in the Tetons). And this is true even though the warrant for your belief that he was vacationing on the Cape didn’t arise as a result of its properly explaining data of one kind or another. So it isn’t just scientific hypotheses that can be called into question by virtue of facing a relevant evidential challenge.

Suppose we look a bit deeper here. That I am in my study (and not at the dog pound), that Maynard is a cat, that you are a human being—these are all subject to an evidential challenge; of course that doesn’t suggest for a moment that there is something irrational or problematic in these beliefs, or that they are improbable with respect to our epistemic situations. Why not? Because each of these propositions has a good deal of warrant for me, warrant that is independent of its probabilistic relationships to the beliefs involved in the evidential challenges. In cases like this, being evidentially challenged comes to very little. And it isn’t even necessary that the belief in question have a high degree of warrant. I believe rather infirmly (I have a relatively weak memory belief here) that the population of greater New York City is more than 17 million; I also believe that the area of 478 greater New York City is 1,384 square miles; that proposition is many times more likely with respect to the serious alternative hypothesis that the population of New York City is less than 10 million. My belief that the population of New York City is more than 17 million is therefore subject to an evidential challenge; that fact doesn’t provide me with a prima facie defeater for it, even though it doesn’t have a high degree of warrant. Perhaps most of what I believe faces an evidential challenge, but most of what I believe (so one thinks) also has warrant of one kind or another; and when it does, an evidential challenge doesn’t amount to much. With respect to most of what I believe, being evidentially challenged does not threaten to serve as a defeater for the proposition in question, and neither does my knowing, if I do, that it is evidentially challenged. Neither the challenge nor the knowledge, in the case of the propositions mentioned, puts me in a condition where, if I continue to believe the challenged proposition, I am irrational or in some other way out of line, epistemically speaking. And that is because the propositions in question get warrant from such sources as perception, memory, sympathy, testimony, a priori intuition, and the like; they do not depend, for their warrant, on their relation to such propositions as those furnishing the evidential challenge.

Well then, how does it stand with theism? According to Draper, “Establishing the truth of H [that theistic belief faces an evidential challenge] would be insignificant if the typical theist could rationally continue to believe that God exists after learning that H is true.”⁵⁹⁶⁵⁹⁶   “Evil and the Proper Basicality of Belief in God,” p. 138. What I propose to argue here is that the typical theist can rationally continue to believe that God exists after learning that theism faces an evidential challenge. Suppose I accept traditional Christian belief, including, of course, theistic belief. Now suppose I come to believe that in fact theistic belief is subject to an evidential challenge. I don’t as a matter of fact believe that the pattern of pain and pleasure in the world does provide such a challenge—at any rate I don’t think Draper’s argument for this conclusion is successful—but suppose I come to believe that there is an evidential challenge of this or some other kind for Christian or theistic belief. Would that give me a defeater for my theistic belief? Would it make it irrational for me to continue believing?

Not if that belief has any significant degree of warrant for me. Suppose Christian and theistic belief has a good deal of warrant for me by way of faith and the internal instigation of the Holy Spirit (IIHS) (see above pp. 249ff.); then the fact that theism is evidentially challenged doesn’t give me a defeater and doesn’t bring it about that 479my theistic belief is irrational. Compare the case of Maynard and my belief that he is a cat. You point out that this belief suffers from an evidential challenge: that he likes cooked green beans is much less likely on his being a cat than on his being a Frisian. I agree, but am undeterred, continuing in full rationality to believe that he is indeed a cat. This belief is rational for me in these circumstances because it has warrant for me quite independent of its relationship to the proposition that Maynard likes cooked green beans. There is of course no cognitive malfunction involved in my continuing to hold a belief with significant warrant from such sources as memory, perception, IIHS, and the like, even when I learn that the belief is subject to an evidential challenge. Our cognitive design plan permits, indeed, requires maintaining such a belief in the face of such a “challenge.” And clearly the same goes for my theistic belief, if, in fact, it has warrant in the way proposed in chapter 8.

And this is true even if I don’t myself believe that theistic belief has warrant for me. Perhaps I have never thought much about epistemology, have at best a hazy idea as to what warrant is, and have never considered such questions as whether a proposition’s being evidentially challenged gives me a reason for rejecting it. You point out that my belief that Maynard is a cat is evidentially challenged; I continue (in my epistemological innocence) to believe as firmly as before that Maynard is a cat; neither the rationality nor the warrant of that belief is diminished. Again, the same goes for theistic belief, if it has significant warrant for me. You point out that theism is evidentially challenged for me: I agree that that is so and continue to believe as firmly as before; if Christian belief and hence theism do have significant warrant for me, my continuing so to believe is wholly rational and remains warranted. It is perfectly rational, internally, because it still seems obviously true to me; it is perfectly rational, externally, because the belief in question is held under the conditions of warrant. If it had sufficient warrant for knowledge before you made your point about its being evidentially challenged, it still has sufficient warrant.

So if theistic belief has significant warrant for me, then (in the typical case and provided I do not believe that it lacks warrant) my coming to believe that it faces an evidential challenge does not provide me with a defeater for it. Here we see a special case of a pattern we have seen before. I argued in chapters 6 and 8 that if theistic and Christian belief is true, then very likely it has warrant. A consequence is that if Christian belief is true, then very likely (in the typical case) an evidential challenge to theism is an insignificant challenge.

But even if theism has little or no warrant, it could still be (and in the typical case would still be) that an evidential challenge doesn’t provide a real challenge or a prima facie defeater. Analogy: perhaps I am once more told by my friend that she has a pet named Maynard 480who is a cat but nonetheless loves cooked green beans; having never met this Maynard, I believe on the basis of my friend’s testimony that Maynard is a cat. As it turns out, my friend is indulging (unbeknownst to me) her penchant for telling whimsical (and false) stories. Then my belief that Maynard is a cat has little by way of warrant: the epistemic minienvironment (see above, pp. 158ff.) isn’t right, being polluted by my friend’s thus lying to me, so that the environmental condition for warrant is not met. Still, my belief that Maynard is a cat is (all else being equal) entirely rational (even if not warranted), both internally and externally; and that holds even though I am quite aware that it is evidentially challenged. The same can be true for theistic belief. Perhaps I mistakenly but rationally believe that it has warrant; I rationally believe that some of the theistic arguments, for example, are very strong, or I believe, mistakenly, in some story like the one told in chapters 6 and 8, according to which theistic belief does indeed have warrant. Under those conditions, my theistic belief does not, in fact, have warrant; nevertheless, my learning that it is subject to an evidential challenge does not compromise its rationality and does not give me a defeater for it.

So when could a belief’s being evidentially challenged (more exactly, my knowledge that it is evidentially challenged) actually offer me a defeater for a belief and make it irrational for me to continue to hold it? I can see two sorts of cases in which learning that theism is subject to an evidential challenge could be a defeater for it. First, suppose I am a theist, am rational in accepting this belief, but hold it with little firmness and furthermore think my reasons for it are absolutely minimal—barely sufficient for holding the belief rationally. Then if I learn that theism is subject to an evidential challenge, perhaps I have a defeater for it. I say ‘perhaps’ advisedly; the situation isn’t really clear.

The second sort of situation is clearer. Consider a belief B I accept because I think it the best explanation of a certain range of data D; B has no warrant apart from its properly explaining D, and I am aware of this fact. Finding that B is subject to an evidential challenge, one thinks, gives me a defeater for it—provided that the belief that is more probable with respect to the alternative hypothesis is one that B is supposed to explain. I believe the butler did it: my only reason for so believing is that this hypothesis best explains all the facts and circumstances of the crime. Now I come to see that the hypothesis that Lady Fauntleroy did the deed better explains some of those facts and circumstances.⁵⁹⁷⁵⁹⁷   And in this context perhaps we can gloss ‘explanation’ in terms of probability. Then my belief that the butler did it faces a relevant evidential challenge, a challenge which is prima facie a defeater 481for that belief. (Of course it doesn’t matter if some alternative hypothesis better explains the fact that Beijing is a large city.) So suppose I accept theism as a hypothesis; I accept it because I think it the best explanation of some range of phenomena including the origin of the universe, the reality and objectivity of right and wrong, and also the distribution of pain and pleasure. Suppose, furthermore, I rightly believe that I have no other sort of reason for theistic belief—no promptings from the IIHS, or from the sensus divinitatis, or from the testimony of others. I believe that I have no other source and am correct in that belief. Now suppose I come to think that the Indifference Hypothesis, or naturalism, or something else does a better job of explaining the magnitude, duration, and distribution of pain and pleasure: then my theistic belief would be subject to an evidential challenge, a challenge that is prima facie a defeater for theistic belief and a reason for giving it up. (Even then, however, I might conclude that theism did a better job of explaining some other relevant phenomena.)

So there are some situations in which an evidential challenge—not just any old evidential challenge, but a relevant one—does furnish a defeater: cases where someone believes that the warrant enjoyed by his theistic belief is minimal, and cases where he believes that the warrant theism has for him depends just on its explaining a certain range of phenomena. Most theists, however, are not in either of these conditions. Are there other conditions in which theists often find themselves, conditions in which coming to see that theism faces an evidential challenge really does provide a defeater or a prima facie defeater for theistic belief? I doubt very much that the typical theist is in any such condition. I therefore think Draper’s challenge, subtle and sophisticated as it is, fails; in his own words, “the typical theist could rationally continue to believe that God exists after learning that H [that theism is evidentially challenged] is true.”