Sceptical theism

Here is a simple evidential argument from evil:

Gratuitous evil is evil such that there is no God-justifying reason for him to permit it. Why suppose the second premise of the above argument is true? Proponents sometimes offer a ‘noseeum inference‘Footnote ¹ in its support: if, after thinking hard, we cannot think of any God-justifying reason for his permitting some evil, then it is likely that no such God-justifying reason exists.

Sceptical theists attempt to block this noseeum inference. According to sceptical theism, those goods, evils, and entailment relations between them that we know of may, for all we know, constitute the unrepresentative tip of a vast, largely hidden iceberg of goods, evils, and entailment relations. The latter, if they exist, will be known to an omnipotent God, and may, for all we know, justify him in allowing those evils we observe.

Thus (call this the anti-noseeum argument): if sceptical theism is true, then the fact that we cannot think of any God-justifying reason for his permitting some evil does not justify us in concluding that no such reason exists. If by inscrutable evil we mean evil for which we can discern no sufficient God-justifying reason, then what the anti-noseeum attempts to block is, in effect, the inference from inscrutable to gratuitous evil.

Analogies are often used to explain and justify the sceptical theist's scepticism regarding the noseeum inference. The fact that we cannot detect any insects in the garage (when taking a look from the street) does not allow us justifiably to conclude that there are no insects in the garage. The fact that a chess novice cannot think of a good reason why a chess Grandmaster made a particular move does not allow the novice justifiably to conclude that no reason exists. Similarly, given our ignorance concerning the completeness of our knowledge of goods, evils, and the entailment relations between them, the fact that we cannot identify a reason that would justify God, if he exists, in permitting some evil does not allow us justifiably to conclude that no such reason exists. Thus we cannot reasonably assign any probability to the second premise of our argument from evil: neither high, nor low, nor middling. The probability remains, in this sense, inscrutable. But then our argument fails.

Wielenberg on divine lies

In his article ‘Skeptical theism and divine lies’ (2010), Erik Wielenberg points out what appears to be an interesting and, for many theists, deeply worrying, consequence of sceptical theism. If the fact that we cannot think of a divine justification for a given evil fails to justify the belief that no such justification exists, then, presumably, the fact that we cannot think of a justification for God lying to us fails to justify the belief that no such justification exists. So, if the sceptical theist is to apply her scepticism consistently, she should acknowledge that the probability that God is lying to us is similarly inscrutable. But then, according to Wielenberg, sceptical theism has the consequence that we cannot know any proposition that has word-of-God justification only. If the only reason I have for believing that p is that God asserts that p, but for all I know God is lying to me, then I do not know that p. Sceptical theism, according to Wielenberg, is thus ‘at odds with any religious tradition according to which there are certain claims that we can know to be true solely in virtue of the fact that God has told us they are true’ (ibid., 509).

Such claims appear to include, for example:

A Christian who, in response to the problem of evil, expresses scepticism about our ability to discern what reasons God might have to allow some evil, but, in response to God's utterances, fails to be similarly sceptical about our ability to discern what reasons God might have to lie to us about (L), would appear to be employing her scepticism selectively – in an inconsistent and partisan way. Once a theist employs sceptical theism to deal with the evidential problem of evil, it appears she cannot also rationally maintain belief in propositions that have word of God justification only, including (L). If, on the other hand, she drops the sceptical theism to avoid this unpalatable consequence, she again faces the evidential problem of evil.

This article examines responses made by defenders of sceptical theism to Wielenberg's argument. To my knowledge, the only published responses are to be found in McBrayer (Reference McBrayer2010), McBrayer & Swenson (Reference McBrayer and Swenson2012), and Segal (Reference Segal2011). I argue that all these responses either fail or else can be sidestepped by modifying Wielenberg's presentation of his argument, modifications I therefore recommend. I begin with Segal.

Segal's response to Wielenberg

The sceptical theist argues that if sceptical theism is true, then we cannot know, for any existent evil, that there is not, or is unlikely to be, an adequate God-justifying reason for it. Segal (Reference Segal2011) notes that Wielenberg presents a parallel argument for the conclusion:

As Segal points out, Wielenberg maintains that the premises of his parallel argument for LIES are at least as plausible as those of the anti-noseeum argument. Thus, concludes Wielenberg, if the sceptical theist accepts the conclusion of the anti-noseeum inference, then she should accept LIES too. In response, Segal argues that Wielenberg has not

I begin by noting that the conclusion Segal draws in the above-quoted passage does not follow. Suppose Segal is correct that the premises of Wielenberg's parallel argument are not as plausible as those of the anti-noseeum argument. Still, Wielenberg's premises might be eminently plausible. But then Wielenberg might still have established beyond reasonable doubt that sceptical theists must indeed choose between their argument and knowledge of propositions having word-of-God justification only.

However, as we will see, Segal believes he can show not just that the premises of Wielenberg's argument are less plausible than those of the anti-noseeum, but that some are implausible. If Segal is right about that, Wielenberg's argument does indeed fail.

The evolution of Wielenberg's argument

Wielenberg's argument has evolved. Consider, to begin with, the following two-premise version presented to the Fourth Philosophy of Religion conference at Baylor in 1989Footnote ² (let ST=sceptical theism):

1. (i) If ST is true, then, for any given divine assertion D, we have no good reason to deny that God's act of intentionally speaking falsely in making D has beyond-our-ken justification.

2. (ii) If, for any given divine assertion D, we have no good reason to deny that God's act of intentionally speaking falsely in making D has beyond-our-ken justification, then we do not know any proposition that has word-of-God justification only.

3. (iii) Therefore, ST implies that we do not know any proposition that has word-of-God justification only.

As it stands, the argument faces a fairly obvious objection: even if we have no reason to deny with respect to any divine assertion that God has good reason to lie, it doesn't follow that we cannot know any proposition that has word-of-God justification only. For perhaps we can still know that, even when God is justified in lying, he rarely if ever does so. Thus, even if sceptical theism is true, we might still know propositions having word-of-God justification only. Call this the Justified-But-Doesn't-Act (JBDA) objection.

By the time it was published in ‘Skeptical theism and divine lies’ (2010), Wielenberg had amended his argument to deal with the JBDA objection. The published version runs:

1. (1) If ST is true, then, for any divine assertion that p, we lack justification for believing that it is false or unlikely that God's act of intentionally asserting that p when p is false has beyond-our-ken justification.

2. (2) If, for any divine assertion that p, we lack justification for believing that it is false or unlikely that God's act of intentionally asserting that p when p is false has beyond-our-ken justification then we do not know p if p has word-of-God justification only (unless we have good reason for thinking that, even if God has some justification for lying about p, God doesn't act on that justification).

3. (3) So, ST implies that we do not know any proposition that has word-of-God justification only (unless we have good reason for thinking that, even if God has some justification for lying, God doesn't act on that justification).

4. (4) We do not have good reason for thinking that, even if God has some justification for lying, God doesn't act on that justification.

5. (5) Therefore, sceptical theism implies that we do not know any proposition that has word-of-God justification only.

The above version of Wielenberg's argument is clearly more complex than the original. It involves more premises. The second premise is also internally more complex: a parenthetical clause has been added. This additional complexity is obviously introduced in large measure to deal with the JBDA objection.

Segal (Reference Segal2011) considers the published version of Wielenberg's argument ambiguous and provides two readings. Here I consider only the second reading (that which Segal considers the more likely). Segal's presentation involves the following predicates: ‘Axp’=x is a divine assertion that p; Lxp=x is a divine lie that p; ‘LJGx’=there is some good g, such that if x were a divine lie, g would justify x. The argument runs:

1. (1b) If ST is true, then for any x and p such that Axp, we lack justification for believing that Pr(LJGx/Axp) is low.

2. (2b) If for any x and p such that Axp, we lack justification for believing that Pr(LJGx/Axp) is low, then for any divine assertion that p, we do not know that p if p has word-of-God justification only (unless we have good reason for thinking that, even if God has some justification for lying about p, God doesn't act on that justification).

3. (3) So, if ST is true, then we do not know any proposition p that has word-of-God justification only (unless we have good reason for thinking that, even if God has some justification for lying about p, God doesn't act on that justification).

4. (4) For any divine assertion that p, we do not have good reason for thinking that, even if God has some justification for lying about p, God doesn't act on that justification.

5. (5) Therefore, if ST is true, then we do not know of any proposition that has word-of-God justification only.

Segal goes on to target premises (1b) and (2b) of the above argument. He considers neither premise plausible. I turn first to Segal's claim that (2b) is implausible.

Segal's first criticism

Why does Segal consider (2b) is implausible? According to Segal, because it relies on:

But, says Segal, CLAIM is implausible, for it is an example of the schema: if we lack justification for believing that p and we lack justification for believing that q, then we lack justification for believing either p or q. This schema has obvious false instances (given I lack justification for believing p and also lack justification for believing not-p, it does not follow that I lack justification for believing either p or not-p). Thus, says Segal, Wielenberg cannot rely on CLAIM.

Notice that Segal's first criticism exploits some of the complexity Wielenberg introduced to deal with the JBDA objection. Specifically, it exploits Wielenberg's introduction of that parenthetical clause within the second premise.

I believe Segal's first objection is fairly easily sidestepped. The complexity Segal targets in premise (2b) can be dropped. Wielenberg can instead narrow down on the kind of reasons God might have to lie. Some reasons may be reasons such that though they would justify God in performing an action, it is by no means likely God will so act. Others reasons may such that, if they obtain, then it is highly likely – even guaranteed – that God will act on them.

The thought that there are reasons of the latter sort might well play a role in setting up our evidential argument of evil in the first place. The first premise is typically based on the thought that we can sure that God, if he exists, will act to prevent gratuitous evil. God, if he exists, has reason to prevent gratuitous evil, reason we can be sure he would act on. That's why the existence of gratuitous evil guarantees there is no God. Call reasons such that God is guaranteed to act on them AG reasons (act-guaranteeing reasons).

Now presumably Wielenberg will suppose that if sceptical theism is true, then not only can we not justifiably conclude that there are no reasons for God to perform a given action a given only that we cannot think of any, neither can we justifiably conclude that there are no AG reasons for God to perform action a given only that we cannot think of any. But then surely, given sceptical theism, there may, for all we know, be not just reasons, but AG reasons, for God to lie to us on a regular basis. But then, by recasting Wielenberg's original argument so that the focus is now specifically on AG reasons to lie, the JBDA objection can be dealt with without introducing the complexity targeted by Segal. Let ‘LAGRGx’=there is some good g, such that if x were a divine lie, g provides God with an AG reason to assert x. Then, otherwise retaining Segal's style of presentation, we might rework Wielenberg's argument so:

1. (1c) If ST is true, then for any x and p such that Axp, we lack justification for believing that Pr(LAGRGx/Axp) is low.

2. (2c) If for any x and p such that Axp, we lack justification for believing that Pr(LAGRGx/Axp) is low, then for any divine assertion that p, we do not know that p if p has word-of-God justification only.

3. (5) Therefore, if ST is true, then we do not know of any proposition that has word-of-God justification only.

Call this the LIES argument.Footnote ³ The LIES argument is structurally closer to Wielenberg's two-premise original than the published version. And, crucially, premise (2c) of the LIES argument is immune to Segal's first criticism (because it lacks that troublesome parenthetical clause).

However, the LIES argument won't quite do as it stands. Premise (1c) is false. Suppose God tells me that Paris is the capital of France. Then, notwithstanding sceptical theism, given that I nevertheless possess good, independent, non-word-of-God justification for supposing Paris is indeed the capital of France, I can reasonably conclude that God lacks AG reason to lie to me about this (for otherwise he would have done so). More generally, notwithstanding the truth of sceptical theism, if God asserts that p and we possess non-word-of-God justification for p, then we possess justification for believing that Pr(LAGRGx/Axp) is low. This flaw in the LIES argument is easily corrected, however, by tweaking its premises slightly. Here is one suggestion:

1. (1d) If ST is true, then for any x and p such that Axp and we lack non-word-of-God justification for p, we lack justification for believing that Pr(LAGRGx/Axp) is low.

2. (2d) If for any x and p such that Axp and we lack non-word-of-God justification for p, we lack justification for believing that Pr(LAGRGx/Axp) is low, then for any divine assertion that p, we do not know that p if p has word-of-God justification only.

3. (5) Therefore, if ST is true, then we do not know of any proposition that has word-of-God justification only.

Call this the LIES+ argument. The LIES+ argument is immune to Segal's first criticism. I suggest it has some prima-facie plausibility.

To summarize: Segal's first criticism of Wielenberg's argument targets a complexity introduced by Wielenberg to deal with the JBDA objection. However, it appears Wielenberg might easily deal with the JBDA objection by other means, thereby sidestepping Segal's criticism.

Segal's second criticism

Let us now turn to what I consider to be Segal's more substantive criticism. It turns on the thought that the sceptical theist's justification for believing what God asserts need not be rooted in any noseeum inference. The sceptical theist need not infer that, because he cannot think of a reason why God would lie about p, therefore there is unlikely to be a reason for God to lie about p. True enough, sceptical theism would indeed block that inference, and thus that way of justifying belief in God's pronouncements. But perhaps the sceptical theist might offer some other, independent justification for believing that p given only that God asserts that p? Segal illustrates as follows:

Segal then points out that, notwithstanding the truth of sceptical theism, if we do know that p, then, assuming premise (2b) is sound, we can use this knowledge to infer that either we have justification for believing that Pr(LJGa/Aap) is low or that we have justification for believing that God would not lie about p even if he had a justification. But given premise (4) of Segal's version of Wielenberg's argument, we lack justification for believing God would not lie about p even if he had justification. So we can in turn infer that premise (1b) is false.

This is an ingenious move. If the sceptical theist can know, for something like the reasons Segal outlines, that p based on no more than the divine assertion that p, then she can use other premises of Wielenberg's argument to infer that (1b) is false. I note a similar but more direct inference might be made to refute premise (1d) of the LIES+ argument: if I know that p given only that God asserts that p, then I can infer that the probability God has AG reason to lie about p is low.

Segal concludes that, before Wielenberg is entitled to hold (1b), he needs to justify the claim that, if sceptical theism is true, then none of these suggested accounts of how one might know p given only the divine assertion that p are adequate. And this is something Wielenberg has not done or even attempted to do.

As I explain below, this second criticism overlooks that fact that Wielenberg presumably considers sceptical theism to supply a defeater for beliefs having word-of-God justification only. Moreover, it appears that, even if Segal's various suggested justificatory grounds are invoked, this defeater remains effective.

Testimony and defeat

Consider the following simple principle of testimony. For any testifier t and proposition p:

This is an intuitively plausible principle. Perhaps it might be justified by induction over all assertions. Or perhaps, even if no justification can be provided, it constitutes a basic Reidian principle to which we are properly entitled. Perhaps (PT) is a principle of ‘common sense’ epistemology we can rightly just take for granted. Given (PT), does it follow that I can consider myself to know some proposition given only that someone – such as God – has asserted it?

Not necessarily. (PT)'s ceteris paribus clause is obviously intended to cover situations in which one has a defeater for one's beliefs that p and that one knows that p. So, for example, if I know, in a given instance, that t has both a track record of dishonesty and excellent reason to lie to me on this occasion, then it is no longer reasonable for me to suppose I know that p given only that t asserts p. Under these circumstances, I possess a defeater both for the belief p and the belief that I know that p.

Suppose Sally tells me that p. (PT) says that, ceteris paribus, it is now reasonable for me to believe both that p and that I know that p. But of course, I might yet come to possess a defeater for these beliefs. Suppose I go on to discover that Sally's assertion that p was a result of the following process. Sally drew a ball randomly from an urn. I have no idea whether or not all the balls in the urn are black, or none are black, or some fraction between 0 and 1 are black. What I do know (perhaps because I know Sally is currently involved in some elaborate game of dare) is that if Sally's ball was black, she told me a lie and if it was not black, she told me the truth.

So, setting aside the fact that Sally did in fact assert that p, the probability that Sally lies on this occasion is otherwise inscrutable to me. I cannot otherwise reasonably assign any probability to the claim that Sally lies – neither high, nor low, nor middling. Given this information, can I still reasonably believe both that p and that I know that p given only that Sally asserts that p?

Surely not. Why not? Because for all I otherwise know Sally lies when she asserts that p. The fact that the probability that Sally lies is otherwise inscrutable to me provides me with a defeater for both those beliefs.

But now suppose that, in response to the above objection, I attempt to defend my claim to know that p in the following manner. Given (PT) plus the fact that Sally asserts that p, it is reasonable for me to believe I know that p. But then (given that I know that p, and that I cannot know that p if the probability that Sally lies is not low) I am entitled to conclude that the probability Sally lies is not inscrutable, but low.

Clearly, such a response is muddled. It overlooks the fact that it appears to be a condition of my knowing that p given only that t asserts that p that the probability that t lies is not otherwise inscrutable.

Consider a similar condition. It seems clear that, notwithstanding (PT) and the fact that t asserts p, if I know the probability that t lies on this occasion is otherwise high, then I cannot reasonably believe I know that p (given only that t asserts p). For I now possess a defeater for that belief. Notice that I certainly cannot reasonably infer that, given both (PT) and that t asserts p, then the probability that t lies is actually not high but low.

But then it also seems clear that, notwithstanding (PT) and the fact that t asserts p, if the probability that t lies on this occasion is otherwise inscrutable to me, then, again, I cannot reasonably believe I know that p (given only that t asserts p). Again, it seems I possess a defeater. That's why my discovering the backstory to Sally's assertion provides me with a defeater for what she asserts.

Now consider God's divine assertion that p. Does (PT) allow me reasonably to believe that I know that p given only that divine assertion? Not, it seems, if I also know that sceptical theism is true. For then I know that the probability that God has AG reason to lie on this occasion, and thus lies, is indeed otherwise inscrutable. If my only reason for believing p is that God asserts that p, then sceptical theism provides me with a defeater for both my belief that p and my belief that I know that p.Footnote ⁴

So, even if (PT) can be justified by induction over all assertions and/or even if (PT) can rightly be accepted with justification, it appears that, given sceptical theism, I cannot reasonably believe propositions having word-of-God justification only.Footnote ⁵

Note that the above point concerning defeaters for beliefs that would otherwise be justified by (PT) suggests an alternative argument for Wielenberg's conclusion. It appears, intuitively, that something like the following principle of defeat is correct. For any proposition p and testifier t:

I take it that something like (D) was presupposed by Wielenberg even in his original presentation (presumably, something like (D) lies behind Wielenberg's second premise, and something like (D) might in any case be used to justify premise (2d) of the LIES+ argumentFootnote ⁶). (D)'s role can be made explicit in a Wielenberg-style argument by combining it with (as a second premise):

In the absence of any defeater-defeater, Wielenberg can now conclude:

The second premise can, in turn, be supported by:

Call this the DEFEATER argument. (NB the parenthetical clauses are required because, as we noted earlier, if we have independent reason to think p is true, then we have independent reason to think God does not lie, and thus independent reason to think God lacks AG reason to lie.)

Of course, in response to the DEFEATER argument, the sceptical theist might attempt to supply some defeater-defeater. Perhaps there is some consideration she can point to that has the consequence that the defeater generated by her sceptical theism for propositions having word-of-God justification only is itself defeated. But, assuming the DEFEATER argument is sound, the onus is now very much on the sceptical theist to supply such a defeater-defeater. Until she does so, her belief in propositions having word-of-God justification remains defeated.

In short, it appears that, given (D) and a plausible no-defeater condition on knowledge, sceptical theists cannot know propositions having word-of-God justification only.Footnote ⁷

Induction over divine assertions

What of Segal's suggestion that we might still justify belief in God's assertion that p by induction, if not over all assertions, then at least over divine assertions? Of course, this would require that we have grounds for believing God tells the truth on at least some occasions. But where we have independent reason to believe that what God has asserted is true, we do possess such grounds. Arguably, this is the case for some, perhaps even many, divine assertions. So, if we know that God has not lied when it comes to those assertions we can independently verify, then we know God lacks AG reason to lie on those occasions (for he would otherwise have lied). And if we know God lacks AG reason to lie on those occasions, why can we not inductively generalize and justifiably conclude that he is unlikely to possess AG reason to lie on those occasions when he makes assertions the truth of which we cannot independently verify?

Unfortunately for the sceptical theist, this inference is also blocked by his sceptical theism.Footnote ⁸

Suppose God makes ten assertions of which we are able independently to verify that seven are true. Given God's established track record of honesty, are we not justified in believing his remaining three assertions? Not, it appears, if sceptical theism is true. For consider the possibility that God has AG reason to deceive us by lulling us into a false sense of security, much as confidence tricksters do. Given sceptical theism, it won't do to argue that if we are unable to think of an AG reason for God to engage in such deception then there's unlikely to be such a reason. But then how are we able reasonably to believe that the probability that such an AG reason exists is low?

Suppose Bert makes ten assertions, seven of which I am able independently to verify are true, the remaining three of which I am otherwise in the dark about. Can I reasonably believe the remaining three assertions? Ceteris paribus, of course I can. But now suppose I discover that, for all I know, Bert is a scammer engaged in a confidence trick of which I am the unwitting victim. For all I know, my verification of Bert's other seven assertions is intended by Bert to play a confidence-building role in his deception. The probability that Bert is engaged in such a deception is otherwise inscrutable to me. I can reasonably assign no probability to Bert's thus deceiving me: neither high, nor low, nor middling. Surely, given the inscrutability of this probability, I possess a defeater for those remaining three beliefs, notwithstanding the fact that I have independently verified the truth of Bert's other seven assertions.

But then similarly, given the inscrutability of the probability that God has AG reason to engage in a similar deception, we possess a defeater for those of God's assertions we have not independently verified, notwithstanding the supposed fact that we have been able to verify some or even most of God's other assertions.

The relevance of God's moral perfection

Segal makes one other suggestion regarding how we might come to know that p given only that God asserts that p. As Segal points out in the above quotation, God is not any old testifier. He is a morally perfect testifier. But then, given the wrongness of lying, perhaps the sceptical theist can still reasonably believe God is unlikely to lie even when he has reason to? And thus the sceptical theist can still reasonably believe propositions having word-of-God justification only?

There may be some plausibility to the suggestion that, when it comes to non-AG reasons to lie, a morally perfect God is unlikely to act on them. However, the LIES+ and DEFEATER arguments both appear immune to the above line of attack. That is because both arguments focus specifically on AG reasons to lie. Perhaps the fact that God is a morally perfect being should lead us to assign a low probability to God's lying when he has non-AG reason to lie. However, that same fact should vastly increase our confidence that God lies if there is a reason for God to lie such that lying is morally the right thing to do (and why shouldn't there be?). We can be confident God acts on such reasons not in spite of but because of his morally perfect nature. Indeed, trivially, if there are AG reasons for God to lie, then he is guaranteed to lie. And it appears that, if sceptical theism is true, then (except when there is some independent reason to think that what is asserted is true) the probability that there exist AG reasons for God to lie is inscrutable. But then, given (D), sceptical theism provides a defeater for propositions having word-of-God justification only, notwithstanding God's moral perfection.

Conclusions regarding Segal's critique of Wielenberg

I draw the following conclusions regarding Segal's objections to Wielenberg's argument (as Segal presents it above).

Segal's first criticism targets premise (2b). More specifically it exploits a complexity introduced by Wielenberg in order to deal with the JBDA objection. I have explained that there is another way of developing Wielenberg's original argument to avoid the JBDA objection that sidesteps Segal's first criticism (the LIES+ argument) yet still produces a plausible-looking argument.

Segal's second criticism involves his outlining various justifications that may be open to theists for believing propositions having word-of-God justification only – justifications Segal believes Wielenberg must rule out before he can claim to have shown that sceptical theism undermines knowledge of such propositions. I believe this second criticism fails. Segal's suggested justifications all run into the same fundamental problem: that, even granted (i) the plausibility of (PT), (ii) God's supposed track record of telling the truth, and (iii) the fact that God is morally perfect, etc., sceptical theism nevertheless appears to supply a defeater for propositions having word-of-God justification only. That sceptical theism supplies such a defeater is, I take it, Wielenberg's fundamental point.Footnote ⁹ In which case, Segal fails to deal with that point.

Perhaps many theists do reasonably believe propositions possessing word-of-God justification only. Perhaps they do quite properly justify their belief in such propositions in the ways sketched out by Segal. However, if such a theist goes on to embrace sceptical theism in order to try to deal with the problem of evil, then, notwithstanding the fact that she happens to employ one or indeed all of Segal's various suggested justificatory grounds, her sceptical theism nevertheless appears to supply her with a defeater for beliefs having word-of-God justification, just as Wielenberg supposes. Indeed, it appears that, given this defeater (and, of course, no defeater-defeater), it is irrational for such a sceptical theist to continue to maintain she knows propositions having word-of-God justification only.

The LIES+ and DEFEATER arguments appear immune to Segal's objections.Footnote ¹⁰ Of course, it remains open to Segal to argue that the premises of the LIES+ and DEFEATER arguments are implausible, or at least not as plausible as those of the sceptical theist's anti-noseeum argument. However, even if Segal could show that the premises of the LIES+ and DEFEATER arguments are not as plausible as those of the anti-noseeum, they might still successfully justify Wielenberg's conclusion. Suppose it is highly plausible that the premises of either the LIES+ or DEFEATER argument are all true. Then, even if those premises happen not to be quite as plausible as those of the anti-noseeum, they might still establish beyond reasonable doubt that sceptical theism generates a defeater for propositions having word-of-God justification only.

The McBrayer/Swenson response to Wielenberg

I turn now to the criticism of Wielenberg's argument presented in McBrayer (Reference McBrayer2010) and then McBrayer & Swenson (Reference McBrayer and Swenson2012). McBrayer and Swenson maintain that, for religiously mainstream monotheists who employ sceptical theism to deal with the problem of evil, Wielenberg's argument ‘is not as scary as it first appears’ (McBrayer & Swenson (Reference McBrayer and Swenson2012), 148). They begin by acknowledging that the sceptical theist should grant the possibility of divine lies:

McBrayer and Swenson then suggest that, given that we at least know that, ceteris paribus, God would tell us the truth, we can remain justifiably confident about the truth of his pronouncements, just as we can remain justifiably confident about the pronouncements of other people even while acknowledging the possibility that they are lying:

Why the McBrayer/Swenson response to Wielenberg fails

This response to Wielenberg also fails. To begin with, let us examine McBrayer and Swenson's ceteris paribus claim about what God would do. The claim is:

How should (G) be understood? Ceteris paribus claims often take the form of generalizations that license predictions. Consider:

On perhaps the most natural reading of (T), the claim is that, as a general rule (usually, setting aside a few exceptions), cats do live more than six years. Thus understood, (T) licenses predictions. It allows me justifiably to conclude that my cat Tiddles will, or will probably, live more than six years (assuming, of course, that I have no reason to believe or suspect that the ceteris paribus clause applies).

However, ceteris paribus claims do not always license predictions. Consider:

Given its subjunctive mood, (J) would not ordinarily be understood to license the prediction that, as John is home, he is, or is probably, naked. The suggestion, presumably, is not that, as a general rule (setting aside a few exceptions) John is naked at home. (J) allows for the possibility – and might well be taken to imply – that other things rarely, if ever, are equal. Perhaps, though being naked is John's strong preference, John does not live alone and, out of courtesy to his easily offended cohabitees, he usually remains clothed.

Now consider McBrayer and Swenson's (G). How should it be understood? Like (J), it is subjunctive in mood, speaking not of what is the case, but of what would be the case. Thus, on its most natural reading, (G) does not assert or imply that as a general rule God tells the truth. Even granted (G), God's telling the truth may be the exception rather than the rule.

But then, thus understood, it's hard to see how (G) provides McBrayer and Swenson with a basis for an effective response to Wielenberg. (G) no more justifies our believing that, as God asserts that p, p is true than (J) justifies our believing that, as John is home, John is naked.

But perhaps, appearances to the contrary, McBrayer and Swenson intend (G) to be understood as asserting or supporting a generalization about what God does: as a general rule God tells the truth. But if that is how (G) should be read, then, given sceptical theism, it is hard to see how McBrayer and Swenson can know that it is true. For sceptical theism does not merely have the consequence acknowledged by McBrayer and Swenson: that it is possible that God has AG reason to lie about things that we otherwise lack reason to think are true (a possibility that is consistent with our nevertheless having good reason to think that the probability he lies is not inscrutable or low). Sceptical theism also appears to have the far more serious consequence that the probability that there are such AG reasons is inscrutable. And, given the principle of defeat (D), that in turn provides the sceptical theist with a defeater for beliefs having word-of-God justification.

What of the suggestion made in the second quotation above? McBrayer and Swenson suggest that it is, ceteris paribus, reasonable to accept the testimony of other people. Thus it is reasonable to accept the testimony of God.

This manoeuvre was dealt with earlier. McBrayer and Swenson are here appealing to something like the principle of testimony (PT). Perhaps that principle is correct. However, (PT) does not allow us reasonably to believe propositions having word-of-God justification if sceptical theism nevertheless supplies a defeater for such beliefs, which it appears it does.