paying the cost of skeptical theism

Int J Philos Relig (2011) 69:45–56DOI 10.1007/s11153-010-9235-8

Paying the cost of skeptical theism

Jeff A. Snapper

Received: 3 December 2008 / Accepted: 16 November 2009 / Published online: 13 July 2010© Springer Science+Business Media B.V. 2010

Abstract In this paper I show that two arguments for the inconsistency of skepticaltheism fail. After setting up the debate in “Introduction” section, I show in “Theinitial debate” section why Mylan Engel’s argument (Engel 2004) against skepticaltheism does not succeed. In “COST” section I strengthen the argument so that it bothavoids my reply to Engel and parallels Jon Laraudogoitia’s argument against skepticaltheism (Laraudogoitia 2000). In “COST*” section, I provide three replies—one by anevidentialist theist, one by a closure-denying theist, and one by a necessitarian theist,and argue that the necessitarian’s reply successfully rebuts the inconsistency charge.I conclude that skeptical theism which accepts God’s necessary existence is immuneto both kinds of arguments for its inconsistency.

Keywords Problem of Evil · Skeptical theism · CORNEA · Evidential atheism ·Epistemic closure · Counterfactuals

Introduction

In this paper I show that two arguments for the inconsistency of skeptical theism(theism which is skeptical of inductive inferences from evil to atheism) fail.1 After

1 The locus classicus of the skeptical theism I address here is Wykstra (1984), but there are others. WilliamAlston’s skeptical theism makes the modest claim that “examining the interconnections of good and evilin the world by our natural powers cannot suffice to establish either [that there is gratuitous evil] or itsnegation” (Alston 1991, reprinted in Howard-Snyder 2004, p. 99). Although he does not advocate skepticaltheism per se, Peter Van Inwagen says things akin to it. In particular, he claims to tells a story S such thatan audience of ideal neutral agnostics would, if they heard it, agree that (i) S entails that God exists andhuge amounts of animal suffering occurred, (ii) for all they know S is true, (iii) no particular probabilityassignment to S is more epistemically defensible than any other. Further, that ideal audience would agree that

J. A. Snapper (B)Department of Philosophy, University of Notre Dame, 100 Malloy Hall, Notre Dame, IN 46556, USAe-mail: [email protected]

123

46 Int J Philos Relig (2011) 69:45–56

setting up the debate, I show in “Introduction” section why Mylan Engel’s argumentagainst skeptical theism does not succeed. In “The initial debate” section I strengthenthe argument so that it both avoids my reply to Engel and parallels Jon Laraudogoitia’sargument against skeptical theism. In “COST” section, I provide three replies—oneby an evidentialist theist, one by a closure-denying theist, and one by a necessitariantheist—and argue that the necessitarian’s reply successfully rebuts the inconsistencycharge. I conclude that skeptical theism which accepts God’s necessary existence isimmune to both kinds of arguments for its inconsistency.

The initial debate

The evidential atheism-skeptical theism debate uses a number of terms of art. Gratu-itous evil is evil which God is not morally justified in allowing.2 The obtaining of agood state of affairs G morally justifies God in allowing an evil E if and only if (i) Godcan bring about G only by allowing E and (ii) G is more valuable than, or prevents anevil at least as bad as, E.3 Call a good that morally justifies God in allowing an evila God-justifying good. The difference between gratuitous and non-gratuitous evil isontic. In contrast, the difference between scrutable and inscrutable evil is epistemic.Inscrutable evil is evil for which we cannot find a God-justifying good. In these terms,the debate is over whether awareness of inscrutable evil(s) makes it rational to believethere is gratuitous evil.

Here is an argument for the rationality of atheism from inscrutable evil.4 Considera fawn’s suffering following a forest fire. Because we are not able to discern a God-justifying good for it, the suffering is an inscrutable evil. Further, some inscrutableevils are such that, our inability to discern God-justifying goods for them makes itrational to believe they lack such goods. That is,

(1) There are inscrutable evils the seeing of which makes it rational to believe theyare gratuitous.

Next, suppose S sees an inscrutable evil that satisfies (1) so that it is rational for S tobelieve there is gratuitous evil. Since (i) S knows that if there is gratuitous evil there is

Footnote 1 continuedgiven (i)–(iii), any argument for God’s nonexistence from evil cannot move ideal neutral agnostics to assignits conclusion a higher probability than they assigned it prior to considering the argument; consequentlyany such argument is not a success (Van Inwagen 2006, chap. 7, esp. 114)2 Equivalently, gratuitous evil is evil for which there is no God-justifying good. Analogously, immunityto measles is a “parent-justifying good” of immunization because it morally justifies parents in submittingtheir children to the pain of injections.3 A possible worry here is that one should add to the definition of a God-justifying good a clause to theeffect that G is a good that is worth it for God to instantiate/cause to obtain. Perhaps, absent such a clause,the definition looks to imply that God is morally required to instantiate every good, no matter how trivial.But all the definition implies (in conjunction with God’s nature) is that God is morally required, by Hisnature, to instantiate enough sufficiently valuable goods to morally justify the evils He allows—which iscertainly a proper subset of all the goods there could be.4 I model it roughly after Rowe (1979).

123

Int J Philos Relig (2011) 69:45–56 47

no God,5 and (ii) rational belief is closed under known implication,6 it is rational forS to believe that there is no God.7 The argument’s key premise is (1), which implies

(2) There are inscrutable evils the seeing of which makes it more rational to believe(increases the rationality of believing) they are gratuitous.8

If seeing that an evil E is inscrutable makes it rational (simpliciter) to believe thereis gratuitous evil, then seeing that E is inscrutable makes it more rational to believethere is gratuitous evil. This is because of the more general fact that if seeing X makesit rational simpliciter to believe p then seeing X makes it more rational to believe p.If seeing storm clouds approach makes it rational (simpliciter) to believe it will rain,then so seeing likewise makes it more rational to believe it will rain.9

Skeptical theism’s beef is with (2). Stephen Wykstra argues against (2) using hisCondition of Reasonable Epistemic Access (CORNEA) conjoined with a skepticismabout the sorts of goods it is rational for a theist to expect to be able to see.10 CORNEAis an epistemic principle governing claims of appearance—i.e., when a person is andis not epistemically entitled to claim ‘it appears that p’. It states:

On the basis of cognized situation s, human H is entitled to claim ‘It appears thatp’ only if it is reasonable for H to believe that, given her cognitive faculties andthe use she has made of them, if p were not the case, s would likely be differentthan it is in some way discernible by her.11

Wykstra has clarified the meaning of CORNEA in response to criticism.12 Its “Onthe basis of s, H is entitled to claim ‘It appears that p”’ locution is meant to be under-stood, says Wykstra, dynamically. As such, CORNEA governs changes only in therelative degree of rationality which ‘appears that’ claims enjoy. For instance, supposeI say ‘it appears that we are in a recession’, based on my general background knowl-edge which is severely impoverished because all I do is read philosophy. Let n be the

5 This follows from the definition of gratuitous evil and God’s being perfectly morally justified in all Hisactions if God exists. Still, note that some philosophers deny this premise—most notably Peter Van Inwagen.6 “Rational belief is closed under known implication” means, for any subject S and propositions p and q,if it is rational for S to believe p and S knows that p implies q, then it is rational for S to believe q.7 There are certainly many and varied objections to closure involving many and varied logically possibleradically skeptical states of affairs the obtaining of which preclude the obtaining of commonsensible statesof affairs we all pretheoretically accept. This is not a paper on closure, but even if it were we could justpretend to be Mooreans for the duration of this argument. Whether rational belief is closed under knownimplication, and if so which formulation is the correct one, are both worthy topics outside this paper’s scope.8 Stephen Wykstra clarifies the distinction between making rational (simpliciter) to believe p and makingmore rational to believe p in (Wykstra 2007).9 Note that (2) does not imply (1). After all, seeing a terrible murder might make it more rational for me tobelieve there is gratuitous evil without making it rational (simpliciter) for me to so believe. I could have,prior to my seeing the murder, have had lots of evidence for the nonexistence of gratuitous evil, e.g., lotsof cogent arguments for God’s existence, lots of strong religious experiences, etc., that made it irrational(simpliciter) to believe in gratuitous evil. If so, then seeing the murder, though it might make it more rationalto believe in gratuitous evil, might nonetheless fail to make it rational simpliciter to believe in gratuitousevil, because it could fail to override all my positive evidence against it.10 Wykstra (1984).11 Ibid.12 Wykstra (2007).

123


degree to which I am epistemically entitled to make that ‘appears that’ claim. Nowsuppose I go read the last year’s worth of The New York Times, thereby gaining moreevidence that we are in a recession. Letting s = my cognized readings and p = we arein a recession, CORNEA says that, after reading the Times, my epistemic entitlementto the ‘appears that’ claim is greater than n only if it is rational for me to believe that ifwe were not in a recession I would notice a difference in what the Times said about theeconomy. Since it is rational for me to believe this, reading the Times does increase myepistemic entitlement to the ‘appears that’ claim. Nonetheless, despite the epistemicbump I glean from my readings, my later ‘appears that’ claim could remain irrationalsimpliciter. So CORNEA governs only relative changes in rationality.13

To streamline discussion, I propose a paraphrase of CORNEA—principle C, whichuses the intellective sense of ‘seeing’ and states that, for any obtaining state of affairsp and any proposition q,

C Seeing that p makes it more rational to believe that q only if it is rational to believethat if it were not the case that q, then probably p would seem different.14

Given Wykstra’s recent clarification of CORNEA, C retains CORNEA’s essenceand brings out its dynamic emphasis. Like CORNEA, C states that adding a cognizedsituation (seeing that p) to one’s noetic structure increases the relative rationality ofbelieving a proposition (that q) only if it is rational simpliciter to believe the appropriatesubjunctive conditional.

Using C, let us conclude the initial debate. The skeptical theist points out that,according to C, seeing that an evil E is inscrutable makes it more rational to believethat E is gratuitous only if it is rational to believe that if E were not gratuitous, prob-ably E would seem different (i.e., probably E would be scrutable). Further, the theistbelieves that there is an omniscient, omnipotent, all-good God, and that His purposesfor allowing evil are beyond our ken—hidden from human view. If so, then for anyevil E, regardless of whether E were gratuitous or not, the theist has good reason tobelieve it would probably remain inscrutable. Therefore, it is not rational for a theist

13 That CORNEA and its paraphrase C, govern only relative changes in rationality does not imply thatthey govern only inductive inferences. CORNEA governs any inference whose conclusion takes the form“On the basis of cognized situation s, human H is entitled to claim ‘It appears that p”’. C governs anyinference whose conclusion takes the form ‘Seeing that p makes it more rational to believe that q’. Sinceboth inductive and deductive inferences can take conclusions of those forms, C and CORNEA govern bothinductive and deductive inferences. It is better to say that CORNEA and C place necessary conditions on thetruth of some varieties of epistemic propositions—namely, epistemic propositions of the form ‘On the basisof cognized situation s, human H is entitled to claim ‘It appears that p” and ‘Seeing that p makes it morerational to believe that q’, respectively. In contrast, C does not govern any inference whose conclusion takesthe form ‘Seeing that p makes it rational simpliciter to believe that q’. Both CORNEA and C cross-cut theinductive-deductive distinction because they are epistemic, not logical, principles of inference-governance.14 One might worry about appropriately substituting into C. Here is an illustration to allay that worry.Italicizing states of affairs and putting propositions in brackets, if we let p = the house is blue and q = [thehouse is blue], C provides “Seeing that the house is blue makes is more rational to believe that [the houseis blue] only if it is rational to believe that if it were not the case that [the house is blue], then probably thehouse is blue would seem different.” We can make the reading of that substitution instance of C even moreexplicit as follows: “Seeing the state of affairs consisting in the house’s being blue makes it more rationalto believe the proposition [the house is blue] only if it is rational to believe that if the proposition [the houseis blue] were not true then probably the state of affairs consisting in the house’s being blue would seemdifferent.” So there is no problem appropriately substituting into C.

123


to believe of an inscrutable evil that if it were not gratuitous, probably it would bescrutable. Consequently, by C, there is no inscrutable evil the seeing of which makesit more rational to believe there is gratuitous evil, contra (2).15

COST

Mylan Engel argues that skeptical theism is inconsistent. Engel calls his argumentthe Cost of Skeptical Theism (COST). In an important sense, Engel’s strategy hereis burden-of-proof shifting. He points out that theism implies there is no gratuitousevil. Given that implication, a successful defense of theism must do more than merelyshow that there is no good argument for the existence of gratuitous evil. Rather, afull defense of theism must go the next mile and provide a good argument for thenonexistence of gratuitous evil. Keeping this strategy in mind sheds lights on COST,which he formulates as follows.16

The theist claims to rationally believe that God exists. The theist knows that ifGod exists then there is no gratuitous evil. Rational belief is closed under knownimplication. So, the theist must also claim that it is rational to believe that there is nogratuitous evil. The skeptical theist claims that God’s deep purposes for evil are hiddenfrom human view, so that seeing an inscrutable evil does not make it more rationalto believe there is gratuitous evil. If seeing an inscrutable evil does not make it morerational to believe there is gratuitous evil, then seeing inscrutable evil likewise doesnot make it more rational to believe there is no gratuitous evil. But, argues Engel,

(3) If seeing inscrutable evil does not make it more rational to believe there is nogratuitous evil, then it is not rational for the theist to believe there is no gratuitousevil.

After all, explains Engel, “Inscrutable evil is doubly inscrutable-we can’t scrut that ithas a divine purpose and we can’t scrut that it doesn’t—and because we can’t do theformer, it is not rational to believe [that there is no gratuitous evil].”17

I take Engel’s point to be this. By examining an inscrutable evil, we cannot telleither (a) that it is gratuitous or (b) that it is not gratuitous. Since, when we examineinscrutable evil we cannot discern (b), it is not rational to believe that there is no gratu-itous evil. Therefore, it is not rational for the theist to believe there is no gratuitous evil.But, of course, the skeptical theist claims it is rational to believe there is no gratuitousevil. Consequently,

(4) It is both rational and not rational for the skeptical theist to believe that there isno gratuitous evil.

15 The same point is made by appreciating the epistemic difference between ourselves and God. Godknows every God-justifying good serving every inscrutable evil. Due to our limited epistemic powers,humans should expect to be able to find a God-justifying good serving at most, very few of the non-gratu-itous evils they perceive. Hence, it is not rational to believe that if a particular evil were not gratuitous thenprobably we would be able to identify a God-justifying good serving it. Therefore, for any evil E, seeingthat E is inscrutable does not make it more rational to believe that E is gratuitous.16 Engel (2004).17 Ibid., p. 412.

123


Thus, skeptical theism is inconsistent.But COST fails because (3) is false. That examination of inscrutable evil fails to

make belief in the nonexistence of gratuitous evil rational does not imply that nothingcan make it rational. To illustrate,18 suppose that in examining a needle, Jones cannotdiscern either (a′) that there are no viruses on it or (b′) that there are viruses on it. Itsurely does not follow that Jones is rationally required to be agnostic with regard to theexistence of viruses on that needle. Jones could plug in his microscope, take a closerlook, and settle the issue. Likewise, there may be some method, other than examiningthe inscrutable evil, by which a theist could obtain rational support for her belief thatGod is morally justified in allowing the inscrutable evils she encounters. Examinationof arguments for God’s existence and religious experiences are both means by whicha theist could obtain rational support for her belief that there is no gratuitous evil,neither of which includes examining inscrutable evil.

So COST’s (3) is false because it fails to take into account the fact that a singleproposition can be (dis)believed from, or one the basis of, a variety of distinct evi-dence sets. Consequently, when examining the rationality of a belief (in this case therationality of the belief that there is no gratuitous evil), the particular basis upon whichit is believed needs to be taken into account.19

COST*

Let us incorporate this insight into the argument and see if what results causes a prob-lem for skeptical theism. The problem with COST was that it failed to distinguishbetween distinct evidence bases for holding a belief. So a strengthened version ofCOST, COST*, asks upon what basis a typical skeptical theist believes that there isno gratuitous evil. As COST rightly points out, it cannot be the theist’s cognization ofinscrutable evil. So it must be upon some other basis. And what would that other basisbe besides the theist’s evidence for God’s existence? Construing evidence broadlyto include religious experiences and arguments, let us suppose that skeptical theistsdeny the existence of gratuitous evil on the basis of their evidence for God’s existence(“God Evidence” for short).20

Using this insight, COST* proceeds in three steps. It first uses C against the skep-tical theist. It then recalls the skeptical theist’s skepticism about her ability to identifyGod-justifying goods. Finally, it claims that an instance of C and the theist’s skepticismabout her powers of good-identification generate an inconsistency.

18 I borrow this example from Wykstra.19 So one key issue at stake here is whether the overall evidence for God’s existence exceeds the overallevidence for the existence of gratuitous evil.20 “God Evidence” is shorthand for the conjunction of the state of affairs a theist has for believing thatGod exists. Examples of states of affairs conjoined in theist Tim’s God Evidence are that there seems tobe good evidence for Jesus’ resurrection, that I (Tim) have read and thought hard about arguments for andagainst God’s existence and overall, it seems that those in favor of God’s existence are stronger than thoseagainst it, that I know lots of smart intellectually honest people who retain their theism despite enduringespecially nasty suffering, that it seems obviously true that if God exists then God necessarily exists, thatI seem to consistently talk and share my life with a being whose properties line up with those ascribed toGod by Scripture. Notice that no part of God Evidence entails that God exists.

123


Consider a skeptical theist who both regards some event as an inscrutable evil anddenies the existence of gratuitous evil. Given the above considerations, the skepticaltheist holds that

(5) Seeing that God Evidence makes it more rational to believe that there is no gra-tuitous evil.

Because skeptical theism accepts C, we can test the truth of (5) by letting p = GodEvidence and q = there is no gratuitous evil, in C. This produces

(6) Seeing that God Evidence makes it more rational to believe that there is no gra-tuitous evil only if it is rational to believe that if there were gratuitous evil thenprobably God Evidence would seem different.21

(5) and (6) imply

(7) It is rational to believe that if there were gratuitous evil, probably God Evidencewould seem different.

Given the skeptical theist’s skepticism about powers of good-identification, she iscommitted to the thesis that

(8) It is not rational to believe that if an evil were not gratuitous then probably wewould be able to identify a God-justifying good for it.

Finally, argues COST*, if the skeptical theist accepts (8)—if she strongly doubts herability to identify goods for non-gratuitous evils—then she must deny (7). Since theskeptical theist denies that there is any gratuitous evil, to evaluate (7) she must inquireas to the rationality of believing, of a closest possible (non-actual) world W containinggratuitous evil, whether, in W, her God Evidence would seem different. This requiresthat she keep everything in W as close to everything in the actual world as she can;according to her, the only difference between the actual world and W is the fact thatW contains gratuitous evil, as well as any consequences thereof. Now recall that inthe actual world, she forthrightly admits that she cannot tell, by looking for a goodserving it, whether an evil is gratuitous or not. This rules out there being epistemicconsequences in W of the sort that would enable her to distinguish W from the actualworld on their basis. Hence, though, according to her, the actual world and W in factdiffer in that the latter contains gratuitous evil while the former does not, nonetheless,to her, the evil in W looks exactly like the evil in the actual world. Since the existenceof that gratuitous evil makes no epistemic difference to her, she cannot claim that it isrational for her to believe that, if there were gratuitous evil, probably her God Evidencewould seem different; by (8) her God Evidence would seem exactly the same. Andhere the skeptical theist finds herself in a pickle. Per her original reply to the evidentialargument for atheism, the skeptical theist must accept (8). By (5) and C the skeptical

21 Presumably, the God Evidence would seem different in that it would provide less evidence for God’sexistence. Further, “Seeing that God Evidence” means seeing the obtaining state of affairs composed of thecombination of all the see-er’s evidence for God’s existence. It need not include seeing that all the argumentstherein are sound, or that all the experiences or testimonies therein are verdical. Rather it need include onlythat seeing that all the arguments therein seem sound and that all the experiences and testimonies containedtherein seem veridical.

123


theist is also committed to (7). But as COST* claims to show, the skeptical theist whoaccepts (8) must deny (7). Therefore, skeptical theism is inconsistent.

Jon Laraudogoitia’s argument22 against CORNEA is similar enough to COST* tomerit discussion here. Instead of invoking God Evidence in favor of the non-existenceof gratuitous evil, Laraudogoitia supposes the theist takes his cognization of the actualworld to be the basis upon which he makes the ‘it appears that there is no gratu-itous evil’ claim. Laraudogoitia argues that, instead of successfully defending theism,CORNEA implies agnosticism.

His argument occurs in three steps. First, in CORNEA he lets p = there is nogratuitous evil and s = the actual state of the world, producing

(9) On the basis of cognized situation [the actual state of the world], human H isentitled to claim ‘It appears that [there is no gratuitous evil]’ only if it is reason-able for H to believe that, given her cognitive faculties and the use she has madeof them, if [there is no gratuitous evil] were not the case, [the actual state of theworld] would likely be different than it is in some way discernible by her.

Second, given the skeptical theist’s skepticism about her ability to detect God-justify-ing goods for non-gratuitous evils, “it is not probable”, writes Laraudogoitia, if therewere gratuitous evil “that the actual state of the world would be different from the onethat really is in some way discernible by H.”23 Consequently,

(10) “It is not reasonable for H to believe that, given her cognitive faculties and theuse she has made of them, if p were not the case, s would likely be different thanit is in some way discernible by her.”24

Third, since (10) appropriately substituted is the negation of (9)’s consequent, itfollows that

(11) On the basis of cognized situation [the actual state of the world], human H is notentitled to claim ‘It appears that [there is no gratuitous evil]’

Since it reasonable for a person to believe that God exists only if that person is entitledto the claim that ‘it appears that there is no gratuitous evil’, CORNEA, appropriatelysubstituted, implies that theism is unreasonable. Thus, argues Laraudogoitia, COR-NEA is fuel for agnosticism.

Reply to COST* and Laraudogoitia

There are at least three replies open to the skeptical theist.First, an evidential skeptical theist could reply to COST* by doing a little more

imagining about how a world containing gratuitous evil would look different fromthe actual world. This kind of theist might think that, in a possible world containinggratuitous evil (despite his ignorance of the existence of that evil) he would remainan evidential theist. Such a theist would take textual evidence for things like Jesus’s

22 Laraudogoitia (2000).23 Ibid., p. 84.24 My ‘(10)’ is Laraudogoitia’s ‘(13)’. I change the number here for clarity’s sake.

123


resurrection and the disciples’ testimony to probabilistically support God’s existence,and would take the absence of such evidence to probabilistically undermine God’sexistence. But, this evidential theist also knows that any world containing gratuitousevil does not contain God, and hence (probably) Jesus does not resurrect nor (prob-ably) do the disciples testify about his resurrection, and hence (probably) there is notextual evidence for those events in a world in which they (probably) do not occur.Therefore, probably his God Evidence would seem different in a world containinggratuitous evil. Due to the wealth of “probably”’s peppering this reply, I am not fondof it, but will not argue against it.

Second, a skeptical theist might try to avoid COST* by denying (5) and instead arguedirectly for the nonexistence of gratuitous evil from God’s existence.25 And without(5), COST* cannot generate an inconsistency. But this reply is problematic. To seewhy, suppose a skeptical theist T denies (5) and argues directly for the nonexistenceof gratuitous evil from God’s existence as follows:

Simple(S1) (T’s seeing his God Evidence makes it rational to believe that) Godexists(S2) (T knows that) If God exists then there is no gratuitous evil(S3) Therefore, there is no gratuitous evil

The problem with Simple arises when we add a closure principle stating, for anyperson S, obtaining state of affairs p and propositions q and r,

Closure 1 If [(S’s seeing p makes it rational to believe q) and (S knows that ifq then r) and (S correctly deductively infers r from q and believes r on the basisof that deduction)] then S’s seeing p makes it rational to believe r.

Simple and Closure 1 imply that T’s seeing his God Evidence makes it rational tobelieve that there is no gratuitous evil. Since anything that makes it rational to believep makes it more rational to believe p, Simple and Closure 1 imply that T’s seeinghis God Evidence makes it more rational to believe that there is no gratuitous evil.They imply (5). So if Simple is the direct argument this reply has in mind, then ratherthan allowing the skeptical theist to avoid COST* by denying (5), it commits him to(5) and so subjects him to COST*.

Further, retreat to a weaker argument like

Simple*(S1*) (T’s seeing his God Evidence makes it more rational to believe that)God exists(S2) (T knows that) If God exists then there is no gratuitous evil(S3) Therefore, there is no gratuitous evil

is no improvement. For, the skeptical theist who employs Simple* is committed to

25 I thank an anonymous referee for this suggestion.

123


(S4) T knows that if God exists then there is no gratuitous evil and correctly deducesthe nonexistence of gratuitous evil from God’s existence.26

As before, trouble occurs when we add the appropriate instance of a closure principle,

Closure 2 If (S1*) and (S4) then T’s seeing his God Evidence makes it morerational to believe that there is no gratuitous evil.27

Closure 2 is the thesis that if the theist’s seeing God Evidence makes theism morerational and the theist correctly deductively infers the nonexistence of gratuitous evilfrom God’s existence then the theist’s seeing God Evidence makes it more rational tobelieve that there is no gratuitous evil. And Closure 2 seems right because whatevermakes it more rational for S to believe that God exists makes it more rational for Sto believe p, for any p S knows that God exists entails and which S correctly deducesfrom that God exists. Thus, employing Simple* also commits the skeptical theist to(5) because (S1*), (S4), and Closure 2 entail (5). Therefore, a rational skepticaltheist cannot deny (5) and argue directly for the nonexistence of gratuitous evil fromGod’s existence unless he denies Closure 1 or Closure 2. Since I find both quiteplausible, I do not accept this reply to COST*.28

Instead, I prefer a third, necessitarian reply which makes use of our semantics forcounterpossibles in evaluating the rationality of our beliefs. This necessitarian replyto COST* first employs the following argument:

(N1) God exists,(N2) If God exists then necessarily God exists,(N3) If necessarily God exists then necessarily there is no gratuitous evil,(N4) If necessarily there is no gratuitous evil then every closest possibleworld containing gratuitous evil is a world in which probably one’s GodEvidence would seem different,29

26 That is, if T employs Simple* then (S2) is true and T performed the inference from (S1*) and (S2) to(S3).27 Spelled out: If [(T’s seeing his God Evidence makes it more rational for T to believe that God exists) and(T knows that if God exists then there is no gratuitous evil) and (T correctly deduces the nonexistence ofgratuitous evil from God’s existence)] then T’s seeing that God Evidence makes it more rational to believethat there is no gratuitous evil.28 David and Warfield (2008) object to a principle like Closure 1, except for knowledge. The closureprinciple to which they object is (p)(q)[(Kp ∧ K(p entails q) ∧ (Bq based on deduction from p and (pentails q))) → Kq]. They deny it on the grounds a belief can be overdetermined—can be held both on thebasis of correct deductive inference and upon some other, nutty basis like a vision or a talking beer can,and that if the belief is sufficiently overdetermined by sufficiently nutty sources, then a belief based onlypartly on a correct deductive inference fails to be knowledge. First, David and Warfield might be right aboutknowledge and yet Closure 1 be true since it claims only that justified belief, not knowledge, is closedunder the right sort of conditions. Second, overdetermination seems irrelevant to whether a belief basedupon a correct, known, deduction from justified premises is justified—which is all that Closure 1 claims.If we justifiedly believe p, know p entails q, correctly infer q from p, and base our belief that q upon thatdeduction, no matter what else we base our belief that q upon, the correct deduction justifies our belief thatq. Adding other, crazy reasons to its basis muddies the waters, but does not removes q’s justification. SinceClosure 2 is weaker than Closure 1, denying either one looks problematic for the skeptical theist.29 This is just an instance of the schema every instance of which is true: If it is necessary that there isno gratuitous evil then every closest possible world containing gratuitous evil is F, for any F. For, on theLewisian semantics for counterfactuals, counterpossibles are trivially true (Lewis 1973, p. 16).

123


(N5) Every closest possible world containing gratuitous evil is a world inwhich probably one’s God Evidence would seem different if and only if itis true that if there were gratuitous evil then probably one’s God Evidencewould seem different, (this is just a Lewisian semantics for counterfactuals)whereby,(N6) If there were gratuitous evil then probably one’s God Evidence wouldseem different.

Next, the skeptical necessitarian theist argues that he rationally believes N1, knowsN2–N5, and hence, by a series of applications of appropriate instances of closure,rationally believes N6. That is, the necessitarian theist has an independent argumentfor (7).30 Since the question is whether a rational theist can be a consistent skepticaltheist, the proponent of COST* should grant the theist rationally believes N1, andthere is no problem granting the skeptical theist knowledge of N2, N3, and N5. Thepressure occurs at N4 where the fan of COST* might question whether the skepticaltheist really knows that counterpossibles are (trivially) true. In reply, it is certainlyopen to the skeptical theist to claim he does, in fact, know that counterpossibles are(trivially) true because he knows of no more successful analysis of the truth-condi-tions of counterfactuals than David Lewis’s, that Lewis’s analysis seems correct andintuitive, that on that analysis counterpossibles are trivially true, and that by N1–N3,any subjunctive conditional of the form ‘if there were gratuitous evil then q’ is acounterpossible.

This argument for the rationality of believing N6 is just an argument for (7), or, whentranslated back into CORNEA/Laraudogoitian terminology, an independent argumentagainst (10). Recall that in (10), p= there is no gratuitous evil and s= the actual world,such that (10) translated is

(10t) It is not reasonable for H to believe that, given her cognitive faculties and theuse she has made of them, if there were gratuitous evil, the actual world would likelybe different than it is in some way discernible by her.

Then notice that (10t), and hence (10), claim that it is not reasonable for the skep-tical theist to believe a certain subjunctive conditional, the antecedent of which is ‘ifthere were gratuitous evil’. But as we have seen in evaluating the rationality of belief insuch a conditional, since, according to the necessitarian theist, any conditional whoseantecedent is ‘if there were gratuitous evil’ is a counterpossible, any such conditionalis (trivially) rational for that theist to believe. Thus, the necessaritarian theist correctlydenies (10), contra Laraudogoitia.

One might complain at this point that it is not sensible to claim that we rationallybelieve counterpossibles due to their triviality. But our options are limited. Unless wedeny the relevant instances of closure, or deny that counterpossibles are (trivially)true, we are committed to rational belief in counterpossibles. And, since from thenecessitarian theist’s point of view, any subjunctive conditional whose antecedent is‘if there were gratuitous evil’ is a counterpossible, the necessitarian theist has a goodargument for (7) and against (10).

30 (7) is just ‘It is rational to believe (N6)’.

123


The problem with COST*’s argument from (8) to the denial of (7) [or, in Laraudo-goitia’s case, for (10)] is its assumption that the theist can imagine a possible worldcontaining gratuitous evil and compare it to the actual world. But the necessitarian the-ist cannot rationally imagine any such world (much less compare it to the actual world)since he cannot rationally imagine any possible world containing a state of affairs andits complement. Thus there is no conflict between (7) and (8) for the necessitariantheist. Such a theist can be rightly skeptical about his powers of good-identificationwithout inconsistency.

Acknowledgements I would like to thank the editors of this journal, as well as William Demsar, MylanEngel, Klaas Kraay, Michael Rea, Calvin Van Reken, and Stephen Wykstra for valuable discussion andcomments.

References

Alston, W. (1991). The inductive argument from evil and the human cognitive condition. PhilosophicalPerspectives, 5, 29–67.

David, M., & Warfield, T. (2008). Knowledge-closure and skepticism. In Q. Smith (Ed.), Epistemology:New essays (pp. 137–188). Oxford: OUP.

Engel, M. (2004). Theism and inscrutable evil. In W. Lffler & P. Weingartner (Eds.), Knowledge andbelief: Wissen und Glauben Proceedings of the 26th international wittgenstein symposium (pp. 408–414). Vienna: Obv & Hpt.

Howard-Snyder, D. (Ed.). (2004). The evidential argument from evil. Bloomington: Indiana UniversityPress.

Laraudogoitia, J. (2000). Cornea against theism. International Journal for Philosophy of Religion, 48,81–87.

Lewis, D. (1973). Counterfactuals. Oxford: Blackwell.Rowe, W. (1979). The problem of evil and some varieties of atheism. American Philosophical Quar-

terly, 16(4), 335–341.Van Inwagen, P. (2006). The problem of evil. Oxford: OUP.Wykstra, S. J. (1984). The humean obstacle to evidential arguments from suffering: On avoiding the

evils of appearance. International Journal of the Philosophy of Religion, 16, 73–93.Wykstra, S. J. (2007). Cornea, carnap, and closure befuddlement. Faith and Philosophy, 24(1), 87–98.

123

paying the cost of skeptical theism

Documents