Virtual Colloquium: Luis Oliveira, “Skeptical Theism and the Paradox of Evil”
December 2, 2016 — 6:00

Author: Kenny Pearce  Category: Problem of Evil  Tags: , , , ,   Comments: 7

Today’s Virtual Colloquium paper is “Skeptical Theism and the Paradox of Evil” by Luis Oliveira. Dr. Oliveira recently received his PhD from the University of Massachusetts Amherst and is now a Visiting Lecturer at the University of Massachusetts Lowell. His papers have appeared in journals such as Philosophical Studies, Australasian Journal of Philosophy, and Pacific Philosophical Quarterly.


Skeptical Theism and the Paradox of Evil

Luis Oliveira

Let me begin by thanking Kenny Pearce for hosting this Virtual Colloquium and for inviting me to contribute a paper to it. I have enjoyed reading and discussing each of the papers presented so far. I hope my paper will continue the trend of substantive and constructive exchanges in the comments section. Here is a preview, from my introductory section:

According to the evidential problem of evil, our seeing no justifying-reason for many instances of suffering is sufficient evidence for the belief that the traditional (maximally great) God does not exist. According to skeptical theism, however, it is not at all likely that we would see a justifying-reason for instances of suffering, were such a God and such reasons to really exist. Given plausible assumptions about the nature of evidence and undercutting defeat, it has seemed to many that the force of the evidential problem of evil therefore depends on skeptical theism being false. If we cannot expect to see God’s justifying-reasons, were Him and them to truly be there, then our not seeing them can hardly count as evidence against His existence.

In this paper, I argue that there is a way of understanding the evidential problem of evil where it is compatible with skeptical theism. I show that skeptical theism blocks the evidential problem of evil only given certain natural assumptions about how the evidence from evil accrues, I show that these assumptions are not essential to the problem, and I show which alternative assumptions can take its place. I do not, however, go as far as endorsing the evidential problem of evil on the basis of these alternative assumptions. Nonetheless, if I am right about all this, the result is that the stalemate between the many who defend skeptical theism and the many who criticize it can be altogether sidestepped.

Here is how I proceed. I begin, in section 1, by clarifying two essential features of William Rowe’s justly famous original formulation of the evidential problem of evil. Next, in section 2, I articulate what I call its inductive justification, which I argue is widely presupposed by Rowe commentators, according to which Rowe’s argument depends on the accumulation of little bits of evidential support from particular instances of apparently pointless suffering. Then, in section 3, I argue that skeptical theism, properly formulated, resists Rowe’s argument by denying that these particular instances provide even a modicum of support against God. With this dialectic clarified in the background, in section 4, I suggest an alternative justification for Rowe’s original argument. On what I call its collective justification, Rowe’s argument turns on the evidential support provided by the collection of instances of apparently pointless suffering in a way that is compatible with each particular instance failing to provide any support at all. Drawing on the evidential dimension of the preface paradox, I call this result the paradox of evil. I conclude, in section 5, by arguing that skeptical theism, based as it is on a claim about our cognitive limitations, is compatible with the collective justification of Rowe’s argument. Whether Rowe’s argument is sound remains open for debate, though a debate that does not center around skeptical theism anymore.


The full paper is here. Comments welcome below!

Comments:
  • Tim Perrine

    Hi Luis,

    Thanks for the interesting paper. There’s a lot to digest, and I’m not sure I understood it all. Nonethless, I hope my two comments can be helpful.

    First comment. You consider two versions of skeptical theism One holds that the existence of apparently pointless evil (APe) does not increase the likelihood of pointless evil (Pe). The other more modest position holds that APe does increase the likelihood of Pe but not enough to make it reasonable (p. 8). You criticize this position by saying, “In order to deny that apparently pointless evil can provide strong evidence for pointless evil, one must deny that it provides any evidence whatsoever in the first place. One cannot grant the evidential power of the particular instances, that is, while denying the increased evidential power of their accumulation.”

    Though we may be speaking past one another, I think that modest position is (at the very least) formally consistent and immune to this criticism. For one might hold that learning, for the first time, of the existence of APe is some evidence for Pe and thus some evidence against theism. However, learning the existence of APe might confirm some version of theism on which we should not expect to see justifying reasons for evils. Consequently, learning of subsequent APe would not disconfirm theism. The accumulation strategy would then fail.

    Here’s an analogy. Watching the sun move across the sky one day while sitting on my porch is some evidence against the view that the earth revolves around the sun. But that evidence might make probable that if the earth revolves around the sun, then the earth rotates on its own axis every 24 hours giving the appearance of movement by the sun. Consequently, my initial evidence might make probable a version of heliocentricism that is not disconfirmed by further days of me sitting on my porch watching the sun “move”.

    Second comment. I’m also a little perplexed by your discussion of the preface paradox. I’m perplexed for two reasons. First, suppose as many participants in this dispute will, that being reasonable is closely related to probability; it is reasonable to believe p only if the probability of p is quite high (say .95) reasonable to believe ~p only if the probability of p is quite low (say .05) and neither if its probability is in some unspecified “middling.” Now we already know that when we start conjoining certain proposition (e.g. non-tautologies/contradictions, logically independent), then very quickly the probability of the conjunction begins dropping. Given these two claims, we have good reason for reject the conditional claim, PP3 (p. 12) in the preface paradox. But, so far as I can tell, the same exact reason holds for the condition claim, PE3 (p. 14) in the paradox of evil. So why isn’t the “reversing” condition in both situations the same? (I’m assuming that your points about epistemic humility can be glossed in this probabilistic way.)

    Second, it does not strike me that the preface analogy is analogous to your collective reading of Rowe. Consequently, I’m not sure if drawing on the analogy helps your argument. Here is my preferred gloss on the preface paradox. Let’s suppose we have a set, S, with a bunch of propositions. Assume each proposition begins with a high probability (e.g. .95). Let ‘&S’ be the conjunction of the elements S. Your point on p. 13 is that no single element of S is evidence against &S. (That is, For all p ∈ S, ~[Pr(&S| p) Pr(Pe)]. That is, you seem willing to concede that for any particular evil, our inability to see a justifying reason is not evidence that it lacks a justifying reason. Nevertheless, it is also claimed, on this reading, that: Pr{[Pe1 v Pe2 v … v Pen] | [APe1 & APe2 &… & APen]} could still be quite high. That is, the fact that no particular inability to see a justifying reason is evidence against there being one for that evil does not stop the collection of inabilities from being very strong evidence that there is one pointless evil.

    But when put this way, the collective reading seems quite different from a preface paradox. First, your own formulation of the paradox of evil (p. 14) says nothing about seemingly pointless evil, though that is important to your collective reading of Rowe. Second, in the preface paradox, the hypothesis that is not disconfirmed initially by various claims, but does have a low probability in the end, is the same hypothesis–&S. But that’s not the case in the collective reading. Third, the conclusions are different: one is that a non-conditional statement has a low probability; the other is that a conditional statement has a high probability. Finally, and perhaps most importantly, what generates the result of the preface paradox (as I’ve presented it) are just well-known features of probability. But that doesn’t seem to be the case for the collective reading. Rather, it turns on some additional claims about (e.g.) God’s maximal greatness.

    All of this is by way saying that I’m not sure how the analogy from the preface paradox helps your argument. Why not just argue directly that God’s perfect goodness gives us very good reason for thinking that we should expect more insight into God’s plans for the evil we observe than what we do, in fact, observe?

    Best,
    Tim

    December 2, 2016 — 13:58
    • Tim Perrine

      Oops! I see some of my comments were accidentally mangled. Let’s try again:

      Second, it does not strike me that the preface analogy is analogous to your collective reading of Rowe. Consequently, I’m not sure if drawing on the analogy helps your argument. Here is my preferred gloss on the preface paradox. Let’s suppose we have a set, S, with a bunch of propositions. Assume each proposition begins with a high probability (e.g. .95). Let ‘&S’ be the conjunction of the elements S. Your point on p. 13 is that no single element of S is evidence against &S. (That is, For all p ∈ S, ~[Pr(&S| p) Pr(Pe)]. That is, you seem willing to concede that for any particular evil, our inability to see a justifying reason is not evidence that it lacks a justifying reason. Nevertheless, it is also claimed, on this reading, that: Pr{[Pe1 v Pe2 v … v Pen] | [APe1 & APe2 &… & APen]} could still be quite high. That is, the fact that no particular inability to see a justifying reason is evidence against there being one for that evil does not stop the collection of inabilities from being very strong evidence that there is one pointless evil.

      December 2, 2016 — 14:02
  • Hi Tim,

    Thanks for these two excellent comments! Here are my initial thoughts on them.

    First:

    In reply to my claim that one cannot hold that both (a) an apparently pointless evil is evidence against theism, and (b) lots of apparently pointless evil do not add up to strong evidence against theism, you say this:

    “One might hold that learning, for the first time, of the existence of [an apparently pointless evil] is some evidence for [there being pointless evil] and thus some evidence against theism. However, learning the existence of [an apparently pointless evil] might confirm some version of theism on which we should not expect to see justifying reasons for evils. Consequently, learning of subsequent [an apparently pointless evil] would not disconfirm theism. The accumulation strategy would then fail.”

    I’m not sure how that would work. I defined theism as the claim that “a maximally great God exists” (T). You say here that the experience of an apparently pointless evil might be evidence against T. That can only be the case if T is less likely, given that experience, and if that experience is more likely given ~T. But then you say that there is a version of theism (T*) according to which it is not true that the experience of an apparently pointless evil is more likely given ~T* than T*, and therefore that the accumulation of these experiences would fail to produce strong evidence against T*. As far as I can tell, however, in this case the experience of an apparently pointless evil would not be evidence at all against T* in the first place. The experience of apparently pointless evil either is or is not expected, given a certain H, and therefore it either confirms or disconfirms it. So it seems to me that if we talk about T, then we can explain (a) but not (b), and if we talk about T*, then we can explain (b) but not (a). But neither of them manages to explain both. I might be failing to see exactly what you meant.

    Second:

    The analogy with the preface paradox is not supposed to be with *how* or *why* the probabilities change from consideration of particular instances and their collection. Instead, the analogy is supposed to be with the evidential dimension of the preface paradox. What the two paradoxes share is the fact that the collection of particular instances is evidence against a certain hypothesis, while the particular instances by themselves are not. What the reversing condition is supposed to help us understand, then, is not the paradox in general, but only its evidential dimension.

    Now, I think you are right that I must make reference to apparently pointless suffering in my formulation of the paradox of evil. I will change that. But notice that the hypothesis that I am not reasonable in believing in the paradox of evil is not a conditional claim. It is rather the claim that “at least some of the apparently pointless evils are truly pointless evils”. This is parallel to the hypothesis that I am not reasonable to believe in the preface paradox: “at least some of the apparently true sentences are false”. In both the preface paradox and the paradox of evil, then, the hypothesis that is not disconfirmed initially but does have a low probability in the end is the same.

    I don’t think my discussion of these issues is sufficiently clear just yet. Thanks again for focusing my attention on them.

    Luis

    December 3, 2016 — 10:43
    • Tim Perrine

      Hi Luis,

      Let me try to make my points a little more clearly.

      First, suppose we have a set of propositions, e1…en. Suppose we have a hypothesis, T, but there are two ways of articulating T: one according to S, one according to ~S. Let me us make three further assumptions: Initially, S is much more likely that ~S given T but T does not entail S; no member of e1…en is evidence against S&T, in fact each is somewhat likely given S&T; each member of that set is evidence against ~S&T, in fact each s somewhat unlikely given ~S&T.

      Now suppose we learn e1. It will be evidence against ~S&T, i.e. one way of articulating T. But it will not be evidence against the other way of articulating T, namely S&T. Consequently, learning e1 will be some evidence against T, by being evidence against one way of articulating T. But it will also be confirming evidence for S&T. Consequently, the proportion of T, on which S is true, will increase. However, learning e2 will be less evidence against T than e1. This is because e2 is only evidence against one way of articulating T, namely ~S&T, and that way of articulating T has a lower probability than it did initially. This pattern will repeat so that the evidential force of learning new members of that set may quickly taper off, so that learning more and more of e1…en may never make the probability T very low. Now so long as none of e1 is actually inconsistent with ~S&T, it will always be the case that learning an element of that set is evidence against T. But the evidential cash value of each subsequent one will be much less than the previous one (We can secure this formal result just using Bayes’ theorem, the total probability theorem, and various assignments needed for using those theorems.)

      One might think of skeptical theism analogously. Perhaps initially learning of seemingly unjustified evils decreases the probability of theism; perhaps it does this by disconfirming a version/articulation of theism on which we should expect to see God’s reasons for evils. But it would not disconfirm versions on which we should not expect to see God’s reasons for evils; in fact, it would confirm them. Consequently, initially learning of seemingly unjustified evils decreases the probability of theism by decreasing the probability of one version of theism. But exactly because learning that information disconfirms a particular version of theism but not another, subsequent information of the same kind will have less evidence force—perhaps tapering off long before making it unreasonable to believe theism on the basis of it. Thus, one can hold the following, formally consistent position: initially learning of seemingly unjustified evils is evidence against theism, but the accumulation of them is not sufficient evidence for making it unreasonable to believe theism.

      Second, thanks to your response, I think I better understand how you imagine the paradoxes to work and how they are analogous. In the preface case, none of my individual assertions p1…pn is evidence against ~(p1&..&pn). But the collection of p1…pn is evidence against it. In the evil case, none of my failures to identify a God-justifying reason is evidence for their being at least one pointless evil, but the collection is evidence against it.

      However, I still have worries. The first is that, when put this way, I think you’ve simply misidentified the reversing condition in the preface case. It has nothing do with epistemic humility. It has to do with the simple fact that a collection of statements entails its conjunction. Consequently, the negation of its conjunction, given that collection, is 0. Thus, so long as ~(p1&…&pn) is logically consistent, its probability with go down to 0, given p1…pn. Consequently, the “reversing condition” is just a humdrum fact about the probability calculus. That does not hold in the case involving evil. Rather, it seems we need some argument for their being a “reversing condition” in that case at all: some reason for thinking that it is very improbable that there would not be any pointless evil, given the amount of seemingly pointless evil we observe. But if you had an argument for that condition, then once again it seems you could just run the argument straightforwardly without talking about the preface paradox.

      Second, p is of course evidence against ~p. If p is true, ~p must be false. Suppose that ~q is independent of p; p does not make it more likely or less likely. Why couldn’t I reason as follows: “p is evidence against ~p v ~q. For while p is not evidence against ~q, it is evidence against ~p. Thus it is evidence against one of those outcomes occurring.” But since ~p v ~q is equivalent to ~(p&q), it follows that p is evidence against ~(p&q). So why couldn’t I reasonably claim that, as a matter of fact, one of the assertions of my book is evidence against one of the assertions of my book being false: namely, evidence against that very claim being false! Thus, contrary to your claim about the preface paradox, each individual assertion is evidence against at least one of my statements being false.

      Again, there are complexity and intricacies that I don’t feel like I’m fully following. And I can’t rule out that these last two points are just based on a misunderstanding. But maybe it will be helpful nonetheless if, in the very least, helping identify the source of my errors!

      Best,
      Tim

      December 4, 2016 — 12:56
  • Hi Tim,

    Thanks again for these excellent comments. Here’s what I’ve got so far to say in reply.

    About your first worry:

    I understand the formal features of your example using two versions of a certain hypothesis H: one where it is combined with S and one where it is combined with ~S. But I’m not sure about its necessary application to my discussion of theism (the believe in the existence of an all-good, all-knowing, and all-powerful being, with a personal interest in the lives of his creatures). There are two routes to my uncertainty.

    Route #1: Suppose I just want to discuss the evidential force of evil on theism (H) without considering its combination with further mutually exclusive propositions (S & ~S). It seems to me that we can stipulate (after careful reflection, as best we can) what can or cannot be expected given H by itself, and so find out what can and cannot count as evidence for or against it. Of course, we *could* talk about combining H with further mutually exclusive propositions. But I don’t see why we should be forced to do so. So while I see that we could fashion a model where the relevant bits of evidence have diminishing marginal value, I don’t see a reason to think that this model forces itself on our discussion of theism.

    Route #2: Suppose you insist that we cannot separate a discussion of theism (H) from a discussion of theism in combination with further mutually exclusive propositions (H&S vs H&~S). (I would like to hear why, though.) Then any evidence against H&~S (that is not also evidence against H&S) will be evidence against H that has a diminishing marginal value as you’ve mentioned. Okay. For clarity, however, I would then suggest distinguishing between H&S and H&~S by calling them H1 and H2 and by treating them as altogether separate hypothesis. I want to know the evidential impact of a bit of evidence on each. Once again, we have escaped the clutches of the model you suggested for the discussion of theism.

    Your second worry:

    What I call the evidential dimension of the preface paradox is the dual fact that (a) none of my individual assertions p1, p2, …, pn is evidence against ~(p1 & p2 & … & pn), but (b) the collection of (p1, p2, …, pn) is evidence against ~(p1 & p2 & … & pn). I say that this evidential reversal is due to a reversing condition: our modest fallibility. You note this, and then say:

    “I think you’ve simply misidentified the reversing condition in the preface case. It has nothing do with epistemic humility. It has to do with the simple fact that a collection of statements entails its conjunction. Consequently, the negation of its conjunction, given that collection, is 0. Thus, so long as ~(p1&…&pn) is logically consistent, its probability will go down to 0, given p1…pn.”

    But I’m not sure of what you mean. I guess I’m not seeing the connection that you are seeing between the fact that (p1, p2, …, pn) has a probability of 1 for me and the reversing evidential behavior captured by (a) and (b).

    I have to think more about your last point involving the transformation of (~p v ~q) into ~(p & q)!

    Thanks again for these probing comments!

    Luis

    December 7, 2016 — 9:33
  • Joe Milburn

    Dear Luis,

    Thanks for this interesting and informative paper. If I understand correctly, the purpose of your paper is to move debates about the evidential argument from evil beyond debates about skeptical theism.

    I think the force of the evidential argument as you present it however, might still rest on the truth of skeptical theism. In particular, I think skeptical theism may give us reason for at least suspending judgment concerning the following claims which I take it are central to your interpretation of the evidential argument.

    P: It behooves God to explain to us, at some point and in broad outline His justifying reasons for allowing for *so many* instances of apparently pointless suffering fro so long. At a minimum, it behooves Him to either give us some intelligible indication of the existence of such justifying-reasons as proxy — some form of assurance that there are, in fact, justifying-reasons — or to make His comforting presence unmistakenly known to us in compensation.’

    Why should we believe P? Here is one explanation.

    It seems that God’s not giving us indications that there are justifying reasons for the seemingly pointless suffering in the world, or His not consoling us with an appreciation of His presence as a compensation for these sufferings, is an instance of pointless suffering. But it behooves God not to allow pointless suffering. So it behooves God to either give us indications that there are justifying reasons for the seemingly pointless suffering in the world, or to consle us with an appreciation of His presence.

    If this is the explanation for why we should believe P, then skeptical theist can say: I have given you reason for rejecting or at least suspending judgment regarding whether it behooves God to explain Himself to us or to console us in dint of not explaining Himself to us.

    Thanks again!

    Best,

    Joe

    December 13, 2016 — 11:34
    • Hi Joe,

      Thanks for reading my paper!

      That’s a good point. In switching the basis for the evidential problem of evil to (P), I must be careful to avoid the suggestion that my support for (P) come from its denial consisting in pointless evil. I agree with you that this would be bad news. But I don’t think I am cornered into this suggestion.

      I think can say instead that support from (P) comes from our understanding of the terms of God’s maximal greatness. In particular, our understanding of parental care and love. So in the context of God’s other perfect abilities, our notion of God as having a personal interest in the well-being of its creatures (where this includes a loving relationship with Him) is what justifies (P).

      I need to think more carefully about this and I need to be clearer about it in the paper. Thanks for bringing this up!

      Best,

      Luis

      December 14, 2016 — 9:27
  • Leave a Reply

    Your email address will not be published. Required fields are marked *