Monday, February 14, 2011

Addendum to “The End of Skeptical Theism?”: On What God Would Do



I know have already written, at nauseating length, about the topic of skeptical theism (ST). But I recently read an article by Rob Lovering entitled “On What God Would Do” and he lays out the implications of ST for theistic belief in such a well-structured manner that I thought I would share it here. This will serve as a nice addendum to the original series of posts on ST.


1. Three Epistemic Attitudes to God
As I explained in part one, ST in its modern form was introduced as a response to Rowe’s evidential problem of evil. Its proponents argued that given certain features of God (e.g. his tri-omni properties), and certain limitations in our cognitive abilities, there is no reason for us to endorse the central inference in Rowe’s argument.

Why not? Because God could, for all we know, have beyond-our-ken reasons for allowing apparently gratuitous evil to exist. Thus we should endorse an attitude of skepticism toward God and his reasons for action.

As Lovering notes, this creates problems for theists because many of the arguments they adduce for the existence of God rely, implicitly or explicitly, on claims about what God is likely to do. One example might be Robin Collins’s fine-tuning argument which relies on the claim that God would want to create morally significant conscious agents. Another example would be the various arguments to and from miracles which depend for their success on assumptions about how God is likely to intervene in the world.

So theists are left in a bit of a bind: are they to endorse skepticism and wriggle their way out of Rowe’s argument, or are they to enthusiastically assume to know God’s mind? Lovering identifies three options:

  • (i) Broad Skeptical Theism: The view that, in each and every case, we cannot know what God would do.
  • (ii) Broad Epistemic Theism: The view that, in each and every case, we can know what God would do.
  • (iii) Narrow Skeptical Theism: The view that, in some cases, we can know what God would do, and in other cases, we cannot.

Lovering sets out to examine the plausibility of each of these options.


2. Intrinsic and Extrinsic Dependence
Lovering begins by arguing that acceptance of broad skeptical theism would force the believer to give up every positive argument for the existence of God. This is because every such argument depends on at least one claim about what god would do. One of the great virtues of Lovering’s article is his attempt to show exactly how this dependence arises.

First, he says that an argument for the existence of God can intrinsically depend on a claim about what God would do if such a claim can be derived from it after one of the following conceptual claims is added to the argument:

  • (a) Conceptual Claim 1 (CC1): If God exists and X is/was the case, then God allows/allowed X to be the case.
  • (b) Conceptual Claim 2 (CC2): If God allows X to be the case, then God would allow X to be the case.

These conceptual claims might seem strange, but they make sense if we think about the kind of being god is supposed to be -- the omnipotent, omniscient, sovereign creator of all reality. Given his nature, if there are red cars in existence, then it must be the case that God allows them to exist (CC1); and if there are red cars that God allows to exist, then it must be the case that God would or is likely to allow such things to exist (this is a claim about what God would allow across different possible worlds - i.e. CC2).

Lovering shows how these conceptual claims can be slotted into general arguments for the existence of God and used to derive from them a claim about what God would do. I will offer an abstract version here. Let P = some fact about the world that God is supposed to explain or account for; and let G = the claim that God exists.

  • (1) If P, then G.
  • (2) P.
  • (3) Therefore, G.
  • (4) Therefore, G and P.
  • (5) If G and P, then it must be the case that God allows P to be the case (CC1).
  • (6) Therefore, God allows P to be the case.
  • (7) If God allows P to be the case, then God would allow P to be the case (CC2).
  • (8) Therefore, God would allow P to be the case.

Although this may be a little abstract, the upshot of it is that a claim about what God is likely to do (or what God would do) can be derived from a general argument for God’s existence by means of CC1 and CC2.

The second way in which theistic arguments depend on claims about what God would do is through extrinsic dependence. This arises when a claim about what God would do can be derived from an argument if, in addition to adding CC1 or CC2 to it, we add the following:

  • Intelligibility Claim (IC): We understand argument A (where “A” is the argument for God’s existence).

To see how this works in the abstract, let P = some feature or attribute of God; and let G = the claim that God exists (this is like the ontological argument).

  • (1) If God is P, then G.
  • (2) God is P.
  • (3) Therefore, G.
  • (4) We understand the argument for G in (1)-(3) (via IC).
  • (5) Therefore, G and we understand the argument for G.
  • (6) If G and we understand the argument for G, then God must allow us to understand the argument for G (CC1).
  • (7) Therefore, God allows us to understand the argument for G.
  • (8) If God actually allows us to understand the argument for G, then God would or is likely to allow us to understand the argument for G (CC2).
  • (9) Therefore, God would or is likely to allow us to understand the argument for G.

So, once again, from an argument for the existence of God we can derive a claim about what God would do.

Lovering goes on to provide examples of how existing arguments for the existence of God exhibit intrinsic and extrinsic dependence. I’ll just provide one example of each.



3. Intrinsic Dependence and the Cosmological Argument
Consider the following version of the Cosmological Argument:

  • (1) There exist things that are caused to be.
  • (2) Nothing that is caused to be can be the cause of itself.
  • (3) There cannot be an infinite regress of causes.
  • (4) Therefore, there exists an uncaused first cause.
  • (5) Probably, God is the uncaused first cause.
  • (6) Therefore, probably God exists.

This should be a fairly familiar generic form of the CA to most people. Now see how we can derive a claim about what God would do from this argument. As follows (the language is convoluted, but read through it slowly and it should make sense):

  • (7) Probably, God exists and God caused the universe to be the case.
  • (8) If God exists and God caused the universe to be, then God must have allowed the universe to be caused to be (CC1).
  • (9) Therefore, probably, God allowed the universe to be caused to be.
  • (10) If God allowed the universe to be caused to be, then God would allow the universe to be caused to be (CC2).
  • (11) Therefore, God would allow the universe to be caused to be.

Et voila, we have a claim about what God would allow to be the case. The method employed in this analysis of the CA can be employed across a range of teleological, ontological and historical arguments for the existence of God.


3. Extrinsic Dependence and the Argument from Miracles
As an example of an argument that is extrinsically dependent on a claim about what God would do, consider the following versions of the argument from miracles:

  • (1) Extraordinary events occur.
  • (2) At least some of these extraordinary events could not have been the result of natural law or natural causes.
  • (3) In such cases, the extraordinary events must be the result of a supernatural cause.
  • (4) Probably, God is the supernatural cause of such extraordinary events.
  • (5) Therefore, God probably exists.

Making use of the intelligibility claim, we can expand on this as follows:

  • (6) We understand the argument for God’s existence in (1) - (6). (IC)
  • (7) Therefore, God probably exists and we understand this argument for his existence.
  • (8) If God probably exists, and we understand this argument for his existence, then God must allow us to understand this argument for his existence (CC1).
  • (9) Therefore, God allows us to understand this argument for his existence.
  • (10) If God allows us to understand this argument for his existence, then God would allow us to understand this argument for his existence (CC2).
  • (11) Therefore, God would allow us to understand this argument for his existence.

And so, once again, we have an argument depending on a claim about what God would do.



4. The Untenability of Broad Skeptical Theism
Through the various other examples he has offered, Lovering thinks he has shown how every positive argument for God’s existence intrinsically or extrinsically depends on a claim about what God would allow to be the case. But since broad skeptical theism denies that we have any knowledge about what God would allow to be the case, it follows that a theist who accepts God’s existence on the basis of one or more of these arguments cannot endorse this brand of skepticism.

Or to put it more succinctly: for the majority of theists, broad skeptical theism is completely untenable. What about the other epistemic attitudes?


5. Broad Epistemic Theism
As set out above, this is the view that, in each and every case, we can know what God would do. At first glance, this is more plausible because it does not force the believer to relinquish every positive argument for God’s existence. But it also creates new problems.

First, it changes the nature of the response to Rowe’s evidential problem of evil. The believer cannot offer vague or aspirational rebuttals to Rowe’s main premise. For example, they cannot say “some reason for allowing prima facie gratuitous evil to exist will be forthcoming”. Instead, they must make a positive case for why God would allow such things to occur. Or, they must turn to other arguments for God’s existence and show how they tip the evidential weighing scales in favour of theism.



Second, it forces the theist to accept that they can know far more about God than they traditionally thought they could know. For instance, if in each and every case we can know what God would do, then we should be able to say whether:
  • God would allow the Red Sox to win the World Series five years in a row;
  • Whether God would allow another Holocaust; and so on.
Furthermore, it makes atheistic arguments such as the argument from divine hiddenness or non-belief more pressing.

It is unlikely then that a believer would be willing to embrace Broad Epistemic Theism.


6. Narrow Skeptical Theism
That leaves us with what many perceive to be the most plausible position: Narrow Skeptical Theism. This is the view that, in some cases, we know what God would do, but in other cases, we do not. This might allow the theist to keep their positive arguments and reject atheistic arguments.

But this requires some principled reason for distinguishing between claims about what God would do in those different cases. Lovering does not think any theist has yet risen to the challenge of developing such a principled distinction, so he tries to meet it himself.

He takes the claims about what God would do in Robin Collins’s teleological argument and in the evidential argument from evil as his reference point and asks the question: why is Collins allowed to claim that God would want to create conscious intelligent agents but the atheist is not allowed to claim that God would not allow apparently gratuitous evil to exist?

One potential difference is that Collins’s case just requires a general affirmation of the prima facie goodness of a particular state of affairs (“the existence of intelligent conscious agents”), whereas the argument from evil requires knowledge of whether something is an all-things-considered good or evil.

So this leads us to formulate the following principle:

  • Principled Narrow Skeptical Theism - We can know whether something is a prima facie good or evil, but we cannot know whether something is an all-things-considered good or evil (only God can know that).

There are problems with this principle.

First, in relation to the problem of evil, skeptical theists are demanding a deep skepticism about our knowledge of the good. They are claiming that we cannot know whether or not a prima facie evil is necessary in order to realise a greater good because we are epistemically closed from the possible worlds in which such evils might be necessary.

But why does the same kind of skepticism not apply to our consideration of the possible worlds that God might choose to create? In particular, why doesn’t our lack of access to different possible worlds, and the forms of life possible within them, shut down any claim that this particular universe is fine-tuned for life.

A second problem with the principle is that Collins’s argument actually does require knowledge of all-things-considered goods. To see this, consider Collins’s own defence of his argument:

“Since God is an all good being, and it is good for intelligent conscious beings to exist, it is not surprising or improbable that God would create a world that could support intelligent life.”

To accept this line of argument, we must be able to say that intelligent beings are an all-things-considered good; that their existence is not outweighed by some greater, countervailing evil. To say that they are merely a prima facie (or “all else being equal”) good is to open up the possibility that their existence serves some ultimately evil purpose.

And the problem is that this isn’t only true of Collins’s argument. Since God is a morally perfect being, then every event or state of affairs that is explained by reference to him, must be compatible with his moral perfection. And since moral perfection must mean “acting so as to achieve all-things-considered goods”, it follows that we must know that the events and states of affairs explained by reference to God are all-things-considered goods.

The conclusion then is not only that the principled distinction offered above fails, but that all such principles are likely to fail.


7. Summary
Lovering began by identifying three different epistemic attitudes theists could take with respect to God. The following conclusions have been reached about each of them:

  • (i) Given Broad Skeptical Theism, theists must relinquish every positive argument for God’s existence since every such argument depends on at least one claim about what God would do.
  • (ii) Given Broad Epistemic Theism, theist must (a) relinquish one of the principal grounds upon which they object to Rowe’s argument from evil and other similar atheistic arguments; and (b) embrace far more knowledge claims about what God would allow to be the case than they have traditionally been willing to do.
  • (iii) Given Narrow Skeptical Theism, the theist must find some principled ground for distinguishing between the set of cases in which we have knowledge of what God would do and the set of cases in which we do not have knowledge of what God would do. Such a distinction is unlikely to be forthcoming.

6 comments:

  1. Why does our lack of knowledge about all things considered goods not extend to Gods nature? It seems like no matter how a theist tries to define the limits of their knowledge about the good God will always be outside it and His nature will just be assumed.

    ReplyDelete
  2. I'm not sure I fully understand.

    I think I may have addressed that point in the second-to-last paragraph of section 6.

    If not, perhaps you could rephrase the question.

    ReplyDelete
  3. Actually you're exactly right, I misread that paragraph. While I'm here though... If a theist rejects ST then do you know how they might try to justify their knowledge of God? It seems like the usual position is that given God has done X and the best explanation for a being doing X is that he has some set of desires and intentions Y we can conclude fairly safely that he has Y. But that kind of inference can only be made when we have some idea of the range of possible desires and intentions the being might have. So why should it apply to something so totally alien to our minds? I think we have reason to be skeptical about Gods nature even without ST.

    ReplyDelete
  4. Adito,
    in her interview with Luke over at commonsenseatheism, Lydia McGrew says that you don't always need to know an agents desires to realize that something was designed by him. She uses as an example the discovery of a deep-space Volkswagen and mentions that W L Craig makes the same point but with ruins on another planet (if I remember correctly). I think Lydia says more on the Volkswagen in her paper Testability, Likelihoods, and Design. The link is available over at CSA, but for some reason I cannot access the pdf. Below is the link to the interview.

    http://commonsenseatheism.com/?p=10555

    ReplyDelete
  5. Two things.

    First, re: McGrew's argument. I've skimmed that paper so this is preliminary, but she appears to make the argument that you can determine the likelihood that a particular object (or state of affairs) is the result of design based on (frequency) data from known designers (i.e. human or animal agents).

    That seems fine to me, but it does rely on the ability to build the analogy (say that one class of agents must be like another). And it would also seem to open the door Broad Epistemic Theism (discussed in the post above), which makes the problem of evil more pressing. Why? Because working with the data that we have from known agents, we could easily argue that the evil we observe in the world is the result of a malevolent agent as opposed to a beneficent one.

    Second, I think some theists might like to rely on a Plantinga-style properly basic account of their belief in God. This might be thought to sidestep the problem of skeptical theism, but I've summarised arguments to the contrary on this blog before.

    Also, Bogdan recommended that I cover Steve Maitzen's article on "Anselmian Atheism" on my podcast. Reading over it last night I think it has something interesting to say about this issue. Namely: that God as defined by theists must be either (a) utterly beyond the comprehension of any human being; (b) non-existent or (c) not maximally great. Maitzen argues that (b) is the most plausible of these options.

    ReplyDelete
  6. That seems fine to me, but it does rely on the ability to build the analogy (say that one class of agents must be like another).

    Which is where Philo's (and hence probably Hume's) arguments in Dialogues Conerning Natural Religion are going to trip up proponents of Christian theism: isn't it more likely, based on our analogies, that there were a team of designers and builders, for example? And, as for evil, as Hume has it, "as this goodness is not antecedently established, but must be inferred from the phenomena, there can be no grounds for such an inference, while there are so many ills in the universe, and while these ills might so easily have been remedied, as far as human understanding can be allowed to judge on such a subject. I am Sceptic enough to allow, that the bad appearances, notwithstanding all my reasonings, may be compatible with such attributes as you suppose; but surely they can never prove these attributes."

    He also says that a person "may be fully convinced of the narrow limits of his understanding; but this will not help him in forming an inference concerning the goodness of superior powers, since he must form that inference from what he knows, not from what he is ignorant of. The more you exaggerate his weakness and ignorance, the more diffident you render him, and give him the greater suspicion that such subjects are beyond the reach of his faculties", which sounds like the beginnings of the modern responses to sceptical theism.

    ReplyDelete