Tuesday, August 25, 2015

On the Limitations of General Skeptical Theism

Erik Wielenberg

Erik Wielenberg has just published a great little paper on skeptical theism and the problem of evil. I don’t mean to use the word ‘little’ in a pejorative sense. Quite the contrary. I use that descriptor because the paper manages to pack quite a punch into a relatively short space (a mere 12 pages of text). The ‘punch’ consists of two interesting arguments. The first is a meta-argument about standards of success in the debate between skeptical theists and proponents of the problem of evil. The second is a strengthened version of the problem of evil, which focuses specifically on the problem of suffering and abandonment.

The second argument is the real centrepiece of the article and I will cover it in a future post. Today, I want to deal with the meta-argument. I do so because it sets the stage for the argument from suffering and abandonment, and because it is an interesting methodological point in its own right. I won’t delay any further; I’ll get straight into it.

1. The Problem of Evil and the Noseeum Inference
Everyone is familiar with the problem of evil. They all know that God is supposed to be a maximally powerful, maximally knowledgeable, and perfectly good being. They also know that there are many real world instances of evil. This evil can take many forms, with the most commonly lamentable form being the suffering of conscious creatures. The problem of evil simply points to the difficulty of reconciling the existence of such suffering with the existence of God.

Of course, it’s a little bit more complicated than that, and I don’t want to completely rehash the centuries-long debate about the problem of evil here. Instead, I want to hone-in on its most popular modern form. Back in 1979, the (sadly) recently-deceased William Rowe published an influential article entitled ‘The Problem of Evil and Some Varieties of Atheism’. In it, he presented an evidential version of the problem of evil, which has become the most widely-discussed contemporary variant on the problem.

Rowe’s argument appealed to the concept of gratuitous evil. This is a type of evil that is not logically or metaphysically necessary for some greater good. In other words, it is a type of evil that a perfectly good being could not permit. This is widely accepted by both theist and atheist alike. What is disputed is whether there are any actual instances of gratuitous evil. Rowe tried to argue that there are. He did this by highlighting examples of real-world suffering that don’t seem (in light of everything we know) to have any God-justifying reason for their existence. His famous example of such evil is a fawn who suffers horribly in a forest fire, with no one around to help or learn from the experience. He argues that we can infer the likely existence of actually gratuitous evils from the existence of such seemingly gratuitous evils.

To put it more formally, Rowe’s version of the problem of evil takes (roughly) the following form:

  • (1) If there are any actually gratuitous evils, then God does not exist.
  • (2) There are seemingly gratuitous evils.
  • (3) We can warrantedly infer the likely existence of actually gratuitous evils from the existence of seemingly gratuitous evils.
  • (4) Therefore, God is unlikely to exist.

The critical premise here is (3). As Rowe’s critics point out, this premise relies upon a ‘noseeum’ inference. In other words, it relies upon the assumption that if there were God-justifying reasons for allowing some evil we could expect to see them. This is something skeptical theists take issue with. The question is whether they are right to do so. To figure this out we need to consider their position in a little bit more detail.

2. Skeptical Theism and the Noseeum Inference
As Wielenberg points out, skeptical theism has two components: (i) a theistic component and (ii) a skeptical component. The theistic component is relatively straightforward. It consists in either the belief in God as classically understood (i.e. as a perfect being), or as understood by holders of some particular faith. Wielenberg uses a specifically Christian version of theism in his analysis because that is the version held by those toward whom he directs his arguments.

The skeptical component is slightly more complicated. The general gist of it is that we should be skeptical of our ability to fully know what God knows and that this skepticism undercuts the noseeum inference at the heart of Rowe’s argument. A number of more specific conceptualisations of the skepticism have been offered over the years. There is, for example, William Alston’s version, which focuses on different parameters of cognitive limitation that seem to apply to humans; and there is Michael Bergmann’s version which focuses specifically on the representativeness of our knowledge of good and evil and the entailment relations between the two.

Wielenberg doesn’t weigh the pros and cons of these different conceptualisations. Instead, he suggests the following as a version of skeptical theism that captures the core idea and does justice to some of the leading conceptualisations (most particularly the Bergmannian form):

SC1: It would not be surprising if there are possible goods, evils, and entailments between good and evil that are beyond our ken (but not beyond the ken of an omniscient God).

Skeptical theists think that a principle like SC1 is sufficient to undermine Rowe’s argument from evil. Are they right to do so? Here’s where Wielenberg’s meta-argument enters the fray.

3. The Need to Distinguish between General and Specific Noseeum Inferences
Wielenberg’s argument is that, to date, participants in the debate about skeptical theism and Rowe’s argument have paid insufficient attention to the difference between general and specific versions of the evidential problem of evil. The failure to do so means that the ability of skeptical theism to undercut the problem of evil is overrated, at least when that view is proffered in response to more specific versions of the problem.

Allow me to explain. The general and specific versions of the evidential problem work like this:

General Evidential Arguments: There are many instances of seemingly gratuitous evil; therefore there are probably some instances of actually gratuitous evil; therefore God does not exist.

Specific Evidential Argument: Specific instance of evil E is seemingly gratuitous; therefore E is probably actually gratuitous; therefore God does not exist.

To put it another way, general evidential arguments say ‘Look, there are all these instances of evil that seem to be gratuitous. They cannot all be necessary for some greater good. Therefore, it is likely that at least one of them is actually gratuitous.’ And specific arguments say ‘Look, there is this specific instance of evil. We have tried really hard and we cannot come up with a God-justifying reason for allowing this evil. Therefore, it is likely that this specific instance of evil is gratuitous.’
These argumentative forms rely on different noseeum inferences:

General Noseeum inference: Moves from the existence of some seemingly gratuitous evils to the existence of at least one actually gratuitous evil.

Specific Noseeum inference: Moves from the seemingly gratuitous nature of E to its actually gratuitous nature.

The differences are crucial because it is much easier to be skeptical about general noseeum inferences than it is to be skeptical about specific ones. The general noseeum inference confidently assumes we should be able to ‘see’ god-justifying reasons for allowing evil wherever they may arise. A principle like SC1 successfully undermines such confidence. But the specific noseeum inference does not share this feature. It assumes merely that we should be able to see god-justifying reasons in some particular case. A principle like SC1 cannot undermine our confidence in inferring from that particular case.

This can be demonstrated more formally. Let’s take Rowe’s case of the fawn suffering in the forest fire as an example of a specific evidential argument from evil. It fits the bill because it points to one particular instance of evil and makes inferences about its likely gratuitous nature (Wielenberg calls this the ‘Bambi’ Argument). Now consider the following two variations on skeptical theism. The first is SC1, which we already had, and the second is SC1a which is a more detailed variant on SC1:

SC1: It would not be surprising if there are possible goods, evils, and entailments between good and evil that are beyond our ken (but not beyond the ken of an omniscient God).

SC1a: It would not be surprising if there are possible goods, evils, and entailments between good and evil that are beyond the ken of human beings (but not beyond the ken of an omniscient God) but it would be surprising if any such possible goods, evils, or entailments had anything to do with fawns.

SC1 and SC1a are logically compatible. SC1 is a general and vague type of skepticism; it doesn’t rule out the possibility of sound moral knowledge in particular cases (indeed, that possibility is something skeptical theists need to preserve if they are to avoid other problems with their position). SC1a is merely adding to SC1 a specific case in which we can expect to have pretty sound moral knowledge.

And here’s the critical point: because SC1 and SC1a are logically compatible, SC1 cannot by itself undermine Rowe’s specific evidential argument from evil. If a proponent of SC1 tried to challenge the argument, they could always be rebuffed on the grounds that SC1a (which is consistent with their general skepticism) does not undermine the argument. This is illustrated below.

In other words, to defeat a specific version of the evidential problem you need to have a specific version of skeptical theism — one that accounts for our inability to make warranted inferences about the likely gratuitous nature of some specific type of evil. You cannot simply fall back on general formulations of skeptical theism.

That’s Wielenberg’s meta-argument and he tries to leverage it to his advantage in formulating the argument from abandonment and suffering. I’ll talk about that some other time.

No comments:

Post a Comment