Philipse on the Probability of Theism

So, in Chapter 11, Philipse starts talking about whether or not theism is probable, and what it might mean to determine that. However, what we see here — and have already been seeing in previous chapters — is an odd sort of issue based on the fact that Philipse himself both seems to want to go after theism in general but focus on Swinburne specifically. Thus, Philipse ends up focusing very much on Swinburne’s specific views while still talking about what theists would do or problems they would have in general. It seems that Philipse wants to focus on Swinburne at least in part because Swinburne accepts some of the issues Philipse has with theism and so at least in general more directly addresses those concerns. In fact, we see on a number of occasions Philipse using Swinburne to argue for Philipse’s main points. The problem with this approach is that Swinburne’s view isn’t that of all or potentially even most theists, and so if someone isn’t convinced by Philipse’s arguments they aren’t likely to be convinced by Swinburne’s either, and also won’t find the discussions of the specific solutions Swinburne advances and the problems Philipse has with them all that interesting. Yes, there are solutions and issues with them, but those solutions are addressing problems that many theists think are only issues if you buy somewhat dubious premises and propose a rather odd solution to those problems. This makes chapters that claim to make general points but that focus on Swinburne specifically seem somewhat irrelevant.

Here, in this chapter, I want to ignore all of the points about Ultimate Explanations and Swinburne’s specific use of that and the problems of it. I’m not convinced that even empirically any of this matters, especially considering that I rejected the need for “immunization” of theistic belief last time. I’m also not convinced that the right way to determine which is the more reasonable theory is by using probability — I’m inclined towards Quine’s “Web of Belief” model and so think general fit is better — and am certainly not convinced that Bayesian approaches are the right ones. So most of the chapter is predicated on my accepting premises that I don’t accept, and so the specific arguments aren’t that interesting. Thus, I’m going to focus somewhat briefly on two specific points. The first is a discussion over how probable a belief must be before we are justified in accepting or believing it, and the second is a discussion over the empirical background, which is about as close as Philipse gets to actually arguing for the theistic belief being improbable.

So, let’s start with the first point. Philipse talks constantly about “religious belief” in that section, and talks about “justified” in that context. When he talks about how philosophers view “justified”, and particularly when he talks about it having to be “highly probable”, he ends up shifting definitions here, talking about justified as it is used in “justified true belief” … which is to say, in terms of knowledge. But all we’re talking about here is which theory is to be preferred given the evidence we currently have. That’s probably not a knowledge claim. Sure, if we want to claim that we know God exists based on that reasoning, we’d want a very high probability, but to merely say that it is the most probable given what we know at the moment should only require it to be more probable than all of the competing theories … and Swinburne wanting it to have a higher probability than 1/2 guarantees that. So unless Philipse wants to demand that before we can reasonably think that a theory is the best candidate we have to know that it is the true one, or else wants to insist that one cannot reasonably believe that the best candidate we have is true, he’s just confused here about what belief and justification for belief has to be, conflating belief and knowledge.

Now, the next issue is when Philipse discusses the “empirical background knowledge”, which is critical for a Bayesian analysis and, it seems to me, provides the best reasons to think that Bayesian analysis is less than useful. While Philipse points out that some Bayesians think that the analysis can be subjective, in order to work as an argument against anyone else the probability of a theory given the empirical background knowledge has to be objective enough that your opponent can’t just reject your probability and blunt your argument. Thus, it can’t depend on things that you believe but that others might not. So let’s look at a couple of possible arguments that might make theism improbable based on the empirical background knowledge.

I’ll start with Philipse’s. His main argument is that God, as defined, is a personal being with consciousness, but he argues that “… all empirical investigations suggest that mental phenomena cannot exist without neural substrata.”[pg 205] In short, his big argument here is that you can’t be conscious unless you have neurons and thus are physical, and God is a non-physical spirit. Philipse has ridden this rather dubious argument for the entire book, and it’s still dubious here. First, if we are talking about an “ultimate” consciousness, then if it exists it would have to be able to compute without any limits. But any physical implementation of consciousness would have limits. Thus, an ultimate consciousness would have to be non-physical to avoid physical limits. Philipse could reply, then, that this would mean that an ultimate consciousness is impossible, but then he’d need far more evidence than “So far, all the conscious things that we’ve found are physical!” to demonstrate that. Second, AI is not going to have a neural substratum and we think that it is at least possible that we could get a conscious AI, and there is no empirical evidence that it can’t and at least some empirical evidence that it might be able to from AI implementations. So we have good reasons to find this purported piece of empirical background knowledge a bit dubious.

Richard Carrier, in a recent post addressing Swinburne and another Bayesian theist, brought in another argument:

If we count up all the things in history we at some point couldn’t explain, or thought was explained by magic or ghosts, and then securely found out what the actual cause was (so that it is now approximately a universally accepted fact of science or history), how many of those things turned out to be magic or ghosts? If the answer is zero (and it is…and anyone who denies that, is literally insane), and the number of those things is in the millions (which have reached that degree of investigation, so that it is now a known fact of the world what causes them; not just a belief or speculation), then the prior probability the next thing you ask the cause of will have been caused by magic or ghosts is logically necessarily millions to one against. And if the number of such things is in the billions, it’s billions to one against; if in the billions of trillions, then billions of trillions to one against. There is no rational escape from this consequence.

Well, there is indeed one: call it what it is, the inductive fallacy. This is essentially like saying that you’ve examined millions and millions of swans and so if someone says they’ve seen a black swan the empirical background knowledge makes that radically improbable. The problem here is not so much in saying that it is reasonable to believe that there are no black swans given what we’ve empirically examined, but is instead in choosing to use that as an argument that black swans don’t exist when someone gives you a reason to think that they might. The same thing applies here: if we think that in this specific case that a supernatural explanation makes the most sense for other reasons, saying that we’ve never had one of those doesn’t impact that assessment because it doesn’t — and can’t — address the reasons we had for preferring the supernatural explanation.

But the point here is not to refute these two arguments, or even to say that it is unreasonably for them to hold them (although I think that these particular arguments are so bad that it is indeed unreasonable to hold them). What is important is that these are justifications for very important premises that will greatly impact the prior probability Philipse and Carrier assign to the theistic hypothesis … and are premises that someone might, at least, reasonably not accept. No one need accept Philipse’s idea that mental activity must have a neural substrata or Carrier’s idea that the success or failure of previous supernatural explanations are relevant to this one. And as soon as someone does that, the whole Bayesian analysis — at least, one using priors, and it seems like there is little reason to use Bayesian reasoning if you don’t include priors — goes out the window. All I have to do is say “I don’t think this piece of empirical background knowledge is true” or “I don’t think this piece of empirical background knowledge is relevant” and the whole analysis collapses. Thus, the empirical background knowledge, well, has to be knowledge for it to work here: things known to be true and known to be true by all parties. And the implications have to strongly follow. This is, in fact, a pretty difficult thing to achieve, and neither Philipse nor Carrier achieve it.

Now, the thing is that it’s reasonable — or at least, not unreasonable — for Philipse and Carrier to hold their beliefs. Philipse is a physicalist and denies the existence of immaterial things, and Carrier is a naturalist who denies the existence of the supernatural. As beliefs, they certainly have sufficient reason to believe those things. But others have sufficient reasons to believe otherwise, or at least to withhold judgement on those propositions. And as soon as they do, the priors falter and their arguments for why theism is improbable evaporate.

This result is consistent with my general view on belief, which is that we assess the “likelihood” of a proposition or theory being true based on how well it aligns with our current “Web of Belief” (which includes but is not limited to what we know). If someone is a naturalist, any supernatural explanation will seem incredibly unlikely. If someone is not, then that it is supernatural might count in its favour, or at least will be neutral. And I will argue that this is perfectly reasonable. Moreover, it’s all we can do. Any objective Bayesian reasoning will try to make the assessment give an initial assessment that is the only reasonable one to accept and then move by objective steps to new probabilities as new evidence is introduced, but to do so it can’t rely on anything that we don’t solidly know and so can’t account for differing beliefs. Either we can’t believe what we don’t know — which is wildly impractical — or it will splinter into subjective Bayesian as soon as there’s a belief that is in dispute that at least one party thinks is relevant. And subjective Bayesianism is nothing more than a mathematical complication of what we’d do naturally anyway, as the number it comes up with is meaningless without the context that spawned it.

This, then, is the issue with arguing that theism is “improbable”. You need an objective standard for that to have meaning, but that standard has to be based on subjective beliefs. In general, the insistence of probabilities strikes me as a way to claim objectivity while hiding the subjective premises that underlie the assessment … which explains why these arguments always devolve to arguing over those specific premises in the end.

Next up, cosmological arguments.

Advertisements

Tags:

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: