Tuesday, January 05, 2010

A response to Mike on the Argument from Desire

I am wondering what your response is to the Bayesian calculation that I gave would be.

H1) Humans are constructed by God is such a way that they can be fulfilled only in relationship to God.

H2) It is not the case that humans are constructed by God in such a way that they can be fulfilled only in relationship to God. This can be true if atheism is true, or if God doesn't care,

D= the existence of "heavenly" desires

Now, notice I don't need the claim that these desires couldn't arise through evolution. In fact, as I did the calcuations, it's 70% likely that these desire would arise through evolution, and the calculation still works!

Now doesn't it seem to you that D is more likely given H1 than H2?

22 comments:

Doctor Logic said...

Victor,

Consider how the "heavenly desire" is defined. It's an abstraction of having all of our desires attained at once. Yet the way the desires are attained is left unspecified. The heavenly desire is nothing more than a "meta-desire", not for something specific, but for something non-specific.

We have a lot of desires, D1, D2, D3,..., etc. Some of these desires conflict. For example, the desire to meet and overcome a challenge is frustrated by attaining our desire to succeed in our endeavors. It's not possible to meet all these desires at once.

This creates two problems for the argument. First, if I have powers of abstraction, I will always abstract to an ultimate desire of meeting all my desires. That is, the probability of a heavenly desire is unity under naturalism. Second, it raises the possibility that meeting the heavenly desire is impossible because the desires from which it was abstracted are conflicting.

Finally, the God theory which you are arguing for is both vague and fine-tuned. You can play the same game with anything.

Consider the explanation for, say, a bank robbery:

H1) The actual bank robber, whoever he may be.

H2) Local ne'er-do-well Fred Bloggs, who had motive and opportunity, is the same height and build as the masked robber, but the lack of other evidence makes us only 70% confident he is the culprit.

In a Bayesian analysis, H1 will win. H1 always gives you a 100% chance of predicting the observed robbery. However, it wins because of how much we don't know about the proposed solution. In effect, H1 is merely defined as the solution to the problem, but there's nothing to the theory at all.

Fine-tuning a solution to the observations is a standard practice, but only when you get some highly specific predictions in the process. Theorizing that Fred is the culprit makes very specific predictions about fingerprints, money laundering, etc. H1 predicts nothing because you don't know who the bandit is. Moreover, if you let the bandit be capable of impossible things (like, say, shrinking 10 million in cash into his earring, or being able to wish away his fingerprints) the argument for H1 becomes even more absurd.

Clayton said...

Victor,

Consider two hypotheses:
(H1) There is a powerful, invisible, and very wise creature that exists outside of time that loves large, orange fruit.
(H2) There is no such thing.

P = Pumpkins exist.

I'd say that P is more likely given H1 than H2, but I don't know why this matters since no one assigns a very high prior probability to the existence of the being described in H1. Isn't the atheist reply going to be basically the reply I imagine you'd offer having considered my (er, Hawthorne's) pumpkin example?

Steve said...

DL,

I'm not sure about your bank robber example. The actual bank robber did do it, didn't they? Well, clearly they did. So there is nothing wrong with such the bayesian inference implicit in your example.

So what is the point of this example exactly? It's not a very interesting or informative "theory" but that doesn't seem to be your point.

Steve

Victor Reppert said...

Clayton: Yes, the existence of pumpkins does confirm the existence of an orange-fruit-loving supernatural being. The existence of bananas and cherries disconfirms it, if we are thinking of a special liking for orange fruit as opposed to red fruit or yellow fruit. But of course, if the prior on this particular supernatural being is low to begin with, then the "confirmation," such as it is, doesn't do much for us. I believe I calculated the confirmation in a previous post for someone who rates theism very, very low, and even though you get a shift in the numbers, it's very little skin off the nose of the unbeliever. But if we are asking "Does it count at all" the answer would have to be yes. Some people think the Christian God and naturalism are the two main "live options," and if they do, does "heavenly desire" count in favor of Christian theism? I think the answer is yes, just as I think certain types of evils count in favor of atheism even though I am a theist.

Doctor Logic said...

Steve,

So what is the point of this example exactly? It's not a very interesting or informative "theory" but that doesn't seem to be your point.

It's not a theory at all. It's the original question.

Question: What's the tallest building in the world.

Answer: The building which has the greatest height from lobby to top floor.

Question: Who's the prettiest actress in the world.

Answer: Julie. That may not be her real name, but that is what we will call the prettiest actress by convention.

Question: Who is responsible for holding up the First Savings & Loan?

Answer: The actual bank robber, whoever he (or she, or it) might be.

None of these answers are answers. They are all descriptions of the answers. They're just restatements of the question. You know. Like on Jeopardy.

Jesse said...

Dr. Logic, you write:

::Consider how the "heavenly desire" is defined. It's an abstraction of having all of our desires attained at once. Yet the way the desires are attained is left unspecified. The heavenly desire is nothing more than a "meta-desire", not for something specific, but for something non-specific.

I really don't think you can read Lewis' illustrations of this desire, which is a felt, concrete experience, and say this is a desire to have all temporal desires attained at once. For instance, as Lewis is reading about the Garden of the Hesperides and finds himself gripped, caught in the throes of an intense longing for something that's clearly not just the Garden of the Hesperides, can we really say the object of that longing is just an abstraction for something indefinite? Just that fact that he's desiring something means it must be something definite, "The form of the desired is in the desire", as Lewis put it.

::We have a lot of desires, D1, D2, D3,..., etc. Some of these desires conflict. For example, the desire to meet and overcome a challenge is frustrated by attaining our desire to succeed in our endeavors. It's not possible to meet all these desires at once.

You can abstract a theory of happiness, defined as “the possession of all good things in aggregate,” and pose it as a normative end, that is, considering life as a temporal whole (and this is done based on the needs of human nature, which don't always rise to the level of felt desires, and are contradicted by unnatural desires); this, however, is done as an approximation based the fact that there is some one thing we desire as an end in itself.

::This creates two problems for the argument. First, if I have powers of abstraction, I will always abstract to an ultimate desire of meeting all my desires. That is, the probability of a heavenly desire is unity under naturalism. Second, it raises the possibility that meeting the heavenly desire is impossible because the desires from which it was abstracted are conflicting.

Taking your second point first (and touching on my last point), fulfilling all natural desires would preclude any conflicting desires, otherwise our natures aren't one; therefore the question doesn't imply all possible desire. Second, though you can abstract to an ultimate desire, or desire-at-large, you cannot desire desire-at-large, that is, you cannot desire an abstraction. What you’re leaving out of the picture is that the abstraction is coming from the fact that we desire something as an end in itself – you’re painting it the opposite way.

Jesse

Steve said...

DL,

So does that mean that you think the explanation for the existence of the world (and the 'heavenly desire') is, trivially, God?

Therefore it's trivial that God exists and you're obviously a theist.

Something seems to be wrong here. I agree that there is something wrong with the "theories" you are mentioning. But they do at least have the virtue of being true.

Like jesse, I also disagree with your reading of the relevant desire as a "meta-desire" but we can think about that later.

Steve

Doctor Logic said...

Jesse,

Just that fact that he's desiring something means it must be something definite, "The form of the desired is in the desire", as Lewis put it.

I don't agree. Suppose that I am seeking a new girlfriend. I don't long to be with my first love (who I broke up with). Rather, I long to feel the way I did when I fell in love for the first time. That's not a longing for someone specific. It's more like a longing for a feeling.

The same goes for a great meal. I may love a good turkey dinner above all else, but I don't want it every day. I want variety, but I also don't want anything less delicious than a turkey dinner. I want to feel as if I am eating my favorite turkey dinner after having been deprived of turkey dinner for a while. How is that a desire for something definite?

this, however, is done as an approximation based the fact that there is some one thing we desire as an end in itself.

Not me.

Taking your second point first (and touching on my last point), fulfilling all natural desires would preclude any conflicting desires, otherwise our natures aren't one; therefore the question doesn't imply all possible desire.

Why should I think our natures must be unified? That just sounds like a way of dismissing the our conflicting desires.

Maybe you are suggesting that denying my "unnatural desires" is one of my desires?

And what exactly is an unnatural desire?

Doctor Logic said...

Steve,

(It's SteveK, right?)

So does that mean that you think the explanation for the existence of the world (and the 'heavenly desire') is, trivially, God?

No! It means God is word representing some questions. God is not an answer to the questions, but just a restatement of the questions.

It's not the "right answer" because it's not an answer at all.

Imagine you're at the airport, and you go to the information desk.

You ask "Where is the American ticket counter?"

The nice lady behind the desk responds "It is the place where people pick up tickets and passes for American Airlines."

Did the person at the information desk answer your question? No! She just defined the American ticket counter in your question. She didn't tell you where it is. And your response would be "Well, duh! But where is it?"

One way to tell that the "nice" lady behind the counter is hosing you is that she cannot be contradicted. If the ticket counter is upstairs, she's not contradicted. If the ticket counter is in the basement, she's not contradicted. In fact, no matter where the ticket counter turns out to be, she won't contradict that location.

With God, it's the same thing.

Walks to theological information desk.

"Excuse me, Miss. Can you tell me what caused the universe?"

"Oh yes, Sir. God."

"What is God?"

"God is that which caused the universe, Sir."

"Mm Hmm. How does that work?"

"Whatever God wills, happens. God willed the universe, and so it was."

"I see, Miss. And what does God will?"

"I already told you, Sir. God wills whatever happens, past, present or future."

"Ah, so what will happen tomorrow?"

"Whatever happens tomorrow."

"Are you being paid for this?"

Steve said...

DL,

Actually it's Steve Lovell not SteveK, but no worries.

I understand what you're saying but it doesn't seem like an objection to the bayesian argument from desire, it seems like an objection to all "God-talk".

Now it may be a good objection, but I'm really not seeing how to get an objection to VRs argument from it.

You admit that (VR's) H1 makes D much more likely than does H2. Indeed you seem to allow that H1 is sufficient and perhaps even necessary for D ... since you seem to allow that H1 may be just a re-description of D. So are you saying that despite this, H1 isn't confirmed (to any extent) by D?

Certainly, if you allow D and H1 to be identical, then if you admit D you must also admit H1. This is why I cheekily suggested that you must be a theist. But this whole line of argument looks like it's supposed to be independent of denying the truth of D ... which is why I'm saying I don't understand your objection. I'm almost tempted to say that there isn't an objection there, and you're just labouring under the misconception of their being an objection to the argument in this general area.

Steve Lovell

Doctor Logic said...

Steve,

Sorry about the identity mix-up.

Whenever you talk about a hypothesis, you have to view your hypothesis as one class of specific theories out of the total set of all specific theories. This is going to be a little verbose...

The easiest way to explain this is to consider fitting a curve to some data points, since most people have done this in high school or college algebra.

As you probably know, you can solve N equations in N unknowns. If I have 2 data points, I can fit my data points to a 2-parameter polynomial (a line). If I try to fit a larger order polynomial to my 2 data points, I'll have too many unknowns, and I won't be able to complete the fitting of my theory, nor confirming the theory's correctness.

Now, suppose there are two researchers, Nate and Theo. Both Nate and Theo have multiple different realms in which they are collecting data.

Nate tries to fit data in each realm to the lowest-order polynomial he can, e.g., a line or a quadratic, etc. As it turns out, Nate is quite successful, and Nate discovers many regularities in the data in each realm. He even begins to piece together connections between the realms, creating a unified theory. Since Nate is working with low-order polynomials, he is able to accumulate enough data to fully fit his curves, and get into confirming his theories in each realm. This gives Nate confidence in his inferences.

In contrast, Theo eschews low-order polynomials. Theo jumps right out and claims that the world is governed by an infinite-order polynomial that governs all domains. There's no point in Theo doing any math. It would be pointless. If he fits his dozen data points, he still has infinitely more data points to collect before he's done fitting to data points, and can move on to confirming his theory. He will never get to the point of confirming his theory.

And, if we're considering infinite parameter theories, there are an infinite number of alternative theories consistent with whatever finite observations we accumulate.

To some it might appear that Nate is at a disadvantage. Nate can be wrong, whereas Theo can never be wrong. If Nate's data accidentally led him to make the wrong inference, he could get a data point tomorrow that would force him to revise his inferences. Nate might have to up the order of his polynomials (by as small an order as possible, of course).

On the other side of the lab, Theo can sit back and admit any data point that comes along. In fact, Theo can do this in any realm he wants. No point in him doing anything until he has infinite data.

Now, which of these two investigators is acquiring any knowledge or making any inferences from experience? Only Nate. Theo is basically just sitting on his hands. Since there are an infinite number of infinite-parameter theories consistent with any data set, Theo should have zero (infinitesimal) confidence in any particular theory.

Theo is just guessing that the solution is infinite-order. It could be second-order, tenth-order, or millionth-order, or quadrillionth-order, but Theo isn't interested. Indeed, Theo will never know anything because he'll never collect enough data points to validate his theory. Since Theo is forever in the curve-fitting stage of his enterprise, he can never be wrong. Alas, he can never be right, either, because if no data point is ever a test of his theory, neither can any data point ever be a confirmation of his theory. In fact, Theo doesn't have a theory at all. He will never have one because he will never get enough data points to know anything or explain anything at all.

cont...

Doctor Logic said...

Steve,

Suppose that Nate and Theo both receive unusual reports that some rocks fell skyward instead of Earth-ward. Nate, with his commitment to the smallest order polynomials, will have a tough time accounting for the strange report. He will be looking at the statistics of false reports to see if it is more likely that the report is false than that rocks fell towards the sky. Eventually, if the phenomenon is observed with sufficient statistical weight, Nate will have to revise his theory and go to a higher-order polynomial theory.

However, Theo has no trouble with the strange report. With an infinite number of parameters to play with, the laws of physics are free to change from moment to moment. Indeed, Theo really has no reason to suppose that things would fall Earth-ward in the first place. He's not in a position to confirm that things fall downward.

There's something even more peculiar that Theo can do. Theo can make claims about his theory that can only be validated after an infinity of data points have been collected. For example, Theo could claim that the number of protons that will ever be created in the universe is even. Theo can afford to make such claims with impunity because he's got an infinite number of parameters to play with - he can make today's data jive with any eventual conclusion he wants. Theo can claim his theory says you should rub your tummy and pat your head if you know what's good for you, and that you'll only realize this fact after the infinity of data points have rolled in.

Obviously, Theo is the theist. No matter how much information we record, it won't contradict a theory of God. There's always the infinity of data points we have yet to collect. For example, Theo can claim that God is good in the (infinite) long run, even if he looks evil in the short run.

Now consider H1 and D. Why should God will that we desire him? It's easy to imagine a God who doesn't will us to desire him. And, had we found ~D, Theo would have been fine with that, too. Theo doesn't have to explain anything. He's perpetually in the curve-fitting stage. If Theo can't predict with his theory, he can't confirm with his theory. Yet he advances H1 as if the truth of D is confirmation of H1. In effect, Theo is defining H1 so that P(H1|D) = 1.

Meanwhile, Nate is past the curve-fitting stage, and so D is potentially confirming evidence for H2.

If we were to even the playing field, we should define H2 as that naturalistic theory which is consistent with physics and in which P(H2|D) = 1.

Steve said...

Hi DL,

Feel we're going round in circles here both thinking the other has lost the plot.

(1) P
(2) P --> Q
============
(3) Q

What you're saying sounds very like saying the following in response to the above.

"I'll accept (1) for the sake of argument, my problem is with (2). You've just defined Q so that it follows from P, but if you can define it that way, and accept this inference you're heading for epistemic madness."

The problem with such a reply is that it doesn't deny that the argument is sound. If the argument is sound, then the conclusion is true.

To escape this you're going to have to seriously fine tune your objection.

Steve Lovell

Doctor Logic said...

Steve,

Let me put it this way.

H1) Humans are evolved by evolution such that they have D.

H2) Humans evolved without D, or humans are designed by God (either with or without D).

Clearly, H1 wins, right? P(D|H1) = 1. P(D|H2) < 0.5.

If H1 wins, then that means I've constructed an argument from desire for unguided evolutionary biology (the kind that yields D)!

Now, I may have misinterpreted your criticism of my comments. Maybe you mean that VR's argument is valid, just as is my argument above.

However, VR's argument is certainly not a sound argument for theism unless my version is a sound argument for unguided evolutionary biology.

What I am saying is that one can always pick H1 to be fine-tuned to predict precisely D. I could even choose

H1) Humans were designed by chickens to have D.

H2) All the alternatives.

And again H1 wins, and I end up with an argument from desire that we were designed by chickens.

So, clearly, we can't go around fine-tuning H1 willy nilly. VR's argument is not an argument for theism, unless my argument is an argument for chickenism.

(Besides, unless VR offers a predictive theory of God, he shouldn't be offering Bayesian arguments of any kind. He's still fitting his theory to the data, and so he can't claim it predicts D.)

J said...

looks a bit like a subtle version of the supposed "ontological argument."

Believer X may have a concept of something he takes to be a monotheistic God, and desires to be godly, or please God, etc. We have no way to prove those concepts match up with something apart from the mind...... Osama Bin Laden may have a concept of God as well... obey, so forth. Then Lady Gaga may have her own god-concept as well........

Steve said...

DL,

Okay, I think we're getting somewhere now. You seem to have stepped back from one of your claims though.

You were comparing God and "the actual bank robber". Since, by definition, the actual bank robber did it, then also by definition God would be the explanation for our heavenly desires.

You have taken this back and replaced it with a similar sounding but logically different, and much better objection.

If I understand correctly you think the AfD is illicitly doing something like this:

H1: God exists and we desire him
H2: Either God doesn't exist or we don't desire him

Since H1 entails that we desire God, then it is confirmed by the fact that we desire him.

This is of course true relative to H2. Unfortunately it is not true relative to

H3: God doesn't exist and we desire him.

or indeed relative to

H*: P and we desire God.

Now the question is whether the AfD really works that way, is it just hamfistedly sticking together the concept of God with the concept of a desire for God in an arbitrary way? To my mind the answer is obviously 'no'.

This concept of God as our "end" goes all the way back to Aristotle and Plato (if not further) and stems from a unified metaphysics.

Logically there is no difference between prediction and retro-diction ('accomodation' as philosophers of science call it), so I disagree that bayesian arguments must be predictive. The problem with accomodation tends to be with distinguishing which cases are like H1 and which represent a unified account of things. This is a matter of the simplicity of the theory in question ... which isn't easy to judge and with appropriately defined terms almost anything can be made to seem "simple" ... I don't know of any decent metric for simplicity, and I'm not sure anyone else does either (happy to be condradicted here) but my intuitive sense is that a theism which predicts a desire for God is not significantly more complex than a theism which is silent on the matter ... so that the desire can be used in the way that VR suggests, though I agree that one should use such arguments with care.

Steve Lovell

Jesse said...

Dr. L,

::I don't agree. Suppose that I am seeking a new girlfriend. I don't long to be with my first love (who I broke up with). Rather, I long to feel the way I did when I fell in love for the first time. That's not a longing for someone specific. It's more like a longing for a feeling.

First, as Lewis said, "The longings which arise in us when we first fall in love, or first think of some foreign country, or first take up some subject which excites us, are longings which no marriage, no travel, no learning, can really satisfy." However, the feeling you seek you seek as a result of finding a definite girlfriend, not of finding the abstraction girlfriend. Again, Lewis also went on to say, though we can then desire the desire itself, the value of the original desire lay in the object of which it was a desire.

::The same goes for a great meal. I may love a good turkey dinner above all else, but I don't want it every day. I want variety, but I also don't want anything less delicious than a turkey dinner. I want to feel as if I am eating my favorite turkey dinner after having been deprived of turkey dinner for a while. How is that a desire for something definite?

Because only something definite will fulfill it!

::::this, however, is done as an approximation based the fact that there is some one thing we desire as an end in itself.

::Not me.

Everyone desires happiness for the sake of itself and nothing else, and all else we desire we do so as a means to this end.

::Why should I think our natures must be unified? That just sounds like a way of dismissing the our conflicting desires.

Because we're one being in ourselves.

::Maybe you are suggesting that denying my "unnatural desires" is one of my desires?
And what exactly is an unnatural desire?

Any desire which frustrates the attainment of the end our natures necessarily seek. That is, any desire not intrinsically a means to happiness.

J said...

Believers misuse Bayesian arguments, like, routinely (as Dr L may have suggested).

Bayes shows that a belief in an outcome might be justified (tho' not certain) given undisputable facts. Say guilt of a person accused of a crime. But barring actually like video of God (and reliable) or real miracles, there are no facts, no evidence to input into the little bayesian formula .

Anecdotal reports, or beliefs, mysticism (someone thinks he perceived God, etc) are not really data. Really Bayesian arguments used correctly actually show the great unlikelihood of the supernatural (ie, considering everyday when nothing miraculous or mystical occurs....). That was one of Rev. Bayes unsettling inferences was it not.

Doctor Logic said...

Steve,

Now the question is whether the AfD really works that way, is it just hamfistedly sticking together the concept of God with the concept of a desire for God in an arbitrary way? To my mind the answer is obviously 'no'.

Obviously or traditionally?

I think that to make it obvious, you need a framework that describes the mind of God, not just what he has willed. You need to say "God intends X, therefore he had to will Y". And this must lead to "And a side effect of God's intent to bring about X is Z." (Otherwise, you're effective identifying X and Y.) You're going to end up making predictions about what God would or would not do.

Logically there is no difference between prediction and retro-diction ('accomodation' as philosophers of science call it), so I disagree that bayesian arguments must be predictive.

There has to be a mechanism (e.g., the thinking of God), even if that mechanism is non-physical, or else the inference doesn't work.

Suppose I seek an explanation for X. It goes in steps, starting with laying out the problem.

1) I wish to explain X with a theory.
2) The theory I seek is one which will predict X.

This happens before I even have a hypothesis. The problem I see begins when people do the following:

3) When I find this theory, it shall be called "Albert", by definition.

Nothing wrong with coming up with an arbitrary name for the theory one is looking for. However, they sometimes follow this up with:

4) "Albert" explains X!
or
4b) "Albert" predicts X!

The fallacy is when people think that (4) is equivalent to having found the explanation for X. Explanations are not mere restatements of the problem to be explained. "Albert" doesn't predict (or retro-dict) anything at all.

Really, (4) should be:

4') "Albert" is what we will call the explanation for X.
or
4b') "Albert" is what we will cal the theory that predicts X.

If you're going to explain D, you have to have a model with a mechanism that predicts D (or at least prefers D).

If we declare that mechanisms aren't needed, I can substitute an unknown unguided principle in place of God in the AfD, and conclude there is no God.

So what is the mechanism behind God's will to implement D? That mechanism has to predict something else. If it doesn't, then, again, I can create the very same argument using an emergent physical principle that created D and nothing else.

This is why the problem of evil is such a big problem for theism. As soon as you try to make a predictive model out of God's thinking, the theory of God just breaks down into contradictions. The common response to the problem of evil is that God and his reality is so complex that we cannot hope to understand how he acts. But if that is so, then we cannot predict (or retro-dict) that he would create the universe or implement D either.

Steve said...

DL,

Thanks for your detailed response. I agree about this sort of problem with "place-holder-explanations" though I think you need to be careful or you'll end up with an infinite regress.

At some point we have to stop and say, X is the explanation of Y and there is no underlying mechanism. Some things have inherent causal powers which cannot be reduced any further. This is one of the reasons I've never been convinced, for example, that "mind-body" causation is a problem for the dualist. In principle it's no more problematic than any other brute form of causal connectivity. Anyway, that's another issue.

Still, When we say that "Albert explains X" we may mean at least two different things.

"Albert" like the name for any theory (placeholder or otherwise) is systematically ambiguous when used in conjunction with terms like "explains". Explanations
can be either

(1) real world causal or quasi-causal connections, or
(2) satisfying answers to "why" questions

Now until it is further developed, "Albert" certainly isn't an explanation in sense (2). However, it is, or the realities which Albert when fully formed will posit are, by definition, an explanation in sense (1).

Now again, it seems clear that we can say with certainty, that "Albert" is a true theory ... so I think you've shot yourself in the foot again with the problem I've been pointing out, but I still agree that there are real issues around Bayesian arguments in this area ... I just don't think you're quite putting your finger on them as precisely as seem to think you are.

On the more relevant matter of the real connection between the concept of God and the desire for him, I can do little more than suggest that you read some Plato, Aristotle, Aquinas, Boethuis, Augustine and indeed the Psalms. If you don't see the connection then you don't understand the traditions you are criticising.

Note I do use the word "traditions" here. I think the connection is obvious relative to the relevant traditions. I also think, like you I'm guessing, that a mere abstract God of theism without any broader metaphysical or religious context would be very difficult to bring into such discussions ... there's no telling what He/She/It might be like.

This will naturally run into further questions about the relevant "priors" of a worked out metaphysical system. And these are difficult questions ... one of the reasons I try to steer clear of Bayesian arguments. Still, any well worked out metaphysical system is detailed enough to run into issues in relation to "priors" and this applies as much to naturalistic systems as to theistic ones.

Steve Lovell

Doctor Logic said...

Steve,

My hat is off to you. I've been making this argument for a long time, and I think you're the first theist who actually figured out what the heck I was talking about, or acknowledged that there was an issue at all.

As you say, consideration of priors will play into any overall Bayesian picture. However, it can be helpful to look at individual Bayesian arguments to see how things stack up. The question is, does theism stack up in any of them? I don't think it does.

For example, in this case, you're suggesting that H1 is not artificially narrowed because, traditionally, theists assumed H1 rather than a more generic theism. However, why was H1 assumed? Is there another Bayesian argument somewhere that justifies narrowing theism to H1?

If H1 was narrowed only because of D, then the traditions don't help the argument. It just means that ancients fell for the same fallacy.

For the "traditions" to help the argument in a Bayesian sense, there must be something about the world that the traditional God hypotheses actually predict. If we line up a dozen theistic arguments that all are narrowed as fallaciously as the AfD, they won't lend each other mutual support. I can't think of a single Bayesian argument that supports theism.

Steve said...

DL,

Sorry to take a while coming back, I had to think a little more about this and didn't have the time.

I'm not sure that I understand your latest comment, so please do correct me if I have it wrong.

You seem to be saying that we need prior reason to go with the "traditions" rather than with a generic theism. If we don't have that, then (you seem to say) we will still end up with an H1 which gets much of it's prior probability from the generic theism and the bayesian argument will still fail for the reasons you explained above.

That seems like a sensible criticism to me.

However I do wonder whether you're playing fair. The same problems will infect any bayesian arguments for naturalism. Now I'm sure that naturalists would have a response to this ... and I can just about imagine how it might go (though having briefly tried, I've decided not to attempt it here). The issue is that for any naturalist response to such a critique there will be a parallel theistic response to your critique, and that seems to leave an impasse. Bottom line: A generic naturalism no more makes predictions that an generic theism.

All that being said, I think that certain bayesian arguments fair better in the context of a generic theism than do others. A bayesian argument from desire seems to me to assume a fairly rich form of theism. However certain versions of the argument to/from design have much more generality. They assume intelligence and power, and little else. I'd be tempted to say the same about arguments from miracles, but in so far as that is true I think it will just turn into a species of the argument to/from design. The intelligent design movement has had a lot of stick in the press ... and I've not kept up with all the arguments, but the basic ID line is that we can detect intelligence by solid empirical means. If this isn't true then we as a race are in serious trouble. So it seems to me that formally there can be no real problem with versions of the argument to/from design which argue for a fairly generic theism on broadly bayesian grounds. To my mind the main questions are not about the form of the argument but the details of the phenomena appealed to. Anyway, this is getting pretty fair off topic ... it was only intended as a response to your comment about not being able to think of a single Bayesian argument that supports theism. Perhaps I should just have said: "Well, I can".

Steve Lovell