Friday, May 8, 2009

Distinguishing multiple universes from design

One way to respond to certain design arguments, including both fine-tuning arguments and arguments from apparent biological design, is to make use of a multiple universe (MU) hypothesis. If there are enough universes (say, infinitely many), and there is the right kind of latitude in the random parameters, it is unsurprising that there be a world exhibiting just about kind of complexity you like, including intelligent life. And there is no further puzzle about why our universe exhibits this complexity, because there is a selection effect—only universes that have observers can be observed.

One might think that MU and Design hypotheses cannot be distinguished. That's why some design arguments are formulated with a disjunctive conclusion. But they can be distinguished: they have distinct predictions. The MU hypothesis basically says that we should expect to see just enough complexity and tuning needed to produce observers. The Design hypothesis makes it moderately probable that there would be more complexity and tuning because of the value of living beings that are non-observers. More genearlly, the difference is that the MU hypothesis involves only a tropism towards intelligence, while the Design hypothesis involves a tropism towards the instantiation of values. So in theory the two hypotheses could be distinguished.


Palamas said...

Alex: You state, "One might think that MU and Design hypotheses cannot be distinguished. That's why some design arguments are formulated with a disjunctive conclusion. But they can be distinguished: they have distinct predictions."

So I take it that you disagree with with Elliott Sober, who, in his book _Evidence and Evolution_, claims "The problem with the hypothesis of intelligent design is . . . that it doesn't predict much of anything . . . [because] we have no independent knowledge of the goals and abilities that the designer of organisms would have if such a being existed."

Why do you disagree with him and what predictions do you think ID makes?

Alexander R Pruss said...

Sober is commenting on Dembski-style ID. I am commenting on the theistic argument from design. Dembski-style ID makes no posits about the possible motives of the designer. And Sober's criticism seems quite justified.

But the theistic argument from design supposes a designer with good motives. And that leads to predictions, such as that there are no evils that God wouldn't be justified in permitting, etc.

Palamas said...

I'm not so sure Sober is limiting his critique to Dembski-style ID; for he states, "auxiliary propositions can be invented about the putative designers goals and abilities, but . . . [w]hat is needed is not invention of auxiliary propositions but the identification of auxiliary information that is independently supported."

Here I take him to be implying theistic design arguments that make God an omni-being.

I suppose the options here are to either (a) reject the idea that independently supported auxiliary information is needed or (b) provide the support. Do you think that other theistic arguments could provide the support needed (e.g., the ontological argument)?

What do you think?

Alexander R Pruss said...

I don't see why the auxiliary hypotheses need independent support. Consider the case of the electron. When the electron was first introduced into the scientific discussion, it was introduced as having two properties: being a small particle, and being negatively charged. Now it would be absurd to say: "Positing a small particle does not have empirical consequences. Of course, you get empirical consequences when you add that it is negatively charged. But that's just an auxiliary hypothesis, and hence needs independent support." And of course there is no independent support available. We have support only for the full hypothesis that there is a small negatively charged particle.

All that said, it is partly and maybe even wholly definitory of agents that they are responsive to reasons. There is an at least moderately high likelihood that a being smart enough to design the universe would be smart enough to see at least some of the objectively correct reasons, and there is some likelihood that he'd act on at least some of them them (three hypotheses: he is good, he is neutral, he is bad; let's suppose that they are approximately equally likely; then the likelihood he acts on objectively correct reasons most of the time is about 1/3).

Adam said...

I would be very interested to hear if you had any thoughts in response to Alister McGrath's perspective on natural theology, as articulated in his recent Gifford lectures (published as A Fine Tuned Universe), and also in his The Open Secret: A New Vision for Natural Theology. He makes use of abduction and "inference to the best explanation" (influenced largely by Peter Lipton, I think) with respect to the "fine-tuned universe." however, he shies away from seeking to establish "proofs" for the existence of God, aiming instead for "resonance."

Alexander R Pruss said...

No, sorry, I haven't read that yet.

Alexander R Pruss said...

Also, in regard to Sober, here is a dilemma. Either the inductive argument from evil has no evidential force against theism or it has evidential force against theism.

Suppose it has no evidential force against theism. Well, that might in the end be true--I actually think it is true--but to hold that simply on the basis of general considerations of the inconfirmability of theism is problematic. And I bet there are very few atheists who would accept the evidential irrelevance of evil.

Alright. Suppose, then, the problem of evil has evidential force against theism. Then there is logically possible empirical data that would, if it obtained, provide some evidence for theism: namely the data that there is no evil. As a matter of fact, the data does not obtain. But its possibility of obtaining is sufficient to show that theism is not the sort of thing for which there cannot be evidence.

Palamas said...

Interestingly, Sober anticipates this objection and responds in the following way: "Perhaps the design argument requires more knowledge of the designer's goals than the argument from evil does. Maybe the hypothesis that an [omni-God] exists makes predictions concerning how much evil there should be even though it does not predict the architecture of the vertebrate eye or the panda's paw" (167).

The problem I have with this response is this. Earlier in his book, he states that an hypothesis must be tested according to the principle of total evidence. But I think by parity a theistic hypothesis must be tested within the total theological system in which it is embedded. In other words, a God hypothesis become more or less probable given our observations of the world depending on the "auxiliary" theological hypothesis accompanying it.

What do you think?

Adam said...

You can read the full texts of his Gifford lectures here:

...if you are interested.

graham veale said...

I'm trying to put a paper on Sober together at the moment. (Just for the fun really, I'm just a High School teacher.)

One point on which I think Sober's critique falls down is that he assumes (without argument) that God is not a rational agent. That is we cannot make predictions about God, because God does not sufficiently resemble human beings in the correct way.
He tries to motivate this belief from Theistic responses to the Problem of Evil. But I think he has misread these. I don't think that the Theist is committed to any proposition that would make God sufficiently mysterious for Sober's purposes.

I'm working on the assumption that there is no way to bridge the gap between freqency data and degrees of rational belief that can avoid some subjectivity. And that if God can be considered as a rational agent and humans are rational agents then we can use frequency data to make predictions about what God might do, as long as the predictions are not too specific in their descriptions.
(I'm not sure that frequency data is the way to go in a design argument, but this is how Sober seems to want to construe it.)

That's not a great description of the paper, but class at door so no time to rewrite.

Has someone already made these citicisms? Is this paper worth pursuing?


graham veale said...

Oh, I agree his criticisms work for ID


Alexander R Pruss said...


He may be right in the special case of those who make the sceptical theist response to the problem of evil.

In general, I don't think we need frequency data. We need conceptual data about what good agents choose.

graham veale said...

Well, the paper concedes that. But I'm not even sure that Wykstra's response (which Sober references) needs to be put in that category.

There is a difference between "we can't see the whole picture" and we can't use human categories to evaluate God."

I agree with you about frequency data. I don't think that subjectivity can be avoided by this route in any case, and I try to point this out. But at the same time God is an agent, and we're agents. SO unless Sober can provide an argument for ruling this out (say an argument for Divine Simplicity - I think that would work) I think his critique fails. I want to saty as close t his argument as possible.
I'm also unhappy with the way he construes the Chance hypothesis (I think you&Gale do better in Wainwrights volume.
And I can't see that his argument from OSE's meets the teleological argument.