In this post I will be commenting on issues in the fine-tuning argument (FTA) from the problem of evil (PoE) as Robin Collins presents it in his chapter in the Blackwell Companion to Natural Theology.
I find the evidential PoE to be a persuasive argument against God, but I find it philosophically boring. By contrast, I think the fine-tuning argument (FTA) is philosophically interesting. Robin Collins is probably the premiere defender of the FTA, and for Collins the PoE is very relevant to the FTA; so we’ll have to interact with the PoE.
Why is the PoE relevant to the FTA? Ultimately, Collins argues that the probability of life-permitting constants (Lpcs) is more probable under theism (T) than the naturalism (N), thus supporting theism. The reason is that God qua good being wants to create embodied moral agents (EMAs); it’s his goodness along with his other omni-attributes that incline us to think that he’d create EMAs and thus the Lpcs. This is where the PoE comes in, because if the creation of EMAs don’t lead to an overall good, then it would be unlikely that God create Lpcs for the EMAs.Collins says:
Thus, in order for God to have a reason to adjust [the constants] so that [the universe] contains our type of embodied moral agents, there must be certain compensatory goods that could not be realized, or at least optimally realized, without our type of embodiment. This brings us directly to the problem of evil.
If we have an adequate theodicy, then we could plausibly argue that [we] would have positive grounds for thinking that God had more reason to create the universe so that EMA is true, since it would have good reason to think that the existence of such beings would add to the overall value of reality. … On the other hand, if we have no adequate theodicy, but only a good defense – that is, a good argument showing that we lack sufficient reasons to think that a world such as ours would result in more evil than good – then [the probability of Lpc would be indeterminate]. (p 255)
Collins thinks that with an adequate theodicy the Lpcs would be very favorable to T over N. And, surprisingly, even if we lack a theodicy and only have a good defense, the Lpcs would still favor T over N. By “good defense”, I suspect Collins means either the free will defense or skeptical theism. (I’m not sure it matters to the FTA which one it is.) Let’s look at the case where we only have a “good defense.” In that case we have:
- P(Lpc|T) = indeterminate
- P(Lpc|N) << 1 (close to zero).
Normally, according to the likelihood principle, if P(Lpc|T) > P(Lpc|N) then the observation Lpc supports T over N. In the above case where P(Lpc|T) is indeterminate, Collins still thinks it would support T over N. This doesn’t seem right to me. It seems to me that when you’re using the likelihood principle, as Collins does, you compare a probability with a probability and not a probability with something indeterminate.
Collins explains his motivation for this in footnote 40.
One might challenge this conclusion by claiming that … a positive, known probability exist .… This seems incorrect, as can be seen by considering cases of disconfirmation in science. For example, suppose some hypothesis h conjoined with suitable auxiliary hypotheses, A, predict e, but e is found not to obtain. Let ~E be the claim that the experimental results were ~e. Now, P(~E|h & A & k) << 1, yet P(~E|h & A & k) ≠ 0 because of the small likelihood of experimental error. Further, often P(~E|~(h & A) & k) will be unknown or indeterminate, since we do not know all the alternatives to h nor what they predict about e. Yet, typically we would take ~E to disconfirm h & A in this case because P(~E|h & A & k) << 1 and ~P(~E|~(h & A) & k) << 1.
To paraphrase, Collins says that in science we often disconfirm some hypothesis when it is highly unlikely that we don’t see e given that hypothesis and it turns out that we don’t see e. I think Collins is reasoning that if P(~E|h & A & k) << 1, then ~(h & A) is confirmed even when it leads to the observation being indeterminate. Presumably, ~(h & A) is an infinite disjunction of mutually exclusive hypotheses, or what’s called a catch-all hypothesis. Even if this is right (which I doubt), there is a disanalogy because h and ~h are dichotomies, while, oddly, N and T are not. That’s because, as Collins runs the argument, T is a good God; so T doesn’t include an evil God, or a non-omnipotent God, among other things. If anything it seems Collins should say ~N is confirmed; and that doesn’t necessarily mean that T is confirmed, since T is a small subset of ~N.
I don’t think Collins’s reasoning here is consistent with the likelihood principle (keep in mind he runs the core FTA based on the likelihood principle), as the principle seems to rule out comparing something with indefinite probabilities. Collins’s reasoning would make more sense if indeterminate meant .50, but it doesn’t. Suppose there are marbles in a vase where each marble has a number from 1 to 10 written on it. What is the probability that we pick a marble with a number greater than 1 if it is indeterminate how the numbers are assigned to each marble. The mistake would be to think indeterminate means that we can apply a principle of indifference so that each number has a 1 in 10 chance. If that were so, we’d have a high probability that the marble picked would be greater than 1. But that’s not what indeterminate means; indeterminate means that we’re not in a position to know if it’s randomly generated or otherwise. We’re simply “in the dark” to borrow the skeptical theist’s phrase. So I can’t see how you can compare a “positive, known probablity” with something indeterminate, just as you can’t say if the marble picked will probably be greater than 1. If you think that it will be greater than 1, then it’s not indeterminate after all.
I’m not sure science should be rationally reconstructed as Collins does in footnote 40. It seems Collins is saying that we can compare and confirm hypotheses that have indeterminate observations. I don’t think we test hypotheses against their catch-alls, as is implied by the footnote. If Elliott Sober is right, we test hypotheses against non-catch-all hypotheses, like the general theory of relativity over Newton’s theories; neither theory being tested are catch-alls.
It seems one can’t be a skeptical theist and support the FTA; you do need to say something about what God would do with some probability.