Millican and Fine Tuning
Here is the letter I wrote to Prof. Peter Millican after his debate with William Lane Craig, asking him for his thoughts on a Collins-ian fine tuning argument. Hopefully the beginning of an extended dialogue on it!
I hope you’re well – was good to see you again briefly at the Sheldonian the other week! Sorry I’ve taken a while to write, I’ve spent the last week in hospital for my leg, so have had lots to catch up on. There were a few things we talked about (Galileo, Genesis, etc.) but for now I’d like to hear your thoughts on the following kind of fine tuning argument, which follows Robin Collins’ work very closely (some of it being verbatim). I much prefer a probabilistic approach, which I think this kind of version exemplifies. I’ll try to keep it brief, though this will inevitably be at the expense of some cogency and completeness:
F = the laws and values of the constants of physics, and the initial conditions of any universe with the same laws as our universe, must be set in a seemingly very precise way for the universe to support life.
LPU = there exists a material spatiotemporal reality that can support embodied moral agents, not merely life of some sort.
NSU = there is only one universe, the existence of which is an unexplained, brute given, and that within that universe the laws and constants of physics do not significantly vary from one space-time region to another.
T = there exists an omnipotent, omniscient, everlasting or eternal, perfectly free creator of the universe whose existence does not depend on anything outside itself.
1) F obtains.
2) Given F, LPU is very, very epistemically unlikely under NSU, that is, P(LPU|NSU & k’) << 1, where k’ represents some appropriately chosen background information.
3) Given F, LPU is not very, very epistemically unlikely under T, that is, ¬[P(LPU|T & k’) << 1].
4) T was advocated prior to the discovery of F, and has independent motivation.
If these were true (and LPU obtains), it would seem to follow, given the restricted likelihood principle (i.e. the likelihood principle, but with ad hoc hypotheses excluded), that LPU strongly supports T over NSU. Arguably, this can be extended to the conclusion that LPU also supports T over ¬T, and hence provides some evidence for T – though this would not make any claim about whether T was overall more probable than not.
Now, (4) seems palpably true – theism was advocated long before F came to light, and has other sincere putative arguments in support, even if they also fail. Similarly, LPU clearly obtains. So the potential problems will be justifying the first three premises.
This can come in 3 forms – the laws of nature, the constants of physics and the initial conditions of the universe. I’ll save spelling out the details, and simply assert that it seems probable, given a material universe, that this universe would have to have very particular parameters to be capable of sustaining life. It seems that laws comparable to those of gravity, the strong nuclear force, the electromagnetic force, Bohr’s rule of quantization and the Pauli Exclusion Principle are necessary for the specific kind of materiality needed for embodied moral agents. Similarly, the life permitting range of many values of the constants of physics (e.g. the strength of gravity, and the cosmological constant) seems to be a small proportion of the epistemically illuminated comparison range (i.e. that range for which we can determine whether a particular value for the constant would be life permitting). And, finally, the extraordinarily low entropy at the beginning of the universe seems to be a very strict requirement for LPU. For brevity, I’ll leave it here – but do let me know if you see this premise as one of the more contentious points, and perhaps then I can go into greater detail.
Obviously a lot of work has to be done here in developing the notion of epistemic probability, defining background information adequately and coming up with some way of judging epistemic probability without always having clear quantitative measures. But it does not seem as though quantitative measures are strictly necessary, and that there is quite a large, yet rational, role for intuition in many instances. As Collins puts it, “[i]n science, many times epistemic probability is determined by an appeal to intuition … for example, those arising in conjunction with the Thesis of Common Ancestry, continental drift theory, and atomic theory. These probabilities clearly were not justified by an appeal to statistical improbability – for example, we have no statistics regarding the relative frequency of life on a planet having those features cited in favour of evolution either under the evolutionary hypothesis or under some nonevolutionary hypothesis. Indeed, these judgments of epistemic probability were never rigorously justified in any way. Rather, after (we hope) doing their best job of looking at the evidence, scientists and laypersons made judgments of what kind of world we should expect under each hypothesis, and then they simply trusted these judgments. This sort of trust in our judgments of epistemic probability is a pervasive and indispensable features of our intellectual life … Of course, the more widely shared these judgments are by those who are relevantly informed, the more seriously we take them. In this regard, it should be noted that, given the fine-tuning data, the judgment that LPU is surprising under naturalism is widely shared by intelligent, informed individuals, as evidenced by the various attempts to account for it, such as the multiverse hypothesis.”
Moreover, I think we can reasonably have some rough quantitative evidence. The restricted principle of indifference would be that “when we have no reason to prefer any one value [or range of values] of a variable p over another in some range R, we should assign equal epistemic probabilities to equal ranges of p that are in R, given that p constitutes a “natural variable””. This seems to me reasonable, and would allow us to come up with some crude quantitative measure for the constants of physics and, arguably, the initial conditions of the universe. When we use this principle, it does indeed turn out that the probability of getting a life permitting value for a constant is particularly low on NSU, which seems to justify this premise. The only problem here might be the normalizability problem as explained by McGrew, McGrew and Vestrup – that is, the difficulty of assigning these probabilities when the comparison range is infinite. This does not seem to me to be a problem, both because we can choose a non-arbitrary comparison range (for example, the epistemically illuminated region, or the ‘level’ of physics in which a particular law/constant makes sense – it is dubitable whether it makes sense for the strong nuclear force to be 10^1000 times its current strength, and in any case we cannot know to any reliable extent what would happen if it was increased that much, and so we can limit our comparison range to a maximum of 10^1000 times the current strength) and because I’m not entirely convinced that there can be no assigning of epistemic probability given an infinite range.
Premise 2, therefore, seems to me to be correct. The background information is most relevant in discussions of the anthropic principle, but if I have understood your position (only from your debate), then you don’t think this would have much force as an objection unless it was conjoined with the multiverse hypothesis. That is, I think you would agree that we can reasonably exclude the information “we exist” from the background information.
This is not a particularly strong claim – only that it is not as extremely improbable on T (relative to NSU) that LPU would obtain. All that needs to be done to justify it, I think, is to give some plausible reason why God might prefer a universe that contained embodied moral agents, ceteris paribus. This is easily done – embodied moral agents, plausibly, contribute to the overall moral and aesthetic value of reality, and they are a unique good which a good god (ex hypothesi) would plausibly want to bring about.
I note that your rejection of the fine tuning argument in your debate with Bill wasn’t immediately dismissive or brash, as it could well have been. It seemed to emphasise caution and scepticism of the argument rather than anything else, but you did proffer some potential points of contention. I’ll try to look at where these fit in and address them, respectively.
“No idea of the range of possible scenarios” – this seems to address the issue of appropriately chosen comparison ranges, and could be dealt with as follows: contrarily, we do have some idea of the range of possible scenarios – we have a natural limit to our varying the constants of physics beyond particular points, and it is questionable whether varying them beyond that would make sense. Indeed, the epistemically illuminated region and the “proper level of physics” provide two examples of possible non-arbitrary comparison limits. In any case, we do know that there is a comparison range which is at least very large and finite – it is epistemically possible (excluding our antecedent knowledge to the contrary), for example, that the force of gravity be 10 times its actual strength. The question then becomes whether the comparison range is very large and finite or infinite. But, unless one thinks that the aforementioned normalizability problem (for an infinite range) is substantive, it is difficult to see how this would diminish the force of the argument. If the comparison range is large and finite, then the epistemic probability of LPU on NSU is very small – if the comparison range is infinite, then the epistemic probability of LPU on NSU is smaller still, and so the argument would still work. What matters is whether or not there is a large comparison range which is not life-permitting, and this does seem to be the case.
Of course, your objection might be that there might be possible universes far beyond the epistemically illuminated region which are life-permitting. But the response to this could be two-fold: firstly, that the epistemically illuminated region is a sample of possible universes, and therefore (by the same rules of normal sampling in science) ought to be taken as at least indicative of the density of life permitting universes unless we have reason to believe that there is a bias – that they are more abundant outside of the EI region. But we have no reason to believe this, and we even have evidence against it, viz. that the universe in which we exist is likely to be in a group of universes which are more propitious for life. So, although it is clearly a very limited sample, I don’t think this objection holds much weight. But secondly, it is not clear why we could not include the information Q, that the universe’s constants fall into the EI region somewhere, in our background information k’. Unless Q is intrinsically biased towards theism, then its modal priority (or lack of posteriority) to LPU should allow us to include it in the background information. In that case, the argument would be that P(LPU|NSU & k’ & Q) << 1, and that P(LPU|T & k’ & Q) is not comparably low. This would entirely avoid the objection, and is not an evident abuse of the likelihood principle or epistemic probability.
I’ve realised that I’ve probably given you plenty to read through already, so will save my response to your other objections for the future – all I would note now is that this formulation of the design argument makes those objections difficult to place, and may make them inapplicable. For example, this version of the design argument doesn’t claim that fine tuning supports T over anti-God, but only that it supports T over NSU (though I could conceivably extent that to T over ¬T). Some of the objections seem to be responding to the unfortunate tradition within natural theology that tried to get every argument to prove every attribute of God. Taking a quasi-Bayesian approach seems to avoid these entirely, and seems to me a lot more reasonable.
Anyway, your thoughts will be much appreciated and mulled over, and I look forward to hearing which objections you think would be most pertinent to this formulation of the fine tuning argument.
Take care, and speak soon,