Great post. I'm more partial to Sobel's theory of intrinsic probability as outlined in Logic and Theism, but I think even on the theory of probability you outline here, Naturalism comes out ahead.
It's clear to me that Naturalism comes out on modesty. As Lorkowski points out in his 2013 article:
"The naturalistic position, in its simplest form, makes two claims, only one of which is unique: the natural world exists, and nothing outside of the natural world affects the natural world. The theistic position, in its simplest form, makes several claims: the shared claim that the natural world exists, and several unique claims, such as the claims that at least one supernatural entity exists, it has affected the natural world, this entity is personal, omnipotent, omniscient, morally perfect, and so forth. It, therefore, appears that the theistic position (or the relevant deistic position) must have a lower prior probability than the naturalist position due to the sheer number of assertions. The theist makes more unique claims and therefore has more ways to be incorrect. Setting aside the shared claim that the natural world exists, the only way that the naturalistic position can be incorrect is if there is at least one supernatural entity that has affected the natural world. The theistic position, on the other hand, can be incorrect in a numerous ways. Therefore, the prior probability of the theistic position is much lower than that of the naturalistic position." (Atheism, pgs. 524-525)
On Brute limitations you say Naturalism is disadvantaged in light of the fact that we can ask: "Why are these particular laws and physical structures instantiated, instead of all the other conceivable laws and physical structures which there might have been?"
But notice, this exact question arises for Theism as well. As Graham Oppy notes in his review of John Foster's "The Divine Lawmaker"
"....there are many different universes that God might have made: we are not to suppose that the existence of our universe is necessary. Furthermore, since God has libertarian freedom, there are possible worlds in which God makes those other universes. When we consider a particular regularity in our universe, the existence of that regularity is explained in terms of God's desire (or intention, or whatever) to make a universe in which that regularity is instantiated. Moreover, when we compare our actual world with a world in which God makes a universe in which some other regularity obtains, ex hypothesi, there is no explanation in either world of why God has the one set of desires (or intentions, or whatever) rather than the other. So, it seems, the naturalist is being asked to trade in (putatively) unexplained regularities in the universe for unexplainable desires (or intentions, or whatever) in God. I do not think that naturalists should accept this deal: if you really think that there must be a satisfying explanation for the holding of the regularities-one that does avoid unnecessary complexities and that minimizes residual sources of puzzlement-then you have very good reason to deny that Foster has found it. (pg. 115)
It seems that whatever considerations you raise about Naturalism in terms of brute limitations (i.e. questions about particular laws and physical structures), the same considerations can be raised in the context of God's choice of instantiating a world with those particular structures. J.L. Mackie put it best: "The particularity has not been removed, but only shelved; we should have to postulate particularities in God, to explain his choice of the particular universe he decided to create." (p. 100). And given that Naturalism comes ahead on modesty, a wash on brute limitations is ultimately going to disadvantage Theism.
On coherence, I think Naturalism has an edge as well. Almost all the properties of God (omnipotence, omniscience, necessary existence, etc) have been challenged in terms of their coherency by a wide variety of Theists and Atheists. Trying to save God's coherency, by invoking Divine Simplicity ironically makes things worse, because just ends up adding more contentious properties to your Theistic ontology (usually it's alleged that timelessness and immutability follow from ADS).
- I agree that naturalism is more modest than theism. No argument there.
- I don't agree with Oppy's comments on brute limitations. I think that when we're assessing the prior probability of a theory, we should consider only that theory's fundamental posits (i.e. the entities which the theory posits but does not explain). For discussion (and justification) of this approach, see e.g. Climenhaga, "The structure of epistemic probabilities" (specifically section 3.6.2 on simplicity). If I'm right that this is the correct way to assess priors, then Oppy's point simply doesn't go through, since our universe isn't a fundamental posit for the theist (after all, it's explained by God, so it can't be explanatorily fundamental). The right way to compare theism and naturalism is to compare the theist's fundamental posit (i.e. God) to the naturalist's fundamental posit (i.e. whatever part or parts of the universe a particular naturalist takes to be fundamental). That's why our ability to ask "why *this* universe?" is a problem for the naturalist, but not for the theist: only the naturalist takes the universe to be a fundamental posit. So the naturalist has brute limitations *in their fundamental posit*, while the theist does not.
- I think it's also worth noting that Oppy has a non-Bayesian epistemology, so he isn't thinking in terms of priors anyway. He's thinking exclusively about simplicity (which he understands to be something like "how many entities and concepts does this theory involve") and explanatory power (which for him basically means "how little bruteness does this theory require"). This causes him to give some rather strange responses to some theistic arguments; for example, he thinks he can rebut the fine-tuning argument by simply saying that the fine-tuning is necessary (since on his view this counts as explaining it, and it doesn't require invoking any extra entities). But of course, a Bayesian would say that this is worthless, since all it does is shift the improbability of fine-tuning back into Oppy's theory.
- I don't agree with Mackie that we have to "postulate peculiarities in God, to explain his choice of the particular universe he decided to create." I think the theist should simply reason as follows: God is (ex hypothesi) a being with all perfections; perfect goodness is a perfection; so, God is perfectly good. A perfectly good being always does the best action, or (if there is no unique best action) an action of the best kind (this is Swinburne's account of perfect goodness). Creation is a unique best action (since created things are valuable, and it costs God nothing to produce them), so God would create. But there is no unique best creation (i.e. there is no "best possible world"), so a perfectly good God would simply pick one of the infinite number of possible good universes to create. That's where explanation comes to an end: there probably won't be a satisfying answer to the question of why God made *this* particular good creation rather than any other; since there is no best creation, he just had to pick one, and so he did. You might think this is a problem, but I'd disagree: after all, the naturalist also doesn't have any explanation for why *this* particular universe exists, so that'll simply be a wash between the two views. The real battle will involve appeal to the *features* of the universe (e.g. its complexity, its laws, its constants, and so on), and these (or so I claim) theism predicts better than naturalism.
- Yeah, if one thinks that theism is incoherent, then obviously it's game over for the theist. But I don't think theism is incoherent, and if it's coherent, then we can assess it in terms of the criteria given above :)
I'm more partial to Sobel's way of cashing out interesting probabilities (see his exchange with Swinburne in his 2006 article "To My Critics with Appreciation; "On Subjective and Objective Probabilities,." I haven't given as deep a reading to Climenhaga's theory as I'd like, but I'd offer a few comments:
1. There are many models of Naturalism where nothing is really fundamental (see Ross P. Cameron's Chains of Being (OUP 2022), so I think that Cilmenhaga's objection doesn't apply to those models.
2. Even taking into account Climenhaga's model and assuming Oppy's methodology of comparing "like for like," I still think Oppy's point goes through when we correctly asses the Theistic hypothesis. The explanatory power of Theism is not just derived from God as a fundamental posit but also his creative act. So the model is not just God, but God + his intentions/creative act, and when we construe things this way, the Theist also ends up having bruteness in their model in light of God's intentions/creative act as Anthony O'Hear notes in his 1984 book "Experience, Explanation, and Faith":
"...some of those theists, such as Hartshorne, who see God as having to create something, stress that God has created only one of the many possible universes open to him to create. So it seems that if the cosmological argument is setting out to explain the existence and nature of the world, its terminus of explanation cannot be a logically necessary truth. It must be something, such as the Big Bang, or God's intentions or existence, which is regarded as a brute fact in the sense that it is not logically necessary and presumably also such that there is no sufficient reason for it." (pg. 109)
Therefore, if Sobel, Rowe, and others are right that any contingency in the universe implies the existence of brute contingency, then it seems that every metaphysical hypothesis has brute contingency. It's just a matter of where we shelve it. Since you've conceded that the Naturalistic view is more modest than the Theistic one, it seems that Oppy's argument goes through.
Regarding Oppy's methodology, his general approach takes theory to come before arguments. Instead of engaging in Theistic arguments individually (which he has done over his career), we should take the best versions of Theism and Naturalism and compare them on respective theoretical virtues. I think there are plenty of good responses to fine-tuning (see "Fine-Tuning and Confirmation Theory: Some Complications" by David Manley), but I don't see what's wrong with Oppy's response. Either fine-tuning is going to be a contingent feature or a necessary feature of reality, and this set of options is going to be true on Naturalism and Theism. In either picture, we will have similar explanations of the data, with Naturalism being a more modest view and, therefore, coming out ahead.
Of course, a Bayesian FTA proponent like Rota, Swinburne, Collins, Barnes, etc., may disagree, but as Oppy shows in Arguing About Gods (see pgs. 200-28), it's generally quite difficult to justify both ends of the bayesian formulation; showing that a life-permitting universe is quite low on Naturalism and that the same datum is quite high on Theism.
We can bracket the contentions I'd take with your analysis (see Jeff Speak's "The Greatest Possible Being" for good critiques of a being with all perfections, and also see Sobel, Rowe, and Fales on God's creating the best), and I still believe Mackie's point has teeth. As you indicate, God's creative act based on the considerations you note only establishes that there is a smaller infinite range of possible universes from which God had to choose. But notice that if we evaluate the full modal space between Naturalism and Theism, it seems clear that there are infinite types of worlds that God could create that would be incompatible with Naturalism (Manson's 2019 paper "How Not to be Generous to Fine-Tuning Skeptics pgs. 11-14) has a good discussion on this. Therefore, the fact that we live in a world compatible with Naturalism would be evidence for Naturalism vis-a-viz Theism, since the set of worlds that are possible on Naturalism will be limited in virtue of what Naturalism entails.
One can also adopt an Aristotelian Approach to modality as Oppy and Pruss do and say that the necessary initial state will limit the possibilities of the universe we'd expect from Naturalism. So just as the Theist can try to limit the possible worlds that God can create based on considerations around God's creative act, the Naturalist also has tools in their worldview that limit the set of possible worlds we'd expect on Naturalism, leading to equal explanatory power which brings us back to modesty/simplicity as tie tie-breaker.
I agree with you that any comprehensive assessment of Naturalism and Theism is going to involve more than just the datum of the universe's existence as a contingency entity; however, there are plenty of apologists who take this issue to be a decisive consideration against Atheism (Bishop Barron had plenty of mileage in the early days of ministry against the New Atheists continuously emphasizing that Atheists had no way to respond to the Contingency argument). Hence, the concession of this issue not fully deciding the debate from more sophisticated Theists is a welcome one.
To be clear, I don't think Theism is incoherent either; I'm just saying that when it comes to evaluating coherence, Theism is going to have a lot more challenges to its coherence compared to Naturalism. I also think that when Theists revise their view of God/the Divine Attributes in light of these coherence arguments, it will lower the epistemic probability of Theism since it forces the Theist to take a more specific way that God could be.
- I don't agree that free actions have to be brute. I think there can be indeterministic explanation, so even libertarian free choice can be explained. So I don't agree that God's free choice would be brute. (Note that this is independent of the probabilistic point.)
- I don't think the relevant thing is necessarily whether a theory posits something as *ontologically* fundamental; when I say "fundamental posits," I just mean the things posited by a theory which the theory itself does not explain. Example: Suppose you want to know where the bird-waste on your window ledge came from, so you posit a bird to explain it. That bird is your theory's fundamental posit, even though he probably won't be ontologically fundamental!
- To avoid getting into a lenghty exchange, I will have to defer to Sebastian Montesinos's post "Why I'm an Atheist" that engages with considersations regarding God's LFW act in the context of non-entailing explanations. I think attempts to remove bruteness from Theism just aren't going to work.
- Sure, but my point is that even in that consideration, the respective hypotheses being compared are not just Naturalism and Theism, but Naturalism and God + God's intentions/desires, and whatever bruteness we find in the Naturalistic picture of reality is just going to be shelved back into Theism with regard to why this universe was picked instead of something else.
Hello again! A few brief points (perhaps I'll say a bit more later):
- I don't think the Manson critique of the FTA works; see Donahue's new paper "Divine psychology and fine-tuning" (Religious Studies, forthcoming).
- On Oppy's critique of fine-tuning: suppose we found that every atom had the words "made by God" inscribed on it in tiny letters. Nathan the Naturalist says "this is no evidence for God; I claim it's *necessary* for atoms to be like this. Therefore I've explained it, and without having to posit God." We'd think that response is awful: all Nathan has done is take the improbablity of the god particles, and put it into his theory. But that's what Oppy does with fine-tuning. (I'm not the first person to make this critique; Mike Huemer makes it in his debate with Oppy.)
- I agree that <this exact universe exists> will have an extremely small probability conditional on theism, since there's infinitely many universes God could have made. But it will *also* have an extremely small probability conditional on naturalism, since there's an infinite number of universes consistent with naturalism. So we should appeal to a broad feature of reality (such as <there is a complex universe>), and evaluate *that*. I claim that theism then predicts those features better than naturalism does.
Here's an analogy. Suppose you found a rock sculpture of a well-dressed man in your town, and wanted to compare two theories: H1 says that there is an artist in town while H2 says that there is not. Now, the probability of *this exact sculpture* existing is vanishingly small on both theories: even on H1, the artist could have made anything, not just this sculpture. But the probability of *any* sculpture existing is massively higher on H1, so it's clearly evidence for H1. The same is true of God and the universe.
I've skimmed over Donahue's paper and don't find his response to the consideration raised by Manson and Sinhababu (that God has infinite ways to instantiate life without needing to engage in fine-tuning) that convincing. He tries to get around this with an appeal to considerations about background knowledge, but as I will outline later, I don't think we have the relevant background knowledge for the inference required by the FTA to go through. The response also runs into some complications, which Manley points out in the manuscript I referenced earlier. Imagine if instead of discovering the laws to be fragile (i.e., life permitting under certain constants), we discover what Manley calls Resilience: The relevant parameters are set up so that the universe would produce life even for pretty significant changes to the values of a given parameter.
Now, this seems like it would naturally be evidence for Theism given God's preference for life that has objective value as outlined by Donahue's paper however, it follows through Bayesian reasoning that for any e, P(h|e) > P(h) iff P(h|~e) < P(h) and so if the Theist wants to say that Fragility counts in favor of Theism, then they have to concede that Resilience count *against* Theism, which seems completely contradictory if God is supposed to care about life.
For Oppy's critique, first, as Oppy responded in the video, if you take a view of an Aristotelian modality, then the thought experiment Huemer describes would be metaphysically impossible given the constraints imposed by the initial state. Of course, the FTA proponent is interested in epistemic probabilities, but I think the analogy goes awry in this realm. Interestingly, in Hume's Dialogues, Philo is faced with a similar counter-example from Cleanthes about "voices from the clouds," here it is presented in "Hume's Critique of Religion" by Dan O'Brien and Alan Bailey:
"Cleanthes' first thought experiment invites Philo to consider what conclusion he would draw 'if an articulate voice were heard in the clouds, much louder and more melodious than any which human art could ever reach' (ibid.). If it is supposed that this voice speaks at the same instant to everyone in the world in their own language and provides profound and morally improving instruction, then Cleanthes purports to be confident that humankind's lack of acquaintance with previous examples of heavenly voices and the circumstances in which they arise would not prevent Philo from inferring that this voice is the product of intelligent agency." (pg. 113)
But as O'Brien and Bailey note, the reasons for confirming the inference in this example have little to do with considerations about design:
"The truly problematic aspect of Cleanthes' example would appear to be its appeal to a phenomenon that is exceptionally well-correlated in our experience with intelligent agency. We have, it might be suggested, no experience whatsoever of articulate, meaningful speech emerging from any source that is not itself intelligent or at least constructed by intelligent agents. Thus the link between meaningful sounds and intelligent sources is a tight and well-confirmed one." (pg. 114).
I think a similar consideration can be made in the context of the atom example. The inference we make in this case doesn't have anything to do with fine-tuning; they would still be evidence for Theism even if we found out the laws that produced them were not fine-tuned. Rather, the reason why the inference towards Theism is secure has to do with our considerations regarding our background knowledge (e.g., our knowledge about written works, cultural context, etc). The improbability of probability has little to do with the inference. The problem for the FTA proponent is that when it comes to fine-tuning, we don't have the required background knowledge to make the inference. To quote from Dan Baras' "Calling for Explanation":
"While it is fairly easy to understand what our expectations should be regarding the explanation of [ordinary cases] and why our expectations should be such, it is far from clear that these considerations translate to the fine-tuning of our universe. ... background information based on previous experience, which heavily informed our understanding of these three examples, is not available with regard to the fundamental physical properties of our universe. ... In a way, we have come full circle back to David Hume's famous objections to teleological arguments. Hume discussed a version of such arguments that is based on an analogy between human artifacts (such as watches and houses) and the universe as a whole, and he argued that the analogy is weak. He says that because the universe as a whole is so different from human artifacts, the knowledge that we have about human artifacts should not be projected onto the universe as a whole." (pg. 168-169) (see also Keith Parsons and Robin Le Poidevin for further discussion of this point).
So, if we don't have the background knowledge to assign the relevant epistemic probabilities for Theism, we're thus back to Oppy's point about the constants being either contingent or necessary for Theism. On both options, Naturalism will be a more economical position that can explain the relevant data.
You say: "But it will *also* have an extremely small probability conditional on naturalism, since there's an infinite number of universes consistent with naturalism."
But as I pointed out earlier, the amount of possible universes on Theism is going to be *much* greater on Theism than on Naturalism because for Theism, the range of possible universes includes not only the majority of those compatible with Naturalism but also infinitely many universes not possible on Naturalism. The range of possible universes is much greater on Theism, which means the ratio between the actualized universe vs. possible universes will be much higher than Naturalism.
This is why I don't think the rock sculpture example is analogous to the situation since the ratio between H1 and the datum of the rock sculpture is going to be quite small in the way that the ratio between Theism and the possible worlds that can be actualized on Theism isn't. Here is a better example in my view:
Imagine you've been told that someone ran a 19-minute mile. Now, we have two hypotheses:
H1: An out-of-shape person who had never run before ran this mile
H2: Someone who has set the world's fastest mile time ran this mile.
I think it's fair to say that H1 is the best explanation of the data. Now note, it's completely compatible with H2 that they they also ran the 19-minute mile. However, we know that H2 can run a whole range of more faster times, which just aren't possible for H1. H1 has a much shorter range of potential mile times than H2 does, and the 19-minute mile falls in that range, meaning the ratio between the actual time and possible times for H1 is much smaller than the ratio between the actual time and possible times for H2. I think we're in a similar position vis-a-vis Theism and Naturalism.
Hello once more! A few final points (unless you'd like to discuss things further):
- Yes, Oppy's response is that the scenario Huemer proposes would be metaphysically impossible on his view. But that doesn't actually answer the objection: Huemer's point is that Oppy's epistemology would allow him to dismiss the god-particles, and since someone in that scenario *shouldn't* dismiss the god-particles, there must be a problem with Oppy's epistemology. Oppy simply saying "well, we won't find any god-particles" completely misses the point of the objection.
- Thanks for bringing Manley's manuscript to my attention! I've only skimmed it so far, but if I'd be happy to discuss it with you at a later date.
- I don't actually think that explanation per se is the relevant thing. My argument isn't "features of the universe cry out for explanation, only God can explain them, therefore God"; I'm appealing to probabilistic claims, to the effect that P(Universe and its features/Theism) >> P(Universe and its features/~Theism). So even if Oppy is right that he's "explained" fine-tuning (or whatever piece of data is under discussion), I don't think he's actually succeeded in undermining the argument.
- I don't think we need to employ any sort of argument from analogy, such as the one to which Hume was objecting. I think we can say enough about what a perfect being would be like that we can make various probabilistic claims, such as that a perfect being would value the existence of intelligent embodied life. And that's all we need to run the argument; any sort of analogy between human artifacts and the universe is totally otiose.
- I take the point about the range of universes being much greater on theism. However, I don't think your runner analogy accurately captures out situation; after all, it's not just *compatible* with your H1 that they ran a 19 minute mile; it's positively *probable* that an out-of-shape person might achieve such a slow time. By contrast, there are (I claim) many features of our universe which, though not strictly incompatible with naturalism, are nonetheless extremely *improbable* on naturalism. Similarly, the statue of a man is not actually inconsistent with the random chance hypothesis; it's in-principle possible that you could get a statue-like rock formation by random chance. But it's still enormously improbable.
Feel free to send me a message, and I'd be happy to discuss some of the other issues that were touched upon. I'm always glad to correspond with theists who are knowledgeable about the literature on the analytic philosophy of religion.
- Re: Oppy; the point I'm making is that there are a few relevant differences between the god-particles scenario and fine-tuning. In the former, we have the relevant background knowledge to be able to make a well-correlated inference from marks of design to a potential designer (which incidentally has nothing to do with fine-tuning). In the latter, as Oppy (and others) have argued, we don't have the requisite background knowledge/epistemic probabilities at hand to be able to justify the inference, so both the Naturalist and Theist are forced to look at the modal scope of how our respective worldview can account for the datum. I think Oppy was gesturing to this point when he mentioned Cleanthes's thought experiments, which I mentioned before.
- Re Manley: along with Oppy's chapter in "Arguing About Gods," Manley's manuscript is perhaps the most decisive critique of fine-tuning arguments I've read, and it's generally my go-to critique on the subject. Manley also has another excellent paper, "God and the Bayesian Conception of Evidence," which is a good overview of how to think about evidence in the context of the philosophy of religion in Bayesian terms:
- Re P(U|T) >> (U|~T); on the probabilistic claims, my point (following Oppy, Manson, Manley, and others) is that it's generally hard to generate the relevant epistemic probabilities on Naturalism and Theism given the datum of fine-tuning. ("Naturalism, Fine-Tuning, and Flies" by Aron Lucas is also an excellent critique to look at). Therefore, as noted earlier, both the Naturalist and Theist are left with the same options; either fine-tuning is necessary or contingent, and it seems that the Naturalist has the same range of options as Theist does and posits less. The epistemic situation we are in when it comes to fine-tuning will be a lot different than in the case of the god-particles.
- Re Analogy: I recognize that the contemporary Bayesian forms of the FTA are different from the classical design arguments from Analogy; however, my reason for citing the Dialogues was to point out that Cleathes offers a similar thought experiment as Huemer does in response to the general skepticism that Hume offers towards design arguments, and I intended to show how one can respond to the respective thought experiments offered in a way that still allows them to be skeptical of the inferences that the FTA makes. I think the excerpt from Baras outlines the relevant differences between inferring design in ordinary cases and what the FTA tries to argue.
- Re: A perfect being's expectations: I agree that the hypothesis of a perfect being has enough conceptual content to be able to allow us to make some predictions on what we would expect regarding Theism; however, I think in the case of the FTA, there are just so many possibilities that far better than what we currently have. For example, I see no read why intelligent life has to be embodied; I think intelligent life could have been a lot more impressive than it actually is (see the section "Humans" from my colleague's case for Atheism below: https://naturalismnext.blogspot.com/2022/12/why-im-atheist_6.html), as well as a host of other considerations. Now, of course, the Theist can build in these further stipulations into their hypothesis, but I think doing so leads to a lot of problems (see the post below for a good overview of "stalking-horse" objections to the FTA: https://fric.substack.com/p/cosmological-fine-tuning?utm_source=profile&utm_medium=reader2)
- Statue Analogy: Not surprisingly, I disagree with the categorization of the relevant datum as being improbable on Naturalism, but I recognize that arguing as such would be beyond the scope of the post. My point was just to outline why I consider Theism's modal scope being much larger than Naturalism to be a problem for Theism as a theory. Such a problem also impacts the overall intrinsic probability of Theism as a whole, and that's all my comments were intending to argue for: that given the considerations you've offered, it seems Naturalism has a higher intristic probability than Theism. Of course, I recognize that this is something that could be overturned by the relevant posterior considerations such as what you've offered.
I echo your sentiments on the discussion and look forward to more in the future.
Thanks for the post! I personally do not see the need for an appeal to your third prong when assessing intrinsic probability. There are an infinite number of velocities at which the particles in worlds m and n could have been moving and each theory asserts a particular velocity and, in that sense, they are both equally modest. moreover, neither theory contains within itself more or less inductive support relations between its parts than the other so they also appear roughly equally coherent.
So here I don’t see the problem with saying they are equally probable intrinsically. Why should infinite velocity be understood as being more in need of explanation than any other particular speed? The idea seems to be that you are understanding infinite velocity as being a hard limit in the upward direction such that there is no possibility of an increase in velocity. But it’s not clear why that should matter. The difference here is only between where on the velocity continuum or range the particle sits. Theory n posits a particle with a particular velocity somewhere between the highest and lowest while theory m posits a particle with the highest. I’m having difficulty understanding what you are considering more brute, unexplained, or arbitrarily limiting about n relative to m.
Great post. I'm more partial to Sobel's theory of intrinsic probability as outlined in Logic and Theism, but I think even on the theory of probability you outline here, Naturalism comes out ahead.
It's clear to me that Naturalism comes out on modesty. As Lorkowski points out in his 2013 article:
"The naturalistic position, in its simplest form, makes two claims, only one of which is unique: the natural world exists, and nothing outside of the natural world affects the natural world. The theistic position, in its simplest form, makes several claims: the shared claim that the natural world exists, and several unique claims, such as the claims that at least one supernatural entity exists, it has affected the natural world, this entity is personal, omnipotent, omniscient, morally perfect, and so forth. It, therefore, appears that the theistic position (or the relevant deistic position) must have a lower prior probability than the naturalist position due to the sheer number of assertions. The theist makes more unique claims and therefore has more ways to be incorrect. Setting aside the shared claim that the natural world exists, the only way that the naturalistic position can be incorrect is if there is at least one supernatural entity that has affected the natural world. The theistic position, on the other hand, can be incorrect in a numerous ways. Therefore, the prior probability of the theistic position is much lower than that of the naturalistic position." (Atheism, pgs. 524-525)
On Brute limitations you say Naturalism is disadvantaged in light of the fact that we can ask: "Why are these particular laws and physical structures instantiated, instead of all the other conceivable laws and physical structures which there might have been?"
But notice, this exact question arises for Theism as well. As Graham Oppy notes in his review of John Foster's "The Divine Lawmaker"
"....there are many different universes that God might have made: we are not to suppose that the existence of our universe is necessary. Furthermore, since God has libertarian freedom, there are possible worlds in which God makes those other universes. When we consider a particular regularity in our universe, the existence of that regularity is explained in terms of God's desire (or intention, or whatever) to make a universe in which that regularity is instantiated. Moreover, when we compare our actual world with a world in which God makes a universe in which some other regularity obtains, ex hypothesi, there is no explanation in either world of why God has the one set of desires (or intentions, or whatever) rather than the other. So, it seems, the naturalist is being asked to trade in (putatively) unexplained regularities in the universe for unexplainable desires (or intentions, or whatever) in God. I do not think that naturalists should accept this deal: if you really think that there must be a satisfying explanation for the holding of the regularities-one that does avoid unnecessary complexities and that minimizes residual sources of puzzlement-then you have very good reason to deny that Foster has found it. (pg. 115)
It seems that whatever considerations you raise about Naturalism in terms of brute limitations (i.e. questions about particular laws and physical structures), the same considerations can be raised in the context of God's choice of instantiating a world with those particular structures. J.L. Mackie put it best: "The particularity has not been removed, but only shelved; we should have to postulate particularities in God, to explain his choice of the particular universe he decided to create." (p. 100). And given that Naturalism comes ahead on modesty, a wash on brute limitations is ultimately going to disadvantage Theism.
On coherence, I think Naturalism has an edge as well. Almost all the properties of God (omnipotence, omniscience, necessary existence, etc) have been challenged in terms of their coherency by a wide variety of Theists and Atheists. Trying to save God's coherency, by invoking Divine Simplicity ironically makes things worse, because just ends up adding more contentious properties to your Theistic ontology (usually it's alleged that timelessness and immutability follow from ADS).
Hello there!
- I agree that naturalism is more modest than theism. No argument there.
- I don't agree with Oppy's comments on brute limitations. I think that when we're assessing the prior probability of a theory, we should consider only that theory's fundamental posits (i.e. the entities which the theory posits but does not explain). For discussion (and justification) of this approach, see e.g. Climenhaga, "The structure of epistemic probabilities" (specifically section 3.6.2 on simplicity). If I'm right that this is the correct way to assess priors, then Oppy's point simply doesn't go through, since our universe isn't a fundamental posit for the theist (after all, it's explained by God, so it can't be explanatorily fundamental). The right way to compare theism and naturalism is to compare the theist's fundamental posit (i.e. God) to the naturalist's fundamental posit (i.e. whatever part or parts of the universe a particular naturalist takes to be fundamental). That's why our ability to ask "why *this* universe?" is a problem for the naturalist, but not for the theist: only the naturalist takes the universe to be a fundamental posit. So the naturalist has brute limitations *in their fundamental posit*, while the theist does not.
- I think it's also worth noting that Oppy has a non-Bayesian epistemology, so he isn't thinking in terms of priors anyway. He's thinking exclusively about simplicity (which he understands to be something like "how many entities and concepts does this theory involve") and explanatory power (which for him basically means "how little bruteness does this theory require"). This causes him to give some rather strange responses to some theistic arguments; for example, he thinks he can rebut the fine-tuning argument by simply saying that the fine-tuning is necessary (since on his view this counts as explaining it, and it doesn't require invoking any extra entities). But of course, a Bayesian would say that this is worthless, since all it does is shift the improbability of fine-tuning back into Oppy's theory.
- I don't agree with Mackie that we have to "postulate peculiarities in God, to explain his choice of the particular universe he decided to create." I think the theist should simply reason as follows: God is (ex hypothesi) a being with all perfections; perfect goodness is a perfection; so, God is perfectly good. A perfectly good being always does the best action, or (if there is no unique best action) an action of the best kind (this is Swinburne's account of perfect goodness). Creation is a unique best action (since created things are valuable, and it costs God nothing to produce them), so God would create. But there is no unique best creation (i.e. there is no "best possible world"), so a perfectly good God would simply pick one of the infinite number of possible good universes to create. That's where explanation comes to an end: there probably won't be a satisfying answer to the question of why God made *this* particular good creation rather than any other; since there is no best creation, he just had to pick one, and so he did. You might think this is a problem, but I'd disagree: after all, the naturalist also doesn't have any explanation for why *this* particular universe exists, so that'll simply be a wash between the two views. The real battle will involve appeal to the *features* of the universe (e.g. its complexity, its laws, its constants, and so on), and these (or so I claim) theism predicts better than naturalism.
- Yeah, if one thinks that theism is incoherent, then obviously it's game over for the theist. But I don't think theism is incoherent, and if it's coherent, then we can assess it in terms of the criteria given above :)
Thanks again for your comment!
Thanks for the response:
I'm more partial to Sobel's way of cashing out interesting probabilities (see his exchange with Swinburne in his 2006 article "To My Critics with Appreciation; "On Subjective and Objective Probabilities,." I haven't given as deep a reading to Climenhaga's theory as I'd like, but I'd offer a few comments:
1. There are many models of Naturalism where nothing is really fundamental (see Ross P. Cameron's Chains of Being (OUP 2022), so I think that Cilmenhaga's objection doesn't apply to those models.
2. Even taking into account Climenhaga's model and assuming Oppy's methodology of comparing "like for like," I still think Oppy's point goes through when we correctly asses the Theistic hypothesis. The explanatory power of Theism is not just derived from God as a fundamental posit but also his creative act. So the model is not just God, but God + his intentions/creative act, and when we construe things this way, the Theist also ends up having bruteness in their model in light of God's intentions/creative act as Anthony O'Hear notes in his 1984 book "Experience, Explanation, and Faith":
"...some of those theists, such as Hartshorne, who see God as having to create something, stress that God has created only one of the many possible universes open to him to create. So it seems that if the cosmological argument is setting out to explain the existence and nature of the world, its terminus of explanation cannot be a logically necessary truth. It must be something, such as the Big Bang, or God's intentions or existence, which is regarded as a brute fact in the sense that it is not logically necessary and presumably also such that there is no sufficient reason for it." (pg. 109)
Therefore, if Sobel, Rowe, and others are right that any contingency in the universe implies the existence of brute contingency, then it seems that every metaphysical hypothesis has brute contingency. It's just a matter of where we shelve it. Since you've conceded that the Naturalistic view is more modest than the Theistic one, it seems that Oppy's argument goes through.
Regarding Oppy's methodology, his general approach takes theory to come before arguments. Instead of engaging in Theistic arguments individually (which he has done over his career), we should take the best versions of Theism and Naturalism and compare them on respective theoretical virtues. I think there are plenty of good responses to fine-tuning (see "Fine-Tuning and Confirmation Theory: Some Complications" by David Manley), but I don't see what's wrong with Oppy's response. Either fine-tuning is going to be a contingent feature or a necessary feature of reality, and this set of options is going to be true on Naturalism and Theism. In either picture, we will have similar explanations of the data, with Naturalism being a more modest view and, therefore, coming out ahead.
Of course, a Bayesian FTA proponent like Rota, Swinburne, Collins, Barnes, etc., may disagree, but as Oppy shows in Arguing About Gods (see pgs. 200-28), it's generally quite difficult to justify both ends of the bayesian formulation; showing that a life-permitting universe is quite low on Naturalism and that the same datum is quite high on Theism.
We can bracket the contentions I'd take with your analysis (see Jeff Speak's "The Greatest Possible Being" for good critiques of a being with all perfections, and also see Sobel, Rowe, and Fales on God's creating the best), and I still believe Mackie's point has teeth. As you indicate, God's creative act based on the considerations you note only establishes that there is a smaller infinite range of possible universes from which God had to choose. But notice that if we evaluate the full modal space between Naturalism and Theism, it seems clear that there are infinite types of worlds that God could create that would be incompatible with Naturalism (Manson's 2019 paper "How Not to be Generous to Fine-Tuning Skeptics pgs. 11-14) has a good discussion on this. Therefore, the fact that we live in a world compatible with Naturalism would be evidence for Naturalism vis-a-viz Theism, since the set of worlds that are possible on Naturalism will be limited in virtue of what Naturalism entails.
One can also adopt an Aristotelian Approach to modality as Oppy and Pruss do and say that the necessary initial state will limit the possibilities of the universe we'd expect from Naturalism. So just as the Theist can try to limit the possible worlds that God can create based on considerations around God's creative act, the Naturalist also has tools in their worldview that limit the set of possible worlds we'd expect on Naturalism, leading to equal explanatory power which brings us back to modesty/simplicity as tie tie-breaker.
I agree with you that any comprehensive assessment of Naturalism and Theism is going to involve more than just the datum of the universe's existence as a contingency entity; however, there are plenty of apologists who take this issue to be a decisive consideration against Atheism (Bishop Barron had plenty of mileage in the early days of ministry against the New Atheists continuously emphasizing that Atheists had no way to respond to the Contingency argument). Hence, the concession of this issue not fully deciding the debate from more sophisticated Theists is a welcome one.
To be clear, I don't think Theism is incoherent either; I'm just saying that when it comes to evaluating coherence, Theism is going to have a lot more challenges to its coherence compared to Naturalism. I also think that when Theists revise their view of God/the Divine Attributes in light of these coherence arguments, it will lower the epistemic probability of Theism since it forces the Theist to take a more specific way that God could be.
Two other things:
- I don't agree that free actions have to be brute. I think there can be indeterministic explanation, so even libertarian free choice can be explained. So I don't agree that God's free choice would be brute. (Note that this is independent of the probabilistic point.)
- I don't think the relevant thing is necessarily whether a theory posits something as *ontologically* fundamental; when I say "fundamental posits," I just mean the things posited by a theory which the theory itself does not explain. Example: Suppose you want to know where the bird-waste on your window ledge came from, so you posit a bird to explain it. That bird is your theory's fundamental posit, even though he probably won't be ontologically fundamental!
- To avoid getting into a lenghty exchange, I will have to defer to Sebastian Montesinos's post "Why I'm an Atheist" that engages with considersations regarding God's LFW act in the context of non-entailing explanations. I think attempts to remove bruteness from Theism just aren't going to work.
- Sure, but my point is that even in that consideration, the respective hypotheses being compared are not just Naturalism and Theism, but Naturalism and God + God's intentions/desires, and whatever bruteness we find in the Naturalistic picture of reality is just going to be shelved back into Theism with regard to why this universe was picked instead of something else.
Hello again! A few brief points (perhaps I'll say a bit more later):
- I don't think the Manson critique of the FTA works; see Donahue's new paper "Divine psychology and fine-tuning" (Religious Studies, forthcoming).
- On Oppy's critique of fine-tuning: suppose we found that every atom had the words "made by God" inscribed on it in tiny letters. Nathan the Naturalist says "this is no evidence for God; I claim it's *necessary* for atoms to be like this. Therefore I've explained it, and without having to posit God." We'd think that response is awful: all Nathan has done is take the improbablity of the god particles, and put it into his theory. But that's what Oppy does with fine-tuning. (I'm not the first person to make this critique; Mike Huemer makes it in his debate with Oppy.)
- I agree that <this exact universe exists> will have an extremely small probability conditional on theism, since there's infinitely many universes God could have made. But it will *also* have an extremely small probability conditional on naturalism, since there's an infinite number of universes consistent with naturalism. So we should appeal to a broad feature of reality (such as <there is a complex universe>), and evaluate *that*. I claim that theism then predicts those features better than naturalism does.
Here's an analogy. Suppose you found a rock sculpture of a well-dressed man in your town, and wanted to compare two theories: H1 says that there is an artist in town while H2 says that there is not. Now, the probability of *this exact sculpture* existing is vanishingly small on both theories: even on H1, the artist could have made anything, not just this sculpture. But the probability of *any* sculpture existing is massively higher on H1, so it's clearly evidence for H1. The same is true of God and the universe.
I've skimmed over Donahue's paper and don't find his response to the consideration raised by Manson and Sinhababu (that God has infinite ways to instantiate life without needing to engage in fine-tuning) that convincing. He tries to get around this with an appeal to considerations about background knowledge, but as I will outline later, I don't think we have the relevant background knowledge for the inference required by the FTA to go through. The response also runs into some complications, which Manley points out in the manuscript I referenced earlier. Imagine if instead of discovering the laws to be fragile (i.e., life permitting under certain constants), we discover what Manley calls Resilience: The relevant parameters are set up so that the universe would produce life even for pretty significant changes to the values of a given parameter.
Now, this seems like it would naturally be evidence for Theism given God's preference for life that has objective value as outlined by Donahue's paper however, it follows through Bayesian reasoning that for any e, P(h|e) > P(h) iff P(h|~e) < P(h) and so if the Theist wants to say that Fragility counts in favor of Theism, then they have to concede that Resilience count *against* Theism, which seems completely contradictory if God is supposed to care about life.
For Oppy's critique, first, as Oppy responded in the video, if you take a view of an Aristotelian modality, then the thought experiment Huemer describes would be metaphysically impossible given the constraints imposed by the initial state. Of course, the FTA proponent is interested in epistemic probabilities, but I think the analogy goes awry in this realm. Interestingly, in Hume's Dialogues, Philo is faced with a similar counter-example from Cleanthes about "voices from the clouds," here it is presented in "Hume's Critique of Religion" by Dan O'Brien and Alan Bailey:
"Cleanthes' first thought experiment invites Philo to consider what conclusion he would draw 'if an articulate voice were heard in the clouds, much louder and more melodious than any which human art could ever reach' (ibid.). If it is supposed that this voice speaks at the same instant to everyone in the world in their own language and provides profound and morally improving instruction, then Cleanthes purports to be confident that humankind's lack of acquaintance with previous examples of heavenly voices and the circumstances in which they arise would not prevent Philo from inferring that this voice is the product of intelligent agency." (pg. 113)
But as O'Brien and Bailey note, the reasons for confirming the inference in this example have little to do with considerations about design:
"The truly problematic aspect of Cleanthes' example would appear to be its appeal to a phenomenon that is exceptionally well-correlated in our experience with intelligent agency. We have, it might be suggested, no experience whatsoever of articulate, meaningful speech emerging from any source that is not itself intelligent or at least constructed by intelligent agents. Thus the link between meaningful sounds and intelligent sources is a tight and well-confirmed one." (pg. 114).
I think a similar consideration can be made in the context of the atom example. The inference we make in this case doesn't have anything to do with fine-tuning; they would still be evidence for Theism even if we found out the laws that produced them were not fine-tuned. Rather, the reason why the inference towards Theism is secure has to do with our considerations regarding our background knowledge (e.g., our knowledge about written works, cultural context, etc). The improbability of probability has little to do with the inference. The problem for the FTA proponent is that when it comes to fine-tuning, we don't have the required background knowledge to make the inference. To quote from Dan Baras' "Calling for Explanation":
"While it is fairly easy to understand what our expectations should be regarding the explanation of [ordinary cases] and why our expectations should be such, it is far from clear that these considerations translate to the fine-tuning of our universe. ... background information based on previous experience, which heavily informed our understanding of these three examples, is not available with regard to the fundamental physical properties of our universe. ... In a way, we have come full circle back to David Hume's famous objections to teleological arguments. Hume discussed a version of such arguments that is based on an analogy between human artifacts (such as watches and houses) and the universe as a whole, and he argued that the analogy is weak. He says that because the universe as a whole is so different from human artifacts, the knowledge that we have about human artifacts should not be projected onto the universe as a whole." (pg. 168-169) (see also Keith Parsons and Robin Le Poidevin for further discussion of this point).
So, if we don't have the background knowledge to assign the relevant epistemic probabilities for Theism, we're thus back to Oppy's point about the constants being either contingent or necessary for Theism. On both options, Naturalism will be a more economical position that can explain the relevant data.
You say: "But it will *also* have an extremely small probability conditional on naturalism, since there's an infinite number of universes consistent with naturalism."
But as I pointed out earlier, the amount of possible universes on Theism is going to be *much* greater on Theism than on Naturalism because for Theism, the range of possible universes includes not only the majority of those compatible with Naturalism but also infinitely many universes not possible on Naturalism. The range of possible universes is much greater on Theism, which means the ratio between the actualized universe vs. possible universes will be much higher than Naturalism.
This is why I don't think the rock sculpture example is analogous to the situation since the ratio between H1 and the datum of the rock sculpture is going to be quite small in the way that the ratio between Theism and the possible worlds that can be actualized on Theism isn't. Here is a better example in my view:
Imagine you've been told that someone ran a 19-minute mile. Now, we have two hypotheses:
H1: An out-of-shape person who had never run before ran this mile
H2: Someone who has set the world's fastest mile time ran this mile.
I think it's fair to say that H1 is the best explanation of the data. Now note, it's completely compatible with H2 that they they also ran the 19-minute mile. However, we know that H2 can run a whole range of more faster times, which just aren't possible for H1. H1 has a much shorter range of potential mile times than H2 does, and the 19-minute mile falls in that range, meaning the ratio between the actual time and possible times for H1 is much smaller than the ratio between the actual time and possible times for H2. I think we're in a similar position vis-a-vis Theism and Naturalism.
Hello once more! A few final points (unless you'd like to discuss things further):
- Yes, Oppy's response is that the scenario Huemer proposes would be metaphysically impossible on his view. But that doesn't actually answer the objection: Huemer's point is that Oppy's epistemology would allow him to dismiss the god-particles, and since someone in that scenario *shouldn't* dismiss the god-particles, there must be a problem with Oppy's epistemology. Oppy simply saying "well, we won't find any god-particles" completely misses the point of the objection.
- Thanks for bringing Manley's manuscript to my attention! I've only skimmed it so far, but if I'd be happy to discuss it with you at a later date.
- I don't actually think that explanation per se is the relevant thing. My argument isn't "features of the universe cry out for explanation, only God can explain them, therefore God"; I'm appealing to probabilistic claims, to the effect that P(Universe and its features/Theism) >> P(Universe and its features/~Theism). So even if Oppy is right that he's "explained" fine-tuning (or whatever piece of data is under discussion), I don't think he's actually succeeded in undermining the argument.
- I don't think we need to employ any sort of argument from analogy, such as the one to which Hume was objecting. I think we can say enough about what a perfect being would be like that we can make various probabilistic claims, such as that a perfect being would value the existence of intelligent embodied life. And that's all we need to run the argument; any sort of analogy between human artifacts and the universe is totally otiose.
- I take the point about the range of universes being much greater on theism. However, I don't think your runner analogy accurately captures out situation; after all, it's not just *compatible* with your H1 that they ran a 19 minute mile; it's positively *probable* that an out-of-shape person might achieve such a slow time. By contrast, there are (I claim) many features of our universe which, though not strictly incompatible with naturalism, are nonetheless extremely *improbable* on naturalism. Similarly, the statue of a man is not actually inconsistent with the random chance hypothesis; it's in-principle possible that you could get a statue-like rock formation by random chance. But it's still enormously improbable.
Thanks again for the interesting discussion!
Feel free to send me a message, and I'd be happy to discuss some of the other issues that were touched upon. I'm always glad to correspond with theists who are knowledgeable about the literature on the analytic philosophy of religion.
- Re: Oppy; the point I'm making is that there are a few relevant differences between the god-particles scenario and fine-tuning. In the former, we have the relevant background knowledge to be able to make a well-correlated inference from marks of design to a potential designer (which incidentally has nothing to do with fine-tuning). In the latter, as Oppy (and others) have argued, we don't have the requisite background knowledge/epistemic probabilities at hand to be able to justify the inference, so both the Naturalist and Theist are forced to look at the modal scope of how our respective worldview can account for the datum. I think Oppy was gesturing to this point when he mentioned Cleanthes's thought experiments, which I mentioned before.
- Re Manley: along with Oppy's chapter in "Arguing About Gods," Manley's manuscript is perhaps the most decisive critique of fine-tuning arguments I've read, and it's generally my go-to critique on the subject. Manley also has another excellent paper, "God and the Bayesian Conception of Evidence," which is a good overview of how to think about evidence in the context of the philosophy of religion in Bayesian terms:
https://static1.squarespace.com/static/55d3621de4b07a4744ce4a23/t/57f50862b8a79bda5001ede3/1475676259380/God+and+the+Bayesian+conception.pdf
- Re P(U|T) >> (U|~T); on the probabilistic claims, my point (following Oppy, Manson, Manley, and others) is that it's generally hard to generate the relevant epistemic probabilities on Naturalism and Theism given the datum of fine-tuning. ("Naturalism, Fine-Tuning, and Flies" by Aron Lucas is also an excellent critique to look at). Therefore, as noted earlier, both the Naturalist and Theist are left with the same options; either fine-tuning is necessary or contingent, and it seems that the Naturalist has the same range of options as Theist does and posits less. The epistemic situation we are in when it comes to fine-tuning will be a lot different than in the case of the god-particles.
- Re Analogy: I recognize that the contemporary Bayesian forms of the FTA are different from the classical design arguments from Analogy; however, my reason for citing the Dialogues was to point out that Cleathes offers a similar thought experiment as Huemer does in response to the general skepticism that Hume offers towards design arguments, and I intended to show how one can respond to the respective thought experiments offered in a way that still allows them to be skeptical of the inferences that the FTA makes. I think the excerpt from Baras outlines the relevant differences between inferring design in ordinary cases and what the FTA tries to argue.
- Re: A perfect being's expectations: I agree that the hypothesis of a perfect being has enough conceptual content to be able to allow us to make some predictions on what we would expect regarding Theism; however, I think in the case of the FTA, there are just so many possibilities that far better than what we currently have. For example, I see no read why intelligent life has to be embodied; I think intelligent life could have been a lot more impressive than it actually is (see the section "Humans" from my colleague's case for Atheism below: https://naturalismnext.blogspot.com/2022/12/why-im-atheist_6.html), as well as a host of other considerations. Now, of course, the Theist can build in these further stipulations into their hypothesis, but I think doing so leads to a lot of problems (see the post below for a good overview of "stalking-horse" objections to the FTA: https://fric.substack.com/p/cosmological-fine-tuning?utm_source=profile&utm_medium=reader2)
- Statue Analogy: Not surprisingly, I disagree with the categorization of the relevant datum as being improbable on Naturalism, but I recognize that arguing as such would be beyond the scope of the post. My point was just to outline why I consider Theism's modal scope being much larger than Naturalism to be a problem for Theism as a theory. Such a problem also impacts the overall intrinsic probability of Theism as a whole, and that's all my comments were intending to argue for: that given the considerations you've offered, it seems Naturalism has a higher intristic probability than Theism. Of course, I recognize that this is something that could be overturned by the relevant posterior considerations such as what you've offered.
I echo your sentiments on the discussion and look forward to more in the future.
Thanks for the post! I personally do not see the need for an appeal to your third prong when assessing intrinsic probability. There are an infinite number of velocities at which the particles in worlds m and n could have been moving and each theory asserts a particular velocity and, in that sense, they are both equally modest. moreover, neither theory contains within itself more or less inductive support relations between its parts than the other so they also appear roughly equally coherent.
So here I don’t see the problem with saying they are equally probable intrinsically. Why should infinite velocity be understood as being more in need of explanation than any other particular speed? The idea seems to be that you are understanding infinite velocity as being a hard limit in the upward direction such that there is no possibility of an increase in velocity. But it’s not clear why that should matter. The difference here is only between where on the velocity continuum or range the particle sits. Theory n posits a particle with a particular velocity somewhere between the highest and lowest while theory m posits a particle with the highest. I’m having difficulty understanding what you are considering more brute, unexplained, or arbitrarily limiting about n relative to m.