A legacy post originally published on AUGUST 13, 2010 at 8:40 AM
🔗 Open Challenge to Atheists by Joshua Rasmussen
Well, this is really a challenge for skeptics of a necessary being (a necessarily existing, causally powerful entity), but I wanted a catchy title, and as a matter of sociological observation, atheists are typically (though not always) skeptics of a necessary being.
The challenge is this: come up with a general, non-ad hoc causal (or explanatory) principle that’s evidently more plausible than any of the ones that imply that there is a necessary being.
It is widely agreed (by the experts) that each of these principles implies the existence of a necessary being:
- Contingency implies explicability
- The existence of any contingent x‘s has a cause
- Every contingent fact has an explanation
Some lesser known routes to a necessary being make use of any of these:
- Every explicable contingent fact has an explanation
- Any possible beginning is explicable
- Every wholly contingent fact has a cause
- Any possible contingent arrangement is causable
Each principle above seems to be supported by a vast array of instances, and we’ve never observed any counter-examples (though it is controversial whether or not quantum mechanism and/or the “taxicab objection” might imply counter-examples to (3)). The principles are also relatively simple and so don’t seem to have low prior probabilities. The challenge for a skeptic of a necessary being is to present a causal principle that accounts for all known (or apparent) cases of causation (explanation, causability, or explicability) and that is at least as simple or non-ad hoc as any causal or explanatory principle that implies a necessary being. Good luck.
The challenge is also open to believers in a necessary being.
Related challenge (inspired by work by J. Schmid): find a theoretically more virtuous (simpler, less ad hoc, wider scope, etc.) principle that *competes* with this:
Principle of Explaining Imperfect Things: every imperfect xx (individual or collective) have an explanation.
If we are to avoid circular explanations, this gives us a perfect foundation of the total of imperfect things.
Different replies are available. One idea (which may be best, not sure) is to simply decline to meet the challenge. Instead, motivate reasons to make exceptions (e.g., from Platonism, problem of evil, problems with the concept of “perfection”, etc.).
Just to give a response to it if I put on my naturalist hat for a minute. I would reject 3 as it is the most suspect as you point out in the challenge I would argue that quantum fluctuations could be uncased. I would argue that the universe itself is uncaused and that the causal relationships we observe within it are the result of natural laws and processes, rather than the direct effect of a necessary being. This would reject the idea that everything that exists must have a cause, and instead propose that some things, such as the universe itself, exist without a cause. Then defend the idea of the universe is a brute fact by arguing that it is a more parsimonious explanation for why the universe exists. Positing a necessary being, or a cause for the universe requires introducing additional entities or concepts that are not directly observed and which do not provide any additional explanatory power. The concept of God is too vague and ill-defined to provide a satisfactory explanation for the existence of the universe. Then point out that scientific materialism, which holds that everything in the universe can be explained in terms of matter and energy and that there is no need for supernatural or metaphysical explanations. Another, more modest route is to defend the idea of the universe being a brute fact by arguing that it is simply unknowable whether or not the universe has a cause, and that it is therefore reasonable to posit that the universe is a brute fact until we have more evidence one way or the other.
My response to the challenge would be to argue that causality and explanation are emergent properties of the universe and are not fundamental concepts that can be reduced to a single principle or entity. This could be based on the idea that the universe operates based on complex and interconnected systems, rather than a single cause-and-effect chain.
All great points. Indeed, there are ways of granting an explanatory chip to the theist, while still adding potential chips on the other side. For my part, I think simplicity considerations involve trading posits. More on that here: https://joshualrasmussen.blogspot.com/2021/02/does-theism-multiply-complexity.html
Josh,
Interesting points were made in your blog post. How would you respond to the counterpoint that God lacks explanatory power because God is all-powerful and able to do anything? You can just say God did it. Now I have heard the response that some theists want to appeal to God’s character. God will do what is morally good. However, I believe this will run into the Euthyphro dilemma for God. I am not sure any theist offers a slam dunk case against it. If this works then the naturalist has shown that theists claim they have the explanatory advantage over naturalism to be very problematic.
Thanks, John, very great question and point. I offer a couple thoughts.
First, a standard view is that theism is not specific enough to explain everything. For example, theism (on its own) doesn’t say anything about why there is a coffee cup. Theism isn’t that specific. We could imagine God exists with no coffee cups. If a theist *adds* to theism, “God made that coffee cup,” then that makes the theory more complicated. “God did it,” where “it” refers to making a coffee cup, isn’t included in the hypothesis that God exists. The hypothesis that God exists doesn’t even mention coffee cups. So theism doesn’t explain that on its own.
Instead, theism predicts that there is an explanation of more general things, like why there are contingent, limited things (since God would be a non-contingent, non-limited prior cause). So theism need not explain everything to explain some things.
Second, and similarly, a theist can grant that theism (on its own) doesn’t explain everything about goodness (per the Euthyphro Dilemma). For example, God doesn’t need to make the feeling of being loved a good feeling. Rather, it’s just the nature of the feeling of being loved that there is something good about it. Similarly, God could be loving, by nature, without first thinking, “I will make myself loving” or “I will make being loving a good thing.” On this account, some things are just good by nature or in relation to their effects on conscious beings. They are not good only because God says so.
In case you may be interested, I say more about God’s relationship to morality in “Foundation of Morals,” in How Reason Can Lead to God.
A nomological-explanation is just a more compact description of some phemomena relative to a world-model, where the laws + initial conditions entail (make more probable) the result.
For example, Ohm’s Law, which holds in our actual world, is a compressed statement of a very large number of observations. But there are plenty of logically-possible world models where it does not hold. But it explains “why” adding a resistor to a circuit with constant voltage results in an amperage decrease.
To say my theory explains some contingent fact X is just to say that my best actual-world model is a member of the set of logically-possible worlds whose dynamic laws make the result more probable, given the antecedent conditions.
If EVERY contingent fact had a nomological-explanation, then the algorithmic complexity of the world model would ipso facto be zero. You can’t have an entailment relation from a zero-length string. That’s like asking “what follows deductively from the sound of one hand clapping?” The threat of Modal Collapse lurks just around the corner.
Explanation is, and has to be, a relative measure, not an absolute one. Explanations can be simple, but they cannot be absolutely simple, otherwise they wouldn’t be explanations.
I’m not sure I’m understanding your comment. It seems like you think explanations are just compact statements of an array of observations. (I highly doubt that is true…. but I wasn’t sure if that’s what you were saying.
Also, you argued against every contingent fact having a nomological explanation. But I’m not sure that exclusively nomological explanations were being discussed. Any causal explanation, even a non-lawlike one would fit the bill.
Yes, that is basically correct. Causal explanations[*] are compact (i.e. parsimonious i.e. low algorithmic-complexity) statements which make the observations more probable (where the limit case of strict entailment just means “has a probability of 1.0”).
We live in a world that is a subset of those logically possible worlds where the action of wind and waves on sand tends to make the spontaneous formation of mechanical watches improbable, but also one in which there are upright apes who are interested in things like metallurgy and getting to the airport on time, which makes their existence substanitally more probable.
Yes, I am using nomological here to refer to all causal explanations, even when they are probablistic rather than the strict logical entailment in nomological-deductive systems, (hence my parenthetical above).
It’d be rude of me to start playing burden-of-proof-tennis after your thoughtful response here, but really, you can see for yourself by just testing out a couple non-controversial examples of causal explanations and trying to find a counterexample.
To assert that smoking causes lung cancer is just to assert that the best world-model to adopt tells you to raise your expectation i.e. consider more probable the observation of cancer temporally following smoking. And this is true even though the relation here is not a strictly deductive one from the nomological model.
You still have a relative reduction in the total randomness of your world-model compared to someone who has all the same observations as you but lacks the causal compactness, and for whom the appearance of lung cancer among smokers is an inexplicable (i.e. random with respect to the model i.e. non-compressible) fact.
Key word in the above being “relative”, which is all an explanation can ever be. You can derive a surprising amount of observations (deductively or probablistically) from surprisingly simple world-models, but you can’t derive anything from an “absolutely simple” (zero-length) world model.
Given that your world model must have non-zero complexity, you can immediately see that either all the statements in this model will be logically necessary (in which case, hello modal collapse!) or it will contain at least some logically contingent facts whose randomness cannot be further reduced.
There you have it. It is knowable as an a priori truth that logical contingency does not entail explicability.
[*]I don’t even really have any problem if someone wants to use the term “explanation” to refer to the the epistemic virtue of something simple like the Peano Axioms “explaining” loads and loads of mathematical truths, so long as no one is going around thinking this is a causal explanation.
Another way to motivate this is to realize that what we want out of an explanation is fundamentally a reduction in randomness, which is the same thing as saying a reduction in algorithmic complexity, which the same thing as saying a reduction in surprise.
If you got plopped into a different logically possible world, you wouldn’t have the foggiest idea what to expect if you shot a cannonball with a certain mass with a certain force. There’s LPWs where it just hovers there, LPWs where it spells out your name in cursive, LPWs were it turns into flowers… your world-model would be stuck just describing the anterior and posterior conditions with no compression, no reduction in surprise value.
It just so happens that in the actual LPW we live in, the best model is one that has lots and lots and lots of non-randomness in it. Is it completely non-random, or do something-something quantum-somethings present a genuine case of irreducible randomness? I have no idea. Above my pay grade. The point is there is no a priori reason not to accept a LPW-model that contains non-compressible randomness, given how infinitely many LPWs there are that are nothing BUT randomness. We should simply be grateful for what we’ve been given and not look a gift horse in the mouth.
Notice how this analysis ports over seamlessly to probablistic (non-strictly-ND) causes. I cannot predict with the certainty of classical physics, what the next Democratic nominee for president will say at any given moment. But there are a certain set of things everyone would be astonished (surprised) to hear them say, and a certain class of things everyone would be unsurprised to hear them say. My world model rules out them saying that e.g. billionaire hedge fund managers deserve a tax cut because they’re salt of the earth people who just work so darn hard, or that Brown v. Board of Education should be overturned.
Theists do this sort of thing all the time when you ask them to evaluate the sorts of things their god or gods would or wouldn’t do, even though in the set of all LPWs there are by definition some worlds where we get a different sort of thing those gods would or wouldn’t do (again, assuming you’ve avoided modal collapse).
Hundreds of millions of theists belive God WOULD NOT BE EXPECTED to allow the New Testament to be composed only to be filled with doctrinal errors necessitating a do-over in the form of The Koran. And hundreds of millions of other theists believe this is exactly what happened, because that’s just the kind of person Allah is.
Whoever is wrong here, the wrongness isn’t a matter of there not existing a LPW where they’re right.
And so we return to the OP’s questions about logical-contingency and “explicability”. What is it the theist is saying when she says their gods “explain” some thing X? It means they have adopted a world-model under which the result X is made more probable i.e. to be expected i.e. non-random with respect to the character of whatever gods there are in that LPW.