On Humean Caution

Against Jumping The Gun

In the last few weeks, “miracle discourse” has made the rounds on substack. When I shared my thoughts, I thought it was the end of it.

Christians Have Good Reason To Reject Modern Spectacular Miracles

But then Scott Alexander posted his lengthy thoughts, resurrecting the discourse.

The Fatima Sun Miracle: Much More Than You Wanted To Know

At one point, Alexander says this, and it sticks out to me:

I will admit my bias: I hope the visions of Fatima were untrue, and therefore I must also hope the Miracle of the Sun was a fake. But I’ll also admit this: at times when doing this research, I was genuinely scared and confused. If at this point you’re also scared and confused, then I’ve done my job as a writer and successfully presented the key insight of Rationalism: “It ain’t a true crisis of faith unless it could go either way”.

(bold/italics are mine)

Upon reading this, I realized that my brain operates differently from Alexander.

On Hypothesis Testing And Its Shortcomings

In the philosophy of science (or at least how it’s expressed online), thinkers rely heavily on hypothesis testing. My best attempt at expressing hypothesis testing as it’s used on substack is:

“If you have hypothesis A and hypothesis B, and the data better (even if imperfectly) fits hypothesis B, hypothesis B is more likely to be true.”

There are bayesian justifications of this kind of formulation.

For example, if we’re trying to evaluate if someone stole my wallet or I just lost by wallet, we formulate different hypotheses, and think of data we would gather to support either one. In the stolen wallet hypothesis, one data point we’d look for is evidence of forcible entry. If I live in a small apartment and my wallet is missing, but also the door is broken, all of my things are scattered about, and valuable items are missing, it’s more likely that someone stole my wallet than I simply lost it.

There are at least two shortcomings with hypothesis testing. These problems don’t render hypothesis testing useless, but instead demonstrate its limitations:

1. A hypothesis being a more likely match with data does not mean it is correct, just that it is more likely correct than alternatives. In a bayesian sense, you may be obligated to accept a hypothesis because it best fits the data, but you also have the option to be agnostic, saying that there’s not good enough evidence to accept any available hypothesis as plausible.

2. Hypothesis testing (and its logic) is optimal when you’re running an experiment (controlling for variables, measuring everything, etc.), but it becomes less useful when evaluating mysterious circumstances that happened in the past, where we have little data and less direct experiences with the events that have occurred (sun miracles, virgin births, universes beginning, etc.)

Point number one seems easy enough. If you’re operating in an environment with uncertain information and you’re trying to act optimally rational, it makes sense to accept the best hypotheses, even if they’re imperfect (assuming inaction or agnosticism isn’t an option). But at the same time, when we’re evaluating truth claims on religion and science, we are not operating in that environment. We have the option to be agnostic.

Point number two may seem problematic on first blush, but it need not be. Just because hypothesis testing is less helpful when evaluating mysterious, rare, and past occurrences, doesn’t mean that it’s useless for those circumstances, just that we should be aware of its limitations and higher uncertainty.

Humean Caution

In Alexander’s telling, the data briefly appeared to fit the “miraculous” hypotheses more than the naturalistic ones.¹ Under a hypothesis testing framework, if it’s more likely that Fatima was a miracle or evaded current naturalistic explanations, then it’s more likely that Christianity, specifically Catholicism, is true. That’s unsettling if you’ve lived your life as a non-Christian!

I want to be very clear here that Scott Alexander knows (and has forgotten) more science and philosophy of science than I do. Still, I’m genuinely puzzled why the temporary shortcomings of naturalistic explanations for Fatima gave him a crisis of faith. Similarly, I have seen people on substack use hypothesis testing to a fault, especially in circumstances where we just don’t have enough data or knowledge to evaluate an explanation. I think SA’s crisis of faith followed a similar fault.

To avoid that fault in the future, I recommend what I call Humean Caution.

Humean Caution is simply the predisposition to disbelieve supernatural explanations because humans have a hard time discerning even naturalistic explanations. Further, if you look closely at miracles as proof of religious claim, you’ll see that they make much bolder claims than their evidence implies, even assuming the supposed miracle is miraculous. To understand what that means, we have to understand the philosophy of David Hume.

Induction, Observation, And Miracles

The problem of induction is one of Hume’s most famous philosophical problems. The shortest way of communicating it is that humans may observe events happening in sequence, but we don’t actually observe causation.

When a billiard ball hits another billiard ball and they both move, we infer that the first ball making contact with the second ball caused the second ball to move. In reality, the movements could have just as easily been spontaneous. Both explanations would look indistinguishable to the human eye. We affirm the causality hypothesis because it feels right and creates the best explanation for predictable results. Yet still, induction is fallible and any induction could be falsified at any time.

In the context of hypothesis testing, humean caution warns against accepting wholesale explanations of weird, rare phenomenon. That means rejecting hypotheses that themselves aren’t substantiated. To express this, I’ll refer to atheist edgelord Christopher Hitchens and the purported miracle of the virgin birth of Jesus.

Put in a more succinct way, even if we could go back and time and track the Virgin Mary and confirm that she never consorted with a man before becoming pregnant with Jesus, that by itself does not mean that Jesus was the son of God!

Under a hypothesis testing framework, I grant that it is much more likely to be the case that Jesus is the son of God conditional on Mary being a virgin, but it does not mean that Mary being a virgin proves Jesus is the son of God. The evidence we hypothetically observed (that He was born of a virgin) does not prove the outcome (that Jesus is the son of God).

Indeed, many of these supernatural claims entail causal relationships that no human being has observed. How would one demonstrate that Jesus was God causally? How could we show that a sun miracle was caused by God or Mary and not some other powerful entity like aliens? The response to this rhetorical question is usually that we judge God or Mary to be more likely to act in these ways. But again: More likely is not the same as true or certain!

The language of miracles and religion plays fast and loose with the human bias for inference, how we infer causation and agency everywhere. Put more specifically, miracle language collapses the distinction between “I did this special thing” and “Everything I say about myself is true.”

In this way, a man rising from the dead does not itself prove him to be God, as there could be any number of alternative explanations.² The religious often assume a miraculous event automatically proves a maximal explanation of a religious hypothesis (that the event happened, that it was performed by the religious figure, and the the religious figure is all of the things she says she is), when the actual data proves minimal (that the event happened).

The Humean Response

In the context of Fatima and other miracles, the Humean retort is so what? Something weird happened in Portugal over 100 years ago, does that mean I have to join the Catholic Church, or take the Pope’s opinion about current events and politics more seriously, or a host of other propositions? No!

It can be the case that something miraculous happened at Fatima and all of that other claims associated with the Catholic hypothesis are still false. Obviously, under a hypothesis-testing framework the Catholic hypothesis is less likely to be false, but the weirdness of Fatima does not outweigh all the other variables that we would consider when evaluating the truth or falsity of Christianity or Roman Catholicism.

Does Humean Caution Render Me Stubborn?

I understand if a believer reads this and finds it frustrating. It sounds as if I’m saying no miracle could convince me of a religion’s truth, and that there’s no evidence that could lead me to accept a miracle. Upon reflection, I don’t think that’s a straw man of my position. I know that makes me sound stubborn or irrational, but let me explain:

First, I think agnosticism on matters of fact is permissible for at least two circumstances: On events that one does not directly perceive and for facts that have no measurable consequence on one’s life. I like knowing and understanding things as much as the next guy, but the number of facts I don’t know out there (will always!) far eclipse the ones I do, I can’t know it all, and so I have to make a strategic decision to not form opinions all the time.³

Though the question of God’s existence certainly has consequences for my life, the question of Fatima does not.

To the extent you can ever get me to say I don’t have a naturalistic explanation for a purported miracle, I’m going to err on “I don’t know,” before “it was a miracle!” The causal variables where one could evaluate miracles (supernatural powers) are always going to be out of my perception for evaluation. Though I can evaluate and eliminate known natural explanations, I can’t evaluate a supernatural one, even if I believe them. I don’t see how I could accept a supernatural hypothesis without direct evidence for it and also not accept a natural hypothesis without direct evidence for it. I’m just holding the supernatural explanation to an equal standard.

Second, the reason why miracle discourse is hot on substack is because the non-believers who take it seriously are engaging believers on even playing field. Other skeptics are saying the equivalent of: “I don’t believe miracles can happen, but let’s assume they can, and evaluate the evidence, because I can ‘beat’ you on both footings.”⁴ I personally don’t have the energy or patience for this!

But even if I did, I am more interested in examining these formulation questions, that are swept under the rug when we skip directly to hypothesis testing: What does a miracle actually look like, how would we know that it was God that did this, and so on and so on.

Another way of putting this is that I can’t evaluate a hypothesis if the hypothesis doesn’t make sense to me. And many theistic arguments and hypothesis don’t make sense to me! When I say that, I don’t mean that in the edgy internet atheist “I can’t take that opinion seriously” kind of way (that Bentham dunked on as I was writing this). Rather, I mean, “when I apply scrutiny to the language being used here, I find it inapplicable or unsubstantiated.”⁵

Third, I don’t think miracles would be good evidence for God, even if I did believe in God. I basically wrote about this in my last post, so I won’t linger too long. Miracles are always going to be bad pieces of evidence for God’s existence because they are so rare and sketchy. If I were God, I wouldn’t want people to believe in me over miracles, and to the extent I would, I would make those miracles unambiguous and theologically important, as we see in scripture. Put in hypothesis-testing terms, in my God-is-real hypothesis, spectacular contemporary miracles would not be real, so I couldn’t affirm God’s existence and a miracle at the same time.

Fourth and finally, none of this means I’m stubborn or unreasonable. People believe or don’t believe in God for various reasons, some of them self-reinforcing. There are millions of Christians who believe in Christianity and have never heard of Fatima or other direct experiences of miracles outside of the Bible. Some religious people get into their faith through apologetics, others through service, others through pilgrimage, and others still through other means. I have my personal criteria for conversion, but I’m not sharing it here, because that’s none of your business.

In the meantime, I’ll practice Humean Caution whenever people make extravagant miracle claims. In my experience, these claims are usually bunk; when they aren’t, they say much less about God, and more about His proponents’ unfamiliarity with Hume.⁶

1. Though it is possible that Alexander was just being rhetorical. If you read his post like a “story” where he is the protagonist and debunker, his admission of a “crisis of faith,” lines up at around the time you’d expect “the black moment” in a story - where the protagonist is at their lowest, facing defeat, and so on, only to find victory and glory. It’s a storytelling tactic to create tension and satisfaction with the outcome.

2. Among them, that he was resurrected by God

3. And everyone is like this, whether they admit it or not.

4. And from the skeptic perspective, each discourse is formulaic and has the same outcome, much like a power rangers episode.

5. A good example of this that I recently stumbled upon is the goodness of Heaven, and how apologists often talk about how good heaven will be. I can’t really evaluate this hypothesis because happiness as we experience it is psychologically an environmental reaction and behavior-regulating tool, evolved in a context of scarcity, fear, and all sorts of other contexts that are not applicable to heaven. What’s more, happiness fades over time if you’ve ever experienced it. To be in a state of constant happiness absent any sort of social or physical fear is to talk about happiness in ways that are just extremely foreign to me, so foreign that I don’t think the term happiness applies to whatever this would be. The agent experiencing happiness would not be like a human experiencing happiness. And all of this opens up a theological can of worms!

6. Also, I didn’t have space to go into it on this post, but one of my pet peeves is when people say Hume said miracles were impossible or couldn’t happen. He never said that!