Saturday, August 12, 2017

'But it's the best explanation!' - how bullshit beliefs are justified

Folk who believe in fairies, or miracles, or alien visitation, are generally fond of an argument called ARGUMENT TO THE BEST EXPLANATION.
 
Here's an example of argument to the best explanation (or abduction, as it's sometimes known):
 
I see shoes poking out from under the curtain and the curtain twitching slightly above them. I can also hear breathing. I infer there's someone standing behind the curtain. Why? Because that's the best available explanation of what I observe. True enough, the twitching might be caused by the breeze from an open window and the shoes were just coincidentally placed in the same spot. But I reckon that's a bit less likely than that there's someone standing there (for what explains the breathing noise?)
 
Quite what makes an explanation the 'best' is controversial, but there's some agreement that the simpler and more elegant an explanation, the better. So, for example, I could explain that twitching curtain by supposing that there are three dwarves standing on top of each other behind the curtain, but that's a far more complex and less elegant explanation for what's oberved than that there's just a single person there.
 

We use argument to the best explanation a lot. For example, it's used by scientists to justify positing various unobserved entities. We may not be able to directly observe electrons, or a very distant heavenly object, but their existence can be the best explanation of what we can observe, such as certain astronomical or experimental results. In which case, we're justified in supposing these unobserved entities - electrons, distant planets, and so on - exist.
 
So argument to the best explanation seems to be a legitimate form of reasoning - a form of reasoning employed even by scientists.
 
However, argument to the best explanation is often also the first port of call for those who believe in spooky, wacky stuff.
 
For example, conspiracy theorists rely on it a lot. They say: 'Can you explain why the Twin Towers came straight down like that? No? Well I can - it was a controlled demolition! An inside job! See - that's the best explanation!' In reply, we may have to admit that we can't, right now, explain the striking way the Twin Towers collapsed. A controlled demolition, on the other hand, would very neatly explain it. So, if the conspiracy theorist's explanation is currently the best available, shouldn't we accept it ? Aren't we now justified in supposing a controlled demolition took place?
 
Or suppose we can't explain the testimony of various supposed witnesses to a flying object. Suppose it's observed by a number of individuals who describe something like a large flaming object hanging stationary over a building site. They are otherwise reliable witnesses. We cannot easily explain what they saw in terms of it being a planet, or a plane, or a prank, or an illusion, etc. So it seems the best available explanation is that a large fiery object really was spotted in the sky, right?
 
'You can't explain it - I can explain it by appealing to [aliens, gods, ghosts, etc.; insert your preferred woo here]. therefore my explanation is the best available, and thus the most reasonable!' is a popular refrain from those who believe in spooky stuff.
 
This sort of move also crops up a lot in religious contexts. Take the Resurrection of Jesus, for example. Typically, this is argued for using argument to the best explanation. We are told Biblical scholars agree on certain facts: that Jesus' tomb was empty, that Jesus was seen afterwards by several different witnesses, and so on. And then it's suggested that a risen Christ is the best available explanation for these reports - a better explanation that that all the witnesses were lying, or deluded, or that Jesus had not really died, etc.
 
So what, exactly, is wrong with this sort of justification of belief in 9/11 conspiracy, alien visitors, ghosts, fairies, and even The Resurrection? I'll discuss this more in my next post... 

(this post was previously published on the CFI website, where it was followed by this...)

In the previous post I pointed out that while argument to the best explanation is a legitimate form of inference used even by scientists, it's often also the first port of call of those who believe in conspiracy theories, miracles, and other wacky stuff. Let's now take a look at what can go wrong with arguments to the best explanation.
 
Suppose I can't find my keys. I was sure I left them by the door. Then I find them on the mantelpiece. I'm baffled - I just can't explain how they ended up where they did. My friend then suggests the following:
 
'Your house has gremlins! Think about it - such tricksy beings would want to hide your keys, wouldn't they? And they'd have the power to do it, too. So the existence of gremlins in your house would neatly and easily explain how your keys ended up where they did. Gremlins explain what you can't otherwise explain! So gremlins are currently the best available explanation for the disappearance of your keys! So you should believe in gremlins!'
 
Is this a good argument for the existence of gremlins? Surely not. But what's wrong with it?
 
Well, while gremlins would indeed neatly and easily explain what happened to my keys - something I am otherwise struggling to explain - the probability remains much higher that some other much more mundane explanation is correct. True enough, I don't remember putting my keys on the mantelpiece, but maybe I did. My memory is very good, but occasionally I get things wrong. Or perhaps someone else entered the room moved the keys while I wasn't looking. Neither of these explanations might strike us as very likely, but that some such mundane explanation is correct is remains far more likely than that gremlins moved my keys. For we know that people do sometimes misremember, play tricks on each other, and so on; whereas we have very little evidence for, and considerable evidence against, the existence of gremlins.
 
Just because I don't know, and indeed can't come up with a plausible-sounding explanation of, how my keys ended up on the mantelpiece obviously doesn't justify me in supposing gremlins put them there.
   
Precisely what's gone wrong in such cases can vary, but as a general rule, (a) while fraud, or hallucination, or some other mundane explanation might in each case seem pretty unlikely, the fact is, we know these things do nevertheless happen (frauds do occur, people do have astonishing hallucinations, etc. etc.), and (b) there's typically little if any evidence for the existence of the mysterious beings other than such anecdotes, and indeed there's often also a great deal of evidence against their existence (e.g. in the gremlins case: if they existed then we'd expect to occasionally spot them lurking under the floorboards; how could they survive (on what food, etc?); why have they never been photographed of caught on CCTV, etc.?)
 
Thus, on balance, it's much more likely that some sort of fraud, or hallucination, etc. happened, even if, as a matter of fact, no especially plausible particular explanation is forthcoming. I'll illustrate (a) below.
 
BTW. For an interesting case study, see William Lane Craig's The Resurrection of Jesus here. Craig says:

In summary, there are four facts agreed upon by the majority of scholars who have written on these subjects which any adequate historical hypothesis must account for: Jesus’ entombment by Joseph of Arimathea, the discovery of his empty tomb, his post-mortem appearances, and the origin of the disciples’ belief in his resurrection.

Now the question is: what is the best explanation of these four facts? Most sholars probably remain agnostic about this question. But the Christian can maintain that the hypothesis that best explains these facts is “God raised Jesus from the dead.”

We’re amazingly prone to 'see' things that are not really there. One of my favourite examples involves a strange object seen over the building site of a new U.S. nuclear plant back in 1967. Police arrived. One officer confirmed that 'It was about half the size of the moon, and it just hung there over the plant. Must have been there nearly two hours.' The County magistrate said he saw 'a rectangular object about the size of a football field'. There was even a rogue radar blip reported by air traffic control!What on earth could this amazing object be?

We know, pretty much for sure, that what was seen by those police officers and the magistrate was the planet Venus. Journalists arrived on the scene, were shown the object, and chased it in their car. They found they couldn’t approach it. Finally, they looked at it through a long lens and saw it was Venus. That radar blip was just a coincidence.

What does this show? Every year there are countless amazing reports of religious miracles, alien abductions, ghosts, and so on. In most cases, it’s easy to come up with plausible, mundane explanations for them. But not all. Some remain deeply baffling.

So, given this hard core of 'unexplained cases', should we believe in such things, then?
No. For, as my UFO story illustrates, we know that some very hard-to-explain reports of miracles, flying saucers, and so on will likely show up anyway, whether or not there’s any truth to such claims. That 1967 case could easily have been such a baffling case if the journalists had not investigated and found the truth. It could easily have gone down in the annals of UFOlogy as one of the great 'unexplained' cases. UFO enthusiasts would have said: 'That is hardly likely to be an hallucination, or a lie, or just a normal aircraft, surely? The best explanation is that there really was an extraordinary object in the sky,''

So when people say:

'The best available explanation of these amazing reports is that witnesses really did see an alien spacecraft (or a resurrection, or a miracle, or an angel, etc.) - for it's just implausible that they could have all hallucinated, or been tricked by an illusion (what explains that radar blip?!), are lying (they're police officers and a magistrate - and independent witnesses!) etc.',
 
remember: such cases are going to crop up every now and then anyway even if there aren't any alien spaceships visiting us, are no resurrections, etc. But then such reports are not evidence for the existence of such alien craft. Our mere inability to come up with a plausible explanation of such reports shouldn't lead us to conclude that there probably was an alien spacecraft, a resurrection, or whatever.

POSTSCRIPT

I have had some helpful comments on this from Calum Miller, Secular Outpost (on twitter) (i.e. Jeffrey Jay Lowder), and subsequently feel the need to clarify a few things.

1. the above piece is not intend to show there could never be evidence good enough to believe the woo thing (miracle or whatever) happened - there could.

2. Everything I say above can be expressed in Bayesian terms, if you prefer to think in a Bayesian way. In effect, I am focusing on what numbers get put into the Theorem.

3. I focus on the probability of mundane explanations being correct, not on the prior probability of the exotic explanation being correct, which is of course also important (base rate fallacy) - though I do (deliberately) mention the prior probability in my gremlin example where I point out the prior probability probability of gremlins existing is very low (as there's little evidence for their existence and a great deal against).

4. My point here is that we seem systematically prone to giving a lower probability to mundane explanations than we should. I myself feel the tug of this whenever I am presented with best explanation arguments by 9/11 truthers, fairy-ists, miracle claimers, etc., - especially when the various mundane explanations of offer are listed and considered individually. We can easily end up supposing the prob of some such mundane explanation being correct is exceptionally low when it is not nearly as low as we suppose. I am suggesting that this is something like a cognitive bias that we should be on our guard against. I am suggesting that the attractiveness of many woo beliefs is due not just to the fact that we ignore the low prior probability of gremlins existing, miracles occurring, etc. (though we do) but also because we tend systematically to underestimate the prob of some mundane explanation being correct. One way of dealing with this bias is to expose oneself to a lot of different kinds of examples - that's exactly what I try to do here.

9 comments:

Ron said...

Nice post Dr. Law. Shameless plug here: I wrote an article for the Secular Web in which I demonstrate how some prominent "inference to the best explanation" arguments offered by Christian apologists commit straightforward probabilistic fallacies. Even if we grant these apologists their premises, these are often insiufficient to deliver the conclusion "the resurrection probably happened."

https://infidels.org/library/modern/aron_lucas/hume.html

Michael Fullerton said...

Following your line of reasoning, isn't your belief that the best explanation is that certain beliefs are bullshit, therefore bullshit?

Peter (Oz) Jones said...

Dear Stephen
Thanks for so many interesting posts too.

see William Lane Craig's The Resurrection of Jesus here. Craig says:

In summary, there are four facts agreed upon by the majority of scholars who have written on these subjects . . .

Just love how WLC sneaks in "facts" whereas they are "stories" - so best explanation for some stories aint facts!

Peter (Oz) Jones said...

Dear Stephen
Thanks for so many interesting posts too.

. . . see William Lane Craig's The Resurrection of Jesus here. Craig says:

In summary, there are four facts agreed upon by the majority of scholars who have written on these subjects . . .

Just love how WLC sneaks in "facts" whereas they are "stories" - so best explanation for some stories aint facts!

Anonymous said...

You have contradictory beliefs, so logically certain beliefs are bullshit. Key word: certain. You can't conclude that that belief is bullshit based on the claim of that statement.

In short, no.

Jeffery Jay Lowder said...

Hi Stephen. This was and is interesting. I Was disappointed you don't seem to say anything about the prior probability of claims about "wacky stuff." Did I miss it?

Surely part of what goes wrong in arguments to the best explanation is that many of the "wacky" explanations are ad hoc. It would be interesting to see how you would formally characterize your objection to Bayesian versions of arguments to the best explanation.

It seems like your objection is intended to show that Probability(wacky claim is true | wacky claim is best explanation) is ALWAYS going to be less than Probability(wacky claim is false | wacky claim is best explanation). Although I'm a fellow atheist, I'm uncomfortable with saying that fraud, hallucination, etc. is ALWAYS more likely than a "wacky" explanation.

It seems to me that this debate--between proponents of arguments for the historicity of "wacky" claims and critics of such arguments--is hard to resolve without a quantitative assessments of the relevant probabilities. Without quantitative estimates, the wacky-ists will claim "The explanation evidence in this case is so good, it outweighs the general track record of 'wacky' explanations." Meanwhile, anti-wacky-ists will claim, "The explanation evidence is so weak, it doesn't outweigh the general track record of 'wacky' explanations." As I say, without defensible numbers, it's hard to see how either side is going to persuade the other.

For example, the McGrews argue that the Bayes' Factors support the Resurrection over its denial by something like a ratio of 10^43 to 1. For your objection to work, I think you'd need to either show the prior probability of the R is less than 10^-43, OR that their Bayes' Factor estimates are somehow wrong (or unjustified). I realize your article wasn't written with the McGrews in mind. At the same time, I think their argument exemplifies why I'm reluctant to say that a non-wacky explanation is always going to be better than a wacky explanation.

Ondoher said...

This is also where parsimony comes in. Occam's razor. The big problem with these explanations is that they introduce a new entity, a new assumption, into the discussion. Mundane explanations like "you forgot you moved the keys," rely on normal, everyday things we take for granted. However, gremlins adds a whole new assumption, a new entity, into the mix. Occam's razor would suggest, all other things being equal, we favor those explanations less encumbered with assumptions.

No need to invoke gremlins, when forgetfulness relies on less.

Michael Fullerton said...

Your article strongly tries to imply that all unofficial explanations for what happened on 9/11 are BS because they are arguments to the best available explanation. The only image you show is one of the Twin Towers after they've been hit by planes. Yet you never give a concrete example of how 9/11 "conspiracy theorists" use such faulty reasoning.

As you've stated, scientists use argument to the best explanation a lot. In fact, the scientific method is used to find the best available explanation (the one that best explains all available observations) and also to prove false inferior explanations. The article below uses the scientific method to prove that the official 9/11 story is false and that controlled demolition of the Twin Towers is the only scientific (best) explanation available. All without appealing to a "conspiracy theory". Please use your method to show how the article uses faulty reasoning. If you can't, you should admit that it actually is the best explanation we have right now.

http://skeptopathy.com/wp/?p=334

David Whitehead said...

Occam's razor comes to mind in this argument.
Also, you could emphasise more how an explanation which leads to an increase in unexplainable claims is less likely to be the explanation of the original.
Agree re: WLC's assertion of 'fact'. Similar to lots of Christians who state 'laws' of the bible, such as prosperity,as though they are comparable to scientifically tested hypotheses.