A probabilistic argument against deniers of the paranormal

[Laird]

I composed the first draft of this post a few weeks back, and, curiously, not long after (a few days), I saw David B make a similar albeit much briefer argument in the middle of a thread on Skeptiko... David, I haven't cribbed your notes!

Here, I am using "paranormal" as a shorthand for everything that a "skeptic" would deny: psi, UFOs, mediumship, etc etc.

So, here's the (very informally expressed) argument, which in the end is very simple, but I'll explain it via a narrative which hopefully isn't too tedious.

Let's say you show a denier of the paranormal a potential paranormal case - something very strange which is hard to explain - and ask him/her, "What do you think? Could that be a case of a genuine paranormal phenomenon?" - and s/he says, "Very probably not!" Let's say you then ask him/her to quantify that: "How probably not? Would you give it, say, odds of one in ten thousand of being a genuine paranormal phenomenon?"

Let's say the denier of the paranormal agrees: "Yes, odds something like one in ten thousand sound about right".

So far, so good.

Let's say you then come back to this particular denier of the paranormal with another potentially paranormal case, and s/he again agrees that the odds that it really is paranormal are only about one in ten thousand.

What are the odds that neither case is woo given the woo denier's odds for each? Simple probability tells us that those odds are 0.9999 (the odds of each against woo, 9,999 in ten thousand) to the power of 2 (because there are two cases), i.e. about 0.99980001. Thus, the probability that at least one of the two cases is genuine is about 0.00019999 (i.e. 1 - 0.99980001).

Not very hopeful for us proponents. But here's the interesting question: how many similar cases would we need to present to the denier of the paranormal for the odds that none of them are genuinely paranormal to drop (by his/her own admission) below 50%, i.e., for it to be more probable than not that at least one case is genuine, home-spun "woo"?

The answer is log 0.5 with a base of 0.9999, which, on rounding up to the nearest whole case, is 6,932.

Is it that hard to find 6,932 potentially paranormal cases, each of which has odds in favour that a reasonable denier of the paranormal would have to concede were at least one in ten thousand? I think alone on Skeptiko (and hopefully, eventually, this forum) we probably have considered and linked to a number of resources which comes close to that count, many of which an honest denier of the paranormal would have to give better odds than one in ten thousand for.

OK, but not all deniers of the paranormal are going to go that far. So, how about we consider odds of one million to one against: how many cases would we need for the probability that at least one of those million-to-one cases was genuine woo?

The answer rounded up to the nearest whole case is 693,147. Do that many cases, at odds of at least one million to one in favour of being genuine paranormal phenomenon, exist? Given that almost every person I've spoken to has had at least one "weird" thing happen to them, and that this seems to have been going for as long as humans have existed, I think they most certainly do, and that a genuine skeptic would have to admit as much.

But back to the deniers. What's my point, then? That a denier, to maintain his/her denial, cannot realistically give any reasonable odds that any individual case is genuine woo, because to do this consistently, given the number of reports of "strange" things out there, would ultimately undermine his/her position: s/he would be bound to admit that the probability of at least one genuine case of the paranormal is greater than 50%, even given his/her exceedingly low odds for individual cases. And once there is one genuine case, then the door is permanently open: goodbye blanket denial.

To maintain his/her denial, a denier effectively has to say: the odds of any one case being woo are near enough to zero that it's not worth quantifying them - and this is one part of what separates a denier from a genuine skeptic: a genuine skeptic will allow for some sort of reasonable odds that any given case is genuine.

One consequence, then, of this argument is that a genuine skeptic is almost compelled to grant that "genuine woo" probably exists.

This argument might also be one way of contextualising all of the "debunking" by that goes on: the mere possibility of anything paranormal cannot be allowed in any case, lest this principle (of allowing odds at all) be extended in general to all known cases, and upend via multiplied probabilities the plausibility of the denier's blanket denial.

Well, I hope that that wasn't all too tedious for you guys. I welcome your thoughts. Is this a reasonable argument? Does it have unintended consequences? Would it make any headway with a denier of the paranormal? How about with a genuine skeptic?

The main (probably obvious) objection I can see to it is a skeptic/denier saying, "Wait, I do not grant that such a large number of cases worthy of million-to-one odds exists: not until you show each one to me!" And now I'm wondering whether there is some creative way of using Google to count the number of potential cases of "woo"...

In any case: enough words from me - over to you guys.

The main (probably obvious) objection I can see to it is a skeptic/denier saying, "Wait, I do not grant that such a large number of cases worthy of million-to-one odds exists: not until you show each one to me!" And now I'm wondering whether there is some creative way of using Google to count the number of potential cases of "woo"...

In any case: enough words from me - over to you guys.

What about the proponent objecting to the arbitrary high odds against the story being true? How is that determined? It seems to me that debunkers rely on assumptions of deception and/or delusion to announce, with confidence, the probability of woo. Yet I've heard stories from people I consider to be of sound mind and character and would trust their account as being accurate and well considered. What odds do such stories - still anecdotal - deserve if we strip away the a priori assumptions?

I think one could point to two potential problems with this line of reasoning.

1. The probability calculation (i.e. P(A and B) = P(A) * P(B) ) only works if the two cases are (stochastically) independent, it doesn't hold in general.

Say, for instance, that the two events in question are that you and I each ask a psychic how to place a bet at the horse race. But unbeknownst to eachother we have actually picked the same medium. Case A would be that you win, and case B would that I win. But since we have the same source for the information in this case clearly P(A and B) = P(A) = P(B).

In your example it wouldn't be a case of just choosing 6932 cases: one would also have to compare them pairwise and make sure they are independent. And how does one determine whether there is some underlying common causality between to paranormal events (a butterfly's wing etc.)?

2. The argument also breaks down if the skeptic chooses to assing the probability 0 to all paranormal events. I'm sure you know this, so just to be clear: For something to have probability 0 of occuring doesn't imply that it cannot occur.

Yes, I agree with lhl. I think the main problem with the argument is that it requires all the cases to be statistically independent. You don't just need a large number of anomalous cases, you need a large number of anomalous cases that are sufficiently different that they can't be explained by common underlying causes. For example, if one "apparition" turned out to be caused by a disease of the retina, then that would affect the probabilities for other similar apparitions. If one "ufo" was demonstrated to be the planet Venus, then that might dispose of a large number of other "ufos".

[Laird]

What about the proponent objecting to the arbitrary high odds against the story being true? How is that determined? It seems to me that debunkers rely on assumptions of deception and/or delusion to announce, with confidence, the probability of woo. Yet I've heard stories from people I consider to be of sound mind and character and would trust their account as being accurate and well considered. What odds do such stories - still anecdotal - deserve if we strip away the a priori assumptions?

Good questions... the argument I tried to make assumed that we accepted the unreasonably bad odds that deniers would give and show that even then they are bound to accept the likelihood of genuine instances of the paranormal. But yes, if they were willing to be reasonable and offer reasonable odds, then the argument is even stronger (albeit that others have raised reasonable criticisms of it).

[Laird]

Chris and lhl: thank you for your criticism - I accept that we would also need to prove that all the cases were independent... oh well, it had seemed like a good argument for a while there, I'm glad I ran it by you guys before trying to use it "in the field".

lhl, re the skeptic assigning a probability of zero: yes, that was a point I tried to make - that to avoid the cogency of this argument (before you guys had shown me that it is not, after all, so cogent), a "skeptic" would pretty much have to set the odds to zero.

Thanks all for responding!

[Laird]

2. The argument also breaks down if the skeptic chooses to assing the probability 0 to all paranormal events. I'm sure you know this, so just to be clear:  For something to have probability 0 of occuring doesn't imply that it cannot occur.

Revisiting this thread, I see that I fluffed my response to this, in particular the part I've bolded, which I seemed to obliviously ignore. Yes, I am (and was) aware of that - and it's explained nicely in this blog post which I've just found for those who weren't already aware and are interested: https://nolaymanleftbehind.wordpress.com...obability/

That said, I am not really sure how it applies or could apply to paranormal events in general. Can you explain a little more your thinking behind this?

Also, on reflection, I think there is a better response than simple acceptance/concession to the counter-argument from a need for stochastic independence which both you and Chris raised:

I think that this in itself can be reduced a probabilistic argument: often, we don't know whether or not two paranormal events might be stochastically dependent, but they seem not to be, and so we can assign a probability to them being independent. I think that with so many varieties of paranormal events, many of these probabilities would themselves be very high, such that, in the end, my argument still works. That's quite roughly expressed, but I hope is understandable. We can explore in more detail if/as necessary.

Thoughts?

The issue I have with the op is that the odds against given by the skeptic appear to be based on a first reaction assuming that the person relating the incident is being accurate and complete. If this skeptic investigated the claim, I expect there would be grounds for adjusting his guess either up or down based on more evidence (since we're talking about a skeptic, probably down). Are there 6,000+ incidents which are immune to this? I genuinely don't think there is.