9 Comments
User's avatar
Spencer's avatar

Ok… so I have a near death experience experience. I was 19 and got in a motorcycle accident and woke up in an ambulance. Ok, it was a motor scooter accident, but still it broke my collar bone and gave me brain damage. I would have died without a helmet on. Anyway, I have a vivid memory of seeing the EMTs in the ambulance from above. However, I fully realize this was a dream/hallucination. Still kind of funny.

I’m fine now- except I’ll never know how smart I COULD have been and, given how absolutely brilliant I am, it’s probably for the best I’m not any sharper.

Expand full comment
Stuart Ritchie's avatar

There was a study about this last year (I think?) where they tried to hide things above the hospital bed so they could only be seen by people floating outside their body... we absolutely must do it for the next Halloween special!

Expand full comment
Gordon Wells's avatar

This was a fun episode. You speak briefly about how immature science was in the 19 nth century compared to now. But most scientists have never been hyper-rational. In relation to this it's well worth listening to the podcasts Emmet Penney and his friend John have been doing on the theological origins of modernity and now on the esoteric origins of the Enlightenment:

https://nuclearbarbarians.substack.com/p/the-esoteric-origins-of-the-enlightenment

Expand full comment
Mark Kerr's avatar

You might be accused of double standards about marginal p-values. You criticise papers which talk about ‘trending towards significance’ but describe p value of 0.036 as not convincing. I agree that the arbitrary 0.05 as the dividing line between significant and not significant is a bit silly but if you want researchers to say 0.064 is not then you have to allow 0.036 is.

Expand full comment
Pam Param's avatar

I mean this is a straightforward Bayesian evaluation, isn't it? You're starting with a very high prior that psychic powers aren't real, and consequently a rule of thumb usually intended to determine whether a certain medicine works or not is hardly convincing.

Expand full comment
Samuel Lee's avatar

I don't think it's the same. The authors of the study chose to use a threshold of .05, not Stuart and Tom.

Anyways, it's more meaningful to think about the meaning of the p-value. It is more likely to see a p-value of .036 or less than it is to roll double sixes during your first turn of Monopoly. It's not an especially rare result.

Expand full comment
Mark Kerr's avatar

Fair point but if we are playing Monopoly and I tell you before the start ‘I am going to roll a double six on my first turn’, then do so, you would I think be impressed (or suspicious)

Expand full comment
Samuel Lee's avatar

But would you be impressed or suspicious if I said that before every game? Researchers set their significance level before every study.

Regardless, extraordinary claims like these require extraordinary evidence. No matter how much you like significance testing, these results are far from extraordinary.

Expand full comment
Mark Kerr's avatar

If the experiment is replicated and the same result obtained then it’s much stronger. To be clear I think communicating with the dead is nonsense but that’s because there is no theory being tested , just an assertion that if you can’t explain the result with normal physical processes then the alternative paranormal explanation must be true. As a Bayesian Tom will see the flaw in this.

Expand full comment