Monday, August 08, 2016

Scientific Study Shows Mediums Are Wrong 46.2% of the Time

Not a very good showing, eh?


In the study,
“Participants were asked to press a button if they thought the person in a photo was living or deceased. Overall mean accuracy on this task was 53.8%, where 50% was expected by chance (p < 0.004, two-tail). Statistically significant accuracy was independently obtained in 5 of the 12 participants.”

The abstract claims the participants showed better than chance performance, but even if we accept this level of accuracy at face value (so to speak), the mediums were wrong 46.2% of the time. Remember that before your next psychic reading.

And of course we should not accept the results at face value. Let's take a closer look at the paper (Delorme et al., 2016), which was published in Frontiers in Human Neuroscience



Actually, let's take a closer look at the authors first. Arnaud Delorme, Alan Pierce, Leena Michel,  and Dean Radin are all affiliated with the Institute of Noetic Sciences (IONS), a parapsychology research institute in California. Dr. Delorme is also affiliated with UC San Diego. Along with Scott Makeig, he developed EEGLAB, a Matlab toolbox that's widely used to analyze EEG data. Delorme and Makeig (2004) has been cited 5738 times (as of this writing).

Why is Delorme doing parapsychology research?? He's a long-time Zen meditator, according to his IONS biography. Why is Frontiers publishing parapsychology research? Here's one opinion.


Dead or Alive?


Figure 1 (Delorme et al., 2016). Process involved in creating a group of photographs of “Alive” and “Deceased” individuals.


Photographs of known alive and dead people were selected from three internet databases: (D1) school portraits from 1939–1941; (D2) school portraits from 1962–1968; and (D3) politicians (US senators excluded) and businessmen. Why? Why use pictures of US Representatives and state politicians outside of California? Even though the subjects said they didn't recognize them, there could be a vague sense of familiarity with some of these faces.

Photos of 404 individuals were presented, and the 12 participants pressed keys to indicate “deceased,” “living,” or “do not know”. 1

The participants all “claimed to be able to experience feelings of vitality from facial photographs alone. ... They were required to have been performing professional ‘readings’ for clients...” THERE WAS NO CONTROL GROUP.  In other words, participants who did not claim any psychic or clairvoyant abilities were not included in this study. Thus, there was no way to know if the marginal ability to discern whether a person was alive or dead was based on mediumship.

And marginal it was. Basically, they were terrible at determining whether people in old yearbook photos were dead or alive. Terrible. No better than guessing. 2




Given the number of statistical tests, we should only consider values with *** (p<.001), of which there were two (out of 35 possible comparisons). Therefore, the evidence for mortality prediction (clairvoyance) should not be taken seriously, despite the authors' conclusion:
We do not rule out the hypothesis that subjects might have had access to information in ways that are not currently understood by modern physics and could potentially go beyond classical information delivered by facial features.

Paranormal physics do not apply to old photographs, however.

And the EEG data were equally unconvincing. The face-specific N170 component did not differ based on dead or alive, correct or incorrect. The earlier P1 component showed a small difference between correct and incorrect responses for the deceased only, but there was no good explanation for this (“Future research could assess if low-level visual image characteristics and attentional modulation were important factors in leading to this difference in electrocortical activity”).

The truth is out there, but this study provides no proof that the ‪#‎Supernatural‬ actually exists.


Footnote

The “do not know” responses were not included in the analyses, and we have no idea of how many such responses were recorded.

2 Oh here's a fun fact. S06 indicated that 90% of the people in the photos were dead.


Reference

Delorme, A., Pierce, A., Michel, L., & Radin, D. (2016). Prediction of Mortality Based on Facial Characteristics. Frontiers in Human Neuroscience, 10.  DOI: 10.3389/fnhum.2016.00173




Subscribe to Post Comments [Atom]

6 Comments:

At August 09, 2016 6:32 AM, Blogger Neuroskeptic said...

Assuming it is possible to predict current life or death from old photos, the most obvious explanation is that people are able to detect health status or other features which predict future health, from the photos.

The authors do acknowledge this, but they say:

"However, given the counterbalanced design of the photo databases and removal of obvious clues such as skin color (Fink et al., 2006), an adequate explanation may rest upon subtle clues that might have been unconsciously exploited."

Hmm...

 
At August 09, 2016 1:57 PM, Blogger The Neurocritic said...

The authors are trying to have it both ways. As @Foreman1David pointed out, the hypothesis that 'faces can express health-relevant social information' is actually interesting (and investigating this was one of the authors' original goals). But why start out with mediums? Why not test "ordinary" people? Then compare mediums, "ordinary" people, and skeptics on a stimulus set of unknown people, male and female, of many races and ages?

 
At October 03, 2016 3:42 AM, Blogger Neuroskeptic said...

This paper has just been retracted (the authors did not agree to the retraction.)

 
At November 08, 2016 2:53 AM, Blogger Smut Clyde said...

I left a comment at Retractionwatch:

The stimuli were
1. 108 old photographs, of young individuals (1939-1941 school yearbooks), mostly dead (actual live:dead ratio not given, but we are told that “the ratio between alive and deceased individuals matched statistics of average life expectancy in the US for a given age group”).

2. 126 not-so-old photographs, also young individuals (school yearbooks from 1962-1968), maybe 50:50 alive.

3. 160 recent photographs, older individuals, mostly alive, possibly familiar (“photos of state politicians for about two-thirds of the images, as well as from photos accompanying obituaries of businessmen”).

So if the individual is not school age, it’s a recent photograph of someone who’s probably alive.

 
At November 15, 2016 9:53 AM, Blogger all sorts of errors said...

I am not at all sure that even the correct analytic method was used. Did the participants KNOW that half of the people in the pictures were dead and half were alive? If so, then the sign test is not the appropriate test to use. (See Fisher's tea tasting lady. Eight cups of tea, 4 are milk first, 4 are tea first, and the lady is aware of this setup. The outcome probabilities are not binomial because the trials are not independent. Once she has classified four tea first or 4 milk first, the remaining tea cups will be classified as the other. Same with half of the pictures. Once N/2 photographs are classified, the classification of the remaining photographs is determined.) On the other hand, if participants are told that photographs are selected from a bank of photos in which half of the photos are of people who are alive and half are of people who are not alive so that p(alive) = .5 for any given photo, THEN the binomial distribution is the appropriate sampling distribution.

 
At March 22, 2019 10:43 AM, Anonymous Anonymous said...

The comment above would be true if the participants were given feedback which was not the case.

Ih the original comment “Given the number of statistical tests, we should only consider values with *** (p<.001), of which there were two (out of 35 possible comparisons). Therefore, the evidence for mortality prediction (clairvoyance) should not be taken seriously, despite the authors' conclusion:” This comment appear contradictory. Even after accepting a stringent threshold, some results are still significant, and the conclusion of the blogger is “therefore it is not significant.”

 

Post a Comment

<< Home

eXTReMe Tracker