evidence?

[davidsmith73]

http://www.scientificexploration.org/jse/articles/pdf/16.2_mcdonough_don_warren.pdf

[Dr Adequate]

Looks more like a completely blank grey screen to me.

Perhaps you could quote the part you found impressive.

[Paul C. Anagnostopoulos]

The abstract:

Abstract—Event-related brain potentials (ERPs) were recorded from 20 subjects
performing a computerized, forced-choice guessing task. On each of 40
trials, ERPs were elicited by digitized images of 4 playing cards, sequentially
presented on a video monitor for 150 ms. After the last card was presented,
subjects guessed which of the 4 cards would be the target for that trial. Following
the subject’s guess, the computer randomly selected one of the 4 cards
to be the target and presented this as feedback; the remaining 3 cards served as
nontarget decoys for the trial. We found that a negative Slow Wave measured
at 150–500 ms post-stimulus had greater amplitude when elicited by targets
than when elicited by nontarget decoys (p <= .05). This result indicates an apparent
communications anomaly because no viable conventional explanation
of the ERP differential could be identified. It is the fourth study in our laboratory
employing essentially the same design to yield this or a similar ERP effect.

[CFLarsen]

davidsmith73,

What do you think?

[The Don]

A few points:

- What were the results of the peer review process ?
- Could you please provide links to studies replicating these results
- What follow up evidence has been posted in the last two years ?
- Where are the Nobel Prizes ?

[TheBoyPaj]

As I understand it...

They see four cards. They make a guess. The target is then chosen (post guess) and displayed.

The subjects ability to consciously select the target cards before it is chosen was not significantly different from chance. However, they noted a difference in their physiological response when the target card was being initially shown, suggesting an unconscious precognition.

The raw data is not displayed, but I have a few questions. The experimenters decided to limit their analysis of the results such that the sample set contained an equal number of tests where the conscious guess was correct and tests where it was incorrect. They also compared the target physiological response with just one of the three possible non-target responses, picked randomly.

I'm not a biologist, but is this ERP related to the measurable emotion response credited with use in lie detection? So might it be reasnable to assume that a subject might subconsciously "choose" which card they will later select as the cards are being initially shown? And might this be registered on the ERP? If this is so, you might expect the ERP to correlate more closely with the correctly guessed trials. This is what they claim, but I'm struggling to see what's paranormal about it.

[davidsmith73]

Originally posted by CFLarsen
davidsmith73,

What do you think?

Still reading it in detail

[CFLarsen]

Originally posted by davidsmith73
Still reading it in detail

Don't you think it would speed up the discussion, if you had read it first, and then told us what you think?

[davidsmith73]

Originally posted by CFLarsen
Don't you think it would speed up the discussion, if you had read it first, and then told us what you think?

Not really. What's the rush man?

[Interesting Ian]

Originally posted by Dr Adequate
Looks more like a completely blank grey screen to me.

Perhaps you could quote the part you found impressive.

I can't get anything either :(

[davidsmith73]

Originally posted by TheBoyPaj
As I understand it...

They see four cards. They make a guess. The target is then chosen (post guess) and displayed.

The subjects ability to consciously select the target cards before it is chosen was not significantly different from chance. However, they noted a difference in their physiological response when the target card was being initially shown, suggesting an unconscious precognition.

The raw data is not displayed, but I have a few questions. The experimenters decided to limit their analysis of the results such that the sample set contained an equal number of tests where the conscious guess was correct and tests where it was incorrect. They also compared the target physiological response with just one of the three possible non-target responses, picked randomly.

I'm not a biologist, but is this ERP related to the measurable emotion response credited with use in lie detection? So might it be reasnable to assume that a subject might subconsciously "choose" which card they will later select as the cards are being initially shown? And might this be registered on the ERP? If this is so, you might expect the ERP to correlate more closely with the correctly guessed trials. This is what they claim, but I'm struggling to see what's paranormal about it.

The subconscious choosing is definately there. In fact they did a couple of studies that were aimed at showing this specific effect. However, they say that because equal numbers of correct guess and incorrect guess trials were included in the analysis (as you pointed out), this means that the subconscious "choice" ERP cannot produce an artifact. Is that correct?

[davidsmith73]

Originally posted by Interesting Ian
I can't get anything either :(

try going to the Society for Scientific Exploration main site, go to the journal page, select abstracts and articles. This paper is Volume 16, Number 2 (Summer 2002)

[Interesting Ian]

Originally posted by TheBoyPaj


I'm not a biologist, but is this ERP related to the measurable emotion response credited with use in lie detection? So might it be reasnable to assume that a subject might subconsciously "choose" which card they will later select as the cards are being initially shown? And might this be registered on the ERP? If this is so, you might expect the ERP to correlate more closely with the correctly guessed trials. This is what they claim, but I'm struggling to see what's paranormal about it. [/B]

Ummm . . you're confusing me. Oh what the hell, I'll read the article first before commenting. Just managed to get it.

[Ed]

I have done a bit of work with Evoked Potentials collected from surface electrodes. Unless things have changed remarkably the pretty picture that textbooks contain is the result of averageing many, many trials. This is because the signal is buried in noise and ongoing EEg all of which is random with respect to the EP. So you average and all of the crap goes away. With me?

OK. Now we have an average that is compared to something. I submit that if you put error bands around each epoch, that is the variability of the data that was averaged, you will realize that there is bupkus. That is to say that the inherent variability of the EP is greater than the EP itself.

Is that where you came out David? Incidentially, this took about 2 minutes of reading.

[CFLarsen]

Originally posted by davidsmith73
Not really. What's the rush man?

There's no "rush". It just seems to me that you would rather discuss what others think of it.

Read it, and get back.

[davidsmith73]

Originally posted by Ed

OK. Now we have an average that is compared to something. I submit that if you put error bands around each epoch, that is the variability of the data that was averaged, you will realize that there is bupkus. That is to say that the inherent variability of the EP is greater than the EP itself.

Is that where you came out David? Incidentially, this took about 2 minutes of reading.

But they used ANOVA for calculating the significance of any differences between ERP's. Does this address your point?

[Pragmatist]

Originally posted by Dr Adequate
Looks more like a completely blank grey screen to me.

Perhaps you could quote the part you found impressive.

Oh you foolish, evil skeptic can you not see the invisible pink unicorn! Sheesh...some people...

[Pragmatist]

Originally posted by Interesting Ian
I can't get anything either :(

We already noticed that! :D

[davidsmith73]

Originally posted by CFLarsen
There's no "rush". It just seems to me that you would rather discuss what others think of it.


Yes I would like to discuss what others think of it

What do you think of it Claus?

[Dr Adequate]

Originally posted by Pragmatist
Oh you foolish, evil skeptic can you not see the invisible pink unicorn! Sheesh...some people...
Yeah... that clanging noise you just heard was me closing my mind.

Have you noticed that I've PM'd you?

[Interesting Ian]

Originally posted by Pragmatist
We already noticed that! :D

This is a different thread. You're supposed to have me on "ignore".

[c4ts]

Oops, wrong thread.

[Diogenes]

Does the subconcious choosing happen among subjects, who do not know what the target choices are going to be, before they are shown them?

[Ed]

Originally posted by davidsmith73
But they used ANOVA for calculating the significance of any differences between ERP's. Does this address your point?


No, that just provides statistical goobledygook. As far as I can tell they averaged the results for each of 20 subjects and then compared those averages, viz:

For each subject, averaged ERPs were formed by calculating the mean of all artifact-free EEG epochs in the target and nontarget category in order to enhance the underlying
electrical brain waveform.

Italics mine. What they did was throw out the within subject variability to get something clean. Put it this way: Suppose a country has two people, one making $100,000 and one making nothing. Would the average ($50,000) really characterize the income of the population? No, even though an average is an acceptable measure of central tendency. You really cannot throw away data in order to get significance.

I am a bit rusty but I probably would have looked at a trials by subject design of some sort. That is to say, given the variation within each subject (remember those hard-to-see evoked potentials without averaging) are any differences across subjects significant.

I suggest that this experiment is fatally flawed and provides no evidence whatsoever.

I will say that I may have misread and am wrong, but if in fact they compared a number of averages to each other it is garbage and that what they "found" is no more than unsupported conjecture.

[TheBoyPaj]

Originally posted by davidsmith73
The subconscious choosing is definately there. In fact they did a couple of studies that were aimed at showing this specific effect. However, they say that because equal numbers of correct guess and incorrect guess trials were included in the analysis (as you pointed out), this means that the subconscious "choice" ERP cannot produce an artifact. Is that correct?

But what effect are they claiming? I really cannot tell from this description.

If they are saying "people have an unconscious reaction to a card which they will later choose" I'd have to say "duh, yeah!"

If they are saying "people have an unconscious reaction to a card which will become a target card" then I would have to see their data to confirm this.

You see, my basic point is this. I think that it's possible that people have a little reaction when they first see the card they will subsequently choose. That people actually make their choice on first seeing the card and that ERP may register this impulse.

Now, they preselected data so that 50% of the sample was from trials where the subsequent choice turned out to be correct. So why should we be surprised if the unconscious choice turned out to be right too? In fact, if this little impulse is a reliable thing, they could be setting themselves up for a 50% success rate or more.

What we need to see is the correlation between unconscious and conscious choices, to see if my guess is correct. Like I said, I'm not a biologist so this could be hogwash, but this report doesn't clarify it for me. For all I know the answer might be buried in there, but all that hemisphere stuff seems to cloud the issue.

[IXP]

This looks like the barn wall effect (shoot at the blank wall, then draw the target afterwards).

It looks like they gathered all this data for the trials, then post hoc, determined how to massage it to give a positive result.

I would suggest that they re-do this experiment, but this time, they analyze the data before the random choice is made and predict from their analysis which card will be chosen randomly. If they are correct, they will be able to do better than chance at predicting the chosen card.

[Ed]

Originally posted by IXP
This looks like the barn wall effect (shoot at the blank wall, then draw the target afterwards).

It looks like they gathered all this data for the trials, then post hoc, determined how to massage it to give a positive result.

I would suggest that they re-do this experiment, but this time, they analyze the data before the random choice is made and predict from their analysis which card will be chosen randomly. If they are correct, they will be able to do better than chance at predicting the chosen card.

No. Actually it looks like a willfull misapplication of statistics. Before they spend any more precious ERGS on new experiments, I would suggest that they reanalyse the data.

[Paul C. Anagnostopoulos]

Presumably, the guessing-related ERP effect could
also produce artifactual differences between target and nontarget ERPs whenever
these stimulus categories have unequal numbers of guessed stimuli. However,
guessing-related effects cannot readily explain the presently observed
differences between target and nontarget ERPs because the numbers of
guessed stimuli in the target and nontarget categories were balanced.
Ignoring Ed's issue for a moment, could someone explain to me how they think they've controlled for the guessing-related effect by balancing the numbers? I'm not getting it.

~~ Paul

[davidsmith73]

Originally posted by Diogenes
Does the subconcious choosing happen among subjects, who do not know what the target choices are going to be, before they are shown them?

Yes, for each trial it happens when they are first shown the sequence of 4 cards, one after the other, without the option to register their guess. When a particular card appears on the screen they will subconsciously decide to choose it at the later stage (regardless of whether its the correct target card). This will cause an ERP.

[Vim Razz]

Originally posted by Dr Adequate
Looks more like a completely blank grey screen to me. It's a .PDF file. Got Adobe (http://www.adobe.com/products/acrobat/readstep2.html)?

(That is, It'll show up grey without the proper plug-ins on your browser.)

[Vim Razz]

Originally posted by TheBoyPaj
But what effect are they claiming? I really cannot tell from this description. Pre-cog. Fortelling the future.

IMHO, they didn't find anything unusual. (what they were looking for, for example) Their statement in the conclusion that the next step is to formulate a theory to explain the mechanism at work is jumping the gun a bit, kind of throwing the whole paper into perspective: the potential for confirmation-bias is just too strong here for 2% over chance to be convincing.

[Ed]

Excuse me people. They don't have an experiment in the first placeand we are actually discussing follow-on work? We are actually discussing their assumptions?

Please.

Amherst, if you are reading this, you are seeing the problem with paranormal research in microcosm. The words and techniques, in the present case, are misdirection applied so that you can't see the magition work his wiles with the stats the result of said analysis justifying all the blather but there is no justification.

[Diogenes]

Originally posted by davidsmith73
Yes, for each trial it happens when they are first shown the sequence of 4 cards, one after the other, without the option to register their guess. When a particular card appears on the screen they will subconsciously decide to choose it at the later stage (regardless of whether its the correct target card). This will cause an ERP.

That wasn't exactly my point. If they have precog, why can't they guess what card is going to be shown without seeing any cards at all?


Four cards don't really leave much room for deviation. Why don't these tests use 50 or a 100 different symbols? If the effect exists, the number of targets shouldn't matter.

[The Mighty Thor]

Let's say I pick my favourite meal from a displayed selection of four meals.

Then later, when the same selection is offered, I pick my favourite meal again? Is that precognition?

Most people have a "favourite card" -- magicians know that a fair percentage have the same favourite card and can play the odds at guessing what it is.

I'd say, "Ace of Spades" for most guys and "Queen of Hearts" for most girls, but I can't remember if this is indeed the case.

[davidsmith73]

Originally posted by Ed
No. Actually it looks like a willfull misapplication of statistics. Before they spend any more precious ERGS on new experiments, I would suggest that they reanalyse the data.

Isn't the comparison of ERP's used in conventional brain research? I've already found one paper that analyses differences in ERP's in response to auditory stimuli, published in the journal Neuroscience (a highly respected journal). I've included the data analysis methods.

------------------------------------------------------------------------
Blockade of epinephrine priming of the cerebral auditory evoked response by cortical cholinergic deafferentation
Neuroscience, Volume 116, Issue 1, 15 January 2003, Pages 179-186


Abstract
The present study tested hypotheses derived from a neurobehavioral model of anxiety that posits an important role of the basal forebrain cholinergic system in the cortical processing of anxiety-associated stimuli and contexts. We hypothesized that visceral afferent activity induced by systemic administration of epinephrine would enhance the processing of auditory stimuli as evidenced by the cerebral auditory evoked response. We further predicted that selective lesions of the basal forebrain cortical cholinergic projection system would disrupt this processing, and would further block the effects of epinephrine. Results confirmed these hypotheses. Epinephrine was found to enhance the amplitude of the P70 component of the auditory evoked response in rats. Selective lesions of the basal forebrain corticopetal cholinergic projection, by intrabasalis infusions of 192 IgG saporin, delayed and reduced the amplitude of the P70 component, and blocked the potentiating effects of epinephrine on the auditory evoked response.

The present results are consistent with the view that visceral afferent input may modulate cortical processing of sensory signals via the basal forebrain cholinergic system. These considerations emphasize the potential importance of ascending, bottom-up modulation of processing by telencephalic circuits that may impact on a wide range of behavioral functions.


Data analysis
EEG epochs with unstable prestimulus baselines or movement artifacts (greater than ±3 S.D. from the mean) were eliminated from analysis (<8% of traces), and the remaining traces were averaged for each session and intensity. The amplitude and latency of the major negative and positive peaks were then derived for each condition. Preliminary analyses revealed no significant effect of day of test for the two sessions under the same drug condition. Consequently, data from these replicate sessions were collapsed for formal analysis, yielding a between-within design having one between-subjects factor (group: lesion vs. control) and two within-subjects factors (drug: epinephrine or saline; and intensity: 60, 70 and 80 dB). Primary analysis was by analysis of variance (SPSS), with a trends analysis on the three levels of stimulus intensity.

-------------------------------------------------------------------------------


I'm not saying you are wrong Ed, but I would have expected such a respected journal as Neuroscience to reject such a paper if what you're saying is true. Do you find anything wrong with the methods above?

[Vim Razz]

Diogenes
If they have precog, why can't they guess what card is going to be shown without seeing any cards at all?The subjects weren't selected on the basis of claims of paranormal ability -- only a particular interest in gambling.

The authors are sugesting that there are precognative abilities which are unconcious / undeveloped.

Paul C. Anagnostopoulos:
Ignoring Ed's issue for a moment, could someone explain to me how they think they've controlled for the guessing-related effect by balancing the numbers? I'm not getting it.Since the target pool contains XX% guesses (acurate guesses), the non-target pool also contains XX% guesses (wrong guesses). By this logic, "guess-related" effects will show up ~equally on both curves and not contribute to the deviations between the two sets that they're looking for.

TheBoyPaj's coments sugest that data was filtered down to the point where XX% = 50%, though I haven't been able to find reference to that myself.

[TheBoyPaj]

Originally posted by Vim Razz
TheBoyPaj's coments sugest that data was filtered down to the point where XX% = 50%, though I haven't been able to find reference to that myself.

From footnote 2:

"After forming the initial nontarget lists, in order to balance the number of guessed stimuli between the target and nontarget categories for each subject, the initial lists were modified by deleting randomly selected target or nontarget epochs from each subject, as needed, to match the number of guessed stimuli in these two categories."

[Vim Razz]

Originally posted by TheBoyPaj
From footnote 2:

"After forming the initial nontarget lists, in order to balance the number of guessed stimuli between the target and nontarget categories for each subject, the initial lists were modified by deleting randomly selected target or nontarget epochs from each subject, as needed, to match the number of guessed stimuli in these two categories." Gotcha. I'd interpreted their procedure diferently, looks like. (as per my reply to Paul)

TY for taking the time to dig that out, btw -- in fact I feel kinda bad if I made you slog through that whole paper again. ;)

[TheBoyPaj]

S'okay. I have a good memory for that kind of thing so I could jump straight to it.

It's not the easiest paper to read though, is it? More than once I was heard to shout "get to the point!".

[Vim Razz]

It's a bit of a bear, I agree, but at least it apears they made an effort to present what they think they've found honestly, which helps. So it's not a Grizzly, really -- more like a clingy koala. ;)

Something occured to me a few days ago about the 27% guesing acuracy in this setup that I'd decided to drop at the time because:

a) I was too lazy to go back and type it all out :p, and
b) they never claimed that 27% was statisticly significant, so there didn't seem to be any point in pursuing the mater.

But, glancing around a few other threads I think I'll lay it down anyway, as it seems to be relevent not just here but in the popular Ganzfield-type setups as well.


They note that they didn't bother analyzing serial randomization of target-card selection because they felt that trail sets (20-40 trils per subject) were too short for the subjects to pick up any accidental paterns in target-location. (footnote 1)

What concerns me is that many psudo-random proceses can have a tendency not to pick the same card twice in a row as often as one might expect by pure chance. There's also a tendency among people to (naively) assume that random numbers dont repeat much. Puting these to factors together -- and being unable to study the raw data to see whether it may be an issue -- there's a good posibility for better-than-25% guessing acuracy to be staticly likely.

For example, if one were to take an extreme case where values never repeat serialy and people unconciously asumed that they wouldn't, then the chances of guessing the right card (in trials after the first) would be 33% rather than 25%. They have a 1 in 3 chance rather than a 1 in 4 chance because they can effectively preselect form the cards which were not the target last time, giving them better odds.

Asuming more realisticly that values repeat ocasionaly, but perhaps not "enough", so to speak, one would expect anywhere from 25%-30% accuracy by chance alone. Without a good analysis of serial randomization within the test set it's impossible to determine where the expected accuracy really should fall within that range.

So, without better information, it's impossible to say that 27% is better than chance at all -- forget about determining whether it's a significant departure or not.


Just had to get that off my chest so I could stop thinkin' about it. ;)