Wowbagger Reviews a Pro-Parapsychology Book

[Wowbagger]

Limbo and Damien: You must either put up, or shut up! Give us a testable hypothesis that could allow your studies to yield more reliably repeatable results, or quit claiming psi is ready for prime time.

Otherwise, your complaints are just petty.

"Just as unreliable" as what? As chance?

What I meant was that they cannot be reliably replicated, under tightly controlled conditions. Chance has nothing to do with that comment.

He neglects to note that what the participants were aiming at was an aesthetic gestalt, not just a random scatter.

Do you really think it helps your case to mention this?

"Aesthetic gestalt"? Is that the best psi can hope for?

Could you imagine if astronomy worked the same way? "Now, simply point your telescopes in any spot you think might be an aesthetically pleasing direction, and note its star patterns. Repeat 180 times. If a few of the collected star patterns are still aesthetically pleasing, that would be a significant find!!!"

Or biology? "Take some photos of giraffes. Blindly grab two of the photos, while mentally focusing on aesthetic appeal. Note the patterns on its skin. Repeat 180 times..."

and this can be analyzed statistically (as was done).

And, even after all that, they still don't have a theory. Why is that?

but in my estimation you're operating in debunk mode when you should be operating in grok mode.

Damien is claiming that it is time for parapsychology to be considered a real science. I was operating in "Science Mode", when I wrote that review.

It is okay if you want to believe psi is a real phenomena. But, I already outlined what it would take for mainstream science to accept it.

This is not about why I, personally, don't accept the claims. This is about how science examines claims.

Put your mind in neutral, as it were.

I already did. I found the evidence from accepted psychology, physics, and mathematics made a more compelling case than the evidence from psi.

Fear of Psi

Why would I fear something that even Damien admits is "unreliable, puckishly evasive"*? What powers does psi have over me?

(* page 288)

I, for one, would welcome the day that psi-GPS replaces electronic GPS; and psychics can reliably find missing people; or even the day I can watch as all the gambling casinos close up shop. But, I am not holding my breath.

Now, stop wasting our time with ad-homs, and get to work on those testable hypotheses!

 

[Ersby]

Was Ertel's paper published in full? I'd like to see that.

 

[fls]

If Ertel did a statistical analysis, then he must have subjected randomly generated and experimental matrixes to blinded judgements as to "aesthetic gestalt". Evidence for psi aside, empirical measurements of the extent to which humans find "aesthetic gestalt" in random patterns would be very useful information to have. Is the control data discussed in the book or does Ertel have a published paper on this?

Linda

 

[Wowbagger]

Was Ertel's paper published in full? I'd like to see that.

Is the control data discussed in the book or does Ertel have a published paper on this?

I, personally, don't know where Ertel's paper, on that experiment, can be found. The closest I was able to get was the PDF already provided by Limbo. But, that was a different (if somewhat similar) experiment.

The only controls the book talks about, (as far as I recall), were ball-drawing simulations on a computer. He had a computer randomly pull "balls" from two virtual "bags", to see if it could come up with nifty patterns, itself. The book claims the computer could not.

If Ertel did a statistical analysis, then he must have subjected randomly generated and experimental matrixes to blinded judgements as to "aesthetic gestalt". Evidence for psi aside, empirical measurements of the extent to which humans find "aesthetic gestalt" in random patterns would be very useful information to have.

Yes, it would be very helpful to establish "aesthetic gestalt", empirically, in that manner, before you go off searching for it in the results of psi.

I doubt they did that. But, I could be wrong.

 

[fls]

The only controls the book talks about, (as far as I recall), were ball-drawing simulations on a computer. He had a computer randomly pull "balls" from two virtual "bags", to see if it could come up with nifty patterns, itself. The book claims the computer could not.

That's the research that I am interested in. In order to come to that conclusion, a set of blinded judgements as to aesthetic gestalt would have to be made under situations where aesthetic results would be expected. The details of the result of that experiment would be very informative and interesting.

If (disappointingly) that information is not available, it could serve as a test as to whether or not to trust the opinions of the author or of Ertel. Since this is absolutely necessary information to have - that is, the whole claim that the human generated 'patterns' are evidence of psi depends upon knowing what happens when the same judgements are applied to computer generated squares - if neither of them have that information, it demonstrates that their claims are unfounded. I hope that is not the case.

Linda

 

[Beth]

That's the research that I am interested in. In order to come to that conclusion, a set of blinded judgements as to aesthetic gestalt would have to be made under situations where aesthetic results would be expected. The details of the result of that experiment would be very informative and interesting.

If (disappointingly) that information is not available, it could serve as a test as to whether or not to trust the opinions of the author or of Ertel. Since this is absolutely necessary information to have - that is, the whole claim that the human generated 'patterns' are evidence of psi depends upon knowing what happens when the same judgements are applied to computer generated squares - if neither of them have that information, it demonstrates that their claims are unfounded. I hope that is not the case.

Linda

Linda's right. I agree with her. I would also be interested to see that information if it is available.

 

[Wowbagger]

That's the research that I am interested in. In order to come to that conclusion, a set of blinded judgements as to aesthetic gestalt would have to be made under situations where aesthetic results would be expected. The details of the result of that experiment would be very informative and interesting.

Perhaps it would help the psi proponents, on this forum, if you could outline a possible protocol to acquire those results.

 

[fls]

Perhaps it would help the psi proponents, on this forum, if you could outline a possible protocol to acquire those results.

Make 1000 computer-generated matrices. Explain the ball-drawing and matrix-generating procedure to your subject and tell them that the matrices they are looking at were formed as part of an experiment to determine the extent to which psi can influence the ball-drawing so that aesthetically interesting patterns will form. It is important to give them the impression that these matrices represent the possibility of the influence of psi.

Have your subject rate them on a 3-point scale as to aesthetics - "very aesthetically interesting", "somewhat aesthetically interesting", "not aesthetically interesting". Start by giving them some examples of "very aesthetically interesting" and "not aesthetically interesting". Summarize the results.

Present 10% of the patterns twice to test intra-rater reliability.

Repeat the experiment with other subjects to test inter-rater reliability.

For the purposes of the experiments described by Ertel. Equivalent numbers of computer-generated and human-generated matrices would be formed and presented randomly to the subjects. The subjects would be given the impression that all matrices had been formed by humans. Anyone present with the subject before or during the testing must be blind as to the source of the matrices.

Linda

 

[Jeff Corey]

o.o...o....o...o.
...oo.o.o.o..oo.
o...o.o.o.o....o
..o...o....o......
versus
.ooo.o.o...o.o..
o.o...o.o.o..o.o
oo...o.o..o.o...
..o..o..o.o.o....

Good luck on getting any agreement on which of those is prettier.

 

[Wowbagger]

Make 1000 computer-generated matrices...
(snip)

That's a good contribution!

I don't know if Ertel ever did anything like that. I doubt it. But, I could be wrong.

 

[Wowbagger]

I should be able to post my revised review in the morning (the database hasn't granted me access all weekend!)

But, I did upload some new stuff for it.

This is the "nice pattern" image I originally had in the review:

This is the image I was accused of "carefully omitting", in post #3:

This image will be added along side the old one, in the review.

Now, If anyone can figure out what all the fuss was about, please let me know.

 

[Ersby]

Is Ertel saying that "aesthetic qualities" are the same as symmetry?

This really is a case where no conclusion can be drawn without seeing the paper written up in full.

 

[Wowbagger]

Revised Review Now Published!

A new version of my review is now available on my site. The URL is exactly the same as the old one:
http://www.mitchlampert.net/News.aspx?R=191

However, since most of the comments above apply only to the previous version, I thought it would be fair to make it available for comparison. The previous version can be found here: http://www.mitchlampert.net/News.aspx?R=192 . (yes, it has an ID number above the newer version, since I decided to keep the new version on the old URL.)

Here is a list of changes:
* The description of Ertel's experiment has been improved, based on feedback from the author, and others.

* A second "nice pattern" image was added to Ertel's experiment. (See my previous post in this thread.)

* A note that one of the images was generated in a computer simulation was added. Hopefully, the posts, here, will make it clear as to why this was immaterial to my claims. But, if anyone has any questions, they can let me know.

* Added a link to Wikipedia, about Bose-Einstein Condensate, that was meant to be included in the first version.

* I now have Uri Geller's name spelled properly.

* A few minor formatting improvements were made.

Enjoy!

 

[Limbo]

I've been asked to post the following on behalf of Professor Suitbert Ertel:

Questions arose here about my recent parapsychological experiments which encompass creativity and aesthetics. Damien Broderick gave a brief account of them in his book "Outside the Gates of Science".

I will paraphrase three questions that were raised in the previous discussion, and provide answers which I am prepared to expand upon request.

1. "Was Ertel's paper published in full? I'd like to see that" (Ersby).

Unfortunately, an account of this last extension of my ping-pong ball research has not been published yet, for four reasons:

* General problems of publishing such findings in journals whose primary goal is to keep track of topics that can safely be located within the current Gates of Science;

* Lack of funding for research trespassing those gates;

* My intention to try to overcome mental barriers of disbelievers by additional experimental evidence.

The general kind of research I am conducting with ping-pong balls has been described in an article that has already been mentioned here: <http://www.psych.uni-goettingen.de/h...figurespdf.pdf>

2. "What was Ertel's hypothesis?" Wowbagger says: "I believe he designed the experiment to be more 'fun' for participants, than the more grueling Zener-card approach."

I am using ping-pong balls that must be jumbled inside bags and taken out in an "ecological" setting (more familiar conditions, more fun). This material and its use can be applied experimentally in very different ways. The idea underlying my more recent experiments, as Broderick described them, was this:

Psi - if it exists - might be capable of supporting or instigating creative processes. (Creativity is one of the phenomena least understood by mainstream science.)

Wowbagger explained the task of my participants: They are told to draw from each of two bags, simultaneously, a single ball; this is repeated many times, with ball replacement before the next draw.

The numbers written on the two balls signify row and column numbers of a 5x5 matrix of results. The results are recorded by an experimenter, as dots in the respective cells of this matrix. The resulting 25 dot frequencies are transformed into circles whose sizes represent deviations from chance expectation. If only chance is at work, these circles would be rather small. Without psi intervention (or cheating, collusion, etc, which I am satisfied has not occurred), they would exhibit mere random variation.

The participants were instructed to produce, without using their senses, deviations from chance. Any accumulated patterns should be interesting and pleasant--some outstanding pattern or form, whatever this may be.

3. Wowbagger's main objection seems to be replicability: "(Ertel's) statistical analysis is meaningless if he cannot develop a method of reliably reproducing the results". "No matter how 'nice' the pattern looks, it cannot reliably be replicated." "The problem is that the results seem to be just as unreliable, in the end."

Lampert seems to presume that I gathered results from hundreds of participants, and then sent Broderick a small selection of the nicest-looking that arose by chance. He should have asked me what I actually did.

In fact, after screening a larger preliminary set of participants, I selected 11 who showed significant surplus hits in my standard ping-pong ball test (as described in <http://www.psych.uni-goettingen.de/h...figurespdf.pdf>). My formal tests began at this point. There were 39 runs generating 39 independent matrices, each comprising 180 (7 runs) or 360 (30 runs) or 2400 (2 runs) double-handed ball selections. Of those 39 matrices, 15 showed significant deviations from chance, 7 of those significant cases showed extreme levels of significance (p = .00001 or even less). Only 2 of those 39 runs were conducted without control.

I consider another point crucial. Of those 39 runs, 15 were conducted without giving participants feedback about the gradual growth of dot frequencies in the 25 matrix cells. In the feedback condition, participants observe how certain cells are rather rapidly filled with dots while other cells are rarely frequented. The participants are inclined to support mentally these processes, so the final result might be partially due to the "mental guidance" (wishes) of participants. Under the no-feedback condition such guidance by the participants' wishes is not feasible. If forms develop in the matrix during such test session, they develop on their own, "out of the blue".

For my statistical analysis I am using chi squared, which is sensitive to any kind of deviation from chance expectation. It is an objective measure, not requiring subjective assessment of "interestingness" or "pleasantness."

fls proposed an unquestionably feasible method for evaluating aesthetic qualities; this, however, requires many additional pre-screened participants. The results I have already found might grow increasingly convincing if more subtle methods of analysis are developed, but they can hardly be weakened or eliminated by using such methods. (My search for alternative methods in the literature of quantitative aesthetics has not yet been successful, and I welcome any suggestions.)

 

[Beth]

Limbo,

Thank you for posting the response. And please pass on my thanks to Dr. Ertel for responding.

 

[Wowbagger]

I've been asked to post the following on behalf of Professor Suitbert Ertel:

Yes, I do appreciate any response Ertel and others are willing to give me.

To prevent this thread from going off topic, I think we should try to stick to the book's claim: That it is time for these paranormal studies to be accepted by mainstream science.

Therefore, we should focus on evaluating the experiments for suitability within mainstream science.

I would like to focus on this, first:

Of those 39 runs, 15 were conducted without giving participants feedback about the gradual growth of dot frequencies in the 25 matrix cells. In the feedback condition, participants observe how certain cells are rather rapidly filled with dots while other cells are rarely frequented. The participants are inclined to support mentally these processes, so the final result might be partially due to the "mental guidance" (wishes) of participants. Under the no-feedback condition such guidance by the participants' wishes is not feasible.

This, to me, sounds like the most potentially interesting aspect. If you could, indeed, find an empirical difference between the results of participants with feedback vs. no feedback; even though they would still be blindly pulling the balls from the bag; that could be a good starting point in discovering a new phenomena from psi research!

(At the very least, you would be well on your way to winning the Million Dollar Paranormal Challenge, but that's a different discussion.)

If you want mainstream science to accept this work, your best bet is to ask this:
What could be a possible mechanism for these results? By thinking of possible mechanisms, you should be able to develop a predictive hypothesis for psi, that could help sharpen the signal. By tapping into this new-found force, others should more-easily be able to consistently replicate the results.

Without a testable hypothesis, it does not matter how striking you think the results are, nor how statistically significant you measure them to be. You will always be seen as merely playing with fluctuations in the noise.

Remember, I am not describing my own attitude, here. I am describing the general attitude of the scientific community. The scientific community has a penchant for reliable results.

Let me give you an example why:

Damien points out that the more trials a subject does, the less striking the results are. The grids from 2,880 trials seem to be more even than those from only 180 trials. This is a trend we would expect from normal phenomena. With a good psi hypothesis, you should be able to boost this, so that the grids actually get less even with more trials, and more striking in their improbable patterns, over time.

You can see this with coin flips. If you flip a coin ten times, you might expect to get five heads and five tails. But, most of the time, you will not. You might get 6 heads and 4 tails, or 7 tails and 3 heads. With repeated trials, you get closer and closer to 50% each. This assumes fair tossing, of course.

If you could apply psi to the situation, someone could use "only their brain" (or wherever the source of psi's power comes from), to get most of the flips to come up heads, even after a tremendous number of trials.

Do we see why it might not, yet, be time for the paranormal to "come in from the cold"?

* Lack of funding for research trespassing those gates;

Do you really think funding is an issue?

If a benefactor gave you all the funding you needed, what would you do with it? What can you not accomplish with the resources you have now?

I remind you that developing hypothesis is, generally, really cheap. It often requires more imagination than cold, hard cash. Albert Einstein was able to develop the whole Theory of Relativity (actualy, two of them) without much in the realm of funding.

Of course, testing your hypothesis may require tons of cash, depending on its nature. (Physicists are spending billions on the Large Hadron Collider, for example.) However, before you worry about funding your tests, you need to have something to test for, first.

And, you can probably start out small. Remember, the effects of Bose-Einstein Condensate were first predicted and found to be persistent in observing liquid helium, in the 1930's. Only much later did someone gather the funds to actually isolate the stuff. But, at least there was little doubt it would be found, at that point.


Before I respond to the other responses, are there any questions, so far?

 

[fls]

So basically the whole "aesthetic gestalt" thing was a red herring?

Linda

 

[fls]

The explanations provided raise some questions which are not addressed by what Ertel has to say. When I get a chance to go over the information that has been provided in detail, I have some specific questions I would like to ask, but it probably won't be until after the weekend (I am away).

Linda

 

[Ivor the Engineer]

...and most of the tests were done at home with practically no controls against cheating or bias...

Are the bags used for the experiments pictured anywhere?

ETA: And is the ball-by-ball data sequence available? E.g., were participants drawing the same pairs of balls several times in a row?

 

[SpuriusMaximus]

Ok, maybe I'm not getting something obvious, but this claim I really don't understand:

In fact, after screening a larger preliminary set of participants, I selected 11 who showed significant surplus hits in my standard ping-pong ball test (as described in <http://www.psych.uni-goettingen.de/h...figurespdf.pdf>). My formal tests began at this point. There were 39 runs generating 39 independent matrices, each comprising 180 (7 runs) or 360 (30 runs) or 2400 (2 runs) double-handed ball selections. Of those 39 matrices, 15 showed significant deviations from chance, 7 of those significant cases showed extreme levels of significance (p = .00001 or even less). Only 2 of those 39 runs were conducted without control.

Significance or deviation from chance, compared to _what_?

The way I understand statistics and error bars, when I talk about the significance of a data set it's really about "how well it matches my hypothesis." If my hypothesis is, say, j = sigma * T^4 (Stefan-Boltzman), and I plot a graph of energy flux density vs temperature, then how significant the experimental data is, is measured basically in "how well my formula matches that data."

So what was that hypothesis there? 15 of 39 looked like they weren't created by chance... compared to what? What was the formula you were fitting them agains?

Oh, it's aesthetic value?

Well, then that makes it even weirder. Because then a whole matrix becomes a single data point. That's really what you're measuring: the aesthetic value of the whole resulting pattern. It's not a whole data set, it's not a graph, it's just one data point: how aesthetic the whole picture was. Just like, say, the Monal Lisa is one single data point, not a whole data set.

So claiming that one single data point is statistically significant, seems to me absurd to the extreme.

It's like rolling a die once, getting a six, and going "well, now that's statistically signifficant. There's only a 1 in 6 chance of getting _that_, so it can't be by chance."

Or, say, like taking a random number between 1 and 1 billion, and proclaiming it to be signifficant. Let's say I use the following program (in Java):

public static void main(final String[] args) throws Exception {
SecureRandom rand = new SecureRandom();
System.out.println(rand.nextInt(1000000000));
}

Oh, wow, I got 765636775. What are the odds of that? One in a billion, precisely. It's got to be signifficant. Those odds don't happen by sheer chance.

No, it's not. It's not signifficant at all, because it didn't have to match a clear prediction. If I claimed that I can predict the next random number to be 765636775, and exactly 765636775 came out, that would be a heck of a coincidence. But that a random number or graph configuration came out, without any prediction, is just dumb random chance.

But, to give credit where credit is due, all the above is just a variation of Feynmann's famous quote: "You know, the most amazing thing happened to me tonight. I was coming here, on the way to the lecture, and I came in through the parking lot. And you won't believe what happened. I saw a car with the license plate ARW 357. Can you imagine? Of all the millions of license plates in the state, what was the chance that I would see that particular one tonight? Amazing!"

So, again: those pictures were signifficant... how? Did they have to match a given prediction, or are you just trying to assign arbitrary significance to a random result?

If the criterion was "how aesthetic it looks", then each matrix is exactly one data point. How is one single data point signifficant to any hypothesis?

Was there any analysis done on the _whole_ data set, as opposed to cherry-picking individual data points? E.g., were those 15 out of 39 pretty matrices picked by people with psi abilities, while unskilled normal people generate only, say, 5 pretty pictures out of 39?