Review of: "Electrodermal Presentiments of Future Emotions"

[Im a Hedge]

This is the next installment in a series of reviews of paranormal research publications. I am working my way backwards through the Journal of Scientific Exploration. In the Vol. 18, No. 2, pp. 253–273, 2004 issue is the article,
"Electrodermal Presentiments of Future Emotions", by Dean I. Radin. The article is available as a pdf at, http://www.scientificexploration.org...18.2_radin.pdf

I highly recommend that you get the paper, especially so you can see Figure 2, which will be central in my discussion.

I will focus on Experiment 1 from this paper. I may touch on the other experiments in future posts.
The basic structure of the experiment is that subjects are shown images on a computer screen while there Skin Conductance Levels are recorded. There is a 5 second period of blank screen prior to the image being shown. The randomly-selected image is shown for 3 seconds. Then the screen is blank again for 10 seconds. The images vary in their 'emotional content'. The paper contains additional details on the experimental setup.

The results are interpreted by the author as demonstrating 'presentiment'. That is, the subjects react more strongly to the images with higher 'emotional content', and this reaction is observed before they see the image.
The data for Experiment 1 are summarized in Figure 2 in the article. The differences in the Calm and Emotional in the 5 second pre-image period look very small in this figure. No error bars are presented. The only notable feature is that the Emotional points are consistently higher than the Calm points during this period. Both Calm and Emotional show a trend of decreasing values. At time 0 (when the image is first presented) the Emotional series continues to decrease, while the Calm series begins to increase. Just before 2 seconds, there is an apparent 'bump' in the Calm series, peaking around 3 seconds, then slowly tapering off. Just after 2 seconds, the Emotional series begins a sharp climb, peaking around 5 seconds, then dropping off.

Upon initial examination, this plot looks consistent with a the conventional explanation. That is, more Emotional images produce a greater change in Skin Conductance Levels than do Calm images. However, Radin claims that this actually shows evidence of presentiment, because the Emotional series has values greater than the Calm series in the 5 seconds prior to the image being shown. A few things in the data are seem peculiar, given this explanation. The SCL values are decreasing throughout the pre-image period. If there is a presentiment effect, whereby the subject is anticipating an image with high emotional content, one would expect the SCL to be increasing during this period. After the image is shown, the Calm series begins to increase, which the Emotional series continues to decrease. In the Results section, it is stated that " Skin conductance is expected to react 2 to 3 seconds after an emotional stimulus, and this response is evident in Figure 2". However, if there is a presentiment effect, we would expect to see the SCL for the Emotional images begin to increase before this 2-3 seconds has elapsed. This is not seen. Rather, the SCL continues to decrease. This is further complicated by the fact that the SCLs for the Calm series begin to increase before the 2-3 seconds have elapsed.

Why does the Emotional series show consistently higher levels throughout the pre-image period? The presentiment explanation seems contradicted by the data, as discussed above. There are a few possibilities to consider. It could be an artifact of the normalization procedures. There are at least two levels of normalization applied to the data, followed by a averaging of the normalized values (if I have followed the description correctly).

The data could be biased by a small number of trials that are skewing the averages. The data are presented as averages of 860 trials. I do not see the standard deviations associated with these averages, and there are no error bars presented in the figure. It is possible that the small differences observed are due to noise.

The data look as though there is some sort of relaxation trend during the pre-image period (in both Calm and Emotional series). This could be an effect of recovering from the previous image. It would helpful to see that the randomization of images was not producing some artifact, such as an Emotional image being more likely to be followed by another Emotional image. If this were this case, it would be a good explanation of the data that are presented.


In general, the experimental design seems to have promise, but could use some tweaking. It seems that the 'cool down' period between images is not long enough, as the pre-image data seem to show an ongoing relaxation trend. My common concern about negative controls seems to be addressed here, as I think the Calm images serve as a good control. The data aren't compelling evidence of a presentiment effect. After heavy normalization and averaging there is a very small effect seen in a portion of the data. However, the presentiment explanation seems inconsistent with the data from other portions of the experiment. The absence of error bars in the figure, or standard deviation values in the text, renders even this small effect meaningless, as we cannot assess its statistical validity.

The data presented point towards a greater Skin Conductance Level response to images with high emotional content than to images with low emotional content. Any conclusions beyond that do not seem to be supported by the data presented in the article.


I am a Hedge

[jacob]

Dr. Radin posted today a blog entry about presentiments demos he's doing and I wrote in the comments about the issue that you wrote here and also of another critical review of his work.

His reply (in the blog):

I've commented on both of those critiques on this blog. My colleagues and I have been aware of these and other proposed artifacts for many years, we've studied them, and they do not account for the observed results.

E.g., anticipatory models assume that the experiment uses dichotomous targets (only calm vs. emotional). But most of my presentiment experiments (including the one mentioned in this post) select targets at random from a large pool of pictures of widely varying emotionality. That avoids the gambler's fallacy type of strategy, which is what anticipatory models are based upon.

Link to Dean Radin's post:
Entangled Minds: Presentiment demos

[Im a Hedge]

Originally Posted by jacob

Dr. Radin posted today a blog entry about presentiments demos he's doing and I wrote in the comments about the issue that you wrote here and also of another critical review of his work.

His reply (in the blog):


Link to Dean Radin's post:
Entangled Minds: Presentiment demos

Thanks for passing that along. I have discovered that I only scratched the surface of the presentiment subject, and there is a lot more to learn. I'll check out Radin's responses to the various criticisms and see if they they seem convincing.

(I'm sure you've noticed by now, but I've managed to get the forum site to come up again.)


I am a Hedge

[Im a Hedge]

Thanks again, Jacob, for passing information along from Dr. Radin. I searched his blog for "presentiment" and didn't find the post(s) where he addresses the specific criticisms in greater detail. If anyone knows where to find this, I would appreciate if they could post a link here.

Since my last post on this thread, I have checked out the posts at JREF about a computer simulation that produces very similar results to those published by Dr. Radin. It looks like a pretty good explanation of the results in terms of an anticipation effect. Radin says there is a problem due to assuming a dichotomous choice, but I'm not sure why this is a problem. I would be interested in finding his more detailed explanation.


I am a Hedge

[Larry Boy]

Originally Posted by Im a Hedge

Thanks again, Jacob, for passing information along from Dr. Radin. I searched his blog for "presentiment" and didn't find the post(s) where he addresses the specific criticisms in greater detail. If anyone knows where to find this, I would appreciate if they could post a link here.

He may be referring to this post:
https://www.blogger.com/comment.g?bl...62538613990459

Yes. That paper discusses a theoretical bias that might occur under certain conditions when dichotomous stimuli are used (i.e., only calm or emotional targets). While the simulations show a potential bias (based on many assumptions), analysis of actual presentiment data conducted in studies run both before and after that paper do not show any bias. So the potential criticism does not hold up.

In addition, in my more recent presentiment experiments I've been using all pictures (selected uniformly at random with replacement) from the International Affective Picture System, which provides a very broad range of emotionality. The final analysis is based on a subset of the 5% most emotional vs. 5% most calm pictures pooled across subjects. That strategy overcomes the theoretical objection, and the results are still significant.

[Im a Hedge]

Originally Posted by Larry Boy

He may be referring to this post:
https://www.blogger.com/comment.g?bl...62538613990459

Thanks for finding and posting that link, Larry Boy. He seems to be responding to a 2002 paper, which I haven't read. But one of the criticisms was from a more recent post on the JREF form (not exactly what I would consider pee-reviewed publication..). In that JREF thread (Simulating "Presentiment" Experiments - JREF Forum) the poster claims to have closely reproduced the "presentiment" results with a computer simulation using a simple anticipation function.

Dr. Radin's reply to the 2002 paper dismisses it on the basis that is uses "dichotomous stimuli". I don't see why this is essentially different from his paper, which presents the stimuli as having been either "emotional" or "calm".

So, I'm still not understanding what his response really means. I may try to look up the 2002 paper being referenced. I don't want to put too much weight on a computer simulation "published" in an online forum. (Unless I can get the code and evaluate it for myself.)


I am a Hedge

[BannedBySkeptiko]

The paper reports three attempts to replicate a previous result, of which two were failures. The way Radin chooses to spin things, a careless reader might mistake the results as a a case *for* the effect being repeatable.

Radin's dismissal of "selective reporting" is a joke. Why didn't he report the results of experiment two at the time he completed experiment two? One plausible reason is that experiment two was such a huge failure to replicate; with over twice as many trials and results indistinguishable from chance, even psi-pitchman Radin could not spin that one in favor of his psychic theory.

-Bryan

[BannedBySkeptiko]

Originally Posted by Im a Hedge

I don't want to put too much weight on a computer simulation "published" in an online forum. (Unless I can get the code and evaluate it for myself.)

I just checked, and Simulating "Presentiment" Experiments - JREF Forum links to the code. Do you know the C programming language?

My suspicion, after not much investigation, is that the simulation may well produce graphs like in 'figure 2', but use too few trials. The figure is good for visualizing the difference at issue, but not for evaluating its significance. Before the stimulus point, the lines are pretty close. Is the difference just chance, or is it significant? The real outcome is the p-values.


One more thing: while I agree about not putting much weight on this computer simulation, I'd caution not too put much weight on the pseudo-scientists and their journals either. Serious scientific journals have given "peer review" a good name because their peering is at the top; the real experts that get real results.

An arbitrary group of idiots can publish a journal and review their idiot peers, and they can title their rag with any name they fancy. The Journal of Scientific Exploration is not a scientific journal. It exists to publish papers that fail the standards of the reputable journals.

-Bryan Olson

[Im a Hedge]

Originally Posted by BannedBySkeptiko

I just checked, and Simulating "Presentiment" Experiments - JREF Forum links to the code. Do you know the C programming language?

I know a little C, but I don't know C. I spotted the link to the code after my previous post. It's a very short bit of code. I'm a java person mostly, so some of the stuff in the code is foreign to me. I may try to figure it out a bit better, if I can make the time. I may try to re-implement it in java, which will give me more ability to make changes to try out different methods.

(This text editor seem to think 'java' is not word. How odd. It's at least three words)

I am a Hedge

[Larry Boy]

Bryan,

I appreciate your comments and I think they're useful, but I think we could have a lot more productive discussion if you didn't refer to people as idiots.