PSI (or cognitive illusion?)

An evocative title for a thread, I know. But I'm just quoting the paper I'm including.

I'm no militant psi skeptic, however this paper has me stumped. It's apparently related to the Ganzfeld experiments. But that's where my understanding of it's relevancy goes. 

I talked to Dean Radin about it but felt I'd discuss it further here. A paper like this doesn't leave much of a dent in the Ganzfeld experiments, not to mention the different studies in a dozen other kinds of PSI phenomenon and even other experiments into telepathy. But it makes me wonder, how much of it all is PSI and what isn't. I'm just confused

I'd love feedback from anybody on it, but especially from the "proponent" crowd that is around here?

[Laird]

Hey man. I saw you post this on Skeptiko a few days ago but didn't have the motivation to respond. This response, then, correspondingly, needs a disclaimer: I have not read beyond the abstract of that paper. If I do, and if I see a need to modify my response, then I hope to respond again. As it is, here's my "where the rubber meets the road" response:

Let's say that the findings of this paper are correct. Fine: one judge found "remarkable correspondences" as much in decoys as in genuine targets amongst twenty segments. Etc etc for four other studies.

OK, but here's where the rubber meets the road:

1. Twenty is a very, very limited sample size considering the (small) size of the effect.
   
2. More importantly, in the real Ganzfeld experiments, where the judge is required to commit to one out of four potential targets, across many, many experiments, those judges consistently and statistically significantly choose the correct target above chance levels (25%). If this paper contends otherwise, then it needs to explain (away) those results.
   

I hope that's not too simplistic a response, but it's all I've got for you right now. Cheers.

Hey man. I saw you post this on Skeptiko a few days ago but didn't have the motivation to respond. This response, then, correspondingly, needs a disclaimer: I have not read beyond the abstract of that paper. If I do, and if I see a need to modify my response, then I hope to respond again. As it is, here's my "where the rubber meets the road" response:

Let's say that the findings of this paper are correct. Fine: one judge found "remarkable correspondences" as much in decoys as in genuine targets amongst twenty segments. Etc etc for four other studies.

OK, but here's where the rubber meets the road:

1. Twenty is a very, very limited sample size considering the (small) size of the effect.
   
2. More importantly, in the real Ganzfeld experiments, where the judge is required to commit to one out of four potential targets, across many, many experiments, those judges consistently and statistically significantly choose the correct target above chance levels (25%). If this paper contends otherwise, then it needs to explain (away) those results 
   

Thanks Laird,

It seems pretty insignificant then. We all know that psychology and neuroscience rely on very similar sample sizes, and suffer God awful replication rates (psychology taking the lead with a whopping 30 percent!) So that's not great odds for it being repeated. 

Plus the two "rubber meets the road" points of yours. 

On a extra note, I emailed Craig Weiler about it and he made the point that this study, being so subjective, at this size could get tweaked and give you totally different results probably. So obviously the extremely small size is one of it's biggest holes. 

The paper's main author, Westerlund, seems like he used to be quite out to find a hole in the Ganzfeld experiments.

My thoughts were essentially the same as Laird's. The study is interesting, but not large enough to allow firm conclusions to be drawn. After seeing the post on Skeptiko I did mention it on another thread, thinking that it might be consistent with other results mentioned there, which might tell us something about the mechanism of psi:
http://psiencequest.net/forums/thread-wa...6#pid13026

I looked at this related paper by the same group:
https://w3.psychology.su.se/staff/jwd/cs...erlund.pdf

The results were actually above chance, both in terms of the percentage of the "remarkable correspondences" that related to targets rather than decoys, and in terms of the average ratings of the correspondences for targets versus decoys. It's possible that in a larger study these differences might become statistically significant, but this study is too small to tell us anything definite.

My thoughts were essentially the same as Laird's. The study is interesting, but not large enough to allow firm conclusions to be drawn. After seeing the post on Skeptiko I did mention it on another thread, thinking that it might be consistent with other results mentioned there, which might tell us something about the mechanism of psi:
http://psiencequest.net/forums/thread-wa...6#pid13026

I looked at this related paper by the same group:
https://w3.psychology.su.se/staff/jwd/cs...erlund.pdf

The results were actually above chance, both in terms of the percentage of the "remarkable correspondences" that related to targets rather than decoys, and in terms of the average ratings of the correspondences for targets versus decoys. It's possible that in a larger study these differences might become statistically significant, but this study is too small to tell us anything definite.

Jeez, I didn't think my post got much attention at Skeptiko. Thanks Chris. 

I'll echo my own reply to Laird above, in my own point about the small sample size, areas of science that have leaned on similar sizes for much of their papers such as psychology have been suffering horrible replication rates in the last ten years or so. So the odds that the study would get replicated with the same results even at the same size as it originally was are not good. 

So you are saying that this paper actually says something about the mechanisms behind PSI? I'd love for you to elaborate the last paragraph of your reply, I'm kinda confused on it? The paper just confused me in general. 

Would more statistically significant results be in favor of a PSI "hypothesis"? 

(Sorry if I'm making no sense)

Jeez, I didn't think my post got much attention at Skeptiko. Thanks Chris. 

I'll echo my own reply to Laird above, in my own point about the small sample size, areas of science that have leaned on similar sizes for much of their papers such as psychology have been suffering horrible replication rates in the last ten years or so. So the odds that the study would get replicated with the same results even at the same size as it originally was are not good. 

Pretty safe to say that it would yield more significant results if done over a larger sample size. Looking at the nature of just about every other Ganzfeld study and paper before and since

To complicate matters further, the traditional "hit rate" reported in the paper was slightly below chance at 23% based on the judge's choices, but only 14% based on the Ganzfeld receivers' own choices (half the receivers made choices), which is significantly lower than chance. Probably these figures further illustrate the fact that the study was too small.

[Laird]

[H]alf the receivers made choices [...] Probably these figures further illustrate the fact that the study was too small.

And perhaps also that they ought to have been forced to have made a choice? (which is not to take away from your - valid - point)

And perhaps also that they ought to have been forced to have made a choice? (which is not to take away from your - valid - point)

I'm not sure whether only half chose because they were given the option of not choosing, or whether the protocol changed during the course of the study. Maybe a proper reading of the paper would clarify that.

I'm not sure whether only half chose because they were given the option of not choosing, or whether the protocol changed during the course of the study. Maybe a proper reading of the paper would clarify that.

Well it's attached in the OP, but I do think they were given the choice. I think