Saturday, December 08, 2007

Japanese translation of Entangled Minds



The Japanese translation of Entangled Minds is now available. It's beautifully done, with many new, clever illustrations added to enhance the text. I owe special thanks to the translator, Masato Ishikawa.

21 comments:

Book Surgeon said...

Dean, I'm curious what your response to this study is:

http://mindblog.dericbownds.net/2007/12/using-neuroimaging-to-resolve-psi.html

Obviously, the authors are summarily discounting all previous psi evidence, which is a careless, biased approach. Also, their study makes several flawed assumptions, including that psi effects would occur under the conditions they used and that psi effects register in the brain at all. We know through your other work that they do register physiologically.

Wondering what you think...

Dean Radin said...

I'll reserve judgment until I get a chance to read the actual paper. From the abstract it looks interesting, although I hardly think that a single study can hope to resolve anything, pro or con.

Tor said...

I agree book surgeon. From the abstract it looks like they haven't read the literature or just ignored it. It's always frustrating to see such cases of ignorance.

Enfant Terrible said...

Mr. Radin,

i put the full article for download in this link:
http://www.4shared.com/file/32136202/cf7d754/Moulton_Kosslyn.html

Best wishes.

Dave Smith said...

I've read the brain imaging paper provided by book surgeon. Firstly, no behavioural psi was found so it should be no great surprise that they found no differential brain activation. Interestingly, the authors don't make any issue of this possible interpretation in their conclusion and discussion. Secondly, their experimental design might have encouraged the receivers (being scanned) to use a precognitive strategy. If they did, the authors could have been analysing the wrong trial periods. IMO they should have tried to detect differential activation during the periods when the participants were making their binary decisions just before feedback, ie, correct minus incorrect decisions. Instead they analysed differential activation when participants were viewing the pictures before making their choice. One might argue that this is the least likely period to find differential activation because attention is drawn away from internal imagery towards processing an external stimulus. The authors are quite confident that this experiment provides strong evidence against psi, but I think it just highlights the problems of studying a phenomenon we know little about, especially since we know even less about how their dependent variable expresses psi.

Enfant Terrible said...

I noted the link I provided is not full, but there is no problem. The article can be downloaded in this link:

http://www.4shared.com/file/32136202/cf7d754

Dean Radin said...

I congratulate the authors of this paper because unlike many who hold strong opinions about this topic, they actually conducted an experiment. However, I disagree with their assertion that this single study resolves anything. Like any new experiment, all it really does is raise new questions.

There are so many points I could respond to in this paper that I was tempted to write a comprehensive reply. But then I remembered that I've already written one. It's called Entangled Minds, which apparently these authors have not read. Nevertheless, a few comments:

1) The authors overlooked four previously reported fMRI psi studies, all four of which reported significant results.

2) Compelling personal psi experiences are dismissed as fallacious beliefs due to cognitive biases. I fail to see how one or more of the known cognitive biases can conceivably explain even the example they provide of a crisis telepathy experience, to say nothing of thousands of similar experiences. Obviously if someone was constantly reporting such experiences, but only one in a thousand times the experience was verifiable, then such anecdotes wouldn't carry much evidential value. But that is not the case. These are often once in a lifetime experiences, and they shatter previously held beliefs. The irony here is that a case can be made that one of those experiences started the neurosciences!

3) The authors made a common mistake by asserting that independent ganzfeld meta-analyses failed to successfully replicate, citing Milton & Wiseman (1999). Unfortunately, that meta-analysis, which is often used to cast doubt on the repeatability of the ganzfeld results, was statistically flawed and underestimated the overall p-value. When corrected, in fact it did result in a significant overall hit rate.

4)One participant out of 16 showed significant fMRI differences consistent with the psi hypothesis. The authors examined three alternative explanations for this result, and concluded that idiosyncratic responses accounted for the significant results. Unfortunately, this explanation reveals a flaw in the underlying design of the entire experiment. If it is possible to dismiss one individual's results as an artifact, then there is no reason to have confidence that the rest of the data is artifact-free.

5) The experimental task is new, and complex. As far as I know, there is no precedence justifying why we think this procedure might work at all. This reminds me of a paper published in The Humanistic Psychologist a few years ago in which two skeptical psychologists reported a series of eight ganzfeld experiments, which overall produced a significant result. They did not like this outcome and so they conducted another study using a new, untested, ad hoc design, and it resulted in a significantly negative outcome. They then used that last study to dismiss the results of the first eight studies. In the present case, explaining away the one participant who showed a significant result also potentially explains away all other significant results, in which case why did they use this design in the first place?

Book Surgeon said...

The answer to your rhetorical question, Dean is bias on the part of the experimenters. Clearly, they wanted to disprove psi due to their own bias against its existence, so they developed a simplistic study that would "disprove" it. It is very much in line with the ganzfeld study from 2006 that you mention (from Notre Dame, if I'm remembering correctly?) in which the researchers not only couldn't cope with the cognitive dissonance of their own ganzfeld trials showing the same ballpark 32-33% hit rate as the historical hit rate from past trials, but also were woefully ignorant about statistical significance.

This whole affair, and the fact that it was published in a major journal, reminds me of the 1998 JAMA paper published about the study by the 11-year-old girl that "proved" that therapeutic touch is bogus. I won't even get into the many fallacies of this whole thing, such as the ridiculousness of a major journal publishing experimental results from a sixth-grader or the idea that a single study of anything can be conclusive. What's instructive is that that work, just as with this fMRI experiment, was embraced so eagerly by the mainstream, as well as by skeptics.

It's as if when psi or anything quasi-paranormal is in question, there's a rush in the science mainstream to jump on the debunking bandwagon and shout, "Yeah, see, we think this is garbage, too!" It's like a gay high schooler rushing to make the football team so he can deflect any suspicions that he's homosexual. Seems that the conventional research community is eager to prove its skeptical street cred.

Once again we see that if you want a certain result badly enough, you can concoct a study that will deliver it.

julio siqueira said...

-
From the point of view of someone that not only "believes" in ESP-PK but that also knows the available good evidence is indeed compelling for it (Ganzfeld mainly), the question that assaults my mind is why was there so little evidence (if any) for psi in this one study.

Dean Radin said...

If we knew exactly why there wasn't much evidence for psi in this study, we'd be much farther along in creating recipes for easy replication. But there could be many potential reasons. E.g., a new type of experimental task was used, we don't know whether the set and setting were optimal, we don't know what the implicit or explicit expectations of the research team were, perhaps the participant relationships and motivations were non-optimal, etc. E.g., the authors used a few sets of identical twins, but just being twins isn't the issue. Some twins don't like each other and never report any psychic connections; others love each other and report them all the time. Was this experiential factor assessed? Then there's growing evidence that certain geomagnetic field frequencies, which are influenced by all sorts of solar-geophysical reasons, influence psi performance. Does the massively powerful magnet in an fMRI interact with those effects? We don't know.

This paper reports one study. Maintaining a skeptical attitude is always a good idea when it comes to controversial topics, so we should maintain as much doubt about one negative experiment, especially one reporting a new type of design, as we maintain about one positive experiment. Let's see what happens after multiple replications by independent teams.

Enfant Terrible said...

Mr. Radin,

I think it is important that you write to the journal, about the possible flaws in the experiment. The journal has a high impact factor and the article "Using Neuroimaging" can made that the mainstream loose even the little interest that has in psi.

Best wishes.

Dean Radin said...

Does anyone know if this journal has a letters section? It might be useful to point out that this was not the first published fMRI psi study, and of the five now conducted, it's the only one reporting a null result (which is not surprising for all the reasons discussed above).

Book Surgeon said...

Here's the Feedback link for the Journal of Cognitive Neuroscience:

http://jocn.mitpress.org/cgi/feedback

Dean Radin said...

Thanks, but it appears that that link is for website and subscription issues, and not for general commentary on published articles.

Dave Smith said...

On the technical side of things, there are other reasons why fMRI might not have detected anything in this study. Logothetis (2003)studied the BOLD signal that forms the basis of fMRI. He simultaneously measured the BOLD signal and a more direct measure of neural activity, the local field potential. He concludes:

"In all of the measurements, the signal to noise ratio of the neural signal was an average of at least one order of magnitude higher than that of fMRI signals. This observation indicates that the statistical analyses and threasholding methods applied to the haemodynamic response probably underestimate a great deal of actual neural activity related to the stimulus or task..."

In other words, the fMRI study by Moulton and Kosslyn could simply have been insensitive to any relevant activity changes. There have been other successful fMRI studies that Dean has mentioned but they measured a different aspect and time period of a psi task, where the neural activity changes were evidently strong enough to show up on the BOLD signal.

ref: Logethetis NK. The neural basis of the blood-oxygen-level-dependent functional magnetic resonance imaging signal. In Parker A (ed), The physiology of cognitive processes. 2003.

Max Clarke said...

Almost all the books I buy are actually audiobooks, unabridged ones. For instance, the Fred Alan Wolf audiobooks about quantum physics, and also the audio book Entanglement. I would very much like to buy Dean Radin's stuff through an audiobook store such as Audible or iTunes. I've heard him several times discussing entanglement and consciousness, and his story should be heard through an unabridged audiobook, not just read.

Dean Radin said...

Max Clarke said... I would very much like to buy Dean Radin's stuff through an audiobook store ...

I've received a number of similar requests, and I'd certainly like to see an audio book made, but that decision is up to the book publishers, not me.

I often listen to audio books and lectures while traveling.

John said...

Hi, Dean,

This link might be useful in contacting the journal:

jocn@jocn.berkeley.edu

Sam Moulton said...

Greetings,

      Thank you all for your thoughtful comments on our experiment, and please see below for our replies. We didn't respond specifically to those who questioned our intent (e.g., likening us to gay high school athletes) because we believe that it should go without extended discussion that, as scientists, our intent was and is always this: to discover the truth. We did not presuppose the outcome of this experiment, nor do we consider ourselves as advocates for or against the existence of psi. We set out to help inform the reality of psi, whatever that may be.

      Critique ("dave smith"). "no behavioural psi was found so it should be no great surprise that they found no differential brain activation. Interestingly, the authors don't make any issue of this possible interpretation in their conclusion and discussion."
      Response. From our article (p, 184): "The behavioral task served two purposes: it motivated participants to detect the psi stimuli, and it provided the structure necessary for the BOLD contrasts. We did not include the behavioral task because we expected an explicit (guessing rate) or implicit (response time) effect. Unfortunately, there are no behavioral tasks that can be used as "standard metrics" of psi; it is precisely this failure of behavioral research that has motivated our neuroimaging work. This highlights a critical asymmetry in cognitive neuroscience: all behavior requires neural activity, but not all neural activity yields behavior. Thus, in the current experiment, although the unexpected presence of a behavioral effect mandates a neural correlate, the expected absence of a behavioral effect does not necessarily imply null neuroimaging results."

      Critique ("dave smith"). "their experimental design might have encouraged the receivers (being scanned) to use a precognitive strategy. If they did, the authors could have been analysing the wrong trial periods. IMO they should have tried to detect differential activation during the periods when the participants were making their binary decisions just before feedback, ie, correct minus incorrect decisions. Instead they analysed differential activation when participants were viewing the pictures before making their choice. One might argue that this is the least likely period to find differential activation because attention is drawn away from internal imagery towards processing an external stimulus."
      Response. We did in fact conduct this analysis suggested above, and found nothing. Moreover, assuming that precognition occurred, we would still have expected it to come through in the analysis of psi versus non-psi stimuli.

      Critique (Dean Radin). "The authors overlooked four previously reported fMRI psi studies, all four of which reported significant results."
      Response. We have only found one study that arguably investigates psi: the Richards et al. (2005) study in the Journal of Alternative and Complementary Medicine. Because they never mentioned "psi", "ESP", "telepathy", "precognition", "clairvoyance", or "paranormal", we did overlook this study initially when reviewing the literature; we agree that this study is relevant to parapsychology. One might contest whether or not it is the "first psi fMRI" study based upon its failure to address directly psi, ESP, telepathy, precognition, or clairvoyance (or even use any of those terms). But whether or not our experiment was the "first ESP fMRI study" is ultimately irrelevant because our intention here was not historical, but scientific: to develop a novel, maximally-informative method for studying ESP. We would be grateful if you could inform us of the three other fMRI psi studies.

      Critique (Dean Radin). "Compelling personal psi experiences are dismissed as fallacious beliefs due to cognitive biases. I fail to see how one or more of the known cognitive biases can conceivably explain even the example they provide of a crisis telepathy experience, to say nothing of thousands of similar experiences. Obviously if someone was constantly reporting such experiences, but only one in a thousand times the experience was verifiable, then such anecdotes wouldn't carry much evidential value. But that is not the case. These are often once in a lifetime experiences, and they shatter previously held beliefs. The irony here is that a case can be made that one of those experiences started the neurosciences!"
      Response. We do NOT dismiss compelling psi experiences as fallacious beliefs due to cognitive biases. Instead, we merely raise (as many others have, for good reason) the possibility that such experiences may result from cognitive biases. As stated in our article (p. 182): "Cognitive biases... may explain many apparently paranormal events that people report [emphasis added]". These types of experiences -- as well as experimental evidence -- inspired us to study psi in the first place. But without experimental control, one cannot know with any certainty how to interpret anecdotal accounts of psi.
      We would be happy to explain how some of the cognitive biases we mentioned in our paper might explain "crisis telepathy" experiences. For instance, take the example from our paper of the woman who reported waking up in the middle of the night under extreme physical and emotional duress, only to find out the next day that her son was killed at approximately the same time that she had woken up the night before. The clustering illusion is the bias to find patterns in random data; applied here, we might underestimate the probability that a temporal "clustering" of the two events (mother's crisis, son's death) might occur by chance. The "availability error" is the tendency to make mistakes on the basis of the mental salience of a particular event; applied here, we (or the mother) might underestimate the probability of this crisis illusion because we cannot bring to mind the less salient events of parents not waking up when their children die, parents waking up when their children don't die, and parents not waking up when their children don't die. The confirmation bias is the bias to search for or interpret information in a way that supports our prior beliefs; applied here, the mother might have interpreted and reported this event in terms of telepathy because of her prior belief in the paranormal, and "skeptics"/"believers" interpret this story differently because of their prior beliefs. And finally, the bias blind spot refers to the finding that our prior knowledge and awareness of cognitive biases doesn't eliminate their effects; as applied here, even well-informed researchers can misinterpret the probability that this event might occur by chance alone.
      As for the claim that psi experiences may have started the neurosciences, we don't understand its relevance. One could make the claim that accounts of animal magnetism helped to found psychology, but that doesn't mean that animal magnetism exists!

      Critique (Dean Radin). "The authors made a common mistake by asserting that independent ganzfeld meta-analyses failed to successfully replicate, citing Milton & Wiseman (1999). Unfortunately, that meta-analysis, which is often used to cast doubt on the repeatability of the ganzfeld results, was statistically flawed and underestimated the overall p-value. When corrected, in fact it did result in a significant overall hit rate."
      Response. First, our argument is that past experimentation suffers from three fatal flaws, not merely a lack of replication. Thus, even if we agreed that the psi-Ganzfeld produces replicable effects, we would still disagree on its interpretability. As stated in our article (p. 183): "positive evidence that has been reported as merely 'anomalous'. As many have noted previous (e.g., Mumford & Rose, 1995), the absence of a normal explanation does not justify the presence of a paranormal explanation."
      Second, regarding Milton & Wiseman's meta-analyses, we think it's unfair to characterize it as "flawed". One may disagree with some of their meta-analytic methods, but the truth is that there is no one "right" way to conduct a meta-analysis: there are many right ways. A true effect is robust enough to survive alternative analyses, and doesn't depend upon particular meta-analytic parameters. The debate over the psi-Ganzfeld analyses has played out (for now, anyways), and we have no desire to rehash it; for anyone interested on this topic, we recommend the 2002 exchange between Milton, WIseman, Storm, and Ertel in the Journal of Parapsychology, as well as John Palmer's 2003 review in the Journal of Consciousness Studies .

      Critique (Dean Radin). "One participant out of 16 showed significant fMRI differences consistent with the psi hypothesis. The authors examined three alternative explanations for this result, and concluded that idiosyncratic responses accounted for the significant results. Unfortunately, this explanation reveals a flaw in the underlying design of the entire experiment. If it is possible to dismiss one individual's results as an artifact, then there is no reason to have confidence that the rest of the data is artifact-free."
      Response. The fact that positive results from single-participant data required a non-parametric analysis does not mean that our experiment was "flawed" or that these results support the existence of ESP. Our non-parametric analysis of those single-participant data ruled out ESP as a viable explanation for those results. Those spurious results occurred because we used a counterbalanced design: Counterbalancing the stimuli and conditions (i.e., switching around stimuli for different participants, so that each stimulus occurs equally often in each condition) is a feature of a *good* experimental design (so that there is not a confounding between the specific stimuli used and the conditions), not a flawed experimental design. But by its very nature this counterbalancing procedure means that different people receive different stimuli in the different conditions -- and there is no way to ensure that the stimuli are perfectly equated on many different dimensions (such as arousal) for any one person. The point of counterbalancing, in this case, is that since we can't be sure that we even know all the relevant ways in which stimuli differ, we take care of the problem by using the same stimuli equally often in the different conditions (for different participants, which means that the averaged data cannot be explained by the confoundings present for any one person).

      Critique (Dean Radin). "The experimental task is new, and complex. As far as I know, there is no precedence justifying why we think this procedure might work at all. This reminds me of a paper published in The Humanistic Psychologist a few years ago in which two skeptical psychologists reported a series of eight ganzfeld experiments, which overall produced a significant result. They did not like this outcome and so they conducted another study using a new, untested, ad hoc design, and it resulted in a significantly negative outcome. They then used that last study to dismiss the results of the first eight studies. In the present case, explaining away the one participant who showed a significant result also potentially explains away all other significant results, in which case why did they use this design in the first place?"
      Response. The irony here is that we designed the study with the goal of trying to maximize the likelihood of finding something. The design was intended to allow us to detect evidence of three distinct types of anomalous cognition - which we took to be a strength of the method. The experimental task is neither new nor complex. In fact, it was exceedingly simple (binary guessing), and based off of past parapsychological research that reported positive psi findings (Don, McDonough, and Warren's studies)!

      Critique ("dave smith"). "the fMRI study by Moulton and Kosslyn could simply have been insensitive to any relevant activity changes"
      Response. If fMRI in general was insensitive to subtle psychological effects, it is difficult to explain the hundreds of subtle psychological effects that have been revealed by fMRI. And if our fMRI design in particular was insensitive to psychological effects, it is not clear why we were able to detect numerous non-psi effects in our data.

Dean Radin said...

Thank you Sam for your comprehensive responses.

On your comment, "We have only found one study that arguably investigates psi," there is no ambiguity in these four studies. They all explicitly studied psi effects, as reported in Entangled Minds. The remaining three citations are:

Achterberg J, Cooke K, Richards T, Standish LJ, Kozak L, Lake J. Evidence for correlations between distant intentionality and brain function in recipients: a functional magnetic resonance imaging analysis. J Altern Complement Med. 2005 Dec;11(6):965-71.

Standish LJ, Johnson LC, Kozak L, Richards T. Evidence of correlated functional magnetic resonance imaging signals between distant human brains. Altern Ther Health Med. 2003 Jan-Feb;9(1):128, 122-5.

Bierman, D. J. & Scholte, H. S. (2002). Anomalous anticipatory brain activation preceding exposure of emotional and neutral pictures. In Proceedings of Toward a Science of Consciousness, Tucson IV.

On your comment "regarding Milton & Wiseman's meta-analyses, we think it's unfair to characterize it as 'flawed.'" It was flawed in that the statistical method used in that paper differed from the extremely simple hit vs. miss analysis used in previous meta-analyses. It's true that for very robust effects differences in analytical procedures might not matter much, but when comparing meta-analyses reviewing the same type of experiment I should think the best approach would be to use exactly the same analysis. The important upshot in this case is the difference between reporting "failure to replicate" vs. "success in replicating."

On: "Unfortunately, there are no behavioral tasks that can be used as 'standard metrics' of psi...", could you elaborate on what you mean? E.g., in the case of the ganzfeld task, the hit rate is standard in the sense that this is an easily understood measure of performance that researchers have been reporting for several decades.

On: "As for the claim that psi experiences may have started the neurosciences, we don't understand its relevance." Hans Berger was motivated to develop the first human EEG, and in so doing set the future course of the neurosciences in motion, because of a personal psi experience.

On: "Those spurious results occurred because we used a counterbalanced design". Yes, counterbalancing is a good idea, but why then weren't the stimuli counterbalanced within subjects? By only counterbalancing across subjects the possibility arises that each individual subject could have ended up with spurious results because of a fortuitous distribution of targets. In that case you could have ended up with a curious outcome where each person showed independently significant results, but the grand average result was not significant! Or have I overlooked something?

Sam Moulton said...

Hi Dean,

By "standard metrics" of psi we meant simply that there is no known behavioral method that can reliably assess the presence of psi in any given participant (or even in group data). At best, effects are statistically replicable over a large number of experiments and participants.

Regarding counterbalancing, we counterbalanced stimuli as much as we thought prudent within participants. Psi and non-psi stimuli occurred equally often in the first and second positions of the trial sequence, and we used the normative IAPS ratings to equate psi and non-psi stimuli on arousal and valence. We chose not to show the same picture twice for the same participant, once as the psi stimulus and once as the non-psi stimulus, because that would have produced large recognition effects that may have swamped any psi effects; doing so may have also complicated or eliminated any precognitive effects. Even if we had chosen to counterbalance psi condition and stimuli within participants, we still would have wanted to conduct a non-parametric analysis on single-participant data; a given participant may have happened to recognize some psi stimuli (i.e., on the second exposure) more than non-psi stimuli, or vice-versa. And no, individual psi effects would not cancel each other out in the group analysis; to the contrary, they would add together (only counterbalancing effects would cancel each other out).