Hyperbole is not normally ‘my thing’. In this case, I feel it is justified. Polite academic language simply doesn’t cut it.
Remember you read it here first: the psi hypothesis has been blown sky high – it’s nothing less than catastrophic – 100 years of laboratory research has yielded zero confirmable findings. The current review demonstrates an almost complete lack of confirmability in Parapsychology. This is not only a replicability crisis – it is a existential crisis for the entire Parapsychology field.
This author has been expressing doubts for a good many years. Now the die is cast. It is looking 100% certain that the laboratory is the last place on Earth to observe the influence of ‘psi’. In previous posts here and here I discuss the crucial, recent findings from Parapsychology. After almost 100 years of laboratory research, there are sufficient numbers of confirmatory studies to reveal the truth about the (non) existence of psi.
If a ‘Dr Smith’ says ‘they’ve’ done an exciting new study that found evidence of psi, we call that an ‘exploratory study’. We don’t get over-excited until Smith study’s been replicated. There have been tens of thousands of ‘Dr. Smith’ studies, which in reality mean nothing. None of such exploratory studies can be reliably replicated. Reviewers have attempted to integrate dozens or hundreds of such findings in narrative reviews or meta-analyses, but it is a thankless and futile task because without the ability to reliably replicate, it’s simply a case of ‘rubbish’ in, ‘rubbish’ out and we remain none the wiser.
Since the launch of the Open Science Framework, scientists have attempted to do better by running confirmatory studies that specify their hypothesis in advance, register it along with the procedures and statistical analyses, and then carrying out an attempted replication of one or more of the exploratory studies. For Parapsychology, Professor Caroline Watt of the Koestler Parapsychology Unit at the University of Edinburgh has led the way by establishing a Registry of Parapsychological Experiments (RPE). The registry is an essential data-base for evaluating not only single hypotheses but for evaluating the entire Parapsychology field.
I report here my analysis of the RPE and the discoveries that can be made there. I have been a little surprised at what the collection of registered studies is revealing. In a previous post, I indicated how the RPE suggests that almost all of the positive confirmatory studies on psi are emanating from the same investigator, Dr Patrizio Tressoldi, of the University of Padua in Italy. Padua is Europe’s oldest university and once the base of one of the world’s most revolutionary scientists, the astronomer Galileo Galilei (1564 – 1642). An analysis of Tressoldi’s confirmatory studies of laboratory psi suggests that ‘Error Some Place’ is the explanation. I summarize below brief comments on Tressoldi’s nine published studies and eleven Confirmatory Hypotheses.
Brief Notes on Why Tressoldi’s ‘Confirmatory Studies’ Are 100% Non-confirmatory
The numbers in the following notes are the ID numbers of the studies in the RPE. Full details of each study are provided in the Appendix.
1049 DISCONFIRMATION. This study should probably be voided due to irregularities in the description of the Confirmatory Hypotheses: the registry document specifies a single hypothesis while the published report specifies two Confirmatory Hypotheses. In spite of this apparent ‘fishing expedition’, both confirmatory hypothesis in the published study are reported as disconfirmed.
1013 VOIDED Because the investigators changed the hypothesis after the data had been collected, which invalidates the conditions of a preregistered study.
1012 DISCONFIRMATION. Two Confirmatory Hypotheses were both disconfirmed. In an earlier post, I inadvertently misclassified the findings as confirmatory.
1011 VOIDED because the investigators tested 20 participants prior to registration. The publication reporting the results is in a journal with questionable peer-reviewing procedures. The paper’s five citations are all self-citations by Tressoldi.
1010 DISCONFIRMATION of two Confirmatory Hypotheses, according to the publication, although the Registry document specified only one.
According to the registration document (15th May 2014):
The planned number of participants and the number of trials per participant. We plan to recruit 34 participants who will contribute for three sessions each, for a total of 102 experimental sessions. The number of bits for second will be set to 200. 6. A statement that the registration is submitted prior to testing the first participant, or indicating the number of participants tested when the registration (or revision to the registration) was submitted. At the 10th of May 2014, we have recorded 30 experimental sessions contributed by 10 participants and 30 control sessions.
It is a serious irregularity that allows data collection on 10/34 participants to have been completed prior to registration or a pre-registered study. The study does not meet the standards of peer review accepted by the majority of scholarly journals. A further issue with this study is that its publication seems to have been hurriedly reviewed and is poorly described in all respects.
The publication carries the citation: Tressoldi P et al., (2014). Mind-Matter Interaction at a Distance of 190 km: Effects on a Random Event Generator Using a Cutoff Method. NeuroQuantology | September 2014 | Volume 12 | Issue 3 | Page 337-343.
The paper is listed by Google Scholar but not by PubMed. The article states that the manuscript was “Received: 3 July 2014; Revised: 14 July 2014; Accepted: 28 July 2014′ leaving only 11 days for review, revision, second review and final acceptance. That may be a world record for a study reporting a phenomenon that, if real, would revolutionize Science. The results section consists of 8 short lines of text and one small table and are described so barely that it is impossible to evaluate exactly what was done to the raw data. The Results section states:
Descriptive and the inferential statistics for the number and percentages of experimental and control sessions when the cutoff was met, are reported in Table 2. The average duration of the sessions was 62 seconds, range 60-71, whereas the average time period during which the cutoff was achieved, was approximately 34 seconds, range 10-66.
Table 2. Descriptive and the inferential statistics for the number and percentages of experimental and control sessions when the cutoff was met.
|Condition||N (% out 102)||Effect size d [95% CIs]||Bayes Factor H1/H0*|
|Mental interaction||84 (82.3)||0.97 [0.73, 1.21]||7.3×1011|
*=using the same priors of the pilot study
1008 DISCONFIRMATION of two Confirmatory Hypotheses.
The citation for the publication reads: Tressoldi PE, Pederzoli L, Bilucaglia M et al. Brain-to-Brain (mind-to-mind) interaction at distance: a confirmatory study [version 3; peer review: 1 approved, 1 not approved]. F1000Research 2014, 3:182. Latest published: 23 Oct 2014, 3:182.
Following peer review by two independent reviewers, F1000Research has three versions of the paper. The final outcome was that only one reviewer approved publication while a second reviewer adamantly did not approve. This reviewer stated: “Unfortunately, while they carried out some of the analyses we discussed, the evidence still does not support their claims, even though the claims have now been toned down.” Although the paper remains listed in Google Scholar it is not listed by Pub Med.
On this basis, it can be concluded that the findings did NOT support the hypothesis of mind to mind interaction at a distance.
Both hypotheses were disconfirmed. In an earlier post, I inadvertently misclassified the findings as confirmatory.
1001: TWO STUDIES FAIL TO REACH THE CRITERIA FOR CONFIRMATION
Publication 1 of 2: F1000
Tressoldi PE, Martinelli M and Semenzato L. Pupil dilation prediction of random events [version 2; peer review: 2 approved with reservations]. F1000Research 2014, 2:262 (https://doi.org/10.12688/f1000research.2-262.v2)First published: 02 Dec 2013, 2:262 (https://doi.org/10.12688/f1000research.2-262.v1)Latest published: 09 May 2014, 2:262 (https://doi.org/10.12688/f1000research.2-262.v2)
Version 1: peer review 1 approved 1 approved with reservations. Version 2 peer review: 2 approved with reservations. Thus, a lower level of approval occurred for Version 2 than for Version 1 because the reviewers remained concerned about the statistical analysis, which one claimed was circular. The peer reviewers’ critical comments do not allow the conclusion that the hypothesis was confirmed.
Publication 2 of 2: SSRN
Tressoldi, Patrizio E. and Martinelli, Massimiliano and Semenzato, Luca, Pupil Dilation Predictive Anticipatory Activity: A Conceptual Replication (February 9, 2014). Available at SSRN: https://ssrn.com/abstract=2393019 or http://dx.doi.org/10.2139/ssrn.2393019
Results were at chance level. Thus, the experimental hypothesis was disconfirmed.
Summary Table of Tressoldi’s Findings
|1013||no||no||yes||Investigators changed hypothesis after data collection|
|1011||no||no||yes||Prior testing of 20 Ss|
|1009||no||no||yes||Prior testing of approximately one third of Ss|
|1002||no||no||yes||Not a confirmatory study|
A Loose End
A previous post included a single confirmatory study by somebody other than Tressoldi that offers partial support to the psi hypothesis. This is study ID 1026 by Dr. David Vernon.
A test of reward contingent precall (C)
|David Vernon||19th Oct 2016||1026||KPU Registry 1026|
Study Results Summary
KPU 1026 data
KPU 1026 data definition PARTIAL
There were two Confirmatory Hypotheses:
HA1 = Precall effect: Post recall practise of images will lead to greater recall of those images
compared to those not practised.
HA2 = Contingent reward: Those offered a contingent reward of £10 will exhibit greater levels of
precall (precall score – baseline score) compared to those not offered a reward.
It should be noted that both hypotheses predict the direction of the difference and so a one-tailed t test is permissable (α <.10). According to Vernon’s report on the RPE website:
A repeated measures t test comparing Precall to Baseline scores showed that the level of accuracy for the Precall condition was significantly higher than the Baseline condition (respective means: 5.78 vs. 5.22), t(98)=2.352, p=0.021, 95% CI (0.0836, 0.987), d=0.32.
The study was later published in the Journal of Parapsychology:
Vernon, D. (2018) Test of reward contingent precall. Journal of Parapsychology, 82
(1). pp. 8-23. ISSN 0022-3387..
There is a need to correct for multiple testing because Vernon carried out two t-tests. The correction can be made using a Bonferroni Correction by adjusting the alpha (α) level to control for the probability of committing a type I error. The formula for a Bonferroni Correction is: αnew = αoriginal / n. For Vernon’s two Confirmatory Hypotheses α should be .025. Unfortunately the link to the data for study 1026 is broken and so I have been unable to check the statistics that Vernon ran. I note that the difference between the two means is approximately one-third of a standard deviation so the difference between the two conditions appears to be marginal and could have been a type I error (see below).
A Second Loose End
Another study by David Vernon contained two Confirmatory Hypotheses, study ID 1046: An implicit and explicit assessment of morphic resonance theory using Chinese characters, registered by David Vernon, 23rd May 2018, 1046, KPU Registry 1046, Study Results Summary, KPU 1046 data
KPU 1046 data definition
A reviewer who, for now, I will call ‘Dr X’, has suggested that one of Vernon’s hypotheses was confirmed:
For registration 1046, two confirmatory hypotheses were preregistered. One was reported as significant in a negative direction and the other test was not significant. The tests were preregistered as two-tailed and therefore one was a significant outcome. Your summary had the study as unsuccessful.
Confirmatory hypothesis for the implicit preference task: Participants will prefer (i.e., select) real Chinese characters at a level significantly greater than chance (i.e., 50%). Vernon observed that participants implicitly preferred real Chinese characters less than would be expected by chance. Observing the exact opposite to what the hypothesis predicted means that the hypothesis was DISCONFIRMED. Dr X’s claim that finding the exact opposite to a Confirmatory Hypothesis is a “significant outcome” appears a bit bizarre.
Confirmatory hypotheses for the explicit identification task: Participants will identify real Chinese characters at a level significantly higher than chance (i.e., 50%). This hypothesis was DISCONFIRMED.
Parapsychology’s laboratory studies of psi are hanging by a single thread, a study by Vernon on the precall of images (study ID 1026). This is an interesting study but it isn’t a life saver – it’s more likely a Type I error.
Avoidance of Type I Error
In my previous post about the non-existence of laboratory psi, I reviewed 27 confirmatory studies listed at the RPE. Allowing for the fact that a few of these studies tested two Confirmatory Hypotheses, we can say that, in round figures, there have been roughly thirty Confirmatory Hypotheses investigated in these 27 studies. Given that all of them have been testing a single hypothesis about the existence of laboratory psi, there is a significant risk of Type I error, i.e. falsely rejecting the null hypothesis. As noted above, to minimize the risk of such an error, it is necessary to make a Bonferroni Correction to the alpha value used as the criterion for rejecting the null hypothesis. With 30 different statistical tests, an appropriate alpha level is defined by the formula: αnew = αoriginal / n, i.e. .05/30 = .00167.
Taking this correction into account, it is necessary to conclude that David Vernon’s p value for study 1026 of p=0.021 fails to reach statistical significance. It is concluded that Vernon’s one and only confirmed hypothesis is most likely attributable to a Type I Error.
My review of Patrizio Tressoldi’s 11 confirmatory hypotheses on laboratory psi reveals a bleak picture of Parapsychology. The overall quality of the studies can be viewed as nothing other than extremely poor. For 4 of 11 hypotheses listed on the Koestler Parapsychology Unit’s data-base it has been necessary to void the studies for one of three reasons: pre-registration collection of a portion of the data; changing the hypothesis after collection of the data; or, lacking any statistically significant exploratory finding as a basis for confirmation. Not one ‘confirmatory study’ provides statistically significant evidence that psi is real. For 7 of 11 hypotheses a clear and resounding disconfirmation is demonstrated. Considering the entire data-base of confirmatory studies, none of the findings was statistically significant at the corrected significance level of p < .00167. The 100% confirmatory failure across the entire gamut of laboratory studies is nothing less than catastrophic. Parapsychology has lost its way and a more fruitful approach is necessary. Ideas about this will follow.
|NO.|| STUDY TITLE||LEAD AUTHOR||DATE SUBMITTED||STUDY ID||LINK TO REGISTERED INFORMATION||REVIEW OF FINDINGS|
|PT11||PsyPhotos (C)||Patrizio Tressoldi & Luciano Pederzoli||3rd Oct 2019||1054||KPU Registry 1054||NO PUBLISHED FINDINGS.|
|PT10||Mind-matter interaction at distance on a standalone device (C)||Patrizio Tressoldi & Luciano Pederzoli||21st Feb 2019||1049||KPU Registry 1049 Published Results DISCONFIRMED||Tressoldi, P. E., Pederozoli, L., Prati, E., & Semenzato, L. (2020). Mind Control at Distance of an Electronic Device: A Proof-of-Concept Preregistered Study. Journal of Scientific Exploration, 34(2), 233-245. https://doi.org/10.31275/20201573 On Google Scholar but not PubMed. The published report differs from the registration document in the number of hypotheses and the measures to be used. The registration document specifies one confirmatory hypothesis: “the percentage of triggered electronic signals emitted by the MindSwitch during the periods of mental interaction at distance, will exceed that observed prior and following this interaction.” The published report specifies two confirmatory hypotheses as follows: “a) the samples obtained during distant mental interaction contain a higher number of data that exceed the probability cutoff of the Frequency or Runs tests of non-randomness and/or b) that the means of the absolute differences between the zeros and ones is greater during the mental interaction than in the preinteraction and the control phases.” Note that the change in the specification of the dependent variable from “the percentages of triggered signals” in the registration document to a) and b) above. There was no empirical support for either of the two hypotheses.|
|PT9||Telephone telepathy, an Italian independent exact replication (C)||Patrizio Tressoldi & Rupert Sheldrake||3rd Nov 2018||1048||KPU Registry 1048||NO PUBLISHED FINDINGS.|
|PT8||Can our mind emit light? A confirmatory experiment of mental interaction at distance on a photomultiplier (C)||Patrizio Tressoldi & John Kruth||6th Jul 2015||1013|| KPU Registry 1013|
Published Results VOIDED OWING TO THE INVESTIGATOR(S) CHANGING THE HYPOTHESIS AFTER REGISTRATION OF STUDY
|Tressoldi, Patrizio E. and Pederzoli, Luciano and Ferrini, Alessandro and Matteoli, Marzio and Melloni, Simone and Kruth, John, Can our Mind Emit Light? Mental Entanglement at Distance with a Photomultiplier (July 1, 2015). Available at SSRN: https://ssrn.com/abstract=2625527 or http://dx.doi.org/10.2139/ssrn.2625527 On Google Scholar but not PubMed. On August 15, 2015, the investigators changed the hypothesis after the data had been collected, which invalidates the conditions of a preregistered study. In addition, a re-analysis by a post-publication peer reviewer indicates that an incorrect statistical analysis was carried out by Grote (2017; https://www.neuroquantology.com/article.php?id=1699) who indicated incorrect statistical assumptions by the authors.|
|PT7||Biophotons as physical correlates of mental interaction at distance: a new confirmatory study (C)||Patrizio Tressoldi & John Kruth||29th Apr 2015||1012||KPU Registry 1012|
Published Results AMENDED TO DISCONFIRMED
|SAME PUBLICATION AS FOR 1013. Two confirmatory hypotheses: The number of photons detected by the PMT in the 30 minutes after the MI will outperform those detected in the 30 minutes before the MI. These differences will hold subtracting the number of photons in the corresponding 60 minutes of the control sessions.|
|PT6||CardioAlert: A portable assistant for the choice between negative or positive random events (C)||Patrizio Tressoldi||17th Mar 2015||1011||KPU_Registry_1011|
Published Results VOIDED OWING TO TESTING OF 20 PARTICIPANTS PRIOR TO REGISTRATION
|SSRN: Tressoldi, P. E., Martinelli, M., Torre, J., Zanette, S., & Duma, G. M. (2015). CardioAlert: A heart rate based decision support system for improving choices related to negative or positive future events. Available at SSRN 2604206. On Google Scholar but not PubMed. Paper has 5 citations, all self-citations by Tressoldi.|
|PT5||Biophotons as physical correlates of mental interaction at distance: a confirmatory study (C)||Patrizio Tressoldi & John Kruth||8th Oct 2014||1010||KPU_Registry_1010|
Published Results DISCONFIRMED
|SSRN: Tressoldi, P. E., Pederzoli, L., Ferrini, A., Matteoli, M., Melloni, S., & Kruth, J. (2015). Can our Mind emit light? Mental entanglement at distance with a photomultiplier. Mental Entanglement at Distance with a Photomultiplier (July 1, 2015). Appears on Google Scholar but not PubMed. According to the Registry document, there was one confirmatory hypothesis: The number of photons detected by the PMT in the experimental sessions will outperform those detected in the control sessions with an expected standardized effect size of 1.5 and a raw difference of 0.3 photons x second. According to the publication of the study, there were two pre-registered confirmatory studies. NEITHER HYPOTHESIS WAS CONFIRMED.|
|PT4||Mind-matter interaction at distance on a random events generator (REG): a confirmatory study (C)||Patrizio Tressoldi||15th May 2014||1009||KPU_Registry_1009|
Published Results VOIDED OWING TO SOME DATA COLLECTION PRIOR TO REGISTRATION AND THE SUBSTANDARD REPORT OF FINDINGS.
|Tressoldi P et al., (2014). Mind-Matter Interaction at a Distance of 190 km: Effects on a Random Event Generator Using a Cutoff Method. NeuroQuantology | September 2014 | Volume 12 | Issue 3 | Page 337-343. Appears on Google Scholar but not PubMed. This paper was rapidly reviewed and resubmitted (in 11 days) and accepted 14 days later: Received: 3 July 2014; Revised: 14 July 2014; Accepted: 28 July 2014. One-third of the data collection had already been completed prior to registration. The quality of the publication is sub-standard.|
|PT3||Brain-to-brain (mind-to-mind) interaction at distance: a proof of concept of mental telecommunication (C)||Patrizio Tressoldi||23rd Apr 2014||1008||KPU_Registry_1008|
Published Results REVISED TO DISCONFIRMED
|Tressoldi PE, Pederzoli L, Bilucaglia M et al. Brain-to-Brain (mind-to-mind) interaction at distance: a confirmatory study [version 3; peer review: 1 approved, 1 not approved]. F1000Research 2014, 3:182 Latest published: 23 Oct 2014, 3:182 (https://doi.org/10.12688/f1000research.4336.3) The paper is listed on Google Scholar but not by Pub Med. F1000 Research has three versions of the paper following peer review by two independent reviewers. The final outcome was that one reviewer approved publication and a second reviewer did not approve. This reviewer stated: “Unfortunately, while they carried out some of the analyses we discussed, the evidence still does not support their claims, even though the claims have now been toned down.” On this basis, it is concluded that the findings did NOT support the hypothesis of mind-to-mind interaction at a distance.|
|PT2||Pupil dilation prediction of random negative events. Can they be avoided? (C)||Patrizio Tressoldi||1st Feb 2013||1002||KPU_Registry_1002|
Published Results VOIDED BECAUSE IT WAS AN EXPLORATORY STUDY NOT A CONFIRMATORY ONE.
|VOIDED According to the registration document, this study tests a single confirmatory hypothesis, namely that pupil dilation can predict and avoid potential negative stimulation. SSRN: Does Psychophysiological Predictive Anticipatory Activity Predict Real or Future Probable Events? This paper is published in EXPLORE. It states: “Experiments 1 and 2:The first two experiments are conceptual replications of studies by Tressoldi et al.,2, 3 using heart rate (HR) as PAA, instead of PD. “Experiments 3 and 4:The following two experiments are a variant of the experiments of Tressoldi et al.3 The only difference being that the negative event predicted in the anticipatory phase was skipped instead of presented. Comparing the results with the previous experiments and the following ones, it is possible to test further the “bilking paradox,” that is, whether it is possible to avoid predicted future negative events, giving more support to the results observed in the experiment 2.” “Experiment 4 This is an exact replication of the experiments by Tressoldi et al.,3 aimed at testing if the observed prediction accuracy holds even when the alerting stimuli get skipped when predicted from the measurement of the PD before their presentation.” Note that reference 3 is from F1000: Tressoldi PE, Martinelli M, Semenzato L. Pupil dilation prediction of random events [v2; ref status: approved with…] the disconfirmed study 1001. A study cannot be confirmatory when the study being replicated did not itself find any significant outcomes and was not fully approved for publication by the reviewers at F1000 (see below).|
|PT1||Pupil dilation accuracy in the prediction of random events (C)||Patrizio Tressoldi||26th Nov 2012||1001||KPU_Registry_1001|
(1 of 2)
1)DISCONFIRMED Published Results
(2 of 2) 2) DISCONFIRMED.
|TWO STUDIES Publication 1 of 2: F1000 Tressoldi PE, Martinelli M and Semenzato L. Pupil dilation prediction of random events [version 2; peer review: 2 approved with reservations]. F1000Research 2014, 2:262 (https://doi.org/10.12688/f1000research.2-262.v2)First published: 02 Dec 2013, 2:262 (https://doi.org/10.12688/f1000research.2-262.v1)Latest published: 09 May 2014, 2:262 (https://doi.org/10.12688/f1000research.2-262.v2) On Google Scholar but not PubMed. Version 1 (2 Dec 2013) peer review 1 approved 1 approved with reservations Version 2 (9 May 2014) peer review: 2 approved with reservations. Thus, a lower level of approval occurred for Version 2 than for Version 1 because the reviewers remained concerned about the statistical analysis, which one claimed was circular. In this author’s opinion, the peer reviews of this study do not support the conclusion that the hypothesis was confirmed. Publication 2 of 2: SSRN Tressoldi, Patrizio E. and Martinelli, Massimiliano and Semenzato, Luca, Pupil Dilation Predictive Anticipatory Activity: A Conceptual Replication (February 9, 2014). Available at SSRN: https://ssrn.com/abstract=2393019 or http://dx.doi.org/10.2139/ssrn.2393019 On Google Scholar but not PubMed. Results were at chance level. The experimental hypothesis was disconfirmed.|