Quantum Measurement as Pragmatic Information Transfer: Observer Effects on (S)Objective Reality Formation

Maier, M. A., Dechamps, M. C., & Rabeyron, T. (2022). Quantum Measurement as Pragmatic Information Transfer: Observer Effects on (S)Objective Reality Formation. Journal of Anomalous Experience and Cognition, 2(1), 16–48. https://doi.org/10.31156/jaex.23535

AI Assessment

A large, preregistered micro-psychokinesis study testing whether intention can bias a quantum random number generator, with an unusual central prediction: the effect should appear only when scientific objectivity is reduced. With 898 participants, the authors found strong Bayesian evidence for an intention-congruent bias in the condition where the objective computer record was never inspected and then erased, so the measure was the participant’s recalled count (Bayes factor 17.14), while the conditions that kept an objective record showed evidence for no effect. A follow-up confirmed the pattern: the bias appeared only where data were permanently erased, not where they were restored. This is exactly what the guiding theory predicts, but it is also its central vulnerability: an effect that exists only in subjective memory and vanishes whenever an objective record is kept is precisely the pattern that memory bias, demand, or the recall-order confound the authors themselves found could produce. This audit reports what the study found and how; it takes no position on whether micro-psychokinesis is real.

Provenance

DOI. 10.31156/jaex.23535 · Journal of Anomalous Experience and Cognition 2022, 2(1), 16 to 48.

Study type. A preregistered, within-participants online micro-psychokinesis experiment, framed by von Lucadou’s Model of Pragmatic Information and a proposed “sobjective” reality lying between the subjective and the objective.

Authors. Markus A. Maier (Ludwig-Maximilians-Universität Munich), Moritz C. Dechamps, and Thomas Rabeyron (University of Lorraine).

Data availability. Preregistered on OSF (osf.io/cr42j, with a second preregistration at osf.io/tsz5p) and data deposited on OSF (osf.io/nveaq).

Source basis. Every figure below is taken from the article’s own Abstract, Highlights, and Results.

What the paper reports

The study conceptualizes quantum measurement as a transfer of “pragmatic information” that occurs when an intentional observer perceives a motive-relevant, quantum-based outcome, and it predicts, following the underlying theory, that any resulting bias can only be documented under conditions of reduced objectivity.¹ In practice this means the micro-psychokinesis effect is expected to show up when the objective computer record is not inspected and is erased (leaving only the participant’s memory), and to disappear when a full objective record is kept. The paper reports exactly that pattern and interprets it as evidence for intention-based reality formation under reduced objectivity.

An intentionally congruent bias in quantum-based outcomes was observed using subjective memory data from the observations when additional objective computer-stored data were not inspected and finally erased … Quantum randomness was confirmed in a maximum objective data collection context.

How it was run

• Task. An online experiment (jsPsych) in which a quantum random number generator (Quantis by ID Quantique) determined a positive or negative image on each trial, preceded by a subliminal priming sequence. Each participant ran 40 within-subjects trials, 20 under maximum objectivity (C-max) and 20 under reduced objectivity (C-reduced).
• The objectivity manipulation. In the reduced-objectivity condition the objective computer-stored outcomes were not inspected and were finally erased, so the analyzed measure was the participant’s recalled count of positive images; the maximum-objectivity condition retained the stored record.
• Analysis. Bayesian sequential t-tests (Cauchy prior centered at 0.05, r = 0.05) with a preregistered stopping rule at a Bayes factor of 10. Data were collected in October 2020 by the Kantar polling company.
• Follow-up (Study 2). A second, preregistered “eraser” analysis used a pseudo-random split of the C-reduced subsample into cases whose data were permanently erased versus restored.

Results, as reported

Values are reproduced from the article’s Results. The evidence for an effect is confined to the reduced-objectivity, erased-data conditions and is measured on recalled counts; every condition that preserved an objective record showed evidence for no effect, and the raw deviations are tiny (about 10.58 to 11.08 positive images against a chance value of 10).

Eleven-dimension audit

Pre-registration

A genuine strength: the study is preregistered twice (the main design and a second eraser analysis), with data on OSF. The main caveat is that a recall-order confound was identified only after the fact, which means the primary reduced-objectivity result is not entirely clean of unplanned influences.

Randomization

The positive/negative image on each trial is set by a quantum RNG, and the objectivity conditions are varied within participants. In Study 2 the erased-versus-restored split used a pseudo-random generator. The randomization of the quantum source itself is confirmed (the maximum-objectivity data show no deviation).

Sensory leakage

Not applicable in the ganzfeld sense; the deeper concern is measurement source. Because the effect lives in the participant’s recalled counts under erased-data conditions, the analogue of leakage here is memory bias and demand, which are not excluded and are compatible with the disclosed recall-order confound.

Blinding

The experiment is automated and online, and participants are not told which theoretical outcome is predicted. But in the decisive condition the dependent measure is the participant’s own memory of the trials, which no automation can make objective.

Optional stopping

Sequential Bayesian testing with a preregistered stopping rule (Bayes factor 10) is a principled way to handle stopping, and preregistering it is the right approach. It remains a sequential design, so the stopping rule’s preregistration is doing important work.

Outcome measure

This is the crux. The measure that shows the effect is the recalled count of positive images under reduced objectivity; the objective stored record shows no effect. An outcome that appears only in subjective recall, and specifically only when the verifiable record is destroyed, is inherently harder to trust than one that survives objective measurement.

Effect size

The Bayesian evidence is strong (Bayes factors of 15 to 27 in the erased conditions), but the raw effect is minute, a recalled mean of roughly 10.58 to 11.08 against a chance value of 10. Strong Bayes factors on tiny deviations reflect the large sample as much as the effect.

Multiple comparisons

Several conditions and a follow-up split are analyzed, but the structure is preregistered and the predictions are directional and theory-derived, which constrains fishing. The pattern (effect only under erased reduced-objectivity) is consistent across the main study, the eraser follow-up, and the robustness check.

Internal replication

The eraser follow-up (Study 2) reproduces the main finding within the same dataset, and the paper is described as a partial replication of the group’s earlier Dechamps et al. (2021) work. Both are same-group internal replications rather than independent ones.

External replication

No independent replication is reported; the supporting prior work is from the same laboratory. Given the extraordinary and theory-specific nature of the claim, independent replication by neutral groups is the decisive missing step.

Transparency

High: double preregistration, open data, a preregistered follow-up analysis, robustness checks, and explicit disclosure of the recall-order confound. The authors do not hide the fact that the effect is confined to reduced-objectivity conditions; they present it as the theory’s central prediction.

The adversarial record

• The effect exists only where it cannot be objectively checked. The decisive objection is structural: the bias appears only in recalled counts under conditions where the objective record is erased, and disappears whenever that record is kept. This is precisely the signature that memory bias, demand characteristics, or the disclosed recall-order confound would produce, and it is very hard to distinguish those mundane explanations from the proposed one when no verifiable record exists.
• Near-unfalsifiability. The guiding theory predicts that objective measurement will show no effect, so every objective null is counted as confirmation. A framework in which objective disconfirmation is expected rather than damaging is difficult to test in the ordinary scientific sense, and a skeptic will regard that as the theory’s core weakness rather than its strength.
• Same-group support, tiny effects. The replications are internal and from the same laboratory, and the raw deviations are minute (about half a positive image above chance), with the strong Bayes factors driven substantially by the large sample.
• What is genuinely strong. Methodologically this is careful work: preregistered twice, open data, a clever preregistered eraser follow-up, robustness checks, a validated quantum RNG whose randomness is confirmed under objective conditions, and honest disclosure of the confound. The honest bottom line is a well-executed study whose positive evidence is inseparable from the very condition, destroyed objective records, that makes it hardest to interpret.