Discovery Lab Answers

Peter A. Bancel: Experiments and Data

Peter A. Bancel is a statistical analyst and parapsychology researcher affiliated with the Institut Métapsychique International in Paris. His work spans two distinct but related areas: long-duration statistical analysis of the Global Consciousness Project (GCP) and exploratory laboratory experiments testing psychophysical interactions with quantum systems.

The Global Consciousness Project: Statistical Analysis

Bancel’s most sustained body of work involves the GCP, a network of random event generators (REGs) running continuously since 1998 and tested for anomalous deviations during major world events. His role has been primarily that of a rigorous statistical analyst — and, notably, one whose conclusions evolved significantly over time.

Early and mid-period findings. An extended analysis covering approximately 236 events over nine years yielded a cumulative result favoring the GCP hypothesis by more than 4.5 standard deviations, with secondary analyses identifying correlations across the REG network at global distances [from the Bancel page on ESP-Nexus]. A subsequent examination covering roughly 500 events over seventeen years produced a cumulative Stouffer Z-score of approximately 7 — a statistically very large result [from the Bancel page and the Goal-oriented page on ESP-Nexus].

The interpretive reversal. Despite co-authoring the GCP’s principal statistical defenses in 2008 and 2011, Bancel eventually concluded that the data do not support a “global consciousness” field explanation. His published analysis determined that the pattern is consistent with a goal-oriented (GO) effect — meaning that psi-mediated selection by experimenters, rather than a physical influence on hardware, explains the anomalies. Eight algorithmically defined surrogate event categories (large earthquakes, full moons, registered plane crashes, rock concerts, World Cup matches, religious observances, and others) all yielded null results, which Bancel used to argue that the positive GCP result traces to the selection of events rather than to the events themselves [from the Goal-oriented page].

This connects to Decision Augmentation Theory (DAT), in which a psi faculty guides the experimenter’s conscious choices toward data segments that already deviate — an unconscious selection process rather than a physical disturbance of the REGs.

Roger Nelson‘s response. Nelson directly contested Bancel’s GO interpretation ^[4]. Nelson argued that several structural features of the GCP data — distance dependence, time-of-day variation, autocorrelation patterns, and orthogonal correlation statistics — are inconsistent with a simple goal-oriented experimenter effect. He characterized the GO model as, taken to its logical limit, unfalsifiable: if any anomalous structure found in data can be attributed to experimenter or analyst intention, the model absorbs all findings without explanatory constraint ^[4]. Nelson acknowledged that many of the structural findings are post-hoc secondary analyses, but contended that their logical coherence makes a GO explanation implausible ^[4].

Bancel’s reply to May and Spottiswoode. In 2011, Bancel responded to a separate DAT-based reanalysis of the GCP data by May and Spottiswoode ^[5]. In that reply, Bancel demonstrated that the DAT model is statistically rejected when a robust estimator (the Theil-Sen estimator, which has higher power than ordinary least squares for this application) and the appropriate regression variable are used — a methodologically careful rebuttal that predates his own later endorsement of a goal-oriented interpretation ^[5]. The trajectory across these papers is therefore not a simple shift from “believer” to “skeptic” but a sustained, technically detailed reexamination that produced an unexpected conclusion from within the analysis itself.

Laboratory Experiments: Entangled Photons

Bancel co-authored a series of five exploratory studies testing whether human intention could influence the degree of quantum entanglement in photon pairs ^[3]. These were conducted with Dean Radin and Arnaud Delorme across two laboratories — the Institute of Noetic Sciences (IONS) and the Institut Métapsychique International (IMI).

Design. Participants observed a real-time display of an entanglement-strength metric (S-value) derived from a quantum entanglement device (quED system) and were asked to mentally influence that metric. The primary hypotheses tested for increases in entanglement strength under intention conditions.

Results. Across the four laboratory studies, combining hypotheses H1, H2, and H3 (while accounting for test dependencies) yielded an overall p = 0.017 ^[3]. Results across individual hypotheses were p = 0.045, 0.090, and 0.043, respectively ^[3]. A feedback hypothesis (H4) was not supported, and control data tests were insignificant ^[3].

A stark discrepancy between the two labs is the central feature of the dataset: the IONS data showed a clear effect (p = 0.003 for the online study; p < 0.0002 for the three IONS lab studies combined), while the IMI data showed no effect ^[3]. The authors noted this as an open question — participant selection, differences in experimental environment, and unidentified procedural variables are all candidate explanations, but none was established.

These five studies are explicitly labeled as exploratory, and the authors themselves flagged participant selection as “more of an art than a science” at this stage ^[3].

Methodological Contributions

Beyond his own experiments, Bancel has published substantive methodological commentary. A 2024 paper in JAEX ^[1] examines an implicit micro-PK experiment by Jakob, Dechamps, and Maier, walking through their statistical chain of reasoning step by step. He identifies concerns about Bayesian prior justification, the treatment of correlated data sets (where treatments share structure, information about one group provides information about others), sensitivity checks, and errors of omission — cases where gaps in reported information leave the statistical chain incomplete ^[1]. Jakob, Dechamps, and Maier responded by arguing that statistical independence across their three micro-PK tasks was empirically documented (chi-squared tests showing no significant association between tasks, with 99% power to detect even small violations at w = 0.1), and that sequential Bayesian designs do not require multiplicity correction when independence holds ^[2]. The exchange illustrates a live methodological dispute about when Bayesian multiplicity control is necessary in psi research.

Skeptical Critiques

What critics argue. May and Spottiswoode (2011) argued that the GCP’s September 11, 2001 result — one of the project’s most cited findings — collapses when the chosen time window is shifted by only a few minutes: shortening or lengthening it returns the result to chance, which they interpreted as evidence that the window was effectively selected post-hoc to maximize the effect. Their proposed mechanism was DAT — experimenter psi as selective attention, not a physical field. Bancel himself later endorsed a version of this argument for the broader dataset.

What the experimental data show. Nelson’s response ^[4] argued that multiple independent structural features of the GCP data (not just the September 11 window) require explanation and are not easily absorbed by a GO or DAT model. Bancel’s 2011 reply ^[5] showed that DAT is statistically rejected under a robust regression estimator when the appropriate regressor is used — a result that Bancel himself produced before his subsequent shift in interpretation. The entangled-photons series ^[3] showed a significant laboratory result at one site and a null result at another, with no identified cause for the discrepancy.

Analysis. The most distinctive feature of Bancel’s contribution is that the sharpest methodological critique of the GCP’s field-consciousness interpretation came from within the GCP’s own analytic team. Nelson and Bancel used the same dataset and reached different mechanistic conclusions — Nelson pointing to structural features inconsistent with GO, Bancel pointing to surrogate-event nulls consistent with it. Independent replication of the entangled-photon effect outside the original two-lab setting has not been published in the retrieved sources.

For a deeper look at how these interpretive questions developed across Bancel’s work, the ESP-Nexus pages on Peter A. Bancel and the goal-oriented versus collective consciousness interpretation cover both the experimental record and the mechanistic debate in detail.