Roger D. Nelson, PhD Sources: off on

Mass Consciousness Events: September 11 and Global Emotional Coherence

On September 11, 2001, as billions of people around the world watched the same images of horror and grief, the Global Consciousness Project’s network of random event generators exhibited measurable deviations from expected randomness, deviations that Nelson and colleagues argued were correlated with the unprecedented global coherence of human attention and emotion. This spoke examines the September 11 analysis in detail, alongside earlier landmark events such as the deaths of Princess Diana and Mother Teresa, as case studies in what Nelson terms mass consciousness effects on physical random systems.

Deeper dives — Nelson:

Back to hub Healing Intention and Energy Medicine Consciousness-Correlated Anomalies Meta-Analysis and Statistical Methods PEAR Lab: Operator Intention and REG/RNG FieldREG and Collective Consciousness Global Consciousness Project: Network 📖 Connected

Key findings

On September 11, 2001, two formal pre-specified analyses of the GCP network yielded a combined Stouffer Z of approximately 4.01 (p ≈ 0.000027), indicating a highly significant departure from expected randomness correlated with the attacks and their immediate aftermath.¹
Five independent analysts examined the September 11 data post hoc, each finding consistent patterns of non-random structure concentrated in the hours of peak global attention and emotional reaction.¹
Nelson ruled out electrical disturbance and elevated mobile phone use as explanations for the September 11 anomaly through comparative analysis of control periods and network node behavior.¹
Multiple FieldREG devices recorded anomalous deviations during the week of Princess Diana’s death and funeral in 1997, providing an earlier landmark case for mass-consciousness effects on REG networks.²
Over a twelve-year GCP database, the cumulative formal statistic reached a 6-sigma departure from chance expectation, with the September 11 event contributing one of the largest single-event signals.³
The GCP 2.0 initiative, building on Nelson’s original framework, is extending the network to 4,000 RNG nodes with the aim of increasing sensitivity to mass-consciousness signals.⁴

Overview

A critical Type-II vulnerability in this research domain is that mass-consciousness effects, if real, are expected to be very small in absolute magnitude, the GCP literature consistently reports effect sizes in the range of fractions of a standard deviation across the full network, meaning that any single event analysis is substantially underpowered to detect the effect reliably, and that dismissal of individual null or marginal results would be premature without considering the cumulative multi-event database. Nelson’s research program addresses this by aggregating across hundreds of pre-specified events rather than relying on any single case, though the September 11 event remains the most widely discussed because its signal was among the largest observed.³

The Global Consciousness Project, which Nelson created in 1997 and has directed since, maintains a world-spanning network of hardware random event generators (REGs), also called random number generators (RNGs) or, informally, “eggs”, that continuously stream data to a central server in Princeton, New Jersey. The core hypothesis is that when large numbers of people share intense, synchronized attention and emotion in response to a global event, the normally independent outputs of geographically separated REG nodes become slightly but detectably correlated with one another.⁵ The September 11 analysis became the most prominent single-event test of this hypothesis because the scale of synchronized global attention on that day was arguably without precedent in the modern era.¹

GCP Network Architecture and the Mass-Consciousness Hypothesis

The GCP network as described in the September 11 paper comprised over 40 host sites worldwide, each running a hardware REG that generates a continuous stream of random bits using quantum tunneling processes. Data are transmitted to a Princeton server and archived. The formal hypothesis for any given event is that inter-node correlations, measured as the variance of the network’s composite output, will exceed chance expectation during the event window. The mechanism proposed by Nelson is that globally shared attention and emotion somehow impose a slight structure on otherwise independent random sources, though Nelson explicitly labels this interpretation speculative and notes that no physical mechanism has been identified.¹ The GCP 2.0 initiative is extending this to a network of 4,000 RNG nodes with more advanced quantum-random hardware, aiming to increase statistical sensitivity.⁴

The September 11 Analysis

The September 11, 2001 analysis stands as the most extensively documented single-event study in the GCP literature. Nelson conducted two formal, pre-specified analyses based on standardized GCP procedures for hypothesis specification and statistical evaluation, yielding a combined result that represented one of the largest single-day deviations in the network’s history at that point.¹ The non-random structure was concentrated in the hours corresponding to the attacks themselves and the period of peak global media coverage, rather than being uniformly distributed across the day.

September 11 Statistical Results and Formal Analyses

The two formal analyses of the September 11 GCP data yielded a combined Stouffer Z of approximately 4.01, corresponding to p ≈ 0.000027, a result Nelson described as a substantial departure from the null expectation of no network-level structure.¹ The network at the time comprised over 40 REG nodes distributed across multiple continents. The formal analyses tested whether the variance of the composite network output exceeded chance during the event window defined by the timeline of the attacks (approximately 8:45 AM to the collapse of the North Tower at 10:28 AM Eastern time, with extended windows also examined). Post hoc analyses by five independent analysts, examining the same archived data, consistently found non-random structure concentrated in the hours of peak emotional reaction, though these post hoc analyses are exploratory and cannot be treated as independent confirmations of the formal result.¹ The same data were also reported in a companion paper examining continuous random data correlations with major world events more broadly.⁵

A central methodological concern in any REG-network study is whether the observed deviations could be explained by environmental artifacts rather than consciousness-related effects. For the September 11 analysis, Nelson specifically examined two candidate non-psi explanations: electrical disturbances caused by the attacks and their aftermath, and elevated mobile phone use creating electromagnetic interference with REG hardware.¹

Artifact Testing: Electrical Disturbance and Mobile Phone Interference

To address the electrical disturbance artifact, Nelson examined the geographic distribution of the anomalous signal across network nodes. If the effect were caused by power-grid disruptions or electromagnetic events localized to the northeastern United States, nodes in Europe, Asia, and the southern hemisphere should show no anomaly. The reported finding was that nodes distributed globally, including those far from any plausible electromagnetic influence from the attacks, contributed to the anomalous signal, partially mitigating (though not fully eliminating) the localized-disturbance explanation.¹ For mobile phone interference, Nelson noted that the GCP REG hardware uses quantum tunneling sources that are shielded from the frequency ranges used by mobile networks, and that the timing of the anomaly tracked the emotional timeline of events rather than the periods of peak mobile phone congestion. These analyses are described as mitigating rather than fully eliminating the artifact concern, as no direct electromagnetic measurement was taken at node sites on that day.¹

Princess Diana, Mother Teresa, and Early Landmark Events

Before the September 11 analysis, the deaths of Princess Diana (August 31, 1997) and Mother Teresa (September 5, 1997) provided the first major test cases for the GCP’s mass-consciousness hypothesis. Multiple FieldREG devices were operating during the week of Diana’s death and funeral, and Nelson and colleagues reported anomalous deviations in the REG data correlated with the periods of most intense global mourning.²

Princess Diana and Mother Teresa, FieldREG Results

Nelson and colleagues reported that multiple FieldREG devices recorded anomalous deviations during the week of Princess Diana’s death and funeral in 1997, with the pattern described as showing “global resonance of consciousness”, a phrase used to describe the inter-node correlation structure.² A separate analysis of recordings during the same period examined the concurrent death of Mother Teresa, which occurred just days after Diana’s death, providing an unusual natural experiment in which two globally significant deaths occurred within the same week. The analyses were exploratory in nature, as the GCP’s formal prediction and analysis protocols were still being developed at this early stage of the project. Nelson also reported on multiple FieldREG recordings specifically during the Princess Diana events in a companion paper examining the network behavior in detail.⁶ These early analyses established the template for the event-based analysis methodology that would later be applied to September 11.

The Twelve-Year Database and Cumulative Evidence

Nelson has consistently argued that the evidential weight of the GCP program rests not on any single event, including September 11, but on the cumulative pattern across hundreds of pre-specified events analyzed over more than a decade. By 2011, the formal cumulative statistic across the full twelve-year GCP database had reached a 6-sigma departure from chance expectation, with the September 11 event contributing one of the largest individual signals in that database.³

Twelve-Year Cumulative Statistics and the Role of September 11

Nelson reported that the full twelve-year GCP database, analyzed through 2011, showed a cumulative Stouffer Z corresponding to approximately 6 standard deviations above chance expectation across all pre-specified events.³ The database at that point included hundreds of events ranging from natural disasters and terrorist attacks to major celebrations and sporting events. The September 11 event was among the largest single contributors to this cumulative signal. Nelson’s preferred interpretation is that the cumulative pattern reflects a genuine mass-consciousness effect, that globally shared attention and emotion produce detectable structure in otherwise independent random sources, but he explicitly labels the full interpretation as speculative and notes that no consensus physical mechanism has been proposed.³ Alternative interpretations include publication bias in event selection (events with larger signals may be more likely to be reported or discussed), multiple-comparisons inflation across the event database, and undetected systematic artifacts in the REG hardware or data transmission pipeline. Nelson addresses the multiple-comparisons concern by noting that all events in the formal database were pre-specified before analysis using a standardized protocol, partially mitigating (though not fully eliminating) the multiple-comparisons artifact.³

Modern Context

The GCP’s mass-consciousness research program sits within a broader landscape of consciousness and global-coherence science. The Global Consciousness Project 2.0, developed by the HeartMath Institute in collaboration with the Institute of Noetic Sciences, explicitly builds on Nelson’s original GCP framework, extending the network to 4,000 RNG nodes and incorporating measurements of Earth’s magnetic field environment alongside the RNG data stream.⁴ This successor initiative frames the original GCP’s demonstration that “global events and shared intentions could influence the correlation of random patterns” as the established foundation from which GCP 2.0 departs, while also introducing new measurement modalities and a larger, more sensitive network architecture designed to improve detection sensitivity for small mass-consciousness signals.⁴

Beyond the parapsychology-internal successors to Nelson’s framework, the GCP’s 9/11 finding intersects with mainstream methodological debates about event-based statistical inference. Retrospective analyses of single dramatic events face well-known inferential challenges, including multiple-comparison correction (the GCP analyses 500 pre-specified events and reports an aggregate test, but secondary post-hoc windowed analyses around 9/11 require explicit correction).⁷ Mainstream statistics has converged on family-wise-error or false-discovery-rate control (Benjamini and Hochberg’s FDR procedure being the modern standard) and, more broadly, on pre-registration of hypotheses and analysis pipelines as the structural defense against researcher-degrees-of-freedom in event-based studies.⁸⁹ A broader movement in the statistical literature has called for abandoning bright-line significance thresholds in favor of effect-size estimation with uncertainty intervals—an editorial reframing that bears directly on how small per-event GCP deviations should be communicated to a mainstream-science audience.

Skeptical Critiques and Discussion

Critique 1: The September 11 result reflects post hoc event-window selection and multiple implicit comparisons rather than a genuine pre-specified effect

Skeptic source: A recurring methodological concern in the GCP literature, applicable to the September 11 analysis, is that even formally pre-specified analyses require the analyst to define the event window, its start time, end time, and the specific network metric, before examining the data. Critics have noted that the choice of which hours to include in the “event window” for September 11 involves judgment calls that, if made after inspecting the data, would constitute implicit multiple comparisons. The companion paper examining continuous random data with major world events acknowledges that multiple analyses were performed on the September 11 data.⁵

Response: Nelson’s response, detailed in the September 11 paper, is that the two formal analyses used event windows defined by the objective timeline of the attacks, the first plane impact at 8:45 AM and the collapse of the North Tower at 10:28 AM, rather than by inspection of the REG data. The post hoc analyses by five independent analysts are explicitly labeled as exploratory and are not counted toward the formal result. The standardized GCP prediction protocol, applied across all events in the database, is designed to prevent window-shopping by requiring event parameters to be logged before analysis.¹ This partially mitigates the multiple-comparisons concern for the formal analyses, though the exploratory post hoc analyses remain susceptible to it.

Analysis. What is in dispute is whether the September 11 event window was independent of REG data inspection. The two formal analyses cite event boundaries defined by the objective attack timeline (first plane impact at 8:45 AM and North Tower collapse at 10:28 AM), with five independent analysts’ subsequent examinations explicitly labeled as exploratory and not counted toward the formal result. Whether the boundary-timing decisions for an unanticipated event were fully insulated from inspection of the underlying REG data is a question the published record does not exhaustively settle, and it remains a methodological discussion in the literature.

Critique 2: The cumulative 6-sigma result may reflect publication bias or selective event inclusion rather than a genuine mass-consciousness signal

Skeptic source: A structural concern with any long-running event-based research program is that the selection of which events to include in the formal database may be influenced, consciously or not, by prior knowledge of which event types tend to produce larger signals. If events with larger deviations are more likely to be nominated for the formal database, the cumulative statistic will be inflated relative to the true effect. This concern applies to the twelve-year cumulative analysis reported by Nelson.³

Response: Nelson addresses this concern by noting that the GCP maintains a public archive of all pre-specified events, with event nominations logged before data analysis, and that the formal database includes events with null or negative results as well as positive ones. The standardized prediction protocol is designed to prevent selective inclusion by requiring event parameters to be registered in advance. Nelson acknowledges that the full interpretation of the cumulative result remains speculative and that no consensus physical mechanism has been identified, but argues that the pre-specification protocol renders the cumulative statistic a valid measure of the overall evidence.³ The GCP 2.0 initiative is designed in part to address sensitivity limitations of the original network.⁴

Analysis. The selective-event-inclusion critique raises the question of whether the cumulative 6-sigma statistic reflects prior knowledge of which event types produce larger signals. Nelson cites the GCP’s public archive of pre-specified events, including null and negative-result events, and the standardized protocol that requires event parameters to be registered before analysis. The degree to which event nominators’ awareness of GCP signal patterns across event categories influences the formal database — a question that the published record alone cannot fully resolve — remains an active methodological discussion, and the GCP 2.0 initiative is in part designed to address related sensitivity questions.

References

Nelson, R. D., Radin, D. I., Shoup, R., & Bancel, P. A. (2002). Correlations of continuous random data with major world events. Foundations of Physics Letters, 15(6), 537-550. https://doi.org/10.1023/A:1023981519179 R001 [Nelson 2002] ↩︎
Nelson, R., Bösch, H., Boller, E., Dobyns, Y., Houtkooper, J., Lettieri, A., et al (1998). Global resonance of consciousness: Princess Diana and Mother Teresa. Electronic Journal for Anomalous Phenomena, eJAP, 22, 425–454. R002 [Nelson 1998] ↩︎
Nelson, R. D., & Bancel, P. A. (2011). Effects of mass consciousness: Changes in random data during global events. EXPLORE: The Journal of Science and Healing, 7(6), 373-383. https://doi.org/10.1016/j.explore.2011.08.003 R003 [Nelson 2011] ↩︎
Nelson, R. D. (2025). 20240627211415 The Path to Global Coherence The Role of the Global Consciousness Project 2.0. Journal of Management, Spirituality & Religion, 22(6), 696–714. https://doi.org/10.51327/udiy4331 R004 [Nelson 2025] ↩︎
Nelson, R., Radin, D. I., Shoup, R., & Bancel, P. A. (2002). Correlations of continuous random data with major world events. Foundations of Physics Letters, 15(6), 537–550. https://doi.org/10.1023/a:1023981519179 R005 [Nelson 2002] ↩︎
Nelson, R. (n.d.). Multiple Field REG/RNG Recordings during a Global Event, Princess Diana. Electronic Journal Anomalous Phenomena. R006 ↩︎
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B, 57(1), 289–300. https://doi.org/10.1111/j.2517-6161.1995.tb02031.x R007 [Benjamini 1995] ↩︎
Nosek, B. A., Alter, G., Banks, G. C., Borsboom, D., Bowman, S. D., Breckler, S. J., et al. (2015). Promoting an open research culture. Science, 348(6242), 1422–1425. https://doi.org/10.1126/science.aab2374 R008 [Nosek 2015] ↩︎
McShane, B. B., Gal, D., Gelman, A., Robert, C., & Tackett, J. L. (2019). Abandon statistical significance. The American Statistician, 73(sup1), 235–245. https://doi.org/10.1080/00031305.2018.1527253 R009 [McShane 2019] ↩︎

Deeper dives — Nelson: