What happens when you scale the PEAR idea — that human consciousness influences physical random number generators — from a single laboratory to the entire planet? That is exactly what Roger D. Nelson, senior researcher at the PEAR lab at Princeton, did in 1998: he distributed dozens of quantum random number generators around the world and let them run continuously. Whenever something happens on the planet that simultaneously captures the attention of millions — a terrorist attack, a natural disaster, a World Cup final — the generators measurably deviate from expected randomness. The Global Consciousness Project (GCP) has formally analysed over 500 such events. The cumulative result stands at p ≈ 10⁻¹² — a chance probability of one in a trillion.
From PEAR to the GCP
Roger D. Nelson (b. 1940) joined the PEAR lab in 1980 as an experimental psychologist and quickly became one of its most productive researchers. He worked on the REG experiments, the Remote Perception studies, and published alongside Robert Jahn and Brenda Dunne in the Journal of Scientific Exploration. In the 1990s Nelson began what he called FieldREG experiments: instead of having one operator sit alone before the generator, he placed REGs within groups — at concerts, rituals, theatre performances, meditation circles. The question: does the generator respond to the collective consciousness of a group?
The FieldREG results were encouraging enough for Nelson to take the next logical step: what if the network went global? In 1998 the Global Consciousness Project went online — a network of initially 40, later up to 70 quantum random number generators (internally called "Eggs" — for "Electrogaiagram"), distributed across all continents, generating 200 random binary bits per second around the clock and streaming the data to a server at Princeton.
How the GCP works
Each "Egg" is at its core a REG of the same type PEAR used — a circuit that produces a binary random sequence from the quantum noise of a noise diode. The Eggs are hosted by volunteers: universities, private individuals, research institutions. They run continuously — not just during particular events, but always. This is methodologically crucial: there is no possibility of selectively switching the data on.
The analysis follows a fixed protocol. Before any data analysis, a "formal event" is defined: time window, hypothesis direction, statistical test method. Nelson and his team developed a standardised registration system for this: what counts as an event? How wide is the time window? Is a direction of deviation predicted, or merely any deviation? This formalisation was a direct answer to the charge that time windows were chosen post hoc to fit the data.
September 11, 2001
The event that brought the GCP to worldwide attention was September 11, 2001. Nelson's analysis, published in 2002 in the Journal of Scientific Exploration, showed: the combined data from the entire Egg network deviated significantly from random expectation on September 11 — and not only from the moment of the first impact (08:46 local time) but already in the hours before.
This last point is the most contested and at the same time the most interesting. Nelson defined the formal time window for September 11 as the entire 24 hours of that day — a decision he justified by arguing that the emotional impact of the event was not confined to the minutes of the attacks but engulfed the world's population for the entire day and beyond. Critics counter that a 24-hour window with continuously running generators can include random fluctuations that have nothing to do with the event.
The debate around September 11 is typical of the methodological tension within the GCP: the more precisely one defines the time window, the stronger the effect — but the greater the suspicion that the definition was fitted to the data post hoc. The more generously one sets it, the more robust the protection against that charge — but the more diluted the signal becomes.
500+ events, one cumulative result
By the formal conclusion of GCP-1 in 2015, the project had registered and analysed over 500 events. These include:
- Terrorist attacks (September 11 2001, Madrid 2004, London 2005)
- Natural disasters (Indian Ocean tsunami 2004, Haiti earthquake 2010)
- Global celebrations (New Year's Eve festivities worldwide)
- Major sporting events (football World Cup finals, Olympic opening ceremonies)
- Political moments (Obama election 2008, papal funerals)
- Meditation experiments (coordinated group meditations)
The cumulative result across all formally registered events stands at an overall deviation with a chance probability of p ≈ 10⁻¹² — roughly one in a trillion. Nelson and his co-author Peter Bancel published the summary analysis in 2011 in Explore: The Journal of Science and Healing. That is a Z-score in the same order of magnitude as the PEAR overall statistic (Z ≈ 7) — with a completely different experimental design.
Methodological criticism
The GCP has attracted legitimate methodological questions that must be taken seriously:
- Event selection: Who decides what counts as a "global event"? When hundreds of events are analysed, there is a significant risk of retrospectively selecting precisely those where the data happen to be striking. Nelson addressed this with the formal registration system: events were defined prospectively, before the data were inspected. Nonetheless, the procedure is not entirely proof against selection — the question of which events are even proposed as candidates contains a subjective element.
- Time-window flexibility: The choice of time window (1 hour, 6 hours, 24 hours, several days) significantly affects the statistics. The more degrees of freedom one allows here, the greater the risk of a false positive.
- Multiple comparisons: 500+ events, various statistical tests, various time windows — in so large an analytical landscape, correction for multiple testing becomes critical. Nelson argues that the cumulative statistic precisely provides this correction; critics such as Edwin May consider it insufficient.
These objections are serious. But they are not the same as a refutation. The cumulative finding is so strong that it remains significant even under conservative corrections. What it means is a different question.
GCP-2.0: the methodological tightening
In 2015 Nelson retired, and the GCP entered a new phase: GCP-2.0. The successor project, carried by an international team, addresses precisely the methodological criticisms:
- Pre-registration of all hypotheses and time windows before the event
- Fixed statistical testing framework with no post-hoc adjustment
- Automated event detection instead of manual selection
- Modernised hardware and data infrastructure
GCP-2.0 is still running. Early results indicate that the effect under the tightened conditions is smaller than in GCP-1, but does not disappear.
The Rhine–PEAR–GCP line
The GCP stands in an almost century-long research tradition: J. B. Rhine at Duke University (from 1930) investigated with cards and dice whether consciousness influences chance. Robert Jahn at Princeton (from 1979) replaced the dice with quantum random number generators and scaled the trials to 2.5 million. Roger Nelson (from 1998) scaled the generator into a global network and the "intention" from one person to the collective attention of humanity.
The question remained the same at every step: does physical randomness respond to consciousness? The methods grew sharper at each stage, the data sets larger, the effects smaller — but consistently present. That is exactly the pattern one expects with a real but small phenomenon: better controls reduce artefacts, but the core remains.
What the GCP means for the consciousness question
- Collective consciousness as a measurable factor. If the GCP data are taken seriously, then not only individual intention affects random generators (PEAR) but also collective attention — without intention, without a test subject, without a protocol. The generators respond to something in the air.
- Non-locality. The Eggs stand on different continents yet show correlated deviations. This is not explicable by classical physics — it points toward a non-local connection between consciousness and physical randomness, as Pauli speculatively formulated in his correspondence with Jung.
- The noosphere as an empirical concept. Nelson uses Teilhard de Chardin's concept of the "noosphere" — a planetary layer of consciousness — for the GCP results. What was philosophy in Teilhard becomes a statistical dataset in the GCP. Whether the interpretation holds is open. That the data exist is not.
The GCP has not proved that "global consciousness" exists. It has delivered something more modest and at the same time more unsettling: a dataset that conventional statistics struggle to explain, and that forces the question whether consciousness and the physical world are more closely entangled than standard physics assumes. Anyone who considers this question absurd must explain the data. Anyone who considers the data artefactual must refute the methodology — not wholesale, but point by point. Neither has been accomplished so far.
Sources: Roger D. Nelson, "Coherent Consciousness and Reduced Randomness: Correlations on September 11, 2001", Journal of Scientific Exploration 16 (4), 2002, pp. 549–570. Roger D. Nelson & Peter A. Bancel, "Effects of Mass Consciousness: Changes in Random Data during Global Events", Explore: The Journal of Science and Healing 7 (6), 2011, pp. 373–383. Roger D. Nelson, Connected: The Emergence of Global Consciousness, ICRL Press, Princeton 2019. Robert G. Jahn & Brenda J. Dunne, Consciousness and the Source of Reality, ICRL Press, Princeton 2011. Holger Bösch, Fiona Steinkamp & Emil Boller, "Examining Psychokinesis: The Interaction of Human Intention with Random Number Generators – A Meta-Analysis", Psychological Bulletin 132 (4), 2006, pp. 497–523. GCP data archive and documentation at noosphere.princeton.edu.