Dean Radin, Cassandra Vieten, Joseph Burnett, Arnaud Delorme, Tam Hunt
Have you ever experienced a time when the collective enthusiasm of a large event seemed to rise to such a peak that you could almost feel a crackle in the air? Or felt a haunting sense in the air while visiting a place that caused sadness or suffering for thousands of people? Provocative evidence suggests that there are significant departures from chance expectation in the outputs of random number generators (electronic devices that produce truly random bits, or sequences of zeros and ones) during times of collective upheaval, global crises and major celebrations.
This year, the Institute of Noetic Sciences, along with several collaborators, conducted an exploratory experiment at Black Rock City, the temporary city created each year in the Nevada desert for the festival known as Burning Man. Burning Man is a week-long event that attracts upwards of 50,000 people. It is unique in its concentrated intensity, isolation, and collective intention, culminating with the burning of a large man-shaped effigy at the center of Black Rock City on Saturday night. See this article in the Atlantic magazine to get a feeling for the event, or these pictures in Rolling Stone magazine.
Our experiment tested the prediction that a random number generator (RNG) placed on the playa would demonstrate significant deviation from randomness during the period of highest collective intensity, i.e., during the burning of the man. In addition, the Global Consciousness Project (GCP) made a prediction that their global network of random number generators would also show a deviation from randomness. That prediction was based in part on a previously successful exploratory analysis that examined the average of eight years of global RNG data at the time of Burning Man (1999 – 2006).
The underlying hypothesis in studies of this type, dubbed “field consciousness” experiments, is that mind and matter are complementary aspects of a more fundamental, holistic reality. As an analogy, the idea is that subjective mind and objective matter may be like heads and tails on a coin. These two aspects of nature may appear to be quite different from one other when examined separately, but from a broader perspective they may be seen as part of an intimate relationship by virtue of being connected to, or part of, the same “substance.” Based on this proposed relationship, when minds experience a period of unusually high collective coherence, such as during the burning of the man, then perhaps matter will also show a period of high coherence. The latter effect might be detectable in an RNG through an unexpected arising of statistical order. To test this idea, we look for a synchronicity: a meaningful event at a specific time. In the present case the synchronicity would appear as a non-chance fluctuation in an RNG output close in time to the burning of the man.
Hypothesis 1 was that an RNG placed on the playa would demonstrate a statistically significant deviation from randomness at the time of the burning of the man. Hypothesis 2 was that the Global Consciousness Project RNGs would demonstrate statistically significant departures from randomness at the same time. Hypothesis 3 was that the RNG placed on the playa might demonstrate departure from randomness at the time of a second major burning event known as the Temple Burn, on Sunday night.
Data were collected from Wednesday night August 29th, 2012 through Monday morning, September 3rd, 2012 using a Psyleron unit (a truly random number generator based on quantum randomness), with periodic small gaps in data collection due to brief power interruptions (the experiment was solar and battery powered in the middle of nowhere in the Nevada desert). Data were collected at a rate of 800 random bits per second. To compute the deviation from randomness, first we calculated a standard normal deviate (i.e., a z score) for each second of data (800 bits) , then we created a cumulative z score starting one hour before the ignition of the burning man to one hour after (i.e., a cumulative Stouffer z).
To evaluate Hypothesis 1 we asked the following question: How likely is it to observe the maximum odds against chance actually obtained within this two hour period within the time period between that maximum and the moment of the ignition? That is, say we observed a statistical deviation associated with odds against chance of 20 to 1 within two minutes of the moment of the ignition. Is that observation a meaningful deviation from chance given the data we collected, or is it consistent with chance?
To answer this question, we used a computational statistical method called a bootstrap. To do this we selected a 2-hour block of data randomly from within the entire data stream collected on the playa. We found the maximum odds against chance and the distance in time to the center of the segment (which represented the moment of ignition). If the odds against chance during the random two-hour block were equal to or larger than that observed during the actual burn, and the temporal distance was equal to or less than that observed during the burn, then we incremented a counter. This process was then repeated 5,000 times, each with a newly selected random block of data. At the end of this process the counter value divided by 5,000 provided the probability of observing a statistical deviation equal to or greater than the one we actually obtained during the experiment. The same method was used to evaluate the Temple Burn event.
Data from the GCP was analyzed similarly. The only change for the GCP data is that that database consisted of 55 RNGs running simultaneously at different locations around the world. To provide a single stream of data to apply to our bootstrap technique, the 55 GCP data streams were combined into a single z score by creating a composite chi-square (sum of z-squares with degrees of freedom equal to the number of elements in the sum) across all RNGs, per second, and then converting the p-value associated with each chi-square back into a z score. The resulting sequence of z scores was then evaluated using the same bootstrap technique described above.
Hypothesis 1. As shown in Figure 1, the RNG located on the playa showed a statistically significant deviation with modest odds against chance of 57 to 1, based on two-tailed probabilities, some 8 minutes after the moment of ignition (the flames died down about 10 minutes after ignition). The bootstrap analysis showed that this outcome, or better, would have occurred by chance with a probability of p = 0.004, confirming Hypothesis 1.
Figure 1. Two-tailed odds against chance and minutes to ignition of the burning man, for the RNG on the playa, along with photos of what this event looked liked.
Hypothesis 2. As shown in Figure 2, the combined GCP RNGs showed a statistically significant deviation with odds against chance of 9,461 to 1 about 45 minutes before the moment of ignition. The bootstrap analysis showed that this outcome, or better, would have occurred by chance with a probability of p = 0.002, confirming Hypothesis 2.