The Princeton Engineering Anomalies Research (PEAR) laboratory experimentally explored the effects of human intention on the behavior of random physical systems for more than a quarter century. Most often this involved microelectronic random event generators (REGs) that produce a string of random binary samples, or bits, at a rate of 1,000 per second, in trials of 200 bits each. A human operator without any particular or noted “psychic” abilities attempted to influence the distribution of the output of the device as displayed on a computer monitor in accordance with his or her pre-recorded intentions (“higher” or “lower”). The generated data were then examined for correlations between the intentions and the output of the device.
PEAR’s “benchmark results,” which involved some 840,000 trials per intention by 91 different individuals over a 12-year period, showed statistically significant correlations between operator intentions and the mean counts of 200 sample trials. The likelihood of the degree of separation between “high” efforts and “low” efforts being due to chance alone was less than 5 in 100,000, an extraordinarily high degree of significance.
While most PEAR experiments involved REGs, other devices, such as a pendulum and a water fountain whose output was also random, were also used. Included among these was a random mechanical cascade (RMC) of small polystyrene balls bouncing randomly down among a set of dowels to end up in one or another of 19 plastic bins. Designated “Murphy,” this device yielded an output similar to that obtained with the REG and other devices. In the following excerpt from their forthcoming book, Consciousness and the Source of Reality (ICRL Press), Robert Jahn and Brenda Dunne describe Murphy and their experimental results:
Early in 1979, several months before the PEAR lab was formally established, we had an opportunity to visit the Museum of Science and Industry in Chicago. There we noticed and admired a large random mechanical cascade device, modeled after the well-known Galton desk design that demonstrated the development of random Gaussian distributions by the compounding of a multitude of binary events, or what is commonly known as the “bell curve.” As we stood before the apparatus and playfully attempted to encourage it to shift its distribution of cascading marbles to the right, we were amused and intrigued by the clearly right-shifted distribution it produced in response. During this time a group of school children looked on, listening in disbelief as their instructor, whose back was to the device, explained how it would always generate a properly centered normal curve. We decided on the spot that we needed to have such a machine in our new laboratory, and shortly thereafter we designed and built a version of it in our own engineering school machine shop.
Originally, this had seemed like a relatively simple task, but it actually took the better part of three years to
complete. During that period it presented an incessant sequence of technical challenges, confirming Murphy’s
Law “Anything that can go wrong, will,” and resulting in the device acquiring the affectionate nickname of
“Murphy.” It even surpassed that law by demonstrating a few things that couldn’t possibly go wrong, such as occasionally slicing some of the balls in half. There was one memorable occasion when after our technician had spent several months unsuccessfully trying to design a funnel system that would preclude the balls jamming in the bin counters, he was given warning that if the problem wasn’t solved more expeditiously “heads would roll.” As this message was being transmitted, the platform supporting the ball distribution suddenly collapsed
and all of the balls that were in the collecting bins crashed loudly to the floor of the storage reservoir, terrifying the poor man!
While our various microelectronic REGs, which permitted the rapid accumulation of large bodies of data, ultimately became the workhorses of the PEAR laboratory, it was Murphy, the random mechanical cascade (RMC), that turned out to be its most popular experimental device, and ultimately, its best public relations representative. Over the years, numerous TV producers were fascinated to film Murphy for their programs,
and he even appeared on the front page of the New York Times. In addition, many of the countless school children who visited PEAR used it as a model for designing their own experiments in probability. More than any of our other devices, Murphy took on a distinctly anthropomorphic character, and our operators usually addressed him by name. On one occasion, when the machine was down for repairs for a few days, one of his operators actually sent him a “get well” card.
Our earliest REG results had clearly posed the categorical question of whether similar phenomena could be demonstrated using a broader range of random processors, in particular with devices of macroscopic scale, and Murphy provided an ideal opportunity to address this. Ten feet high and six feet wide, the machine was mounted on a wall in the reception area of the laboratory, facing a comfortable couch. In operation, 9,000 precision-cast polystyrene balls, ¾" in diameter, trickled downward from an entrance funnel into a quincunx array of 330 nylon pegs, also of ¾" diameter, mounted on 2¼" centers. The balls bounced in complex random paths through the array, colliding elastically with the pegs and with other balls, ultimately accumulating in nineteen parallel collecting bins across the bottom. The fronts of the peg chamber and the collecting bins below it were made of transparent plastic sheets so that the cascade of balls and their developing distributions of bin populations were visible as feedback to the operators. After considerable empirical modifications to determine appropriate combinations of peg spacing, ball inlet arrangement, and material properties, the resulting distribution of ball populations in the collecting bins could be tuned to a good approximation
of a Gaussian distribution.
The entrance to each collecting bin was equipped with a photoelectric sensor that detected and recorded the arrival of each ball, and the growing populations of all bins were displayed on LED counters below each bin, and graphically on a computer terminal screen. The disposition of each of the 9,000 balls in every run was recorded on-line in an appropriately coded computer file that later could be accessed to yield a faithful reproduction of the complete history of all of the bin fillings for more detailed study, or to calculate statistical properties of the terminal distributions. In addition, a photograph of the distribution and LED-displayed bin counts was taken after every run. As with all PEAR experimental devices, extensive calibrations were performed to provide background statistical data and to explore possible sensitivities to temperature
and humidity, which were routinely measured and recorded before each run.
The experimental tripolar RMC protocol called for the operator, seated on the couch approximately eight feet from the machine, to attempt to distort the distribution of balls in the bins toward the right or higher numbered bins (RT), or to the left or lower numbered bins (LT), or to generate baselines (BL) with no conscious intention. These efforts were interspersed in concomitant sets of three runs, each lasting approximately twelve minutes. The hundreds of experimental data sets thus obtained displayed similar anomalies in their overall concatenations to those achieved in the REG studies, including strongly operator-specific patterns of achievement. Detailed tabulations and cumulative deviation and structural graphs of the results can be found in a number of the archival references. Above we include only the cumulative deviation plots of all data acquired in an extended sequence of these experiments.
Unlike the REG experiments, where theoretical baselines confirmed by calibration were available for comparison with the operator-generated data, the internal mechanics of the RMC were too complex to submit to detailed theoretical prediction. This forced us to utilize a differential criterion based upon comparison of the empirical means of the RT and LT distributions with the local baseline of the same experimental set. This strategy had the advantage of minimizing any spurious effects of short- or long-term drift in the machine
operation, but introduced the confounding possibility that an operator might inadvertently influence the empirical baseline distribution as well.
And indeed, as we examined the overall cumulative deviation graphs (previous page), plotted as RT – LT, RT – BL, LT – BL differences, an intriguing secondary anomaly appeared. Whereas it was abundantly evident that the overall RT – LT mean separation was statistically highly significant (ɛµ = 1.93, zµ = 3.89; ρµ = 5 × 10–5), it also displayed a curious asymmetry in the LT direction. Namely, virtually all of the compounding RT – LT anomalous deviation was attributable to the LT – BL separation alone; the RT and BL evolutions were statistically indistinguishable!
For some time we attempted to resolve this asymmetry empirically: operators changed their positions on the couch, closed and opened laboratory doors, and one mounted a mirror on the facing wall and observed the reflected runs. One even stood on his head, but to no avail! It was several years later, in the course of our study of gender differences, that it was discovered that this propensity was entirely attributable to the tendency of many of the female operators to produce baselines that were strongly shifted in the right-going direction, thus producing results that showed a significant deviation in their LT – BL efforts, but a null result in the RT – BL. A similar gender-related trend subsequently was found to prevail in several other PEAR experiments.
As in the REG experiments, the total number of runs conforming to the intended direction to any degree was found to be considerably higher than the chance prediction, so that once again we concluded that the overall patterns of anomalous mean shifts of the mean were being compounded from an overall accumulation of small individual anomalous effects. We also again observed some operator-specific dependencies of the results on the secondary parameters of the experiment, such as the time of day, the volitional vs. instructed assignment of run order, or whether the LED count display was on or off.
Perhaps of higher importance, however, was the similarity of many of the individual operator cumulative deviation patterns with those they demonstrated in the microelectronic REG experiments. Despite their inherently stochastic character, the evident gross similarities of their signatures had major implications for experimental design and theoretical modeling. Namely, although the observed anomalous effects were clearly operator-specific and in many cases condition-specific, they appeared not to be nearly so device-specific, a feature later confirmed over a much wider range of physical processes, scales, and energies. Thus, once again, it appeared that any direct influence of operator consciousness on these widely different physical processes, e.g. the flow of electrons in the REG noise diode, or the cascade of balls in the macroscopic RMC, are less likely to be direct dynamical mechanisms than more holistic interactions with the statistical information common to both these systems.
Finally, we might note that although the RMC differed substantially from the REG devices in its scale and physical process, it retained a certain quasi-digital character in the manner in which it generated information. Specifically, each falling ball, upon collision with a peg, might be diverted either to the right or to the left, and it was the compounding of these binary right/left options that primarily determined the terminal distributions in the bins. To be sure, in this machine the binary right/left probabilities were not simply .50/.50, since the balls did collide with one another as well, and therefore their subsequent trajectories were not at all uniform, but nonetheless, a synthetic binary quantification could be, and actually was, imposed in the analyses.
A further step in tracking the ubiquitousness of operator related anomalies, therefore, was to apply similar protocols to physical systems that were yet more analogue in character, even to those whose central random processes and outputs lent themselves to continuum representation. All of these experiments utilized similar tripolar protocols to those followed for their digital counterparts, and from this array of studies we broadened our conclusion that the specific character of the physical random sources employed was not a primary correlate of their anomalous responses.
When the PEAR laboratory closed in 2007, perhaps the most emotionally poignant moment was Murphy’s disassembly. He had played a vital role in our program, both in the valuable data he had produced and in his contribution to the laboratory’s physical and subjective ambiance. Fortunately, he has found a new home with an organization in California, Index Fund Advisors, whose staff seems to find him just as engaging and instructive as we did.
References
Benedict Carey. “A Princeton Lab on ESP Plans to Close Its Doors.” New York Times. February 6, 2007.
Brenda J. Dunne, Roger D. Nelson, and Robert G. Jahn. “Operator-related anomalies in a random mechanical cascade.” Journal of Scientific Exploration, 2, No. 2 (1988). pp. 155–179.
Brenda J. Dunne. “Gender differences in human/machine anomalies.” Journal of Scientific Exploration, 12, No. 1 (1998). pp. 3–55.
Brenda J. Dunne and Robert G. Jahn. “Experiments in remote human/machine interaction.” Journal of Scientific Exploration, 6, No. 4 (1992). pp. 311–332.
Brenda J. Dunne, York H. Dobyns, Robert G. Jahn, and Roger D. Nelson. “Series position effects in random event generator experiments.” Journal of Scientific Exploration, 8, No. 2 (1994). pp. 197–215.
Robert G. Jahn and Brenda J. Dunne. “The PEAR proposition.” Journal of Scientific Exploration, 19, No.2 (2005). pp. 195–246.
Robert G. Jahn is Professor of Aerospace Sciences and Dean, Emeritus
of Princeton University’s School of Engineering and Applied Science,
founder of the PEAR Laboratory, and Chairman of International Consciousness
Research Laboratories (ICRL ). Brenda J. Dunne holds degrees
in psychology and the humanities, was the manager of the PEAR
laboratory from its inception in 1979, and is currently President of ICRL .
