readings> psi - the micro-PK results
Does the mind have paranormal powers? After the ganzfeld experiments, the micro-PK work of Robert Jahn at Princeton is the psi community's best hopes of providing evidence. However, fingers are being pointed at a worrying aspect of the results.
In the lobby of the Flamingo Hilton, Las Vegas, slot machines - one armed bandits - stretch in serried ranks to the far horizon. Hanging over the machines nearest the entrance stands a sign stating: "97.4 - the hottest slot percentage in town." With characteristic Yankee bluntness, what the Flamingo's management are telling customers is that the machines have been adjusted so they will cream off "only" 2.6 cents of every dollar they spend.
This promise of steady, if unspectacular, loss is supposed to draw in the punters from the gritty heat of the Las Vegas strip. And attract them it does. So impatient do many seem to shed their money at the guaranteed rate that they can be seen feeding pairs of machines with either hand, shovelling in coins and barely waiting for the clacking reels to come to rest.
If ever proof were needed against the existence of telepathy, psychokinesis, precognition or any other form of psychic power, the gambling halls of Las Vegas seem to provide the perfect, living experiment. The odds on every game of chance - from the slot machines and crap games to the blackjack and roulette tables - have been fine-tuned to fractions of a percent. Judging by the faces masked in concentration, it can hardly be said that the gamblers are not exerting every psychic effort to win. And yet still the cash flows into the pockets of the casino owners in an even, predictable stream.
Despite such everyday evidence, people continue to believe in the power of the mind. Public opinion polls commonly find that as many as a quarter of the population are convinced that they personally have experienced premonitions or moments of telepathic understanding. Belief in the psychic seems impossible to shake. What appears needed is the perfect experiment, a foolproof laboratory test that could settle the matter once and for all, either revealing believers to be dupes or forcing sceptics finally to start taking mental powers seriously.
The dream of a conclusive test has led parapsychology - the science of psychic research - to experiments which, in fact, mirror the very games of chance which have made the gambling industry so profitable. Under tightly-controlled conditions, subjects try to influence the outcome of a random event such as the roll of a dice, the radioactive decay of an atom, the diffraction pattern of a beam of light, the fall of a cascade of polystyrene balls, or the "direction" taken by electrical noise.
What is more, many parapsychologists claim to be seeing an anomalous effect. They are reporting a deviation from chance which is vanishingly small - just a tenth of a percent - but when measured over millions of trials, this faint effect multiplies into a hugely significant distortion of the apparent odds.
Already, a few researchers are leaping ahead to the day when this slight "micro-PK" effect (for microscopic psychokinesis) will be harnessed for commercial application. Dean Radin of the University of Nevada (ironically, situated in Las Vegas) says: "It may be a small difference, but if we can find a way of amplifying it, we could build thought-controlled switches. Perhaps in 50 years we will be using psi to open our garage doors or change channels on our TVs!" Radin even claims that several household name electronics companies are, at this moment, quietly sponsoring further research into the possibility. But before becoming entranced by visions of an effortless, thought driven world, just what is the scientific status of the micro-PK work?
Psychic experiments with random systems date back to at least the 1930s. But most of the early research relied on dice or mechanical devices which, because of slight imperfections of manufacture, could never be truly random, and which were also rather susceptible to fraud. Reviews of this work showed that the tighter the controls, the less likely an experimenter was to report an effect.
In the 1970s, Helmut Schmidt of the Mind Science Foundation in San Antonio made a major advance with the introduction of experiments based on radioactive decay. Then in the 1980s, Robert Jahn, an engineering professor at Princeton University, New Jersey, began studies using the random white noise generated by an electrical diode.
Jahn's work is currently the most respected because of its scale and technical sophistication - although as was made plain when Jahn featured in a recent BBC2 TV series, "Heretic", his move into parapsychology has horrified Princeton's authorities. When Jahn, a rocket propulsion specialist, went public with his research in 1986, he was demoted from dean of the engineering faculty and left in no doubt that he would have been booted right off campus if it were possible.
Despite this rough treatment by fellow academics, Jahn - like most parapsychologists - is surprisingly open and helpful when questioned about his research. His first remark is that common-sense examples like gambling are not a particularly good argument against paranormal powers. Jahn points out that in group situations, such as race courses and roulette games, many people would be willing different outcomes and these are likely to cancel each other out. Then, of course, there are the management's wishes to consider. Another confounding factor is the possibility of "psi-missing" where some people might consistently get the opposite of what they urge. Finally, the size of the effect being argued for - just a tenth of a percent - is so small that it could easily be built into the odds on gambling devices like slot machines. Jahn believes questions about psychic ability can only be settled in the controlled conditions of the laboratory.
Jahn's basic experiment, which he has been running for 14 years, is simple. He built a random event generator which acts like an electronic coin toss. A thousand times a second, the white noise produced by a diode is sampled and its phase will produce either a positive or a negative value. On average, these readings should be split exactly 50/50. Jahn gets people to sit in front of the generator and will it to produce either more "heads" or "tails". The subjects - or operators, as Jahn calls them - can see how well they are doing from a cumulative plot rising or falling on a computer screen.
The most common criticism of random event generator experiments is that either the machine is probably not truly random in performance or that the recording of the results leaves too much scope for mistakes and even plain fraud. Jahn has gone to great lengths to counter these possibilities. The design of the random event generator does not seem to be in question. Measured over many days and millions of readings, its output has been perfectly well behaved - even to the point where it throws up the occasional "excursion" into apparently significant deviations from chance. If left to run long enough, a properly random system should sometimes stray quite a way from the mean and Jahn's box has produced the expected number of such excursions during its calibration trials.
The generator also has safeguards against tampering. Subjects are normally left alone with the box during trials and sceptics have suggested that its output could be affected by something as crude as it being given a kick, to more subtle effects like waving a magnet near it or even just leaning towards the machine and creating some sort of weak capacitance effect thought the static on a subject's clothing. To guard against such possibilities, Jahn has fitted the generator with various warning bells and temperature gauges.
But more importantly, the sampling method does not rely on the raw output of the noise diode. Instead, the definition of what counts as a head or tail is alternated with each trial, so a positive signal will be counted as a head on one trial, but a tail the next. This added layer of filtering would cancel out any inherent bias that the equipment might develop during the course of an experiment. Switching the polarity criteria a thousand times a second would also seem to rule out any deliberate, or even inadvertent, tampering by subjects.
As yet a further precaution, the performance of subjects is measured against three conditions. They have to move the numbers both up and down. There is also a control condition where they sit by the box, leaving it to perform on its own. Jahn says it is difficult to think what kind of equipment failure or environmental interference could change its direction as the subject had to switch between each of the three conditions.
The control over recording data seem equally stringent. One complaint against many earlier parapsychology experiments was that subjects could begin and end trials as they wanted. By recording trials that seemed to be going in the desired direction, and aborting sessions once they began to produce a downward turn on the excuse of having a headache or suddenly feeling uninspired, subjects could manipulate an experiment to create a result. But Jahn guarded against the perils of optional stopping by specifying the number of trials to be completed in advance and insisting that all results be recorded in the final database. In addition, the initiation of each session and the logging of results was controlled by computer software. Not only were results automatically dumped to tape, but the computer printed out a separate paper record and subjects wrote up their scores in the lab log-book.
With an apparently water-tight design, Jahn reported his first major batch of results in 1986 after completing a quarter million experimental trials (a trial consisting of 200 "coin-flips" in each of the three conditions). This was already several hundred times more data than collected by any other micro-PK researcher. But Jahn and the small team he assembled at the privately-funded Princeton Engineering Anomalies Research (PEAR) lab, in the basement of his engineering department, kept on going and by last year, Jahn had reached 14 million trials using over 100 different subjects.
In brief, the results he has found are tiny but highly significant. The effect size is about 0.1 percent, meaning that for every thousand electronic tosses, the random event generator is producing about one more head or tail than it should by chance alone. However, while microscopic, the effect is so constant that there is only a one in 5,000 chance that Jahn's result are a statistical fluke rather than some kind of anomaly.
So it seems like game, set and match to the parapsychologists. An experiment which was designed to meet all the standard criticisms of psychic research has come up with a steady, replicable result. Certainly Jahn's work appears to have put noted sceptics, like James Alcock of York University, Ontario, and Ray Hyman of the University of Oregon - both members of the self-appointed policing body, the Committee for the Scientific Investigation of Claims of the Paranormal (CSICOP) - on the backfoot. Yet a closer look at the detail of Jahn's findings raises some worrying questions - if only because if they turn out to be true, their ultimate implications seem too bizarre to be believed.
Since reporting his early results in 1986, Jahn has progressively extended the scope of his experiments. What he has found is that the anomalous effect appears astonishing insensitive to changing circumstances. For example, the effect size remains much the same when instead of testing the influence of subjects on a physical process - the random thermal movement of electrons across a transistor junction - Jahn asks them to disturb the output of a pseudo-random source. The pseudo-random number generator is just a repetitive mathematical calculation, so it would seem that the mind is as effective at influencing arithmetic as a real- life event.
The effect size also appeared constant when Jahn tested subjects with a random mechanical cascade. This device is a pin-ball machine, looking rather like a giant version of the popular Japanese arcade game, Pachenko, in which 9,000 polystyrene balls are dropped through a grid of nylon pegs, bouncing and skittering to collect in bins at the bottom. In an unbiased system, the balls should end up with a classic Gaussian bell-shaped distribution. But Jahn claims that when subjects sat in front of this ten foot high "macro-PK" device, they were able to produce slight deviations to the side.
More implausibly still, the effects on all three systems seemed impervious to distance and time. Over the past few years, Jahn has reported the results of large scale trials in which 30 operators have attempted to influence the devices from as far away as Kenya, New Zealand, England and Russia. Subjects would sit down for an hour at an agreed time and try to alter output according to a prearranged pattern. The results seemed to show no fall-off of effect with distance, being much the same whether the subjects were a mile away from the Princeton lab or right across the other side of the world.
In a further batch of trials using the same remote operators, Jahn asked them to make their efforts at a completely unrelated time - up to several days before or after the running of the machine. If anything, says Jahn, the effect was slightly stronger under such extreme conditions.
Jahn is not perturbed by such a pattern of results. He says that on the face of it, it might seem that if psychic powers existed, they should be strongest when subjects are closest to the equipment. Also it seems likely that feedback on success rates and the kind of device being used should have an effect. However Jahn believes that micro-PK is misnamed as what is actually happening is not a mental interference with a physical event but something much more subtle - a distortion of the very laws of statistics themselves. Jahn thinks that what subjects bend is the probability envelope of an outcome and so time and distance do not count.
Jahn has written about how such a view ties in with a quantum mechanical view of consciousness in a book, Margins of Reality (Harcourt Brace, 1987), co-authored with his lab manager, Brenda Dunne. Jahn says consciousness, like quantum systems, appears to have both a "particle" and a "wave" aspect. Consciousness is at its most concrete and particle-like when involved in ordinary rational thought, but becomes fluid and wave-like when thinking is creative and holistic. Extending the analogy, Jahn argues that just as the wave aspect of quantum systems allows them occasionally to penetrate physical barriers - a phenomenon familiar to microelectronics as quantum tunnelling, in which particles can be made to hurtle insulated junctions - so the mind might be able to reach beyond the brain and have a faint, resonant influence on the surrounding world.
Sceptics, however, treat such talk as mumbo-jumbo. They point out that, for a start, statistics are something that emerge after the fact. The "laws" describe the behaviour of random processes, they do not create them. Instead, sceptics see the surprising insensitivity of the claimed PK effect as being rather fishy. It suggests there is something else going on, a basic flaw in the experimental set-up, which is responsible for the significant results.
Suspicions have hardened as sceptics have looked more closely at the fine detail of Jahn's results. Attention has focused on the fact that one of the experimental subjects - believed actually to be a member of the PEAR lab staff - is almost single-handedly responsible for the significant results of the studies. It was noted as long ago as 1985, in a report to the US Army by a fellow parapsychologist, John Palmer of Durham University, North Carolina, that one subject - known as operator 10 - was by far the best performer. This trend has continued. On the most recently available figures, operator 10 has been involved in 15 percent of the 14 million trials yet contributed a full half of the total excess hits. If this person's figures are taken out of the data pool, scoring in the "low intention" condition falls to chance while "high intention" scoring drops close to the .05 boundary considered weakly significant in scientific results.
Sceptics like James Alcock and Ray Hyman say naturally it is a serious concern that PEAR lab staff have been acting as guinea pigs in their own experiments. But it becomes positively alarming if one of the staff - with intimate knowledge of the data recording and processing procedures - is getting most of the hits.
Adding fuel to the controversy, sceptics have pointed to the strange behaviour of the baseline condition results. Theoretically, the baseline condition should show the same gently wandering pattern as the calibration trials with occasional excursions into areas of apparent significance. Given the number of baseline trial that have been run, the scoring should have broken through this statistical envelope at least half a dozen times by now. Instead, the baseline result has stuck unnaturally close to a zero deviation from chance.
In noting these results, Jahn himself has remarked that what makes the situation even odder is that when the baseline statistics and the high and low scores are all added together, the abnormally wide variance of one nicely cancels out the abnormally narrow variance of the other to create a well-behaved Gaussian distribution. It is almost as if the extra hits found in the high and low scores had been taken from what would otherwise have been outliers of the baseline condition.
Alcock says this is exactly the sort of pattern that might be expected if some sort of data sorting had been going on. If just a handful of extreme baseline trials had been wrongly identified as high or low trials - or, alternatively, a few middling high and low trials had been reassigned to the baseline pool - then it would be easy to create an apparently significant result. Given an effect size of just one in a thousand, it would not take many such swaps to distort Jahn's results.
Little of this speculation has been discussed openly by CSICOP members - to do so would be virtually to accuse Jahn's lab of fraud and sceptics admit they certainly have no proof of that. Alcock also stresses that Jahn is widely respected and such alterations need not be deliberate, they could happen as the result of honest mix-ups.
Jahn, however, says he is well aware there has been a whispering campaign and he welcomes the chance to put the record straight. With candour, Jahn says no experimental design can ever rule out fraud. But he believes that the recording procedures at the PEAR lab are unusually tight and any fiddling with results would have to be systematic, including as it would the lab's computer database, the paper print-outs and subjects' entries in the lab's log-book. Jahn adds that sceptics have had a long-standing invitation to check his work at first-hand and the few that have dropped by seem to have left relatively impressed.
Jahn admits that operator 10 - whom he insists must remain anonymous - has been responsible for a large proportion of the lab's significant findings. But he makes two points. Firstly, at least four or five other of the 100 subjects show a more powerful effect than operator 10. What is difference is that they have been involved in far fewer trials. Jahn says if these better performers had been able to do as many runs as operator 10 - and if the strength of their effects persisted - then operator 10's results would have dropped away into the background.
A second point is that when the contributions of all the operators are graphed, they form a smooth continuum. Just as there are a few high performers like operator 10 at one end of the spectrum, so there are an equal number of poor performers - even psi-missers - at the other end who drag the overall numbers down. With over 100 subjects, statistically speaking there would have to be a few high-end scorers like operator 10, so no sinister conclusions should be drawn from the fact.
As to the "too perfect" baseline, Jahn says this fits in neatly with his argument that what subjects are doing is bending statistics rather than having a direct influence on physical events. It seems that, in the short term, subjects can pull the scoring in one direction. But this has to be balanced by a shortfall in later extreme scores. The statistics can be stretched but not broken.
However in the end, says Jahn, sceptics will always be able to dismiss positive results from a parapsychology experiment. Suspicions of fraud, faulty machinery or plain mistaken recording of data can never be completely countered. Jahn says the only way forward is to have the same experiment replicated by other labs. For this reason, he has recently built a cheap, solid-state version of his random event generator and over the past year, he has been farming them out to other interested investigators.
Yet even replications may not be the answer given the strength of entrenched views. Hardened sceptics are just as likely to find reasons to suspect a successful replication. And, of course, the same doubts work the other way. If a scientist gets a negative result (see below) then it may be the parapsychologists who start talking about incompetence and faulty procedures. Recent experience seems to show that there may never be a simple, conclusive test of the existence of psychic powers.
However, Jahn's work does seem to narrow the boundaries somewhat, for if such abilities exists, then their effects appear microscopically small. They also seem quite bizarrely resistant to the constraints of time, place and logic. Knowing what science is not looking for, at least is knowing something.
a different experiment fails
While successful parapsychology experiments grab attention, failures to find a result rarely get any press. However one recent experiment - modelled closely on Robert Jahn's micro- PK studies - is worth mentioning.
Stan Jeffers, a physicist at York University, Ontario, says his curiosity was piqued when he stumbled upon an old report of Jahn's research. He says Jahn's methodology sounded impressively solid and inspired him to mount his own parapsychological research. Soon he discovered that CSICOP member, James Alcock, worked at the same university, and this was a great help in producing a strong experimental design.
Jeffers's idea was to test people's ability to bend a beam of light and so distort the interference pattern created as it passed through a diffraction slit. Jeffers says the experiment was a straight optical analogue of Jahn's polystyrene ball cascade, except that because he used photons, subjects were dealing with "zillions" of events per second and so he expected any effect to show up quickly.
After testing over 80 people - including martial artists and self-proclaimed psychics - Jeffers found only chance results. He even took his apparatus down to Princeton in case it was the particular ambience of Jahn's lab that helped produce effects. Jahn himself admits that he expected Jeffers's experiment to work and was puzzled when it did not. Jahn has since lent Jeffers one of his new miniature random noise generators and Jeffers is planning further investigations.