Since seeing last week’s Through the Wormhole episode on precognition, I’ve been reading more on the subject. Valid criticisms have been raised: were the experiments that seemed to show precognition really better than random chance? Can we be sure of the honesty of investigators who claim to have shown that people can see into the future? More tests of precognition must exist; what do they show?
It turns out that, yes, hundreds of experimental tests of precognition have been conducted since the 1930s. Yet after reading Bem’s work and a meta-analysis of 309 other experiments, I’m still unsure what to believe. About 30% of studies on the subject show slight precognitive effects; the other 70% don’t. And yet the combined findings of the 300+ studies consisting of over 1 million trials still yield a statistically significant precognitive effect. If, that is, the statistical analyses of these trials can be believed.
Unfortunately, study of the subject is also riddled with difficulty from skeptics and believers alike. Incidents of fraud by precognition “investigators” have cast doubt on all studies of the subject; and prejudice from mainstream science has prevented respected journals from publishing or encouraging research on the subject.
Those who have brought to light legitimate problems with precognition work have undoubtedly done a great service to science. Serious incidents of outright fraud have been exposed, and so have subtler manipulations of statistics.
Yet the attitude of the accusers turns me off. One psychologist opened his criticism of Dr. Bem’s methods by declaring that he “didn’t believe a word” of the reported findings because the claim of precognition was too “outrageous.” He questioned whether precognition research should ever be published in major scientific journals, and the reasons he cited were not experimental evidence, but rather conventional wisdom.
It’s not unheard of for scientists to accuse others of fraud or error because their results are simply too strange. Sometimes, these accusations turn out to be right; such was the case when seemingly-compelling experiments showing precognition were found to be completely fake in the 1950s On the other hand, sometimes they’re terribly wrong; the first papers describing REM sleep were also accused of error and rejected by scientific journals, because they violated conventional wisdom.
And sometimes, disturbingly enough, fraud occurs and nobody catches it. We saw a horrifying example of that recently when prominent social psychology researcher Diederik Stapel admitted to having simply made up reams of false data, which he used to write “groundbreaking” papers that were published by leading psychological journals.
So how exactly do we determine who to believe?
That’s an especially interesting question in the field of precognition research. It turns out that dozens of researchers have tested over 50,000 participants for precognitive abilities over the course of the last century. And they’ve been published almost exclusively in journals with names like Journal of Parapsychology and Journal of Psychical Studies. Never in Psychology or Social Psychology, or even Physics (which already acknowledges ripples in spacetime created by matter; simulation of neutron-star induced ripples above by NASA), which would undoubtedly be effected by findings related to precognition. Why?
The immediate instinct of most people, seems to be that the topic is not worth examining. The justification seems to be this: because industries like insurance and gambling turn profits, clearly people cannot see the future. If they could, our whole society would be different. But is that really sufficient? Could not mechanisms of precognition exist that are so slight that they are often missed by our conscious minds, or are only effected by a certain class of events?
I must suspect that there is something deeper going on here. Perhaps scientists are bothered by the idea of something once thought to be supernatural, even if it has a plausible physical mechanism. Goodness knows the stranger theories of physics, though supported by excellent mathematics, are often met with discomfort by scientists in other fields.
But to decide anything, we need to evaluate the quality of the actual research on precognition that exists to date. Here’s a little more about Bem’s research, and the findings of 62 other precognition researchers who have never gained mainstream recognition:
There have been three attempts by other researchers to replicate Dr. Bem’s findings that study participants could predict the future locations of erotic images with better-than-chance accuracy. All have failed; the three other scientists who have used Bem’s methods have shown no precognitive effect at all.
More troubling still, another scientist did his own analysis of Dr. Bem’s numbers and found that even Bem’s own reported results may not have been outside the realm of chance at all. He found that Bem made some confusing choices about how to analyze his results, using methods that may have skewed the data.
The numbers Bem reported are real: 100 test subjects correctly guessed the future location of erotic images 53.1% of the time, while they guessed the future locations of non-erotic images only 49.8% of the time. But for the number of trials run, the 53.1%, this could just be a fluke happening; it could have greater than a 5% chance of occurring randomly, which would make it insufficient proof of anything.
These problems with Bem’s results cast a troubling light on the other studies of precognition. Or, perhaps those additional studies could shore up Bem’s work. In the mire of statistical conventions and accusations of fraud, it seems difficult to find the truth. (Image at right in U.S. public domain.)
Either way, let’s take a quick look at a meta-analysis of 20th century studies of precognition:
I had the great good fortune of finding a wonderful paper by Charles Honorton and Diane Ferrari, which analyzes 309 precognition studies for quality of experimental methods and significance of results. They even controlled for “publication bias” favoring studies with significant results, and included some subtle fraud-checking measures, like testing both studies and researchers for see for “outliers” with unusual findings.
The results are striking. Honorton and Ferrari found that about 30% of these 309 studies reported statistically significant precognitive effects, similar to those reported by Dr. Bem. 70% found no evidence of precognition. But despite the high rate of “failed” experiments, the findings as a whole–that is, the results of over 1 million trials using 50,000 participants– are still found to show a slight yet significant precognitive effect.
This means two things: for one, if researchers studying precognition were committing fraud, were still reporting honest failure to show evidence of precognition more often than not. Furthermore, because investigators reported similar rates of success and failure, if fraud was occurring, all 62 of the researchers studied must have been reporting false positives at approximately the same rate across the decades.
And from comparing these to Bem’s studies and the attempts to replicate them, we also know that, if funny statistical analysis methods were used, results that actually don’t support the existence of precognition could be made to appear to do so.
There are also some interesting trends to be gleaned from this research. All experiments seem to show that people are better at predicting near-future events than far-future ones; the nearer in time the event, the higher the chance of showing a positive precognitive effect. In addition, if the results are accurate, they suggest that some people are better at precognition than others, or at least that people have different kinds of precognition skills; studies that used test subjects selected for their high accuracy rate in previous trials found stronger precognitive effects in those groups.
These seemingly logical trends would not appear if all positive results were accounted for by statistical misinterpretation of random chance. If only random chance is going on, no group of test subjects should score consistently higher than any others; if only random chance is going on, the distance or proximity to the predicted event should make no difference in the accuracy of predictions. These findings suggest either concerted fraud, or a real precognitive effect. Which is it?
We don’t know. It would be devilishly hard to painstakingly comb through the records of the studies showing these trends, some of them many decades old. The consistency in the rates of reported successes make fraud seem less likely, but they do not rule it out; perhaps precognition researchers collectively figured out that there was an “acceptable” rate of positive results which would arouse interest without arousing suspicion. This is not beyond the realm of possibility.
So what’s the conclusion, after decades of research and over 1 million trials of precognitive abilities in humans? We still don’t know. The field has been stymied, both by fraud from supposed “investigators” and a complete unwillingness on the part of the mainstream scientific community to publish results or encourage investigation in this area.
If precognitive abilities do exist, they appear to be very weak within the average person, yielding little better than chance guesses at what the future holds. And it’s entirely possible that precognition doesn’t exist at all; that all positive results are in fact derived from funny math, fabrication, and pure chance. But the similar reports of slight precognition down the ages by dozens of different investigators are just enough to keep me, and the entire scientific community, guessing. More investigation is definitely warranted.
On a personal note, while I was researching this, two more people in my life came forward to report their own seemingly precognitive experiences. One former classmate of mine reported having felt distress and a feeling of impending doom on the morning of 9/11/2011, before the first plane had hit the World Trade Center.
Another friend recounted her belief that she once witnessed the same car crash twice. The first time she saw it, the passersby near the collision didn’t react; while she was looking around to try to figure out why no one seemed to notice it, the crashing sound came again. She turned around to see the exact same collision in progress again, and this time the people around her responded.
Anecdotal evidence is, of course, no proof of precognition. Given all the tricks the mind plays on us, it’s entirely possible that most people will at some point in their lives misremember having a precognition, just as most people healthy people will experience at least one hallucination not associated with illness. Perhaps my own feeling of impending doom around 9/11 was sheer coincidence; goodness knows such feelings of impending doom are one kind of hallucinatory experience. Perhaps what my friend with the car crash experienced was merely deja vu.
But the repletion of these types of experiences, and the powerful realism of them, keeps us guessing as it has in cultures throughout all of history (ancient Chinese fortune telling is being practiced in the image at right) And that’s a good thing for scientific research.
I’d be glad to hear of any research or relevant information that I’ve missed.