Soon Bem was led to a soundproof room, where he sat in a reclining chair with Ping-Pong ball halves covering his eyes and headphones delivering white noise. For the next half hour, as a sender in another room watched a one-minute clip on a TV monitor several times, he described the images that went through his mind. Later, shown four clips, including the one that the sender had watched, Bem ranked them in order of how closely they coincided with his own mental images.
Because he gave the second-highest rating to the clip the sender had watched, the trial didn’t result in a “hit.” But Bem was taken with the rigorous protocol. “This is quite sound,” he told Honorton. He also made an offer: “If you get positive results, then I have one major talent, and that is getting published in mainstream journals.”
Honorton continued to collect data until he had tested 240 participants, some more than once. He found that of the 329 individual sessions with these subjects, 32 percent produced a hit, significantly above the 25 percent that would be expected by chance alone. True to his word, Bem wrote up the results with Honorton in a paper they submitted to the peer-reviewed Psychological Bulletin. It was accepted days after Honorton died in 1994 and signaled the beginning of Bem’s career as a serious researcher of psi.
By the end of the 1990s, Bem had changed his focus from clairvoyance to precognition, the most mind-
boggling of psi phenomena. “It has the biggest wow factor,” he says. Although telepathy, or straightforward mind reading, is hard to believe, at least it seems remotely scientifically possible. Electromagnetic waves travel over vast distances, so perhaps there is some way the electrical impulses that generate thoughts could be transmitted from one person to another. Precognition is different. Sensing events that have not yet occurred requires that information move backward in time. “I thought, my god, that is fascinating,” Bem says, “because it means that our classical view of the physical world is wrong.”
So Bem devised a series of experiments to test precognition. In the simplest, subjects were asked to click on either of two curtains on a computer screen to find an erotic image hidden behind one of them. A computer program randomly assigned the image to one of the curtains only after the subjects had made their choice. Bem found the subjects chose correctly 53.2 percent of the time, notably higher than chance.
Another test was a flipped version of a memory experiment, a reversal of cause and effect. In the standard experiment, subjects see a list of words on a computer screen. Then they are shown another list containing half of those original words and are asked to type them. In the last step, all the words disappear from the screen, and the subjects are asked to type all the words they can recall from the full list. Not surprisingly, subjects do better at recalling words that they typed in the interim step.
In Bem’s version of the experiment, the subjects were first asked to type all the words they could recall from a list of 48 words shown on the screen. Then they were presented with half the words from the full list and asked to type them. Bem found the subjects were better at recalling the words in the first round that they would later be asked to type again in the second round. That is, the memory-reinforcing effect of typing the words seemed to work backward in time.
Yet another Bem experiment played with a phenomenon called priming. In a typical priming test, subjects are flashed a positive or negative word, such as beautiful or ugly, before being shown an image that they must judge pleasant or unpleasant. Studies show that subjects respond more quickly when the word and the image are congruous—that is, if both are pleasant or unpleasant. In Bem’s reversal of the procedure, the word was flashed after the subjects had judged the picture, yet the results were the same.
Bem ran five experiments on precognition (four of them twice), all variations of his priming and reversal design, testing more than 1,000 subjects. The positive effect detected in the studies was small—only about 3 percent greater than chance—but statistically significant nonetheless. A roulette wheel at a casino has a similar edge over players, in that the casino wins 53 percent and players 47 percent of the time. “And casinos are not complaining that that is too small,” says Bem. “They are making plenty of money with that edge.”
When Bem’s paper was accepted by the Journal of Personality
and Social Psychology, a hailstorm of criticism erupted in the normally measured field. Ray Hyman, by now an emeritus professor of psychology at the University of Oregon, denounced the work as “an embarrassment.” Some of Bem’s colleagues were impressed by the paper’s elegance and rigor but were still unwilling to take its claim seriously. “These are the best ESP studies I’ve ever heard of, and they make clever use of paradigms used in mainstream psychology,” says Tom Gilovich, the former chair of Cornell’s psychology department. “But I don’t believe the conclusions even for a second.”
The most extensive critique came from Eric-Jan Wagenmakers, a mathematical psychologist at the University of Amsterdam, who believes that Bem’s findings reveal fundamental problems in how statistics are applied to test ideas. In classical statistics, scientists evaluate whether their data fit the null hypothesis, in which a statement is negated or disproved. Specifically, Bem analyzed whether the results of his studies could be explained if precognition did not exist. The analysis found the odds of getting all of those results in a world without precognition, due to chance alone, were “one in about 74 billion,” according to Bem.
With this impressive-sounding result, Bem ruled out the null hypothesis, that precognition does not exist. Following the standard rules of hypothesis testing, he concluded that the alternate hypothesis—the reality of precognition—must therefore be valid.
Wagenmakers thinks this kind of analysis is misleading whenever it is applied in the social sciences, however, and especially when applied to an extraordinary claim like the existence of ESP. “Evidence is a relative concept, and although some data may be unlikely under one hypothesis, this does not mean that therefore we should accept the other,” Wagenmakers says. He analyzed Bem’s data using a different system, called Bayesian statistics, which compares how well data might fit both the null and the alternate hypothesis. In this analysis, Bem’s psi effect was no longer evident.
Bem quickly teamed up with two experts on Bayesian statistics and did his own Bayesian analysis of the data, which showed the psi effect was still intact. The problem with the rival analysis, Bem claims, was that Wagenmakers set a very high bar for how strong the psi effect had to be to meet the Bayesian test.
Bem himself has never experienced anything extrasensory, nor has he had any spiritual awakenings that might predispose him to belief in the paranormal. But just because psi lacks an obvious scientific explanation does not mean it does not exist, he argues, adding that the history of science is replete with examples of phenomena like electromagnetism that went unexplained for decades or centuries after they were discovered.