From mind to matter: data analysis challenges psychokinesis.

Scientists have long considered claims that people can manipulate the physical world with their minds. Yet numerous experiments conducted over the past 35 years, in which people try to influence the output of computers that generate random sequences of 1s and Os, have overall failed to demonstrate the existence of so-called psychokinetic effects, according to a new analysis. Some individual experiments, however, have seemed to indicate an impact.

The random-number studies were inspired by controversial mid-20th century investigations in which volunteers attempted to affect how a rolled die would land. In a typical recent experiment, a participant mentally tries to influence a computer to produce, say, more 1s than 0s over a predefined sequence. The participant receives visual cues that give constant feedback on hits and misses.

Participants shift their focus from 1 to 0 on alternating runs to account for any tendency of the digit device to produce one number slightly more often than the other.

Data collected from 380 random-number studies gleaned from journals, conference proceedings and reports, dissertations, and book chapters were analyzed by psychologist Holger Bosch of University Hospital Freiberg in Germany and his colleagues. The results appear in the July Psychological Bulletin.

In the majority of the random-number experiments, there are fewer than 100 participants, who try to influence sequences that total between 1,000 and 10,000 numbers, Bosch's team says. Taken together, these small-scale studies, which yield positive results more easily than large-scale investigations do, show a small but significant tendency for pre-designated 1s or 0s to appear. Typically, the effect is 1 to 2 percent above the distribution predicted by chance.

However, that effect disappears when data from three larger investigations are added to the mix, the scientists assert. Each of those studies assessed efforts to influence sequences totaling tens of millions of numbers.

The scientific literature on psychokinesis seemed to support the credibility of the effect, the investigators say. However, small studies, which are more likely than large ones to show statistically significant effects, are much more common, they note.

Large-scale studies, which rely on high-speed random-number generators and quicken the pace of volunteers' responses, seem to interfere with any psychokinetic effects, contends a group of psychokinesis researchers led by Dean Radin of the Institute of Noetic Sciences in Petaluma, Calif. Small-scale studies have yielded "a genuine psychokinetic effect," they say in the same Psychological Bulletin.

Psychokinesis researchers need to go beyond the statistics and explain how the mind might influence a computer, then test that prediction, say psychologists David B. Wilson of George Mason University in Manassas, Va., and William R. Shadish of the University of California, Merced in the same journal issue.