Academic journal article The Journal of Parapsychology

Evaluation of a Conventional Interpretation of Helmut Schmidt's Automated Precognition Experiments

Academic journal article The Journal of Parapsychology

Evaluation of a Conventional Interpretation of Helmut Schmidt's Automated Precognition Experiments

Article excerpt

Among the strongest bodies of evidence for psi anomalies are the ESP and PK experiments of Helmut Schmidt in which electronic random event generators (REGs) were used. Palmer (1985) found that 25 of 33 experimental series (76%) yielded results significant at the .05 level (two-tailed) as based on directional CRs(1) of the pooled trials in the series. Of the 29 series where the direction of the result could be determined, scoring was in the direction presumably intended by the subject in 21 of them (72%).

The first major critic of Schmidt's research was Hansel (1980, 1981), who focused primarily on the possibility of subject or experimenter fraud. These criticisms have been addressed by Rao and Palmer (1987). More recently, criticism has tended to focus on the randomness of the output of Schmidt's REGs. Schmidt routinely conducts and frequently reports the results of control tests with his machines, but these tests are not procedurally or temporally isomorphic with the experimental trials. This has led some critics, most notably Hyman (1981), to speculate that the machines might be susceptible during test sessions to short-term (local) biases that either did not occur in the control sessions or were not picked up by the statistical evaluations of those sessions.

A more specific version of this criticism has been proposed by Alcock (1990), who explains how subjects might have been able to take advantage of such possible local biases in two of Schmidt's early precognition experiments (Schmidt, 1969b). Subjects in these experiments pressed one of four buttons on the face of the machine to predict which target the REG would select for that trial. The REG then made its selection, and one of four colored lamps lit up revealing the target selected. Thus, subjects had immediate feedback after each trial. The machine also automatically recorded the number of trials and hits. Alcock wrote:

The subject was free to "play" with the equipment (with paper punch and nonresettable counters disconnected) and to decide when to start and stop a given session. If there were short-term biases in the generator that lasted, suppose, for ten minutes, and which were not detected during the randomization tests (which were of much greater length), these play sessions, the feedback, and the freedom to choose when to start and stop a session would provide magnificent opportunity to exploit, consciously or unconsciously (and most likely the latter), that bias. After all, the subject would want to begin a session, presumably, when it appears that he or she is "hot," while, if the subject's scoring rate declines, he or she may well want to end the session and start again later. The subject was given immediate feedback by means of a set of resettable counters (distinct from the nonresettable ones mentioned above) that displayed the number of trials and the number of hits. (Alcock, 1990, p. 133)

The possibility of local biases in test sessions cannot be conclusively eliminated in Schmidt's PK experiments because the hypothesis is that the test sequences themselves will be nonrandom. However, this problem may not apply to Schmidt's ESP experiments, where the target sequences would be expected to be random. In such cases, one does not need to resort to control sequences at all to resolve the dispute. Moreover, since Schmidt's ESP tests involved thousands of trials, tests can be conducted of sufficient power to demonstrate any biases strong enough to be detected and utilized by an ESP subject, and to reveal whether such biases were in fact so utilized.

I am grateful to Dr. Schmidt for granting me access to the raw data from two of the ESP experiments to which Alcock's criticism was directed (Schmidt, 1969a, 1969b). Both were precognition experiments in which subjects were asked to predict for each trial which of four lamps would light, as determined by the REG immediately after the subject's button-pressing response.

Three subjects (J. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.