ESP at Random

My review of Mind-Reach, by Russell Targ and Harold Puthoff—a book on the testing of clairvoyance—appeared in The New York Review of March 17. Shortly thereafter the NYR received an interesting letter from Aaron Goldman, Sherman Stein, and Howard Weiner, three top mathematicians at the University of California, at Davis. Although their letter does not deal with P and T (as Puthoff and Targ are called), it concerns the closely related work of Charles Tart, a colleague of the three mathematicians at Davis.

Tart’s reputation as a parapsychologist is even higher than that of P and T. When his latest book, Learning to Use Extrasensory Perception, was published last year under the imprimatur of the University of Chicago Press, it was widely hailed as a major breakthrough. (See The New Yorker’s “Talk of the Town,” December 13, 1976.) Tart shares with P and T the conviction that ESP powers can be markedly strengthened by electronic teaching machines. The major work of P and T in this field, made possible by an $80,000 grant from NASA, was with a four-choice machine designed by Targ. I discussed this test, considered a failure by almost everybody except P and T, in my Scientific American column, October 1975.

Tart’s work is more elaborate. He uses a ten-choice trainer, or TCT as he calls it, of his own invention. He believes it to be superior to the Targ machine. Indeed, his book contains severe criticism both of Targ’s machine and of the protocols of P and T in their NASA experiment.

This is how Tart’s TCT operates. A “sender” sits in one room in front of a console bearing a circle of playing cards from ace through ten. Beside each card is a button and a pilot light. An electronic randomizer selects a digit (zero counting as ten), then the sender pushes the button that turns on the light next to the selected card.

The “receiver” sits in front of a duplicate console in a room across the hall. A “ready” light informs him when a card has been chosen. After moving his hand around the ring of cards, searching for the “hot” one, he pushes a button to register his choice. This procedure is repeated in runs of twenty-five choices each, and twenty runs per test. Hits and misses are automatically recorded. There is no recording of the time at which any button is pressed.

As soon as the choice is made, a light beside the actual target card goes on to provide instant feedback. If the choice is a hit, a “pleasant chime” sounds inside the console. Above the console is a TV camera joined by cable to a TV screen above the sender’s console. On this screen the sender can see the hand of the receiver as he searches for the hot card. This is so the sender can concentrate more intently on where he wants the hand to stop, although Tart concedes there is no way to tell whether the receiver is getting the target information by telepathy or clairvoyance. Senders and receivers are usually college students, unpaid but sometimes given academic credit for their help.

Tart’s superstar was an unnamed girl who scored so high that the results are against normal odds of more than a billion to one. She worked with unusual slowness, taking about forty-five minutes to complete each run.

Now for the letter from the three mathematicians:

To the Editors:

Readers of the review of ESP research in the March 17, 1977 issue of The New York Review of Books may be interested in the details of an ESP experiment conducted by Charles Tart, our colleague at U. C. Davis. Much more easily analyzed, it concerns the transmission and reception of the ten digits….

[I have omitted a paragraph in which the TCT is described.—M.G.]

Ten receivers in Tart’s experiment had a total of 722 hits out of 5,000 trials. Statistically this is far above chance, which would be around 500 hits (since there is one chance in ten of guessing a digit). The simplicity of the experiment, together with our natural curiosity about ESP, led us to ask Tart for the raw data, which were even more spectacular than the sessions. One receiver had 124 hits in 500 trials, far above the expected fifty. This was truly astonishing.

Then, as we scanned the target digits produced by the random-number generator, we noticed that twins seemed to be sparse—that is, a digit zero following a zero, or a one followed by a one, etc. Since there is a 10 percent chance that the machine, having produced the digit X, will immediately produce X again, there ought to be some 500 such twins in the run of 5,000. Instead there were only 193. Yet, when we tested the generator, pressing the button about every three seconds, they appeared approximately 10 percent of the time. However, when we pressed the button at ninety-second intervals—which corresponds more closely to the rate in the actual ESP experiment—the generator again avoided twins as it had during the experiment.

It isn’t clear how many of the 222 hits above those expected by chance can be “explained” by the peculiarities of the generator. ESP may indeed be the reason for some of these 222 hits. However, the raw data make it difficult to quantify the relative proportion of hits due to the machine’s nonrandomness and those due to ESP. One way to indicate the delicacy of the analysis required is to note that there were 4,278 misses while pure chance yields some 4,500, if the generator’s were random.

Tart is anxious to redo the experiment, using the Rand table of random numbers, and we hope to work with him. Until the experiment is done again, we are in the position of a chemist who at the end of an experiment discovers that his test tube was dirty. Whether it was only a little contaminated or a lot doesn’t matter. The experiment has to be executed with a clean test tube. Tart hopes to carry out the modified experiment before the end of the year.

I have several comments.

Tart had earlier recognized anomalies in the workings of his randomizer, although he did not realize that some of them also appeared when the randomizer was operated under non-test conditions. In recent lectures he has attributed the anomalies to the ability of strongly psychic subjects to influence the randomizer by psychokinesis, the mind literally altering the operation of the device.

Tart also recognizes in his book (page 164) a clever method by which sender and receiver could have cheated, though he argues that this is unlikely. Assume that the sender, immediately after the receiver registers a target choice, operates the randomizer. Suppose it selects seven. The sender multiplies seven by, say, five to get thirty-five, then waits exactly thirty-five seconds before pressing the next target button. The receiver, watching a second hand on a wrist-watch, knows the target before he starts to move his hand.

“This possibility should be eliminated in future work.” Tart writes in a thumping understatement (page 104), by making “the time delay between switching off one target and selecting the next uniform and beyond the experimenter’s control.” Time delay techniques of secret information transmission are well known to magicians familiar with modern “mentalism.” Had Tart consulted such an expert, the most glaring defect of his machine could have been remedied before he began testing.

Tart is to be commended for his willingness to provide raw data; this in marked contrast to P and T. In 1975 Scientific American requested permission to let a statistician inspect the raw data of their NASA experiment. The request was refused. (See the letters department of Scientific American, January 1976.) Tart is also to be praised for his candor in describing in his book the chief defect of his experimental design.

In an article on “ESP Training” in Psychic magazine, March 1976, Tart wrote: “For the 10 TCT subjects, the average number of hits per run was well above chance, with odds against chance being a million billion billions to one” (Tart’s italics). Nothing so sensational has ever been claimed before by a professional parapsychologist. The above facts make clear that until Tart repeats his tests under controlled conditions—adequate randomizing and rigid exclusion of all possible methods (there are others!) of secret coding between subject and sender—the staggering results reported in his book cannot be taken seriously even by other parapsychologists.