A recent paper by Dr. Charles T. Tart, a parapsychologist at the University of California at Davis, casts some revealing light both on Tart and on a sensational earlier experiment by him that was the topic of a spirited debate in 1977 in two issues of The New York Review.
The debate began with my note, “ESP at Random” (NYR, July 14), in which I criticized Tart’s book, Learning to Use Extrasensory Perception (University of Chicago Press, 1976). In this book Tart reported success in ESP scoring that far exceeded anything obtained before in the history of parapsychology.
My note reproduced a letter from three of Tart’s colleagues at Davis, mathematicians Aaron Goldman, Sherman Stein, and Howard Weiner. Impressed by the results in Tart’s book, they had asked to see the raw data. Going over it they found that the alleged random-number device did not produce random numbers for the target sequence. “Until the experiment is done again,” they wrote in their letter, “we are in the position of a chemist who at the end of an experiment discovers that his test tube was dirty…. The experiment has to be executed with a clean test tube.”
My note also pointed out a glaring flaw in Tart’s experimental design. His “Ten-Choice Trainer” (TCT) was constructed so that a “sender” and “receiver” who wished to cheat could easily do so by signaling with what magicians know as a time-delay code.
To see how such a code could have been used it is necessary to describe again the TCT’s basic working. A sender in one room is in front of a console that displays a circle of playing cards from ace through ten. Next to each card is a pushbutton and a pilot light. When the electronic randomizer selects a card value, the sender pushes the button that turns on the light by the corresponding card. This actuates a “ready light” on a duplicate console in another room where the receiver is stationed. As soon as the receiver sees the ready light he begins an ESP search for the target card, usually by moving a hand around the dial. The sender observes this sweep on a TV monitor above his console. This arrangement is intended to help him telepathically “urge” the receiver to stop at the target. When the receiver has made a choice he pushes a button by the card. The target card’s pilot light immediately goes on and a chime sounds if the guess is correct. This immediate feedback is supposed to keep the receiver’s interest high and to stimulate the training of his psi powers.
In my 1977 note I explained how a sender could transmit the value of each card by varying the time between the receiver’s last choice and the activation of the ready light. It would, however, be foolish to send individual numbers because what is wanted is not a perfect score but only a significant score. This permits such simple coding that wrist watches are not even necessary, just a little practice in counting seconds mentally. For example, a delay of under ten seconds could mean an even card, a delay of more than ten seconds an odd card. Transmission of just this one bit of information raises an expected score of 50 hits to 100. If the cards are divided into three groups the code is almost as simple. This would increase the expectation from 50 to 150 hits. To raise a score higher than that would be too suspicious.
Tart himself recognized the possibility of time-delay coding in a footnote that begins: “After the completion of the Training Study, I realized that this procedure allowed a possibility of sensory cueing. If a particular experimenter showed a differential time delay between reading the output of the random number generator and switching on various newly selected targets, a subject might become sensitive to this and artifactually increase his score…. This possibility should be eliminated in future work.”
Tart’s superstar was a girl identified only as S3. Her overall TCT score far exceeded that of any other subject: 124 hits when only 50 were expected by chance. Her experimenter-sender was Gaines Thomas, one of Tart’s students. John Sladek, a London correspondent, sent me a plausible conjecture that derives from Thomas’s vivid account of how he fixed the targets in his mind. First he entered the number on a score sheet, then he “silently repeated the number” to himself. Finally he “positioned” the card in his mind by keeping it just “posterior to the upper part of my ears.” Success, he adds, “very often corresponded with a numbing feeling in that location. Once I felt I had the number positioned, I would turn on the proper target switch….”
Is it possible, Sladek asks, that Thomas took less time to fix behind his ears an image of a card with a low value? It certainly is easier to visualize one of the five or six low cards than to visualize a seven, eight, nine, or ten. If this were the case, a sensitive receiver might unconsciously learn, as testing continued, a correlation between certain cards and a short or long delay. Thomas’s five subjects, each of whose scores were better than those of any student who worked with other senders, had an overall expectation of 250 hits. A binary code would raise this to 500. The actual hits were 466. S3 would be the student with the greatest skill in picking up, subliminally or otherwise, Thomas’s time-delay pattern.
After Tart became aware of time-delay coding he examined the scores of Thomas’s five subjects without finding any consistent pattern “as to which targets they scored best or worst on.” But of course such a check would not catch a binary code unless one had a record of the delay for each target, which Tart apparently does not have. However, Sladek points out, the following examination of the raw data should be made. Check the misses to see if these cards tend to be related to the target by a two-part grouping of the ten numbers.
The most likely division that Thomas would unconsciously make would be into high versus low cards. If a subject responded to such a grouping, misses would tend to be displaced a short distance clockwise or counterclockwise around the dial. This was actually the case. Tart reports significant “displacement” scores on adjacent cards. For Tart this was the result of poor ESP “focusing,” but such displacements are just as easily explained by a binary code. Sladek ran a simulation test of 500 trials, using a binary code. After sixty-seven trials his subject had correctly divided the numbers into two groups. Results for hits, as well as for positive and negative displacement to an adjacent number, were remarkably close to the overall scores of Thomas’s five subjects.
Thomas also tells us that during the subject’s sweep, as he watched it on the monitor, he would sometimes “orally coax the image on the screen, or swear at the near misses.” The receiver’s cubicle was just across a four-foot-wide corridor. (The doors of the two cubicles were ten feet apart.) One assumes that the sender’s cubicle, which Tart calls a “sound-attenuated” Faraday cage, was sufficiently sound attenuated to prevent a subject with sharp ears from hearing Thomas’s swearing, though I know of no checks by outsiders that were made on this. I suppose it depends on how loud Thomas swore. Of course any kind of sensory feedback from Thomas would explain high scores on both direct hits and displaced targets as readily as the time-delay hypothesis. In either case, it is significant that Thomas’s five students scored 466 hits when only 250 were expected, whereas the other five students in the TCT test, who had other senders, scored 256 hits when 250 were expected.
Tart replied to my NYR note in a letter (NYR, October 13, 1977) to which I responded in the same issue. I asked if videotapes had been made of S3’s performance. If so, a study of them could confirm or refute the time-delay theory. If videotapes were not made, this would be another design defect. I urged that S3 be tested under better controlled conditions in another laboratory. I assume no tapes were made and that the girl has not been tested again even though her scores, if based on genuine ESP, would make her one of the most talented psychics in the history of parapsychology.
When Tart wrote about his TCT experiment in Psychic magazine (“ESP Training,” March-April, 1976) he began as follows:
Strong criticism has been leveled at ESP research over the years because the phenomena could not be repeated regularly. Since they could not, skeptics gloated that they did not exist.
Now, a research breakthrough soon may shelve such criticism. A study carried out under my direction at the University of California at Davis Psychology Department has taken a big step toward repeatability of ESP by helping people understand how ESP works and how it can be controlled.
Tart did replicate his “breakthrough,” and the report, “Effects of Immediate Feedback on ESP Performance: A Second Study,” by Tart and two associates, appeared in last April’s issue of The Journal of the American Society for Psychical Research. The study was funded by est and by the Parapsychology Foundation of New York. Sixteen students participated. They were selected by a screening of high scorers from 2,424 tested persons. Overall scores for the screening process were at chance levels. As the authors put it: “There is no evidence that more percipients scored significantly above chance than would be expected if no ESP were operating.”
The high scorers were used in replicating the earlier experiment even though the screening provided no basis for assuming they were better at ESP than any of the others. The replication used three machines: a four-choice trainér called Aquarius, the TCT in its improved form, and ADEPT (Advanced Decimal Extrasensory Perception Trainer) which Tart describes as a ten-choice machine similar to the TCT but with more sophisticated circuitry.
Overall results for both ten-choice machines did not differ significantly from chance. There was some evidence of ESP improvement on the part of some subjects who used the Aquarius, but since only three students completed this part of the experiment it contributed nothing of statistical value.
In view of the astonishing contrast between the replication’s chance results and the near miraculous scores of the original experiment, one would expect Tart to end his report by withdrawing his former results. A chemist or physicist would feel ethically obliged to do so, especially if flaws had been pointed out in the original experiment. On the contrary, Tart expresses his belief that the second experiment failed mainly because “too few talented percipients were selected by the screening process.”
Here are Tart’s three reasons for the failure:
With respect to psychological interpretations, several people who have had close contact with students at the University of California, Davis, over the past three or four years have told us of a dramatic change in the attitudes of students during that period. In the last year or two, students have become more serious, competitive, and achievement-oriented than they were at the time of the first experiment. Such “uptight” attitudes are less compatible with strong interest and motivation to explore or develop a “useless” talent such as ESP. Indeed, we noticed that quite a few of our percipients in the present experiment did not seem to really “get into” the experiment and were anxious to “get it over with.”
The situation also was different for the student experimenters in the two experiments. Experimenters in the first experiment could legitimately feel that they were embarking on a new adventure. Despite our best efforts to create the same enthusiasm in the second group of experimenters, there was no way to deny the fact that we were asking them to simply repeat an experiment designed and executed by others before they ever arrived on the scene. It is understandable that they did not feel as intensely involved in the experiment as did the first group of experimenters, and this factor could have been responsible for the relatively poor performance of their percipients. Indeed, several of the more seriously involved experimenters later told C.T.T. [Charles T. Tart] that they were quite disturbed by the attitude of some other experimenters who “just wanted to get it done with.”
Finally, we were constantly plagued by machine malfunctions…and this was a source of continual annoyance and inconvenience to all concerned.
Let me summarize. Tart reported in a book, written with unbounded confidence, results so extraordinary that they far exceeded those obtained in similar testing by any other researcher. His TCT machine was found to have a flawed randomizer, and a design that permitted time-delay coding. A replication of the experiment, with both flaws eliminated, showed no significant departures from chance. Tart attributes this primarily to his inability to find sufficiently psychic students. As for the original experiment: “Because the level of scoring in the first experiment was so high, it would be absurd to argue that the results of the second experiment mean that the results of the first experiment were a mere statistical fluke.” Nowhere does he even mention the possibility that the first experiment was invalid because of a defective randomizer, or fraud, or unconscious time-delay or sensory cuing.
Tart’s last statement leaves me so staggered that I can respond only with a parable. A parapsychologist finds a psychic who can levitate a table forty feet. He investigates this under poorly controlled conditions, but is so convinced the phenomenon is genuine that he writes a book about it. The book is published by a gullible university press. After skeptical magicians—those terrible spoilsports!—patiently explain how the levitation could have been accomplished by trickery, the parapsychologist agrees to test the psychic again, this time with adequate controls. The table does not rise at all. The parapsychologist then writes a formal report that concludes: “In view of how high the table rose during the first experiment, it would be absurd to contend that the failure of the second experiment in any way casts doubt on my previous observations.”
During the academic year 1978-1979, Tart was on leave from Davis to work with Harold Puthoff and Russell Targ, at Stanford Research Institute, on remote viewing (clairvoyance) experiments presumably funded by the military. SRI is a private research organization not affiliated with Stanford University. As a matter of policy it will not reveal the sponsors of research projects. Experiments in parapsychology, which has yet to establish itself as a science, should be open to the scrutiny of the scientific community. When open research in parapsychology is conducted in shoddy fashion, we have a right to be concerned about the quality of clandestine ESP research. From what we know of their previous work, the grant in support of the new work by Targ and Puthoff will likely be a total waste. Unfortunately, that money comes from taxpayers such as you and me.
May 15, 1980