电话：010-85195300
E-mail：cassoa@cass.org.cn
地址：中国北京建国门内大街5号中国社会科学院1#楼
扫一扫关注微信
国家哲学社会科学文献中心版权所有
PK with prerecorded random events and the effects of preobservation - psychokinesis
Helmut SchmidtPK experiments with prerecorded targets (Schmidt, 1976) are interesting because they suggest a variety of new test arrangements to study the basic PK mechanism. In addition, they permit the inclusion of independent observers who, with little time investment, can obtain first-hand evidence of any anomalous effects that may occur.
Four such experiments with different independent observers, different subject populations, and minor variations have been performed previously (Schmidt & Braud, 1993; Schmidt, Morris, & Hardin, 1990; Schmidt, Morris, & Rudolph, 1986; Schmidt & Schlitz, 1988) with the first author as main experimenter and the other one or two authors acting as independent observers.
The results of PK experiments with prerecorded random events appear most interesting and most puzzling because the subject's mental effort is made long after the random events to be affected have occurred. One tentative viewpoint (Schmidt, 1978; Walker, 1979) is that the subject's mental effort could act backward to the time when the random events were generated and recorded. This would imply a noncausal mechanism in the sense that the effect (the biasing of the random events) occurs before the cause (the mental effort). It might be this element of noncausality that makes psychic phenomena so intuitively implausible and at odds with the known principles of physics.
Another viewpoint (Schmidt, 1982) is based on the quantum theoretical notion that there is no reality if there is no observer. Thus, the decisions of the random generator might become "reality" only when the subject observes the outcome during the test session. The subject's effort would not have to act into the past because the decision on the outcome would not yet be made until the subject received feedback during the PK test session. In this case, a preobservation of the recorded events should inhibit the subject's success, because the preobservation would make the random decisions real so that the subject's subsequent mental effort would come too late.
For this reason it seemed particularly interesting to conduct an experiment in which half of the prerecorded random events had been preinspected so that one could compare the possible PK effects on the preobserved and not-observed events.
One previous study (Schmidt, 1985) has indicated an inhibition of the PK effect by preobservation. In view of the central role of the question, however, further studies of the preobservation effect seem most important.
In this and the four previous studies, the question of highest priority was whether a PK effect could be demonstrated under tightly controlled conditions. Because preobservation might inhibit the PK effect, we decided in advance to base the possible evidence for the existence of PK only on the results from the not-preobserved events.
Although the independent observer was somewhat skeptical about the existence of PK, he was very interested in a possible preobservation effect because of its possible implications for quantum theory.
Our subject pool consisted primarily of martial arts students as a matter of practical convenience. This population was relatively easily accessible and one that the experimenter had not yet explored. No attempt was made, however, to compare martial arts training and skill with PK performance systematically.
METHOD
The study was structured into 20 units such that a unit could be considered as a mini-experiment. Each unit was subdivided into 100 blocks of binary data. The sequence of binary data in each block determined two partial results, an even result and an odd result, arising from the even- or odd-numbered bits in the sequence. The partial results were defined so that by inverting the bits in the sequence (interchanging 1s and 0s) the signs of the partial results were inverted.
In following the procedure for each unit, the experimenter used a true random generator to record the random bit-sequences for the 100 blocks on floppy disk. Next a computer produced a 100-line printout, each line showing the two partial results for the corresponding block. Without looking at the data, the experimenter covered the numbers with labels that could be removed later and mailed the printouts to the independent observer.
The independent observer then used a random method of his own choice to specify for each block (a) whether the PK effort should be directed toward high- or low-valued results, and (b) whether the even or the odd result should be subject to preobservation. In accordance with the latter decision, he peeled the label from the corresponding number in each line and "preobserved" these partial results. The first decisions (a), also called the target assignments, were communicated to the experimenter. On the basis of these assignments, a computer derived from the original bit-sequence a secondary sequence in which the bits for all blocks with low-result assignments were inverted. The assigned PK goal was now the appearance of predominantly high results in all the blocks derived from the secondary sequence.
The secondary sequence was stored in a computer chip inside a small test machine, and the test machine was then sent out to a subject. For a PK test run, the next unused bits were accessed consecutively with a continuous display of the subject's performance at the moment. At the end of the run, a total score representing the sum of the two partial scores was displayed, with positive values indicating success. Thus, the subject received no information on the individual partial scores and was not even aware that part of each score had been preobserved.
For half of the units (Units 11-20), the partial scores of each single run corresponded to the partial results in the initial printout. For the other half (Units 1-10), the scores of two runs combined determined the printed results. (See the section on Data Preparation by the Experimenter.)
Some time after the subjects had completed their PK efforts on all runs in a unit, the independent observer removed the covers from the not-yet-observed results and evaluated these data by the same method applied previously to the preobserved data. The first question was whether (as indicated by previous studies) the not-preobserved results showed a tendency toward the direction (high or low) specified by the initial target assignments. Such a possible bias was expressed in terms of a z value measuring the deviation from chance in units of one standard deviation. At the end, the z values from all 20 units were combined to form a total z value as the final indicator of a possible PK effect. A corresponding calculation was carried out for the preobserved part.
Test Machines and Feedback Display
Each of several identical test machines used in the experiment showed a row of 27 LED lamps, which could be lighted individually in red or green. Internally, the system was controlled by a microprocessor (the DS5000 version of the 8051 microcontroller) that held an internal back-up battery and a 32K memory for the storage of the program, the prerecorded random events, and the options chosen by the subjects for the individual test runs.
For each run, the subject had a number of options concerning the form and speed of the display. Depending on these choices, one run could last from about 5 sec to 1 min. For each selected choice, the display pattern that the subject tried to affect and the resulting score were fully determined by a prerecorded binary sequence in the chip memory. A detailed description of the available display options is given in the Appendix. These display options fell into two classes that used the prerecorded bits in slightly different ways and led to different relationships between the bit-sequence and the resulting score that measured the subject's success in the PK effort.
Considering the runs of Class 2 first, each run used a prerecorded sequence of 128 bits. The score was simply given by:
Score = [(number of 1s) - (number of 0s)]/2.
In a typical display mode, a light started at the center of the display and jumped for each 1 by one step to the right or left, respectively, as the 128 bits were gradually read. Once the right or left margin was reached, the light was reset to the center to continue its random walk. The PK goal was a high score, that is, the appearance of an excess of 1s, indicated in this display by a bias of the random walk toward the right. As an additional stimulus, the color of the lights turned green for lights moving in the desired right half and red for lights in the left half of the display. For subjects who preferred to move the light to the left into the red color zone, the display could be inverted so that the desired excess of 1s now implied an overall motion to the left. While the standard run length was 128 bits, some of the display modes could subdivide the 128 bits into smaller sections so that the subjects could do the run in the form of several short "bursts of energy."
To study the effect of preobservation, we divided the total score for the 128 bits into one partial score resulting from the 64 even-numbered bits and one partial score from the odd-numbered bits,
Score = [Score.sub.even] + [Score.sub.odd],
with one of these scores having been preobserved. With the bits from the observed and not-observed scores interlaced, the subject could by no ordinary means distinguish between the two types of bits. Assuming a random bit-sequence, the standard deviation for the fluctuations of each partial score is seen to be [Sigma] = 4.
For the runs of Class 1, a typical display showed a light swinging over the one-dimensional 27-lamp array with varying amplitudes, at a typical rate of one cycle per second. Basically, a 1 or a 0 in the bit-sequence slightly increased or decreased the swing amplitude, subject to the upper limit (swing over the whole lamp range) and lower limit (zero swing amplitude). The subject's goal was a high average swing amplitude. Accordingly, the score that measured the success of the subject's effort was defined as proportional to the average swing amplitude in a run. Although a prevalence of 1s in the sequence would generally favor high scores, the score could no longer be expressed simply as the difference between the numbers of 1s and 0s in the sequence.
Each run was based on 256 prerecorded bits. To ensure that a preobserved and a not-observed part would contribute evenly to the display, we used two pointers to indicate the successive swing amplitudes updated after each half-cycle. The even- and odd-numbered bits were shifted into two corresponding 128-bit sequences, which in turn regulated the changes of the two corresponding pointers as follows:
Starting at the pointer value P = 3, a 1 or a 0 in the corresponding bit-sequence increased or decreased the pointer value by one step, with the restriction that the pointer value could not go higher than 6 or lower than 0. Thus, the 128-bit sequence generated a sequence of 128 pointer values, with the pointer changing gradually (by 1, 0, or -1) in the range from 0 to 6.
During the run, the pairs of even- and odd-numbered bits were successively read, and the sum of the two current pointer values, [P.sub.tot] = [P.sub.even] + [P.sub.odd], was displayed to the subject in the form of the current swing amplitude of the light.
Partial scores resulting from the odd- and even-numbered bits were defined as the sums of the corresponding 128 P values.
The expectation value of each partial score ([Score.sub.even] or [Score.sub.odd]) is seen to be 3 x 128 = 384. Because the exact expression for the theoretical variance would be cumbersome to calculate, it seemed easier to obtain the variance experimentally as the average from a total of 40,000 generated control sequences. This gave for the standard deviation of each partial score the value [Sigma] = 94.1. At the end of a run a display showed the deviation of the total score from chance, which is
Display = [Score.sub.even] + [Score.sub.odd] - 768.
To get familiar with the different display options and as a "warm-up," the subject could switch the machine to a play mode. In this setting, the required random bits resulted from a quasi-random algorithm with a seed number for each run determined (in a practically random manner) by the timing of the start button push. The results of these play runs were not recorded or evaluated.
Data Preparation by the Experimenter
In preparing each block, the experimenter used a PC computer with an attached random number generator to record the random bit-sequence required for this block onto a floppy disk.
The random generator (similar to the one in Schmidt, 1970) used the random timing of radioactive decays (average of 30 signals per second from an Sr-90 source registered by a Geiger counter) to stop a fast counter (0.5 MHz) randomly at even or odd counts. The reliability of this generator had been previously tested extensively. As an additional precaution against the possibility that a malfunction of this generator might bias the output, each generated random bit was combined through the XOR operation with a bit provided by a quasi-random algorithm. It should be noted, however, that the independent observer introduced an additional random element of his own design, so that he did not have to concern himself with the precautions taken by the experimenter in generating the random numbers.
The length of the prerecorded bit-sequence depended on two factors: (a) whether the unit should be used for Class 1 or Class 2 runs (a Class 1 run used 256 bits whereas a Class 2 run used 128 bits); and (b) whether each result to be printed out for the independent observer should be compiled from one or two runs. The experimenter made these decisions for each unit. It gave him some freedom regarding the displays to be used and the time (measured by the number of runs) to be invested in each unit. As it turned out, the first 10 units used 200 runs each (2 runs per displayed result). But then the experiment had dragged out longer than anticipated so that the experimenter felt the need for a speedy conclusion and used for the next 10 units only 100 runs per unit (1 run per displayed result). Note that for the integrity of the statistical evaluation the preset number of 20 units had to be maintained.
Table 1 shows for each unit the type of feedback used (Class 1 or Class 2) and the number of runs completed. Furthermore, the table lists the expected standard deviation per printed partial result score. (Note that there are cases with one and two runs per result and that the standard deviations per run are [Sigma] = 91.1 for Class 2 and [Sigma] = 4 for Class 1.)
TABLE 1 NUMBER OF RUNS AND DISPLAY TYPES FOR THE UNITS Unit no. Class No. of runs SD 3, 4, 5, 8, 9, 10 1 200 133.1 1, 2, 6, 7 2 200 5.66 16, 17, 18, 19, 20 1 100 94.1 11, 12, 13, 14, 15 2 100 4.0
From the prerecorded bits, the experimenter prepared a printout for the independent observer. We will describe this first for the case of Class 2 units with 100 runs per unit.
In this case there were 100 blocks of 128 prerecorded bits. For each block a computer program read the prerecorded bits and printed two numbers into one line, the score for the 64 even-numbered bits at the left, and the score for the 64 odd-numbered bits at the right. At this stage nobody looked at the printed numbers, but Xerox copies were made and sealed for later use. The experimenter then covered the individual numbers on the original printout with removable labels. To avoid an accidental look at the printed numbers, the experimenter used dark glasses in a semi-darkened room for this procedure. The covered print-out was sent to the independent observer. In addition, the independent observer received the sealed copy as well as a sealed disk with the original bit-sequence. These additional data were intended only as back-up in case the later removal of a label should have damaged the underlying printout or the independent observer might want to perform some additional analysis on the original data.
The data preparation was similar in the other cases. For units with 100 runs of Class 1, the computer calculated and printed the partial scores as described previously, and for units with 200 runs the computer added the partial scores from two successive runs for the printout.
Initial Role of the Independent Observer
The experiment was designed so that the independent observer could verify the presence of anomalous effects that might occur without having to depend on the reliability of the experimenter or his procedures.
Having received the covered printout, the independent observer used a randomizing method under his own control to make two binary decisions for each line: (a) the decision on the target assignment, that is, whether the PK effort should be directed toward producing high (positive) or low (negative) printouts in this line, and (b) the decision of whether the right or the left number in the line should be preobserved.
The process for making these decisions was prespecified as follows. When the independent observer received the printout with the hidden numbers, he waited one week and then obtained The New York Times of the following Tuesday. From the data in a prespecified weather column, he then obtained a seed number and used it as an entry point into a prespecified quasi-random algorithm, which, in turn, provided the sequence of binary decisions required for the unit.
The independent observer then lifted the label from that number in each line that was to be preinspected and noted the value.
Next, he sent to the experimenter the sequence of 100 binary events, which specified the target assignments for the 100 blocks.
Subsequent Procedures of the Experimenter
Having received the target assignments, the experimenter entered the 100 bits into a computer program that derived from the original bit-sequence a secondary sequence for which all bits of the blocks with low-score assignment had been inverted. This secondary sequence was transferred into the memory of the test machine to be used for the subsequent test runs. Then the subject had to aim throughout toward high (positive) score values.
The test machine, with written instructions, was given out or mailed to participants. In some cases a single person completed a whole unit. In other cases several subjects contributed to the total number of runs in the unit.
When all blocks of a unit were completed, the experimenter informed the independent observer, who was now free to remove all labels and inspect all the results of this unit.
Evaluation of the Data, As Seen by the Independent Observer
We shall discuss first those data that were directly accessible to the independent observer. Other matters that the independent observer could not check, such as the selection of subjects and scores obtained by specific subjects, will be discussed later.
At the completion of the whole experiment, the independent observer sent the experimenter the information on which results had been preobserved and the key by which these decisions had been derived from the weather data.
Then both of us, independent observer and experimenter, had available for each unit two sets of 100 numbers representing the preinspected and not-inspected results. We also had the list of target assignments. The aim of the PK effort was to produce a bias toward the specified target directions in both of these sets.
Let us look, for example, at the 100 not-preobserved values for one unit. From these values--let us call them P(1), P(2),..., P(100)--we derived a secondary set of numbers P[prime](1), P[prime](2),..., P[prime](100) by inverting the signs of the P(n) values with low-aim target assignment. Then success in the PK effort could be measured by a positive bias of the numbers P[prime](n). Previous studies (for example, Schmidt, Morris & Rudolph, 1986) evaluated such a bias with the help of a rank-order test. For the present study we decided to use a somewhat simpler, but equally valid, method based on the evaluation of the random variable,
V = P[prime](1) + P[prime](2) + . . . P[prime](100).
The evaluation makes no assumptions about the reliability of the experimenter's random generator. It is sufficient that the target assignments by the independent observer are random.
Then for any given set of values P(1), . . ., P(100) supplied by the experimenter, the expectation value of Vis:
[Mathematical Expression Omitted].
Furthermore, the variance of V (for the given P(n) values) under all possible target assignments is seen to be:
[Mathematical Expression Omitted].
Thus, the expected value and the variance of V under the null hypothesis (no PK effect) are known, and we can measure the statistical significance of a possible PK effect in terms of the deviation from chance in units of one standard deviation:
[Mathematical Expression Omitted].
Let z(1), . . ., z(20) be the z values obtained in the not-observed cases for the 20 units; then we measure the total significance of the result in terms of the total z value:
[z.sub.tot] = [z(1) + . . . + z(20)]/[square root of 20].
Because of the large number of contributing terms, the distribution of [z.sub.tot] can be taken as practically normal (with unit variance and zero expectation value). Our final decision on whether the experiment showed an overall PK effect will be based on this number.
A corresponding calculation was performed for the preobserved part of the experiment.
RESULTS
Results as Seen by the Independent Observer
The results obtained in this manner for the preobserved and not-observed part of each unit are listed in Table 2. At the bottom the table also gives the final z values to tell us about the existence of PK under two conditions.
Some additional information, which was also directly available to the independent observer, is provided by Table 3. There we have sorted the units according to the average standard deviation of the results. As expected from Table 1, these values fall into four clusters with [Sigma] values close to the theoretical values given in Table 1. In addition to the z values taken from Table 2, Table 3 provides the number of results that deviated from chance in the desired (hit) or not desired (miss) direction. The bottom line shows that, in all, 50.89% of the not-preobserved results tended in the desired direction but only 48.89% of preobserved results did.
Discussion of the preceding results. The most central question in this study was whether PK effects could be demonstrated with prerecorded, not-preobserved, random events and under the supervision of an independent observer. The present experiment produced a deviation from chance in the predicted direction, but, with z = 1.23, not at a level of statistical significance.
TABLE 2 MAIN RESULTS FOR UNITS 1-20 Unit no. Observed condition Not-observed condition 1 -1.83 0.36 2 0.56 -0.24 3 0.33 1.54 4 0.43 1.34 5 -0.44 0.67 6 2.47 1.91 7 0.23 0.38 8 -0.84 0.27 9 -1.03 -0.18 10 -1.21 0.64 11 -1.11 -0.64 12 0.27 -1.37 13 0.81 0.93 14 -1.89 -1.22 15 0.44 0.14 16 -0.43 0.83 17 1.56 0.27 18 -0.94 -0.61 19 -1.43 1.18 20 -0.09 -0.69 z(total) -0.93 1.23
It might be fair to consider, however, that this is the fifth study conducted by the present experimenter under similar conditions, under the supervision of independent observers. Each of these experiments was similarly structured, using a prespecified number of units that were independently evaluated in terms of z values from which a total z value was derived. Table 4 gives a summary of these studies. A more detailed compilation and comparison of them is in preparation.
These five z values obtained up to the present combine for a total z value of z = ([z.sub.1] + [z.sub.2] + . . . + [z.sub.5])/[square root of 5] = 3.69, which is statistically quite significant and not easily discarded as a result of chance fluctuations.
None of the preceding four studies explored the effect of preobservation of the random events. With the difference between not-observed and preobserved events in the present study not significant, [z.sub.diff] = (1.23 + 0.93)/[square root of 2] = 1.52, further studies will be required to address this question.
TABLE 3
RESULTS FOR UNITS ORDERED BY VARIANCE
z Preobs. Not-obs.
Unit no. Mean [Sigma] Preob Not-obs. Hit Miss Hit Miss
8 142.9 -0.84 0.27 51 49 53 47
5 136.9 -0.44 0.67 50 50 55 44
9 131.1 -1.03 -0.18 48 52 45 54
10 131.1 -1.21 0.64 48 52 48 52
4 129.7 0.43 1.34 52 48 55 45
3 119.9 0.33 1.54 54 46 53 46
Subtotal 303 297 309 288
Hits 50.50% 51.76%
17 936.0 1.56 0.27 53 47 52 48
16 934.7 -0.43 0.83 47 52 57 43
19 908.2 -1.43 1.18 45 54 57 42
18 893.5 -0.94 -0.61 42 58 45 55
20 866.1 -0.09 -0.69 46 54 52 48
Subtotal 233 265 263 236
Hits 46.79% 52.71%
1 61.4 -1.83 0.36 36 54 49 41
2 59.2 0.56 -0.24 46 51 47 48
6 57.7 2.47 1.91 55 40 57 36
7 56.6 0.23 0.38 52 40 41 50
Subtotal 189 185 194 175
Hits 50.53% 52.57%
11 39.0 -1.11 -0.64 41 49 38 49
13 38.9 0.81 0.93 46 45 47 42
14 38.6 -1.89 -1.22 38 56 43 51
15 38.2 0.44 0.14 46 44 43 46
12 38.0 0.27 -1.37 45 43 37 53
Subtotal 216 237 208 241
Hits 47.69% 46.69%
Grand Total 941 984 974 940
Hits 48.89% 50.89%
The remainder of the paper is concerned with data that the independent observer could not check, such as the scores of individual subjects or groups of subjects.
TABLE 4 SUMMARY OF PK EXPERIMENTS WITH PRERECORDED NOT-PREOBSERVED RANDOM EVENTS CONDUCTED UNDER INDEPENDENT SUPERVISION Study z Schmidt, Morris & Rudolph (1986) [z.sub.1] = 2.71 Schmidt & Schlitz (1988) [z.sub.2] = 1.66 Schmidt, Morris & Hardin (1990) [z.sub.3] = 0.62 Schmidt & Braud (1992) [z.sub.4] = 1.98 Schmidt & Stapp (present study) [z.sub.5] = 1.23
The Test Subjects
One of the objectives of the study was to identify individuals or groups of subjects who might be exceptionally successful in PK tasks. In particular, the study focused on students of martial arts, on a few highly proficient martial arts teachers, and one psychic who had done very well in a previous experiment. Getting the cooperation of these subjects turned out to be more difficult than anticipated. Although most subjects said they greatly enjoyed the experiment, they took an unduly long time in completing tests and returning the mailed-out test machines.
Before the results were known, the experimenter decided to group the subjects, with the score for each group to be reported separately, as follows:
Category F. This group consisted of a martial arts teacher, Don Finley, and six of his students. D.F. (age 36) teaches T'ai Chi and Kung Fu and has practiced meditation for 20 years. The 6 students (average age, 38 years) had between 2 and 3 years of training. The experimenter paid several visits to D.F. to explain the tests and check on the progress. D.F., in turn, explained the experiment and the operation of the test machine to his students.
Category G. Gilbert Leal, a martial arts teacher, and 13 of his students made up the group. G.L. (age 37) teaches T'ai Chi and Wing Chun. The students (average age, 28 years) had practiced typically for one year. G.L. received instructions in the test operation and then passed this information on to his students.
Category L. This was a group of four subjects tested by a psychology student, Linda Reszeck. These subjects were meditators and healers with at least two years of regular practice (average age, 39 years).
Category M. Nine volunteers, who visited the laboratory and were introduced by the experimenter to the tests, made up this group. These subjects (average age, 45 years) included martial arts practitioners and meditators. They had practiced for an average of 7 years and were chosen because of their serious interest in psi and their eagerness to participate.
The following persons were selected to contribute large numbers of trials and were accordingly evaluated individually:
Category HH. Hoang Ho is an acupuncturist and T'ai Chi teacher with many years of experience. He was selected because of his high scores in a preceding experiment (Schmidt & Braud, 1993). The experimenter visited HH, discussed the experiment briefly, and left a test machine with instructions.
Category HL. Horacio Lopez is a T'ai Chi teacher and massage therapist who has spent a good part of his life on the study of the martial arts. He was selected because of his students' high recommendations and his apparent dedication to mental disciplines. The experimenter met with HL, explained the experiment, and handed over a test machine with instructions.
Category IS. Ingo Swan is an artist and "psychic" who has worked with numerous researchers. He was selected because of his success in a previous experiment (Schmidt & Braud, 1992). A test machine with instructions was mailed to IS.
Category GL. George Leonard is an Aikido teacher with a long-time dedication to mental disciplines. He had performed well in a previous experiment (Schmidt & Braud, 1993). He readily agreed to participate again and received a test machine with instructions by mail.
Category HS. Helmut Schmidt, the present experimenter, had performed well as a subject in one similar experiment (Schmidt, Morris, & Rudolph, 1986). He contributed initially short sections that other subjects had failed to complete. Toward the end, when two subjects had dropped out, he completed longer sections to finish the experiment within the projected time frame.
Results Obtained by Groups and Individuals
The total results of the groups or individuals are given in Table 5. The z values listed in this table were calculated from the deviation from chance, and the variance was calculated from the number of runs and the variance per run as given in Table 1.
Considering only the not-preobserved parts (Not obs. columns), it appears that the Class 1 runs produced generally more positive results than the Class 2 runs did. Note that a Class 1 run required a more extended effort of typically one minute's duration, whereas the Class 2 runs represented typically a very short effort with the subject trying to exert a burst of energy. The experimenter's expectation that the latter runs would be particularly successful for martial arts students and teachers (F, G, HH, HL, GL) turned out to be wrong.
TABLE 5
RESULTS OBTAINED BY INDIVIDUALS OR GROUPS
Class 1 trials (z) Class 2 trials (z)
Category Runs Obs. Not obs. Runs Obs. Not obs.
F 155 -0.09 0.47 143 -0.04 -0.34
G 388 -2.67 0.69 322 -1.41 -0.22
L 94 0.53 1.39 84 -0.41 -0.57
M 440 0.03 0.60 96 2.43 1.94
HH 100 -0.44 0.81 100 -1.08 -0.63
HL 100 1.55 0.26 100 0.25 -1.33
IS 100 -0.09 -0.65 100 0.45 0.13
GL 50 -0.03 0.98 50 -1.45 -0.88
HS 273 -0.85 0.81 305 1.22 1.96
Note. The table lists for each category the numbers of contributed runs and
the corresponding z values for the preobserved (Obs.) and not-preobserved
(Not obs.) parts.
The only notable scores in Class 2 were obtained by the experimenter himself (HS) and by the group (M) of subjects who came to the laboratory and worked under the experimenter's direct supervision.
Although the z values indicate the statistical significance of the different contributions, one might feel that the performance of the different groups can be compared better in terms of a scoring rate or an effect size (ES) proportional to the average score per run. A convenient general measure for this effect size is given by
ES = z/[square root of (number of contributing runs)].
Table 6 lists the ES values next to the z values for the not-preobserved part of the study. One should be aware, however, that these ES values can provide a reliable measure of a PK ability only in cases where the corresponding z values are sufficiently large to achieve statistical significance.
CONCLUSION
The experiment had several objectives: (a) the gathering of more evidence for the existence of PK effects on prerecorded events under external supervision, (b) the study of the effect of preobservation of the random events, and (c) the test of the hypothesis that martial arts practitioners might be particularly successful in PK tests.
TABLE 6
EFFECT SIZES FOR THE NOT-PREOBSERVED PARTS
Class 1 trials Class 2 trials
Category z 100 ES z 100 ES
F 0.47 3.77 -0.34 -2.80
G 0.69 3.51 -0.22 -1.24
L 1.39 14.32 -0.57 -6.23
M 0.60 2.87 1.94 19.84
HH 0.81 8.06 -0.63 -6.25
HL 0.26 2.65 -1.33 -13.25
IS -0.65 -6.47 0.13 1.25
GL 0.98 13.90 -0.88 -12.50
HS 0.81 4.93 1.96 11.22
(a) The positive outcome of the tests with not-preobserved events, although not significant in itself, lends further support to the existence of PK effects under these strictly supervised conditions.
(b) The preobserved and the not-preobserved random events led to deviations from chance in opposite directions. With the effects not statistically significant, continuation of this kind of work is required to see whether, indeed, preobservation has an effect, and how different kinds of preobservation (more or less meaningful to the observer) might have different effects.
(c) The martial arts practitioners whom the experimenter was able to obtain showed no outstanding abilities. Surprisingly, they scored negatively (opposite to the direction aimed for) in a novel arrangement (Class 2 tests) where the subjects made short intense mental efforts.
APPENDIX
DISPLAY OPTIONS OF THE TEST MACHINE
The PK tester showed a row of 27 LED lamps arranged over a length of 30 cm. Each lamp could be lighted in red or green. Each tester was programmed by the experimenter for one of two classes of display options.
Class 1 Displays
The Class 1 displays were based on the motion of two pointers, each varying randomly in single steps in the range from 0 to 6. The sum(S) of the two pointers was displayed after each pointer change, 128 times in a test run. The subject could select one of several operating speeds that determined the duration of the 128-step runs. Typically, a run lasted from half a minute to one minute. The following display options were available:
1. In one display the successive S values determined the successive amplitudes of a light swinging around the center lamp. For the maximal value S = 12, the light would swing over the whole lamp range, whereas for S = 0 the light stayed at the center lamp. The amplitude changes occurred after each half-cycle. The color of the light changed from green for above-average amplitudes (S [is greater than] 6) to red for lower amplitudes. By the setting of a switch, the subject could opt for the aim of a high or a low swing-amplitude.
2. This display was similar to the previous one, but instead of the one light, the subject saw two lights swinging symmetrically with respect to the center lamp. The goal was to have these lights "bounce against each other" with maximal or minimal amplitude.
3. In this display the consecutive S values determined the current positions of the light, with higher S values corresponding to positions more to the right. The goal was to keep the light most of the time on the selected side of the display.
4. A light moved at varying speeds, proportional to the current S value, across the display, with S = 12 and S = 0 corresponding to maximal speeds toward the right and toward the left, respectively. When the light left the display on one end, it immediately entered again from the other end. Motion to the right and left were accompanied by green or red lamp colors, respectively.
5. Here the light moved only in one direction, at 12 different speeds corresponding to the 12 possible S values. The goal was to speed up or to slow down the motion.
6. Here a red and a green light were swinging over the whole lamp range. The varying element was the phase between these two sine wave motions. The phase angle was linearly dependent on S. For small S values the lights moved close together, whereas for S = 12 the lights were moving in opposite directions, bouncing against each other.
Class 2 Displays
In Class 2 displays, each run was based on a 128-bit sequence. The runs in Class 2 took less time than the runs in Class 1. Furthermore, with one form of display, a run could be completed in several shorter sections of "bursts of mental energy." The primarily used forms of display were these:
1. The light started a random walk from the center lamp, moving one step to the right or to the left for each successive 1 or 0 in the binary sequence. Whenever one edge of the display was reached, the light was reset to the center to continue its random walk. The aim was to move the light predominantly toward a selected side.
2. This display was similar to No. 1, but the light stopped after reaching one side, or after a maximum of 32 steps. Thus, the 128-bit run was subdivided into several short periods. It was hoped that this mode would appeal in particular to martial arts students accustomed to exerting short, intense mental efforts.
3. The light began with one pass from left to right and then continued cyclically in a stop- (for a 0) and go- (for a 1) manner. The goal was to keep the light moving.
REFERENCES
SCHMIDT, H. (1970). Quantum mechanical random number generator. Journal of Applied Physics, 41, 462-468.
SCHMIDT, H. (1976). PK effect on pre-recorded targets. Journal of the American Society for Psychical Research, 70, 267-291.
SCHMIDT, H. (1978). Can effect precede its cause? Foundations of Physics, 8, 463-480.
SCHMIDT, H. (1982). Collapse of the state vector and psychokinetic effect. Foundations of Physics, 12, 565-581.
SCHMIDT, H. (1985). Addition effect for PK on prerecorded targets. Journal of Parapsychology, 49, 229-244.
SCHMIDT, H., & BRAUD, W. (1993). New PK tests with an independent observer. Journal of Parapsychology, 57, 227-240.
SCHMIDT H., MORRIS, R. L. & HARDIN, C. L. (1990, September). Channeling evidence for a psychokinetic effect to independent observers: An attempted replication. (Mind Science Foundation Research Report.) San Antonio, TX: Mind Science Foundation. (See also abstract in Research in Parapsychology 1991.)
SCHMIDT, H., MORRIS, R. & RUDOLPH, L. (1986). Channeling evidence for a psychokinetic effect to independent observers. Journal of Parapsychology, 50, 1-15.
SCHMIDT, H., & SCHLITZ, M. (1988, December). A large scale pilot PK experiment with prerecorded random events. Mind Science Foundation Research Report. San Antonio, TX: Mind Science Foundation. (See also abstract in Research in Parapsychology 1989.)
WALKER, E. H. (1979). The quantum theory of psi phenomena. Psychoenergetic Systems, 3, 259-299.
Mind Science Foundation 7979 Broadway, #100 San Antonio, TX 78209
Lawrence Berkeley Laboratory 50A-3115, Physics Division 1 Cyclotron Read Berkeley, CA 94720
COPYRIGHT 1993 Parapsychology Press
COPYRIGHT 2004 Gale Group