Information Acquisition Rates in Forced-Choice ESP Experiments: Precognition Does not Work as Well as Present-Time ESP(NOTE 1) Charles T. Tart Department of Psychology University of California, Davis Davis, Calif. 95616 Published in the Journal of the American Society for Psychical Research Vol.77, October 1983, pp. 293-310. ----------------------------------------------------------------- NOTE 1: A brief version of this paper was presented at the Twenty-Fourth Annual Convention of the Parapsychological Associa tion, Syracuse, New York, August 18-22, 1981.--Ed. -----------------------------end of footnote _________________________________________________________________ ABSTRACT: Quantification of the amount of information acquired in successful, forced-choice ESP experiments is possible using a measure of the average number of bits per trial. Using this measure, 53 studies of present-time ESP, where the percipient attempted to call currently existing targets, and 32 studies of precognitive ESP, where the percipient attempted to call targets that would only be generated (by a random process) at some later time, were reviewed. A striking and robust performance dif ference was found: present-time ESP can work up to 10 times as well as precognitive ESP in forced-choice tests. Three theories are proposed to account for this difference: a psychological theory that there are generally held biases against precognition in Western culture, so percipients don't try as hard: a two- process theory that present-time ESP and precognition are two basically different processes, with inherently different charac teristics: and a temporal-break theory that ESP is a unitary process, but something about the nature of time itself attenuates ESP performance that extends into the future. _________________________________________________________________ INTRODUCTION In 1980 I became interested in the question "How much infor mation can the ESP channel carry when it is working well?" In some card-guessing experiments, where only a slight (even if statistically significant) deviation from mean chance expectation exists, it seems obvious that little information is being con veyed on each trial; but occasional star performances suggest much greater possibilities. This paper is an initial approach to quantifying the amount of information that a well-functioning ESP percipient can get, with unexpected and important observations about present-time versus precognitive ESP functioning. Measurement Procedures Given our present state of knowledge, estimating the amount of information conveyed in a free-response study is very diffi cult and largely subjective. In this paper, therefore, I decided to deal solely with forced-choice ESP test data: card-guessing studies and similar procedures. In a forced-choice decision, standard information make a correct decision. Thus we need only one bit to make a binary decision, but 2.32 bits, on the average, to correctly make a five-choice decision. The number of bits given by this information theory convention is a standardized representation of the amount of information needed, i.e., it is standardized to the convention that all of the information is transmitted in a binary code. We know that some proportion of correct calls on any forced- choice ESP test are correct by chance alone, however, so we must factor out this chance baseline. Timm's (1973) psi coefficient (Psi[subscript+]) (Note: I use English apprxomations of greek math symbols in this ASCII text) is appropriate for this. It is calculated for ESP hitting as follows: Psi[subscript +] = (H-Np) / Nq Here H is the number of hits, N is the number of trials, p is the a priori probability of a hit by chance alone, and q = 1 - p. The psi coefficient for ESP hitting gives us a number between zero and one, and this coefficient represents the proportion of time hits were made on trials after the mean chance baseline is factored out. Perfect ESP performance on every trial of a run would give Psi+ = 1, while no ESP (only chance expectation perfor mance) would give Psi+ = 0. In comparing different experiments, we cannot simply compare statistical significance levels to see which was more successful unless the a priori probability of a hit and the number of trials in both experiments were identical. The same proportional devia tion levels in a 10-choice test as compared with, say, a two- choice test. The psi coefficient provides a straightforward comparison of the proportion of the time psi was presumably operating in two or more experiments, but it totally ignores the fact that one test may be more difficult, requiring more informa tion to be conveyed for a hit, than another. Thus I devised the following measure, R[subscript psi]^2 , (NOTE 2) to measure the average amount of information conveyed per trial in a particular experiment. R[bar][subscript psi] = (R[subscript T])*(Psi[subscript +]) The average number of bits conveyed per trial (R[bar][subscript psi]) is equal to the number of bits needed for success on an individual trial( )multiplied by the proportion of trials on which ESP was presum ably operating after chance hits are factored out (Psi[subscript +]). Maximum Channel Capacity Being an optimist about our potential ability to make ESP work well, I am interested in just how well it can work; what indications do we have of maximum possible ESP channel capacity? ----------------------------- NOTE 2: My R[bar][subscript psi] measure is not identical to the information-theoretic measures proposed by Timm (1973) or by Beloff and Bate (1971), although I have undoubtedly been stimulated to derive this measure by earlier sources in the parapsychological jour nals. -----------------------------end of footnote I model the degree to which ESP functions for given individual at a given time as a function of the individual's innate ability, how strongly this ability was (usually) sup pressed or cultivated in his or her development, and multiple psychological factors (internal state and external testing condi tions) affecting a given performance. Since we seldom seriously select for high innate ESP ability or ESP ability that has been thoroughly cultivated before a percipient comes to the lab, and since our current understanding of and ability to control psy chological factors affecting performance are very poor, it fol lows that the typical ESP performance in the laboratory is under far from optimal conditions, and so is almost useless in estimat ing maximum possible ESP performance ability. Only the best performances will be useful for this estimation. Possible channel capacity must be at least as high as (or higher than) our highest observed capacities. To estimate maximum ESP channel capacity, I quickly scanned the literature for studies that I recalled from memory that were methodologically sound and highly successful. My memory turned out to implicitly define "highly successful" primarily on the basis of extremely high (often 10^6 to 1 and better) odds against the results being due to chance. I reviewed some dozen studies selected this way and calculated the mean bit rate per trial (R[bar][subscript psi]) for them. Since I was interested in maximum performance possi- bilities, whenever studies were broken down into individual percipient data or individual run or condition data, I used the most successful unit (individual and/or condition) of analysis. Thus I used four known perfect Zener card runs (conveniently summarized in Rhine, 1964) to get four indications that maximum channel capacity might go as high as 2.32 bits per trial, rather than submerging these runs into the larger overall studies of which they were a part. As would be expected, some of the studies that were very statistically significant were not impressive at all in terms of average information conveyed per trial. Ryzl's percipient, Pavel Stepanek, for example, scored 11,978 hits in 19,350 binary calls (Ryzl, 1966), a result which has enormous odds against its being due to chance(p<< 10^-20), but this performance represents a quite low bit rate per trial: R[bar][subscript psi] = .04. I found 10 instances of bit rates that were greater than one bit per trial, the highest being two instances of 5.7 bits per trial on confidence calls for exact hits, using ordinary playing cards as targets (Kanthamani and Kelly, 1974).(NOTE 3) Thus we can tentatively conclude that maximum channel capacity for ESP may be in the five to six bits-per-trial range, and possibly higher. ___________________ NOTE 3: My own study of the out-of-body experiences of Miss Z(Tart, 1968), in which she correctly called a five-digit number in one try, suggests that bit rate might go as high as 16.6 bits per trial, but I am not including it in any data analyses. A figure based on one trial seems too unsure. -----------------------------end of footnote Is Precognition Harder? In examining the previous studies in order to estimate maximum channel capacity, I retrospectively noted that none of the very successful studies involved precognitive ESP, only present-time GESP or clairvoyance. A dozen studies is not a large sample, but it still seemed odd that not a single precogni tive study showed up. I therefore decided to systematically review the bulk of the published, successful ESP studies so that I could compare present-time ESP with precognitive ESP. LITERATURE REVIEW I reviewed the literature, subject to the following restric tions: First, I reviewed only studies in which there was clear evidence of ESP hitting as a main effect, significant at the .05 level of significance or better. I call these "successful" studies. Studies whose only significance involved differential effects, psi-missing, etc., were excluded.(NOTE 4) Second, I continued using only forced-choice studies, which lent themselves to accurate quantitative evaluation of bit rate per trial.(NOTE 5) Third, I assumed that the methodology for sound ESP experiments had been adequately worked out by the late 1930s, so for quality control I reviewed articles published in main stream, refereed parapsychology journals, namely, the Journal of Parapsychology from 1937 to the present (1980), the Journal of the American Society for Psychical Research from 1941 (the earli est date in the University library) to the present, and the European Journal of Parapsychology from 1975 (its earliest publi cation date) to the present. Since I scanned the journals by reading titles, it is possible that I have missed a few relevant studies whose titles did not make it immediately clear that they were experimental works, but I believe I have found the vast majority of successful studies. Where I already knew of relevant data published in books I turned to them, but this was rare. Fourth, I continued to use only the peak performance data from each study. And finally, I did not use data contributed by any experimenters whose integrity had later been brought into serious question, such as Soal or Levy. ______________________ NOTE 4: Given my earlier model of factors affecting ESP perfor mance in a given situation, it follows that occasionally a para psychological experiment will not show any significant signs of ESP because of some combination of the percipients' innate abili ty being too low or too thoroughly suppressed, and of the psy chological conditions of the experiment being unfavorable for ESP manifestation. Since I am interested in evaluating how strongly ESP manifests when it manifests, rather than arguing about the existence of ESP, I view statistically insignificant experiments as more representative of failures of technique for eliciting psi than of samples of levels of manifestation of ESP. ---------------------end of footnote _______________________ NOTE 5: While free-response material often seems to imply very high information acquisition rates, so that leaving it out of this study may create an important gap, we should note that the high information acquisition rates may be apparent rather than real. In forced-choice studies, percipients generally devote only a few seconds to each trial, but in free response studies they may often devote 20 minutes or more to a single response. It may be that the same low information acquisition rates apply to free-response studies, and it is just the summing up of many items over the longer time periods that creates the impression of high information acquisition rates. Obviously research is needed here that takes the time devoted to the ESP task into account. -----------------------------end of footnote Given these criteria, I found 53 successful studies of present-time ESP (either GESP or clairvoyance procedures) and 32 successful studies of precognition. All the studies are included in the reference list. As single published studies sometimes reported several experiments, there are fewer studies listed than there are data points on the graphs. Surveying the bulk of a body of published literature in this manner necessarily results in the inclusion of studies with widely varying conditions and procedures. Ideally, subgroups should be formed of relatively comparable studies. But this is not possible for the parapsychological literature here surveyed, for it would have the effect, due to the wide variations, of making the subgroups too small for useful analysis. Surveying all the relevant literature selected as described above has, on the other hand, an important advantage: small effects will not show up; only large-magnitude effects that can overcome the effects of random variation will survive such a survey. RESULTS Distribution of Bit Rates in Experimental Studies Figure 1 shows the relationship between level of ESP perfor mance and the time interval between response and target genera tion for all 85 experiments. Logarithmic scales(NOTE 6) are used for both vertical and horizontal axes because of the wide range of data to be plotted. R[bar][subscript psi] ranges from a low of .02 bits per trial to a high of 5.7 bits per trial, and the time intervals in precognition studies range from the fractional second intervals in electronic machine-tested precognition to one year. Many of the electronic machine-based precognition studies do not specify the exact interval that elapsed between the time the response was recorded and the time the random generator generated the precog nitive target, so as a graphical convention I have assumed this interval to be .2 second in all such studies. NOTE ON FIGURES: TOO COMPLEX TO APPROXIMATE IN ASCII CHARACTERS, GO BY VERBAL DESCRIPTION OR CONSULT ORIGINAL JOURNAL ARTICLE ___________________ NOTE 6: Readers used to linear coordinate graphs should remember that logarithmic graphs visually compress differences that would be striking on linear coordinates: Comparing Figure 1 and Figure 2 will be helpful. -----------------------------end of footnote As can be seen, there are many studies, both present-time and precognitive, in which the measured channel capacity is quite low. If we want to characterize the best of successful (given the sampling restrictions) experimental parapsychology as a whole, combining both the present-time and precognitive studies, we could say that clearly successful forced-choice ESP studies show an information rate of .49 bits per trial on the average, with a standard deviation of +or- 1.04 bits. The very large standard deviation here reminds us of what is shown graphically, viz., that bit rates are not normally distributed. Judging from the graph, most successful ESP experiments show an information rate somewhere between .04 and 4 bits per trial, an order of magnitude variation. The most successful conditions in success ful ESP experiments have yielded another order of magnitude better results, with occasional results up to 5.7 hits per trial. Present-Time ESP versus Precognition. Figure 1 suggests a dramatic difference in maximum bit rate per trial between pres ent-time and precognitive ESP. To test the significance of this difference, I used a Mann-Whitney U test (Siegel, 1956). (The U test is almost as powerful as a conventional t test, but does not make the assumptions of normality of distribution that would invalidate the use of a t test on this data.) As Figure 1 shows, present-time ESP results can range an order of magnitude higher than precognition tests in mean bit rate per trial. The overall distribution differences are significant (U = 462, transformed Z = -3.50) at p < 5 x 10^-4, two-tailed. Bearing in mind the above cautions about lack of normality, present-time ESP studies show a mean of .70 bits per trial (standard deviation is 1.27), while precognitive ESP studies show a mean of only .13 bits per trial (standard deviation of .14). Precognition studies show a maximum bit rate of .66 bits per trial, with most (73%) of them not exceeding .20 bits per trial. While many (56%) of the present- time ESP studies also fail to exceed .20 bits per trial, a sub stantial number show up to an order of magnitude greater bit rates. In terms of maximum channel capacity, then, present-time ESP can apparently convey up to 10 times as much information per trial on the average as precognitive ESP. GESP versus Clairvoyance. I was also curious as to whether present-time GESP procedures might be more effective than pres ent-time clairvoyance procedures: my personal bias is to expect that the act of "sending" can be an active psi process. The quality of the "sending" act would be quite variable in GESP studies in general, but still present. One of the 53 studies did not lend itself to a clear clas sification as clairvoyance or GESP, and so was omitted. Other wise there were 14 GESP studies and 38 clairvoyance ones. In spection of the distributions showed strong departures from normality, so the Mann-Whitney U test was again used for compari son. Contrary to my expectations, there is a suggestion that clairvoyance procedures transformed Z = -1.88;p < .06, two- tailed). I would not make too much of this finding, however, as a change in a single top data point from the clairvoyance to the GESP column would destroy this suggestive difference. It seems most conservative to conclude that GESP versus clairvoyance conditions seem to make no difference in this sample of success ful ESP studies. This should not be generalized to parapsy chological studies in general, however, given the restrictive sampling criteria used in collecting the present data. Precognition and Temporal Distance Since most ordinary physical energies fall off in intensity with increasing spatial distance, the question whether precogni tive performance might, analogously, fall off with increasing temporal distance(NOTE 7) between response and target creation is of interest. Ideally this should be investigated by systematic comparisons of various temporal distances within a single experi ment, where other factors are held constant; but almost no ex periments of this type have been done, so we will examine the present data. Figure 1 strongly suggests that present-time ESP and precog nitive ESP may be different processes; or, if they are a single process, there is a tremendous fall-off in performance in the interval from the present to only .2 second. For all the data plotted in Figure 1 (both present-time and precognitive), the correlation of R[bar][subscript psi] and temporal distance is not significant (r = -.05), although a linear correlation coefficient may not be appropriate here. If precognitive ESP is a different process than present-time ESP, then a possible relationship between performance and tempo ral distance should be examined for the precognitive data alone. Figure 2 plots only the precognitive data, and plots R[bar][subscript psi] on a linear, rather than a logarithmic, scale so that differences are more readily grasped. (The time scale is still logarithmic.) Again, however, the linear correlation coefficient is insignifi cant (r = -.05). Given my interest in maximum performance potential, a third way of modeling the data suggests itself: Consider the largest R[bar][subscript psi] at each time interval to be the variable of interest, with R values below that representing less than maximum performances that were determined by variables other than the inherent nature of ESP. That is, less than maximum performances may represent lowered motivation, psychological conflict, nonoptimal experimen tal conditions, etc., and only the maximal performances need be examined for their relationship to time. If we take the 13 maximal data points in each temporal category (including the present) having at least one data point, we then find that r = -.14, which is still insignificant. If we examine the precogni tive data alone, with 12 data points, r = - .17, also insignifi cant.(NOTE 8) ___________________ NOTE 7: As Douglas Dean and I realized in the course of a con versation about precognition some years ago, however, we never deal with only temporal distance in precognition. The movement of the earth around the sun and the drift of the whole solar system through space adds a spatial distance component of about 193.5 miles per second of time. -----------------------------end of footnote _________________________ NOTE 8: For those who want to speculate on possible relation ships in spite of the very low N, the following regression equa tions have been fitted to the 12 precognitive maximum data points, where t = time in seconds. Linear: R[bar][subscript psi]=(at)+b a=.19 b=3.7*10^-9 Exponential: R[bar][subscript psi]=(a)(c^bt) a=.11 b=(-7.04)(10^-9) Logarithmic: R[bar][subscript psi]=a+(b ln t) a=.42 b=-.02 Power Curve: R[bar][subscript psi]=a(t^b) a=.39 b=-.13 -----------------------------end of footnote The above correlations should be taken with reservations, however: 32 data points are too few to map a range of .2 second to 3.15 x 10^-7 seconds (one year) adequately. Therefore, while a strong difference between present-time and precognitive ESP performance is clear, the question of possible fall-off of precognitive performance with temporal difference is still open. How Frequently ESP is Used Psi coefficients were computed for the same 85 studies: the results are plotted in Figure 3. The psi coefficient is a meas ure of how frequently the ESP process is activated, regardless of target difficulty, and so can be used as a rough measure of how hard percipients tried (and succeeded) in using ESP. As with the mean bit rate per trial data, the psi-coefficient analysis shows that ESP is activated much less frequently in precognitive stud ies than in real-time ESP experiments. Because the range of psi-coefficient values is not so enor mous as that for R[bar][subscript psi] , descriptive statistics based on assumptions of normality are moderately accurate. For present-time ESP, the mean psi coefficient is .28 with a standard deviation of .32, while for precognitive ESP it is .06, with a standard deviation of .06. The ranges of the psi coefficients displayed are also quite different: no precognitive study exceeds = .33 and most (77%) are below .10, while the present-time studies show six instances of perfect ESP functioning and most (56%) show psi coefficients greater than .10. The differences are highly sig nificant (U = 389, transformed Z = -4.16;p <_ 3 x 10^-5, two- tailed). DISCUSSION To recapitulate this study's findings, an examination of the bulk of successful, quantitative studies of forced-choice ESP shows that present-time ESP experiments (GESP or clairvoyance) generally have a significantly higher mean information transmis sion rate per trial than do experiments in precognition. The most successful present-time studies show 10 times more bit acquisition than the most successful precognitive studies. I shall propose three possible theories to account for this dif ference. A Psychological Theory If precognition studies generally don't work as well as present-time ESP studies, it might be due to some actual dif ference in the processes used for information gathering in the two modes of ESP, or it might be due to a psychological factor. We could hypothesize that because precognition appears harder, given our cultural preconceptions, percipients don't try as hard, and so don't do as well, in precognitive experiments. A Two-Process Theory Another theory to account for these data, which I shall call the two-process theory, is that present-time ESP is a different mechanism or process than precognitive ESP. We could further theorize that when the precognitive process itself is operative, it is unaffected by temporal distance. This second theory must be qualified, however, by reminding ourselves that it is based on very few data points over an eight-orders-of-magnitude range of time. A Temporal-Break Theory An alternative hypothesis, which I shall call the temporal- break theory, is that there is only one, unitary ESP process, but external reality itself is strongly different for the present moment and any future moments, so that it is much more difficult to use ESP to get information about the future. Again, we could postulate as an additional theory that temporal distance per se does not matter, only the distinction between present-time and any future time, subject to the paucity of data points mentioned above. Comments on the Three Theories As shown in Figures 1 and 3, both the mean bits per trial and the psi-coefficient analyses show that ESP is activated much less frequently in precognitive than in real-time experiments. This could be interpreted as supporting the psychological theory that percipients don't try as hard (or are not as successful in activating the ESP process in spite of their trying), or as supporting a theory that precognition is inherently harder, for unknown reasons, than present-time ESP. The psychological argu ment seems moderately plausible, but one would think that of the many percipients involved in 32 successful precognition experi ments, surely some would not have an antiprecognition bias. Is cultural conditioning that uniform? Nevertheless, the psycholog ical theory deserves closer examination; it may be that the difference is best explained by some other psychological inter pretation not requiring such an antiprecognition bias. If present-time and precognitive ESP are indeed basically different processes, then they will probably show other kinds of differences in addition to the one reported here. This conse quence of the two-process theory is difficult to test at present, as ESP is characterized now as "characterless"--i.e., we do not know of any bounds of quantitative and qualitative aspects of ESP functioning that would limit it in any way and thus constitute characteristics. Given the very few studies of ESP that have been done, however, compared with the number of questions we could ask about it, we may find that ESP has many characteristics that we simply haven't observed yet. Thus a consequence of the two-process theory is potentially testable: if we do begin to identify characteristics of ESP, then the theory's implication of other differences should become testable in practice. If, on the other hand, the temporal-break theory is a better fit to reality, we would not expect to find differences in other qualities of present-time and precognitive ESP apart from differ ences inherent in the nature of time itself. What kind of theory of time can we envision that gives reality to both the present and the future, but makes the present much stronger or more psychically detectable than the future? If present-time ESP does produce stronger "signals" than precognitive ESP, might this make the discriminatory process I have called trans-temporal inhibition (Tart, 1978) more efficient? It is also possible that precognition per se may not be a single process but two or more processes. Russel Targ (Personal communication, 1981) has suggested that the plot of R[bar][subscript psi] in Figure 2 resembles radioactive decay plots where two different kinds of atoms, with differ ent half lives, are present. Thus there might be a short-term precognitive process that functions from a fraction of a second to somewhere in the neighborhood of five minutes, and a long-term precognitive process that functions at a lower level of informa tion flow but at intervals of up to one year. It is possible to carry out a post hoc analysis that shows a suggestive difference (t = 1.93, df = 28; p < .05, one-tailed) between precognitive data gathered over temporal intervals of more than five minutes versus data gathered over intervals of five minutes or less. This is an elaboration of Targ's earlier model of precognition (Targ, 1973), which predicted a fall-off in precognition with increasing temporal distance. Concepts of immediate, short-term, and long-term memory have been very useful in understanding human memory phenomena, so this type of theorizing may be of value in understanding ESP. Predictively based research is needed to explore such theories. The present finding of a large difference between present time and precognitive ESP in forced-choice tests is quite robust: it is not a small difference that would drop into insignificance with the shift of only a few data points. It might be argued that some qualitative free-response precognition data suggest high bit rates, but until we know how to adequately quantify that kind of data and take response time into account, we cannot draw on it very usefully. If new studies demonstrate strong precogni tive ESP functioning, with bit rates in the one to five bits-per- trial range, the present finding will be challenged. In addition to general research on the nature of precognition then, attempts to produce very high level precognitive performances are also necessary, as is research to determine whether we are dealing with psychological factors inhibiting precognition, two kinds of ESP, or a real quality of time. REFERENCES ANDERSON, M. A precognition experiment comparing time intervals of a few days and one year. Journal of parapsychology, 1959, 23, 81-89. ANDERSON, M., AND GREGORY, E. A two-year program of tests for clairvoyance and precognition with a class of public school pupils. Journal of Parapsychology, 1959, 23, 149-177. BANHAM, K. M. Temporary high scoring by a child subject. Journal of parapsychology, 1966, 30, 106-113. BASTIN, E. W., AND GREEN, J. M. Some experiments in precognition. Journal of Parapsychology, 1953, 17, 137-143. BELOFF, J., AND BATE, D. Psi efficiency: A formal comparison of three different measures. Journal of Parapsychology, 1971, 35, 273-289. BIERMAN, D. J. Exploratory test on the proscopic theory known as the "Advance Wave Hypothesis." European Journal of Parapsycholo gy, 1977, 1, 33-36. BLOM, J. G., AND PRATT, J. G. A second confirmatory ESP experi ment with Pavel Stepanek as a "borrowed" subject. Journal of the American Society for Psychical Research, 1968, 62, 28-45. CASLER, L. The improvement of clairvoyance scores by means of hypnotic suggestion. Journal of Parapsychology, 1962, 26, 77-87. CASLER, L. The effects of hypnosis on GESP. Journal of Parapsy chology, 1964, 28, 126-134. FAHLER, J., AND OSIS, K. Checking for awareness of hits in a precognition experiment with hypnotized subjects. Journal of the American Society for Psychical Research, 1966, 60, 340-346. FISK, G. W., AND MITCHELL, A. M. J. ESP tests with clock cards: A new technique with differential scoring. Journal of the Socie ty for Psychical Research, 1953, 37, 1-14. FISK, G. W., AND WEST, D. J. ESP tests wth erotic symbols. Journal of the Society for Psychical Research, 1955, 38, 1-7. FORWALD, H. An experiment in guessing ESP cards by throwing a die. Journal of Parapsychology, 1963, 27, 16-22 FREEMAN, J. A. An experiment in precognition. Journal of Para psychology, 1962, 26, 123-130. HARALDSSON, E. Subject selection in a machine precognition test. Journal of Parapsychology, 1970, 34, 182-191. HONORTON, C. Relationship between EEG alpha activity and ESP card-guessing performance. Journal of the American Society for Psychical Research, 1969, 63, 365-374. HONORTON, C. Automated forced-choice precognition tests with a "sensitive." Journal of the American Society for Psychical Re search, 1971, 65, 476-481. HONORTON, C., RAMSEY, M., AND CABBBO, C. Experimenter effects in extrasensory perception. Journal of the American Society for Psychical Research, 1975,69, 136-149. HONORTON, C., TIERNEY, L., AND TORRES, D. The role of mental imagery in psi-mediaton. Journal of the American Society for Psychical Research, 1974, 68, 385-394. HUMPHREY, B. M., AND NICOL, J. F. The feeling of success in ESP. Journal of the American Society for Psychical Research, 1955, 49, 3-37. HUTCHINSON, L. Variations of time intervals in pre-shuffle card- calling tests. Journal of Parapsychology, 1940, 4, 249-270. KANTHAMANI, H. Psi in relation to task complexity. Journal of Parapsychology, 1974, 38, 154-162. KANTHAMANI, H., AND KELLY, E. F. Awareness of success in an exceptional subject. Journal of Parapsychology, 1974, 38, 355- 382. KEIL, H. H. J. A GESP test with favorite music targets. Journal of Parapsychology, 1965, 29, 35-44. KELLY, E. F., AND KANTHAMANI, H. A subject's efforts toward voluntary control. Journal of Parapsychology, 1972, 36, 185-197. McCOLLAM, E., AND HONORTON, C. Effects of feedback on discrimina tion between correct and incorrect ESP responses: A further replication and extension. Journal of American Society for Psychical Research, 1973, 67, 77-85. MORRIS, R. L., ROLL, W. G., KLEIN, J., AND WHEELER, G. EEG pat terns and ESP results in forced-choice experiments with Lalsingh Harribance. Journals of the American Society for Psychical Research, 1972, 66, 253-268. NASH, C. B., AND NASH, C. S. Checking success and the relation ship of personality traits to ESP. Journal of the American Society for Psychical Research, 1958, 52, 98-107. NIELSEN, W. An exploratory precognition experiment. Journal of Parapsychology, 1956, 20, 33-39. (a) NIELSEN, W. Mental states associated with success in precogni tion. Journal of Parapsychology, 1956, 20, 96-109. (b) OSIS, K. Some explorations with dowsing techniques. Journal of the American Society for Psychical Research, 1960, 54, 141-152. OSIS, K., AND FAHLER, J. Space and time variables in ESP. Journal of the American Society for Psychical Research, 1965, 59, 130- 145. PRATT, J. G. Further significant ESP results from Pavel Stepanek and findings bearing upon the focusing effect. Journal of the American Society for Psychical Research, 1967, 61, 95-119. PRATT, J. G., AND BLOM, J. G. A confirmatory experiment with a "borrowed" outstanding ESP subject. Journal of the Society for Psychical Research, 1964, 42, 381-389. RHINE, J. B. Experiments bearing on the precognition hypothesis: I. Preshuffling card calling. Journal of Parapsychology, 1938, 2, 38-54. RHINE, J. B. Experiments bearing upon the precognition hypothe sis: III. Mechanically selected cards. Journal of Parapsychol ogy, 1941, 5, 1-57. RHINE, J. B. Special motivation in some exceptional ESP perfor mances. Journal of Parapsychology, 1964, 28, 42-50. RICE, G. E., AND TOWNSEND, J. Agent-percipient relationship and GESP performance. Journal of Parapsychology, 1962, 26, 211-217. ROLL, W. G., AND KLEIN, J. Further forced-choice ESP experiments with Lalsingh Harribance. Journal of the American Society for Psychical Research, 1972, 66, 103-112. ROLL, W. G., AND PRATT, J. G. An ESP test with aluminum targets. Journal of the American Society for Psychical Research. 1968, 62, 381-386. ROSE, L., AND ROSE, R. Psi experiments with Australian aborigi nes. Journalof Parapsychology, 1951, 15, 122-131. ROSE, R. A second report on psi experiments with Australian aborigines. Journal of Parapsychology, 1955, 19, 92-98. RYZL, M. Training the psi faculty by hypnosis. Journal of the Society for Psychical Research, 1962, 41, 234-252. RYZL, M. A model of parapsychological communication. Journal of Parapsychology, 1966, 30, 18-30. RYZL, M. Precognition scoring and attitude. Journal of Parapsy chology, 1968, 32, 183-189. RYZL, M., BARENDREGT, J. T., BARKEMA, P. R., AND KAPPERS, J. An ESP experiment in Prague. Journal of Parapsychology, 1965, 29, 176-184. RYZL, M. FREEMAN, J., AND KANTHAMANI, B. K. A confirmatory ESP test with Stepanek. Journal of Parapsychology, 1965, 29, 89-95. RYZL, M., AND OTANI, S. An experiment in duplicate calling with Stepanek. Journal of Parapsychology, 1967, 31, 19-28. RYZL, M., AND PRATT, J. G. Confirmation of ESP performance in a hypnotically prepared subject. Journal of Parapsychology, 1962, 26, 237-243. RYZL, M., AND PRATT, J. G. A further confirmation of stabilized ESP performance in a selected subject. Journal of Parapsycholo gy, 1963, 27, 73-83. SARGENT, C.L. Hypnosis as a psi-conducive state: A controlled replication study. Journal of Parapsychology, 1978, 42, 257-275. SCHECHTER, R., SOLFVIN, G., AND McCOLLUM, R. Psi and mental imagery. Journal of the American Society for Psychical Research, 1975, 69, 321-326. SCHMEIDLER, G. R. ESP and tests of perception. Journal of the American Society for Psychical Research, 1962, 56, 48-51. SCHMEIDLER, G. R. An experiment on precognitive clairvoyance. Part I. The main results. Journal of Parapsychology, 1964, 28, 1-14. SCHMEIDLER, G. R. High ESP scores after a swami's brief instruc tion in meditation and breathing. Journal of the American Socie ty for Psychical Research, 1970, 64, 100-103. SCHMEIDLER, G. R., AND LINDEMANN, C. ESP calls following and "ESP" test with sensory cues. Journal of the American Society for Psychical Research, 1966, 60, 357-362. SCHMIDT, H. Precognition of a quantum process. Journal of Para psychology, 1969, 1969, 33, 99-108. SCHMIDT, H., AND PANTAS, L. Psi tests with internally different machines. Journal of Parapsychology, 1972, 36, 222-232. SHIELDS, E. Comparison of children's guessing ability (ESP) with personality characteristics. Journal of Parapsychology, 1962, 26, 200-210. SIEGEL, S. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill, 1956. STEILBERG, B. J. Investigation of the paranormal gifts of the Dutch sensitive Lida T. Journal of Parapsychology, 1971, 35, 219- 225. TARG, R. Precognition in everyday life: A physical model. In W. G. Roll, R. L. Morris, and J. D. Morris (Eds.), Research in Parapsychology 1972. Metuchen, N.J.: Scarecrow Press, 1973. TART, C. T. A psychophysiological study of out-of-the-body ex periences in a selected subject. Journal of the American Society for Psychical Research, 1968, 62, 3-27. TART, C. T. Learning to Use Extrasensory Perception. Chicago: University of Chicago Press, 1976. TART, C. T. Space, time and mind. In W. G. Roll(Ed.), Research in Parapsychology 1978. Metuchen, N.J.: Scarecrow Press, 1979. TIMM, U. The measurement of psi, Journal of the American Society for Psychical Research, 1973, 67, 282-294. WOODRUFF, J. L., AND RHINE, J. B. An experiment in precognition using dice. Journal of Parapsychology, 1942, 6, 243-262.