TESTING CLAIRVOYANCE AND PRECOGNITION BY MANIPULATING PROBABILITIES: A CONCEPTUAL ASSESSMENT OF THE EXPERIMENTAL LITERATURE.
Conceptually, precognition is a notoriously difficult topic. Not only does it appear to imply backwards causation because a future event seems to impinge on my present mental activity, but it also seems to imply that future events are fixed, because if the future can have an impact on my brain state, the future must have sufficient reality in order to make that impact in the first place. These and similar issues have been debated extensively in both the philosophical (e.g., Broad, 1937; Dummett, 1954; Rankin, 1957) and parapsychological (e.g., Beloff, 1990; Roberts, 1993; Struckmeyer, 1970) literature.
Experimentally, the problem of whether foreseen events are fixed or alterable has been examined by distinguishing between actual and probable futures. Such precognition experiments have some items in their target pool that are more likely to become the target than others. The concept is that if participants in these experiments make more calls on the high-probability targets than on the low-probability ones, the idea that precognition directs itself towards probable (hence, preventable) futures rather than towards actual (and thus fixed) ones will arguably have received some experimental support.
However, even disregarding the conceptual problems behind such designs for now, attempting to draw conclusions from the experimental paradigm above is more complicated than the given description suggests. The first experiments to use targets of varying frequencies (when items that appeared more frequently had a higher probability of actually being the target if called) were clairvoyance ones. This alone indicates a problem in varying target probabilities to see whether foreseen events are of probable or actual futures; if clairvoyance tests vary the frequency of symbols in the target set in order to gain a better idea of how psi might operate, those conclusions about the operation of psi may hold for similar experiments in the precognition mode as well.
For example, one reason for varying the target frequency in clairvoyance tasks might be to see whether psi homes in on the most conspicuous item in the set (which could either be the most frequent or the least frequent symbol depending on their overall distribution). Alternatively, other clairvoyance experiments might examine whether varying target probabilities encourages participants to use psi in the form of a "best bet" by provoking participants to make more calls on the more frequently occurring symbols. Similarly, then, precognition experiments examining the issue of whether people foresee probable or actual futures may not be able to show definitively whether precognition is of a future that is determined or free. Instead, any accrued results could be due to ESP in general directing itself to the most interesting aspect of the future environment.
The purpose of this article is to review clairvoyance experiments using unequally distributed target sets and then to apply those findings to the experimental precognition literature on actual and probable futures. The aim is to get a clearer and more thoroughgoing conception of a direction that precognition research might take to gain insight into the problem of actual and probable futures. Consequently, this review will examine six clairvoyance studies and two precognition ones in the light of those clairvoyance studies. I will conclude by placing this discussion in the broader context of precognition research in general.
THE CLAIRVOYANCE EXPERIMENTS
The first clairvoyance/telepathy experiment to use a target set containing some symbols of greater frequency than others was that of Soal and Goldney (1942). Although Soal's work is in question because some of his results were suspected to be fraudulent (Markwick, 1985), this experiment is of interest because it is the first of its kind. The experimental procedure involved an agent and a receiver, "B. S." (Basil Shackleton), located in different rooms. The receiver would try to guess which of the 5 ESP cards the agent was looking at. A run consisted of 25 trials (using an open deck). Both agent and receiver were observed by separate people (pp. 37-40). On C. A. Mace's suggestion, on three separate dates and unknown to B. S., one run with nonrandom symbols was interspersed amongst the usual runs (pp. 68-69). This made a total of three runs with nonrandom symbols. Each of these nonrandom runs comprised 12 consecutive trials of all one symbol followed by 13 all of a different symbol. The results showed that whe n B. S. called one of the two symbols in the target set, his call would be situated significantly more often in the correct half of the run than in the wrong half. Analyses also indicated that call frequency on those symbols had been adjusted upwards correspondingly.
However, B. S. was known as a participant who putatively obtained chance results on direct hits, but who consistently showed forward displacement--that is, he consistently achieved significant psi-hitting if the card immediately following the target was taken as the target. When the targets are the same over 12 or 13 trials, though, even forward displacement hits become direct hits. Thus, in 75 trials B. S. was able to score 32 direct hits. Even if the results are in doubt, it is of theoretical interest to note that if a participant cannot focus properly on the actual target, this inability to focus can be overcome by having targets repeated in a run. Soal and Goldney made no hypotheses about these results.
Introducing Global Psi and Its Models
The first experiment to explore target sets with symbols of varying frequencies with a specific research question in mind was not undertaken until 1973. This was Child and Kelly's experiment which used an exceptional participant (Lalsingh Harribance, "L. H.") and the standard "down through" (DT) technique for calling ESP cards. Namely, a shuffled deck of ESP cards was placed face down on the desk and L.H. wrote down his 25 guesses for that pack. In this experiment, though, the packs of cards did not have five cards of each symbol, but instead had nine cards of one symbol, seven of another symbol, five of yet another, three of one of the other two remaining symbols and just one of the last symbol. There were 120 differently composed decks, these 120 decks comprising all possible permutations of this composition (with different symbols having different frequencies in different combination in each deck). All permutations were used three times, thus there were 360 runs in total. L. H. knew that the decks would n ot be of the usual composition, but he did not know in what way their composition would differ. He also knew that each deck would have a different composition in this experiment. Child and Kelly wanted to see if psi operates solely by positioning calls correctly (as must be the case in successful standard ESP experiments) or whether it can also function by adjusting calls to reflect the frequency of symbols. If L. H. called the most frequently occurring symbol more often than the other less frequently occurring symbols, he would therefore increase both his likelihood of obtaining a hit and of gaining overall positive results.
However, Child and Kelly's results were ambiguous. L. H.'s calls did not correspond exactly to target frequency. L. H. did make significantly more calls on the seven-target symbols than on the three-target ones, but he did not make significantly more calls on nine-target symbols than on one-target symbols. Even when only erroneous calls (i.e., those times when L. H. called the frequently occurring symbol and the call was wrong) were considered, the same relation between three- and seven-target symbols and the one- and nine-target symbols obtained. That is, erroneous calls of seven-target symbols exceeded mean chance expectation (MCE) for that symbol frequency, whereas erroneous calls of three-target symbols fell short of the MCE for their symbol frequency, thus indicating that call frequency was adjusted to target frequency. For the one- and nine-target symbols, again, this relation did not hold.
From these results Child and Kelly concluded that ESP was not operating exclusively by correctly positioning the calls; there was some adjustment of calls to target frequency.
To see whether the hits exceeded chance by virtue of L. H. having made more calls on the more frequent targets, Child and Kelly calculated how many hits would be expected for each target once the number of calls for it was taken into account. The results were ambiguous. Their hypothesis was that if the number of hits in relation to the number of calls exceeded MCE then psi in this experiment would not necessarily be operating just by adjusting call frequency to target frequency. Conversely, if hits did not exceed MCE in relation to the number of calls, then the extra hits were due simply to the greater number of calls for the more frequent targets. What actually happened is that hits just missed significance once the number of calls for each target was taken into account, leaving any conclusion ambiguous. Although call frequency corresponded well--but not exactly or linearly--to target frequency, successful calls could be due either to positioning the calls correctly or to adjusting the calls to the target f requency. Moreover, the raw results showed that there was high scoring on the least frequent symbol.
Child and Kelly concluded that in this experiment, call frequency was adjusted to target frequency and that this adaptation probably improved L. H.'s overall results. They believed that psi as manifested in this experiment is a global process in which the cards are seen as a whole rather than one at a time. If the cards are "seen" as a whole, they say, the single-target symbol could be the most distinctive one, thus explaining why there were more hits on the single-target symbol. However, they admit that it is difficult to explain why the nine-target symbol did not provoke extra calling. Additionally, the results are from only one participant and they may not be generalizable.
Twenty-five years later Palmer (1998) attempted to replicate Child and Kelly's findings using the same participant (L. H.). Due to lack of time, only 60 runs (compared to Child and Kelly's 360) were possible. To increase the power of the planned experiment Palmer omitted the one-and nine-target symbols because this particular distribution pair had not provoked any significant difference in calling in Child and Kelly's study. Thus Palmer used 60 packs of 25 ESP cards with a distribution of 7-7-5-3-3 with the permutations of symbols perfectly counterbalanced across the packs.
However, the results from Palmer's study showed no tendency for the seven-target symbols to be called more frequently than the three-target symbols and no evidence for psi-hitting at all. The only procedural difference between Child and Kelly's study and Palmer's--apart from the omission of the nine- and one-target symbols--is that in Child and Kelly's experiment L. H. wrote down his guesses, and in Palmer's L. H. called them out aloud.
Nevertheless, even disregarding Palmer's failure to replicate Child and Kelly's results, Child and Kelly's conclusions are contradictory. Child and Kelly decided that in their experiment L. H. had adjusted call frequency and that this helped him gain better overall results. Child and Kelly are thereby essentially promoting a strategic model of psi in which L. H. unconsciously decides to increase his calls on more frequently occurring targets in order to augment his chances of getting a hit. However, this model is in conflict with their explanation for the increase in hits on the least frequent symbol. If hits on the least frequent symbol are explained by the distinctiveness of that symbol, the model is then a purely perceptual one in which attention is drawn to the more prominent target and accurate calling on that target accordingly results. Here there is no call strategy on the part of the receiver at all; there is just direct perception of the target. It is on this perceptual model of the results that Sch meidler decided to focus.
In 1985 Schmeidler conducted a conceptually similar clairvoyance experiment but used a perceptual model of psi to explain her results. She had found post hoc that some of her data indicated that participants' responses would echo any clusters in the target set. That is, if one target appeared more frequently than others (which will happen occasionally by chance), participants would call that target more often. Because this was reminiscent of Child and Kelly's findings, Schmeidler conducted two experimental series with thirty different participants in each (thus making a total of sixty participants) to investigate this further. Schmeidler thought that if something in the target set were particularly prominent (such as a repeated target) this would draw a person's ESP to it, just as it would perceptually. That is, whereas Child and Kelly used biased decks predominantly to see whether participants can use psi strategically--by increasing calls on frequently occurring targets to enhance their chances of getting a hit--Schmeidler's interest in unbalanced decks was primarily to support a perceptual model of psi. In this model, the increased frequency of some targets would attract participants' attention and participants would make more calls on those symbols as a result.
In Schmeidler's experiment the target set consisted of 495 pages of 12 clock faces. Each page had one hour (e.g., three o' clock) represented four times with the eight other faces being of eight randomly selected different hours. One of these pages was selected and placed in a well-sealed envelope. A response sheet of 12 blank clock faces was placed over the envelope in the same position as the target page inside. Each participant filled out three different response sheets. They were informed that it was an open deck and that some targets could appear more than once and others may not turn up at all. The participants filled out the first sheet in their own time (control condition); for the second sheet they were told that sometimes scores improve if people don't think too hard about their guesses (fast condition); and for the third sheet they were informed that some participants perform better if they spend more time over their responses (slow condition). These three conditions were not counterbalanced, and participants were timed. The ESP scores took the repeated hour as the target for all twelve clock faces and each guess was scored for its difference from the target hour. The term "psi-hitting" for this study therefore refers to participants selecting significantly more often than would be expected by chance alone an hour that was relatively close to the target hour.
Each participant also took the matching familiar figures test (MFFT) after the experiment. In this test people have to select which figure they think matches the figure at the top of the page. For this test, too, participants were timed and classified accordingly as either fast or slow.
Results in the first, exploratory series showed chance results on both the repeated targets and the eight individual targets. Under the fast condition, though, when (and only when) the repeated target was taken to be the target for the eight individual targets, there was evidence of psi-hitting (N = 30, p [less than] .05). Moreover, for this condition and when only those who were categorized as fast for the MFFT were considered, the effect was yet greater (N = 18, p [less than] .005).
A second series was carried out in the same way as Series 1, this time with formal hypotheses corresponding to the results found in Series 1. The results in Series 2 were similar to (but not exactly the same as) those in Series 1. Again, participants' performance on the repeated targets and on the individual targets yielded only chance results. This time, however, there was no overall significant evidence of psi-hitting under the fast condition when the repeated target was taken to be the target for the eight nonrepeated targets (N = 30, p = .11), although the finding of psi-hitting on these targets when only those who were fast at the MFFT were considered was replicated (N =14, p [less than] .02). No ANOVAs were performed in either series to confirm that the superior performance of those fast at the MFFT really was due to individual differences.
Schmeidler concludes that these results support Child and Kelly's suggestion that ESP can respond in a global way to the target field. However, in Schmeidler's experiment the support is limited to fast ESP responses on nonrepeated targets by those who were fast at the MFFT when the repeated target was taken to be the target overall. Only this very specific group scored significantly well in both series. Schmeidler notes that it is puzzling that the scoring occurred only on the nonrepeated targets. However, psi-hitting may have been evident only on individual targets because they had greater statistical power than the repeated targets. In her paper Schmeidler reports only that scores on repeated targets were nonsignificant; she does not provide the statistical details. In respect to her findings, Schmeidler argues that a significant effect might show up only on fast responses because in normal perception, too, quick glances comprehend any overriding pattern and overlook individual differences. That is, Schmei dler's model is essentially one in which the use of global psi in a "fast" setting provokes certain ("fast") participants to call but it does not enable them to place their call at the correct time.
It is possible that one reason for Palmer's failure to replicate Child and Kelly's findings is that the experimental procedure in Palmer's study allowed L. H. to make very fast calls, sometimes so fast that the recorder could not keep up. In Child and Kelly's study, L. H.'s calls would have to have been slower because he had to write his calls down. Thus, if individual differences based on timing are crucial to success or failure, this difference between verbal and written calls may have been crucial in testing L. H.
Although Schmeidler was inspired by Child and Kelly's findings, it is not easy to compare the two sets of results. Both conclude that ESP can be a global process that scans the whole target set and that participants can adjust their calling pattern in a suitable manner. What is less clear is the extent to which call frequency does adjust to target frequency and the model of psi that any adjustment of call frequency is supposed to support.
In fact, it appears that underlying these experiments there are at least three models--two perceptual and one strategic--and that the papers simply switch between these models as convenient. Moreover, these models are in conflict with each other.
The first of the two perceptual response models, the "homing-in" model, holds that perceptually distinct targets draw the participant's attention to them. The participant is thereby able to "home in" on the target(s). This is, effectively, the model that Child and Kelly use to explain the successful hitting on the single-target symbols.
However, it is also possible to use another perceptual response model, the "call-provocation" model, to explain Child and Kelly's findings on the single-target symbol. The call-provocation model holds that perceptually distinct targets provoke more calls because participants can get the information about the targets, but participants are not necessarily able to time (and hence locate) their calls well. This appears to be the model of psi in Schmeidler's study insofar as it focused on how relatively quickly or slowly participants filled out their response sheets and their speed at completing the MFFT in relation to their ability to call the target hour relatively accurately. The call-provocation model differs from the homing-in model purely by allowing calls to be either accurate or inaccurate. The central issue for the call-provocation model is one of the timing of the calls--the targets provoke a suitable response, but the timing or location of those calls may or may not be appropriate. Thus Child and Kelly 's finding on the single-target symbols can also be explained by the call-provocation model: the participant is provoked to make a call because the single-target symbol is perceptually prominent. In Child and Kelly's experiment the participant made the call at the right time (hence at the correct location on the response sheet) and thus gained more hits.
Nevertheless, the call-provocation model cannot be used to explain L. H.'s increase in calls for the seven-target symbols. Normally on the call-provocation model, one would argue that L. H. had responded to the seven-target symbols but that he was not able to make his responses at the right time. This is why L. H. did not achieve a significant number of hits once the number of calls was taken into account. However, using the call-provocation model to explain L. H.'s psi-hitting on the one-target symbol, it was assumed that L. H. was good at timing his responses. Consequently, the call-provocation model cannot explain all of Child and Kelly's findings with L. H.
In fact, Child and Kelly appear to opt for a strategic model. They believe that by perceiving the target set globally, L. H. was able to sense that seven-target symbols were relatively frequent and he (unconsciously) called more often on those symbols (strategically) to increase his chances of doing well. Nevertheless, the strategic model does not explain why L. H. failed to increase his calls on nine-target symbols or why L. H. obtained psi-hitting on single-target symbols.
Additionally, the strategic model cannot explain Schmeidler's results. Although it could be argued that the fast participants' success when the repeated target is taken to be the target for the individual faces is due to those participants making more calls that are closer to the repeated target on the individual faces, this does not support the strategic model. Under the strategic model, participants would sense the predominance of repeated targets and would increase their calls to enhance their number of hits. However, if the aim is to get direct hits, the strategy does not work because there was no significant scoring on the repeated targets themselves. Moreover, if the strategic model is revised so as to allow for the possibility of a strategy failing (and nevertheless remaining a strategy), the allegedly better performance of those fast at the MFFT on the fast page still makes the call-provocation model more tenable. A strategy should fail because the strategy was wrong and not because of the type of pe rson who used that strategy. The way in which a target set affects a percipient (and the strategy depends on participants in some sense successfully perceiving the target set), however, may well depend on the individual idiosyncrasies of that person. Thus a purely perceptual model fits Schmeidler's results better.
In conclusion, there appears to be no one model that can explain all the results, and thus these experiments are not as complementary as they may first seem. Moreover, Child and Kelly's study used L. H. as a special participant, whereas Schmeidler's experiments used only ordinary volunteers. It is not clear that these two population types will be comparable--especially given that Schmeidler's findings appear to emphasize individual differences as a factor for success in such experiments.
In addition, although the perceptual response and strategic models are closely linked (having a call strategy requires the participant to have perceived the target set), the inability to merge the two sets of results under any one particular model--other than psi as putatively capable of working as a global process--suggests that none of the models is the complete story. Figure 1 illustrates the structure of the models outlined so far; it is beyond the scope of this article to discuss how the models may or may not relate to each other. My aim is simply to bring these models into view so that future experiments can be better oriented conceptually. Moreover, in both Schmeidler's and Child and Kelly's experiments the postulated aim was ambiguous and it is possible that this may have some bearing on the results. That is, in Child and Kelly's experiment there was a conscious aim (psi-hitting) and an unconscious aim (adjusting calls to target frequency), and in Schmeidler's experiment there were the actual targets (the four repeated targets and the individual targets) and the intended targets (all targets to be classified as the same as the four repeated targets).
Although both Child and Kelly and Schmeidler conclude that psi appears to be able to work as a global process, neither of them set out to test explicitly this hypothesis alone. Child and Kelly wanted to see if L. H. would adjust his calling pattern; their idea was that L. H. would make such an adjustment because it would be to his advantage. Schmeidler, on the other hand, wanted to see if an unbalanced target set would provoke more calls; she wanted to see if the global process involved was a perceptual one. The primary question of whether psi can operate as a global process at all rather than as a process directed to single targets irrespective of any particular model of how such a global process might work did not arise until Carroll B. Nash's experiment in 1985.
The Basic Question--Is Global Psi Possible?
In 1985 Nash supervised each of three experimenters conducting a clairvoyance experiment with twenty participants. The aim of the experiments was to see if (a) participants would be able to detect the most frequently occurring card in a set of five random ESP cards; (b) the scoring would be sufficiently high to indicate that psi was operating as a single ESP act over five cards rather than as separate ESP perceptions of each card; (c) results would be better when there was a greater difference between the frequencies of symbols in the set (e.g., when there are three of the same symbol and two different ones than when there are three of one symbol and two of another); and (d) participants would perform better when the repeated cards were the first and last cards in the set than when they occurred in the middle positions of the set.
The cards used were open decks of 100 ESP cards composed from random number tables. The experimenter would place the deck face down on the table, take five cards off the top and lay them face down in front of the participant. Experimenters were asked to ensure that the faces of the bottom cards were not exposed to participants. The participant would then guess the most frequently occurring card in that set of five. There was trial-by-trial feedback; sets in which no card occurred more frequently than others were bypassed.
Out of the three experimenters only one produced significant results (significant enough to render the whole study significant) and only the results of this experimenter (after correcting for selection) were analyzed for questions (b)-(d). However, there was no formal statistical evaluation of the difference in results across experimenters. Nevertheless, after correcting for selection, the findings from this single experimenter still remained highly significant (p [less than] [10.sup.-4], two-tailed). All results were reported as supporting the formal hypotheses listed above. Namely, Nash claims that the results indicate that participants did not get information from each card to make their guesses and he therefore suggests that the ESP act in this experiment was a single one over the five cards.
However, it is difficult to know whether the global-psi hypothesis can be supported through the analysis that Nash provides. There is no condition in which participants guess single targets to compare with the global-psi condition. Nash also assumes that the only alternative to global psi would be for participants to guess all five cards individually. However, a participant could get good results by waiting until they had perceived just two repeated cards and then guessing that this card would be the most frequently occurring one. Thus, although the results from one experimenter were impressive with a scoring rate far greater than would normally be expected from a forced-choice experiment (.28 hits per call, MCE = .20), it is difficult to draw any firm conclusions from these results in respect of the global-psi hypothesis.
Nash reports that hypothesis (c) was also supported: the number of hits declined as the difference between the number of most frequently occurring cards and the number in the next most frequently occurring one in that same set declined.  This suggests that the greater the difference, the easier it is to perceive the more frequently occurring card. The final hypothesis was likewise confirmed: when the repeated cards were on the outside rather than the inside (supporting the view of primacy and terminal salience effects) the number of hits was greater.
Nash notes that the evidence for psi was spread equally among all the experimenter's participants. Nash concludes with Tyrrell that ESP is a two-stage process in which targets are perceived unconsciously and then the information gets distorted as it filters into consciousness. Thus, according to Nash, the difficulty observed when the number of the cards' difference decreases is due not to a decline in ESP function, but to a difficulty in introducing the information to normal consciousness. Nash presumably holds this view because he thinks the results indicated that psi operated in this experiment as one ESP act over five cards. He notes that postulating an information filtering process to explain the concomitant decline of hits with the decline of the difference in card frequency is consistent with the super-psi hypothesis, whereas if one maintained there was a decline because it is harder to "see" the difference when that difference is small the super-psi hypothesis would be untenable. In other words, if re sults are worse when the difference between the most frequently occurring symbol and the next most frequently occurring one is less, this does not necessarily entail that it is harder to perceive the most frequently occurring set; the psi process could be perfectly clear in both cases--the information just gets more distorted when it enters consciousness if the two sets of frequently occurring targets are similar in number.
Although the results are hard to generalize because they come from only one experimenter and thus from only one third of all trials conducted, and although there are a number of problems in the design of the experiment and in the assumptions behind the statistical analyses, it nevertheless overcomes some of the conceptual problems inherent in Child and Kelly's and in Schmeidler's designs. A hit is based on perceiving the most frequent card; if perceptual distinctness (the ability to receive the symbol into consciousness) is defined as better when the number of cards' difference between frequently occurring cards in a set is greater, then Nash's experiment confirms a perceptual model of psi. Nash's experimental design also tests more specifically the idea that ESP can be a global process than either Child and Kelly's or Schmeidler's studies did, although because of the lack of a control condition it is hard to say whether or not the experiment showed what Nash claims it did.
However, Nash's experiment does not examine (nor was it meant to examine) if
the operation of global psi can be manifested behaviorally by participants making more calls to the target in question. This is mainly because Nash is not concerned with the finer-grained question of whether psi can operate by taking advantage of existing probabilities (i.e., the possibility of psi being strategic), or whether an excessive number of one symbol will provoke more calling on that symbol, or even whether the greater frequency of that symbol helps the percipient to "home in" on that target, but is concerned purely with seeing if psi can operate at all as a global (and perceptual) process.
As a result, the outcome of Nash's experiment differs from Child and Kelly's and Schmeidler's because Nash's results--if valid--would be relatively unambiguously in favor of a global-psi process. One reason why the combined outcomes of both Child and Kelly's and Schmeidler's experiments were more ambiguous is because both sets of investigators were, in part, interested in the psychology behind any adjustment of call frequency to target frequency. For Child and Kelly, the person adjusts their calls strategically because they want to get hits; for Schmeidler, the target frequency has a psychological effect on the participant who will respond according to their own individual psychological make-up. It was the idea of the psychological effect of unbalanced decks that prompted Crandall and Covey's (1986) interest.
Global Psi: Another Model
Crandall and Covey were primarily interested in the psi-missing displacement effect (PMDE). Influenced by Child and Kelly's and Schmeidler's work, Crandall and Covey conducted some post-hoc analyses on their data. They hypothesized that perhaps the PMDE exhibited in their previous experiments had been due to psi-missers having been more influenced by background stimuli than psi-hitters had. Thus, subsequent analyses on their experimental results should demonstrate that the psi-missers had been more prone to distraction by differing target frequencies than psi-hitters.
These previous experiments had originally tested the hypothesis that psi-missers are improperly focused. They thought that instead of hitting the target, psi-missers pick up on the symbols immediately preceding or succeeding the actual target. Crandall and Covey's research had led them to believe that when using an open deck, psi-missers would score significantly above chance on such displacements, whereas psi-hitters and those scoring at chance would not. Because Tart (1983) had reported that precognition does not work as well as real-time ESP, they had used a precognitive design. They had hoped all displacements would thereby be channeled into the preceding targets only and that forward displacement would be unlikely to occur.
In Crandall and Covey's experiments, participants had guessed on a computer which target would occur two seconds after their guess (targets were the numbers 1-5). Participants had not been given any feedback until they had completed all the runs. In this way, any forward displacement would have been by precognition (and thus unlikely, according to Crandall and Covey). Participants had also been cautioned that targets may not occur with equal frequency in any short run. There were three series: the first with 30 participants, the second with 15 participants, and the third with 25 participants.
When Crandall and Covey saw Schmeidler's and Child and Kelly's work, they returned to the results of these experiments. In these experiments hitters and missers had been defined as those who had scored above or below chance (20%) respectively over 100 trials, but displacement scores had been used only from runs of 25 trials in which hitters or missers had scored above or below chance (5) respectively. For the new analyses, Crandall and Covey investigated only those runs in which there had been a difference of five or more between the most and least frequently occurring targets in that run (e.g., if in a given run the most frequently occurring target appeared seven times and the least frequently occurring target occurred only twice: 7-2 = 5). They tried to see whether participants had celled the more frequently occurring target significantly more often than the least frequently occurring target. Crandall and Covey expected missers and not hitters to have adjusted their call frequency, for they expected only m issers to have been distracted by background stimuli (i.e., the targets in the previous trials). All analyses conducted were reported (but did not include any computer simulations to safeguard against possible statistical artifacts).
The results showed that missers in Series 1 and 2 had indeed called the more frequently occurring targets significantly more often than the least frequently occurring ones. Hitters had shown no such significant adaptation to target frequency. However, Series 3 showed that missers had no tendency to have increased their number of calls on the more frequently occurring targets.
Crandall and Covey's post-hoc analyses are somewhat the converse of Nash's experiment. Nash's experiment examined only the ability to detect precisely which was the most frequently occurring target, whereas Crandall and Covey's analysis looked purely at the ability to adapt the frequency of calls to the target frequency, without noting whether these particular calls match the targets. Thus, whereas Child and Kelly's and Schmeidler's work intermingle these two possibilities by looking at both call frequency and direct hits, Nash's and Crandall and Covey's studies each consider just one of these possibilities. Interestingly, both Nash's and Crandall and Covey's results are limited to only one experimenter; Series 3 in Crandall and Covey's analysis was conducted by a different experimenter from Series 1 and 2.
It is also notable that what Crandall and Covey characterize as "background distraction" for psi-missers (i.e., an increase in target frequency producing an increase in call frequency--and decreasing direct hits) is what Child and Kelly essentially characterize as "strategic enhancement" for a psi-hitter (i.e., a psi-hitter increased call frequency on seven-target symbols over three-target symbols in order to increase hits). For Crandall and Covey, however, psi-hitters appeared to be oblivious to target frequency. It thus appears that what were supposed to be complementary findings are in fact at odds with each other.
Crandall and Covey also offer a different interpretation than Schmeidler does. In Schmeidler's experiment, the increase in the proximity of the calls to the target face would have been counted as a psi-hitting trend (taking the frequently occurring target to be the target for the individual clock faces as well). For Schmeidler, this in turn would have implied that the "call-provocation" model was at work, according to which a higher frequency of a given clock face provokes more calls closer to that clock face. Although Crandall and Covey seem to take a similar perspective to Schmeidler, they decide to characterize the provocation to call as a "distraction" rather than as an essentially successful, if misplaced, response. Thus, once more, an analysis that aims to replicate previous findings about global psi actually fails to support them in any direct manner. Although the global-psi hypothesis is yet again supported, their understanding of global psi conflicts with the previously suggested models.
Indeed, Crandall and Covey's model of global psi seems to be one in which psi-hitters are able to block frequently occurring targets in order to make more successful calls, whereas psi-missers cannot. Thus, on this "blocking" model psi-hitters and psi-missers should be equally successful in global-psi experiments such as Nash's, because psi-hitters will know that they do not have to block, and it will be advantageous to psi-missers to perceive the set of targets globally. In experiments with individual targets alone, however, psi-hitters perform better because they can strategically "block" any background noise such as frequently occurring targets (whereas psi-missers cannot block). Figure 2 now shows a revised summary of the models of global psi outlined in this paper.
Nevertheless, Crandall and Covey looked only at relatively extreme changes in target frequency and did not look at the relation of hits to calls on those specific targets. In Child and Kelly's experiment, the most frequently occurring target did not provoke L. H. to make the same strategic adjustment of calls as he did on the middle frequencies. Thus it is possible that Child and Kelly's findings did not apply to the most frequently occurring target because L. H. was attempting to focus and thus partly managed to block out the most distracting (i.e., the most frequently occurring) target, but did call more often--and relatively accurately--on the middle frequency. It may also be that Palmer failed to replicate Child and Kelly's findings because the very high frequency target was not there for L. H. to block and to help him focus. Thus, whereas in Child and Kelly's experiment, putatively blocking the most frequently occurring target (the nine-target symbol) seems to have made the seven-target symbols more con spicuous to L. H. and the three-target symbols less visible, the same breadth of difference (i.e., 7 - 3 = 4) is not available in Palmer's study. In Palmer's study, if L. H. were to block the seven-target symbol, only the five- and three-target symbols would remain for L. H. to focus on. These give a difference of only two and such a small difference may be perceptually negligible. This explanation for Palmer's failure to replicate Child and Kelly's findings would support Crandall and Covey's contention that psi-hitters are able to focus better and that they are less distracted by background patterning. Similarly, Schmeidler's results could have showed up only on nonrepeated targets simply because there was an inability to focus properly. Indeed, one should not expect studies in which both psi-hitting and adaptive call frequency are applied to have consistent results, because the former is applicable to a clear focus and the latter is applicable to an inability to center on the target.
If this way of unifying the results from all the experiments outlined so far holds, then the model of psi at play is a global one of a perceptual nature with those who can focus well being capable of blocking frequently occurring targets when necessary. As noted before, the strategic models rest on the target set being globally perceived in the first place. However, only the blocking model can explain why Child and Kelly should not find an increase in call frequency for the highest frequency target. Moreover, if psi-hitters block out other targets in order to score well, this would suggest that psi is primarily a global process. Nevertheless, the blocking strategy has not yet received much experimental attention and remains to be tested explicitly. In addition, only a little experimental work has tested the idea of whether psi is primarily a global process. Thus only carefully designed future research will show whether or not the refined model of global psi offered in this paper may hold.
In sum, then, a review of the clairvoyance experimental literature pertinent to unbalanced target sets has revealed some support for the notion that psi can operate in some way as a global process--for example, by scanning the whole target environment rather than by perceiving each individual item separately. It remains for future experiments to show if psi operates better as a global process and/or if psi even prefers to work globally in the laboratory situation. Nevertheless, even if it is discovered that psi works better and/or preferentially as a global process, not only would questions remain as to how globally psi can operate (how widely across space and/or over time), there would also be questions about what theoretical implications global psi has. This survey alone has revealed four models implicit in the literature--two perceptual response models ("call provocation" and "homing-in") and two strategic ("best bet" and "blocking") and no thoroughgoing attempt has yet been made to devise experiments to separate them out and compare them. However, one of the main purposes of this review is to bring out these possible models. The second major aim is to show how precognition experiments using varying target probabilities also fail to home in on the issues they hoped to address. Moreover, precognition research is further complicated by the possibility of global psi, and hence by the various models outlined in this section. In the following section I will outline the precognition experiments using varying target probabilities, discuss the questions underlying them, and I will then focus carefully on one of these questions alone. From this subsequent discussion a program for precognition research will emerge.
THE PRECOGNITION EXPERIMENTS
The Experiments Outlined
Only two precognition experiments have examined whether precognition is of actual or probable futures. The first one was conducted by Targ and Targ in 1986. Interestingly, Targ and Targ explain their experiment as one that examines the difference between the actual and probable existence of future events, rather than as one that transfers the clairvoyance experiments above into a new setting. For example, they do not ask whether precognitively available background stimuli can cause the psi response or whether precognition acts globally rather than directing itself towards individual targets, etc. Instead, Targ and Targ seem to assume that future events are not real in the same sense as clairvoyant targets are and that the perceptual analogies examined in varying target probabilities in clairvoyance experiments therefore will not be of relevance. I will return to this point later.
In their paper, Targ and Targ contrast two possible views of future events. The first view is that until a future event happens, many different futures coexist. What distinguishes one possible future from another is the probability of their occurrence. Some futures, according to present circumstances, may be more likely to occur than others. In this view, precognition of a future event will depend on the probability of that future occurring.
Targ and Targ, however, favor an alternative viewpoint. In their view, the actual future event has some property distinct from all the others. Here targets should be perceptible even if they are relatively unlikely to come about because the actual future target will have a distinct (if unknown) property.
In their experiment, Targ and Targ used two experienced viewers: Keith Harary (K. H.) and Hella Hammid (H. H.). Each viewer completed six trials in which they tried to view which object in a set of six would be chosen randomly an hour later. Both viewer and experimenter were blind to the set of target objects. One of these objects had a 50% chance of being selected later and the other five had only a 10% chance each of being selected. It turned out that in five trials, a low-probability object was the one to be randomly selected later, and in seven trials, it was a high-probability one. There was trial-by-trial feedback for both receiver and experimenter. A single pair of judges each evaluated the viewers' taped mentation, drawings, and notes for one trial only on a scale of 1-100 to describe the amount of correspondence between the viewers' impressions and each of the target objects. The judges then ranked the six objects for that trial accordingly; the judges were of course blind both to the assigned probability to each object and to what the target had actually been.
Using the sum of ranks, the results showed significant psi-hitting on the five targets assigned a low probability and nonsignificant results on the seven targets assigned a high probability of being selected. In addition, when targets were of low probability, there was no excess of high rankings for the high-probability target.
Targ and Targ conclude that in this experiment, an actual event makes itself recognizable regardless of the probability of its occurrence.
This experiment differs from all the others reviewed so far not only because it has a precognitive rather than a clairvoyance design, but also because it uses a free-response protocol rather than a forced-choice one. It is notable that the probabilities for the targets' selection were defined at the time of the remote viewing. Thus, in contrast to the clairvoyance experiments summarized above, in Targ and Targ's experiment what is clairvoyantly available is only the probability of each object's possible occurrence and not its actual repetition. To this extent, the experiment was more similar to a traditional ESP experiment in which all targets are equally available. For the receiver to perceive clairvoyantly which item was most likely to be selected, the receiver would have had to have focused accurately on the probability attached to each target and then inferred which item was most likely to become the target. Obviously, in this particular experiment this is not what happened.
However, although this particular experiment indicated that the actual future event made itself recognizable, the number of trials was small (N= 12) and only two experienced remote viewers were used. So, even if the results were an accurate reflection of what these particular viewers foresee, the results are not necessarily easily generalizable to a larger population. Moreover, if Targ and Targ claim that what was accessed was the future that actually occurred, it is curious that the viewers did not accurately perceive those targets that had a high probability of being selected. If only an actual future is accessed, the fact that the target had a high probability of occurring should be irrelevant: the high-probability targets should still be just as accessible as the low-probability ones. It thus seems that motivational factors--such as the challenge of successfully getting a "difficult" target--may account for the superior performance of the low-probability targets in this experiment.
Moreover, had the viewers succeeded predominantly with those objects that had a high probability of being selected, the interpretation of the results would have been ambiguous, especially if one takes the view that future events already have some form of perceptible reality. Indeed, by promoting the view that the future event has some distinct property, Targ and Targ seem to be claiming that future events are real enough to be perceptible in the present. But if high-probability items in some sense coexist in alternative futures and if they (by definition) turn up in more futures than low-probability ones, the same questions that arose in clairvoyance experiments with unbalanced decks will arise in respect of precognition experiments. For example, will the multiplicity of futures in which a high-probability target comes about make that target more perceptually distinct? I will return to this problem in some detail below.
The only other experiment to date examining actual or possible futures is that of Radin in 1988. Radin postulated that although Schmeidler refers to the problem of unbalanced target sets as one of a difficulty in separating figure from ground (i.e., of correctly timing one's response to a perceptual stimulus), such experiments could also be viewed as people adopting a strategy to select the most likely target (i.e., like the "best bet" model). The latter is what Radin wanted to investigate; his hypothesis (and hunch) was that precognition is of probable rather than of actual futures.
In this experiment Radin was both participant and experimenter and a total of 60,000 trials were conducted. The following is a simplified approximation of the experimental procedure. The experiment was computer-based. There were six boxes on the screen; any of these boxes may later hold the target. One box (randomly selected on each trial) is assigned a high probability (this high probability itself being randomly selected between 0.3, 0.5, 0.7, or 0.9)and the other five boxes were each assigned an appropriate low probability (1 - high probability/5). The participant would select one of the boxes, not knowing which box housed which probability. Once the selection had been made, the PRNG would choose a target between 1 and 6 using the biases allocated to the six boxes for that particular trial. The computer would then put the target into the randomly selected corresponding box so that the participant could have feedback of their failure or success. Appropriate measures were in place to counteract potential re sponse bias effects.
Radin's hypotheses were that (a) there would be evidence for precognition overall; (b) there would be more calls of high-biased targets (supporting both probable and actual futures) and that the number of times a high-bias box was selected when it did not house the target (i.e., the number of erroneous high-bias calls) would be greater than chance (supporting a probable futures interpretation); and (c) for erroneous high-bias calls there would be a positive relationship between the magnitude of the bias and the effect size.
Radin's results showed overall hitting (taking into account the number of hits expected for each target probability) and thus provided evidence for precognition. There were also significantly more calls of high-biased targets than would be expected by chance and evidence of a significant number of erroneous high-bias calls. The latter, therefore, supported the probable futures hypothesis. However, the third hypothesis (a relationship between the size of bias and effect size) was not confirmed.
Radin's results are, therefore, the reverse of Targ and Targ's findings. Interestingly, Targ and Targ seem to believe that had they had success on high-probability targets, this would have supported the view that many different futures coexist. Radin, however, believes that the success of probable futures in his experiment indicates that precognition is possibly due to extrapolation from real-time information to the most probable outcome. That is, Radin believes that he may have known through real-time psi where the high-probability target would be and that he guessed more times on that target to increase his chances of getting a hit.
Clarifying the Experimental Questions
Interestingly, neither Targ and Targ's experiment nor Radin's study are well-designed to answer the questions that motivated their research. Targ and Targ appear to be primarily interested in determining whether the actual future has a distinct property and/or whether a future that is highly probable involves the coexistence of many different futures. It is hard to know what Targ and Targ mean either by the "coexistence" of many futures or by their supposition that an actual future may have a "distinct" property. However, it seems that ultimately their question is one about the reality of future events; whether the actual future has a distinct property or whether probable futures coexist, the implication either way is that somehow the future has some kind of perceivable reality.
Radin's focus, however, seems to be on whether apparent precognition really is of future events or whether it is simply a sophisticated form of inference from the present to the most likely future outcome. However, the answer to this question is found more readily in experimental comparisons between a condition in which the future event would at best be very difficult to predict and one in which it would be relatively easy to infer the future outcome. For instance, it would be relatively difficult for someone correctly to predict the target using real-time information alone if the target were determined by using stock market figures on a prespecified future date. Stock-market figures are determined by many varying human decisions, and thus the closing stock-market price on a future date should be extremely difficult to predict with accuracy. In contrast, a target in which the information and conditions for the future target selection are already in place would be relatively easy for a person to infer using r eal-time ESP.
Simply using some items in the set that are highly likely to become the target, as Radin did, will not answer the question of whether precognition is actually due to real-time psi. As noted when discussing Child and Kelly's experiment, good results on high-probability targets may simply be a strategic move. Although this is in keeping with the idea that true precognition is not possible, it may be that the participant prefers to use a strategy even if they could in fact have foreseen the actual futures directly. Moreover, Radin reports that his own predisposition was to think that precognition is of probable futures. If psi reflects or serves a person's motivations, it is hardly surprising that Radin's experiment indicates that a psychic strategy may have been at play so that probable targets outperformed the low-probability ones.
Radin's leading question about whether true precognition is possible, however, is well taken. It does not make sense to ask anything about the type of reality that foreseen events have until we know whether or not events really can be foreseen. If true precognition is possible, though, then Targ and Targ's question about the reality of foreseen events is one of the first that comes to mind. As outlined at the beginning of this article, if true precognition is possible, the problem immediately arises as to what kind of reality the foreseen event can have if it can have an impact on my present cognition.
In the following, I will assume that true precognition is possible and that it is a future event that is foreseen. Given this assumption, and the assumption that only true precognition was at work in Targ and Targ's experiment, a clearer conception of the question that Targ and Targ's (or a similar) experiment may have attempted to tackle can emerge. It appears that the experiment asked whether (a) the future is perceptible because it does come about and this gives the future event a special quality, or (b) potential futures have a perceptible reality prior to their fulfillment quite independent of whether they ever do come about. Thus, had both low-and high-probability objects been equally successful, Targ and Targ's (or a similar) experiment would have indicated that it is the fulfillment of the future that is perceived (i.e., that fulfilled futures have a distinct property). However, if high-probability objects had been more successful, this might have indicated that futures have a perceptible reality pri or to their fulfillment, because prior to their fulfillment high-probability objects will appear in more potential futures than low-probability objects; that is, many different futures coexist.
Nevertheless, this latter interpretation of the question underlying Targ and Targ's research assumes that global psi is possible (insofar as it presupposes that all potential futures are scanned by the viewer), that true precognition need not direct itself to a future that will be fulfilled (even though the viewer's aim is to foresee the fulfilled future and not the unfulfilled ones) and, by implication, that futures can exist prior to their fulfillment. Moreover, when considered in detail, possible results would not always clearly differentiate between the two types of future event--fulfilled and unfulfilled. (I am using the term "unfulfilled futures" to refer to the idea that futures may have a perceptible reality prior to and independently of their potential fulfillment. Fulfilled futures, by contrast, are only those which do actually come about.) My contention is that Targ and Targ's experiment cannot differentiate between fulfilled and unfulfilled futures. I will show that this is at least in part due t o competing models of global psi, any one of which may potentially explain the results.
For instance, still assuming that only true precognition is operative, if high-probability targets obtained more hits than the low-probability ones, this could be due to any of the four global-psi models I outlined earlier. First, the viewer could have scanned unfulfilled futures and adopted a successful "best bet" strategy by calling high-probability targets more often. Here it appears from the results that the viewer is directly seeing the fulfilled high-probability futures, but in fact the viewer is scanning only the prior range of potential unfulfilled futures and placing an appropriate bet. Second, the prior predominant set of potential unfulfilled high-probability futures may have helped the viewer to "home in" on the fulfilled high-probability target more easily by highlighting the inherent likelihood of that high-probability target. In this case, the viewer foresees both the potential unfulfilled futures and the fulfilled high-probability one. Third, a background of unfulfilled futures may prompt the participant to call more frequently on those targets--especially when the fulfilled future is itself a high-probability target, thus rendering the high-probability targets that much more dominant again. The participant would thereby gain more hits on high-probability items (i.e., a "call-provocation" model might apply). Here, again, it is hard to say whether the viewer foresees the fulfilled future, the unfulfilled futures, or both. Last, if the participant were a psi-misser, the participant may be unable to block the high-probability unfulfilled futures. The participant may therefore make more calls on high-probability targets and thus score well on those items, but not on the low-probability items. Here the problem is not one of not being able to see the fulfilled future, but an inability to focus on that future because of other competing distractions. Other interpretations of the results had they favored high-probability targets are doubtless also possible. However, in all four of the interpretations abov e, both fulfilled and unfulfilled futures could play a role in the formulation of the participant's guess.
As a result, findings from experiments that vary target probabilities to determine the reality of fulfilled and unfulfilled future events will be ambiguous on two accounts. First, such experiments will always fail to show whether precognition is of fulfilled or unfulfilled futures in any given case because it is always possible that both the fulfilled and unfulfilled futures are foreseen and not just one or the other. Second, even if results could indicate that unfulfilled futures have their own perceivable reality, it would be difficult to interpret such a finding because so many models of global psi are available. It is possible that in some cases analyses of call frequency, for example, may help to distinguish between various models. However, as it stands, precognition experiments using targets of varying probabilities will not even be able to show whether people foresee fulfilled or unfulfilled futures, and thus to this extent any question about what model of global psi would be at play is moot.
This discussion about the question that motivated Targ and Targ's and Radin's experiments serves to show not only how hard it can be to formulate a research question clearly, but also how truly difficult it is to test that question in the right manner. Whereas the clairvoyance experiments using varying target probabilities generally appeared to support the global-psi hypothesis, they, too, were not clearly linked to each other conceptually, despite their having been conducted in the backdrop of previous research, and therefore the questions were often not targeted in an entirely appropriate and unifying manner.
The aim of the discussion in this section has been to show that if true precognition is possible, then a fundamental issue is whether futures can have some kind of perceptible reality prior to their fulfillment. If they do, then--as shown previously--models pertaining to global psi may be pertinent in interpreting the results of experiments using varying target probabilities. A related issue is whether people foresee the crucial decision point and perhaps also its immediate consequences (i.e., the as-yet-unfulfilled futures), or whether people look directly into the future (i.e., the future event). In experimental work this could perhaps be understood as the distinction between target selection (the point at which the target is decided) and target fulfillment (the way in which the target is realized in real-life). The following section will outline a series of experiments designed to address this issue.
TESTING WHETHER WHAT IS FORESEEN IS THE FUTURE DECISION OR THE FUTURE EVENT
In the following I shall sketch a series of experiments that could be performed to test whether it is a future decision that is foreseen by precognition or a future event. The proposed experiments will be outlined in only enough detail to make the necessary conceptual points. For the sake of argument, I will assume throughout that true precognition has already been sufficiently well-demonstrated. Thus the first issue now will be to see whether or not psi does operate globally and whether it operates globally across both space and time.
To test the hypothesis that psi operates globally across space using a free-response precognition setting with video clips as targets, two conditions are required. In both conditions the participant is asked to guess which of four video clips they will see playing on the TV screen in two days' time. In the first (standard, direct psi) condition, the participant is shown the target as feedback. In the second (global psi) condition, the participant would be shown the target as feedback as before, but this time the target would also be displayed simultaneously on (say) six screens in one or more other empty room(s) and without the participant's knowledge that this is happening.
If results favored the global-psi condition, this will indicate not only that psi operates better in the global-psi condition but also that it acts preferentially as a global process (because if the option is there for it to act globally, it does). In both conditions the participant's aim and feedback is the same. The only difference between the two conditions is that in one condition, at the time of feedback the target clip is also displayed simultaneously on a number of screens in one or more other rooms without the participant's knowledge. Thus, if there were better results when the target was displayed on a number of screens, it would be due to psi operating globally (scanning a larger spatial environment than the feedback alone) in order to improve its chances of getting the required information. Therefore, it would appear that if psi can operate globally in space, it will.
A second experiment would test whether psi can operate globally over time. In this second experiment, the two conditions would be the same as above, but this time in the global-psi condition, the target would not be displayed on a number of screens in one or more rooms; instead, the target would be displayed nonsimultaneously a number of times the day after the participant had received their feedback (unknown to the participant). If the global-psi condition provided better results in this experiment, these results would presumably be due to the participant using their psi globally over time to see if looking further forward into time gave any more indication about what the correct target would be. If the global-psi hypothesis were supported in both its spatial and temporal forms, it would subsequently be possible to test whether or not it is a future decision or a future event that is foreseen.
For instance, there could be an experiment using two conditions. In the first condition there are perhaps six items. The participant is asked to guess which item they will later see. A while after the participant's guess and some time before they get feedback (say, two days before feedback), one of the items is selected as the target. To guess the target even at this stage, the participant would have to use precognition. However, at this point no one knows which item it is. Two days later for example), the participant is shown the target. In the second condition, the procedure is the same, only here, at the time of feedback and unknown to the participant, the target item is displayed simultaneously a number of times in one or more empty room(s). The participant gets only the same feedback as in the first condition (i.e., they are shown the target item once). Unlike the global-psi experiments, though, the condition itself (i.e., multiple or single display) is chosen randomly only at the time at which the targ et is about to be displayed.
This experiment to examine global psi differs from the ones previously outlined because in this new experiment there is a two-day time lag between the target selection and the participant's feedback. By contrast, in both of the other global-psi experiments, all items were as likely to become the target right up to the last minute before the participant's feedback. The time-lag also allows a difference to come in between target selection and condition selection (the condition being selected at the time of feedback). If, as in this proposed experiment, in one condition the target is fulfilled or displayed more times than in another, under the global-psi hypothesis one would normally expect the condition in which the target is displayed more times to be the more successful one.
Consequently, if only the second condition (in which the target is displayed on several screens at the time of feedback) provided good results, this would suggest that participants look forward to their future feedback (i.e., target fulfillment). Indeed, if psi is preferentially a global process, as would have been demonstrated beforehand in the global-psi experiments above, participants should normally be able to make more accurate guesses when the target is also displayed simultaneously on several screens in another room.
However, if participants performed equally well in both conditions, it would appear that what is foreseen is the target selection and not the target fulfillment. In this experiment, the target selection is not equivalent to the condition selection (which occurs only after the time-lag period). Thus in this new experiment, both conditions could produce similar results only if participants look forward to the target selection (i.e., the crucial decision point) and its immediate consequences (i.e., the fact that once the target selection has been made, the target item's possible future becomes p = 1 and all other target possibilities effectively disappear), rather than to the fulfillment of that decision as such (the condition selection and subsequent display). In this new experiment, the condition selection is irrelevant to the target selection and it is, after all, the target selection that is the participant's goal.
In the previous global-psi experiments, it was always determined right from the beginning of any given trial what condition that trial would be in. Thus, from the outset of a trial in the global-psi condition, for instance, it is already determined that when the target is selected, it will be followed by a multiple display of that target. The global-psi experiments may have been successful because participants looked forward to the selection of the target and the immediate or intrinsic potential of that target (i.e., the fact that when that item was selected it would be displayed many times) instead of the multiple display itself.
However, because there is a difference in the new experiment between target selection and condition selection, if both conditions (i.e., multiple and single display of target) produced equally good results, precognition must focus on the crucial decision point (i.e., the target selection) and the immediate or intrinsic consequences of that selection, rather than on the fulfillment of that decision. At the point at which the target is selected, it is not intrinsically likely to be displayed more or less times because the condition selection regarding the display occurs only later. Therefore, at the target selection point, the target in both (to be decided later) conditions becomes only that which is going to be shown to the participant; all other target possibilities at this point are, in effect, no longer possibilities at all. If what is foreseen is the target selection and the target's immediate or intrinsic potential, both conditions should provide the same results, because it is only at the condition-sele ction point that there is a difference between the conditions. Moreover, the similarity in the two conditions' results could not be due to the participant merely having directly foreseen their feedback alone--which is the same in both conditions--because they would have previously performed better in the experiment to test global psi when there was a multiple display of targets elsewhere. At the time of feedback, the target item is intrinsically tied up with the condition selection. Thus in this new experiment, if participants look forward to the time of feedback, they should perform better under the second condition (with a multiple display). If they do not do better in the second condition, then it appears that what is foreseen is the target selection and not the target feedback (or fulfillment).
The differentiation between target selection and target feedback is an important one. If it can be shown that target selection is foreseen and not target feedback, this would indicate that precognition does not foresee how things will be; it looks forward to the relevant information about what will be decided. This is relevant to the idea of whether or not we can intervene in future events, because if we look forward to how things will be, these things can presumably not be changed; however, if we precognize what the inherent implications of a given decision are (in a broad sense of the term "decision"), it may be possible for us in some cases to alter those decisions or their surrounding circumstances if we so desire.
A simple experiment investigating feedback alone would not be able to make the same distinction between target selection and target fulfillment because, as in the global-psi experiments above, the target selection is intrinsically tied up with the type of feedback. In a feedback! no-feedback experiment it would be impossible to know if, for example, the feedback condition performed better because the feedback itself was what was foreseen, or whether the feedback condition did better because participants looked forward to the target selection and its intrinsic or immediate potential. If participants looked forward to the target selection and also psychically knew that this time the target would be followed by feedback, this alone might be enough to give them more motivation. That is, they would still be looking forward to the target selection--including its intrinsic potential (here: feedback)--and not necessarily to the feedback itself.
If the proposed experiment separating target selection and target fulfillment produced results in which the first condition (where the target is displayed only once) performed better, the findings would be difficult to assess; if the future multiple display becomes irrelevant (as this outcome would suggest), the two conditions are basically identical and there would be no reason to think that either one should perform better than the other. The structure of the whole series of experiments is given in Figure 3.
As noted before, any experiments investigating what it is that precognitive experiences foresee rest on the assumption that true precognition is possible. However, what the proposed set of experiments shows is that even if true precognition is possible, it is a long and complicated line of research that will pave the way to understanding precognition in any depth.
In addition, as noted at the beginning of this paper, precognition experiments that vary target probabilities have sometimes linked themselves conceptually to the question of whether it is possible to intervene in foreseen events. Nevertheless, such experiments do not necessarily have a bearing on this question at all. What is foreseen might well be highly improbable at the time of the precognition and the foreseen event might already exist in some (as yet undefined) sense, but it need not necessarily be fixed as such. Indeed, in some respects, "probable" futures could be regarded as more "fixed" than a relatively improbable one.
For example, if I can take paths A, B, D, or E to get to F, then it is relatively likely that I will select an appropriate path; where I end up is more "fixed" on this account than if only path A led to F. The real issue is whether the foreseen future is fixed, and we cannot know this until we know whether there are any breakable links that surround and lead up to that event.
In another example, someone may predict that I will be in London tomorrow, even though I hadn't planned to go. The next day I get on the bus to go to work and someone hijacks it and takes it to London. This would be an extraordinary occurrence, but, on its own, it does not necessarily appear to be fixed that I am in London that day. For example, I could have walked to work or I could have decided to take the day off. In both of these cases I would not have gotten on the bus and I would not have gone to London.
However, it may be that even if I had walked to work I would have been run over and rushed by ambulance to a specialist hospital in London. Alternatively, it may have happened that had I decided not to go to work that day, I would have found that I had won a competition and I may have been whisked to London in a taxi to collect my prize.
This account now makes it appear as if my going to London is set in stone. However, given that we do not know the limits of psi, we also cannot know the limitations or encouragement to any given event's coming about. Thus, if I wanted to avoid going to London and if psi is not limited, I may have been able to use my PK to affect the environment in some way so that the various forms of transport fail to arrive in London. Here, then, the event is not fixed after all. We are unable to know what is within or outside the realm of possibility until we know the limits of psi. And if we do not know what is or is not possible, we cannot tell whether any given event can be avoided.
The more important issue in asking whether we can intervene in future events is to decide why it is we want to know. Is it because we need to know whether or not the experience of free will is illusory, either quite generally or for a given set of circumstances? But we cannot hope to know whether we can make things turn out otherwise until we know more about the limits of psi. Thus, here the primary aim might be to determine whether psi does have any limits. The aim of the experiments outlined in this article address this question in part insofar as they attempt to determine whether what is foreseen is something that in principle could be changed; they do not address the question of whether they could be altered in fact.
However, there are other legitimate questions underlying the interest in the intervention problem. For example, the underlying question may be to learn something about the nature of time. I have argued elsewhere (Steinkamp, 1997) that this is fundamentally the question that lies behind the intervention paradox.
Alternatively, someone's interest in intervention may really be one about whether precognitive experiences do actually serve a purpose. At least two possibilities spring immediately to mind here: precognitive experiences may serve as a mental preparation for what is ahead so that the percipient is better able to cope with the event when it does occur; another possibility is that precognitive experiences serve as warnings about events that will happen unless the person does something to avoid it. Both of these possibilities are experimentally testable, although it is not this paper's aim to outline research strategies for all possible questions. There are of course many other potential definitions of purpose (not all of which may be experimentally testable); for example, precognitive experiences may be of evolutionary value. All the questions are nevertheless interlinked and it is possible that, together, answers to at least some of these questions may point to an overall picture of how and why precognition m ight work.
This article has shown that precognition research may do well to conceive of itself in the broader context of other related experiments. It has demonstrated that research findings are not always as complementary as may first be believed and that differing models of psi often underlie people's conclusions. These models need to be brought to the fore before any coherent and programmatic research can proceed.
I am grateful to the Fundacao Bial for their financial support of this project. This paper has also benefited greatly from a number of helpful comments from Julie Milton, Bob Morris, John Palmer, and Caroline Watt. An anonymous referee also provided careful, considered and detailed feedback that helped to improve this paper considerably.
(1.) For example, the difference between three cards of one symbol and two of another symbol (one Set of 2) is 1 (i.e., 3 - 2) and the difference between three cards of one symbol and two different cards is 2 (i.e., 3 - 1).
BELOFF, J. (1990). Backwards causation. In J. Beloff, The relentless question: Reflections on the paranormal (pp. 78-87) .Jefferson, NC: McFarland.
BROAD, C. D. (1937). The philosophical implications of foreknowledge. Proceedings of the Aristotelian Society. Supplementary Volume 16, 177-209. (Also in J. Ludwig, Ed., 1978, Philosophy and parapsychology, 287-312. New York: Prometheus Books.)
CHILD I. L., & KELLY, E. F (1973). ESP with unbalanced decks: A study of the process in an exceptional subject. Journal of Parapsychology, 37, 278-297.
CRANDALL, J. E., & COVEY M. K. (1986). A model for receptivity to background stimuli in ESP research, with further data. Journal of the American Society for Psychical Research, 80, 393-408.
DUMMETT, M. (1954). Can an effect precede its cause? Proceedings of the Aristotelian Society. Supplementary Volume, 28, 27-44.
MARKWICK, B. (1985). The establishment of data manipulation in the Soal-Shackleton experiments. In P. Kurtz (Ed.), A skeptic's handbook of parapsychology (pp. 287-312). Buffalo, NY: Prometheus Books.
MORRIS, R. L. (1982). Assessing experimental support for true precognition. Journal of Parapsychology, 46, 32 1-336.
NASH, C. B. (1985). Clairvoyant determination of the most frequent of five cards. Journal of the Society for Psychical Research, 53, 26-30.
PALMER,J. (1998). ESP and REG PK with Sean Harribance: Three new studies. Proceedings of Presented Papers: The Parapsychological Association's 41st Annual Convention, 124-134.
RADIN, D. I. (1988). Effects of a priori probability on psi perception: Does precognition predict actual or probable futures? Journal of Parapsychology, 52, 187-212.
RANKIN, K. W. (1957). Order and disorder in time. Mind, 66, 363-378.
ROBERTS, F. S. (1993). Time-perception and precognition: A resolution of the intervention paradox. Journal of the Society for Psychical Research, 59, 141-148.
SCHMEIDLER, G. R. (1985). Field and stream: Background stimuli and the flow of ESP responses. Journal of the American Society for Psychical Research, 79, 13-26.
SOAL, S. G., & GOLDNEY, K. M. (1942). Experiments in precognitive telepathy. Proceedings of the Society for Psychical Research, 47, 21-150.
STEINKAMP, F. (1997). Backwards causation, precognition and the intervention paradox. Proceedings of Presented Papers: The Parapsychological Association 40th Annual Convention, 410-425.
STRUCKMEYER, F. R. (1970). Precognition and the "intervention paradox." Journal of the American Society for Psychical Research, 64, 320-326.
TARG, E., & TARG, R. (1986). Accuracy of paranormal perception as a function of varying target probabilities. Journal of Parapsychology, 50, 17-27.
TYRRELL, G. N. (1947). The "modus operandi" of paranormal cognition. Proceedings of the Society for Psychical Research, 49, 65-120.
|Printer friendly Cite/link Email Feedback|
|Publication:||The Journal of Parapsychology|
|Date:||Jun 1, 1999|
|Previous Article:||Where Reincarnation and Biology Intersect.|
|Next Article:||ANOTHER VIEW OF THE PARANORMAL BELIEF SCALE.|