This feature was first published as one of CSI's special web exclusives. The link below will take you to the article.

It takes a special kind of person to write insightfully about quantum mechanics and mathematics—and literature and religion and pseudoscience and conjuring and philosophy. And it takes a very special kind of person to be able to do so in a way that is comprehensible, enlightening, entertaining to the intelligent layperson, and worthy of the respect of experts. Such a rare person was Martin Gardner, and his achievements are all the more impressive given that he was largely self-taught and without advanced degrees in physics, mathematics, literature, or philosophy.

I knew Martin Gardner the icon rather well, and I owe him a considerable debt for what I have learned from him over the years. When I was an undergraduate physics student, my classmates and I avidly devoured his “Mathematical Games” column in Scientific American, along with his published collections of mathematical puzzles and enigmas and his other books on science and mathematics. He helped make mathematics and physics delightful to pursue. Later on, when I switched disciplines and became a graduate student in psychology, I turned to his writings once again when I was asked—nay, virtually ordered—by the department chair to prepare a critical examination of extrasensory perception (ESP) for presentation to undergraduate students who were clamoring for such a talk. I knew nothing of the subject at the time; so where was I to begin, given the apparent paucity of critical literature on the subject? I dug out Fads and Fallacies in the Name of Science and let Martin introduce me to the subject. That simple beginning unexpectedly led me into decades of critical commentary and debate with regard to parapsychology.

Later on, as a young psychology professor, I was researching how people maintain their beliefs in the face of contradictory evidence. I wanted a demonstration that would confront subjects with ostensible evidence that something they held to be absolutely true was apparently false. Where to begin? I turned to Martin Gardner once again: I began re-reading some of his books and articles and soon came across the perfect vehicle for my research—a puzzle, invented a century earlier by Sam Lloyd but preserved and analyzed by Gardner, in which a piece of paper of a certain area, when cut into pieces and the pieces rearranged, appears quite clearly to have gained in area. This was ideal for my purpose, for psychologists had long believed that all of us acquire in childhood a firm belief in “conservation of area,” so that we “know” that area cannot be changed by the rearrangement of component parts.

I had always been very impressed by Martin Gardner the icon, but I was fortunate enough to be able to meet Martin Gardner the man. This came about when I was made a member of the CSICOP executive council. With this appointment, I was delighted that I would rub shoulders with the man himself, for he was one of the founders of CSICOP and a member of its executive council. I soon learned, however, that he was averse to travel and did not usually attend council meetings. I therefore had to wait to meet him until a meeting was held in Atlanta, which was near enough to his home at the time that he did indeed attend. Martin the man proved to be as impressive as Martin the icon. He was gentle, intelligent, witty, modest, curious, and filled with creative energy and imagination. A longtime fan such as I could not avoid feeling a little star-struck, although it was very clear that stardom was the last thing that he wanted. I remember our first conversation very well: he was a major contributor to the conjuring literature, and when he learned that I was an amateur magician, he immediately and graciously responded by sharing with me a new magical effect that he had just invented. I was struck by his warmth, his lack of pretense, and his excitement about sharing new ideas.

As I reminisce, I see that Martin has had an important influence on me—as he no doubt has had upon countless others who have been devoted to his scholarship—for a very long time. Whether as Martin the icon or Martin the man, he has enriched our lives. We shall all miss him.

This paper will also appear in the March/April 2011 issue of Skeptical Inquirer.

A flurry of media attention is being directed toward the prepublication distribution of Daryl Bem’s forthcoming research paper “Feeling the Future: Experimental Evidence for Anomalous Retroactive Influences on Cognition and Affect.”1 Bem claims to have found evidence of marvelous psychic abilities that transcend time and allow the future to reach backward to change the past. Both the academic stature of its author, a respected emeritus professor of psychology at Cornell University, and the fact that it was to be published in the American Psychological Association’s (APA) Journal of Personality and Social Psychology have made this report particularly newsworthy.

Daryl J. Bem

Parapsychology has long struggled, unsuccessfully, for acceptance in the halls of science. Could this article be its breakthrough? After all, the article apparently provides evidence compelling enough to persuade the editors of the world’s preeminent social-psychology journal of its worthiness. However, this is hardly the first time that there has been media excitement about new “scientific” evidence of the paranormal. Over the past eighty-odd years, this drama has played out a number of times, and each time parapsychologists ultimately failed to persuade the scientific world that parapsychological phenomena (psi) actually exist. Recalling George Santayana’s now-clichéd dictum, “Those who cannot remember the past are condemned to repeat it,” we should approach Bem’s work using a historical framework to guide us. Consider the following:

1. In 1934, Joseph Banks Rhine published Extra-Sensory Perception (Rhine & McDougall, 1934/2003), summarizing his careful efforts to bring parapsychology into the laboratory through application of modern psychological methodology and statistical analysis. Based on a long series of card-guessing experiments, Rhine wrote: “It is independently established on the basis of this work alone that Extra-Sensory Perception is an actual and demonstrable occurrence” (p. 210). Elsewhere, he wrote: “We have, then, for physical science, a challenging need for the discovery of the energy mode involved. Some type of energy is inferable and none is known to be acceptable . . .” (166).

Despite Rhine’s confidence that he had established the reality of extrasensory perception, he had not done so. Methodological problems with his experiments eventually came to light, and as a result parapsychologists no longer run card-guessing studies and rarely even refer to Rhine’s work.

2. Physicist Helmut Schmidt conducted numerous studies throughout the 1970s and ’80s that putatively demonstrated that humans (and animals) could paranormally influence and/or predict the output of random event generators. Some of his claims were truly extraordinary: for example, that a cat in a garden shed, which was heated only by a lamp controlled by a random event generator, was able—through psychokinetic manipulation—to turn the lamp on more often than would be expected by chance. Schmidt’s claim to have put psi on a solid scientific footing garnered considerable attention, and his published research reported very impressive p values.2 In my own extensive review of his work, I concluded that Schmidt had indeed accumulated impressive evidence that something other than chance was involved (Alcock 1988). However, I found serious methodological errors throughout his work that rendered his conclusions untenable, and the “something other than chance” was attributable to methodological flaws.

As with Rhine, excitement about Schmidt’s research gradually dwindled to the point that his work became virtually irrelevant, even within the field of parapsychology itself.

3. The 1970s gave rise to “remote viewing,” a procedure through which an individual seated in a laboratory can supposedly receive psychic impressions of a remote location that is being visited by someone else. Physicists Russell Targ and Harold Puthoff claimed that their series of remote-viewing studies demonstrated the reality of psi. This attracted huge media attention, and their dramatic findings (Targ and Puthoff 1974) were published in Nature, one of the world’s top scientific journals. At first, their methodology seemed unassailable; years later, when more detailed information became available, it became obvious that there were fundamental flaws in procedure that could readily account for their sensational findings. When other researchers repeated Targ and Puthoff’s procedure with the flaws intact, significant results were obtained; with the flaws removed, outcomes were not significant (Marks and Kamman 1978, 1980).

Add Targ and Puthoff to the list of “breakthrough” psi researchers whose work is now all but forgotten.

4. In 1979, Robert Jahn, then dean of the School of Engineering and Applied Science at Princeton University, established the Princeton Engineering Anomalies Research (PEAR) unit to study putative paranormal phenomena such as psychokinesis. Like Schmidt, Jahn was particularly interested in the possibility that people can predict and/or influence purely random subatomic processes. Given his superb academic and scientific credentials, his claims of success drew particular attention within the scientific community. When his laboratory closed in 2007, Jahn concluded that “over the laboratory’s 28-year history, thousands of such experiments, involving many millions of trials, were performed by several hundred operators. The observed effects were usually quite small, of the order of a few parts in ten thousand on average, but they compounded to highly significant statistical deviations from chance expectations” (PEAR, n.d.).

However, parapsychologists themselves were among the most severe critics of his work, and their criticisms were in line with my own (Alcock 1988). More importantly, several replication attempts have been unsuccessful (Jeffers 2003), including a large-scale international effort led by Jahn himself (Jahn et al. 2000).

5. In the 1970s, the ganzfeld, a concept borrowed from contemporaneous psychological research into the effects of sensory deprivation, was brought into parapsychological research. Parapsychologists reasoned that psi influences may be so subtle that they are normally drowned out by information carried through normal sensory channels. Perhaps if a participant were in a situation relatively free of normal stimulation, then extrasensory information would have a better opportunity to be recognized. The late Charles Honorton carried out a large number of ganzfeld studies and claimed that his meta-analysis3 of this work substantiated the reality of psi. Hyman (1985) carried out a parallel meta-analysis that contradicted that conclusion. Hyman and Honorton (1986) subsequently published a “joint communiqué” in which they agreed that the ganzfeld results were not likely to be due to chance, but they thought that replication involving more rigorous standards was essential before final conclusions could be drawn.

Daryl Bem subsequently published an overview of ganzfeld research in the prestigious Psychological Bulletin (Bem and Honorton 1994), claiming that the accumulated data were clear evidence of the reality of paranormal phenomena. That effort failed to be convincing, in part because a number of meta-analyses have been carried out since then with contradictory results (e.g., Bem et al. 2001; Milton and Wiseman 1999). Recently, the issue was raised again in the pages of Psychological Bulletin, with papers from Storm et al. (2010a, 2010b) and Hyman (2010). While Storm and coworkers argued that their meta-analyses demonstrate paranormal influences, Hyman pointed to serious shortcomings in their analysis and reminded us that the ganzfeld procedure has failed to yield data that are capable of being replicated by neutral scientists.

Because of the lack of clear and replicable evidence, the ganzfeld procedure has not lived up to the promise of providing the long-sought breakthrough that would lead to the acceptance of psi by mainstream science.

Add Honorton (and Bem the first time around) to the list.

The lesson in this history is that new claims of impressive evidence for psi should give one pause. Early excitement is often misleading, and as Ray Hyman has pointed out, it often takes up to ten years before the shortcomings of a new approach in parapsychological research become evident.

One must also keep in mind that even the best statistical evidence cannot speak to the causes of observed statistical departures. Statistical deviations do not favor arbitrary pet hypotheses, and statistical evidence cited in support of psi could as easily support other hypotheses as well. For example, if one conducted a parapsychological experiment while praying for above-chance scoring, statistically significant outcomes could be taken as evidence for the power of prayer just as readily as for the existence of psi.

Another key consideration is that parapsychology’s putative phenomena are all negatively defined: to claim that psi has been detected, all possible normal influences must be ruled out. However, one can never be certain that all normal influences have been eliminated; the reader of a research report has only the experimenter’s word for it.

This point brings us to a related concern. Research reports involve an implicit social contract between experimenter and audience. The reader can evaluate only what has been put into print and must presume that the researcher has followed the best practices of good research. We assume that the participants did actually participate and that they were not allowed to use their cellular telephones during the experiment or to chat with other participants. We assume that they were effectively shielded from cues that might have inappropriately influenced their responses. We assume that the data were as reported—that none were thrown out because they did not suit the experimenter—and that they were analyzed appropriately and in the manner indicated. We assume that equipment functioned as described and that precautions reported in the experimental procedure were carefully followed. We take for granted that the researcher set out to test particular hypotheses and did not choose the hypotheses after looking at the data. We must take all this on faith, for otherwise any research publication might simply be approached as a blend of fact, fantasy, skill, and error, possibly reflecting little more than the predilections of the researcher. Obvious methodological or analytical sloppiness indicates that the implicit social contract has been violated and that we can no longer have confidence that the researcher followed best practices and minimized personal bias. As Gardner (1977) wrote, when one finds that the chemist began with dirty test tubes, one can have no confidence in the chemist’s findings and must wonder about other, as yet undetected, contamination. So, when considering Bem’s present research, not only do we need to look at the data, but—following the metaphor—we need to assess whether Bem used clean test tubes.

Bem’s Research

Bem describes a series of nine experiments that “test for retroactive influence by ‘time-reversing’ well-established psychological effects so that the individual’s responses are obtained before the putatively causal stimulus events occur.” His stated goal is “to provide well-controlled demonstrations of psi that can be replicated by independent investigators.” He defines psi as denoting “anomalous processes of information or energy transfer that are currently unexplained in terms of known physical or biological mechanisms.”

EXPERIMENT 1: Precognitive Detection of Erotic Stimuli

Each trial in this experiment involved the presentation of an erotic, negative, or neutral picture. The participant sat in front of a computer screen and was tasked to predict which of two curtains had a picture behind it. Only after the participant had chosen a curtain by depressing a key did the computer select a picture at random and present it behind either the left or the right curtain.

Each participant was presented with thirty-six of these trials and was given feedback on each one. The erotic pictures were considered to be “explicit reinforcement for correct ‘precognitive’ guesses,” although no effort was made to determine whether they were indeed reinforcing anything. The main hypothesis was that participants would be able to identify the position of the hidden erotic picture significantly more often than by chance.

So far, clear enough. But then things become quite messy: we learn that “most” of the pictures used in the experiment were selected from a databank, the International Affective Picture System. Bem then states that each session (a “session” refers to all the trials of an individual participant) involved thirty-six trials of randomly intermixed erotic and non-erotic pictures (eighteen of each). However, we soon learn that not all sessions were conducted in this way: the first forty of the one hundred sessions (that is, those of the first forty participants) involved twelve trials of erotic pictures, twelve of negative pictures, and twelve of neutral pictures! (The distinction between the “non-erotic” pictures seen by the majority of the participants and the “neutral” pictures seen by only the first forty is unclear.) To muddle things even more, Bem then states that the remaining sixty sessions involved “18 trials of erotic pictures and 18 trials of non-erotic positive pictures with both high and low arousal ratings. These included pictures featuring couples in romantic but non-erotic situations. . .” (emphasis added). How many were of high or low arousal weighting, or what those terms even mean, he does not say.

What is going on here? Setting aside the confusion about the stimulus, no competent researcher dramatically modifies an experiment two-fifths of the way into it! To do so is to seriously compromise any subsequent analysis and interpretation.

But that is not all. Bem next indicates that in all the experiments using highly arousing erotic or negative stimuli, “a relatively large number of non-arousing trials must be included to permit the participant’s arousal level to ‘settle down’ between critical trials. This requires including many trials that do not contribute directly to the effect being tested.” This leaves us not knowing how many trials were actually run and wondering by what method the researcher determined the number of non-arousing trials that were needed to ensure that even the most randy of participants would “settle down.”

So by this point, it is not clear how many trials were actually presented to each participant or even whether they all received an equal number of trials. It is unclear just what the stimulus materials were, and we are faced with a procedure that was changed partway through the experiment.

Just when one thinks that this study cannot be made any more confusing, Bem informs us that he discovered in Experiment 5 (which turns out to have been conducted prior to Experiment 1!) that “women showed psi effects to highly arousing stimuli but men did not.” In light of this odd complication, Bem states that “we introduced different erotic and negative pictures for men and women in subsequent studies, including this one using stronger and more explicit images from Internet sites for the men. We also provided two additional sets of erotic pictures so that men could choose the option of seeing male-male erotic images and women could choose the option of seeing female-female erotic images” (emphasis added).

By now, a careful reader is totally confused as to what went on in this experiment. Now, we find that participants were allowed to choose their target set! This is one of the most baffling descriptions of research materials and procedures that I have ever encountered.

In reflecting on the extremely unusual change in procedure during the experiment—when the appropriate course would be to run two different experiments—one cannot help but wonder if two experiments were indeed run, and when each failed to produce significant results the data from them were combined with the focus shifted to only the erotic pictures common to all participants. Surely that was not done, for such an action would make a mockery of experimental rigor.

Data Analysis: Bem states that “the main psi hypothesis was that participants would be able to identify the position of the hidden erotic picture significantly more often than chance (50%).” At first, this claim is puzzling. Although sixty of the participants completed eighteen trials with erotic pictures and eighteen trials with “non-erotic positive pictures”—therefore making the chance outcome 50 percent of the thirty-six trials—the other forty participants received twelve trials with erotic pictures, twelve with negative pictures, and twelve with neutral pictures. For them, the chance outcome would be 33.3 percent. However, it turns out that Bem combined the data for success or failure, but on the erotic pictures only, from all one hundred sessions (i.e., from all one hundred participants) and then applied t-tests4 to assess whether identification of the future position of erotic pictures occurred significantly more frequently than the 50 percent rate expected by chance. We are also informed that the hit rate for non-erotic pictures—whether they were neutral, negative, positive, or romantic and non-erotic—did not differ significantly from chance. (This is the first mention of “romantic but non-erotic,” which adds to the confusion.)

Now we have learned that the focus of the experiment is on the erotic pictures presented to the participants, but no information is provided regarding how participants with three choices scored on erotic pictures as compared with those who had only two choices; one wonders why this is so.

The data analysis was conducted through multiple t-tests without any correction for that multiplicity. We are informed that there were at least seven such t-tests, and the only significant outcome was that the one hundred participants “identified the future position of erotic pictures significantly more frequently than the 50% hit rate expected by chance: 53.1%.” This was stated to be statistically significant at p = .01. However, that significance level is simply incorrect. This kind of error (Type I)5 increases with the number of t-tests conducted, and given that there were at least seven such t-tests with a criterion of p ≤ .01, the actual probability associated with each of these t-tests is 1 –(.99)7=.06 one-tailed.6 Thus, none of these t-tests is actually statistically significant, not even at a more generous .05 p value. It is simply unacceptable that Bem did not correct for multiple testing, despite indications later in his report that he is familiar with one such correction technique, the Bonferroni t-test.7

Another reason for concern is Bem’s deliberate use of one-tailed t-tests, which provide a simpler criterion to meet than the two-tailed tests generally employed by parapsychologists. (Parapsychologists typically interpret both above-chance and below-chance scoring as indicative of psi, and thus they do not make specific predictions about the direction of the extra-chance scoring.) When we say that something is significant at the .01 level two-tailed, this means that we would expect these results to occur by chance alone only 1 percent of the time. But, given that either above-chance or below-chance results are considered to be meaningful, this 1 percent must be distributed in both directions. Thus, above-chance results would be significant at the .01 level two-tailed only if they are so extreme that they would be expected to occur by chance only half of 1 percent, or 0.5 percent, of the time. The same applies for below-chance results.

Bem also reports that he carried out a nonparametric binomial test8 on the overall proportion of hits on erotic targets across all trials and sessions, but he offers no adequate rationale for using more than one type of significance test for the same data. The test is redundant and offers nothing beyond the t-test.

Then, after having examined the data, he introduces the possibility that introversion/extroversion may play a role in presumed precognitive ability. He suggests that it may be an extrovert’s “susceptibility to boredom and the tendency to seek out stimulation” that underlies observed correlations between extroversion and psi performance reported in the literature. However, rather than using existing, well-documented measures of stimulus-seeking, he constructed his own such scale comprising two statements, reversed in scoring: “I am easily bored” and “I often enjoy seeing movies I’ve seen before.” The content and construction of this scale is bewildering. Proper scale construction involves precise and often difficult work, including operationalization of the construct, finding items that can be demonstrated to relate to the construct, endeavoring to ensure that the increments on the response measure are of approximately equal size, and establishing satisfactory reliability and validity of the final scale. Bem has ignored these considerations. As a result, the arbitrary assignment of numbers to participants’ responses on this “scale” is unjustified and misleading.

Nonetheless, Bem correlates responses on the scale with the participants’ “psi scores” and reports a significant correlation, but only for those participants whose scores on his scale fall above the midpoint. Participants who score below the midpoint on the scale did not score significantly above chance on either erotic or non-erotic trials.

Overall Evaluation: Just about everything that could be done wrong in an experiment occurred here. And even if one chooses to overlook that methodological mess, Bem’s data still do not support the claimed above-chance effect because of the multiple-testing problem.

It is difficult to have confidence that the other eight experiments, some of which were carried out earlier than the one just described, were conducted with appropriate attention to experimental rigor: We have toured the laboratory; we have found the dirty test tubes and the mislabeled vials; we have observed inappropriate methodology and analysis. We have lost confidence in the chemist, and there seems little need to poke about further.

Nonetheless, go on we must.

EXPERIMENT 2: Precognitive Avoidance of Negative Stimuli

This study involved 107 female and forty-three male undergraduate students. Using a computer, each participant first responded to Bem’s two-item stimulus-seeking scale and then completed a sequence of thirty-six trials in which a “low arousal affectively neutral” picture was presented side by side with its mirror image. The participant depressed a key to indicate which picture he or she liked better. Only after the preference was registered did the computer randomly choose which of the two pictures would be considered the “target.” If the participant had chosen this target, the computer thrice flashed a reportedly subliminal “positively valenced picture.” If the participant chose the non-target, then a “highly arousing negatively valenced picture” was flashed three times.

A hit was defined as choosing the “target-to-be.” However, as in Experiment 1, the description of the situation is difficult to unravel: For the first one hundred sessions (the first one hundred participants), “the flashed positive and negative pictures were independently selected and sequenced randomly.” Then there was a change in procedure. For the next fifty participants, “the negative pictures were put into a fixed sequence, ranging from those that had been successfully avoided most frequently during the first 100 sessions to those that had been avoided least frequently.” When the participant selected what was to later be designated as the target picture, the positive picture was flashed, subliminally as before, and the negative picture was retained for the next trial. However, when the participant selected the non-target, “the negative picture was flashed and the next positive and negative pictures in the queue were used for the next trial.”

This presents the same problem as before—the procedure has been changed partway through the experiment. Bem states that this was done to evaluate the possibility that “the psi effect may be stronger if the most successfully avoided negative stimuli were used repeatedly until they were eventually invoked.” It is difficult to get one’s head around this justification, and in any case, this should have been examined in a separate study. Again, given the inherent unreasonableness of changing the procedure in an ongoing experiment, one cannot help but wonder if two separate experiments were run and then combined after neither produced significant results on its own.

As in the first experiment, simple t-tests were used to compare participants’ hit rates against the chance hit rate of 50 percent, and a nonparametric binomial test was used to assess the proportion of hits across all sessions. (A third statistic was also calculated; it is said to correct for unequal frequencies of left/right target positions within each session.) In this instance, we are not told how many other t-tests were carried out; if there were other tests, as is likely, this again would have required a correction for multiple comparisons. Of course, because all the data were pooled, we have no information about how many participants actually scored at a level significantly above chance. It seems odd that this information was not of interest.

Bem reports a significant correlation between the score on his two-item stimulus-seeking test and psi performance, but once again the effect was non-significant for “low stimulus seekers.” (Could it be that Bem has serendipitously invented a two-item scale that predicts psi ability?)

EXPERIMENT 3: Retroactive Priming I

This experiment involved a “priming” paradigm borrowed from contemporary psychological research: participants indicate as quickly as possible whether a picture is pleasant or unpleasant, and their response time is measured. Just prior to the presentation of the picture, a positive or negative word (a “prime”) is presented briefly (“subliminally”) on the screen. This prime has been shown to have an effect in that participants usually respond more quickly when a positive picture is preceded by a positive word, or a negative picture is preceded by a negative word, than when picture and word are incongruent. Bem refers to this as a contrast effect.

Bem has taken this procedure and changed it so that the prime is presented after the participant has responded. He reports a significant contrast effect. His data analyses are very complex, involving two transformations as well as outlier9 cutoff criteria; without access to the actual data, it is difficult to evaluate the adequacy of the analysis. However, it is obvious once again that multiple comparisons were carried out without any control for multiple testing.

EXPERIMENT 4: Retroactive Priming II

Experiment 4 is described as a replication of Experiment 3 “with one major change and two timing changes.” Similar positive results were reported. Again, one would need access to all the data, including the discarded outliers, before one could properly evaluate the stated conclusions.

EXPERIMENT 5: Retroactive Habituation I

After the presentation of the previous four experiments, we are now informed that Experiment 5 was a pilot for the basic procedures used in the other experiments in this article. Why it is presented as the fifth experiment is not explained.

The experiment employed a mere-exposure protocol10 borrowed from experimental psychology, but Bem “runs it backwards.” The participant is presented with two pictures side by side. One is the “habituation target” and the other is “closely matched” to the habituation target. The participant is then instructed to indicate which picture he or she likes better. Only then is the participant repeatedly exposed, subliminally, to a picture of the “habituation target.” If it turns out that the habituation target is the one that was earlier chosen, this is considered a “hit;” it is assumed that the effect of the repeated subliminal exposure to the target after the participant had made a choice operated backward in time to influence that original choice.

The habituation target was chosen 53.1 percent of the time, which is reported to be significant at the .014 level one-tailed. However, once again multiple t-tests (six) are reported, which means that the actual p values need to be adjusted. (Suppose that Bem had begun with .014 as the criterion value; then the actual Type I error would be 12(12.014)6=.08, which is not significant).

Incidentally, Bem reports that the hit rate was significantly above chance for women but not for men. Nonetheless, he also states that there was not a significant sex difference! Though this seeming contradiction can arise statistically, it is up to the researcher to make sense of it—which Bem does not.

EXPERIMENT 6: Retroactive Habituation II

Experiment 6 is described as a replication and extension of Experiment 5. Trials with erotic picture pairs were added, and it was hypothesized that the outcome for erotic pictures would be the opposite of that for the negative pictures and that the participants would prefer the habituation picture in fewer than 50 percent of the trials. Bem does not explain his reasoning. There was also another change: on the basis that men may have simply been less aroused than women by the erotic pictures in Experiment 5, thus leading them to not produce a significant effect, it was decided to use stronger and more explicit negative and erotic images obtained from Internet sites for male subjects. Men were also given the choice of male-male erotic images and women the choice of female-female erotic images. (The reader will recall that this was also done in Experiment 1, which was run after Experiment 5.) Such matters should be investigated in further pilot studies rather than incorporated into what is billed as a replication experiment.

Bem also tells us that he had not yet introduced by this point his two-item stimulus-seeking scale into his series of experiments (remember, Experiments 5 and 6 were at the beginning of this series of nine). Instead, he constructed another ad hoc scale by converting two items from Zuckerman’s (1974) well-known Sensation-Seeking Scale into true-false statements: “I enjoy watching many erotic scenes in movies” and “I prefer to date people who are physically exciting rather than people who share my values.” He gives no reason for choosing only these statements, but he does not hesitate to treat them as a reliable and valid measure. While showing no concern for the psychometric properties of these two statements, he then arbitrarily defines only those who endorse both statements as “erotic stimulus seekers.” Thus, an individual who enjoys “many erotic scenes in movies” but prefers to date people who share his/her values was not considered to be an erotic stimulus seeker. This is purely an ad hoc and unacceptable procedure, again suggesting a cavalier attitude about the rigors of proper experimentation.

As for the data analysis, once again there were numerous t-tests without any control for multiple testing, thereby rendering erroneous the claimed significance levels.

EXPERIMENT 7: Retroactive Induction of Boredom

The hit rate was not reported to be significant in this experiment. The reader is therefore spared my deliberations.

EXPERIMENT 8: Retroactive Facilitation of Recall I

This experiment was an attempt to test the hypothesis that the future rehearsal of a set of words can make them easier to recall in the present. (Students would be delighted if this effect could be verified and harnessed, for they could then do further study following a difficult exam and presumably improve their performance on the examination already taken). The design was simple. Participants were shown a set of words and then were tested for their recall of the words. Subsequently, they were given practice exercises with a randomly selected subset of those words, and the hypothesis was that as a result of this subsequent practice, their performance on the test (in the past, remember) would be enhanced and they would have (in the past) recalled more of the words that were practiced in the present.

The participants were one hundred undergraduates. Again, they first responded to the two stimulus-seeking statements. Next, forty-eight common nouns were presented serially for three seconds each. The participant was then asked to type out all the words he or she could recall. The computer then selected twenty-four words at random, and the participant was now instructed to type each of the selected words. It was hypothesized that these practiced words would turn out to be the ones that had been better recalled (before the practice).

Each recalled word was deemed to be a trial and was scored as either a practice or a control word. The actual difference—recall of practice words minus recall of control words—was not analyzed; only a weighted score was given, which was that difference multiplied by the participant’s overall score (on both practice and control words). We are told that this was done to give more weight to the scores of those participants who recalled more words. No appropriate justification is given for this awkward analysis; an analogy is drawn with the practice of weighting studies by their sample size in a meta-analysis, but this is a spurious analogy. The apparently arbitrary weighting of scores, when the more direct-difference scores would offer less ambiguity, renders these findings extremely difficult to interpret. Making choices about data analysis after the data are collected introduces unacceptable opportunity for bias and allows selection of a method that suits one’s hypothesis.

Making matters more complicated, Bem then informs us that another twenty-five “control” sessions were run, similar to the sessions outlined above but without any practice sessions. These control sessions were interspersed among the experimental sessions. The overall recall of words in his control sessions was no different than that in the experimental sessions, and so he concluded that “the enhanced recall of practice words came at the expense of diminished recall of control words.”

Again, it was found that participants who scored low in terms of his stimulus-seeking scale scored at the chance level in the recall test, while those high in stimulus seeking scored above chance.

EXPERIMENT 9: Retroactive Facilitation of Recall II

This is described as a replication of Experiment 8, with one procedural change: a new practice exercise was introduced “immediately following the recall test in an attempt to further enhance the recall of the practice words.” Again, weighted scores were calculated, and on this basis a significant result was obtained. However, on this “replication,” the stimulus-seeking questions did not correlate with psi success. My concerns about the data analysis in Experiment 8 similarly apply in this case.

Overall, then, this is a very unsatisfactory set of experiments that does not provide us with reason to believe that Bem has demonstrated the operation of psi. All that he has produced are claims of some significant departures from chance, and these claims are flimsy given the many methodological and analytical problems that I have touched on in this review. Moreover, Ray Hyman has noted (in my personal communication with him) that the correlation of effect size (as well as significance level) with sample size is highly significant across this set of Bem’s experiments, but it is in the wrong direction! “Effect size,” simply put, refers in this case to the magnitude of the difference between the observed scoring rate and the chance rate. Larger samples provide a better opportunity to detect such a difference if it is truly there, and thus effect size should increase with increased sample size. However, in Bem’s experiments, the effect size correlates negatively (−.91) with sample size, indicating that the claimed effect is smaller when the sample size is larger.

Statistical power is a related concept that refers to the ability to detect an effect when it is actually there. Hyman notes that while power (he uses the log of significance probability as a proxy for power) should be positively correlated with sample size (technically with the square root of sample size), in this series of studies the correlation is approximately .80—in the wrong direction once again. This raises a bright red flag and further erodes confidence with regard to the conduct of this research.

* * *

Having presented his nine experiments, Bem then discusses a number of general issues in parapsychology research and then turns to quantum mechanics! Even if one were to take his interpretation of his results at face value, the claimed results are small and hardly justify an incursion into quantum mechanical theory in the pursuit of accommodation of psi phenomena within modern scientific theory.

While it may seem puzzling that this distinguished psychologist has produced such flawed research, anyone who has read his “Writing the Empirical Journal Article” (published on his website at http://dbem.ws/WritingArticle.pdf) would not be surprised. There he provides advice to students regarding the conduct of research. A few revealing selections (all emphasis added):

Once upon a time, psychologists observed behaviour directly, often for sustained periods of time. No longer. Now, the higher the investigator goes up the tenure ladder, the more remote he or she typically becomes from the grounding observations of our science. If you are already a successful research psychologist, then you probably haven’t seen a participant for some time. Your graduate assistant assigns the running of a study to a bright young undergraduate who writes the computer program that collects the data automatically. And like the modern dentist, the modern psychologist rarely sees the data until they have been cleaned by human or computer hygienists.

To compensate for this remoteness from our participants, let us at least become intimately familiar with the record of their behaviour: the data. Examine them from every angle. Analyze the sexes separately. Make up new composite indexes. If a datum suggests a new hypothesis, try to find additional evidence for it elsewhere in the data. If you see dim traces of interesting patterns, try to reorganize the data to bring them into bolder relief. If there are participants you don’t like, or trials, observers, or interviewers who gave you anomalous results, drop them (temporarily). Go on a fishing expedition for something—anything—interesting...

When you are through exploring, you may conclude that the data are not strong enough to justify your insights formally, but at least you are now ready to design the ‘right’ study. . . . Alternatively, the data may be strong enough to justify re-centering your article around the new findings and subordinating or even ignoring your original hypotheses...

Your overriding purpose is to tell the world what you have learned from your study. If your research results suggest a compelling framework for their presentation, adopt it and make the most instructive findings your centerpiece. Think of your data set as a jewel. Your task is to cut and polish it, to select the facets to highlight, and to craft the best setting for it. Many experienced authors write the results section first.

But before writing anything, Analyze Your Data!

Reflections of this advice appear to be writ large throughout Bem’s research article.

* * *

The publication of this set of experiments will serve no one well. Parapsychology is not honored by having this paper accepted by a mainstream psychology journal. Neither does Bem’s paper serve the public well, for it only adds to confusion about the scientific case for the existence of psi. And it does no service to the reputation of the Journal of Personality and Social Psychology. Although Bem has failed to demonstrate the existence of mysterious intellectual powers that defy the normal constraints of time and space, there seem nonetheless to have been mysterious intellectual powers at play here. I refer to the decision by the editors of an esteemed psychology journal to publish this badly flawed research article.

“Think of your data set as a jewel,” Bem instructs. However, with these nine experiments, Bem did not end up with a polished jewel. Rather, to extend his metaphor, the jewel cracked under the intense pressure used to try to shape it to fit expectation. One is left with nothing but useless fragments that reflect not the light of knowledge but the biases of the researcher.

Rhine, Schmidt, Targ, Puthoff . . . the list grows on. Plus ça change, plus c’est la même chose.

Acknowledgements

Thanks to Ray Hyman, Scott O. Lilienfeld, Timothy Moore, and Benjamin Wolozin for their very sage comments on an earlier draft of this article.

Notes

1. My discussion is based on the pre-publication version of Professor Bem’s article that appears on his website at http://www.dbem.ws/FeelingFuture.pdf.

2. The p value is the likelihood of having concluded that there is a significant effect when in fact there is not. The lower the p value, the less likely it is that the null hypothesis (that there is no effect) is true.

3. Meta-analysis is a statistical process for testing the combined results of a number of studies that were based on similar research hypotheses.

4. The t-test is a statistical test typically used either to compare two means or to compare a mean with a theoretical expectation—for example, to assess the difference between an observed average success rate and a hypothetical chance rate of 50 percent.

5. A Type I error occurs when the null hypothesis (that there is no effect) is rejected when it is in fact true.

6. In a two-tailed test, one assesses the data to see whether they significantly differ from what would be expected by chance in either direction, that is, whether they are greater than or less than what would be expected by chance alone. When we say that something is significant at the .01 level two-tailed, this means that we would expect these results to occur by chance alone only 1 percent of the time. But, given that either above-chance or below-chance results are considered to be meaningful, this 1 percent must be distributed in both directions. Thus, above-chance results would be significant at the .01 level (two-tailed) only if they are so extreme that they would be expected to occur by chance half of 1 percent, or 0.5 percent, of the time or less. The same would apply for below-chance results.

With a one-tailed test, one also assesses the data to see whether they significantly differ from what would be expected by chance but in only one direction, that is, whether they are either greater than expected by chance or less than expected by chance, but not both. A one-tailed test is properly used if one has good reason to predict the direction of the data in advance. Again, using the example of the .01 level, for a one-tailed test the data only need be extreme enough that they would be expected by chance alone 1 percent of the time or less (compared to 0.5 percent with a two-tailed test). This makes it much easier to claim statistical significance.

Parapsychologists normally employ two-tailed tests because results that are either significantly above chance or significantly below chance are taken to reflect psi. Although Bem indicates that he predicted that the erotic pictures would lead to above-chance scoring, which could justify using a one-tailed test, what would he have done had the participants scored at a below-chance rate that would have been significant had he predicted that the results would indeed be below chance? Apparently committed to a one-tailed test and having made only the above-chance prediction, he properly would have had to ignore those data—something that parapsychologists do not want to do. By using two-tailed tests, parapsychologists avoid the problem and also avoid any suspicion of having changed the direction of their prediction after having examined the data.

7. The Bonferroni t-test is a modified t-test that adjusts for the number of tests being carried out so that the overall likelihood that one of them produces significance by chance alone is kept at a specified level, such as 5 percent.

8. A nonparametric binomial test deals with data divided into two categories and examines the statistical significance of deviations from a theoretically expected distribution. It is referred to as “nonparametric” because it does not rely on the parameters of a distribution, such as the mean.

9. An outlier is a datum that is numerically distant from all the other data in the sample, either as a result of measurement error or because the data are not distributed in the manner that was assumed.

10. Mere-exposure protocol is a research approach in which participants’ responses are assessed with the assumption that having simply been exposed (perhaps subliminally) to a stimulus object will cause an effect.

References

Alcock, J.E. 1988. A comprehensive review of major empirical studies in parapsychology involving random event generators and remote viewing. In Commission on Behavioral and Social Sciences and Education, Enhancing Human Performance: Issues, Theories and Techniques, Background Papers. Washington, D.C.: National Academy Press, 601–719. Available online at http://books.nap.edu/openbook.php?record_id=778&page=601.

Bem, D.J., and C. Honorton. 1994. Does psi exist? Replicable evidence for an anomalous process of information transfer. Psychological Bulletin 115: 4–18.

Bem, D.J., J. Palmer, and R. Broughton. 2001. Updating the ganzfeld database: A victim of its own success. Journal of Parapsychology 65: 1–6.

Gardner, M. 1977. ESP at random. New York Review of Books, July 14.

Hyman, R. 1985. The ganzfeld psi experiments: A critical appraisal. Journal of Parapsychology 49: 3–49.

———. 2010. Meta-analysis that conceals more than it reveals: Comment on Storm et al. [2010a]. Psychological Bulletin 136(4): 486–90.

Hyman, R., and C. Honorton. 1986. A joint communiqué: The psi ganzfeld controversy. Journal of Parapsychology 50: 351–64.

Jahn, R., B. Dunne, G. Bradish, Y. Dobyns, A. Lettieri, R. Nelson, J. Mischo, E. Boller, H. Bosch, D. Vaitl, J. Houtkooper, and B. Walter. 2000. Mind/machine interaction consortium: PortREG replication experiments. Journal of Scientific Exploration 14: 499–555.

Jeffers, S. 2003. Physics and claims for anomalous effects related to consciousness. In J.E. Alcock, J. Burns, and A. Freeman (Eds.), Psi Wars: Getting to Grips with the Paranormal. Exeter, UK: Imprint Academic, 135–52.

Marks, D., and R. Kamman. 1978. Information transmission in remote viewing experiments. Nature 274: 680–81.

———. 1980. The Psychology of the Psychic. Buffalo, NY: Prometheus Books.

Milton, J., and R. Wiseman. 1999. Does psi exist? Lack of replication of an anomalous process of information transfer. Psychological Bulletin 125: 387–91.

PEAR (Princeton Engineering Anomalies Research). n.d. Experimental research. Available online at http://www.princeton.edu/~pear/experiments.html.

Rhine, J.B., and W. McDougal. 1934/2003. Extra-Sensory Perception. Whitefish, MT: Kessinger Publishing.

Storm, L., P.E. Tressoldi, and L. Di Risio. 2010a. Meta-analysis of free-response studies, 1992–2008: Assessing the noise reduction model in parapsychology. Psychological Bulletin 136: 471–85.

———. 2010b. A meta-analysis with nothing to hide: Reply to Hyman. Psychological Bulletin 136: 491–94.

Targ, R., and H. Puthoff. 1974. Information transmission under conditions of sensory shielding. Nature 251: 602–4.

Zuckerman, M. 1974. The sensation seeking motive. In B.A. Maher (Ed.), Progress in Experimental Personality Research (Vol. 7). New York, NY: Academic Press, 79–148.

Note: This post is a response to Daryl Bem's comments, which can be found here. Alcock's original article may be viewed here.

Outrage and ad hominem condescension. Bem’s response brings to mind an old adage from the legal world: “If the facts are against you, argue the law; if the law is against you, argue the facts; if the facts and the law are against you, yell like hell.” And yell like hell he does. Rather than deal with much of the substance of my critique, he directs his attack at my abilities. However, the issue is not about my intelligence or his or my character or his. It is about the data and the way in which they were gathered and analyzed.

Not only does he shoot the messenger; much of his defense rests on an appeal to authority – in this case, the authority of the reviewers at the Journal of Personality and Social Psychology who recommended that his article be published. While I do not understand why they decided as they did, their decision cannot make a silk purse out of a sow’s ear; it does not make the serious flaws in this research go away. Bem also employs the tired defence that it is not enough for a critic to find flaws in a study; the critic must also show that the obvious flaws could themselves account for the observed results. This is of course simply nonsense. The burden of proof is on the individual who presents the data, and significant flaws in the research militate against confidence that the researcher did not make other undetected and unreported errors as well. This is all the more of concern when one is claiming evidence for phenomena that contradict well-established knowledge in physics, neurology and psychology.

In response to his various points:

1. He accuses me of an imaginative rewriting of parapsychological history. However, in essence, I simply pointed out that none of the much-touted supposed parapsychological “breakthroughs” of the past have succeeded in establishing the reality of psychic phenomena so far as the larger scientific community is concerned. If he believes otherwise, then it is his imagination and not mine that is at play.

2. Bem chooses not to address most of my criticisms about his procedure, but instead misleadingly states that my major criticism is the “selection and deployment of the pictorial stimuli used in six of the nine experiments.” My major methodological criticism actually has to do with the chaotic, careless nature of his procedures; his “selection and deployment of stimuli” only reflects the more general problem, and is not the key problem itself. Whether gay participants were provided with homosexual pictures is not in and of itself at issue. However, his response actually adds further obfuscation. Let me explain: Although the presumed impact of erotic stimuli is central to several of his experiments, he made no effort to assess whether any of the erotic stimuli are actually erotic. Instead, begging the question in the true philosophical sense of the term, he simply assumed that because females “psychically” scored above chance with the erotic pictures, while males did not, this gender difference must have come about because the males did not find the pictures to be erotic enough. He states, in regard to my critique of Experiment 1:

“Male and female raters differed markedly in their ratings of negative and erotic pictures. Male raters rated every one of the negative pictures as less negative and less arousing than did the female raters, and they gave more positive ratings than the female raters to the most explicit erotic pictures. Possibly reflecting this sex difference, female participants showed significant psi effects with negative and erotic stimuli in my earliest experiment but male participants did not. Accordingly, I decided to introduce different sets of pictures for men and women in subsequent experiments, choosing more extreme and more arousing pictures for the men.”

This would seem to suggest that he directly assessed the erotic nature of the stimuli, that his participants actually rated the pictures. They did not. His conclusion about gender differences in erotic response is based on the ratings provided with the IAPS set of pictures that he used, ratings that were available to him before he began his experiments. He apparently had no concern about those gender differences until after he had collected his data, at which time, having found that males did not show what he refers to as “significant psi effects” (actually, just significant deviations from the chance response rate), he decided to modify his stimuli set for males for Experiment 6, and in other subsequent experiments including the strangely-numbered Experiment 1. However, in his original article, there is nothing in his description of Experiment 1 that discusses male and female reactions to the stimuli, apart from a reference to Experiment 5, (which was run before Experiment 1!). It is in Experiment 6 where he makes clear where the “ratings” came from. He states (in his original article):

“... we decided to use sets of negative and erotic pictures that were different for men and women. As noted above, women showed a significant psi effect on the negative trials in Experiment 5, but men did not. ... The ratings supplied with the IAPS pictures reveal that male raters rated every one of the negative pictures in the set as less negative and less arousing than did female raters. … So, for this replication, we supplemented the IAPS pictures for men with stronger and more explicit negative and erotic images obtained from Internet sites.” (My underlining)

So, after relying on a standardized set of pictures, he now chose who-knows-what pictures from the internet to “supplement” the IAPS pictures (How many internet pictures were involved, he does not say). Again, let me make clear that it is not the picture set per se that is my concern; it is the arbitrary way in which pictures were chosen and switched about, which reflects the systemic experimental carelessness. .

3. Bem points out that correctly that I was confused with regard to the procedure in Experiment 1. I was initially mistaken in interpreting the chance rate for erotic pictures as being 33% for some of the participants, although I did indeed figure out that his primary analysis was with regard the hit rate on the erotic targets for all 100 participants, with a chance rate of 50%. Upon review, I can assure the reader that none of this has any bearing on my overall conclusions.

Compare the relative clarity of his present description to the description in the original article: (Keep in mind that by “session,” he is referring to the trials conducted with a single participant).

His present response to me:

“There were 100 sessions in this experiment, and on each of 36 trials, participants saw images of two curtains side-by-side on the computer screen. They were told that a picture would be behind one of the curtains and a blank wall would be behind the other. Their task on each trial was to click on the curtain they felt concealed the picture. After they made their selection, the selected curtain opened, revealing either a picture or a blank wall. ... On randomly selected trials, the picture was erotic; on other trials, it was a nonerotic, and the participant had no (non-psi) way of knowing which kind of picture would be used on any given trial. Because there were two alternatives on each trial—left curtain or right curtain—the probability that the participant would correctly select the location of the picture by chance was always 50%.”

His original article:

“For this purpose, 40 of the sessions comprised 12 trials using erotic pictures, 12 trials using negative pictures, and 12 trials using neutral pictures. The sequencing of the pictures and their left/right positions were randomly determined by the programming language’s internal random function. The remaining 60 sessions comprised 18 trials using erotic pictures and 18 trials using nonerotic positive pictures with both high and low arousal ratings. These included eight pictures featuring couples in romantic but nonerotic situations (e.g., a romantic kiss, a bride and groom at their wedding). The sequencing of the pictures on these trials was randomly determined by a randomizing algorithm … Although it is always desirable to have as many trials as possible in an experiment, there are practical constraints limiting the number of critical trials that can be included in this and several others experiments reported in this article. In particular, on all the experiments using highly arousing erotic or negative stimuli a relatively large number of nonarousing trials must be included to permit the participant’s arousal level to “settle down” between critical trials. This requires including many trials that do not contribute directly to the effect being tested...”

However, thinking again about his procedure, another problem becomes apparent. There was a relatively large data set for the erotic pictures, since all participants were exposed to them, but the data set for each of the various categories of non-erotic images was considerably smaller, which makes it less likely that one would detect small but significant effects in those data even were they to exist. Thus, the procedure is weighted such that it is less likely that significant above-chance rates, if they exist, would be detected for the non-erotic stimuli because smaller sample sizes provide less power in the statistical tests. So, while he makes much of having found significant “psi” effects only with the erotic stimuli (notwithstanding the statistical problems associated with the multiple testing), this difference might well be due only to differences in the power of the tests, reflecting the differences in sample size.

4. Bem deals with my critique of his statistical analysis in a purely ad hominem manner. He completely fails to respond to one of my most serious criticisms, about his use of one-tailed tests. However, he sneers at my concerns about the multiple t-tests that he relies upon, and argues that:

“…multiple t tests demonstrated that participants did no better than chance on any of the subcategories of nonerotic pictures. It is here that Alcock first complains about my performing multiple tests without adjusting the significance level for the number of tests performed. In this case, Alcock is almost right. Suppose that in testing each of the four subcategories of nonerotic pictures, I had found that one of them (e.g., romantic pictures) showed a significant precognitive effect. Because this finding would have emerged post hoc, only after I had first performed separate tests on four different picture types, I would have had to adjust the significance level to be less significant. If I did not, I would be illegitimately capitalizing on the likelihood that at least one of the four tests would have yielded a positive result just by chance. But there was no psi effect on any of the subcategories of nonerotic pictures.”

Bem thus justifies the decision not to control for multiple testing on the basis of an examination of the data themselves! Because he found no psi effect for the non-erotic categories, he argues that there was no problem with conducting the several tests without controlling for multiple testing!! One simply cannot make one’s choice of statistical procedure after looking at the data. One must adjust the statistical criterion in advance to reflect the number of tests to be done, taking into account not only the t-tests for the non-erotic images, but the test for the erotic images as well. Since he obviously did not do so, he consequently employed the wrong criterion value, and that, combined with the use of one-tailed testing, greatly magnifies the likelihood of finding statistical significance when none exists.

5. With regard to his comments about the Priming studies, the reader should note that, contrary to Bem’s complaint, I did not object to the use of complex data analyses involving two transformations and outlier cut-off criteria. I simply said that because of the complexity of the analyses, one cannot fairly judge his interpretations of the data without seeing the data and the analyses that were done.

It is interesting that Bem states that:

“At least one expert in priming experiments has also argued that one should always perform several analyses using different transformations and different cut-off criteria to ensure that the priming effects hold up across these variations. That is precisely what I did.”

That may be just fine, but the cautious observer will reserve judgment in the absence of detailed information about those various analyses, and the transformations and cut-off criteria upon which they relied.

For all his sound and fury, this seriously flawed set of experiments is still a seriously flawed set of experiments. If Bem wants the scientific world to pay attention to his claims of psi, he must first produce meaningful data from a well-designed, well-executed and well-analyzed experiment. Neither excuses for careless research, nor angry defences of it, will achieve this; he must simply do it right.

What if, instead the dead could speak to us directly, without the middle person?

If You Survive Death?

Imagine for a moment that you are the dead person, that your body has died, but your mind / personality / soul lives on. You are surprised by this, and you want to tell people, especially your skeptical friends, all about it — you want to communicate with us.

What would you do?

You have no voice box therefore you cannot speak. You have no arms or legs or any means of moving objects. But you are — as they say — an “energy field.” Could you reach us by interference with devices that rely upon other energy fields, a radio or tape recorder, for example?

But if you were able to generate some sounds on a tape recorder, would any one even detect them, or pay attention if they did? It’s often hard to detect weak signals — and you are but a wraith, a spirit, and probably without a lot of energy.

However, there is hope for humans, as Ray Hyman points out, because humans are the best pattern detectors in existence. Pattern detection, in this example, would be the ability to discriminate signal from noise.

Voices of the Dead?

This is exactly what is happening, according to some people. If you listen carefully, they say, you can hear the voices of the dead in tape recordings.

What do the voices of the dead sound like? Here are two examples of actual recordings where people claim to hear spoken words, the words of the dead.

This from the webpage http://members.tripod.com/~GSOLTESZ/evp.htm#listen:

“In this EVP recording, you can hear a voice saying, almost in agony, ...“Save Me.”

You might have to play it a couple of times but you can’t miss it. This recording has only been enhanced by myself using a sound editing program. The reason was to cut down on the “noise” and “bring out” the actual voice.

• "Save Me” (WAV)

Here are two other examples from Dr. Michael Daniels, psychologist and parapsychologist. (www.mdani.demon.co.uk)

• Example 1 (WAV)
• Example 2 (WAV)

The website instructs:

“To hear the voices at their best you should play them at maximum volume through headphones. In both cases you should be able to hear a definite “English” male voice. You may need to replay the recordings several times in order to make out the words, which are quite indistinct. The first clip seems to be saying something like “do you like potatoes?”. The second clip sounds to me rather like “five thirty and four-eye”. Different words may suggest themselves to you.” [Dr. Daniels points out that there is divided opinion about the reality of EVP.]

Electronic Voice Phenomena

So — it’s not so easy to hear the voices, is it? These are examples of what are called electronic voice phenomena, or EVP.

We are informed by another website that:

“EVP is a process whereby unexplained snatches of  voice or voices are embedded onto magnetic recording tape by a process that is not yet fully understood. The embedded “ghost” voice can be heard when the magnetic audio tape is played back on a standard tape recorder/player.” [hauntedhike.com]

Again, the Web informs us that:

“Recordings typically last only for a few minutes. This is because intense concentration is required in order to hear the voices on the tape, which usually has to be replayed several times in order to decipher the speech. Use of headphones is recommended.” [www.mdani.demon.co.uk]

The best way to understand the development of EVP it is to go back a little in time.

With the rise of Spiritualism beginning with the “mysterious rappings” of the Fox sisters in the nineteenth century, there have been many attempts to “contact the dead,” while claiming to be engaged in scientific study.

Thomas Edison saw new technology, part of which he invented, as a means by which spirits might try to contact us. Apparently, he strove to make contact through some sort of phonograph device in the 1890s. Then, in the late 1920s, he tried to make contact with the souls of the dearly departed by means of some sort of special chemical equipment. It is claimed that spirit voices were first captured on phonograph records in 1938, seven years after his death.

However, it was with Friedrich Jürgenson (1903-1987) that the study of EVP really begins. Jürgenson was in some ways a Renaissance Man — an archeologist, a philosopher, a linguist, a painter who was commissioned by Pope Pius XII, a singer and recording artist, and a film documentary maker. . Jürgenson’s interest in Electronic Voice Phenomena apparently began when, after having recording bird songs with a tape recorder, he could hear human voices on the tapes, even though there had been no one in the vicinity.

This surprising event naturally piqued his interest, and he turned his attention to making recordings of nothing — that is, recordings made in a quite place with no one around. He continued to detect voices on these tapes, and his studies led to the 1964 publication of his book Rosterna fran Rymden (“Voices from space”).

He subsequently recognized some of the voices that his tape recorder picked up, including that of his mother, who called him by her pet nickname for him. However, as we say where I grew up, his mother was already “on the wrong side of the grass;” that is, she was deceased. It seemed natural to him to assume that she was communicating from beyond the grave. Thus, he came to the conclusion that all the voices that he had recorded were voices of the dead. In 1967, he published Sprechfunk mit Verstorbenen (“Radio-link with the dead”).

Dr Konstantin Raudive (1906-1974), a student of Carl Jung, was a Latvian psychologist who taught at the University of Uppsala in Sweden. He was preoccupied with parapsychological interests all his life, and especially with the possibility of life after death, and he kept in close contact with leading British psychical researchers

In 1964, Raudive read Jürgenson’s book, Voices from space, and was so impressed by it that he arranged to meet Jürgenson in 1965. He then worked with Jürgenson to make some EVP recordings, but their first efforts bore little fruit, although they believed that they could hear very weak, muddled, voices.

However, one night, as he listened to one recording, he clearly heard a number of voices- and when he played the tape over and over, he came to understand all of them — some of which were in German, some in Latvian, some in French. The last voice on the tape — a woman’s voice — said “Va dormir, Margarete” ("Go to sleep, Margaret”).

Raudive later wrote (in his book Breakthrough): “These words made a deep impression on me, as Margarete Petrautzki had died recently, and her illness and death had greatly affected me.” Amazed by this, he then started researching such voices on his own, and spent much of the last ten years of his life exploring electronic voice phenomena. With the help of various electronics experts, he recorded over 100,000 audiotapes, most of which were made under what he described as “strict laboratory conditions.” He collaborated at times with Hans Bender, a well-known German parapsychologist. Over 400 people were involved in his research, and all apparently heard the voices. This culminated in the 1971 publication of his book Breakthrough, mentioned above. His impact was such that these phenomena are now often referred to simply as “Raudive voices.”

Raudive developed several different approaches to recording EVP, and he referred to:

1. Microphone voices: one simply leaves the tape recorder running, with no one talking; he indicated that one can even disconnect the microphone.
2. Radio voices: one records the white noise from a radio that is not tuned to any station.
3. Diode voices: one records from what is essentially a crystal set not tuned to a station.

Raudive delineated a number of characteristics of the voices, (as laid out in Breakthrough):

1. “The voice entities speak very rapidly, in a mixture of languages, sometimes as many as five or six in one sentence.”
2. “They speak in a definite rhythm, which seems forced on them.”
3. “The rhythmic mode imposes a shortened, telegram-style phrase or sentence.”
4. Probably because of this, “grammatical rules are frequently abandoned and neologisms abound.”

Of course, to the skeptic, these characteristics are what one might expect if indeed the “voices” are simply misinterpretations of random, “white” noise.

EVP Today

Serious parapsychologists today show virtually no interest in EVP, and modern reports in the parapsychological literature find no evidence of anything paranormal in such recordings. That does not deter the devoted, of course: it is claimed that there are more than 50,000 sites on the internet devoted to EVP! Again, an example from the Internet:

“Briefly, electronic voice phenomenon (EVP) is the process of capturing messages from the spirit world, including our loved ones in Heaven, by using an ordinary tape recorder. Yes, someone in your family or your special friend who has passed on, can record or imprint their voice onto your tape. It is not the scope of our [web] site to fully explain EVP, but please feel free to visit the learning links below for more information. Our site is designed to help you, the beginner, succeed with EVP” [www.paranormalnetwork.com]

And now it is claimed that one does not even need to be quiet while making the recordings — the voices often show up in the background while one is recording a conversation. Consider these examples (from www.paranormalnetwork.com).

Examples:

Voice of the dead?

Here, we are told to listen for a whispery woman’s voice saying “we all turn this way” or “we all turn that way” recorded by “Karen and Jill” at Edgar Allen Poe’s grave on his birthday. (The salient part of the recording is repeated five times so that you can catch it.) [Listen]

Voice of the dead?

In the middle of the recording, we are told that a voice whispers “Pat!” [Listen]

There is no end to the efforts people will make to find “voices.” For example, it is claimed that:

“Some voices of spirits or entities are very close to the background level of static; Others may be clearly heard. If the speech is difficult to understand, remember that the spirit talking may be talking in a language or dialog that is not in common usage today. The voice can also be in reverse, you would need a computer to reverse this to hear it.” [www.blueskies.org]

As yet another example of the unbridled enthusiasm and creativity associated with finding voices, consider the American Association of Electronic Voice Phenomena. Its website informs us that:

“The membership includes people who record paranormal voices, pictures and information from friends and loved ones on the other side through tape recorders, telephones, fax machines, television, computers and video recorders.”

“EVP has been featured in such technical publications as “Popular Mechanics” and “wireless world.” It was recently shown in a movie called “the sixth sense". Sarah Estep, one of the world’s foremost EVP recorders, has been featured on cable channels such as discovery and Sci-Fi with her numerous EVP recordings. Why EVP remains unknown by the general public continues to astound us. EVP can provide a huge sense of relief for the bereaved and documented proof for paranormal investigators.”

And if one surfs the web, sooner or later, one finds sites that offer to sell devices to help you obtain better recordings!

Possible Explanations

Well, if the voices aren’t spirits, what are they?

1. Cross-modulation. This is a common phenomenon; I first became aware of it in the 1960s when my “record player” clearly picked up a local radio station, which one could hear between cuts.

   But Raudive dismissed this possibility, saying that it cannot be radio since one never hears music or other obvious elements of radio transmission.

2. Apophenia. This refers to a common perceptual phenomenon whereby we spontaneously perceive connections and find meaningfulness in unrelated things. In other words, it involves seeing or hearing patterns where in reality, none exist. A visual example is the Rorschach Inkblot test.

   We may be the best pattern detectors that exist, but not all the patterns we find have any objective meaning. However, once we think we have detected a pattern, it is hard to ignore it, and generally, we take it to be meaningful. A common example of apophenia occurs when people are in the shower, and mistakenly think that they hear their door bell or telephone ringing. The white noise produced by the shower contains a broad spectrum of sounds, including those that make up ringing bells. The ear picks out certain sounds from the spectrum, and we “detect” a pattern corresponding roughly to a bell.

   (Apophenia is virtually synonymous with what has been called Pareidolia, an illusion involving misperception of an external stimulus; an obscure stimulus is viewed as something clear and distinct. Examples include instances such as when thousands of people in New Mexico saw the face of Jesus on a Tortilla chip in 1978. This perception, or misperception, does not involve conscious effort or any particular mental set, and the illusion does not vanish even when one pays closer attention to the stimulus because it is so ambiguous that it has no objective meaning at all.

   (See http://thefolklorist.com/ for many examples)

While you might accept apophenia as an explanation for voices barely discernable from static, as in the earlier examples above, can it account for the “clear voices’ in the later examples (e.g. — recall the word “Pat” from the tape)? First of all, of course, the extraneous voices, if really there, could be the result of intended or unintended background interruptions by real people — the recordings were not made under any sort of controlled conditions. Secondly, as is discussed below, it is fascinating just how easy it is for our brains to come to interpret certain noise patterns as words, once we know what the words are supposed to be.

What is going on?

Perception is a very complex process, and when our brains try to find patterns, they are guided in part by what we expect to hear. If you are trying to hear your friend while conversing in a noisy room, your brain automatically takes snippets of sound and compares them against possible corresponding words, and guided by context, we can often “hear” more clearly than the sound patterns reaching our ears could account for. Indeed, it is relatively easy to demonstrate in a psychology laboratory that people can readily come to hear “clearly” even very muffled voices, so long as they have a printed version in front of them that tells them what words are being spoken. The brain puts together the visual cue and the auditory input, and we actually “hear” what we are informed is being said, even though without that information, we could discern nothing. Going one step further, and we can demonstrate that people can clearly “hear” voices and words not just in the context of muddled voices, but in a pattern of white noise, a pattern in which there are no voices or words at all.

Given that we can routinely demonstrate this effect, it is only parsimonious to suggest that what people hear with EVP is also the product of their own brains, and their expectations, rather than the voices of the dearly departed.

We can describe the process, leading from mental set to expectation to perception to amazement to belief in the following general way (see graphic): We are told that tape recordings made with no one around contain mysterious voices. This sets up a mental set that motivates us to try to discern voices. That is, we must presume that there may be something there, or we would not waste our time in listening. If others have told us what the voices seem to say, this expectancy influences our auditory perception, so that our brains match up bits of random noise to the words that we expect to hear. Of course, if we play the same piece of tape over and over, as is explicitly recommended by some of the web sites cited earlier, and if we do everything we can to focus our attention on the “noise” (perhaps by listening through headphones, again as recommended by the web sites), then we not only increase the likelihood of discerning voices if they really are there, but we maximize the opportunity for the perceptual apparatus in our brain to “construct” voices that do not exist, to detect patterns that match up with our expectations. Then, once we “hear” the voices, then it is easy, given the mental set that is usually involved, to attribute them to deceased individuals. This interpretation is likely to produce an impressive emotional reaction, and since we have now heard what we set out to hear (our expectancy is fulfilled) our belief in the reality of the voices of the dead grows, and this may be rewarding in various ways. Such an outcome is likely to heighten the expectation that we will hear more voices the next time we listen to such tapes.

How to disabuse the believer

How can someone who has heard the voices be persuaded to be more critical and to examine more mundane possibilities?

A rational, deliberative discussion is rarely helpful because clear evidence or logic is not involved. Believers are reporting an experience that was highly meaningful and perhaps highly emotional to them — not something that is easily challenged by logic. Moreover, there is a self-selection of people predisposed to believe — the voices are compatible with their belief system.

Remember — we process information in two different ways through two more or less separate parts of our brain and nervous system. On the one hand, part of our brain works on a very intuitive / emotional / automatic level, and on the other hand, another part of our brain works according to the logic and rationality that we develop over our lifetimes. These two systems often produce contrary results, and this is especially so where paranormal phenomena are involved. The “believer” removes the contradiction by bringing the intellect into line with the intuitive interpretation, that is, by coming to accept the paranormal — in this case, the voices — as reality, and thereby reshaping the intellectual understanding of the world so that belief in such phenomena appears to be rational. Over time, an impregnable belief system develops which is supported by a very substantial base of personal experience (interpreted in such a way as to support the paranormal belief), as well as anecdotal evidence provided by others.

It is very difficult to change such deeply held beliefs, especially if they include a significant emotional component. Consider this example: In my work as a clinical psychologist, a father wanted me to “cure” his gay son. I asked the father how easy would it be for me to turn him (the father) into a gay person? “No way !!!,” he said. I told him that it would likely be as difficult to turn his son into a straight person as it would be to make him, the father, turn gay. Fortunately, he saw the point and came to accept his son as he was. My point is this: When we ask how to turn believers into skeptics, let us ask instead: “How easy would it be for me to persuade you that voices on a tape really are spirits of the dead?” Well, that is probably just how easy it would be to persuade devoted believers that their beliefs about the voices are wrong.

What the Raudive voices teach us is that intelligent people — for Raudive was no doubt an intelligent man - can come to believe fervently in phenomena which in all likelihood do not exist. There is a lesson in this for all of us, for we just as surely may be mistaken in some of our own deeply held convictions. This is why we must rely on science as the avenue to truth rather than personal experience or other people’s anecdotal reports. Science, with its reliance on data and its insistence on looking for sources of error and for alternative explanations, provides the best method that humans have produced for protecting against error and self-delusion. Electronic Voice Phenomena are the products of hope and expectation; the claims wither away under the light of scientific scrutiny.

Imagine, if you will, the following scenario-admittedly an unlikely one:

• A new super-psychic has burst onto the world stage. He has amazing powers that go far beyond mere “cutlery distortion” and telling people things about themselves that they already know.
• This is a psychic whose powers apparently have already been tested in scientific experiments that produced highly statistically significant results.
• This is a psychic who has already personally entertained many of the world’s leaders, and impressed them with his powers.

Well, so far, this is not that different from what we have seen before perhaps, but now, suppose the following:

• This psychic has offered to use his powers to negotiate a lasting peace amongst warring factions in the Middle East, in Afghanistan, in Iraq and in Kashmir. He has told members of the United Nations Security Council, who gave him a private audience, that by being able to read the minds of the leaders of the protagonists in the various conflicts, he will bring about better communication and better understanding of issues. He might even resort to altering the mindsets of recalcitrant leaders by means of psychokinesis, turning them into peace-seekers and compromisers. The members of the Security Council have become so convinced of his powers that they are about to name him Ambassador-At-Large and Chief Negotiator for World Peace.

The Committee for the Scientific Investigation of Claims of the Paranormal is not amused by this news.

CSICOP and its supporters protest loudly, and finally we are informed by the Security Council that we may send an emissary-one person only-who will be allowed to evaluate the supporting research and to test the psychic directly, and then present his or her findings and conclusions directly to that world body.

Well, this is pretty important stuff-much more important than the usual CSICOP work. After all, if, as we suspect, this man is a phony, there is the likelihood of tremendous harm being done to the cause of world peace if he is allowed to mess about in these seemingly intractable conflicts.

Let’s not quibble about the likelihood of such a scenario, but let’s focus instead on what qualities we would want our emissary to possess in such a case. I've made a list:

1. Since scientific evidence has been adduced to support claims of the psychic’s powers, our emissary should be a scientist, preferably a social scientist, someone who knows how to conduct and evaluate research involving human subjects; someone who knows how to detect flaws and biases in such experiments.
2. We need an expert in statistics, since statistical analysis was part of the scientific support offered for the psychic’s powers.
3. We need an expert in the psychology of belief and deception, someone who knows all about how people can both deceive and be deceived.
4. We need someone with sound academic credentials, for credibility is going to be a very important if our emissary is to have an influence on the Security Council.
5. We need someone with expert knowledge of magic and mentalism, for if the psychic is using the magician/mentalist’s craft, only someone experienced and knowledgeable in this craft will be able to detect this. As the saying goes, it takes a thief to catch a thief.
6. We need someone who is experienced in evaluating supposed psychics and the research adduced in their support. Without such experience, even a very good social scientist may overlook important sources of error and bias.
7. We need someone who has a track record for fairness, someone who has gained the respect of skeptics and believers alike, so that our emissary will not be seen as some sort of hit man for CSICOP and skepticism.
8. We need a good communicator. It is not enough just to be able to show that the psychic is not really psychic at all. We need someone who can cogently present the skeptical case in such a way as to have an impact on the members of the Security Council.
9. We need someone who does not antagonize others, someone whose personality and charm will ease the sting of whatever critical commentary he or she has to offer to the Security Council.

Well, that’s quite a list, and one that is almost impossible to fill, one might think.

I know of only one person in the entire world-and believe me, I am not exaggerating here; I really mean it-who measures up to all these criteria. And he is in our midst tonight.

Let me tell you about Ray Hyman:

• Ray earned his Ph.D. at Harvard University, where he then taught statistics, amongst other subjects. He is now Professor Emeritus of Psychology at the University of Oregon, where he taught for many years.
• He is an expert in cognitive psychology.
• He is an expert in statistical analysis.
• He is an expert in research design as applied to the study of human subjects.
• He is an expert in the study of deception and self-deception and has dedicated most of his professional career to the study of why people come to believe strange things, how they can be fooled, and how some people set out to fool them.
• He has published books, book chapters, and over 200 articles that critically evaluate studies of the paranormal and related domains. His article on cold reading, so Paul Kurtz informs me, has generated more requests for reprints than any other article in the history of the Skeptical Inquirer.
• He is an expert in magic and mentalism. Few people outside of magic circles are aware of just how accomplished he is as a magician/mentalist. Indeed, he started out by earning his living as a mentalist. He is also an accomplished inventor of magical effects and routines. He is highly respected by other magicians, and has had the rare honor of twice appearing on the cover of the Linking Ring, the monthly magazine of the International Brotherhood of Magicians.
• He has a long and unparalleled history of investigating psychics. Some of this work has been for official agencies such as the United States Defense Department and the United States National Research Council. He knows personally all of the leading, and many of the not-so-leading, parapsychologists in the world, and is respected by virtually all of them.
• He is, of course, a champion among skeptics, and one of the founders of CSICOP.
• He is known above all for being fair-minded-so much so that on more than one occasion I have been approached by people at CSICOP conferences who have complained that he had gone soft on parapsychology. They believed that he had “gone soft” because, rather than taking a debunking stance as they had wanted, he had approached the topic of paranormal claims from the point of view of scientific objectivity.
• In terms of personality, Ray is able to be highly critical of parapsychologists and psychics without antagonizing them. Indeed, no doubt every one of us who has talked with Ray knows of his warmth and charm, and knows as well that no matter how stupid or ill-informed our questions might be-and I've asked my share-he never makes us feel foolish for having asked. He is always patient in his explanations and never condescending.

Ray Hyman clearly meets all the criteria in my list. I can think of no other individual in the entire world who could do the same. The In Praise of Reason Award is CSICOP’s highest honor, and is given to those rare individuals who have made truly outstanding contributions to the promotion of science and the defense of reason. Previous recipients include such stellar scientists and communicators as Carl Sagan, Stephen Jay Gould, Nobel laureate Murray Gell-Mann, and our own outstanding editor, Ken Frazier.

Tonight, I take great personal pride in being able to present, on behalf of CSICOP, the In Praise of Reason Award to my friend and colleague Ray Hyman, from whom I-and I am sure all of us-continue to learn so much.

Congratulations, Ray.

The following beliefs are strongly held by large numbers of people. Each of them has been hotly disputed by others:

• Through hypnosis, one can access past lives.
• Horoscopes provide useful information about the future.
• Spiritual healing sometimes succeeds where conventional medicine fails.
• A widespread, transgenerational Satanic conspiracy is afoot in society.
• Certain gifted people have been able to use their psychic powers to help police solve crimes.
• We can sometimes communicate with others via mental telepathy.
• Some people have been abducted by UFOs and then returned to earth.
• Elvis lives.
• Vitamin C can ward off or cure the common cold.
• Immigrants are stealing our jobs.
• Certain racial groups are intellectually inferior.
• Certain racial groups are athletically superior, at least in some specific sports.
• Crime and violence are linked to the breakdown of the traditional family.
• North Korea’s developing nuclear capability poses a threat to world peace.

Despite high confidence on the part of both believers and disbelievers, in most instances, neither side has much — if any — objective evidence to back its position. Some of these beliefs, such as telepathy and astrology, stand in contradiction to the current scientific worldview and are therefore considered by many scientists to be “irrational.” Others are not at all inconsistent with science, and whether or not they are based in fact, no one would consider them to be irrational.

Nineteenth-century rationalists predicted that superstition and irrationality would be defeated by universal education. However, this has not happened. High literacy rates and universal education have done little to decrease such belief, and poll after poll indicates that a large majority of the public believe in the reality of “occult” or “paranormal” or “supernatural” phenomena. Why should this be so? Why is it that in this highly scientific and technological age superstition and irrationality abound?

It is because our brains and nervous systems constitute a belief-generating machine, an engine that produces beliefs without any particular respect for what is real or true and what is not. This belief engine selects information from the environment, shapes it, combines it with information from memory, and produces beliefs that are generally consistent with beliefs already held. This system is as capable of generating fallacious beliefs as it is of generating beliefs that are in line with truth. These beliefs guide future actions and, whether correct or erroneous, they may prove functional for the individual who holds them. Whether or not there is really a Heaven for worthy souls does nothing to detract from the usefulness of such a belief for people who are searching for meaning in life.

Nothing is fundamentally different about what we might think of as “irrational” beliefs — they are generated in the same manner as are other beliefs. We may not have an evidential basis for belief in irrational concepts, but neither do we have such a basis for most of our beliefs. For example, you probably believe that brushing your teeth is good for you, but it is unlikely that you have any evidence to back up this belief, unless you are a dentist. You have been taught this, it makes some sense, and you have never been led to question it.

If we were to conceptualize the brain and nervous system as a belief engine, it would need to comprise several components, each reflecting some basic aspect of belief generation. Among the components, the following units figure importantly:

1. The learning unit
2. The critical thinking unit
3. The yearning unit
4. The input unit
5. The emotional response unit
6. The memory unit
7. The environmental feedback unit.

The Learning Unit

The learning unit is the key to understanding the belief engine. It is tied to the physical architecture of the brain and nervous system; and by its very nature, we are condemned to a virtually automatic process of magical thinking. “Magical thinking” is the interpreting of two closely occurring events as though one caused the other, without any concern for the causal link. For example, if you believe that crossing your fingers brought you good fortune, you have associated the act of finger-crossing with the subsequent welcome event and imputed a causal link between the two.

Our brain and nervous system have evolved over millions of years. It is important to recognize that natural selection does not select directly on the basis of reason or truth; it selects for reproductive success. Nothing in our cerebral apparatus gives any particular status to truth. Consider a rabbit in the tall grass, and grant for a moment a modicum of conscious and logical intellect to it. It detects a rustling in the tall grass, and having in the past learned that this occasionally signals the presence of a hungry fox, the rabbit wonders if there really is a fox this time or if a gust of wind caused the grass to rustle. It awaits more conclusive evidence. Although motivated by a search for truth, that rabbit does not live long. Compare the late rabbit to the rabbit that responds to the rustle with a strong autonomic nervous-system reaction and runs away as fast as it can. It is more likely to live and reproduce. So, seeking truth does not always promote survival, and fleeing on the basis of erroneous belief is not always such a bad thing to do. However, while this avoidance strategy may succeed in the forest, it may be quite dangerous to pursue in the nuclear age.

The learning unit is set up in such a way as to learn very quickly from the association of two significant events — such as touching a hot stove and feeling pain. It is set up so that significant pairings produce a lasting effect, while nonpairings of the same two events are not nearly so influential. If a child were to touch a stove once and be burned, then if the child were to touch it again without being burned, the association between pain and stove would not automatically be unlearned. This basic asymmetry — pairing of two stimuli has an important effect, while presenting the stimuli unpaired (that is, individually) has a much lesser effect — is important for survival.

This asymmetry in learning also underlies much of the error that colors our thinking about events that occur together from time to time. Humans are very poor at accurately judging the relationship between events that only sometimes co-occur. For example, if we think of Uncle Harry, and then he telephones us a few minutes later, this might seem to demand some explanation in terms of telepathy or precognition. However, we can only properly evaluate the co-occurrence of these two events if we also consider the number of times that we thought of Harry and he did not call, or we did not think of him but he called anyway. These latter circumstances — these nonpairings — have little impact on our learning system. Because we are overly influenced by pairings of significant events, we can come to infer an association, and even a causal one, between two events even if there is none. Thus, dreams may correspond with subsequent events only every so often by chance, and yet this pairing may have a dramatic effect on belief. Or we feel a cold coming on, take vitamin C, and then when the cold does not get to be too bad we infer a causal link. The world around us abounds with coincidental occurrences, some of which are meaningful but the vast majority of which are not. This provides a fertile ground for the growth of fallacious beliefs. We readily learn that associations exist between events, even when they do not. We are often led by co-occurring events to infer that the one that occurred first somehow caused the one that succeeded it.

We are all even more prone to error when rare or emotionally laden events are involved. We are always looking for causal explanations, and we tend to infer causality even when none exists. You might be puzzled or even distressed if you heard a loud noise in your living room but could find no source for it.

The Critical-Thinking Unit

The critical-thinking unit is the second component of the belief engine, and it is acquired — acquired through experience and explicit education. Because of the nervous-system architecture that I have described, we are born to magical thinking. The infant who smiles just before a breeze causes a mobile above her head to move will smile again and again, as though the smile had magically caused the desired motion of the mobile. We have to labor to overcome such magical predisposition, and we never do so entirely. It is through experience and direct teaching that we come to understand the limits of our immediate magical intuitive interpretations. We are taught common logic by parents and teachers, and since it often serves us well, we use it where it seems appropriate. Indeed, the cultural parallel of this developmental process is the development of the formal method of logic and scientific inquiry. We come to realize that we cannot trust our automatic inferences about co-occurrence and causality.

We learn to use simple tests of reason to evaluate events around us, but we also learn that certain classes of events are not to be subjected to reason but should be accepted on faith. Every society teaches about transcendental things — ghosts, gods, bogeymen, and so on; and here we are often explicitly taught to ignore logic and accept such things on faith or on the basis of other people’s experiences. By the time we are adults, we can respond to an event in either a logical, critical mode or in an experiential, intuitive mode. The events themselves often determine which way we will respond. If I were to tell you that I went home last night and found a cow in my living room, you would be more likely to laugh than to believe me, even though there is certainly nothing impossible about such an event. If, on the other hand, I were to tell you that I went into my living room and was startled by an eerie glow over my late grandfather’s armchair, and that the room went cold, you may be less likely to disbelieve and more likely to perk up your ears and listen to the details, possibly suspending the critical acumen that you would bring to the cow story. Sometimes strong emotion interferes with the application of critical thought. Other times we are cleverly gulled.

Rationality is often at a disadvantage to intuitive thought. The late psychologist Graham Reed spoke of the example of the gambler’s fallacy: Suppose you are observing a roulette wheel. It has come up black ten times in a row, and a powerful intuitive feeling is growing in you that it must soon come up red. It cannot keep coming up black forever. Yet your rational mind tells you that the wheel has no memory, that each outcome is independent of those that preceded. In such a case, the struggle between intuition and rationality is not always won by rationality.

Note that we can switch this critical thinking unit on or off. As I noted earlier, we may switch it off entirely if dealing with religious or other transcendental matters. Sometimes, we deliberately switch it on: “Hold it a minute, let me think this out,” we might say to ourselves when someone tries to extract money from us for an apparently worthy cause.

The Yearning Unit

Learning does not occur in a vacuum. We are not passive receivers of information. We actively seek out information to satisfy our many needs. We may yearn to find meaning in life. We may yearn for a sense of identity. We may yearn for recovery from disease. We may yearn to be in touch with deceased loved ones.

In general we yearn to reduce anxiety. Beliefs, be they correct or false, can assuage these yearnings. Often beliefs that might be categorized as irrational by scientists are the most efficient at reducing these yearnings. Rationality and scientific truth have little to offer for most people as remedies for existential anxiety. However, belief in reincarnation, supernatural intervention, and everlasting life can overcome such anxiety to some extent.

When we are yearning most, when we are in the greatest need, we are even more vulnerable to fallacious beliefs that can serve to satisfy those yearnings.

The Input Unit

Information enters the belief engine sometimes in the form of raw sensory experience and other times in the form of organized, codified information presented through word of mouth, books, or films. We are wonderful pattern detectors, but not all the patterns we detect are meaningful ones. Our perceptual processes work in such a way as to make sense of the environment around us, but they do make sense — perception is not a passive gathering of information but, rather, an active construction of a representation of what is going on in our sensory world. Our perceptual apparatus selects and organizes information from the environment, and this process is subject to many well-known biases that can lead to distorted beliefs. Indeed, we are less likely to be influenced by incoming information if it does not already correspond to deeply held beliefs. Thus, the very spiritual Christian may be quite prepared to see the Virgin Mary; information or perceptual experience that suggests that she has appeared may be more easily accepted without critical scrutiny than it would be by someone who is an atheist. It is similar with regard to experiences that might be considered paranormal in nature.

The Emotional Response Unit

Experiences accompanied by strong emotion may leave an unshakable belief in whatever explanation appealed to the individual at the time. If one is overwhelmed by an apparent case of telepathy, or an ostensible UFO, then later thinking may well be dominated by the awareness that the emotional reaction was intense, leading to the conclusion that something unusual really did happen. And emotion in turn may directly influence both perception and learning. Something may be interpreted as bizarre or unusual because of the emotional responses triggered.

Evidence is accumulating that our emotional responses may be triggered by information from the outside world even before we are consciously aware that something has happened. Take this example, provided by LeDoux (1994) in his recent article in Scientific American (1994, 270, pp. 50-57):

An individual is walking through the woods when she picks up information — either auditory, such as rustling leaves, or visual, such as the sight of a slender curved object on the ground — which triggers a fear response. This information, even before it reaches the cortex, is processed in the amygdala, which arouses the body to an alarm footing. Somewhat later, when the cortex has had enough time to decide whether or not the object really is a snake, this cognitive information processing will either augment the fear response and corresponding evasive behaviour, or will serve to bring that response to a halt.

This is relevant to our understanding of paranormal experience, for very often an emotional experience accompanies the putatively paranormal. A strong coincidence may produce an emotional “zing” that points us toward a paranormal explanation, because normal events would not be expected to produce such emotion.

Our brains are also capable of generating wonderful and fantastic perceptual experiences for which we are rarely prepared. Out-of-body experiences (OBEs), hallucinations, near-death experiences (NDEs), peak experiences — these are all likely to be based, not in some external transcendental reality, but rather in the brain itself. We are not always able to distinguish material originating in the brain from material from the outside world, and thus we can falsely attribute to the external world perceptions and experiences that are created within the brain. We have little training with regard to such experience. As children, we do learn to distrust, for the most part, dreams and nightmares. Our parents and our culture tell us that they are products of our own brains. We are not prepared for more arcane experiences, such as OBEs or hallucinations or NDEs or peak experiences, and may be so unprepared that we are overwhelmed by the emotion and come to see such experience as deeply significant and “real” whether or not it is.

Ray Hyman has always cautioned skeptics not to be surprised should they one day have a very strong emotional experience that seems to cry out for paranormal explanation. Given the ways our brains work, we would expect such experiences from time to time. Unprepared for them, they could become conversion experiences that lead to strong belief. When I was a graduate student, another graduate student who shared my office, and who was equally as skeptical as I was about the paranormal, came to school one day overwhelmed by the realism and clarity of a dream he had had the night before. In it, his uncle in Connecticut had died. It had been a very emotional dream, and was so striking that Jack told me that if his uncle died anytime soon, he would no longer be able to maintain his skepticism about precognition — the dream experience was that powerful. Ten years later, his uncle was still alive, and Jack’s skepticism had survived intact.

The Memory Unit

Through our own experience, we come to believe in the reliability of our memories and in our ability to judge whether a given memory is reliable or not. However, memory is a constructive process rather than a literal rendering of past experience, and memories are subject to serious biases and distortions.

Not only does memory involve itself in the processing of incoming information and the shaping of beliefs; it is itself influenced strongly by current perceptions and beliefs. Yet it is very difficult for an individual to reject the products of his or her own memory process, for memory can seem to be so “real.”

The Environmental Feedback Unit

Beliefs help us to function. They guide our actions and increase or reduce our anxieties. If we operate on the basis of a belief, and if it “works” for us, even though faulty, why would we be inclined to change it? Feedback from the external world reinforces or weakens our beliefs, but since the beliefs themselves influence how that feedback is perceived, beliefs can become very resistant to contrary information and experience. If you really believe that alien abductions occur, then any evidence against that belief can be rationalized away — in terms of conspiracy theories, other people’s ignorance, or whatever.

As mentioned earlier, fallacious beliefs can often be even more functional than those based in truth. For example, Shelley Taylor, in her book Positive Illusions, reports research showing that mildly depressed people are often more realistic about the world than are happy people. Emotionally healthy people live to some extent by erecting false beliefs — illusions — that reduce anxiety and aid well-being, whereas depressed individuals to some degree see the world more accurately. Happy people may underestimate the likelihood of getting cancer or being killed, and may avoid thinking about the ultimate reality of death, while depressed people may be much more accurate with regard to such concerns.

An important way in which to run reality checks on our perceptions and beliefs is to compare them with those of others. If I am the only one who interpreted a strange glow as an apparition, I am more likely to reconsider this interpretation than if several others share the same view. We often seek out people who agree with us, or selectively choose literature supporting our belief. If the majority doubts us, then even if only part of a minority we can collectively work to dispel doubt and find certainty. We can invoke conspiracies and coverups to explain an absence of confirmatory evidence. We may work to inculcate our beliefs in others, especially children. Shared beliefs can promote social solidarity and even a sense of importance for the individual and group.

In Conclusion

Beliefs are generated by the belief engine without any automatic concern for truth. Concern for truth is a higher order acquired cognitive orientation that reflects an underlying philosophy which presupposes an objective reality that is not always perceived by our senses.

The belief engine chugs away, strengthening old beliefs, spewing out new ones, rarely discarding any. We can sometimes see the error or foolishness in other people’s beliefs. It is very difficult to see the same in our own. We believe in all sorts of things, abstract and concrete — in the existence of the solar system, atoms, pizza, and five-star restaurants in Paris. Such beliefs are no different in principle from beliefs in fairies at the end of the garden, in ghosts in some deserted abbey, in werewolves, in satanic conspiracies, in miraculous cures, and so on. Such beliefs are all similar in form, all products of the same process, even though they vary widely in content. They may, however, involve greater or lesser involvement of the critical-thinking and emotional-response units.

Critical thinking, logic, reason, science — these are all terms that apply in one way or another to the deliberate attempt to ferret out truth from the tangle of intuition, distorted perception, and fallible memory. The true critical thinker accepts what few people ever accept — that one cannot routinely trust perceptions and memories. Figments of our imagination and reflections of our emotional needs can often interfere with or supplant the perception of truth and reality. Through teaching and encouraging critical thought our society will move away from irrationality, but we will never succeed in completely abandoning irrational tendencies, again because of the basic nature of the belief engine.

Experience is often a poor guide to reality. Skepticism helps us to question our experience and to avoid being too readily led to believe what is not so. We should try to remember the words of the late P. J. Bailey (in Festus: A Country Town): “Where doubt, there truth is — ‘tis her shadow.”