As described in the previous chapter, our poor appreciation of probability sometimes leads us to draw faulty conclusions and see significance in events where we should not. This chapter will consider other cognitive biases that may underlie our tendency to misinterpret events as involving paranormal forces when, in fact, mundane explanations will suffice.
Back in the 1970s, Russell Targ and Harold Puthoff published sensational results in the prestigious journal Nature in support of the idea that it was possible to accurately perceive information at remote locations without the use of the known sensory channels.1 They were reporting the findings from investigations using the technique of remote viewing. A typical remote viewing study involves one or more individuals visiting a randomly selected location, such as a park, a bridge, a shopping center, or a library, at a prespecified time and attempting to telepathically transmit information about that location to an individual acting as a “receiver” back at base.2 The receiver describes any impressions going through their mind at the time and may also make drawings to record any visual images. An experimenter, who is also blind to the target location, may ask the receiver to clarify anything ambiguous or unclear. A number of such trials are carried out. Transcripts of the receiver’s impressions are then issued to independent judges, who visit the randomly selected sites and assess the degree of match between each transcript and each site. The receiver themselves may also sometimes be involved in such judging.
Targ and Puthoff claimed that the results they had obtained from the study reported in Nature were highly statistically significant. Across nine locations, their independent judge had correctly matched transcripts to locations on no fewer than seven out of nine trials. Intrigued by these impressive results, David Marks and Richard Kammann attempted to replicate the study. Their results offered no support whatsoever for the original claim of paranormal information transfer. Furthermore, after persistent detective work, they discovered that there were major flaws in the original study that fully explained the apparently significant results reported by Targ and Puthoff without any need to invoke psychic powers. The debate regarding whether remote viewing studies really do provide strong evidence for extrasensory perception rumbles on, but I want to focus here on an interesting phenomenon noted by Marks and Kammann in the course of carrying out their attempted replication.
Over a series of thirty-five experiments, Marks and Kammann noted strong similarities between transcripts and locations on many occasions, as did their independent judges and the receivers themselves. This initially gave all concerned the very strong impression that some kind of psychic communication had taken place. The only problem was that the transcripts that appeared to match were rarely the ones generated when the senders were visiting that particular location. Any correspondences noted were purely the result of coincidental matches between transcripts and locations. Marks and Kammann coined the term subjective validation to refer to this phenomenon, which they described as occurring “when two unrelated events are perceived to be related because a belief, expectancy, or hypothesis demands or requires a relationship.”3
In the case of remote viewing studies, the transcripts (and accompanying drawings, if any) contain a large number of elements. Similarly, looking around at any randomly selected location, one might view a large number of distinct elements that provide potential matches to those in the transcripts. It is very likely that some matching elements will be found. Subjective validation comes into play in many other paranormal contexts. When a psychic or other type of diviner gives a reading for a client, there is huge potential to find matches between the many elements in a reading, on the one hand, and the rich tapestry of a human life, on the other. As we shall see in chapter 11, subjective validation is an important factor in explaining why dreams sometimes appear to match future events.
Probabilistic reasoning is not the only type of reasoning that human beings find difficult. We also often struggle with certain types of deductive reasoning, including syllogistic reasoning.4 A syllogism consists of two statements, known as premises, and a third, the conclusion. Performance on syllogistic reasoning tasks is typically assessed by presenting participants with a set of syllogisms and asking them to judge whether the given conclusions are valid or invalid, assuming the premises to be true.
For example, one type of syllogism is known as a categorical syllogism. A typical example is:
The conclusion drawn here is valid, meaning that it follows logically from the two premises. If the first two statements are true, then the third statement must also be true.
Note that in syllogistic reasoning tasks one is not being asked whether the premises or conclusion are true, simply whether the conclusion follows logically from the premises if the premises are true. For example, the following syllogism is valid even though we might disagree with the conclusion:
We tend to be influenced in our judgments about the validity of syllogisms by the degree to which we believe the conclusion, an effect known as belief bias. Thus, for example, we are more likely to judge invalid reasoning as valid if we agree with the conclusion given and vice versa.
Another type of syllogism is known as a conditional syllogism. An example would be:
It is a trivially easy task to confirm that it follows from the first two premises here that the conclusion must be true. Sometimes, however, it is not so easy to judge the validity of conditional syllogisms. A common error is what logicians refer to as affirmation of the consequent. Here is an example:
This logical error is very common among conspiracy theory believers where a lack of evidence supporting a conspiracy is often interpreted as evidence of a cover-up.
Readers may have spotted that this is the error made by those claiming that 3.7 million Americans have been abducted by aliens, as described in chapter 5. Even if people really were being abducted by aliens, and even if these abductions really did cause people to experience paralysis, a sense of presence, and so on, the conclusions drawn from the responses to the Roper poll would still not be justified. Here is their argument presented in typical syllogistic form:
The conclusion would only logically follow in this case, as with all similar conditional syllogisms, if the if in the first premise meant if and only if. In other words, if there are other possible reasons why the second premise might be true, they cannot logically be ruled out. In this case, there is a very plausible alternative explanation for reports of paralysis, a sense of presence, and so on—our old friend sleep paralysis.
Might it be the case that believers in the paranormal are somewhat poorer at deductive reasoning than nonbelievers, and therefore somewhat more prone to drawing erroneous conclusions from the available evidence? An early study by Michael Wierzbicki supported this hypothesis, but a subsequent replication attempt by Harvey Irwin failed to find any difference between paranormal believers and nonbelievers in terms of their reasoning ability.5 Irwin speculated that the different pattern of results obtained in the two studies might be due to an experimenter effect.
Both studies used student samples. Irwin proposed that the brighter students in Wierzbicki’s study may have been aware that Wierzbicki was likely to be rather skeptical regarding paranormal claims, in line with the general attitude of most psychologists. These students may therefore have tended to be reluctant to admit to any pro-paranormal beliefs they held, thus producing a negative correlation between reasoning ability and (professed) level of belief. In contrast, Irwin’s own students would know him to be more sympathetic toward paranormal claims and thus would feel free to express their true attitudes toward the paranormal. As a consequence, Irwin claimed, no artifactual correlation would result. Although this suggested explanation is certainly plausible, Maxwell Roberts and Paul Seager highlighted other possible reasons for Irwin’s failure to find any correlation between reasoning ability and paranormal belief, claiming that Irwin’s reasoning task differed in important ways from that used by Wierzbicki.6 In their own study, Roberts and Seager did indeed replicate the negative correlation reported by Wierzbicki.
One study of the relationship between paranormal belief and syllogistic reasoning, by Caroline Watt and Richard Wiseman, did indeed find experimenter effects.7 Watt and Wiseman followed identical procedures and yet, whereas Watt obtained a large negative correlation between the two variables, Wiseman failed to find any correlation. Unfortunately for Irwin’s hypothesis, however, these results are the opposite of what that hypothesis would have predicted. It is Caroline Watt who is known to be more sympathetic toward paranormal claims whereas Richard Wiseman is a well-known and vocal skeptic—another case of nice theory, shame about the data!
It is generally accepted within psychology that we are not consciously aware of most mental processes. Typically, we are aware of the products of mental processing rather than the processing itself. As a simple demonstration, mentally answer the following question: What was the name of your first school? The answer, no doubt, simply automatically popped into your head, but you had no conscious access to the processes involved in reading and understanding the question and then retrieving the correct piece of information from all the information stored in your long-term memory. The same is true for other psychological processes. For example, during waking consciousness we feel as though we have immediate direct access to the visual and auditory world around us, whereas in fact that mental model of the world and our place in it is the result of complex interactions between bottom-up and top-down processing, as described in chapter 3.
The distinction between conscious and nonconscious processing is at the heart of an interesting concept that was first proposed by Michael Thalbourne and Peter Delin in 1994 as a possible explanation for the fact that measures of paranormal belief and experience are found to consistently correlate with a range of other psychological variables, including creativity, susceptibility to mystical experiences, and certain psychopathological tendencies.8 Their hypothesis was that a single common factor might underlie all of these intercorrelations. They labeled this factor transliminality, which they initially defined as “the extent to which the contents of some preconscious (or ‘unconscious’ or ‘subliminal’) region of the mind are able to cross the threshold into consciousness (in its sense of ‘awareness’).” Subsequently, they defined it as “a hypersensitivity to psychological material originating in (a) the unconscious, and/or (b) the external environment. ‘Psychological material’ is taken to cover ideation, imagery, affect and perception, and thus a rather broad concept.”9
A useful way to think about transliminality is to imagine that the mind contains both conscious and unconscious areas and that they are separated by a semipermeable membrane through which material must pass in order to cross the threshold from unconsciousness into conscious awareness. The idea is that people differ in terms of the permeability of this membrane. Thus, a highly transliminal individual may have conscious access to material that would not normally be accessible to a low transliminal person. Thalbourne, a believer in the reality of psi, originally developed this concept as a possible explanation for differences between individuals in terms of their psychic abilities. If ESP signals are received initially by the nonconscious part of the mind, he reasoned, such differences may be explained by inherent differences in transliminality.
Although I personally do not believe that psi exists, I too found the concept of transliminality of interest. A number of commentators had already suggested that certain experiences that a person might interpret in paranormal terms might in fact be explicable in terms of nonconscious processing.10 For example, in some of the early studies of ESP by J. B. Rhine using Zener cards (figure 8.1), as described in chapter 8, it is said that some of the cards were so heavily embossed that some participants could just about make out what the symbol was from the back of the card.11 One might expect participants would let the experimenter know that they were not actually trying to use ESP but instead could see the symbol using normal vision. Is it possible that these participants were not consciously aware that they had processed this information even though it was influencing their guesses on each trial? Furthermore, might it be the case that highly transliminal individuals would be more likely to score above the level expected by chance by this means compared to low transliminals?
Such a possibility is supported by the results of a study that I carried out in collaboration with Susan Crawley as part of her PhD research.12 Participants were asked to take part in a computerized ESP task. On each trial, the computer would randomly select one of the five Zener card symbols, and then the image of a swirly pattern, like those often seen on the back of playing cards, would appear on the screen. Participants would then enter their guess. What participants were not told was that on half the trials the correct answer had been subliminally presented to them by flashing it on the screen so briefly before the swirly pattern appeared that it was below the threshold of conscious awareness. Participants also completed a questionnaire designed by Thalbourne to assess transliminality. As we had expected, the scores on this scale correlated with the number of correct guesses on the computerized task—but only on the subliminally primed trials. On the unprimed trials, there was no correlation. Although no evidence for actual ESP was found, it would not be surprising if our high transliminal participants seriously entertained the idea that they might be psychic. After all, how else could they explain their above-average performance on the task?
If the numerous cognitive biases identified in recent decades are as ubiquitous as they appear to be, they may provide some sort of advantage in terms of our evolutionary history. At first glance, this may appear to be somewhat surprising. Surely any kind of cognitive bias that leads us to misinterpret, misperceive, or misremember events would be a disadvantage in evolutionary terms and thus would have been eliminated by natural selection? The solution to this puzzle is provided by an appreciation of what has become known as System 1 and System 2 thinking.13
It is now widely accepted that human thought operates in two distinct modes, as reflected in the title of Daniel Kahneman’s bestseller, Thinking, Fast and Slow.14 System 1 thinking corresponds to what we might loosely refer to as intuition. Such thinking is fast, nonconscious, effortless, emotional, and automatic and involves no sense of voluntary control. A typical example of System 1 thinking is meeting someone for the first time and taking an instant liking or disliking to them. By contrast, System 2 thinking is slow, conscious, effortful, and unemotional and feels under voluntary control. Such thinking corresponds loosely to what might be called rational thought. A typical example of System 2 thinking is attempting to solve a mental puzzle. Many of the cognitive biases discussed in this book are based on System 1 thinking. A great deal of evidence has been produced showing that System 1 influences our attitudes, beliefs, and behavior to a much greater degree than we might assume.
In an evolutionary context, one of the most important differences between System 1 and System 2 thinking is speed. Our brains have evolved to maximize the chances of us staying alive long enough to pass on our genes to the next generation. Our ancestors lived in a world full of dangers from enemies and predators. Relying primarily on System 1 thinking, which was both fast and usually delivered the right answer, was a better bet than relying primarily on System 2 thinking, which, although likely to deliver the right answer a bit more often, was much slower.
It simply made sense for our ancestors to rely on the quick and dirty rules of thumb of System 1, technically known as heuristics, in their day-to-day interactions with the world around them rather than on System 2. For example, a rustling in the bushes may or may not have been a predator. By relying on System 1 thinking and assuming it was indeed a threat, our ancestors were more likely to stay alive long enough to procreate. If they were correct and responded appropriately in terms of either fight or flight, they maximized their chances of staying alive. If they were incorrect, the cost was minimal. To make the opposite error of assuming that there was no threat or simply taking too long to make any decision at all, they maximized their chances of becoming lunch. Reliance on System 2 thinking was clearly not the better option.
Evolutionary pressures such as these were likely to result in cognitive systems with a bias toward what scientists refer to as false positives. In science, these are also known as a Type I errors. For example, a medical researcher may wrongly conclude that a new form of treatment is effective, perhaps because the results came from a clinical trial that was affected by confounding variables. Type II errors, in contrast, are false negatives, for example, a medical scientist wrongly concluding that a treatment was ineffective, perhaps as a result of basing their conclusion on unreliable data.
Our success as a species depends on our ability to spot meaningful patterns and recognize cause-and-effect relationships in our environment. Our ancestors would have used such skills in hunting, avoiding predators, and developing agriculture, for example. The problem is that sometimes, in line with our general bias toward Type I errors, we may think we see meaningful patterns or cause-and-effect relationships when in fact there are none, as previously discussed with respect to pareidolia, our poor appreciation of the true nature of randomness, and subjective validation.
Michael Shermer refers to our tendency to see meaningful patterns where there are none as patternicity.15 Others have used the term apophenia to describe this tendency, although it should be noted that when Klaus Conrad first coined the latter term in 1958, it was intended not only to refer to the tendency to see meaningful patterns in randomness and to make connections between unrelated things but also to perceive deep personal significance in such occurrences, as may occur in schizophrenia.16 Thus, although we may all notice a pattern resembling a face in a stain on a church wall, only some of us will see this as more than an accidental configuration of marks. These days, apophenia is used more loosely than Conrad intended and, in some contexts, is used interchangeably with patternicity and pareidolia.
Humans also appear to have an in-built tendency to assume that when something happens in their vicinity, it happens because someone or something made it happen for some specific reason. Again, this would make sense in terms of human evolution, as it would lead us to routinely assess any event occurring nearby in terms of the degree of threat that it might pose, assuming it was caused by some sentient agent with specific intentions toward us.
Justin Barrett has plausibly argued that belief in gods was a natural consequence of this tendency.17 He hypothesizes that humans possess a mental module that he refers to as a hyperactive agency detective device (HADD) that provided an evolutionary advantage in ensuring that we would be more inclined to make Type I errors in ambiguous but potentially threatening situations and thus more likely to survive. We would also be naturally inclined to assume that some sentient being or beings caused thunder and lightning, natural disasters, and crop failures, not to mention disease and injuries, perhaps as punishment for our wrongdoing. Such reasoning clearly can be applied to belief in a range of supernatural beings, not just gods.
Our brains evolved over millennia, and the pace of evolutionary change is slow. Our brains are therefore pretty much indistinguishable from those of our recent evolutionary ancestors. When considered in this context, it is not at all surprising that much of our thinking is based on the heuristics that characterize System 1 as opposed to the more rational System 2 with all of the advantages and disadvantages that follow from that.