7
In the present chapter I discuss the causal history of beliefs, postponing to Chapter 13 a discussion of the normative principles of belief formation.
What is it to “believe” something?
To understand the role of beliefs in generating action, we have to understand their nature, their causes, and their consequences. As I mentioned in the introductory remarks to Part II, it is not always clear what it means to “believe” that something is the case. Did the followers of Communism who “believed” that the party could do no wrong really believe it?1 How can we tell the difference between the congenital pessimist who tends to believe the worst and the prudent decision maker who merely acts as if the worst-case scenario were true? How can we tell the difference between risk aversion (a formal preference) and pessimism (a belief)?
Also, in everyday language “belief” suggests less than full endorsement. I believe it will rain tomorrow, but I also know I might be wrong. I do not merely believe that I am married; I know it. In philosophical analyses, knowledge is usually defined as justified true belief, a belief that stands in a particular relation both to the world (it is true) and to the body of evidence the agent possesses (it is justified). Yet neither of these features of knowledge captures the subjective certainty that often underlies the phrase “I know” in ordinary discourse. This certainty is not simply the limit of 97 percent probability, 98 percent, 99 percent, 99.9 percent, and so forth. It is qualitatively different from anything short of certainty.
This “certainty effect” shows up in the following experiment. One group of subjects was asked to express their preferences over various options. (Numbers in parentheses indicate the proportion of subjects preferring a given option.)
A 50 percent chance to win a three-week tour of England, France, and Italy (22 percent).
A one-week tour of England, with certainty (78 percent).
Another group was given the following options:
A 5 percent chance to win a three-week tour of England, France, and Italy (67 percent).
A 10 percent chance to win a one-week tour of England (33 percent).
Members of the first group tend to prefer the “England only” option because it is available for sure. Once it is deflated by the same probability as the alternative, the latter looks more attractive. Soldiers who are asked whether they will volunteer for highly dangerous missions may have disproportionately fewer hesitations than those who are asked to volunteer for suicide missions. The former may also, of course, be subject to wishful thinking (“It won't happen to me”), which has no purchase on the latter.
Four cognitive attitudes
Even setting aside these problems, the idea of belief remains ambiguous. We may distinguish among four cognitive attitudes to the world, with decreasing strength. First is the mode of certainty. Second is the mode of risk, in which agents assign probabilities, whether based on past frequencies or their own judgment, to each of a set of mutually exclusive and jointly exhaustive outcomes. Third is the mode of uncertainty, in which people know the set of mutually exclusive and jointly exhaustive outcomes but find themselves unable to attach any (cardinal) probabilities to them. They may be able to attach ordinal probabilities to the outcomes, that is, to say that one outcome is more likely to occur than another, without being able to say how likely they are (see Chapter 13 for an example). For practical purposes, that situation is no better than uncertainty. Finally is the mode of ignorance, in which both the range of possible outcomes and their probability of occurrence are unknown or incompletely known.2 In the memorable words of former Defense Secretary Donald Rumsfeld, we are facing not only known and unknown quantities, but also “unknown unknowns.”
I focus on certainty and risk, not because these are always the appropriate cognitive attitudes, but because they are the most common ones. Even when people have no grounds for having any belief on a given topic, they often feel irresistibly compelled to form an opinion – not a specific opinion (as in wishful thinking), but some opinion or other. This propensity is to some extent determined by cultural factors. Albert Hirschman has said that most Latin American cultures “place considerable value on having strong opinions on virtually everything from the outset.” In such societies, to admit ignorance is to admit defeat. But the tendency is really universal. Montaigne said that “many of this world's abuses are engendered – or to put it more rashly, all of this world's abuses are engendered – by our being schooled to be afraid to admit our ignorance and because we are required to accept anything which we cannot refute.” The intolerance of uncertainty and ignorance flows not only from pridefulness, but from a universal human desire to find meanings and patterns everywhere (see Chapter 9). The mind abhors a vacuum.
Social agents are averse to uncertainty because it makes them feel uncomfortable. Scholars can be averse to uncertainty because it makes it difficult to prove theorems. The notion of expected utility maximization that is at the core of economic theory is undefined if expectations are. Economists, therefore, often attribute subjective probabilities to the agents without providing evidence that these have any kind of psychological reality. As I mentioned in the Introduction to Part II, they may for instance assume that in a situation of uncertainty, an agent will think all options equally likely (a “uniform” or flat probability distribution). As I argued, this procedure is arbitrary. Natural scientists also appeal to uniform distributions when trying to understand natural phenomena, for instance when trying to predict climate change. One study criticizes this procedure by observing that “physically, we can equally well use a parameter labeled ‘ice fall rate in clouds’ or its inverse (‘ice residence time in clouds’) and achieve identical simulations. Sampling uniform distributions under each of the two different labels however, yields completely different results.”
In other cases, scholars assume that agents have rational expectations (Chapter 6), that is, that they share the information and risk assessments of the modeler. Often, this procedure is manifestly adopted because of the need to prove theorems rather than because it is supported by evidence. In the Conclusion, I cite a particularly egregious example in which scholars impute to revolutionary agents a sharp probability about the future state of the economy. After the recent financial crisis, however, many economists have been more willing to follow Keynes in his rejection of the idea that agents act to maximize “a weighted average of quantitative benefits multiplied by quantitative probabilities.”
These error-generating mechanisms rely in one way or another on motivation. Yet error can also arise from ignorance. The point seems obvious but is actually a bit subtle. Darwin noted, for instance, that “ignorance more frequently begets confidence than does knowledge.” Ignorance together with confidence is a good recipe for error. Conversely, when the circle of light expands, so does the surrounding area of darkness, inducing greater humility. As Adam Smith noted, it “is the inferior artist only, who is ever perfectly satisfied with his own performances.” Experiments suggest in fact that incompetence not only causes poor cognitive performance, but also the inability to recognize that one's competence is poor. The incompetent are doubly handicapped.
Subjective assessments of probability
Probability judgments can stem from observation of objective frequencies or be purely subjective evaluations.3 When the agent can draw on a large number of observations of similar situations, the frequentist method can yield good results. If I plan to have a picnic on my birthday next month and need to form an opinion about the likely weather, the best I can do is probably to look up the weather statistics for the same day in previous years. But if I need to form an opinion about the weather tomorrow, the best single predictor is today's weather. It is not, however, the only predictor. Past records can tell me whether sunny weather on that day is a rare or normal event. If it is rare, today's sunny weather loses some of its predictive value. I may consult the barometer on my wall to see whether the air pressure is rising or falling or look at the evening sky, the flight of the swallows, and so on.
To integrate all this information into an overall probability judgment about tomorrow's weather is a difficult task. Most of us are not very good at it. Often, the problem is not lack of information, but an abundance of it, combined with the lack of a formal procedure for integrating it into an all-things-considered opinion. Some people, though, are better than most of us at integrating vast and diffuse information with varying degrees of relevance into an overall assessment. They possess the elusive but crucial quality of judgment. Successful generals, businesspeople, and politicians tend to have it – that is why they succeed. A good central banker needs to have it, but most economists do not.4 The best the rest of us can do is to recognize that we do not have it and learn not to trust our intuition. I may learn, for instance, that I often distrust people for reasons that, when I come to understand them, are irrelevant. (“He looked like a bully I knew in fifth grade.”) Hence I may come to distrust my distrust.5
We tend to think, however, that judgment is possessed not only by successful generals, politicians, and businesspeople, but also by trained experts. In complicated matters of diagnosis or prognosis, such as identifying psychotic individuals or assessing how likely it is that a person who requests early release from prison will commit a second offense, we trust the expert. Because of their experience, experts are sensitive to telltale signs that untrained observers might ignore or whose significance they might not understand. Moreover, when different pieces of evidence point in different directions, experts can draw on their experience to decide which, in any given case, should be given most weight. This at least is how we think about experts. As most of us consider ourselves experts in some domain or other, if nothing else in predicting the behavior of our boss, spouse, or children, we have a great deal invested in this image of the superior cognitive skills of the expert.
Unfortunately, this image is thoroughly false. In many studies the diagnostic or prognostic performance of experts has been compared with the performance of a simple mechanical formula based on a few variables. Essentially, this amounts to comparing objective (frequentist) methods and subjective ones. The weights assigned to the variables are derived by statistical techniques that assign the weights most likely to predict observed outcomes. Almost without exception, the formula performs at least as well as the expert and usually better.6 In a study of the diagnosis of progressive brain dysfunctioning based on intellectual testing, to cite only one example, a formula derived from one set of cases and then applied to a new sample correctly identified 83 percent of the new cases. Groups of experienced and inexperienced clinicians correctly identified 63 percent and 58 percent, respectively. Moreover, experts often disagree strongly with one another. In another study, highly experienced psychiatrists who viewed the same psychiatric interview could not agree on the patient's diagnosis, motivations, or feelings. Some psychotherapists use responses to ambiguous inkblots as cues to diagnoses. It appears, however, that the patients are as ambiguous to them as the inkblots to the patients.7
Some errors of statistical inference
Experts no less than laypersons often go wrong because they ignore obvious or not-so-obvious principles of statistical reasoning. In one study, subjects were given a description of a young man with long hair and a habit of reading poetry and asked whether they thought it more likely that he was an orchestra violinist or a truck driver. Most said he was more likely to be a violinist, thus ignoring the base rate of the two groups, that is, the absolute number of individuals in each. There are so many more truck drivers than orchestra violinists in the nation (and so much variation among truck drivers) that the poetic young man is in fact more likely to drive a truck. This mistake is referred to as the base-rate fallacy.
Another source of mistakes in belief formation is selection bias. Think of a man arriving at a railway station and examining a map of the local area with a large red dot on it labeled “You Are Here,” and being amazed that the railway company knew he would be there at that time. More seriously, patients in dialysis centers are often surprisingly reluctant to be on the waiting list for a kidney transplantation. One reason is that all the transplanted patients they ever see are those for whom the operation failed so that they had to go back on dialysis. Montaigne cited a bias of this kind when he referred to Diagoras as being “shown many vows and votive portraits from those who have survived shipwrecks and … then asked, ‘You, there, who think that the gods are indifferent to human affairs, what have you to say about so many men saved by their grace?’ – ‘It is like this,’ he replied, ‘there are no portraits here of those who stayed and drowned – and they are more numerous!’” Similarly, a psychiatrist who claims that “no child abusers ever stop on their own” neglects the fact that if any do he is unlikely to have met them. In À la recherche du temps perdu the Narrator observes about Charlus that “everything made him become Germanophile, because … he lived in France. He was very keen-witted and in all countries fools outnumber the rest; no doubt, if he had lived in Germany the German fools defending an unjust cause with passion and folly would have irritated him; but living in France, the French fools, defending a just cause with passion and folly, irritated him no less.” As a final example of this very common reasoning flaw, consider an inference by American officers during World War II: observing that there were fewer bullet holes in the engines of planes that came back than in other parts, they concluded that armor needed to be concentrated in what appeared to be the most vulnerable parts. The statistician Abraham Wald observed, however, that the reason planes were coming back with fewer hits to the engine was that most planes that got hit in the engine were not coming back. In the words of the writer from whom I take this example, “The armor goes where the bullet holes aren't” rather than where they are.
Israeli air force leaders made a less obvious mistake when assessing the relative efficacy of reward and punishment in the training of pilots. Noting that the performance of pilots improved when they were punished for a bad performance but not when they were rewarded for a good one, they concluded that punishment was more efficient. In doing so, they ignored the phenomenon known as regression to the mean. In any series of events that are fully or partly determined by chance, there is a tendency for an extreme value on one occasion to be followed by a less extreme value on the next. Tall fathers get sons who are shorter than they are, and bad pilot performances are followed by less bad ones, independently of reward and punishment. When athletes who have done exceptionally well in one season do less well the next, fans and coaches often say they have been spoiled by success, when what we observe may only be regression to the mean.
Many scares about incidence of disease or harm in a particular community or profession are due to what epidemiologists call the “Texas sharpshooter effect”: blast a barn door with a shotgun and then find the holes that are closest together. Draw a target around them and it looks like you hit a bull's-eye. When there was a cluster of suicides in the French postal service some years ago, newspapers blamed a toxic work environment, until statisticians pointed out that any random process will generate clusters (see below). Many cancer scares are also spurious. A specialist on epidemiology calculated that given a typical registry of eighty different cancers, you would expect 2,750 out of California's 5,000 census tracts to have statistically significant but perfectly random elevations of cancer. In Chapter 9, I return to the causes and effects of such spurious pattern finding.
The gambler's fallacy and its (nameless) converse offer another example. The purchase of earthquake insurance increases sharply after an earthquake but then falls steadily as memory fades. As do gamblers who make the mistake of believing that red is more likely to come up again if it has come up several times in a row, the purchasers form their beliefs by using the availability heuristic. Their judgment about the likelihood of an event is shaped by the ease with which it can be brought to mind, and recent events are more readily available than earlier ones. The decay of emotion over time (Chapter 8) might also be a factor. Conversely, people living in areas that are subject to frequent floods often believe that a flood is less likely to occur in year n + 1 if one has occurred in year n. As do gamblers who make the mistake of believing that red is less likely to come up again if it has come up several times in a row, they form their beliefs by relying on the representativeness heuristic. They believe, or act as if they believe, that a short sequence of events is likely to be representative of a longer sequence in which it is embedded.
People often fail to grasp the relation between random processes and the distribution of outcomes. During World War II, many Londoners were certain that the Germans systematically concentrated their bombing in certain parts of their city, because the bombs fell in clusters. They did not understand the basic statistical principle that random processes tend to generate clustering, and that bombs falling in a neat gridlock pattern would have been stronger evidence of deliberate target selection. A fact that never fails to surprise those who have not come across it before is that in a group of as few as twenty-three people, the probability that two of them have the same birthday (day and month) is more than 50 percent. Out of a thousand investment firms, thirty are statistically likely to offer good advice on five successive occasions even if they basically perform the equivalent of tossing a coin. Clients who invest with these firms on the basis of their past performance, wrongly understood as the result of skill rather than luck, may experience that simple induction can lead one astray.
Magical thinking
Consider next various forms of magical thinking, that is, the tendency to believe one can have a causal influence on outcomes that are actually outside one's control. Many readers, afraid of tempting fate, will have had the thought, “If I don't take my umbrella, it's sure to rain.” People will also place larger bets on a coin that has not yet been tossed than on a coin that has already been tossed and for which the outcome has been concealed. In Proust, the Narrator's friend Robert Saint-Loup was subject to “a sort of superstitious belief: that the fidelity of his mistress to him might depend on his to her.” (It is clear from the context that he was not referring to a causal influence.) Also, people may fail to grasp the distinction between causal and diagnostic relevance. In one experiment, subjects who were led to believe that the length of time they could hold their arms in painfully cold water was the best indicator of longevity held their arms in the water longer than those not given this (false) information.8 Also, using their own behavior as a predictor of how others will act, people may choose the cooperative strategy in a Prisoner's Dilemma as if they could somehow bring it about that others cooperate too. In one experiment, cooperating subjects who were asked to predict the choice of their interaction partner as well as that of a non-partner who was matched with another person were more likely to predict (and had greater confidence in their prediction) cooperation by their interaction partner than the non-partner.9 Public authorities sometimes seem to count on the susceptibility of citizens to magical thinking. Thus in Paris buses one finds a sign saying: “Qui salit le siège à l'aller risque de se tâcher au retour” (if you dirty the seat going out, you risk getting stained coming back).
Calvinism offers an example of this kind of magical thinking (Chapter 3). Given the Calvinist belief in predestination, there would seem to be no reason for a Calvinist not to indulge in all sorts of worldly pleasures, which by assumption cannot affect their fate after death. Max Weber claimed that Calvinism nevertheless made its followers adopt an ascetic lifestyle, not to gain salvation but to acquire the subjective certainty of being among the elect. We may read him as saying that the Calvinists confused the causal and diagnostic relevance of their behavior. This is made quite explicit in a letter circulated by English Baptists in 1770: “Every soul that comes to Christ to be saved … is to be encouraged … The coming soul need not fear that he is not elected, for none but such would be willing to come.” We may think of this as retroactively forcing the hand of God, as opposed to prospective forcing by good works.
These errors (and many others that have been extensively documented) are for the most part “cold” or unmotivated mistakes, similar in some respects to optical illusions. Other errors, or “hot” mistakes, arise because the beliefs of the agents are motivated, that is, unduly influenced by their desires. As we shall see in Chapter 13, a causal influence of desires on beliefs is not intrinsically irrational. A desire can provide a reason for investing a specific amount of resources in information acquisition. The information thus obtained may serve as a reason for holding a certain belief. Although the desire does not provide a reason for holding the belief, it enters into a rational complex of belief formation. What drives in the wedge between the initial desire and the final belief is the fact that the outcome of the search for information is, by definition, not known at the time the decision to search is made.
Motivated belief formation
The direct influence of desires on beliefs I just cited is, uncontroversially, consistent with rationality. A more controversial idea was provided by Pascal's wager. As I explained in the last chapter, Pascal argued that an agent who believes that there is a non-zero probability, however small, that God exists, should for the purely instrumental reason of maximizing expected value try to acquire a firm belief (in the mode of certainty) that God exists because, if he does exist, that belief will ensure eternal bliss. The premises for the argument are (1) that certain belief is certain to provide salvation and (2) that the instrumental origin of the belief does not detract from its efficacy for salvation. Although both premises may be dubious from a theological point of view, especially as Pascal also believed in predestination, this need not concern us here. The question is whether this “decision to believe” is a rational project. In one sense it is not: I cannot decide to believe at will the way I can decide to raise my arm at will. One might, however, use an indirect strategy. By acting as if one believed, Pascal argued, one will end up believing. The mechanism by which this might happen is, however, somewhat unfathomable.
There are other cases in which one might want to acquire a belief one believes to be false, because of the good consequences of holding it. If I want to cut down on my drinking but find myself insufficiently motivated by the risk of becoming an alcoholic, I may desire to believe that the risk is larger than I now believe it to be. By and large, however, there is no reliable technology for acquiring such beliefs. Unless the process has a self-erasing component, by which the origin of the belief in the desire to acquire it is eliminated from the conscious mind, the desire is likely to remain a mere wish.
In the “uncontroversial” case, the agent's desire induces a certain level of information gathering that will in turn induce some belief or other. In the “controversial case” the desire induces specific behavior that will in turn induce a specific belief the agent wants to hold. Both are indirect strategies. I now turn to beliefs that are directly shaped by motivation. This can come about in one of two ways, corresponding to two basic features of motivations: arousal and content. Just as we say that the stone broke the ice by virtue of its weight, not of its color, we may say that a motivation affects belief not by virtue of its content, but by virtue of the accompanying arousal level. Moderate physiological arousal can improve the quality of belief formation, by focusing attention and stimulating the imagination. “When a man knows he is to be hanged in a fortnight,” Dr. Johnson said, “it concentrates his mind wonderfully.” Beyond a certain level of arousal, however, cognition deteriorates. In states of extreme hunger, stress, fear, or addictive craving, it is hard to think straight because the arousal makes it difficult to keep previous reasoning steps in mind. Presumably, mental concentration is blunted when the hanging is but one day away. In scholastic aptitude tests a very strong motivation to get it right may actually cause one to get it wrong, just as a shooter's strong desire to hit the target may cause her hands to shake so that she misses (see Chapter 10). In the next chapter I argue that because of the urgency of many emotions, they may cause the agent to bypass the normal machinery of rational belief formation. Thus beliefs may be shaped by motivation yet not be motivated, because the agent has no particular desire to believe they are true. Arousal clouds the mind but does not bias it in favor of any particular belief.
Motivated beliefs are of two main varieties. As I noted earlier, the agent may be motivated to hold some belief or other on a given topic, because of a need for closure or an intolerance of admitting ignorance. Alternatively, he may be motivated to hold some specific belief, such as the belief that his spouse is being faithful to him. The most important mechanisms generating this variety are rationalization, wishful thinking, and self-deception.
Rationalization
Rationalization can provide the agent with a belief that justifies her mistakes or serves as a reason for doing what she would want to do anyway. The second mechanism is similar to transmutation (Chapter 9), and might also have been discussed under that heading.
Rewriting the past in order to absolve oneself (or one's affiliates) from blame is extremely common. In divorce proceedings, the two spouses often try to shift the blame for the breakdown of the marriage on the other. They may produce motivated and incompatible narratives, for instance by claiming percentages of time spent on housework or with children whose sum exceeds 100 percent. At the other end of the spectrum of importance, citizens of Germany, France, England, Austria, Serbia, and Russia tend to blame another country than their own for the outbreak of World War I.10 A common mechanism is what one might call “reverse hindsight bias.” When a risky choice fails, observers often claim that the failure was foreseeable (the choice had negative expected value). Even when the claim is inaccurate, it may be supported by hindsight bias. The decision maker might claim that it “looked like a good idea at the time” (the choice had positive expected value). Even when that claim is inaccurate, it may be supported by rationalization. Thus when it became clear that the critics of the Vietnam War had been right from the beginning, defenders such as Walt Rostow rationalized it by redefining the goal as “buying time” or creating “breathing room” for other countries in the region.
A very common form of rationalization is framing self-interested behavior as disinterested. In Jane Austen's Persuasion, we find an exchange between Sir Walter Elliot, one of the most finely drawn egotists in fiction, and an unnamed female interlocutor. He asks how his daughter Mary is doing, adding that “the last time I saw her she had a red nose, but I hope that may not happen everyday.” Upon being reassured that his daughter was in very good health and very good looks, he responds that “If I thought it would not tempt her to go out in the sharp winds, and grow coarse, I would send her a new hat and pelisse.” The heroine of the novel, his daughter Anne, considered “whether she should venture to suggest that a gown, or a cap, would not be liable to any such misuse.” To rationalize his reluctance to spend money on his daughter, Sir Walter limits his options to gifts that he can reject as not being in her objective interest. Similarly, stingy parents may rationalize their refusal to help their children out financially by claiming that it will detract from their incentive to work, and rich nations may use the same argument to justify their refusal to aid poor countries. The fact that these arguments are in the interest of those who make them does not, of course, prove that they are wrong or are motivated by self-interest. To establish such claims, more evidence would be needed.
Wishful thinking
Let me turn to wishful thinking and self-deception. These two ill understood phenomena have in common that a desire that p be the case causes the belief that p is the case. In wishful thinking this is a simple one-step process: the wish is the father of the thought. The evidence is not so much denied as ignored. As a result, the wishfully formed belief might happen to be the very same one that would be justified by the evidence, had it been consulted.11 Self-deception as usually conceived involves four steps: first, the evidence is considered; second, the appropriate belief is formed; third, this belief is rejected or suppressed because it is inconsistent with our desire; and last, the desire causes another and more acceptable belief to be formed in its place. Self-deception is a paradoxical phenomenon, whose existence and even possibility have been called into doubt, so let me begin with the simpler issue of wishful thinking.
Before suggesting a mechanism by which wishful thinking is brought about, let me first state that, unlike what is the case for self-deception, it is impossible to deny its existence. One may deny that it occurs in high-stake situations or that it affects aggregate behavior such as stock markets or elections, but not that it occurs. If nothing else, world literature would testify to its existence. Moreover, many wishfully formed beliefs serve as premises for action, and hence are more than mere “quasi-beliefs.” Some smokers who fool themselves into believing that smoking is not dangerous, in general or for them specifically, would have quit or tried to quit had they held more rational beliefs.12 Overconfident individuals, who wishfully believe they are more capable than they really are, may embark on ventures they would otherwise have avoided. People who fool themselves into thinking they are as successful as others may lose a spur to improve themselves. A common mechanism is the following. First, a person is motivated to believe he is successful. Second, he finds some areas in his life in which he does in fact do well. Third, he enhances the importance of those areas to be able to tell himself that he is successful overall. Finally, he relaxes his efforts to succeed in other walks of life.
To navigate in life, it is instrumentally useful to have accurate beliefs. At the same time, beliefs may be intrinsically pleasant or unpleasant, that is, cause positive or negative emotions. If told that I have cancer, I can seek treatment, but the belief will also make me feel horrible. In Freud's language, those governed by the reality principle seek accurate beliefs, whereas those subject to the pleasure principle seek pleasant beliefs. This distinction applies only to beliefs in the strict sense, not to quasi-beliefs. People who form unrealistic beliefs about receiving a big monetary prize for their achievements yet do not spend the prize money before they have received it are at worst subject to a harmless form of the pleasure principle. In the more noxious variety, their conviction that they will receive the prize actually causes them to go into debt. The institutions of tontines and rentes viagères, in which one person receives a sum of money or a property upon the death of an unrelated person, probably owe some of their popularity to wishful thinking (and their notoriety to the crime novels in which they are part of the plot).
Wishful thinking cannot produce just any kind of pleasant belief, as it is somewhat subject to constraints. An agent who begins smoking may be tempted to form the wishful belief that smoking is not dangerous, or at least not dangerous for her. In doing so, however, she may be constrained by her prior beliefs about the dangers of smoking. The first time a person does badly on an exam, he may tell himself a story about bad luck, but if the same outcome occurs on the next four occasions the story is less likely to work if he fails for a sixth time. Or consider the example of the expensive Broadway show tickets that I introduced in Chapter 1. If I have paid $75 for the ticket but the show is lousy, my recollection of what I paid is likely to be too vivid to be subject to wishful downward revision. Given the intangible and multidimensional nature of aesthetic appreciation, it is easier to adjust my evaluation of the show upward. Similarly, although there is evidence both that likely events are seen as more desirable and that desirable events are perceived as more likely, the latter effect is more heavily constrained than the former.
In an instructive experiment, subjects expected to participate in a history trivia game with a given person either as their partner or as their opponent. After exposure to a sample of the person's performance, in which he got a perfect score, those who expected the person to be their partner (and therefore wished him to have high ability) judged him as better at history than those who expected him to be their opponent (and who therefore wished him to have low ability). At the same time, subjects were clearly constrained by the nature of the information they received, since even subjects expecting him to be their opponent judged him as better than average. A limitation of the experiment is that it did not offer the subjects an opportunity to act on these beliefs, with the potentially costly consequences that might follow from underestimating an opponent. For all we know, they might be mere quasi-beliefs. If they had been playing for money, the subjects might have been more circumspect.
In the examples just given, wishful thinking is constrained by prior factual beliefs. In other cases, it may be constrained by plausible causal beliefs. Wishful thinking often involves “telling oneself a story,” the idea of a story being closely related to the idea of a mechanism that I discussed in Chapter 2. The plethora of mechanisms makes it easier to find some story or other that will justify any belief one might want to be true. I may dismiss an unwelcome rumor with the proverb “Rumors often lie” and embrace a welcome one by the proverb “Rumors rarely lie.” Or suppose I read in the application material for a school of social work that emotional stability is highly desirable for people in that profession. If my mother left the workforce to take care of me when I was born, I may bolster my belief in my stability by telling myself a story that children benefit from the full-time attention of their parents. If she kept her job and sent me to day care, I may instead adopt a story that children benefit from being with other children and from having parents who have professional fulfillment outside the home.13 If my favorite soccer team does badly, I can maintain my belief in its superiority if the other team won by (what can be construed as) a fluke event. “If the ball hadn't been deflected by the referee, the wing player would have received the pass in a position to score.” If my horse finishes second, I can maintain my belief in my betting skills by saying that it “almost won.” In an even more blatantly irrational piece of wishful thinking, if I put money on 32 and 33 comes up, I can also say that I “almost won” even if the two numbers are far from each other in the roulette wheel.14
Sometimes, however, there is no readily available and plausible story. Suppose a person places his money on 24. The number that did come out was 15, which is adjacent to 24 on the number wheel; hence his belief in his gambling skills is confirmed. Probably he would have considered other outcomes, such as 5, 10, and 33, also confirmations, because they are nearby on the wheel. Also he could have taken the outcomes 22, 23, 25, and 26 as confirmations because their numerical value is closer, or the numbers 20, 21, 26, and 27 because they are adjacent on the table. Thus 13 out of 37 possible outcomes could be taken as confirmations of his betting ability. But that also means that there are 24 outcomes for which no simple story is available. If one of these occurs, even a person prone to wishful thinking and highly motivated to adopt a specific belief might have to face the facts.
Self-deception
Consider now the thorny issue of self-deception. The canonical definition is:
1.The individual holds two contradictory beliefs (that p and that not-p).
2.These two contradictory beliefs are held simultaneously.
3.The individual is not aware of holding one of the beliefs.
4.The act that determines which belief is and which belief is not subject to awareness is a motivated act.
The simultaneity condition is similar to condition (3) in the strict definition of weakness of will (Chapter 6). If these conditions are relaxed, both self-deception and weakness of will lose their paradoxical character. In Chapter 6, I argued that if we adopt a broader definition of weakness of will, actions that appear to go against the better judgment of the agent may simply be due to a preference reversal. In the case of self-deception, the paradox would disappear if the motivated adoption of the belief that not-p went together with the erasure from the mind of the belief that p rather than its relegation to the unconscious. Although this situation may be frequent, I shall focus on the full-blown paradoxical case. Standard examples include the denial that one's spouse is having an affair, that one is exhibiting the symptoms of cancer, that one is gaining weight, that one is drinking more than x glasses of wine per day, and so on.
Before I proceed, let me comment on the ambiguous role of the desires (the motivation) in self-deception. Does the suspicious husband desire that his wife be faithful, or desire to believe that she is? He might in fact want her to be unfaithful, to justify his belief, but also want his belief to be false, to save his marriage. Although this example may be contrived, the following is not. A person who predicts that something he wants not to happen is nevertheless going to happen (a war, an earthquake) might be reluctant to give up his belief even when disconfirming evidence comes to light, because his amour-propre would suffer, but at the same time might welcome evidence that disaster will not strike. In other words, people make a psychic investment in their beliefs that, independently of their content, makes them reluctant to face the facts. As La Rochefoucauld said, “Nothing blights our self-esteem so much as to disapprove of what we once approved.” Here I ignore this complication, to which I return in Chapter 9.
To my knowledge, experimental psychologists have not tried to verify the existence of weakness of will, in the strict sense, in the laboratory. By contrast, they have tried to demonstrate the existence of self-deception, in the strict synchronic sense. I shall discuss four such studies, and conclude by discussing some literary treatments of self-deception.
The authors who proposed the “canonical definition” of self-deception cited earlier also offer a putative example of the phenomenon. Their experiments turned on the belief of subjects as to whether a voice recording was of their own speech or of another person's. Subjects with a poor self-image tended to attribute their voice to another person, just as some people avoid looking at themselves in a mirror; at the same time, their galvanic skin response was stronger when they heard their own voice than when they heard that of another person. These facts suggest that conditions (1) and (3) of the definition are satisfied. Condition (2), simultaneity, was ensured by the experimental set-up. The satisfaction of condition (4), the motivated character of the non-awareness that they were hearing themselves, was assured by the finding that subjects who tended to ascribe the voices of others to themselves scored highly on a narcissism score. Although the experiments were more complex and subtle than I have indicated, my simplifications do not affect the objection that the experiment did not prove that the unconscious response had a propositional basis. One cannot exclude that the subjects reacted with disgust upon hearing their own voice, without formulating to themselves the proposition “This is my voice.” Disgust is, in fact, one of the few emotions that do not require a propositional trigger (see Chapter 8).
The same comment applies to another experiment that was not designed to prove the reality of self-deception, but has been used as an argument for its existence. As background, let me cite a remark put in the mouth of Roy Cohn, the legal adviser of Joseph McCarthy, in the play Angels in America: “Roy Cohn is not a homosexual. Roy Cohn is a heterosexual man, who fucks around with guys.” This combination of homosexuality and homophobia suggests self-deception. In the experiment, homophobic and non-homophobic men were exposed to homosexual stimuli. The former were significantly more aroused, as measured by changes in the circumference of the penis. Their self-reports of arousal thus measured were low, however, and not different from those of non-homophobic subjects. Assuming that the self-reports were sincere, these findings might seem to provide evidence for self-deception: the conscious belief of the homophobic subjects that they were not turned on by homosexual material was inconsistent with their bodily responses. Yet once again, the findings do not provide evidence of an unconscious propositional belief about their homosexual tendencies, since the cause of their denial may have been disgust at their arousal.
I have already referred to the third experiment, in which some subjects were told that the time they could tolerate holding their hand in very cold water was an indicator of a heart condition predicting their life expectancy. They tended to hold their hands in the water longer than subjects who were not given this (false) information. These facts, together with self-reports, suggest the presence of self-deception. On the one hand, the subjects could hardly hold the conscious belief that they could modify the cause by acting on the symptoms. On the other hand, their behavior seemed to reveal that they unconsciously did hold that magical belief. Also, they were clearly motivated to keep the belief unconscious, since otherwise they could not, as most of them did, form and report the gratifying belief that their heart condition and life expectancy were indeed good. In this experiment, then, the proposed evidence for a repressed belief was behavior, not, as in the previously discussed experiments, a somatic response. Moreover, the inference from the behavior to the belief seems more reliable than the inferences from the somatic responses to the beliefs.
The fourth set of experiments, on patients suffering from anosognosia, in this case paralyzed patients who believed that they are not paralyzed, provide the most compelling scientific evidence for the existence of self-deception. Specifically, the patients suffer from paralysis of the left side of the body, notably the arm and the hand, caused by a stroke in the right brain hemisphere. One the one hand, these patients firmly believe that they are not paralyzed, as shown not only by what they say, but also by what they do. Thus when given the choice between a well-remunerated task that requires the use of both hands (tying their shoe laces) and a less well-remunerated task that can be performed with one hand (screwing in a light bulb), they invariably choose the former. If, following a failure, they try again ten minutes later, they make the same choice and claim to have succeeded on the first try.
On the other hand, the patients know that they are in fact paralyzed, as shown by several observations. For ill-understood reasons, the anosognosia disappears when the left ear of a patient is irrigated with cold water. In this state, the patient affirms that he is paralyzed and has been so for several days. When the effect of the irrigation wears off and the patient is reminded of this statement, he claims that he had stated having the use of both hands. The patient seems to remember the fruitless effort to his shoelaces, but the access to this memory is blocked until it is unblocked by the cold water. Although these facts do not show that the blocking is motivated, other findings point in that direction. Thus a woman who had tried in vain to tie her shoelaces later affirmed that she had done so “with both hands,” a detail that a normal person would omit (“the lady doth protest too much, methinks”). Other patients come up with farfetched explanations for their failures to respond to requests to move their left hand. Finally, when the experimenter injects a saline solution in the paralyzed arm and tells the patient, untruthfully, that it is an anesthetic that will paralyze his arm for a few minutes, he answers “No” to the question whether he can move his arm, presumably because a temporary paralysis is much less threatening than a permanent one. Whether these findings can illuminate self-deception in non-damaged persons remains an open question.
The experimental studies illuminate sharply, perhaps too sharply. Real self-deception, outside the laboratory, is a matter of clair–obscur rather than of black and white; of half-believing or averting one's gaze rather than of full awareness contrasted with total unawareness. By the nature of the case, a good novelist is better equipped than the experimenter in capturing these fluid or half-crystallized states. I shall briefly discuss a passage from Stendhal, and then at greater length some analyses from À la recherche du temps perdu.
In Stendhal's unfinished novel Lucien Leuwen, we encounter the phenomenon of self-deceptive ignorance of love. In the early part of the novel, Lucien and a young widow in the garrison town where he is serving, Mme de Chasteller, have come to love each other deeply, yet are uncertain about each other. She fears that he may be no more than a rake, he that she does not really love him. Whenever he makes a clumsy and tentative advance, she sees it as a reason for doubt about his character; she grows haughty, he is made desperate, and, through his desperation, redeems himself for a while. Gradually they grow closer to each other. Lucien writes her a letter; she, after some soul-searching replies in what she believes to be a severe and uncommunicative tone. In an authorial aside, Stendhal comments as follows: “What would be the point of noting that her reply involved a studied attempt at the haughtiest turns of phrase? Three or four times Leuwen was urged to abandon all hope, the very word hope was avoided with an infinite adroitness that made Mme de Chasteller very pleased with herself. Alas, without knowing it she was the victim of her Jesuitical education; she deceived herself, in applying badly, and unawares, the art of deceiving others which she had been taught at the Sacré-Coeur. She answered: everything lay in this word, which she preferred to ignore.”
Rather than saying that Mme de Chasteller knew, at an unconscious level, that the discouraging letter would be read as an encouragement, Stendhal seems to suggest that she preferred not to explore the implications of her act. I now discuss some passages where Proust makes a similar suggestion. Because of the conceptual richness and psychological acuity of his analyses, I shall reproduce them at some length.
The question I shall discuss is whether Swann, a friend of the Narrator's family, deceived himself when he thought that his mistress Odette was faithful to him. The stage is set in a passage where Swann reflects on the reasons Odette might have for staying with him, apart from an uncertain and fragile personal attraction:
For the moment, by lavishing presents upon her and performing all manner of services, he could rely on advantages not contained in his person or in his intellect, and forego the exhausting effort to please by himself. And the price he paid … for this delight in being in love, in living by love alone, a delight in whose reality he sometimes doubted, enhanced its value in his eyes – as one sees people who are doubtful whether the sight of the sea and the sound of its waves are really enjoyable, become convinced that they are, as also of the rare quality of their own disinterested taste, when they have agreed to pay a hundred francs a day for a room in an hotel, from which that sight and that sound may be enjoyed.
Pursuing this train of thought, Swann reverses the argument:
One day, when reflections of this order had brought him once again to the memory of the time when someone had spoken to him of Odette as of a “kept” woman, and when, once again, he had amused himself with contrasting that strange personification, the “kept” woman … with that Odette upon whose face he had watched the passage of the same expressions of pity for a sufferer, indignation of an act of injustice, gratitude for an act of kindness, which he had seen, in earlier days, on his own mother's face, and on the faces of friends; that Odette, whose conversation had so frequently turned on the things that he himself knew better than anyone, his collections, his room, his old servant, his banker, who kept all his title-deeds and bonds; – the thought of the banker reminded him that he must call on him shortly, to draw some money. And indeed, if, during the current month, he were to come less liberally to the aid of Odette in her financial difficulties than in the month before, when he had given her five thousand francs, if he refrained from offering her a diamond necklace for which she longed, he would be allowing her admiration for his generosity to decline, that gratitude which had made him so happy, and would even be running the risk of her imagining that his love for her (as she saw its visible manifestations grow fewer) had itself diminished.15
Then, self-deception makes an appearance:
And then, suddenly, he asked himself whether that was not precisely what was implied by “keeping” a woman (as if, in fact, that idea of “keeping” could be derived from elements not at all mysterious nor perverse, but belonging to the intimate routine of his daily life, such as that thousand-franc note, a familiar and domestic object, torn in places and mended with gummed paper, which his valet, after paying the household accounts and the rent, had locked up in a drawer in the old writing-desk whence he had extracted it to send it, with four others, to Odette) and whether it was not possible to apply to Odette, since he had known her (for he never imagined for a moment that she could ever have taken a penny from anyone else, before), that title, which he had believed so wholly inapplicable to her, of “kept” woman. He could not explore the idea further, for a sudden access of that mental lethargy which was, with him, congenital, intermittent and providential, happened, at that moment, to extinguish every particle of light in his brain, as instantaneously as, at a later period, when electric lighting had been everywhere installed, it became possible, merely by fingering a switch, to cut off all the supply of light from a house. His mind fumbled, for a moment, in the darkness, he took off his spectacles, wiped the glasses, passed his hands over his eyes, but saw no light until he found himself face to face with a wholly different idea, the realization that he must endeavor, in the coming month, to send Odette six or seven thousand-franc notes instead of five, simply as a surprise for her and to give her pleasure.
In a sense, Swann “believes” that Odette is a kept woman, and in a sense he is motivated to suppress that belief.16 It would be misleading to say, however, that the belief is relegated to the unconscious. Like the sun, the belief is too hard to look in the face. As a French historian said about his adherence to the Communist Party in the 1950s, he and his friends “knew what the Soviet Union was like; so, to remain Communists, we made an effort not to think about it” (his italics).
Motivated framing
These are examples of motivated blindness. I conclude by considering what one might call motivated framing. People may be opposed to an action or a policy if it is framed in one way, but accept it if it is brought under a different heading. The classical discussion of the issue occurs in Pascal's Provinciales, perhaps the greatest satirical work ever written and a blow from which its target, the Jesuit order, never fully recovered. The focus of his criticism is precisely the Jesuitical idea of making forbidden actions appear as licit by directing one's intention in the appropriate way. The Jesuits claimed, for instance, that even though the Bible prohibits revenge, “a military man may demand satisfaction on the spot from the person who has injured him – not, indeed, with the intention of rendering evil for evil, but with that of preserving his honor.” In other words, when A suffers an injury at the hands of B, A may licitly inflict an injury on B by directing his intention to the fact that he would be blamed by C, D, E … if he failed to retaliate.17 In this case as well as in the other examples Pascal considers, the framing is done by the Jesuit confessor, not by the agent himself. I now consider cases in which the agents themselves were responsible for the framing.
In the making of the American constitution of 1789 and the French constitution of 1791, the issue of slavery was prominent in several ways. Yet in both constituent assemblies, the framers took great care to avoid the words slave or slavery in the constitution they adopted. In America, the framers were simply hypocritical. In their internal deliberations at the Federal Convention, they did not shy away from using the terms, which occur more than a hundred times in Madison's highly compressed notes. Yet to accommodate Northern and perhaps international audiences, they substituted euphemisms such as “other persons” in the text of the document. In France, Robespierre's near-hysterical insistence on using the term “unfree persons” about the slaves in Santo Domingo suggests that, for him, the framing was all-important. He was willing, he said, to let the colonies perish rather than having the word “slave” appear in the law, since, if it did, the declaration of the rights of man and of liberty would be undermined.
These two cases, especially the first, are closer to deception than to self-deception. A case closer to the latter end of the continuum is provided by the way the Soviet elite under Stalin framed their privileges. A scholar of the period notes that since according to Marxist doctrine class societies were based on property, “the fact that the amenities of life – car, apartment, dacha – were not owned but were state issue was very important in enabling Communists of the nomenklatura to see themselves as something different from a new nobility or ruling class. On the contrary, they were people who owned nothing! … It was comparatively easy for elite members to see themselves as indifferent to material things when there was no personal property at stake.”
As a further example, consider the framing of subsidies and taxes. The aluminum industry in western Norway demands and gets huge subsidies in the form of cheap energy, partly because the workers do not want wage subsidies. The government has tried to offer the fishermen in northern Norway direct labor subsidies, only to be met with the response that they prefer subsidies to be given to the shipowners. In both cases, observers emphasize that accepting wage subsidies is perceived to be like begging. Workers in the textile industry, where direct wage subsidies are given, envy the aluminum industry its energy requirements, because these justify less transparent income transfers. For similar reasons, farmers in western Germany favor a subsidy through price over a direct subsidy of income although this involves astronomical dead-weight losses. Somehow, what would be rejected as a one-step operation becomes acceptable as a two-step operation. Similarly, pacifists may be willing to pay taxes that are used for military purposes. In the War of 1812, the American “government exempted religious pacifists – Quakers, Mennonites, and Dunkers – from the militia, but they had to pay annual fines of five pounds per man and had to provide draft animals, wagons, cars and sleighs on military demand. The Mennonites and Dunkers accepted that compromise, but the Quakers balked at contributing anything that promoted bloodshed.”
As the examples of the two previous paragraphs show, social agents can be motivated to frame their situation in a way that is consistent with their preferred self-image as non-exploitative, non-parasitical or non-murderous. As in many other cases of self-deception, the alternative frame is not relegated to the unconscious, but simply not activated or explored. For a more adequate analysis, one would need a Proust.
Bibliographical note
Evidence for many of the findings reported here can be found in the following source books: D. Kahneman, P. Slovic, and A. Tversky (eds.), Judgment Under Uncertainty (Cambridge University Press, 1982); D. Bell, H. Raiffa, and A. Tversky (eds.), Decision Making (Cambridge University Press, 1988); T. Connolly, H. Arkes, and K. R. Hammond (eds.), Judgment and Decision Making (Cambridge University Press, 2000); D. Kahneman and A. Tversky (eds.), Choices, Values, and Frames (Cambridge University Press, 2000); T. Gilovich, D. Griffin, and D. Kahneman (eds.), Heuristics and Biases: The Psychology of Intuitive Judgment (Cambridge University Press, 2002); C. Camerer, G. Loewenstein, and M. Rabin (eds.), Advances in Behavioral Economics (New York: Russell Sage, 2004); I. Brocas and J. Carillo (eds.), The Psychology of Economic Decisions, vols. I and II (Oxford University Press, 2003, 2004). The quote on the veil of ignorance in Rome is from H. Scullard, From the Gracchi to Nero (London: Routledge, 1982), p. 35. The comments on the “Texas sharpshooter effect” and spurious cancer statistics are from G. Johnson, The Cancer Chronicles (New York: Knopf, 2013), and A. Gawande, “The cancer-cluster myth,” The New Yorker, February 8, 1999. The observation about the Londoners’ perception of bombing patterns is from W. Feller, An Introduction to Probability Theory and its Applications (New York: Wiley, 1968), p. 160, and the comment on investors who rely on past performance of firms is from N. Taleb, Fooled by Randomness (New York: Random House, 2005). A study of magical thinking in actions that “tempt fate” is A. Arad, “Avoiding greedy behavior in situations of uncertainty: the role of magical thinking,” Journal of Behavioral and Experimental Economics 53 (2014), 17–23. The cold-water study is in G. Quattrone and A. Tversky, “Causal versus diagnostic contingencies: on self-deception and the voter's illusion,” Journal of Personality and Social Psychology 46 (1984), 237–48. The greater tendency to impute cooperation to partners than to non-partners is documented in L. Messé and J. Sivacek,” Predictions of others’ responses in a mixed-motive game: self-justification or false consensus?” Journal of Personality and Social Psychology 37 (1979), 602–7. The double incompetence of the ignorant is documented in J. Kruger and D. Dunning, “Unskilled and unaware of it,” Journal of Personality and Social Psychology 77 (1999), 1121–34. A sophisticated study of the unreliability of the judgments of some experts (“hedgehogs”) and the somewhat more reliable judgments of others (“foxes”) is P. Tetlock, Expert Political Judgment (Princeton University Press, 2005). For Abraham Wald's insight, see J. Ellenberg, How Not to be Wrong (New York: Penguin, 2014), pp. 5–7. The example of the man who was puzzled about the railway company's knowledge about his location is lifted from D. Sand, The Improbability Principle (New York: Scientific American, 2014), p. 122. For the data about earthquakes and floods, see P. Slovic, The Perception of Risk (Sterling, VA: Earthscan, 2000). For the (il)logic of conspiracy theories, see B. Keeley, “Of conspiracy theories,” Journal of Philosophy 96 (1999), 109–26. On theories of famine see S. Kaplan, “The famine plot persuasion in eighteenth-century France,” Transactions of the American Philosophical Society 72 (1982), and F. Ploux, De bouche à oreille: naissance et propagation des rumeurs dans la France du XIXe siècle (Paris: Aubier, 2003). A study of conspiratorial thinking is R. Hofstadter, The Paranoid Style in American Politics (Cambridge, MA: Harvard University Press, 1964). The study of the constraints on motivated reasoning is W. Klein and Z. Kunda, “Motivated person perception: Constructing justifications for desired beliefs,” Journal of Experimental Social Psychology 28 (1992), 145–68. The best overview of self-deception is a special issue of Behavioral and Brain Sciences 20 (1997), organized around an article by A. Mele, “Real self-deception.” The “canonical definition” is that of R. Gur and H. Sackeim, “Self-deception: A concept in search of a phenomenon,” Journal of Personality and Social Psychology 37 (1979), 147–69. The study of homophobia and homosexuality is H. Adams, L. Wright, and B. Lohr, “Is homophobia associated with homosexual arousal?” Journal of Abnormal Psychology 105 (1996), 440–5. The findings about anosognosia are from V. S. Ramachandran and S. Blakelee, Phantoms in the Brain (New York: William Morrow, 1999). I discuss Proust's analyses of self-deception at greater length in Chapter 15 of L'irrationalité (Paris: Seuil, 2010). The quote from the French historian is taken from P. Veyne, Et dans l’éternité je ne m'ennuierai pas (Paris: Albin Michel, 2014), p. 93. For the verbal denials of slavery in the American and French constitutions, see my “Throwing a veil over inequality,” in C. Sypnovich (ed.), The Egalitarian Conscience: Essays in Honour of G. A. Cohen (Oxford University Press, 2006). On the framing of privilege by the Soviet elites, see S. Fitzpatrick, Everyday Stalinism (University of Chicago Press, 1999), pp. 104–5. For references to the framing of subsidies to Norwegian and German workers, see my Alchemies of the Mind (Cambridge University Press, 1999), pp. 252–3. The payment of taxes to finance the 1812 War is cited from A. Taylor, The Civil War of 1812 (New York: Knopf, 2010), p. 310.
1 After the fall of Communism, a woman from the former East Germany said at a public meeting that her generation had been raised from childhood on to conform, to stay in line. A long-term schizophrenia had hollowed them out as people. So, this woman said, now she could not just suddenly “speak openly” or “say what she thought.” She did not even really know precisely what she thought.
2 The state (or “veil”) of ignorance may be deliberately induced, to prevent agents from acting according to their self-interest. If the electoral law is included in the constitution, one might specify that any changes will take effect only after n years + 1, where n is the length of the electoral cycle. Elected officials might be required to put their financial assets in a blind trust. A rare use of a veil-of-ignorance procedure in politics occurred when, in the words of an historian, Gaius Gracchus “compelled the Senate, which decided to which provinces consuls should be sent (normally after their year of office in Rome), to fix the provinces before the consuls were elected instead of during their consulships. Thus the Senate could less easily reward its favourites with the best provinces.” He adds that “since the provinces would now have to be allocated eighteen months in advance, the new arrangement would not make for efficiency.”
3 In a deeper analysis, the first (objective) method boils down to the second (subjective) one, since to be useful objective data always need a subjective interpretation. For many practical purposes, though, the distinction is clear and useful.
4 Writing about Alan Greenspan (New York Times, October 28, 2005), Paul Krugman noted that while distrusting formal models he had “the ability to divine from fragmentary and sometimes contradictory data which way the economic wind was blowing.” In his more recent writings, subsequent to the financial crisis, Krugman has been more critical of Greenspan.
5 Knowing that one may be subject to bias is one thing; being able to correct it is another. Studies show that deliberate attempts to debias one's judgment are of little value, since one easily falls into the traps of insufficient correction, unnecessary correction, or overcorrection. One may learn to distrust one's judgment, but it is harder to improve it. If one were able to, there might be no need to.
6 This superiority remains even if we simply assign equal weights to all variables!
7 In the 2012 trial of the Norwegian mass murderer Anders Breivik, the first team of psychiatrists concluded that he was a “paranoid schizophrenic,” who could not be held legally responsible for his actions. Following a public outcry, the court appointed a second team, which concluded that Breivik was a right-wing fanatic who could be held responsible. In her summing-up, the judge agreed with the second team. Public confidence in forensic psychiatry was shattered.
8 This distinction between cause and symptom is not always evident. As late as 1959, the great statistician R. A. Fisher, assuming a genetic trait that predisposed the individual both to smoking and to cancer, argued that smoking was diagnostic of lung cancer rather than its cause. (It is true that he was in the pay of tobacco companies at the time.) Or consider the finding, discussed in Chapter 2, that the longer an individual has been out of work the less likely it is that he will find a job in a given time span. The duration of unemployment might be simply diagnostic of employability, or it could make a causal contribution (through demoralization, etc.) to the chances of finding employment.
9 This discrepancy allows us to exclude that the imputation of cooperative behavior to the interaction partner could have been due merely to the “false consensus effect” (Chapter 22).
10 The idea that no single state might be to blame is undermined by the need to find meaning and order in the universe, specifically by the agency bias (Chapter 9). Even if this bias is overcome, the tendency to absolve one's own country might persist.
11 Ignoring this point could be a source of irrational belief formation. Because it is often easy to detect the operation of motivated belief formation in others, we tend to disbelieve the conclusions reached in this way, without pausing to see whether the evidence might in fact justify them. Until around 1990 I believed, with most of my friends, that on a scale of evil from 0 to 10 (the worst), Communism scored around 7 or 8. Since the recent revelations I believe that 10 is the appropriate number. The reason for my misperception of the evidence was not an idealistic belief that Communism was a worthy ideal that had been betrayed by actual Communists. In that case, I would simply have been victim of wishful thinking or self-deception. Rather, I was misled by the hysterical character of those who claimed all along that Communism scored 10. My ignorance of their claims was not entirely irrational. On average, it makes sense to discount the claims of the manifestly hysterical. Yet even hysterics can be right, albeit for the wrong reasons. Because I sensed and still believe that many of these fierce anti-Communists would have said the same regardless of the evidence, I could not believe that what they said did in fact correspond to the evidence. I made the mistake of thinking of them as a clock that is always one hour late rather than as a broken clock that shows the right time twice a day. Later, I made the same mistake about members of the ecology movement.
12 As in the case of alcohol, quitting might require the irrational belief that smoking is more dangerous than it actually is.
13 As a matter of fact, no consistent differences are found in the later development of children brought up in these two environments.
14 At the same time, people may be more disappointed if their number is close to the winning number. Some national lotteries offer small “consolation prizes” to those who “almost won.”
15 Comparing the phrases I have italicized in these two passages, we note a complete inversion. First, Swann's generosity is explained by its magical capacity to increase Odette's value in his eyes. Next, Swann makes the more down-to-earth observation that the generosity might increase his value in her eyes.
16 The word “providential” is puzzling, since the extinction of Swann's mental lights is surely motivated rather than accidental. Other passages from the novel confirm this interpretation.
17 The claim that one may legitimately use a bad means (killing an adversary) to achieve a good end (defending one's honor) must be distinguished from the doctrine of double effect, according to which it is permissible to bring about as a foreseeable side effect of action what it would not be permissible to bring about as the intended result of action. As philosophers have noted, that doctrine also creates a potential for self-deception.