Introduction: Fixing Prices on the Cognitive Market
1 The cognitive market belongs to a group of social phenomena (to which the economic market also belongs), where individual interactions more or less blindly converge toward emerging and stable (without being commodified) forms of social life. This involves a market in which people exchange what could be termed cognitive products: hypotheses, beliefs, knowledge, and so on. Just as in economic phenomena, pure competition between cognitive products (requiring a series of criteria that are impossible to bring together: exhaustiveness of information, etc) does not exist, although the competition may be oligopolistic, or even monopolistic, some sort of law of supply may be in effect,  and so on.
2 Of course, the term market is not neutral. It has a long history that some claim began with Aristotle, although many consider Smith or Turgot to be at its origin.  My aim here is not to import this metaphor into the sociology of beliefs by searching for “term-by-term” correspondences to realities of the economic market. On the contrary, I had the opportunity to highlight some of the limitations of this metaphor, and I largely developed the possibility of this importation in L’empire des croyances. For this reason, and due to the space limitations of this article, I will remain allusive on this point. I will merely point out that the three factors determining the success of a cognitive product on the market are the mediator of the belief (who can also be its producer), the receiver (who becomes a believer and often, in turn, a mediator), and the product itself.
3 The characteristics of the mediator are fundamental. According to whether he is considered qualified or not,  whether he is known and reliable or not,  and finally whether it is seen to be in one’s best interests to believe him, or, on the contrary, one is simply uninterested, one will more or less readily agree “to consume” the product on offer—that is to say, to endorse the belief that he promotes.
4 The characteristics of the receiver are no fewer, since a belief is all the more easily endorsed when it marries with the curves defining the universal limits of individual understanding, which I consider to be three in number. Firstly, the individual is imprisoned in an eternal present and embodied in a restricted space, which forms the conditions for the dimensional limits of his rationality (time and space) and often strips him of his ability to be a knowing subject. Furthermore, the individual is never neutral vis-à-vis phenomena, because he interprets them according to what Schutz termed “a preorganized stock of knowledge,” thus defining the representational (or cultural) limits of rationality. Last, because his abilities of memorization, abstraction, deliberation, computation, and so on are limited, he may develop a vision of the world that is misrepresentative of the reality and so succumb to the cognitive limits of rationality.  The subject of the debate introduced in this article does not generally concern the characteristics of the mediator or receiver of the belief; the questions opposing the three paradigms, which I will present here, instead relate to the contents, form, and nature of the cognitive product itself, given their decisive influence on its successful dissemination in the market.
1. The Crypto-Functionalist Interpretation
5 The commonly defended idea concerning the emergence of beliefs and their success, be it in sociology, psychology, or anthropology, is associated with a more or less explicit functionalism. This crypto-functionalism underlies very different scientific leanings, which nevertheless share the assumption that the generation of a collective belief and its longevity owe much to the social function that it assumes.
6 These beliefs may serve as tools of domination and so assume a function of reproducing the distinctions between social categories (Marx, Bourdieu, etc.). Less precisely, they may endorse the responsibility of “speaking” or collectively remind people of social truths that they are unable to discern easily (the structuralist interpretation of myths, the categories of thought in Durkheim or Mauss, etc.). They may also serve as moralizing fables, revealing social interrogations, crises, and consensus (as well as the orthodox interpretation of the phenomena of rumors or urban legends, which try to account for them),  or they may allow the collective control of an environment (the usual stance of social psychology [Jodelet 1989; Moscovici 1976]). For all these authors, the reasons for the existence of these semantic objects come down to their social utility.
7 When the crypto-functionalist position purports to describe the selection process of beliefs by referring to a function that is allegedly targeted, it is, in spirit, related to that of Lamarckism in biology. It imparts to social function the role that Jean-Baptiste de Monet, Chevalier de Lamarck, attributed to a mysterious “vital force” contained in all life.
8 The same arguments that were previously used against the functionalism of Malinowski (1968) can immediately be called on against these functionalist theories of belief. If these theories were valid, then every product emerging on the cognitive market would have an identifiable social function. However, it is evident that many have none, and quite often, they are rather counterproductive when judged from the perspective of their social utility. Some self-fulfilling prophecies, for example, are downright harmful, but they owe their damaging effect to their own success.
9 Likewise, it may be assumed that any cognitive product with a potential social utility would be bound to emerge on the market, but this is not the case.
10 A more significant problem may be cited. This theory more or less explicitly supposes that the generation and success of a belief are subject to the teleological form of causality. Indeed, once fully developed, the belief assumes a social function, but something must initially give it this inclination. Here the end is the cause of the origin of the entire process. Hence, if the teleological form of causality is willingly used in order to clarify the actions and decisions of individuals who are gifted in reason, then its utilization in this case is somewhat awkward. If this crypto-functionalist hypothesis is maintained, it should be supposed that from their genesis, beliefs have within them a force inclining them toward whatever they must become, if their destiny is to assume a social function. This idea strongly resembles Lamarck’s concept regarding the vital principle, which, to recall the most well-known illustration, caused the giraffe’s neck to lengthen from one generation to another, thus serving a fundamental biological function: alimentation. In both cases, this idea carries a certain metaphysical load.
2. The Selectionist Interpretation
11 In this section, I distinguish two leanings around which the methodological debate—the reason behind this article—is organized. These two tendencies, though very different, may at least be used to show the unconvincing nature of crypto-functionalist explanations for beliefs. Both defend the notion that a selection process occurs when the observer notes the emerging form of an idea so that the less “adapted” ideas are eliminated. The dominant items may thus give the illusion that they are inclined toward a social function from the start. In the same way as crypto-functionalist explanations are related to Lamarckism, it can be said that they introduce the phenomena of belief selection, a Darwinian logic. Darwin saw the evolution of the species to be the consequence of a natural selection process, allowing the survival of the best-adapted individuals. In other words, individuals do not adapt to their environment biologically and within their own lifetime. If they survive, it is because, through the luck of genetic combinations, they are better adapted than others. Compared with finalism, Darwin proposed a much more satisfactory solution to the enigma of the morphological adaptation of living beings to their environment, because it was not based on any ad hoc hypothesis. Life forms are so well adapted to their environment that they may give the impression of having an ultimate cause, as the observer can only become aware of nature’s successes, without seeing the innumerable mass of its failures, as the theory of evolution explains. This logic is found in the interpretations that I described as selectionist because, rather than suppose that a mysterious social functionality governs the emergence, development, and dissemination of beliefs, they choose to conform to the principle of sufficient reason (and so charge the cannons of scientific reasoning). 
12 This selection occurs in the cognitive market. It can therefore be supposed that among the hypotheses put forward in any given enigmatic situation, only a handful will turn out to be of value.
13 A product appears and then disappears, simply because it ceases to be believed or the conditions of its dissemination go unmet. The diverse collective beliefs regarding crop circles provide an illustration. These large circles mysteriously appear, usually in cornfields. They may be simple circles or outlines of more complex figures. Nobody doubts the material existence of these phenomena. However, there are several concurrent interpretations. The most immediate explanation is that of a hoax, all the more so because these phenomena, mainly appearing in the south of England in the 1980s, could easily be reproduced artificially. Furthermore, in September 1991, two artists, Doug Bower and Dave Chorley, claimed to have been the creators of more than 200 crop circles since 1976. Around the world, groups known as circle makers have laid claim to the creation of these “works of art.”
14 Others maintain that these circles cannot all be hoaxes since some are too complex or too regular to be hand drawn in a single night. There is no lack of suggestions that try to explain this enigma.
15 The theory of landing strips for extraterrestrials is one of most common, being popularized in M. Night Shyamalan’s film Signs.
16 There exist other theories: that these crop circles, for example, were the work of marasmius oreades mushrooms, which left a toxic substance on the ground, creating circles that widened each year.
17 Some even claim the circles to be the result of prototype weapons testing, microwave laser cannons, and so on.
18 Another theory, advanced in 1989 by a meteorologist, George Terence Meaden, explained this phenomenon to be the result of plasma-vortex effects. His theory gained some success when it entered the cognitive market. It was based on a known physical phenomenon that could produce, in record time, these mysterious circles which were still puzzling the world. However, the theory’s success was short lived. In the 1990s, it was quickly eliminated from the cognitive market when the crop circles assumed the form of pictograms with much more complex designs in comparison with the simple circle. The physical hypothesis of the vortex henceforth became untenable. This is an example of elimination from the cognitive market. The same applies, and for the same reasons, to the interpretation based on the marasmius oreades.
19 However, this selection, before occurring on the cognitive market, may take place in the intimacy of individual thought, as recalled by Boyer (2001, 37): “Anthropology explains the origin of many cultural phenomena by moving, not from One to the Multiple, but from Many to the Many less, that is to say, from the multiple alternatives that our brain constantly produces to many fewer alternatives, which can certainly be transmitted to others and stabilize in a human group…. But these alternatives do not all suffer from the same fate. The majority only remain in the mind for a short moment. A small number linger for longer, but remain difficult to formulate or communicate. An even more restricted number are memorized and communicated to others, but not retained by them. Only a very small number of these alternatives are fixed in memory, communicated to third parties, memorized by them, and then communicated to others in a form that more or less respects the original concept.” A serious divergence exists among selectionist programs on the question of knowing what governs the selection of cognitive products. Some may be known as naturalistic/mechanistic, others as cognitive/mixed.
A. “Naturalistic–Mechanistic” Programs
20 The program mentioned here is defended mainly by anthropologists and psychologists (Boyer, Sperber, Atran, Clément, Barrett, Comsides, Nyhof, Tooby, etc.). I borrow the terms “naturalistic” and “mechanistic” from Dan Sperber, who claims them and provides the following definition (1997, 125–6):
An explanation is mechanistic when it analyzes a complex process as an articulation of more elementary processes, and it is naturalistic to the extent that there are good reasons to think that these more elementary processes could themselves be analyzed in a mechanistic fashion down to the level where their natural character would be wholly evident…not all mechanistic decomposition of mental phenomena is appropriate: a psychological decomposition that “speaks” to neurology is necessary.
22 The authors adhering to this program hope to contribute to the naturalization of the mind, which inspires in them a mode of argument based only on efficient causality,  thus distinguishing them from the mixed cognitive program, as we will see. According to these authors, the success of a belief depends on the ability to respect a certain number of mental rules, while producing an easily remembered cognitive “effect.” Rubin (1995), for example, emphasizes the idea that cultural success, particularly in oral traditions, can be achieved only through a cognitive product that is memorable, and that out of two beliefs, the most easily memorized will pervade.
23 Boyer (2001, 82–3) distinguishes five ontological categories that combine to form a story: an animal, person, artifact, natural object (river, mountain, etc.), and plant. These categories may be staged in a way that is said to be counterintuitive, or, on the contrary, intuitive. A counterintuitive item violates the natural expectations that we associate with an object. For example, a story about a man who walks is said to be “intuitive,” while another about a man who flies is “counterintuitive.” Several experiments have shown that people remember stories that violate their intuitive expectations better than those that do not. Thus, Barrett and Nyhof (2001) had their experimental subjects memorize and narrate Amerindian tales, noting that 92 percent of slightly counterintuitive items were reproduced compared with 71 percent of intuitive items.
24 These authors certainly realized that beliefs that successfully passed the cognitive selection process were not just those composed of counterintuitive elements (Atran, 2006, 68). Working on the memorization of several versions of Grimm’s fairy tales, Atran and Norenzayan (2004) showed that “cultural success” was a curvilinear (bridge-shaped) function of the number of counterintuitive elements found in a story. Too many or too few counterintuitive elements rendered the memorization of a cognitive product more difficult. Two or three supernatural events or objects conferred an optimal cognitive efficiency to the stories. Such results are even more fascinating as they were obtained from culturally diverse populations (American students, Yucatec Mayas, etc.).
25 In its final analysis, according to the working of this program, the success of a cognitive product on the market is thus based on the mnemic factor.
26 Firstly, for the adherents to this program, it is necessary for a proposition to be memorizable in order to be able to be disseminated, but this does not imply that this criterion alone is sufficient. Indeed, it seems reasonable to suppose that a belief, before being memorized, must be imagined and then believed so as to be able to spread. It is undoubtedly because the authors used stories that were and were not in the process of being formed that this dimension escaped the adherents to the naturalistic/mechanistic program. It is as if their research were struck by some kind of retrospective illusion, which Slovic, Fischhoff, and Lichtenstein (1984) termed creeping determinism or hindsight bias,  showing that knowing the outcomes of a past event increases the probability of these outcomes. In fact, because of the protocol types adopted by the adherents to the naturalistic/mechanistic program, only the mnemic factor appears in the analysis. With all other selection criteria invisible within the chosen experimental framework, the hyperbolic conclusion reached is that the mnemic factor is the only discriminant for the cognitive success of stories. In other words, it is ironic that they accept a specious argument not so dissimilar from that inspiring the partisans of crypto-functionalism, who, when focused on the characteristics of the constituted cognitive products, believe that they can interfere with the rules governing their formation.
27 Furthermore, in using Grimm’s fairy tales, for example, Atran and Norenzayan do not need to consider the story’s persuasiveness from the perspective of their experimental subjects, who obviously accept the tales as simple legends and not as accounts that should be accepted as true.
28 Lastly, the mnemic hypothesis is not neutral: it accompanies the tendency of these paradigms and their aim of naturalizing the human spirit. Indeed, with the mnemic criterion, these authors are not concerned with the way in which individuals “argumentatively” validate the statements to which they adhere. The memorization of a statement is an activity that evades, for the most part, the approval of individuals, and offers the chance to see its credulity as controlled by mechanisms,  that is, by efficient causes and in no way by reasons.
B. A Mixed Cognitive Program
29 The program that I would like to defend here may be called mixed, as it claims to describe the phenomena of belief dissemination through a dual mode of causality, implying both efficient causes and reasons. I defend elsewhere (Bronner 2006b, 2007, and 2008) the idea that mental phenomena (particularly those relating to beliefs and errors of reasoning) could, with difficulty, be reduced to algorithms. This ambition, however, seems somewhat premature, taking into account the current state of knowledge, especially in neuroscience, although it also encounters considerable descriptive difficulties. Considering the limitations of this article, I will not develop these points further. In this discussion with the adherents to the naturalistic/mechanistic program, I aim largely to test the idea that the ability to memorize an account is an ideal variable to describe its capacity for dissemination.
30 In 2003, 2004, and 2005, I conducted a series of experiments  aimed at testing the selectionist hypothesis regarding the generation and dissemination of beliefs. These studies took advantage of real enigmatic situations, which had created collective beliefs in the past. Voluntary subjects were confronted with statements presenting these situations, with the aim of investigating, in the manner of Ward (1994), the limited logical space of assumptions likely to be aroused and the power relations established between them. Initially, 144 face-to-face interviews were conducted.  The interviewees were given a statement that briefly described an enigmatic situation. After all the solutions imagined by the interviewee  were recorded (criterion of evocation),  he or she was asked to choose the most credible (criterion of credibility). In the second phase, collective interviews were conducted with 240 individuals (60 groups of 4)  according to the same protocol. Finally, voluntary subjects (120 individuals:  60 for enigma 1 and 60 for enigma 2) were presented with an enigmatic situation along with a series of hypotheses  proposing to explain it, the number of which exceeded participants’ memorization capacities.  The individual was recontacted twenty-four hours later and asked to state any recollected hypotheses. Two pieces of data were collected: the memorized scenarios and their order. Fourteen points, for example, were attributed to the first scenario, thirteen to the second, and so on. The different hypotheses were always presented in a different order and selected at random (in order to avoid first impression bias, cf. Luchins  and Asch , or introductory bias, Drozda-Senkowska ).
a. Two Enigmatic Situations
31 Here is enigma 1 as found in the three phases of the investigation:
32 Normally, there are approximately 100 deaths per year in a population of 10,000 inhabitants. However, in the 1980s in the Chinese neighborhood of the 13th district of Paris, which has approximately 10,000 inhabitants, there were only two or three deaths per year. How should this difference be explained?
33 These true facts  generated a remarkable belief in Paris during the 1980s, allowing the enigma to be credibly resolved through a solution likely to grab people’s attention at a time when the subject of immigration was at the forefront of many debates. The solution thus affirmed:
Deaths are not declared in this community in order to recover the identity papers of the dead. Bodies are transported and anonymously buried in neighboring countries (Holland and Belgium). In this way, it is possible and lucrative for new emigrants to enter the country as “real-false” clandestines. These papers are then resold, allowing new cheap labor to arrive.
35 The scenario and argumentation that dissolve this belief seem acceptable and even rather convincing, but apart from very rare exceptions,  they do not correspond to any reality. They are presented here as a paralogism. In a population of 10,000 people, there are approximately 100 deaths per year. In the population of the 13th district, there are fewer than 10 deaths per year. There are deaths in this community that are not declared. Where is the argumentative flaw here? It is found in the initial comparison. To predict the number of deaths in two distinct populations supposes their comparability, particularly in terms of their age structure. It is the clause that all things being equal that is forgotten since premise 1 neglects the term “on average.” On average, there are approximately 100 deaths in a population of 10,000 people. According to the demographic structure of this population, for example, this figure can vary widely. The Asian community under consideration was characterized by a high number of infants and young people in general. This demographic situation was thus not comparable with that of the aging French population. Consequently, there was nothing astonishing in the lower death rate in this community.
36 During the first research phase, the diverse responses of subjects made it possible to establish a typology distinguishing the logical space from the possible responses to the enigma.
- Lifestyle: in this scenario, individuals are supposed to have a better life expectancy because they eat better, do not smoke, do not drink, and so forth.
- Demography 1: this scenario proposes the correct solution, namely, if the mortality rate is low in this neighborhood, it is because its inhabitants are younger than the average age for the city.
- Demography 2: this equivalent idea supposes that the elderly leave the neighborhood before dying (returning to China, for example).
- Medical infrastructure: the mortality rate is low in this neighborhood because the inhabitants benefit from a better medical infrastructure.
- Environmental safety 1: the neighborhood in question has an exceptional natural protection.
- Environmental safety 2: The neighborhood in question is protected from human-caused dangers (delinquency, lack of road safety if the neighborhood is pedestrian, etc.).
- Psychological state: the neighborhood population has better mental health, is less anxious, less depressed, and so on.
- Disappearance of the dead: this scenario unfolded in the 1980s in Paris.
- False statistics: the figures proposed in the statement are simply disputed, their calculation method, and so forth.
- Social status: a well-off population, which would have a better life expectancy, lives in this neighborhood.
- Gender: this neighborhood houses more women, who have a longer life expectancy.
- Religion and culture: the population of this neighborhood has a culture and religion promoting a longer life expectancy (for example, encouraging people to take care of the elderly).
38 I will not expand on the results from phases 1 and 2 of the research here, as I have previously done so (Bronner, 2006a). Instead, I will discuss the phase 3 results, which directly relate to the issue of memorization. The following graph presents the memorization of the different scenarios proposed to resolve enigma 1 according to a criterion that allows both spontaneity and mnemic recurrence to be explained. A simple glance at these results reveals that scenario 8, the disappearance of the dead, was not the most easily memorized response for the subjects, even though this scenario is evident in vivo. On the contrary, in vitro, and taking into account only the criterion of memorization, scenarios 1 and 3 (lifestyle and departure of the elderly) are at the fore.
Enigma 1 and memorization
Enigma 1 and memorization
39 A similar result was obtained using the same protocol for the second enigma, which is expressed in the following statement.
In the 1950s, many people in Seattle (United States) observed unexplainable cracks on their car windshields. The phenomenon assumed such proportions that it aroused the interest of the highest authorities in the country.
How would you explain the enigma of these windshields in Seattle?
41 No less real than the first enigma, this mystery occupied public opinion for some time, assuming such proportions that President Eisenhower, at the request of Washington’s governor, thought it advisable to set up a team of experts in order to clarify the situation. Two competing beliefs were proposed to resolve this enigma in Seattle in the 1950s.
42 According to the first theory, known as “fallout,” the situation resulted from Soviet nuclear tests that had polluted the atmosphere, the fallout of this pollution taking the form of light rain that was corrosive to glass, aided by the humid climate of the Seattle area.
43 The second theory, known as “tarmac,” accused the large-scale reconstruction of the road networks launched by Governor Rosellini. This highway program was supposed to cause frequent acid splashes due to the recent laying of tarmac, furthered by the wet climate (once again) of Puget Sound.
44 The investigators were well advised not to pay too much attention to these explanations. They sought to determine why the inhabitants of Seattle claimed their car windshields to be pockmarked. They subsequently discovered that the windshields in Seattle were no more damaged than those in any other city, and that the number was not increasing. In fact, as the rumor spread throughout the city, its inhabitants started to do what they normally did not do: they meticulously examined the windshield of their vehicle and were thus able to see that it was occasionally damaged with small cracks. However, as the experts said, it was nothing other than normal wear and tear, which generally went unperceived. Watzlawick was thus right in pointing out the following with regard to this matter:
“What erupted in Seattle was not an epidemic of damaged windshields, but of examined windshields”
46 Here is the typology of the scenarios imagined for enigma 2, which was conducted in the same way as enigma 1.
- Natural climatic phenomena: The climate, rain, or other natural phenomena deteriorated the windshields.
- Pollution (nuclear/chemical): nuclear, chemical, or other pollution damaged the windshields (acid rain, for example).
- Condition of the roads: The roads were in poor condition and, in the end, this caused damage to the windshields.
- Road debris: Public works in the region or passing vehicles gave rise to spattering, which would have damaged the windshields.
- Paranormal: All scenarios involving the intervention of extraterrestrials, ghosts, and so on, were placed in this category.
- Intentional manufacturing issue: The windshields were of poor workmanship, so the problem was linked to their manufacture; manufacturers hoped motorists would be forced to change their windshields more frequently.
- Unintentional manufacturing issue: The windshields were of poor workmanship, so the problem was linked to their manufacture without any commercial intent.
- Vandalism: The windshields were damaged by one or more vandals during the night.
- Self-vandalism: The windshields were damaged by their owners, who sought to defraud their insurer.
- Natural wear and tear: The windshields were not more damaged in Seattle than elsewhere; the situation rather involved an epidemic of observed windshields.
- Natural vibrations: An earthquake or other phenomenon of natural vibration deteriorated the windshields.
- Human-induced vibrations: The proximity of an airport created vibrations that deteriorated the windshields.
- Maintenance issue: The cleaning products that people used to wash their windshields were corrosive.
- Animal interference: Bird droppings, a swarm of locusts, or any other animal interference could explain the windshields’ deterioration.
Enigma 2 and memorization
Enigma 2 and memorization
48 Similarly to the first enigma, the two scenarios that had, in reality, attracted overwhelming support, were not the most remembered in the experiment (scenarios 1 and 2), which were instead in fifth and sixth position behind scenarios 11, 1, 6, and 8, respectively. These results lead me to make two remarks, which will serve as my conclusion.
49 First, the mnemic criterion does not appear sufficient to account for the generation and dissemination of beliefs. By retaining this criterion alone, we do not obtain a satisfactory simulation of what occurred on the cognitive market. In other words, these results lead me to dispute the prominent proposal of the mechanistic/naturalistic program: the mnemic criterion is the appropriate index for evaluating the cultural success of a product. In my opinion, the mnemic variable should not be completely revoked, but rather supplemented, specifically, by the variable of evocation (i.e., the facility with which an account can be invented) and the variable of credibility (i.e., the facility with which an account can be believed). This results in a simulation that is much more in line with reality (Bronner, 2006c). The intellectual challenge is thus to determine how these variables “interlock.” Indeed, evocation precedes credibility, which precedes memorization; these factors are not disjoined; they adjust to each other, but in a certain order. The investigations mentioned above show that evocation and credibility are, on average, much more selective than memorization. However, in phase three of the investigation, the protocol presented subjects with all the scenarios. On the cognitive market, a selection would probably have already taken place, and thus, memory would only have to apply to some of the scenarios. It can also be assumed that an individual, who is confronted with competing beliefs that are already in existence, would not have much trouble inventing others (which limits the influence of the factor of evocation). Indeed, why bother imagining and defending rival solutions to products that already offer an effective solution to an enigmatic situation without any costs (imagination or invested mental energy)? The issue of temporality is particularly difficult to simulate in vitro.
50 Second, the aspect which naturalists do not take into account, but which sociologists, whatever their leaning, will be delighted to stress, relates to the context of the statement, which is obviously crucial for the mnemic variable, but also for the evocation of an account or its credibility. In the two enigmas presented above, we see that the beliefs that emerged on the cognitive market constituted a hybridization between the intellectual solutions to a problem and a particular “social context.”  Thus, the legend of concealing dead people in the 13th district owes much to the depictions emerging from the debate on clandestine immigration in the 1980s and to the stereotypical view of the Chinese community. Enigma 1 was thus presented without its context to half of the sample population, summarized in the following manner:
Normally, there are approximately 100 deaths per year in a population of 10,000 inhabitants. However, it was noted that in a particular neighborhood with 10,000 inhabitants in a large city X, there were only two or three deaths per year. How should this difference be explained?
52 The results obtained for the variables of both evocation and credibility change considerably according to whether the subjects are presented with a statement with or without its context. This is not surprising, but it enables us to emphasize the fact that the logical skeleton of a statement is not enough to imitate the possible responses of the social actors, which are also conditioned by contextual beliefs giving a social embodiment to this statement. Thus, the scenario regarding the disappearance of dead people will benefit the most from the context of the statement, because it tends to focus the mind on the ethnocultural characteristics of the population. For this reason, but conversely, the situation suffering the most from the inclusion of the context is the explanation by exogenous causes (the low mortality rate being explained by the characteristics of the neighborhood and not by those of the population: medical infrastructure, environmental safety 1 and 2). For the variable of evocation, these explanations move from second position without context, to last place with context. For the variables of credibility or spontaneity (first scenario cited), this explanatory mode literally collapses, occupying the last position by far. From a cognitive perspective, the mention of the population’s nationality renders the categorization according to social space less fitting.
53 What has just been said for enigma 1 could just as easily apply to enigma 2. It is obvious that the two scenarios that gained “cultural success” in Seattle in the 1950s, be it the Soviet nuclear pollution or the roadwork initiated by the governor, which was expensive in terms of local taxation, owe much to contextual information and/or beliefs.
54 The descriptive and explanatory issues of this hybridization between the intellectual solutions and the information relating to the social context are completely absent in the analysis of the mechanistic/naturalistic program. Several reasons may explain this. First, the materials used in their analyses (Grimm’s fairy tales, great religious systems, etc.) make it awkward to reconstruct the enunciative context. Furthermore, they place believable and unbelievable statements on the same analytical level, thus neglecting the multifaceted character of a product’s success on the cognitive market. Finally, their naturalistic ambition contributes to their concentration on a single factor (memorization), which responds to the criteria of efficient causality while concealing the fact that they succumbed to a retrospective bias.
55 The success of a product on the cognitive market may be explained according to three elements: the characteristics of the transmitter of the belief, those of the receiver, and the nature of the belief itself. This may be methodologically approached using three criteria: evocation, credibility, and memorization. The criterion of evocation expresses the facility with which individuals, either alone or in a group, evoke one scenario or another. On this point, I was influenced by research conducted in psychology on social representations. It was thus necessary to take into account both the spontaneity with which an account appeared in conversation (by identifying the order in which the scenarios were mentioned) and the recurrence of the topic during the interview. The idea is that the stronger the factor of evocation is, the more likely an account is to appear on the cognitive market. Of course, once formed, there is nothing to say that the account will become established. However, the sine qua non condition enabling a cognitive product to gain success is that it is available. The criterion of credibility expresses the subjective evaluation of individuals regarding the credibility of a cognitive product. It is assumed that the force of their conviction is associated with the nature of the arguments supporting the accounts. I was influenced here by Boudon’s conception of cognitive sociology, which considers that the cause behind an individual’s adherence to a belief is most often motivated by his reasons to endorse it. Finally, the criterion of memorization effectively expresses, if the naturalistic/mechanistic program is believed, the cognitive effect produced in people’s minds, a point that I do not dispute and that, on the contrary, I borrow in order to propose a mixed interpretative model for the success of a belief on the cognitive market. This proposal forms part of a more general program that can be described as cognitive sociology, which proposes to account for social facts by seeing them as the consequence of a hybridization between the invariants of human thought and the variables of social life.
Aristotle (1962, 11, 1259a, 11–28) provides the foundations for the law of supply and demand, and while obvious traces appear in the work of Smith, Turgot, or, later, Walras, it is Marshall (1906–1909) who generalized the concept through the intersection of two curves of supply and demand. This law is in effect the synthesis of two laws. The first, known as the law of demand, states that the price of a commodity increases in line with its demand. The second, the law of supply, states that the price of a commodity decreases when the supply is increased. The law of supply and demand is thus the convergence of these two laws. For several reasons, only the law of demand has an equivalent on the cognitive market. It is stated as follows: There is an inversely proportional variation between the strength of the cognitive supply and the necessary cost (cognitive and social) of adhering to it.
On this point, refer to the work of Chaiken (1980).
Shérif and Hovland (1961), for example, highlight the fact that individuals tend to overestimate the skills and assessments of individuals whom they like, even whom they love, whereas they underestimate those of individuals whom they do not like. On this subject, see also the ingenious experiments of Bovard (1953).
I developed these three aspects on the limits of rationality in Bronner (2003), and particularly explored the cognitive limits of rationality in Bronner (2007).
For example, Campion-Vincent and Renard (2002, 12): “These creations are anonymous and collective, because, while born from individual innovations, they are transmitted and constantly re-elaborated by the social group in which they play a functional role”; or Renard (1999) alone, who considers rumor to be a phenomenon whose function is to respond to collective questions, which can be reduced to six categories (new technologies, otherness and foreigners in particular, wilderness, urban violence, development of morals, and the supernatural). Even Reumaux (1998) perceives the effects of a “captivated collective conscience” in the phenomenon of rumors, attributing to it the virtues and function of collective exorcism. Kapferer (1995, 140) partly maintains this idea when he explains “the raw sensitivity of the social body” owing to the fact that it can “spontaneously speak in order to express itself, even outside of any event.”
For this reason, I cannot follow J.-B. Fox (2006, 215) when he writes, “And regardless of whether this phenomenon is ‘Lamarckist’ rather than ‘Darwinist.’” I cannot subscribe to the benevolent reproach directed at me: “This concept is exposed to the same criticism as that leveled against functionalism: how are counterproductive, socially harmful beliefs ‘more adapted.’” Obviously, selection mechanisms should not lead to counterproductive beliefs on the cognitive market, instead sanctioning the appearance of such phenomena, as is well known in the field of biology. The mechanisms of nature preserve many things that are not always useful. This is well known in the case of our appendix as well as our craving for sugar. Thus, during the Pleistocene, our distant ancestors undoubtedly benefited from their propensity to consume sweet things and so constitute reserves of rapidly available energy. But this propensity has become suboptimal in a society in which sugar can be produced in large quantities, often endangering our health.
This term is of course borrowed from Aristotle’s Metaphysics.
This error has often been commented on in the social sciences, being diversely identified and named. Sometimes, one speaks of “hindsight bias,” “teleological dependence,” or “a posteriori illusion,” but generally the term “retrospective illusion” is used.
As is also indicated, for example, by the title of Fabrice Clément’s book, Les mécanismes de la crédulité (2006).
I would like thank here the students enrolled in the multidisciplinary bachelor’s degree at the University of Nancy II in 2003–2004 and 2004–2005, without whose assistance this research would have been greatly diminished.
Among 72 men and 72 women without any other distinction.
The interview grid was designed to encourage subjects to imagine the greatest possible number of hypotheses.
The criterion of evocation expresses the facility with which individuals, either alone or in a group, evoke one or another scenario. It was necessary to take into account both the spontaneity with which an account appeared in conversation (by identifying the order in which the scenarios were mentioned) and the recurrence of the topic during the interview. This method relates to what social psychologists call “prototypical and categorical analysis” (for more details, see Vergès 1992 and 1994). Here, I draw from the idea that the stronger the factor of evocation is, the more likely an account is to appear on the cognitive market. Of course, once formed, there is nothing to say that the account will become established.
In the second phase of this investigation, the criterion of gender was abandoned, as it was not a discriminating factor in the first phase. On the other hand, the variables of age and profession were retained: the average age of the groups was to be greater than thirty-five years, while each group was to be composed of individuals from different professions.
Taking into account the difficulties inherent in the protocol, the only sampling rule retained was that of age distribution, controlled by comparing the average to the ratio of the interquartile range/interval.
These hypotheses corresponded to those imagined in the first two experimental phases.
Since the famous article of Miller (1956), we know that short-term memory is limited, and under usual conditions, it can register more or less seven elements.
I would like to thank J.-B. Fox and V. Campion-Vincent for the information that they kindly gave me on this subject.
The newspaper Libération (11/1/83) seems to admit that several cases of fraud were found, without commenting further. Of course, this is not impossible, but, as the daily newspaper specifies, the phenomenon is so marginal that it cannot account for the enigma.
I use this term metaphorically, and on this point, align myself with the opinion expressed by Raynaud (2006).