2.1 “Supernatural” as a Qualifier

Let us begin by revisiting traditional attempts to define divination through its non-ordinary nature. Taking “supernatural” as an example (with other qualifiers following a similar logic), we immediately encounter the question of whether to define it emically (from an insider’s perspective) or etically (from an outsider’s perspective). Mainstream anthropological thinking often privileges the emic approach of adopting the native’s point of view, but this approach is problematic with the concept of the “supernatural” because it likely does not exist in many cultures.Footnote ⁷ There have been heated discussions in anthropology regarding whether the supernatural constitutes a valid emic category (Dein, Reference Dein2016; Lohmann, Reference Lohmann2003; Winzeler, Reference Winzeler2012), in particular whether the natural versus supernatural dichotomy is a Western construct imposed on many traditional, small-scale societies (Hallowell, Reference Hallowell and Diamond1960). A cursory look at the ethnographic literature reveals much direct commentary on the lack of natural versus supernatural distinction in premodern, non-Western societies: “few preindustrial cultures make a neat distinction between natural and supernatural phenomena … [as they] … may simply lack emic categories for ‘natural’ and ‘supernatural’” (Petrus & Bogopa, Reference Petrus and Bogopa2007); “our use of the notion “supernatural” does not correspond to any Gururumba concept: they do not divide the world into natural and supernatural parts” (Newman, Reference Newman1965, p. 83); “supernatural persons … [if applied to characters in the myths of the northern Ojibwa] … is completely misleading, if for no other reason than the fact that the concept of ‘supernatural’ presupposes a concept of the ‘natural.’ The latter is not present in Ojibwa thought” (Hallowell, Reference Hallowell and Diamond1960). In his masterpiece Witchcraft, oracles and magic among the Azande, Evans-Pritchard (Reference Evans-Pritchard1937) similarly makes the following comment on the lack of the natural versus supernatural distinction of the Azande people:

Of course, we need to approach these ethnographic accounts with caution, mindful of ethnographers’ biases and the selective nature of their reporting (Geertz, Reference Geertz1973). We should also bear in mind that there are accounts pointing to the opposite: Malinowski, for example, suggests that the Trobriand islanders may indeed have the emic concept of “supernatural”Footnote ⁸ and would apply magical and rational means selectively to achieve desired outcomes in different situations (Malinowski, Reference Malinowski1992). However, the overwhelming consensus from these ethnographic descriptions is that many societies do not explicitly recognize the distinction between natural and supernatural, and the concept of the “supernatural” often does not constitute a valid emic category. Indeed, early social theorists like Durkheim have posited that the notion of the supernatural is relatively recent in the history of human thought, emerging alongside the rise of Enlightenment science in the seventeenth and eighteenth centuries (Durkheim, Reference Durkheim1915). Thus, for the vast majority of human history, the world was mostly viewed as an integrated whole, without clear divisions between different ontological realmsFootnote ⁹ (Sahlins, Reference Sahlins2022).

We are then left with the etic approach, but defining the supernatural in etic perspectives is not without problems, either. For one thing, the modern Western understanding of “supernatural” is far from clear-cut (Saler, Reference Saler1977). Delineating the meaning of the term can involve sophisticated theological discussions on the nature of reality (Berger, Reference Berger1977), and its usage by social scientists is similarly ambiguous, as it is often used as a convenient shorthand to refer to a broad cluster of loosely related phenomena and events. A systematic review of psychologists’ employment of “supernatural” and related terms by Lindeman and Svedholm (Reference Lindeman and Svedholm2012) reveals that these concepts are variably defined as either domain-general (e.g., false beliefs, scientifically impossible phenomena, associative biases, irrational acts) or domain-specific (e.g., specific beliefs that violate our intuitive ontology). The authors conclude that domain-general definitions are problematic because they are either too narrow (e.g., associative bias does not encompass belief in devils) or too broad (e.g., false beliefs would include the belief that dolphins are fish which does not appear supernatural), and propose that supernatural beliefs are best described as a confusion of ontological categories, that is, misattributing properties that belong to objects/events in one category to those in another category. For example, this can include beliefs that thoughts can move external objects or that force or energy can possess life (e.g., Feng Shui, Chi), as well as the idea that minds can exist independently of bodies and operate as animate entities (e.g., angels, devils, and ghosts).

The examination ofthe definitional issue of “supernatural” has two important implications for our discussion of divination. Firstly, it highlights the need for an etic approach in defining divination, one that leans on a scientific understanding of the world and its causal structures. Secondly, we must recognize that the use of the term “supernatural” – along with similar terms like “occult,” “obscure,” or “non-ordinary” – is not without its complications. As I will argue, while Lindeman and Svedholm’s (Reference Lindeman and Svedholm2012) definition of “supernatural” makes it a good qualifier for “magic,” it requires some adjustments to suitably apply to “divination.”

2.2 Demarcating Divination from Non-divination and the Two Senses of Divination

As with most complex sociocultural phenomena, any general theory of divination faces a demarcation problem. In her book Ancient Greek Divination, Sarah Johnston (Reference Johnston2009, p. 5) explicitly points this out:

Now, if we remind ourselves of the dictionary definition of divination, can we say in any sense that the use of dark clouds to predict rainfall is natural while the use of a night owl to predict a political insurrection is supernatural? One way to make the distinction is to appeal to the involvement of the divine, as Johnston herself seems to suggest: If the signs are believed to be provided by (typically anthropomorphized) deities, then the method of interpreting these signs may be deemed divination. After all, the etymological root of “divination” has an unambiguous link to the idea of divine involvement, a concept frequently assumed, either explicitly or implicitly, within scholarly discussions (Brown, Reference Brown2006; Zuesse, Reference Zuesse1975). In discussing the status of prophecy as a possible form of divination, for example, Kitz (Reference Kitz2003) explicitly states the “fundamental principle” of divination: “ … divination is based on one very simple premise: all divine action causes material reaction.” Thus, divination can be viewed as a practice of communicating with anthropomorphized superhuman entities such as Gods, spirits, or deceased ancestors.Footnote ¹⁰ Throughout this Element, I will label this “divination as communication with the divine” as the “narrow-sense” definition of divination. This definition aligns well with Tylor’s (Reference Tylor1871) minimal definition of religion as a belief in spiritual beings, offering the benefits of precision and elegance. However, defining divination in this narrow sense would leave out many practices that do not invoke any superhuman agent yet are traditionally recognized as forms of divination, both in ancient and modern contexts such as Yijing (I Ching), tarot cards, and palm reading.Footnote ¹¹ This calls for the formulation of a “broad-sense” definition of divination, one that encompasses these practices while simultaneously distinguishing them from ordinary information technology.

Let’s refine the use of the term “supernatural” to better suit our purposes. Lindeman and Svedholm (Reference Lindeman and Svedholm2012) define the supernatural as mistaking ontological categories – an etic perspective, since practitioners and believers likely do not view their beliefs as mistaken. While this approach works fine for magic, it doesn’t fully encompass broad-sense divination, which is fundamentally rooted in a belief in the interconnectedness of the universe (French, Reference French2005, p. 135; Hong, Reference Hong2024), often contradicting our current scientific understanding of worldly causality. This notion bears resemblance to the concept of “ontological category confusion” but extends beyond it by positing causal links between events and phenomena that do not necessarily belong to different ontological categories yet are unrelated from the perspective of modern science (Hong & Henrich, Reference Hong and Henrich2021). Consider the use of astrology to predict personal outcomes, based on the assumption that the positions and movements of celestial bodies influence human lives. This belief does not conflate ontological categories, as both celestial entities and human experiences exist within the physical world. Yet, it posits a causal relationship deemed implausible by our current (scientific) understanding of the world due to the absence of known mechanisms linking the two types of entities. As such, “supernatural” in this context signifies more than mere incorrect beliefs; it refers to beliefs in causal relationships lacking any scientifically plausible mechanism. Conversely, some cultural practices are based on factually incorrect beliefs but are not typically categorized as divination.

Let us consider fetal sex prediction as an example to illustrate how the proposed demarcation scheme works in practice. In modern societies, ultrasound technology can determine fetal sex with nearly 100 percent accuracy, and no one would consider it divination – its legitimacy is firmly grounded in the scientific understanding of biology and physics. More interesting are the various “folk” methods of fetal sex prediction. On one hand, some methods deviate significantly from our modern understanding of causality. For example, popular numerological methods in China calculate fetal sex based on the mother’s age and the month of conception (Hong & Zinin, Reference Hong and Zinin2023), and the use of dreams to foretell a baby’s sex is a recurring belief in many traditional societies (Hong, Reference Hong2022b). Under the proposed demarcation scheme, these methods would be classified as “divination” because they are scientifically implausible. On the other hand, there are methods, such as associating maternal food cravings or abdominal bumps with fetal sex, that appear biologically plausible but do not perform better than chance (Forbes, Reference Forbes1959; Hong & Zinin, Reference Hong and Zinin2023). In this framework, these biologically plausible but scientifically unsupported methods would not qualify as “divination.” To drive the point home, consider the use of fundal height to predict fetal sex. While its predictive validity is highly questionable, I trust that most readers would intuitively agree that it does not belong in the category of “divination.”

Figure 1 shows a general topology of divinatory practices in this framework. To reiterate, “narrow-sense” divination refers to communication with the divine (anthropomorphized superhuman deities), and in accordance with the existing literature, I differentiate between two primary methods of divination: inspired divination, where the diviner directly receives and communicates information from a deity, often as a specific message, and technical divination, which involves the diviner interpreting the hidden meanings behind natural occurrences, signs, or omensFootnote ¹² (Flower, Reference Flower2008; Kitz, Reference Kitz2003). This is an ancient classification which traces at least back to Plato, who famously expressed a preference for inspired divination, arguing that human interpretations of divine signs are inherently prone to errors and misjudgments (Landry, Reference Landry2014).

Figure 1 A typology of divination practices.

Broad-sense divination, on the other hand, includes both narrow-sense divination and scientifically implausible information technology that does not involve communication with the divine. There has been some inconsistency in whether to classify the latter kind of divination as “technical divination” (illustrated by a dashed line with a question mark in Figure 1), and I have no intention of policing the scholarly usage of divination here. I do, however, wish to highlight the implicit assumption of “divination as implausible information technology” in contemporary scholarly discourse. Phrenology, for example, has often been referred to as a form of “divination” by modern scholars (Robertson, Reference Robertson2018). Anthropologists also sometimes use the term “divination” pejoratively to critique what they view as questionable, in particular ethically problematic information technologies, such as genetic testing (Lock, Reference Lock2005; Palmié, Reference Palmié2007) and big data/algorithmic predictions (Cabrera, Reference Cabrera2020; Lazaro, Reference Lazaro2023).

This topology is a decidedly etic approachFootnote ¹³ that aims to capture its usage within (Western) academic discourse. Although not a “carving nature by its joints” way of demarcating divination from non-divination, it provides an analytically useful framework for understanding various cognitive and social factors influencing different types of divinatory practices. Such a definition of divination also implies that whether we categorize a particular information-generating technology as divination hinges on the perceived presence of a scientifically plausible mechanism linking the sign and the outcome, rather than its empirical record of predictive success and failures. Indeed, an intuitive understanding of the validity of technological practices often precedes the need for “data” (Hong & Henrich, Reference Hong and Henrich2021).

3.1 Divination as Dispute Resolution

A common way in which divination has been explained is mechanism for resolving disputes. This perspective posits divination as an arbitrative tool to facilitate agreement in challenging situations (Johnston, Reference Johnston, Johnston and Struck2005). For example, Victor Turner (Reference Turner1968) highlights divination’s role in social redress among the Ndembu people. According to Turner, Ndembu diviners try to elicit from their clients’ responses which give them clues to the current tensions in their groups. Divination, in this case, becomes a form of social analysis that reveals hidden conflicts so that they can be dealt with by institutional norms and procedures. Turner notes that diviners, through experience, have learned to “reduce their social system to a few basic principles and factors, and to juggle with these until they arrive at a decision that accords with the view of the majority of the clients at any given consultation” (p. 51). Furthermore, the public nature of divination, especially in emotionally charged contexts, allows diviners to reaffirm social norms in impactful ways. Similarly, Park (Reference Park1963) argues that the people employ divination in the selection of a house site in which “the diviner in effect provides a legitimating sanction upon a process of structural realignment which … would be difficult indeed to sanction in any remarkably different manner.” According to Park, selecting a house site influences not only the actor him/herself but also their kinship network, and involves the important decision of whether or not to stay with or depart from one’s kinship network. In this scenario, divination “depersonalizes” an otherwise deeply personal and impactful decision and thus precludes potential disputes and conflicts.

Another prominent cultural phenomenon that falls into the category of dispute resolution that involves the divine is trial by ordeal, in which God(s) is/are believed to condemn the guilty and exonerate the innocent through formally conducted physical tests (Radding, Reference Radding1979). Trial by ordeal itself was rather ancient and has been suggested to have existed as early as the time of Hammurabi in ancient Babylon (Bell, Reference Bell2017), and has been extensively studied by legal scholars (Gross, Reference Gross1937; Kamali, Reference Li2018). Though not always classified as divination in the literature, trial by ordeal fits nicely in the topology of divination in Section 2 as a way of determining the guilt/innocence of suspects with the help of the divine. The specific methods of ordeal can vary, but the general idea is that if the suspect is able to go through the tests unharmed, then they are proven innocent as God is on their side (Leeson, Reference Leeson2012). Various theories have been proposed to explain this seeming irrationality, and functional approaches (particularly in legal scholarships) emphasize the role trial by ordeal plays in forming consensus and unity in communities. Hyams (Reference Hyams, Arnold, Green, Scully and White1981), for example, suggests that trial by ordeal in medieval England often served to release tensions and reinforce the community’s standards of proper behavior by having God publicly cast his judgment. Bartlett (Reference Bartlett1986) similarly argues that ordeal was a reasonable instrument for settling disputes in medieval Europe, and it was mostly used in extraordinary situations when other modes of proof failed. Trial by ordeal, or more broadly the invocation of the divine to help distinguish the guilty and the innocent similarly exists in many traditional societies, in particular in the context of identifying witchcraft and sorcery (Evans-Pritchard, Reference Evans-Pritchard1937; Hiltunen, Reference Hiltunen1986; Lambert, Reference Lambert1956).

3.2 Divination as Mechanism to Legitimize Political Power

In societies with complex political hierarchies, divination is often a highly political matter and under tight regulations of the state (Cooley et al., Reference Cooley, Pongratz-Leisten and Strine2014; Pecírková, Reference Pecírková1985; Smith, Reference Smith1986). As such, a common functional interpretation of divination in these contexts is that it helps legitimize and solidify political power. For example, elaborate divination procedures in the Shang and Zhou dynasties of ancient China were argued to support the bureaucratic institutions as a source of state power (Flad, Reference Flad2008), and ancient Mesopotamian kings often employed professional diviners who would offer favorable interpretations of existing omens regarding the kings’ rule out of political expediency (Brown, Reference Brown2006).

A particularly noteworthy form of divination is celestial (or astral) divination, where heavenly phenomena are believed to be associated with earthly (political) affairs (Reiner, Reference Reiner1995; Swerdlow, Reference Swerdlow1999). Historically, this form of divination was frequently employed to justify the legitimacy of the rulers of the state by showing auspicious signs from Heaven. A memorable example occurred in ancient Rome, where the appearance of a comet in 44 BCE, shortly after the assassination of Julius Caesar, was interpreted by many Romans as a sign of Caesar’s deification. Augustus (then Octavian) used this celestial event to bolster his political position by claiming that the comet was Caesar’s soul ascending to the heavens, thereby implying divine favor for himself as Caesar’s heir (Gurval, Reference Gurval1997).

A more salient political use of divine messages is perhaps oneiromancy, the deciphering of dreams to extract meaningful information. In traditional China, dreams of suns and dragons were typically associated with imperial power, and historical records have shown a plethora of instances where princes and other competitors to the throne used (or fabricated) such dreams to enhance political legitimacy (Hong, Reference Hong2022b; Yu, Reference Yu2022), and historians similarly had an incentive to retrospectively fabricate such dreams to justify the emperors’ political power of their own dynasty (Fang, Reference Fang2015). For example, the official dynastic record of the Eastern Han Dynasty documented a story of its founding father, the Guangwu Emperor, recounting a dream to one of his generals as he was achieving tremendous military success yet had not officially declared himself emperor:

As modern readers, we cannot know exactly whether Guangwu indeed had such a dream or was merely fabricating it, but its political significance is obvious. In a cultural context where dreams were believed to be a valid channel for divine messages, it is no surprise that political actors strategically resort to it for their own advantage. More generally, divinatory signs could be used as political arguments for persuasion purposes (Hong & Chen, Reference Hong and Chen2024; Ramsey, Reference Ramsey2023). In the Neo-Assyrian period, for example, royal advisors often counseled their kings using the celestial omen series (Rochberg, Reference Rochberg2004, p. 220).

3.3 Divination as a Method to Reduce Uncertainty and Alleviate Anxiety

Of course, divination does not always occur in the public and may be utilized in rather private settings, particularly where its usage is deemed illegitimate by the authorities,Footnote ¹⁵ or when the information sought is highly personal. Scholars have also offered individual-level explanations, the most common being that divination helps alleviate anxiety by reducing uncertaintyFootnote ¹⁶ in what appears chaotic and mysterious. Such examples are widespread across different historical contexts. Professional dream interpreters, for instance, were believed to alleviate the anxiety of worried rulers in traditional China by clarifying the meaning of otherwise vague dreams (Vance, Reference Vance2017), and ancient Greek divination has been described as a “cultural heuristic to help people overcome the unpleasant feelings of uncertainty and the fear of the unknown” (Jouan, Reference Jouan1990). In the context of decision-making where clients desperately need practical information to handle pressing matters, divination can also be used to directly guide actions, thereby relieving the clients’ anxiety. This type of explanation often emphasizes the reduction of psychological discomfort as a result of the reduction of uncertainty, and implies that the participants in divination understand that their efforts do not “work” in a tangible sense. Instead, divination primarily provides psychological comfort in situations where no other solution is available. This line of thinking traces back to Malinowski’s analysis of magic, where he suggests that people recognize the limits of their empirical knowledge and capacity, yet are driven by strong emotional factors such as anxiety, fears, and hopes when they resort to magic (Malinowski, Reference Malinowski1992). In other words, magic functions to preserve human confidence (even if illusory) in threatening situations (Kippenberg, Reference Kippenberg, Schäfer and Kippenberg1997).

Such explanations, which focus on individuals’ psychological needs for certainty and control, are also common in the anthropological literature, and often have an added dimension of “meaning-making.” Victor Turner (the same author that describes divination as dispute resolution), in his analysis of divination in rural Africa, highlights that in harsh environments with high morbidity and mortality rates, low nutritional levels, plagues, droughts, and famines, diviners can counteract the fears and anxieties produced by such indeterminacy though the frequent interposition of their overdetermined schemata provided by divination, which restores coherence and meaning. (Turner, Reference Turner1975, p. 25). Similarly, Denham (Reference Denham2015) gives the following description of Nankani (an ethnic group in Northern Ghana) divination that involves seeking answers from ancestral spirits:

This particular account emphasizes the role divination plays in making sense of the apparently unexplainable (e.g., “Why does this misfortune happen to me?”), and even if there are no immediate action recommendations, obtaining a coherent and sensible explanation could confer affective benefits. Other scholars have explicitly emphasized the therapeutic effect of divination (Ajala, Reference Ajala, Peek and van Beek2013; Bohannan, Reference Bohannan, Beattie and Lienhardt1975; Zempléni, Reference Zempléni1975); in fact, divination sessions have sometimes been compared to modern psychological counseling (Chuang, Reference Chuang2011; Kohol & Akuto, Reference Kohol and Akuto2019).

Some psychologists similarly suggest that a psychological need to relieve anxiety may be a sufficient explanation for certain types of divination (Jahoda, Reference Jahoda1970). While studies in psychology are typically not about divination per se, there have been a number of studies on how the need for psychological certainty contributes to superstitious thinking and behaviors generally. Baseball players, for example, are suggested to be particularly superstitious due to the inherent risks and uncertainties in their games. These superstitions often manifest themselves in the form of routinized rituals that may be idiosyncratic to individual players (Gmelch, Reference Gmelch, Moro and Myers2010). The literature on pathological gambling also frequently invokes cognitive biases such as the illusion of control to account for superstitious behaviors in games with significant uncertainty (Cocker & Winstanley, Reference Cocker and Winstanley2015; Griffiths, Reference Griffiths1990). Similarly, theories such as compensatory control propose that individuals employ “compensatory strategies” to manage uncertainty – such as affiliating with external systems perceived to act on their behalf or seeking out simple, clear, and consistent interpretations of the social and physical environment – to regain a sense of control over their lives (Landau et al., Reference Landau, Kay and Whitson2015). Unlike anthropological approaches, psychological perspectives typically assume that individuals might genuinely hold false beliefs due to faulty reasoning – a point we will revisit – and that these beliefs lead to tangible behavioral consequences.

3.4 Divination as a Special Way of Knowing

Though functional and individual-psychological theories of divination currently dominate the scholarly discourse, there have been attempts to describe divination as a special way of knowing (Tedlock, Reference Tedlock2001). These accounts are “cognitive” in so far as the authors are dissatisfied with functional accounts of divination that emphasize only their sociopolitical aspects or anxiety-relieving/therapeutic effects, and call for greater attention to the practice of divination itself as an information-generating activity. Divination is “special,” however, in that this kind of knowing is to be distinguished from ordinary information technology.

A recent account that attempts to explain divination in informational terms concerns how it helps us express intuitions. The idea itself was quite novelFootnote ¹⁷ when the classicist Peter Struck (Reference Struck2016) offered a comprehensive account with an elaborate analysis of how ancient Greek employed divination for such a purpose, and there have been a few follow-up studies since (Baratz, Reference Baratz2022). According to Struck, divination does not produce new information in a technical sense but serves to uncover and convey knowledge already held subconsciously. In cognitive science, it is well established that we know certain things without understanding how (Gigerenzer, Reference Gigerenzer2007; Radman, Reference Radman2012), and divination is “the most robust ancient version in a long series of attempts” to express this “surplus knowledge” (Struck, Reference Struck2016, p. 15). Struck argues at length that ancient philosophers often attributed sudden, inexplicable insights to divine sources, and this type of involuntary, gut-feeling sort of knowledge parallels the modern concept of cognitive intuition. Note that Struck labels his approach “cognitive,” as clearly seen in the title of his book Divination and Human Nature: A Cognitive History of Intuition in Classical Antiquity. By “cognitive” he means that divination, especially inspired divination, can be viewed as a method of acquiring information through intuition rather than rational inference or conscious deliberation.

More generally, divination is sometimes suggested to operate as a “non-normal” mode of cognition (Peek, Reference Peek and Peek1991). Spirit possession in shamanism exemplifies this, where diviners enter altered states of consciousness (Beattie & Middleton, Reference Beattie, Middleton, Beattie and Middleton1969). Anthropologists generally accept the sincerity of the diviners (i.e., that they are not charlatans merely faking it) in these contexts, and there is plenty of evidence showing that professional shamans do genuinely seek to enter altered states and sometimes would resort to the use of hallucinogenic plants and other psychedelics (Hatsis, Reference Hatsis2018; Metzner, Reference Metzner1998).

A feature of many anthropological accounts that suggest divination as a special way of knowing is their somewhat elusive and ambiguous stance on the truth status (in the ordinary sense) of information generated by divination and the motivations behind individuals engaging in it (Hong & Henrich, Reference Hong and Henrich2024). Some examples include “ … [divination] is itself a mode of discovery that makes a truth-claim with regard to how it represents the world … the diviners’ claims are incommensurate with those of anthropologists” (Myhre, Reference Myhre2006), “The role of the truths that diviners pronounce is not to make a claim about the world but rather to change it – to interfere, in other words, in its ontological constitution” (Holbraad, Reference Holbraad2009), “within the divinatory space, past experience and potential futures are brought together, framed by the querent’s current questions and needs and interpreted through the lens of the present. In doing so, the authentic is resituated; it is not linear but experiential … ” (Sawden, Reference Sawden2018). I submit that these statements can be challenging to comprehend even for professional social scientists, let alone readers without much expertise in recent theoretical developments in anthropology (e.g., postmodernism and the ontological turn). The basic point is really that while divination seeks to generate information, it is not concerned with truth in the correspondence sense – where the truth or falsity of a statement is determined by how accurately it describes the world.Footnote ¹⁸ This is why Holbraad (Reference Holbraad2019) claims that Cuban divination always produces truth by emphasizing the transformative aspect of divination, that divination is “indubitable” because divinatory statements are not merely (or even primarily) representations of the world but help the client reconstruct her experiences in her particular contexts.

4.1 Ethnographic and Historical Evidence for the Instrumentality of Divination

Few would deny that, at face value, the objective of divination appears to be obtaining information. Yet as I have mentioned, many (especially anthropological) accounts of divination have maintained that divination is not really about knowledge acquisition but rather aims to uncover the deeper meaning or hidden function of divination. While these perspectives offer valuable insights, they sometimes overlook or misinterpret the substantial ethnographic and historical evidence that underscores the face-value instrumentality of divination. It’s evident not only in how diviners market their services as crucial information providers but also in the willingness of clients to pay significant fees for these services. Moreover, both diviners and clients are aware that divination may produce inaccurate results and considerable efforts are made to ensure the accuracy of divinatory verdicts. Here, I summarize empirical evidence focusing on two aspects: the use of repetition to enhance the reliability of information and the strategies employed to distinguish competent diviners from potential quacks and charlatans.

In many ethnographic and historical records, we observe that divinatory procedures are repeated until a consistent verdict appears. In a paradoxical sense, such repetition “ensures” that the revealed information isn’t merely due to chance.Footnote ¹⁹ For example, in both ancient China and ancient Greece, military generals would perform divination procedures multiple times before making important battlefield decisions. As Raphals (Reference Raphals2013) notes:

If we interpret the generals’ actions as merely performative or strategic, such repetition would appear puzzling. However, these actions become immediately sensible if we recognize that some generals genuinely wished to follow divine instructions given the great uncertainty and high stakes involved in war. In less serious situations, we similarly observe the tendency to repeat. Among the Nupe people in Nigeria, for instance, diviners in commercial settings might repeat the casting of shells – similar to the Yoruba’s Ifa divination – multiple times, rationalizing that “when you are told something important once only, do you believe it? No – you would wish to hear it twice or three times before you are satisfied that it is the truth. Similarly, you must consult the cords several times.” (Nadel, Reference Nadel1954).Footnote ²¹ My own fieldwork among the Nuosu people in southwest China similarly shows that diviners, when using twig divination to generate a yes/no answer, would repeat the same procedure at least twice to confirm the validity of the verdict (Hong & Henrich, Reference Hong and Henrich2024). In Taiwanese Poe divination, divinatory confirmation typically requires three consecutive positive responses from poe throws (statistically similar to coin flipping), and people are aware that the outcomes of these throws are subject to influences other than godly ones (Homola, Reference Homola, Good, Berti and Tarabout2016; Jordan, Reference Jordan1982). The Quiché Mayan requires the same divinatory procedure to be repeated four times with the same result in determining the causes of illness (Bunzel, Reference Bunzel1952). Perhaps the most striking example of the measures taken to ensure the accuracy of divinatory verdicts is the Azande chicken oracle, as documented by Evans-Pritchard (Reference Evans-Pritchard1937). The Azande pose a question to the oracle and feed fowls (chickens) with poison, observing whether the fowl dies or survives to determine the verdict. Notably, multiple tests were often conducted to confirm the verdict. Evans-Pritchard (Reference Evans-Pritchard1937) provides a concrete example:

This example illustrates the logic of Zande divination: By posing the question in the opposite ways, they confirm the consistency and reliability of the outcome generated by the divinatory procedure. The ingenuity of this setup is that it eliminates the possibility of the poison being too strong or too weak in which case the fate of the fowl would be determined by the physical-chemical properties of the poison itself rather than the oracle which supposedly answers the question through controlling the fate of the fowl.Footnote ²²

In addition to the doubts regarding the reliability of individual divinatory instances, people often hold critical attitudes towards individual diviners. They are acutely aware of the possibility that the works of diviners may be deceptive or fraudulent (Sawden, Reference Sawden2018), and sometimes would frankly acknowledge that many diviners are liars (Evans-Pritchard, Reference Evans-Pritchard1937; Faki et al., Reference Faki, Kasiera and Nandi2010; Mbiti, Reference Mbiti1969). This skepticism drives diviners to demonstrate their expertise by commenting on the client’s past and present situations – remarks that, though merely confirming what the client already knows or can easily verify, serve as “retrodictions” to showcase their supposed insight (Homola, Reference Homola, Good, Berti and Tarabout2016). Clients might even test a diviner’s abilities by asking questions to which only they know the answer or by withholding information, a strategy aimed at validating the diviner’s responses (Homola, Reference Homola, Good, Berti and Tarabout2016; van Beek, Reference van Beek, Olsen and van Beek2015). Heald (Reference Heald1991) documents a case among the Gisu people where a schoolboy suffering from an illness went to a diviner who attempted to elicit information by running through one potential causative agent after another with the associated symptoms:

What’s perhaps more telling is that the client later admitted that he himself suspected that his illness was likely due to witchcraft, and that one of the names that the diviner suggested had indeed been possible. However, he was not going to tell Juma; he wanted Juma to tell him. In Heald’s (Reference Heald1991) words, “the diagnosis had to meet the acid test of plausibility, usually by confirming suspicions already entertained by the enquirer.”

The skepticism toward individual diviners is evident from the common practice of consulting multiple diviners, a practice that is customary and sometimes even recommended. In many African traditions, many clients would “shop around” for suitable diviners (Fernandez, Reference Fernandez1967). Among the Nuosu in southwest China, for example, divination has become a commercial activity in marketplaces, where it is not uncommon for people to seek advice from several diviners (Hong & Henrich, Reference Hong and Henrich2024). This is especially true in medical contexts where patients and their families are eager to restore health. Ample ethnographic records have shown that a second opinion may be sought for additional confirmation (Bunzel, Reference Bunzel1952; Gelfand, Reference Gelfand1956; Mendonsa, Reference Mendonsa and Bharati1976; van Beek, Reference van Beek, Olsen and van Beek2015), or when one diviner’s diagnosis and proposed treatment (usually involving sacrifice to appease the ghosts/spirit) fails (Berglund, Reference Berglund1976; Wilson, Reference Wilson1959). In fact, many medical anthropologists have long pointed out the similarity between divinatory and medical diagnosis (Bunzel, Reference Bunzel1952; Fortes, Reference Fortes1966).

Another type of situation where more than one diviner is needed is where potential social conflict is involved and there is pressure for the verdict to appear objective, such as witchcraft and sorcery accusations (Holleman, Reference Holleman1969; Junod, Reference Junod1927; Turnbull, Reference Turnbull1965). Holleman (Reference Holleman1969) gives a generic description of such practice among the Shona people in Zimbabwe:

Witchcraft accusations carry serious consequences, making it crucial for those involved to be cautious about the final verdict. To avoid potential conflicts of interest, it is not uncommon for an “outside” diviner to be consulted (Turnbull, Reference Turnbull1965, p. 75). More generally, familiarity between the diviners and clients is often seen as problematic because it makes it difficult to distinguish a “genuinely capable” diviner from a quack who appears to offer accurate predictions simply because they know their clients better (also a form of “ostensive detachment”). In eastern and southern Africa, diviners seldom know their clients because “foreign” diviners are intentionally sought to ensure objectivity (Colson, Reference Colson, Swartz, Turner and Tuden1966). For instance, among the Chagga-Rombo people of Kilimanjaro, clients prefer diviners living in villages other than their own, and people may travel considerable distances to see someone in particular (Myhre, Reference Myhre2006). Similarly, the Gisu people in Uganda tend to consult distant diviners who do not know about their cases via local gossip (Heald, Reference Heald1991). These practices highlight the emphasis on the accuracy and objectivity of divination as an information-generating activity.

4.2 The Primacy of Cognition in Understanding Divination

Human actions usually have some foundation in cognition (Comaroff & Comaroff, Reference Comaroff, Comaroff, Moeran and de Wall Malefyt2018). While it is often difficult to distinguish genuine belief from political or social expediency based on manifested action in historical and ethnographic studies (Hankinson, Reference Hankinson1988), to deny the cognitive aspect is to treat people engaging in these actions as mindless zombies, which is not only factually incorrect but in a paradoxical sense disrespectful.Footnote ²³ On the other hand, emphasizing the role cognition plays in divination – and religious actions in general – does not negate the symbolic or functional effects of these practices, but rather highlights that it is inappropriate to completely ignore the cognitive aspects of the actors involved. As Driediger-Murphy (Reference Driediger-Murphy, Driediger-Murphy and Eidinow2019) comments on much contemporary theorization on religious sacrifice in ancient Rome:

Indeed, functional accounts often omit the actors’ emic understanding of their own cultural practices, sometimes to the point of treating such understanding as irrelevant. I would make a further point that functional explanations make sense only in light of the actors’ belief in the efficacy of these practices. This is because the functional consequences of technological practices crucially depend on people’s belief in these practices fulfilling their instrumental goals. In the case of divination, this means that their social and political functions depend on people placing some faith in divination’s epistemic value.

For example, there is ample evidence for the use of divination in military settings in ancient Greece (Anderson, Reference Anderson1970). While one could argue that the generals manipulated omens strategically, such as to launch the attack at the optimal timing (Burn, Reference Burn1962; Hignett, Reference Hignett1963), the utility of such manipulation depends on the soldiers’ belief that reading the body of sacrificial animals is a sensible way of receiving divine messages (the reader could imagine how the soldiers in a secular, modern army would react if a military commander did this). A more dramatic example is the use of inauspicious astronomical phenomena for political purposes in early China. In the Chinese tradition, yinghuoshouxin (Mars at Antares) refers to the astronomical phenomenon where the planet Mars appears to be very close to the star Antares in the constellation Scorpius and was considered an extremely inauspicious sign for rulers, who sometimes would push the responsibility to his ministers (Sun & Kistemaker, Reference Sun and Kistemaker1997). Zhang & Huang (Reference Zhang and Huang1990) analyzed a detailed case where a “Mars at Antares” sign was fabricated in the year 7 BCE (modern astronomical calculations show this phenomenon was impossible that year) during the Han dynasty, and the chancellor was forced by the emperor to commit suicide to claim responsibility. Again, the effectiveness of such manipulation lies in the audience’s recognition of the validity of information generated by divination. Elvin (Reference Elvin1998) makes this point more explicitly when describing how Yongzheng Emperor of the Qing dynasty interpreted an unusual natural sign – the unusually clear flow of the usually muddy Yellow River – to enhance his political legitimacy:Footnote ²⁴

We may never know the emperor’s personal stance on the validity of such signs; indeed, Elvin (Reference Elvin1998) comments that Yongzheng’s own real thought on this matter remains “a mystery.” But what matters for us is the belief of the common people whom the emperor was trying to impress.Footnote ²⁵ Ultimately, we need a theory to explain why a significant portion of the population believed in the connection between the signs and the emperors’ reign.

4.3 The Psychological Basis of Divination

Recall our definition of divination as implausible forms of information technology – given that people seriously intend to employ divination to obtain accurate information, what sustains their belief in these practices if they do not achieve their professed aims? In this section, from a psychological perspective, I discuss both the content-specific factors and general cognitive and cultural-transmission biases that contribute to the belief in the efficacy of divination practices.

Beliefs come in degrees (Frankish, Reference Frankish, Huber and Schmidt-Petri2009; Moss, Reference Moss2018). Probabilistically speaking, our belief in the efficacy of some technology may be viewed as a real number between 0 and 1, where 0 represents the belief that the technology will never work and 1 represents the belief that it will always work. In the most general form, such beliefs may be viewed as having an “intrinsic plausibility” component and an “objective reality” component (Figure 2, adapted from Hong and Henrich, Reference Hong and Henrich2021). Intrinsic plausibility refers to the extent that the mechanism via which the technology is supposed to work fits our larger, background worldview regarding the causal structures of the universe. Importantly, this includes both evolved intuitions and culturally transmitted worldviews – a point we will return to later in the discussion. The implication here is that “intrinsic plausibility” may vary significantly among people with different worldviews. An individual growing up in a society with a polytheistic belief system, where ghosts and spirits are believed to exist and affect the living in concrete ways, will likely find consulting the divine (narrow-sense divination) quite sensible. In contrast, someone growing up in a scientific culture that denies the existence of supernatural entities may find such practices suspicious. The “objective reality” component, on the other hand, refers to the objective (etic) proportion of the time that the desired outcome is produced when the technology is employed. Crucially, this component takes into account the role of “chance”; for example, a divinatory method that randomly predicts the sex of the fetus will still have a roughly 50 percent success rate. Of course, humans cannot directly perceive objective reality and are subject to various cognitive and transmission biases (Nickerson, Reference Nickerson1998; Skowronski & Carlston, Reference Skowronski and Carlston1987; Stahlberg & Maass, Reference Stahlberg and Maass1997).

Figure 2 The composition of subjective perception of efficacy (belief).

This account of belief composition has two advantages. First, it clearly acknowledges that people’s belief in divination – indeed any technological practices in general – is a function of a multitude of factors, and the breaking down of these factors into two larger categories provides greater analytical clarity on the sources of belief-constituting information and their psychological basis. Second, it allows for a more nuanced understanding of how different individuals and cultures may perceive and evaluate the efficacy of the same practice differently, depending on the idiosyncratic or cultural-specific factors that affect the perception of intrinsic plausibility and objective reality. In the rest of the section I will focus on the intrinsic plausibility aspect of divination and leave the cognitive and transmission biases that distort our perceptions of reality in Section 5.

5.1 Efficacy Assessment as a Social Process

As an ultra-social species, we rely heavily on socially transmitted information (Boyd et al., Reference Boyd, Richerson and Henrich2011). Due to the cumulative nature of human cultural knowledge, many cultural products and technologies cannot be fully understood by individuals alone and necessarily require faith and social signals of their utility (Henrich, Reference Henrich2016). In other words, Individuals’ firsthand, personal experiences are often insufficient to form beliefs that guide actions. In divination, this means that in addition to personal experiences, both testimonialFootnote ³⁵ (transmitted stories of divination yielding accurate information) and observational (people turning to divination in cases of uncertainty) information is taken into account when we assess the overall efficacy of some divination method.

Let’s first consider the influence of testimonials using fetal sex prediction as an example. Imagine an individual experiencing a 50 percent (i.e., chance level) success rate of some fetal sex prediction method through firsthand encounters. The individual may nonetheless encounter stories told by others where 90 percent of divination instances are reported as yielding accurate information as a result of selective reporting. To analyze how these two types of information influence an individual’s beliefs, we can apply Bayesian belief updating, a statistical method that updates the probability estimate for a hypothesis as additional evidence is learned. In this context, “priors” refer to the initial beliefs about a parameter or hypothesis before new data are observed, while “posteriors” refer to updated beliefs after new data have been taken into account. From this perspective, integrating firsthand experiences and anecdotal reports will result in a “posterior” belief that lies between 50 and 90 percent, depending on the relative weight assigned to each type of evidence.Footnote ³⁶ This is, of course, not to say that humans are fully rational Bayesian updaters; they are emphatically not (Albert, Reference Albert2009). What matters for our purposes, however, is that the direction of human belief updating should be in line with Bayesian rationality. In other words, we should generally expect individuals’ prior belief of 50 percent due to personal experience to increase rather than decline in response to testimonial evidence of a 90 percent success rate. The preferential reporting of divinatory successes, combined with the well-known confirmation bias where individuals selectively remember and recall information that aligns with their prior beliefs (Nickerson, Reference Nickerson1998), thus could greatly enhance individuals’ subjective belief in divination’s efficacy and create an overwhelming impression of the accuracy of these predictive technologies.

But do divinatory successes really get preferentially transmitted? In a series of studies, I and my colleagues have shown that individuals have a strong tendency to report the outcomes of successful magic and divination rituals (Hong, Reference Hong2022b, Reference Hong2022c; Hong et al., Reference Hong, Slingerland and Henrich2024; Hong & Zinin, Reference Hong and Zinin2023), consistent with research in social medicine where favorable and positive outcomes are preferentially reported (De Barra, Reference de Barra2017; De Barra et al., Reference De Barra, Eriksson and Strimling2014). On the very topic of fetal sex prediction, we performed a systematic examination of documented instances of fetal sex prediction across various genres of historical texts – including oracle bone inscriptions, dynastic histories, encyclopedias, and local gazetteers – and found that the reported predictive success rates are consistently high, around 90 percent (Hong & Zinin, Reference Hong and Zinin2023). A naïve reader exposed to these stories that highlight the efficacy of fetal sex prognostication methods is likely to make the false generalization that these methods are effective, at least in theory.

In addition to testimonial information that enhances people’s estimates of efficacy, observing others engaging in divinatory activities can also increase efficacy estimates through an inference chain: “other people perform divination when they need information → other people believe in divination’s efficacy → divination is efficacious.” The first inference is relatively straightforward: Goal-directedness is a basic psychological tendency, and individuals naturally interpret an agent’s actions as purposefully directed towards achieving a particular objective (Harris, Reference Harris, Damon, Lerner, Kuhn and Siegler2006). More generally, humans are predisposed to interpret other people as intentional agents whose actions aim to accomplish specific goals (Dennett, Reference Dennett1987; Mackey, Reference Mackey, Laes and Vuolanto2016), and the most direct inference from observing people resort to divination is often their belief in its efficacy.

The second inference, however, is less tenable from a strictly philosophical perspective, and adopting beliefs simply because others hold them is often viewed as the antithesis of critical thinking – an Enlightenment ideal that traces back to philosophers like Locke and Descartes (Levy, Reference Levy2022). Yet, as a practical heuristic for mimicking others, this type of inference generally functions well, especially when the actions observed are costly (Henrich, Reference Henrich2009). In fact, our day-to-day functioning heavily relies on our default trust in others – that they are not intentionally deceiving us. In some cases, actions may even carry more epistemic weight than words in influencing our own decision-making, as reflected in the saying “actions speak louder than words.”

We therefore should expect individuals’ subjective belief in efficacy to be further boosted as a result of such observational learning. Note that in our stylized example, we haven’t considered the epistemic influence of intrinsic plausibility; in reality, certain fetal sex prediction methods may align with people’s intuitive understanding of the world (e.g., methods based on yin-yang theory in traditional China) and may additionally enhance individuals’ subjective belief. In Hong and Henrich (Reference Hong and Henrich2021), we formally model the belief updating process within a dynamic population setting where individuals are constantly observing and learning from one another. The main takeaway from the study is that personal experience is just one of many factors affecting individuals’ subjective estimates of efficacy. Once social learning is accounted for, it becomes clear how subjective beliefs can be amplified by various types of transmitted information.

5.2 Human Culture as Imposing “Priors” and Evidential Standards for Divination

In the previous section, I discussed how socially transmitted information, when combined with other types of data, contributes to the belief updating process. However, human culture in a broader sense can also directly influence the intrinsic plausibility of various divination methods through culturally transmitted worldviews, or meta-understandings of the causal structures of the world. In Section 4.3.2, I examined how a holistic worldview may facilitate promiscuous causal learning. I now wish to highlight that in most parts of the modern world, this holistic understanding has been largely supplanted by a worldview that actively denies the causal relevance between events that do not have plausible physical connections (Hong & Henrich, Reference Hong and Henrich2021). It is important to note that this new worldview does not deny promiscuous causality in the absolute sense; indeed, according to Newton’s law of universal gravitation, any two objects are constantly exerting forces upon each other, and such interconnectedness is a fundamental feature of the universe. Rather, it posits that many events and objects are practically causally irrelevant, in the sense that the causal impact may be too weak to be observable or meaningful (Hong, Reference Hong2024). One prominent contemporary critique of astrology, for instance, argues that celestial bodies are too distant to exert any meaningful influence on us (Bok et al., Reference Bok, Jerome and Kurtz1975).

The influence of culturally transmitted worldviews on the plausibility of divinatory methods is also clearly manifested in narrow-sense divination. By definition, the validity of narrow-sense divination depends on a background belief in supernatural entities, and is sometimes further legitimized by the recognition of witchcraft and sorcery as realities that humans have to deal with (Faki et al., Reference Faki, Kasiera and Nandi2010). Surely, there is no point in consulting the divine if such entities do not exist. Though in many ways anthropomorphic thinking and the resultant religious representations may be considered human beings’ cognitive and cultural defaults, with psychologists fond of noting that magico-religious thinking is often present in modern, contemporary humans as it was in the ancients (Nemeroff & Rozin, Reference Nemeroff, Rozin, Rosengren, Johnson, Harris, Johnson and Harris2000; Subbotsky, Reference Subbotsky2010), the reality is that a significant portion of today’s population does not believe in ghosts, spirits, or other supernatural entitiesFootnote ³⁷ (Bullivant et al., Reference Bullivant, Farias, Lanman and Lee2019), and divination has largely disappeared from at least the surface of daily life in modern societies (Burkert, Reference Burkert, Johnston and Struck2005). Our understanding of the world has shifted from a theistic one to a mechanistic one which leaves little room for anthropomorphic deities that communicate with humans (Fourie, Reference Fourie1988).

The profundity of the shift from a holistic, interconnected world where deities interacted with humans to a disconnected, materialistic one where possible superhuman entities are no longer acknowledged and possible causal relationships constrained, cannot be overstated. This transition underpins the intuition behind the etic definition of divination as implausible information technology; as modern observers, certain information-generating methods strike us as implausible because they violate our culturally transmitted assumptions about the ontological structures of the world. Indeed, the decline of divination in many parts of the world must be considered in the context of the diffusion of Western science (Hong & Henrich, Reference Hong and Henrich2021). In both China and Japan, for example, the superiority of Western science and technology was quickly recognized during extensive cultural contact with the West in the nineteenth century (Waley, Reference Waley1958; Hones & Endo, Reference Hones and Endo2006), and many intellectual elites increasingly began to critically assess their own traditions. During the Meiji restoration in Japan, with the establishment of Western style scientific and educational institutions (Bartholomew, Reference Bartholomew1989), many traditional cultural practices and beliefs were labeled “superstitious,” and some were banned at the legislative level (Figal, Reference Figal1999). In China, divination, along with a range of other traditional cultural practices, came to be viewed as an irrational superstition to be eradicated (Zhiwei, Reference Zhiwei2009). Progressive scholars in particular attacked the theoretical basis of divination (and superstitions in general) and often made explicit contrast with modern science. Chen Duxiu, a leader of the New Culture Movement in the early twentieth century in China, famously criticized these practices: “If one believes that science is the compass that points to truth, then things that contradict science like ghosts, spirits, alchemy, talismans, fortune telling, divination, spirit writing, feng shui, and the theory of Yin and Yang are all utter nonsense and absolutely not to be trusted” (Chen, Reference Chen1915).

In addition to blunt attacks, naturalistic explanations were sometimes provided for both why people resort to superstition and the insidious intentions of their practitioners. In a Shenbao article published in 1939, the author made such an argument explicitly:

It’s important to note that the shift in worldview was gradual, and for a considerable time, even the literati continued to engage in various superstitious activities (Xiong, Reference Xiong2015). However, as science became the dominant framework to understand the world, divination and other superstitious practices were increasingly marginalized (Li & Lang, Reference Li and Lang2012). In the Republic of China era, such marginalization was sustained by state-level legislative efforts and educational programs aimed at eradicating superstitions. Materialistically, this was most salient in the destruction of temples or their repurposing into schools or secular public venues, as well as regulations outlawing ritual activities (Zhu, Reference Zhu2013). Although we do not have definitive evidence of the extent to which people’s beliefs were affected and much literature on this topic tends to emphasize how folk religious practices persisted in such anti-superstition movements (Katz, Reference Katz2013), it is undeniable that divination, magic, and other superstitious practices have since been relegated to the margins of societies. In the West, such transition occurred much earlier (Thomas, Reference Thomas2003) and these “past irrationalities” were already treated as ancient survivals and documented by folklorists during Victorian times (Walsham, Reference Walsham2008).

The significant cognitive consequence of this worldview change is elegantly summarized by the sociological concept of a “disenchanted world,” originally proposed by Max Weber (Greisman, Reference Greisman1976). Here, “disenchantment” refers to the devaluation of religion and the emphasis on rationality, characteristic of modernized, bureaucratic, secularized Western societies, which contrast sharply with traditional societies where “the world remains a great enchanted garden” (Weber, Reference Weber1922). Although recent scholarship has challenged the Weberian thesis, arguing either that magic has survived disenchantment (Hanegraaff, Reference Hanegraaff2003) or that the world is experiencing a re-enchantment (Landy & Saler, Reference Landy and Saler2009), we can’t ignore the brute fact divination and magic have largely lost the cognitive appeal that they once had for a significant proportion of humanity, as people become increasingly skeptical towards the postulated superhuman entities and the presumed causal mechanisms that magicians and diviners rely upon. A person with a scientific outlook does not need empirical data to be deeply suspicious of the claim that illnesses can be diagnosed by examining the holes thigh bones of a sacrificed chicken (Hong & Henrich, Reference Hong and Henrich2021). The very features that once bolstered the plausibility of divinatory practices now often raise red flags in the modern context.

In addition to affecting the intrinsic plausibility of divination through modifying the “priors,” a scientific culture also influences how we recognize, collect, and process evidence in subtle yet significant ways. Modern societies feature a distinct division of labor in knowledge production and transmission: scientists as the producers and laypeople as the consumers of knowledge. Scientists systematize personal and anecdotal experiences, transforming personal experiences into randomized, controlled trials and anecdotes into meta-reviews and meta-analyses, and in doing so largely avoid the underreporting of negative evidence and other biases. As a result, for lay people, accepting expert opinion is a far more dependable route to truth since our epistemic environment is structured to be more reliable (Levy, Reference Levy2022). Additionally, even those not professionally involved in science develop an understanding of reliable knowledge production; for instance, one does not need to be a scientist to know that the evaluation of the efficacy of drugs requires randomized, controlled trials (minimally, some kind of experimentation), or that a neighbor’s anecdotal story of her horoscope correctly predicting her personality does not qualify as evidence for the validity of astrology (Hong & Henrich, Reference Hong and Henrich2021). More crucially, the very concept of “chance-level performance” often requires a substantial amount of statistical education, and people without such concepts may not even consciously consider whether the perceived efficacy of some divinatory practice exceeds chance. For example, many Nuosu individuals with little education in contemporary southwest China think that a diviner with a 50 percent success record in fetal sex prediction is demonstrating mediocre to good ability, not recognizing that this is merely the expected success rate for random guessing (Hong, Reference Hong2022c). In contrast, a modern reader with some understanding of statistics would easily recognize that a 50 percent success rate for an outcome with dichotomous, equiprobable options merely matches the expectations of random chance, indicating no actual divining ability. Without such a conceptual benchmark for evaluating efficacy, divination methods that produce correct outcomes at sufficiently high frequency due to natural stochasticity may persist as they appear to “work” from time to time.

It is worth noting that even systems explicitly designed to minimize false positives, such as modern science, can become unreliable when safeguards are weakened. Practices like p-hacking and other questionable research practices (QRPs) – such as selectively reporting results or testing multiple hypotheses without corrections – can inflate the likelihood of statistically significant but unreliable findings (Fiedler & Schwarz, Reference Fiedler and Schwarz2016). P-hacking, which involves manipulating data analysis to achieve statistically significant results, often produces findings that appear meaningful but fail to replicate, undermining the credibility of the research (Head et al., Reference Head, Holman, Lanfear, Kahn and Jennions2015). However, the scientific community is acutely aware of these issues and actively works to address them through initiatives such as open science practices (Banks et al., Reference Banks, Field and Oswald2019), preregistration of studies (Nosek et al., Reference Nosek, Ebersole, DeHaven and Mellor2018), and replication efforts (Lindsay, Reference Lindsay2015). These measures aim to reduce biases and improve the reliability of scientific findings. Such awareness and commitment to improve integrity and reliability contrast sharply with divinatory practices, where mechanisms for evaluating validity are often absent.

6.1 The Precision versus Accuracy Trade-off in Divination

In Section 4.3.3, I highlighted how the vagueness of divinatory verdicts, such as the Barnum effect in astrology, may enhance people’s belief in the efficacy of divinatory practices. The advantage of vague predictions is clear: They evade falsification, allowing clients to affirm their existing trust in the diviner’s competence or the divination method’s validity. However, ethnographic and historical evidence suggests that clients are often more impressed by precise predictions.

In this context, “precise predictions” refer to what philosopher of science Karl Popper describes as “risky predictions” – specific and bold forecasts that risk being disproven (Popper, Reference Popper2005). Ethnographic studies highlight a preference among clients for diviners and divination systems that can provide more specific answers (van Beek et al., Reference Van Beek, Blakely and Thomson1994, p. 221). These diviners sometimes demonstrate impressive accuracy, revealing knowledge of clients’ personal experiences that are so private and rare that it is unlikely for a stranger to correctly guess them. In contemporary Chinese divination, clients would often acknowledge that they are impressed by diviners’ precision, saying things like “ … the diviner said I broke my elbow when I was five years old. I didn’t remember that and doubted his words but I was very surprised when my mother later confirmed it” or “the diviner claimed I had a mole on my back. This is amazing, as even my wife had not noticed it” (Li, Reference Li2018). As mentioned, these insights typically do not directly facilitate decision-making but rather serve to showcase the diviner’s expertise or skill. An extreme example can be seen in Bourdillon’s (Reference Bourdillon1976, p. 175) description of a good diviner among the Shona people:

Conversely, vague predictions can be frowned upon. During fieldwork among the Nuosu in southwest China, I encountered complaints about diviners who make broad, nonspecific predictions that could apply to anyone – such as having red clothes or experiencing headaches – leading to suspicions of deceit (Hong & Henrich, Reference Hong and Henrich2024).

The intuition is straightforward. A diviner who offers more specific predictions, such as forecasting a misfortune in January rather than just “next year,” when correct, seems more impressive and trustworthy. Yet, precise predictions are inherently riskier and more likely to fail. While many divinatory failures are rationalized away (which relates to the second outstanding question), they nonetheless can diminish confidence in the diviner’s skills.

Consequently, a trade-off emerges between precision and accuracy in divination. Diviners aim to showcase their skill by offering precise predictions, yet also strive to enhance their accuracy by crafting predictions that are broad enough to withstand falsification. Ideally, both precision and accuracy are desired, but the nature of divination as an implausible information technology necessitates a compromise. Could there be some optimal balance where accurate predictions are precise enough to impress, yet the number of incorrect predictions doesn’t undermine trust excessively?

It is unlikely that there will be a single “sweet spot” across the board, but exploring the trade-off between precision and accuracy in divination may reveal the complex interplay between client expectations and diviner strategies. The preference for precision, while potentially enhancing the reputation of a diviner when predictions are correct, comes with heightened risks of disconfirmation. Vague predictions, on the other hand, minimize the risk of being proven wrong but may not sufficiently impress clients to inspire confidence or repeat consultations.

This dynamic raises several important questions for future research. First, what mechanisms do diviners employ to navigate this trade-off? It would be beneficial to investigate whether diviners consciously adjust the level of precision based on factors such as the perceived skepticism of the client, the seriousness of the inquiry, or their own confidence in the information at hand. Second, how do clients’ experiences with both types of predictions influence their long-term beliefs about divination? Understanding this could illuminate the processes through which cultural beliefs about divination are reinforced or challenged over time.

Additionally, the impact of cultural context on this trade-off is profound. In cultures where divination is widely accepted and integrated into the daily life, the balance might lean more towards precision to maintain and enhance the diviner’s status, especially given that divinatory failures tend to be underreported (see Section 5.1). Conversely, in more skeptical environments, the strategy might shift towards safer, more ambiguous predictions to avoid direct challenges and maintain a baseline level of client engagement.

The precision versus accuracy trade-off in divination is not just a theoretical concept but a practical challenge that diviners navigate through every interaction. This ongoing negotiation between diviner and client, shaped by broader cultural and social dynamics, continues to sustain divination as a fascinating subject of study in understanding human cognition, culture, and social interaction. Future research in this area can provide deeper insights into the cognitive underpinnings of belief systems and the complex ways in which humans generate and interpret information about the world.

6.2 Divinatory Failures and the Limit of Post-hoc Rationalization

In the previous section, I briefly touched upon how divinatory failures are often rationalized to protect the core belief in divination’s efficacy. Such post-hoc (after the fact) rationalizations are ubiquitous across cultures and historical periods. The key idea is that there are always factors that can be invoked to account for failed predictions without challenging the legitimacy of divination itself. Keith Thomas (Reference Thomas2003) famously argued that no amount of failed predictions would shake the faith of true believers because they can always rationalize these failures within their existing belief systems. Evans-Pritchard (Reference Evans-Pritchard1937) documents a similar mode of thinking among the Azande, where individuals were acutely aware of the myriad of factors that could lead to an incorrect verdict in chicken oracle: wrong variety of poison, age of the poison, breach of a taboo, anger of the ghosts, or sorcery (Evans-Pritchard, Reference Evans-Pritchard1937). Evans-Pritchard refers to the ways in which people try to justify and explain the inconsistencies and contradictions within their belief systems “secondary elaborations,” a phenomenon that is prevalent and has been noted by ethnographers (Jordan, Reference Jordan1982). More often, the blame shifts to individual diviners, who are usually considered fallible. The working assumption of the clients is typically that there is, in theory, a correct answer that could be revealed by divination, but that not all diviners are equally skilled to be able to (Bascom, Reference Bascom1941). As a result, clients may suspect individual diviners, but seldom the system as a whole.

In the philosophy of science, these additional factors that account for empirical failures of some theoretical prediction are termed “auxiliary hypotheses” (Hempel, Reference Hempel1966) and have long worried philosophers because they suggest a slippery slope toward unfalsifiability (Harding, Reference Harding1976). Gershman (Reference Gershman2019) recently proposed a Bayesian solution to this conundrum: Observations that appear to contradict a central hypothesis can, under the right circumstances (e.g., if the prior belief in the core theory is sufficiently strong), be “explained away” by changing auxiliary hypotheses in a Bayesian rational manner. In other words, it may be perfectly rational to look for “excuses” instead of abandoning the theory when an unexpected outcome occurs. Consider the discovery of Neptune as an illustration: When anomalies in Uranus’s orbit contradicted Newtonian physics, the hypothesis of an unseen planet was posited rather than discarding Newtonian principles, leading to Neptune’s discovery in 1846 (Smart, Reference Smart1946). Indeed, such rationalization is common in contemporary scientific discourse (Blanton et al., Reference Blanton, Jaccard, Christie and Gonzales2007) and everyday reasoning (Cushman, Reference Cushman2019; Summers, Reference Summers2017), ensuring theories are not discarded prematurely.

What does this mean for our understanding of divination? The existence and utility of auxiliary hypotheses in science suggest that employing them to explain inconsistencies in divination isn’t inherently irrational. However, Gershman’s Bayesian analysis indicates that even when these hypotheses absorb most of the blame, the core theory still suffers epistemically. In other words, if we think our belief in the efficacy of some divinatory method as a value between 0 and 1, auxiliary hypotheses such as the incompetence of the individual diviner or the insincerity of the client can never fully protect the core theory that the divination method generates the truth, and our belief in its efficacy should always decrease with each failure, even if only slightly. Over time, this can lead to a cumulative decline in belief. This poses the question: Is there a threshold of disbelief, or must skepticism and change come from outside the belief system itself? Evidence on this topic is mixed. On one hand, numerous observations suggest that faith in divination and religious practices is often insulated from contradictory evidence. Beyond the anecdotal evidence from ethnography, which shows that predictive failures are frequently rationalized, there exists a body of research in sociology and social psychology focused on how religious cults handle failed prophecies. In a seminal study by Festinger et al. (Reference Festinger, Riecken and Schachter1956), the leader of a cult predicted a catastrophic flood that would destroy much of the United States, with followers to be rescued by aliens. After the prophecy failed, rather than abandoning their beliefs, a small core of the group intensified their proselytizing efforts. Follow-up studies have generally supported these findings, indicating a pattern where beliefs persist despite disconfirming evidence (Dawson, Reference Dawson1999). Some philosophers propose a more radical view, suggesting that religious beliefs function as forms of make-believe that are fundamentally different from factual beliefs and thus impervious to empirical counter-evidence (Van Leeuwen, Reference Van Leeuwen2014).

On the other hand, some research suggests that religious adherents often exhibit significant disappointment at failed prophecies, and in some cases, these disillusionments may lead to the disintegration of the group (Stark, Reference Stark1996). For example, many ethnographers and historians have attributed the decline of the Ghost Dance movement among the Cherokee in the 1800s to unfulfilled predictions, such as a hailstorm that was supposed to destroy the earth and threats of death to nonbelievers, which never materialized (McLoughlin et al., Reference McLoughlin, Conser and McLoughlin1984, p. 113). In the Millerite movement in nineteenth-century America, similarly, followers expected the world to end on a specific date based on prophetic calculations. When the world did not end – a day referred to as the “Great Disappointment” – the movement experienced profound turmoil, with many members departing in disillusionmentFootnote ³⁹ (Knight, Reference Knight1993). Most divinatory verdicts, of course, are not as dramatic as prophecies, but we’d nonetheless expect them to influence our beliefs in some way. In fact, there are theoretical reasons to expect divinatory failures to carry significant epistemic weight: Researchers in social psychology have identified a “negativity bias,” where humans have a propensity to give more weight to negative information than to positive information in contexts such as news selection (Soroka et al., Reference Soroka, Fournier and Nir2019) and impression formation (Skowronski & Carlston, Reference Skowronski and Carlston1989).

The manner in which believers cognitively process counter-evidence has profound implications not only for divination but also for instrumental practices in general. While instrumental religious actions such as rainmaking and fetal sex prognostication occasionally prove successful, what happens in scenarios where failures significantly outnumber successes? Could the cumulative impact of these failures gradually erode faith in the system?Footnote ⁴⁰ Should we expect a cultural evolutionary process in which only those practices and belief systems that effectively withstand counter-evidence survive? Furthermore, if a negative bias is present in divination, how does it reconcile with the selective reporting of positive outcomes, as discussed in Section 5.1? Future research could explore the dynamic interaction between belief change and post-hoc rationalization, in particular, possible thresholds at which repeated divinatory failures begin to outweigh the confirmation and reporting biases that typically maintain such beliefs. Understanding these dynamics could provide deeper insights into how cultural and belief systems evolve or dissolve over time, especially in the face of contradictory evidence.

6.3 Looking Ahead: A Cognitive Theory of Human Technology

Many of the characteristics of divination and features that could enhance people’s faith in it in this Element are not exclusive to divination per se. In fact, this is necessarily the case given that divination is not qualitatively different from other ordinary information technology in my proposed commonsense, cognitive framework. Thus, I challenge theories that claim divination possesses a unique rationality (Rapisarda, Reference Rapisarda, Herbers and Lehner2022; Zeitlyn, Reference Zeitlyn2012). All technological practices combine aspects of intrinsic plausibility with elements of objective reality, and in competitive market environments, practitioners must effectively signal their competence to attract clients. What demarcates divination from ordinary information technology – their “supernaturalness” so to speak – is our worldview which no longer acknowledges anthropomorphized deities and a deeply interconnected universe.

The implication is that the theoretical framework and tools that we employ to make sense of ineffective technologies such as divination and magic can also be used to explain the effective ones. For instance, financial analysts and economists are sometimes sarcastically compared to shamans when they fail to accurately predict market movements (Caldararo, Reference Caldararo2008; Coggin, Reference Coggin2006). While the original intent of these comparisons may be to highlight the objective ineffectiveness of both financial forecasting and shamanistic divination, the more relevant point for our discussion is their structural similarities as information technologies. Both involve elements of intrinsic plausibility, though these vary significantly depending on the observer’s worldview. Unlike divination, which gains plausibility through the assumption of divine influence and a holistic worldview of an interconnected universe, financial analysts in contemporary societies are trusted largely due to our familiarity with the capitalist economic system and our constant interaction with numbers and probabilities. Financial forecasting is further legitimized by endorsements from academic and research institutions, similar to how traditional divination methods draw legitimacy from ancient wisdom. Meanwhile, individual financial analysts, much like diviners, face skepticism when their predictions fail. To explain these failures, analysts might cite limitations in their models, technical issues, or unforeseen market events, mirroring how diviners might attribute their inaccuracies to violations of taboos or sorcery influences.

Can there be a cognitive theory of human technology? I think yes, and I envision such a unified theory would not only elucidate not only the structural parallels between divination and modern technologies like financial or weather forecasting but also offer insights into the cognitive processes that lend both ancient and contemporary technologies their perceived efficacy. By understanding these cognitive underpinnings, we can better comprehend how certain technologies – whether labeled as divinatory or scientific – maintain their credibility and influence across various cultures and historical contexts.