Cognitive Science of Religion

D Jason Slone. Science, Religion, and Society: An Encyclopedia of History, Culture, and Controversy. Editor: Arri Eisen & Gary Laderman. Volume 2. Armonk, NY: M.E. Sharpe, 2006.

The cognitive science of religion is a new approach to the scientific study of religion. Cognitive science is the set of disciplines that investigate the mind-brain processes involved in human thought and behavior. Scholars in the cognitive science of religion explain features of religious thought and behavior that recur across cultures and eras in terms of the mental processes involved in their production and transmission. Religion, along with other forms of culture like music, art, and literature, is understood by cognitive scientists as a natural by-product of the ways our mind-brains function.

Importantly, cognitive scientists of religion are primarily interested in explaining religion, which they distinguish from theology. Religion is the set of actual religious concepts people have in their heads and behaviors they perform. Theology is the set of creeds clergy instruct people to believe. Also, the cognitive science of religion is methodologically agnostic in nature and universal in scope. We are interested in neither the veracity of theological claims nor the religious experiences of mystics. What is of primary interest to cognitive scientists of religion is explaining why most people in most places at most times have strikingly similar types of religious thoughts in their heads.

The cognitive science of religion began with the publication of E. Thomas Lawson and Robert N. McCauley’s Rethinking Religion: Connecting Cognition and Culture in 1990. Its foundations, however, are much older. The broad theoretical framework of cognitive science was formally established in the 1950s by scholars like George Miller, Noam Chomsky, Allen Newell, and Herbert Simon. These scholars challenged the dominant theory in psychology at the time, behaviorism, which avoided mentalist constructs—thoughts—in explanations of behavior on the grounds that human thought processes could not be, and need not be, studied scientifically. The early cognitive scientists argued that it was both necessary and possible to do so to fully understand human behavior. This argument was based in part on the success of mentalist theories in computer science, linguistics, and other fields.

The article usually credited for starting the “cognitive revolution” in the human sciences is Miller’s 1956 “The Magical Number Seven, Plus or Minus Two.” Through simple recall experiments, Miller showed that people could overcome the short-term memory constraints of 7 + 2 bits of data at a time by “recoding” those bits into smaller “chunks” of data. To illustrate, you can more easily remember the string of numbers 1492177618652005 by chunking them into 1492, 1776, 1865, and 2005. Using similar types of recall experiments, Miller showed in a scientific way that the reorganization processes of memory could not be accounted for unless one invoked mental constructs like the processes of “recoding.” Thus he showed that the scientific study of mental constructs was both necessary and possible for understanding human thought and behavior.

The dominant model in the cognitive sciences today is the computational-representational model of mind, which imagines the mind as an information-processing computer. But some evolutionary psychologists, cognitive scientists whose theories are informed by the selectionist framework of evolutionary biology, have gone further and theorized that the mind is composed of numerous “modules,” each of which is an evolutionary adaptation that solved a particular problem in our species’ past. Such modules include a module for perception, a module for detecting social cheaters, a module for acquiring language, and so on. These modules are theorized to be hardwired in our minds, and include a host of representational information about the world that gets triggered by environmental cues.

The modularity model allows scholars studying cultural phenomena to employ an epidemiological framework to explain how certain cultural concepts become widespread in populations. All human societies, as far as we know, contain similar types of art, music, religion, and other cultural forms. The reason is that these cultural forms trigger cognitive systems (designed for other tasks) and therefore are good “fits” for consumption. Music, for example, is like auditory cheesecake because it fits so well with various cognitive systems. The sounds are pleasing to hear and the stories that the lyrics tell are pleasing to follow, even though our cognitive “sound systems” and “story systems” evolved to solve adaptive problems, not for acquiring music per se. Such cultural forms spread easily from mind to mind across populations, and in turn are, in evolutionary terms, “selected for” over generations. Similar to music, the same types of religions show up all over the world because they are also good fits for cognitive consumption.

Most cognitive scientists of religion adopt at least a weak version of the modularity model. The mind is not, most believe, a single all-purpose, problem-solving organ but rather is composed of various cognitive “systems” that are designed to solve particular problems. Importantly, religion is not a cognitive adaptation itself; there is no “religion module” or “religion system” in the mind. Rather, religion is a by-product of cognitive systems that have other primary functions. The central goal of the cognitive science of religion is, therefore, to connect the recurrent features of religion that anthropologists and historians of religion have documented across cultures and eras to the cognitive processes involved in their production and transmission.

An important first step in cognitive explanations of religion is to understand that “religion” is a collection of features that coalesce, more or less, into (symbolic) cultural systems. Most religions include a belief in supernatural agents such as gods, goddesses, angels, and demons. They also include ritual actions in which people traffic with those supernatural agents; ethical mandates that regulate behavior; social commitments that result in the separation of people into in-group and out-group members; afterlife beliefs and death rituals; metaphysics that provide teleological explanations for personal and world events; and in religions with official theologies, attempts to centralize beliefs and practices across disparate populations. These features of religion all trigger cognitive systems that evolved for other purposes, namely to solve problems posed by our ancient evolutionary environment (including problems of group living). The cognitive explanations for these seven features of religion are discussed in the sections that follow.


The central feature of religion is the belief in supernatural agents. We will call them “gods” in this discussion, just for convenience, but the label includes goddesses, ghosts, demons, saints, sorcerers, and so forth. Despite their apparent differences, gods are represented quite similarly across cultures. That is, there is a limited catalogue of god concepts that recur. Cognitive scientists of religion argue that this is due to the constraints of our “intuitive ontology.”

An intuitive ontology, or folk understanding of what things in the world are like, develops very early in life and without explicit cultural instruction. For example, our intuitive ontology allows us to distinguish “agents” and “objects.” We know that objects (e.g., rocks and chairs) don’t move on their own, don’t need to eat, and don’t respond to communicative gestures. In contrast, we know that agents (humans and many animals) move on their own (and their behaviors are goal driven), need to eat, and respond to communicative gestures. The most important difference between the two, however, is agents have minds that govern their behavior and objects don’t.

More specifically, objects and agents can be classified into four categories. Objects are either (1) natural objects (e.g., rocks) or (2) artifacts (e.g., statues), and agents are either (3) animals, or (4) humans. Each category contains a host of tacit information about the properties of its members, information that delineates what a particular thing is. In other words, the tacit information about the properties of each domain’s members is “default” information. For example, as described by Hirschfeld and Gelman, tacit information about rocks is that they are solid, don’t have minds, don’t eat, and won’t move unless pushed by another (solid) object. Importantly, this ontological knowledge is part of our natural cognition. Intuitive ontology is not acquired from culture.

God concepts are constrained by this intuitive ontology in the sense that they violate a small number of the intuitive default expectations about agents. They either breech expectations of a single category or they transfer expectations from one category to another. For example, gods are persons that can see everywhere at once, a concept that breeches our expectations about human vision. And a talking statue transfers expectations from the category of “person” (can talk) to the category of “artifact.” God concepts are not, it turns out, supernatural; they are just modestly counterintuitive.

It is important to emphasize that successfully transmitted concepts, religious or otherwise, tend to contain only modest (vs. minimal [one] or radical [many]) violations of intuitive ontology. Modestly counterintuitive concepts are attention grabbing yet retain all of the other expectations of natural kind concepts, making them easy to represent. The more violations a concept contains, the more counterintuitive it is and thus the harder it is to represent.

Cognitive scientists who study memory have shown that humans are better at remembering intuitive and modestly counterintuitive ideas than maximally counterintuitive ideas. These findings suggest that the types of god concepts that recur across religions constitute a memory optimum. God concepts are attention grabbing but easy to process, represent, and store.

In addition to being attention grabbing and believable, god concepts are also salient because of their personal and social relevance. By having powerful epistemic abilities, or super-knowledge, gods are represented as having full access to what we think and do. As discussed by Boyer, this feature of god concepts triggers important biases in our “social intelligence” system.

Our social intelligence system evolved to handle the various adaptive problems posed by small-group living (fewer than 150 people) in ancient evolutionary environments where resources were scarce. In such environments, your social reputation was critical to survival and successful mating. If you became known as selfish or untrustworthy (a “cheater”), you risked not having allies or mates. In contrast, if you developed a reputation as generous and trustworthy (a “cooperator”), you improved your chances of gaining allies and mates. What other persons knew about you was of vital importance because a great deal of your fate rested in the hands of others. Someone who possessed strategic knowledge of your thoughts and behaviors held great power over you. Pleasing that person would be of great benefit, while displeasing that person could result in great costs either directly (a potential mate turned you down) or indirectly (that person spread knowledge about your “bad character” to others).

God concepts trigger this system because gods are represented as “full access agents.” Gods know everything you think and do, even when you (think you) are alone. This postulation has enormous relevance for people, for obvious reasons. Gods are like Santa Claus, whose relevance is captured in the Christmas song “Santa Claus Is Coming to Town”:

He sees you when you’re sleeping.
He knows when you’re awake.
He knows if you’ve been bad or good,
So be good for goodness sake.

Cognitive scientists of religion think that god concepts recur across cultures because of the personal and social relevance their postulated epistemic powers trigger.


The cognitive system that underpins ritual performance is an “action representation system.” This system represents rituals as an agent performing an action on a patient: agent → action → patient

Religious rituals conform to this structure, with the additional variable that gods serve as either the agent or patient of a ritual, or gods are instrumental in the action:

gods → action → patient
agent→ action w/ gods’ instrument → patient
agent → action → gods

The constraints of this action representation system mean that there are only three types of ritual actions possible. Lawson and McCauley call these “special agent,” “special instrument,” and “special patient” rituals:

  1. Special agent rituals : gods are the agents and people are the patients. Examples include weddings and baptisms.
  2. Special instrument rituals : agents use an instrument that represents the gods on patients. An example is a blessing using holy water.
  3. Special patient rituals : gods are the patients and people are the agents. Examples include taking communion and making offerings.

Gods, of course, do not actually perform or receive rituals “in person.” Clergy, as representatives of gods (a status obtained via ordination rituals), act on their behalf.

Typically, special agent rituals are performed relatively infrequently, and they are surrounded with relatively high levels of “sensory pageantry.” That is, they are usually performed only once in a lifetime to the patient, are exciting, and involve significant material accoutrements. In contrast, special patient rituals are performed frequently and have low levels of sensory pageantry.

This “ritual form hypothesis” makes additional predictions, for example, about the potential longevity of religious systems altogether. Religions with ritual systems that are “balanced”—contain enough special patient rituals to keep folks busy, punctuated by the occasional special agent ritual to generate excitement—are likely to be long-lived. In contrast, religions with ritual systems that have no special agent rituals are “unbalanced” and as such are susceptible to schisms, if not outright dissolution. New religious movements are particularly interesting test cases for this hypothesis, as those that survive long enough to become part of mainstream society are likely to have balanced ritual systems. Those that do not are likely to have unbalanced ritual systems. In this way the ritual form hypothesis provides a kind of evolutionary explanation for which religions survive.


Cognitive scientists of religion believe that ethical mandates recur across religions because they trigger our intuitive “morality” system. This morality system develops naturally in (normal) human beings. Religious ethics do not create morality; they trigger the natural moral sense.

The moral sense is governed by the social intelligence system. As noted above, the social intelligence system is designed to detect and punish “cheaters” (those who steal, are aggressive, don’t share) and to detect and reward “cooperators” (those who don’t steal, are not aggressive, share). Whether or not an act is moral depends on its social impact. For example, honesty is generally good because it encourages trust by eliminating the problem of deception. But even some lying is acceptable if no one is harmed: “Yes, Virginia, there is a Santa Claus.”

Importantly, human ethical judgments are not necessarily conscious deliberations. Instead, humans possess a moralsense. Our ethical judgments are based on “gut feelings” about whether acts are good or bad, feelings that are reinforced by emotions like guilt (which follows from having done something wrong, even if never detected), joy (which comes from giving to others), anger (at cheaters or their injustices, even if they have no impact on your life), and other emotions. If they do so at all, people search for ethical creeds post facto to justify their intuitive feelings about morality. For this reason, ethical creeds are culturally selected, based on how well they map onto our moral sense.

Religious ethics, at least those found in most long-lived religions, tend to map neatly onto this intuitive moral sense. Not coincidentally, what religious ethical systems usually deem “good” tends to involve socially desirable behaviors and avoidance of socially undesirable behaviors. Religious ethical systems benefit society by encouraging good acts by individuals, who are rewarded for being good with improved chances of gaining mates later on (by signaling that they are trustworthy cooperators). Thus the social nature of morality is ironic; being good is an effective strategy for our species’ “selfish genes.”

Finally, the socially desirable behaviors our moral-sense-based religious ethics motivate result in social “cohesion.” The social commitments that religious ethics generate might be so strong, in some cases, as to trigger the kinship selection biases that are normally reserved for genetic relatives. Religions function as fictive families, which is evident by the use of familial language in many religions—brothers and sisters of Islam; God the Father and Jesus the Son; Kali, the Mother Goddess of Shakta Hindus. Regardless, the same types of religious ethical schemes recur across religions because they trigger our moral sense and thereby help to maintain group solidarity by encouraging socially desirable behavior and discouraging socially undesirable behavior.

Sects: In-Groups and Out-Groups

Religions tend to be “sectarian.” That is, most religions claim that they are the one true faith and that others are false. Committing to a sect, therefore, involves particular judgments about the “rightness” of one group over another. Membership in such groups is gained through costly social commitments, both resources (e.g., tithes) and behaviors (e.g., circumcisions) that signal one’s choice to commit to cooperate with the sect. Such “in-group” representations in turn trigger the representation, sometimes explicit other times implicit, that outsiders are foes. This insider-outsider feature of religion explains the seemingly paradoxical (given the “golden rule” to treat others as you wish to be treated) nature of religiously motivated violence toward members of other religions. Cognitive scientists of religion believe that sectarianism triggers our cognitive system of “coalitionism.”

Humans are naturally coalitional, which is explained by the theory of “inclusive fitness” discussed by Hamilton. According to inclusive fitness theory, individuals act in ways that promote not only their own well-being but also the well-being of those with whom they share genes, their kin. Additionally, people categorize nonkin into one of two groups, allies or foes, based on their potential to help or harm. Allies are those we trust to be cooperators, and foes are those we distrust as cheaters or predators. Thus we possess an evolved predisposition for creating coalitions that include kin and allies.

There is, however, a computational difficulty in judging whether another person is trustworthy or not—the problem of deception. Because a person’s “character” must be inferred from her or his behavioral patterns, we are susceptible to being deceived. Cheaters can perform a few good deeds as a strategy for winning you over temporarily in order to steal from you later. To manage the problem of deception, people stereotype others through the processes of “essentialism” and “group entativity.” Essentialism involves representing people as having an “essence,” either good or bad, which governs their behavior and determines whether they are trustworthy or not. Group entativity involves representing all members of a group as possessing the same essence. Thus a single person shares the essence of the group, and the group shares the essence of the person. For examples, Hirschfeld mentions “men are jerks” and “Muslims are terrorists.”

Combined, essentialism and group entativity lay the foundations for racism, sexism, ethnocentrism, and religious bigotry. As noted above, religions trigger our kinship selection biases by, for example, using familial language. Non-relatives in the sect become “brothers and sisters”—a designation that is purchased with costly acts of commitment. On the other hand, the representation of other sects as out-groups triggers the fear of foes that can’t be trusted. In cases of intergroup conflict, enemies are even represented as different species altogether, being dehumanized as “evil beasts,” “pigs,” “snakes,” “dogs,” “rats,” and so forth. Cognitive scientists of religion believe that religions are sectarian in such ways because they trigger our natural predisposition toward coalitionism.

Afterlife Beliefs and Death Rituals

Most religions claim that death is not final but rather is a transition to existence somewhere else. The cognitive systems involved in afterlife beliefs are mental essentialism and object permanence. The belief in afterlife recurs across religions because humans have difficulty representing the psychological cessation of agents. We think that if an agent exists, it must always exist.

The developmental psychologist Jean Piaget theorized that humans develop the capacity to know that things exist even if we can’t see them. He called this capacity “object permanence.” Object permanence develops in the first year of life (thus accounting for toddlers’ fascination with the games of “peek-a-boo” and “hide and seek”), long before the acquisition of religion. Thus, the theory goes, the biases about objects’ permanence that stem from this capacity result in the representation that people don’t die per se. Instead, dead people just go somewhere else.

This Piagetian theory is insightful but not entirely accurate. Recently, psychologists have refined this theory by showing that people don’t represent agents as singular units. Rather, we are natural “dualists.” We represent people as being composed of two substances, a body and a mind. And among the two, we tend to privilege the mind as the real essence of what makes a person a person. The insight that persons have bodies and minds, but that the mind is the “essence” of a person, better explains the precise types of afterlife beliefs that recur across cultures and eras. When people die, their bodies might cease but their souls, or minds, live on.

This neo-Piagetian theory has not only empirical support from world religions, but also experimental support. In a particularly clever psychological experiment, children saw a puppet show in which a mouse became lost from its family and at the end of a long, hard day was eaten by an alligator. After making sure that the children understood that the mouse was dead, the experimenters asked the children questions about what was happening—if anything—to the (dead) mouse. The results were striking. When asked questions about the mouse’s biological functioning (e.g., “Will the mouse eat dinner tonight?”), the children overwhelmingly responded, “no.” However, asked questions about the mouse’s psychological functioning (e.g., “Is the mouse angry at the alligator for eating it?”), most children responded, “yes.” The results, as pointed out by Bering and Bjorklund, suggest that children intuitively represent death as the cessation of biological functioning but not psychological functioning. In our minds, the mind lives on.

If the mind lives on, what happens to the body? In most cultures, we dispose of dead bodies. In most cases, the disposal is performed publicly, in burial or cremation. Boyer has theorized that death rituals are widespread because dead bodies trigger our “toxin avoidance” system. The human toxin avoidance system involves feelings of “disgust” toward potentially toxic substances. Substances that trigger very strong feelings of disgust are feces, rotting meat, and dead bodies, in part because the toxins in these substances are invisible and can infect even in the smallest of doses and through various means of contact. We evolved, presumably, a hypersensitive disgust of these substances because the risk is so great. Thus, the disposing of dead bodies publicly is an effective way to assure everyone that the toxic threat of the corpse has been eliminated.

Thus, while death rituals and afterlife beliefs might perform the functions, as Freud said, of reducing neurotic tensions and mortality anxiety, that is not why they are so widespread. Rather, cognitive scientists of religion believe that these features of religion are widespread because they trigger ordinary cognitive systems. Afterlife beliefs recur across religious systems because our mind has difficulty representing the psychological cessation of agents, and death rituals recur across cultures because corpses trigger our toxin avoidance system.


Religions provide people with explanations of where we came from and why things happen in the ways they do. That is, religions provide metaphysics. Importantly, a limited number of metaphysical schemes recur across religious systems. This suggests that religious metaphysics are constrained by our more basic intuitive systems, notably our “folk physics” system that includes intuitions about proximate and ultimate causality.

The conceptual schemes of most religions state that things happen “for a reason.” In other words, events happen not only as a result of a discernible cause, but also because a causer caused the event to happen when, where, why, and how it did. The developmental psychologist Deborah Keleman has termed the tendency to overattribute purpose in the world “promiscuous teleology.” And she has shown experimentally that this tendency is a natural developmental capacity.

Keleman showed children pictures of various items, including fictive animals (e.g., “footles,” horselike creatures with a long sticklike protrusion on the back) and relatively unusual objects (e.g., pointy rocks). She then asked the subjects how those things got the way they are. More often than not, children responded with teleological explanations—those things were that way for a reason (e.g., “to scratch with”). Even when primed with the evolutionary biological principles of random mutation and natural selection and then asked “Do you think that thing could have gotten that way by accident?”, children still preferred teleological explanations; they answered, “no.”

Similarly, E. Margaret Evans has shown experimentally that the tendency to attribute intentionality as the ultimate cause of things and events accounts for the widespread belief in creationism. This also explains the correspondingly low levels of comprehension of Darwinian theories of evolution by natural selection. Tapping children’s intuitions about how things got the way they are, Evans found little difference in creationist inferences by children from fundamentalist Christian families and children raised without strong religious instruction. Subjects from both groups tended to say that someone had to have made things the way they are now.

This research suggests that we naturally infer agents at work in the world, even where data do not warrant such inferences. In this regard, Barrett has theorized that humans might possess a “hyperactive agency detection device” (HADD) that is primed to detect agents. It makes good evolutionary sense, of course, to possess a HADD, given the possibility of lurking predators in our ancient evolutionary environments. Being hyperactive in detecting agents would be of great benefit in avoiding predators and come at little cost in false positives. Occasionally mistaking rocks for bears would be relatively harmless. Mistaking bears for rocks could be deadly.

Thus, cognitive scientists of religion believe that certain metaphysical themes recur across religious systems because they trigger our ordinary “folk physics.” We are primed to overrepresent agents as causes of events, as creators of existing things, and as pulling strings behind the curtain of reality. In addition, the biases of hyperactive agency detection might help to account for the widespread recurrence of anthropomorphism—representing gods as special persons or animals. Regardless, we don’t find metaphysics recurring across religions because they give us explanations per se, but rather because religious metaphysics tend to trigger our “folk physics” cognitive system.

Theological Incorrectness

One other feature that seems to recur across religions is “theological incorrectness.” Theological incorrectness is another term for intrasystem conceptual variation: specifically, when people have religious ideas that differ from the official creeds of their religions—ideas they “shouldn’t” have. Such conceptual variation seems to be the rule, not the exception, in religious systems with established official theologies. But if religion provides people with ideas to believe, and people believe in those ideas, why would they believe other ideas that they shouldn’t?

The simplest explanation for why theological incorrectness recurs across cultures is that theological concepts are too maximally counterintuitive to be employed in everyday reasoning tasks. Theological concepts, like scientific concepts, can be memorized but have little effect on people’s daily lives. For example, physicists insist that there are no true straight lines in the universe. However, while memorizable, this concept has little effect on how people hang picture frames in their homes. Likewise, religious people can learn theological concepts like “god controls everything” and yet live their lives as if they in fact have control. As noted above, people simply have great difficulty representing concepts that violate too many of the expectations in our intuitive ontology (in this case, self-agency).

Theological god concepts are hard to represent because it is difficult to generate (concrete) images of maximally counterintuitive concepts. Allan Paivio has shown that any images with low levels of concreteness (e.g., beauty, justice) are harder to construct in memory than concepts with high levels of concreteness (e.g., house, tree). Theological concepts typically have low levels of concrete imagery. It is literally impossible to image what “a god who exists everywhere at once” actually looks like. In contrast, “folk” religious concepts generally have high levels of concrete imagery. A concept like “god is a big guy in the sky” is much easier to represent than a god who exists everywhere at once. This, too, might account for the widespread tendency for people to anthropomorphize images of deities.

Of course, the fact that theology is a poor fit with cognition begs the question of how it develops in the first place. Importantly, however, theology is relatively unusual in the history of religion. Like other cognitively burdensome concepts, like scientific concepts, theology requires the use of external (to cognition) storage devices, such as texts, for successful transmission. With texts, any concept, no matter how complex, can be stored and transmitted for a long time. Moreover, storing concepts “artificially” (in artifacts) allows for the ratchet effect in cultural elaboration, as subsequent generations expand on previous creations. In this way, theological concepts tend to drift relatively far from ordinary cognition as they accumulate complexity over generations of transmission. Examples from Christianity include the doctrines of the Trinity, original sin, Calvinist predestination, and Christian evolution (e.g., a biological era is one day in biblical time).

Thus theological incorrectness is a natural product of the constraints of ordinary cognition. This does not mean, of course, that people are incapable of memorizing theology, nor even of acquiring it with enough training and effort. All else being equal, anyone can become an expert, in other words a theologian. But even if acquired (stored in memory), theology is unlikely to be employed in everyday thought.

The cognitive science of religion seeks to explain such recurrent features of religion by connecting them to cognitive systems involved in their production and transmission. This approach both divides religion into various features and deflates religion into a by-product of natural mental processes. There are at least two significant advantages to such an approach to the study of religion. This approach naturalizes the study of religion by making it a tractable scientific subject rather than a philosophical mystery, and connecting religion with cognition allows scholars to employ the full resources of the cognitive sciences. This means that scholars interested in understanding religion have robust metatheoretical and theoretical frameworks from which to work, and powerful methodological techniques to employ. For these reasons, the cognitive science of religion is poised to be one of the most important developments in the scientific study of religion.