In the 1980s, archaeologists excavating a cave at Qafzeh in Israel discovered something in a burial dating to approximately 130,000 years ago. The graves held red ochre, shells, and objects placed with apparent deliberateness alongside the dead. The people who created these graves were anatomically modern humans — the oldest known burial sites with evidence of ritual intention.

They did not have writing. They did not have agriculture. They did not have metal. But they appear to have had something approaching religious behavior: the treatment of the dead in ways that suggest beliefs about what death means, and perhaps about what — if anything — might come after it.

This is remarkable not because it tells us what they believed but because of what it shows about human cognition: that the impulse toward ritual, toward marking death as significant, toward engaging symbolically with the invisible — appears to be as old as the modern human brain itself.

Religion is universal. In every human culture ever documented — from isolated Amazonian tribes to Arctic hunters to urban industrial societies — researchers find beliefs in supernatural agents, ritual practices, and structures for handling the great contingencies of human life: birth, death, illness, suffering, and uncertainty. The universality demands an explanation that mere cultural diffusion cannot provide.

The cognitive science of religion, a field that emerged in the 1990s, offers one: religion is not something humans are taught. It is something human cognition naturally produces.

"We are born believers." — Paul Bloom, Descartes' Baby

Key Definitions

Cognitive byproduct theory — The dominant framework in cognitive science of religion: religion emerges not from a specific evolved "religion module" but as a natural byproduct of cognitive systems evolved for other purposes — agent detection, theory of mind, teleological thinking, and intuitive dualism.

Agent Detection Device (ADD) — Pascal Boyer's term for the cognitive system that rapidly and hyperactively detects agents (entities with goals and intentions) in the environment. Produces false positives — detecting agents where there are none — as an adaptive asymmetry (missing a predator is more costly than falsely detecting one). The system naturally generates supernatural agents from ambiguous environmental stimuli.

Theory of Mind (ToM) — The cognitive capacity to attribute mental states (beliefs, desires, intentions, knowledge) to other agents, including agents with different mental states from one's own. Allows social coordination; also naturally extended to supernatural agents, enabling "relationship" with gods.

Intuitive dualism — Paul Bloom's term for the natural, pre-theoretical sense that mind and body are distinct kinds of things, arising from separate cognitive systems for tracking physical objects vs. agents. Provides the cognitive foundation for afterlife beliefs.

Teleological thinking — The tendency to perceive the world in terms of purpose and design. Deborah Kelemen calls children "intuitive theists" because they naturally interpret natural phenomena as designed for purposes. Underlies design arguments for God.

Minimal counterintuitiveness — Pascal Boyer's principle that cognitively "sticky" concepts — most transmissible across generations — are mostly consistent with natural intuitions with a small number of counterintuitive violations. Explains why religious concepts are memorable and culturally persistent.

Big Gods — Ara Norenzayan's term for the moralizing, omniscient, and omnipresent deities associated with major world religions, as opposed to local spirits and capricious deities of smaller-scale societies. Norenzayan argues Big Gods enabled the large-scale social cooperation of civilizations.

Costly signaling — The display of commitment through behaviors that are genuinely costly (and therefore hard to fake), providing credible signals of group loyalty. Religious rituals, dietary restrictions, and other costly religious behaviors are proposed to function as credible commitment signals.

The Universality Problem

Why does religion need a special explanation? Because if it were simply a culturally transmitted set of ideas, we would expect to find cultures without it — cultures where the transmission chain was broken or where the ideas simply never took hold.

We don't.

The ethnographic record stretching back to the 19th century documents no human culture entirely lacking in supernatural belief of some kind. The specific forms vary enormously — polytheism, monotheism, animism, ancestor veneration, shamanism, totemism — but the category of supernatural agency is constant. There are always agents whose behavior matters and who are not physically present.

This pattern is precisely what we would expect if religion were not primarily transmitted but generated — if human cognition, given the right (which is to say, any) environment, naturally produces religious concepts.

Before Formal Religion: The Archaeological Trace

The Qafzeh burials are among the oldest evidence, but the archaeology of religion extends across the entire span of anatomically modern humans:

130,000 years ago: Qafzeh, Israel — deliberate burials with grave goods suggesting ritual intention 77,000 years ago: Blombos Cave, South Africa — ochre processing, shell beads, possibly symbolic 45,000 years ago: Lion-Man of the Hohlenstein-Stadel — a figurine with lion head and human body: a being that does not exist in nature, created by a brain that could imagine it ~11,000 years ago: Göbekli Tepe, Turkey — the oldest monumental architecture yet found, built before agriculture, apparently for ritual purposes. The site suggests that the impulse to build for gods came before — not after — the impulse to build for food storage

The Cognitive Architecture of Religious Belief

The cognitive science of religion does not propose a "religion module" — a specialized evolved structure for producing religious belief. It proposes something more parsimonious: religious concepts naturally emerge from cognitive systems evolved for entirely different purposes.

1. The Hyperactive Agent Detector

Pascal Boyer, an anthropologist at Washington University in St. Louis, developed what remains the most influential cognitive theory of religion: the byproduct theory.

Its starting point is the observation that humans are systematically hypersensitive to agents — entities with goals, intentions, and the capacity for directed action. This hypersensitivity was adaptive: in ancestral environments, the cost of failing to detect a predator or a rival far outweighed the cost of false positives (treating wind in the grass as a predator when it wasn't). Natural selection produced a system biased toward over-detection.

The consequence: we see agents everywhere. The wind that moves in an unusual pattern, the cloud that takes a human shape, the shadow at the edge of vision, the run of bad luck that feels personally directed — all can trigger agent-detection responses.

Supernatural agents — gods, spirits, ancestors, demons — fit the agent-detection template perfectly. They are agents: they have intentions, they observe behavior, they respond to it. They are, for the agent-detection system, exactly the right kind of thing to detect.

Boyer's key insight: religious concepts are not taught to an otherwise blank cognitive slate. They are variations on cognitive natural kinds — the kinds of representations the human mind naturally produces and finds compelling.

2. Theory of Mind Extended

The capacity to attribute mental states to others — beliefs, desires, intentions, knowledge — is essential for human social life. We cannot navigate the social world without modeling what others think, want, and know.

This system operates automatically and irresistibly. We extend theory of mind to fictional characters, to pets, to wind-up toys, and to weather. We cannot help ascribing mental states to things that move purposefully.

Supernatural agents are represented as having mental states — God knows what you did, the ancestors want to be honored, the spirits are displeased. This is not a religious teaching that people accept cognitively; it is what the theory-of-mind system naturally generates when applied to agents, including supernatural ones.

The concept of prayer — communicating with a supernatural agent — is specifically processed by the brain as social communication. Uffe Schjoedt's neuroimaging found that devout Christians' brain activity during personal prayer to God was indistinguishable from patterns during communication with a close friend — but distinct from patterns during ritualized prayer. The brain represents the relationship with God as a personal relationship, not an abstract concept.

3. Intuitive Dualism and the Soul

Paul Bloom at Yale has argued that humans are natural dualists: we intuitively experience the mind and body as distinct kinds of things.

This is not a religious teaching. It is a consequence of the fact that human brains use largely separate cognitive systems for tracking physical objects (constant, persistent, causally governed by physics) and agents (intentional, unpredictable, governed by mental states). These systems give us very different experiences of bodies vs. minds.

Bloom demonstrated this with children: 5-year-olds told that a mouse had just died would correctly say the mouse no longer needs to eat or move (bodily states) but would also often say the mouse still knows what its mother looks like and still wants to go home (psychological states). The distinction between body-death and mind-persistence is intuitive, not taught.

This intuitive dualism provides the natural cognitive foundation for afterlife beliefs. If the mind is experienced as separate from the body, its persistence after the body dies is cognitively natural — it requires no special teaching, only cultural direction about what form this persistence takes.

Jesse Bering at Cornell found that even adults with explicitly materialist, secular worldviews show intuitive afterlife attribution in indirect measures — the cognitive default persists even after deliberate commitment to contrary philosophical positions.

The Big Gods Hypothesis: Religion and Cooperation

Ara Norenzayan at the University of British Columbia developed a different but complementary account: whatever its cognitive origins, religion's persistence and specific forms have been shaped by its functional role in enabling human cooperation.

The Small-Scale Baseline

Small-scale hunter-gatherer societies typically feature local spirits, capricious deities, or ancestral presences that are not particularly interested in human moral behavior. These supernatural agents may be powerful and potentially dangerous, but they are not omniscient moral monitors.

This pattern — spirits that don't care much about whether humans treat each other fairly — is the norm in small-scale societies.

The Emergence of Moralizing Gods

The major world religions — the Abrahamic traditions, Hinduism, Buddhism, Zoroastrianism — feature deities that are explicitly interested in human moral behavior: omniscient, omnipresent, and likely to punish defectors and reward cooperators.

Norenzayan's argument: these "Big Gods" who monitor and punish moral behavior provided a solution to the cooperation problem in large-scale anonymous societies. When you live in a band of 50 people who all know each other, reputation mechanisms are sufficient to maintain cooperation. In a city of 10,000 strangers, they are not.

An internalized belief that an omniscient, morally engaged deity is monitoring your behavior provides the psychological equivalent of perpetual reputation monitoring even in anonymous contexts — the "watched by God" effect.

Norenzayan's lab ran economic games (games where participants could share resources or defect) across 15 diverse societies. Participants primed with their culture's big god concept before the game were significantly more generous, particularly toward distant strangers. Participants in societies with more moralistic gods showed higher fairness in cross-cultural economic games.

The implication: the specific forms that religion takes in large complex societies may reflect cultural selection for religions that solve the cooperation-among-strangers problem.

Why Beliefs Resist Updating

Religious beliefs show unusual persistence in the face of counter-evidence. This is not primarily a failure of intelligence — highly educated, analytically sophisticated people hold deep religious beliefs — but reflects the architecture of how religious beliefs are held.

Intuitive vs. Analytical Holdings

Daniel Kahneman's dual-process framework distinguishes fast, automatic, intuitive processing (System 1) from slow, deliberate, analytical processing (System 2). Many religious beliefs are held intuitively — they feel obviously true, self-evident, naturally given — rather than as conclusions of deliberate reasoning.

Intuitive beliefs are not easily updated by counter-evidence presented to System 2, because they are maintained by System 1 processes that run automatically and prior to conscious deliberation.

Will Gervais and Ara Norenzayan found that experimental manipulations inducing analytical thinking (priming analytical processing through exposure to Rodin's "The Thinker") produced temporary reductions in religious belief — suggesting System 2 can override System 1 religious defaults, but only when actively engaged.

The Social Scaffold

Religious beliefs are not held in isolation. They are embedded in communities, identities, relationships, and rituals that provide constant reinforcement and that make belief-change socially costly.

Updating a religious belief often means:

  • Threatening membership in a community that provides social support
  • Challenging an identity that is central to self-concept
  • Breaking shared symbolic systems that maintain family and community bonds
  • Confronting the existential questions (death, suffering, meaning) that the belief addressed

The social and psychological costs of belief change are often far more salient than abstract evidential considerations, which is why rational counter-arguments often strengthen rather than weaken religious commitment.

What Secularization Tells Us

The natural experiment of European secularization — one of the fastest belief changes in human history — provides important evidence.

Highly secular societies like Sweden, Denmark, and Norway function well: they score high on trust, life satisfaction, and civic health. This challenges the view that religion is irreplaceable.

But the secularization of formal religion has not eliminated supernatural belief. Phil Zuckerman's Scandinavian fieldwork found persistent belief in fate, luck, karma, and vague "something more" even among people who never attend church and explicitly reject formal religion. "Spiritual but not religious" remains widespread.

This is consistent with the cognitive byproduct view: formal religion can be secularized; the cognitive systems that generate supernatural-flavored beliefs cannot be, or can only be suppressed with active philosophical effort.

For related articles, see why cultures think differently, how different cultures think about death, terror management theory explained, and what causes anxiety.

References

  • Boyer, P. (2001). Religion Explained: The Evolutionary Origins of Religious Thought. Basic Books.
  • Barrett, J. L. (2004). Why Would Anyone Believe in God? AltaMira Press.
  • Norenzayan, A. (2013). Big Gods: How Religion Transformed Cooperation and Conflict. Princeton University Press.
  • Bloom, P. (2004). Descartes' Baby: How the Science of Child Development Explains What Makes Us Human. Basic Books.
  • Bering, J. (2011). The Belief Instinct: The Psychology of Souls, Destiny, and the Meaning of Life. W. W. Norton.
  • Gervais, W. M., & Norenzayan, A. (2012). Analytic Thinking Promotes Religious Disbelief. Science, 336(6080), 493–496. https://doi.org/10.1126/science.1215647
  • Schjoedt, U., et al. (2009). Highly Religious Participants Recruit Areas of Social Cognition in Personal Prayer. Social Cognitive and Affective Neuroscience, 4(2), 199–207. https://doi.org/10.1093/scan/nsn050
  • Kelemen, D. (2004). Are Children 'Intuitive Theists'? Psychological Science, 15(5), 295–301. https://doi.org/10.1111/j.0956-7976.2004.00672.x
  • Mehr, S. A., et al. (2019). Universality and Diversity in Human Song. Science, 366(6468). https://doi.org/10.1126/science.aax0868

Tags

religion cognitive science anthropology psychology evolution culture belief

Frequently Asked Questions

Is religion universal across human cultures?

By virtually all empirical measures, yes — religion or something closely resembling it has been found in every known human culture, including those with minimal contact with other societies. The evidence comes from multiple converging sources. Archaeological record: burial practices, grave goods, and evidence of ritual date back at least 130,000 years in early Homo sapiens; Neanderthal graves suggest possible proto-religious behavior by 60,000-100,000 years ago. Ethnographic record: anthropologists studying cultures across every inhabited continent, including isolated hunter-gatherer societies, consistently find beliefs in supernatural agents, ritual practices, and structures for handling birth, death, illness, and uncertainty that involve some appeal to non-physical causes. Cross-cultural surveys: the Pew Research Center's Global Religious Landscape survey (2012) found 84% of the world's population identifying with a religious group; the remaining 16% (largely concentrated in specific countries) includes many who hold supernatural beliefs without formal religious affiliation. This universality is itself theoretically significant: universal features of human behavior typically reflect either (a) universal environmental pressures producing convergent cultural solutions, or (b) features of human cognition that naturally produce that behavior across diverse cultural contexts. The cognitive science of religion argues for the latter: religion is universal not because it is hardwired in a specific 'religion module' but because it emerges naturally from cognitive systems that evolved for other purposes.

What is the cognitive byproduct theory of religion?

The cognitive byproduct theory (also called the cognitive by-product hypothesis or the 'religion is a spandrel' view) is the dominant framework in the cognitive science of religion. Associated with Pascal Boyer, Justin Barrett, David Sloan Wilson (though Wilson advocates a different functional version), and Ara Norenzayan, it proposes that religion is not an evolutionary adaptation directly selected for its religious functions but rather a byproduct — an emergent consequence — of cognitive systems that evolved for entirely different purposes. The key cognitive systems proposed as generating religious belief: (1) Agent Detection Device (ADD): humans are hypersensitive detectors of agents — things with goals, intentions, and the ability to act on the world. This was adaptive in ancestral environments where under-detecting a predator or enemy was far more costly than over-detecting one. The system generates false positives — detecting agents in random noise, wind, shadows, or coincidences. Supernatural agents (gods, spirits, ancestors) are cognitive natural kinds precisely because they fit the agent-detection template exactly: they have intentions, observe behavior, and respond to it. (2) Theory of Mind (ToM): the ability to attribute mental states (beliefs, desires, intentions) to other agents, enabling social navigation. Humans naturally and irresistibly apply ToM to agents including supernatural ones, modeling what God or spirits want, believe, and intend. (3) Intuitive dualism (or body-soul dualism): the intuitive sense that the mind is separable from the body, which Paul Bloom at Yale argues is a natural consequence of the separate cognitive systems we use to track physical objects vs. agents. This intuitive dualism makes afterlife beliefs cognitively natural: if the mind-soul is distinct from the body, it can persist after the body dies. (4) Teleological thinking: the tendency to see the world in terms of purpose and design — clouds are for giving rain, lions are for hunting, the world is for something. Deborah Kelemen has documented this in children (who are 'intuitive theists') and adults.

Does religion serve evolutionary functions — did it help groups survive?

This is the central debate in the evolutionary anthropology of religion, with several competing positions. The byproduct view (Boyer, Barrett): religion is a cognitive byproduct, not directly adaptive. It costs resources (ritual, dietary restrictions, costly signals of commitment) and many of its beliefs are false; natural selection would not directly favor it. Any adaptive benefits religion produces are incidental to the cognitive systems that generate it. The cultural group selection view (David Sloan Wilson, Ara Norenzayan): religion evolved as a mechanism for producing prosocial behavior at the group level, enabling large-scale cooperation between non-relatives. Groups with strong shared religious commitments — enforced by belief in moralizing gods who punish defectors — were more cohesive, more willing to sacrifice for group benefit, and more successful in competition with other groups. Norenzayan's 'Big Gods' thesis (2013): the shift from animistic religions with capricious local spirits to 'Big God' religions with omniscient, moralizing deities who care about human behavior may have been culturally selected for because it enabled large-scale cooperation in complex societies. Survey data across 15 cultures found that people primed with their god concept were more generous in economic games — suggesting active behavioral effects of religious belief. The costly signaling view (William Irons, Randolph Nesse): religious rituals and beliefs function as costly, hard-to-fake signals of group commitment. Participating in expensive rituals (fasting, circumcision, pilgrimage) demonstrates genuine group loyalty because the cost is prohibitive for free-riders. Current consensus is pluralistic: religion likely serves multiple functions simultaneously, with byproduct mechanisms generating the initial content and cultural and possibly biological selection shaping which religious forms become dominant.

Why do people believe in life after death — is this universal?

Afterlife beliefs are among the most cross-culturally widespread religious beliefs, documented in virtually every human culture — though the specific form varies dramatically (heaven/hell, reincarnation, ancestor spirits, shadowy underworld, continued community membership). Paul Bloom's 'intuitive dualism' provides the primary cognitive explanation. We are born with two largely separate cognitive systems: one for tracking physical objects (constancy, persistence, physical causation) and one for tracking agents (mental states, intentions, goals). These systems are computationally distinct — we understand how bodies work very differently from how we understand minds. The result is a natural, intuitive sense that the mind and body are distinct kinds of things. Bloom's experiments with children found that they intuitively understood that a dead mouse no longer had bodily needs (hunger, cold) but did still have psychological states (it still knew its mother's name, still wanted things). This suggests an intuitive 'soul' concept that persists after body death is not taught but emerges naturally from dual-system cognition. Cross-cultural research by Jesse Bering found that children as young as 4-5 years old, across cultures and regardless of religious instruction, attributed continued mental states to dead individuals. Adults with secular backgrounds show similar intuitive afterlife attribution in indirect measures even when explicitly endorsing materialist views. This suggests afterlife belief is a cognitive default that requires specific suppression through deliberate philosophical commitment — not a belief that requires active teaching.

Why do religious beliefs resist evidence — why don't people update them rationally?

Religious beliefs show unusual resistance to counter-evidence compared to empirical beliefs, and this has been studied extensively from cognitive and social perspectives. Several mechanisms contribute. Dual-process architecture: many religious beliefs are held intuitively (System 1, fast, automatic, associative) rather than analytically (System 2, slow, deliberate, evidence-evaluating). Intuitive beliefs are less subject to standard rational updating because they are maintained by fast, automatic processes rather than deliberate reasoning. Willard and Norenzayan have found that higher reflective thinking predicts lower religious belief in some populations — suggesting analytical thinking can override intuitive religious beliefs, but requires deliberate effort. The social dimension: religious beliefs are embedded in community, identity, and relationship. Updating a religious belief often means threatening one's social relationships, community membership, and identity narrative — costs that are psychologically far more salient than abstract evidential considerations. Motivated reasoning: people process evidence about beliefs they are motivated to hold differently than beliefs they are neutral about. Disconfirming evidence is subject to heightened scrutiny; confirming evidence is accepted more readily. Religious beliefs, being identity-relevant and social, recruit more motivated reasoning than low-stakes empirical beliefs. The 'minimal counterintuitiveness' principle: Pascal Boyer notes that religious concepts are memorable and transmissible precisely because they are mostly intuitive with a small number of counterintuitive violations (an invisible being that is everywhere simultaneously; a person who can walk through walls). Concepts that are mostly but not entirely consistent with natural intuitions are cognitively 'sticky' — memorable and transmissible — in ways that either fully intuitive or fully counterintuitive concepts are not.

Are religious and non-religious brains actually different?

Neuroimaging and cognitive studies have explored whether religious individuals process the world differently from non-religious individuals, and the answer is a nuanced yes — with important caveats about causation. Prayer and religious experience activate a distributed network overlapping with theory-of-mind regions (medial prefrontal cortex, temporoparietal junction), self-referential processing regions (default mode network), and reward regions (nucleus accumbens). Prayer activates the same regions as intentional social communication directed toward a perceived agent — suggesting the brain processes prayer as communication with an agent. Uffe Schjoedt and colleagues found that devout Christians' brain activity during prayer to God closely resembled brain activity during communication with a close friend — distinct from the pattern for ritualized prayer or communication with an unknown person. This suggests the brain genuinely represents the deity as a personal relationship rather than an abstract concept. Structural differences: Michael Ferguson and colleagues found specific structural brain differences associated with religious upbringing — thicker cortex in several regions associated with self-regulation and reflection. Some studies find that religious/spiritual individuals show lower stress reactivity (reduced cortisol response to threat), potentially reflecting the security provided by relationship with a personal deity, though the literature is mixed. The 'God helmet' research: Michael Persinger's claims that weak magnetic field stimulation to the temporoparietal region produced 'sensed presence' experiences were widely reported but have not reliably replicated. The temporal lobe has been proposed as particularly involved in religious experience (the 'God spot' hypothesis) but this oversimplifies a complex distributed process. Key caveat: all neuroimaging findings in religion are correlational — they show which neural systems are engaged by religious cognition, not whether religion 'causes' specific brain organization.

Can religion be replaced — what happens to societies that secularize?

Secularization — the decline of religious belief and practice — has proceeded rapidly in many Western countries over the last 50 years, providing natural experiments about what happens when religion recedes. The Nordic countries (Sweden, Denmark, Norway, Finland) are the most studied cases: they have among the world's lowest religious observance rates while maintaining high social trust, life satisfaction, and civic health. Phil Zuckerman's extensive fieldwork in Scandinavia ('Society Without God,' 2008) found that secular Nordic societies had developed alternative structures for meaning, community, and moral framework — strong welfare states, civic organizations, secular rituals (solstice celebrations, secular funerals), and humanistic moral frameworks. This challenges the view that religion's social functions are irreplaceable. However, secularization is uneven: while formal religious belief and practice have declined, supernatural belief has been more persistent. Many people in highly secular societies retain beliefs in fate, karma, luck, ghosts, or vague spirituality — consistent with the cognitive byproduct view that these beliefs arise naturally from human cognition and require active suppression. 'Spiritual but not religious' has become the fastest-growing religious category in many Western countries — arguably reflecting the natural-default status of some form of supernatural belief in human psychology. Research by Lanman, Coleman, and others on the consequences of religion loss suggests that religious people who lose their faith without acquiring replacement meaning structures show elevated depression and social isolation. This suggests religion's functions are important even if they are replaceable — which means secular societies need deliberate alternatives for community, ritual, and meaning.

Share this article