1.1 Definitions

A good place to start discussion of children’s understanding of God and religion is to give some definitions of how psychologists define ‘religion’, ‘belief’ and ‘God/gods’. When psychologists talk about religion, we often define it as a complex set of cognitive, behavioural and socio-cultural phenomena. When we examine cognition, we are often concerned with understanding the cognitive tools and biases that give rise to beliefs, or the structured systems of ideas about supernatural agents (e.g., God/gods), supernatural events and other beliefs. When we examine behaviour, we are interested in what people do as a consequence of beliefs, such as looking at religious practices, rituals, ceremonies and morals. Third, psychologists are also interested in socio-cultural influences and how they shape cognition and belief and behaviour and practice. In particular, communities and families shape the beliefs and practices of that group and, within cultures, large social groups and institutions (e.g., churches and religious playgroups) serve to encourage and transmit these beliefs and practices.

Most of the research in this Element will capture a particular concept of God. When God is mentioned, this is usually to describe the God of Christian traditions (God as three persons in the Divine Trinity, including God as Father, Jesus Christ the Son and the Holy Spirit), God as YHWH or G-d in Jewish traditions and God as Allah in Islamic traditions. These Abrahamic traditions conceptualise God as what scholars of religion call a ‘high God’, a Divine being attributed with all-knowing, moralizing and punitive, all-present and all-powerful qualities (Norenzayan et al., Reference Norenzayan2016; Swanson, Reference Swanson1960).

We also will look at lesser deities, ‘gods’ with a lower case ‘g’ or other supernatural agents that do not fit the ‘High Gods’ status. These supernatural agents may have one or two supernatural characteristics, such as lacking a physical body (e.g., ghosts), immortality (e.g., elves) or have a super-ability to hear the requests of relatives (e.g., ancestor spirits). In the following sections, past and present work will highlight how children in different religious traditions and cultures conceptualise these very different G/gods.

A few final definitions are worth noting regarding the language used to discuss ‘biases’ in thinking about God and religion. This Element will talk about different kinds of biases, such as ‘cognitive biases’ and ‘transmission biases’. ‘Cognitive biases’ are tendencies to act in a certain way or patterns in thinking that influence how we process information. Some of the other biases discussed in this Element stem from ‘naïve’ or ‘folk’ knowledge, such as ‘naïve or folk psychology’, which is someone’s intuitive understanding of others. For example, as will be discussed in Section 2, even infants have an early understanding of mental states of others. This Element, and mainly in Section 6, also discusses ‘transmission biases’, also termed ‘social learning strategies’, which are normally discussed in research focussing on cultural evolution, or how humans and other animals learn and acquire new information and also transmit ideas and information. Acquisition of these ideas often occurs via ‘model-based’, ‘context’ and ‘content biases’. ‘Model-based’ biases are particular characteristics of a person, such as someone with high competence, expertise, prestige and familiarity; that are particular cues that one may use in order to choose to learn from these individuals compared to other individuals lower in those characteristics. ‘Context biases’ are those where one might make a decision to learn from someone else depending on the context, situation or environment. Finally, ‘content biases’ are ideas or concepts one is drawn to because they are concepts that can be interesting, surprising, memorable and grab attention. Content biases, such as Minimally Counterintuitive (MCI) ideas, are discussed in Section 6.

1.2 Developmental Origins of Belief

Where do religious beliefs come from? Why are some religious beliefs (e.g., belief in God/gods) so common across different cultures and why are some beliefs so different (e.g., Boyer & Bergstrom, Reference 70Boyer and Bergstrom2008; Sperber & Hirschfeld, Reference Sperber and Hirschfeld2004)? Why do some beliefs persist across generation after generation? This Element will not be answering these questions from an evolutionary or historical point of view. Instead, this Element will be describing how religious beliefs originate, develop and persist from the perspective of child development.

The ontogenetic origins of religious ideas trace the fundamental questions and human experiences that have shaped the development of religious belief systems throughout history. At its core, this exploration delves into the origins of religious concepts and beliefs, seeking to understand how human cognition, cultural evolution and existential inquiries converge to form diverse ideas, beliefs and culture. Exploring the ontological roots of religious thought provides insights into the profound ways in which human societies have grappled with existential questions and sought meaning through spiritual narratives and beliefs.

1.3 Early Cognitive Mental Tools That Predispose Understanding of Supernatural Agents: Hyperactive Agency Detection Device, Teleological Reasoning, Theory of Mind

This section focusses on three cognitive tools that influence children to think about supernatural agents in particular ways. It will introduce research that describes how intuitive and domain-specific knowledge helps children to learn about the natural and supernatural worlds. It will cover biases such as teleological reasoning, theory of mind and hypersensitivity agency detection device (HADD).

One of the main concerns of CSR is to explain why particular kinds of ideas and behaviours persist across time, and in forms that are often similar across different cultural groups. The concept of God is a good example. Across many cultures, there could have been the potential for a large pantheon of supernatural agents. However, representations of God, and even other supernatural agents, are very similar and have remained stable and prevalent concepts across a diverse set of societies and across generations. Indeed, the divine characteristics are prevalent in many societies and religious traditions: for example, attributes such as omnipresence or immortality are common among supernatural agents. The early founders of CSR suggested that the acquisition of a concept of God, and other gods, are a natural consequence supported by cognitive processes used in everyday activities to understand and interact with the world (Barrett & Nyhof, Reference Barrett and Nyhof2001; Bloom, Reference Bloom2007; Boyer & Ramble, Reference Boyer and Ramble2001; Guthrie, Reference Guthrie1995; McCauley, Reference McCauley, Keil and Wilson2000).

Many CSR scholars claim that supernatural agent concepts emerge early in development because they tap into intuitive processes (and cognitive tools) we hold about the social world (McCauley, Reference McCauley, Keil and Wilson2000). The logic goes that God and supernatural agents are in every society, and people conceptualise these agents as persons. So, it stands to reason that cognitive tools that pertain to social reasoning are relevant to concept formation concerning God. Thus children (and adults) are cognitively constrained by default inferences and assumptions humans make about agents, and specifically supernatural agents. These cognitive defaults are part of an intuitive psychology, also known as social cognition. Social cognition is our ability to perceive, interpret and respond to the behaviour of others. Adults and children use several cognitive tools to perceive and interpret agency. These cognitive tools begin to develop in infancy and in early childhood and serve to help form many of children’s beliefs and knowledge about agency. Three of the most crucial mental tools to understanding the development of religious cognition are described further.

1.4 Transmission Biases Important for Acquiring Religious Concepts

The previous text depicts particular cognitive tools that scaffold how children and adults think about religious ideas. However, it’s worth mentioning here (although it will be discussed more fully in Section 6) that there are other transmission biases (also called ‘social learning’ strategies) that influence how children and adults learn cultural and religious information (Wood et al., Reference Wood, Kendal and Flynn2013). Indeed, because religious ideas are abstract, one of the ways to learn about them is to discuss and transmit them. There are two main types of transmission biases. They are called ‘content’ or ‘direct’ biases and also ‘context’ or ‘indirect’ biases (and also known as ‘model-based biases’).

Content biases are any characteristic or property (the ‘what’) of a cultural trait that is likely to be copied. For example, it could be an idea that is particularly engaging and interesting. Content biases work by drawing one’s attention to an interesting idea so much that it is worth talking about and transmitting to someone else. Many theorists believe that religious ideas and concepts are particularly interesting because they are counterintuitive and violate natural laws (e.g., supernatural agents that can fly or live forever) (Boyer & Ramble, Reference Boyer and Ramble2001). And the reason we see so many of these counterintuitive concepts in cultural and religious stories is because they have been passed down and transmitted because of their compelling content (Barrett & Nyhof, Reference Barrett and Nyhof2001).

Context biases, on the other hand, focus on the person (the ‘who’) or the situation (the ‘when’) in a particular context. In other words, research has shown that one is more likely to copy a new behaviour from someone who is prestigious (Berl et al., Reference Berl, Samarasinghe, Roberts, Jordan and Gavin2021; McGuigan & Cubillo, Reference McGuigan2013; McGuigan, Reference McGuigan2013), familiar (Corriveau & Harris, Reference 72Corriveau and Harris2009), successful or has particular expertise (Hermes et al., Reference Hermes, Behne, Bich, Thielert and Rakoczy2018), or through copying a consensus (Corriveau, Fusaro, & Harris, Reference Corriveau, Fusaro and Harris2009). Studies confirm that children are sensitive to learning religious ideas and concepts from familiar people and also when they profess confidence in their beliefs. For example, a recent study found that caregivers who discussed religious concepts with greater certainty than caregivers who discussed with less certainty were more likely to have children who confidently endorsed those concepts (e.g., God, angels and Heaven) (McLoughlin et al., Reference McLoughlin, Davoodi, Cui, Kelly, Jennifer, Harris and Corriveau2021).

To conclude this section, these three cognitive tools, in conjunction with transmission biases, help explain the origin and spread of supernatural agency. HADD biases us to identify agency and may cause us to attribute supernatural agency to events we cannot explain. ToM supports thinking about and imagining agents and their behaviour. Finally, teleological reasoning helps to explain why we may be biased to think that a supernatural agent designed the world and all the animals and plants within it. Transmission biases influence when and what are transmitted, and who children pay attention to in learning new cultural information.

The next section builds on the discussion of these cognitive tools and transmission biases and describes several approaches for how children use these cognitive processes, specifically their ToM, to acquire supernatural agent concepts. The acquisition of supernatural agent concepts is debated, so this next section describes each of four approaches in turn.

2.1 Anthropomorphism

The oldest theory is the anthropomorphic hypothesis. This account originated with the early work of Freud and Piaget, and although there is still commonality from these old theories that children first think of God in human terms, the way in which children develop the concept is different. Instead of seeing God as a human father figure, new work from developmental psychology suggests that the development of understanding human limitations is the prototype for understanding all other agents (including animals and supernatural agents).

This account suggests that as children develop an understanding of human limitation, children use humans as a template to attribute human-like characteristics to other non-human agents. Indeed, preschool children understand that they do not know everything and learn that others have limited knowledge (Gopnik & Slaughter, Reference Gopnik and Slaughter1991). Based on this, children first assume that all minds are fallible, including God. Then, in combination with growing older, cognitive development, and interactions with peers, family members and strangers (Carpendale & Lewis, Reference Carpendale and Lewis2004; Lane et al., Reference Lane, Wellman and Evans2010); children develop a rich understanding of agency. For example, young children are likely to attribute God with human-like (fallible) knowledge and as they get older, they use deliberate reasoning to attribute God as all-knowing (Heiphetz et al., Reference Heiphetz, Lane, Waytz and Young2016; Kiessling & Perner, Reference Kiessling and Perner2014; Lane et al., Reference Lane, Wellman and Evans2010, Reference Lane, Wellman and Evans2012; Lane et al., Reference Lane and Harris2014). By four or five years they can differentiate human ability from supernatural ability (e.g., God and Tooth Fairy) and non-human ability (e.g., animals).

Further, this account suggests that cultural learning is necessary for the development of God concepts (Heiphetz et al., Reference Heiphetz, Lane, Waytz and Young2016; Lane, Reference Lane2021). As children gain better understanding of others through experience and cognitive maturation, they come to a theologically correct understanding of God. And children raised in religious contexts have been shown to differentiate God from humans much earlier than children not raised in religious contexts (Lane et al., Reference Lane, Wellman and Evans2012).

2.2 Preparedness

The anthropomorphic hypothesis is in contrast to the preparedness account. As the name suggests, the second theory proposes that young children are ‘prepared’ to conceptualise supernatural agents before they grasp the limitations and abilities of human agents. According to this approach, certain cognitive tools bias children to attribute supernatural abilities to all agents and find this the default. This approach was named and first promoted by Justin Barrett based on studies that showed preschool children will attribute both their caregivers and God with knowing everything until they are aged five or six and then realize that their mother or father have human limitations. Specifically, these studies showed that three-year-old children, when asked whether their mother and God will know particular information that they cannot see (e.g., knowing an item in a closed box), will respond that both know the contents of the box even when they are encountering the box for the first time (Barrett & Richert, Reference Barrett, Richert and Driesenga2001; Knight et al., Reference Knight, Sousa, Barrett and Atran2004; Wigger et al., Reference Wigger, Paxson and Ryan2013). And, not until about five years, did children recognise that their mother would not actually know the contents of the box.

Those who support the preparedness hypothesis have used these results as an explanation for why so many supernatural agent concepts exist and for the persistence of religion. They suggest that young children are ‘intuitive theists’ and are able to give a theologically correct response that God would know everything, and that religious ideas are ‘natural’ (Kelemen, Reference 77Kelemen2004; McCauley, Reference 81McCauley2011), based on their intuitive knowledge of agents (using ToM and HADD).

Contrasted with the anthropomorphic account, the preparedness account promotes that preschool-aged children first conceptualise all agents as all-knowing. From this early standpoint, cultural learning is needed to gain direct experience and knowledge of other humans and animals so that children understand the limitations of infallible minds.

2.3 Agent Specific or Anthropocentrism

A third approach is the agent-specific or anthropocentric approach. This approach suggests that from infancy children have a much more general understanding of agency, rather than drawing from a human or all-knowing (Godlike) perspective. Here social learning is important for the child to gain experience of all types of agency to help them differentiate among human and non-human agents. This approach assumes that from infancy, children gain experience interacting with others and begin to classify the types of agents that they encounter, differentiating human from other types of agents. Infant researchers have shown that infants track the gaze of non-human-like blobs (who have eyes) and seemingly track their intentions, as they would in experiments where they track people (Csibra & Gergely, Reference Csibra and Gergely1998; Luo & Baillargeon, Reference Luo and Baillargeon2010; Woodward et al., Reference Woodward, Sommerville, Gerson, Henderson and Buresh2009).

Support for this approach is slim. Very few infant studies have been conducted focussing on supernatural agency and, in general, some researchers are sceptical of using solely behavioural measures to infer cognition and conceptual understanding (Kunin et al., Reference Kunin, Piccolo, Saxe and Liu2024; but see also Oakes, Reference Oakes2014 for other methodological issues), especially abstract cognition as supernatural agents.

2.4 Constructivism or Socio-Cultural Approach

A final approach is a constructivist or socio-cultural approach. This approach differs from the previously three as it more heavily suggests that children rely on socio-cultural learning to understand God concepts than on early cognitive biases. The constructivist viewpoint proposes that children acquire knowledge through experience and this knowledge both shapes distinctive concept development of various agents and also constrains cognition by shaping the development of ToM through experience and reasoning.

This is not a new theory. Vygotsky (Reference Vygotsky1934/1986) originally proposed the importance of focussing on socio-cultural factors. He said that children first form pseudo-concepts where they begin with a particular word and build on making associations for that term. Thus, the environment and communities that children live in greatly influence the way these concepts are refined (Gauvain, Reference Gauvain2001). This approach recognises the importance of both intuitive cognitive biases that aid in the thinking of other beings and also learning biases (such as context biases) that acknowledge the necessity of learning from others about specific religious ideas and doctrine that contextualise and refine the concept (Burdett et al., Reference Burdett, Wigger and Barrett2021; Lane, Reference Lane2021; Legare & Sousa, Reference Legare and Souza2012; Richert et al., Reference Richert, Saide, Lesage and Shaman2017; Shtulman et al., Reference Shtulman, Foushee, Barner, Dunham and Srinivasan2019). For example, research has shown that children raised in religious contexts develop a rich understanding of differences in different types of minds (Burdett et al., Reference Burdett, Wigger and Barrett2021; Lane et al., Reference Lane, Wellman and Evans2012; Richert et al., Reference Richert, Saide, Lesage and Shaman2017).

The socio-cultural account is growing in support as recent work demonstrates how cognition influences cultures, but also that social and cultural learning scaffolds cognition (e.g., Lane, Reference Lane2021; Richert & Corriveau, Reference Richert and Corriveau2022). Current researchers are using more diversified samples (Lesage & Richert, Reference Lesage and Richert2021) and understudied populations (Richert et al., Reference Richert, Weisman, Lesage, Ghossainy, Reyes-Jaquez and Corriveau2022; Weisman et al., Reference Weisman, Ghossainy, Williams, Payir, Lesage and Reyes-Jaquez2024). They are also recognising the value of collaborating with communities to understand the behaviours and ideas that are important in that community (Bang et al., Reference 68Bang, Faber, Gurneau, Marin and Soto2016). The focus of this new research will be discussed in more detail in Section 6.

2.5 Why Do Researchers Come to Different Conclusions About this Work?

After reading these different accounts, one might be confused as to how so many theorists have come up with quite different conclusions. Some of the differences may be because of the small size of the samples, the lack of participant diversity, the variations in methodologies and potentially the biases from the researchers themselves. Some clarifications for each of these four reasons are merited.

Small sample sizes, in addition to the non-diversity of samples, put a study at risk by not giving a sample enough statistical power, such that results could suggest a significant result that is not true. So, if a particular study had sampled a large population, this added power may show that the result when using a small sample size is not actually true for this population. Psychologists call this a Type 1 error. Another possibility is that results could suggest that there are no significant results when, if there was a larger sample, the results would actually be true of the population. This is called a Type II error. In addition to sample sizes, the diversity of religious traditions and background of the children has certainly limited prior studies and could explain the variance in results. Without diversity, there is risk of not characterising the population correctly.

Of the four issues mentioned earlier, the most likely reason for the differences in results is because of the methods used in each study and consequently the interpretation given by the researchers. In order to understand how methodological design can impact results and interpretation, a useful exercise would be to describe key studies that have been used as evidence and rationale for each approach mentioned earlier. The next section presents these studies historically.

3.1 Studying Children’s God Concepts

The study of children’s concepts of God launched with a seminal study by Justin Barrett, Rebekah Richert and Amanda Driesenga in 2001. While research into children’s understanding of God is not new, this study was ground-breaking as the first study to apply a theory-of-mind task, a task that had already been used for over forty years to understand children’s understanding of the abilities of human minds. There are many theory-of-mind tasks, such as the false belief (or Sally-Anne) task or the surprising contents task. In these tasks, children are asked to predict what another person (usually their mother or another peer) knows or believes, based on the person’s perspective. For example, in the surprising contents task, children are shown a familiar box (such as a crayon box) of which they would normally know the contents of the box (e.g., crayons). The experimenter reveals to the child that instead of crayons inside the box, the experimenter has put something surprising: rocks. The experimenter usually checks with the child to make sure they know what is inside the box. Then, the experimenter shuts the box and asks the child the key questions. They first ask the child what they originally thought was in the box. Three-year-olds commonly say they always knew rocks were in the box, whereas five-year-olds say that they originally thought the box contained crayons. The experimenter then asks if their mother walked in the room and the experimenter showed them the box for the first time, what would their mother say is inside the box. Three-year-olds typically say that their mother would respond there are rocks inside the box, whereas five-year-olds understand that their mother would have a false belief and respond that there are crayons inside the box. Children who pass this task respond correctly that their mother would not know what was really inside the box. Younger children have difficulty with this task because once they know the reality that there are rocks inside the crayon box, they have a hard time remembering a time when they held a false belief and also cannot imagine that another person would not know the real contents.

Barrett and colleagues (Reference Barrett, Kurian and Johnson2001) adapted the ‘surprising contents’ task to investigate children’s understanding of human, non-human and supernatural minds. In their study, children were shown a cracker box that contained rocks and were then asked to predict what various agents (e.g., a tree, a bear, a snake, a mother and God) would think was inside the box if they had never seen it before. Consistent with prior research (e.g., Wellman et al., Reference Wellman, Cross and Watson2001), children under five years typically believed (incorrectly) that a human would think there were rocks inside the box. From five years, children stated correctly that humans would have a false belief (e.g., there were crackers inside the box). However, all aged children also stated (correctly) that God would know the box contained rocks. This study was replicated with Yucatec Mayan children (Knight et al., Reference Knight, Sousa, Barrett and Atran2004), showing that children of all ages think that God knows the correct contents, and only the older children respond correctly that their mother would not know the contents.

Based on these results, Barrett and colleagues concluded that young children may be cognitively flexible and prepared conceptually to reason about supernatural minds, particularly omniscient ones. Thus, they proposed the ‘preparedness’ hypothesis. They argued that children may find it easier to think about an agent like God – who knows everything – than to grasp the many and diverse limitations of human minds.

Other research has found similar results while using other types of theory-of-mind tasks. Following the publication of this initial study, Barrett, Newman and Richert (Reference Barrett, Newman and Richert2003) published another study where they gave children three tasks (the doodle task, the secret code and the ‘new game’) where children did not know (1) what a drawing could be because they just saw what looked to be a doodle, (2) the meaning of a secret code nor (3) the rules of a new game. In their ignorance, they answered questions about whether their mother, a dog or God would know the drawing, the code or the rules of the game. Similar to their first study, Barrett and colleagues found that by age four, children responded that God would have more knowledge than a dog or their mother.

Further experiments (Richert & Barrett, Reference Richert and Barrett2005) expanded on this by exploring how children reason when using a different theory-of-mind task that focusses on sensory knowledge. They gave children three tasks where children at first were not able to see (a drawing), smell (peanut butter) or hear (music). Then, the researcher revealed the stimuli to children. Once they could see the drawing, smell the peanut butter in the canister and hear the music on a tape recorder, children responded whether God, their mother and animals with particularly sharp sensory skills (e.g., an eagle who had special vision, a fox who had special hearing and a dog who had special olfaction). Again, young children attributed to God unique cognitive abilities and differentiated that God would have knowledge of the stimulus, compared to the other agents, even when it was far away, occluded or silent.

The previous four studies highlight four different methods and samples of children and results seem to converge that very young children differentiate a supernatural mind (God) from human minds. However, other researchers have found different results. For instance, Giménez-Dasí and colleagues (Reference Giménez-Dasí, Guerrero and Harris2005) used the same task as Barrett et al. (Reference Barrett, Kurian and Johnson2001) and found that Spanish children under age five did not attribute omniscience to God in the same way: the youngest children (three- and four-year-olds) did not attribute omniscience to God like the previous studies, and children did not differentiate between a human mind and God.

Likewise, two researchers from Greece (Makris & Pnevmatikos, Reference Makris and Pnevmatikos2007) found similar results to Giménez-Dasí et al. (Reference Giménez-Dasí, Guerrero and Harris2005). They questioned whether the children in Barrett’s studies were biased by their own knowledge of the contents of the box and attributed knowledge to all agents because that was their own knowledge state. Makris and Pnevmatikos (Reference Makris and Pnevmatikos2007) proposed that a better test would be an ignorance theory-of-mind task where children were not made aware of the surprising contents, the rules of a new game, the details of the picture and so on. They proposed that instead of being ‘prepared’ to think of supernatural minds, understanding of different minds develops gradually across the preschool years and follows the trajectory of understanding of human minds. Indeed prior work with humans show that children develop an understanding of their own ignorance by 4.5 years, before understanding ToM (Wellman & Liu, Reference Wellman and Liu2004). Makris and Pnevmatikos (Reference Makris and Pnevmatikos2007) found that when they gave three- to seven-year-old children the surprising contents test, results were similar to Barrett et al. (Reference Barrett, Kurian and Johnson2001) in that children younger than age five years, attributed God with knowledge of the contents of the box and attributed a human with no knowledge. However, when children were given the ignorance task, where children remained ignorant of the contents of the box, children did not consistently attribute knowledge to God until age five. This study highlights that methods influence the way in which children respond to tasks, and based on the differences in responses between the two tasks, it seemed that children responded based on their own knowledge state rather than taking the perspective of another agent.

Further studies, including the work by Kiessling and Perner (Reference Kiessling and Perner2014) and Lane et al. (Reference Lane, Wellman and Evans2010), supported this gradual development hypothesis, showing that children under five years do not differentiate human from supernatural minds and tend to be more anthropomorphic in their reasoning, applying their own knowledge to both human and supernatural agents. In other words, younger children are more likely to attribute their own reality status to other minds, regardless of being human or a supernatural agent.

However, as children age, they become less anthropomorphic in their attributions to God. This may be because they become increasingly aware of and exposed to other’s beliefs and theological teaching. One study demonstrated that awareness of their caregiver’s beliefs in addition to a maturing ToM influences concepts of God: older children are less likely to anthropomorphise supernatural agents (including God) than younger children (Saide & Richert, Reference Saide and Richert2020; Reference Saide and Richert2021).

To summarise, these seminal studies were important as they demonstrated that methodology influences the way in which children answer questions about perspective taking. Specifically, these studies showed that the youngest children (three-year-olds) are easily influenced by their own knowledge state and apply this to other agents, rather than reasoning that other minds might have a different perspective or knowledge state based on whether the other mind has limited or extraordinary abilities.

These studies underscore the importance of methodology and children’s cognitive biases when exploring their understanding of other minds. All researchers, regardless of whether they promote any of the previous accounts, agree that very young children can think about supernatural minds. Further, all agree that conceptualisation of God is possible because of our cognitive biases (e.g., ToM, HADD and teleological reasoning) that are used to understand the natural world (of people, non-human animals, plants), and are also used (some might say as a by-product) to think about the supernatural.

The findings also suggest that cultural and social context play a crucial role in shaping children’s understanding of supernatural agents. As research in this section has shown, and much more current work will show in Section 6, that children’s understanding of God concepts is susceptible to cultural influences. Religious or cultural concepts are not in themselves inherent; they require experience, observation and learning from caregivers, peers and their community. For example, children raised in religious communities, such as those attending religious schools or living with religiously active parents, tend to develop more theologically accurate concepts of God at an earlier age (Lane et al., Reference Lane, Wellman and Evans2012; Lam & Guerrero, Reference Lam and Guerrero2020, see Section 6 for more examples).

And so far, in the last ten years, new studies are showing striking diversity and variability in children’s development of God concepts, and socio-cultural influence and context have a large role to play. For example, many young children (by four years) are more likely to differentiate human minds from God’s mind if they attended a religious school over a state school (Lane et al., Reference Lane, Wellman and Evans2012) or had religiously active parents (Lam & Guerrero, Reference Lam and Guerrero2020).

Recent studies by Burdett et al. (Reference Burdett, Wigger and Barrett2021) have demonstrated that children from different cultures (e.g., Israeli Modern Orthodox Jewish children and Protestant children in Kenya, the Dominican Republic and the UK) show complex and varied responses about God and other supernatural agents. This suggests that, far from being universal, the way children conceptualise God is deeply influenced by their cultural and religious context. However, being ‘Jewish’ or ‘Christian’ in these contexts does not mean that these children think the same about God or humans. Each of these children may have very different social interactions, experiences and theological conversations. This variability highlights the importance of considering socio-cultural factors in the development of God concepts.

3.2 Shifting Perspectives: From Cognitive Biases to Social Learning Biases

In the past decade, there has been a shift in developmental research from the ‘preparedness’ and ‘anthropomorphic’ accounts to a more constructivist or socio-cultural approach. Researchers now emphasise that religious and cultural concepts are not innate, but are shaped by experience, socialisation and learning from caregivers and peers. As children mature, their cognitive capacities, such as ToM, evolve to support more complex understandings of God, theological terminology and religious practices.

This socio-cultural perspective has led to greater recognition of the diversity in children’s development of God concepts. For instance, children in different countries (e.g., Albania, Austria, Japan, Indonesia, Israel, Kenya, the UK, Dominican Republic, Spain, Greece, Yukatec Mayan and the U.S.) all eventually develop an understanding that God possesses special knowledge that differs from human minds (Burdett et al., Reference Burdett, Barrett and Greenway2020; Burdett et al., Reference Burdett, Wigger and Barrett2021; Gimenez-Dasi et al., 2007; Kiessling & Perner, Reference Kiessling and Perner2014; Knight et al., Reference Knight, Sousa, Barrett and Atran2004; Lam & Guerrero, Reference Lam and Guerrero2020; Makris & Pnevmatikos, Reference Makris and Pnevmatikos2007; Moriguchi et al., Reference Moriguchi, Takahashi, Nakamata and Todo2019; Nyhof & Johnson, Reference 83Nyhof and Johnson2017; Wigger et al., Reference Wigger, Paxson and Ryan2013). However, the development of a God concept is not uniform, and social factors such as religious upbringing and cultural teachings are key to shaping children’s views of God.

The seminal studies mentioned earlier demonstrated that young children can differentiate between human and divine minds, though the extent and nature of this differentiation depend on both the task methodology and the child’s socio-cultural context. While debates continue about whether children’s reasoning is due to an inherent ‘preparedness’ to understand supernatural beings or a gradual development of perspective-taking abilities, the growing body of research underscores the complex and context-dependent nature of children’s understanding of God. The next section broadens the description of research from just the God concept of omniscience to discuss several other attributes of God.

4.1 Immortality

A key distinction between many supernatural and natural agents is immortality, or the ability to live forever. Immortality is a common and prevalent supernatural attribute. For example, many supernatural agents that children (in the Western Hemisphere) hear about are immortal, such as angels, ghosts, elves, the tooth fairy, Santa Claus and God.

Unlike the attribute of omniscience, where children may understand knowledge states in their preschool years, the attribute of immortality develops later in childhood. To comprehend immortality fully, children must differentiate basic biological concepts that apply to humans, such as that many biological processes start after conception/before birth (Emmons & Kelemen, Reference Emmons and Kelemen2015) and cease upon death, in addition to understanding that biological attributes do not apply to some supernatural agents.

Some work has shown that children as young as four years old can differentiate the differences in biological phenomena between humans and robots (Saylor et al., Reference Saylor, Somanader, Levin and Kawamura2010) and human and non-human animals (Herrmann et al., Reference Herrmann, Waxman and Medin2010). These researchers argue that children need not think about other agents from an anthropomorphic perspective, but are able to draw distinctions among these different agents. However, many researchers argue that experience and learning about the natural world is a key to understanding biological phenomena, as many phenomena are not intuitive, such as understanding death.

Some researchers claim that an understanding of biology requires that the child develops a coherent concept of life (and death) and bodily function (Inagaki & Hatano, Reference Inagaki and Hatano2002; Slaughter & Lyons, Reference Slaughter and Lyons2003). Research indicates that a mature concept of death includes, but is not limited to, five subcomponents: (1) universality, the notion that all living things will die; (2) inevitability, or the idea that a living thing will eventually die; (3) irreversibility, or that once a living thing is dead, it cannot be brought to life again; (4) cessation, or that all mental and bodily functioning ceases with death; and (5) causality, or that death is caused by the breakdown of the body. These subcomponents develop with age: between the ages of three and five, children understand universality and irreversibility (Brent, Speece, Lin, Dong, & Yang, Reference Brent, Speece, Lin, Dong and Yang1996; Slaughter & Lyons, Reference Slaughter and Lyons2003), by age seven or eight, children understand cessation and causality (Bering & Bjorklund, Reference Bering and Bjorklund2004; Slaughter & Lyons, Reference Slaughter and Lyons2003) and between the ages of five and seven years, children understand a concept of inevitability (Speece & Brent, Reference Speece and Brent1984).

The previous text suggests that children understand death better after the preschool years. However, this work does not point to whether understanding death hinges on their understanding of immortality. Two studies suggest that preschool children can understand a concept of immortality and perhaps before they fully understand death. In an initial study, Giménez-Dasí and colleagues (Reference Giménez-Dasí, Guerrero and Harris2005) asked three-to-five-year-old children to respond to four questions about a human and God: (1) ‘When there were dinosaurs in the world, did ____exist?’, (2) ‘Will____get older and older or stay the same?’, (3) ‘Will ____die or go on living forever and ever?’ (4) ‘Was____ a baby a long time ago?’ Answers were summed to form a ‘mortality score’ for the human agent and God. They found that three-year-olds did not clearly distinguish between God and the human, and although older children differentiated between agents, their scores for God were at chance. However, results are not clear whether older children were at chance because of lack of understanding of God, lack of understanding of biological concepts or perhaps children were confused whether they should answer questions about God or Jesus. If children were from a Christian or Catholic tradition, Jesus (as God) was once a baby, aged and eventually died (although now lives forever). To clarify this, Burdett et al. (Reference Burdett and Barrett2016) replicated these questions but asked both a Christian and Jewish sample to see if Jewish children (who do not have Jesus in their theology) would have clearer differentiation between human and God. Burdett et al. (Reference Burdett and Barrett2016) also added one more question: (5) ‘Do you think____had a mom and dad?’ to create an overall index score of 5. Burdett et al. (Reference Burdett and Barrett2016) found that three-year-olds differentiated between God and a human but that older children were much better at attributing correct biological phenomena to each agent. Additionally, they found that Jewish Israeli children were much better at attributing correct immortal/non-biological characteristics to God than the Christian English children. They rationalise that Christian theology is more complex as Christians learn both about Jesus (Divine human) and God. Although just two studies, this work suggests that children are able to use their biological experience and cultural knowledge to reason about immortality amongst human and supernatural agents.

4.2 Omnipotence

Another divine attribute is omnipotence, or being all-powerful. Although there are only two studies that directly test children’s understanding of omnipotence, understanding supernatural versus human power draws on literature and work examining children’s growing understanding of whether events are improbable, impossible, fantastical or real. In the following text I lay out the background of how children understand human limitations to contrast with how children understand the divine capabilities of omnipotence.

4.3 Omnipresence

The concept of omnipresence – being present everywhere at once – is the final attribute discussed in this section. Compared to the other attributes of God (such as omnipotence and omniscience), omnipresence has received less experimental attention. This attribute also differs from the others because it relies on children’s understanding of physical space and how objects are constrained to specific locations.

Studies with infants show that even very young children have a basic understanding of the physical world. For example, studies demonstrate that by five to nine months, infants understand object permanence – the idea that objects continue to exist even when they are out of sight (Moore et al., Reference Moore, Borton and Darbly1978). Researchers know this based on looking-time methodology, where infants show consistent surprise and prolonged looking when an object disappears or is not in the location it should be (counterintuitive event), compared to conditions where the object is where it should be and infants lose interest (intuitive event). Additionally, five-month-old infants can grasp basic arithmetic concepts, such as not being surprised when seeing a display that “1 puppet + 1 puppet = 2 puppets” and are surprised when 3 puppets remain after subtracting one from a group of two (Wynn, Reference Wynn1992). By the end of infancy, children also understand object individuation – the ability to distinguish one object from another (Baillargeon, Reference Baillargeon2004).

These early cognitive abilities suggest that children intuitively understand that objects and entities are bound by space and can only be in one place at a time. This makes the idea of someone being in multiple places at once – especially everywhere at once – counterintuitive to young children.

4.4 Other Attributes Given to Supernatural Agents

The previous attributes are largely ones given to God or ‘High Gods’; however, belief in other supernatural agents is widespread, and the characteristics attributed to them can vary widely. Indeed, many supernatural agents are more limited. For example, some are anthropomorphic and embodied, while others are disembodied with a few supernatural characteristics. This section details further research regarding whether children attribute other types of human properties to supernatural agents, such as various psychological, biological and physical properties.

In general, studies have shown that children readily attribute different types of human attributes to supernatural agents. For example, unlike American adults, who attributed more psychological properties (e.g., thinking, talking and dreaming) to supernatural and fictional agents, five-year-olds attributed psychological, physiological (e.g., sitting, stretching and jumping), and biological properties (e.g., eating, growing and sneezing) to both types of agents (Shtulman, Reference Shtulman and Schulz2008).

Other patterns emerged when children were recruited based on cultural backgrounds and religion. Similar to the above study with American children (Shtulman, Reference Shtulman and Schulz2008), American Christian children (Richert et al., Reference Richert, Saide, Lesage and Shaman2017) attributed human properties to God, but Muslim children did not attribute human properties to Allah. In contrast, both Muslim and Hindu Indian children anthropomorphised fictional supernatural agents (ghosts and fairies) and Hindu gods (Ganesha and Krishna), but responses varied for whether they attributed human characteristics to the more abstract Islamic agents (Muhammad and Allah): Muslim children attributed Muhammad with the most human properties, whilst Allah was attributed with the least (Shtulman et al., Reference Shtulman, Foushee, Barner, Dunham and Srinivasan2019). On the other hand, Hindu children attributed more human attributes to both agents. Thus, whilst Muslim children differentiate Muhammad and Allah, Hindu children do not.

Researchers conclude that the differences in responses may be reflective of the different theological teachings of each religion. For example, Jesus and the Hindu gods, Ganesha and Krishna, are all depicted in human form, whereas Islamic theological teaching stress that Allah should not be depicted in human form. Further work has also shown that some children may be more likely to anthropomorphise based on how much the child ascribes ‘animacy’ (or the degree to which the agent acts) and the perceived ‘aliveness’ of the agents (or whether the child sees that agent as alive or dead) (Saide & Richert, Reference Saide and Richert2021). Thus, there are interesting patterns that emerge depending on age (i.e., younger children are more likely to attribute more human-like properties than adults), religious background (i.e., Muslim children anthropomorphise less than children of other traditions) and individual variation in concepts (i.e., children who see God as ‘alive’ attribute anthropomorphic properties). This pattern suggests other variables are influential in concepts of God beyond just cognitive biases.

This section highlighted a broader picture of how young children view various aspects of God and other supernatural agents. This work confirms that children still use intuitive biases to think about supernatural agents, but these concepts are shaped by culture. The next section broadens the scope of work on God concepts discussed in the previous section by discussing work on children’s understanding of other concepts concerning religion.

5.1 The Influence of Supernatural Agency: Princess Alice

The first developmental study to examine how behaviour is influenced by perceived supernatural presence is the ‘Princess Alice’ study. This early study launched work on children’s early understanding of how children perceive supernatural agency and whether perceived presence influences moral behaviour. This idea is rooted in what has been called the supernatural monitoring hypothesis, which suggests that people may be motivated to behave better if they believe a supernatural agent is monitoring them (Gervais & Norenzayan, Reference Gervais and Norenzayan2012; Shariff & Norenzayan, Reference Shariff and Norenzayan2007). The hypothesis links the idea of supernatural surveillance to larger explanations about the origin of religion and its role in fostering cooperation within societies (i.e., that surveillance from a supernatural being influences prosocial behaviour among people).

The supernatural monitoring hypothesis was inspired by prior work showing that adults alter their behaviour when they believe a supernatural agent could punish bad behaviour. Work has shown that adults have better behaviour when primed (shown a particular stimulus before a task) with religious words on a computer task (McKay et al., Reference McKay and Efferson2010; Shariff & Norenzayan, Reference Shariff and Norenzayan2007), attributes of God (Lin & Suarez, Reference Lin and Suárez2020) or when they believe a ghost is present (Bering et al., Reference Bering, Blasi and Bjorklund2005). For example, when adults were primed first with words like ‘God’ or ‘sacred’, they behaved more generously in economic games (Shariff & Norenzayan, Reference Shariff and Norenzayan2007), compared to adults in a control group who heard neutral words. Similarly, adults who believed a ghost was present were less likely to cheat on a competitive task (Bering et al., Reference Bering, Blasi and Bjorklund2005).

Children also behave similarly. Piazza et al. (Reference 84Piazza, Bering and Ingram2011) gave five- to six-year-old and eight- to nine-year-old children a task that was so difficult that when left alone, there was the potential to cheat. They also created a fictional, non-threatening supernatural agent, Princess Alice, to ensure participants had no prior knowledge or experiences with the agent. Children were told they could win a prize if they could throw a ball and have it stick to the centre of a Velcro target. However, children were given specific rules: they could only throw from a particular distance (marked by a line on the carpet), throw with their non-dominant hand and have their back to the target. That main manipulation was the presence of an agent during the experiment via three conditions where children completed the task: (1) with the experimenter sitting in a chair in the room, (2) by themselves and with an empty chair in the room or (3) with Princess Alice sitting in the chair in the room. The dependent variable was whether children cheated or not. If the supernatural monitoring hypothesis is true, the researchers predicted that children in the Princess Alice condition, who believed Princess Alice was present, would be less likely to cheat during the task compared to children who were either in the conditions where they were alone or with a researcher. They found that in the Princess Alice condition, children who professed belief in Princess Alice were significantly less likely to cheat than children who did not profess belief in Princess Alice and were significantly less likely to cheat than those who were in the alone condition. This finding suggests that belief in a supernatural presence can reduce dishonest behaviour in a way similar to being monitored by a person.

These results, along with adult studies (Gervais & Norenzayan, Reference Gervais and Norenzayan2012; Shariff & Norenzayan, Reference Shariff and Norenzayan2007), support the idea that belief in a supernatural presence – whether it is a high God or a low-level supernatural agent – in addition to one’s fear of supernatural retribution, can motivate people to behave morally. This can help explain why belief in moralising gods (as in Christianity, Judaism and Islam) may encourage prosocial behaviours, like cooperation and honesty, if one feels that God is watching your behaviour.

Other work has further explored how belief in supernatural agents influences cooperation. One large study examines this idea across eight diverse communities with different religions, including Christianity, Hinduism and Buddhism (Purzycki et al., Reference Purzycki, Finkel, Shaver, Wales, Cohen and Sosis2012). Participants were given a coin distribution task where they could choose to allocate coins to themselves or to an anonymous person from either their own community or a distant one. They hypothesised that people would be most likely to put money in their own cup and the cup of someone from their own village, but that belief in a moralising God would encourage rule-following of the game. The study found that people from communities with moralising gods – those that punish or oversee human behaviour – were more likely to follow the rules and distribute coins fairly. In contrast, people from communities without moralising gods were more likely to keep the coins for themselves or favour members of their own community.

These findings support the idea that belief in supernatural punishment, whether from a high God or other types of supernatural agents, encourages prosocial behaviour (Lightner et al., Reference Lightner, Bendixen and Purzycki2022; Watts et al., Reference Watts, Greenhill, Atkinson, Currie, Bulbulia and Gray2015). The belief that one might be monitored or judged by a divine figure seems to promote rule-following, generosity and cooperation, even with strangers, enabling people to live and work together in larger groups. As demonstrated by the Princess Alice study, it seems children’s moral behaviour is also influenced by the perceived presence of a supernatural agent.

The idea that belief in supernatural punishment drives moral behaviour raises important questions about atheists and cultures without punitive gods. Interestingly, research suggests that explicit belief in a supernatural agent may not always be necessary to trigger prosocial behaviour (Watts et al., Reference Watts, Greenhill, Atkinson, Currie, Bulbulia and Gray2015). For instance, priming participants with the idea that a ghost had been in the room was enough to increase their heart rate and galvanic skin response, suggesting that implicit belief in supernatural monitoring may also influence behaviour (Bering et al., Reference Bering, Smith, Stojanov, Halberstadt and Hughes2022).

The ‘Princess Alice’ study and related research suggest that the belief in being observed by a supernatural agent can lead to more ethical behaviour, much like being monitored by another person. This aligns with the idea that belief in moralising gods or the concept of divine surveillance can encourage prosocial behaviours that promote cooperation in large societies. However, as studies on atheists show, the mere perception of supernatural monitoring – whether or not one explicitly believes a god – can also influence moral behaviour. However, the link between belief in God and supernatural agents, their perception of their presence and how this influences behaviour is still understudied in children. Much more work should be done to understand this link.

5.2 Children’s Understanding of Origins

The next group of studies were inspired after the initial studies demonstrating children’s teleological bias (Kelemen, Reference Kelemen1999a), or a default to thinking that objects and people were created for a purpose. These studies are central to understanding children’s concept of biological origins, particularly when it comes to the natural world and existence.

One of the by-products of people holding a teleological bias is that children and adults may be biased to see the origin of the world and humans to be designed by someone, and in particular, God. Prior work has shown that both children and adults often misunderstand biological origins. For example, even college undergraduates with nearly two years of biology education misinterpreted natural selection concepts, even after participating in extra lectures, activities and problem sets (Bishop & Anderson, Reference Bishop and Anderson1990). This raises the question: If adults struggle to grasp the origins of species, how do children understand these complex concepts?

Margaret Evans (Reference Evans2000) asked five- to twelve-year-old children two questions: ‘How do you think the first dinosaur got here on earth?’ and ‘How do you think the first humans got here on earth?’ She also told a story to the children where scientists discovered a novel animal that looked like a pig and a squirrel, called the ‘spiggle’, on an island. She then asked, ‘how do you think the first spiggle got to be on the island?’ Children’s responses were coded according to four different explanatory types:

1. 1. Spontaneist generationist: responses that mentioned these animals just ‘appeared’ or ‘grew’,

2. 2. Creationist: responses that mentioned God, a human or another creature created or put the animal on earth,

3. 3. Evolutionist: responses that discussed one species becoming another species, or

4. 4. Hybridizationist: responses that suggested two animals mated to become a different species (e.g., the squirrel and pig gave birth to a spiggle).

Results showed that the youngest children (5.5–7.5 years) were more likely to give spontaneous and creationist explanations, the middle group (7.5–10.5 years) were more likely to give creationist explanations and the oldest children (10.5–12.5 years) and a sample of adults were more likely to give evolutionist accounts. Comparing these beliefs with their own knowledge of natural history and their parents’ views revealed that children’s beliefs about origins often reflected their parents’ beliefs. Older children who expressed creationist views were more likely to have low natural history knowledge and parents who endorsed creationism. Conversely, older children who expressed evolutionist views were more likely to have higher natural history knowledge and parents who believed in evolution. Parents’ beliefs (regardless of whether they were evolutionist or creationist) were not related to responses about origins for the younger children.

Further research by Evans (Reference Evans2001) and Tenenbaum and Hohenstein (Reference Tenenbaum and Hohenstein2016) confirms this pattern. For example, Evans (Reference Evans2001) asked five- to thirteen-year-old American children raised either in fundamentalist Christian or non-fundamentalist homes about origins of various animals and a human. Children aged five to ten from fundamentalist Christian homes were more likely to favour creationist explanations. Only children aged eleven to thirteen years from non-fundamentalist homes showed a shift toward evolutionist explanations. In a more recent study, Tenenbaum and Hohenstein (Reference Tenenbaum and Hohenstein2016) recorded conversations about origin beliefs between caregivers and their seven- to ten-year-old children. They found that children’s origin beliefs often mirrored those of their parents – children of evolutionist parents tended to adopt evolutionist views, while children of creationist parents adopted creationist views.

Tenenbaum and Hohenstein (Reference Tenenbaum and Hohenstein2016) asked further questions to determine how children rationalised their responses. They found that the youngest and mid-group children struggled with the concept of existence. In other words, children seemed to think that current species have always existed in their current form. Part of this erroneous thinking comes from another bias that young children hold: essentialism – the belief that species have an unchanging, inner essence that determines their traits and behaviours. This bias leads children to believe that species are fixed and cannot change over time (Gelman et al., Reference Gelman, Coley, Gottfried, Hirschfeld and Gelman1994; Springer & Keil, Reference Springer and Keil1989).

Shtulman (Reference Shtulman2006) argues that this essentialist bias makes it difficult for children and adults to understand variability within species, which is a core concept in natural selection. Children make inferences about an animal’s appearance or behaviour based on the species with failure to recognise that members of one species vary. To examine whether children accept variability in behaviour and anatomy, Shtulman and Schulz (Reference Shtulman and Schulz2008) asked children (ages four to nine years) and adults questions about the variability of anatomical and behavioural traits within a species. They found that both children and adults were resistant to accepting within-species variation, though they were more likely to accept variability in behaviour than in anatomy. Adults who were more open to variability were also more likely to understand natural selection.

Recognising that misperceptions about natural selection are rooted in early developing cognitive biases, such as essentialist and teleological thinking, Kelemen et al. (Reference Kelemen, Emmons, Seston Schillaci and Ganea2014) set out to see if she could find a way to teach children the principles of natural selection. She designed a storybook called, ‘How the Piloses Evolved Skinny Noses’ to see if she could encourage learning of adaptation by natural selection through a theoretically coherent storybook. After reading the storybook, five- to eight-year-old children answered questions correctly and explained adaptation by natural selection correctly a day later and also three months later. Furthermore, when another group of children was given the same storybook but a storyline that included a more gradual process of adaptation across multiple generations, children were still able to learn these concepts and generalise them. This storybook intervention reveals that a theoretically rich, coherent and comprehensive instruction can help children suppress competing cognitive biases and learn complex concepts like natural selection.

Other work has shown that the way in which people present information about animals and species influences children’s understanding of variation. Emmons and Kelemen (Reference Emmons and Kelemen2015) told five- to eight-year-old children that scientists had discovered a new animal (e.g., a hergob or orina) and described three traits: an internal trait, external trait and a behavioural trait. They found that when the focus was on the function of the traits, children used more essentialist reasoning and were not likely to say that there would be much variation in the newly discovered animal species. However, in another experiment where the researchers did not mention the function of the trait, seven- to eight-year-old children were more likely to accept variation within the species. This suggests that when children are primed with teleological reasoning (thinking that traits exist for a purpose), they are more likely to adopt essentialist beliefs and think of species as fixed. Without this priming, children are more open to the idea that species can vary.

Kelemen (Reference Kelemen2019) argues that understanding cognitive biases like teleology and essentialism is crucial for teaching children about evolution and natural selection, but also helping them understand how humans originate. She suggests that instruction that includes explanation-based biological and evolutionary processes helps children integrate these concepts early and build mental models of how natural selection works, reducing the likelihood of misconceptions that persist into adolescence and adulthood.

5.3 Children’s Understanding of the Afterlife

The final studies in this section focus on how children understand death and what they think comes after death. Many theorists think that to understand the afterlife, one must first understand the life-cycle processes, and that life begins and one day ceases. Recall the studies mentioned in Section 4, that by age five, Spanish, Israeli and British children universally understood that God had never been a baby, did not have parents and would not die, but that humans can and experience each of these things (Gimenez-Dasi et al., Reference Giménez-Dasí, Guerrero and Harris2005; Burdett & Barrett, Reference Burdett and Barrett2016).

An understanding of where people come from or intuitions of where people were before birth may help in the understanding of afterlife. To explore this, Emmons, Smith and Kelemen (Reference Emmons, Smith and Kelemen2016) examined how children from different cultural backgrounds – urban children in Ecuador and indigenous Shuar children – conceptualised two different periods: the foetal period, ‘when they were in their mother’s tummy’ and the infancy period, ‘when they were a baby’. The questions covered different categories: biological (e.g., could your eyes work?), psychobiological (e.g., could you be thirsty?), perceptual (e.g., could you watch something?), epistemic (e.g., could you think of others?), emotional (e.g., could you feel sad?) and desire (e.g., could you want something?). Results showed that both cultures only attributed desire and emotional capacities to foetuses and infants. In another study, Emmons and Kelemen (Reference Emmons and Kelemen2014) found that children from these same cultural groups in Ecuador also attributed desire and emotional capacities to the ‘prelife’ period, the period before a person is conceived. American children aged seven to ten years also make the same inferences that only emotional and mental life exist before conception (Kelemen et al., Reference Kelemen, Emmons, Brown and Gallik2021). However, this study also showed that religious belief systems can shape the beliefs of children and adults. Consistent with Mormon theological doctrine that says one experiences phenomena in premortal life, Mormon children from eleven years and Mormon adults were more likely to attribute emotional, epistemic, perceptual and bodily states to a being before conception.

These results on pre-life converge to suggest that children across cultures find some aspects of mentality (emotions and desires) to be an essential and inviolable core of human mentality. These results also suggest that mentality is both present pre- and after-life. However, cultural and religious doctrine can shape these beliefs. If children think that desires and emotions exist before birth, do children attribute them after death? And do children understand that biological processes cease upon death?

To answer these questions, first I will describe the comparative endpoint with adults. When American, Brazilian and British adults were asked what processes cease upon death, they were likely to say that biological (e.g., eating food, using the toilet), psychobiological (e.g., feeling hungry or sick) and perceptual processes (e.g., hearing or seeing things) cease but entertained the possibility that emotional (e.g., feeling angry or happy), desire based (e.g., wanting or wishing for something) and epistemic (e.g., believing or remembering something) processes exist and are part of the person (Bering, Reference 69Bering2002; Cohen et al., Reference Cohen, Burdett, Knight and Barrett2011).

Studies with children in the UK, Israel, Greece, United States, Madagascar and Spain show that children also think that mental capacities continue even after death (Astuti & Harris, Reference Astuti and Harris2008; Bek & Lock, Reference Bek and Lock2011; Bering, Reference 69Bering2002; Bering & Bjorklund, Reference Bering and Bjorklund2004; Bering, Hernandez Blasi, & Bjorklund, Reference Bering, Blasi and Bjorklund2005; Harris & Gimenez, Reference Harris and Gimenez2005; Misailidi & Kornilaki, Reference Misailidi and Kornilaki2015). However, cultural variation exists and often this variation is influenced by exposure to death and the cultural teachings and beliefs of that community. For example, one study interviewed seven- to eighteen-year-old children and adults in America and in Vanuatu (Watson-Jones et al., Reference Watson-Jones, Busch, Harris and Legare2017). They asked seven biological and seven psychological questions after priming them (exposing participants to a stimulus that is likely to influence their subsequent behaviour). Children were either given a non-theistic prime (e.g., ‘I know a person named David…I’m going to ask you a few questions now that he is dead’) and a theistic prime (e.g., ‘I know a person named David…I’m going to ask you a few questions now that he is with God’). Although American children and adults responded that only psychological processes continued, adults and children from Vanuatu (aged seven and older) believed that both psychological and biological processes continue after death. In their explanations, children mentioned that a new body will be resurrected after death and that the dead will be given a new body in Heaven (Watson-Jones et al., Reference Watson-Jones, Busch, Harris and Legare2017). These explanations are based on their knowledge of scripture and teachings.

In another study, Mexican children (aged 3.5 to 7 years) reported that biological and psychological processes cease at death (Gutierrez et al., Reference Gutiérrez, Menendez, Jiang, Hernandez, Miller and Rosengren2020). Mexican children are surrounded by symbols of death and participate in Dia de los Muertos, an event where they believe that the dead come back to receive food offerings. Children in these communities spend time with their families gathering food and favourite objects of their dead relatives. They help decorate the graves and prepare the food, flowers and photographs to display in a location in their home and public spaces for the dead. Over 50 per cent of the sample of children had also experienced the death of a loved one and had attended a funeral. These Mexican children understood very early that biological and psychological processes cease much earlier than European and American children (Rosengren et al., Reference Rosengren, Miller, Gutierrez, Chow, Schein and Anderson2014). Gutierrez et al. (Reference Gutiérrez, Menendez, Jiang, Hernandez, Miller and Rosengren2020) speculate that this difference is in large part due to early socialisation and this early exposure to symbolism and rituals/celebration of the dead.

There are other cultural differences. In another study, Lane and colleagues (Reference Lane, Evans, Brink and Wellman2016) asked Chinese and American children aged four to twelve years and adults to answer questions about whether psychological or biological processes cease after death. They found that Chinese children up until age five years believe that psychological processes will continue, but from age five, Chinese children believe that all processes cease, even when primed to think about Qingming, a festival involving ancestors where families make ritual offerings to ancestors with the belief that they receive them. American seven- to eight-year-old children, but not Chinese children, believed in the persistence of psychological attributes, and this belief was related to hearing a religious narrative about death (e.g., U.S. version: the person was very sick and died. This person’s children go to the church to pray to God. Chinese version: the person was very sick and died. This person’s children mourn for them in the Qingming festival – a festival to honour ancestors) rather than a medical one (e.g., this person got very sick and was taken to a hospital). Interestingly, however, from five years, children in both countries responded that biological processes would cease upon death. When children were asked, ‘what happens after a person dies?’, the youngest children (five and six years old) mentioned the cessation of biological or psychological properties. Older children (seven to twelve years old) from both countries tended to answer this question by discussing a ritual or burial. The second most common response was to discuss supernatural states, such as ‘the person is now in heaven’ or ‘if they’re good, Allah will take them to some place’. However, older Chinese children were much less likely to give these responses and the researchers suggest that this is due to Chinese political and cultural discouragement of formal religious expression.

These seminal studies made a huge impact in understanding children’s God concepts and the understanding of religion. This body of work shows that children’s behaviour changes based on their belief that a supernatural agent is watching them, the teleological bias shapes children’s perception of the world to see it as created by a designer and that ideas about the afterlife are shaped by children’s developing understanding of pre-life and life processes. These studies contribute to the literature on God concepts by showing that a broader understanding of agency influences key concepts of the afterlife and origins (with connections to a God that has a hand in these processes).

The focus of the next section will be on recent work showing how much culture and experience shape religious cognition. Indeed, the early work on the development of religious cognition focussed on examining the cognitive biases of religion so that researchers could better understand universal aspects of religious development in children across cultures and religious traditions. Whilst understanding cognitive mechanisms is very useful, this approach failed to appreciate that cultural concepts require the integration of social and cultural learning. Understanding how children conceptualise supernatural agents not only sheds light on their cognitive development but also highlights the profound impact of cultural and religious contexts on these beliefs.

6.1 How Are Concepts Influenced by Caregivers?

The previous work focussed on how children judge the reality of various agents and how the influence of their community impacts these beliefs. Here, I focus on how the influence of caregivers and experiences within the home impact belief.

Firstly, the way in which caregivers talk about belief impacts how children interpret their belief in supernatural, religious and even scientific phenomenon. Religious caregivers are keen to support supernatural beliefs in their children (Braswell et al., Reference Braswell, Rosengren and Berenbaum2012; Yoo et al., Reference Yoo, Jiang and Rosengren2025) and they acknowledge that their beliefs and conversations impact the beliefs of their children. For instance, research by McLoughlin et al. (Reference McLoughlin, Davoodi, Cui, Kelly, Jennifer, Harris and Corriveau2021) examined caregivers in Iran, China and the U.S. and found that in both Iran and the U.S., caregivers believed their views on religion strongly influenced their children. However, in China, where religion is less emphasised, caregivers were more confident in their influence on science education than on religious beliefs. Caregivers in all three countries affirmed that they influenced their children’s knowledge of science. Further cultural differences emerged, in that caregivers from Iran were likely to say that religious influence mainly comes from themselves and family, whilst caregivers in the U.S. acknowledged that religious learning could come from other sources, including other family members and church.

Interestingly, further work in China has revealed some interesting differences looking at religious and secular families. Even though China is a secular country, Chinese caregivers and children (aged five to six and nine to eleven years) who identified as Christian were confident of the existence of religious entities and this confidence was seen even in the older children in the state school system. This Christian group in China also identified people in their immediate social circle when identifying testimony sources as justifications for their own beliefs (Davoodi et al., Reference Davoodi, Cui, Clegg, Yan, Payir, Harris and Corriveau2020). In contrast, Chinese caregivers and their five- to six-year-old and nine- to eleven-year-old children who identify as secular were sceptical about religious concepts (Cui et al., Reference Cui, Clegg, Yan, Davoodi, Harris and Corriveau2020).

Beliefs can also vary based on how caregivers talk about religion. Dore and colleagues (Reference Dore, Woolley and Hixon2019) and Harris and colleagues (Reference Harris, Pasquini, Duke, Asscher and Pons2006) pointed out that implicit statements like ‘God lives in Heaven’ tend to sound more certain than explicit ones like ‘I believe in God’. Children are sensitive to these linguistic cues and can detect when there is doubt. Canfield and Ganea (Reference Canfield and Ganea2014) found that parents who use phrases like ‘I believe’ or ‘I think’ when discussing religious concepts may unintentionally signal uncertainty to their children. Thus, the difference between implicit and explicit ways of saying something can mean the difference of sounding more certain or not. Conversely, when parents speak with certainty about religious entities (e.g., God, angels and Heaven), their children are more likely to believe in them (McLoughlin et al., Reference McLoughlin, Davoodi, Cui, Kelly, Jennifer, Harris and Corriveau2021).

Indeed, the influence of caregivers on children’s understanding of God is especially significant. How children hear and experience God through religious teachings from their caregivers shapes how children conceptualise supernatural agents. In a study by Richert et al. (Reference Richert, Saide, Lesage and Shaman2017), children from different religious backgrounds – Muslim, Protestant Christian, Roman Catholic and non-affiliated – were interviewed alongside their caregivers. The study found that children’s ideas about God were closely tied to how their parents conceptualised God. For instance, children whose parents had anthropomorphic (human-like) views of God were more likely to view God in the same way. In contrast, children whose parents held less anthropomorphic views saw God as more supernatural and distinct from humans. This was especially true for Muslim children, who tend to see God as all-powerful and abstract, and teachings are explicit that Allah has never taken human form. In comparison, children from non-affiliated homes often did not differentiate between God and humans, viewing them as more alike. Christian children fell in between, likely due to the influence of Christian theology, which teaches that God became human in the form of Jesus. Studies have found that Christian children are less likely to distinguish between God and humans compared to children from other religious traditions, such as Judaism (Burdett, Wigger & Barrett, Reference Burdett, Wigger and Barrett2021). This suggests that the presence of human-like depictions of God (such as Jesus) complicates children’s understanding of God as a supernatural being.

And secular, non-affiliated children also attribute human-like attributes. Non-affiliated children rely on an anthropomorphic understanding for God when they do not have the cultural scaffolding to conceptualise God in any other way (Heiphetz et al., Reference Heiphetz, Lane, Waytz and Young2016; Lane et al., Reference Lane, Wellman and Evans2012; Richert et al., Reference Richert, Saide, Lesage and Shaman2017).

Beyond supernatural attributes, how children view and feel about God often matched their caregiver’s attitudes. For example, a study of Dutch pre-schoolers found that children who saw God as distant had mothers who shared that view, while children who saw God as loving and caring had mothers with similar beliefs (de Roos et al., Reference de Roos, Ledema and Miedema2001; Reference de Roos, Ledema and Miedema2004). Parenting practices also correlated. Strict parenting styles correlated with views of a God as punishing, whereas a more ‘loving and caring’ style correlated with children who believed God to be loving and caring (de Roos et al., Reference de Roos, Ledema and Miedema2004). These studies highlight how caregivers’ conceptualisations of God directly influence children’s beliefs, often reflecting exposure to broader cultural and theological teachings.

In sum, children develop their concepts of supernatural entities, including God, through a combination of direct exposure to cultural teachings and the selective testimony of trusted figures like caregivers and community members. They pay attention to how confidently these figures speak about religious concepts, and their beliefs are shaped by both the degree of certainty in the testimony and the broader cultural consensus around them. However, more research is needed to understand how these concepts evolve over time and how other social and cultural influences – such as religious texts, religious leaders and peer groups – play a role in shaping children’s religious cognition as children grow.

6.2 The Perception of Religious Belief Varies Across Cultures

Harris and Corriveau (Reference Harris and Corriveau2021) propose that both children and adults come to believe in scientific and religious phenomena largely through hearing about them from trusted adults and community members. Research shows that both children and adults express more confidence in scientific concepts, like germs and electricity, than in unobservable phenomena, such as God (Harris et al., Reference Harris, Pasquini, Duke, Asscher and Pons2006). One reason for this difference is that there tends to be greater cultural consensus around the reality of scientific phenomena, while belief in supernatural entities like God is more culturally divided. Cross-cultural studies support this idea. For example, in Iran people generally show greater confidence in the existence of natural (scientific) entities than in supernatural (religious) ones (Davoodi et al., Reference Davoodi, Jamshidi-Sianaki, Abedi, Payir, Cui, Harris and Corriveau2019; Payir et al., Reference Payir, McLoughlin, Cui, Davoodi, Clegg, Harris and Corriveau2021). In contrast, in China, where religion is less emphasised, children’s beliefs in supernatural entities tend to reflect their parents’ beliefs (Cui et al., Reference Cui, Clegg, Yan, Davoodi, Harris and Corriveau2020).

Children are also aware of cultural consensus of these entities within their communities. As noted earlier, how caregivers and community members talk about religious beliefs shapes children’s perceptions. This awareness usually comes from explanations of belief from caregivers or others in their community about these entities. For example, some caregivers might say, ‘We believe in Jesus, but others may not’, signalling that belief in Jesus is not universal. In contrast, other caregivers might say, ‘Everyone believes in God’, signalling a more widespread cultural consensus (Canfield & Ganea, Reference Canfield and Ganea2014; McLoughlin et al., Reference McLoughlin, Davoodi, Cui, Kelly, Jennifer, Harris and Corriveau2021). In cultures where there is low consensus about the reality status of religious entities, such as in China, children justified their beliefs by identifying that their caregiver and family were the source of their information (Davoodi et al., Reference Davoodi, Cui, Clegg, Yan, Payir, Harris and Corriveau2020). In contrast, in Iran, where there is majority consensus, children did not readily identify a source for where they learned their information (Davoodi et al., Reference Davoodi, Cui, Clegg, Yan, Payir, Harris and Corriveau2020).

In summary, children learn about the supernatural entities in their communities by listening carefully to the testimonies of others within their family and communities. They pay attention to who they heard it from (e.g., their caregivers) and what they said (e.g., noting confidence, degree of belief among their community and the valence of how these beliefs are conveyed (e.g., loving or pushing God).

However, the previous work largely focusses on transmission of religious concepts between children and parents. Much more work is needed to understand how these concepts develop over time and what other social and cultural influences shape religious concepts. For example, an analysis of the relationship between religious concepts and the stories and texts that children read in their locations of worship or at home would be fruitful to understand how important texts and stories shape religious conceptualisation. In addition, understanding who the other influential people that children pay attention to for religious knowledge is important to understand how children select and trust the testimony they are being exposed to in their individual communities. Of course these authorities and spheres of influence will change as the child grows up and understands the importance of these figures to the development of religious concepts.

6.3 The Influence of Ritual on Religious Cognition

Children also learn about cultural traditions, social norms and religious practices through rituals. Rituals are a collection of behaviours that can be highly elaborate (e.g., some religious rites) to simple actions (e.g., crossing oneself), and they can range from highly emotional to very boring and tedious. Ritual serves many different purposes. Many psychologists and anthropologists agree that one of the main functions of ritual is its role in generating social glue, a strong individual and group identity, and motivates cooperation (Swann et al., Reference Swann, Jetten, Gómez, Whitehouse and Bastian2012; Wen et al., Reference Wen, Willard, Caughy and Legare2020; Whitehouse, Reference Whitehouse2004). A second function is that by imitating specific actions, children internalise the social, religious and cultural expectations of their community (Whitehouse, Reference Whitehouse2004).

Children are very capable learners of ritual. Many developmental studies have documented how well children imitate and learn by copying others. In fact, they not only imitate but often ‘overimitate’, or imitating all actions shown to them even when they know that some of them might be unnecessary or irrelevant to the task at hand (Horner & Whiten, Reference Horner and Whiten2005; Lyons, Reference Lyons, Young and Keil2007). In a now classic developmental study, Horner and Whiten showed three- to four-year-old children an opaque puzzle box. The box had an opening on the top with two dowels obstructing the opening. There was another opening at the front of the box with a door. Children watched as the experimenter used a tool to slide the dowels out of the way of the top of the box and then used the tool to tap inside the top opening three times. Then they watched as the researcher opened the front door, inserted their tool and retrieved a small prize from inside the box. The majority of children, when given a turn with the box and the tool, repeated the actions perfectly. The experimenter then showed the child a clear box and repeated the same actions. Children could see in this transparent box that the actions of removing the dowels and tapping were unnecessary. A ceiling at the top of the box prevented any action on the top of the box from any actions below it, where the prize was located. To the surprise of the researchers, children still faithfully copied every action. Follow-up studies reveal that children still ‘over-imitate’ even when they tell the researcher that they know the first actions are unnecessary (Lyons et al., Reference Lyons, Damrosch, Lin, Macris and Keil2011; Lyons et al., Reference Lyons, Young and Keil2007), when they are left alone (Whiten et al., Reference Whiten, Allan, Devlin, Kseib, Raw and McGuigan2016), when they have been given time to explore the task before the demonstration to see how the task works (Nielsen & Tomaselli, Reference Nielsen and Tomaselli2010) and when they are under competitive or timed pressure (Lyons et al., Reference Lyons, Damrosch, Lin, Macris and Keil2011; Lyons et al., Reference Lyons, Young and Keil2007). These researchers claim that children are doing the smart thing and learning from an adult even when they do not understand the full reasons for copying all actions (Horner & Whiten, Reference Horner and Whiten2005). By faithful imitation of these actions, children acquire skills and particular knowledge to be competent adults in their cultural context (Berl & Hewlett, Reference Berl and Hewlett2015; Nielsen et al., Reference Nielsen, Mushin, Tomaselli and Whiten2014; Nielsen & Tomaselli, Reference Nielsen and Tomaselli2010).

Indeed, imitation is an important social tool. Imitation provides a means for forming and maintaining relationships with others (Nielsen & Blank, Reference Nielsen and Blank2011; Over, Reference Over2020; Watson-Jones et al., Reference Watson-Jones, Legare, Whitehouse and Clegg2014); by imitating someone else, one shows that you have chosen to copy them and the other person may feel closer to you. Additionally, imitation can strengthen group membership; if all members of a group do things a certain way, the group feels a better sense of group identity (Buttelmann et al., Reference 71Buttelmann, Zmyj, Daum and Carpenter2013; DiYanni et al., Reference DiYanni, Corriveau, Kurkul, Nasrini and Nini2015; Schleihauf et al., Reference Schleihauf, Pauen and Hoehl2019). And, finally, imitation is fundamental for learning cultural conventions and norms (Herrmann et al., Reference Herrmann, Legare, Harris and Whitehouse2013; Legare & Nielsen, Reference Legare and Nielsen2015) and, in particular, plays a significant role in the transmission of religious beliefs and rituals.

6.4 MCI: Minimal Counterintuitiveness Theory

As introduced in Section 2, a content bias is a particular characteristic of a trait that attracts attention. Because it attracts interest, this trait might be more likely to be remembered and then transmitted to someone else. Indeed, examining what characteristics are recurrent and prevalent across successful, culturally religious concepts is important for understanding their attraction, persistence and transmission.

One prominent theory in CSR research is called the Minimal Counterintuitiveness Theory and is one such content bias. ‘Counterintuitive’ refers to events and ideas that conflict with intuitive assumptions grounded in lay theories of core knowledge domains, such as psychology, biology and physics. We have expectations for how persons and animals should behave or function according to these domains. For example, biological intuitions guide one to think that a cat births kittens rather than dogs or dragons. And psychological intuitions guide one to think about mental states to predict how another would behave and think. Further, a theory of physics helps one predict how physical objects move or react to movement. With these intuitive theories and expectations in mind, MCI theory proposes that counterintuitive content is appealing because they violate one or two ontological expectations about persons, non-human animals, plants and natural objects. For example, a ghost is a person that violates physical intuitive expectations (e.g., walking through walls). These tweaks to natural expectations of our intuitive assumptions about the natural world are what make counterintuitive concepts so interesting. These violations of our intuitive expectations demand more attention because we try to process something that is counter to our understanding of the intuitive concept. Because we are thinking more about them, they become advantageous for memory recall and transmission.

However, not just any counterintuitive concept is memorable and interesting. The key characteristic of MCI theory is that these concepts are mostly intuitive and only contain one or two violations. Indeed, work with adults suggests that the optimal concept has minimal counterintuitive violations, ideally one or two violations, for it to be both interesting and memorable (Barrett & Nyhof, Reference Barrett and Nyhof2001; Boyer & Ramble, Reference Boyer and Ramble2001). One or two violations seem to be the cognitively optimal number for being interesting enough to demand attention to require a bit more time to process (as this concept is inconsistent with intuitive schema). Indeed, a concept that has more than two violations (e.g., a talking plant that changes colours and gives birth to kittens) is hard to fathom and remember. In further support of this, one study confirmed that MCI concepts with only one or two counterintuitive properties were the most common concepts (79 per cent of concepts were MCI) among counterintuitive concepts in folktales from eight different geographical regions (Barrett et al., Reference Barrett, Burdett and Porter2009).

Work with children confirms that MCI concepts have a recall advantage. Children aged seven to nine years who listened to a story were much more likely to recall counterintuitive concepts containing only one or two violations compared to three violations and intuitive concepts (Banarjee et al., Reference Banarjee and Bloom2013). These results held for both immediate recall and recall one week later. Authors concluded that one or two violations of expectation may be a cognitive optimum for children: They are more inferentially rich and therefore more memorable, whereas three or more violations diminish memorability for target concepts. Other work confirms that older children and adolescents recall proportionally more MCI concepts than older persons (above the age of fifty) (Gregory & Greenway, Reference Gregory and Greenway2017). Indeed there seems to be a peak for remembering MCI concepts in adolescence (thirteen–twenty years) and early adulthood (twenty–twenty-eight years).

One interesting question to come out of this research is whether there is a cultural transmission advantage for counterintuitive concepts and narratives and what advantages there might be from good recall during this period of adolescence and young adulthood. Recently, researchers have called for more studies to focus on adolescence and young adulthood (Nakkawita & Heiphetz, Reference Nakkawita and Heiphetz2021). This period is known for enhancements in cognitive flexibility which can enhance adaptability and application of new knowledge (Zheng et al., Reference Zheng, Akaliyski, Ma and Xu2024). Perhaps the combination of this flexibility and increased religious exploration (Layton et al., Reference Layton, Hardy and Dollahite2012) makes this period ripe for learning and engaging with counterintuitive concepts and religion in general.

7.1 Coexistence of Religious and Scientific Ideas

While previous sections focussed on how children acquire religious concepts, this section emphasises how children integrate these concepts with other knowledge they gain, such as scientific, historical and cultural information. Children often encounter a wide range of explanations (e.g., historical, cultural, religious and scientific) for the same events, and one key question is how they manage to hold multiple ideas or explanations simultaneously.

Early theorists, such as Piaget (Reference Piaget1928), thought that young children gradually understand the world through natural rather than through supernatural explanation. Over time, it was assumed that technological advancement and education would lead societies to increasingly favour scientific explanations over supernatural ones (Vygotsky, Reference 88Vygotsky1978). However, these ideas have not proven to be true so far. Studies of adults show that many people continue to endorse both natural and supernatural explanations for various phenomena (Evans et al., Reference Evans, Legare, Rosengren, Ferrari and Taylor2010; Legare & Gelman, Reference Legare and Gelman2008).

Both children and adults integrate scientific and supernatural information into explanations of natural phenomena. Legare and colleagues (Reference Legare and Souza2012) identified three areas – illness, death and the origin of human life – where U.S. adults often provide both scientific and supernatural explanations (Astuti & Harris, Reference Astuti and Harris2008; Evans, Reference Evans2001; Harris & Gimenez, Reference Harris and Gimenez2005; Legare & Gelman, Reference Legare and Gelman2008). Further, adults may also find meaning in supernatural explanations to meet social, emotional and moral needs or functions (Davoodi & Lombrozo, Reference Davoodi and Lombrozo2022). However, these explanations are deeply influenced by cultural narratives. For example, a study of Christian adults in Vanuatu found that while they provided natural explanations for illness and death (such as illness being caused by eating something harmful), they also included supernatural explanations (such as illness being a punishment for a wrong they had done) (Watson et al., Reference Watson-Jones, Busch and Legare2015). In contrast, when asked about human origins, Vanuatu adults overwhelmingly endorsed a Biblical Creationist view without referencing evolutionary theory.

Similarly, studies show that children integrate both scientific and religious explanations. For instance, U.S. children often combine biological and religious explanations for illness, human origins and death (see Legare et al., Reference 79Legare, Evans, Rosengren and Harris2012, for a review). Although seven- to twelve-year-old children from Vanuatu also combine both types of explanations for illness like U.S. children, they only endorse theistic explanations to explain their origin (e.g., God made them and put them on the earth), and refer to scientific explanations (e.g., biological causes) to talk about death (Busch et al., Reference Busch, Watson-Jones and Legare2017). Busch and colleagues (Reference Busch, Watson-Jones and Legare2017) suggest that responses may be explained by the level of formal education. Children in Vanuatu have minimal exposure to evolutionary explanations, so a scientific account of human origins would not be familiar or an endorsed belief. However, Vanuatu has seen several health campaigns to eradicate tuberculosis and malaria, so communities are aware of how germs can cause illnesses. It seems children combine both supernatural (e.g., behaving badly) and natural (e.g., germ contact) explanations for the causes of illness. These cultural differences highlight the important role of cultural context in shaping children’s religious cognition.

7.2 The Role of Culture in Religious Beliefs

One of the central points of this Element is the importance of culture in shaping how children develop religious beliefs. As earlier sections have shown, culture is not a new consideration, but it is essential to understanding how children form religious ideas. Doing cross-cultural research presents unique challenges (e.g., organisation, long-distance collaboration, translation, conceptual differences, language barriers, expense and time in the field), but it is necessary to draw accurate conclusions about human development more broadly.

The issue of cultural representation in psychological research was brought to the forefront in 2010 by Joe Henrich and colleagues, who pointed out the over-reliance on WEIRD (Western, Educated, Industrialised, Rich and Democratic) societies in psychological studies. Their analysis revealed that 96 per cent of psychological studies were based on participants from WEIRD societies, which make up just 12 per cent of the global population and the majority of these studies comprise North American undergraduate samples (Henrich et al., Reference Henrich, Heine and Norenzayan2010). This overwhelming percentage of studies is an inaccurate representation of the world’s population. The consequences of not examining diverse populations are legion: without diverse studies, the conclusions from psychological studies are at risk of making inaccurate assumptions and incomplete understandings about human behaviour.

7.3 Diversifying Religious Research

Since the publication of Henrich et al. (Reference Henrich, Heine and Norenzayan2010), many researchers have emphasised the need to diversify their samples. Despite this, a review of recent studies shows that much of psychological research still relies heavily on non-diverse, Western samples in the industrialised world (Apicella et al., Reference Apicella, Norenzayan and Henrich2020; Singh et al., Reference Singh2024), and an increase in cross-cultural and diversified sampling of data has not yet occurred. Because of this slow uptake many researchers claim that samples without cultural diversity are no longer acceptable and that generalisation of results beyond the socio-cultural context of that particular sample is bad science (Nielsen et al., Reference Nielsen, Haun, Kartner and Legare2017; Singh et al., Reference Singh2024).

Diversification is key for understanding human development. To address this, researchers like Amir and McAuliffe (Reference Amir and McAuliffe2020) suggest strategies for diversifying samples, such as developing research questions driven by theoretical frameworks and formal modelling, creating standardised protocols, seeking out interdisciplinary collaborators that can help with ethnographic background, choosing samples that are hypothesis driven and analysing data to capture variation by multiple levels of analysis. Further, Rad et al. (Reference Rad, Martingano and Ginges2018) suggest that journals require reporting of sample characteristics, justifying the sampled population, tying the conclusions of results to the sampled population and thoughtful discussions of how culture and context might influence the results.

Although cross-cultural research is challenging and costly, there are many ways to ensure research is more culturally inclusive. For example, researchers could consider the environmental factors of the participant population and record the lived experiences of those children (Rogoff et al., Reference 85Rogoff, Dahl and Callanan2018). Researchers could also consider creating testing paradigms that are more naturalistic to increase validity and generalisability (Dahl, Reference Dahl2017).

In addition to diversifying cultural samples and designing ones that are inclusive, there is a growing recognition of the need to expand research on religious beliefs beyond monotheistic traditions, especially the Abrahamic religions of Christianity, Islam and Judaism (Norenzayan, Reference Norenzayan2016; Richert & Corriveau, Reference Richert and Corriveau2022). Research to date has been disproportionately focussed on Christianity, with limited attention given to the diversity of global religious traditions. Barrett, Kurian and colleagues (Reference Barrett, Kurian and Johnson2001) estimate that there are over 10,000 different religious traditions worldwide, and yet much of the research focusses on a narrow subset of these.

To address this gap, new projects are emerging that examine a wider range of religious traditions and cultural contexts. A good example is the work that Rebekah Richert is doing in her lab. She incorporates multiple methodologies and multiple religious traditions. Some of this recent work (many studies are described in the sections of this Element) examines the degree that intra- and socio-cultural factors influence conceptual development of, and belief in, supernatural agents (Richert et al., Reference Richert, Saide, Lesage and Shaman2017). This work explores the variation within Western communities rather than the variation across cultures (although see Richert et al., Reference Richert, Saide, Lesage and Shaman2017). In these studies, they recruited Protestant Christian, Roman Catholic, Muslim and secular families and used surveys, measures and tasks for both children and their caregivers to understand religious background and socio-cultural input.

Another excellent example is the work coming out of the Developing Belief Network (DBN). The DBN formed in 2020 and involves a consortium of developmental psychologists and anthropologists from forty-seven cultural and religious sites (www.developingbelief.com). This large-scale project collected longitudinal experimental data on children’s religious cognition across diverse cultures and religious traditions, in addition to rich ethnographic data that will contribute to a more comprehensive understanding of religious belief development worldwide. Together the consortium designed collectively a full study protocol in three different waves that incorporated surveys and tasks given to both children (aged four–ten years) and their caregivers, recorded child–caregiver conversations about various topics, and key demographics that included cultural background and ethno-histories (Weisman et al., Reference Weisman, Ghossainy, Williams, Payir, Lesage and Reyes-Jaquez2024; Williams-Gant et al., Reference Williams-Gant, Weisman, Amin, Ghossainy, Soueidan and Nissel2025). Tasks were designed with specific pilots in various cultural communities. Tasks were also carefully translated and back-translated to ensure reliable measures, and everything was pre-registered in the spirit of ensuring that all practices were transparent, open and culturally relevant.

This work will be a huge contribution to the field and tackles the pressing need for diverse samples, multiple religious traditions, rich ethnographic and observational data and the first studies of its kind to track children over two to three years. Although there have been two publications on the DBN’s Wave 1 and 2 Protocols (Weisman et al., Reference Weisman, Ghossainy, Williams, Payir, Lesage and Reyes-Jaquez2024; Williams-Gant et al., Reference Williams-Gant, Weisman, Amin, Ghossainy, Soueidan and Nissel2025), papers using data from this project are currently being written. Many publications will be surfacing from this project. Hopefully, this pioneering project will be the catalyst for stimulating future research and funding and be a model for showing how other research groups might work as a collective to do rigorous cross-cultural research in large teams.

In conclusion, this section has outlined the importance of cultural diversity and using innovative methodologies in understanding the development of religious beliefs in children. By integrating these approaches, future research can provide a more accurate and comprehensive understanding of how religious cognition develops across cultures and societies.

8.1 The Future of Other Methodologies

The future of developmental psychology and the study of religious cognition looks bright with a host of new technologies and methods available to shed light on new and old. Further I detail a few promising techniques and studies that have already contributed much to the field.