RESEARCH DOSSIER
Religious Belief Acquisition, Doubt, and Revision
Abstract. Religious belief acquisition, doubt, and revision constitute an interconnected field of psychological inquiry into how propositions about gods, sacred realities, revelation, supernatural causation, and religious authority become credible, remain uncertain, and change through experience and social interaction. This critical narrative synthesis examines developmental testimony, cultural learning, credibility-enhancing behavior, cognitive representations, reflective reasoning, epistemic trust, interpersonal conflict, religious doubt, longitudinal trajectories, and methodological disputes. The scientific literature demonstrates substantial effects of religious exposure on children's judgments about supernatural narratives; identifies cultural models and community relationships as important contexts of belief maintenance and questioning; and documents the limits of a simple causal account linking analytic-thinking primes to religious disbelief. Longitudinal studies show that alterations in belief, religious practices, institutional affiliation, and subjective well-being may follow distinct trajectories. The central synthesis is that religious belief change is best examined as multidimensional, socially situated, temporally extended epistemic activity. Methodological priorities include separating assent from behavior, distinguishing certainty from distress, testing cultural measurement invariance, measuring change prospectively, and identifying causal mechanisms through transparently reported experiments.
Keywords: religious belief; belief acquisition; religious doubt; belief revision; epistemic trust; testimony; cultural learning; credibility-enhancing displays; analytic thinking; deconversion; cognitive science of religion; longitudinal psychology; belief certainty; religious socialization; epistemic identity.
Introduction: The Research Object
Religious belief is sustained through several partially independent processes: acceptance of propositional claims, confidence in interpreters and institutions, habitual interpretation of experience, membership in communities, and the practices through which religious meanings become consequential. Acquisition concerns the emergence and stabilization of such commitments. Doubt concerns uncertainty, questioning, or conflict about their warrant. Revision concerns their modification, differentiation, reorganization, reinforcement, or abandonment over time. These phenomena belong to the psychology of belief formation and change, with distinctive religious content and social institutions as their objects.
The same person can endorse a religious claim with considerable confidence, intermittently question its meaning, continue participating in a ritual, and gradually change the grounds on which the claim is believed. Accordingly, a scientific analysis must avoid treating assent, certainty, spiritual practice, group membership, and subjective attachment as interchangeable measures. The large Dutch longitudinal study by Bleidorn et al. (2024) Bleidorn (2024) provides an especially useful demonstration: changes in religious practices around conversion need not coincide with a parallel increase in belief in God. A single religious-status label would obscure this temporal differentiation.
This monograph investigates the epistemic and psychological sequence through which religious propositions acquire authority and are subsequently maintained or reconsidered. Its central questions are how children and adults select trustworthy sources, how ordinary cognition represents extraordinary claims, how interpersonal relationships alter the costs and meaning of doubt, which kinds of evidence are associated with belief revision, and which research designs can identify the causal contributions of cognition and culture. Its object is belief-specific psychological transformation. Broad religious identity, institutional conversion, ritual regulation, enduring spiritual distress, and the psychometrics of religiosity have adjacent specialized dossiers.
An adequate synthesis incorporates the experience of religious believers, uncertain believers, agnostics, atheists, people with multiple religious affiliations, and people whose positions change without public disclosure. Religious traditions also vary in the role assigned to propositions, testimony, ritual, inherited obligation, and individual assent. This diversity is part of the empirical problem. An English-language sample of respondents assessing a personal creator God cannot automatically represent traditions in which sacred practice, ancestral relations, or non-theistic commitments organize religious life.
Conceptual Architecture of Religious Belief
Belief acquisition is the process by which a person comes to treat a religious claim as credible enough to influence judgments, expectations, speech, or action. Operational indicators include agreement with an explicit statement, judgments of the factual status of a religious narrative, confidence ratings, reports of religious experiences, and choices made in situations where a belief is relevant. No single indicator exhausts the construct. Especially in childhood, a correct retelling of a story demonstrates recall and cultural competence but may provide limited evidence of personal conviction.
Religious doubt refers to a state or process in which the truth, authority, coherence, or personal significance of religious claims becomes an object of uncertainty or critical inquiry. Doubt may be episodic or sustained, private or publicly expressed, emotionally neutral or distressing. The Religious and Spiritual Struggles Scale developed by Exline et al. (2014) Exline (2014) includes a doubt dimension defined specifically by distress over religious questions. That scale is valuable for the psychology of struggle, but it does not measure every form of epistemic uncertainty. A person can question a religious proposition calmly and systematically without experiencing religious or spiritual struggle.
Belief revision denotes a meaningful change in the content, strength, justification, or organization of a religious commitment. Revision can include strengthening an inherited doctrine after reflection, interpreting a literal proposition symbolically, changing beliefs about divine intervention, reducing certainty, adopting a different theological account, or ceasing belief. A change of institutional affiliation may accompany any of these, but affiliation is a different variable. A longitudinal model should track both belief direction and the stability of individual claims rather than assume a linear trajectory from faith toward disbelief or the reverse.
The proposed research architecture distinguishes five analytically separable levels: proposition (what claim is entertained); attitude (what confidence, certainty, or credence is attached); justification (what testimony, argument, experience, or authority is credited); expression (what is said publicly under specific incentives); and practice (what a person actually does). These five levels constitute a measurement and analytical framework for comparing research traditions, not an experimentally established five-stage theory. Their utility lies in preventing mistaken inference from a change in one observable variable to a change in all other dimensions.
Historical Formation of the Research Field
Scientific inquiry into the transformation of religious commitment developed through psychology of religion, developmental psychology, social psychology, anthropology, the study of conversion, and cognitive science. Early work asked how religious experience and conversion altered the organization of personal life. Subsequent survey and sociological traditions investigated intergenerational transmission, religious participation, community networks, and institutional change. The late twentieth-century development of experimental cognitive psychology made possible controlled tests of the representation of supernatural agents, belief testimony, and intuitive religious concepts.
A decisive example of the experimental turn is Barrett and Keil (1996) Barrett (1996), who investigated a discrepancy between participants' explicit theological descriptions and the concepts used during story comprehension. Their experiments showed that representations employed in an immediate processing task could anthropomorphize God despite more abstract doctrinal commitments. This result shifted attention from professed theological content to the cognitive operations involved in actually using a concept.
Another foundational development was research on testimony. Harris and Koenig (2006) Harris (2006) argued that many things children learn about science and religion are inaccessible to direct firsthand observation and depend on the evidence supplied by other people. That perspective reframed religious learning as part of a more general developmental problem: children must judge what and whom to believe when the relevant objects cannot be inspected directly. Later work could compare the treatment of religious narratives with that of historical and fantastical stories.
The expansion of cultural-evolutionary models created a third line of explanation. Henrich (2009) Henrich (2009) proposed that learners may treat behavior costly to a model as evidence of the model's commitment, a proposal known as credibility-enhancing displays. Lanman (2012) Lanman (2012) situated belief acquisition within a discussion connecting sociology and cognitive science. These contributions are historically important theoretical accounts; evidence about particular displays and their social context remains necessary before attributing an observed belief transition to that mechanism.
The twenty-first century also brought research into the relation between cognitive reflection and religious disbelief, along with direct replication, cross-cultural testing, and increasingly informative longitudinal designs. The field now possesses evidence that requires a differentiated account of cognitive processes, culturally transmitted credibility, and trajectories of questioning rather than a single universal psychological explanation.
Scientific Method and Evidence Boundaries
This work is a critical, structured narrative review of original experiments, longitudinal observational studies, cultural-comparative research, psychometric development, explicit theoretical arguments, and selected synthesis articles. The evidence was located through academic publisher records, PubMed, PubMed Central, and accessible full texts, with attention to study populations, operational definitions, design, causal reach, and replication. This synthesis has no preregistered search protocol, exhaustively screened database corpus, independent duplicate coding of eligibility, formal risk-of-bias matrix, or pooled quantitative effect estimation. It is therefore presented as a fundamental research monograph and critical narrative synthesis.
Sources are assigned different inferential roles. Controlled experiments can test the causal effect of a specific manipulation on a measured response within the studied setting. Cross-sectional associations can estimate relationships while leaving temporal and confounding explanations unresolved. Longitudinal within-person models help establish ordering and trajectories but still require causal identification when exposure and belief change are jointly influenced by social processes. Psychometric studies establish what a particular instrument measures under the populations and conditions examined; measurement quality cannot be assumed for new traditions and languages.
A productive comparison requires scrutiny of sampling. A psychology undergraduate sample in one country gives evidence about that setting. Representative national panels strengthen population description but may still be limited by item wording, attrition, institutional definitions, and the cultural location of the research. Religious belief sometimes receives a single numerical score, while complex traditions require distinctions among metaphysical propositions, practices, lineage commitments, institutional credibility, and confidence in doctrinal interpreters. This is a substantive threat to construct validity.
This research places special emphasis on failed replications and disagreements about religious belief as a cognitive attitude. Evidence that complicates an elegant theory belongs at the center of the synthesis, because revisions in theoretical confidence should follow a transparent appraisal of the data rather than the prominence of an initially striking finding.
Testimony, Authority, and Epistemic Trust
A child is rarely positioned to verify every religious proposition independently. Religious histories, divine attributes, miraculous interventions, and expected afterlife states are normally transmitted through conversation, texts, teaching, ritual narration, or observation of trusted members of a community. This is an instance of the broader human dependence on testimony. The developmental account in Harris and Koenig (2006) Harris (2006) shows why religious learning should be examined alongside learning about scientific entities: both may involve real-world claims about entities beyond the learner's immediate experience.
Epistemic trust is selective reliance on another person or source as a credible guide to what is true. A psychologically adequate model should separate trust in a person, trust in a particular proposition, and trust in the institution that authorizes the speaker. A child may regard a parent as loving and competent yet later question one religious claim. An adult may reject an institution's political or moral authority while retaining theological beliefs. Conversely, a person may participate in religious education for social reasons without strong assent to all its teachings.
Source selection changes with experience. Contradictory messages from family, school, congregations, peers, media, and independent scholarship create opportunities for comparison. New information may be assimilated into prior belief, discounted because its source is distrusted, or used to alter the interpretation of a religious claim. The inferential pathway depends on how source accuracy is evaluated and how the recipient understands the social consequences of publicly disagreeing.
Research should therefore elicit reasons for source trust, source-specific confidence, prior exposure, recall, and attitudes toward the claim before presenting new testimony. If a participant merely repeats a familiar authoritative formulation, the investigator cannot infer personal conviction without additional measures. Similarly, treating all testimony as credulity obscures the adaptive function of learning from others in an information-rich social environment.
Cultural Learning and Credibility-Enhancing Displays
Religious instruction transmits words and practices, but the credibility of those words is also shaped by the conduct of people who transmit them. In Henrich's (2009) cultural-evolutionary account Henrich (2009), actions that would be difficult to sustain without commitment may serve as evidence that a model genuinely holds the beliefs professed. The proposal is especially relevant to domains that cannot easily be settled through direct observation. A learner may judge a caregiver's sacrifices, consistency, or ritual participation as more informative than an isolated verbal assertion.
A credibility-enhancing display is a hypothesized source cue, not a guarantee of truth or evidence that every costly action conveys sincere belief. Costliness depends on resources, cultural expectations, incentives, and the possibility of strategic performance. An act that is costly to one person may be routine for another. Research must distinguish observation of behavior from the observer's inference that it certifies a proposition. It must also compare displays against parental warmth, institutional resources, repeated exposure, peer norms, and other covariates.
Lanman's (2012) treatment Lanman (2012) relates religious belief acquisition and secularization to the interaction of cognitive and sociological processes. This is a productive framework for thinking about differing contexts of transmission: a religious proposition may be memorable yet rarely accepted if few trusted adults publicly treat it as credible, whereas a community may reproduce a claim through dense and consistent networks of practice and testimony.
A strong causal test would randomize or experimentally vary perceived credibility while holding the proposition's wording constant, examine whether the manipulation changes belief rather than simple liking, and measure persistence after an interval. Where experimentally manipulating family religious conduct would be infeasible or unethical, longitudinal observational studies and carefully matched natural settings may provide complementary evidence, accompanied by explicit statements of their inferential limits.
Childhood Development and Supernatural Possibility
The experiments by Corriveau, Chen, and Harris (2015) Corriveau (2015) tested five- and six-year-old children's judgments about protagonists in realistic, religious, and fantastical stories. Across two studies, children exposed to religion through church attendance or parochial schooling were more likely to classify protagonists in religious stories involving divine intervention as real, whereas children without such exposure tended to classify them as fictional. The findings linked children's reality judgments to the contexts in which they had encountered religious accounts.
The significance lies in the specificity of the comparison. All groups generally classified protagonists in ordinary realistic stories as real; their judgments diverged markedly in stories involving religious and extraordinary events. Cultural exposure influenced how the children represented the boundary between an event that can happen and a story convention. The result therefore challenges a simple thesis that identical supernatural ontological commitments arise spontaneously in every child regardless of testimony and social learning.
The study measures responses to stories, not the entire structure or future permanence of individual faith. Church attendance and school context also reflect correlated features of families and communities, so an observed group difference cannot isolate the causal contribution of any one component of religious upbringing. The experimental organization of story stimuli strengthens interpretation of narrative-related judgment; the exposure comparison remains an observational contrast.
Developmental studies must also avoid confusing familiarity with endorsement. A child may recognize a miracle as a religious narrative, judge that adults in a community consider it real, and still vary in private certainty. Interview wording, parental presence, interviewer identity, and the perceived expected answer can all influence children's performance. Cross-tradition extensions are essential because both story conventions and theological conceptions of possibility vary widely.
Conceptual Representation and Theological Cognition
The contents of a religious proposition can be stored, interpreted, and used through representational processes that vary with the task. Religious traditions often provide elaborate descriptions of divine attributes, but immediate reasoning may depend on intuitive person-like representations. In three story-processing experiments, Barrett and Keil (1996) Barrett (1996) found that participants often interpreted God in anthropomorphic terms despite professing theological claims incompatible with those interpretations. Increasing attention to explicit theology could modify the pattern of anthropomorphization.
For belief acquisition, the important implication is that knowing a religious formula is a different achievement from consistently deploying the same conceptual structure in comprehension, memory, explanation, and action. A respondent who verbally describes an omniscient or nonspatial deity may process a narrative using ordinary assumptions about the capacities and location of agents. Researchers must determine which type of representation a question recruits. A survey of explicit theological endorsement and a timed comprehension task can yield different results without either measure being meaningless.
This distinction becomes crucial when people report that their beliefs have changed. A doctrinal revision may be visible in language before intuitive representations alter; implicit narrative expectations may remain familiar even after public disaffiliation. Such temporal dissociation is a plausible psychological research question arising from the representational literature. However, persistence of an automatic association cannot by itself prove that a person secretly endorses a religious proposition. Familiarity, learned language, and affective salience can generate task responses independent of current conviction.
The work of Norenzayan, Gervais, and Trzesniewski (2012) Norenzayan (2012) explored associations between mentalizing and personal-God belief across several samples. The findings illustrate that characteristics of cognition may predict some religious beliefs under some conditions. Mentalizing performance is not an all-purpose index of religious capacity, and lower belief in a personal God is not interchangeable with a lack of spiritual commitment. Generalizing from a personal-agent ontology to all religious traditions would exceed the operational scope of the reported tests.
Belief, Credence, and Context-Sensitive Assent
A major conceptual dispute asks whether religious assent should be modeled as the same cognitive attitude ordinarily called factual belief. Van Leeuwen (2014) Van Leeuwen (2014) argued that religious credence has a different psychological profile from mundane factual belief: the two attitudes may differ in their origins, governing roles, and contextual sensitivity. His account invites researchers to examine how religious propositions guide inference and behavior, rather than classify attitudes solely by what participants say they believe.
Levy (2017) Levy (2017) challenged the proposed distinction, arguing that context-dependent deployment can occur with factual beliefs too and that intuitiveness and processing fluency may better explain variation in which representations become active. Van Leeuwen's reply (2017) Van Leeuwen (2017) defended the theoretical contrast and called for further empirical discrimination. This is a substantive scientific disagreement about predictive categories and mechanisms, and it cannot be settled by choosing an appealing definition of faith.
The argument has direct consequences for belief-revision studies. Suppose a respondent ceases participating in worship yet continues to report confidence in the existence of God. Another respondent maintains ritual practice while increasingly treating theological descriptions as symbolic. A third adopts a religious label primarily as a communal identification. The word belief may conceal differences in judgment, participation, motivational commitment, and the inferential use of claims. Laboratory designs can compare how people reason from religious premises across tasks; observational studies can measure which claims shape everyday choices.
A useful research strategy is to record both reported content and practical inferential commitments, with careful attention to competing explanations. A behavioral decision that seems inconsistent with a stated belief may reflect immediate costs, social constraints, conflict between values, or uncertainty about the experimenter's intentions. It is therefore methodologically premature to claim that discrepancy always reveals a distinct hidden attitude. The dispute between Van Leeuwen and Levy shows why empirical discriminating tests are necessary.
Religious Commitment as a Socially Embedded Process
Religious commitment may persist because an individual finds a claim convincing, because the claim is embedded in longstanding patterns of family and collective practice, or because several such influences reinforce one another. Social networks supply opportunities for testimony, interpretation, reassurance, and challenges. Institutions distribute authoritative explanations, regulate entry and departure, and attach public meanings to signs of assent. These processes are psychologically consequential even when the proposition being considered remains conceptually unchanged.
The social context also determines the informational environment. When a claim is affirmed by many trusted people, hearing yet another endorsement may add little independent evidence but reinforce perceptions of consensus. When alternative accounts are rarely encountered or disclosing questions has costs, the available reports may not represent the distribution of private beliefs. Researchers studying acquisition must consequently observe social exposure and source independence as well as total repetition.
The distinction between institutional conformity and personal conviction matters especially in surveys. Attendance may result from household routines, obligations, cultural celebration, family harmony, or individual devotion. Similarly, public disengagement may follow relocation, constraints on time, or dissatisfaction with a congregation without any corresponding doctrinal shift. An accurate psychological model keeps behavioral membership and proposition-specific assent as separate outcomes.
Credibility within a community can also be reconstructed after disappointment. A person may retain religious belief while revising judgments about a leader; another may interpret institutional failure as evidence against previously held doctrines. The same social event can therefore produce diverse epistemic responses. Empirical analyses should elicit attribution and perceived significance, rather than assume that institutional scandal or interpersonal conflict has an identical effect on everyone.
The Structure and Experience of Doubt
Religious doubt encompasses questions of evidence, authority, internal coherence, moral interpretation, religious experience, and relevance to life. Its phenomenology varies. Someone may question whether a particular sacred account is historically accurate, whether an experience should be regarded as divine, or whether an inherited doctrine should continue to command assent. Others may worry that having such questions itself violates a religious norm. A scientific account can distinguish the object of uncertainty from the emotional consequences of that uncertainty.
Doubt can be operationalized as reduced confidence, increased perceived alternatives, inconsistency between propositions, self-reported questioning, or distress related to belief. Each definition generates a different dataset. The doubt domain of Exline and colleagues' (2014) Religious and Spiritual Struggles Scale Exline (2014) measures troubled doubt within a larger set of struggles, including divine, interpersonal, moral, demonic, and ultimate-meaning concerns. A psychological survey that uses that measure should call its outcome distressing religious doubt or struggle, rather than treating it as a universal indicator of curiosity or openness.
The relationship between belief certainty and psychological well-being is therefore an empirical question that depends on both the individual and the surrounding social setting. Doubt experienced as a legitimate inquiry can accompany learning and reinterpretation. Doubt experienced as shameful or socially dangerous can produce concealment, rumination, and anxiety. Both cases involve uncertainty, yet their consequences differ because people attach different meanings and costs to questioning.
This distinction also guards against equating skepticism with mental disorder. Religious questions and changes in religious commitment ordinarily belong to the diversity of human meaning-making. Mental-health evaluation becomes relevant when distress, impairment, compulsive behavior, or other independently assessable clinical concerns are present. A philosophical disagreement about supernatural propositions is not a diagnostic criterion.
Interpersonal Conflict and the Doubting Process
Krause and Ellison (2009) Krause (2009) explicitly examined religious doubt as a longitudinal process involving a precipitating experience, doubt, coping, and health-related outcomes. They reported that negative interactions with fellow congregants predicted more subsequent religious doubt, whereas spiritual support and prayer-group involvement were associated with less doubt. They also distinguished suppression of doubt from efforts to seek spiritual growth, finding less favorable health associations for suppression in their analysis.
This study demonstrates the value of identifying a temporal process rather than making doubt a static personality trait. It also illustrates why the interpersonal climate matters: a relationship that ordinarily certifies religious belonging may become a source of conflicting information, making the credibility of teaching and the reliability of fellow members difficult to separate. In some settings, the believer's question concerns doctrine; in others, the question concerns whether a community embodies its stated commitments.
The longitudinal design helps investigate ordering, but the findings should not be read as proving that particular congregational interactions universally cause religious skepticism or health problems. People who are already questioning may notice negative interactions differently, and unmeasured characteristics may affect both social experiences and later outcomes. Statistical adjustment, multiple time points, and transparent models improve inference while leaving some alternative explanations possible.
The work also points toward a more specific research question: does the availability of safe, respectful discussion moderate the association between belief uncertainty and distress? Participants could be studied across communities with differing opportunities for disclosure. Such research would need measures of perceived openness, actual interpersonal responses, private belief, publicly reported belief, and well-being, with adequate attention to self-selection into communities.
Evidence Evaluation and Motivated Reasoning
Belief revision involves encountering information, judging its relevance, deciding how much weight it deserves, and integrating it with previously held commitments. In religious settings these judgments are situated within established accounts of what counts as authoritative testimony or legitimate evidence. A scriptural argument, a personal experience, historical scholarship, a scientific explanation, or testimony from a religious leader may carry different weights across traditions and persons. Epistemic methods and social commitments are therefore part of the causal background of belief change.
Motivated reasoning is relevant because people can have a personal stake in the outcome of inquiry. The possibility of losing relationships, authority, moral orientation, or a coherent biography can affect which questions are pursued and how conflicting evidence is appraised. These mechanisms should be specified and tested rather than assumed from religious identity alone. Skeptical and nonreligious commitments can also be attached to identity and community, so a balanced research design examines comparable processes across positions.
The integrative review by Zein, Altenmüller, and Gollwitzer (2024) Zein (2024) examines how epistemic beliefs, identity, and contextual experience organize perceived relations between science and religion. The analysis illustrates why disagreements about religious propositions can also involve beliefs about which forms of knowledge deserve authority. Theoretical models must separate directional motivation, which favors a particular conclusion, from accuracy motivation, which favors information judged valid. Both may operate in one decision. A person may resist an unwelcome religious conclusion while devoting substantial effort to verification. Conversely, revision may occur without deliberate evidential comparison, following a change in trusted relationships or the salience of a once-familiar experience.
The most informative research records what counts as disconfirming evidence for the participant, measures confidence before exposure, independently codes the new information, and observes change after a meaningful delay. Such designs can detect whether an apparent shift is temporary compliance, durable reconceptualization, stronger conviction, or selective uncertainty about one subclaim. They also prevent an untested claim that all doubt is generated by purely analytic deliberation.
Analytic Thinking and the Replication Problem
The relationship between analytic thinking and religious belief has been a central controversy in the cognitive science of religion. Gervais and Norenzayan (2012) Gervais (2012) reported studies in which prompts intended to encourage analytic thought were associated with lower reported religious belief. Their findings supported the possibility that increased reflection could causally attenuate particular intuitions supporting belief in God. This was an experimentally framed hypothesis, not a universal demonstration about religious change.
A large preregistered direct replication by Sanchez, Sundermeier, Gray, and Calin-Jageman (2017) Sanchez (2017) failed to reproduce the expected effect of a visual analytic-thinking prime. Across four sites, the authors observed an estimated standardized effect of d = 0.07 in the opposite direction to that predicted, with a 95% confidence interval from -0.12 to 0.25 and an analyzed sample reported as N = 941 in the abstract. This result is specifically evidence against a dependable effect of that manipulation in those settings; it does not disprove every possible influence of reflective thinking on religious belief.
Pennycook, Ross, Koehler, and Fugelsang (2016) Pennycook (2016) reported four studies and a meta-analysis finding an association between cognitive reflection and lower religious belief. Associations remain a distinct evidentiary class. Performance on cognitive reflection tasks can correlate with education, numeracy, cultural attitudes, and other characteristics that complicate attempts to infer the causal direction of belief change.
The scientific response to apparently conflicting findings is to distinguish the questions. One concerns the cross-sectional association between reflective performance and belief. Another concerns whether a brief laboratory prime changes reported belief. A third concerns whether improved reasoning skills predict within-person revision over years. An association can be present even when a particular experimental prime fails. Experimental null findings, likewise, need assessment of manipulation strength, measurement precision, confidence intervals, and the size of effects that the design could detect.
This case provides a methodological lesson with wider importance. An elegant and socially resonant result can be overstated when the original experimental manipulation is not independently reproduced. Mature evidence assessment weighs preregistered replications and heterogeneous samples alongside influential original studies rather than treating historical prestige as evidentiary priority.
Cross-Cultural Variation and Generalizability
Cross-cultural testing is necessary because both belief and cognitive performance develop within institutions, languages, educational systems, and culturally differentiated accounts of religious authority. Gervais and colleagues (2018) Gervais (2018) examined cognitive reflection and belief in God or gods among 3,461 participants drawn from 13 societies. Their analysis found that the negative relation commonly discussed as analytic atheism was modest and varied across countries; clear country-specific evidence was limited to a minority of sampled contexts.
The sample included diverse religious backgrounds, but several national subsamples were convenience samples of students. This limits population-level generalization even as the research improves on claims derived only from a single North American setting. A cross-cultural contrast in average religiosity cannot, without additional study, demonstrate that a culture itself causes differences in cognition. Measurement equivalence, educational experience, age composition, religious history, and sampling processes must also be considered.
Recent research can extend this analysis temporally. A 2026 Polish investigation Łowicki (2026) examined analytic thinking and exposure to credible childhood religious models across younger and older adults. Its reported cross-lagged analyses over six and twelve months did not establish the bidirectional changes expected under a simple reciprocal causal account after stable individual differences were considered. The study is informative about its population, measures, and interval; neither cultural learning nor analytic reflection should be reduced to one deterministic driver.
The key generalization is methodological: a theory of belief acquisition or revision must state the populations and religious constructs for which it is intended. Research in settings with widespread theistic belief should not be silently extended to non-theistic traditions. Likewise, an instrument asking whether God exists may confound the respondent's theology with the researcher's operational definition of religion.
Identity, Trust, and Social Costs of Revision
Belief revision can be an epistemic change with consequences for belonging, loyalty, and self-understanding. A religious proposition sometimes connects a person to family history, moral exemplars, relationships with ancestors, or a community of practice. Questioning that proposition can become intertwined with questions about who the person is, which relationships are dependable, and how prior choices should be interpreted. This intertwining is psychologically consequential because the same evidential information may carry different practical costs across social environments.
The social costs of change can take several forms. Some people anticipate disappointment from parents, the loss of trusted relationships, reduced access to institutional resources, or the need to reconstruct familiar celebrations. Others encounter communities where questioning is expected and theological revision is treated as a legitimate dimension of religious growth. A comparative design must measure anticipated sanctions, actual interpersonal responses, available alternatives, and perceived sources of support separately. A participant's expected loss may be a stronger predictor of public disclosure than the factual strength of conviction itself.
The distinction between private belief and public religious identity is essential. Public assent may continue after confidence diminishes because a person values peace within a family or cannot safely reveal uncertainty. Conversely, public disaffiliation may occur while a person retains substantial supernatural belief. An anonymous survey is therefore preferable to direct group interviews for some sensitive questions, while confidential longitudinal interviews can illuminate how disclosure and personal conviction change on different timelines. Both approaches require credible privacy protections.
Chen and Streib (2026) Chen (2026) examine characteristics, predictors, and outcomes of deconversion with cross-sectional and longitudinal data. Their contribution belongs to the emerging literature that treats a religious exit as a temporally structured transformation rather than a single event. It also highlights the need to distinguish the social consequences of changing one's identity from the epistemic consequences of changing specific metaphysical commitments. Measures of affiliation alone would not establish whether someone changed their beliefs, the reasons for change, or the interpersonal circumstances that made change possible.
Change across Adolescence and Emerging Adulthood
Adolescence and emerging adulthood often involve widening social networks, increased exposure to alternative interpretations, and changing relationships with parents and institutions. Such developmental transitions provide opportunities for religious belief revision, but the direction and extent of change depend on individual histories, cultural expectations, educational environments, and opportunities for open inquiry. Age should therefore be treated as a developmental context, not a predetermined stage of declining or increasing religious belief.
A longitudinal study by Stoppa and Lefkowitz (2010) Stoppa (2010) followed emerging-adult college students and examined trajectories in religious belief and behavior. The study found declines in several indicators of religious participation while some dimensions of religious attitudes were more stable. The methodological lesson is that frequency of attendance, prayer, religious importance, and doctrine-related confidence should be traced separately. A decline in behavior may reflect relocation, new schedules, or institutional distancing without an equivalent change in propositional conviction.
Developmental research also needs to account for transition-specific reporting. Students who are newly away from family may report less religious practice because previous participation depended on household routines, while the meaning assigned to religious claims remains unchanged. Peer interaction can expose inconsistencies or provide new interpretive resources. Educational experiences may strengthen convictions for some students, weaken them for others, or redirect them from literal to symbolic interpretation. A single precollege/postcollege difference cannot isolate which experience produced a change.
Research on religion in young adulthood often draws disproportionately from university populations. Such samples provide an informative institutional setting but are not representative of young adults whose life trajectories include work, caregiving, religious training, migration, or interrupted schooling. To generalize developmental claims, studies need broader sampling frames, repeated observations beyond one academic year, clear descriptions of baseline socialization, and measurement that remains comparable when the participant's religious vocabulary changes.
The acquisition-revision relationship also has a developmental history. Early testimony may give a child the categories through which later doubts become expressible. Subsequent revision may involve reconsidering what previously trusted speakers intended, reevaluating historical claims, or distinguishing inherited community practices from personal assent. Longitudinal models should examine continuity in the structure of questions, not simply transitions between religious-status labels.
Prospective Studies of Conversion and Deconversion
Longitudinal evidence offers a direct opportunity to compare beliefs before and after major religious transitions. In a large Dutch longitudinal panel, Bleidorn, Lenhausen, Schwaba, and Hopwood (2024) Bleidorn (2024) analyzed psychological changes around religious conversion and deconversion. The study identified 181 participants who converted and 450 who deconverted in an eleven-wave panel of more than 20,000 individuals. By distinguishing trajectories in religious practice and belief from other psychosocial variables, the research avoided assuming that an institutional transition automatically produces a uniform internal transformation.
One central finding concerned asymmetry in the dimensions of religion. Religious conversion was associated with an increase in religious practices, but a corresponding sustained increase in belief in God was not clearly established in the same way. Deconversion was associated with decreases in religious practices and belief. Such patterns show that the temporal relation between participation and assent can depend on the type of transition. They also demonstrate why studying conversion solely through retrospective narratives risks missing differences that were present before the named event.
Within-person analysis reduces some sources of bias because the same individual contributes repeated observations, but it does not transform naturally occurring conversion into random assignment. A change in employment, relationship status, geographical environment, religious community, or personal crisis may affect both the probability of conversion and the recorded psychological outcomes. Panel studies must document timing, attrition, missing-data assumptions, sensitivity analyses, and the possibility that participants differ before the observed transition.
Prospective work can be strengthened by recording exact claims and reasons for assent before a transition, rather than only global belief in God. For example, a person may remain theistic but revise beliefs about revelation, divine punishment, salvation, providence, or religious institutions. Another may relinquish affiliation while retaining practices that provide meaning. Repeated assessments of certainty, evidential sources, everyday practice, and the willingness to disclose beliefs could distinguish these trajectories.
The science of belief revision benefits from studying people who never formally convert. Changes within a tradition may be more frequent than major status transitions, while new forms of uncertainty may remain private. A panel design that focuses only on converts and deconverts can miss changes in how a participant understands the authority of scripture, the meaning of prayer, or the credibility of particular sacred histories. Future designs should treat institutional transition as one possible event within a broader series of belief-specific modifications.
Distinguishing Belief Revision from Religious Switching
Religious conversion and deconversion are valuable research settings because they provide observable turning points. Nevertheless, they are composite outcomes that can involve changes of affiliation, social membership, practices, personal identification, and metaphysical judgments. For the purposes of the present dossier, the target variable is the transformation of propositions and the grounds on which they are accepted. An affiliation switch belongs to the explanation insofar as it supplies information about that transformation.
Conceptual differentiation allows researchers to describe several distinct trajectories. A person may convert from one tradition to another while preserving belief in a personal deity. A person may become institutionally unaffiliated but continue to affirm supernatural agents or an afterlife. Someone may maintain an existing tradition while reinterpreting miracles, prophecy, or divine attributes. A person raised without religious affiliation may develop a new metaphysical commitment without joining any organization. These trajectories involve different combinations of beliefs and social changes.
Operational indicators should consequently include proposition-specific endorsement; confidence and uncertainty; interpretive mode, such as literal or symbolic construal; reported grounds for belief; institutional identity; and practice. These variables should be measured at multiple time points. Repeated single-item measures can be useful in broad national panels, but they should be interpreted in relation to their narrow wording and the stability of their meaning across traditions.
A further difficulty is retrospective reconstruction. After conversion, people may describe their earlier views in light of their present identity and current account of what the change meant. Memory for reasons may be selective, and stories of sudden transformation may compress a longer sequence. Prospective records are therefore especially valuable because they preserve the person's earlier wording and the contemporaneous context of questioning. Qualitative interviews remain indispensable for interpreting the significance of change, but retrospective accounts require temporal triangulation.
The resulting scientific distinction is precise: belief revision concerns changes in assent and justification; switching concerns a change in social or institutional classification. Their association is an empirical question. The difference matters not only for classification but also for the interpretation of purported causes. A life event associated with church attendance may have a different relation to theological assent than to public identity.
Emotions, Existential Concerns, and Religious Uncertainty
Doubt is frequently discussed in connection with anxiety, guilt, existential threat, and uncertainty about death. These associations require differentiation between the cognition of uncertainty and its emotional interpretation. Doubt can be experienced as curiosity, liberation, painful disorientation, betrayal, or a manageable component of faith. An individual's response is shaped by prior beliefs, expected consequences, relationships, and the available language for describing uncertainty.
In a longitudinal study of later life, Upenieks (2023) Upenieks (2023) investigated transitions in religious doubt and death anxiety. The research treated doubt as a changeable religious experience and examined its association with existential concerns rather than assuming a fixed relationship across all individuals. Its relevance lies in the capacity of repeated assessments to identify changing states and in the importance of age-specific context. The results should not be extrapolated as a universal claim that religious doubt causes fear of death.
Religious struggles research provides additional distinctions. The measure developed by Exline et al. (2014) Exline (2014) differentiates troubled doubt from other dimensions such as divine, interpersonal, and moral struggles. This architecture clarifies how distress about uncertainty can occur alongside concerns about a perceived relationship with God or conflict with religious people. The dimensions can co-occur, but each requires its own operational description.
A person who questions a belief after bereavement may be responding to grief, altered expectations of divine action, new evidence about an event, or the strain between these factors. A person who doubts under threat of rejection may experience anxiety largely because of anticipated interpersonal loss. Evidence about emotional correlates must therefore include context, preexisting vulnerability, coping opportunities, and time ordering. A survey linking doubt and distress at one point cannot establish which process preceded the other.
The scientific purpose of this distinction is explanatory and ethical. Researchers should avoid pathologizing religious uncertainty, inferring the truth of a proposition from emotional comfort, or treating changes toward belief and disbelief as psychologically equivalent events. They can instead investigate the mechanisms through which uncertainty acquires meaning, becomes privately manageable, or is associated with distress and change.
Measurement Architecture and Construct Validity
Measurement is the decisive link between a theory of religious change and a claim about psychological evidence. Studies sometimes use the words religious belief for a direct response to whether God exists, for agreement with a collection of doctrines, for the importance assigned to religion, or for a scale that also measures ritual practice and organizational participation. The resulting operational definitions are sufficiently different that their findings cannot be combined without careful qualification. A respondent who assigns high personal importance to religious identity may nonetheless have uncertain metaphysical commitments, and a confident believer may have little religious participation.
A proposition-sensitive research design begins by specifying the precise content under study. It then measures strength of endorsement, uncertainty, reasons for endorsement, interpretive framing, source credibility, and practical consequences. These variables allow researchers to distinguish a revision of doctrine from a change in subjective importance. Explicit self-reports remain fundamental because they capture an articulated position. Measures of recall, implicit association, reaction time, or spontaneous inference can provide supplementary evidence about accessibility and habitual processing. Their interpretation requires task-specific validation: an automatic response may signal conceptual familiarity rather than present conviction.
The psychometrics of doubt presents a particularly instructive example. Exline et al. (2014) Exline (2014) produced an instrument to measure religious and spiritual struggles, with a subdomain targeting emotional conflict around doubt. Its construct differs from a neutral confidence rating. Using a distress-oriented item and then concluding that all uncertain believers suffer psychological distress would introduce construct contamination. Similarly, asking respondents whether they are religious may chiefly capture self-identification in one cultural setting and institutional membership in another.
Reliability has several distinct components. Internal consistency addresses relations among items at a given occasion. Test–retest reliability addresses score stability when the target construct is expected to remain stable. Longitudinal measurement invariance asks whether the meaning and measurement properties of indicators are sufficiently similar across time to permit interpretation of change. Cross-cultural invariance and careful translation are necessary when comparing religious traditions, and the heterogeneity of theological vocabulary may demand culturally specific modules rather than a single universal scale.
Social desirability and disclosure risk require additional design attention. Anonymous responses may differ from in-person interviews, but anonymity does not remove all normative expectations. Participants may change what a question means to them after a shift in belief. Longitudinal cognitive interviewing can identify such response shifts. Useful designs pair structured questionnaires with open-ended explanations, collect confidence alongside categorical responses, and allow distinctions such as uncertainty, active rejection, practical indifference, and symbolic interpretation.
Experimental and Longitudinal Research Designs
Experimental designs can isolate the effects of specific information or social cues on belief-related judgments. A study might vary the apparent reliability of a narrator, the consistency of an adult's actions with stated belief, or the presentation of conflicting interpretations. Such manipulations should be accompanied by preregistered primary outcomes, attention to manipulation checks, sufficient statistical power, careful debriefing, and tests of persistence. A transient response to a laboratory stimulus is a different outcome from durable religious belief revision.
The failure of a prominent analytic-thinking prime to replicate in Sanchez et al. (2017) Sanchez (2017) illustrates why a manipulation should be evaluated separately from the broader construct it seeks to influence. The original result can motivate further tests, but the replication narrows confidence in a particular causal claim. A rigorous program compares multiple manipulations with independent validity checks, specifies outcomes and exclusions in advance, and reports effect sizes and confidence intervals rather than relying on isolated significance tests.
Longitudinal observational designs capture changes that unfold over years, including shifts that are subtle or privately concealed. The study by Bleidorn et al. (2024) Bleidorn (2024) provides a concrete model of repeated assessment around religious transition. By measuring belief and practice separately, it revealed differences that a single conversion classification would have hidden. Related research on emerging adulthood by Stoppa and Lefkowitz (2010) Stoppa (2010) demonstrates the importance of recognizing within-person heterogeneity even when an aggregate trend is visible.
A useful longitudinal protocol collects baseline beliefs, source networks, experienced conflict, social supports, reasons for change, and disclosure opportunities before an identified turning point. Event-centered designs can relate assessment waves to conversion, migration, institutional conflict, marriage, bereavement, or new religious education. The researcher must also inspect whether participants were already changing before the event. Without pre-event observations, a difference observed afterward may be misattributed to the most visible transition.
Causal inference remains challenging. People select religious communities, social relationships, and sources of evidence in ways partly shaped by prior belief. Those beliefs also change their future social exposure. This produces time-varying reciprocal influence. Models should consider baseline differences, attrition, confounding, measurement error, and the plausibility of comparison groups. Statistical sophistication is valuable only when the design supplies an interpretable causal question.
Qualitative and mixed methods contribute a different kind of precision. A repeated interview series can reveal exactly what a participant meant by certainty, how a religious narrative was reinterpreted, and which interpersonal events were experienced as credible challenges. Structured coding increases transparency, while preserving contradictory cases prevents premature generalization. Jointly analyzing changes in questionnaire responses and narrative accounts can reveal whether an apparent reversal reflects actual revision, changed language, or public disclosure of an earlier private position.
Evidence Appraisal and Alternative Explanations
The strongest conclusion of the present literature is that religious belief change is multidimensional and context-dependent. This claim rests on several independent strands of evidence. Developmental experiments associate religious exposure with children's interpretations of extraordinary narratives; cognitive experiments demonstrate discrepancies between explicit theological accounts and representations deployed in comprehension; cross-cultural studies show variation in the association between cognitive reflection and belief; and repeated observations document divergent trajectories of religious practice, assent, and affiliation.
The causal strength of these strands is uneven. Corriveau et al. (2015) Corriveau (2015) provide experimental variation in story type but observational differences in prior religious exposure. Barrett and Keil (1996) Barrett (1996) identify task-sensitive conceptual representation; their results do not show that anthropomorphic processing causes religious acquisition. Gervais and colleagues (2018) Gervais (2018) report a cross-cultural correlational pattern of variable strength. Sanchez et al. (2017) Sanchez (2017) directly test a specific priming claim with preregistered replication and find little support. These are different questions and should remain distinguishable in any evidence synthesis.
Several alternative explanations recur. Family and community exposure may operate through repetition, testimony, perceived authority, emotional security, shared activity, or the selection of already committed families into religious institutions. Cognitive reflection may correlate with educational opportunity, sociocultural identity, numeracy, or interest in philosophical inquiry. Reported doubt may reflect changed convictions, changed willingness to disclose those convictions, or changes in perceived acceptability of uncertainty. Institutional conversion may accompany other life transitions that are responsible for some observed outcomes. Each explanation calls for a different design or measurement strategy.
An important unresolved issue concerns the temporal relationship between private belief and external practice. An individual may first change assent and later change affiliation, or begin attending a community before affirming its doctrines. The Dutch longitudinal panel analyzed by Bleidorn et al. (2024) Bleidorn (2024) supplies evidence of these dimensions moving differently around transitions, but its setting—conversion to or deconversion from Christianity in the Netherlands—places limits on generalization to other religious systems or political environments.
The field also needs a more precise account of epistemic trust. A source can lose moral legitimacy while retaining informational credibility, or remain emotionally important while losing authority over theological claims. Empirical measures that merge trust, affection, and deference may overlook these distinctions. Research capable of separating them would clarify whether change arises mainly from new propositions, new evaluations of the source, or altered incentives to disclose disagreement.
Finally, the philosophical question of whether religious propositions are true is outside the evidentiary reach of most psychological experiments. A task that measures how people acquire or revise a claim can describe its cognitive mechanisms without resolving its metaphysical validity. The scientific subject here is the formation and modification of human attitudes toward religious propositions, assessed through observable methods and explicit operational definitions.
Psychological Change in the Artificial Era
The Artificial Era introduces additional channels through which religious knowledge and uncertainty are encountered. Search engines, recommendation systems, digital archives, social platforms, and conversational artificial systems make theological interpretations, historical arguments, and testimony available in new configurations. They can expose people to multiple traditions and conflicting sources while simultaneously mediating what information appears salient and credible. This creates a scientifically relevant extension of the psychology of epistemic trust.
A conversational system may present a religious argument with fluent language, apparent neutrality, or interpersonal responsiveness. Users can attribute competence and authority to such interfaces, and those attributions may affect exposure, confidence, and questioning. These possibilities should be studied as testable hypotheses involving information presentation and user behavior. Current research on religious belief acquisition does not establish that artificial systems possess religious convictions, subjective faith, or an intrinsic capacity for spiritual experience. Claims about the system's internal status require a separately defined research program.
The relevant questions concern human belief processes in mediated environments. Does conversational presentation increase perceived credibility relative to an identical written argument? Do source disclosures alter the effect? Does repeated interaction encourage genuine belief revision, temporary verbal agreement, or selective information seeking? Do people use artificial interfaces to rehearse difficult questions they hesitate to raise in their communities? Answering these questions requires experiments with transparent source attribution, privacy protections, longitudinal follow-up, and independent assessment of the quality of information provided.
Artificially mediated inquiry can also complicate measurement. A participant may reproduce a compelling answer encountered in a conversation without adopting the underlying proposition. Exposure to an algorithmic explanation may alter vocabulary or perceived consensus while leaving private certainty intact. A study should therefore capture initial confidence, the interaction record with informed consent, delayed reassessment, and the distinction between informational persuasion and relational trust.
Within the broader research horizon From Homo to Artificial, the psychology of religious belief provides a case for examining how nonbiological systems become participants in human epistemic environments. The presently established psychological evidence concerns people evaluating, trusting, resisting, and revising claims. The proposal to investigate Artificial as an additional category of historical agency belongs to a wider interdisciplinary program. In this dossier it serves as a source of specific future questions, with empirical warrant assessed independently.
A Research Program for Belief Revision
The preceding evidence supports a research program organized around temporal resolution, proposition-level specificity, contextual validity, and causal identification. A first priority is to follow the same individual through acquisition, doubt, and revision using repeated measures of particular claims rather than broad classifications. Researchers can record belief content, certainty, confidence in the source, interpretive style, and practical consequences. They can then test whether these dimensions change together or in different sequences.
A second priority is to examine credible testimony through controlled comparisons that vary the reliability and social conduct of models. Such studies should distinguish the strength of a proposition from the perceived credibility of its advocate. Where behavior is observed rather than experimentally manipulated, researchers can use prospective designs and record plausible alternative explanations including warmth, familiarity, group membership, and institutional incentives.
A third priority is to study doubt in multiple forms. A useful battery would distinguish neutral uncertainty, active evidential inquiry, moral disagreement, conflict over authority, distress about doubt, and fear of public disclosure. This would prevent the mistaken inference that questioning is necessarily a form of spiritual struggle. It could also establish which forms predict durable doctrinal revision, renewed commitment, or changes limited to public identity.
A fourth priority is cross-cultural expansion. The field needs instruments sensitive to non-theistic traditions, ritual-centered religious practice, plural affiliations, and culturally variable meanings of certainty. Strong comparisons depend on measurement equivalence as well as diverse samples. The 2026 Polish study by Łowicki and colleagues Łowicki (2026), which considers analytical reflection and credible religious modeling across generations and time, illustrates the value of jointly analyzing cognitive and cultural accounts while attending to the limits of causal inference from observational designs.
A fifth priority is explicit replication. High-profile experimental claims should be tested in preregistered, adequately powered, multi-site projects using transparently validated manipulations. Null outcomes and heterogeneity should be reported with the same precision as statistically significant effects. Replication must address the causal claim made by the original experiment, while theoretical synthesis should also consider the wider evidence on correlations, naturally occurring changes, and development.
A sixth priority is the study of digital information environments. Researchers should compare human, institutional, and artificial sources, measure actual rather than assumed exposure, and establish how interface features affect perceived authority. Ethical design should make deception proportionate and necessary, preserve participants' autonomy, and avoid creating uncontrolled social harm through sensitive identity disclosures.
These priorities share a common requirement: the scientific explanation of religious belief change must be anchored in precise observations of what a person believes, how firmly, why, under what social circumstances, and at what point in a longitudinal sequence. Progress will depend on refining those distinctions rather than compressing them into one global measure of religiosity.
Conclusion
Religious belief acquisition, doubt, and revision form a continuous but heterogeneous domain of psychological change. Acquisition depends on learning from testimony, cultural models, shared narratives, and the credibility assigned to sources. Doubt arises when a claim, its justification, or the authority of its interpreters becomes uncertain. Revision alters the content, confidence, organization, or practical application of a belief. These processes can occur within a religious tradition, across institutional boundaries, or without any public change in affiliation.
The evidence shows that childhood exposure is associated with how children classify religious narratives, and that theological concepts may be processed differently across explicit and immediate cognitive tasks. It also shows that reported belief and ritual participation can follow distinct trajectories. Experiments on analytic-thinking primes require particular caution: the influential initial causal claim met an important failure of preregistered replication, while correlational and cross-cultural findings support a more limited, heterogeneous relationship between reflection and religiosity.
Scientific understanding advances when the dimensions of belief are kept analytically separate. A change in church attendance cannot substitute for a measure of conviction; troubled doubt cannot represent all uncertainty; an affiliation change cannot establish a uniform theological transformation; and a participant's automatic response to a familiar concept cannot be equated with private assent. These distinctions improve historical interpretation, developmental research, experimental inference, and the ethics of studying religious transitions.
The field's central unresolved questions concern the precise causal contribution of credibility, the temporal organization of doubt and revision, culturally valid measurement, the relationship between disclosure and private conviction, and the effects of digital and artificial sources on epistemic trust. A rigorous answer requires integrated developmental, experimental, cross-cultural, longitudinal, and qualitative evidence. This monograph establishes the research object and critical evidentiary architecture for that work.
References
- Barrett, J. L., & Keil, F. C. (1996). Conceptualizing a nonnatural entity: Anthropomorphism in God concepts. Cognitive Psychology, 31(3), 219–247. https://doi.org/10.1006/cogp.1996.0017
- Bleidorn, W., Lenhausen, M. R., Schwaba, T., & Hopwood, C. J. (2024). Psychological change before and after religious conversion and deconversion. Journal of Personality, 92(4), 1193–1210. https://doi.org/10.1111/jopy.12881
- Chen, Z. J., & Streib, H. (2026). Characteristics, predictors and outcomes of religious deconversion: A cross-sectional and longitudinal analysis. The International Journal for the Psychology of Religion, 36(1), 97–114. https://doi.org/10.1080/10508619.2025.2520104
- Corriveau, K. H., Chen, E. E., & Harris, P. L. (2015). Judgments about fact and fiction by children from religious and nonreligious backgrounds. Cognitive Science, 39(2), 353–382. https://doi.org/10.1111/cogs.12138
- Exline, J. J., Pargament, K. I., Grubbs, J. B., & Yali, A. M. (2014). The Religious and Spiritual Struggles Scale: Development and initial validation. Psychology of Religion and Spirituality, 6(3), 208–222. https://doi.org/10.1037/a0036465
- Gervais, W. M., & Norenzayan, A. (2012). Analytic thinking promotes religious disbelief. Science, 336(6080), 493–496. https://doi.org/10.1126/science.1215647
- Gervais, W. M., van Elk, M., Xygalatas, D., McKay, R. T., Aveyard, M., Buchtel, E. E., Dar-Nimrod, I., Klocová, E. K., Ramsay, J. E., Riekki, T., Svedholm-Häkkinen, A. M., & Bulbulia, J. (2018). Analytic atheism: A cross-culturally weak and fickle phenomenon? Judgment and Decision Making, 13(3), 268–274. https://doi.org/10.1017/S1930297500007701
- Harris, P. L., & Koenig, M. A. (2006). Trust in testimony: How children learn about science and religion. Child Development, 77(3), 505–524. https://doi.org/10.1111/j.1467-8624.2006.00886.x
- Henrich, J. (2009). The evolution of costly displays, cooperation and religion: Credibility enhancing displays and their implications for cultural evolution. Evolution and Human Behavior, 30(4), 244–260. https://doi.org/10.1016/j.evolhumbehav.2009.03.005
- Krause, N., & Ellison, C. G. (2009). The doubting process: A longitudinal study of the precipitants and consequences of religious doubt in older adults. Journal for the Scientific Study of Religion, 48(2), 293–312. https://doi.org/10.1111/j.1468-5906.2009.01448.x
- Lanman, J. A. (2012). The importance of religious displays for belief acquisition and secularization. Journal of Contemporary Religion, 27(1), 49–65. https://doi.org/10.1080/13537903.2012.642726
- Levy, N. (2017). Religious beliefs are factual beliefs: Content does not correlate with context sensitivity. Cognition, 161, 109–116. https://doi.org/10.1016/j.cognition.2017.01.012
- Łowicki, P., Tokarz, J., Laskowski, P., Kiszkiel, Ł., & DeLaRosa, J. M. (2026). (Dis)belief in God among younger and older Poles: Analytic thinking and cultural learning between generations and over time. Journal for the Scientific Study of Religion. https://doi.org/10.1111/jssr.70072
- Norenzayan, A., Gervais, W. M., & Trzesniewski, K. H. (2012). Mentalizing deficits constrain belief in a personal God. PLOS ONE, 7(5), e36880. https://doi.org/10.1371/journal.pone.0036880
- Pennycook, G., Ross, R. M., Koehler, D. J., & Fugelsang, J. A. (2016). Atheists and agnostics are more reflective than religious believers: Four empirical studies and a meta-analysis. PLOS ONE, 11(4), e0153039. https://doi.org/10.1371/journal.pone.0153039
- Sanchez, C., Sundermeier, B., Gray, K., & Calin-Jageman, R. J. (2017). Direct replication of Gervais & Norenzayan (2012): No evidence that analytic thinking decreases religious belief. PLOS ONE, 12(2), e0172636. https://doi.org/10.1371/journal.pone.0172636
- Stoppa, T. M., & Lefkowitz, E. S. (2010). Longitudinal changes in religiosity among emerging adult college students. Journal of Research on Adolescence, 20(1), 23–38. https://doi.org/10.1111/j.1532-7795.2009.00630.x
- Upenieks, L. (2023). Uncertainty in faith, fear of death? Transitions in religious doubt and death anxiety in later life. Omega: Journal of Death and Dying, 87(3), 814–837. https://doi.org/10.1177/00302228211029475
- Van Leeuwen, N. (2014). Religious credence is not factual belief. Cognition, 133(3), 698–715. https://doi.org/10.1016/j.cognition.2014.08.015
- Van Leeuwen, N. (2017). Two paradigms for religious representation: The physicist and the playground (a reply to Levy). Cognition, 164, 206–211. https://doi.org/10.1016/j.cognition.2017.03.021
- Zein, R. A., Altenmüller, M. S., & Gollwitzer, M. (2024). Longtime nemeses or cordial allies? How individuals mentally relate science and religion. Psychological Review, 131(6), 1459–1481. https://doi.org/10.1037/rev0000492