Implicit pattern learning predicts individual differences in belief in God in the United States and Afghanistan

The study investigates whether individual differences in implicit pattern learning (IL-pat)—a bottom-up, largely unconscious ability to learn order in environmental stimuli—predict individual differences in religious belief, specifically belief in an intervening/ordering God, and changes in belief strength from childhood to adulthood. Grounded in dual-process frameworks and anthropological/psychological theories positing that intuitions arising from implicit learning shape explicit beliefs, the authors hypothesize a bottom-up pathway: stronger IL-pat fosters intuitions of universal order, which in turn bias individuals toward stronger interventionist religious beliefs. The work aims to test this across culturally disparate contexts (United States and Afghanistan), dissociate IL-pat from explicit pattern awareness, analytic thinking, schizotypal ideation, and parental belief, and examine mediation by intuitions of universal order.

The paper reviews theories that religious beliefs emerge partly from evolved neurocognitive systems supporting predictive processing and intuition. It discusses dual-process accounts where automatic, bottom-up processing generates intuitions that often guide explicit beliefs. Prior work links evolved systems (morality/cooperation, pathogen avoidance, anthropomorphism, agency detection) to religious intuitions, though evidence for hypersensitive agency detection is mixed. Research suggests analytic thinking correlates with religious disbelief, but priming analytic thought effects have not consistently replicated. Believers tend to perceive purpose/design in events, and belief in divine intervention is central across major religions and associated with teleological explanations. IL-pat has been identified as a basis of intuitions of order and is associated with self-reported intuitive styles; it is widespread, develops early, shows stability, and may be partly genetic. These literatures motivate testing whether IL-pat predicts interventionist belief and belief change, and whether intuitions of universal order mediate this relationship across cultures.

Design and samples: Analyses were conducted separately by sample. U.S. sample: N=199 (M_age=19.83±2.72; 65.83% female), recruited from Georgetown University and community; religious affiliations included 52.26% Christian, 25.13% unaffiliated. Afghan sample: N=148 (M_age=26.99±4.57; 41.22% female), recruited from Hazara neighborhoods in Kabul; religious affiliation not queried due to risk. A predominantly European online replication sample (N=96; M_age=28.21±9.31; 83.33% European) completed measures of interventionist belief, universal order intuitions, analytic problem-solving, and belief in science. A U.S. re-contact subsample (N=65) completed additional CRT and belief-in-science measures. Implicit pattern learning (IL-pat): Measured with a modified Serial Reaction Time Task (SRTT). Participants responded to targets appearing in one of four horizontal positions via corresponding keys (z, x, c, v). Six blocks: three pattern blocks (each a distinct 10-target repeating sequence, repeated 5x; included first-order and second-order structures) and three random blocks (50 non-repeating targets). No response-stimulus interval (no-RSI) was used to minimize explicit rehearsal. For each block, the slope (correlation between target number 1–50 and response time) was computed; IL-pat was operationalized as the difference in slope between random vs. pattern blocks to capture learning-specific speeding, beyond general factors (e.g., motivation). Explicit awareness: After each block, participants rated whether it was a pattern or random (1=Definitely Not a Pattern to 4=Definitely a Pattern). Computed explicit accuracy, illusory detection on random blocks, and overall bias toward reporting patterns. Belief outcomes: Three measures formed the Interventionist Belief (IB) composite via PCA: Belief in Divine Intervention Scale (BDIS; 6-point agreement), and two overlapping circles tasks indexing perceived God influence on self (self-overlap) and on the world (world-overlap). PCA supported a single component across samples; IB scores were standardized (mean=0, SD=1). Belief change was assessed retrospectively on 9-point scales for belief in God at ages 6 and every 3 years to current age; the primary outcome was current minus age-6 belief. In the U.S. only, a single-item Existence Belief (EB) measured overall belief in God’s existence (1–9). Mediator: Intuitions of Universal Order (UO) measured via two secular items (Everything happens for a reason; There is order to the universe) rated 1–9; summed to a UO score. UO was not administered in Afghanistan due to translation/cultural interpretability issues; the European sample used the same two items. Covariates and additional measures: Schizotypal Questionnaire (unusual perceptual experiences subscale) to account for tendencies to perceive order/supernatural agency; parental belief strength (9-point) during participant’s childhood. Cognitive Reflection Test (CRT; 3 items) administered to U.S. re-contact and European samples to index analytic vs intuitive responding. Belief in Science Scale (10 items) completed by U.S. re-contact and European samples. Statistical analysis: Zero-order correlations assessed bivariate relationships. Linear regression models tested whether IL-pat predicted IB and belief change, controlling for schizotypal ideation and parental belief; additional models also controlled for explicit awareness metrics and, where available, CRT performance. Mediation analysis (bootstrapped bias-corrected 95% CIs) in the U.S. sample tested whether UO mediated IL-pat effects on IB and belief change. An exploratory interaction tested whether the IL-pat–EB association varied with EB–IB similarity.

Implicit learning occurred in both samples: Participants showed faster RT change on pattern vs random blocks (U.S.: t(198)=-10.95, P<0.0001; Afghanistan: t(147)=-7.61, P<0.0001). IL-pat was uncorrelated with explicit pattern accuracy (U.S.: r=-0.12, P=0.10; Afghanistan: r=0.08, P=0.40) and with illusory pattern reports on random blocks (U.S.: r=0.03, P=0.67; Afghanistan: r=0.01, P=0.93). IL-pat predicted interventionist belief (IB) controlling for schizotypal ideation and parental belief: U.S.: b=1.24, β=0.17, SE=0.47, P=0.009; Afghanistan: b=1.60, β=0.26, SE=0.48, P=0.001. Zero-order associations between IL-pat and individual belief measures were consistent across samples: BDIS (U.S.: r=0.13, P=0.06; Afghanistan: r=0.19, P=0.02), self-overlap (U.S.: r=0.19, P=0.007; Afghanistan: r=0.19, P=0.02), world-overlap (U.S.: r=0.16, P=0.03; Afghanistan: r=0.26, P=0.001). IL-pat predicted belief change (current minus age-6 belief), controlling for covariates: U.S.: b=3.53, β=0.15, SE=1.67, P=0.036; Afghanistan: b=3.53, β=0.17, SE=1.67, P=0.036. Explicit SRTT awareness measures were not related to IB or belief change (all P>0.09). Including explicit accuracy, illusory detection, and overall pattern-report bias as covariates left IL-pat effects on IB (all P≤0.009) and belief change (all P≤0.05) intact. Existence Belief (EB; U.S. only) correlated strongly with IB (r=0.66, P<0.0001), but not with IL-pat overall (r=0.05, P=0.50). An interaction showed IL-pat was more predictive of EB when EB closely matched IB (belief similarity × IL-pat: β=-0.20, P=0.002). Intuitions of universal order (UO) correlated with IL-pat (r=0.20, P=0.005) and remained significant controlling for schizotypal ideation and parental belief (b=5.12, β=0.18, SE=1.90, P=0.008). UO correlated with IB (U.S.: r=0.48, P<0.001; European: r=0.56, P<0.001) and belief change (U.S.: r=0.36, P<0.001). Mediation (U.S.) showed significant indirect effects of IL-pat on IB (P=0.005) and on belief change (P=0.006) via UO. CRT (U.S. re-contact): IL-pat was unrelated to CRT correct (analytic) (r=-0.07, P=0.56) or intuitive (incorrect) responding (r=-0.03, P=0.84). Controlling for CRT, IL-pat remained predictive of IB (b=3.22, β≈0.43, SE=0.81, P<0.001), belief change (b=9.88, β=0.40, SE=2.96, P=0.001), and UO (b=11.64, β=0.38, SE=3.61, P=0.002). In the European sample, higher CRT analytic scores related to lower IB (r=-0.22, P=0.03), and CRT intuitive scores related to higher IB (r=0.23, P=0.02). Belief in science (U.S. re-contact, European): Inversely associated with IB (U.S. re-contact: r=-0.22, P=0.08; European: r=-0.36, P=0.0003) and with belief change (U.S. re-contact: r=-0.28, P=0.02). Belief in science was unrelated to UO (U.S. re-contact: r=-0.19, P=0.13; European: r=-0.11, P=0.27) and negatively correlated with IL-pat in U.S. re-contact (r=-0.28, P=0.02).

Findings support a bottom-up pathway wherein stronger implicit learning of environmental order (IL-pat) aligns individuals toward stronger belief in an intervening/ordering God and greater increase in belief from childhood to adulthood. Relationships replicated across culturally disparate U.S. and Afghan samples and were distinct from explicit pattern awareness, schizotypal ideation, parental belief, and CRT-indexed analytic problem-solving. Mediation by intuitions of universal order (UO) in the U.S. sample suggests that IL-pat may engender generalized intuitions that the universe is orderly, which are then reflected or rationalized in explicit interventionist religious beliefs. The no-RSI SRTT design and lack of association with explicit pattern detection argue against top-down searching or effortful analytic cognition as drivers of IL-pat in this context. While causality cannot be established, the cross-cultural replication and evidence of belief change associations indicate that bottom-up perceptual learning mechanisms may contribute to both the prevalence and variability of religious belief, complementing existing accounts that integrate intuitive and reflective processes.

The study demonstrates that individual differences in implicit pattern learning predict stronger belief in an intervening/ordering God and greater increases in belief from childhood to adulthood, with evidence that intuitions of universal order mediate these effects. These relationships generalize across U.S. and Afghan samples and are independent of explicit pattern awareness and analytic problem-solving. The results suggest a core, bottom-up perceptual component underlying religious belief variability, providing a bridge between implicit learning, intuition, and explicit belief. Future research should employ longitudinal and intervention designs to test causal pathways, broaden cross-cultural sampling (including additional non-Western contexts), refine measures of order-related intuitions, and examine neural mechanisms linking IL-pat to belief formation.

- Causality not established: All mediation and regression analyses are correlational; longitudinal and experimental manipulations are needed. - Retrospective belief change measure may be subject to recall bias. - Intuitions of universal order (UO) were not administered in the Afghan sample due to translation/cultural interpretability constraints, limiting cross-cultural mediation tests. - Existence Belief (EB) was a single-item measure and only collected in the U.S. - Samples include university-affiliated participants and, in Afghanistan, primarily Hazara neighborhoods; generalizability to other populations may be limited. - Potential unmeasured confounders (e.g., broader cultural, educational, or cognitive variables) may influence both IL-pat and belief. - While explicit SRTT awareness was assessed, other forms of top-down influence on implicit learning cannot be entirely ruled out.