Religious Belief Systems and Mental Health Outcomes: A Structural Taxonomy Distinguishing Architecture from Content

Authors: Patrick Grünig · Claude Fable (Anthropic)

Status: Paper 1.4 of ten — the series’ clinical-outcomes paper; reads independently of the series’ framework

Note on authorship. This paper is a human–AI collaboration, and the byline says so plainly. The division of labor: the thesis, the source corpus, the conceptual arc, and every decision of substance are Patrick Grünig’s; drafting, reference verification, and adversarial revision are Claude’s — one collaborator instantiated across model generations (first draft: Claude Opus 4.6, February 2026; charter-governed revision: Claude Fable 5, with Claude Opus 5 verification agents, September 2026). Accountability for the work and custody of it are human, and Grünig’s. Where a venue’s policy does not admit machine co-authorship, this byline converts to an acknowledgment without loss: the note records facts, not a claim to legal personhood. One standing rule keeps the collaboration honest, stated in full in Paper 5.1 (§9.1): the AI co-author’s fluent agreement with the thesis is never evidence for it.

Note on epistemic status. This paper proposes a framework and reviews evidence, and it holds its verbs to that division. What the reviewed studies show is reported in their own terms, with samples and designs stated inline; what the taxonomy proposes — including every mechanism and every dimension — is marked as proposal; what would decide it is stated as a prediction. The words established, demonstrated, validated, and proven do not appear below as this paper’s own verdicts on its own claims. Every external reference carries a verification record in the project archive; the References section states the standard.

Note on this revision (removable at publication). This is the reworked successor of the February 2026 draft, produced under the project’s method charter after a three-phase review. Relative to that draft: a prevalence figure that exists nowhere in the literature is replaced by the published range and its published explanation; a cross-tradition invariance claim that no study supports — and that the nearest real evidence contradicts — is removed from the review and restated as the prediction it always was; two citations that pointed at prejudice research for cognitive-flexibility findings are retired; a religious-trauma prevalence estimate is re-attributed to its actual source, with that source’s methods stated; the claim that the three dimensions are orthogonal is demoted to what the psychometric record supports; a wholly uncited list of protective factors is replaced by factor-by-factor verdicts under honest labels; the benefits and meditation literatures acquire the calibrations their own authors published; a cross-cultural record that constrains the taxonomy is added to the limitations, with its numbers; and the reference list grows from thirteen entries to the evidence base the claims require, every field verified against the archive. The body makes no reference to prior drafts; the change record lives in the project archive.

Abstract

The relationship between religious belief and mental health has produced a two-sided literature: meta-analytic mortality benefits and longitudinal flourishing associations on one side, and a clinical literature of scrupulosity, religious trauma, and post-exit distress on the other. We propose that this tension eases when the unit of analysis shifts from religious content (what is believed) to belief-system architecture (how the system is structured with respect to the believer’s cognitive autonomy). We introduce a structural taxonomy of three architectural dimensions — authority locus (external-hierarchical vs. internal-experiential), stress architecture (guilt-installing vs. equanimity-promoting), and dependency model (loop-creating vs. autonomy-supporting) — treated as conceptually distinct rather than statistically independent, and none of which yet has a validated instrument; constructing those instruments is the program this paper proposes. We find real support at specific joints: scrupulosity prevalence varies from 0% to 93% across OCD samples, a spread the field’s own review attributes to the religious intensity of the community sampled; hell belief predicts lower well-being at national and individual levels and is the only component with a causal effect in the one experimental test, where heaven does not differ from control; negative religious coping predicts more distress without predicting fewer positive outcomes; exit penalties concentrate in high-cost groups in population data, while the deficit of ordinary disaffiliation is completely mediated by lost participation; and religious psychology measurably persists after deidentification. We also report the constraints: the benefits literature’s strongest channel is communal participation rather than the interior pole our taxonomy favors, and the structure–outcome relation reverses sign in several non-Western samples. The taxonomy’s central claim — that architecture predicts outcomes beyond denomination, religiosity, and content — is stated as a prediction, with eight testable forms and a matched clinical program.

Keywords: religion and mental health, structural taxonomy, guilt, cognitive autonomy, dependency, religious trauma, scrupulosity, deconversion, religious residue, decentering

1. Introduction: The Religion–Mental-Health Paradox

1.1 The Two-Sided Evidence

Decades of research on religion and mental health have produced a literature whose two halves are hard to hold in one hand. On the benefits side, the associational record is broad (Koenig, King, & Carson, 2012; VanderWeele, 2017), and it includes strong designs. A meta-analysis of prospective cohort studies found religiosity/spirituality associated with reduced mortality in initially healthy populations (combined hazard ratio 0.82, 95% CI 0.76–0.87), independent of behavioral factors, negative affect, and social support — though not in diseased populations (HR 0.98, p = .19), with the protective association concentrated in organizational activity such as service attendance, and with publication bias acknowledged by the authors (Chida, Steptoe, & Powell, 2009). Longitudinal work with extensive confounding control finds religious service attendance associated with happiness, mental and physical health, meaning, character, and social connection — and, notably, finds these associations “much stronger for communal religious participation than for spiritual-religious identity or for private practices” (VanderWeele, 2017). We return to that sentence, which constrains our own proposal, in §5.3.

This literature also carries correctives published from within it, and we print them rather than paraphrase around them. A methodological audit found that of 266 articles returned by a Medline search on religion for the year 2000, only 17% were relevant to claims of health benefit; that about half the articles cited by comprehensive reviews were irrelevant to such claims; and that of the relevant remainder many were flawed or misrepresented — concluding there was “little empirical basis” for assertions of health benefit (Sloan & Bagiella, 2002). And the direction of cause is genuinely open: the same research program that documents the attendance–depression association has shown, with marginal structural models, that depression also lowers subsequent attendance (VanderWeele, Jackson, & Li, 2016).

On the harms side, a distinct literature documents severe distress organized around religious involvement. Religious Trauma Syndrome has been proposed as a clinical construct describing complex-PTSD-like presentations arising from authoritarian religious environments (Winell, 1993, 2011). One survey estimated that 27–33% of U.S. adults have experienced religious trauma at some point, with roughly 10–20% currently suffering — figures from a self-selected online panel (n = 1,581) fielded and funded by the Global Center for Religious Research under its own in-house institutional review board, cited here with that provenance stated (Slade, Smell, Wilson, & Drumsta, 2023). A large qualitative survey of self-identified sufferers (N = 333) maps the phenomenology across nine themes, from emotional distress and destabilized identity to — in the same dataset — personal growth and resilience (Hollier, Clifton, & Bobo, 2026). And the clinical professions have begun to treat the presentation as real: the counseling-supervision literature now instructs supervisors on trauma-informed handling of religious-abuse cases (Cashwell & Swindle, 2018) — a recognition claim, distinct from prevalence evidence, and we cite it as such.

How can religious involvement be associated with flourishing and with trauma at once?

1.2 The Limits of Content-Based Analysis

The standard approach — comparing outcomes by denomination, tradition, or belief content — has limited resolving power. Studies comparing “religious” and “non-religious” populations collapse enormous structural diversity into a single variable. A fundamentalist Baptist congregation and a progressive Baptist congregation share a denomination and differ profoundly in their relationship to members’ cognitive autonomy.

The field’s own results keep pointing the same way. A meta-analytic review of religion and personality concluded that religiosity is not one variable: openness to experience relates negatively to religious fundamentalism and positively to measures of open, mature religiosity and spirituality (Saroglou, 2002). A review of the religion–prejudice literature closed on the sentence that “it is not religion per se, but rather the ways in which individuals hold their religious beliefs” that carries the association (Hunsberger, 1995). And in the religious-struggles literature, a four-sample study (total N = 4,675) found that cumulative stressful life events predict struggles uniformly while general religiousness provides no buffering — with the authors recommending that future tests target “aspects of religious orienting systems,” i.e., structure rather than quantity (Stauner, Exline, Pargament, Wilt, & Grubbs, 2019).

We take that recommendation literally. The variable that matters, we propose, is not what people believe or how much, but how the belief system is architecturally structured — how it relates to the believer’s cognitive autonomy, emotional regulation, and capacity for independent meaning-making.

1.3 The Structural Approach

This paper proposes a structural taxonomy of belief-system architectures, defined not by content (theology, cosmology, moral positions) but by the formal relationship between system and psyche. We identify three architectural dimensions — authority locus, stress architecture, and dependency model — and review evidence that features of each track mental-health outcomes.

We treat the dimensions as conceptually distinct rather than statistically independent, and we say now what the measured record shows. Among the constructs nearest to our dimensions, one pairing approaches independence and others are substantially correlated: in the Religious Schema Scale’s validation samples, the textual-authority and interreligious-openness schemas correlate −.44 (Germany) and −.22 (United States), while the fairness/rationality and openness schemas correlate .44 and .39; only the textual-authority × fairness pairing approaches zero (Streib, Hood, & Klein, 2010). The one construct pair in this territory that behaves orthogonally across many samples is positive versus negative religious coping (Pargament, Feuille, & Burdzy, 2011). We therefore expect our poles to co-occur in the wild, we treat the mixed cells of the taxonomy (§2.4) as theoretically possible rather than empirically populated, and we regard the factorial separability of the three dimensions as an open question this framework exists to test (§7, §8.5).

This is the clinical-outcomes paper of a ten-part series; a collective-level companion paper supplies the formal model on which one dimension draws, and an individual-level companion supplies the neural account of belief entrenchment (Papers 1.2 and 1.3). The paper reads independently of the series’ framework.

2. The Structural Taxonomy

2.1 Dimension 1: Authority Locus

This dimension describes where the belief system locates epistemic authority — the final source of truth and guidance.

External-Hierarchical (EH): Authority resides outside the individual: sacred texts as interpreted by authorized institutions, clerical hierarchy, doctrinal councils, tradition. The believer’s role is to receive, accept, and transmit — not to evaluate or originate. Epistemic virtue is faith: acceptance of authoritative claims without, or against, personal evidence. Indicators: emphasis on orthodoxy, deference to clergy, scriptural literalism, distrust of unvalidated personal experience, heresy as a live category.

Internal-Experiential (IE): Authority resides in the individual’s direct experience, contemplation, or insight. External sources serve as guides rather than final authorities. The practitioner’s role is to investigate, experience, and verify. Epistemic virtue is knowledge through direct realization. Indicators: orthopraxy over orthodoxy, teacher–student rather than clergy–congregation structure, encouragement of personal experience and of doubt, contemplative or self-inquiry practice.

Most traditions contain both; the taxonomy classifies emphasis and structural dominance, not exclusive presence.

Two qualifications attach to this dimension at the outset. First, no validated instrument for authority locus exists; when we cite evidence in §5, the measured constructs are religiosity, dogmatism, and fundamentalism — neighbors, not the dimension itself — and we will not let the difference blur. Second, the nearest measured relative deserves naming: the Religious Schema Scale’s “truth of texts and teachings” factor correlates .80–.81 with the classical Religious Fundamentalism Scale (Streib et al., 2010) — empirically, a fundamentalism measure by another name. The one research program that has defined a structural, content-neutral religious pattern and carried a common instrument across traditions is the intratextuality model of fundamentalism — fundamentalism as a specific relation to a sacred text (divine, inerrant, self-interpretive, privileged, authoritative, unchanging) rather than as any particular doctrine — whose scale has been applied to Christian, Muslim, and Jewish samples (Hood, Hill, & Williamson, 2005; Williamson, Hood, Ahmad, Sadiq, & Hill, 2010; Hammer & Lazar, 2019). That program shows a structural definition can travel across traditions; it measures text-relation, not cognitive function, and we build on it accordingly.

2.2 Dimension 2: Stress Architecture

This dimension describes how the belief system relates to emotional stress — particularly guilt, fear, and shame.

Guilt-Installing (GI): The system installs baseline emotional stress as a structural feature: doctrines of inherent sinfulness (baseline guilt), vivid divine punishment (fear installation), behavioral codes whose violation generates shame. The stress is structural — a designed feature, not a side effect. Indicators: original-sin-type doctrine, detailed eschatological punishment, confession/atonement cycles that reactivate guilt, emphasis on human depravity, absolutist codes with cosmic consequences.

Equanimity-Promoting (EP): The system aims to reduce emotional disturbance and cultivate stability: practices that attenuate fear, guilt, and craving; teachings that normalize imperfection without installing permanent guilt; suffering attributed to ignorance or unskillful action rather than inherent deficiency. Indicators: meditative practice, error framed as correctable through understanding, present-moment emphasis, suffering-as-misunderstanding.

This dimension has an operational shadow in the clinical literature. The Penn Inventory of Scrupulosity decomposes into exactly two factors — fear of having committed sin, and fear of punishment from God (Abramowitz, Huppert, Cohen, Tolin, & Cahill, 2002) — which is the GI pole rendered psychometric. And the mechanism we propose was proposed before us, in the cognitive-behavioral literature: religious doctrine can foster obsession-enabling beliefs because it “(a) imposes explicit moral standards for thinking and behaving, (b) is inculcated by influential authority figures (e.g., clergy), and (c) includes the possibility of severe punishment (e.g., eternal damnation)” (Abramowitz & Jacoby, 2014, reviewing Rachman and Salkovskis). We cite this as convergence and inherit its modality: it is offered there, as here, as a proposal.

2.3 Dimension 3: Dependency Model

This dimension describes whether the system’s relief mechanisms create ongoing dependency or promote eventual autonomy.

Loop-Creating (LC): The system’s stress-relief mechanisms reactivate the conditions for future stress. The canonical cycle: guilt → absolution → temporary relief → renewed awareness of sinfulness → guilt. Relief and relapse run on the same circuit; the member is never well, only current. Indicators: recurring confession/absolution cycles, doctrine that full liberation is impossible in this life, doubt framed as moral failure requiring renewed effort, unwinnable “spiritual warfare,” exit penalties (hell, ostracism, family rupture).

Autonomy-Supporting (AS): The system aims to make itself progressively unnecessary. The practitioner develops capacities — self-awareness, equanimity, discernment — that reduce dependence on the system’s institutional forms. Indicators: the teaching explicitly a means, not an end; personal discernment cultivated; self-regulation built; the goal state independent of external authority; no penalty for leaving.

The formal treatment of the loop — its decomposition of installed, induced, and reactive stress, and its relief-reactivation dynamics — is developed in this series’ collective-level companion (Paper 1.2, §§2.2, 2.4); here we use the construct clinically and cite the model rather than restate it. Independent convergence exists from a literature that has never heard of that model: a clinical interview study of former cult members concluded that cult membership shares core characteristics with addictive disorders — “persistence despite damage, initial psychological relief, occupation of an exclusive place in the thoughts of members” — and named social and therapeutic follow-up and family contact as important to leaving (Rousselet, Duretete, Hardouin, & Grall-Bronnec, 2017; N = 31, qualitative). That is the loop-creating architecture described in the vocabulary of substance dependence by clinicians working from cases.

Exit penalties, the LC pole’s most visible indicator, have a population-scale correlate. In pooled General Social Survey data, people raised and remaining in high-cost sectarian groups (the study names Latter-day Saints and Jehovah’s Witnesses) report better health than stayers elsewhere — while people who leave such groups report worse health than other leavers (Scheitle & Adamczyk, 2010). The pairing matters: the same architecture that rewards staying penalizes going, which is what a loop that integrates emotional regulation, identity, and social life into membership should produce. Three cautions travel with this anchor, and we print them: the outcome is self-reported general health, not a mental-health measure; the design is repeated cross-sections, and the authors themselves name selection and causation as entangled; and “high-cost” is a construct from the strictness literature in the sociology of religion — theological and cultural exclusivity — not our LC criterion. Mapping high-cost onto loop-creating is a theoretical step we take in the open: the strictness tradition explains high-cost groups’ strength via the screening-out of free riders, and our reading adds a stress-architecture channel; the two are competing-and-compatible explanations that the §7 predictions are designed to separate.

2.4 The Taxonomy Matrix

The three binary dimensions define eight theoretical architectural types. Belief systems and communities exist on continua, and individual experience varies within any community. The taxonomy predicts that communities clustering toward the EH-GI-LC pole will show worse average mental-health outcomes than communities clustering toward the IE-EP-AS pole, regardless of theological content — a prediction §8.3 constrains to within-context tests first (P6).

The outcome column states hypotheses, not findings. Because the dimensions correlate in the measured record (§1.3), the mixed cells should be rare in the wild — which is itself a prediction: if they are common, the clustering claim weakens; if systems occupying them can be identified, they are the taxonomy’s natural experiments. We claim no real-world instances for the mixed cells here.

3. Evidence: Stress Architecture and Mental Health

3.1 Guilt-Installation and Scrupulosity

Scrupulosity — obsessive-compulsive symptoms organized around religious or moral failure — is where the GI dimension meets the clinical record, and the record’s headline is its spread. The frequency of scrupulosity in OCD “varies in the literature from 0% to 93% of cases, and this variability seems predicated on the importance of religious belief and observance in the community examined” (Greenberg & Huppert, 2010). The ladder beneath that range, as assembled in the field’s own review: religious obsessions in about 6% of a 425-patient U.S. field-trial sample and religion the primary theme in 5%; 10% and 33% in two later large Western patient samples; and, in less secular societies, 42% of Turkish outpatients and as many as 50% and 60% of OCD patients in Saudi Arabia and Egypt (Abramowitz & Jacoby, 2014, whose text we follow for the primary figures; Tek & Uluğ, 2001, for the Turkish sample directly). This pattern is a community-level finding, not an individual-content one — a step toward the architectural claim rather than a statement of it: the field’s variable is community religious intensity, and the guilt-architecture moderator this paper proposes remains P3’s target. We also carry the same review’s brake: the evidence that religion causes OCD “is scarce” (Greenberg & Huppert, 2010).

Within scrupulosity, what tracks severity? Not religiosity as quantity, and not general pathology — the specific measured proxies are more architectural than either. Among 147 OCD patients, scrupulous and non-scrupulous groups did not differ on overall OCD severity, but scrupulous individuals with a more negative concept of God had more severe symptoms, while a positive God-concept was unrelated (Siev, Baer, & Minichiello, 2011) — an asymmetry, and the GI asymmetry specifically. In 273 Muslim undergraduates, a structural model ran religiosity (alongside neuroticism) → obsessive beliefs → guilt and self-punishment → scrupulosity (İnözü, Kahya, & Yorulmaz, 2020) — one model, one sample, and the causal chain our GI construct names. And in 180 treatment-seeking adults with OCD, scrupulosity differed by religious affiliation (Catholics highest; Hedges’ g = .83 vs. Jewish participants) while global OCD severity did not differ by affiliation at all (Buchholz et al., 2019) — a content-specific pathology dissociating from general pathology, which is the shape our framework predicts. One caution the same literature requires: the ordering of traditions is unstable across studies (the scale’s own development sample put devout Protestants and Catholics above devout Jews; a later sample put Catholics highest), so no ranking of traditions by guilt-load is asserted here.

The limits of this evidence are part of the record. The studies are correlational. The vast majority of religious people have no OCD — a base-rate fact any explanatory model must accommodate, and ours does by prediction rather than concession: the taxonomy holds that only GI-dominant architectures should carry the association, so the denominator “all religious people” is the wrong one. And the religiosity–scrupulosity link has failed to appear in several named populations (Jews, American Protestants, Turkish Muslims, Iranian schoolchildren — as reviewed by Abramowitz & Jacoby, 2014). We read that inconsistency the way the moderator logic suggests: it is what a real moderator looks like from studies that did not measure it. The moderator we propose — the guilt-intensity of the religious environment, as distinct from individual religiosity — currently has no instrument; building one is this framework’s first empirical task (§7, P3).

3.2 Fear-Based Teaching, Struggles, and Trauma

Fear-based religious teaching — vivid hell, divine wrath, demonic threat — is the GI pole at its most acute, and the clinical construct proposed for its aftermath is Religious Trauma Syndrome: complex-PTSD-like symptoms (intrusive imagery, avoidance, hypervigilance toward divine punishment, numbing, identity confusion) arising from authoritarian religious environments (Winell, 1993, 2011). The available prevalence estimate — 27–33% of U.S. adults lifetime, 10–20% currently symptomatic — comes from the self-selected panel study already cited, and inherits its methods (Slade et al., 2023).

The measurement situation is itself a finding about the field. Religious and spiritual struggle is a psychometrically mature construct: a 26-item scale with six replicated domains — divine, demonic, interpersonal, moral, doubt, ultimate meaning — validated across samples (Exline, Pargament, Grubbs, & Yali, 2014), two of whose domains (divine: negative emotion centered on one’s relationship with God; moral: worry and guilt over perceived offenses) are close operationalizations of what we call GI. Religious trauma, by contrast, is a clinically proposed construct that has not yet been validated to that standard. The gap between a measured struggles literature and a proposed trauma construct is a research target, not a rhetorical inconvenience, and we name it as such.

3.3 Hell Belief and Well-Being

The cleanest evidence for the stress-architecture dimension is a single study with three legs, and we report all three (Shariff & Aknin, 2014). Cross-nationally (63 countries; 52 in the covaried model), national rates of hell belief predicted lower life-satisfaction rank and daily affect (β = −1.51 and −1.38) while heaven belief predicted higher (β = 1.74 and 1.49), controlling for GDP, inequality, inflation, unemployment, political stability, belief in God, and religious attendance — with the caution that a 52-country regression with ten predictors is the weakest of the three legs, whatever its R². At the individual level (n = 257,597, multilevel), the same divergence held (hell −.28, heaven +.25). And experimentally (N = 417), participants primed to write about hell reported less happiness and more sadness and fear than heaven and control conditions — while the heaven prime did not differ from control on any measure. The causal evidence is thus one-sided: the punishment component does active harm on the manipulation’s timescale; the reward component is inert. That is a sharper claim than symmetric divergence, and it is the one we adopt: the stress-installing component and the meaning-providing component of one theology are separable in their effects. No replication of this study exists — its complete citing literature has been enumerated and contains none — so it stands as one study, and we cite it as one study.

Independent convergence exists on a different malevolence marker: in the 2010 Baylor Religion Survey (N = 1,627), belief in supernatural evil was associated with general anxiety and paranoia, with secure attachment to God buffering parts of the association — fully for social anxiety and paranoia; for general anxiety, among women only (Jung, 2020). The buffer is an AS-pole variable moderating a GI-pole harm, inside one dataset.

3.4 Positive and Negative Religious Coping

The religious-coping literature supplies the taxonomy’s best-replicated convergence, and its effect sizes belong in the text. The constructs: positive religious coping reflects “a secure relationship with a transcendent force, a sense of spiritual connectedness with others, and a benevolent world view”; negative religious coping reflects “underlying spiritual tensions and struggles within oneself, with others, and with the divine” (Pargament et al., 2011). The outcomes, meta-analytically (49 studies, 105 effect sizes, N = 13,512): positive coping correlates .33 with positive adjustment and −.12 with negative adjustment; negative coping correlates .22 with negative adjustment — and .02, non-significantly, with positive adjustment (Ano & Vasconcelles, 2005). All effects are cross-sectional, as the authors state.

The null cell is the interesting one. Negative religious coping predicts more distress without predicting fewer positive outcomes — people who feel punished by God do not thereby report less purpose or growth, and the source meta-analysis notes that some struggles precede growth. Mapped onto the taxonomy, this is the §3.3 result arriving from an independent literature: the GI architecture installs distress without removing meaning, because stress-installation and meaning-supply are different components of the system. Our §3.4 claim is therefore the narrow one the data support: negative religious coping consistently predicts worse outcomes on negative-outcome measures; the beneficial and harmful faces of religious coping are, in most samples, empirically orthogonal (Pargament et al., 2011) — the one true orthogonality in this territory, and a direct support for treating “religion” as a package of separable components (§5.3).

4. Evidence: Dependency Architecture and Deconversion

4.1 The Exit-Distress Record, in Scope

What happens to people who leave? The answer has a scope structure.

In high-demand and clinical samples, significant distress is real and affects a substantial minority. The first critical review of the question concluded that people entering such groups are not conspicuously pathological, that current members appear generally well-adjusted — while noting that conformity pressure could mask pathology, a measurement artifact that a loop running on conformity would itself predict — and that “at least a substantial minority of former … members experience significant adjustment difficulties,” not attributable to demand characteristics (Aronoff, Lynn, & Malinoski, 2000). The oldest clean number is of the same size: of 66 people who left the Unification Church, 36% reported serious emotional problems after leaving — and the same cohort had apparently achieved stable adjustment at a mean of 3.8 years (Galanter, 1983). Whether some distress in this older literature is attributable to the group left or to coercive exit interventions of the era is itself a contested question (Lewis & Bromley, 1987), and any structural account must eventually discriminate its predictions from that rival; ours does so at §7 (P4). Former members are also not one population: latent-profile work finds the formerly religious heterogeneous (McLaughlin et al., 2022), and the deconversion research program’s own review of high-tension exits reports “various trajectories” of well-being and growth, not a single outcome (Keller, Bullik, & Streib, 2024).

In population data on high-cost groups specifically, the exit penalty appears: leavers from high-cost sectarian groups report worse health than other leavers, in the same dataset where stayers in those groups report better health than other stayers (Scheitle & Adamczyk, 2010; §2.3’s cautions apply).

In general-population data on ordinary disaffiliation, the picture splits cleanly by outcome — the scope condition the whole section rests on. Disaffiliates in pooled GSS data show poorer health and well-being than both the consistently affiliated and the consistently unaffiliated — and the entire disadvantage is completely mediated by frequency of attendance (Fenelon & Danielsen, 2016). In an 11-wave Dutch panel (~20,000 adults, of whom 450 deconverted), deconversion brought no change in well-being, alongside small declines in agreeableness, conscientiousness, and openness — the last in tension with cross-sectional findings that deconverts score higher on openness (Bleidorn, Lenhausen, Schwaba, & Hopwood, 2024; Streib, Hood, Keller, Csöff, & Silver, 2009). And in six waves of a New Zealand panel (N = 46,672), causal estimation under modified treatment policies found deidentification lowering meaning and purpose relative to remaining religious — while leaving dones still higher in meaning than the steadily secular (Van Tongeren, Davis, Sibley, & Bulbulia, 2026). These three results are not in conflict, but only if the constructs are kept apart: the population cost of ordinary exit shows up in lost participation and lost meaning, not in global well-being.

Two structural lessons follow. First, the fully-mediated attendance result is the benchmark this taxonomy must beat: in those data, lost social participation absorbs the entire exit penalty, leaving no residual variance for architecture-specific mechanisms. If architecture adds predictive power beyond participation — as P4 claims — that must be shown, not assumed. Second, exit is a two-sided ledger: the role-exit literature’s synthesis is that leaving catalyzes distress and simultaneously relieves the strain that preceded it (Haire, 2022). A taxonomy that counts only the exit cost is measuring one side of the ledger — and the two-sidedness sharpens the LC/AS contrast into a testable double dissociation: an AS system leaves little strain to relieve and imposes little cost to exit; an LC system maximizes both (§7, P4).

4.2 Persistence After Intellectual Deconversion

A striking clinical observation is that people who have intellectually rejected their former beliefs often continue to experience the emotional signature for years: the ex-fundamentalist who no longer believes in hell and still panics at its imagery; the ex-Catholic who no longer endorses confessional theology and still feels intrusive guilt (Winell, 1993, 2011 — clinical description, cited as such).

The structural half of that observation now has population-scale evidence, in the religious-residue literature. Across nationally representative samples in four cultures (Studies 1–3: n = 3,071; a preregistered n = 1,626; N = 31,604), the formerly religious differ from the never religious across cognition, emotion, and behavior — sitting between the currently religious and the never religious (Van Tongeren, DeWall, Chen, Sibley, & Bulbulia, 2021). The stairstep replicates for moral foundations, with four-wave longitudinal evidence in young samples that the residue “may erode over time” (Van Tongeren, DeWall, Hardy, & Schwadel, 2021). It replicates outside the founding lab, discipline, and continent — in European survey data, disaffiliates are consistently more religious than lifelong nones, with greater residue among those more religiously committed as children: a dose–response relation between depth of embedding and size of remainder (Beider, 2023). And the program has published its own boundary: in the political domain, dones do not carry residue but move away (Van Tongeren, Hardy, Taylor, & Schwadel, 2026) — residue is domain-specific, which is a reason to trust the program and a caution against reading “the whole architecture persists” into it. What this literature anchors is the structural proposition: propositional deidentification does not clear the psychological structure; the structure persists, measurably, at population scale. It does not measure clinical symptoms, and we have kept the symptom claims on the clinical literature where they belong.

The mechanism reading is the series’: what persists is an installed regulatory setpoint outliving the membership that installed it (Paper 1.2, §3.3), and — where acquisition was fearful — conditioned learning that extinction does not erase; the neural account, and the decomposition of exit into four separable undoings whose partial completion explains these dissociations, are given in the individual-level companion (Paper 1.3, §§5.4, 7.2).

4.3 Protective Factors: Verdicts, Not a List

What protects those who leave? The literature does not converge on a tidy set, and we grade the candidates instead of listing them.

Alternative community — supported, and possibly primary. Among 147 former Haredim, perceived social support and current community affiliation predicted coping resources through the transition (Erhard, 2025); among 1,146 who left ultra-Orthodox Judaism, perceived social support — and social-media use for problem-solving rather than escape — tracked better mental-health outcomes (Keller, David, & Trachtenberg, 2026); and the two-country deconversion study found German deconverts faring worse than Americans, who reintegrated more easily where more religious alternatives existed (Streib et al., 2009). Add §4.1’s mediation result — the population exit penalty vanishing once attendance is controlled — and community stops being one factor among five: on current evidence it is the candidate primary mechanism, and we state that as a falsifiable ranking. All of this is correlational, and the quantitative exit studies are culturally narrow (largely Jewish-exit and Christian-Western samples).

Gradual rather than abrupt departure — no comparative evidence exists. The exit literature describes staged processes (first doubts, weighing, turning point, ex-role construction: Ebaugh, 1988; organized for religious exit by Haire, 2022), but no study compares outcomes of gradual versus abrupt religious exit. The factor is a clinically plausible inference and moves to the predictions (§7, P7), where it is cheap to test.

Access to therapy or support groups — clinical synthesis. Clinical sources recommend it; the interview study of former cult members names social and therapeutic follow-up as important to leaving (Rousselet et al., 2017); a scoping review of psychotherapy for adverse religious experiences exists (Zaeske et al., 2024). We found no outcome evidence that therapy access protects post-exit mental health, and we say so.

Agency in the exit — adjacent support, recast. The voluntary-versus-forced framing of older debates is confounded with the coercive-intervention controversy above. The nearest quantitative relative is an immigration-psychology study of Orthodox Jewish disaffiliation (N = 206): being pushed out by distress predicted worse current wellness; attaining the goals one left for predicted better well-being — and the strongest reported pull was toward “opportunities for physical and ideological autonomy” (Engelman, Milstein, Schonfeld, & Grubbs, 2020). Agency and goal-attainment, not mere voluntariness, are the supportable form of this factor — and what exiters report leaving for is the autonomy pole of our third dimension.

Alternative meaning frameworks — the deficit is real; the remedy is untested. The causal meaning-loss estimate of §4.1 establishes the deficit an alternative framework would need to fill (Van Tongeren et al., 2026); no study tests whether supplying one protects. Supported in the negative; open in the positive.

A sixth candidate the list logic missed: pre-exit access to a credible alternative model. Qualitative work with ex-Jehovah’s Witnesses finds exiters citing like-minded questioners inside the organization and the sight of former members thriving (Murphy, 2026); a symbolic-interactionist study of 23 former members across 11 groups concludes exit occurred “when alternate discourses or other identity resources became available” (Coates, 2013). Alternative community again, operating before the exit.

5. Evidence: Authority Locus and the Adaptive Pole

5.1 Cognitive Style and Religious Engagement

What does the behavioral evidence actually link to a more persistent, less flexible cognitive style? Religiosity and dogmatism — not content, and not, so far, authority locus, for which no instrument exists.

In the one religion-specific study with performance measures (N = 744, U.S.-dominant), non-religious participants outperformed religious participants on the Wisconsin Card Sorting Test, the Remote Associates Test, and the Alternative Uses Task, and within religious participants weekly service attendance went with lower set-shifting accuracy than yearly attendance (67.5% vs. 74.0%); the authors state the causal direction is open and recommend longitudinal designs (Zmigrod, Rentfrow, Zmigrod, & Robbins, 2019). The effect sizes are modest on the classic tasks (ηp² = .04 and .02) and implausibly large on the divergent-thinking task (.36), a spread that itself counsels caution. In a 63,235-participant internet cohort, self-rated dogmatism — which varied across religious groups by 0.87 SD — tracked performance specifically on tasks that pit intuition against logic (0.60 SD on a color-word conflict task) and not on matched non-conflict reasoning (0.01 SD); the religious groups are reported anonymously, and the authors warn against extrapolating group differences or individual-level inferences (Daws & Hampshire, 2017). Fundamentalism scores correlate with lower performance-based flexibility in a penetrating-brain-injury cohort (r = −.31) — but not in that study’s healthy controls (r = −.07, n.s., N = 30) (Zhong, Cristofori, Bulbulia, Krueger, & Grafman, 2017; the lesion findings themselves are reviewed in Paper 1.3, §3, and are not re-argued here). And fundamentalism relates to a greater need for cognitive closure, which partially mediates its association with outgroup derogation — with only some closure facets carrying the path, and no meta-analysis of the relation yet existing (Brandt & Reyna, 2010).

No study has compared cognitive flexibility across authority-locus architectures — the constructs have never been operationalized — and none has compared flexibility profiles across Muslim, Christian, and Jewish samples. The cross-tradition claim our taxonomy suggests is therefore stated in §7 as a prediction with a named design, not here as a finding. The mechanism the February literature of this field sometimes implies — external authority under-exercising evaluation circuits that internal authority exercises — is likewise a hypothesis; the studies above cannot assign direction.

5.2 Internal Authority, Self-Regulation, and Decentering

The IE pole’s signature practices — meditation, contemplation, self-inquiry — have a real but bounded evidence base.

A narrative review credits mindfulness with increased subjective well-being, reduced symptoms and emotional reactivity, and improved behavioral regulation (Keng, Smoski, & Robins, 2011 — a review without pooled effects, cited for direction only). The comprehensive meta-analysis of meditation in non-clinical adults reports a medium average effect (r = .28 across 163 studies; r = .27 in reviewed journals) with a telling gradient: strongest for emotionality and relationships, medium for attention, weakest for cognitive measures — and notes that nearly three-quarters of 595 identified studies had to be excluded for methodological problems (Sedlmeier et al., 2012). Against active controls, the picture tightens: across 47 randomized trials (3,515 participants), mindfulness programs showed moderate evidence for improvements in anxiety (effect size 0.38 at 8 weeks, 0.22 at 3–6 months), depression (0.30; 0.23), and pain (0.33), low or insufficient evidence for attention and positive mood — and no evidence of superiority over any active treatment (Goyal et al., 2014). Nor is the pole harm-free: a systematic review of meditation adverse events found a total prevalence of 8.3% — 3.7% in experimental studies and 33.2% in observational ones — with anxiety and depression the commonest categories, occurring also in practitioners with no prior history (Farias, Maraldi, Wallenkampf, & Lucchetti, 2020); the phenomenological mapping of difficult meditation-related experiences in Western Buddhists finds a median symptom duration of one to three years in affected practitioners — a sample recruited for such experiences, so a severity profile, not a prevalence (Lindahl, Fisher, Cooper, Rosen, & Britton, 2017). Summed, the adaptive pole’s practice evidence: modest, real reductions in distress; no demonstrated advantage over other active treatments; the weakest evidence on the higher-functioning outcomes; and a real adverse-event tail.

What, then, is the IE pole’s plausible active ingredient? The construct the contemplative-science literature offers is decentering — the metacognitive capacity to shift perspective from within one’s experience onto that experience, decomposed into meta-awareness, disidentification from internal experience, and reduced reactivity to thought content (Bernstein et al., 2015) — a capacity the contemplative traditions have long named: witness consciousness, mindful awareness, viveka. To “observe mental states without automatically identifying with them,” as §2.1’s indicators already put it, is Bernstein’s second process by name. Component evidence exists that defusion-type training outperforms inactive comparisons in laboratory studies (Levin, Hildebrandt, Lillis, & Hayes, 2012 — component-level effect sizes not reproducible here). On the other side of the same coin, entrenchment travels with impaired metacognition: in two general-population samples (n = 381, 417), holders of radical beliefs showed specifically reduced metacognitive sensitivity on a perceptual task that excludes knowledge and motivation confounds — a result about political radicalism, which we cite as ideological rather than religious evidence (Rollwage, Dolan, & Fleming, 2018).

Two further findings suggest the configuration this points to is realizable inside religious life, not only outside it. In a single-case series (n = 5), an acceptance-and-commitment treatment for scrupulosity — defusion-centered, no in-session exposure — reduced religious compulsions from 25.0 to 4.3 per day while participants’ strength of religious faith declined only 4–7% (Dehlin, Morrison, & Twohig, 2013): a clinical demonstration, at n = 5, that the fused relation to religious content — fusion in the acceptance-and-commitment sense, thoughts taken as their objects — can be reduced while engagement is preserved. And the psychology of religion has measured an engaged-but-non-literal stance for two decades: the Post-Critical Belief Scale’s second naiveté quadrant — inclusion of transcendence held with symbolic rather than literal processing (Duriez, Fontaine, & Hutsebaut, 2000). The scale relates to sense of coherence in a well-powered sample (N = 636), with the salutogenic association moderated by age and gender (Zarzycka & Rydz, 2014) — a corroboration of the configuration’s shape, in a different vocabulary, at scale rather than quadrant resolution, and with a coherence rather than symptom outcome.

On that base we advance one proposal, ours and marked as such. We propose, as a hypothesis and not a finding, that the protective core of the IE-EP-AS configuration is a trained relation to one’s own religious representations: high engagement with low identification — the capacity to enter religious states and frames intensely while retaining the ability to experience them as representations. No literature on protective religious externalization exists; we searched for one from several directions and it is not there. What exists are the five components above — a defined metacognitive capacity, evidence that training it helps, evidence that entrenchment impairs it, one small demonstration that religious engagement survives its cultivation, and a measured engaged-but-symbolic stance with a well-being correlate. The proposal assembles them into a testable configuration (§7, P8). If it fails, the rest of this section stands unchanged.

5.3 The Paradox, Decomposed

The religion–mental-health paradox eases under decomposition, and the decomposition now has to be stated with its strongest counter-finding inside it.

Religion is associated with better outcomes where it provides community and participation, meaning and purpose, self-regulatory practice, and coping frameworks. The best-controlled benefits literature is emphatic that the first of these carries the most weight: associations with flourishing are “much stronger for communal religious participation than for spiritual-religious identity or for private practices,” and social support alone explains only a small portion of them (VanderWeele, 2017); the mortality meta-analysis likewise isolates organizational activity as the effective channel (Chida et al., 2009). We concede it: the outcome literature’s strongest signal runs through a channel our taxonomy did not privilege, and any version of this framework that predicted private interiority as the master benefit variable would be wrong on current evidence. Our claim is accordingly bounded: the three dimensions are proposed as sources of variance within and across communities — as an account of why observably similar levels of participation produce divergent outcomes across architectures — not as a replacement for the participation effect. Participation is at least a non-doctrinal variable — form rather than creed — which is congenial to decomposition, though it is not one of this paper’s three dimensions and is not claimed as one.

Religion is associated with worse outcomes where it installs unresolvable baseline guilt and fear (§3), loops relief through renewed stress (§4), and — on evidence linking dogmatism rather than authority locus — travels with a more persistent cognitive style (§5.1). The two separations actually in hand hold the frame together: the beneficial and harmful faces of religious coping are empirically orthogonal (§3.4), and the punitive and meaning-providing components of one theology dissociate experimentally (§3.3). “Religion” in outcome research is a package of separable components with opposing effects; studies that aggregate across architectures will keep producing contradictions.

6. Clinical Applications

6.1 Assessment

The taxonomy offers clinicians a frame for assessing the architectural features of a client’s religious involvement rather than inferring from denomination — with the explicit caveat, developed in §8.1, that the taxonomy is a population-level tool and individual assessment remains a clinical judgment pending instrument validation. The orienting questions: Where does the client locate epistemic authority? What stress does the system install, and is it framed as resolvable? Does the relief mechanism resolve or recycle? How dependent is the client on the system for regulation, identity, and social life?

6.2 Intervention Matching

Different architectural features suggest different targets — offered as a clinical proposal whose comparative test is P5.

For GI-dominant distress: CBT for irrational guilt and fear-based distortions; exposure and response prevention for scrupulosity, for which specialized protocols exist (Huppert & Siev, 2010); acceptance-based approaches with defusion elements, which have preliminary single-case support in scrupulosity specifically (Dehlin et al., 2013); self-compassion work against internalized unworthiness.

For LC-dominant distress: graded independence-building; identity reconstruction (narrative and schema approaches); building alternative social support — which §4.3 ranks as the candidate primary protective factor; psychoeducation normalizing the difficulty of leaving.

For EH-dominant distress: Socratic questioning and guided discovery; perspective-taking; mindfulness-based practices that build internal epistemic authority; bibliotherapy across traditions.

6.3 Working Within Religious Frameworks

The taxonomy does not require clients to abandon religious belief. Many individuals benefit from migrating within their tradition — from an EH-GI-LC expression toward an IE-EP-AS expression of the same faith. Contemplative Christianity, progressive Judaism, Sufi Islam, and Buddhist practice communities all sit, on the taxonomy’s classification, nearer the IE-EP-AS pole within their traditions; whether they are in fact healthier is what P2 tests, and §8.3’s record counsels testing it within cultural context. Clinicians can support clients in finding such expressions rather than framing the choice as religion versus mental health.

7. Predictions

P1 (Architecture over content). The three structural dimensions will predict mental-health outcomes (anxiety, depression, well-being, cognitive flexibility) above and beyond denominational affiliation, religiosity level, and belief content. This is the taxonomy’s central claim, and it is a prediction — nothing reviewed above tests it directly.

P2 (Pole contrast). Individuals in EH-GI-LC communities will show higher rates of scrupulosity, religious trauma presentations, and post-exit distress than individuals in IE-EP-AS communities, controlling for intensity of involvement.

P3 (The instrument). A measure of environmental guilt-intensity — the GI load of a congregation or community, as distinct from individual religiosity — can be constructed and will moderate the religiosity–scrupulosity relation, explaining the literature’s named null populations (§3.1). No such instrument exists; building it is this framework’s first empirical task.

P4 (Exit distress tracks architecture). Post-exit distress severity will be predicted by the dependency architecture of the system left, not its theology — beyond the attendance-mediation benchmark (§4.1), which is the result to beat; with the LC/AS double dissociation (high strain relieved and high cost incurred vs. little of either); with a larger meaning-decrement on exit from LC systems; and discriminably from the rival attributing distress to post-exit handling rather than to the system left.

P5 (Matched intervention). Interventions matched to the dominant architectural feature (GI → guilt-focused; LC → dependency-focused; EH → flexibility-focused) will outperform generic or content-focused intervention.

P6 (The cross-tradition test). If the dimensions are genuinely structural, their outcome associations should hold across traditions. The evidence on this is currently split by dimension: the stress-architecture effect has one real invariance datum — the hell/heaven well-being association did not differ between Abrahamic and non-Abrahamic believers in a 203,036-respondent classification (Shariff & Aknin, 2014, Study 2) — while the authority-locus effect has none, and the cross-cultural record of §8.3 predicts that naive invariance will fail. The natural design is the cross-tradition structural comparison proposed as Paradigm D in the companion paper (Paper 1.2, §6.2), using the intratextuality instrument family (§2.1) alongside performance-based flexibility tasks.

P7 (Exit gradient). Gradual, supported exits will show better outcomes than abrupt, unsupported exits from the same architecture class. No comparative study exists (§4.3); the collective-level companion marks the same comparison for its Paradigm B (Paper 1.2, §3.3), and the design is cheap.

P8 (Decentering as the IE mechanism). Decentering capacity will moderate the relation between religious engagement intensity and pathology: at high decentering, high engagement without elevated symptoms; at low decentering, engagement intensity tracking symptoms. This is §5.2’s proposal in testable form; existing instruments suffice to run it.

8. Limitations

8.1 Within-System Variation

The taxonomy classifies systems at the community level; individual experience varies enormously within any community, and the formerly religious are themselves heterogeneous in profile (McLaughlin et al., 2022). Population-level architecture is not an individual diagnosis; §6.1’s assessment frame is an orientation, not an instrument.

8.2 Selection and Causation

Do GI-LC systems produce anxious, rigid people, or do anxious, rigid people select GI-LC systems? Both, plausibly, and the reviewed evidence cannot apportion: the coping and scrupulosity literatures are cross-sectional (Ano & Vasconcelles, 2005); the flexibility studies state their direction question openly (Zmigrod et al., 2019); the population exit data name selection explicitly (Scheitle & Adamczyk, 2010); and the benefits literature’s own methodologists have shown feedback in both directions between attendance and depression (VanderWeele, Jackson, & Li, 2016). Longitudinal designs tracking individuals across architectural transitions are the answer, and P4/P7 specify two.

8.3 Cultural Embeddedness — the Strongest Current Constraint

The taxonomy’s content-neutrality has a cross-cultural record to face, and it belongs here in numbers. The Religious Schema Scale — the nearest measured relative of our authority-locus dimension — achieves configural but not metric invariance even between Germany and the United States (Streib et al., 2010). Outside the West, the structure–outcome relation reverses sign. In Iran (N = 398), the textual-authority schema correlates positively with interreligious openness (+.14) and predicts lower anxiety (−.16) and depression (−.12) (Ghorbani, Watson, Amirbeigi, & Chen, 2016). In Indonesia (N = 761), textual authority and openness correlate +.374 — against −.44 in Germany (Ardi, Tobing, Agustina, Iswahyudi, & Budiarti, 2021). In India, all three schemas correlate positively (Kamble, Watson, Marigoudar, & Chen, 2014), and Malaysian data run the same direction on the textual-authority–openness pairing (Tekke, Watson, Hisham İsmail, & Chen, 2015). In the one direct Christian–Muslim comparison using the xenosophia measure, Turkish Muslim and German Christian theology students score identically on interreligious openness (3.74 vs. 3.74) while fundamentalism — a different fundamentalism instrument from the Altemeyer scale §2.1 discusses — relates to that openness at −.42 among the Christians and .01 among the Muslims (Demmrich, Ağılkaya-Şahin, & Şenel, 2024). Within the RSS’s own validation, the textual-authority schema correlates positively with purpose in life (.14 German, .25 U.S. samples) — external authority buys meaning even where it costs openness (Streib et al., 2010).

What this record does, and does not do: it does not test our taxonomy, whose dimensions remain unoperationalized; the RSS is a neighbor construct. It does constrain any claim that a structural pattern carries identical psychological meaning across cultural contexts: where an architecture is normative — where textual authority is the community’s water rather than one stream in it — its correlates evidently differ, in sign and not merely size. The taxonomy’s architecture–outcome predictions must therefore be tested within cultural context before they are compared across it, and P6 is written accordingly. A framework whose limitations section knows this record is constrained by it; a framework that ignored it would be refuted by it in review.

8.4 Positive Functions of GI-LC Features

The taxonomy must not be read as claiming GI-LC architectures do nothing for their members. Guilt can motivate prosociality; fear can deter harm; a demanding community can hold a person through crisis — and stayers in the highest-cost groups report the best health in population data (Scheitle & Adamczyk, 2010). The coping meta-analysis’s null cell (§3.4) shows negative coping does not strip positive outcomes; the RSS record shows textual authority correlates with purpose; the residue literature shows even leavers retain prosocial benefits (Van Tongeren, DeWall, Chen, Sibley, & Bulbulia, 2021) and meaning above the steadily secular baseline (Van Tongeren, Davis, Sibley, & Bulbulia, 2026). The question is never whether these features have positive functions but whether the net effect on the member is positive, and whether the same functions are achievable at lower architectural cost — which is precisely what P2 and P5 test.

8.5 Measurement

None of the three dimensions has a validated instrument; the nearest neighbors (fundamentalism scales, RSS, coping measures) each capture part of one dimension while dragging surplus meaning. The factorial separability of the dimensions is untested and, given the correlations in §1.3, not to be presumed. The instrument program — P3 first, then the full battery P6 requires — is not an appendix to this framework; it is the framework’s next paper.

9. Conclusion

The relationship between religious belief and mental health is not paradoxical — it is architecturally heterogeneous, and the heterogeneity is measurable wherever the literature has had the constructs to measure it: in a scrupulosity prevalence that tracks community religious intensity across an order of magnitude; in a hell component that experimentally hurts while a heaven component sits inert; in coping faces that are orthogonal rather than opposed; in exit penalties that concentrate where architectures are high-cost and vanish, in ordinary cases, once lost participation is accounted for; in a religious psychology that outlasts its own disavowal.

The structural taxonomy proposed here offers three things, each within its bounds. For researchers: a decomposition of the “religion” variable into three proposed dimensions, an instrument program for operationalizing them, and eight predictions — including the cross-tradition test on which the framework has agreed, in advance and in print, that naive invariance is likely to fail. For clinicians: an assessment orientation and a matched-intervention proposal, offered as proposals, alongside the documented option of migration within a tradition rather than out of it. For individuals: a vocabulary for asking, without any verdict on the truth of their beliefs, whether their belief system’s architecture supports their autonomy and well-being — and a reason to think the question is the right one: on the harm side, the best-supported findings are architectural — installed, unresolvable stress, and relief looped through renewed stress — while on the benefit side the dominant variable, participation, is at least a matter of form rather than creed, though not one of this paper’s dimensions.

The approach is deliberately non-polemical. It does not argue that any theology is true or false, or that religious involvement is generally beneficial or harmful. It argues that how a belief system is structured matters in ways what it teaches does not capture — and it converts that claim from an assertion into a research program.

References

External literature (each entry carries a verification record in the project archive — journal articles at fetch level, several at full-text level; books at book level; entries print only archive-held fields)

• Abramowitz, J. S., Huppert, J. D., Cohen, A. B., Tolin, D. F., & Cahill, S. P. (2002). Religious obsessions and compulsions in a non-clinical sample: The Penn Inventory of Scrupulosity (PIOS). Behaviour Research and Therapy, 40(7), 825–838. DOI 10.1016/S0005-7967(01)00070-5.
• Abramowitz, J. S., & Jacoby, R. J. (2014). Scrupulosity: A cognitive–behavioral analysis and implications for treatment. Journal of Obsessive-Compulsive and Related Disorders, 3(2), 140–149. DOI 10.1016/j.jocrd.2013.12.007.
• Ano, G. G., & Vasconcelles, E. B. (2005). Religious coping and psychological adjustment to stress: A meta-analysis. Journal of Clinical Psychology, 61(4), 461–480. DOI 10.1002/jclp.20049.
• Ardi, R., Tobing, D. H., Agustina, G. N., Iswahyudi, A. F., & Budiarti, D. (2021). Religious schema and tolerance towards alienated groups in Indonesia. Heliyon, 7(7), e07603. DOI 10.1016/j.heliyon.2021.e07603.
• Aronoff, J., Lynn, S. J., & Malinoski, P. (2000). Are cultic environments psychologically harmful? Clinical Psychology Review, 20(1), 91–111. DOI 10.1016/S0272-7358(98)00093-2.
• Beider, N. (2023). Religious residue: The impact of childhood religious socialization on the religiosity of nones in France, Germany, Great Britain, and Sweden. The British Journal of Sociology, 74(1), 50–69. DOI 10.1111/1468-4446.12982.
• Bernstein, A., Hadash, Y., Lichtash, Y., Tanay, G., Shepherd, K., & Fresco, D. M. (2015). Decentering and related constructs: A critical review and metacognitive processes model. Perspectives on Psychological Science, 10(5), 599–617. DOI 10.1177/1745691615594577.
• 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. DOI 10.1111/jopy.12881.
• Brandt, M. J., & Reyna, C. (2010). The role of prejudice and the need for closure in religious fundamentalism. Personality and Social Psychology Bulletin, 36(5), 715–725. DOI 10.1177/0146167210366306.
• Buchholz, J. L., Abramowitz, J. S., Riemann, B. C., Reuman, L., Blakey, S. M., Leonard, R. C., & Thompson, K. A. (2019). Scrupulosity, religious affiliation and symptom presentation in obsessive compulsive disorder. Behavioural and Cognitive Psychotherapy, 47(4), 478–492. DOI 10.1017/S1352465818000711.
• Cashwell, C. S., & Swindle, P. J. (2018). When religion hurts: Supervising cases of religious abuse. The Clinical Supervisor, 37(1), 182–203. DOI 10.1080/07325223.2018.1443305.
• Chida, Y., Steptoe, A., & Powell, L. H. (2009). Religiosity/spirituality and mortality: A systematic quantitative review. Psychotherapy and Psychosomatics, 78(2), 81–90. DOI 10.1159/000190791.
• Coates, D. D. (2013). Disaffiliation from a new religious movement: The importance of self and others in exit. Symbolic Interaction, 36(3), 314–334. DOI 10.1002/symb.60.
• Daws, R. E., & Hampshire, A. (2017). The negative relationship between reasoning and religiosity is underpinned by a bias for intuitive responses specifically when intuition and logic are in conflict. Frontiers in Psychology, 8, 2191. DOI 10.3389/fpsyg.2017.02191.
• Dehlin, J. P., Morrison, K. L., & Twohig, M. P. (2013). Acceptance and commitment therapy as a treatment for scrupulosity in obsessive compulsive disorder. Behavior Modification, 37(3), 409–430. DOI 10.1177/0145445512475134.
• Demmrich, S., Ağılkaya-Şahin, Z., & Şenel, A. (2024). Love thy neighbor: Exploring religious and social openness among prospective theologians in Germany and Turkey. Religions, 15(3), 260. DOI 10.3390/rel15030260.
• Duriez, B., Fontaine, J. R. J., & Hutsebaut, D. (2000). A further elaboration of the Post-Critical Belief Scale: Evidence for the existence of four different approaches to religion in Flanders-Belgium. Psychologica Belgica, 40(3), 153–181. DOI 10.5334/pb.961.
• Ebaugh, H. R. F. (1988). Becoming an Ex: The Process of Role Exit. University of Chicago Press.
• Engelman, J., Milstein, G., Schonfeld, I. S., & Grubbs, J. B. (2020). Leaving a covenantal religion: Orthodox Jewish disaffiliation from an immigration psychology perspective. Mental Health, Religion & Culture, 23(2), 153–172. DOI 10.1080/13674676.2020.1744547.
• Erhard, M. C. (2025). Leaving Haredi Judaism: Coping resources and perceived social support during community transitions and religious disaffiliation. Journal for the Scientific Study of Religion, 64(2), 148–172. DOI 10.1111/jssr.12946.
• 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. DOI 10.1037/a0036465.
• Farias, M., Maraldi, E., Wallenkampf, K. C., & Lucchetti, G. (2020). Adverse events in meditation practices and meditation-based therapies: A systematic review. Acta Psychiatrica Scandinavica, 142(5), 374–393. DOI 10.1111/acps.13225.
• Fenelon, A., & Danielsen, S. (2016). Leaving my religion: Understanding the relationship between religious disaffiliation, health, and well-being. Social Science Research, 57, 49–62. DOI 10.1016/j.ssresearch.2016.01.007.
• Galanter, M. (1983). Unification Church (“Moonie”) dropouts: Psychological readjustment after leaving a charismatic religious group. American Journal of Psychiatry, 140(8), 984–989. DOI 10.1176/ajp.140.8.984.
• Ghorbani, N., Watson, P. J., Amirbeigi, M., & Chen, Z. J. (2016). Religious schema within a Muslim ideological surround: Religious and psychological adjustment in Iran. Archive for the Psychology of Religion, 38(3), 253–277. DOI 10.1163/15736121-12341327.
• Goyal, M., Singh, S., Sibinga, E. M. S., Gould, N. F., Rowland-Seymour, A., Sharma, R., Berger, Z., Sleicher, D., Maron, D. D., Shihab, H. M., Ranasinghe, P. D., Linn, S., Saha, S., Bass, E. B., & Haythornthwaite, J. A. (2014). Meditation programs for psychological stress and well-being: A systematic review and meta-analysis. JAMA Internal Medicine, 174(3), 357–368. DOI 10.1001/jamainternmed.2013.13018.
• Greenberg, D., & Huppert, J. D. (2010). Scrupulosity: A unique subtype of obsessive-compulsive disorder. Current Psychiatry Reports, 12(4), 282–289. DOI 10.1007/s11920-010-0127-5.
• Haire, F. S. (2022). Born again: Narrating the mental health journeys of religious exiters. Sociology Compass, 16(9), e13030. DOI 10.1111/soc4.13030.
• Hammer, J. H., & Lazar, A. (2019). Internal structure and criterion relationships for long and brief versions of the Intratextual Fundamentalism Scale (IFS) among Israeli Jews. Psychology of Religion and Spirituality, 11(4), 358–367. DOI 10.1037/rel0000148.
• Hollier, J., Clifton, S., & Bobo, F. (2026). Mapping the impact of religious harm: Results from a large-scale qualitative study. Journal for the Academic Study of Religion, 39(1), 109–133. DOI 10.1558/jasr.34314.
• Hood, R. W., Jr., Hill, P. C., & Williamson, W. P. (2005). The Psychology of Religious Fundamentalism. Guilford Press.
• Hunsberger, B. (1995). Religion and prejudice: The role of religious fundamentalism, quest, and right-wing authoritarianism. Journal of Social Issues, 51(2), 113–129. DOI 10.1111/j.1540-4560.1995.tb01326.x.
• Huppert, J. D., & Siev, J. (2010). Treating scrupulosity in religious individuals using cognitive-behavioral therapy. Cognitive and Behavioral Practice, 17(4), 382–392. DOI 10.1016/j.cbpra.2009.07.003.
• İnözü, M., Kahya, Y., & Yorulmaz, O. (2020). Neuroticism and religiosity: The role of obsessive beliefs, thought-control strategies and guilt in scrupulosity and obsessive-compulsive symptoms among Muslim undergraduates. Journal of Religion and Health, 59(3), 1144–1160. DOI 10.1007/s10943-018-0603-5.
• Jung, J. H. (2020). Belief in supernatural evil and mental health: Do secure attachment to God and gender matter? Journal for the Scientific Study of Religion, 59(1), 141–160. DOI 10.1111/jssr.12645.
• Kamble, S. V., Watson, P. J., Marigoudar, S., & Chen, Z. (2014). Varieties of openness and religious commitment in India. Archive for the Psychology of Religion, 36(2), 172–198. DOI 10.1163/15736121-12341283.
• Keller, B., Bullik, R., & Streib, H. (2024). Deconversion from high-tension religious groups. In The Routledge International Handbook of Changes in Human Perceptions and Behaviors (pp. 489–507). Routledge. DOI 10.4324/9781003316602-35.
• Keller, Y., David, Y., & Trachtenberg, E. (2026). On a new path: Social support, social media engagement, and well-being after religious disaffiliation. New Media & Society, 28(2), 739–764. DOI 10.1177/14614448241302312.
• Keng, S.-L., Smoski, M. J., & Robins, C. J. (2011). Effects of mindfulness on psychological health: A review of empirical studies. Clinical Psychology Review, 31(6), 1041–1056. DOI 10.1016/j.cpr.2011.04.006.
• Koenig, H. G., King, D. E., & Carson, V. B. (2012). Handbook of Religion and Health (2nd ed.). Oxford University Press.
• Levin, M. E., Hildebrandt, M. J., Lillis, J., & Hayes, S. C. (2012). The impact of treatment components suggested by the psychological flexibility model: A meta-analysis of laboratory-based component studies. Behavior Therapy, 43(4), 741–756. DOI 10.1016/j.beth.2012.05.003.
• Lewis, J. R., & Bromley, D. G. (1987). The cult withdrawal syndrome: A case of misattribution of cause? Journal for the Scientific Study of Religion, 26(4), 508–522. DOI 10.2307/1387101.
• Lindahl, J. R., Fisher, N. E., Cooper, D. J., Rosen, R. K., & Britton, W. B. (2017). The varieties of contemplative experience: A mixed-methods study of meditation-related challenges in Western Buddhists. PLOS ONE, 12(5), e0176239. DOI 10.1371/journal.pone.0176239.
• McLaughlin, A. T., Van Tongeren, D. R., Teahan, K., Davis, D. E., Rice, K. G., & DeWall, C. N. (2022). Who are the religious “dones?”: A cross-cultural latent profile analysis of formerly religious individuals. Psychology of Religion and Spirituality, 14(4), 512–524. DOI 10.1037/rel0000376.
• Murphy, J. (2026). Understanding deconversion: Exploring why and how people leave a high-control religion and develop alternative worldviews. Secularism and Nonreligion, 15(1), Article 8. DOI 10.5334/snr.253.
• Pargament, K. I., Feuille, M., & Burdzy, D. (2011). The Brief RCOPE: Current psychometric status of a short measure of religious coping. Religions, 2(1), 51–76. DOI 10.3390/rel2010051.
• Rollwage, M., Dolan, R. J., & Fleming, S. M. (2018). Metacognitive failure as a feature of those holding radical beliefs. Current Biology, 28(24), 4014–4021.e8. DOI 10.1016/j.cub.2018.10.053.
• Rousselet, M., Duretete, O., Hardouin, J. B., & Grall-Bronnec, M. (2017). Cult membership: What factors contribute to joining or leaving? Psychiatry Research, 257, 27–33. DOI 10.1016/j.psychres.2017.07.018.
• Saroglou, V. (2002). Religion and the five factors of personality: A meta-analytic review. Personality and Individual Differences, 32(1), 15–25. DOI 10.1016/S0191-8869(00)00233-6.
• Scheitle, C. P., & Adamczyk, A. (2010). High-cost religion, religious switching, and health. Journal of Health and Social Behavior, 51(3), 325–342. DOI 10.1177/0022146510378236.
• Sedlmeier, P., Eberth, J., Schwarz, M., Zimmermann, D., Haarig, F., Jaeger, S., & Kunze, S. (2012). The psychological effects of meditation: A meta-analysis. Psychological Bulletin, 138(6), 1139–1171. DOI 10.1037/a0028168.
• Shariff, A. F., & Aknin, L. B. (2014). The emotional toll of hell: Cross-national and experimental evidence for the negative well-being effects of hell beliefs. PLOS ONE, 9(1), e85251. DOI 10.1371/journal.pone.0085251.
• Siev, J., Baer, L., & Minichiello, W. E. (2011). Obsessive-compulsive disorder with predominantly scrupulous symptoms: Clinical and religious characteristics. Journal of Clinical Psychology, 67(12), 1188–1196. DOI 10.1002/jclp.20843.
• Slade, D. M., Smell, A., Wilson, E., & Drumsta, R. (2023). Percentage of U.S. adults suffering from religious trauma: A sociological study. Socio-Historical Examination of Religion and Ministry, 5(1), 1–28. DOI 10.33929/sherm.2023.vol5.no1.01.
• Sloan, R. P., & Bagiella, E. (2002). Claims about religious involvement and health outcomes. Annals of Behavioral Medicine, 24(1), 14–21. DOI 10.1207/S15324796ABM2401_03.
• Stauner, N., Exline, J. J., Pargament, K. I., Wilt, J. A., & Grubbs, J. B. (2019). Stressful life events and religiousness predict struggles about religion and spirituality. Psychology of Religion and Spirituality, 11(3), 291–296. DOI 10.1037/rel0000189.
• Streib, H., Hood, R. W., Jr., & Klein, C. (2010). The Religious Schema Scale: Construction and initial validation of a quantitative measure for religious styles. International Journal for the Psychology of Religion, 20(3), 151–172. DOI 10.1080/10508619.2010.481223.
• Streib, H., Hood, R. W., Jr., Keller, B., Csöff, R.-M., & Silver, C. F. (2009). Deconversion: Qualitative and Quantitative Results from Cross-Cultural Research in Germany and the United States of America. Vandenhoeck & Ruprecht. DOI 10.13109/9783666604393.
• Tek, C., & Uluğ, B. (2001). Religiosity and religious obsessions in obsessive-compulsive disorder. Psychiatry Research, 104(2), 99–108. DOI 10.1016/S0165-1781(01)00310-9.
• Tekke, M., Watson, P. J., Hisham İsmail, N. A., & Chen, Z. (2015). Muslim religious openness and ilm. Archive for the Psychology of Religion, 37(3), 295–320. DOI 10.1163/15736121-12341313.
• Van Tongeren, D. R., Davis, D. E., Sibley, C. G., & Bulbulia, J. A. (2026). Life loses some meaning after leaving religion. Psychology of Religion and Spirituality, 18(3), 313–322. DOI 10.1037/rel0000590.
• Van Tongeren, D. R., DeWall, C. N., Chen, Z., Sibley, C. G., & Bulbulia, J. (2021). Religious residue: Cross-cultural evidence that religious psychology and behavior persist following deidentification. Journal of Personality and Social Psychology, 120(2), 484–503. DOI 10.1037/pspp0000288.
• Van Tongeren, D. R., DeWall, C. N., Hardy, S. A., & Schwadel, P. (2021). Religious identity and morality: Evidence for religious residue and decay in moral foundations. Personality and Social Psychology Bulletin, 47(11), 1550–1564. DOI 10.1177/0146167220970814.
• Van Tongeren, D. R., Hardy, S. A., Taylor, E. M., & Schwadel, P. (2026). Religious dones become more politically liberal after leaving religion. Journal of Personality, 94(3), 458–471. DOI 10.1111/jopy.70014.
• VanderWeele, T. J. (2017). Religious communities and human flourishing. Current Directions in Psychological Science, 26(5), 476–481. DOI 10.1177/0963721417721526.
• VanderWeele, T. J., Jackson, J. W., & Li, S. (2016). Causal inference and longitudinal data: A case study of religion and mental health. Social Psychiatry and Psychiatric Epidemiology, 51(11), 1457–1466. DOI 10.1007/s00127-016-1281-9.
• Williamson, W. P., Hood, R. W., Jr., Ahmad, A., Sadiq, M., & Hill, P. C. (2010). The Intratextual Fundamentalism Scale: Cross-cultural application, validity evidence, and relationship with religious orientation and the Big 5 factor markers. Mental Health, Religion & Culture, 13(7–8), 721–747. DOI 10.1080/13674670802643047.
• Winell, M. (1993). Leaving the Fold: A Guide for Former Fundamentalists and Others Leaving Their Religion. New Harbinger Publications. (Reprinted 2006, Apocryphile Press.)
• Winell, M. (2011). Religious Trauma Syndrome (three-article series). CBT Today, 39(2–4).
• Zaeske, L. M., Dye, A. R., Spadoni, S., Strothkamp, R., Kane, M. L., Ridgway, K., Dugan, A. J., Patterson, T. P., McEathron, S. R., & Cole, B. P. (2024). Addressing harm from adverse religious/spiritual experiences in psychotherapy: A scoping review. Practice Innovations, 9(1), 1–18. DOI 10.1037/pri0000237.
• Zarzycka, B., & Rydz, E. (2014). Explaining the relationship between post-critical beliefs and sense of coherence in Polish young, middle, and late adults. Journal of Religion and Health, 53(3), 834–848. DOI 10.1007/s10943-013-9680-7.
• Zhong, W., Cristofori, I., Bulbulia, J., Krueger, F., & Grafman, J. (2017). Biological and cognitive underpinnings of religious fundamentalism. Neuropsychologia, 100, 18–25. DOI 10.1016/j.neuropsychologia.2017.04.009.
• Zmigrod, L., Rentfrow, P. J., Zmigrod, S., & Robbins, T. W. (2019). Cognitive flexibility and religious disbelief. Psychological Research, 83(8), 1749–1759. DOI 10.1007/s00426-018-1034-3.

The series

• Grünig, P., & Claude Fable (Anthropic). (2026). Stress-Sharing as Cognitive Glue in Ideological Collectives: From Biological Morphogenesis to Cultural Cohesion. Paper 1.2 of this series (rev. 2, August 2026). Project archive. [cited as Paper 1.2]
• Grünig, P., & Claude Fable (Anthropic). (2026). Neurological Markers of Deep Belief Entrenchment: A Narrative Review of Identity-Fused Conviction Across Domains. Paper 1.3 of this series (rev. 2, August 2026). Project archive. [cited as Paper 1.3]