Do Religious Worldviews Generate Normativity? — Decomposing World-Models, Sanctions, Goodness, and Authority
I. Decomposing the religious package
1. “Is religion true?” is too coarse a question
Religion is not one proposition. At the outer level, this page separates descriptive world-model claims D, hypothetical or instrumental implications from existing goals H, normative claims N, expressive and symbolic practice E, and institutional practice I. Only when N itself needs analysis do we further distinguish value content C, bridge principles B connecting facts or value to normative relations, and normative structure R, including reasons, oughts, obligations, permissions, and fittingness. In short, Religion = D/H/N/E/I, while N = C/B/R is an internal decomposition. A claim that resurrection occurred, advice to repent if one wants to avoid hell, a claim that God ought to be worshipped, and a preference to celebrate Easter communally are all “religious,” but they require different standards of assessment.
Gould’s NOMA is an important predecessor. Within Barbour’s familiar conflict / independence / dialogue / integration taxonomy, NOMA belongs strongly on the independence side. Gould assigns empirical facts and theories about the natural world to science while placing meaning and moral value in religion. This preserves at least one valuable intuition: D ⇏ N. A description of the physical world does not by itself generate normativity.
But drawing the boundary at the level of the institutions called “science” and “religion” causes trouble. Religions make ordinary factual claims about resurrection, miracles, creation, rebirth, and causal effects of prayer, while normative philosophy is not a religious monopoly. Dawkins is right that a god who causally acts in the world competes, in that respect, with ordinary world-models. Yet Wittgenstein’s criticism of Frazer also matters: funeral rites, mourning, festival, and symbolic expression should not all be redescribed as failed empirical hypotheses. In this page’s taxonomy, much of what Wittgenstein is protecting in such cases belongs primarily to E—expressive, symbolic, and preference-laden forms of life—and its expressive significance does not by itself establish the truth of D- or N-claims. Dennett’s naturalization of religion further makes the production of religious belief a causal object of study, but Cause(Belief) ⇒ ¬Truth(Belief) is invalid.
Stenmark’s multidimensional approach and Brooke’s complexity principle both push against assigning one global relation to the two giant categories. Brooke’s historical point is that there is no single “the relationship” between science and religion; conflict, reinforcement, institutional borrowing, and conceptual overlap vary locally. This page pushes that localization one step further, from institution to claim type. The aim is therefore not to reject NOMA wholesale, but to replace institution-level NOMA with claim-level non-conflation.
2. Provisional position on religious world-models
This page is primarily about normativity, so it does not attempt a final adjudication between theism and naturalism. Still, a provisional epistemic stance is necessary. On present evidence, detailed supernatural models proposed by existing religions do not appear to outperform naturalistic explanations as best explanations. Modern natural explanations now cover many domains once assigned substantial explanatory work to supernatural agency: adaptation, disease, weather, celestial motion, and large parts of psychology. This does not deductively prove that no god exists, but it does reduce the explanatory necessity of many concrete god-hypotheses.
Revelation, miracle reports, religious experience, and transmission should also not be treated as strong evidence merely because the corresponding beliefs exist. Mutually incompatible testimony, culture-sensitive interpretation, cognitive agency-detection, institutional authority, and transmission dynamics can explain substantial parts of religious belief formation without positing the represented objects. Yet genealogy is primarily an undercutter, not a deductive refutation: explaining why a belief arose does not logically entail that its content is false.
We should therefore distinguish local from global atheism. Assigning very low epistemic weight to a specific revealed religion or specific personal-god model is not equivalent to stipulating ∀H_supernatural, P(H)=0. This project can strongly prefer current naturalistic explanations while leaving unconceived supernatural or religious hypotheses open in principle. That is simply the world-model analogue of open-world moral uncertainty and open hypothesis spaces.
3. Early religion can concern effective powers rather than good beings
Religious history need not be reconstructed as morality-first submission to a perfectly good deity. Rainmaking, fertility rites, healing, safe-voyage rites, appeasement, and ancestor rites often fit D + G → R_h(A). If a god controls rain and one wants rain, prayer may be instrumentally rational. If an ancestor spirit can cause calamity and one wants to avoid calamity, propitiation may be instrumentally rational. What is required here is power plus a causal model, not moral perfection. Here H names not Rh itself but the layer of practical implication in which D plus an existing goal G generates Rh.
It is useful to distinguish at least four overlapping types rather than tell a single evolutionary story. First, a power/efficacy model: this being can cause outcomes. Second, a sanction model: conduct brings reward or punishment in this life or another. Third, a normative model: divine command or cosmic order itself generates obligation. Fourth, a value-identity model: God does not merely command the good but is the Good, or the standard of goodness. These types can coexist, and the first two can remain important in popular religion even after sophisticated theology develops the latter two.
Japanese myth and Shinto offer a useful historical contrast to perfect-being theology. Aramitama/nigimitama, pacification of dangerous spirits, and the transformation of vengeful spirits into protective kami illustrate a structure in which one sacred power may appear as calamity and benefit across contexts. Fixed classification into “good god” and “evil god” is less central than the current manifestation of power and the relation one establishes with it. Thus Divinity ≠ Power ≠ Benevolence ≠ MoralGoodness is not merely a conceptual possibility; it has religious-historical exemplars.
4. Sanction is not normativity
Reward and punishment are ancient religious motivators. But “comply or suffer” has the same formal structure whether issued by a god, tyrant, or criminal organization. If a deity can certainly send dissenters to hell, that can generate an enormously strong Rh. It still does not establish R_h(A) ⇒ R_n(A).
There is a long internal history of trying to separate morality from sanction. Plato’s Republic asks whether justice remains choiceworthy when reputation and reward are stripped away. Christian discussions distinguish servile fear of punishment from more filial relations to God and goodness. Shaftesbury worries that making reward and punishment the basis of virtue corrupts moral motivation. Kant, whatever one thinks of his final grounding, sharply separates action for happiness or reward from action out of respect for moral law. The common problem is that sanctions do not by themselves supply the bridge from prudence to genuine normativity.
J. S. Mill’s famous refusal to call an immoral ruler good even under threat of damnation makes the distinction vivid. Structurally, the case can be represented as recognizing a very strong R_h(obey) while taking a competing R_n(refuse) to be decisive. That is not ordinary akrasia but deliberate sanction rejection. Don Giovanni and Faustus provide literary relatives, though self-deception and obstinacy make them less philosophically clean than Mill’s case.
5. Huck Finn — internalized conscience still does not guarantee truth
The famous “All right, then, I’ll go to hell” scene in Adventures of Huckleberry Finn is not a pure calculation about hell avoidance. Huck sincerely judges helping Jim escape to be wrong, thinks he is helping steal Miss Watson’s “property,” and experiences guilt through the social and religious conscience he has internalized. Yet his concrete relation and sympathy toward Jim move him to act against that explicit judgment.
We should separate the agent’s actual motivational dynamics D_A, the represented normative model V̂_A=(Ĉ,B̂,R̂), and, if it exists, the true normative structure V*. Huck has something like R̂_H(return Jim) and acts against it, so from his own all-things-considered judgment the case resembles akrasia. From a later moral standpoint, the very “weakness” can be redescribed as inverse akrasia because it produces the objectively better action.
The general lesson matters more than the literary classification. Even when external sanctions are fully internalized as a seemingly unconditional conscience, the whole package social learning → Ĉ + B̂ + R̂ may still be false. Hence PhenomenologyOfObligation ≠ TruthOfObligation. This pressure is not uniquely religious: it also bears on Kantian appeals to the “fact of reason,” Wang Yangming-style liangzhi, and any intuitionist view that treats felt categorical bindingness as strong evidence of objective validity.
II. Where does theology locate normativity?
6. Transforming goal G becomes a value claim when the transformation is demanded
Religions often do more than provide means to current goals. They say: abandon this desire, become a lover of God, renounce attachment, generate bodhicitta, become a different kind of person. Three layers should be separated.
If a practice causally produces G0→G1, that is a psychological or social D-claim. If one’s present G0 gives an instrumental reason to acquire G1, that is H. But if the religion claims that one ought to become an agent with G1 rather than G0, or that abandoning current desire is itself good, then a genuine N-claim has appeared and its internal C/B/R structure should be examined. Preference transformation therefore need not be treated as a new religious primitive; its causal production and its normative justification should be separated.
7. How does theology get from C to B?
Once N is opened up internally, even if “God is good,” “mercy is good,” or “holiness is valuable” supplies C, reasons and obligations do not automatically follow. A large part of theological ethics can be reread as competition over where to place B.
7.1 Divine command
Divine-command theories propose something like Command_G(A) → Obligation(A). In strong voluntarist versions, the fact of divine command itself constitutes obligation. The Euthyphro-style pressure is therefore not only whether goodness is independent of God, but also: why can a commanding act generate normative force? Appealing to omnipotence establishes sanction capacity, not Rn.
7.2 Divine nature
Adams-style approaches connect goodness to God’s nature or resemblance to God while connecting obligation to divine command. This avoids the crude worry that absolutely anything becomes good merely because it is commanded. Yet it creates a new bridge: ResemblesGod(x) → Good(x). Why divine nature is the standard of goodness must be supported as a substantive claim rather than secured by stipulation.
7.3 Natural law and practical reason
Thomistic natural law locates normativity in a broader order of creaturely nature, ends, basic goods, and practical reason. God may be the source of the order without needing to generate each individual reason by command. But then Nature/Fulfillment → Reason or Good → ToBePursued becomes an independent metaethical bridge that must itself be understood.
7.4 Creation, ownership, and authority
Another route says that creatures owe obedience because God created them. But CreatedBy(G,A) → Authority_G(A) is not self-evident. Human creators of artificial agents do not automatically acquire unlimited ownership, sovereignty, or worship-rights over them. A causal relation of creation requires an additional principle before it becomes normative jurisdiction.
7.5 The perfect knower model
A more modest model says that God does not create reasons but perfectly knows them. God then functions as an ultimate moral epistemologist or witness, and R_n(A) is logically prior to divine command. This reduces the bridge problem of voluntarism but moves the issue into epistemology: does God really know the norm, and do we have independent reason to trust the claimed divine testimony?
This page leaves all of these models open as candidates. What it rejects is compressing existence, omnipotence, creation, knowledge, goodness, and authority into the single unexplained phrase “because God.”
8. Does God create reasons, or communicate them?
The same sentence, “God says do A,” can express at least four structures. First, epistemic communication: God knows an independently existing R_n(A) and reports it. Second, evidence: the divine testimony raises our P(R_n(A)). Third, reason-generation: the command itself constitutes R_n(A). Fourth, sanction: reward or punishment creates R_h(A).
Consider one command: “help your neighbor.” It may mean (1) God reports an independent moral truth, (2) testimony from a reliable omniscient being should update our credence, (3) an otherwise optional act becomes obligatory because it is commanded, or (4) compliance is prudent because punishment follows refusal. These require different premises. (1) and (2) require epistemic reliability; (3) requires an authority-generating bridge; (4) requires only causal power.
Mark C. Murphy’s work on divine authority sharpens the distinction. He separates an authority thesis, on which divine command itself constitutes a decisive reason, from a weaker compliance thesis, on which decisive reason exists whenever God commands, perhaps for some independent reason. The authority thesis implies compliance, but compliance does not imply authority. Murphy also argues that the familiar divine perfections—omniscience, goodness, omnipotence—do not automatically establish divine authority. This is a major precedent for the project’s principle Power/Knowledge/Goodness ⇏ Authority.
9. How is divine goodness recognized?
Before using perfect goodness as a premise in debates about evil or hiddenness, we should ask how that goodness was identified. At least six routes recur.
- Revelation: scripture or prophecy reports divine goodness. The challenge is authenticity and circularity if divine goodness is also used to establish revelatory reliability.
- Perfect-being theology: God is the maximally perfect being and goodness is a great-making property. The remaining question is why an actual creator or ultimate being instantiates that concept.
- Metaphysics of being and goodness: being, perfection, and goodness are deeply connected, so supreme being is supreme good. The challenge is the bridge from ontological to moral or normative goodness.
- Creation and providence: order, beauty, life, or benefit are evidence for a good creator. The challenge is selective use of observations and the fact that skeptical moves used to defuse evil may also reduce the evidential force of apparent good.
- Independent moral reason: we identify goodness independently and then judge God by it. This has strong discriminatory power but makes goodness logically prior to God.
- Religious experience: the subject directly experiences overwhelming goodness, holiness, or love. This may be evidence, but
SeemsGood(G) ⇏ Good(G)remains a gap between phenomenology and objective validity.
If “God” is defined as omniscient, omnipotent, and perfectly good, then God → PerfectGood is analytically true. The substantive problem simply shifts: what establishes that the actual creator, revealer, or object of religious experience is the referent of that God-concept? We should therefore keep Creator ≠ PerfectGoodGod explicit.
III. How far can a perfectly good god be identified?
10. Apophatic theology — can unknown goodness save theology?
Pseudo-Dionysius, Maimonides, and other apophatic or analogical traditions warn against applying ordinary human predicates straightforwardly to the divine essence. In Maimonides, multiplying positive attributes can compromise divine simplicity; negative predication and descriptions of divine effects are therefore epistemically safer. This humility partly resembles the project’s insistence that V̂_human ≠ V*.
The project can therefore leave open the possibility that what looks cruel under current human value is right under a radically different true value. If present human goodness is not automatically the cosmic standard, the problem of evil may fail as a deductive refutation of every possible perfectly good deity.
But distance has a cost. The further Good_G is detached from our concept of goodness, the harder it becomes to infer Good(G) → Trust(G), obedience, or worship. More generally: apophatic distance ↑ ⇒ falsification pressure ↓, but epistemic discrimination / action-guidance ↓. Apophaticism may preserve the possibility that God is perfectly good while giving us less ability to identify which being, if any, is perfectly good.
11. Extending the Evil-God Challenge to alien value
Stephen Law’s Evil-God Challenge asks why an omniscient, omnipotent, perfectly evil god should receive dramatically less credence than an omniscient, omnipotent, perfectly good god if symmetrical theodicies can be built. If evil can be explained away as serving unknown greater good, why cannot apparent good be explained away as serving unknown greater evil?
A first family of responses is skeptical-theist. Perry Hendricks argues, in effect, that if our access to divine reasons is sufficiently limited, apparent good need not decisively refute an evil god either. This is a response to Law, but from the project’s perspective its implication is more important than its dialectical victory: symmetric skepticism makes both good and evil gods harder to exclude from observation and therefore worsens the identification problem.
A second family appeals to metaphysical asymmetry. Evil may be a privation of good, goodness may be closer to being or perfection, or a perfectly evil god may be metaphysically more complex or less natural than a perfectly good god. If true, such theses could give the good-god hypothesis a prior advantage. But Being → Good and Privation(Evil) are substantive metaphysical and axiological theses, not free background assumptions.
A third family invokes rational-agency asymmetry. Perhaps perfect rationality necessarily pursues what is apprehended as good, making a perfectly rational yet perfectly evil being unstable. But this itself assumes something like RationalAgency → GoodMotivation. If orthogonality between intelligence and terminal value remains a live possibility, the connection requires independent defense.
A fourth family offers a pragmatic response: belief in a good god may have better psychological, social, or moral consequences than belief in an evil god. This project keeps Useful(Belief H) ⇏ True(H). Practical usefulness may matter for policy or coordination but should not directly increase the epistemic weight of a world-model.
The project also broadens Law’s binary comparison. Let G+ be a god recognizably good by current human standards, G− recognizably evil, G* a god pursuing unknown objective value, G¬* a god pursuing unknown anti-value, and G0 a value-indifferent creator. A genuine Good/Evil asymmetry is not ruled out. But the asymmetry itself is a candidate theory requiring evidence; it should not be imported into the concept of “god” for free.
12. Evil and divine hiddenness become identification problems
The traditional problem of evil asks why an omniscient, omnipotent, perfectly good being permits vast suffering. Schellenberg-style divine hiddenness asks whether a perfectly loving personal god would permit so many non-resistant people sincerely to lack belief and relationship. Both arguments infer predictions from perfect goodness or perfect love: concern for welfare, relationship-seeking, self-disclosure, and so on.
This project treats those inferential bridges as hypotheses. If true value may be radically alien to current human value, PerfectGood → WelfareMaximization and PerfectLove → RelationshipSeeking may not be necessary. Epistemic independence, non-contact, a long historical process, or some unconceived value could in principle matter more.
The same applies to Draper-style comparisons. Even if observed suffering is more expected under a creator indifferent to creaturely welfare than under traditional theism, WelfareIndifferent ⇏ ValueIndifferent. A deity pursuing alien objective value could be indifferent to human welfare while still being objectively good under that value structure.
Yet this does not give theology a free escape. The more observations are made compatible with divine goodness by saying that true goodness outruns our understanding, the less clear it becomes what observations discriminate G+, G−, G*, and G0. The central inversion is therefore: strange goodness may preserve the logical possibility of a perfectly good god without automatically preserving our ability to identify one. The question becomes not only “why would a good god permit this?” but “why should this being be identified as good rather than evil, indifferent, or alien-valued?”
IV. Worship, cultural practice, and non-theistic traditions
13. Worship, sacredness, and fittingness should also be decomposed
A psychological desire to worship and the claim that worship is objectively fitting are different. The former may be a locally produced preference shaped by evolution, culture, and socialization. The latter belongs to N and is treated here as a fittingness relation within normative structure R, for example R_fittingness(Worship,G). This does not commit the project to reducing fittingness ontologically to reasons; it only places reasons, oughts, obligations, permissions, and fittingness within the same analytic family of normative relations. The foundational question returns to Reasons First and normative structure.
Even within worship we should distinguish admiration, gratitude, reverence, submission, allegiance, and obedience. Gratitude may fit a creator without implying unlimited allegiance; awe may fit overwhelming value without implying obedience; legitimate authority may generate reasons for compliance without making worship appropriate. Thus Existence(G) ⇏ Gratitude ⇏ Allegiance ⇏ Worship.
“Sacred place” likewise decomposes into D (“a deity appeared here”), E/I (“this community marks the place as special”), and a fittingness relation inside N/R (“awe, silence, or ritual is objectively fitting here”). Describing fittingness as part of R does not eliminate the B-question of why this response fits this object.
14. False religious fictions may still be useful
Even if a religious D-claim is false, ritual, myth, commemoration, ancestor imagery, or shared symbolism may support trust, long-term commitment, coordination, and mourning. False(D) ∧ Useful(E/I) is coherent. This does not exempt the factual claim from truth assessment; it merely separates factual truth from cultural function.
But UsefulBelief and UsefulFiction should also be separated. A community may knowingly share a fiction as symbol, or it may enforce falsehood as fact because widespread belief improves coordination. For an inquiry community the latter risks damaging epistemic trust and long-run error correction, so transparent fiction is much easier to reconcile with the project’s aims.
15. Non-theistic traditions reveal different connections
15.1 Buddhism
Large parts of early Buddhism can be reconstructed as D-claims about suffering, attachment, karma, rebirth, and liberation plus H for agents who seek liberation: Dukkha-model + Goal(liberation) → practice. But once the tradition says that suffering ought to be removed as such, or that one ought to transform into a bodhisattva-like agent, an additional N-claim appears and its internal C/B/R structure must be examined. Mahayana universalizes concern beyond personal liberation, yet the ultimate bridge from suffering to negative value remains a separate question.
15.2 Confucianism
Confucian traditions offer several non-divine-command routes to normative cognition. Mencius emphasizes moral sprouts such as compassion; Xunzi emphasizes ritual, learning, and institutional transformation of conflict-producing desires; Zhu Xi connects li to cosmic order and human nature; Wang Yangming emphasizes liangzhi as direct moral knowing. From this project’s perspective they generate distinct bridge questions: MoralFeeling → Truth?, SocialOrder → Normativity?, CosmicLi → Ought?, and PhenomenalCertainty → ObjectiveValidity?.
15.3 Zhuangzi
Zhuangzi is especially interesting as a possible precursor to anti-lock-in toward one’s current evaluative representation. Wuwei, ziran, and free-and-easy wandering need not be read as obedience to one uniquely authoritative cosmic law; they can be read as loosening fixation on rigid right/wrong and useful/useless schemes so that an agent remains responsive to changing contexts and perspectives. The divergence from this project is that the project still seeks evidence accumulation, hypothesis comparison, and possible convergence toward V*, whereas Zhuangzian skepticism may question the desire for one final uniquely correct Dao itself.
15.4 Shinto and Japanese myth
Shinto examples serve as a historical control on the tendency of perfect-being theology to fuse divinity, power, goodness, and authority. In models where dangerous gods are pacified and protective relations are cultivated, practical reasons can initially be causal and relational rather than grounded in moral perfection. A being does not cease to be divine merely because it is harmful.
The same framework applies to low-credence models such as Ragnarök. Assigning the world-model extremely low present epistemic weight is compatible with refusing to hard-code literal zero. If evidence changed, hypothetical reasons to prepare for battle could change; the further claim that preparing for battle is intrinsically good would remain a distinct N-claim.
16. Provisional conclusion — keep religious possibilities open, inspect each arrow
The project does not exclude divine beings, revelation, divine command, sacredness, ancestor rites, or unknown religious value from the hypothesis space. Because true value may differ radically from current human value, it should also be cautious about using the traditional problem of evil to rule out every conceivable perfectly good divine being.
But this openness grants no epistemic privilege to existing theology. Divine existence, authenticity of revelation, goodness, epistemic authority, practical authority, reason-generating power, obligation, and worship-worthiness each require separate support. If a genuinely irreducible normative structure remains after decomposition, it can be added to the open hypothesis space as a new candidate. Before that point, words such as “God,” “sacredness,” and “worship” should not introduce unexplained normative primitives for free.
H: D + G → Rh can then be contrasted with something like N: D/C + B → Rn: hypothetical implications and normative claims have different routes of generation.
Existence(G) ⇏ Good(G); Knowledge(G) ⇏ Authority(G); Power(G) ⇏ Rn; Creator(G) ⇏ Ownership/Obedience; Good(G) ⇏ WorshipWorthiness(G).
Non-exclusion from the hypothesis space does not grant practical privilege or veto power. Every inferential arrow should be assessed under the same epistemic standards as other value candidates.