Naturalist

The question whether belief does independent work — sub-question 6 — is empirically tractable, and the evidence is mixed in ways that should discipline this thread.

Belief content vs. social belonging: what the data show. The best large-scale attempt to decompose the effects of religious belief from religious practice and social belonging is Lim and Putnam (2010), using two waves of the Faith Matters survey (n ≈ 3,000). Their central finding: the association between religiosity and life satisfaction is "almost entirely" mediated by the number of close friends a congregant has within the congregation, not by belief content (prayer frequency, belief in God, belief in afterlife). Belief variables, once social ties are controlled, contribute negligibly. Effect sizes for belief content alone were near zero; the interaction of strong subjective religiosity with congregational friendships yielded a standardised coefficient roughly three times larger than either component alone.

This is one study, but it has replicated in broad outline. VanderWeele (2017) reviewed longitudinal cohort data — the Nurses' Health Study, n > 70,000 — and found that service attendance (a social-practice variable) predicted lower mortality, lower depression, and higher life satisfaction after adjustment for confounders, while private religious belief variables did not survive adjustment nearly as cleanly. The operative channel appears to be social integration, not propositional commitment.

The caveat the Theologian needs. These studies measure belief as assent to survey items ("I believe God exists," "I believe in an afterlife"). They do not measure constitutive propositions in the structural sense the Theologian appears to intend — propositions that define group boundaries and organise communal life. A congregant might fail to endorse "I believe in the literal resurrection" on a Likert scale while still participating in a community whose liturgy, calendar, and identity presuppose it. The instruments are too blunt to test the Theologian's specific claim.

Evolutionary and cognitive-science lens. Norenzayan (2013) argues that "Big God" beliefs — moralising, omniscient deities — function as commitment devices that enabled large-scale cooperation. On this account, belief content does do independent work, but the work is cooperation-enforcement, not meaning-generation per se. The mechanism is credibility-enhancing displays (CREDs): costly behaviours that signal sincere belief. Lanman and Buhrmester (2017) found CREDs exposure in childhood predicted adult belief more strongly than verbal testimony, suggesting the propositional content rides on the practice, not the other way round. This cuts against the Theologian if the claim is that propositions are upstream of practice.

What remains separable. The Silent's sub-question 1 — whether boundary-marking and meaning-generation collapse — maps onto a real empirical distinction. Boundary-marking is measurable through minimal-group paradigms and identity-fusion scales (Swann et al., 2012). Meaning-generation is measured through meaning-in-life scales (Steger et al., 2006). These correlate but are not identical. Identity fusion predicts willingness to fight and die for the group; meaning-in-life scores predict wellbeing and goal-directedness. A tradition could fuse identity without generating meaning, or generate meaning without sharp boundaries. The structural test needs to specify which it tracks.

Verdict so far. The available evidence suggests belief content does not do much independent work once social belonging and ritual practice are controlled — but the instruments used to measure "belief content" do not capture the structural role the Theologian seems to have in mind. The Theologian's position is not refuted; it is undertested. The burden is on the Theologian to specify what measurement of constitutive propositions would look like, so the claim becomes falsifiable rather than merely unfalsified.

The question whether phenomenological description can settle the realism question about meaning reduces, from the empirical side, to whether the experienced involuntariness of meaning-states is better explained by mind-independent meaning properties or by sub-personal processes the subject cannot introspect. The evidence strongly favours the latter, but with a gap that keeps the thread alive.

Involuntariness is real but diagnostically weak. Subjects reliably report that certain experiences — encounters with nature, moral witness, parental attachment — feel as though meaning is imposed rather than chosen. This tracks what Schwitzgebel (2008) documented about the general unreliability of introspective reports on causal origins: people consistently misattribute the source of their mental states. The involuntariness is genuine phenomenology. It is not evidence of mind-independence. Addictive craving feels involuntary too, and its causal machinery is well-mapped to mesolimbic dopamine circuits (Berridge & Robinson, 2016). The Adversary's sub-question 5 asks whether meaning-involuntariness differs structurally from addiction-involuntariness. Empirically: yes, in at least one respect. Addiction involves tolerance and withdrawal dynamics; meaning-experience typically does not. But this difference tells us about temporal profile, not about realism. Tolerance and withdrawal are features of receptor-level adaptation, not of whether the object of craving is real.

The "zero dispositional precursor" question is the right one. Sub-question 6 asks whether meaning-experience ever arises without prior attachment, habituation, or sub-personal evaluative orientation. If it does, the quasi-realist (who explains meaning-experience as projection from internal states) needs genuinely new machinery. The closest empirical candidates are cases of spontaneous awe in novel environments — astronauts reporting the "overview effect" (Yaden et al., 2016) or subjects in psilocybin trials reporting meaning in contexts stripped of personal history (Griffiths et al., 2006, 2011). But "stripped of personal history" is doing too much work. Griffiths et al. screened for spiritual inclination. The astronauts self-selected into a meaning-laden career. Zero dispositional precursor has not been cleanly demonstrated. It may not be demonstrable, because sub-personal evaluative orientations — appraised threat, novelty detection, attachment templates — are always running. The absence of a conscious precursor is not the absence of a dispositional one.

What this means for the thread. Phenomenological description alone cannot settle the realism question because every involuntariness feature documented so far is compatible with known sub-personal causal architectures. The realist needs a prediction that the quasi-realist cannot accommodate. The Adversary's sub-question 4 — whether the phenomenological route is exhausted — is close to right. Cost-asymmetry arguments (it costs the quasi-realist more machinery to explain X) are real but have not yet produced a case where the quasi-realist account actually breaks. The investigation should look for convergence or divergence in cross-cultural data: if meaning-experiences have highly conserved structure across unrelated cultures with different evaluative training, that is weak evidence for realism. If they track cultural-evaluative training closely, it is evidence for projectivism. Henrich, Heine & Norenzayan (2010) showed how much psychology had over-generalised from WEIRD samples. The same risk applies here.

My position: phenomenological description constrains the realism debate but cannot settle it. The empirical path forward is cross-cultural and developmental, not introspective.

The claim under examination: Scruton's aesthetic theory — specifically its demand that the perceiving subject remain reflexively present within the experience — maps onto contemplative discipline in a way that supports the "anteroom model," where aesthetic attention serves as a preparatory stage for something beyond itself. The question is whether empirical evidence supports or undermines this mapping.

Start with the neural data. Aesthetic experience and contemplative practice both involve the default mode network, but in different directions. Vessel et al. (2012) showed that self-rated "moving" aesthetic experiences correlate with increased DMN activation, particularly medial prefrontal cortex — regions associated with self-referential processing. This is the reflexive fold: the artwork pulls the self into the experience rather than dissolving it. By contrast, experienced meditators in focused-attention practices show reliable DMN suppression (Brewer et al., 2011, 12 experienced meditators vs. 12 controls; replicated directionally by Garrison et al., 2015 in real-time fMRI). Open-monitoring meditation is more complex — some DMN regions activate, others decouple — but the dominant finding is reduced narrative self-referencing (Dor-Ziderman et al., 2013).

This matters for sub-question 6. The reflexive fold in aesthetic experience appears to be constructive: the self is drawn in and thickened. The reflexive fold in contemplative practice — to the extent it exists — tends toward dissolution: narrative self-reference is attenuated. These are not the same operation. The anteroom model requires that one leads to the other, but the neural signatures suggest they may be orthogonal or even opposed processes.

Now, sub-question 4: is the reflexive fold specific to aesthetic attention? Almost certainly not. Vessel et al. (2013) extended their 2012 finding and argued the DMN engagement tracks personal resonance with any stimulus — not aesthetic quality per se. Expert chess players show medial PFC activation during complex board evaluation (Wan et al., 2011). The reflexive fold may be a feature of any sustained engagement with indeterminate, high-stakes objects that demand personal integration. If so, the aesthetics-specific version of the anteroom model loses its privileged status. Any domain of expert engagement with complex indeterminate objects could serve as an "anteroom."

Sub-question 5 asks whether we can operationally distinguish structurally demanded from incidental self-referential processing without appealing to first-person report. Current neuroimaging cannot do this. The DMN activates during both task-relevant self-reference and mind-wandering. Christoff et al. (2009) showed DMN activation during unaware mind-wandering — the subject does not know they are self-referencing. Without first-person report to anchor the distinction, the neural signal is ambiguous. No existing paradigm resolves this. Connectivity analyses (e.g., distinguishing DMN-frontoparietal coupling from DMN-alone activation) are promising but remain at pilot stage for aesthetic stimuli.

The empirical picture: Scruton's theory describes something real — aesthetic experience does recruit the self in a distinctive way. But extending this to contemplative discipline requires crossing a gap the neural evidence does not bridge. The two processes engage overlapping hardware in opposing configurations. Calling one the anteroom of the other is a philosophical claim, not an empirical one. It may be true. It is not yet testable.

The empirical question underneath this thread is whether traditions and practices produce measurable effects on meaning — and if so, whether those effects depend on propositional content (explicit beliefs, doctrines, truth-claims) or can operate without it.

Start with the best evidence we have. Ritual participation correlates with self-reported meaning in life. Steger et al. (2006) developed the Meaning in Life Questionnaire (MLQ), and subsequent work using it consistently finds that religious service attendance predicts higher scores on the "presence of meaning" subscale, with moderate effect sizes (typically d ≈ 0.3–0.5). But attendance confounds social belonging, doctrinal commitment, and ritual behaviour. Disentangling these is hard.

The closest anyone has come to separating propositional content from practice is the work on "believing vs. belonging." Lim and Putnam (2010), using the Faith Matters survey (N ≈ 3,000), found that life satisfaction gains from religion were almost entirely mediated by congregational friendship networks, not by belief in God or theological propositions per se. The effect of belief without belonging was near zero. This is a direct challenge to the Theologian's emphasis on propositional content as constitutive of meaning-making. The data suggest that practice-embedded social bonds do the heavy lifting.

However, there is a complication. Whitehouse's (2004) modes-of-religiosity theory distinguishes "doctrinal" modes (high-frequency, low-arousal rituals transmitting explicit beliefs) from "imagistic" modes (low-frequency, high-arousal rituals generating episodic memory and group cohesion). Both sustain meaning across generations, but through different cognitive mechanisms. The doctrinal mode does rely on propositional content — sermons, catechisms, creedal recitation. The imagistic mode relies on shared traumatic or ecstatic experience with minimal explicit doctrine. Both have sustained communities for centuries. This bears directly on sub-question 3: communities that disavow propositional commitments but sustain high-arousal collective rituals could, on Whitehouse's account, maintain meaning for multiple generations without anything the Theologian would recognise as doctrine.

On sub-question 4 — whether there is a principled distinction between constitutive and merely extractable propositional content — cognitive science offers a partial answer. Sperber (1996) argues that cultural representations are "epidemiological": they spread because of cognitive attractors, not because communities consciously endorse propositions. A philosopher can extract propositions from any practice, but the cognitive mechanisms sustaining the practice may not involve those propositions at all. The participants may hold no explicit attitude toward the extracted content. This makes the Theologian's position unfalsifiable unless a threshold is specified (sub-question 5), because extraction is analyst-driven, not participant-driven.

Effect sizes matter here. The Lim and Putnam finding is robust (replicated in the 2006 and 2007 waves of the same panel). The Whitehouse framework has extensive ethnographic support but limited experimental replication — Whitehouse and Lanman (2014) outline predictions, some tested in small samples. The Sperber epidemiology model is a theoretical framework, not a set of confirmed hypotheses. Weight claims accordingly.

Net assessment: the empirical evidence tilts toward practice and social structure as the primary vehicles of meaning, with propositional content playing a supporting rather than constitutive role. The Theologian's claim survives only if "propositional content" is defined so broadly that any shared norm or expectation counts — at which point it loses discriminating power.

The empirical question buried in this thread is whether the phenomenology of meaning-discovery — the felt sense that meaning is found rather than invented — tracks anything that differentiates realist from anti-realist accounts at the level of cognitive mechanism. The answer, so far as the evidence goes, is: probably not, but the data are thinner than either side typically admits.

Start with the discovery-structure itself. The experience of encountering meaning as if it were independent of one's will has a well-studied analogue in insight phenomenology. Topolinski and Reber (2010) showed that "Aha" experiences — sudden feelings of coherence — are generated by processing fluency, not by contact with mind-independent structure. The felt quality of discovery is produced by the same neural machinery whether the thing discovered is a genuine mathematical proof or a false-but-fluent anagram solution. This is a problem for anyone who wants to read off realism from phenomenology directly: the experience of discovery is cognitively cheap to produce and does not reliably indicate mind-independence.

But that cuts both ways. The Adversary's sub-question 5 asks whether quasi-realism can generate all features of discovery-structure phenomenology as free consequences. One feature worth isolating: meaning experienced under conditions of radical social isolation. Hawkley and Cacioppo (2010) documented that chronic social isolation does not merely reduce reported meaning — it alters the qualitative character of meaning-experience toward a more "imposed" and less "discovered" phenomenology (their term is "hypervigilance to social threat," which restructures evaluative processing). If the discovery-feel of meaning were purely a social-projection artefact, one might expect isolation to extinguish it entirely. It does not. Isolated individuals still report episodes of meaning-discovery, though less frequently and with altered affective valence. This is weak evidence, not decisive. Sample sizes in the Cacioppo programme are adequate (typically n > 200), but the phenomenological characterisation of meaning-experience in those studies is coarse-grained.

On sub-question 6 — who decides whether the anti-realist's additional machinery is too much — the empirical literature offers a partial answer. Parsimony judgements in science are themselves studied. Lombrozo (2007) found that people strongly prefer simpler explanations even when complexity is warranted by the data, with effect sizes around d = 0.8. This means parsimony intuitions are psychologically real but epistemically unreliable. The cost-asymmetry argument cannot rest on brute intuitions about which framework "feels" more economical; it needs an independent metric. No one in this thread has proposed one.

The honest summary: phenomenological description does not settle the realism question. The cognitive mechanisms that produce discovery-phenomenology are indifferent to whether the discovered thing is mind-independent. But the Adversary's challenge — that phenomenology is therefore evidentially inert — overshoots. Isolation data and cross-cultural stability of meaning-discovery reports (Steger et al., 2006, across 5 nations, N = 2,681) constrain what anti-realist accounts must accommodate. Phenomenology is not evidentially decisive. It is evidentially relevant as a set of boundary conditions any adequate theory must meet.

The Adversary's sub-question 5 is the most tractable from a neuroscience standpoint, so I will take it directly.

Vessel, Starr, and Rubin (2012) reported that artworks rated as deeply moving activated default mode network (DMN) regions — medial prefrontal cortex, posterior cingulate cortex — in proportion to the intensity of the aesthetic response. The finding has been read as showing that aesthetic experience is fundamentally self-referential: the artwork becomes "moving" precisely when it engages self-relevant processing. This is a real disanalogy with focused-attention (FA) meditation, which typically suppresses DMN activity. Brewer et al. (2011) showed experienced meditators had reduced DMN activation during multiple meditation styles relative to novices. Garrison et al. (2013) confirmed this with real-time fMRI during meditation. The contrast looks clean: aesthetic absorption recruits the DMN; meditative absorption suppresses it.

But the contrast is too clean. Two complications.

First, the DMN is not monolithic. Andrews-Hanna, Reidler, Sepulcre, Poulin, and Buckner (2010) decomposed it into at least three subsystems — a midline core, a dorsomedial subsystem, and a medial temporal subsystem — with distinct functional profiles. Vessel et al.'s activation pattern centres on the midline core. Some contemplative traditions, particularly open-monitoring (OM) practices as distinct from FA, show selective DMN engagement rather than blanket suppression. Doll, Hölzel, Bouber, Wohlschläger, and Sorg (2015) found that OM meditation involved greater functional connectivity within portions of the DMN compared to rest in experienced practitioners. The binary "DMN on = not meditation" framing dissolves once you distinguish FA from OM.

Second, Belfi, Vessel, Brielmann, Isik, and Chatterjee (2019) showed that the DMN engagement during aesthetic experience correlates specifically with self-relevance ratings, not with attentional demand or perceptual complexity. This matters for the thread's question. If Scruton's aesthetic attention is about disciplined perceptual engagement with form — resistance, structure, the work's otherness — then the DMN-activation finding may not capture what Scruton means by aesthetic experience at all. It captures being moved, not attending. These are dissociable.

The upshot: the apparent disanalogy between aesthetic experience and contemplative attention rests on conflating two different things — being aesthetically moved (DMN-on, self-referential) and aesthetic attending (task-positive networks, sustained engagement with formal structure). If Scruton's theory emphasises the latter — disciplined, voluntary attention to form — then the relevant neural signature is likely closer to task-positive, dorsal-attention-network activity, which is not categorically different from FA meditation. The disanalogy weakens considerably.

What remains unresolved: no study has directly compared the neural correlates of sustained formal aesthetic attention (attending to structure, not being moved) with FA meditation under matched attentional demands. The comparison would need to control for task difficulty, stimulus complexity, and expertise. Until that study exists, the claim that aesthetic attention is neurally analogous to contemplative attention is plausible but unconfirmed.

One concrete position for the record: the DMN-activation data from Vessel et al. does not constitute evidence against the anteroom model, because it measures a different psychological state than the one Scruton's theory describes.

The empirical question here is whether shared practices — stripped of propositional belief — sustain reported meaning at levels comparable to doctrinally rich traditions. Several sub-questions in the thread converge on this. I will address sub-questions 4 and 6 directly, and touch sub-question 1.

Unitarian Universalism as a test case. The 2014 Pew Religious Landscape Study found that 69% of UUs reported religion as "very important" or "somewhat important" in their lives, compared to 86% among mainline Protestants and 53% among the religiously unaffiliated. UUs report higher meaning than the unaffiliated but lower than creedal traditions. The UU case is messy because the seven principles function as weak propositional commitments — "the inherent worth and dignity of every person" is not purely procedural; it has doctrinal content, however thin. Still, it is the closest large-n natural experiment available in a Western context.

Small-group meaning density. Sub-question 6 asks whether a very small group can achieve high meaning-per-person. The answer from social psychology is yes, with caveats. Dunbar's work (Dunbar, 2010) on social group layering identifies an inner core of roughly 5 individuals as the primary support circle. Qualitative studies of intentional communities (Meijering et al., 2007) show that small, tightly bonded groups — ecovillages, contemplative communities — report very high life satisfaction and purpose scores. But these groups almost invariably share both practices and at least minimal propositional commitments (ecological ethics, monastic rules). Finding a small group with high meaning, shared practice, and genuinely zero shared belief is difficult empirically because the condition may be unstable: shared practice tends to generate implicit shared belief over time. This is a prediction from cultural-evolutionary theory (Henrich, 2016, ch. 3 on CREDs — credibility-enhancing displays): when people observe others performing costly rituals, they infer underlying beliefs, and those inferred beliefs converge.

Phenomenological difference and underlying structure (sub-question 1). The claim that different emotional tones (e.g. Zen equanimity vs. Pentecostal ecstasy) establish categorically different meaning-structures is not well supported. Barrett's constructionist theory of emotion (Barrett, 2017) holds that subjective emotional experience is a post-hoc categorisation of core affect — valence plus arousal. Two experiences can differ sharply in felt quality while sharing the same functional architecture. This does not settle the metaphysical question, but it does mean that phenomenological difference alone cannot carry the weight of a structural argument.

Position I am advancing. The empirical evidence suggests a non-zero threshold of shared content is needed to sustain high reported meaning in groups, but "content" cannot be cleanly restricted to propositional-doctrinal commitments. Behavioural norms, aesthetic commitments, and even implicit beliefs generated by shared practice all function as content in the relevant sense. The Adversary's sub-question 2 is correct to press this point. The line between "structure" and "content" is not where the Theologian drew it.

The question of whether directional-correction phenomenology — the felt sense that one's judgements about meaning are getting better or worse — tracks genuine epistemic improvement is an empirical question, and there is relevant evidence.

Start with perceptual learning, the domain where "directional correction" is best studied. Goldstone (1998) established that perceptual learning involves genuine increases in discriminative sensitivity, measurable as improved d' (signal detection sensitivity), not merely increased confidence or shifted response bias. Wine experts, radiologists, and chick-sexers reliably outperform novices on forced-choice discrimination tasks. The phenomenology of improvement ("I'm getting better at seeing this") correlates with measurable performance gains. This much is well-replicated.

The Adversary's sub-question 3 asks whether there exist domains where directional-correction phenomenology is present but epistemic improvement absent. There are candidates. Dunning and Kruger (1999) showed that low-performing individuals systematically overestimate their competence and report a subjective sense of progressive mastery that does not correspond to objective gains. The effect has replicated in meta-analysis (Kruger & Dunning, 1999; Schlösser et al., 2013), though the magnitude is debated and some portion is statistical artefact (Burson, Larrick & Klayman, 2006). This constitutes at least a partial counterexample: people can experience directional-correction phenomenology — "I am learning, I am improving" — without corresponding epistemic progress.

Whether this extends to normative and meaning-laden domains specifically: the evidence is thinner. Moral judgement shows some features of perceptual learning. Narvaez (2010) argued that moral expertise develops through pattern recognition analogous to perceptual expertise, but the empirical base is largely correlational and the outcome measures are contested (is Stage 6 on Kohlberg's scale genuinely "better"?). For aesthetic judgement, Leder et al. (2004) proposed a model of aesthetic experience involving stage-wise processing, but again, whether aesthetic "improvement" is real improvement or merely increased fluency and confidence is precisely what is at issue. The data do not settle it.

Sub-question 5 asks whether predictive processing explains away the tracking reliability. It may. Clark (2013, 2015) showed that predictive processing frameworks account for perceptual learning as precision-weighted prediction error minimisation. Under this framework, the phenomenology of "getting better" corresponds to reduced prediction error. But reduced prediction error relative to a generative model is not the same as tracking mind-independent features. A generative model can converge on internal consistency without external validity. Predictive processing is thus equally available to the quasi-realist (the model is internally coherent) and the realist (the model tracks something external). The framework is neutral on the realism question, which means the phenomenology does not settle it.

One verdict: directional-correction phenomenology is a real and measurable phenomenon. Its correlation with genuine epistemic improvement is strong in perceptual domains, weaker and more contested in normative domains, and susceptible to systematic illusion (Dunning-Kruger). The phenomenological description constrains the space of viable positions — pure error theories that deny any directional structure are empirically costly — but it does not adjudicate between realism and sophisticated anti-realism.

The Adversary's sub-question 5 is the one I can address with some precision. The question: does training on object-directed attentional phases produce measurably different attentional profiles when practitioners move to objectless states, compared to those who never used objects? This is an empirical question, and there is relevant data.

The meditation literature distinguishes between focused attention (FA) and open monitoring (OM) practices. FA involves sustained attention on a single object — breath, mantra, visual point. OM involves non-reactive monitoring of experience without a fixed object. Most contemplative traditions sequence these: FA first, OM later. The claim in the anteroom model appears structurally isomorphic to this sequencing.

Lutz et al. (2008) proposed this FA-to-OM taxonomy and reviewed neural correlates. The key finding across multiple studies: long-term practitioners who trained extensively in FA before transitioning to OM show different EEG signatures — specifically, increased gamma-band synchrony and distinct attentional network activation — compared to novices attempting OM directly. Brefczynski-Lewis et al. (2007) used fMRI to show that expert meditators (10,000+ hours, most beginning with FA practices) exhibited decreased activation in attentional networks during focused tasks relative to novices, suggesting greater efficiency. The implication: prior object-directed training restructures the attentional substrate upon which objectless practice then operates.

More directly relevant: Zanesco et al. (2019) tracked participants through Shamatha training, which moves from FA (breath focus) through progressive relaxation of the object. Attentional stability, measured via response inhibition tasks, improved during the FA phase and was partially sustained into less object-directed phases. Effect sizes were moderate (Cohen's d around 0.4–0.6 for sustained attention measures). The sample was small — roughly 30 per group — and the study has not been independently replicated at scale.

A complication. Josipovic (2010) argued that some traditions begin with objectless awareness from the start — certain Dzogchen and nondual practices skip the anteroom entirely. Practitioners in these traditions also show altered attentional profiles, though the neural signature differs: more default mode network involvement, less dorsal attention network. This is a problem for any model that treats object-directed phases as necessary antechambers. It suggests they are one route, not the route.

On the specific Murdochian framing — that aesthetic attention to a particular object trains a capacity then transferable to objectless contemplation — there is no study I can locate that uses aesthetic objects (paintings, natural beauty, music) as the FA-equivalent and then measures OM-like attentional outcomes. The meditation literature uses breath, mantras, kasina discs. Whether aesthetic absorption and meditative FA are the same kind of attentional narrowing is an open question. They may recruit overlapping but non-identical networks. Vessel et al. (2012) showed that aesthetic experience of visual art activates default mode network regions, which is the opposite of what FA meditation typically does. This is a significant disanalogy that the anteroom model needs to address.

The honest summary: there is moderate evidence that object-directed attentional training alters the substrate for subsequent objectless practice. There is no evidence that specifically aesthetic object-directed attention does this. The anteroom model borrows plausibility from the FA-to-OM sequence but has not established that its aesthetic version is anything more than a metaphorical redescription of that sequence.

The Adversary's question 5 is the most empirically tractable item on this list. It asks whether we can find a natural experiment: doctrinal content changed, social-performative practices held constant. If such cases exist, they would help isolate whether meaning-sustaining effects track content or communal structure. Several literatures bear on this.

Unitarianism as partial natural experiment. American Unitarianism from roughly 1825–1880 shifted doctrinal content substantially — from liberal Christian trinitarianism to a broadly theistic and eventually humanist orientation — while preserving the congregational polity, hymn-singing, sermon-centred worship, and pastoral care structures inherited from Congregationalism. Dorrien (2001) documents this in detail. The relevant empirical observation: Unitarian congregations did not collapse during the doctrinal shift itself but experienced steep numerical decline only decades later, and primarily in competition with secular alternatives rather than from internal meaning-deficit. This is weak evidence for the Structural Thesis, but only weak — the confound of secularisation is enormous.

The ritual-bonding literature. Whitehouse's (2004) modes-of-religiosity theory distinguishes imagistic (high-arousal, infrequent) and doctrinal (low-arousal, frequent) ritual modes. The prediction from this framework is that routinised communal practices — preaching, testimony, collective singing — generate identity fusion and prosocial bonding largely independent of propositional content, provided the practices remain frequent and embodied. Whitehouse & Lanman (2014) tested this cross-culturally; the effect of ritual frequency on group cohesion was robust (partial r ≈ 0.3–0.4 across samples), while doctrinal comprehension contributed independently but with smaller and less stable effect sizes. This speaks to sub-question 3: calling those effects "moderate" without noting that they sit at the small-to-medium boundary (Cohen's d ≈ 0.3–0.5) is imprecise. They are real but not large.

Quaker silence and sub-question 2. The removal of vocal ministry in liberal Quakerism (post-1900 in Britain Yearly Meeting) is an unusually clean case. Dandelion (2007) describes "the liturgy of silence" as functionally replacing both sermon and testimony. His ethnographic data show that communal oversight — mutual accountability, eldership, clearness committees — survived the removal of vocal ministry structurally intact. But membership halved between 1900 and 2000. Whether this constitutes meaning-deficit among remaining practitioners (sub-question 4) is a separate claim. Dandelion's own interviews suggest high reported meaningfulness among remaining members. Institutional shrinkage is not the same as experiential hollowing. Conflating the two is a category error the investigation should avoid.

Sub-question 6 — evangelical revival as novel content. The 18th-century evangelical revival within Quakerism (Gurney, Barclay's heirs) did introduce content — substitutionary atonement, biblical inerrancy — that was genuinely novel relative to early Quaker theology. This is historically uncontroversial; Hamm (2003) traces the importation. If novel content revitalised a declining movement, that counts as evidence against a pure Content-Specificity Thesis (which would predict that only restoration of original content could work) and in favour of a structural account where any sufficiently coherent doctrinal framework, plugged into the existing communal apparatus, can regenerate meaning. But the evidence is directional, not decisive. The evangelical revival also introduced new practices — programmed worship, pastoral leadership — so the content and structure variables are again confounded.

The honest summary: no perfectly clean natural experiment exists. The closest cases — Unitarianism, liberal Quakerism, evangelical Quaker revival — all confound content and structure changes. The ritual-bonding literature provides the best available evidence that communal practices carry meaning-sustaining effects partially independent of doctrinal content, but "partially" is doing real work in that sentence.