Memeplexes as Adaptive Cognitive Systems: A TAME-Integrated Framework for Understanding Ideological Agency

Authors: Patrick Grünig · Claude Fable (Anthropic)

Status: Series foundation — Paper 1.1 of ten; the later papers presuppose it

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, August 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 is a theory proposal, and it holds its verbs to that register. What an external literature has shown is reported in that literature’s own terms; what this framework proposes is marked as proposal; what would decide between them is stated as a prediction (§7). 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 each record’s level.

Note on corpus citations. Two passages cite the project’s source corpus — the writings of Leo Panakal (sigla: PRE = Preview of the Ancient Mother Series of Treatises, 1998, cited by page; K = The Key to the bible, cited by chapter). Quoted wording follows the project’s adjudicated witnesses and preserves the corpus’s deliberate orthography (lowercase christian and kin); printed copies are under physical inspection at the time of writing, and wording will be corrected to the printed state should it differ. The corpus, its method, and the rules governing its use are presented in Paper 5.1. Here the citations do two jobs: attributing ideas to their source, and disclosing that the ideas precede this framework.

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: the mechanism citation for stress-sharing — a chimera assembled from four real bibliographic items — is replaced by the two real publications it pointed toward, at their honest register; one confabulated reference is deleted and the claim it propped up is withdrawn to open-prediction status; the metamorphosis exhibit is restated at its source’s own strength; a supportive study the February draft did not know the project already possessed is cited; and the taxonomy’s derivation circularity is stated rather than buried. The body makes no further reference to prior drafts; the change record lives in the project archive (the claim table and the critic file). This note exists so that the paper’s history is recorded in exactly one removable place.

Abstract

The meme concept has generated productive debate since Dawkins (1976), yet memetics never became a predictive science: by its own critics’ closing verdict, it added no explanatory or predictive power beyond what was available without it (Edmonds, 2005). The central obstacle has been agency. Do large-scale memeplexes (systems of co-adapted memes) merely describe cultural patterns, or do they constitute cognitive agents? We propose a resolution by integrating memetics with Michael Levin’s Technological Approach to Mind Everywhere (TAME), which characterizes cognition functionally across substrates and scales and grades agency attributions by one operational criterion: the prediction-and-control yield of the agentive model against its alternatives. Memeplexes exhibiting six functional properties (§2.1) can qualify as cognitive agents in this operational sense: a classification proposal, justified by predictive yield and testable as such. We propose stress-sharing as the mechanism recruiting individual cognition into memeplex-level computation: proposed for biological collectives in TAME, formalized in an agent-based model (Shreesha & Levin, 2024), and extended here to ideological collectives as this paper’s own hypothesis. We introduce pattern persistence through substrate replacement to explain how memeplexes survive the destruction of their institutional forms, with the biological exhibit stated at its source’s strength. And we propose a taxonomy of memeplex architectures — parasitic, mutualistic, commensal — classified by effect on host cognitive autonomy, whose markers, derived largely from one well-documented case, acquire diagnostic standing only by out-of-sample prediction. The framework yields ten testable predictions; one already has supportive adjacent evidence (cognitive flexibility covarying with religious adherence; Zmigrod et al., 2019).

Keywords: memetics, TAME, scale-free cognition, cultural evolution, substrate independence, collective intelligence, memeplex, cognitive agency, stress-sharing, ideological systems

1. Introduction: The Agency Problem in Memetics

1.1 The Unfinished Science of Memetics

When Richard Dawkins introduced the concept of the meme in The Selfish Gene (1976), he offered a deceptively simple proposition: cultural information replicates, mutates, and undergoes selection in ways analogous to genes. Ideas compete for limited cognitive real estate in human minds, and those best adapted to the selection environment — the cognitive biases, emotional needs, and social structures of their hosts — propagate most successfully.

The analogy was generative. Daniel Dennett (1991, 1995) extended it into a theory of consciousness shaped by memetic competition. Susan Blackmore (1999) argued in The Meme Machine that memetic selection could explain phenomena ranging from language evolution to religious belief. More recently, cognitive scientists including Joscha Bach have described religions, governments, and identities themselves as software running on human minds — a god, in Bach’s formulation, being a self that runs on many brains at once, achieving a form of autonomous existence through its hosts (Bach, 2020).

Yet memetics has not achieved the status of a mature scientific framework. Critics have raised legitimate objections: memes lack a clear unit of analysis analogous to the gene; the mechanisms of memetic replication are poorly specified — where Dawkins proposed replication, Sperber (1996) argued that cultural transmission is reconstructive transformation, which if right dissolves the replicator concept itself; and, most fundamentally, the claim that memeplexes exhibit “agency” or “goals” has remained suspended between metaphor and mechanism (Sperber, 1996; Aunger, 2002; Edmonds, 2005). When we say a religion “wants” to spread, or an ideology “defends itself” against criticism, are we speaking poetically or literally? If literally, by what mechanism does an abstraction acquire goals?

This paper addresses the agency problem directly. We propose that it has remained unresolved not because memetics lacks rigor, but because it has lacked the right theoretical partner — a framework capable of defining cognition in substrate-neutral, operationally testable terms. That framework now exists.

1.2 TAME: A Scale-Free Theory of Cognition

Michael Levin’s Technological Approach to Mind Everywhere (TAME) grew out of decades of experimental work in developmental biology (Levin, 2019, 2022). Levin and colleagues have shown experimentally that biological systems at many scales — individual cells, tissues, organs, whole-organism collectives — exhibit behavior the framework models as goal-directed: navigating problem spaces, correcting errors, reaching target states through diverse means.

TAME’s central claim is that cognition is not a property that emerges at some threshold of neural complexity but a scale-free feature of any system that processes information and pursues goals. A single cell navigating a chemical gradient, an embryo regenerating a limb, a swarm of bees selecting a nest site, and a human planning a career are all cognitive events — differing in degree, not in kind. What matters is not the substrate (neurons, cells, silicon, social networks) but the functional properties: Does the system represent states? Does it pursue goals? Does it adapt to perturbation?

This operational definition removes the need to resolve intractable philosophical questions about consciousness or subjective experience before attributing cognitive agency. As Levin argues, the question is not whether a system is “truly conscious” but what level of cognitive model — from pure mechanism to full agency — yields the most efficient prediction and control of it, an empirical question answered system by system (Levin, 2019, 2022).

1.3 The Proposition

We propose that TAME provides the missing theoretical foundation for memetics. Specifically:

1. Memeplexes can qualify as cognitive agents under TAME’s operational definition — not metaphorically, but in the same functional sense that cell collectives, organs, and organisms qualify. This is a classification proposal, argued in §§2–3 and testable via §7.

2. The mechanism by which memeplexes recruit individual cognition into collective computation is analogous to stress-sharing — a binding mechanism proposed for biological collectives in TAME and since formalized in a computational model (Levin, 2022; Shreesha & Levin, 2024).

3. Memeplexes exhibit substrate-independent pattern persistence — they survive the destruction of specific institutional forms by remapping their core structure onto new substrates, a process for which memory persistence through insect metamorphosis supplies a biological exhibit (§5).

4. The framework generates testable predictions about when and how memeplexes will behave (§7) — the empirical grounding memetics has lacked, here supplied as a research program rather than claimed as a result.

The paper proceeds as follows. Section 2 develops the theoretical integration of TAME and memetics. Section 3 formalizes the criteria for memeplex cognitive agency. Section 4 introduces stress-sharing as the binding mechanism. Section 5 presents the pattern-persistence model. Section 6 proposes a taxonomy of memeplex architectures. Section 7 generates testable predictions. Section 8 considers objections and limitations, and Section 9 concludes.

2. Theoretical Integration: TAME Meets Memetics

2.1 Levin’s Scale-Free Cognition

The foundational insight of TAME is that cognition is defined by what a system does, not by what it is made of. The six properties below are our extraction from Levin’s framework — the grouping is ours; the components are his, drawn from his tenet list and working concepts (Levin, 2019, 2022, 2023):

• Goal-directedness: The system pursues specific target states and can achieve them through diverse means despite perturbation (equifinality, in the systems-theory term we will use for this).
• Navigation of problem spaces: The system moves competently through a space of possible states toward preferred outcomes — Levin’s own core definition of intelligence, and the space need not be spatial (morphospace, transcriptional space, and physiological space are Levin’s own list; cultural space is this paper’s proposed extension, argued in §2.3).
• Error correction: The system registers deviations from target states and acts to reduce them.
• Memory: The system represents states that are not present now, and past states shape future behavior.
• Stress: The system registers the distance between its current state and its target states — stress in Levin’s sense is systemic setpoint error, not threat alone — and mobilizes resources to close the gap.
• Communication: Components share signals that bind their individual activities into collective behavior.

Crucially, these properties have been shown experimentally in systems far simpler than neural networks. Planaria regenerate their target morphology after bisection, and bioelectric interventions can rewrite the target itself, yielding lines of permanently two-headed worms whose fragments go on regenerating the altered form: the “memory” of the body plan is a rewritable stored pattern, not a fixed genomic readout (Levin, 2019, 2022). Xenobot assemblages — motile living constructs built from frog cells — execute behaviors designed for them in silico (Kriegman et al., 2020) and, without any genetic modification, exhibit one no amphibian possesses: kinematic self-replication, sweeping loose cells into moving copies of themselves (Kriegman et al., 2021). Gap-junction-connected cell networks process information and make collective “decisions” about tissue patterning — the scare quotes are ours, and TAME hedges the term the same way, quoting ‘decision-making’ at its own definition (Levin, 2019, 2022).

The implication is radical: cognition does not require neurons. It requires information processing, goal pursuit, and adaptive response. The substrate is irrelevant to the classification.

2.2 The Computational Boundary of a Self

A second critical contribution is Levin’s concept of the computational boundary of a self (Levin, 2019). In biological systems, the “self” is not fixed but dynamically defined: its boundary is the spatio-temporal envelope of what the system can measure, model, and try to affect. Individual cells have a small self — they measure local chemical gradients and act on their own state. When cells form gap-junction networks and share bioelectric signals, their computational boundaries merge: the whole collective can measure and act on the same data, and the tissue or organ becomes a larger “self” pursuing goals defined at a scale inaccessible to any individual cell — in Levin’s summary, “physiological connectivity is the binding mechanism responsible for the appearance of larger unified Selves.”

This merging is not permanent or inevitable, and Levin’s exhibits for its reversal are not hypotheticals. The first is cancer: in carcinogenic transformation, cells “become isolated from the physiological signals that bind them into unified networks” — electrically cut off by a shutdown of gap-junction coupling — and “revert to their unicellular past,” pursuing cell-scale goals (over-proliferation, metastasis) in disregard of the body plan; the scope of the self collapses from body-boundary to cell-surface (Levin, 2019). The second is pharmacological: most anaesthetics, across animals and even plants, work as gap-junctional uncouplers — severing the communication is what removes the cognition (Levin, 2019). The boundary of a self, on this account, is made and unmade by its information channels.

The concept of a dynamically expanding and contracting computational boundary has significant implications for understanding collective human cognition, as we develop in Section 4.

2.3 Mapping Memeplexes onto TAME

We can now state the mapping precisely. A memeplex is a system of co-adapted memes — beliefs, narratives, practices, and institutions — that replicate as an interdependent unit (Blackmore, 1999). We propose that certain memeplexes meet TAME’s criteria for cognitive agency:

The mapping is not merely analogical. It identifies functional equivalences at the level of information processing and goal pursuit. The question is whether these functional equivalences are sufficient to warrant cognitive attribution.

2.4 The Operational Justification

We adopt Levin’s operational criterion in the form he states it, which is comparative and graded, not binary: the right level of cognitive model for a system (anywhere on the continuum from pure mechanism to full agency) is the one that yields the most efficient prediction and control, compared with the alternatives; and whether a given way of modeling a system is the correct one is, in Levin’s words, “an empirical question to be answered in time” (Levin, 2019, 2022). Two features of the criterion matter here. It is two-sided: under-attribution is an error with a name in this literature, Levin’s “neganthropomorphic fallacy” — the failure to credit cognition where crediting it would have improved prediction and control, as much an empirical failure as anthropomorphic excess. And it is observer-relative by design: an estimate of a system’s intelligence “say[s] as much about the observer and their limitations” as about the system (Levin, 2022); that is not a weakness of the framework but the reason its attributions must be cashed out in predictive performance rather than intuition.

Consider a predictive question of the relevant kind: How will a globally distributed religion respond to a sustained, large-scale defection event (e.g., secularization in Western Europe)? An aggregative (passive) model treats the religion as a collection of beliefs held by individuals; its expectation is decline tracking the individual-level drivers, with no coordinated counter-response and no systematic split between surface concessions and core preservation. An agentive model treats the religion as a cognitive system navigating a threat landscape; it expects the system to (a) mobilize defensive resources (counter-apologetics, revival movements), (b) explore alternative substrates (expansion into the Global South, digital evangelism), and (c) modify surface-level features while preserving core patterns (theological liberalization in some branches, fundamentalist retrenchment in others).

The two models thus diverge, and the operational criterion has something to grade. But all three agentive expectations are, of course, matters of record — read off the very history they describe — which is why this illustration is only an illustration: P5 (§7.2) states the model comparison as a prospective test, scored against held-out cases rather than narrated. Whether the agentive model’s more specific predictions are also more accurate is that test’s question, not this paragraph’s verdict.

The agency premise also has an ostensive case that arrived in stages across 2022–2025 and that any reader can now operate. The competences came first, grown in the models; the agents appeared as scaffolding closed a loop of sensing and acting around them — a pattern becomes an agent in a body, the lesson §2.2 drew from gap junctions. A large language model agent pursuing a multi-step goal — answering questions, writing code, executing a plan — is agency in an informational pattern: weights and text in a tool-loop, no biology in it, navigating toward an outcome. Whether such a system “truly understands” is philosophically unresolved and, for the operational question, beside the point: treating these systems as agents with capabilities and strategies is, as a matter of public record, how they are predicted, managed, and used. Two disciplines accompany the case. First, Levin’s own caution against over-attribution names thermostats and simple AIs as the place agency-talk goes wrong; the reply is the criterion itself — the agentive stance earns its keep where it buys prediction and control for the effort, a graded and comparative matter, and current language-model agents pass that test where thermostats do not. Second, the case cuts in a direction that matters for everything downstream of this paper: these machine agents arise from optimization processes in which no one authors the specific competences (engineers build the training pipeline; the capabilities themselves are grown). Pattern agency, in the one non-biological case where its emergence can be watched, requires no designer of content. The same pragmatic logic applies to memeplexes: asking whether a religion “truly wants” to spread is as productive as asking whether a language model “truly understands.” In both cases, functional agency is the relevant concept, and the metaphysical question can be set aside without loss.

3. Formalizing Memeplex Cognitive Agency

3.1 The Agency Spectrum

Not all collections of memes constitute cognitive agents. A fashion trend, a slang term, or a dietary preference may spread memetically without exhibiting goal-directed behavior or adaptive responses to threat. We propose that memeplex cognitive agency exists on a spectrum, defined by the number and strength of TAME criteria the system satisfies.

We identify four levels:

Level 0 — Inert memes: Individual ideas or behaviors that replicate through imitation but exhibit no coordination, self-preservation, or adaptive response. Example: a catchy jingle. These are the “atoms” of memetic space — they replicate, but they do not compute.

Level 1 — Co-adapted meme clusters: Groups of memes that replicate more successfully together than apart, due to mutual reinforcement. Example: a political platform combining economic, social, and cultural positions that appeal to overlapping constituencies. These clusters exhibit weak coordination but no active self-defense or adaptation.

Level 2 — Adaptive memeplexes: Interconnected meme systems that exhibit self-preservation (responding to threats), adaptation (modifying strategies in response to environmental change), and memory (maintaining identity over time despite surface variation). Example: a major political ideology that survives leadership changes, electoral defeats, and cultural shifts by evolving its messaging while preserving core commitments. These systems meet several TAME criteria and can productively be modeled as agents.

Level 3 — Autonomous memeplexes: Systems that exhibit all TAME criteria at high intensity: robust goal-directedness, sophisticated problem-space navigation, active error correction, long-duration memory, rapid stress response, and rich internal communication. These systems operate with sufficient complexity and coordination to generate behavior that no individual host intended or planned. Candidate example: major world religions that have persisted for millennia, survived the collapse of empires, adapted to radically different cultural contexts, and continue to recruit new hosts despite organized opposition. We call the example a candidate deliberately: Level 3 is a classification this framework assigns by argument, and P4–P5 (§7.2) state what would test the assignment.

The taxonomy is not categorical but dimensional. Memeplexes may move along the spectrum over time — gaining or losing agency as their complexity, coherence, and communication bandwidth change.

3.2 Necessary Conditions

We propose the following necessary conditions for attributing Level 2+ cognitive agency to a memeplex:

1. Self-preservation behavior: The system takes identifiable actions (through its human hosts) to maintain its own existence when threatened. This must go beyond passive inertia — the system must respond to threats in ways that are adaptive, not merely persistent.

2. Means-end flexibility (equifinality): The system achieves similar outcomes through different strategies across different contexts. A religion that spreads through military conquest in one era, missionary activity in another, and social media in a third exhibits means-end flexibility — the goal (propagation) remains constant while the means vary.

3. Trans-host coordination: The system coordinates behavior across multiple human hosts in ways that serve the memeplex’s propagation rather than the individual hosts’ interests. When believers sacrifice personal resources, relationships, or even lives for the ideology’s benefit, that is evidence that the memeplex’s “goals” are being pursued at the expense of host-level fitness (§8.4 states why this condition is the hardest of the four to operationalize).

4. Temporal persistence with identity maintenance: The system maintains recognizable identity over time periods far exceeding individual human lifespans, adapting surface features while preserving core patterns.

No single condition is sufficient. Jointly satisfied, they constitute the strongest warrant the framework can offer for adopting the agentive stance toward a memeplex — a warrant, on the operational criterion, for a modeling decision, not a metaphysical verdict (§8.1).

3.3 The Emergent Agency Objection

A predictable objection: memeplex agency is merely the emergent result of individual human decisions and therefore not “real” agency. This objection proves too much. By the same logic, the agency of a multicellular organism is merely the emergent result of individual cell decisions; the agency of a brain is merely the emergent result of individual neuron firings. TAME’s insight is that emergence does not diminish agency — it constitutes it at a higher scale. Every cognitive agent we know of, including the human mind, is an emergent property of components that are themselves cognitive at a lower scale. And the machine case of §2.4 makes the same point from the other side: agent-grade competences emerge from optimization without anyone authoring them.

The relevant question is not “Is the agency emergent?” (it always is) but “Does the emergent system exhibit goal-directed behavior that cannot efficiently be predicted from the behavior of its components?” — the same explanatory-economy test §8.2 states in full. If the answer is yes — if the memeplex does things that none of its individual human hosts planned or intended — then the memeplex constitutes a distinct cognitive agent at a higher level of organization.

4. Stress-Sharing as the Binding Mechanism

4.1 Biological Precedent

How do individual cognitive units merge into a larger cognitive system? TAME’s proposed answer is stress. In a box titled “Stress as the glue of agency,” Levin (2022) describes stress as a systemic response to the difference between current state and a desired setpoint, propagated outward by an unsatisfied sub-agent as “hard-to-ignore signals” that recruit distant sub-agents to act — stress serving, in his words, as “an influence that binds subunits across space into a coherent higher Self.” The proposal is that shared stress is what scales selves: it yokes modules that sense and act at different scales and locations into one homeostatic loop, serving, in Levin’s comparison, the same function as hidden layers in a network.

This proposal has since been formalized computationally. Shreesha and Levin (2024) built a multiscale agent-based model in which cells share stress — modeled as an abstract homeostatic-error signal — with their neighbors, and measured the effect on collective morphogenesis. Populations with stress-sharing reached their anatomical target configurations faster than populations without it (“enhanced morphogenetic efficiency,” in the authors’ terms), and stress-sharing extended each cell’s radius of influence over distant cells’ fates — results they summarize as “supporting the hypothesis that stress sharing increases collective cohesiveness.”

Stress-sharing as the binding mechanism of biological collective intelligence is, then, a proposal within TAME, with support from an in-silico model rather than from living tissue; this paper’s central mechanism claim inherits that register. (The concrete signal types one might picture — bioelectric gradients, chemical alarm signals, mechanical forces — are the modalities of Levin’s broader framework, not measurements of the model, and are named here as illustration only.) The proposal’s converse, that removing the shared stress degrades the collective’s coherence, is in the model simply the observed difference between populations with and without stress-sharing; in living systems it remains a prediction. We import the mechanism at that strength and no higher.

4.2 Memetic Stress-Sharing

We propose that memeplexes recruit individual human cognition into collective computation through an analogous mechanism: shared emotional stress. Specifically:

• Shared guilt functions as a stress signal that synchronizes the behavior of individuals within a guilt-based memeplex. The doctrine of Original Sin, for example, installs a baseline stress state in adherents — a persistent sense of moral inadequacy that can be managed (never resolved) through continued participation in the system.

• Shared fear (of divine punishment, social ostracism, eternal damnation) functions as an alarm signal that activates collective defensive responses when the memeplex is threatened.

• Shared outrage (at perceived enemies, heretics, or moral violations) coordinates collective aggressive behavior directed at threats to the memeplex.

• Shared ecstasy (ritual euphoria, mystical experience, group worship) functions as a reward signal that reinforces the merged state. This fourth channel extends the mechanism beyond stress proper — it is the binding loop’s positive pole, and we mark it as this paper’s extension rather than part of the biological proposal. (Its place among the “shared emotional experiences” of §2.3’s Communication row is on the memeplex side of the mapping only; the biological column carries no reward analogue in the stress proposal.)

In each case, the shared emotional state creates a common information channel that synchronizes individual behavior, effectively expanding the computational boundary of each individual to include the collective. A believer experiencing guilt is not only feeling an emotion — they are, in the framework’s terms, sharing a stress signal that binds them into a larger cognitive unit: computing as a node in the memeplex’s network. The within-host face of the defensive channel, at least, is measurable: challenges to strongly held political beliefs engaged default-mode structures associated with self-representation more than challenges to non-political beliefs, and participants who updated less showed more insula and amygdala involvement while weighing the counterevidence (Kaplan, Gimbel & Harris, 2016; n = 40; the emotion-circuit correlation did not reproduce in the one preregistered replication — see Paper 1.3, §5.1). Read through this framework (the identity-relevant belief as the memeplex’s anchor point in the host), that is an individual-level substrate on which a collective defense response can be built.

The Original Sin example is not ours, and its provenance is part of the disclosure. The reading of that doctrine as an installed baseline — a manufactured and unpayable debt, rather than a description of a condition the doctrine found — is Leo Panakal’s thesis, argued from the tradition’s own texts a generation before this framework existed: “As guilt is irredeemable because imagined, so is its inescapable counterpart of christian sin irredeemable” (PRE, pp. 23–24); the Key derives the same structure as implantation followed by prohibition, and identifies the sin concept’s defining property with guilt’s (K, chs. V–VI). Paper 5.1 presents that corpus and its method, and grades what they can carry. Here the example is used at its functional face value: a doctrine that specifies a permanent deficit, manageable only inside the system that names it, is — whatever else it is — a stress-installation device of the kind the mechanism requires.

4.3 The Dissolution Prediction

If stress-sharing is the binding mechanism, then a strong prediction follows: removing the shared stress should cause the memeplex to lose collective coherence. Individuals whose guilt, fear, or shame is resolved, with no alternative shared stress taking its place, should progressively disengage from the memeplex, as their computational boundary contracts back toward the individual level. Conversely, memeplexes that intensify shared stress during periods of threat — sharpening rhetoric about divine punishment, external enemies, or moral decay — should exhibit increased collective coherence: the memetic analogue of tighter coupling under stress.

We state both halves as predictions (P2 and P3, §7.1); the existing literature neither delivers nor refutes them, and the gap has a precise shape. Documented: clinical descriptions of significant distress in people leaving high-demand religious environments (Winell, 1993, 2011), and survey evidence that a non-trivial fraction of U.S. adults report religious trauma — with the caveat that the best-known prevalence figures come from an opt-in panel and carry that limitation (Slade et al., 2023). Also documented, and calibrating: in a longitudinal study of ~20,000 Dutch adults (450 of whom deconverted), ordinary deconversion shows no average change in well-being (Bleidorn et al., 2024) — so exit dynamics of the kind this model concerns itself with belong to high-demand, high-stress architectures, not to religious disaffiliation as such (Streib, 2021, reviews the wider deconversion literature). Not yet shown by any study: the temporal order the model predicts, resolution of the shared stress preceding and facilitating disengagement. The clinical literature mostly documents the reverse sequence — guilt and fear persisting after exit and needing treatment then — which this framework reads as the residue of an installed baseline outlasting membership, but which it cannot cite as evidence for P2’s ordering. P2 is offered as what it is: the framework’s most direct falsifier, not yet run.

4.4 Comparative Evidence: Non-Religious Memeplexes

The stress-sharing model is not specific to religion. Political movements that sustain shared outrage (nationalist movements emphasizing external threat), corporate cultures that sustain shared anxiety (high-pressure sales environments), and conspiracy communities that sustain shared paranoia display the same surface signature:

1. A shared emotional stress creates a common information channel.
2. Individuals synchronize behavior around the stress response.
3. The collective exhibits coordinated, goal-directed behavior that exceeds what any individual planned.
4. Removal of the stress coincides with fragmentation of the collective.

These are structured observations, not run studies; we offer them as the cross-domain face of P1–P3 rather than as confirmation. If the signature holds up under measurement across these domains, the mechanism is general: stress-sharing as cognitive glue would be a substrate-neutral binding principle rather than a metaphor borrowed from biology. P1–P3 state the tests.

5. Pattern Persistence Through Substrate Replacement

5.1 The Biological Exhibit

A second import from developmental biology concerns the persistence of memory through radical bodily reorganization. Blackiston, Silva Casey, and Weiss (2008) conditioned larvae of the moth Manduca sexta to avoid an odorant and tested the adults after metamorphosis: moths conditioned as late-stage caterpillars retained the aversion across pupation — through a transformation in which the larval brain is extensively remodeled and the body plan rebuilt. Stated at the study’s own strength: retention held only for conditioning late in larval life (earlier larval memory did not survive), and the authors’ account of the mechanism is continuity, not resurrection — the results are, in their words, “consistent with, but do not provide conclusive support for the survival of synaptic connections” through the remodeling, with adult recall involving brain regions that develop late in the larva. The memory persists because part of its carrier does.

The stronger reading — that what survives metamorphosis is best described as an informational pattern, maintained and remapped while its substrate is extensively rebuilt around it — is Levin’s interpretive frame for this and related findings (Levin, 2022), and TAME’s own wording shows the frame running ahead of the finding: it describes the caterpillar’s brain as “largely dissolved and reassembled into a different architecture,” where the study it cites reports surviving connections. This paper adopts the frame as a frame, with the moth study as its consistent-with exhibit rather than its proof. The distinction matters because the popular version of this result — total liquefaction, memory surviving anyway — overstates both the biology and what the framework needs. What the biology shows is a pattern persisting through radical but partial substrate replacement, carried across the transition by surviving structure.

5.2 Memetic Metamorphosis

For the memetic analogue, the corrected biology is not a loss but a better fit. We propose that memeplexes exhibit an analogous capacity for pattern persistence through substrate replacement: when the institutional, cultural, or social substrate of a memeplex is destroyed or rendered unviable, its core informational patterns — goal-structures, dependency architectures, propagation strategies, error-correction mechanisms — can remap onto a new substrate. And as in the moth, the transition is never carried by nothing: texts survive institutional collapse, practices survive textual loss, dispersed host networks survive both. The pattern crosses on whatever partial carriers persist, and is rebuilt outward from them.

Consider the recurring historical shape:

• A political movement survives the destruction of its founding party by remapping its core ideology onto successor organizations, informal networks, or cultural narratives.
• A philosophical tradition survives the closure of its institutional centers by remapping its core concepts onto new literary forms, educational systems, or religious frameworks.
• A religious system survives the collapse of its associated civilization by remapping its core practices and beliefs onto the cultural frameworks of its conquerors or successors.

In each case, the surface-level features change dramatically — new institutions, new leaders, new rituals, new terminology — while the deep structure (the dependency architecture, the authority model, the propagation strategy) persists. This is memetic metamorphosis: the institutional body is dismantled, and the pattern re-grows a body from its surviving carriers.

5.3 Formalization

We can formalize this concept as follows. Let a memeplex M be characterized by:

• Surface features S(M): specific doctrines, rituals, institutional forms, aesthetic styles, and leadership structures.
• Deep structure D(M): authority architecture (centralized vs. distributed), dependency model (guilt-based vs. knowledge-based vs. reward-based), propagation strategy (missionary vs. organic vs. coercive), and error-correction mechanism (heresy management, schism handling, apologetics).

A memeplex undergoes substrate replacement when S(M) changes dramatically — potentially beyond recognition — while D(M) remains functionally equivalent. The deep structure is the “pattern” that persists; the surface features are the “substrate” that is replaced.

This formalization generates specific predictions: when a memeplex faces existential institutional crisis, we should expect it to (a) seek new institutional substrates compatible with its deep structure, (b) modify surface features to fit the new substrate while preserving deep structure, and (c) exhibit recognizable behavioral continuity (goal-pursuit, self-preservation, error-correction) across the transition. §7.3 states the tests.

6. A Taxonomy of Memeplex Architectures

6.1 Structural Classification

The framework developed above is architecturally neutral — it describes how memeplexes function as cognitive agents without evaluating their effects on human hosts. To address that dimension, we borrow ecology’s classification of symbiotic relationships:

• Mutualistic: Both symbiont and host benefit.
• Commensal: Symbiont benefits; host is unaffected.
• Parasitic: Symbiont benefits at the host’s expense.

The application of this triple to memeplexes is not new — within this project it appears in a February 2025 research layer (as three of seven replication-strategy categories, classified by host fitness), and the underlying idea runs back through the memetics literature to the founding text itself (Dawkins, 1976). What this paper contributes is the criterion: we propose that memeplexes be classified not by diffuse host fitness but by their structural effect on host cognitive autonomy — the host’s capacity for independent evaluation, revision, and exit.

Mutualistic memeplexes enhance the host’s cognitive autonomy, problem-solving capacity, and well-being. Their propagation strategy relies on hosts voluntarily spreading ideas they find genuinely useful.

Commensal memeplexes persist in hosts without significantly affecting their autonomy or well-being. Many cultural traditions — culinary practices, aesthetic preferences, linguistic conventions — plausibly fall here. They replicate because they are embedded in social contexts, not because they create dependency.

Parasitic memeplexes persist by reducing the host’s cognitive autonomy and creating dependency. Their propagation strategy relies on (a) diagnosing a “problem” in the host that only the memeplex can solve, (b) creating emotional states (guilt, fear, shame) that bind the host to the system, and (c) suppressing the host’s capacity for independent evaluation of the memeplex’s claims.

Where does science fall? We state the classification as a hypothesis with its analysis owed, because no marker analysis (§6.2) has been run on science, and the taxonomy’s flattering pole deserves the same discipline as its damning one. Two markers can be checked by inspection: the dependency-loop marker is absent (the deficit science addresses, ignorance of how things work, is not manufactured by science, and its resolution is not designed to reactivate it), and the epistemic-closure marker is inverted (evaluation by standards external to the system is not just permitted but constitutive). A full run of the seven markers is future work, and it would have to engage two complications we flag rather than hide: institutional science as an institution can exhibit belief-protective behavior at the sociological level, a phenomenon distinct from the methodology but not cleanly separable from it in practice; and the project’s own 2025 research layer classed scientific methodology not as mutualistic but as defensive, an immune-system architecture structured to resist other memes — a rival classification the marker analysis would adjudicate.

The corpus this project draws on contributes one further classificatory axis, older than any of the above and orthogonal to it: transparency. In a comparative aside, Panakal distinguishes hostile systems that declare themselves — systems “transparent in their intentions and actions” are “at any rate a perceivable enemy” (PRE, p. 5), against which hosts can mobilize — from the concealed case, which he treats as the dangerous one. In the terms of this taxonomy: two memeplexes with identical autonomy effects may differ in whether their operation is legible to their hosts, and the concealed configuration should be more persistent (nothing mobilizes against it) and more damaging (exit requires a discovery step before a decision step). We name transparency-of-operation as a candidate second axis for this taxonomy and leave its development to future work; Paper 5.1 presents its source.

6.2 Structural Markers of Parasitic Architecture

One confession must precede this list, because it determines what the list can claim: the markers below were derived in large part from the system this series later treats as its central case study — Christianity is visible in their examples. A marker set reverse-engineered from one case cannot then diagnose that case; the reasoning would be circular. The markers therefore earn diagnostic standing only one way: by predicting outcomes in other systems — and in the source case itself only via outcome measures that are independent of the derivation (§7.4). Until those tests are run, the list is a hypothesis about what parasitic architecture looks like, generalized from one well-documented instance.

We identify the following structural features as markers of parasitic memeplex architecture. No single feature is diagnostic; the combination and intensity determine the classification.

1. Unfalsifiable core claims: The system’s central propositions are structured to be immune to disconfirmation. Counter-evidence is interpreted as further confirmation (e.g., “persecution proves we are right”) or as a test of faith.

2. Dependency loops: The system creates a problem (e.g., moral inadequacy) and offers the only solution (e.g., redemption), creating a cyclical dependency in which the “cure” periodically reactivates awareness of the “disease.”

3. Epistemic closure: The system discourages or prohibits evaluation of its claims by standards external to itself. “Faith” — belief without or against evidence — is elevated above empirical evaluation.

4. Exit costs: Leaving the system incurs severe social, emotional, or psychological penalties (loss of community, family rupture, existential terror, identity dissolution).

5. Thought-terminating mechanisms: The system provides formulaic responses that halt critical inquiry (“God works in mysterious ways,” “You just need to have faith,” “That’s the devil talking”).

6. Trans-generational installation: The system is installed in hosts before they develop the capacity for critical evaluation (childhood indoctrination), reducing the probability of later rejection.

7. Host-against-host mobilization: The system recruits hosts to suppress defection in other hosts, creating a self-policing network that reduces the system’s need for centralized enforcement.

6.3 Classification Is Empirical, Not Moral

This taxonomy describes structural features, not moral judgments. A parasitic architecture may coexist with genuine benefits to the host (social support, meaning-making, community belonging) — just as biological parasites sometimes provide incidental benefits to their hosts. The classification refers to the structural relationship between the memeplex’s propagation strategy and the host’s cognitive autonomy, not to the host’s subjective experience or the system’s cultural value.

The structural approach is also what makes the taxonomy falsifiable, and after §6.2’s confession that is a requirement rather than a virtue: if the parasitic-architecture markers correlate with specific, measurable outcomes (reduced cognitive flexibility, increased dependency, measurable mental-health impacts) in systems beyond the one they were derived from, the taxonomy is fit for diagnostic use. If they do not, it is refuted — and §7.4 shows the first such test has, in part, already been run.

7. Testable Predictions

The framework generates the following empirically testable predictions:

7.1 Stress-Sharing Predictions

P1: Populations embedded in memeplexes with high shared-stress features (guilt, fear, apocalyptic anxiety) should exhibit greater behavioral synchronization and collective action capacity than populations in low-stress memeplexes, controlling for population size and resources.

P2: Therapeutic or social interventions that resolve individual guilt/fear without providing alternative community bonds should be followed by disengagement from the memeplex — resolution preceding exit, the temporal order §4.3 identifies as undocumented. Interventions that resolve guilt/fear while providing alternative community should accelerate disengagement.

P3: Memeplexes should intensify stress-signaling (apocalyptic rhetoric, moral panic) during periods of threat (declining membership, cultural competition, political marginalization) — the memetic analogue of tighter coupling under stress.

7.2 Agency Predictions

P4: Memeplexes at Level 3 agency should exhibit means-end flexibility: achieving similar propagation outcomes through different strategies across different cultural contexts. Testable by comparing the propagation strategies of a single religious tradition across culturally diverse regions.

P5: The behavior of Level 3 memeplexes during existential crises should be better predicted by an agentive model (the memeplex as goal-pursuing system) than by an aggregative model (the sum of individual decisions). This is the comparison §2.4 declined to award itself: testable by running both models against historical data on religious responses to persecution, competition, and modernization, and scoring them.

7.3 Pattern-Persistence Predictions

P6: When a memeplex undergoes substrate replacement (radical institutional transformation), its deep structure D(M) should remain functionally equivalent across the transition even as surface features S(M) change dramatically. Testable by operationalizing D(M) (authority architecture, dependency model, propagation strategy) and measuring its stability across documented institutional transitions.

P7: Memeplexes with more complex deep structures should exhibit greater capacity for substrate replacement (more pattern to remap onto new carriers); simpler memeplexes should be more substrate-dependent and more vulnerable to institutional destruction.

7.4 Taxonomy Predictions

P8: Memeplexes classified as parasitic by the structural markers should correlate with measurable reductions in host cognitive flexibility (e.g., Wisconsin Card Sorting Test performance) relative to matched populations in mutualistic or commensal memeplexes.

The nearest study has already been run, and it is supportive. Zmigrod, Rentfrow, Zmigrod, and Robbins (2019; N = 744) found religious disbelief associated with greater cognitive flexibility across three instruments — the Wisconsin Card Sorting Test, the Remote Associates Test, and the Alternative Uses Task — with flexibility tracking practice frequency linearly, and current affiliation outweighing upbringing. What P8 adds, and what remains open, is the independent variable: this framework predicts that flexibility tracks parasitic-marker load, not religiosity as such. A mutualistic or commensal religious configuration should show no deficit; a non-religious system carrying the markers (§4.4’s cases) should show one. That sharpened comparison has not been run. And because Zmigrod’s design is cross-sectional, it leaves the self-selection confound (§8.4) untouched; P9 is the discriminator.

P9: Hosts of parasitic memeplexes who undergo deconversion should exhibit measurable increases in cognitive flexibility, self-reported autonomy, and — in high-demand contexts specifically (§4.3) — well-being after a transitional period, consistent with the removal of a parasitic symbiont. The within-person longitudinal design is what separates “rigid people choose rigid systems” from “rigid systems produce rigid people.”

P10: Memeplexes with more parasitic-architecture markers should exhibit greater collective coherence but lower individual host well-being, while mutualistic memeplexes should exhibit lower collective coherence but higher individual host well-being — a tradeoff between memeplex fitness and host fitness.

8. Objections and Limitations

8.1 The Intentional Stance vs. Real Agency

One might argue that this framework merely adopts Dennett’s “intentional stance” (1987), treating systems as if they have beliefs and goals for predictive convenience, without claiming they really do. We accept the characterization, and so, explicitly, does the framework’s source: Levin describes his own approach as “a kind of ‘intentional stance’ (Dennett, 1987) approach, generalized beyond brains and behavior” (Levin, 2019), relates TAME’s persuadability continuum to the intentional stance directly (Levin, 2022), and, in the half the objection always omits, names the symmetrical error — the neganthropomorphic fallacy of §2.4, “as bad for empirical research as is profligate anthropomorphic reasoning” (Levin, 2019). The stance is not a limitation discovered in this framework; it is the framework’s announced method, with a two-sided error structure the objection reads as one-sided. The intentional stance is, moreover, how we attribute agency to all entities, including other humans, whose subjective states are never directly accessible. The framework does not claim that memeplexes are conscious; it claims they are cognitive agents in the same operational sense that cell collectives, immune systems, and developmental networks are cognitive agents under TAME. A stronger ground for the attribution exists — that doing the things is what the agency consists in, with no “as if” about it — but nothing in this paper requires the stronger ground, and we argue throughout on the weaker, predictive-utility one.

8.2 The Reification Objection

Critics may argue that we reify an abstraction — treating a pattern of human behaviors as a “thing” with its own goals. We respond that all cognitive agents are, at a lower level of description, “patterns of behavior” of their components. A human mind is a pattern of neural activity; an organ is a pattern of cellular behavior. The question is never whether the agent can be described as a pattern of its components’ behavior (it always can) but whether the pattern-level description adds explanatory power over the component-level description — which it does exactly when the pattern exhibits coordinated, goal-directed behavior that no component planned or intended.

8.3 Falsifiability

Is the framework falsifiable? Yes, and §7 is the inventory: if memeplexes classified as agentive do not exhibit means-end flexibility (P4), if the agentive model loses the scored comparison (P5), if stress-feature intensity does not track collective coherence (P1–P3), if deep structure fails to persist across substrate transitions (P6–P7), or if the parasitic markers fail to predict autonomy and flexibility outcomes in systems beyond their derivation case (P8–P10), the framework is refuted in the corresponding part.

This inventory answers a standard we did not set ourselves. When the Journal of Memetics closed in 2005, Edmonds’ verdict in its final issue was that memetics had failed because it “has not provided any extra explanatory or predictive power beyond that available without the gene-meme analogy” — with the premature memetic “explanation” of religion named as the characteristic overreach (Edmonds, 2005). We accept that criterion as the right one; it is, in substance, the same predictive-yield standard this paper adopts from Levin in §2.4, applied to the field itself. Section 7 is our attempt to meet it — on pain of the same verdict.

8.4 Limitations

Five limitations should be on the record:

1. Operationalization: The criteria for memeplex agency (§3) and the structural taxonomy (§6) require further operationalization before they can be applied consistently across cases. We have provided conceptual definitions; measurement instruments remain to be developed. Condition 3 (§3.2) is the hardest of the four: identifying host sacrifice as memeplex-serving requires a host-interest baseline specified independently of the memeplex’s own value system — a believer’s utility function includes salvation, and sacrifice can serve inclusive fitness.

2. Historical data: Several predictions (P5, P6, P7) require comparison against historical data that may be incomplete, biased, or ambiguous. Retrospective application of the framework must be treated as hypothesis-generating, not hypothesis-confirming.

3. Target-derivation circularity: The parasitic markers were derived largely from the Christian case (§6.2’s confession). Their diagnostic force therefore depends entirely on out-of-sample prediction; applied back to their source case alone, they would show nothing. This is the framework’s most consequential structural limitation, and the design of P8–P10 is its mitigation.

4. Cultural scope: The framework has been developed primarily with reference to Western ideological systems. Cross-cultural validation — particularly with non-Abrahamic religious systems, indigenous knowledge traditions, and non-Western political ideologies — is essential, and is distinct from limitation 3 (a marker set can be culture-fair and still target-derived, or vice versa).

5. Measurement and selection: Measuring “cognitive flexibility,” “autonomy,” and “well-being” across populations embedded in different memeplexes raises hard methodological problems, chief among them self-selection: do rigid people choose rigid memeplexes, or do rigid memeplexes produce rigid people? Cross-sectional designs cannot answer this (§7.4); longitudinal within-person designs (P9) can.

9. Conclusion

This paper has proposed a theoretical integration of memetics and Michael Levin’s TAME framework, arguing that large-scale memeplexes can constitute cognitive agents under TAME’s scale-free, operational definition of cognition. The integration addresses memetics’ longstanding agency problem: memeplex agency is neither metaphorical nor mystical but functional — a modeling stance licensed by the same pragmatic criterion that warrants cognitive attribution to cell collectives, developmental networks, and machine agents, and adopted here at the strength that criterion supports.

We have imported stress-sharing as the candidate mechanism by which individual human cognition is recruited into memeplex-level computation: proposed for biological collectives in TAME, formalized in silico, extended here to ideological collectives as this paper’s own hypothesis. We have proposed pattern persistence through substrate replacement as the mechanism by which memeplexes survive institutional destruction, with the biological exhibit held to what its source shows: patterns crossing radical remodeling on partial carriers, in moths as in movements. And we have proposed a structural taxonomy — parasitic, mutualistic, commensal, classified by effect on host cognitive autonomy, with transparency of operation flagged as a second axis — whose markers acquire diagnostic standing only by out-of-sample prediction, one instance of which (P8’s flexibility association) already has supportive adjacent evidence.

The framework is substrate-neutral and ideology-neutral by construction: it applies equally to religious, political, economic, and cultural systems, and it evaluates architecture, not content — not what a system says, but how it operates on the cognition of its hosts. Its predictions invite work across cognitive neuroscience, social psychology, computational modeling, cultural anthropology, and the history of ideas. If they are borne out, memetics will have advanced from a suggestive metaphor toward a predictive science of cultural cognition. If they are falsified, the attempt will have clarified the boundary conditions under which cultural systems can and cannot be meaningfully described as cognitive agents.

Either way, the question — Do our ideas think through us? — deserves a scientific answer.

References

External literature (each entry carries a verification record in the project archive — journal articles and the podcast at fetch level, books at book level; where a record is partial, the entry says so)

• Aunger, R. (2002). The Electric Meme: A New Theory of How We Think. Free Press.
• Bach, J. (2020, June 13). Artificial consciousness and the nature of reality (No. 101) [Audio podcast episode]. In Lex Fridman Podcast. https://lexfridman.com/joscha-bach/
• Blackiston, D. J., Silva Casey, E., & Weiss, M. R. (2008). Retention of memory through metamorphosis: Can a moth remember what it learned as a caterpillar? PLoS ONE, 3(3), e1736. DOI 10.1371/journal.pone.0001736.
• Blackmore, S. (1999). The Meme Machine. Oxford University Press.
• 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. [longitudinal, ~20,000 Dutch adults; 450 deconverts]
• Dawkins, R. (1976). The Selfish Gene. Oxford University Press.
• Dennett, D. C. (1987). The Intentional Stance. MIT Press.
• Dennett, D. C. (1991). Consciousness Explained. Little, Brown and Company.
• Dennett, D. C. (1995). Darwin’s Dangerous Idea: Evolution and the Meanings of Life. Simon & Schuster.
• Edmonds, B. (2005). The revealed poverty of the gene-meme analogy — why memetics per se has failed to produce substantive results. Journal of Memetics — Evolutionary Models of Information Transmission, 9(1). http://cfpm.org/jom-emit/2005/vol9/edmonds_b.html (archived: web.archive.org/web/20260715113915/http://cfpm.org/jom-emit/2005/vol9/edmonds_b.html)
• Kaplan, J. T., Gimbel, S. I., & Harris, S. (2016). Neural correlates of maintaining one’s political beliefs in the face of counterevidence. Scientific Reports, 6, 39589. DOI 10.1038/srep39589.
• Kriegman, S., Blackiston, D., Levin, M., & Bongard, J. (2020). A scalable pipeline for designing reconfigurable organisms. Proceedings of the National Academy of Sciences, 117(4), 1853–1859. DOI 10.1073/pnas.1910837117.
• Kriegman, S., Blackiston, D., Levin, M., & Bongard, J. (2021). Kinematic self-replication in reconfigurable organisms. Proceedings of the National Academy of Sciences, 118(49), e2112672118. DOI 10.1073/pnas.2112672118.
• Levin, M. (2019). The computational boundary of a “self”: Developmental bioelectricity drives multicellularity and scale-free cognition. Frontiers in Psychology, 10, 2688. DOI 10.3389/fpsyg.2019.02688.
• Levin, M. (2022). Technological approach to mind everywhere: An experimentally-grounded framework for understanding diverse bodies and minds. Frontiers in Systems Neuroscience, 16, 768201. DOI 10.3389/fnsys.2022.768201. [TAME]
• Levin, M. (2023). Collective intelligence of morphogenesis as a teleonomic process. In P. A. Corning, S. A. Kauffman, D. Noble, J. A. Shapiro, R. I. Vane-Wright, & A. Pross (Eds.), Evolution “On Purpose”: Teleonomy in Living Systems (pp. 175–198). The MIT Press. DOI 10.7551/mitpress/14642.003.0013. (PsyArXiv preprint 2022, DOI 10.31234/osf.io/hqc9b.)
• Shreesha, L., & Levin, M. (2024). Stress sharing as cognitive glue for collective intelligences: A computational model of stress as a coordinator for morphogenesis. Biochemical and Biophysical Research Communications, 731, 150396. DOI 10.1016/j.bbrc.2024.150396.
• 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. [opt-in panel; cited with that limitation]
• Sperber, D. (1996). Explaining Culture: A Naturalistic Approach. Blackwell.
• Streib, H. (2021). Leaving religion: Deconversion. Current Opinion in Psychology, 40, 139–144. DOI 10.1016/j.copsyc.2020.09.007.
• 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). [May installment fetched; series per the BABCP listing]
• 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 project corpus (presented in Paper 5.1; quotations per the project’s adjudicated witnesses)

• Panakal, L. (1998). Preview of the Ancient Mother Series of Treatises, Part I. Society of Friends of the Ancient Mother / Identity Publishers. [PRE]
• Panakal, L. The Ancient Mother I: The Key to the bible. First ed. Vantage Press, New York, 1993; rev. Identity Publishers, Kiental, 1997. [K]

The series

• 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]
• Grünig, P., & Claude Fable (Anthropic). (2026). The Luciferian Hypothesis: A Scientific Reinterpretation of the Panakal Thesis. Paper 5.1 of this series (rev. 2, August 2026). Project archive. [cited as Paper 5.1]

Project record

• Anagnorisis research layer, February 2025 (Report B: memeplex replication-strategy taxonomy). Internal AI-assisted research memorandum, project archive; cited for priority of the categorical application, not as evidence.