ANAGNORISIS

Stratum 3 · Paper 1.2 rev. 2 · Full text

Reading options This is the complete paper — the series' mechanism paper, in its charter-governed revision. The wider frame at other grains: the series index or the Atlas (the claim structure at card resolution). Every external reference carries a verification record in the project archive — journal articles at fetch level, books at book level. The one source-corpus citation follows the project's adjudicated witnesses; printed copies are under physical inspection, and wording will be corrected to the printed state if it differs.

Stress-Sharing as Cognitive Glue in Ideological Collectives: From Biological Morphogenesis to Cultural Cohesion

Authors: Patrick Grünig · Claude Fable (Anthropic)

Status: Paper 1.2 of ten — the series’ mechanism paper; builds on Paper 1.1

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 (§6). 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. One passage (§2.4) cites 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. The citations do two jobs: attributing an idea to its source, and disclosing that the idea precedes this framework. The model of Section 2 stands as stated without them.

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 this paper is titled after — a chimera assembled from real bibliographic parts — is replaced by the two real publications it pointed toward, at their register; one confabulated reference is deleted; one real reference cited for content it does not contain is replaced by the deconversion research program it was reaching for; a temporal-order claim is withdrawn to prediction status; the digital-amplification prediction is rebuilt against the field experiments that tested its familiar form and did not find it; the mathematics is restated as bookkeeping rather than derivation; two Levin citations are re-pointed to the documents that contain the material; and an unfilled cross-reference placeholder is resolved. The body makes no reference to prior drafts; the change record lives in the project archive (the claim table and the critic file). This note is the history’s single, removable location.

Abstract

Biological collectives, from cell networks to swarms, have been proposed to maintain coherence through shared stress: in Michael Levin’s TAME framework, stress is systemic setpoint error, and its propagation across sub-agents is the influence that binds them into a larger cognitive self. An agent-based model has since formalized the proposal, with stress-sharing populations achieving anatomical targets faster than populations without it (Shreesha & Levin, 2024). This paper extends the mechanism to cultural collectives as its own hypothesis: shared emotional stress (guilt, fear, shame, outrage) is the binding signal of ideological communities, with shared ecstasy as the loop’s reward pole. We state a conceptual cohesion model relating collective cohesion to stress load and narrative density; decompose the stress load into baseline (installed at entry), induced (ritually renewed), and reactive (threat-driven) components; and identify the dependency loop (a relief mechanism that re-sensitizes hosts to the condition it relieves) as the architecture that most reliably keeps shared stress above zero. The framework’s central original account concerns exit: leaving is computational boundary contraction, which is why departure from high-demand systems presents as grief rather than belief revision. Six predictions and four study designs follow, including a deliberately narrowed digital prediction: not that echo-chamber exposure causes radicalization, an effect the field’s largest experiments have not found, but that affective density within already-selected communities drives collective identification and action. The model’s mathematics is offered as bookkeeping, its dynamics as hypotheses, and its instruments as work owed.

Keywords: stress-sharing, collective intelligence, cognitive glue, memeplex, ideological cohesion, computational boundary, dependency loop, deconversion, quorum sensing, affective polarization, TAME

1. Introduction

1.1 The Binding Problem of Collective Cognition

Why do people stay in ideological communities that appear to cause them distress? Why do collectives of believers, partisans, or followers sometimes act with a coordination and purpose that none of the individual members planned? And why does leaving such communities — even when the individual has intellectually rejected their claims — feel less like changing one’s mind and more like losing a part of oneself?

These questions point to a shared puzzle: the binding mechanism of ideological collectives. What holds these groups together is not shared belief alone (people share beliefs without forming coherent collectives) or social incentive alone (many communities offer belonging without ideological rigidity). Something more basic is at work — a mechanism that merges individual cognitive agents into a collective that computes, decides, and acts as a unit.

This paper proposes that the binding mechanism is shared stress.

1.2 Biological Precedent: Stress as the Glue of Agency

The concept comes from developmental biology, and its source is specific: a box in Levin’s TAME paper titled “Stress as the glue of agency” (Levin, 2022, Box 1). Stress, in Levin’s usage, is not threat or suffering but systemic setpoint error: “a systemic response to a difference between current state and a desired setpoint.” The box proposes that this error signal is what scales small selves into larger ones. A dissatisfied sub-agent “propagates its unhappiness outward”; the signal is hard for neighboring units to ignore because it rides on conserved machinery (“my internal stress molecule is the same as your stress molecule”), which erodes the ownership of information across the collective; and the propagating stress recruits distant units into one homeostatic loop aimed at a shared setpoint. Two further features of the proposal matter for everything below. First, the channel is computational, not just adhesive: “stress pathways serve the same function as hidden layers in a network” — the shared-stress traffic is where the collective’s processing happens. Second, the coupling runs both ways: networks scale stress, and stress scales the network’s agency, “a bidirectional positive feedback loop.”

The proposal now has a computational formalization. In the multiscale agent-based model of Shreesha and Levin (2024), simulated cells pass their homeostatic-error signal to neighbors during collective morphogenesis. Populations that shared stress achieved their anatomical targets faster than populations that did not (“enhanced morphogenetic efficiency,” in the authors’ phrase), and sharing extended each cell’s influence over the fates of distant cells; the authors read their results as “supporting the hypothesis that stress sharing increases collective cohesiveness.” Stress-sharing as the binding mechanism of biological collectives is, then, a proposal within TAME, formalized and supported in silico; in living tissue it remains a hypothesis. The extension to cultural systems undertaken here inherits that register and adds a further layer of hypothesis on top.

What of the converse, that severing the shared signal dissolves the collective? Levin’s instances are two real phenomena and one stated speculation, and their registers differ. In carcinogenic transformation, cells “become isolated from the physiological signals that bind them into unified networks” and “revert to their unicellular past,” pursuing single-cell goals against the body plan; the isolation is literal, “a shutdown of gap junction synapses” (Levin, 2019, p. 6). Under anaesthesia the isolation is induced: “most anesthetics used to remove cognition and sensory experience, whether in plants or animals, are gap junctional (bioelectric) uncouplers” (Levin, 2019, p. 11) — cutting the communication is, on Levin’s reading, what removes the larger cognition. And as speculation, Levin proposes that erasing a collective’s setpoint would dissolve “the essential glue that creates a cognitive Self,” leaving “the constituent parts (smaller Selves in their own right)” intact and free — “completely different from killing the individual components” (Levin, 2019, p. 18). No experiment blocks a stress signal and watches a body fall apart. The fragmentation claim, in biology as in what follows, is an inference from pathology, from pharmacology, and from a hypothetical the source states as such.

The mechanism sits inside the hypothesis of scale-free cognition (Levin, 2019): cognitive vocabulary applies wherever information is processed and goals are pursued, at any scale, and what varies between a cell and an organism is not the kind of process but the size of the self that runs it. The size is set by a boundary. Any self, in Levin’s definition, “is demarcated by a computational surface – the spatio-temporal boundary of events that it can measure, model, and try to affect” (Levin, 2019, p. 1). The reach of a self’s binding signals sets that boundary’s extent: couple more units into the signal traffic and the self grows; cut the traffic and it shrinks. Section 3 imports this boundary concept whole.

1.3 The Extension to Cultural Systems

We propose that this biological mechanism has a cultural analogue. Ideological collectives — religious communities, political movements, cults, nationalist formations — maintain coherence through shared emotional stress: guilt, fear, shame, moral outrage, existential anxiety. These states function as the stress signals that synchronize individual behavior and merge individual computational boundaries into a larger collective self.

The parallel is structural. The same functional logic is proposed to operate in both domains:

  1. Individual components (cells / people) possess autonomous cognitive capacity.
  2. Shared stress signals (bioelectric / emotional) synchronize component behavior.
  3. The shared stress expands each component’s computational boundary, merging individuals into a larger cognitive unit.
  4. The collective exhibits goal-directed behavior that no individual component planned — goal-directed in the operational, model-relative sense defined in Paper 1.1 (§2.4).
  5. Withdrawing the shared stress predicts fragmentation back into autonomous individuals — in the biological case an in-silico contrast and an inference from pathology (§1.2), in the cultural case this paper’s dissolution prediction (§3.3, P2).

The bridge is tighter than a loan of vocabulary. Each of the proposed cultural stress states is, in its phenomenology, a registered distance between a current state and a required one: guilt is moral-setpoint error (the state one is in, against the state one is required to be in), fear of damnation is anticipated terminal error, shame is social-setpoint error before the community’s eye. The doctrine specifies the setpoint; the emotion measures the distance; the community shares the measurement. That is stress in Levin’s systemic sense — setpoint error propagated across units — instantiated in affect and narrative rather than in physiology.

The biological half of the analogy is a proposal with computational support (§1.2); the cultural half is this paper’s hypothesis, and nothing below treats it as settled. The framework’s author has since, with a collaborator, carried the concept to one social-scale system, identifying “the price system as the cognitive glue of the economy” (Lyons & Levin, 2024, a preprint cited at abstract strength); the extension of the stress-sharing mechanism to ideological collectives remains this paper’s. Section 2 develops the model, Section 3 its boundary dynamics, Section 4 runs it against documented patterns, Section 5 states its formal properties, Section 6 derives predictions and study designs, Section 7 positions the model among existing frameworks, Section 8 states its limitations, and Section 9 concludes.

2. The Cohesion Model

2.1 Basic Framework

We define the cohesion of an ideological collective as the degree to which its members behave as a coordinated unit rather than as independent agents — operationalized, where measurement is proposed (§6), as behavioral synchronization, self-sacrifice for the group, and resistance to defection. We write cohesion as a function of two variables:

C = f(σ, I)

where σ (stress load) is the aggregate shared emotional stress experienced by members, and I (information density) is the density and coherence of the shared narrative through which stress is experienced as collective rather than individual.

What the notation does and does not deliver should be stated at once. We do not specify f; no functional form is derived anywhere in this paper. The equation is a compact name for a structured qualitative claim: cohesion rises with shared stress over a bounded range, given sufficient narrative density. The value of the formalism at this stage is bookkeeping — it names components and their relations so that predictions can be addressed to them. A formal treatment is deferred (§8.5).

Both variables are proposed as necessary. Shared stress without shared narrative produces panic, not cohesion: a crowd fleeing a fire shares stress but does not form a cognitive collective, because no interpretive frame binds the individual alarms into one representation. Shared narrative without shared stress produces agreement without binding force: academic communities share dense frameworks but rarely show the costly collective behavior characteristic of ideological groups. The claim is that stress supplies the binding energy and narrative supplies the coherence, and that only the conjunction yields a collective that computes.

The central hypothesis: C increases with σ within a viable range, and the relationship is bounded. Below a minimum threshold (σ_min), binding force is insufficient and the collective fragments. Above a maximum threshold (σ_max), stress overwhelms coping and produces collapse — mass defection, breakdown, or the collective’s self-destruction. Section 5.1 develops the resulting three-region landscape.

2.2 The Stress Load Function

We decompose the total stress load into components:

σ_total = σ_baseline + σ_induced + σ_reactive

σ_baseline is the stress installed at initiation into the system. The type case is the doctrine of Original Sin, which specifies a moral deficiency present in every human from birth; whatever else the doctrine does, it sets σ > 0 at t = 0 for each member, so binding force exists from the moment of entry, prior to any act. The biological precedent for entry-time installation is Box 1’s closing observation that stress spreads “not only horizontally in space (across cell fields) but also vertically, in time” — stress responses are among the traits most readily transferred across generations (Levin, 2022, Box 1). σ_baseline is the cultural form of vertical spread: a deficit state transmitted at entry, and for hereditary membership at birth, before any individual transgression exists.

σ_induced is stress generated through ongoing participation: sermons on human sinfulness, confession rituals that reactivate guilt, apocalyptic preaching that renews existential fear.

σ_reactive is stress generated by perceived external threat: persecution narratives, culture-war rhetoric, demonization of out-groups. This component rises sharply under threat, and the model takes the widely described cohesion surge of threatened communities to be this component at work (§4.2; P3).

2.3 The Stress-Relaxation Dynamic

Stress decays when not maintained: a guilt episode fades, a fear subsides, an outrage dissipates. Left to natural dynamics, σ would relax toward zero and the collective would gradually lose cohesion. Durable ideological systems counteract the decay through periodic stress-renewal mechanisms:

As bookkeeping, not dynamics:

σ(t+1) = σ(t) − R(t) + ε(t)

where R(t) is natural relaxation plus any relief the system administers, and ε(t) is renewal from ritual, narrative, and social reinforcement. The design feature of durable stress-maintenance architectures is that ε remains positive at all times: renewal never lets stress fully relax. The result is a permanent state of managed stress. How a system guarantees ε > 0 leads to its most refined architecture.

2.4 The Dependency Loop

A particularly effective stress-maintenance architecture is the dependency loop: a cycle in which the system’s own relief mechanism (confession, redemption, absolution) provides temporary reduction of stress while reactivating the conditions for its return.

The canonical form:

  1. Stress activation: the member experiences guilt, fear, or shame (σ rises).
  2. System-mediated relief: the member engages the system’s relief mechanism — confession, prayer, ritual purification, recommitment (σ falls, temporarily).
  3. Residual stress: the relief is incomplete by design, because the underlying condition (inherent sinfulness, permanent moral inadequacy) is defined as incurable by the member’s own efforts.
  4. Re-sensitization: the relief mechanism itself renews awareness of the condition — confessing sin requires contemplating sin; seeking redemption requires rehearsing the need for it.
  5. Return to step 1: stress rebuilds, driving the member back to the system for relief.

The cycle is self-sustaining: the “cure” perpetuates the “disease.” From a dynamical-systems standpoint the loop describes an oscillation around a non-zero stress level — the shape of a limit cycle. We state that as the hypothesis the loop’s instrumentation would test (P6; Paradigm B), not as a derived property: no equations of motion have been written here, and the formal question of which relief operators cannot reach the baseline is deferred (§8.5).

Why can relief not empty the reservoir? Here the model reaches its one theological input. The floor under σ is doctrinal, not dynamical: in the type case, the deficit is defined as beyond the host’s own means, so no quantity of relief-seeking touches it. The observation that this incurability is by construction — a debt unpayable because imagined — is Leo Panakal’s, stated in 1998, decades before this framework: “As guilt is irredeemable because imagined, so is its inescapable counterpart of christian sin irredeemable” (PRE, pp. 23–24); the Key grounds the sin concept’s defining property in guilt’s (K, ch. VI), and the corpus’s textual derivation (the Genesis 4:7 collation) is presented in Paper 5.1 (§5.4). For the model the attribution is corroborative rather than structural: a reader who sets the corpus aside may take the floor as this paper’s stated axiom. The guarantee σ_steady-state > 0 then holds by construction for any system whose doctrine specifies an unpayable deficit; whether a given system’s doctrine does so is a reading exercise, and whether its hosts actually cycle around the floor is the empirical half (P6).

The loop’s form is not unique to religious systems. Addiction cycles (the substance relieves a withdrawal the substance causes), coercive-relationship cycles (reconciliation relieves a tension the abuser causes), and debt servitude (partial repayment relieves a pressure the interest schedule renews) share the structure: relief administered by the source of the pressure, incompleteness by design. We offer the triple as an observation of common form, not as a documented equivalence.

3. Computational Boundary Dynamics

3.1 The Self as Variable Boundary

Levin’s concept of the computational boundary supplies the mechanistic link between shared stress and collective cognition. A self’s boundary, in the definition §1.2 imported, is the envelope of what it can measure, model, and try to affect; Paper 1.1 (§2.2) develops the biological case. When cells couple into gap-junction networks, “the Umwelt expands to create a larger individual in which the comprising cells share a unifying picture of the world” (Levin, 2019, p. 11): each cell now measures and responds to conditions across the connected network, and the tissue becomes a larger self.

We propose the analogous process in ideological collectives:

Individual boundary (B_individual): the range of concerns, values, and goals a person monitors and acts on when operating autonomously.

Collective boundary (B_collective): the range of concerns, values, and goals shared across the community through common stress, narrative, and ritual.

Boundary expansion: when an individual joins the collective and begins sharing its stress signals — internalizing its guilt, its fears, its narratives — their computational boundary expands:

B_effective = B_individual ∪ B_collective

The member begins monitoring and responding to the collective’s concerns — heresy, external threat, doctrinal purity, recruitment — as personal concerns. They compute as a node in the larger system.

The join needs no altruism and no designer. In Levin’s account of how cells become tissues, apparent cooperation is built “from selfish agents minimizing their stress (surprise) and competing for information”; each unit expands its measurement boundary through communication with neighbors “and thus inevitably becomes part of a bigger self with bigger set points serving as homeostatic attractors. It only looks like cooperation from a perspective of a higher level” (Levin, 2019, p. 14). The cultural join has the same shape. The individual arrives already stressed (grieving, guilty, anxious, meaning-starved), and membership relieves and informs; participation is locally rational at every step. The collective’s goals are what the merged setpoints jointly constitute, chosen by no one. This answers in advance the objection that the model credits collectives with purposes nobody has: on the source framework’s own account, neither does anybody in a tissue.

3.2 Why Leaving Feels Like Self-Amputation

This model explains a documented but under-explained phenomenon: the psychological severity of leaving a high-demand ideological group. If disengagement were belief revision — updating on evidence — it should be uncomfortable but manageable, like correcting a factual error. It is not. People leaving high-demand religious groups and intense ideological communities report grief, identity dissolution, and existential crisis, in clinical descriptions and in the deconversion research literature (Winell, 1993, 2011; Streib et al., 2009; Streib, 2021). Within this framework, the reason is that leaving is not only a cognitive event but a boundary contraction:

B_effective → B_individual (as B_collective is severed)

The individual is not merely changing their mind; they are losing part of their cognitive self. Concerns, values, and purposes that were part of their expanded boundary are amputated. The grief is real because the loss is real — not a loss of facts, but a loss of self.

The contraction has a second dimension the spatial picture misses: the self shrinks in time as well as in reach. In Levin’s cancer case, the isolated cell’s “time horizon of activity shifts from decades … to a much shorter time frame”; contraction “reduces their temporal horizon of concern” (Levin, 2019, p. 6). The ideological case mirrors it. Membership placed the member’s actions on the collective’s timescale — eschatological, generational, historical; exit collapses the horizon to a biographical one. An ex-member describing life after leaving as suddenly small, flat, or pointless is describing temporal contraction, not only social loss: the events their actions used to mean something on no longer exist for them.

3.3 Conditions for Boundary Separation

Under what conditions does the individual boundary separate from the collective without catastrophic loss? The model identifies three:

  1. Stress attenuation. If the member’s experienced stress is reduced — through therapy, alternative community, or intellectual deconversion — the signal binding them to the collective weakens, and the boundary begins to contract.

  2. Alternative boundary expansion. If the member forms new connections (secular community, therapeutic relationships, other meaning-making frameworks) that expand their boundary in new directions, the contraction is compensated rather than absolute.

  3. Gradual rather than abrupt disconnection. Abrupt severance without alternative expansion predicts the worst outcomes: the boundary contracts with nothing to expand into, leaving a drastically reduced cognitive self. This is why the model expects members expelled from groups to fare worse than members who leave gradually with support — a prediction-grade claim, stated for Paradigm B to test.

Condition 2 has an empirical instance already on record. The Bielefeld cross-cultural deconversion program (matched-control, mixed-method, run in Germany and the United States) reports that German deconverts showed more emotional instability and social difficulty after exit than American deconverts, and attributed the contrast to the denser landscape of religious alternatives available to Americans (Streib et al., 2009). The model reads that contrast mechanistically: where substitute expansion is available, contraction is buffered; where it is not, exit is amputation without prosthesis.

One clarification, prompted by the biological source itself, sharpens all three conditions. Levin’s dissolution condition is not stress removal but setpoint erasure: “By erasing the set point toward which the feedback loop expends energy to accomplish, the higher level integrated Self disappears, leaving nothing but the constituent parts” (Levin, 2019, p. 18). Applied here, the distinction separates two things the exit literature tends to merge. Attenuating the stress (condition 1) weakens the binding signal; releasing the setpoint — ceasing to hold the goal-state the doctrine installed, be it sinlessness, salvation, or purity — dissolves the participation itself. The distinction accounts for a documented and otherwise puzzling phenomenon: intellectual deconversion without relief. A host can stop believing and keep hurting, because disbelief revises propositions while the installed setpoint persists as a felt requirement and keeps generating error signal. Clinical descriptions of guilt and fear persisting for years after exit (Winell, 1993, 2011) are, on this account, the signature of a setpoint outliving the membership it served. Paradigm B measures belief, stress, and setpoint retention separately for this reason.

The scope of the grief account is bounded by the same calibration Paper 1.1 carries: in a longitudinal study of roughly 20,000 Dutch adults, 450 of whom deconverted, ordinary religious disaffiliation shows no average change in well-being (Bleidorn et al., 2024). Boundary contraction of the kind described here is a phenomenon of high-demand, high-σ architectures, systems in which the boundary had expanded far and the stress ran deep — not of religious change as such.

4. Comparative Evidence

The model was built to explain a family of documented patterns; this section runs it against them. The comparisons are consistency checks against described phenomena, not tests — the tests are Section 6’s.

4.1 Religious Communities

Strictness and strength. Within the sociology of religion, the observation that demanding churches thrive while lenient ones shrink has half a century of standing: Kelley (1972) documented conservative growth against lenient decline; Iannaccone (1994) recast the pattern as strictness and supplied a rational-choice mechanism, on which strictness screens out free riders and raises average participation; and the dispute remains live, with Marwell (1996) objecting that the data establish neither the pattern’s strength nor its cause. The cohesion model enters this dispute as a third mechanism rather than a bystander: on the present account, what strictness delivers is neither seriousness (Kelley) nor screening (Iannaccone) but stress: behavioral demands, judgment, and boundary policing all raise σ_induced. An unresolved dispute over mechanism is the natural opening for a mechanism-level proposal.

The guilt sharpening. The model’s own sharpening of the pattern is a hypothesis the strict-church literature does not test: that σ installed as guilt — an unpayable deficit in the self, rather than a demanding practice — binds hardest, because it operates continuously and travels with the host. Traditions built on inherent sinfulness and external redemption should, on this account, sustain tighter cohesion than traditions built on self-inquiry and internal authority, at matched levels of behavioral strictness. That comparison is Paradigm D’s; until it is run, the claim is a prediction with adjacent support, not a finding.

Revival as reactive intensification. When religious communities face decline, revival movements characteristically intensify guilt, fear, and apocalyptic messaging; the Great Awakening, twentieth-century Islamic revivalism, and contemporary evangelical renewal movements fit the shape. Revival, in the model’s terms, is σ_reactive engineering: stress renewal escalated in response to collective threat (P3’s historical instance, offered as consistency, not confirmation).

Deconversion. What the deconversion literature documents, and what it does not, divides cleanly. Documented: the exit phenomenology of §3.2 — grief, identity disruption, and long residuals in departures from high-demand systems (Winell, 1993; Streib et al., 2009). Not documented: the temporal order the model predicts, in which resolution of guilt and fear precedes and facilitates disengagement. The clinical literature mostly records the reverse sequence, stress outliving exit (§3.3), and no study known to this project has measured the order prospectively. The model therefore does not cite deconversion as confirmation of its causal claim; it stakes P2 on it.

4.2 Political Movements

The model is not specific to religion; its political cases are the classic subject matter of intergroup psychology.

Nationalist movements sustain cohesion through shared threat narratives: the external enemy generates σ_reactive that binds the collective, and the model expects cohesion to track the threat narrative’s vividness. Social identity theory supplies the frame in which such dynamics are standardly treated (Tajfel & Turner, 1979); Brewer’s (1999) caution that ingroup love and outgroup hate are dissociable — attachment to us does not automatically produce aggression toward them — marks the joint where this model adds a variable. σ_reactive is the proposed coupler: threat-stress converts in-group attachment into out-group hostility, and where σ_reactive stays low the two stay uncoupled. That conversion claim is the political form of P3.

Conspiracy communities sustain cohesion through shared paranoia. The belief that hidden forces work against the group generates persistent σ_reactive, and the epistemic closure of conspiracy belief — counter-evidence read as further evidence of the conspiracy — prevents the stress from being reduced by ordinary epistemic means. Structurally, that is a dependency loop built out of epistemology: the interpretive frame that generates the stress also disables its resolution.

Revolutionary movements exhibit escalation-release cycles: accumulated shared grievance (σ_induced) builds cohesion; the revolutionary act discharges it; and the aftermath — the described listlessness and fragmentation of post-revolutionary movements — is what follows, on the model, when σ collapses and the binding force goes with it. This is a consistency reading of a described historical pattern, not a measured result.

4.3 High-Demand Groups

The model aligns closely with the classic analyses of coercive control, and the alignment is a re-derivation, not a competition.

Lifton’s eight criteria of thought reform (1961, ch. 22) — milieu control, mystical manipulation, demand for purity, cult of confession, sacred science, loading of language, doctrine over person, dispensing of existence — can be re-read as a stress-maintenance architecture. Each criterion either installs stress (demand for purity: σ_baseline), renews it (cult of confession: ε), or blocks its relaxation (milieu control: suppressing the external inputs that would let σ decay). The re-analysis is this paper’s, not Lifton’s; what it adds to Lifton’s descriptive catalogue is a claim about function — the eight practices are one machine, and the machine’s product is σ held inside the viable band.

Hassan’s BITE model (2015) — Behavior, Information, Thought, and Emotional control — describes coercion at the individual level. The present model supplies the collective-level complement: the controlled individual is not only controlled but bound into a larger cognitive unit through shared stress, and the control mechanisms serve that binding function.

Exit counseling, as standardly described in the cult-recovery literature, concentrates on reducing guilt and fear, normalizing doubt, and building alternative community — the three levers §3.3 names. The model did not inform those practices; that they converge on its variables is consistency, recorded as such.

4.4 Non-Ideological Collectives: Sports and Corporate Culture

The binding pattern also appears where no doctrine is in play. Intense sports fandom runs on shared stress (anxiety over outcomes, outrage at officials, fear of relegation) punctuated by shared release, and produces recognizable collective identity and coordinated behavior. High-pressure corporate cultures that sustain shared anxiety (ranking systems, layoff fear, performance escalation) show stronger collective identification than relaxed workplaces, and the intense bonds formed in high-stress occupations (military units, emergency services, startups) are of the same kind the model describes.

These cases are offered as the mechanism’s cross-domain face, not as its confirmation. Their value is scoping: the content of the stress differs completely across the cases (theological, political, competitive, economic), so if the pattern survives measurement across domains (P1), what binds is the stress’s structural properties — shared, persistent, cyclically renewed — and not its subject matter.

5. Formal Properties of the Model

5.1 The Stability Landscape

The bounded-range hypothesis of §2.1 yields three characteristic regions:

Region I (σ < σ_min): insufficient binding. Ideas are shared but coordinated collective action does not arise. Loose intellectual communities, casual cultural affiliation, nominal religious identification.

Region II (σ_min < σ < σ_max): viable collective. Shared stress suffices to maintain cohesion without overwhelming coping. Goal-directed collective behavior, self-preservation, adaptive response. The dependency loop operates here, holding σ inside the band through periodic renewal.

Region III (σ > σ_max): overload. Stress exceeds coping capacity. Expected outcomes: mass defection (binding force turned aversive), collective breakdown (doomsday episodes), or a terminal hardening into authoritarian rigidity as the last coherence mechanism.

Two of the three regions have anchors in the source framework, and they anchor its two sides. Region II’s binding is Box 1’s: stress yoking sub-agents into one homeostatic loop (Levin, 2022). Region III, the point past which binding force becomes expulsive, is a possibility Levin himself raises for cells: stress “could be a factor that motivates cells to reduce their sensory/action surface and contract, in effect abandoning organismal membership when it begins to generate more stress than its presence reduces by its protective action” (Levin, 2019, p. 17). The stress economy has an exit condition, in cells as proposed here for hosts: membership must relieve more stress than it generates. The dependency loop of §2.4 appears, in this light, as a design solution to that constraint (it must keep relief credible while keeping resolution impossible), and σ_max marks where the credibility fails. The anchors supply the landscape’s two-sidedness, not its mathematics; §8.5 states what a derivation must deliver.

5.2 The Stress-Maintenance Problem

From the perspective of the collective as a cognitive agent — the operational, prediction-licensed framing developed in Paper 1.1 — maintaining σ inside Region II is a control problem. The collective must balance renewal (σ_induced) against natural relaxation to stay above σ_min, and reactive stress against coping capacity to stay below σ_max. This framing predicts that long-lived ideological systems are those that have evolved reliable stress-maintenance architectures — mechanisms, identifiable by §2.3’s renewal inventory independently of a system’s age, that keep σ in the viable band across changing environments, demographics, and histories.

The dependency loop is an elegant solution to this control problem because it is self-regulating: the relief mechanism that keeps σ below σ_max is the same mechanism whose re-sensitizing action keeps σ above σ_min. The loop holds the band without external tuning — which is what a control engineer would call good design, and what the rest of this series examines without assuming a designer (Paper 1.1, §2.4, on where agency-talk is licensed; §8.6 below on the ethics of the observation).

5.3 The Quorum-Sensing Analogy

In microbiology, quorum sensing is the process by which bacteria coordinate behavior on population density. Cells release signaling molecules, and when local concentration crosses a threshold, collective behavior switches on: biofilm formation, virulence, antibiotic production (Miller & Bassler, 2001; Waters & Bassler, 2005). Two features of the reviewed biology transfer to the cultural case. First, the threshold’s rationale — quorum-controlled processes are, in the review’s words, “unproductive when undertaken by an individual bacterium acting alone but become beneficial when carried out simultaneously by a large number of cells” (Waters & Bassler, 2005). Second, the sharpness of the transition — though “phase transition” is this paper’s description, not the microbiologists’.

We propose the analogous dynamic in ideological collectives. Individual members express stress; when the local density of expressed stress crosses a threshold, collective behavior triggers: coordinated worship, mobilization, moral panic, collective aggression. The bacterial rationale carries over directly: moral panic and mass mobilization are the class of actions that are futile for one and effective for many. The analogy yields three corollaries:

  1. Density dependence. Collectives should be more cohesive where stress signals propagate efficiently (congregations, rallies, residential communities) and less cohesive where members are dispersed (P4).

  2. Media as signal amplification. Communications technology functions as an artificial amplifier for stress signals, letting collective behavior trigger at physical densities that would never reach quorum unaided. This is the model’s account of why ideological collectives that once required co-location can now form rapidly online. Törnberg’s (2018) simulation gives the corollary a formal precedent: a polarized cluster acts as the seed bandwagon for complex contagions (content requiring multiple exposures spreads from the dense cluster outward), which is density-gated collective behavior in an information network.

  3. Engineered quorum environments. Algorithmic curation can concentrate stress-expressing content, raising perceived density above the population’s real prevalence — an artificial quorum. What such concentration does and does not do to hosts is among the most heavily tested questions in computational social science, and P5 (§6.1) states this model’s prediction against the field’s results.

6. Predictions and Experimental Paradigms

6.1 Core Predictions

P1–P3 restate, at this paper’s finer grain, the mechanism predictions stated at series level in Paper 1.1 (its P1–P3); P4–P6 are specific to the density and loop structure developed here.

P1 (Stress–cohesion correlation). Across ideological communities, measured shared emotional stress (guilt, fear, shame, outrage) should predict behavioral cohesion (self-sacrifice, conformity, resistance to defection, collective action), controlling for group size, resources, and social incentives.

P2 (Therapeutic dissolution). Individuals whose guilt, fear, and shame are resolved without simultaneous integration into an alternative community should show progressive disengagement from the collective — resolution preceding exit, the temporal order §4.1 identifies as undocumented. The setpoint distinction of §3.3 sharpens the prediction: attenuation of stress with retention of the doctrinal setpoint predicts partial disengagement with residual suffering; setpoint release predicts full boundary separation. Interventions that resolve stress while providing alternative community should accelerate both.

P3 (Threat intensification). Communities facing existential threat (membership decline, marginalization, legal challenge) should measurably intensify stress-signaling (guilt and fear rhetoric, apocalyptic messaging, out-group demonization), followed by increased cohesion among remaining members.

P4 (Density dependence). Cohesion should be higher among members participating in regular in-person gatherings than among dispersed members consuming identical content, controlling for belief intensity and commitment duration.

P5 (Affective density). The familiar form of the digital prediction, that echo-chamber exposure causes radicalization, is not this model’s prediction, and the field’s strongest results are the reason. The two largest field experiments on the question changed exposure substantially and moved no measured attitude: replacing algorithmic ranking with chronological feeds for three months during a U.S. election did “not significantly alter levels of issue polarization, affective polarization, political knowledge, or other key attitudes” (Guess et al., 2023); reducing exposure to like-minded sources by about a third for 23,377 Facebook users had “no measurable effects on eight preregistered attitudinal measures” (Nyhan et al., 2023). Trajectory data find little evidence of the recommendation algorithm driving consumption of radical content — the pattern favors demand (Hosseinmardi et al., 2021). Cross-cutting exposure can even harden positions (Bail et al., 2018) — an asymmetry this model interprets as σ_reactive behavior, the out-group’s voice arriving as threat. A systematic review of 129 studies finds the answer tracks the method: network-structure studies find echo chambers, content-exposure studies find little (Hartmann et al., 2025). Echo chambers are real as network structures (Cinelli et al., 2021) and rarer as habitats than the public debate assumes — the Reuters Institute’s review puts partisan echo chambers at roughly 2–5% of the UK public (Ross Arguedas, Robertson, Fletcher & Nielsen, 2022; Dubois & Blank, 2018).

What those experiments manipulated, however, was ideological composition (like-minded versus cross-cutting sources, algorithmic versus chronological order). None manipulated affective density: the concentration of outrage, fear, and grievance in what a community’s members see, which is the variable this model’s §5.3 actually names. On that channel the evidence runs the other way. Moral-emotional language increases a message’s diffusion “within (and less so between) ideological group boundaries” (Brady et al., 2017) — in this model’s reading, a stress signal propagating preferentially inside a boundary; Brady’s own construct is moral-emotional language, not measured stress, and the identification is the model’s. And social feedback measurably amplifies future outrage expression, with users conforming to their network’s expressive norms (Brady et al., 2021: 12.7 million tweets observed, plus preregistered experiments). In the model’s terms, that is ε sustained by reinforcement learning, with no designer required.

P5 is therefore stated narrowly: within an already-selected community, raising the density of stress-expressing content raises collective identification and willingness to act collectively, with attitude positions free to stand still. The nulls above leave the narrowed claim standing almost untouched — they varied composition and measured attitudes — and Brady’s results support its premise; Paradigm C is its direct test. The “almost” must be recorded: in Guess et al. (2023), the chronological feed reduced exposure to uncivil content on Facebook while attitudes stayed flat. There, incivility density was a by-product rather than the manipulated variable, and attitude scales are not measures of identification or mobilization — but the result stands as the nearest existing brush with the narrowed claim, and Paradigm C must beat it, not ignore it.

P6 (Dependency-loop signature). Systems whose relief mechanisms re-sensitize (confession, redemption rituals, purification) should show more stable long-term cohesion than systems whose relief mechanisms genuinely resolve — testable by comparing membership retention curves of loop-bearing and loop-free communities.

6.2 Proposed Experimental Paradigms

Paradigm A: Stress–cohesion measurement. Cross-sectional survey across ideological communities (denominations, political organizations, wellness movements). Stress via existing instruments: the Religious Comfort and Strain Scale (Exline, Yali & Sanderson, 2000) and its successor the Religious and Spiritual Struggles Scale (Exline et al., 2014) — the closest existing instruments to σ, though built to measure individual strain rather than its shared, cyclically renewed character (§8.1); the Intolerance of Uncertainty Scale (Freeston et al., 1994; short form Carleton et al., 2007); the Guilt Inventory (Kugler & Jones, 1992). Cohesion via behavioral indicators: attendance, contribution, network density, willingness to sacrifice personal interests for the group. Prediction: stress measures predict cohesion measures after controlling for belonging, doctrinal agreement, and demographics (P1).

Paradigm B: Longitudinal deconversion study. Prospective study of individuals leaving high-demand groups, with the Bielefeld study’s matched-control design as ancestor (Streib et al., 2009). Measure guilt, fear, and shame; behavioral engagement; identity salience; and, as §3.3 motivates, setpoint retention: whether the host still holds the doctrine’s goal-state as a felt requirement, independent of stated belief. Predictions: stress reduction precedes behavioral disengagement (P2), and setpoint release, not attenuation alone, predicts full separation; disbelieving hosts who retain the setpoint remain bound or symptomatic.

Paradigm C: Affective-density experiment. Randomized trial with the design the existing null results dictate: hold the feed’s ideological composition constant; vary only affective density (the proportion of outrage-, fear-, and grievance-framed presentations of the same ideological content); take as outcomes collective identification and costly collective action (petitioning, donating, attending), not attitude scales. Measure participants’ whole media environment rather than a single platform, the known failure mode of narrow designs (Dubois & Blank, 2018). Prediction: density raises identification and action-willingness at constant composition, and may leave attitude positions unchanged (P5). As redesigned, this is not a smaller replication of Guess et al. (2023) — composition is clamped, density is the treatment, and the outcomes are the ones the model cares about.

Paradigm D: Cross-tradition structural comparison. Compare traditions with high installed σ_baseline (inherent sinfulness, external redemption) against traditions with low σ_baseline (self-inquiry, internal realization) on cohesion, membership stability, and host cognitive flexibility. The styles axis has a candidate instrument in the Religious Schema Scale (Streib, Hood & Klein, 2010), whose “truth of texts and teachings” pole against its openness poles approximates the contrast; the flexibility outcome connects to an adjacent result already in hand, an association between religious disbelief and higher cognitive flexibility on three separate instruments (Zmigrod et al., 2019), and to the taxonomy predictions of Paper 1.1 (P8–P10), which state the marker-load version of this comparison. Prediction: high-σ_baseline traditions show higher cohesion and lower host flexibility; low-σ_baseline traditions the reverse — the binding/autonomy tradeoff.

7. Relationship to Existing Frameworks

7.1 Social Identity Theory

Tajfel and Turner’s social identity theory (1979) explains in-group cohesion through categorization, identification, and comparison. The present model complements it by specifying a binding mechanism that turns categorization into committed, costly collective behavior: SIT explains why people categorize; stress-sharing proposes what holds the category together once formed, and why some identities command sacrifice while others remain labels.

7.2 Terror Management Theory

Terror management theory (Greenberg, Pyszczynski & Solomon, 1986) proposes that awareness of mortality drives adoption of cultural worldviews. In the present model, mortality salience is one form of σ_reactive — existential stress binding individuals to meaning-providing collectives. TMT names one source of binding stress; this model generalizes the mechanism to the full family of shared stress states, of which mortality awareness is a member.

7.3 Durkheim’s Collective Effervescence

Durkheim (1912/1995) described the heightened collective energy of ritual, “collective effervescence,” as the wellspring of solidarity and shared belief. The present model splits Durkheim’s observation into two poles and supplies the connective between them. The effervescent event itself is the reward pole of the binding loop: shared ecstasy, the positive channel Paper 1.1 (§4.2) marks as the framework’s extension beyond stress proper (phenomenologically, the felt merger of boundaries). What Durkheim’s account leaves open is persistence between assemblies: why solidarity survives the crowd’s dispersal. The model’s answer is the tonic layer: σ_baseline and the renewal cycle maintain binding between effervescent peaks, so that ritual both renews stress (ε) and pays reward (effervescence), the loop’s two strokes housed in one institution. Effervescence without a stress floor yields festivals; a stress floor without effervescence yields grim compliance; on this model, the durable architectures run both.

7.4 Attachment Theory and Grief

The experience of leaving an ideological community (grief, identity confusion, existential anxiety) parallels attachment disruption as described by Bowlby (1969), whose account of grief proper occupies the third volume of the trilogy (Bowlby, 1980). The model proposes that the parallel is mechanistic and not only phenomenological: both are the subjective face of a computational boundary contracting after it had expanded to include another — a caregiver in the one case, a collective in the other. §3.2 is, in this sense, attachment theory’s loss account carried to the collective scale.

8. Limitations and Future Directions

8.1 Operationalization

The model’s variables (σ, I, C, B) need instruments. Those named in Paradigm A measure individual strain, uncertainty intolerance, and guilt; none measures what the model actually posits: the shared and cyclically renewed character of the stress. The Exline lineage (Religious Comfort and Strain Scale; Religious and Spiritual Struggles Scale) is the nearest existing work and a natural base to build from; a stress-sharing instrument proper, capturing synchrony across members and renewal across time, is the framework’s first methodological debt.

8.2 Causal Direction

The model predicts that shared stress produces cohesion. The reverse direction — membership producing shared stress — is also plausible, and both plainly operate together. The paradigms of §6.2 are designed to isolate the stress-to-cohesion direction (P2’s temporal order; Paradigm C’s manipulation), but observational versions of these designs cannot settle direction, and §4’s consistency readings do not pretend to.

8.3 Individual Differences

The model treats members as homogeneous nodes; people are not. Need for closure, tolerance of ambiguity, attachment style, developmental history, and neurobiological reactivity all plausibly moderate stress-mediated binding. A fuller model would carry susceptibility as a distribution, not a constant — and the distribution itself may be part of what architectures select for.

8.4 Cross-Cultural Generalizability

The model was developed against Western religious and political material, and its base distinction between individual and collective cognition is itself culturally inflected. In strongly collectivist settings the “autonomous individual” pole of the boundary model may be the wrong zero point. Cross-cultural work — non-Abrahamic systems, indigenous traditions, non-Western political forms — is needed both to test the predictions and to find the model’s own parochial assumptions.

8.5 Formal Development

The mathematics of this paper is bookkeeping throughout: f is unspecified, the update equation names a competition without solving it, and the limit-cycle description of §2.4 is a hypothesis about dynamics, not a derivation. What a formal treatment must deliver is known: a specification of the relief operator R under which the σ_baseline term is unreachable — the mathematical form of a debt defined unpayable — and conditions under which the loop’s oscillation is stable rather than decaying or divergent. That treatment belongs to the series’ dependency-architecture paper (Paper 2.2) and is deliberately not duplicated here.

8.6 Ethics of Dual Use

A model of how stress binds collectives can inform interventions for people leaving coercive systems — and can equally inform the design of better coercive systems. We note the dual-use character without resolving it: it attends any real understanding of social cognition, and the remedy is governance and disclosure, not ignorance. This paper’s contribution to the disclosure side is its own transparency about mechanism.

9. Conclusion

This paper has carried a proposal from developmental biology, stress as the glue that scales selves, into a model of ideological cohesion. The claim is simple: shared emotional stress binds individuals into ideological collectives, and the architecture of stress maintenance determines the character of the bond. Systems that install a baseline deficit at entry, renew it cyclically, and route relief through mechanisms that re-sensitize produce the tightest binding; systems that leave stress resolvable produce looser, more autonomous membership. Stated so, the contrast is a claim about architecture with a prediction attached (Paradigm D), not a moral verdict — and not yet a finding.

What the framework offers is unification with a mechanism: religious commitment, political mobilization, coercive-group dynamics, deconversion grief, and digital outrage economies as one phenomenon — computational boundary dynamics under shared stress — linked across three levels, from individual emotional state through boundary merger to collective behavior. What it owes is on the table in Sections 6 and 8: instruments, temporal order, dynamics, and the narrowed digital test.

If the biological proposal holds, and if its cultural extension survives the tests stated here, then the oldest question about ideological life — why people stay, and why leaving costs so much — has a mechanical answer: the collective is not only something the member believes; while the boundary holds, it is something the member is. Understanding when that merger serves the people inside it, and when it serves only the pattern, is the question the rest of this series pursues.

References

External literature (each entry carries a verification record in the project archive — journal articles, the tests-database record, and the institute review at fetch level; books at book level; the report series at partial level; one preprint at abstract strength; where a record is partial, the entry says so)

The project corpus (presented in Paper 5.1; quotations per the project’s adjudicated witnesses)

The series