Position Statement

This chapter establishes the neural integration architecture through which Appetites (A), Intrinsic Motivation (I), and Mimetic Desire (M) combine to produce unified behavioural output, and identifies the epistemic constraints this architecture imposes on self-knowledge.

The chapter proceeds through six movements:

Movement A (Sections 9.1–9.5) establishes the static architecture: the Decision Hub as common-currency integration system, the AIM Integration Equation as functional description, composition and intensity as the two dimensions of motivated behaviour, neurodiversity as baseline assumption, and the Phenomenological Reality Principle that grounds AIM’s commitment to respecting source-signals as constitutive of felt experience.

Movement B (Sections 9.6–9.8) establishes the dynamic operation: the Decision Hub as continuously operating control system with feedback loops, Structural Restlessness as the emergent consequence of differential satiation dynamics, and habits as entries in the Decision Hub's cached-value store that bypass real-time integration for efficiency.

Movement C (Sections 9.9–9.11) establishes the epistemic constraints: Source Opacity as the architectural consequence of common-currency integration; the Aggregation Gradient as the structural complement that identifies the dimension along which M-detection survives within-Individual concealment; and the timing architecture that places Preconscious Transmission before conscious awareness.

Movement D (Section 9.12) establishes the structural consequence: Confabulation as the Thinker’s response to the information gap created by Source Opacity.

Movement E (Sections 9.13–9.14) establishes compensatory mechanisms and implications: I-Override as the deliberative pathway that achieves source-discrimination by inference, and the implications for self-knowledge including the special challenges posed by habitual behaviour.

Movement F (Sections 9.15–9.20) applies the integration to relationships: romantic and sexual relationships as the joint output of the Body Manager, the Observer and the Thinker, and the Relational Specificity Principle; the three Collaborator subtypes; friendship; kinship; extraction and disengagement, the failure path on the Mimetic Premium axis and the dyadic ground of the Push-Pull Obligation Principle; and divergence and estrangement, the failure path on the Belief-Register axis.

The chapter’s central finding is that Source Opacity is not a contingent limitation but an architectural necessity: the same common-currency integration that enables unified decision-making among incommensurable inputs is the mechanism that destroys source-tags. The Individual cannot have both integrated valuation and preserved source-information: the architecture permits one or the other, and evolution selected for integration. This constraint is permanent and cannot be overcome through willpower, insight, or training. What can be achieved is compensatory discrimination through inferential methods (the Two-Question Diagnostic) that reconstruct source-information from available evidence after integration has occurred.

Dependency Statement

This chapter depends on:

Chapter 2 for the canonical definitions of all capitalised terms
Chapter 4, Section 4.4.12 for the component-function localisations on which the chapter's architectural claims depend
Chapters 4–7 for the disaggregation argument and the established properties of Appetites (A), Intrinsic Motivation (I), and Mimetic Desire (M) as distinct motivational sources
Chapter 8 for the transmission mechanism of M (mirror-neuron systems, preconscious acquisition), Object Parasitism (M operates through A-I Objects, not independent M-Objects), and the Salience-Status Identity (Status is Salience aggregated across observers and averaged over time, derived at Chapter 8, Section 8.1.4)

(See Chapter 3 for the full treatment of OCEP as the scope-setting axiom for AIM, including the Completeness Criterion for General Theory Status and the Replacement Requirement.)

Methodological Note on Cited Evidence

This chapter cites empirical studies as the best available accounts of neural mechanisms underlying the outputs the AIM architecture must explain. The explanatory obligation is imposed by OCEP (Chapter 3): the human motivational system demonstrably produces common-currency integration, source-tag loss, and confabulation. These outputs require neural explanation. Where a cited study faces replication challenges, the relevant question under the Replacement Requirement (Chapter 3) is not whether the study is beyond dispute, but whether a superior account of the same output exists. If it does, AIM adopts the superior account. If it does not, the cited study remains the best available discharge of the obligation. The architecture of AIM depends on the outputs, not on any particular mechanistic study.

Derivation Preview

This chapter derives:

Source Opacity (Axiom 3) as architectural consequence of common-currency integration
Confabulation (Axiom 4) as structural response to the information gap Source Opacity creates
The Decision Hub as the neural locus of A-I-M integration
The AIM Integration Equation as functional description of common-currency valuation
The Phenomenological Reality Principle as normative foundation for respecting source-signals
Structural Restlessness as emergent consequence of differential satiation dynamics
Habits and the cached-value store as efficiency mechanism with compounded opacity effects
I-Override as compensatory mechanism operating through inferential reconstruction
I-Override Capacity as context-dependent variable and proper target of institutional design

Chapter Preview

9.1 The Decision Hub: Common-Currency Valuation: Establishes the neural architecture that integrates incommensurable motivational inputs into unified behavioural output.

9.2 The AIM Integration Equation: Provides the mathematical form of common-currency integration as functional description, not conscious calculation.

9.3 Composition and Intensity: Distinguishes the two dimensions of motivated behaviour: which sources contribute (composition) and how strongly (intensity).

9.4 Neurodiversity and Individual Variation: Establishes that the architecture is universal while the weights are individual; no two brains compute identical AIM profiles.

9.5 The Phenomenological Reality Principle: Establishes why AIM treats source-signals as paramount inputs deserving respect: they arrive involuntarily and constitute the Individual’s real felt experience.

9.6 The Decision Hub as Control System: Reframes the Decision Hub as continuously operating cybernetic architecture with feedback loops and error-correction.

9.7 Structural Restlessness; Derives the architecturally necessary condition whereby stable final satisfaction is unavailable, arising from differential satiation dynamics of A, I, and M.

9.8 Habits and the Cached-Value Store: Establishes habits as stored context-value associations that bypass real-time integration, with implications for System 1/System 2 dynamics and the limits of willpower-based intervention.

9.9 Source Opacity: The Architectural Constraint: Derives Source Opacity from the information-theoretic requirements of common-currency integration; establishes that source-tags are discarded, not merely obscured, and that the principal cost of the discarding is a category error rather than an anonymity, the Salience contribution being delivered under the only interpretive category the Common-Currency Scalar (the Scalar) supports. It closes by establishing the opacity of social valuation (Section 9.9.11), namely that valuing another rightly once certifies nothing about the contributions behind it, so a confident record is not self-certifying, and by establishing that Status enters valuation as a source-opaque input resting on no substrate of its own beyond the A and I it rides upon (Section 9.9.12).

9.10 The Aggregation Gradient as the Structural Complement to Source Opacity: Establishes that Source Opacity is a local claim (single Decision Hub, single moment) and identifies the Aggregation Gradient as the dimension along which M-detection survives within-Individual concealment at higher rungs of representational distance.

9.11 The Timing Architecture: Establishes that Preconscious Transmission (the Transmission Window: 60–340ms) precedes conscious awareness (300–500ms), with cached retrieval even faster, placing fundamental limits on interception.

9.12 Confabulation as Structural Consequence: Derives confabulation as the Thinker’s response to the information gap, with intensification under challenge and special severity for habitual behaviour.

9.13 I-Override Information Architecture: Establishes the compensatory pathway: what information survives integration, how the Two-Question Diagnostic operates, why vocabulary provision is architecturally necessary, and how cache-interruption enables I-Override for habits.

9.14 Implications for Self-Knowledge; Addresses the introspection illusion, the weakness of self-report, the superiority of behavioural signatures, and the double opacity problem for habitual behaviour.

9.15 Romantic and Sexual Relationships: Derives formation from the Minimum Installation Condition and the Body Manager, durability from the Thinker, and the Relational Specificity Principle, under which only the I-engagement whose Object is another Individual is specific to that Individual.

9.16 The Collaborator Subtypes: Derives the mating-pair, parent-child and among-equals readings of the Collaborator coordinate of the Agent-Type Vector and the channels each engages.

9.17 Friendship: Derives formation through co-presence, recovers the three classical kinds of friendship from the decomposition of valuation, and establishes friendship as the relationship most dependent on the I-component.

9.18 Kinship: Derives the one-directional and life-course-reversing Appetitive flow, attachment through care, parents as first Models, sibling rivalry as internal mediation, and what holds kinship where the I-component is weak.

9.19 Extraction and Disengagement: Derives the failure path on the Mimetic Premium axis, in which extracted Mimetic Premium erodes the I-component until the party extracted from disengages, and why parents disengage last.

9.20 Divergence and Estrangement: Derives the failure path on the Belief-Register axis, in which Beliefs load-bearing in either Self-Model diverge, and distinguishes divergence from extraction and from Conflict.

Movement A: Static Architecture

9.1 The Decision Hub: Common-Currency Valuation

9.1.1 The Integration Problem

The Decision Hub faces a computational problem that no engineering system has fully solved: how to compare incommensurable values on a common scale and select a single course of action.

Consider: an Individual is moderately hungry (A-signal), engaged in interesting work (I-signal), and aware that colleagues are leaving for a status-relevant lunch meeting (M-signal). These three inputs have no natural common unit. Hunger is measured in metabolic deficit; engagement is measured in what the work in hand still requires; status-relevance is measured in social-comparison position. Yet the Decision Hub must produce a single output: continue working, eat alone, or attend the meeting.

This is the integration problem. The AIM Framework’s answer is the Decision Hub: the neural architecture that converts incommensurable motivational inputs into a common currency of subjective value, enabling comparison and choice.

9.1.2 Neural Substrate: vmPFC and Ventral Striatum

Convergent evidence from neuroimaging, lesion studies, and single-cell recordings identifies the ventromedial prefrontal cortex (vmPFC) and ventral striatum (VS) as the primary loci of common-currency valuation.

vmPFC encodes subjective value across radically different reward types:

Primary rewards (food, water, temperature regulation): A-relevant
Information and learning opportunities: I-relevant
Social approval, status cues, conformity rewards: M-relevant

Critically, vmPFC does not maintain separate value-tracks for each type. Single neurons respond to subjective value regardless of source. A neuron that fires strongly for a highly-valued food reward will fire with similar intensity for a highly-valued social approval cue, if the Individual values them equivalently. The coding is source-blind: what matters is the integrated value, not where it came from.

Ventral striatum performs a complementary function, encoding reward prediction and motivational Salience. VS activity predicts approach behaviour across reward types and scales with expected value in the same source-blind manner as vmPFC.

Together, vmPFC and VS constitute the Decision Hub, the neural locus that integrates the signed common-currency scalars supplied by the Body Manager (A), the Thinker (I), and the Observer (M input channel) into a single signed integrated valuation that guides choice. The conversion of each source's native signal into a signed scalar is performed upstream within the input-bearing component, not at the Decision Hub itself.

9.1.3 Common-Currency Conversion

The Decision Hub does not receive A, I, and M signals in their native formats. Upstream processing converts each source’s signals into a format compatible with common-currency integration:

A-signals arrive via hypothalamic and interoceptive pathways. Hunger, thirst, temperature deviation, and tissue damage are encoded as deficit-magnitudes that modulate the value assigned to corrective stimuli. The classification of an Object in the field as a future threat is not an A-signal: it is a Thinker prior delivered as the Observer's signed Salience (Chapter 4, Section 4.4.2). A food stimulus is valued highly when the organism is hungry, minimally when sated: the same physical stimulus receives different common-currency values depending on A-state.

I-signals arrive via dopaminergic pathways from midbrain to striatum and prefrontal cortex. What the process the Thinker is running requires and does not yet have is encoded as a reward signal that modulates the value assigned to the Object through which the requirement can be met, and to continuing current activity or pursuing new opportunities. An Object the running process requires receives high common-currency value; an Object that holds no unresolved residue for Ground-Resolution to work on, and so produces Boredom, receives low value.

M-signals arrive via mirror-neuron systems and social-reward circuits. Observed Model behaviour, status cues, and social-comparison outcomes are encoded as value-modulations that inflate or deflate the worth assigned to Objects and actions. An Object pursued by high-status Models receives elevated common-currency value; an Object associated with low-status receives depressed value.

The conversion process is the first site of information loss. Each source’s native encoding (metabolic deficit, process requirement, relative position) is translated into a scalar value contribution. The richness of the original signal is compressed into a number. This compression enables comparison but discards source-identity.

9.1.4 Integration and Selection

Once converted to common currency, signals from all three sources are integrated through weighted summation. The result is a single value for each candidate action or Object: a scalar representing “how much the organism wants this, all things considered.”

Selection follows from comparison. The Decision Hub continuously computes values for available options and selects the highest-valued option for behavioural output. This is not a discrete, deliberated choice but a continuous process: values update in real-time as inputs change, and behaviour shifts when the value-ranking shifts.

The selection mechanism implements a form of winner-take-all dynamics: the highest-valued option captures motor resources and becomes behaviour, while lower-valued options are suppressed. This produces the unity of action that characterises normal behaviour; we do one thing at a time, not three things corresponding to A, I, and M separately.

9.1.5 The Cost of Integration: Source-Tag Destruction

The common-currency architecture solves the integration problem but creates a new problem: source-tags are destroyed in the conversion process.

When a food stimulus is valued highly, the Decision Hub encodes “high value”, not “high value because hungry” (A) versus “high value because Model is eating” (M). The integrated value signal does not carry information about which source contributed what proportion. The integration is a many-to-one map from three source-specific signals onto one Scalar, and that map necessarily discards the source-decomposition.

This is not a design flaw but an information-theoretic necessity. The map is non-injective: many combinations of the three produce each value of the Scalar, so no function of that value recovers which combination produced it (Appendix A, Section A.2). The architecture that enables unified decision-making is the same architecture that destroys source-transparency.

A coin shows the same fact from the outside. A child handed a coin wonders about the hands it has passed through, the tills and pockets and purses, and finds that the coin says nothing about any of them. It carries a denomination and a date: how much, and when it was struck. Nothing on it records where it has been, and nothing could, because a coin that carried its history would not be interchangeable with the next coin, and interchangeability is what lets it discharge the common-currency function.

The Scalar arrives in the same condition. It carries a magnitude and a sign, and it arrives now, from the Decision Hub, so the Individual knows where it was struck and when. What it does not carry is the record of what went into it, the Appetitive, Intrinsic and Mimetic contributions and their weights, and it cannot carry that record for the reason the coin cannot: a value that kept its sources would not be comparable with the next value, and comparability is what the common currency exists to provide. The history cannot be read off either Object. Some of it can be reconstructed from outside, from the coin's wear or from the behaviour the Scalar drives, which is the work of Section 9.9.5, but none of it can be recovered from the number itself. The child, knowing the history is missing, invents one. The Individual, not knowing it is missing, does the same, and the invented history is Confabulation (Axiom 4).

The coin is not an illustration borrowed from another domain. Chapter 12, Section 12.2 establishes that Money is the Scalar itself, once transduced by exchange and held on a substrate that can cross between bodies, so the coin and the Scalar are one quantity in two instantiations, and the coin carries no history for the same reason the Scalar carries none.

This destruction of source-tags is Source Opacity (Axiom 3), the architectural consequence of Common-Currency Integration whereby the Decision Hub's integration operation irreversibly destroys information about which source contributed what weight to the integrated Scalar, leaving the Individual without introspective access to the source-decomposition of any integrated valuation. Compensatory discrimination remains available only through inferential reconstruction from behavioural evidence (the Two-Question Diagnostic), which operates on the Thinker's prior space rather than on any single value of the integrated Scalar. Section 9.9 will develop Source Opacity in full. The present section establishes the architectural basis: common-currency integration necessarily entails source-tag destruction.

A consequence of Common-Currency Integration: the Decision Hub processes other Individuals through exactly the same integration architecture as any other Object. When Individual A encounters Individual B, the Decision Hub generates a scalar valuation of B by integrating A-signals (B's relevance to A's bodily needs), I-signals (B's relevance to A's Intrinsic engagement), and M-signals (B's Status position, prestige cues, social proof from others' observed valuations of B). The output is a single "how much do I value this Individual?" signal in which the three source-components have been irreversibly merged. This means that the PM-component of A's valuation of B (the portion driven by the fact that others visibly value B) is indistinguishable, after integration, from the A-component (B is physically attractive, B provides material security) and the I-component (B is intellectually stimulating, B facilitates creative engagement). The Individual cannot determine how much of their felt regard for another Individual reflects genuine A or I relevance and how much reflects socially transmitted PM-valuation.

This is Source Opacity applied to social valuation (developed in Movement C), and it is the architectural reason why the transfer channels through which Mimetic Premium (PM, the Premium) flows between Individuals (attention, deference, narrative authority, positional concession) operate invisibly to both parties. Neither the giver nor the receiver of PM-transfer can identify it as such from the felt valuation, because the PM-component of each party's valuation has already been integrated and its source-tag destroyed at integration. See Chapter 8 for the full derivation of the Individual-as-Object Principle and its consequences.

9.2 The AIM Integration Equation

9.2.1 Mathematical Form

The Decision Hub’s integration operation can be described mathematically as weighted summation:

U(x) = w_A \cdot U_A(x) + w_I \cdot U_I(x) + w_M \cdot U_M(x)

The weight $w_M$ is the same parameter labelled $\beta$ in Appendix A; the two notations refer to identical content.

Where:

U(x) is the integrated subjective value of option x, a signed scalar.
UA(x), UI(x), UM(x) are the source-specific value contributions from Appetites, Intrinsic Motivation, and Mimetic Desire. Per the Signed Valuation Principle, each is a signed scalar, taking positive values where the relevant component informs attraction toward x, negative values where it informs avoidance of x, or zero where the component is indifferent to x.
wA, wI, wM are non-negative weights expressing the proportional sensitivity of the Decision Hub to each of the three input channels at the moment of integration, and sum to one. The integrated valuation U(x) is therefore a signed scalar whose sign carries the action-selection consequence: a positive integrated valuation selects approach, a negative one selects avoidance (including fleeing and rejection), and one near zero selects disengagement. The signed character of U(x) is carried by the source-specific utilities, not by the weights, which is why a single component (the Observer, for example, returning a strongly negative UM(x) in response to a socially transmitted aversion) can produce avoidance behaviour even when the other components are indifferent or mildly positive.

This equation is a functional description, not a claim about conscious calculation. No Individual multiplies weights by utilities in awareness. The equation describes what the Decision Hub computes, not what the Individual experiences.

9.2.2 Weights Are Latent Variables

The weights $w_A$ , $w_I$ , $w_M$ cannot be directly observed. They are latent variables: theoretical constructs inferred from patterns of behaviour, not measured from neural activity or self-report. The commitment this carries is Representational Realism: the weights refer to real features of the Individual whether or not anyone is measuring them, and their validity rests on the accuracy of the predictions they generate about observable choice rather than on direct observation of the weights themselves, which is the standing the unobserved quantities of any mature science hold.

Weights vary across:

Individuals: Different Individuals have different typical weight distributions (AIM Profiles)
Contexts: The same Individual weights sources differently in different situations
Time: Weights shift as circumstances change (hunger rises, Models change, engagement fluctuates)

The weight-distribution at any moment represents the Decision Hub’s current sensitivity to each source. High $w_M$ means M-signals heavily influence the integrated value; high $w_I$ means I-signals dominate; high $w_A$ means A-signals are currently most potent.

9.2.3 Source-Specific Utilities

The source-specific utilities $U_A(x)$ , $U_I(x)$ , $U_M(x)$ represent how valuable option $x$ is from each source’s perspective:

$U_A(x)$ : How well does $x$ address current Appetitive deficits? High for food when hungry, shelter when cold, safety when threatened. Approaches zero as deficits are corrected (satiation).

$U_I(x)$ : How much does $x$ enable or constitute Intrinsically rewarding engagement? High for activities in the flow-channel, for tools that enable valued processes, for opportunities that expand competence. Does not approach zero with engagement at the level of the I-process, which deepens rather than satiating, though I-valuation of any particular I-enabling Object is bounded by functional sufficiency and terminates on its acquisition (Section 9.6.3; Chapter 12, Section 12.3.1).

$U_M(x)$ : How much does $x$ improve relative position or align with Model behaviour? High for status-signalling Objects, for pursuits that Models are observed pursuing, for choices that maintain or elevate social standing. Does not satiate because relative position is a moving target.

9.2.4 The Equation’s Limitations

The AIM Integration Equation is useful but simplified. It does not capture:

Non-linear interactions: Sources may interact multiplicatively, not just additively. Extreme A-deficit may suppress I-engagement entirely (hard to enjoy learning when starving); M-amplification may be stronger when A-floors are insecure.

Temporal dynamics: The equation represents a snapshot. Real integration involves continuous updating as signals change, with hysteresis effects (recent values influence current values) and momentum (ongoing activities resist displacement).

Threshold effects: Some behaviours require minimum source-contributions to occur at all. An activity may need both adequate I-value and adequate A-security before it becomes viable.

The equation is a first-order approximation; accurate enough for conceptual work and diagnostic application, while acknowledging that the full dynamics are more complex than weighted summation captures.

9.3 Composition and Intensity

9.3.1 Two Dimensions of Motivated Behaviour

Motivated behaviour varies along two independent dimensions:

Composition: The relative contribution of each source to the integrated value. Which sources are driving this behaviour, and in what proportions?

Intensity: The absolute magnitude of the integrated value. How strongly is the Individual motivated, regardless of source?

These dimensions are orthogonal. A behaviour can be:

High-intensity, A-dominant (desperate hunger)
High-intensity, M-dominant (urgent Status Pursuit)
Low-intensity, I-dominant (mild ongoing engagement)
High-intensity, balanced (strong motivation from multiple sources)

Composition answers “why this behaviour?”, which sources are responsible. Intensity answers “how strongly?”: the magnitude of motivational force.

9.3.2 Lead-Tag-Last Heuristic

For practical purposes, AIM uses a Lead-Tag-Last heuristic to characterise composition:

Lead: The source with the highest weight: the dominant driver
Tag: The source with the second-highest weight: the secondary influence
Last: The source with the lowest weight: minimal contribution

This simplification loses precision but gains usability. Rather than specifying exact weights (which are rarely knowable), the heuristic identifies which source is primary and which is secondary.

Examples:

“I-lead, A-tag” = Intrinsic Motivation dominates, with Appetitive contribution (e.g., enjoying a meal that is also satisfying hunger)
“M-lead, I-tag” = Mimetic Desire dominates, with Intrinsic contribution (e.g., pursuing a career that is status-driven but also somewhat engaging)
“A-lead, M-tag” = Appetites dominate, with Mimetic contribution (e.g., eating at a trendy restaurant primarily because hungry but also for the social setting)

9.3.3 Why Composition Matters

Composition determines:

Satiation dynamics: A-lead behaviours satiate episodically; I-lead behaviours deepen without satiation; M-lead behaviours never satiate because reference points shift.

Persistence patterns: I-lead behaviours persist without audience; M-lead behaviours collapse once observability is removed and the loop holding them releases, which can lag the removal (Chapter 4, Section 4.4.13.2); A-lead behaviours persist until deficit is corrected.

Intervention targets: Changing A-lead behaviour requires addressing the deficit; changing I-lead behaviour requires changing the engagement conditions; changing M-lead behaviour requires addressing the Model-tracking or social-comparison dynamics.

Subjective experience: The phenomenology of A-driven wanting (pressure from deficit), I-driven wanting (pull of engagement), and M-driven wanting (urgency to achieve/maintain position) feel different, though Source Opacity often prevents accurate identification.

Composition is the diagnostic target for AIM-informed analysis. Knowing that an Individual is motivated tells you little; knowing which sources are driving the motivation tells you what to expect and how to intervene.

9.3.4 Intensity as Independent Variable

Intensity matters for behavioural prediction but not for source-diagnosis. Two behaviours with identical composition but different intensity will have the same satiation dynamics and persistence patterns but different urgency levels.

High-intensity motivation:

Captures attention more completely
Overrides competing options more easily
Produces stronger subjective experience of wanting
Is more resistant to interruption or redirection

Low-intensity motivation:

Competes less effectively with alternatives
Is more easily redirected or abandoned
Produces milder phenomenology
Is more amenable to deliberative intervention (I-Override is easier)

The relationship between intensity and composition is not fixed. M-signals can produce both high-intensity (desperate Status Pursuit) and low-intensity (mild social conformity) motivation. The same is true for A and I. Source-identity does not determine intensity; context and Individual variation determine both.

9.4 Neurodiversity and Individual Variation

9.4.1 AIM Assumes Neurodiversity

The AIM Framework assumes neurodiversity as a baseline condition. No two brains are structurally identical; therefore no two Individuals will compute identical AIM weights even when encountering identical stimuli in identical contexts. This structural non-identity is only the first of two channels. Even setting brain-structural differences aside, the non-commutative path-dependence of Belief guarantees non-identical AIM profiles and Self-Models (Chapter 15, §15.2; Non-Commutativity), so the absence of a modal type is over-determined: it holds on the genetic-structural channel and on the architecturally-entailed trajectory channel alike. This two-channel non-modality is the descriptive basis of the Diversity Referent (Chapter 14, §14.8.3.1.2), the result that any carve of human variation into ranked natural kinds is held against an available decohered Referent.

This variation is not noise to be averaged away but signal to be preserved. The architectural claim (that all Individuals possess Decision Hubs integrating A, I, and M signals through common-currency valuation) is compatible with wide variation in:

Baseline sensitivities: How strongly each source’s signals register before integration
Weighting tendencies: The typical $w_A$ , $w_I$ , $w_M$ distribution across contexts
Integration dynamics: How rapidly signals are processed and integrated
Override capacity: The strength of I-system deliberative resources
Caching propensity: How readily and durably habits form

AIM is a framework about architecture, not about outputs. The claim is that every sentient Individual capable of making decisions possesses the three-source structure feeding common-currency integration. What that architecture produces (which desires dominate, how intensely, with what behavioural consequences) varies by Individual.

9.4.2 The Architecture Is Universal; The Weights Are Individual

The universality claim and the individuality claim are complementary, not contradictory:

Level	Claim	Status
Architecture	All Individuals have A, I, M sources integrating through a Decision Hub	Universal (species-typical)
Weights	Each Individual’s typical weight distribution differs	Individual (person-specific)
Episodes	Each decision episode produces context-specific weights	Situational (episode-specific)

An Individual may have:

Higher baseline A-sensitivity (stronger interoceptive signals, more readily detected hunger, fatigue and pain)
Lower baseline M-sensitivity (weaker mirror-neuron responsivity, reduced social-comparison activation)
Atypical I-dynamics (different flow-channel parameters, unusual competence-frontier trajectories)

These variations produce different phenomenologies of motivation. The high-A-sensitivity Individual experiences bodily states more intensely; the low-M-sensitivity Individual is less moved by what others pursue; the Individual of atypical I-sensitivity may find flow in contexts that bore or overwhelm Individuals of more typical I-sensitivity.

None of this variation contradicts the architectural claim. The variation occurs within the architecture: different parameters, same structure.

9.4.3 Implications for Diagnosis and Intervention

Individual variation in AIM parameters has direct implications for both diagnosis and intervention:

Diagnostic implications:

Population-level behavioural signatures (Section 9.14.3) may not apply uniformly to all Individuals
An Individual with atypically low M-sensitivity may show reduced audience-dependence even for M-led behaviours
An Individual with atypically high A-sensitivity may show A-signatures (state-dependence, satiation) even when I or M leads
Diagnosis should calibrate to the Individual’s baseline, not to population norms

Intervention implications:

I-Override training may be more or less effective depending on Individual I-system strength
Environment restructuring may be more critical for Individuals with high caching propensity
A-floor security may be more urgent for Individuals with high A-sensitivity
M-management strategies may require adjustment for Individuals with atypical M-dynamics

The AIM Framework provides the diagnostic categories; Individual assessment determines which interventions are appropriate for which Individual. A uniform intervention applied across a neurodiverse population will produce variable results, not because the framework is wrong but because the parameters vary.

9.4.4 Neurodivergent Populations

Clinically recognised neurodivergent conditions may involve systematic differences in AIM parameters:

Autism Spectrum Conditions:

May involve reduced M-sensitivity (lower mirror-neuron responsivity, reduced automatic social-comparison)
May involve atypical I-dynamics (intense engagement with specific domains, different flow-channel parameters)
May involve heightened or reduced A-sensitivity depending on sensory processing differences
The “social deficits” framing may partly reflect reduced M-capture rather than reduced social capacity

ADHD:

May involve atypical I-dynamics (novelty-seeking elevated, sustained-engagement reduced)
May involve altered caching dynamics (habits form differently, cached entries less stable)
May involve different weighting dynamics (rapid weight-shifts, reduced persistence of any single source’s dominance)

Anxiety Disorders:

May involve over-weighted threat-priors (the Thinker's threat-priors conditioning the Observer's signed Salience, Chapter 4, Section 4.4.2)
May involve elevated M-sensitivity (social-evaluation signals amplified)
May reflect I-system weakness (reduced capacity for deliberative override of A/M signals)

Depression:

May involve suppressed I-activation (reduced engagement-reward, flattened competence-seeking)
May involve altered A-dynamics (appetite and energy dysregulation)
May involve M-dysregulation (social-comparison producing consistent negative signal)

These mappings are speculative and require empirical investigation. AIM does not claim to explain these conditions; it provides a framework within which they might be characterised in motivational terms, and within which interventions might be designed that target the specific parameter variations involved.

9.4.5 The Architectural Claim Is Falsifiable

The universality of the architecture is a Tier 1 claim, and its falsification conditions sit on the architectural surface at Appendix B, Section B.2. The discovery of a fourth motivational source that cannot be reduced to A, I, or M, or of Individuals who demonstrably lack one of the three sources, strikes the first commitment stated there; the discovery that motivational signals remain source-tagged through to conscious awareness in typical Individuals strikes the third.

The Individual-variation claim is not falsifiable in the same way; it is a framework assumption that accommodates observed diversity. But specific Hypotheses about which conditions involve which parameter variations are falsifiable through appropriate research designs.

AIM welcomes such research. Discovering that the architecture varies more fundamentally than AIM assumes, that some Individuals have genuinely different motivational structures, not just different parameters, would require framework revision. Such revision would be a scientific advance, not a framework failure.

9.4.6 Firing Sequences and the Architectural Dissolution of Nature versus Nurture

The architectural account of integration developed in this chapter has a consequence that bears directly on the treatment of neurodivergence and, more broadly, on the longstanding debate over the contributions of nature and nurture to individual motivational profiles. The debate presupposes that the neural firing sequence underlying any given cognitive or behavioural output can in principle be decomposed into a genetically-caused component and an environmentally-caused component. No such decomposition exists. The connectome through which any current input is processed is itself the integrated product of all prior genetic expression and all prior input received during the Individual's development, and the current input is itself being processed by that same integrated connectome. There is no stage of processing at which genetic and environmental contributions exist as separable inputs that could be held constant independently.

The dichotomy is therefore not merely empirically intractable. It is architecturally ill-posed at the level of the firing sequence, and therefore at every level that supervenes on it, including the behavioural outputs by which the AIM Framework classifies motivational sources. This matters for neurodivergence in two ways. First, variation in motivational profile across Individuals is not usefully partitioned into innate and acquired components, because no such partition is well-defined for the underlying neural events. Second, AIM's three-source taxonomy does not depend on such a partition. The taxonomy classifies motivations by the operating dynamics of the resulting desires (Differential Satiation Dynamics, audience-dependence, reference-point behaviour), which are observable behavioural signatures that remain available regardless of whether the developmental pedigree of any particular firing sequence can be causally decomposed. Neurodivergent motivational profiles are accommodated at the level of individual-specific weights on A, I, and M inputs to the Decision Hub, without requiring that those weights be explained by reference to a nature–nurture partition the architecture does not support.

9.4.7 Sex as Parameterisation

Biological sex enters the architecture where every other source of Individual variation enters it: in the parameters, not the structure. Sex-typical differences in hormonal milieu act on baseline sensitivities and weighting tendencies (Section 9.4.1). They alter the gain on bodily signals, the reactivity to threat and provocation, and the readiness with which aggression is selected as a response. These are differences in the weights the Decision Hub integrates. The three-source structure and the common-currency integration are the same in every Individual.

Hormonal action is not a channel separate from experience. The hormonal state at any moment is partly a product of the Individual's history, rising and falling with competition, standing and social context. The argument that dissolves the nature–nurture partition at the level of the firing sequence (Section 9.4.6) therefore applies here. There is no stage at which a hormonal contribution to an act exists as an input that could be held constant while the Individual's history varies.

Sex therefore changes two things at once: the weighting and the informational experience of the Individual. A different hormonal milieu produces a different weight distribution. A different body, treated differently by others from birth, produces a different history of evidence, and so different Beliefs and a different Self-Model.

None of this fixes what an Individual can do. Differences between the sexes in mean upper-body strength, or in average rates of physical aggression, are distributional observations, and the Diversity Referent does not defeat them (Chapter 14, Section 14.8.3.1.2). What the Diversity Referent defeats is turning a distributional difference into a kind. The claim that women are not warriors or hunters is a kind-claim, and the overlap of the distributions does not support it.

Where women do not become warriors or hunters, the architecture places the constraint in the culture's Beliefs, not in capacity. A culture's Beliefs about the proper pursuits of each sex decide which Models are placed before a girl and which pursuits draw approval or sanction. They therefore decide which M-signals reach her Decision Hub when she moves toward the hunt or the fight. The same Beliefs are installed in her own Belief Register as priors about herself. Opportunity and encouragement are constrained before ability is ever tested. A constraint that operates before the test cannot be read from the outcome as evidence about the ability.

9.5 The Phenomenological Reality Principle: Why Sources Must Be Respected

9.5.1 The Involuntary Arrival of Motivational Signals

Motivational signals from A, I, and M arrive at the Decision Hub without the Individual’s consent or choice. The Individual does not decide to feel hungry, decide to find an activity engaging, or decide to want what Models pursue. These signals are generated by subsystems operating below conscious awareness and fed into the integration process before the Individual has any opportunity to evaluate, endorse, or reject them.

This involuntary arrival is not a design flaw; it is the architecture operating as evolved. The A-subsystem must generate hunger signals without waiting for conscious permission; starvation cannot await deliberative approval. The I-subsystem must register what its running process requires without waiting for deliberative approval; a requirement the process has cannot be willed out of existence by fiat. The M-subsystem must track relative position and Model behaviour continuously; coalitional standing in ancestral environments could not wait for conscious social-comparison decisions.

The timing architecture (Section 9.11) establishes the neurological basis: source-specific processing occurs in the 0–300ms window, integration in the 200–400ms window, and conscious awareness arrives only at 300–500ms. By the time the Individual is aware of wanting something, the wanting has already been generated, transmitted, and integrated. The Individual’s first conscious contact with the motivational signal is reception, not generation.

This means the Individual cannot be held responsible for having the desires they have. Desires arrive; they are not chosen. The hungry Individual did not choose to be hungry; the curious Individual did not choose to be curious; the status-anxious Individual did not choose to feel status-threatened. Moral frameworks that treat desire-occurrence as itself blameworthy (“you should not want that”) misunderstand the architecture. The Individual has no pre-desire vantage point from which to prevent desires from arising.

9.5.2 Felt Experience as Real Experience

AIM adopts what may be called the Phenomenological Reality Principle: the felt intensity and character of a motivational signal constitutes the Individual’s real experience, regardless of the signal’s source or “objective” justification.

The M-driven desire for a status object feels as real, as urgent, and as authentically “mine” as the A-driven desire for food when hungry or the I-driven desire to continue an engaging activity. Source Opacity ensures the Individual cannot distinguish these by introspection, but even if they could, the felt experience would remain. Knowing that a desire is mimetically transmitted does not make it feel less urgent. Knowing that status anxiety is “just” relative-position tracking does not make the anxiety feel less real.

This principle has three implications:

Felt experience cannot be argued away. Telling an Individual that their desire is “merely mimetic” or “just status-seeking” does not extinguish the desire. The M-subsystem continues generating signals regardless of the Individual’s intellectual assessment. The conscious recognition “this is M-driven” does not reach back into the M-subsystem and deactivate it. The signal continues; only the Individual’s response to the signal can change.

Dismissal of felt experience is a form of violence. When external authorities (whether institutional, interpersonal, or ideological) dismiss an Individual’s felt desires as illegitimate, unreal, or unworthy of consideration, they deny the Individual’s phenomenological reality. The Individual does feel what they feel; asserting otherwise is gaslighting. AIM insists that all source-signals be acknowledged as real experiences, even when the appropriate response is not to act on them.

Paternalism that denies experience fails. Interventions designed on the assumption that Individuals can simply stop having certain desires (that M-desires will disappear if labelled as such, that A-desires can be overcome by willpower, that I-engagement can be commanded into existence) systematically fail because they misunderstand the architecture. The signals continue arriving regardless of what external authorities declare about their legitimacy.

9.5.3 Respecting Sources vs. Obeying Sources

AIM’s insistence that sources be respected is distinct from the claim that sources should be obeyed. The framework distinguishes:

Respect = acknowledging the signal as real, understanding its source, treating it as valid input to deliberation

Obedience = acting on the signal without scrutiny, treating it as command rather than input

The I-Override mechanism (Section 9.13) exists precisely to create space between reception and action. The Individual who receives an M-signal can:

Acknowledge the signal as real felt experience (“I genuinely want this”)
Diagnose the signal’s likely source through the Two-Question Diagnostic (“This want diminishes when I remove the audience: likely M-led”)
Evaluate whether to act on the signal given its source (“Do I endorse pursuing M-satisfaction in this domain?”)
Choose to act or refrain, with full recognition that the signal will persist regardless of choice

This sequence respects the signal at every stage (it is never denied or dismissed), while retaining the Individual’s capacity to choose their response. The M-desire remains real; the Individual simply decides not to act on it. The A-hunger remains real; the Individual on a medically-necessary fast acknowledges the hunger while choosing not to eat. The I-engagement remains real; the Individual who must attend to other obligations acknowledges the pull of the engaging activity while redirecting attention.

Respect without obedience is the structure of mature agency in the AIM framework. The alternative, attempting to not-have the desires one has, is architecturally impossible and psychologically damaging.

9.5.4 Why AIM Does Not Advocate Source Elimination

Some readers may expect AIM to advocate eliminating M-influence; training Individuals to become “M-free” and operate purely on A and I. This misunderstands both the architecture and the framework’s normative commitments.

Architectural impossibility: The M-subsystem cannot be surgically removed or deactivated. Mirror-neuron systems are anatomically integrated with motor and valuation circuits. Social-comparison processing is not an optional module but a pervasive feature of human cognition. The Individual cannot choose to stop receiving M-signals any more than they can choose to stop receiving visual input while their eyes are open.

Normative undesirability: Even if M-elimination were possible, it would be undesirable. The M-subsystem enables:

Cultural learning and transmission
Coordination and cooperation
Emotional attunement
Shared meaning and collective experience

An M-free Individual would be incapable of learning language, acquiring cultural knowledge, coordinating with others, or participating in shared human experience. The M-subsystem is not a parasite to be excised but a fundamental component of human social cognition. Though it does act as a host to a parasite called the Mimetic Premium, which will be explored at length in later chapters.

The target is M-management, not M-elimination. AIM’s normative project is to:

Help Individuals recognise M-signals as M (vocabulary provision)
Enable Individuals to choose which M-signals to act on (I-Override)
Structure environments to reduce harmful M-amplification (visibility governance)
Channel M toward positive coordination rather than destructive rivalry (positive mimesis)

This is management of a permanent architectural feature, not elimination of a contingent flaw.

9.5.5 Implications for Intervention Design

The Phenomenological Reality Principle constrains how AIM-informed interventions should be designed:

Interventions must acknowledge felt experience. Any intervention that begins by denying that Individuals really feel what they feel will fail and will be experienced as invalidating. Effective intervention starts from “Yes, you genuinely want this” and proceeds to “Let’s understand why and whether acting on it serves you.”

Interventions must work with the architecture, not against it. Telling Individuals to “just stop wanting status” is like telling them to “just stop seeing colour.” The instruction is architecturally meaningless. Effective intervention provides tools for recognising, evaluating, and choosing responses to desires, not tools for preventing desires from arising.

Interventions must respect Individual variation. The Phenomenological Reality Principle applies to each Individual’s actual experience, which varies (Section 9.4). An Individual with high M-sensitivity experiences more intense M-signals; dismissing this as “oversensitivity” denies their phenomenological reality. Intervention must be calibrated to actual experience, not to population averages.

Interventions must accept that signals persist after diagnosis. The Individual who successfully identifies a desire as M-led will continue to feel that desire. I-Override provides choice, not extinction. Interventions should prepare Individuals for this persistence rather than promising that insight will eliminate wanting. The goal is freedom to choose despite the desire, not freedom from the desire itself.

Structural interventions are often more effective than individual-level interventions. Because signals arrive involuntarily and persist after diagnosis, reducing the generation of problematic signals (through environment design) may be more effective than increasing Individual capacity to override signals that have already been generated. This is why AIM emphasises visibility governance, A-floor security, and I-domain protection alongside Individual-level I-Override training.

9.5.6 The Dignity of Felt Experience

AIM’s commitment to respecting sources is ultimately grounded in a claim about human dignity: the Individual’s felt experience is constitutive of their existence as a conscious being and must be treated as real and significant regardless of its causal origin.

The reductive move, “your desire is merely mimetic, therefore not really yours”, fails on two counts:

Causation does not determine ownership. All desires have causal histories. A-desires are caused by physiological processes; I-desires are caused by engagement dynamics; M-desires are caused by social observation. None of these origins makes the desire less “the Individual’s own.” The desire is felt by this Individual, in this body, at this moment; it is theirs by the criterion that applies: phenomenological presence.

The distinction between “authentic” and “inauthentic” desires cannot be drawn at the source level. Some M-desires lead to flourishing (positive mimesis, cultural participation, cooperative coordination); some I-desires lead to harm (obsessive engagement that damages health or relationships); some A-desires are manipulated (engineered food products that exploit hunger signals). Source does not determine value. What matters is whether acting on the desire serves the Individual’s wellbeing and aligns with their reflectively-endorsed values, and this can only be determined through the kind of scrutiny that I-Override enables.

AIM therefore treats all three sources with equal ontological respect: each generates real signals that constitute real experiences that deserve acknowledgment. The evaluative question, whether to act on any particular signal, is subsequent to and dependent on this foundational respect.

The Phenomenological Reality Principle applies symmetrically across the signed range. Negative-valence felt experience (revulsion, fear, contempt, dread) is as architecturally real, and as deserving of respect as input data, as positive-valence felt experience. The normative implication is symmetric: institutions and interlocutors must take avoidance signals seriously as Decision-Hub outputs of an Individual with intact architecture, while remaining alert to the fact that Source Opacity applies to those signals as fully as to attraction signals, and that Confabulation will return grounds for them within whatever Hypothesis Space the Thinker holds.

Movement B: Dynamic Operation

9.6 The Decision Hub as Control System

9.6.1 Cybernetic Architecture: Goals, Feedback, Error-Correction

The Decision Hub is not a one-shot calculator that computes a value and stops. It is a control system: a continuously operating architecture that monitors current states, compares them against goal-states, computes error signals, and generates corrective outputs.

The cybernetic framework, developed by Norbert Wiener (1948) and elaborated by Powers (1973) and by Carver and Scheier (1998) in control-theory approaches to psychology, provides the conceptual vocabulary for understanding this dynamic operation:

Goal-state (reference value): A target condition the system attempts to achieve or maintain. For the A-subsystem, this is homeostatic equilibrium: appropriate levels of nutrients, hydration, temperature, safety. For the I-subsystem, this is the state the running process was set to reach. For the M-subsystem, this is relative position: acceptable standing within the reference group.

Current-state (perception): The system’s representation of present conditions, derived from sensory and interoceptive inputs. The A-subsystem monitors bodily states; the I-subsystem monitors the state of the running process against what the process requires; the M-subsystem monitors social position.

Error signal (discrepancy): The difference between goal-state and current-state. Positive error means the current state falls short of the goal; negative error means it exceeds the goal (relevant for A, where oversatiation can occur). Error signals generate motivational urgency proportional to the discrepancy magnitude.

Output (behaviour): Action designed to reduce the error signal, to move current-state toward goal-state. The Decision Hub integrates error signals from all three subsystems and generates the behavioural output most likely to reduce aggregate discrepancy.

Feedback loop: The output affects the environment; the changed environment is perceived; the new perception updates the current-state representation; the error signal is recomputed. The loop runs continuously, not in discrete episodes.

This cybernetic framing explains why motivation feels like pressure or drive. The phenomenology of wanting is the conscious registration of error signals: the system detecting discrepancy between where it is and where it should be. The urgency of wanting reflects error magnitude: large discrepancies produce intense pressure; small discrepancies produce mild inclination; zero discrepancy produces the temporary quiescence we experience as satisfaction.

9.6.2 The A-Subsystem as Homeostatic Controller

The A-subsystem operates as a classic homeostatic controller, the archetype that inspired cybernetic theory.

Goal-state: Physiological and safety set-points: blood glucose within range, core temperature within range, resource buffer above threshold. These set-points are biologically determined, though they can be modulated by context (e.g., fever raises the temperature set-point as an immune response).

Current-state: Interoceptive monitoring through hypothalamic and brainstem circuits. The system continuously samples internal conditions: nutrient availability, hydration status, temperature, pain signals.

Error signal: Deficit or surplus relative to set-point. Hunger is the error signal for nutrient deficit; thirst for hydration deficit; cold for temperature deficit; fear for safety deficit. The signal intensifies with discrepancy magnitude and duration.

Output: Behaviour oriented toward deficit-correction: eating, drinking, warming, fleeing, securing resources. The A-subsystem drives approach toward Objects that promise set-point restoration and avoidance of conditions that would increase deficit.

Feedback: Consumption reduces the error signal. Eating diminishes hunger; drinking diminishes thirst; reaching safety diminishes fear. The loop closes when the current-state reaches the goal-state (satiation), at which point the error signal drops to zero and A-driven motivation terminates, until the next perturbation reintroduces discrepancy.

The A-subsystem’s control dynamics explain A’s episodic satiation property (Axiom 6). A-motivation is inherently self-terminating because the control loop is designed to reach equilibrium. The system does not want for its own sake; it wants to eliminate discrepancy. When discrepancy is eliminated, wanting ceases.

9.6.3 The I-Subsystem as Process-Requirement Controller

The I-subsystem operates as a controller on the requirements of whatever process the Thinker is running.

Goal-state: the state the running process is for. Not a set-point and not a level of engagement, but whatever the process was set to reach: the finished chair, the committed posterior, the meal on the table. The goal-state is not fixed by the architecture, a process being one the Thinker set itself, one the Body Manager's signal set for it, or one arriving through the loop the Belief Register closes (Chapter 7, Section 7.5.1.1).

Current-state: what the process has achieved and what it still requires. Ground-Resolution applied to the requirements of the current stage returns the shortfall as its result.

Error signal: the requirement the current stage has not yet met. The signal is scaled to what the stage needs and enters the Common-Currency Scalar alongside the Body Manager's. Boredom is not an error signal on this loop, reporting as it does that the Object under attention holds no unresolved residue, which is a fact about the Object rather than about the process's requirements, and which can arrive with every requirement met.

Output: action directed at the Object through which the requirement can be met. The action is the transduction of the Scalar into a medium in which it can act on the Object (Chapter 4, Section 4.4.1), and the Object carries the signal for as long as action is directed there. Where the requirement is a condition of the Individual's own body, the action meets it directly and no intermediary is interposed (Chapter 4, Section 4.4.2).

Feedback: the work done on the Object closes that stage's requirement and leaves the process at the stage registering the next. Where the state the process requires is possession, obtaining the Object closes the requirement; in the general case the requirement closes when the action has done its work. The loop closes on completion of the process rather than on the first error reaching zero, and unlike A the closure does not reinstate itself: a set-point drifts back and re-opens the same loop, whereas a completed process does not re-open, and what comes next is whatever process the Thinker sets.

The I-subsystem's control dynamics explain I's non-satiating deepening property (Axiom 6). Each of the loop's requirements is met and the process itself completes, but the prior-maintenance and ground-resolution function that runs the loop does not stop, and a committed posterior raises the resolution at which a mismatch is detectable, so a domain worked through at one grain presents unresolved structure at a finer one. Deepening is that regeneration of the activation condition by its own satisfaction. There is no final state at which the system has enough, not because no closure is available to it, but because each closure makes the next one available.

The I-subsystem produces satiation for I-enabling Objects: tools, instruments, and resources the I-process requires. The guitarist's desire for a guitar terminates upon acquiring a functionally adequate instrument. These are the stages at which the state the process requires is possession, and every other stage closes in the same form by work rather than by acquisition, the wood carved and the joint nailed as much as the sandpaper bought. The process continues across all of those closures, and it is the process rather than any of its requirements that does not satiate.

9.6.4 The M-Subsystem as Relative-Position Controller

The M-subsystem operates as a relative-position controller oriented toward acceptable standing within reference groups.

Goal-state: Relative position within the reference group, not absolute resource level but rank. The set-point is inherently comparative: “at least as high as relevant comparisons” or “higher than current position.” This comparative constitution is fundamental to M-dynamics.[^kappa-S]

[^kappa-S]: The salience-to-M coupling described here is given the formal symbol $\kappa_S$ in Appendix A, where it appears as the fast-layer parameter governing how strongly Observer-borne Salience loads the Decision Hub's M-input. The chapter uses prose throughout; the formalisation in $\kappa_S$ is supplied for readers who pursue the appendix.

Current-state: Social-comparison circuits continuously monitor relative position. Mirror-neuron systems track what Models pursue; social-reward circuits track prestige cues; status-monitoring circuits compute rank within reference groups. The M-subsystem maintains a running representation of “where I stand.”

Error signal: Discrepancy between current position and goal position. Feeling “behind” one’s reference group produces negative M-error (status-anxiety, envy, urgency to catch up). Feeling “ahead” produces positive M-feedback (status-satisfaction, superiority), but this positive state is unstable because reference points shift.

Output: Behaviour oriented toward improving or maintaining relative position. This includes pursuit of status-signalling Objects, display behaviours, competitive effort, Model-tracking (shifting attention toward what successful others pursue), and reference-group management (adjusting which comparisons are Salient).

Feedback: Status-achievement temporarily reduces the error signal. But successful achievement shifts the reference point. What was “ahead” becomes “normal.” New comparisons become Salient. The goal-state ratchets upward. The error signal regenerates.

The M-subsystem’s control dynamics explain M’s non-satiating property (Axiom 6), but through a different mechanism than I’s deepening. I does not satiate because competence generates new frontiers; the system pursues an advancing positive target. M does not satiate because the Observer's Salience assessment over the perceived field has no completion condition, and the reset of reference points is how that shows: the system pursues a moving target that recedes with each achievement. The Individual who achieves a status goal finds that the goal no longer feels like an achievement; it feels like a baseline against which new deficits are measured.

This is the hedonic treadmill mechanism specific to M. A-satiation produces stable equilibrium (deficit corrected, wanting terminates). I-deepening produces positive expansion (competence grows, new frontiers open). M-reset produces perpetual deficit (position achieved, reference point shifts, new deficit appears). The same architecture (goal-state, current-state, error-signal, feedback) produces radically different experiential trajectories depending on which subsystem dominates.

9.6.5 Integration of Three Control Loops

The Decision Hub integrates error signals from all three control loops simultaneously. At any moment, the Individual may have:

A-error: moderate hunger (deficit from last meal)
I-error: mild boredom (current activity below optimal challenge)
M-error: status anxiety (colleague recently promoted)

The integrated output reflects weighted combination of these errors. The behaviour selected is the one that best addresses the aggregate discrepancy, given the current weights and available options.

This integration explains motivational conflict; the experience of being pulled in different directions. Conflict occurs when different sources’ error-corrections point toward incompatible behaviours:

A-error says “eat now”; I-error says “keep working on this engaging problem”
I-error says “pursue this fascinating but low-status project”; M-error says “pursue the high-status project that bores you”
M-error says “attend the networking event”; A-error says “rest, you’re exhausted”

The Decision Hub resolves conflict through integration: the behaviour with the highest weighted-sum value wins. But subjectively, the losing sources’ error signals persist. The Individual who chooses work over lunch continues to feel hunger; the Individual who chooses the status project over the interesting one continues to feel the pull of engagement. Conflict resolution does not eliminate the losing signal: it merely results in behaviour that addresses the dominant signal.

9.6.6 Continuous Recalculation

The control system operates continuously, not in discrete episodes. Values update in real-time as:

Internal states change (hunger rises, fatigue accumulates, pain subsides)
Engagement conditions shift (task becomes easier or harder, novelty diminishes)
Social information arrives (Model behaviour observed, status feedback received)

Behaviour can therefore shift mid-action. The Individual who began eating because A-dominated may stop before satiation if M-signals suddenly spike (an important Individual enters the room). The Individual engaged in flow-state work may abandon it if A-signals become overwhelming (acute pain, extreme hunger). The Individual pursuing a status goal may lose interest if the relevant Models shift their attention elsewhere.

This continuous recalculation is the substrate on which habits (Section 9.8) and I-Override (Section 9.13) operate. Habits represent cached prior-configurations that bypass real-time recalculation for efficiency. I-Override represents deliberate intervention in the recalculation process, introducing I-system scrutiny that can modify weights before action.

9.6.7 Signed Operation Across Approach and Avoidance

The control-system description above is stated for approach behaviour, but the same architecture operates symmetrically for avoidance. By the Signed Valuation Principle, each input-bearing component supplies a signed scalar: positive (informing approach), negative (informing avoidance), or zero (informing indifference). The Decision Hub integrates signed terms, and the integrated valuation can therefore be negative, in which case the Decision Hub selects an avoidance response. Source Opacity applies symmetrically: the Individual cannot introspect whether felt revulsion is A-grounded (bodily disgust), I-grounded (value-incongruity), or M-grounded (socially transmitted aversion). Confabulation likewise generates avoidance-side grounds (this is harmful, this is wrong, this is beneath me) under the same architectural mechanism. The avoidance side of the Camouflage Architecture is the architectural source of socially transmitted contempt and the affective substrate on which the Scapegoat Mechanism (Chapter 11) operates.

9.7 Structural Restlessness: The Aggregate Consequence

The differential satiation dynamics of the three control systems produce an aggregate consequence: Structural Restlessness; the architecturally necessary condition whereby stable, final satisfaction is unavailable to any Individual.

This is not a contingent feature of modern life, a symptom of consumerism, or a failure of gratitude. It is a direct consequence of how the three subsystems operate:

A-restlessness is cyclical. The A-subsystem achieves equilibrium (satiation) but cannot maintain it indefinitely. Homeostatic perturbations recur: metabolism depletes nutrients, circadian rhythms generate fatigue, environmental changes create new deficits. The Individual who has eaten will hunger again; the Individual who has rested will tire again. A-satisfaction is episodic, not terminal. The best achievable outcome is reliable maintenance of dynamic equilibrium, not permanent elimination of A-need.

I-restlessness is expansive. Each process the I-subsystem runs reaches the state it was set to reach, and the subsystem then sets another. Competence growth does not terminate interest: a committed posterior raises the resolution at which a mismatch is detectable, so the guitarist who masters one technique discovers new challenges, the scientist who answers one question generates new questions, and the craftsperson who achieves one level of skill perceives new possibilities for refinement. I-restlessness is therefore not the persistence of an unmet deficit but the arrival of a further one: not “I lack something still” but “more is now possible.” This is the phenomenology of deepening rather than the phenomenology of deprivation.

M-restlessness is treadmill-driven, and the operative mechanism is the Mimetic Treadmill, the reference-set reset whereby achievement shifts the comparative reference point so that what was ahead becomes the new baseline. The M-subsystem pursues relative position, but relative position is defined against a moving target. Achievement shifts the reference point; what was “ahead” becomes “normal”; new comparisons become Salient; the felt deficit regenerates. The Individual who achieves a status goal experiences brief satisfaction followed by reference-point reset and renewed pursuit. M-restlessness is structurally interminable because the goal recedes with each step toward it.

The three forms of restlessness are phenomenologically distinct:

Source	Phenomenology	Resolution Path
A	“I need” → deficit-pressure	Provision (satiation, then recurrence)
I	“I can go further” → frontier-pull	Engagement (deepening, no terminus)
M	“I’m behind” → comparison-anxiety	None (reference-point resets)

Diagnostic and intervention implications follow directly. Misidentifying which source generates restlessness produces interventions that address the wrong system:

Treating A-restlessness (material insecurity) with M-interventions (status reassurance) fails: the Individual needs provision, not perspective
Treating I-restlessness (desire for new challenges) with A-interventions (more consumption) fails: the Individual needs opportunity, not resources
Treating M-restlessness (status anxiety) with A-interventions (more acquisition) fails; more Objects cannot satisfy a deficit defined by relative position

The AIM Framework predicts that M-restlessness is the most pernicious because it lacks any resolution mechanism. A-restlessness can be addressed through reliable provisioning; I-restlessness is experienced as positive (the pull of possibility rather than the pressure of deficit); M-restlessness generates continuous felt-deficit that no achievement can permanently resolve.

This explains the persistent observation across wisdom traditions that humans are “never satisfied”, and why moralistic exhortations to “be grateful” or “want less” systematically fail. The restlessness is not a character flaw amenable to willpower correction; it is an architectural feature of motivational systems that evolved for adaptive function, not for subjective contentment. The A-system that re-generates hunger keeps the organism fed; the I-system that opens new frontiers drives cumulative learning; the M-system that tracks relative position maintained coalitional standing in ancestral environments.

Structural Restlessness is therefore not a design flaw but a design feature whose consequences have become maladaptive in environments that maximally activate M while providing minimal I-protection. Chapter 10 will develop this evolutionary mismatch in detail. The present chapter establishes the architectural basis: the three control systems, operating as designed, produce an Individual who can never achieve stable final satisfaction, and whose restlessness takes different forms depending on which system dominates.

9.8 Habits and the Cached-Value Store

9.8.1 The Efficiency Rationale for Caching

Real-time integration is computationally expensive. Converting source-specific signals to common currency, weighting them, summing them, comparing options, and selecting behaviour requires neural resources and time. For novel decisions in uncertain environments, this expense is justified; careful integration produces better outcomes than hasty response.

But most decisions are not novel. The Individual encounters recurring situations (morning routines, commute choices, mealtime patterns, social rituals), where the optimal response has been computed many times before. Re-computing from scratch each time would waste resources without improving outcomes.

The brain solves this efficiency problem through caching: retaining the value that repeated integration in a given context has settled on, and acting on it directly when that context recurs. A habit is an entry in that store, an association between a context-pattern and a cached value-plus-response, acted on without a fresh integration cycle. The store is intrinsic to the Decision Hub, which is where action-selection occurs.

9.8.2 The Structure of a Cached Entry

A cached entry contains:

Context-signature: The pattern of cues that triggers retrieval. This may include environmental features (location, time of day, sensory cues), internal states (hunger level, fatigue, mood), social context (who is present, what they are doing), and recent behaviour (what was just completed).

Cached value: The value that repeated integration in this context settled on. It is a single Common-Currency Scalar of the kind the Decision Hub emits, carrying no record of which components contributed to it, because the integration that produced it destroyed the source-tags before the value was ever stored.

Response: The behavioural output associated with that value. This may be a specific action (reach for phone, eat breakfast, take usual route) or a more general orientation (attend to social cues, prioritise efficiency, seek stimulation).

When the context-signature is matched, the entry is acted on directly: the stored value enters action-selection without a fresh integration cycle, and the associated response is executed. This bypass is what makes habits fast, because the slow integration process is skipped entirely.

The Belief Register is a separate mechanism, and the two are easily conflated. The Belief Register is the architectural channel through which the Thinker's posteriors reach the Observer's preconscious Salience computation. What the Observer draws from it is prior-configuration: who counts as a Model, what counts as comparative position, which Objects are status-charged or aversion-coded, and the categories through which the field is parsed. That configuration conditions the Salience the Observer assigns to what it perceives, and the Observer reads prior-configuration, not values, with its signed Salience output feeding the Decision Hub on the live cycle. Habit-routed selection does not run through the Observer at all; it runs on the Decision Hub's own cached-value store. The two routes are revision-limited in different ways, and together they are the architectural reason a deliberative I-update does not produce immediate behavioural change. The Thinker's posteriors must be rewritten, and the rewrite is propagation-rate-limited, before the Observer's salience-conditioning tracks the update. The cached-value store never tracks it, being read against no posterior at all, and is reconditioned only by repetition.

9.8.3 System 1 and System 2 Mapping

The AIM caching architecture maps onto dual-process models of cognition (Kahneman’s System 1/System 2):

System 1 (fast, automatic, effortless):

Cached retrieval pathway
Context-triggered activation
No deliberation required
Cached value acted on directly rather than computed
Executes before conscious awareness

System 2 (slow, deliberative, effortful):

Real-time integration pathway
Novel or conflicted situations
Deliberation and comparison
Source-weights computed from current signals
Conscious awareness of decision-process

Most behaviour is System 1: habits firing in response to context-matches. System 2 engages when:

No cached entry exists (novel situation)
Multiple cached entries compete (ambiguous context)
Cached response conflicts with current goals (habit vs. intention)
Strong signals demand attention (high-intensity A, I, or M error signals)
Deliberate intervention occurs (I-Override)

What the architecture adds to the dual-process picture is a mechanism rather than a further description. System 1 and System 2 are not two faculties but two routes through one architecture, distinguished by whether action-selection acts on the cached-value store or runs a fresh integration cycle. How the two routes are coupled, and why the system errs in the way it does, is given in 9.8.4.

9.8.4 Kahneman and the Fast-Slow Loop

Kahneman's distinction between System 1 and System 2 (Kahneman, 2011) can be read as an important phenomenological identification of what AIM describes more precisely as the fast and slow loop. What that framework did not provide was a substrate-level or information-theoretic account of how those loops are coupled, where the coupling occurs, and why the overall system is systematically prone to error. Chapter 4 supplies that account. The fast loop delivers to Conscious Cognition only a tag-stripped, signed integrated Scalar generated by the Decision Hub. The slow loop feeds back through prior-maintenance and Bayesian ground-attribution in the Thinker, whose posteriors are propagated via the Belief Register to condition the Observer's subsequent Object-field Salience computation, which in turn feeds the next round of integration at the Decision Hub. The characteristic errors of introspection and deliberation follow directly from that architecture.

9.8.5 Habit Formation: Crystallising Integration Episodes

Habits form through repeated integration episodes in similar contexts. The process:

Initial encounter: Novel context triggers real-time integration. Sources are weighted, options compared, behaviour selected.
Outcome feedback: The behaviour produces consequences. Reward signals (positive or negative) update the value of the selected response.
Repetition: The same context recurs. Integration is performed again, but prior experience biases the weights toward previously-successful configurations.
Consolidation: After sufficient repetition, the context-value-response association is cached. The integration computation is no longer performed; the cached value is acted on directly.
Automaticity: The behaviour executes without deliberation, often without conscious awareness. The habit is established.

The integrations that laid the entry down were tag-stripped before their results were stored, so the entry carries a value and no record of what produced it. A habit formed when M-signals were dominant, such as a behaviour adopted because Models were doing it, is stored as a value like any other, and when it fires it fires on that value without consulting any current signal from any component.

This explains why habits can persist after the original motivation has changed. The Individual who started checking social media because everyone else was doing it (M-led) may continue checking after they no longer care about social comparison: the cached entry fires regardless of current M-signals because it does not consult current signals at all.

9.8.6 The Limits of Willpower-Based Habit Change

Willpower models of habit change assume the Individual can simply “choose differently” when a habit fires. This misunderstands the timing:

Cached retrieval completes in 100–200ms
Motor preparation begins at 150–250ms
Behaviour execution starts at 200–300ms
Conscious awareness arrives at 300–500ms

By the time the Individual is aware of acting, the action is already underway. “Choosing differently” requires intervening in a process that has largely completed before choice becomes possible.

This is why willpower-based habit change fails so frequently. The Individual resolves to change, encounters the triggering context, and finds themselves performing the habitual behaviour before they’ve had a chance to deploy their resolution. The habit is not “overpowering” their will; it is executing in a time-window where will has not yet come online.

Effective habit change requires intervention at a different level:

Context modification: Remove or alter the triggering cues so the cached entry is not activated
Implementation intentions: Pre-form competing cached responses that activate in the same context
Habit stacking: Attach new behaviours to existing cached entries
Environment design: Structure surroundings to make desired behaviours easy and undesired behaviours difficult

These strategies work with the caching architecture rather than against it. They recognise that the battle is not won in the moment of habit-execution (where willpower is too slow) but in the conditions that determine whether and how cached entries activate.

9.8.7 AIM-Specific Implications of Caching

The caching architecture has several AIM-specific implications:

Behaviour outlives the conditions that formed it. A habit laid down during a period of high M-sensitivity, status-anxious adolescence being the characteristic case, persists as a stored value into contexts where that sensitivity has passed. The Individual is behaviourally locked into a pattern that no longer reflects their motivational state, and cannot recognise it as such by introspection, because the entry driving the behaviour never carried a record of what installed it.

Cached habits bypass current A-I-M signals. This can be adaptive (efficient response without recomputation) or maladaptive (behaviour mismatched to current needs). The Individual in a cached routine may not notice that their A-state has changed (now hungry, now tired) because the cache doesn’t check.

I-Override cannot operate on habits directly. I-Override requires a desire to scrutinise. Habits execute without generating experienced desire: only experienced behaviour. I-Override for habits must therefore target the cache itself (through interruption and reconsolidation) rather than the desire (which doesn’t exist in the moment). Section 9.13.6 develops this point.

Habit-heavy lives can be source-opaque even with AIM vocabulary. An Individual whose behaviour is largely habitual may find the Two-Question Diagnostic unhelpful, not because the diagnostic is flawed but because there’s no live desire to diagnose. The diagnostic applies to real-time integration; habits bypass integration entirely.

Movement C: Epistemic Constraints

9.9 Source Opacity: The Architectural Constraint

9.9.1 From Integration to Opacity

Section 9.1 established that the Decision Hub integrates A, I, and M signals through common-currency conversion. The conversion process compresses three source-specific signals into a single Scalar, enabling comparison and selection.

This compression necessarily discards information. The integrated value $U(x)$ does not encode whether it was primarily contributed by $U_A$ , $U_I$ , or $U_M$ . The Scalar cannot carry its own decomposition. The source-tags that identified which channel generated which portion of the value are lost in the summation.

Source Opacity is this information loss, experienced from the first-person perspective: the Individual receives the integrated value as a unified wanting-signal without access to its source-composition.

9.9.2 What Source Opacity Is

Source Opacity is the architectural feature whereby motivational signals from A, I, and M are integrated in the Decision Hub with source-tags discarded during common-currency valuation, leaving the Individual without introspective access to the source-decomposition of the wanting that results. Because the three signals are not of one kind, the loss carries a consequence beyond anonymity. Two of them encode deficits over Objects, each with a termination condition; the third encodes Salience, a signed reading of which Objects the environment marks as valued and which as to be kept from, with no termination condition at all. The integrated Scalar admits one interpretive category, a signed demand for correction, so what integration delivers is not an unlabelled signal of indeterminate kind but a Salience reading in the form of a lack, or, where the reading is negative, in the form of a threat.

Key features:

Its principal cost is a category error, not an anonymity. The Individual does not merely fail to know which channel produced the pull. They receive a reading of the environment under the only category the Scalar supports, and act on it as though something were missing. This is the Salience-Deficit Conflation Principle. A deficit closes when its Object is obtained and the signal stops; a Salience reading closes on nothing, there having been no shortfall for the acquisition to repair. That M does not satiate and that M is felt as deficit are therefore one result stated twice rather than two findings that happen to coincide. Three conditions produce this jointly, and each is separately necessary: the Mimetic signal is a Salience reading rather than a deficit, integration destroys the tag that would have marked it as one, and the Mimetic channel leaves no phenomenological residue against which the resulting mis-categorisation could be checked (the Legibility Gradient, developed at Section 8.2.5 and restated in the key feature below). Remove any one and the result does not follow: a tagged Salience reading could be held as a reading, a Salience reading that left a residue could be examined against the wanting it accompanied, and a traceless deficit signal would still terminate of its own accord once its Object was obtained.

It is architectural, not psychological. Source Opacity is not a failure of attention, a motivated blindness, or a cognitive bias. It is a consequence of how the integration hardware works. The information is destroyed, not merely hidden.

It is universal wherever Common-Currency Integration is intact. All Individuals with intact common-currency integration systems experience Source Opacity. Variation exists in degree (some Individuals may have slightly more access to source-signals before integration) but not in kind (no Individual has reliable post-integration source-access).

It affects the formal source-tag symmetrically and bears asymmetrically. The integrated Scalar U carries no marker identifying which source contributed what, and this applies to A, I, and M equally. What is not symmetric is what each channel leaves behind for the Individual to work with once the tag is gone. A-signals produce bodily states (hunger, pain, cold, fatigue) that persist during conscious awareness independently of the Scalar. I-signals produce engagement states (flow, curiosity, absorbed attention) that persist during conscious awareness and are observable in private. M-signals produce no correlate of either kind: the Observer’s reading is integrated before awareness of the wanting arrives, and nothing of it survives into the conscious timeframe. This ordering of the three channels by what each leaves behind is the Legibility Gradient (Section 8.2.5), and the timing established at Section 9.11 is what produces the tracelessness rather than what constitutes it; the constraint on the Individual is that introspection has nothing to examine, not that the transmission was quick. Source Opacity’s practical consequence is therefore most severe for M, because M is the one channel that is both mis-categorised at integration and left with no residue against which the mis-categorisation could be checked. Object Parasitism and Confabulation compound this concealment but do not create it.

It is permanent and irreducible. Source Opacity cannot be overcome through training, insight, or willpower. The Individual cannot learn to “see” source-tags that have been destroyed. Compensatory strategies (Section 9.13) can infer likely sources from available evidence, but this is reconstruction, not direct access.

9.9.3 What Source Opacity Is Not

Not mere difficulty. Source Opacity is not “it’s hard to know why I want this.” It is “the decomposition of why I want this was destroyed at integration, and no reading of the wanting can return it.” Difficulty implies the information exists but is hard to access; opacity means the information does not exist in accessible form.

Not motivated self-deception. The Individual is not “fooling themselves” about their motivations. They genuinely lack access to source-information. Self-deception requires a self that knows and a self that is deceived; Source Opacity involves no hidden knowledge to deceive about.

Not a bias to be corrected. Cognitive biases involve systematic distortions of available information. Source Opacity involves the absence of information, not its distortion. Debiasing techniques address biases; they cannot address absences.

Not symmetric in practical consequence, despite symmetric tag-stripping. The formal source-tag is destroyed for all three sources in the integrated Scalar. However, A and I both produce phenomenological correlates that persist during conscious awareness: the hungry Individual experiences hunger pangs (A-phenomenology), the engaged Individual experiences absorbed interest that continues in private (I-phenomenology). These correlates provide diagnostic access to A and I even after tag-stripping. M produces no such correlate. The Individual’s practical opacity is therefore concentrated on M. This is not because A and I are “transparent” (their formal tags are equally destroyed) but because A and I offer compensatory phenomenological channels that M does not. What matters here is the absence rather than the timing that produces it: an Individual who had somehow witnessed the mimetic acquisition would still hold no residue to examine afterwards, and would still hold no category other than deficit in which to place what they had witnessed.

9.9.4 The Information-Theoretic Basis

Source Opacity follows from basic information theory. Consider the integration equation:

U(x) = w_A \cdot U_A(x) + w_I \cdot U_I(x) + w_M \cdot U_M(x)

This maps three source-specific valuations to one integrated value. The mapping is many-to-one: infinitely many different source-combinations can produce the same integrated value.

Example: an integrated value of 0.6 could result from:

an Appetitive valuation of 0.6 carrying the whole weight (pure A)
an Intrinsic valuation of 0.6 carrying the whole weight (pure I)
a Mimetic valuation of 0.6 carrying the whole weight (pure M)
Appetitive and Intrinsic valuations of 0.6 each carrying half the weight (equal A-I blend)
Infinitely many other combinations

Given only the integrated value of 0.6, it is mathematically impossible to recover the input decomposition. The information has been irreversibly lost, because the map is non-injective.

This is not a limitation of neural implementation; it is a mathematical necessity. Any system that integrates multiple inputs into a single scalar will destroy information about input-decomposition. The Decision Hub’s common-currency architecture guarantees Source Opacity.

Note (Appendix A cross-reference). The Source Opacity Theorem derived in this chapter is treated in Appendix A Section A.2 (M5a restatement) as a consequence of the Common-Currency Integration operation M3, and in Section A.6.1 as an essential (load-bearing) assumption of the fast valuation formalism. The chapter's derivation via the Data Processing Inequality remains canonical; the appendix treats it as a load-bearing entailment of M3 via the DPI, on which the fast-layer dynamics depend. The strict information loss follows from non-injectivity of the map; the Data Processing Inequality ensures that no downstream processing of the Scalar recovers what the integration discarded (Appendix A, Section A.2). The preconscious timing of the Observer's signal (Axiom 7) is a separate barrier to self-knowledge of the M-contribution, the Timing Barrier of Section 9.9.9, and it forms no part of Source Opacity.

9.9.5 Non-Recoverability Is Not Non-Existence

The Data Processing Inequality result invites a misreading. What the inequality establishes is that recovery of the source decomposition from the integrated Scalar alone cannot be done. It does not, and cannot, entail that any source contribution failed to shape the Scalar, nor that the contribution is beyond reconstruction by any means whatever. These are three different claims, and the distinction between them is load-bearing for everything the framework builds on Source Opacity.

Consider them side by side. The first is a claim about legibility from the output: given $U(x)$ alone, no function of $U(x)$ can return which of $U_A(x)$ , $U_I(x)$ , $U_M(x)$ produced which portion of the total. The second would be a claim about causation, that a source which is unreadable from the output therefore did no work in producing it. The third would be a claim about reconstruction in principle, that the source contribution can never be estimated by any route at all. Only the first is true. The integration map is a weighted sum, and every term in that sum contributes its full weighted value to $U(x)$ causally. Nothing is deleted in the causal sense. What is deleted is only the record, carried in the Scalar itself, of which term produced which portion of the total. A dominant M-contribution still moves $U(x)$ , still can flip which Object is selected when $U(x)$ crosses a competitor's value, and still leaves a behavioural trace in the choice that results. It is simply that reading $U(x)$ alone never tells you M did it.

The corollary's whole practical point turns on the third claim. The Data Processing Inequality bounds what any function of the Scalar can carry about the sources, and no more. It says nothing about what an estimator can recover when it is supplied with information beyond the Scalar. If one knows in advance that a source exists, knows its characteristic behavioural signature, and knows the context in which the valuation was formed, then the decomposition becomes estimable, not by inverting $U(x)$ (which cannot be done), but by inferential reconstruction that draws on that additional structure. Knowing the number and character of the contributing sources constrains the pre-image of $U(x)$ , and each further piece of prior knowledge (the M-signature of non-satiation, with audience-dependence in its special case, the A-signature of bodily set-points, the I-signature of engagement that persists in private) narrows it further. What results is an approximation of what went into the Scalar, and the approximation improves as the priors sharpen.

The impossible operation is exact inversion from the output on its own. The available operation is estimation of the input from the output plus independent knowledge of the input's structure, and the gap between the two is exactly where the framework's diagnostic and population-scale methods live. That estimation is Source Inference: reconstruction of an integrated valuation's source decomposition from Behavioural Signatures and independent knowledge of source structure, not inversion of the Scalar.

A stripped-down engineering analogy holds the three claims apart. Summing the left and right channels of a stereo recording down to a single mono track does not remove any of the instruments, you can still hear the same guitar, vocal, and drums that were in the stereo mix, because each instrument's sound is additively present in the combined waveform. What the mono track loses is only which channel each instrument came from, the left-right placement that the stereo image carried (this is the non-injectivity, and it is what the Data Processing Inequality rules out; and notwithstanding phase-cancellation effects, where left and right partially offset when combined, the analogy stands, since those effects change how the inputs sum, not the fact that the sum discards which channel each contribution came from). An instrument that plays louder pushes the mix louder, and one that falls silent quiets it, exactly as before. And an engineer who has the individual instruments on separate tracks, and knows which instrument was panned to which channel and at what level, can reconstruct an approximation of the two channels, not by inverting the mono sum, which cannot be done, but by reassembling the channels from those separate stems and their known pan assignments.

The mono track is a real and fully causally-determined fingerprint of both its inputs, and every instrument that went in is still audible in it. What is not legible, read off the track alone, is which channel any given instrument occupied. But even that is estimable, given the separate stems and knowledge of how they were placed. Opacity, in other words, is non-legibility of the source-attribution from the output alone, not non-existence of the source content and not non-reconstructability in principle.

Non-Recoverability Is Not Non-Existence. The Data Processing Inequality applied to the integration map establishes that the source decomposition cannot be recovered by any function of the integrated Scalar alone. It does not establish that any source contribution is causally inert, and it does not establish that the decomposition is unrecoverable in principle. Each source contributes its full weighted value to the Scalar, and those contributions retain their full behavioural consequences (selection, substitution, and choice), even though the Scalar retains no record of which source contributed which portion. Where an estimator is supplied with independent knowledge that a source exists and of its characteristic signature, an approximation of the decomposition can be reconstructed from the Scalar together with that knowledge. What cannot be done is inversion from the output on its own. Source Opacity is a destruction of the source-tag carried in the Scalar, not of the source-effect and not of the source-inference.

This is what makes both the Two-Question Diagnostic and the Aggregation Gradient possible. The diagnostic works at the level of the individual valuation precisely because it does not attempt to read the source-tag off $U(x)$ , it supplies independent behavioural evidence about which source is present and reconstructs from that. The Aggregation Gradient (Section 9.10) works at population scale for the same reason at the level of the outcome, since if opacity meant non-existence rather than non-legibility (if the M-contribution's effects did not survive the compression that destroys its scalar-borne tag), population-scale outcome analysis would have nothing to detect. The Gradient's whole premise is that effects survive while tags do not, and its whole method is inferential reconstruction from those surviving effects. The same premise licenses the later results on manufactured versus reliable valuations, since a single valuation can be produced by offsetting errors that happen to land on a given $U(x)$ , but a reliable pattern across many valuations cannot be so produced, precisely because the underlying source contributions are still doing real causal work in each case even while each remains individually unreadable from its own Scalar. Non-legibility from the output alone coexists with full causal potency and with reconstructability given independent knowledge, and it is that combination, not any one part alone, that the framework relies on.

9.9.6 Why Evolution Selected for Opacity

If Source Opacity creates problems (as later sections establish), why did evolution select for Source Opacity? The answer lies in the adaptive value of unified action.

The ancestral problem: An organism facing a predator while hungry and observing conspecifics fleeing cannot afford to deliberate about which motivation should govern. It must act: immediately, coherently, with full commitment. Paralysis is death.

The integration solution: Common-currency valuation produces a single, action-ready output. The organism doesn’t experience “I have a hunger-want, a fear-want, and a conformity-want; let me compare them.” It experiences “RUN” or “EAT” or “FOLLOW”, a unified imperative that commands behaviour.

The opacity trade-off: Unified action requires integrated valuation. Integrated valuation requires common-currency conversion. Common-currency conversion is a many-to-one map. That map destroys the source decomposition. Source Opacity is the cost of decisive action.

Evolution did not select for Source Opacity as such; it selected for common-currency integration because unified action had survival value. Source Opacity is a by-product: an architectural consequence that was not itself selected for or against because, in ancestral environments, knowing the source-composition of one’s desires rarely mattered for survival.

The environment has changed; the architecture has not. Source Opacity was adaptive-neutral in ancestral conditions and has become maladaptive in conditions where M-signals are artificially amplified and where distinguishing A-needs from M-wants would prevent significant harm. This evolutionary mismatch is developed in Chapter 10.

9.9.7 Source Opacity in Cached Decisions

Source Opacity applies not only to real-time integration but also, and with compounded effect, to cached retrieval.

When a habit forms, the integration that produced the cached entry lost its source-tags through standard common-currency conversion. The cached entry stores the result of that integration (context → value → response) but not the source-decomposition that produced it. The Individual who formed a habit of morning social-media checking does not store “this behaviour was M-dominant”; they store “in morning-context, execute this response.”

When the habit fires, the cached entry is acted on without any integration occurring at all. There is no moment where A, I, and M signals are weighted and combined; the computation was performed previously and the result was stored. The Individual experiences the behavioural impulse directly, without even the post-integration wanting signal that accompanies real-time decisions.

This creates double opacity:

Layer	What is Lost	When Lost
First layer	Source-tags (which channel generated each input)	During original integration that formed the habit
Second layer	Integration process itself (any experience of weighing or deliberating)	During cached retrieval that executes the habit

For real-time decisions, the Individual at least experiences the integrated wanting, “I want X”, even though the source-composition is unavailable. For habitual behaviour, even this is bypassed. The Individual finds themselves doing X without having experienced wanting X. The behaviour emerges from cache-retrieval, not from any process the Individual can inspect.

The phenomenological consequence is that habitual behaviours feel like “just what I do” rather than “what I want.” There is no subjective wanting to examine because the wanting occurred historically, during habit formation, not during habit execution. The Individual who habitually checks their phone upon waking does not experience “I want to check my phone” followed by checking: they experience finding themselves checking, with no antecedent wanting available to introspection.

This has significant implications for I-Override. The Two-Question Diagnostic assumes a desire to examine: “Do I want this because my body needs it?” and “Would I still want this if no one ever saw?” But habitual behaviour presents no desire to the Individual at the moment of execution: only behaviour. The diagnostic questions must be applied retrospectively (“Why did I just do that?”) or prospectively (“Why do I keep doing this?”), not in the moment of action where no deliberative pause occurs.

Source Opacity in cached decisions is therefore more severe than Source Opacity in real-time integration. Real-time integration produces opaque wanting; cached retrieval produces opaque doing. The Individual has even less material to work with when attempting to understand their own habitual behaviour.

9.9.8 Source Opacity in Social Interaction

Source Opacity has a domain of application that extends beyond the Individual's assessment of their own desires. It applies with equal force to the Individual's assessment of their social interactions. When Individual A defers to Individual B, grants B's claims greater credibility, or directs sustained attention toward B, the motivational source of that behaviour is opaque to A. A cannot determine from the valuation itself whether the deference reflects genuine A-relevance (B controls resources A needs), genuine I-relevance (B possesses expertise A values for its own sake), or PM-transfer (A defers because others defer, and the social proof has been integrated into A's valuation of B before conscious evaluation could intervene). The Confabulation system (Movement D) generates plausible explanations for A's deference ("I respect their expertise," "they've earned their position," "they clearly know what they're talking about") that systematically favour A-sounding and I-sounding attributions over the M-accurate account ("I defer because I have observed others deferring, and that social proof was integrated preconsciously into my valuation of this Individual"). This interaction-level Confabulation operates through exactly the same mechanism as desire-level Confabulation: the Information Gap (true source unavailable due to Source Opacity) is filled by the narrative system using available Templates, which do not include the M-Template.

The consequence is that PM-transfer between Individuals proceeds unrecognised by either party. The giver of deference experiences it as earned recognition. The receiver experiences it as appropriate response to their genuine qualities. Both are Confabulating, and the jointly produced explanation that each is responding to the other's earned qualities is a Reciprocity Narrative, the standard dyadic-scale Confabulation that conceals PM-transfer because neither party has introspective access to the source-composition of the valuation driving the interaction.

The implications of Source Opacity extend far beyond self-misunderstanding. Radiation is invisible to every human sense, yet it kills. For millennia, humanity could not detect it, could not name it, and suffered its effects without comprehension. Unmanaged Mimetic Desire occupies the same structural position in the social world: phenomenologically invisible to the Individual, yet producing effects (rivalry, escalation, Scapegoat designation, collective violence) that are exhaustively documented across every culture and epoch. Chapter 17 will establish that this invisibility, under nuclear conditions, is not merely epistemically unfortunate but existentially terminal.

The architectural concealment described in this section operates not only on the PM-component of A's valuation of B and on the PM-component of B's valuation of A, but also on the broader phenomenological output that the same architecture produces: the felt sense that one's desires originate within a sealed motivational interior, with other Individuals experienced as external to that interior. This phenomenological output is Phenomenological Isolation, the default persistent phenomenological state within which all conscious-cognition operations including Empathy are conducted. Section 9.11.1 derives the result from the conjunction of Preconscious Transmission (Axiom 7) and Source Opacity (Axiom 3) and distinguishes it from Empathy, which is a downstream conscious-cognition operation, and from the Dissolution of Individualism (Chapter 19), which is a claim about the architecture itself rather than about the phenomenology it produces.

9.9.9 Three Independent Barriers to Self-Knowledge of the M-Contribution

Self-knowledge of the M-contribution fails three times over. Three logically independent barriers seal the Individual's access to source-decomposition, and the removal of any one leaves the remaining two sufficient to preserve practical opacity for M. The three are the Integration Barrier, which is Source Opacity proper, the Timing Barrier, and the Conceptual Barrier.

The Integration Barrier is connection-invariant. The Decision Hub's common-currency integration maps the three-dimensional source-vector to a one-dimensional scalar, and by the non-injectivity of the weighted-sum map combined with the Data Processing Inequality, the source decomposition is irreversibly destroyed. This destruction is fixed by the integration operation itself, so it is invariant under the inter-component connection structure of the Body Manager, the Thinker, and the Observer. Whether or not direct channels couple these components, the tag-stripping occurs at the point of common-currency conversion, not at the point of inter-component signalling. Connection topology is therefore moot for Source Opacity, even though the Thinker-to-Observer connection (via the Belief Register) remains load-bearing for prior-maintenance and for I-Override leverage across successive rounds.

The Timing Barrier is fixed by the conscious threshold. The compensatory phenomenological correlates that survive tag-stripping (bodily states for A, engagement-persistence for I) are available to noticing only because they persist within the conscious timeframe. The Observer's Salience computation operates within the Transmission Window (60–340ms) and is integrated before the conscious threshold (300–500ms), so the mimetic acquisition event is never registered and M produces no surviving correlate. This is the Speed-Selectivity Trade-Off: the preconscious speed that enables cumulative culture is faster than the conscious evaluation that would filter observed goals, which is why pre-adoption filtering is impossible. The residual diagnostic asymmetry between A and I on one side and M on the other is therefore fixed by the conscious threshold, not by connection.

The Conceptual Barrier is fixed by the Hypothesis Space. Even granting, counterfactually, that the Timing Barrier were removed and a fragment of the acquisition event were available to noticing, noticing is not representing. To ascribe probability to the Hypothesis that a desire was acquired from another, the Individual must possess the M-Hypothesis as an element of the represented Hypothesis Space, because Bayesian updating redistributes probability only across Hypotheses present in that space. Absent the M-Template, the M-Hypothesis is not assigned a low probability; it is unrepresentable, and the question of its probability does not arise. This is Type I Exclusion (Structural Exclusion): the true generative Hypothesis is orthogonal to the representable subspace, so no quantity of evidence can move mass toward it, and the Individual, reasoning correctly from incomplete premises, converges on the least-wrong A-narrative or I-narrative by the Confabulation Rationality Lemma. The Conceptual Barrier corresponds to Layer 7 of the Camouflage Architecture (the Vocabulary Gap), which blocks Hypothesis formulation rather than detection.

The three barriers are logically independent. The Integration Barrier removes the formal tag, the Timing Barrier removes the surviving evidence, and the Conceptual Barrier removes the Hypothesis to which any such evidence could be assigned. Defeating connection topology does not defeat integration, defeating timing does not defeat the Hypothesis-space restriction, and supplying the Hypothesis (the Knowledge Patch (the Patch)) does not restore the destroyed tag but instead routes around it by directing residual probability mass, generated by independently sourced doubt, toward the now-representable M-Hypothesis. This threefold barrier is the structural reason the Knowledge Patch is necessary rather than merely helpful: no improvement in phenomenological access could be self-correcting, because correction requires the M-Hypothesis to be pre-installed before any evidence can be bound to it.

The three conditions of Section 9.9.2 and the three barriers of this section answer different questions. The conditions state what the Individual receives: the Mimetic signal is a Salience reading rather than a deficit, its tag is destroyed at integration, and it leaves no residue behind, so what arrives in awareness is a reading of the environment wearing the form of a lack. The three barriers of this section state why self-examination cannot correct that. The Integration Barrier has destroyed the tag that would name the source, the Timing Barrier has left no evidence for the Individual to notice, and the Conceptual Barrier withholds the Hypothesis to which any such evidence could be assigned, so an Individual who set out to determine where their wanting came from would find, in order, nothing to read, nothing to notice, and nothing to conclude. The first triad describes the signal. The second describes the search that fails to identify it. They share the destruction of the tag, which figures in both, and are otherwise distinct.

9.9.10 The Salience-Deference Identity One Salience Function Across the Agent-Type Vector

Section 9.9.8 established that Source Opacity applies to the Individual's valuation of other Individuals, and that the transfer channels through which the Mimetic Premium flows (attention, deference, narrative authority, positional concession) operate invisibly to both parties because the PM-component of each party's valuation has had its source-tag destroyed at integration. That result generalises. The deference an Individual directs toward a high-Status conspecific and the deference the same Individual directs toward a more powerful non-conspecific agent are outputs of a single Salience-valuation function, integrated and tag-stripped by the same Decision Hub through the same operation. This is the Salience-Deference Identity.

The Decision Hub does not run one valuation procedure for conspecifics and a separate procedure for predators, rivals, or Prey. It processes every agent through the Agent-Type Vector, the four-way classification of any perceived agent along the Collaborator (C), Rival (R), Predator (X), and Prey (Y) axes (developed in Chapter 11). Each axis contributes a signed Salience loading to the Common-Currency Scalar, and the integration operation that combines them is the same weighted summation, subject to the same non-injectivity, that Section 9.9.4 showed destroys source-identity. The output is a single signed valuation of the agent, and the behavioural response that valuation selects (approach, avoidance, deference, display) is chosen on the Scalar alone. The Individual has no more introspective access to why they defer to a powerful agent than to why they want an Object.

The consequence is that deference is not a distinctively social or distinctively mimetic response. It is the general behavioural output that high received Salience selects whenever the agent is valued as standing above the Individual in the relevant asymmetry, and the source of that asymmetry is opaque after integration exactly as the source-composition of any desire is opaque. A predator's capacity to harm loads the X axis and a high-Status Model's position loads the C axis, and an Individual deferring on either can no more read off which, from the felt valuation, than read off why they want an Object.

The asymmetry does not always run that way, and the response follows its direction rather than its magnitude. Y-weight loads on the inverse criteria to X: reduced capacity to retaliate, visibility, and position outside the network that would make targeting costly. An agent carrying Y-weight is valued as standing below rather than above, and what high Salience selects there is not deference but extraction, on the reciprocity-of-extraction profile the Y axis carries. Deference and extraction are one function read at two signs, and the Individual has no more access to the ground of the second than to the ground of the first.

The R axis differs from the other three in what places an agent on it, and the difference changes how R-weight behaves. X, Y and C load on a capacity or a position the agent carries and would carry whether or not this Individual were present: the predator can harm, the Prey-classified agent cannot retaliate, the Model stands where it stands. R loads on a relation rather than a property. An agent occupies the Rival position by directing action at the Object this Individual's own action is directed at, so R-weight is set by that relation rather than by anything the agent is. This is why an agent can occupy the Rival position without having changed in any respect, and why the disposition recorded about that agent during a contest outlasts the Object that produced it (Chapter 16, Section 16.3).

What we observe in human society as status-deference (yielding attention, conceding position, granting narrative authority to the higher-ranked) is the C-and-R-dominant instance of the same salience-computation that, when the X axis dominates, produces the deference an organism shows toward a more powerful agent. The Mimetic Premium is therefore best understood not as a separate quantity but as the received, gradient valuation of emitted Salience in the special case where the dominant axes are conspecific, and it flows unidirectionally toward higher Salience for the same reason deference flows unidirectionally toward the more powerful agent in any asymmetry.

This identity is derived, not postulated. It follows from three results already in place. First, Common-Currency Integration (Sections 9.1.3 and 9.1.4) requires that all agent-relevant inputs be converted to a single Scalar before selection, which forecloses any architecture in which conspecific-Salience and threat-Salience or resource-Salience are carried on separate, independently readable channels. Second, Source Opacity (Axiom 3, Section 9.9) guarantees that the axis-decomposition of the agent-valuation is destroyed in that conversion, so the Individual cannot introspect whether their deference tracks Status or threat. Third, the Agent-Type Vector (Axiom 9, Chapter 11) establishes that the same perceptual apparatus classifies all agents along the same four axes, so there is one input format for the integration operation to consume. Given these three, a single Salience-valuation function across all agent-types is not an additional assumption but the only architecture consistent with the ones already derived. The apparent specialness of social deference is an artefact of the Vocabulary Gap (Section 9.9.9, Conceptual Barrier), which supplies a conspecific-status Template but no unified Salience-deference Template, so the shared computation is described in two vocabularies (submission to power, respect for status) that conceal their common architectural origin.

Two clarifications prevent over-reading. The identity is an identity of mechanism, not of magnitude or valence. The axes carry different signed loadings and select different responses (the X axis, when threat dominates, selects avoidance or submission rather than the attention-and-emulation that C-dominant deference produces), and nothing here collapses the distinction between fearing a predator and admiring a leader at the level of felt experience, which the Phenomenological Reality Principle (Section 9.5) preserves. What is identical is the valuation-and-tag-stripping operation that generates the deferential output from received Salience, whatever the dominant axis. And the identity does not make the Mimetic Premium reducible to threat or resource dynamics, because on the conspecific axis the Premium is non-satiating and unidirectional (the Boundedness Asymmetry Principle, Section 9.10) in a way that A-grounded threat-avoidance and resource-approach are not. The Salience-Deference Identity unifies the computation; the Boundedness Asymmetry Principle preserves why the conspecific instance alone compounds.

A final qualification completes the identity. The Salience loading on the C and R axes is only one of the three source-contributions the Decision Hub integrates when it valuates an agent, and it is not by itself decisive as to which axis effectively governs the response. The agent-valuation is the same weighted summation of Appetitive, Intrinsic and Mimetic contributions as any other, so the A-contribution (the agent's bearing on the Individual's bodily needs and safety) and the I-contribution (the agent's bearing on the Individual's Intrinsic engagement) enter the sum alongside the transmitted Salience and can reinforce it, discount it, or override it. Which axis an agent falls into for the purpose of action-selection is therefore determined by the integrated Scalar summed across all three sources, not by the Salience channel alone. A high-Status Model who is simultaneously an acute bodily threat may draw avoidance rather than deference, because a large negative A-contribution dominates the positive Salience loading in the sum. Conversely, a low-competence agent whom the Individual can directly assess as such may still be deferred to, because a sufficiently strong transmitted Salience loading dominates the low I-contribution in the sum, which is the ordinary architectural explanation for why demonstrably unintelligent high-Status figures continue to command deference, attention, and narrative authority.

The Salience-Deference Identity holds at the level of the mechanism, that received Salience is valuated and tag-stripped identically across agent-types, but the case in front of the Individual is always decided by the summed valuation, in which the Mimetic channel is one input among three. This is why the same architecture that makes status-deference near-automatic in the ordinary case also permits it to be overridden when A or I signals are strong enough to dominate the sum, and it is the agent-classification counterpart of the general point, established throughout Movement C, that no single source is legible or decisive in isolation after integration.

9.9.11 The Opacity of Social Valuation Why Valuing Another Rightly Once Certifies Nothing

The preceding subsections established that when one Individual values another, the source-composition of that valuation is opaque. The Individual cannot introspect how much of their regard for another reflects genuine A-relevance, genuine I-relevance, or transmitted M-valuation (Section 9.9.8), because the same integration that strips source-tags from the valuation of an Object strips them equally from the valuation of an agent (Section 9.9.10). A consequence follows that governs not why an Individual values another but how far that valuation can be trusted as accurate, and it can be established entirely at the level of valuation.

When Individual A values Individual B, the Decision Hub returns a single integrated Scalar, namely how much A values B, all things considered, and that Scalar is the only quantity to which A has access. Underneath it lie the three source-contributions, weighted and summed: B's relevance to A's bodily needs, B's relevance to A's engagements, and the transmitted valuation carried by how visibly others value B. The summation is many-to-one. Distinct combinations of the three contributions produce the very same Scalar, and once the Scalar is formed the combination that produced it cannot be recovered by any inward operation (Section 9.1.5). A therefore holds a number without its composition.

This sets a ceiling on how accurate the valuation can be, and the ceiling is fixed by the contributions, not by the care A takes in forming the Scalar. If the contributions that entered the valuation carry good information about how dealings with B will actually go, that is, if A's regard is built from features of B that bear on what B will do, then the Scalar is a good guide, and A's valuation is accurate. If instead the dominant contribution is the transmitted one, reporting only how others rank B and nothing about what B will do, then the Scalar is an accurate summary of B's standing and a poor guide to B's conduct, and no amount of deliberation applied to the Scalar afterward can put into it information the contributions never contained. The valuation cannot be more accurate about B than the material it was integrated from. Deliberation operates downstream of integration, on the Scalar; it cannot reach back to enrich the contributions that made it.

The sharper consequence concerns what a correct valuation does and does not certify. A can value B correctly, arriving at very nearly the right regard for how a dealing with B will in fact go, while the contributions underneath that valuation are individually wrong, provided their errors offset in the summation. Overvaluing B's relevance in one respect and undervaluing it in another leaves a Scalar close to the mark though neither contribution was sound. Because the summation is many-to-one, the combinations of contributions that yield any given correct valuation are not one accurate composition but a whole family of mutually incompatible wrong ones, each producing the right Scalar by a different route of cancellation. Valuing another rightly on a given occasion is therefore cheap. It can be had by luck, by offsetting error, or by contributions that are wrong in every part but wrong in a way that nets out. What cannot be had cheaply is valuing others rightly reliably, across many people and many dealings, because cancellation is coincidence, and coincidence does not repeat to order. To land rightly again and again as the people and circumstances vary, the contributions themselves must track what those people will do, and there is no route around that which survives repetition.

The two points together dissolve a natural defence. An Individual who has judged others well in the past is inclined to trust their read of the next person, treating the record as evidence that the faculty is sound. But a run of correct valuations is not evidence that the contributions feeding them are informative, because any single correct valuation can be manufactured by offsetting error rather than by sound contributions, and the Individual, by Source Opacity, cannot tell the two apart from the inside. The felt confidence attaches to the Scalar, and the Scalar is silent as to whether it was well-founded or merely lucky. Confidence and a history of being right are both fully compatible with a valuation faculty that has never been accurate about anyone at all, only fortunate in how its errors have so far cancelled.

A further route to unwarranted confidence arises the moment the target is not inert but is itself an Individual valuing back. When A deals with B, A's conduct in the encounter carries A's valuation of B, and B reads that conduct and adjusts B's own behaviour to steer the encounter toward the outcome B wants. The consequence for A is that the conduct A observes from B is not, to the extent it was so adjusted, conduct independent of A's valuation against which that valuation could be tested; it is conduct that A's own signalled valuation has helped to shape.

If A treats B as promising and behaves accordingly, B, reading this, may meet it in the way that makes the dealing go as A expected, not because A's valuation captured anything true about B, but because B has accommodated the valuation A displayed. A then observes an interaction that went as predicted and takes the prediction to have been confirmed, when what has occurred, in the measure that B accommodated rather than revealed disposition, is that the encounter supplied A the very outcome A's conduct invited. The interaction does not to that extent function as an independent test, because the evidence it returns was co-produced with the valuation it appears to confirm. This is a valuation-layer statement of a dynamic that the belief-dynamics analysis derives in full, where the mutual, live-updating version of the loop and its consequences for self-knowledge are set out as the Interactional Confirmation Loop (Chapter 14, Section 14.8.9).

This bears directly on the danger the earlier subsections identified. The contribution that most readily inflates or deflates A's valuation of B is the transmitted one, the visible signal of how much others value B, and it is precisely the contribution least informative about what B will do, because it reports B's standing in others' regard rather than anything about B's conduct. It therefore raises A's confidence in the valuation without raising its accuracy, making A feel better positioned to judge B while leaving A no better positioned at all.

The corrective is the one the framework applies throughout, now with a second and independent justification. Because valuing another rightly once certifies nothing about the contributions behind it, the felt confidence in the valuation must first be held under Epistemic Suspension, withheld precisely because it attaches to a Scalar that cannot certify itself, and the valuation must then be rebuilt on evidence of what an Individual does rather than on evidence of how others regard them. The behavioural signatures of Section 9.14.3, which Epistemic Suspension (Chapter 18, Section 18.5) exists to send the Individual toward, are informative about conduct in the way transmitted regard is not, provided they are read across many instances so that one-off cancellations average out, and provided the conduct read is conduct that reveals the Individual's disposition rather than conduct elicited in accommodation to the reader's own signalled valuation. Valuing another by the standing they carry raises confidence without raising accuracy; valuing them by what they reliably do raises accuracy itself, and the two are indistinguishable from the inside, which is why the discipline must be procedural rather than felt.

The formal treatment of why the transmitted contribution is uninformative, and of how the valuation is revised over successive encounters, belongs to the belief-dynamics analysis, where these results are derived as the fidelity bound and the identifiability result (Chapter 14, Section 14.10.10).

9.9.12 Status as a Source-Opaque Valuation Input and the Substrate It Lacks

Section 9.9.11 established that a valuation of another is source-opaque and that valuing someone rightly once therefore certifies nothing about the contributions behind it. A further question concerns one of those contributions in particular, the transmitted signal of how much others visibly value a target. This contribution, the Status contribution, behaves within the valuation like any other, but beneath it lies a feature that no ordinary Object-contribution shares, and identifying it completes Movement C's account of social valuation.

When A values B, the visible signal of B's standing arrives as an M-input, is assigned a weight, is summed with the A- and I-contributions, and is stripped of its source-tag in the integration exactly as any M-input is (Section 9.1.5). Status is thus not a property A perceives sitting on B and reads off directly. It is a contribution to an integrated Scalar, and A experiences only the Scalar. The phenomenology misrepresents the architecture. The Individual feels themselves to be perceiving another's standing, as though standing were out there to be seen, when what is in fact occurring is the integration of a transmitted contribution into a valuation whose composition is, by Source Opacity, closed to inspection. Standing is not observed but computed and then felt as though observed.

The feature peculiar to this contribution appears when one asks what remains if it is removed. For an ordinary Object, the transmitted contribution, the portion of the Object's valuation driven by the fact that others value it, rides atop an A and I substrate. Strip the transmitted portion away and that substrate remains, the Object's relevance to a bodily need or an engagement, a contribution with a Referent the world can eventually test. A meal valued partly because others prize it is still, underneath, food.

Status is not the exception to this that it first appears. It has no substrate of its own, distinct from the A and I, on which it independently rests; it is the M-contribution riding on the same A and I substrate the Object already has, position in a comparison-constituted ordering carried on the use or need that others are seen to prize. The point is therefore subtractive rather than a matter of Status possessing some emptier substrate than a good does. Strip the Status contribution away and what is exposed is only the A and I to which Status itself added nothing recoverable, together with, in the pure case of conspecific standing, further comparison, standing relative to others who are themselves ranked relative to others. There is no floor of substrate under Status other than the A and I floor that was always the real substrate, and to that floor Status contributes nothing.

This is the individual-scale root of a property the chapter attributes to the conspecific case at Section 9.10. A valuation contribution that rests on a substrate of its own can settle, because it answers to something outside itself: the A-contribution to valuing food subsides as the deficit it answers to is corrected. The Status contribution rests on no substrate of its own, adding nothing to the A and I it rides upon, so it answers to nothing outside the comparison that constitutes it, and each position attained is immediately a position relative to a new set of others, so the contribution regenerates. The non-satiating, unidirectional character that the Boundedness Asymmetry Principle identifies in the conspecific instance is therefore not an additional empirical fact about Status but a direct consequence of its resting on no substrate of its own: a contribution constituted wholly by comparison has no closure condition, because comparison has no terminus.

This content-level account, that non-satiation follows because comparison has no terminus, is the same result seen from the content side that Chapter 8, Section 8.5 derives from the channel side, where the M-channel is shown to be an identification service with no deficit of its own and therefore no closure condition. The absence of a substrate of its own is why the comparison never terminates, and the open-loop channel is how the architecture runs a valuation that never terminates, so the two are one derivation of non-satiation viewed from its content and from its mechanism, not two competing roots.

How the transmitted signal comes to dominate the valuation before any encounter, how weakly the target's own conduct revises it, and how the many individual valuations across a population condense into the single public quantity called a person's Status, are questions about the transmitted contribution's origin and revision that belong to the belief-dynamics analysis, where they are derived as the Belief-Conditioned Status Estimate and its low-pass condensation (Chapter 14, Section 14.8.8).

9.10 The Aggregation Gradient as the Structural Complement to Source Opacity

The Source Opacity Theorem establishes that the Decision Hub's integration of A, I, and M into a Common-Currency Scalar irreversibly destroys the source-tag on the integrated value within the Individual at the moment of integration. This is the architectural floor of the framework. It is the rung at which Confabulation operates as the immediate post-hoc completion of the missing source explanation, and it is the rung at which the Reciprocity narrative and the parasitic attachment of Mimetic valuation to Objects of Appetitive or Intrinsic origin derive their feasibility.

It is also a local claim. It is local to a single Decision Hub, local to a single moment of integration, and local to the representation produced at that moment. None of these localities is the whole world. The information that is destroyed at the Decision Hub is destroyed within the Individual's representation, not erased from the structural record of what the Individual then did, with whom, in what pattern, and to what cumulative effect. This locality is what permits the Aggregation Gradient to operate as the structural complement to Source Opacity.

The Aggregation Gradient is the monotonic relationship between representational distance from any single Decision Hub and the legibility of the Mimetic Premium. It comprises six rungs. Rung 1 is the Individual single moment, where Source Opacity is total. Rung 2 is the dyadic transaction, where two opacified Decision Hubs interact and produce a jointly sincere Reciprocity narrative. Rung 3 is the Individual longitudinal scale, where persistent behavioural patterns across many transactions become legible to the Individual's own deliberative system once the framework's terminology is in place. Rung 4 is the small group or network scale, where mimetic contagion and prestige-biased transmission leave traces in interaction patterns. Rung 5 is the institutional scale, where the institutional defences a society has erected against Mimetic-Premium accumulation, and the structures it has erected to conceal that accumulation, become tractable to outside analysis. Rung 6 is the macro-historical scale, where population-scale outcome residues (concentration of market power, concentration of coercive power, cumulative Net Mimetic Position divergence, A-Floor erosion patterns) display their full diagnostic clarity. That residual observability at population scale, against the architectural invisibility of the Mimetic Premium at the dyadic and transactional rungs, is the Macro-Residue Visibility Principle.

The mechanism by which legibility increases up the gradient is the Low-Pass Filter Mechanism. A is bounded by satiation; A-pursuits cycle but do not compound. I is bounded by sufficiency of the Object to enable the process; an instrument or resource that meets the I-requirement at a given level of competence keeps meeting it (the competence itself can develop indefinitely; what is bounded is the resource requirement at any given level of competence). M is structurally non-satiating and unidirectional. Aggregation across many Decision Hubs and longer time periods averages the bounded source contributions toward stable substrate values, while the unbounded source contribution compounds and produces a divergent signal that aggregation cannot suppress. Aggregation is therefore not merely averaging. It is a change in the measurement object, from "what one Individual reports about why they did one thing" (where Source Opacity is fatal) to "what the distribution of all the things that all the Individuals did looks like" (where Source Opacity has no purchase). Aggregation is source-discriminating where integration is not. The formal apparatus for this claim is set out in Appendix A.

Two corollaries follow. First, the Mimetic Premium is not the dominant component of typical interpersonal transactions. It is a persistent minority component that compounds because it is non-satiating and unidirectional, not because it dominates any single transaction. This is the Per-Transaction Minority Claim, and the Per-Transaction Minority Claim is what makes Source Opacity's concealment feasible: Confabulation has a thin layer to cover, not a thick one. Second, the same eliminative inferential structure that drives source-attribution at the individual scale (ruling out A and I to identify M, by the Boundedness Asymmetry Principle (the Boundedness Asymmetry Principle, the architectural fact that A is bounded by satiation and I is bounded by sufficiency of the Object to enable the process, while M alone never closes while the Individual is conscious and perceiving, because relative position has no terminal value)) drives source-attribution at the population scale through the Low-Pass Filter Mechanism. The diagnostic instrument changes; the inferential structure does not. The Knowledge Patch developed in Chapter 18 operates at every rung of the gradient with rung-specific instruments unified by this shared inferential structure. The formal counterpart of the Boundedness Asymmetry Principle is the functional derivation of Appendix A (M5b): the conspecific instance compounds because the salience channel runs open-loop with respect to the deficits it serves.

The within-Individual concealment that Source Opacity makes architectural is therefore not undone by the gradient. The gradient identifies the dimension along which detection survives within-Individual concealment: not introspective access at the dyad, which is permanently blocked by Source Opacity, but population-scale outcome analysis at higher rungs, which the framework's terminology makes source-discriminating in a way that earlier vocabularies could not.

9.11 The Timing Architecture

9.11.1 The Preconscious Window

Timing adds a separate barrier to the one integration imposes. Source-tags are destroyed during integration, and in addition the entire integration process completes before conscious awareness comes online.

The timeline:

Time Window	Process
0–100ms	Stimulus registered; source-specific processing begins
60–340ms	Preconscious Transmission occurs (mirror-neuron response to observed goals, spanning reactive motor-mirroring onset through predictive goal-state activation)
200–400ms	Common-currency integration in Decision Hub
300–500ms	Conscious awareness of wanting emerges
500ms+	Possible deliberative intervention (I-Override window)

Preconscious Transmission (the Transmission Window: 60–340ms) completes before conscious awareness (300–500ms). When an Individual observes a Model pursuing an Object, the mimetic signal is transmitted, integrated, and contributing to the unified value-signal before the Individual is consciously aware of wanting the Object.

The architectural consequence of this timing, taken together with the Source Opacity result of Section 9.9, is that the Individual's first-person phenomenology of motivational origin is systematically constructed under conditions that prevent it from tracking the actual source-decomposition of the integrated valuation. The mimetic adoption that partly constitutes the valuation occurs preconsciously (Axiom 7); the integrated output the Individual consciously experiences has had its source-tags destroyed (Axiom 3). The phenomenology the Individual receives is therefore one of self-originance, with desires presenting as if they had arisen within an enclosed self, because the architectural facts that would falsify that presentation are inaccessible. This is Phenomenological Isolation, the default persistent phenomenological state, the architectural background within which every conscious-cognition operation including Empathy is conducted.

Empathy, as a conscious-cognition Bayesian inference that compares an observed Individual's configuration against the Empathising Individual's stored memories of similar configurations, is downstream of Phenomenological Isolation; it furnishes inferred content about another Individual's state but does not penetrate the felt boundary of self-originance, because the boundary is generated upstream by mechanisms Empathy cannot reach. The Knowledge Patch (Chapter 18) accordingly operates through inferential reconstruction from behavioural evidence rather than through phenomenological repair; the phenomenology of self-originance does not disappear under the Patch, only the Individual's interpretation of it does. Phenomenological Isolation is distinct from the Dissolution of Individualism (Chapter 19), which is a claim about the architecture itself rather than about the phenomenology the architecture produces, and which Phenomenological Isolation persists despite, because the architectural facts the Dissolution articulates remain inaccessible to introspection regardless of whether the Individual has accepted the Dissolution at the conceptual level.

9.11.2 Libet and the Readiness Potential

Benjamin Libet’s classic experiments (Libet, Gleason, Wright, and Pearl, 1983) demonstrated that neural preparation for action precedes conscious awareness of intending to act. The “readiness potential”, a buildup of electrical activity in motor cortex, begins approximately 550ms before action execution, while conscious awareness of the “urge to act” appears only 200ms before execution.

This finding is often interpreted as challenging free will. For AIM, the relevance is different: it demonstrates that motivational processes operate on timescales faster than consciousness. The Decision Hub computes and prepares before the Individual is aware of deciding.

The Libet findings align with the AIM timing architecture: integration occurs in the 200–400ms window, awareness arrives at 300–500ms, and the gap between neural decision and conscious experience is the space in which the Timing Barrier operates; Source Opacity is fixed by the integration itself, at any timing. By the time “I want X” reaches awareness, the source-integration that produced that wanting has been completed for 100–200ms.

9.11.3 Implications for M-Interception

The timing architecture places severe constraints on intercepting Preconscious Transmission:

M cannot be intercepted preconsciously. The mirror-neuron response to observed goals is automatic and below the threshold of conscious awareness. The Individual cannot choose to “not see” a Model pursuing an Object or to “not register” the mimetic signal. The transmission occurs in hardware that operates before software-level control comes online.

M cannot be filtered during integration. By the time integration occurs (200–400ms), the M-signal is already present and contributing to common-currency conversion. The Individual cannot selectively exclude M from the integration because the Individual is not yet aware that integration is happening.

M can only be addressed post-integration. Conscious awareness arrives after the wanting has been computed. The Individual can examine the wanting, apply diagnostic questions, and choose whether to act, but they cannot prevent the wanting from being generated in the first place.

This is why AIM emphasises I-Override (post-integration scrutiny) rather than M-prevention (pre-integration filtering). Prevention is architecturally impossible; scrutiny is achievable though difficult.

9.11.4 Timing in Cached vs. Real-Time Processing

The timing architecture described in Sections 9.11.1–9.11.3 applies to real-time integration. Cached retrieval operates on a faster timescale that creates additional constraints for I-Override.

Real-time integration timing:

Stimulus encountered: 0ms
Source-specific processing (A, I, M circuits active): 0–100ms
Preconscious Transmission (mirror-neuron response to observed goals): 60–340ms
Common-currency integration in Decision Hub: 200–400ms
Conscious awareness of wanting: 300–500ms
Possible I-Override window: 500ms+

Cached retrieval timing:

Context cues encountered: 0ms
Pattern-matching to stored cache entries: 50–150ms
Cached response retrieval: 100–200ms
Motor preparation initiated: 150–250ms
Behaviour execution begins: 200–300ms
Conscious awareness of behaviour (if any): 300–500ms

Cached retrieval completes and initiates motor execution before conscious awareness arrives. The habit fires in the window where real-time integration would still be processing source-specific signals. By the time consciousness comes online, the behaviour is already underway.

This explains the phenomenology of “catching yourself” in habitual behaviour. The Individual who habitually reaches for their phone may become aware of the action only after their hand is already moving, or after the phone is already in their hand. Conscious awareness arrives too late to prevent the behaviour; it can only observe behaviour already in progress.

The Libet findings established that even for deliberated actions, neural preparation precedes conscious awareness. For habitual actions, this gap is larger: there is no deliberation phase at all, only retrieval and execution. The “readiness potential” that Libet detected for willed actions has an analogue in habit-execution, but the conscious experience of “deciding” is absent entirely.

Implications for I-Override timing:

For real-time M-signals, I-Override can operate in the post-integration window (500ms+): after wanting is felt but before action is taken. The Individual experiences “I want X,” applies the Two-Question Diagnostic, and may choose not to act.

For cached habits, this window does not exist. By 500ms, the behaviour is executing. I-Override for habits must therefore operate at a different point in the sequence:

Pre-emptive override: Before the triggering context is encountered, the Individual establishes implementation intentions or modifies the environment to prevent cache-matching
Interruption override: During the 200–300ms execution-initiation window, inhibitory signals may halt the behaviour, but this requires the Individual to have pre-committed to interruption for this context
Post-hoc override: After the behaviour has executed, the Individual recognises the habit, applies the diagnostic retrospectively, and implements changes for future occasions

None of these provides the real-time deliberative pause that I-Override achieves for non-habitual desires. Habit-change requires structural intervention in the conditions that trigger cached retrieval, not moment-of-action willpower.

Movement D: Confabulation

9.12 Confabulation as Structural Consequence

9.12.1 The Information Gap

Source Opacity creates an information gap: the Individual wants something but lacks access to why they want it. The conscious mind receives the wanting-signal without the source-decomposition that produced it.

Confabulation is not a generic explanation-seeking tendency. It is the specific output of the Thinker's Bayesian ground-attribution operation when the Thinker's Hypothesis Space, evidence base, or weighting is in any way mismatched to the actual source-decomposition of the integrated valuation it has been asked to ground. The Thinker, asked to return a ground for the integrated valuation that the Decision Hub has just delivered to the Thinker, returns the highest-posterior ground available within the Hypothesis Space it currently holds, given its currently available evidence and its current weighting. If that Hypothesis Space, evidence, or weighting is mismatched to the actual source-decomposition, the returned ground will be wrong. The Individual experiences this returned ground as a sincere explanation in every case, because the Thinker is in fact returning its best Bayesian estimate, and Source Opacity has already stripped the weighting information from introspective access before the Thinker is queried. This is the Sincerity Guarantee.

The mismatch can take many forms: incomplete evidence about a known source, misweighting of evidence between known sources, or, in the architecturally most important case, a Hypothesis Space that is missing the M-Hypothesis altogether. The M-absent case is the instantiation of the general mechanism that AIM's Camouflage Architecture analysis treats in detail (Layer 7, the Vocabulary Gap), because civilisations that have not installed AIM-vocabulary into the Thinker's prior space cannot return M-grounds, and therefore Confabulate A-grounds and I-grounds even when the integrated valuation was M-driven. Confabulation is the general phenomenon, of which M-absence is one instantiation, not the defining case. The Thinker does not tolerate unexplained experience and fills the gap created by Source Opacity with plausible narratives, returning a Confabulated ground for the integrated valuation.

Confabulation is the Thinker’s routine tendency to generate sincere but inaccurate stories about why a choice was made, filling in gaps in introspection with plausible reasons that do not match the actual motivational causes in A, I, and M.

9.12.2 Confabulation Is Not Lying

Confabulation differs from lying in a crucial respect: the confabulator believes their explanation. They are not consciously deceiving; they are sincerely reporting what feels like accurate introspection.

The neurological basis for this is well-established. Split-brain patients, whose hemispheres cannot communicate, provide dramatic demonstrations: when the right hemisphere (which cannot speak) is induced to perform an action, the left hemisphere (which speaks) will confidently generate explanations for why the action was performed: explanations that cannot be accurate because the left hemisphere had no access to the right hemisphere’s reasons.

Normal brains show subtler versions of the same phenomenon. The “choice blindness” experiments by Johansson et al. demonstrated that Individuals will confidently explain choices they did not actually make, if the experimenter secretly substitutes options. The explanation-generating system operates on the outcome (I chose X), not on the process (here’s how I came to choose X), and can therefore generate explanations for outcomes it did not produce.

9.12.3 Why Confabulation Fills the Gap with A-I Rather Than M

When the Thinker confabulates explanations for M-driven choices, it systematically produces A-sounding or I-sounding narratives rather than M-sounding ones.

Reasons for this asymmetry:

Cultural availability: Explanations in terms of A (“I needed it”) and I (“I wanted to learn / create / experience”) are culturally sanctioned and readily available. Explanations in terms of M (“I wanted it because others want it”) are culturally stigmatised and less available as templates.

Self-concept preservation: The Romantic Lie, the Belief that desires originate autonomously within the self, is foundational to modern self-understanding. Confabulation that preserves this self-concept (A-I explanations) is preferred to confabulation that threatens it (M explanations).

Phenomenological plausibility: M-desires, by the time they reach consciousness, feel like A-desires (urgent, necessary) or I-desires (genuinely engaging). The confabulation system, working from the phenomenology, naturally generates explanations that match the feeling.

The result: when asked why they purchased a luxury Object, the Individual generates explanations about quality, durability, or aesthetic appreciation (I-frame) rather than “because high-status others have it and I want to signal comparable status” (M-frame). The I-frame explanation feels accurate because it matches the phenomenology; the M-frame explanation feels reductive and insulting even if it more accurately describes the source.

9.12.4 Confabulation Intensifies Under Challenge

Confabulation is not static. When a confabulated explanation is challenged, the Thinker does not simply abandon it: it generates increasingly elaborate defences.

The intensification dynamic:

Initial explanation offered (“I bought this car because it’s reliable”)
Challenge issued (“But there are more reliable cars at half the price”)
Elaborated defence generated (“Reliability includes resale value, and the brand’s dealership network, and...”)
Further challenge issued
Further elaboration, often incorporating the challenge into the narrative (“I’m not the kind of Individual who just buys for status: I carefully researched...”)

This intensification occurs because the explanation has become identity-constituting. “I am an Individual who makes rational decisions based on quality” is not merely a claim about a single purchase; it is a claim about the self. Defending the explanation is defending the self-concept.

The harder the Individual scrutinises their own motivations (attempting I-Override), the more elaborate the confabulated defences become. The I-system’s scrutiny triggers the confabulation system’s defensive elaboration. This is why I-Override requires external tools (the Two-Question Diagnostic, behavioural signatures) rather than pure introspection; introspection activates the very confabulation it is trying to penetrate.

9.12.5 The Romantic Lie as Civilisation-Scale Confabulation

The Romantic Lie, the Belief that desires originate autonomously within the Individual rather than being socially transmitted, is confabulation operating at civilisational scale.

Structure of the Romantic Lie:

Phenomenological basis: Desires feel self-generated because source-tags are destroyed at integration and the acquisition leaves no residue
Cultural reinforcement: Narratives of authenticity, originality, and self-expression assume autonomous desire
Institutional support: Legal and economic systems treat preferences as exogenous and individual
Intergenerational transmission: Each generation teaches the next that “being true to yourself” means following desires that are, in fact, largely mimetic

The Romantic Lie is not a deliberate deception but a civilisation-wide confabulation: the same process that operates in individual minds, operating in cultural transmission. The information gap created by Source Opacity is filled, at the cultural level, with the narrative of autonomous authentic desire.

This narrative is functional: it enables coordination, preserves self-esteem, and provides a basis for moral responsibility. But it is false. And its falsity has consequences when M-dynamics produce collective harm that the Romantic Lie renders invisible.

9.12.6 Confabulation of Habitual Behaviour

Confabulation is especially severe for habitual behaviour because there is no deliberative process to report, even a distorted one.

For real-time decisions, the Individual at least experienced wanting and (possibly) some form of weighing before acting. The confabulated explanation is a distorted report of something that occurred: “I wanted X, and here’s why.” The explanation misattributes the source (claiming A or I when M dominated), but it is an explanation of an experienced wanting.

For habitual behaviour, there was no experienced wanting at the moment of action. The behaviour emerged from cached retrieval without deliberation. When asked “why did you do that?”, the Individual must confabulate an explanation for behaviour that had no reasoning behind it.

The confabulation system handles this by constructing post-hoc narratives that invent the deliberation that didn’t occur:

“I checked my phone because I was expecting an important message” (constructed A/I rationale)
“I had another drink because I was thirsty” (A-attribution for M-driven social drinking)
“I bought it because it was on sale” (A-attribution for M-driven acquisition)

These narratives feel subjectively accurate because the Individual has no competing information. They didn’t experience deliberation, so they cannot compare the narrative against their actual reasoning. The narrative fills a complete void rather than distorting an existing memory.

The confabulation is therefore purer for habits than for deliberated choices. For deliberated choices, the confabulation must be consistent with something the Individual experienced (the wanting, the context, the weighing). For habits, the confabulation is constrained only by plausibility: any narrative that could explain the behaviour will be accepted, because there is no experiential residue to contradict it.

This explains why self-report is especially unreliable for habitual behaviour. The Individual is not reporting (even inaccurately) on their reasoning; they are inventing reasoning that never occurred. The explanation has no connection to the actual causal process (direct action on a historically-formed cached value) because that process left no experiential trace.

Confabulation intensification (Section 9.12.4) operates differently for habits. When a deliberated choice’s explanation is challenged, the Individual generates increasingly elaborate defences of their stated reasoning. When a habit’s explanation is challenged, the Individual may simply generate a different confabulation, since no actual reasoning anchors the explanation, alternative narratives are equally available.

This can produce the appearance of inconsistency or dishonesty when the Individual offers different explanations for the same habitual behaviour on different occasions. They are not being dishonest; they are drawing from a pool of plausible narratives, any of which could explain the behaviour, none of which reflects the actual (cached, non-deliberative) causal process.

Movement E: Compensatory Mechanisms and Implications

9.13 I-Override Information Architecture

The I-Override mechanism described in this section is the within-Individual fast-layer channel of the five-channel Knowledge Patch architecture specified in Appendix A.6.4 (Channel 1) and developed in full at Chapter 18. The five channels comprise: Channel 1 (within-Individual fast-layer I-Override gate, this section), Channel 2 (within-Individual slow-layer Hypothesis Space expansion), Channel 3 (within-Individual slow-layer Epistemic Suspension protocol), Channel 4 (network-layer collective epistemic norms), and Channel 5 (network-layer institutional governance). This chapter develops the architectural foundation for Channel 1 only; the remaining channels are developed in Chapter 18.

9.13.1 What Survives Integration

Source Opacity destroys source-tags, but not all information is lost. The I-Override mechanism operates on information that survives integration:

Desire-content: What is wanted. The Object or state toward which motivation is directed remains accessible. The Individual knows they want this car or that promotion even if they don’t know why (which source drives the wanting).

Desire-intensity: How strongly it is wanted. The magnitude of the wanting-signal is experienced directly. Urgency, compulsion, mild preference; these intensity-levels are available to introspection.

Context-information: The circumstances in which the wanting arose. Where you were, who was present, what you had just observed, what state your body was in; this contextual information is remembered separately from the wanting itself.

Temporal patterns: How the wanting varies over time. Whether it persists or fades, whether it recurs cyclically, whether it tracks specific events; these patterns are observable across episodes.

Behavioural consequences: What happens when you act on the wanting. Satisfaction or continued wanting, depletion or invigoration, social feedback or private experience: outcomes provide information about source.

9.13.2 The Two-Question Diagnostic

The compensatory pathway works through two behavioural questions that an Individual can apply after the fact: 'Do I want this because my body needs it?' and 'Would I still want this if no one ever knew?' If both answers are no, the desire is M by elimination within an exhaustive A, I, M taxonomy. This procedure is the Two-Question Diagnostic.

The Two-Question Diagnostic is AIM’s core tool for inferential source-reconstruction. It uses surviving information to estimate source-composition:

Question 1: “Does my body need this?”

This question probes for A-involvement. Consider:

Am I hungry, thirsty, tired, cold, or in pain?
Would this wanting diminish if my bodily state changed?
Does this wanting correlate with physiological cycles?
Will satisfying this wanting produce satiation that lasts until the next physiological perturbation?

If yes: A-involvement is likely. If no: A is probably not the dominant source.

Question 2: “Would I still want this if no one ever saw or knew?”

This question probes for M-involvement. Consider:

Does this wanting persist in private?
Would I pursue this if I could never tell anyone or display the result?
Is this wanting stronger when certain Individuals are present or watching?
Does this wanting track what Models (Individuals I admire, envy, or compare myself to) are pursuing?

If wanting persists in private and the engagement is rewarding in the doing: I-involvement is likely; M is probably not dominant. If wanting diminishes without audience: M is probably the dominant source.

Interpreting results:

Q1 (Body needs?)	Q2 (Private want?)	Likely Composition
Yes	Yes	A-lead or A-I blend
Yes	No	A-M blend (social eating, etc.)
No	Yes	I-lead
No	No	M-lead

The diagnostic does not provide certainty: it provides evidence for updating Beliefs about source-composition. Repeated application across contexts strengthens the inference.

9.13.3 Why Vocabulary Provision Is Architecturally Necessary

The Two-Question Diagnostic cannot be applied without conceptual vocabulary. The Individual must have the concepts “Appetites,” “Intrinsic Motivation,” and “Mimetic Desire” (or functional equivalents) to formulate the diagnostic questions.

Before AIM: The Individual experiences wanting and, if they introspect, generates confabulated explanations. They have no framework for systematic source-reconstruction. The questions “Is this mimetic?” or “Is this intrinsic?” cannot be formulated because the categories are absent.

After AIM vocabulary acquisition: The Individual can formulate the diagnostic questions. The questions provide structure for analysing surviving information. Confabulation is not eliminated but is now competing with a structured alternative. The Individual can notice when their confabulated explanation fails the diagnostic tests.

This is why AIM is self-applying. The framework provides the conceptual vocabulary that makes I-Override possible. Without the vocabulary, the information that survives integration cannot be systematically interrogated. The framework is not merely descriptive but instrumentally necessary for the compensatory mechanism it describes.

The Self-Application Principle states: The AIM Framework functions as debugging software for human motivation, providing the conceptual vocabulary required for I-Override to infer the leading source of integrated urgency signals by contrast, without restoring the destroyed source-tags. This self-application is architecturally grounded: AIM identifies the conditions for its own effectiveness and the limitations of its reach.

9.13.4 I-Override Capacity as Contextual Variable

I-Override is not a fixed ability but a contextual capacity that varies with:

Cognitive resources: I-Override requires working memory, attention, and processing capacity. When these are depleted (fatigue, stress, cognitive load), I-Override capacity diminishes. The tired Individual is more likely to act on M-signals without scrutiny.

A-state: Appetitive deprivation degrades I-Override. The hungry, cold, or injured Individual cannot sustain deliberative scrutiny: the A-system demands resources. A-floor security is therefore a precondition for reliable I-Override. This is the Bootstrapping Principle: I-Override requires A-sufficiency, because the Body Manager is correctly prioritising survival and will not release the attention that structural analysis needs, so a Scapegoat narrative succeeds under A-deficit for the reason that it is cognitively cheap while systemic analysis is not.

Time pressure: I-Override requires time. Urgent decisions, snap judgments, and high-speed environments reduce the window for deliberative intervention. Social media’s rapid-fire content delivery systematically undermines I-Override.

Social support: I-Override is easier when the social environment supports it. Norms that legitimate questioning (“Why do you want that?”), relationships that tolerate deviation from M-pursuit, and communities that value I-engagement all facilitate I-Override. Isolated Individuals facing unanimous M-pressure have reduced capacity.

Practice: I-Override capacity increases with exercise. Repeated application of the Two-Question Diagnostic builds automaticity, not automatic answers, but automatic question-formulation. The practiced Individual notices M-relevant contexts more readily and initiates scrutiny with less effort. The automatic question-formulation is a self-set question, and a self-set question is one of the conditions that recruits the Thinker (Chapter 4, Section 4.4.3): practice works by recruiting the Thinker in the contexts where I-Override is needed, because I-Override is an operation of the Thinker and is unavailable until the Thinker is recruited (Chapter 4, Section 4.3.3).

These contextual dependencies explain why individual-level I-Override is insufficient. An Individual in an environment that depletes cognitive resources, threatens A-security, imposes time pressure, provides no social support, and offers no opportunity for practice will have near-zero I-Override capacity regardless of their AIM vocabulary.

This is why AIM emphasises structural intervention alongside individual capacity-building. Securing A-floors, reducing time pressure, building supportive communities, and creating practice opportunities are all I-Override enhancement strategies; they operate on the context that determines whether I-Override is achievable. The Bootstrapping Principle also fixes their order. Emergency A-provision comes first, urgency reduction follows from it, only then does the diagnostic get a hearing, only after that can a systemic account be held, and only then can anything be organised around it. This is the Bootstrapping Sequence, and it cannot be entered at the third step: an audience under A-deficit adopts the cheaper explanation before the expensive one arrives.

9.13.5 The Compensatory Pathway Analogy

The relationship between native source-tagging (which doesn’t exist) and I-Override-based source-reconstruction can be understood through a medical analogy:

Native pathway (ideal but absent): An Individual born with full colour vision perceives colours directly. The information is available without effort or inference.

Compensatory pathway (achievable): An Individual born with colour-blindness can learn to distinguish colours through contextual cues: labelling, memorisation, inference from other features. They never perceive colour directly, but they can achieve functional colour-discrimination through learned strategies.

I-Override is the compensatory pathway for source-discrimination. The Individual will never directly perceive “this desire is mimetic” the way they directly perceive “this apple is red.” But they can learn to infer mimetic involvement through the Two-Question Diagnostic and behavioural signatures, achieving functional source-discrimination through learned strategies.

The compensatory pathway is:

Less efficient than direct perception (requires effort and time)
Less reliable than direct perception (inference can err)
Achievable through training (capacity improves with practice)
Dependent on tools (vocabulary, diagnostic questions)
Still valuable (functional discrimination beats no discrimination)

AIM’s goal is not to restore native source-tagging; that is architecturally impossible. The goal is to establish and strengthen the compensatory pathway so that Individuals can achieve functional source-discrimination despite the architectural constraint of Source Opacity.

Architecturally, I-Override is the Thinker's Bayesian ground-attribution operating on an expanded Hypothesis Space that includes the M-Hypothesis. The compensatory pathway works because the Knowledge Patch installs the M-Hypothesis into the Thinker's prior space, making non-zero posterior weight on M reachable from behavioural evidence even though the integrated Scalar at the Decision Hub remains source-opaque.

The compensatory pathway is asymmetrically located in the Thinker for an architectural reason. The minimal architecture exhibits a channel-count asymmetry in which each of the Thinker, the Observer, and the Body Manager supplies one channel of direct input, a physical count of one to one to one, but the Thinker controls two of the three because its posteriors propagate via the Belief Register to the Observer’s Salience computation, while the Body Manager controls one. Counted by independent source the asymmetry is two to one in the Thinker’s favour, the Body Manager being the one genuinely independent input.

The Thinker is the only component that holds priors and performs Bayesian ground-attribution, and it does so over its own outputs as well as over the Observer's, which gives the Thinker a self-feedback property the other components lack. The Thinker's posteriors propagate via the Belief Register to the Observer's preconscious Salience computation, which preloads the Observer's M-Salience field with the Thinker's current best estimate of which Models, contexts, and Objects warrant attention. The Body Manager signals deficit only and does neither of these things. The Thinker therefore has an outsized effect on the weight calculation across successive rounds of integration, even though its instantaneous contribution to any single integration is not numerically dominant. From this structural asymmetry follows the I-Supremacy Result, the A-Disadvantage, the I-Template Override Result, and the I-Corrective Asymmetry, which together entail that source-discrimination, where it is achievable at all, must operate through the Thinker's expanded Hypothesis Space rather than through the Decision Hub or the other input-bearing components.

9.13.6 I-Override for Habitual Behaviour: Cache Interruption

I-Override as described in Sections 9.13.1–9.13.5 assumes a desire available for scrutiny. For habitual behaviour, no such desire is present at the moment of action. I-Override for habits therefore requires a different operation: cache interruption, preventing automatic retrieval so that real-time integration (and standard I-Override) becomes possible. A habit that fires on an adequate cached value selects the action without recruiting the Thinker (Chapter 4, Section 4.4.3), and I-Override is an operation of the Thinker (Chapter 4, Section 4.3.3), so cache interruption is what opens the point at which the Thinker can be recruited.

The goal of cache interruption is to force the Decision Hub back into real-time integration mode. When the cache lacks an applicable entry, or when retrieval is blocked, the system must compute a fresh response. This computation produces the experienced wanting that I-Override can then scrutinise.

Techniques for cache interruption:

1. Implementation intentions (“if X, then pause”)

Implementation intentions are pre-formed plans that link a triggering context to a specified response. By rehearsing “if I reach for my phone in the morning, I will pause and ask why,” the Individual creates a competing cache entry. When the morning context is encountered, two cached responses compete: the original habit and the implementation intention. If the implementation intention fires first or with sufficient strength, it interrupts the habit and creates a deliberative pause.

Implementation intentions work because they operate at the same level as habits; they are cached context-response associations. Fighting a habit with deliberation pits slow System 2 against fast System 1. Fighting a habit with an implementation intention pits cache against cache, on equal temporal footing.

2. Environment restructuring (remove retrieval cues)

Cached entries are triggered by context cues. Removing or modifying those cues prevents cache-matching and forces real-time integration. The Individual who moves their phone to another room eliminates the visual cue that triggers the checking-habit. In the new environment, no cached response exists; the Individual must deliberate about whether to retrieve the phone.

Environment restructuring is often more effective than willpower because it operates upstream of the retrieval process. The habit cannot fire if its triggering context is absent. This is why context-change (moving house, changing jobs, travelling) often enables habit-change that was impossible in the original environment: the new context lacks the cue-associations that triggered cached retrieval.

3. Deliberate novelty introduction

Introducing novel elements into a habitual context can disrupt cache-matching. The cached entry was formed for a specific context; significant novelty may prevent the pattern-match from succeeding. Eating breakfast at a different time, in a different location, or with different Individuals may prevent food-related habits from firing automatically.

Novelty introduction is a temporary intervention: once the new context becomes familiar, new habits will form. But it provides a window during which real-time integration occurs and I-Override becomes possible. The Individual can use this window to establish new patterns before caching reconsolidates.

4. Mindfulness and meta-awareness training

Training in meta-awareness, the capacity to notice one’s own mental states and behaviours as they occur, can reduce the gap between behaviour-initiation and conscious awareness. While this cannot eliminate the timing differential (cached retrieval will always be faster than conscious awareness), it can shorten the delay, allowing the Individual to interrupt behaviours earlier in their execution.

Mindfulness training may also increase sensitivity to the absence of deliberation: the Individual learns to notice when they are acting without having decided, which provides a cue for cache-interruption even after behaviour has begun.

The sequence for habit-change:

Recognition: Identify the habit as habitual (not deliberated): often through retrospective pattern-recognition or external feedback
Cache analysis: Identify the triggering context and the cached response
Interruption strategy: Select appropriate technique(s): implementation intention, environment restructuring, novelty introduction
Deliberative window: When interruption succeeds, apply standard I-Override (Two-Question Diagnostic) to the now-conscious desire
New cache formation: If the new response is repeated sufficiently, it forms its own cached association, replacing the original habit

This process is effortful and failure-prone. Cache interruption requires advance planning; it cannot be performed in the moment because the moment is too fast. The Individual must commit to interruption before the triggering context is encountered, which requires recognising the habit and its triggers in advance. This is why habit-change typically requires multiple attempts; the planning and commitment must be in place before the next occurrence of the triggering context.

9.14 Implications for Self-Knowledge

9.14.1 The Introspection Illusion

Source Opacity and Confabulation together produce what psychologists call the introspection illusion: the confident but false Belief that we have direct access to the causes of our mental states.

The illusion operates as follows:

We experience a mental state (wanting, feeling, deciding)
We generate an explanation for it (confabulation)
The explanation feels like direct access (because we are not aware of the generative process)
We believe we know why we want/feel/decide (illusion of insight)

The introspection illusion is not a sign of stupidity or carelessness. It is a structural feature of minds that integrate motivational signals and fill information-gaps with explanations. Every Individual with intact Common-Currency Integration is subject to the introspection illusion.

Implications:

First-person reports about motivational sources are unreliable
Confidence in self-explanation does not correlate with accuracy
Introspective “insight” may be confabulation completion, not discovery
Self-knowledge requires methods beyond introspection

9.14.2 The Weakness of Self-Report

Self-report is the standard method for assessing motivation in psychology, market research, and everyday inquiry. AIM predicts systematic weakness in self-report for source-composition:

Self-report accurately captures:

Desire-content (what is wanted)
Desire-intensity (how strongly)
Conscious Beliefs about motives (what the Individual thinks is driving them)

Self-report poorly captures:

Actual source-composition (the A-I-M decomposition)
Mimetic influence (systematically under-reported due to confabulation)
Habitual motivation (no experienced desire to report)

This creates a methodological challenge. Research relying on self-report will systematically underestimate M-influence and overestimate I-influence. Market research asking “why did you buy this?” will receive confabulated A-I answers for M-driven purchases. Policy evaluation asking “what do you need?” will receive contaminated responses mixing A-needs with M-wants.

AIM-informed research methods must supplement self-report with:

Behavioural observation (tracking what Individuals do, not just what they say)
Experimental manipulation (varying audience, Models, A-state and observing changes)
Longitudinal tracking (observing persistence, satiation, and reference-point shifts)
Context analysis (assessing visibility, Model-exposure, and A-security)

Third-person motivational diagnosis is structurally more Source-Opacity-robust than first-person introspection precisely because it bypasses the report that Source Opacity compromises. The Third-Person Status Diagnostic (Chapter 2) formalises this asymmetry as a two-stage procedure (a type-level pattern filter across observable behaviour followed by a token-level elimination test parallel to the Two-Question Diagnostic) and identifies the type/token launder-tell as the primary diagnostic signature: genuine I-pursuit persists privately and tracks craft; M-leading behaviour tracks audience and relative position, and voiced status-lowering (calling someone "irrational" or "stupid") is itself a launder-tell for rank competition. Because the third-person form operates on behavioural evidence rather than self-report, the Confabulation that degrades self-report does not degrade it. This inferential asymmetry has direct implications for institutional design: mechanisms for assessing motivation in hiring, resource allocation, and governance that rely on self-report will systematically reproduce the same confabulation architecture they are meant to evaluate; mechanisms that operate on observable type-level patterns are architecturally insulated from this failure mode.

9.14.3 Behavioural Signatures as Evidence

Since self-report is unreliable for source-composition, AIM relies on behavioural signatures: observable patterns that correlate with source-dominance:

A-signatures:

State-dependence: wanting correlates with physiological state
Satiation: wanting terminates after consumption
Cyclicity: wanting recurs on physiological timescales
Substitutability: functionally equivalent Objects satisfy equally

I-signatures:

Private Persistence: wanting continues without audience
Process focus: engagement with the activity, not just the outcome
Competence-scaling: wanting adjusts with skill development
Deepening: engagement intensifies over time rather than satiating

M-signatures (special case):

Audience-dependence: wanting increases with observability
Model-tracking: wanting follows what high-status others pursue
Reference-point sensitivity: satisfaction resets as comparisons shift
Object-specificity: the specific high-status Object is wanted, not functional equivalents

These signatures provide evidence for updating source-composition estimates. An Individual whose purchase behaviour shows audience-dependence and model-tracking (M-signatures) but who reports “I just like quality” (I-explanation) is probably M-led despite their self-report.

Behavioural signatures are not infallible. An Individual might display M-signatures for I-driven reasons (genuinely enjoying an activity that happens to be high-status) or I-signatures for M-driven reasons (performing private behaviour because private authenticity is itself a status-signal in some communities). Signatures provide evidence, not proof. But they provide better evidence than self-report alone.

9.14.4 The Triple Camouflage (Layers 1–3 of the Camouflage Architecture)

Sections 9.9–9.12 established three mechanisms that obscure M-influence:

Source Opacity: M-signals are integrated with source-tags destroyed; the Individual cannot introspect M’s contribution
Confabulation: The Thinker fills the information-gap with A-I explanations; M-involvement is systematically explained away
Object Parasitism (Chapter 8): M operates through A-I Objects; there are no pure M-Objects that would reveal M-involvement

Together, these create the Triple Camouflage (Layers 1–3 of the Camouflage Architecture): a closed system that renders M systematically invisible at the individual level:

Cannot detect M through introspection (Source Opacity)
Cannot detect M through self-explanation (Confabulation)
Cannot detect M through object-analysis (Object Parasitism)

The Individual who attempts to identify M-influence using first-person methods alone is blocked at every entry point. Introspection finds only undifferentiated wanting; explanation produces A-I narratives; examining their desires finds only legitimate A-I Objects.

This is why AIM vocabulary and the Two-Question Diagnostic are necessary: they provide a detection method that routes around the Camouflage Architecture by using contextual and behavioural evidence rather than introspection, self-explanation, or object-examination. The full eight-layer Camouflage Architecture extends beyond the Triple Camouflage (Layers 1-3) through the institutional tier of Disciplinary Priors, Disciplinary Boundaries, and Academic Camouflage (Layers 4-6) and the civilisational tier of the Vocabulary Gap (Hilbert Collapse, Layer 7) and Meta-Architectural Exclusion (Layer 8). The full structure is set out in the Chapter 2 entry for the Camouflage Architecture.

9.14.5 Self-Knowledge of Habits: The Double Opacity Problem

Habitual behaviour presents a double opacity problem that compounds the limitations on self-knowledge described in Sections 9.14.1–9.14.4.

First opacity: The source-composition of the original integration that formed the habit is inaccessible. When the habit was established, the Decision Hub integrated A, I, and M signals into a common-currency value, discarding source-tags. The Individual who formed the habit did not know then (and cannot introspect now) whether the behaviour was A-led, I-led, or M-led at formation.

Second opacity: The current instance involves no deliberative process to inspect. The behaviour emerged from cached retrieval, not from weighing or wanting. The Individual has no first-person data about the current episode because no deliberation occurred.

The compound effect: the Individual cannot introspect why they formed the habit (first opacity) and cannot introspect why they just performed it (second opacity). The only information available is the behaviour itself and the context in which it occurred.

Implications for self-knowledge:

Behavioural signatures become essential. For deliberated choices, the Individual has weak access to their reasoning (distorted by confabulation but not absent). For habits, even this weak access is unavailable. The only evidence about habitual motivation is behavioural: Does the behaviour correlate with A-states? Does it track what Models are doing? Private Persistence cannot be read off the habit itself, because the Decision Hub's cached-value store runs in private whatever formed it (Chapter 4, Section 4.4.8), so Private Persistence is read on the wanting that returns once the habit is interrupted. These signatures (Section 9.14.3) provide the only window into what sustains a habit, since the cached entry itself holds no record of it.

Historical reconstruction is required. To understand a habit’s motivational composition, the Individual must reconstruct the conditions under which it formed. When did this behaviour begin? What was happening in my life? Who was I observing or comparing myself to? What A-states were present? This reconstruction is imperfect (memory is fallible, and the original formation may have been gradual), but it provides more information than introspection on the current instance, which provides none.

Identity claims about habits are especially suspect. When an Individual says “I am an Individual who exercises every morning” or “I am an Individual who values quality,” they are often describing habits rather than deliberated values. The identity claim feels authentic because the behaviour is consistent and long-standing. But the consistency reflects cached retrieval, not ongoing value-endorsement. The Individual may be M-captured in ways that have become invisible precisely because the M-dominant behaviour has been cached and no longer produces experienced wanting that could be examined.

The introspection illusion is maximally active for habits. The Individual confidently explains their habitual behaviour despite having no access to the actual causal process. The explanations are pure confabulation: narratives generated to fill a complete void. Yet they feel as accurate as explanations for deliberated choices. The Individual cannot distinguish “I know why I did this” from “I generated a plausible story about why I did this” because both feel identical from the inside.

Implications for AIM-informed self-understanding:

An Individual seeking accurate self-knowledge should:

Identify habitual vs. deliberated behaviours: Which behaviours occur automatically, without experienced wanting? These are habits, and introspective explanations for them are unreliable.
Apply behavioural signatures rather than introspection: For habits, ask the diagnostic questions about the behaviour pattern, not about the current instance. Does this behaviour correlate with A-states (across many instances)? Does it track Model behaviour (across many instances)? Read Private Persistence on the interrupted instance (point 5), not on the habit.
Reconstruct formation conditions: When and why did this habit form? What was the original context? This historical analysis provides more information than present-moment introspection.
Treat identity claims with suspicion: “I am an Individual who...” claims may describe cached patterns rather than endorsed values. Test them against the Two-Question Diagnostic applied to the behaviour pattern, not to the identity claim.
Use cache-interruption to generate data: By interrupting a habit and forcing real-time integration, the Individual creates an instance where I-Override is possible and where the resulting wanting can be examined. The composition of this wanting is information about what now drives the behaviour, which the cached entry itself cannot supply.

Self-knowledge of habits is difficult but not impossible. It requires shifting from introspection (which provides nothing for habits) to behavioural analysis (which provides patterns) and historical reconstruction (which provides formation context). The Individual who masters these techniques can achieve reasonable understanding of their own habitual motivation, but the understanding will always be inferential, never direct.

Movement F: The Integration Applied to Relationships

9.15 Romantic and Sexual Relationships: Three Channels in One Dyad

9.15.1 Formation Through the Observer and the Body Manager

Every Individual in the field receives a Salience computation. The Configural Update-Rate Asymmetry makes agents the highest-frequency source of configuration change the Observer samples, and the Salience elevation that Individual-detection produces binds to the perceived Individual as an Object in the field (Individual-as-Object). Romantic attraction therefore begins on almost any Individual, and candidates differ in the magnitude of the Salience computed on them rather than in eligibility for it.

The magnitude clause sets that difference. The Observer computes Salience under the Individual's own Belief Register, and the Belief Register carries the templates of desirability the surrounding culture supplies: the configurations of face, body, bearing and position that Models have been observed to approach. Because a culture's Individuals share much of the same template content, the magnitude term predicts consensus, and consensus attractiveness does predict romantic evaluation across studies (Eastwick, Luchies, Finkel and Hunt, 2014). Because the magnitude term sets how strongly each candidate registers, stated preferences predict the traits of the partners Individuals later choose (Conroy-Beam and Buss, 2016; Gerlach, Arslan, Schultze, Reinhard and Penke, 2019). The Thinker commits these templates, but the evidence behind them arrived through the Observer and was integrated before it reached the Thinker, so Source Opacity conceals their provenance and the Individual experiences a culturally installed template as personal taste. The Belief Register also carries the Thinker's priors about the Body Manager's deficit state, and the Body Manager's sexual deficit signal reaches the Decision Hub on its own channel, so Appetitive state amplifies attraction without opening any gate.

Mimetic installation toward a candidate follows the observed directional valence of other Individuals toward the candidate, under the gate clause of the Minimum Installation Condition. Attraction toward a candidate therefore rises when other Individuals are observed approaching the candidate. Human mate-choice copying is established on meta-analysis, with substantial variation across paradigms (Gouda-Vossos, Nakagawa, Dixson and Brooks, 2018), and it operates as a case of social learning general across Object domains rather than as a mating-specific mechanism (Street, Morgan, Thornton, Brown, Laland and Cross, 2018), which is the pattern a general-purpose Observer predicts.

9.15.2 Integration and the Romantic Lie

The Decision Hub integrates the Body Manager's signal and the Observer's signed Salience into one Common-Currency Scalar and strips the source tags, as Section 9.1 established for the valuation of any Individual-as-Object. The Individual feels one undifferentiated pull and, under Confabulation, explains the pull in Appetitive or Intrinsic terms: chemistry, recognition, having simply clicked. Physiological arousal from an unrelated source raises attraction toward a co-present candidate, and the effect weakens when the source of the arousal is made unambiguous (Foster, Witcher, Campbell and Green, 1998), which is the pattern Section 9.13 predicts for the provision of source vocabulary. The Romantic Lie takes its name from the novel of romantic desire, and romantic attraction remains the Romantic Lie's clearest case.

9.15.3 Recruitment of the Thinker and the Durability of the Dyad

Approach produces contact, and contact presents an Individual-as-Object for whom the Belief Register holds no committed posterior. Ground-Resolution is recruited, and what the Individual experiences as interest in the other is Curiosity. Conversation, the exchange of ideas and the discovery of the other are Ground-Resolution running on the other and on the Objects the other presents. Under the Reciprocal Valuation Property the other is also a valuing subject, so the process runs in both directions, and for the reason Chapter 7, Section 7.1.2.1 gives, the other can sustain the I-process without exhausting it.

The I-component is what carries a relationship once formation is complete. Across forty-three longitudinal studies of 11,196 couples, relationship quality is predicted chiefly by relationship-specific perceptions, among them perceived partner commitment, appreciation and sexual satisfaction, while the partner's own traits add almost nothing beyond them (Joel et al., 2020). Where the Individual observes the partner's directional valence toward Objects, the Minimum Installation Condition installs signals toward those Objects, and where the partner expands the Individual's I-capacity the transmission takes the form of Admiration: wanting to develop what the partner is observed developing.

9.15.4 The Relational Specificity Principle

The three components of the valuation each partner assigns the other behave differently over the life of the relationship, as Differential Satiation predicts, and differ in whether another Individual could supply them.

The A-component satiates and returns. Sexual appetite and the needs met by care are cyclical, and any adequate partner could meet them, so the A-component is substitutable across partners.

The Mimetic Premium component does not satiate, and its Referent is comparative. As the Reference Set shifts through new candidates, other Individuals' partners and changed standing, the Mimetic Premium assigned to the partner falls relative to available alternatives and registers as a felt deficit. A partner's value relative to available alternatives predicts relationship satisfaction (Conroy-Beam, Goetz and Buss, 2016), and the perceived quality of alternatives predicts lower commitment (Le and Agnew, 2003). The Mimetic Premium component is substitutable by any candidate carrying more Salience, and the Reference Set always holds such a candidate.

The I-component does not satiate, and one part of it cannot be substituted. The I-contribution a partner makes to engagement with shared Objects is bounded by that contribution, and another Individual contributing equally would supply it. The I-engagement whose Object is the partner is different: the accumulated history of Ground-Resolution on one Individual, and the unresolved residue that recruits the Thinker, belong to that Individual and cannot be transferred to another. The unique component of desirability, the part of one Individual's attraction to another that consensus does not predict, grows as the two become acquainted (Eastwick and Hunt, 2014), and investment in the relationship predicts commitment (Le and Agnew, 2003). This is the Relational Specificity Principle: of the components of valuation directed at an Individual-as-Object, only the I-engagement whose Object is that Individual is specific to that Individual. Only that engagement can therefore carry a relationship indefinitely, and that engagement is why a durable relationship is with one Individual rather than with anyone who fits a template.

9.15.5 Sexual Relationships

Sexual motivation carries all three components. The Body Manager supplies a cyclical, satiating A-component. The I-component is pleasure in the act and mutual exploration, which remain rewarding in private and deepen while the act sustains unresolved structure. The Mimetic Premium component is the partner's Status, conquest and performance against a Reference Set, and the Mimetic Premium component escalates without satiating toward novelty and comparison. Sexual desire declines with relationship duration in cross-sectional data, though not uniformly across the sexes (Klusmann, 2002). A sexual relationship endures where the I-component carries what the A-component cannot sustain between cycles and what the Mimetic Premium component would otherwise turn toward the next candidate.

9.15.6 Five Failure Paths and the Diagnostic

In the first, the Thinker never takes over. The partner was selected on Mimetic Premium criteria, and the two Individuals' Intrinsic Motivations are neither compatible nor complementary, so the relationship performs for audiences and feels hollow in private.

In the second, the relationship moves to Habit. Action toward the partner runs on the Decision Hub's cached-value store without fresh Ground-Resolution (Section 9.8), and the cached route is insensitive to current outcomes, so the relationship persists without satisfaction. The Boredom felt in such a relationship does not show that the partner is resolved. The Thinker is no longer recruited to find out.

In the third, the partners fall into internal mediation. No two Individuals are more socially proximate than partners, so where the Mimetic Premium component dominates, each becomes the other's rival, and the Rival-Constitution Principle embeds each in the other's Self-Model as the reference point against which each measures their own standing.

In the fourth, one partner extracts Mimetic Premium from the other beyond what the relationship can sustain. In the fifth, the partners' Belief Registers diverge on Beliefs that constitute either partner's Self-Model. The fourth and fifth paths are common to every Collaborator subtype, run on different axes, and are derived at Sections 9.19 and 9.20.

The diagnostic is the Audience-Removal Test applied to the relationship: would the satisfaction survive if no one ever saw the pair?

9.15.7 The General Case and the Permissive Architecture

Sections 9.15.1 to 9.15.6 derive the general case: the configuration the three channels produce when one Individual forms and sustains a partnership with one other Individual under ordinary conditions. The architecture fixes which channels exist and how the Decision Hub integrates them. The architecture does not fix the weights, the templates or the Objects over which the channels run, and the gate clause of the Minimum Installation Condition is permissive by construction. Relationship forms and failure paths beyond those derived here are therefore expected outputs of the same architecture rather than exceptions to it: a partnership among more than two Individuals, a partnership in which the A-component is absent, a relationship conducted largely through a medium, attachment directed at an Individual who does not reciprocate. Each is derivable by running the same components under different weights, templates or field conditions, and Chapter 11, Section 11.16 does so for sexual behaviour directed across species and at artefacts.

9.16 The Collaborator Subtypes

The Agent-Type Vector weights every Individual-Object across Collaborator, Rival, Predator and Prey (Chapter 4, Section 4.4.7), and Section 9.9.10 established that one Salience function runs across all four coordinates. The Collaborator coordinate is not uniform. The classifier reads, alongside evidence of reciprocity, evidence of whether the Individual-Object presents reproductive cues and whether the Individual-Object depends on the observer for the repair of its Appetitive deficits, and the Collaborator weight takes one of three subtypes accordingly. Each subtype sets which of the three channels carries weight in the valuation.

The mating-pair subtype is the Collaborator reading under reproductive cues, and it engages the Body Manager's sexual deficit term alongside the Observer and the Thinker. It is the classification under which the romantic and sexual relationship of Section 9.15 runs.

The parent-child subtype is the Collaborator reading under dependency: the Individual-Object's Appetitive deficits are ones the observer can repair and the Individual-Object cannot. It carries an Appetitive flow that runs in one direction, from the capable party to the dependent one.

The among-equals subtype is the Collaborator reading under reciprocity between Individuals of comparable capacity, with neither reproductive cue nor dependency. It is the classification of friendship and of cooperation among adult peers.

The subtypes are readings of one vector, not three relationship modules. One Individual-Object can carry more than one subtype at once, as a partner who is also a friend does, and can move between subtypes as the evidence changes, as an adult child does whose parent becomes dependent. The parent-child subtype applied across species is treated at Chapter 11, Section 11.15, and the mating-pair subtype applied across species at Chapter 11, Section 11.16.

9.17 Friendship

9.17.1 Formation Through Co-Presence

Friendship forms under the among-equals subtype, and the Minimum Installation Condition supplies its opening. Each Individual's engagement with an Object is directional valence toward it that the other perceives, so repeated co-presence repeatedly opens the gate, and shared activity installs, in each Individual, signals toward the Objects the other engages. Where those Objects are I-Objects, each Individual's engagement recruits Ground-Resolution on the Objects and on the other, and Curiosity follows. Randomly assigned seating raises the rate of mutual friendship between the Individuals seated together, though similarity still conditions which pairs become friends (Rohrer, Keller and Elwert, 2021), and the closeness of a friendship rises with the hours spent together (Hall, 2019). Similarity conditions formation because Individuals who engage the same Objects give each other's Ground-Resolution the most to run on.

The Observer's magnitude term also reads the other Individual's standing. An Individual-as-Object carries Status, and the association value of a friend's Status enters the valuation as a Mimetic Premium component. Friendship networks sort by status as by other attributes (McPherson, Smith-Lovin and Cook, 2001).

9.17.2 Three Components and the Three Kinds of Friendship

The valuation of a friend decomposes as every valuation of an Individual-as-Object does, and the decomposition recovers a distinction drawn in the philosophical study of friendship since antiquity. Aristotle separated friendships of utility, friendships of pleasure and friendships of character, and held that the first two dissolve when the benefit ends while the third, in which the friend is loved for who the friend is, endures (Aristotle, 2009, Books VIII and IX). A friendship of utility is carried by the A-component and by the association value of the friend's Status. A friendship of pleasure is carried by the I-contribution the friend makes to engagement with a shared Object: the tennis partner valued for what the partner adds to the game. A friendship of character is carried by the I-engagement whose Object is the friend. Under the Relational Specificity Principle (Section 9.15.4) the first two are substitutable, because another Individual meeting the same deficit, carrying the same Status or contributing equally to the shared Object would serve, and only the third is not. The architecture therefore explains why the first two kinds dissolve and the third endures.

9.17.3 The Relationship Most Dependent on the Intrinsic Component

Friendship carries the smallest A-component of the three subtypes: the minimum social contact Chapter 6, Section 6.1.7 places under the Appetites, and mutual aid. Friendship carries no sexual deficit term and little institutional or genealogical support. Friendship is therefore the relationship most dependent on the I-component, which makes friendship the most voluntary of the three and the most fragile. Because the I-component is recruited by the other's presence, a friendship that loses co-presence loses the condition that recruits the Thinker, and decays. Emotional closeness to friends declines when contact falls, while closeness to kin does not decline in the same way (Roberts and Dunbar, 2011, 2015).

9.17.4 Failure Paths

A friendship drifts where co-presence ends. Drift differs from the Habit path of Section 9.15.6: under Habit the relationship continues without fresh Ground-Resolution, while under drift the relationship stops presenting anything to resolve.

A friendship turns to rivalry under internal mediation. Friends are peers, socially proximate and often engaged with the same Objects, so where the Mimetic Premium component rises, each friend becomes the other's Model and rival at once, and a friend's success in a domain the Individual cares about registers as a loss of standing. A close other's better performance on a task central to the Individual's own standing is rated less generously than a stranger's (Tesser and Campbell, 1982) and produces heightened arousal (Tesser, Millar and Moore, 1988).

A friendship is instrumentalised where the association value of the friend's Status comes to carry the valuation, and the friend becomes network leverage. Extraction and divergence are common to all three subtypes and are derived at Sections 9.19 and 9.20.

9.18 Kinship

9.18.1 A Relationship Given Rather Than Selected

Kinship differs from friendship and romantic partnership in formation. The relationship is not selected through the magnitude clause; it is given by birth, dependence and shared environment, and the classifier places the dependent Individual under the parent-child subtype. The Appetitive flow of that subtype runs in one direction and reverses over the life course. An infant's valuation of the carer is dominated by the A-component: the carer is Salient for an Appetitive Referent, the Individual who repairs the infant's deficits. The carer's valuation runs the other way. The infant's deficits do not reach the carer through the carer's own Body Manager. They reach the carer through the Observer, as signed Salience of high magnitude on the infant's distress, and through the Thinker's priors about which Objects repair which deficits. In old age the flow reverses and the adult child becomes the capable party. Parents continue to support their adult children across adulthood, giving both to children in need and to children who are succeeding (Fingerman, Miller, Birditt and Zarit, 2009).

9.18.2 Attachment Follows Care

The A-component of the infant's valuation is substitutable: any adequate carer repairs the same deficits. What is not substitutable builds up through interaction, as the Relational Specificity Principle requires. Attachment is therefore predicted to follow caregiving rather than genealogy, so that adoptive and genetic bonds form by the same route. Children adopted within their first year are as securely attached as children raised by their genetic parents, and security falls with later adoption (van den Dries, Juffer, van IJzendoorn and Bakermans-Kranenburg, 2009), which tracks the timing of care rather than genealogy.

9.18.3 Parents as the First Models

The first Individuals the child observes approaching and recoiling from Objects are the child's carers, so the Minimum Installation Condition installs the child's earliest Mimetic signals from them, and the Thinker commits the child's first templates of what is worth approaching and avoiding from that evidence. Kinship is the first channel of Mimetic transmission, before the surrounding culture reaches the child directly. Parent and child attitudes correspond across studies (Degner and Dalege, 2013), a correspondence the architecture predicts, though the correlational record does not by itself separate transmission from shared inheritance.

9.18.4 Sibling Rivalry

Siblings meet every condition of internal mediation at once. Siblings are maximally proximate, share the same Models in their parents, and compete for the same scarce Object, parental attention. The Rival-Constitution Principle then embeds each sibling in the other's Self-Model as the reference point against which each measures their own standing, and siblings differentiate from one another while converging on what they contest. Greater parental differential treatment of siblings is associated with more child psychopathology and poorer sibling relationships (Buist, Deković and Prinzie, 2013), and differential treatment partly tracks real differences between the siblings in age and sex (Jensen et al., 2023), so the scarce Object is not distributed at random.

9.18.5 What Holds Kinship Without the Intrinsic Component

Kinship persists where the I-component is weak, as friendship does not. Genealogy is committed in the Belief Register, kin are built into the Self-Model (Chapter 8, Section 8.10.4), and the surrounding culture supplies templates of kin obligation that the Observer reads. Estranged kin therefore remain Salient in a way former friends do not. The same holds raise investment in kin: willingness to help is highest toward siblings even when similarity of values is held constant (Baucal and Lazić, 2022). The same holds make kinship the relationship most exposed to extraction, because a kinship can continue after the I-component that would otherwise carry it has gone. A child whose achievements carry a Mimetic Premium for the parent is the characteristic case, and Section 9.19 derives where it leads. Kinship is also the relationship most exposed to divergence, because the parents supplied the Beliefs from which the child's Self-Model must differentiate, and Section 9.20 derives where that leads.

9.19 Extraction and Disengagement

Every relationship of Sections 9.15 to 9.18 is exposed to two failure paths in common, and the two run on different axes. The first concerns what one party takes from the other: one party extracts Mimetic Premium from the other beyond Appetitive sufficiency, as display, deference, compliance or standing. Three results make the path self-terminating.

First, Mimetic Desire carries no satiation signal (Differential Satiation, Axiom 6), so each concession resets the baseline against which the next demand is made, and the demand escalates.

Second, compliance with a demand replaces I-led engagement, and contingent external pressure undermines Intrinsic Motivation (Chapter 7, Section 7.6.3). Extraction therefore erodes the I-component, which under the Relational Specificity Principle is the only component of the relationship that no other Individual could supply.

Third, once the I-component is eroded, what remains is substitutable. The relationship has become replaceable, and where the cost of the extraction exceeds what the remaining components return, the integrated valuation of the relationship turns negative for the party extracted from, and that party disengages. The threshold is high, because the integrated valuation stays positive for as long as the A-component and the Individual-specific I-engagement outweigh the cost, so extraction can run for a long time before it ends the relationship. Disengagement, the cutting off of a partner, friend or relative, is the architecture's correction of a relationship run in the unstable direction. Feeling underbenefited predicts lower relationship satisfaction (Van Yperen and Buunk, 1990) and, for wives, marital disruption (DeMaris, 2007), and a friend's transgression leads Individuals to withdraw from or end the friendship (Khullar, Kirmayer and Dirks, 2021).

The threshold differs by subtype, and parents sit highest. A parent's valuation of a child carries the longest history of Ground-Resolution on one Individual, begun at birth; the child is built into the parent's Self-Model; the surrounding culture supplies templates of parental obligation; and the parent is the nearest capable party to the child's Appetitive deficits. The child's early valuation of the parent is dominated by an A-component that any adequate carer could supply. Parents therefore absorb more extraction before disengaging than any other party.

Extraction and disengagement are the dyadic ground of the Push-Pull Obligation Principle. Obligation beyond Appetitive sufficiency is stable only when it runs outward from the duty-bearer's capacity, because obligation run inward on extracted entitlement consumes the I-component that sustains the relationship, and the party extracted from disengages. The principle is derived at Chapter 18, Section 18.10, with its dyadic formalism at Appendix 18B.

9.20 Divergence and Estrangement

The second failure path concerns not what one party takes from the other but what each holds. Each Individual's Belief Register is written by that Individual's own Thinker, and nothing in the architecture lets one Individual write another's. Parents supply the child's first templates (Section 9.18.3), but the child's Thinker commits its own posteriors on the evidence the child meets, and the child's Self-Model forms partly against the parents: the child knows that the child is not the parent, and under the Rival-Constitution Principle a proximate Model is also the reference point from which a Self-Model differentiates (Chapter 8, Section 8.10). The Beliefs the child commits will therefore diverge from the parents' somewhere, and where they diverge on Beliefs built into either party's Self-Model, each party's Belief is evidence against the other's Self-Model. Political allegiance is the characteristic case. The relationship becomes a site of rivalry between Belief Registers rather than a channel of extraction.

Estrangement between mothers and adult children is predicted by dissimilarity of values rather than by violation of serious social norms (Gilligan, Suitor and Pillemer, 2015), which is the pattern the Belief-Register axis predicts and the extraction axis does not.

The two axes are independent. Extraction can run between Individuals whose Beliefs agree, and divergence can run between Individuals neither of whom extracts anything from the other. Extraction ends in disengagement once the party extracted from crosses a high threshold. Divergence ends in estrangement where the divergent Beliefs are load-bearing in either Self-Model. Divergence met by each party updating on the other's expressed position is productive friction, and divergence becomes Conflict only where the interface crosses into grinding, which it does when either party's Grievance Belief reads the other as taker or threat (Chapter 16, Sections 16.2 and 16.3). The perturbation that difference imposes on a Self-Model is derived at Chapter 15, Section 15.8.

The architecture predicts that estrangement between parents and adult children is initiated more often by the child than by the parent, on both axes at once: the drive to differentiate runs from child to parent, and the parent's threshold for disengagement is the highest of any party (Section 9.19).

Establishment Summary

This chapter has established:

Claim	Section	Status
The Decision Hub integrates A, I, M through common-currency valuation	9.1	Empirically grounded
The AIM Integration Equation describes this integration functionally	9.2	Framework-specific formalism
Composition and intensity are independent dimensions of motivation	9.3	Derived from integration architecture
Neurodiversity produces individual variation in AIM parameters	9.4	Framework assumption
The nature–nurture partition is architecturally ill-posed at the firing-sequence level; AIM's taxonomy operates on behavioural signatures independent of developmental pedigree	9.4.6	Derived from integration architecture
The Phenomenological Reality Principle requires respecting source-signals as real felt experience	9.5	Normative foundation
The Decision Hub operates as a cybernetic control system	9.6	Theoretical extension
Structural Restlessness emerges from differential satiation dynamics	9.7	Derived from control-system properties
Habits are cached prior-configurations bypassing real-time integration	9.8	Integration of dual-process literature
Source Opacity results from common-currency integration	9.9	Derived from information theory
Source Opacity applies with compounded effect to cached decisions	9.9.7	Extension of Source Opacity
The Aggregation Gradient is the structural complement to Source Opacity: M-detection survives within-Individual concealment at higher rungs of representational distance	9.10	Derived from Source Opacity locality claim and Boundedness Asymmetry Principle
Preconscious Transmission precedes conscious awareness	9.11	Empirically grounded
Phenomenological Isolation: the felt sense of self-originance is the default persistent phenomenological state, the architectural background within which every conscious-cognition operation including Empathy runs	9.11.1 (with forward pointer at 9.9.8)	Derived from Preconscious Transmission (Axiom 7) and Source Opacity (Axiom 3), bridged by the Speed-Selectivity Trade-Off
Cached retrieval is faster than real-time integration	9.11.4	Extension of timing architecture
Confabulation fills the information gap Source Opacity creates	9.12	Derived from Source Opacity
Confabulation is especially severe for habitual behaviour	9.12.6	Extension of confabulation analysis
I-Override operates through inferential reconstruction	9.13	Framework-specific mechanism
Cache-interruption is required for I-Override of habits	9.13.6	Extension of I-Override
Self-knowledge faces the introspection illusion	9.14	Integration of existing literature
Habitual behaviour presents double opacity	9.14.5	Extension of self-knowledge analysis
Romantic attraction forms through the Minimum Installation Condition and the Body Manager; durability is carried by the Thinker's I-component	9.15	Derived from integration architecture, externally supported
Relational Specificity Principle: only the I-engagement whose Object is an Individual-as-Object is specific to that Individual	9.15.4	Framework-Specific Named Result
The Collaborator coordinate takes mating-pair, parent-child and among-equals subtypes, each engaging a different weighting of the three channels	9.16	Framework-Specific Construct
Friendship is the relationship most dependent on the I-component; the three classical kinds of friendship follow from the decomposition of valuation	9.17	Derived from integration architecture
Kinship carries a one-directional Appetitive flow that reverses over the life course, and persists where the I-component is weak	9.18	Derived from integration architecture
Extracted Mimetic Premium erodes the I-component until the party extracted from disengages; parents disengage last; the dyadic ground of the Push-Pull Obligation Principle	9.19	Derived from integration architecture
Divergence of load-bearing Beliefs is a failure path independent of extraction; it ends in estrangement, and becomes Conflict only under a Grievance Belief	9.20	Derived from integration architecture
The Triple Camouflage (Layers 1–3 of the Camouflage Architecture) renders M systematically invisible at the individual level	9.14.4	Synthesis of Source Opacity, Confabulation, Object Parasitism
The Phenomenological Reality Principle applies symmetrically across the signed range; negative-valence felt experience is as architecturally real as positive-valence felt experience.	9.5 (extended)	Extension of the Phenomenological Reality Principle
Signed Operation Across Approach and Avoidance: the Decision Hub's integration is symmetric across signed terms; Source Opacity and Confabulation operate on the avoidance side of the architecture as fully as on the attraction side.	9.6.7	Derived from the Signed Valuation Principle
The Belief Register is the architectural channel through which the Thinker's posteriors propagate to the Observer's preconscious Salience computation. Habit-routed action-selection is not on that channel: it runs on the Decision Hub's cached-value store, which repetition conditions.	9.8.2 (extended)	Extension of the Belief Register
I-Override is the Thinker's Bayesian ground-attribution operating on an expanded Hypothesis Space; the channel-count asymmetry (one to one to one in channels, two to one in control) and the Thinker's self-feedback and Observer-preloading properties entail the I-Supremacy Result, the A-Disadvantage, the I-Template Override Result, and the I-Corrective Asymmetry.	9.13.5 (extended)	Derived from the channel-count asymmetry

Diagram Placeholders

The following diagrams should accompany this chapter:

Diagram 9.1: The Decision Hub Integration Architecture

Three input streams (A, I, M) converging on vmPFC/VS
Common-currency conversion process
Winner-take-all output to motor systems
Source-tags shown as discarded during conversion

Diagram 9.2: The AIM Integration Equation

Visual representation of weighted summation
Weight normalisation constraint
Lead-Tag-Last composition example

Diagram 9.3: The Cybernetic Control Architecture

Three parallel control loops (A, I, M) with distinct goal-states
Error signals feeding into Decision Hub
Feedback from behaviour to current-state updating
Continuous recalculation indicated

Diagram 9.4: Habit Formation and Retrieval

Real-time integration path (slow, deliberative)
Cached retrieval path (fast, automatic)
Context-cue triggering of cached entries
System 1 / System 2 mapping

Diagram 9.5: Timing Architecture

Timeline showing Preconscious Transmission (the Transmission Window: 60–340ms), integration (200–400ms), conscious awareness (300–500ms)
Cached retrieval timeline (50–200ms to execution)
I-Override window indicated
Preconscious M-acquisition highlighted

Diagram 9.6: The Triple Camouflage (Layers 1–3 of the Camouflage Architecture)

Three detection pathways (introspection, object-analysis, reasoning)
Each blocked by corresponding mechanism (Source Opacity, Object Parasitism, Confabulation)
Closed system indicated

Diagram 9.7: I-Override Information Architecture

Information destroyed (source-tags)
Information preserved (context, timing, behaviour, desire-content)
Two-Question Diagnostic as inferential pathway
Compensatory pathway analogy

Forward Connection

Chapter 10 will address the Evolutionary Mismatch Hypothesis: why an architecture that was adaptive in ancestral environments produces maladaptive outcomes in modern conditions. The mismatch is now specifiable with precision:

The caching mechanism that enabled efficient decision-making in stable ancestral environments produces rigid habits in rapidly-changing modern environments
The M-sensitivity that maintained coalitional standing in small-group ancestral environments produces chronic status-anxiety in large-scale anonymous modern environments
The Source Opacity that was inconsequential when A dominated and M-Objects were scarce becomes catastrophic when M-Objects are abundant and A-floors are (for many) secured
Phenomenological Isolation, derived in this chapter as the joint output of Axiom 7 (Preconscious Transmission) and Axiom 3 (Source Opacity), supplies the default persistent phenomenological state within which the Evolutionary Mismatch Hypothesis operates. The mismatch is felt by the Individual as self-originance because the architecture produces self-originance regardless of the environmental conditions under which it is read. Empathy, the downstream conscious-cognition Bayesian inference that models another Individual's state from observable configuration and stored memory, operates within rather than against Phenomenological Isolation; the categorical distinction matters for institutional design, because interventions that rely on Empathy to dissolve the felt boundary of self-originance misidentify the layer at which the boundary is generated. This is why moral exhortation against individualism, like moral exhortation against dehumanisation under the Humanisation Null Hypothesis, operates on the wrong architectural layer.
The obligation architecture of care between partners, in which push-mode care transfers no Mimetic Premium and exercises the carer's autonomy, competence and relatedness, is derived at Chapter 18, Section 18.10 and formalised in Appendix 18B, and the Reciprocal Care Principle, under which each partner holds evidence about the other that Source Opacity hides from the other, is derived at Chapter 18, Section 18.8. Both apply to the relationships of Sections 9.15 to 9.20, and Section 9.19 supplies the dyadic ground of the Push-Pull Obligation Principle.

The architectural features established in this chapter are the substrate on which evolutionary mismatch operates. Chapter 10 will complete the picture by explaining why the architecture was selected and why its consequences have changed.

What Would Falsify This?

This chapter’s claims are falsifiable as follows:

Claim	Falsification Condition
Common-currency integration in vmPFC/VS	Discovery that motivational signals remain source-tagged through to motor output
Source Opacity as architectural necessity	Tier 1: Appendix B, Section B.2, third commitment
Preconscious Transmission timing (Transmission Window: 60–340ms)	Reliable evidence that M-acquisition occurs after conscious awareness
Confabulation as structural	Tier 1: Appendix B, Section B.2, third commitment
Habits as cached prior-configurations	Discovery that habitual behaviour involves real-time integration indistinguishable from deliberated choice
I-Override effectiveness	Controlled studies showing Two-Question Diagnostic provides no improvement over untrained introspection
Structural Restlessness	Tier 1: Appendix B, Section B.2, fourth commitment
Phenomenological Reality Principle†	Discovery that source-signals can be voluntarily prevented from arising
Neurodiversity assumption†	Discovery that AIM parameters are uniform across Individuals

† Framework-assumption checks rather than operational falsifiers; failure would narrow the framework's scope rather than force revision of the architecture.

Appendix 9A: Relationship to Existing Literature

Dual-Process Theories (Kahneman, Stanovich)

AIM’s treatment of habits as cached prior-configurations integrates with dual-process models:

System 1 = cached retrieval + automatic integration
System 2 = deliberative real-time integration + I-Override

AIM adds a mechanism rather than a further description: System 1 and System 2 are two routes through one architecture, distinguished by whether action-selection acts on the Decision Hub's cached-value store or runs a fresh integration cycle.

Self-Determination Theory (Deci & Ryan)

SDT’s autonomy-competence-relatedness framework maps partially onto AIM:

Competence → I-subsystem engagement
Autonomy → I-Override capacity; freedom from M-capture
Relatedness → Complex; involves A (attachment), I (genuine connection), and M (social standing)

AIM extends SDT by providing the integration mechanism and explaining why “controlled” motivation feels different from “autonomous” motivation (different source-compositions, not merely different regulatory styles).

Mimetic Theory (Girard)

AIM operationalises Girard’s literary-anthropological insights:

Triangular desire → Preconscious Transmission through Models
Romantic Lie → Confabulation attributing M to I
Scapegoating → Crisis dynamics (Chapter 13)

AIM adds the neural mechanism, the integration architecture, and the source-taxonomy that Girard’s framework lacked.

Control Theory (Powers, Carver & Scheier)

AIM’s cybernetic framing draws directly on control-theory psychology:

Goal-state, current-state, error-signal, feedback loop
Hierarchical control (sub-goals nested within higher goals)
Continuous operation rather than discrete decision-points

AIM specifies three distinct control subsystems with different goal-types and satiation dynamics, integrating through a common Decision Hub.

Predictive Processing (Friston, Clark)

The error-signal framing is compatible with predictive processing accounts:

A-errors = interoceptive prediction errors
I-errors = competence/engagement prediction errors
M-errors = social-position prediction errors

AIM does not require commitment to the strong predictive-processing thesis but is compatible with it.

Phenomenology (Husserl, Merleau-Ponty)

The Phenomenological Reality Principle draws on phenomenological traditions that treat felt experience as the primary datum requiring explanation. AIM’s commitment to respecting source-signals as constitutive of real experience aligns with phenomenology’s bracketing of “objective” explanations in favour of experiential description.

Appendix 9B: Anticipated Objections

Objection 1: “The three-source taxonomy is arbitrary; motivation is more complex.”

Response: The taxonomy is grounded in distinct neural pathways (hypothalamic-homeostatic, dopaminergic-intrinsic, mirror-neuron-mimetic) that converge on common valuation regions. The claim is not that motivation is only these three sources but that these three are the major tributaries feeding the integration system. Finer-grained source-distinctions (e.g., different types of I-engagement) may be valuable for some purposes but do not invalidate the three-tributary architecture.

Objection 2: “Source Opacity seems unfalsifiable; any failure to introspect is attributed to opacity.”

Response: Source Opacity is falsifiable by discovery of reliable introspective source-discrimination in typical untrained Individuals. The claim is not that introspection always fails but that it systematically fails for source-composition specifically. Introspection succeeds for desire-intensity and desire-content; it fails for desire-origin. This specific failure pattern is the evidence for Source Opacity.

Objection 3: “Habits are just conditioned responses: why invoke AIM?”

Response: Simple conditioning models (stimulus → response) do not explain why habits have motivational character: why they feel like “wanting to” rather than merely “finding oneself doing.” AIM explains habits as cached motivational profiles, not merely cached motor patterns. The habit carries the source-composition from its formation, which is why habitual status-checking feels different from habitual tooth-brushing despite both being automatic.

Objection 4: “The cybernetic framing adds nothing: it just redescribes what we already know.”

Response: The cybernetic framing adds the crucial point that the three sources have different control-loop structures (homeostatic, competence-seeking, relative-position-tracking) with different satiation dynamics. This explains Structural Restlessness as an architectural consequence rather than a character flaw, and it specifies which interventions are appropriate for which type of restlessness.

Objection 5: “I-Override sounds like willpower with extra steps.”

Response: Willpower models assume the Individual can simply “choose” to resist desires. I-Override specifies what makes such resistance possible (conceptual vocabulary, diagnostic procedure, contextual information), what limits it (timing constraints, cognitive load, A-insecurity), and why it systematically fails for habits (no desire present to override: cache-interruption required instead). This is not willpower renamed but willpower decomposed into its architectural components.

Objection 6: “Neurodiversity as assumption makes the theory unfalsifiable for individual cases.”

Response: The neurodiversity assumption makes individual predictions uncertain but does not make the architectural claims unfalsifiable. The claim that all Individuals have A-I-M integration is falsifiable by discovering Individuals who lack one source. The claim that parameters vary is an accommodation of observed diversity, not a retreat from falsifiability. Specific parameter-variation Hypotheses (e.g., “autism involves reduced M-sensitivity”) are independently testable.

Objection 7: “The Phenomenological Reality Principle just validates whatever Individuals feel, including harmful desires.”

Response: The Phenomenological Reality Principle requires acknowledging felt experience as real, not endorsing action on every desire. Section 9.5.3 explicitly distinguishes respect (acknowledging the signal) from obedience (acting on it). The Phenomenological Reality Principle constrains intervention design (don’t deny that Individuals feel what they feel) without constraining evaluation (whether acting on a feeling serves wellbeing). Respect for felt experience is compatible with choosing not to act on desires that would cause harm.

Objection 8: “If M cannot be eliminated, why bother with the framework?”

Response: M-management is possible even though M-elimination is not. The framework enables Individuals to recognise M-signals, evaluate whether to act on them, and choose responses aligned with their reflective values. It enables institutions to reduce harmful M-amplification through visibility governance and structural design. The goal is not a world without M but a world where M is channelled toward positive coordination rather than destructive rivalry.

Appendix 9C: Key Terms Introduced

Term	Definition	Section
Decision Hub	The neural locus (vmPFC/VS) where A, I, M signals are integrated through common-currency valuation	9.1
Common-Currency Integration	The conversion of incommensurable motivational inputs into a single Scalar enabling comparison and choice	9.1
AIM Integration Equation	$U(x) = w_A \cdot U_A(x) + w_I \cdot U_I(x) + w_M \cdot U_M(x)$ ; the functional description of source-weighted integration	9.2
Composition	Which sources contribute to a motivated behaviour and in what proportions	9.3
Intensity	How strongly the integrated motivation drives behaviour	9.3
Lead-Tag-Last	Heuristic for characterising composition by dominant source, secondary source, and tertiary source	9.3
Phenomenological Reality Principle	The commitment that felt intensity and character of motivational signals constitutes real experience regardless of source	9.5
Cybernetic Architecture	The Decision Hub as continuously operating control system with goals, feedback, and error-correction	9.6
Structural Restlessness	The architecturally necessary condition whereby stable final satisfaction is unavailable, arising from differential satiation dynamics	9.7
Cached-Value Store	The Decision Hub's store of context-value-response associations, acted on directly in recurring situations	9.8
Source Opacity	The architectural feature whereby source-tags are discarded during integration, leaving the Individual without introspective access to motivational origins	9.9
Double Opacity	The compounded inaccessibility for habitual behaviour: original integration lost source-tags; current execution involves no deliberation	9.9.7
Confabulation	The Thinker’s routine generation of sincere but inaccurate explanations for choices, filling gaps created by Source Opacity	9.12
Romantic Lie	The civilisation-scale confabulation that desires originate autonomously within the Individual	9.12.5
Triple Camouflage (Layers 1–3 of the Camouflage Architecture)	The combination of Source Opacity, Confabulation, and Object Parasitism that renders M systematically invisible at the individual level; see Camouflage Architecture in Chapter 2 for the full eight-layer structure	9.14.4
I-Override	The deliberative mechanism that scrutinises integrated desires using the Two-Question Diagnostic	9.13
Two-Question Diagnostic	“Does my body need this?” (probes A) and “Would I still want this if no one ever saw?” (probes M)	9.13.2
I-Override Capacity	The contextual variable determining how effectively an Individual can apply I-Override	9.13.4
Cache Interruption	Preventing automatic activation of cached entries to enable real-time integration and I-Override	9.13.6
Introspection Illusion	The confident but false Belief that we have direct access to the causes of our mental states	9.14.1
Behavioural Signatures	Observable patterns (audience-dependence, satiation dynamics, model-tracking) that indicate source-composition	9.14.3
Relational Specificity Principle	Of the components of valuation directed at an Individual-as-Object, only the I-engagement whose Object is that Individual is specific to that Individual and cannot be substituted	9.15.4
Collaborator Subtype	The mating-pair, parent-child and among-equals readings of the Collaborator coordinate of the Agent-Type Vector	9.16

Appendix 9D: Chapter 9 Section Index

Section	Title	Movement
9.1	The Decision Hub: Common-Currency Valuation	A: Static Architecture
9.2	The AIM Integration Equation	A: Static Architecture
9.3	Composition and Intensity	A: Static Architecture
9.4	Neurodiversity and Individual Variation	A: Static Architecture
9.5	The Phenomenological Reality Principle: Why Sources Must Be Respected	A: Static Architecture
9.6	The Decision Hub as Control System	B: Dynamic Operation
9.7	Structural Restlessness: The Aggregate Consequence	B: Dynamic Operation
9.8	Habits and the Cached-Value Store	B: Dynamic Operation
9.9	Source Opacity: The Architectural Constraint	C: Epistemic Constraints
9.10	The Aggregation Gradient as the Structural Complement to Source Opacity	C: Epistemic Constraints
9.11	The Timing Architecture	C: Epistemic Constraints
9.12	Confabulation as Structural Consequence	D: Confabulation
9.13	I-Override Information Architecture	E: Compensatory Mechanisms
9.14	Implications for Self-Knowledge	E: Compensatory Mechanisms
9.15	Romantic and Sexual Relationships: Three Channels in One Dyad	F: The Integration Applied to Relationships
9.16	The Collaborator Subtypes	F: The Integration Applied to Relationships
9.17	Friendship	F: The Integration Applied to Relationships
9.18	Kinship	F: The Integration Applied to Relationships
9.19	Extraction and Disengagement	F: The Integration Applied to Relationships
9.20	Divergence and Estrangement	F: The Integration Applied to Relationships

References

Aristotle. (2009). The Nicomachean ethics (D. Ross, Trans.; L. Brown, Ed.). Oxford University Press.

Baucal, A., & Lazić, A. (2022). Selfish genes or selfish memes: The effect of genetic relatedness versus value similarity on altruism. Psihologija, 55(4), 379–395. https://doi.org/10.2298/PSI211107009B

Buist, K. L., Deković, M., & Prinzie, P. (2013). Sibling relationship quality and psychopathology of children and adolescents: A meta-analysis. Clinical Psychology Review, 33(1), 97–106. https://doi.org/10.1016/j.cpr.2012.10.007

Carver, C. S., & Scheier, M. F. (1998). On the self-regulation of behavior. Cambridge University Press.

Conroy-Beam, D., & Buss, D. M. (2016). Do mate preferences influence actual mating decisions? Evidence from computer simulations and three studies of mated couples. Journal of Personality and Social Psychology, 111(1), 53–66. https://doi.org/10.1037/pspi0000054

Conroy-Beam, D., Goetz, C. D., & Buss, D. M. (2016). What predicts romantic relationship satisfaction and mate retention intensity: Mate preference fulfillment or mate value discrepancies? Evolution and Human Behavior, 37(6), 440–448. https://doi.org/10.1016/j.evolhumbehav.2016.04.003

Degner, J., & Dalege, J. (2013). The apple does not fall far from the tree, or does it? A meta-analysis of parent–child similarity in intergroup attitudes. Psychological Bulletin, 139(6), 1270–1304. https://doi.org/10.1037/a0031436

DeMaris, A. (2007). The role of relationship inequity in marital disruption. Journal of Social and Personal Relationships, 24(2), 177–195. https://doi.org/10.1177/0265407507075409

Eastwick, P. W., & Hunt, L. L. (2014). Relational mate value: Consensus and uniqueness in romantic evaluations. Journal of Personality and Social Psychology, 106(5), 728–751. https://doi.org/10.1037/a0035884

Eastwick, P. W., Luchies, L. B., Finkel, E. J., & Hunt, L. L. (2014). The predictive validity of ideal partner preferences: A review and meta-analysis. Psychological Bulletin, 140(3), 623–665. https://doi.org/10.1037/a0032432

Fingerman, K., Miller, L., Birditt, K., & Zarit, S. (2009). Giving to the good and the needy: Parental support of grown children. Journal of Marriage and Family, 71(5), 1220–1233. https://doi.org/10.1111/j.1741-3737.2009.00665.x

Foster, C. A., Witcher, B. S., Campbell, W. K., & Green, J. D. (1998). Arousal and attraction: Evidence for automatic and controlled processes. Journal of Personality and Social Psychology, 74(1), 86–101. https://doi.org/10.1037/0022-3514.74.1.86

Gerlach, T. M., Arslan, R. C., Schultze, T., Reinhard, S. K., & Penke, L. (2019). Predictive validity and adjustment of ideal partner preferences across the transition into romantic relationships. Journal of Personality and Social Psychology, 116(2), 313–330. https://doi.org/10.1037/pspp0000170

Gilligan, M., Suitor, J. J., & Pillemer, K. (2015). Estrangement between mothers and adult children: The role of norms and values. Journal of Marriage and Family, 77(4), 908–920. https://doi.org/10.1111/jomf.12207

Gouda-Vossos, A., Nakagawa, S., Dixson, B. J. W., & Brooks, R. C. (2018). Mate choice copying in humans: A systematic review and meta-analysis. Adaptive Human Behavior and Physiology, 4(4), 364–386. https://doi.org/10.1007/s40750-018-0099-y

Hall, J. A. (2019). How many hours does it take to make a friend? Journal of Social and Personal Relationships, 36(4), 1278–1296. https://doi.org/10.1177/0265407518761225

Jensen, A. C., Jorgensen-Wells, M. A., Andrus, L. E., Pickett, J. M., Leiter, V. K., Hadlock, M. E. B., & Dayley, J. C. (2023). Sibling differences and parents' differential treatment of siblings: A multilevel meta-analysis. Developmental Psychology, 59(4), 644–654. https://doi.org/10.1037/dev0001406

Joel, S., Eastwick, P. W., Allison, C. J., Arriaga, X. B., Baker, Z. G., Bar-Kalifa, E., Bergeron, S., Birnbaum, G. E., Brock, R. L., Brumbaugh, C. C., Carmichael, C. L., Chen, S., Clarke, J., Cobb, R. J., Coolsen, M. K., Davis, J., de Jong, D. C., Debrot, A., DeHaas, E. C., … Wolf, S. (2020). Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. Proceedings of the National Academy of Sciences, 117(32), 19061–19071. https://doi.org/10.1073/pnas.1917036117

Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.

Khullar, T. H., Kirmayer, M. H., & Dirks, M. A. (2021). Relationship dissolution in the friendships of emerging adults: How, when, and why? Journal of Social and Personal Relationships, 38(11), 3243–3264. https://doi.org/10.1177/02654075211026015

Klusmann, D. (2002). Sexual motivation and the duration of partnership. Archives of Sexual Behavior, 31(3), 275–287. https://doi.org/10.1023/A:1015205020769

Le, B., & Agnew, C. R. (2003). Commitment and its theorized determinants: A meta-analysis of the Investment Model. Personal Relationships, 10(1), 37–57. https://doi.org/10.1111/1475-6811.00035

Libet, B., Gleason, C. A., Wright, E. W., & Pearl, D. K. (1983). Time of conscious intention to act in relation to onset of cerebral activity (readiness-potential): The unconscious initiation of a freely voluntary act. Brain, 106(3), 623–642.

McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444. https://doi.org/10.1146/annurev.soc.27.1.415

Powers, W. T. (1973). Behavior: The control of perception. Aldine.

Roberts, S. G. B., & Dunbar, R. I. M. (2011). The costs of family and friends: An 18-month longitudinal study of relationship maintenance and decay. Evolution and Human Behavior, 32(3), 186–197. https://doi.org/10.1016/j.evolhumbehav.2010.08.005

Roberts, S. G. B., & Dunbar, R. I. M. (2015). Managing relationship decay. Human Nature, 26(4), 426–450. https://doi.org/10.1007/s12110-015-9242-7

Rohrer, J. M., Keller, T., & Elwert, F. (2021). Proximity can induce diverse friendships: A large randomized classroom experiment. PLOS ONE, 16(8), e0255097. https://doi.org/10.1371/journal.pone.0255097

Street, S. E., Morgan, T. J. H., Thornton, A., Brown, G. R., Laland, K. N., & Cross, C. P. (2018). Human mate-choice copying is domain-general social learning. Scientific Reports, 8, 1715. https://doi.org/10.1038/s41598-018-19770-8

Tesser, A., & Campbell, J. (1982). Self-evaluation maintenance and the perception of friends and strangers. Journal of Personality, 50(3), 261–279. https://doi.org/10.1111/j.1467-6494.1982.tb00750.x

Tesser, A., Millar, M., & Moore, J. (1988). Some affective consequences of social comparison and reflection processes: The pain and pleasure of being close. Journal of Personality and Social Psychology, 54(1), 49–61. https://doi.org/10.1037/0022-3514.54.1.49

van den Dries, L., Juffer, F., van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (2009). Fostering security? A meta-analysis of attachment in adopted children. Children and Youth Services Review, 31(3), 410–421. https://doi.org/10.1016/j.childyouth.2008.09.008

Van Yperen, N. W., & Buunk, B. P. (1990). A longitudinal study of equity and satisfaction in intimate relationships. European Journal of Social Psychology, 20(4), 287–309. https://doi.org/10.1002/ejsp.2420200403

Wiener, N. (1948). Cybernetics: Or control and communication in the animal and the machine. MIT Press.

Neural Integration and Source Opacity