AIM Framework glossary
Confabulation Cascade
Canonical definition
Confabulation Cascade is the process by which individual Confabulations propagate through social networks, compound through Bayesian updating (each observed Confabulation becoming evidence that updates others' Beliefs), and converge toward population-level shared narratives that are resistant to revision because: (1) each Individual's posterior is shaped by the collective prior; (2) expressing doubt produces social sanction; and (3) the converged narrative has excluded from Hypothesis Space the structural explanations that would reveal its confabulated nature.
Explanation
The cascade operates through three phases: Phase 1: Expression. Individual A generates a Confabulation and expresses it publicly. The Confabulation enters others' evidential environments. Phase 2: Propagation. Individuals B, C, D observe A's expressed Confabulation and incorporate it as evidence through prestige-biased learning (if A is high-status), social proof (if multiple individuals express similar narratives), and mimetic template transmission (if the narrative provides a deployable schema). Phase 3: Convergence. As more Individuals adopt the Confabulation, the collective prior shifts. New Individuals encountering the narrative observe not one person expressing it but many, treating the distribution as evidence of validity. The cascade becomes self-reinforcing: the more people express the Confabulation, the more others treat it as validated, the more it propagates. At convergence, the population reaches a stable configuration (a collective fixed point) where: the collective prior assigns near-certainty to the confabulated narrative; alternative explanations (especially structural ones) receive near-zero likelihood; dissent is treated as evidence of defection or pathology; and no Individual can exit the trap because the collective prior has made alternative explanations unintelligible. Dissenting Individuals are also locked into the collective Confabulation; they cannot exit by dissent alone. The formal mathematics of cascade dynamics (Bayesian updating, likelihood ratios, convergence theorems, fixed-point analysis) is developed in Chapter 11. Chapter 11 establishes the phenomenology of the cascade: what it feels like to be inside a converging system, why the convergence feels like truth-discovery rather than collective error, and why exit is so difficult. Direction-of-drift and substrate specification. (Subsumed by Instruction A Convention 2.) The Confabulation Cascade operates with the architectural directional asymmetry specified by Convention 2 (preferential A-and-I-attribution, systematic M-avoidance) and produces compounding through the Bayesian content-agnostic mechanism (Axiom 8). The cascade’s destructive consequences (scapegoating, institutional capture, civilisational lock-in) operate through M-channel compounding on A-and-I substrates: the A-precarity, I-domain destruction, and Status displacement that are the felt-cascade phenomenology are produced by M-channel content-cascade dynamics.
Epistemic status
Framework-Specific Construct, with formal development in Chapter 11.* Scale specification. The Confabulation Cascade names the inter-personal, network-propagation dynamic by which one Individual's expressed Confabulation enters another's evidential environment and compounds across the network. Its individual-scale counterpart is the Individual Confabulation Cascade (the intrapersonal Bayesian loop within a single Individual updating on their own self-generated narratives). Its population-scale counterpart is the Collective Convergence Mechanism (the synchronisation of many individual cascades into shared population-level Belief equilibria via social proof, prestige-biased learning, and mimetic template transmission). The three are the same Bayesian compounding mechanism operating at three coupled scales: within an Individual, between Individuals, and across populations.
Derivation
Chapter 11, Section 11.11; Chapter 13, Section 13.4.1; Chapter 14, Section 14.3.6; Appendix A, Section A.5.8.3.