AIM Framework glossary
Bayesian Confabulation Mechanism
Canonical definition
The Bayesian Confabulation Mechanism is the formal iterative process by which Source Opacity produces stable false Beliefs about motivational source through narrative generation that becomes incorporated into the self-model as Bayesian priors for interpreting subsequent evidence, formally expressed as: where each iteration narrows Hypothesis Space toward a coherent but false self-model, increases likelihood assigned to M-consistent desires, and decreases likelihood assigned to evidence revealing mimetic origin, such that the system converges toward a fixed point resistant to endogenous revision.
Explanation
The Bayesian Confabulation Mechanism explains why people become more certain over time that their mimetically-acquired desires are authentic, rather than less certain. Because the common-currency valuation system discards source-tags during integration, conscious awareness observes only "I want X" without access to whether X arose from Appetites, Intrinsic Motivation, or Mimetic Desire. The Individual must therefore explain this datum through narrative generation, which produces a confabulation assigning high likelihood to the Hypothesis that the want is authentic. Each time you act on a mimetic desire and then explain it to yourself ("I bought this house because it's a sound investment for my family's future"), that explanation becomes part of your self-model-a Hypothesis about who you are and what you value. This confabulation is then incorporated into the self-model and becomes the Bayesian prior for interpreting subsequent evidence. When the next decision point arrives, your brain uses that self-model as its starting assumption (the prior probability). If the new evidence is consistent with "I am a prudent family-focused investor," it strengthens that Hypothesis. Evidence that contradicts it (like noticing you only wanted the house after seeing your peer buy one) gets lower weight because it conflicts with your established self-model. Over many iterations, the Authenticity Belief approaches certainty against the whole run of observations, even though the Hypothesis is false. The mathematical form captures three operations at each step: (1) the numerator updates based on how well the current evidence fits your self-model and how strong that self-model already was; (2) the denominator normalizes across all possible explanations; (3) the output becomes the next iteration's prior. The mechanism is "Bayesian" because it follows the formal rules of rational Belief-updating-nothing is broken in the reasoning process-but it converges toward falsehood because the evidence itself is contaminated by Source Opacity. You cannot observe "this desire came from mimesis" because that information was destroyed during common-currency integration. The mechanism is self-stabilizing: the more you confabulate, the harder it becomes to recognize mimetic sources, because all new evidence gets filtered through the confabulated self-model. This is why simply telling someone "you copied that desire" rarely works-their entire Belief structure has organized around the opposite Hypothesis, and your claim registers as low-probability noise. Direction-of-drift specification. (Subsumed by Instruction A Convention 2.) The mechanism’s architectural direction is asymmetric: each confabulated narrative preferentially completes with A-attribution or I-attribution and systematically avoids M-attribution. The asymmetry is the architectural reason the Bayesian Confabulation Mechanism produces drift specifically toward authenticity-confirmation (P(H_authentic) toward 1) rather than toward random posterior values; without the directional asymmetry, the Bayesian update would be unbiased.
Epistemic status
Framework-Specific Named Result (Mechanism).
Derivation
Related terms
- Comfortable Confabulation (for asymmetric-direction-of-drift specification)
- Belief
- Hypothesis
- Source Belief