AIM Framework glossary
Monotone Belief-Race Lemma
Canonical definition
The result that the Belief Register's update dynamics cannot generate oscillation of their own under content-agnostic updating. Because the likelihood each hypothesis receives depends on the delivered Evidence (the integrated Scalar and Salience) and not on the current prevalence of any hypothesis in the Belief Register (Axiom 8; restatement B3), the linearised update block is diagonally similar to a symmetric matrix at every Register state, and its spectrum is therefore real: each hypothesis competes only against the belief-weighted mean likelihood, and such a race sorts monotonically toward the best-supported hypothesis without cycling. Rotation in the Belief layer requires likelihoods that respond to hypothesis prevalence with intransitive structure (each hypothesis gaining support because another is currently held, around a closed cycle), which violates content-agnostic updating and is not a parameter regime of the architecture; the delay-feedback self-coupling preserves the result at every recall lag. Jointly with the one-signed coupling commitments of the fast and network layers, the Monotone Belief-Race Lemma upgrades bifurcation-type preservation from architectural argument to derived result: at decohered operating points the first loss of stability of the coupled system is a steady-state crossing, and the Saddle-Node follows by genericity.
Explanation
Beliefs inside one mind are in a race: whichever hypothesis best explains the evidence gains ground, and the others lose it. The Monotone Belief-Race Lemma establishes that this race can never go in circles, because of one feature of Bayesian updating: the referee is blind. How strongly the evidence supports a hypothesis does not depend on how popular that hypothesis currently is. A race where every runner competes only against the field average always produces a winner by steady sorting. For beliefs to cycle, popularity itself would have to change the evidence, in a rock-paper-scissors pattern where each belief gains because another is currently held, and the architecture does not permit that. The race is not fair in one respect: the track tilts toward the front-runner, whose gains compound and feed back as further evidence, so the leader is biased toward growing its lead, which is Bayesian Lock-In run as a race. The consequence is stark: a belief system has no built-in pendulum, no thermostat that swings it back after it overshoots. It either holds steady or it runs away and falls, and anything that looks like a self-correcting swing in a society's beliefs is the work of an installed corrective structure, never of the bare architecture. The same blindness is what keeps reform open: a better-supported hypothesis gains from the moment it becomes representable, however unpopular, so the architecture withholds self-correction, not correctability, and the Interference Term marks the point where the leader's compounding can be throttled.
Epistemic status
Framework-Specific Named Result (qualitative). Formal statement and proof sketch at Appendix A, Section A.5.6; network-scale application at Appendix A, Section A.5.7.6.
Derivation
Appendix A, Section A.5.6; Chapter 13, Section 13.9; Appendix A, Section A.5.7.6.