Variable frame theory explains focal-point coordination by assigning each description an availability, the probability that the description comes to mind, but it leaves that availability exogenous. This paper makes availability endogenous. It identifies the variable-frame availability of a description with a self-consistent population marginal generated by logit revision over descriptions, and it supplies a finite-population microfoundation: under empirical convergence the limiting population share is the marginal law of a uniformly sampled agent, and under exchangeability it is the marginal law of every labeled agent. The central result is the consensus-coordination gap. Because descriptions have extensions (each description picks out a set of options), agreeing on a description is not the same as agreeing on an action. Complete salience consensus on a description c yields a matching probability of only 1/|Oc|, and the exact shortfall from the sharpest available description is Gamma (c) = 1/mind |Od| - 1/|Oc|. A population can therefore agree perfectly on how to describe a situation and still fail to coordinate reliably on what to do. The effect is invisible to atomic-label voter, naming-game, and Potts consensus models, in which labels already are actions and the matching kernel is the identity; here the kernel is a nontrivial Gram operator built from overlapping extensions. The paper develops the accompanying phase structure with standard statistical-mechanics tools. A global Lyapunov (free-energy) identity rules out spurious flow cycles and justifies deterministic basin selection; the symmetric profile loses stability at lambda*theta/M = 1; the Curie-Weiss-Potts transition is first order for three or more descriptions; payoff-inferior false focal points persist when the mimetic field beats the objective gap; an exact two-description finite-temperature theorem gives the saddle-node window and the seed threshold for branch selection; and a small-field continuation theorem shows that heterogeneous distinctiveness tilts but does not destroy the ordered branches above the spinodal. Multi-concept frames add an exact covariance sign identity and an inference decomposition in which perfect information leaves the coordination-lock core intact while prevalence-based inference only enlarges the false-focal region. The term meconnaissance is used in the Girardian sense: the population mistakes convergent use for independent warrant. This is the first paper in a series; companion papers treat scapegoat coordination and strategic frame manipulation.
K. Fathi (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: