Memory in Situated Learning Accessibility: Record Standing, Scope, and What Actually Ages

A theory of situated learning accessibility that can say where a learning relation is, how it moves, what the evidence supports and what it warrants a system to do, still cannot say for how long any of that remains true. The four earlier layers of the Dynamic Map of Learning Accessibility (MDAA) give every proposition a configuration scope, a status, a provenance and, at the action level, a warrant; none gives a recorded observation a duration. The loop currently reads an observation from yesterday and one from three months ago with the same weight. This paper adds the missing coordinate. The obvious answer, that evidence decays with time, is the one the programme forbids. Much of what looks like decay did not decay: the condition changed, and the record left its scope while remaining true under the condition in which it was made. That is scope, and the first and fourth papers already handle it. The question of this layer is what remains after scope has been discounted: what actually ages, and how one can tell. We first separate two memories that are routinely conflated. The memory of the process is the learner's, the M axis of the model lattice, and the identification bench of the second paper has already shown that it is not identifiable on a single trajectory. The memory of the record is the system's: what the MDAA keeps about a relation, and for how long what it kept may still be read as current evidence. This paper is about the second. We then introduce one new object, the standing of a record, ν(t; r), with four values that the current loop collapses into one: applicable, out of scope, aged, and unevaluable. Each returns by a different mechanism, which is the test that they are distinct states rather than labels. We argue, inheriting the bench and the sociohydrodynamic result that macroscopic memory can emerge from memoryless agents, that aging cannot be inferred from the trajectory it would weight and must therefore be a declared specification, frozen before use, in the same position that the warrant specification occupies in the fourth paper. Standing has two windows, one for what appears in the appraisal and one for what warrants action, with the second contained in the first and governed by a different mechanism; and standing is independent of authorized pause. Three collision cases test whether the distinctions earn their cost: scope read as age, the learner's forgetting read as the record's aging, and vocabulary change read as scope change. A preregistered four-world bench, frozen by hash and then run over four hundred executions, tests whether a reader can distinguish true aging from unlogged condition change, from the learner losing what she knew, and from vocabulary change. All seven committed predictions are sustained. The reader that infers age from the trajectory does not separate true aging from an unlogged change of condition; the reader that reads declared conditions as scope separates them perfectly when the condition was logged and not at all when it was not. The difference between the two is the record of the condition, not the reader. Preprint. Conceptual manuscript with a preregistered simulation study, not peer reviewed. Fifth paper of the MDAA series. The preregistration and its three amendments were recorded before any seed of the grid was drawn, which is what distinguishes this paper from the four that precede it. The bench of Section 12, the preregistration and its amendments, the frozen aging specification and its hash, the generators, the readers and the raw results are archived at https://doi.org/10.5281/zenodo.22728835