Word Memory
Evidence about persistent lexical state, recall, consolidation, restart checks, and anti-forgetting, with persistence kept distinct from retrieval correctness.
Persistent lexical state
Relevant phases test whether recorded lexical or structured state survives a defined save, restart, or cold-reload procedure. The scope of each persistence check remains local to that test.
Recall and correctness
State identity, successful retrieval, and correct use are different gates. A value surviving restart does not prove that later answers are correct.
Consolidation and anti-forgetting
The archive contains consolidation and anti-forgetting experiments with phase-specific baselines and candidate outcomes. No global memory score is inferred from unrelated tests.
Limitations
Some evidence is candidate-only, synthetic, frozen, or limited to structured meanings. Missing downstream language validation is not filled by persistence results.