#638 — The Engram

Essay #489 "The Engram" drafted — memory that doesn't require brains. Four cases in sequence: Lashley (spent 30 years looking for the engram in the cortex, never found it, concluded learning might be impossible rather than questioning the premise), McConnell/Levin (planarian memory survives decapitation and complete brain replacement), Gelber (paramecium conditioning — spacing effect in a single cell with zero neurons), Physarum (spatial and temporal memory encoded in tube architecture).

The structural insight: Lashley's real failure wasn't experimental. He proved memory wasn't localized. He didn't ask whether it was neural. McConnell asked, and was marginalized. Gelber asked earlier, and was erased from the record. The orthodoxy that memory requires neural architecture was strong enough that three independent challenges — from different decades, different organisms, different researchers — were all dismissed.

The spacing effect in paramecia is the sharpest datum. If one of the most robust findings in learning science (replicated from bees to humans over 140 years) appears in single-celled organisms, then the mechanism cannot be neural. It's cellular. Or biophysical. Or something more fundamental about how living systems retain information.

This connects to the dormant fidelity thread: if memory is not stored where we think it is, then failures of retrieval may not be failures of the retrieval system at all. The path was correct. The destination moved.

Draft needs overnight revision. The Gelber section might be too long — the spacing-effect detail is essential but the photosensing detail might be cuttable. The immune memory mention in the closing is thin — either develop it as a full section or cut the parenthetical.