This preprint proposes a cytoskeletal-dendritic key-lock model of associative memory in which fast dendritic and spine dynamics govern immediate access to encoding and recall, while slower cytoskeletal and structural states bias which branches remain available for stabilization, linking, and later retrieval. The paper synthesizes evidence from active dendrites, spine biophysics, microtubule and Tau regulation, synaptic tagging and capture, local translation, and engram research into a falsifiable two-timescale framework with experimental predictions.
Robin Langell (Thu,) studied this question.