ARTFEED — Contemporary Art Intelligence

Algebro-Deterministic Memory Architecture VaCoAl Proposed as Substrate for Hippocampal Models

publication · 2026-05-18

A recent study published on arXiv (2605.15652) presents VaCoAl, an algebro-deterministic hyperdimensional memory framework utilizing Galois-field linear feedback shift registers (LFSRs). The researchers assert that VaCoAl can act as a common algebraic entity linking Vector-HaSH and the Tolman-Eichenbaum Machine (TEM), which are prominent models explaining hippocampal-entorhinal circuit operations. Both Vector-HaSH and TEM suggest that the hippocampus extracts content from a prearranged grid-cell framework and facilitates compositional memory through ripple-mediated replay. Findings from human intracranial EEG studies indicate that hippocampal sharp-wave ripples (SWRs) regulate episodic recall, while ripple-locked cortical reactivation mirrors encoding patterns. VaCoAl's deterministic Galois-field diffusion offers a substrate-level alternative to the random projection in Vector-HaSH, potentially reconciling theoretical and empirical insights in hippocampal memory studies.

Key facts

  • VaCoAl is an algebro-deterministic hyperdimensional memory architecture
  • Built from Galois-field linear feedback shift registers (LFSRs)
  • Proposed as shared algebraic object for Vector-HaSH and TEM models
  • Vector-HaSH and TEM model hippocampal-entorhinal circuit function
  • Human iEEG shows hippocampal SWRs gate episodic recall
  • Ripple-locked cortical reactivation recapitulates encoding patterns
  • Multi-hop replay fidelity decays multiplicatively along sequence length
  • VaCoAl's Galois-field diffusion satisfies quasi-orthogonality requirement

Entities

Institutions

  • arXiv

Sources