The Tolman-Eichenbaum Machine develops hexagonal grid and remapping place cells

Whittington et al. train the Tolman-Eichenbaum Machine, a recurrent generative model, on a range of spatial and non-spatial relational-structure tasks [whittington-etal-2020-tolman-eichenbaum-machine] It develops entorhinal-like units with real hexagonal grid periodicity, plus band, border, and object-vector cells [whittington-etal-2020-tolman-eichenbaum-machine] It also develops hippocampal-like place cells whose remapping between environments follows a structured, non-random pattern matching biological data [whittington-etal-2020-tolman-eichenbaum-machine] No causal intervention is performed; the correct citation is the Cell/bioRxiv DOI, not arXiv:1910.07663 (an unrelated paper) [whittington-etal-2020-tolman-eichenbaum-machine]