Path Integration

Neural mechanism for tracking location through accumulation of self-movement vectors; shown to play the role of position encodings in TEM.

Neighborhood — ranked by edge-count

framework

Tolman-Eichenbaum Machine (TEM)
implements
Neuroscience model of hippocampal formation that the paper shows is mathematically equivalent to a transformer with recurrent position encodings.

method

Recurrent Position Encodings
analogous_to
Key modification to transformers proposed in this paper: position encodings generated by a recurrent network trained on action sequences.

concept

Grid Cells
associated_with
Spatially periodic firing neurons in medial entorhinal cortex; TEM-t learns representations resembling these.
Accurate path integration in continuous attractor network models of grid cells (Burak & Fiete, 2009)
cites
Work on path integration stability using neural manifold constraints; mentioned in context of stabilising recurrent position encodings.

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Integrationconcept0.800
Key property enabling organized collective action; mechanism by which parts coordinate into functional wholes.
Information Integrationmethod0.781
Integration of signals across components to compute holistic states, a hallmark of organismic individuality and basal cognition
Path Patchingmethod0.776
Method by Goldowsky-Dill et al. 2023 for localizing model behavior via targeted activation interventions
Sensory Integrationconcept0.772
Path Expansion Methodmethod0.769
The core analytical technique of expanding transformer computations from layer-by-layer products into sums of end-to-end path terms for independent analysis
Path-Specific Objectivesframework0.760
An approach training agents to avoid unsafe pathways leading to deception, informed by Causal Influence Diagrams
Representation-based Pathconcept0.749
The path in activation space derived by fitting the representation manifold, used to steer along the geometric structure of internal representations.
Euler integrationmethod0.742
Numerical method used to integrate stochastic differential equations of the primordial soup.