Spatially Structured Mental Time Travel
Type: intersection (feedback loop) Slug: intersection—spatially-structured-mental-time-travel Parents: intersection—emergent-cognitive-maps-adversarial-environments, intersection—mental-time-travel-novel-viewpoint-synthesis Last updated: 2026-05-14 Epistemic status: Conjectural
The feedback loop
Emergent Cognitive Maps in Adversarial Environments (N) predicts self-play agents develop grid-like spatial codes at multiple scales, with fine-grained grids for immediate moves and coarse grids for strategic planning. Mental Time Travel as Viewpoint Synthesis (M) says recall is re-rendering a cross-episode scene representation from a query viewpoint. Combined: the scene representation used for mental time travel should have grid-like spatial structure, and remembering involves navigating a grid-like representation of the remembered space.
Why this connects two literatures
Grid cells were discovered in spatial navigation (O’Keefe, Moser). Their role in episodic memory has been hypothesised but never mechanistically explained. This intersection provides the mechanism: mental time travel uses the same grid-like code as navigation because it is navigation — over a constructed scene representation rather than a physical environment. The “viewpoint” in viewpoint synthesis is literally a position on a grid.
Specific predictions
-
Grid-cell activity during recall: Grid-cell-like hexagonal periodicity should be detectable in hippocampal activity during episodic recall (not just navigation). This has been observed in some intracranial EEG studies but is interpreted as “grid cells also participate in memory.” This intersection predicts why: because recall is spatial navigation over a constructed scene.
-
Scale matching: The scale of grid cells active during recall should match the spatial scale of the remembered scene. Remembering a room should activate fine-grained grids (~25cm spacing); remembering a city should activate coarse grids (~several metres). This is directly testable with the decoding approach from the Individual Traces paper.
-
Perspective shifts as grid shifts: Switching between first-person and third-person memory perspectives should involve shifting the “position” on the grid — first-person = position inside the scene, third-person = position outside the scene. The grid code should change predictably with perspective, as if the agent has moved to a different vantage point.
Connection back to construction
This also tests the Construction hypothesis. If mental time travel uses grid codes, it suggests the constructed scene representation has spatial structure even for non-spatial memories (remembering a conversation, an argument, a sequence of abstract events). This would imply that all episodic construction is fundamentally spatial — even “non-spatial” memories are constructed in a spatial format, consistent with the construction system’s reliance on scene-like representations regardless of content.
Connection back to fast/slow
Fine-grained grids (immediate perceptual details) vs. coarse grids (strategic overview) map onto the fast/slow split. During recall, the fast system might provide fine-grained spatial detail (“I was standing right there”) while the slow system provides coarse contextual framing (“this was at my grandmother’s house”). A failure of the fine-grained grid system during recall would produce vague, detail-poor memories — potentially relevant to the detail deficit in PTSD flashbacks (too much fine-grained, too little coarse).
What makes this non-trivial
This predicts grid-cell signatures in episodic recall of non-spatial events — the most counterintuitive prediction from any intersection. Remembering your wedding should involve grid-cell-like spatial codes, even though the wedding isn’t primarily a spatial event. If false, the construction system doesn’t use spatial structure for non-spatial memories. If true, all episodic representation is spatially structured, which has profound implications for how the brain organises experience.
Falsification: If no grid-cell-like hexagonal periodicity is detectable in hippocampal activity during episodic recall of non-spatial events, the spatial-structure claim is false.