9/12 What do our findings reveal about hippocampal computation? We tested several models—but only Clone-Structured Causal Graph (CSCG) @dileeplearning.bsky.social matched the orthogonalized states and learning trajectory, highlighting hidden-state inference as a key learning principle.

8/12 At the single-cell level, cells in expert mice showed a continuum: some cells were “place-like,” while others became “splitters” firing selectively for Near vs. Far trials. Single-cell responses evolve across sessions, driving overall decorrelation. Explore our data: cognitivemap.janelia.org

6/12 Using UMAP on our data, the CA1 manifold evolved over learning. It started as an unstructured cluster, then formed a hub-and-spoke, and finally a ring that split into branches by trial type—refining similar inputs into distinct states that eventually capture the underlying task structure.

5/12 During learning, hippocampal activity became more distinct. Early on, neurons in ambiguous zones responded similarly; over time, responses separated and tuned to specific task states.

4/12 During training, mice showed 4 behavioral strategies:
1. Random licking,
2. Licking at both potential rewards,
3. Stopping after collecting a reward,
4. Only licking at the correct reward location.

3/12 Task: Mice ran along two similar VR corridors—one with a Near reward and one with a Far reward. An indicator cue at the start signaled the active corridor. Expert mice learned to lick only at the correct reward zone.

2/12 We used a custom 2P-RAM mesoscope (designed by @sofroniewn.bsky.social, Dan Flickinger, & Karel Svoboda. @hhmijanelia.bsky.social) to track thousands of CA1 cells across sessions. It let us see changes in single-cell & population activity throughout learning.