Read on Twitter
Is the cognitive map a topological map? Nope!
We recorded from CA1 place cells in rats and changed the connectivity of the environment without changing its geometry. Place cells did not care!
Come in for a thread, or click for the preprint
https://www.biorxiv.org/content/10.1101/2020.10.20.346130v1
1/
We recorded from CA1 place cells in rats and changed the connectivity of the environment without changing its geometry. Place cells did not care!
Come in for a thread, or click for the preprint
https://www.biorxiv.org/content/10.1101/2020.10.20.346130v1
1/
Is the door leading into your room open or locked? You must know this without checking, because you have a memory of the connectivity of your environment.
Where is this represented in the brain?
To test this in rats, we built 4 rooms connected by pushable & lockable doors:
2/
Where is this represented in the brain?
To test this in rats, we built 4 rooms connected by pushable & lockable doors:
2/
First, we found that rats, like us, can remember the connectivity status of their environment!
When we locked some of the doors (without visible changes), they would first push them as if they were open, but quickly stopped and used the open doors instead.
3/
When we locked some of the doors (without visible changes), they would first push them as if they were open, but quickly stopped and used the open doors instead.
3/
So, where is this info stored in the brain? Hippocampal place cells already encode space, so we anticipated that they would encode connectivity.
Perhaps they would over-represent the doors (compared to fake doors or boxes) as these are important connecting points? Nope!
4/
Perhaps they would over-represent the doors (compared to fake doors or boxes) as these are important connecting points? Nope!
4/
Perhaps, changing the connectivity of the environment would create global remapping? Nope (B,C below)!
Perhaps, changing the connectivity would create local remapping? Still nope (E below)!
5/
Perhaps, changing the connectivity would create local remapping? Still nope (E below)!
5/
Perhaps, fields extending through a door will remain on one side of the door once it's closed? Nope!
By the way, most fields near the doors stayed on one side of it, even when the doors were unlocked.
6/
By the way, most fields near the doors stayed on one side of it, even when the doors were unlocked.
6/
ALL these results were confirmed in another condition - "One-Way" - where doors were locked one-way only, either all clockwise or anti-clockwise.
7/
7/
In the task, rats alternated foraging in a box and transitions between boxes. To avoid confounds (speed & movement direction can influence place cell firing), we focused most analyses on the foraging.
However, place cells' spatial firing did not change between these phases.
8/
However, place cells' spatial firing did not change between these phases.
8/
In a previous experiment, place cells were found to remap next to a connectivity change in a Tolman detour task:
https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1460-9568.2011.07653.x
We do not find this. Perhaps these findings are due to the addition of a barrier, which changes the geometry *as well as* the connectivity.
9/
https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1460-9568.2011.07653.x
We do not find this. Perhaps these findings are due to the addition of a barrier, which changes the geometry *as well as* the connectivity.
9/
Our results also seem to contradict the view of the hippocampus as a predictive map (Successor Representation model of place cells): https://www.nature.com/articles/nn.4650
Indeed, this model would predict changes in place cell firing when connectivity (available future states) changes.
10/
Indeed, this model would predict changes in place cell firing when connectivity (available future states) changes.
10/
Conclusion: dorsal CA1 place cells do not encode changes of spatial connectivity in their firing location or firing rate.
Where else might this be encoded then? This is for future experiments to find out but could be in CA3, replay, grid cells, prefrontal cortex...
11/
Where else might this be encoded then? This is for future experiments to find out but could be in CA3, replay, grid cells, prefrontal cortex...
11/
Credits to:
- the magical analysis & visualisation skills of @rmgrieves (co-1st author)
- hard work of @AnyiLiu1 particularly, but also Selim, @NilsNyberg_ , Adam, Jo
- precious input from @JulieM_Lefort, @drkjjeffery, @summerfieldlab, @GiovanniPezzulo, @FranDonnarumma and..
12/
- the magical analysis & visualisation skills of @rmgrieves (co-1st author)
- hard work of @AnyiLiu1 particularly, but also Selim, @NilsNyberg_ , Adam, Jo
- precious input from @JulieM_Lefort, @drkjjeffery, @summerfieldlab, @GiovanniPezzulo, @FranDonnarumma and..
12/
... of course, the awsome supervision, positivity, support and insight of @hugospiers!
Last but not least, this would not have been possible without our rat subjects, the cutest and smartest
13/
Last but not least, this would not have been possible without our rat subjects, the cutest and smartest
13/
Thanks for reading and please do not hesitate to leave feedback, comments or ask questions! This is what Twitter is for after all 
Another thread will come on the other main results from this experiment - the global coding part!
end/
Another thread will come on the other main results from this experiment - the global coding part!
end/
