PS: I also posted an abbreviated version on twitter so you just may have to read both x.com/savuzmir/sta...
please like, share, subscribe, and hit the notification bell
21/ It goes without saying this work is the child of a lot of years of hard work by a lot of very talented and amazing people:
@timothymulleridonthaveyourtag
@elena-gtrrz.bsky.social
@lhuntneuro.bsky.social
@behrenstimb.bsky.social
@idontthinkjamesandstevehavebluesky
20/ piece of chocolate
19/ ... value-based decision-making focused on the frontal lobe and spatial navigation research focused on the temporal lobe, and describe them in a more parsimonious way
It seems both rely on deeply overlapping computations (it’s almost as if the lobes were anatomically connected).
18/ ... how does this all fit together?
(i dont have an answer for that sorry)
Overall though, if the results from this paper are remotely true, it may be the case that we can start bridging two big fields of research that have, for the most part, not interacted much historically
17/ Rodent work has shown neural replay in mPFC [Kaeffer 2020 Neuron] and ppl have also reported ripples in mPFC [Khodagholy 2017 Science]
We didn't have enough cells on a per-session basis to look for funkier replay-like patterns in our data, but this made us think ...
16/ The last thing worth saying - we were quite puzzled by how different the raw LFP signal in vmPFC looked like compared to other channels recorded at the same time.
Often the LFP exhibited “ripple-like” patterns using sharp wave ripple detectors commonly used in hippocampus world:
15/ ... to the generation and transformation of these representations to support economic choice on a trial by trial basis ...
However. The RTs in this task were really quick so it's difficult to conclude this was *exclusively* theta-related.
We're still looking into this with a lot more data now
14/ A hex signal in vmPFC cells appeared at the trough of the underlying θ oscillation and then disappeared. Approx. 200 ms later, a chosen value signal appeared (roughly 1 θ cycle later)
This could imply input from another brain region (you know who you are) contributes ...
13/ ... the vmPFC cell population had a rhythmic firing pattern:
So you may not be surprised to hear that the same hex signals we found in the LFP were also there in cells ...
12/ vmPFC is difficult to record from and cells there didn’t really seem to do much when we plotted the average firing rates (I'll let you infer which color is vmPFC)
However ... it turned out that when we aligned firing rates in vmPFC to LFP θ frequency phase (hi θθθ gang pt. 2) ...
11/ This is exciting because our LFP hex signals behaved similarly to what you’d expect from grid cells, providing additional face validity for the hex signals people have been reporting for the last 16 years.
We also obviously looked at what was going on in the cells across the brain regions.
10/ Why am I mentioning stimuli sets? A niche design choice allowed us to ask something very specific. In spatial navigation, grid cell orientations remap between environments
We found a similar pattern in our data. Orientations within stim sets were more similar compared to across stim sets.
9/ In the simultaneous dataset, stimulus sets changed frequently. Subjects had to learn image --> magnitude/probability associations every couple of days in a separate task
(if you are interested in value learning dynamics in this separate task, stay tuned for a preprint soon (wink wink))
8/ This suggest a reason why Bongioanni et al. found altered estimation of novel choice values when stimulating similar chunks of the brain, may have been due to the subjects' inability to construct such maps ... exactly the kind of deficits you'd expect from not being able to use cogmaps
7/ These hexagonal periodic code have been thought to arise from grid cells (which some people think are related to cognitive maps)
We find these codes in 2 datasets where options were presented simultaneously or sequentially.
In both, the code was present in LFP θ frequency (hi θθθ gang)
6/ What we found in the Cell paper:
Figure 2:
When subjects pick between two options, each associated with reward magnitude and probability, vmPFC embeds these options as “locations” in a 2D choice attribute space and represents the trajectory between these options with a hexagonal periodic code
5/n ... and figuring this out may bring us closer to a cellular blueprint for understanding how PFC organizes abstract knowledge *and* uses it during choice.
But this requires ephys with the electrodes in the right brain regions ... which we happened to have had lying around on some old USB stick.
4/n
We thought vmPFC might represent reward magnitude and probability in a cognitive map during *CHOICE* considering papers like [Bongioanni 2021 Nature, Park et al. 2021 Nat Neuro.
The idea was: such signals during choice would imply we might want to rethink our process models of choice ...
3/n
A big neuroscience theme for the last 20 years has been the research of cognitive maps and grid codes.
We even wrote a brief review for SFN, covering the evidence for these representations in PFC. shorturl.at/caQ3O
In the Cell paper, we wanted to ask if we can frame choice as "navigation"
2/n
b) highlight @elena-gtrrz.bsky.social is presenting an update to this story TODAY AT COSYNE IN LISBON LIMITED TICKETS AVAILABLE (1-193 "Grid cells in the frontal cortex represent abstract knowledge", 8:30-11:30PM)
c) reward you with a piece of chocolate on post #20
1/n
With Cosyne around the corner, I thought I would:
a) make a thread about our “A cognitive map for value-guided choice in the vmPFC” Cell paper (i'm a bit late). shorturl.at/DBA3H
bsky.app/profile/sebv...
5/ If you feel we missed something, please feel free to message here to add visibility to your work and highlight it!
I hope people find this review useful in their future work :).
Do reach out if you have any questions or comments and come see us at SfN today!
4/ We hope this review will provide a common anchor for interpreting what is happening in PFC across species. Because of the space constraints, we could not cover all the cool work that is going and focused on work relevant to the symposium :)
3/ We tried giving equal amounts of space to the amazing research done in human, non-human primate, and rodent research. The main point of the review is to highlight how ubiquitous representations of cognitive maps in PFC are, as well as to showcase how diverse they can be.
2/ PFC is crucial for flexible behaviour but at the same time what it mechanistically does is not exactly clear. Our main goal with this review was to cover some of the key work that inspired this symposium and our own work.
1/ Excited to share that our symposium review covering cognitive mapping in the PFC just came out in the Journal of Neuroscience: www.jneurosci.org/content/45/4...
