#NeuroESC #JournalClub
Reading Mental exploration of future choices during immobility theta oscillations

If you've read it, will you let me know what you think?

The authors look at #ThetaSequences in a working memory task in a radial arm maze. They find theta during immobility (makes sense, e.g. we saw that in our two-goals task). They also find that theta sequences might preferentially represent the next goal (also makes sense, e.g. Hippocampal theta sequences reflect current goals)!

I have only done a quick reading so far, but am confused by a few points:

• the decoding is done on all cells (pyramidals and interneurons), shouldn't it be done on pyramidal or #PlaceCells only?
• the cell counts are quite low (often less than 40 pyrs) when I would have thought at least 50 place cells would be needed for this kind of maze. I guess that shows that #Neuropixels are not the best to record from dCA1!

• the decoded algorithm itself includes a ' position transition matrix' which seems like it would bias decoding towards realistic trajectories that the rat is about to do??? (but I probably missed something)

• also, this study is very related to this other paper, which is not discussed or even cited (😕 ):
  Assembly Responses of Hippocampal CA1 Place Cells Predict Learned Behavior in Goal-Directed Spatial Tasks on the Radial Eight-Arm Maze #CsisvariLab

Let me know what you think!

#LeutgebLab #NeuroRat #Neuroscience #SpatialCognition

@elduvelle_neuro

2/2

Importantly, not preplay, which occurs before any experience, but shortcuts, which occur after experience and allow the animal to learn the connectivity of the map.

One of the useful things lost in the publication is that Gupta et al 2010 (http://www.cell.com/neuron/abstract/S0896-6273(10)00060-7) and Gupta et al 2012 (http://www.nature.com/neuro/journal/vaop/ncurrent/abs/nn.3138.html) are the same data. 2010 is #SharpWaveRipples and 2012 is #ThetaSequences. What's cool is that the theta segments the task, while SWRs link those segments together.

A short commentary by the great Anna Gillespie on awake replay:

Ruminating on replay during the awake state

#GillespieLab (https://www.gillespie-lab.com/)

Interesting bits:

"replay during sleep is generally thought to occur in the forward direction"

I believe this is true too, but has anyone actually analyzed it? It could be done on several existing datasets...

"The idea that awake replay might serve a planning role was broadly adopted by the field, even though direct experimental evidence [...] has been slow to emerge"

YES because replay is not related to immediate planning, however nice it would be if it was

"We, as a field, also need to be more precise with our terminology. For instance, what exactly should we consider evidence of ‘planning’ or ‘deliberation’ in an animal model? "

YES

Related: history of the 'genesis' of #ThetaSequences narrated in this retrospective by @adredish: Mental Time Travel: A Retrospective
(original post:
https://neuromatch.social/@adredish/113662497894186335) #RedishLab

I'd actually argue that replay, and not theta sequences, might be closer to mental time-travel... but we probably don't know enough about either to be sure yet!

#Neuroscience #HippocampalReplay #Hippocampus

📢 New work on the Integration of rate and phase codes by hippocampal cell assemblies to support the encoding of spatiotemporal context
https://www.nature.com/articles/s41467-024-52988-x

It was a pleasure to work with Nadine Becker, Aleks P. F. Domanski, Timothy Howe, Kipp Freud, @DurstewitzLab, and Matt Jones.
@SantAnnaPisa @BristolNeurosci @zi_mannheim @BernsteinNetwork
#Neuroscience #Hippocampus #ThetaSequences #PlaceCells

#JournalClub on Closed-loop modulation of remote hippocampal representations with neurofeedback, from #FrankLab: little summary + comments.

• The goal is to see if rats can deliberately reactivate internal representations, without external cues.
• The authors design a closed-loop system that does clusterless real-time decoding of position and rewards the rats when they reactivate a specific maze arm end.
  
• The rats go through a gradual training process: first getting reward on the actual maze (a T-maze), then getting reward in the home box when they orient towards the target arm of the maze [head-direction task], then getting reward in the home box when they reactivate the representation of the target maze end [neurofeedback task].
• The rat's performance is not too bad given the difficulty of the task, kudos to Rat2 who seemed to really know what he was doing.
• interestingly, the reactivations do not happen during #SharpWaveRipples, and not really during theta either, just during an uncharacterised LFP state.
• Little caveat 1: as far as I can see, the head-direction of the rats is not shown in any plot and it is not possible to say if the rats might still be doing the head-direction task during the neurofeedback task. That being said, the rats clearly reactivate the target arm more in the neurofeedback than head-direction task, which is quite convincing.

• Little caveat 2: the performance is always shown in terms of total rewards collected, and not reward rate. So in different sessions rats might reach the maximum reward but take twice the time. Showing reward rate would be more informative.

• conclusion: this is pretty cool, but we would really need to know about the head-direction.

• question for the audience: what do you think the rats "think about" during those reactivation moments??

#Neuroscience #HippocampalReplay (not really) #ThetaSequences (not really) #PlaceCells #NeuroRat #Hippocampus

@Andrewpapale I thought that too but I think #ThetaSequences tend to be much shorter, and are generally (always) relatively local no? While the example on the left here definitely looks like a replay sequence..

In any case, it’s certainly something to rule out and I’m sure the reviewers will ask them to exclude high theta power periods…

@thetransmitter
Oh yes - like this (#Hippocampal #ThetaSequences):
https://m.youtube.com/watch?v=IU8wKl5PcLU

Or this (#HippocampalReplay):
https://m.youtube.com/watch?v=q0kD2RSejRo

(source: #vanDerMeerLab)

@elduvelle_neuro @BenoitGirard @Raphael_Brito

A few notes.

1. Remember that even the #SharpWaveSequences that Pfeiffer and Foster 2013 [https://pubmed.ncbi.nlm.nih.gov/23594744/] saw going ahead of the animal is a very very small proportion. (Home was twice as likely as the other 35 options. So it's 2/37 vs 1/37 or an increase of 2.7%. ) Still real. But small. Also, lots of people have seen #SWRSequences go to places rats are not going to go. (Such as Gupta et al 2010 [https://pubmed.ncbi.nlm.nih.gov/20223204/], where it was more likely to go to the other side if the animal was NOT going there next - perhaps "keeping the map flat"?)

2. We know that in Wikenheiser and Redish 2016 (https://pubmed.ncbi.nlm.nih.gov/25559082/), #ThetaSequences on the first part of the journey go longer when the rat is going to run longer (where start of journey is same, but distance from rat to goal varies), but on the last part of the journey, the sequences are the same length (when start of journey is different, but distance from rat to goal is the same).

3. We also know that both time of second half of theta and sequences go longer the more time it will take to reach the goal. (Schmidt and Redish 2019 [https://pubmed.ncbi.nlm.nih.gov/30892976/]).

4. That all being said, @elduvelle_neuro is correct that on two-alternative-forced-choice tasks (2AFC) #ThetaSequences while deliberating do not reflect the final choice made in any way we've been able to detect, and that when they do (such as in Johnson and Redish 2007 [https://pubmed.ncbi.nlm.nih.gov/17989284/] or Kay...Frank 2020 [https://pubmed.ncbi.nlm.nih.gov/32004462/]) the rat does not seem to be deliberating anymore.

My suspicion is that this may be because of the 2AFC structure of the task, where knowing you don't want to go one way is just as informative as saying you want to go the other way. We are now starting to try to develop tasks with multiple paths to goals (so not 2AFC). @elduvelle_neuro 's cool star-shaped task is a good one for this. As are a bunch of the cool new hex tasks from the Loren Frank lab that were presented at #SFN2023 this year.

@BenoitGirard @Raphael_Brito Nicely said! #ThetaSequences are where I’ll look next actually :) However, I’m not fully sure they’ll be predicting the behaviour, and @adredish would probably tell you that they don’t, earlier in training (even if the rat ends up deciding where to go), and when they do later in training the rat is already automatized… see also work from the #JadhavLab.

On the other hand (and do correct me if I’m missing something) I’m not sure that anyone has looked at theta sequences in a place / allocentric task, so maybe we’ll see the hippocampus doing something different in that case!

@edeno cool thanks! Putting some screenshots for illustration.. it definitely seems like a lot of events happen at the choice point, but strangely, most if not all of the replay examples shown in the paper are from the reward locations.

Also, as you say, it is not clear that those events were during SWRs and not theta - the theta/delta ratio around “replays” does seem quite similar to the rest of behaviour... it would have been nice if they had made this figure separately for reward vs choice point events. Actually they even say that their choice point events could be similar to @adredish ‘s #VTE activity, which is definitely #ThetaSequences. 🤔

(For reference, here is the paper:
Awake replay of remote experiences in the hippocampus)

Happy to finally see this work published!
https://www.cell.com/neuron/fulltext/S0896-6273(23)00757-2

Hope this inspires more work on spike-timing changes in the context of #neurodegeneration.

#Neuroscience #Hippocampus #Placecells #ThetaSequences

Here’s a really cool-looking paper that I absolutely #Need2Read:

Associative and predictive hippocampal codes support memory-guided behaviors

#Hippocampus #ThetaSequences #HippocamplaReplay #FernandezRuizLab

@NicoleCRust I think that’s proof that #ThetaSequences are essential for flexible decision-making 😁
Also..I can’t think of a particular instance but when I move myself - walk, run, bike - or in the shower, I definitely feel that my mind is clearer and find it easier to take decisions.

Dear @adredish ,
as the #ThetaSequences expert - do you think theta sequences can be remote (and that, like replay, they have an initiation bias which explains why most of them are local), or that they are by nature local, constrained to stay around the rat's current position?
Or maybe the third option: they are local, but can extend to a remote place (perhaps like what you saw in Wikenheiser & Redish, 2015?

tagging @chongxi who was interested in the answer as well
#Hippocampus #NeuroRat #PlaceCells

⤴️ What are good reviews that explain the differences between #Replay and #ThetaSequences ?
Do post your own!

I'd recommend, as the most recent I know which addresses both: Decoding the cognitive map: ensemble hippocampal sequences and decision making by Wikenheiser & @adredish .
Also really like the figures in Hippocampal ripples and memory consolidation (yes, a bit old).
​Will post others as I come up with them but I also count on you!
2/2

So... what are the differences between #ThetaSequences and #Replay (in the rodent #Hippocampus ?

Here is an illustration from real data!

Fig 1 shows ~70 #PlaceCells firing while a rat runs on a maze to reach a reward. See how the activity is different between running and pausing?
Fig 2 shows possible theta sequences.
Fig 3 shows possible replay.

Spot the differences & check alt text for more info!

PS: (very simple) #Matlab code to plot this is available here !

1/2

@adredish I agree!! The most convincing support for a planning role of replay, I’d say, is the increased fit of the replay with the future rat trajectory, but I think that could be caused by a trajectory optimization / reward assignment mechanism (path integrating from current position to previous rewarded position) instead of a planning mechanism going to the future goal.
100% agree that, if anything, #ThetaSequences are more likely to contribute to planning than replay. I’m currently working on something that should help answer this in a place task!

@mrspaghetti Super interesting!!
Does it affect “basic” behaviour in any way (running speed, number /duration of pauses, rearing etc.)?
Does it affect #SplitterCell coding?
Does it affect #ThetaSequences?!
👀 👀👀