Joao Barbosa
@jbarbosa.org
INSERM group leader @ Neuromodulation Institute and NeuroSpin (Paris) in computational neuroscience.
How and why are computations enabling cognition distributed across the brain?
Expect neuroscience and ML content.
jbarbosa.org
How and why are computations enabling cognition distributed across the brain?
Expect neuroscience and ML content.
jbarbosa.org
I think I will miss a meeting at 17.30 Nov 6...🙈
October 9, 2025 at 11:44 AM
I think I will miss a meeting at 17.30 Nov 6...🙈
Finally, Jeanne Sentenac will show how low rank RNNs provide a really neat explanation for information selection within 'orthogonal subspaces' - purely theoretical work but heavily inspired by mattpanichello.bsky.social beautiful paper
This will be P II 58
This will be P II 58
September 26, 2025 at 4:57 PM
Finally, Jeanne Sentenac will show how low rank RNNs provide a really neat explanation for information selection within 'orthogonal subspaces' - purely theoretical work but heavily inspired by mattpanichello.bsky.social beautiful paper
This will be P II 58
This will be P II 58
Philipp Werthmann will use low-rank RNN fit to large scale recordings, to show that even when 'everything is everywhere', different regions have different functions. A lot of push back on the everything is everywhere mantra, as you can see 🤓 @pessoabrain.bsky.social @benhayden.bsky.social
P III 13
P III 13
September 26, 2025 at 4:57 PM
Philipp Werthmann will use low-rank RNN fit to large scale recordings, to show that even when 'everything is everywhere', different regions have different functions. A lot of push back on the everything is everywhere mantra, as you can see 🤓 @pessoabrain.bsky.social @benhayden.bsky.social
P III 13
P III 13
Lubna Abdul Parveen will present her PhD project on analysis of errors in the Mante task and decoding analyses form 6 regions of the monkey brain. She will show clear evidence that everything is not always everywhere.
This will be P IV 22.
This will be P IV 22.
September 26, 2025 at 4:57 PM
Lubna Abdul Parveen will present her PhD project on analysis of errors in the Mante task and decoding analyses form 6 regions of the monkey brain. She will show clear evidence that everything is not always everywhere.
This will be P IV 22.
This will be P IV 22.
On the second (Mon), I will show how low-rank RNN fit to data are a useful tool to ask hard questions, such as studying communication subspaces from large scale recordings or the effect of neuromodulators on population dynamics.
This workshop is organized by @rdgao.bsky.social and Manuel Brenner.
This workshop is organized by @rdgao.bsky.social and Manuel Brenner.
September 26, 2025 at 4:57 PM
On the second (Mon), I will show how low-rank RNN fit to data are a useful tool to ask hard questions, such as studying communication subspaces from large scale recordings or the effect of neuromodulators on population dynamics.
This workshop is organized by @rdgao.bsky.social and Manuel Brenner.
This workshop is organized by @rdgao.bsky.social and Manuel Brenner.
On the first workshop talk (Monday), I'll make the case that everything might be everywhere when you decode, but that is not the end of the story: different regions have very different dynamics and encoding geometries.
This workshop is organized by @aitormg.bsky.social and @ackurth.bsky.social
This workshop is organized by @aitormg.bsky.social and @ackurth.bsky.social
September 26, 2025 at 4:57 PM
On the first workshop talk (Monday), I'll make the case that everything might be everywhere when you decode, but that is not the end of the story: different regions have very different dynamics and encoding geometries.
This workshop is organized by @aitormg.bsky.social and @ackurth.bsky.social
This workshop is organized by @aitormg.bsky.social and @ackurth.bsky.social
Very cool, congratulations!
Ps: that behavioral map looks like the map of Paris 🤓
Ps: that behavioral map looks like the map of Paris 🤓
September 26, 2025 at 6:23 AM
Very cool, congratulations!
Ps: that behavioral map looks like the map of Paris 🤓
Ps: that behavioral map looks like the map of Paris 🤓
Just showing off how my office looks when we DON'T work until late 😎
September 18, 2025 at 4:48 PM
Just showing off how my office looks when we DON'T work until late 😎
Just showing off how my office looks when I work until late🤓
September 15, 2025 at 9:45 PM
Just showing off how my office looks when I work until late🤓
We asked (and partially answered?) this question during covid in this paper
jbarbosa.org/files/Barbos...
Back then @yaelniv.bsky.social's repo was the best
nimh-dsst.github.io/OpenCogData/
jbarbosa.org/files/Barbos...
Back then @yaelniv.bsky.social's repo was the best
nimh-dsst.github.io/OpenCogData/
September 15, 2025 at 8:46 PM
We asked (and partially answered?) this question during covid in this paper
jbarbosa.org/files/Barbos...
Back then @yaelniv.bsky.social's repo was the best
nimh-dsst.github.io/OpenCogData/
jbarbosa.org/files/Barbos...
Back then @yaelniv.bsky.social's repo was the best
nimh-dsst.github.io/OpenCogData/
It's sad to leave Berlin behind but the move to Frankfurt came with no more online Bernstein? Why? 😭
@bernsteinneuro.bsky.social
@bernsteinneuro.bsky.social
September 14, 2025 at 8:29 AM
It's sad to leave Berlin behind but the move to Frankfurt came with no more online Bernstein? Why? 😭
@bernsteinneuro.bsky.social
@bernsteinneuro.bsky.social
I recently learned: w/ lesioned 8A, you can do many WM tasks but not one👇
Guess what happens when you decoding from 8A during each of these tasks? They are all the same.
Decoding is like a quality check, it provides almost no info about function
scholar.google.com/citations?vi...
Guess what happens when you decoding from 8A during each of these tasks? They are all the same.
Decoding is like a quality check, it provides almost no info about function
scholar.google.com/citations?vi...
September 6, 2025 at 3:03 AM
I recently learned: w/ lesioned 8A, you can do many WM tasks but not one👇
Guess what happens when you decoding from 8A during each of these tasks? They are all the same.
Decoding is like a quality check, it provides almost no info about function
scholar.google.com/citations?vi...
Guess what happens when you decoding from 8A during each of these tasks? They are all the same.
Decoding is like a quality check, it provides almost no info about function
scholar.google.com/citations?vi...
Sad to miss #CCN2025. It will be the 1st conference where a PhD working w/ me will speak 😭
go see Lubna's talk (Friday) about distributed neural correlates of flexible decision making in 🐒,
work done in collaboration w/ @scottbrincat.bsky.social @siegellab.bsky.social & @earlkmiller.bsky.social
go see Lubna's talk (Friday) about distributed neural correlates of flexible decision making in 🐒,
work done in collaboration w/ @scottbrincat.bsky.social @siegellab.bsky.social & @earlkmiller.bsky.social
August 10, 2025 at 3:56 PM
Sad to miss #CCN2025. It will be the 1st conference where a PhD working w/ me will speak 😭
go see Lubna's talk (Friday) about distributed neural correlates of flexible decision making in 🐒,
work done in collaboration w/ @scottbrincat.bsky.social @siegellab.bsky.social & @earlkmiller.bsky.social
go see Lubna's talk (Friday) about distributed neural correlates of flexible decision making in 🐒,
work done in collaboration w/ @scottbrincat.bsky.social @siegellab.bsky.social & @earlkmiller.bsky.social
He's right, this is PhD level (lack) of math skills
August 8, 2025 at 8:57 AM
He's right, this is PhD level (lack) of math skills
"it's completely b-less!" 😂
August 8, 2025 at 8:10 AM
"it's completely b-less!" 😂
Hey look, chat gpt 5 is still an idiot.
It would be great to have a control condition for tech hyped BS:
Say openAI just copy pastes gpt N and says it's N+1. How many would say it's amazing?
It would be great to have a control condition for tech hyped BS:
Say openAI just copy pastes gpt N and says it's N+1. How many would say it's amazing?
August 8, 2025 at 12:21 AM
Hey look, chat gpt 5 is still an idiot.
It would be great to have a control condition for tech hyped BS:
Say openAI just copy pastes gpt N and says it's N+1. How many would say it's amazing?
It would be great to have a control condition for tech hyped BS:
Say openAI just copy pastes gpt N and says it's N+1. How many would say it's amazing?
Only a metal band, of the best kind, chooses a text like this for the press release of their new album ✊🖤
August 1, 2025 at 1:16 PM
Only a metal band, of the best kind, chooses a text like this for the press release of their new album ✊🖤
A senior non human primate researcher by chatgpt
July 24, 2025 at 10:27 PM
A senior non human primate researcher by chatgpt
July 23, 2025 at 8:14 PM
the whole blog post is a bit weird, but this in particular makes no sense 😂 this simply shows that you can get attractors in an RNN
July 8, 2025 at 3:44 PM
the whole blog post is a bit weird, but this in particular makes no sense 😂 this simply shows that you can get attractors in an RNN
how is this a proof that attractors are not necessary? this just shows that trained RNN develop attractors
July 8, 2025 at 3:39 PM
how is this a proof that attractors are not necessary? this just shows that trained RNN develop attractors
Yes! It's this one. But does this mean it didn't work?
July 7, 2025 at 9:32 PM
Yes! It's this one. But does this mean it didn't work?