Sign in

Olivier Codol

@oliviercodol.bsky.social
1.6K followers 262 following 143 posts

Neuroscience, RL for motor learning, neural control of movement, NeuroAI. Opinions stated here are my own, not those of my employer.

PostsRepliesMedia
Olivier Codol @oliviercodol.bsky.social · 19/07/2026
I was going to ask if this is a first period or second period statement 😂
010
Reposted by Olivier Codol
Andrew Pruszynski @andpru.bsky.social · 13/04/2026
For those going to @ncmsociety.bsky.social annual meeting next week, come check out our posters! @diedrichsenjorn.bsky.social @gribblelab.org. I know that @mnlmrc.bsky.social @sivanjo.bsky.social @alighavampour.bsky.social @arminpanjehpour.bsky.social and Amin are excited show you what they've done.
03111
Reposted by Olivier Codol
Juan Gallego @juangallego.bsky.social · 14/04/2026
What are the real promises and looming perils of neural foundation models? 🧠 I put my thoughts on (virtual) paper for @thetransmitter.bsky.social following a very energised workshop at @cosynemeeting.bsky.social 2025. It's also my first piece for them 😊
02210
Reposted by Olivier Codol
Roy Eyono @royeyono.bsky.social · 19/03/2026
How do neural circuits in the brain implement normalization? 🧠 In our new paper, we show that just normalizing sensory input isn't enough. Crucially, we must also normalize the error signals! 🧵👇 Paper: arxiv.org/abs/2603.17676
arxiv.org
Inhibitory normalization of error signals improves learning in neural circuits
Normalization is a critical operation in neural circuits. In the brain, there is evidence that normalization is implemented via inhibitory interneurons and allows neural populations to adjust to chang...
16922
Reposted by Olivier Codol
Matt Perich @mattperich.bsky.social · 17/03/2026
New paper! We introduce JEDI, Jointly Embedded Dynamics Inference for neural dynamics. arxiv.org/abs/2603.10489. JEDI flexibly infers dynamical principles (across behaviors/contexts) from neural population data through RNNs constrained at single-neuron resolution to reproduce that data.
arxiv.org
JEDI: Jointly Embedded Inference of Neural Dynamics
Animal brains flexibly and efficiently achieve many behavioral tasks with a single neural network. A core goal in modern neuroscience is to map the mechanisms of the brain's flexibility onto the dynam...
14415
Olivier Codol @oliviercodol.bsky.social · 17/03/2026
That’s wonderful news! Congrats and welcome!
001
Reposted by Olivier Codol
Charlotte Volk @charlottevolk.bsky.social · 11/03/2026
Excited to be at #Cosyne2026! I'll be presenting my poster tomorrow during the Thursday session: [1-106] “A biologically inspired predictive coding algorithm with multiplicative and additive feedback”. With @colin-bredenberg.bsky.social and @tyrellturing.bsky.social
0304
Reposted by Olivier Codol
Matt Perich @mattperich.bsky.social · 10/03/2026
New paper hot off the (pre-)press! We dig into the evolutionary origins of neural computations for behavioral control across mice, monkeys, and humans: www.biorxiv.org/content/10.6.... As our lab's first foray into comparative analysis of neural dynamics, I’m super excited about this work! 1/18
614148
Reposted by Olivier Codol
Patrick Mineault @patrickmineault.bsky.social · 08/03/2026
I want to write a fun little post on what we've learned in neuroscience in the last 20 years. What are the most interesting results you can think of? Biggest trends?
10327
Reposted by Olivier Codol
Anton Sobinov @ansobinov.bsky.social · 05/03/2026
New paper out in Nature Communications about how object identity information evolves across sensorimotor cortex through grasp — and I am really happy with how it turned out. www.nature.com/articles/s41...
nature.com
Evolution of object identity information in sensorimotor cortex throughout grasp - Nature Communications
How the brain maintains object representations during grasping, when complex sensory input rapidly changes, remains poorly understood. Here the authors show that object-identity signals shift and stre...
1123
Olivier Codol @oliviercodol.bsky.social · 31/01/2026
Do you observe these issues with value functions instead of Q-values? I’ve had much less issues with value functions than Q-value based algorithms in the past and wondering if this relates
000
Reposted by Olivier Codol
Jonathan A. Michaels @jonathanamichaels.bsky.social · 13/01/2026
The Neural Control and Computation Lab is seeking a skilled part-time software engineer in Toronto to lead the development of ATHENA (Automatically Tracking Hands Expertly with No Annotations), our open-source, Python-based toolbox for 3D markerless tracking! www.yorku.ca/health/resea...
yorku.ca
01712
Reposted by Olivier Codol
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 21/12/2025
Why motor learning involves multiple systems: an algorithmic perspective www.biorxiv.org/content/10.64898/20…
084
Reposted by Olivier Codol
Blake Richards @tyrellturing.bsky.social · 18/12/2025
Great to see this paper on sensory expectations in motor control from @jonathanamichaels.bsky.social and @andpru.bsky.social out in Nature today! www.nature.com/articles/s41... #neuroscience 🧪
nature.com
Sensory expectations shape neural population dynamics in motor circuits - Nature
Experiments with human volunteers and macaques show that expectations produced by probabilistic cueing of future sensory inputs shape motor circuit dynamics in order to increase the efficiency of move...
14512
Olivier Codol @oliviercodol.bsky.social · 15/12/2025
I’m trying really hard to narrow down who is behind this every-Montreal-cycling-lanes masterpiece of a suit
130
Reposted by Olivier Codol
Andrew Pruszynski @andpru.bsky.social · 02/12/2025
Join us for Fall 2026. In our group, you can run studies from human behavior and neuroimaging, to large-scale NHP ephys, and join them up with a robust computational foundation. Bonus: you can help build the reading list.
13829
Olivier Codol @oliviercodol.bsky.social · 25/11/2025
Wow and in winter, which is even more beautiful!
010
Reposted by Olivier Codol
Jörn Diedrichsen @diedrichsenjorn.bsky.social · 24/11/2025
The Sensorimotor Superlab with @gribblelab.org and @andpru.bsky.social is a unique place to work and learn. We are now accepting MSc and PhD applications for Fall 2026. Join our awesome team at Western University... For application instructions see diedrichsenlab.org and gribblelab.org/join.html!
diedrichsenlab.org
Diedrichsenlab
13124
Reposted by Olivier Codol
Juan Gallego @juangallego.bsky.social · 13/11/2025
Come share your passion about motor control, sensory systems, neurophysiology, neurotechnology, and more at #NCMKobe26 !!
0153
Olivier Codol @oliviercodol.bsky.social · 12/11/2025
As always, thank you to my kind friends and mentors along the way, who make my journey not only possible but also fun and fulfilling.
000
Olivier Codol @oliviercodol.bsky.social · 12/11/2025
In my free time, I am wrapping up (a lot of) work and projects with former colleagues and friends. I will be communicating these as they come, so stay tuned!
100
Olivier Codol @oliviercodol.bsky.social · 12/11/2025
While I'm sad to step away from my full-time academic work, the first few months have been fantastic—I'm enjoying doing exciting research at the scale possible in such an ambitious team and company. There's a lot to learn and I'm grateful for my inclusive colleagues enabling this experience.
110
Olivier Codol @oliviercodol.bsky.social · 12/11/2025
Happy to announce that as of this summer, I've joined the CTRL-Labs group at Meta Reality Labs as a Research Scientist! I've also relocated to the bustling city of New York, where I hope I can do my best work (and enjoy running in Central Park).
The view a Research Scientist may enjoy running in Central Park
3220
Reposted by Olivier Codol
Mitchell Ostrow @neurostrow.bsky.social · 10/11/2025
Our next paper on comparing dynamical systems (with special interest to artificial and biological neural networks) is out!! Joint work with @annhuang42.bsky.social , as well as @satpreetsingh.bsky.social , @leokoz8.bsky.social , Ila Fiete, and @kanakarajanphd.bsky.social : arxiv.org/pdf/2510.25943
47124
Reposted by Olivier Codol
Fred Crevecoeur @fredcrevecoeur.bsky.social · 10/11/2025
🚨🚨 We're hiring !! Looking for postdoc? Come work in an international, collaborative and stimulating environment on mechanisms of human upper limb motor control 👇👇👇 euraxess.ec.europa.eu/jobs/386645
euraxess.ec.europa.eu
Postdoc Position on Systems Neuroscience, Motor Control at UCLouvain, Belgium
Project Title: Multi-disciplinary, multi-lab investigations of the neural bases of human sensorimotor control Project Description:
078
Reposted by Olivier Codol
Juan Gallego @juangallego.bsky.social · 07/11/2025
A very nice contribution to the field, adding more evidence on how our expectations and goals shape upcoming motor commands. Congrats to the wonderful team!
1122
Reposted by Olivier Codol
Shahab Bakhtiari @shahabbakht.bsky.social · 07/11/2025
I’m looking for interns to join our lab for a project on foundation models in neuroscience. Funded by @ivado.bsky.social and in collaboration with the IVADO regroupement 1 (AI and Neuroscience: ivado.ca/en/regroupem...). Interested? See the details in the comments. (1/3) 🧠🤖
ivado.ca
AI and Neuroscience | IVADO
14624
Olivier Codol @oliviercodol.bsky.social · 07/11/2025
Yes! The advantages are much clearer wrt neural computation (memory, expressivity, and gradient propagation) than for exploration per se.
020
Olivier Codol @oliviercodol.bsky.social · 07/11/2025
Learning through motor noise (exploration) is well documented in humans (lots of cool work from Shadmehr and @olveczky.bsky.social) but the scale is rather small. Here if the dynamical regime helps exploration I’d say it should be within these scales as well.
110
Olivier Codol @oliviercodol.bsky.social · 07/11/2025
That being said this is not how we move (execute movements) and in that sense this is a model of learning rather control.
110
Olivier Codol @oliviercodol.bsky.social · 07/11/2025
I would say yes it’s possible. Particularly because a deviation is carried over instead of collapsing back, so the filtering function that non linear muscle activations have will not impact it as much as white noise.
120
Olivier Codol @oliviercodol.bsky.social · 07/11/2025
As in if the edge of chaos regime is a consequence or if it is a cause of RL’s need for exploration?
110
Reposted by Olivier Codol
Matt Perich @mattperich.bsky.social · 06/11/2025
If you're interested in dynamical systems analysis for neuroscience, definitely check out @oliviercodol.bsky.social 's revised version of our RL paper! Very cool results in the new Fig 6, worth it regardless of if you saw our previous version or if it's all new. www.biorxiv.org/content/10.1...
biorxiv.org
Brain-like neural dynamics for behavioral control develop through reinforcement learning
During development, neural circuits are shaped continuously as we learn to control our bodies. The ultimate goal of this process is to produce neural dynamics that enable the rich repertoire of behavi...
03711
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
As always a huge thank you to my colleagues and supervisors @glajoie.bsky.social @mattperich.bsky.social and @nandahkrishna.bsky.social for helping make this work what it is—and making the journey so fun and interesting
060
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
We’re pleased to see RL's role in neural plasticity is increasingly under focus in the motor control community (check out @adrianhaith.bsky.social's latest piece!) I strongly believe motor learning is sitting at the interface of many plasticity mechanisms and RL is an important piece of this puzzle.
160
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
Along the above, we add discussion points that I hope will clarify some of our stance on the topic of RL in neuroscience and acknowledge some past important work that we believe our study complements. We also add several important controls (particularly Figs. S8, S14). Feel free to check it all out!
130
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
“Edge of chaos” dynamics are long recognized as a computationally potent dynamical regime that avoids vanishing gradients during learning and allows greater memory and expressivity of a system. This stark difference surprised us, and we think it can help explain our results on neural adaptation.
241
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
Indeed, Lyapunov exponents show that fixed points for RL models largely stay near 0, showing these networks’ dynamics lie at the edge of chaos. Whereas SL models’ dynamics are contractive and orderly, keeping very little information in memory for long and having stereotyped expressivity.
151
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
Does this mean SL models are very orderly, while RL models lie at the interface between order and chaos? To formally confirm, we looked at Lyapunov exponents, which tell us how fast close-by states diverge. Unlike Jacobians, this tells us about long-horizon, not just local, dynamics.
120
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
We looked at local dynamics around fixed points over time. This showed that SL models’ fixed points are indeed very stable, having nearly all modes of their eigenspectrum <1. RL models showed many more self-sustaining modes ≈1, again demonstrating isometric dynamics.
130
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
A dynamical system could recover perfectly against a state perturbation, or it could expand following that perturbation. It turns out supervised learning (SL) models do the former, while reinforcement learning (RL) models do something in-between; they act as isometric systems.
130
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
But a biological brain receives an ever-changing stream of inputs, rarely ever reducing to steady-state inputs. Our models reflect that, and their inputs are time varying. So we took a slightly different approach, and asked how fixed-points evolved over time and over perturbed neural states.
130
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
Usually, one determines where neural activity naturally settles under a steady-state input regime to find “fixed-point” neural states. Local dynamics around these points provides valuable information about how neural networks process information—that is, what they compute, and how.
130
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
But alignment metrics can overlook the question of what gives rise to the differences they capture. We approached this using a now established framework in systems neuroscience, dynamical systems theory.
120
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
This similarity to NHP neural recordings was true for geometric similarity metrics (CCA), but also for dynamical similarity. Importantly, this was only evident when our models were trained to control biomechanistically realistic effectors.
140
Olivier Codol @oliviercodol.bsky.social · 06/11/2025
A tad late (announcements coming) but very happy to share the latest developments in my previous preprint! Previously, we show that neural representations for control of movement are largely distinct following supervised or reinforcement learning. The latter most closely matches NHP recordings.
1468
Reposted by Olivier Codol
U15 Canada @u15ca.bsky.social · 03/11/2025
Western study reveals brain’s predictive power. Researchers find neurons don’t wait for challenges to arise, they plan how to react. news.westernu.ca/2025/10/sens... #CdnPSE @westernu.ca
news.westernu.ca
Study reveals brain’s predictive power
New study reveals that we rely on sensory expectations to get prepared for unexpected disturbances, helping us react faster and more accurately.
093
Reposted by Olivier Codol
Juan Gallego @juangallego.bsky.social · 04/11/2025
🚨Job alert🚨 The lab has up to *3 postdoc openings* for comp systems neuroscientists interested in describing and manipulating neural population dynamics mediating behaviour This is part of a collaborative ARIA grant "4D precision control of cortical dynamics" euraxess.ec.europa.eu/jobs/383909
euraxess.ec.europa.eu
3 Postdoctoral Research Fellows
Champalimaud Foundation (Fundação D. Anna de Sommer Champalimaud e Dr.
27951
Reposted by Olivier Codol
Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 04/11/2025
What are the salient aspect of the LLM RL problem that could be abstracted into a benchmark? Tremendously large action space, sensitivity to numerical precision, prior policy that you must remain close to, what else?
373