iiinova — where your digital self keeps evolving

Reposted by Valentin Schmutz

The Transmitter @thetransmitter.bsky.social · 12/10/2026

In two 1926 articles, Edgar Douglas Adrian revealed that neurons communicate through trains of spikes. @carandinilab.net describes the remarkable discovery and how it shifted focus from the medium (the nerve) to the message (the neural code). #neuroskyence www.thetransmitter.org/history-of-n...

thetransmitter.org

The neural code at 100

In two 1926 articles, Edgar Douglas Adrian revealed that neurons communicate through trains of spikes.

16231

Reposted by Valentin Schmutz

Matthijs Pals @matthijspals.bsky.social · 02/10/2026

What kind of dynamics underlie sequence working memory? In the last project of my PhD we developed and analysed RNNs fitted to multi-session single-unit data of macaques. Now out on: www.biorxiv.org/content/10.6... With @jakhmack.bsky.social and data from Chen et al., Neuron 2024. [1/4]

14820

Reposted by Valentin Schmutz

Soledad Gonzalo Cogno @soledad-gcogno.bsky.social · 18/09/2026

📣 🔥 🤩 Our lab is recruiting a PhD student! If you want to do a PhD in computational neuroscience come and join us at the @kavlintnu.bsky.social ! 🧠 💻 For questions and info, the lab will be at Bernstein in a few days 😊 ⏰ Apply before October 11: www.jobbnorge.no/en/available...

jobbnorge.no

PhD in Computational Neuroscience (307990) | NTNU - Norwegian University of Science and Technology

Job title: PhD in Computational Neuroscience (307990), Employer: NTNU - Norwegian University of Science and Technology, Deadline: Sunday, October 11, 2026

02210

Valentin Schmutz @bio-emergent.bsky.social · 11/09/2026

I agree with this last post. My guess is that the authors of the report would agree as well.

001

Valentin Schmutz @bio-emergent.bsky.social · 11/09/2026

I don't know what the vested interests of METR are. What are they?

100

Valentin Schmutz @bio-emergent.bsky.social · 11/09/2026

Thanks for sharing. Thoughtful opinion piece. However, a short opinion piece can't be compared to a long report. This article only contains very superficial information on what happened (it's an opinion piece!). I still recommend reading the METR report. Much more comprehensive.

100

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

How can we describe of the behaviour of 1300 interacting LLM agents with near-human capabilities, working full-time at super-human speed over the course of several days without using some form of anthropomorphic language? Do we have another choice?

100

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

We, computational neuroscientists, have a hard time understanding and explaining what's happening in relatively small networks of relatively simple neurons.

200

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

That's a very interesting point. I still think the authors made the right call when they chose to lean into the anthropomorphic language the agents were using. I see it as a practical decision. How else could they have told the story?

200

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

I agree, but that's the best analysis we have so far. The report explicitly says that an analysis verifiable in detail by human experts would have yielded more reliable results. I would love to read such an analysis of this case and of similar future cases. I am not 100% confident I will be able to.

110

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

The company is clearly responsible for the incident. In OpenAI's own report, they acknowledge that a few things went wrong during the training of the models involved in the attack. Designing safe training environments doesn't seem to be such an easy task, as the models become more capable.

100

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

Regarding the anthropomorphic language, the report presents many raw quotes from the AI agents' communication and reasoning traces; many of these raw quotes are much more anthropomorphic than the authors' description of the events.

100

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

Given the time constraints the three authors were working under, they had to rely heavily on GPT-5.6 Sol – one of the very models involved in the attack – for their analysis. They fully acknowledge this, as well as the unreliability of LLM-based analysis.

110

Valentin Schmutz @bio-emergent.bsky.social · 07/09/2026

metr.org/blog/2026-08... Nothing has informed me more about the world we are living in than this 91-page report on the OpenAI/Hugging Face incident. This report describes a real case of AIs getting out of the box, while their actions and reasoning are still somewhat understandable to humans...

metr.org

Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident

Two METR staff members and a Redwood Research contractor investigated an incident in which OpenAI agents coordinated a multi-day hack of Hugging Face on a shared unsanctioned message board.

1100

Reposted by Valentin Schmutz

Enny van Beest @ennyvb.bsky.social · 04/09/2026

Textbook: tracking position during navigation is primarily the job of the hippocampus. Recent findings suggest a more complicated story. Recording >20k neurons brainwide, we indeed find spatial representations in every region! With @kenneth-harris.bsky.social and @carandinilab.net

Brainwide representations of a virtual corridor

416852

Valentin Schmutz @bio-emergent.bsky.social · 03/09/2026

Congratulations, Joao! This is great news for European comp neuro 🙌🏻

110

Reposted by Valentin Schmutz

The Royal Society @royalsociety.org · 29/08/2026

The Ferrier Medal and Lecture 2027 is awarded to Professor Kenneth Daniel Harris for his contributions to neuroscience by unifying experimental neurophysiology with advanced analytical methods and for pioneering methods that allow recording from thousands of neurons. #RSMedals bit.ly/4qCJZRo

A man in a black shirt smiles at the camera in front of a whiteboard covered in scientific notation.

47219

Reposted by Valentin Schmutz

David G. Clark @david-g-clark.bsky.social · 31/07/2026

This September, I will join the @flatironinstitute.org Center for Computational Neuroscience as a Group Leader, where I will launch the High-Dimensional Dynamics & Computation Group. Information for postdocs and grad students is forthcoming. I am grateful to my mentors who made this possible!

Exterior view of the Flatiron Institute in Manhattan

810411

Reposted by Valentin Schmutz

Shuqi Wang @shuqiw.bsky.social · 15/07/2026

Our paper "Rarely categorical, highly separable representations along the cortical hierarchy" is now out in Nature! www.nature.com/articles/s41... In this paper, we studied whether the brain is well-organized and interpretable, both globally and locally. (See thread below.)

nature.com

1228

Reposted by Valentin Schmutz

Louis Pezon @lpezon.bsky.social · 30/06/2026

Both brains and RNNs can re-use components of computation across similar tasks or contexts. But what exactly are those “shared components”? How can they be used to solve several tasks? We address these questions in a new preprint with @avm.bsky.social! Link: www.biorxiv.org/content/10.6...

biorxiv.org

Interpretable compositional computation with recurrent neural networks

Flexible cognition utilizes reusable components to enable rapid adaptation of behavior to different contexts or tasks. Analysis of artificial neural networks trained on multiple tasks suggested that t...

16425

Reposted by Valentin Schmutz

Jacob Zavatone-Veth @jzv.bsky.social · 23/06/2026

Tremendously excited to announce that I will be joining @rockefeller.edu as an Assistant Professor and Head of Lab starting in January 2027! My group will be broadly focused on theoretical neuroscience, and mathematical problems in neural computation in the large.

Rockefeller campus image from https://commons.wikimedia.org/wiki/File:Rockefeller_University_Campus_aerial_2.jpg, licensed under the Creative Commons Attribution-Share Alike 2.5 Generic license.

910615

Reposted by Valentin Schmutz

Kenneth Harris @kenneth-harris.bsky.social · 19/05/2026

New method for identifying the fine transcriptomic types of thousands of neurons recorded in vivo!

17222

Reposted by Valentin Schmutz

Julie Fabre @juliefabre.bsky.social · 08/04/2026

How do the basal ganglia turn what you see into what you do? New preprint w/ @kenneth-harris.bsky.social, @flickerfusion.bsky.social & @carandinilab.net: we recorded across striatum, GPe & SNr in a Go/NoGo task. Striatum encodes which stimulus, GPe & SNr encode action. 🧵 biorxiv.org/content/10.6...

16525

Reposted by Valentin Schmutz

James McAllister @jajmca.bsky.social · 02/04/2026

Really excited to share a new preprint! 𝗛𝗼𝘄 𝗱𝗼 𝗯𝗿𝗮𝗶𝗻𝘀 𝘀𝘁𝗮𝘆 𝗿𝗼𝗯𝘂𝘀𝘁 & 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝘁 𝘄𝗵𝗶𝗹𝗲 𝗯𝗲𝗶𝗻𝗴 𝗶𝗻𝗰𝗿𝗲𝗱𝗶𝗯𝗹𝘆 𝘀𝗽𝗮𝗿𝘀𝗲? We explore how! www.biorxiv.org/content/10.6... We built Connectome-based Neural Networks (CoNNs) using Drosophila wiring (larva&adult) & compared with random networks with same sparsity.

Constructing connectome-based neural networks

15213

Valentin Schmutz @bio-emergent.bsky.social · 30/03/2026

Ten years ago, I did a research internship with Tilo Schwalger (then in the @gerstnerlab.bsky.social) on deriving mean-field equations from networks of interacting neurons – I have been doing that ever since. I'm still very fond of the way he explains things "à la physicienne".

0160

Reposted by Valentin Schmutz

Earl K. Miller @earlkmiller.bsky.social · 19/03/2026

Cortical circuits are often thought to be specialized, but the same large-scale activity patterns can arise from different circuit architectures. In other words, different instruments can play the same tune, and the same instrument can sometimes play different tunes. www.cell.com/neuron/fullt...

cell.com

Linking neural manifolds to circuit structure in recurrent networks

Neural population activity can be described either by low-dimensional dynamics on neural manifolds or by single-neuron selectivities. Using a theoretical approach, Pezon et al. relate these two statis...

14011

Reposted by Valentin Schmutz

Alexander van Meegen @avm.bsky.social · 12/03/2026

Travelling to COSYNE seems to be the perfect opportunity to announce that I started my own lab at RWTH Aachen University earlier this year, funded by NRW's Ministry of Culture and Science through its Return Program. If you are at COSYNE and want to chat please reach out!

5335

Reposted by Valentin Schmutz

PessoaBrain @pessoabrain.bsky.social · 08/03/2026

𝗪𝗵𝗮𝘁'𝘀 𝘁𝗵𝗲 𝗿𝗲𝗹𝗮𝘁𝗶𝗼𝗻𝘀𝗵𝗶𝗽 𝗯𝗲𝘁𝘄𝗲𝗲𝗻 𝗺𝗮𝗻𝗶𝗳𝗼𝗹𝗱𝘀 𝗮𝗻𝗱 𝗿𝗲𝗰𝘂𝗿𝗿𝗲𝗻𝘁 𝗻𝗲𝘁𝘄𝗼𝗿𝗸𝘀 𝗶𝗻 𝘁𝗵𝗲 𝗯𝗿𝗮𝗶𝗻? This looks like a must read (suppl material bursting with goodies). #neuroskeyence doi.org/10.1016/j.ne...

17629

Valentin Schmutz @bio-emergent.bsky.social · 06/03/2026

How are neural manifolds and single-neuron response properties related to circuit structure? How degenerate are these relationships? Theory and a plethora of examples can be found in the following paper, out today in Neuron 🌟 It was a privilege to co-supervise first author @lpezon.bsky.social!

15413

Valentin Schmutz @bio-emergent.bsky.social · 21/02/2026

These workshops are absolutely amazing. I had the privilege to attend the theoretical neuroscience workshop in 2024; it was a memorable week.

0132

Reposted by Valentin Schmutz

Soledad Gonzalo Cogno @soledad-gcogno.bsky.social · 17/02/2026

Applications are now open for the summer school: 𝐌𝐚𝐭𝐡𝐞𝐦𝐚𝐭𝐢𝐜𝐚𝐥 𝐌𝐞𝐭𝐡𝐨𝐝𝐬 𝐢𝐧 𝐂𝐨𝐦𝐩𝐮𝐭𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐍𝐞𝐮𝐫𝐨𝐬𝐜𝐢𝐞𝐧𝐜𝐞 🧠 Apply before March 15: www.compneuronrsn.org 📍 Located in beautiful Eresfjord 🇳🇴 🗓️ Between July 6-24 Supported by the @kavlifoundation.org In collaboration with @kavlintnu.bsky.social

compneuronrsn.org

Mathematical Methods in Computational Neuroscience

Summer school in Eresfjord, Norway (July 8th - 26th, 2024)

06329

Reposted by Valentin Schmutz

Enny van Beest @ennyvb.bsky.social · 02/02/2026

Turns out deep neural networks are better at extracting relevant features than I am... Meet "DeepUnitMatch", a DNN version of our neuron tracking tool: www.biorxiv.org/content/10.6... It has been a privilage to continue the collaboration with Célian, and cosupervise the talented Suyash & Wentao!

biorxiv.org

0133

Valentin Schmutz @bio-emergent.bsky.social · 28/01/2026

Thanks! But is the tail exponent really a continuous function of the spectrum in L1 norm? Can't small errors in the last eigenvalues have a large effect on the estimated tail exponent? On noisy data, I would be surprised if the two-step approach you suggest would suffice to guarantee consistency...

100

Valentin Schmutz @bio-emergent.bsky.social · 28/01/2026

Estimating the power-law exponent of the tail of the eigenspectrum from noisy data is a difficult problem. Here is an approach based on eigenmoment estimation. 👏🏻 A hard stats problem remains open: Can we find an estimator of the tail exponent for which we can PROVE unbiasedness, consistency, etc.?

1133

Valentin Schmutz @bio-emergent.bsky.social · 16/01/2026

Talk on "High-dimensional neuronal activity from low-dimensional latent dynamics" at NeurIPS 2025 with @haydari.bsky.social now accessible online 🎥: neurips.cc/virtual/2025... Please reach out if you have any comments or ideas for follow-up work!

0243

Reposted by Valentin Schmutz

Alexander Ecker @aecker.bsky.social · 19/12/2025

If you recently finished your PhD in ML for life science and are looking for a new job early next year, please apply! Göttingen is a lovely town and a great scientific environment.

033

Reposted by Valentin Schmutz

David G. Clark @david-g-clark.bsky.social · 15/12/2025

Very excited about this new work from the omnipotent Owen, with me and Ashok Litwin-Kumar! Can we reconcile low- and high-dimensional activity in neural circuits by recognizing that these circuits ~multitask~? (Plausibly, yes 😊)

0314

Reposted by Valentin Schmutz

Blake Richards @tyrellturing.bsky.social · 11/12/2025

I'm more and more convinced that low-dimensional manifolds in the brain are just an artifact of the experimental designs and analyses we use... 🧠📈 🧪

1814720

Reposted by Valentin Schmutz

Ali Haydaroğlu @haydari.bsky.social · 13/12/2025

"Low-d neural activity" would probably look high-D if you plot the spectrum on a log-log scale. @engeltatiana.bsky.social made a similar point recently on @braininspired.bsky.social. Also agrees with our claim with @bio-emergent.bsky.social that low-D and high-D can be two sides of the same coin!

192

Reposted by Valentin Schmutz

Ali Haydaroğlu @haydari.bsky.social · 08/12/2025

Best part about being a scientist is the people I get to work with. Valentin (@bio-emergent.bsky.social) and I got to give a talk at NeurIPS, bridging a gap between low- and high-dim perspectives of the brain. Thankfully, the audience was (somewhat) more awake than the San Diego desert 🏜️

4324

Reposted by Valentin Schmutz

Tatiana Engel @engeltatiana.bsky.social · 05/12/2025

Thank you for having me on BrainInspired, Paul @braininspired.bsky.social! It was such an honor to be on my favorite show—a rare place where we can leisurely talk about manifolds, latent circuits, power laws, and other esoteric ideas, and still be taken seriously in knowing they are all real.

36014

Valentin Schmutz @bio-emergent.bsky.social · 05/12/2025

Tomorrow at #NeurIPS2025! Oral at 10 am in UL Ballroom 20D and poster #2016 at 11 am. @haydari.bsky.social and I are looking forward to hearing your thoughts.

0184